Thermal rearrangements of di- and triphenyl-substituted benzocyclobutenes and corresponding o-quinodimethanes

Article abstract of DOI:10.1002/(SICI)1099-0690(199903)1999:3<551::AID-EJOC551>3.0.CO;2-2

7,8-Dimethoxy-7,8-diphenyl- (1c), 7,8-dimethyl-7,8-diphenyl- (1d), 7- methoxy-7,8,8-triphenyl- (1e), 7-methyl-7,8,8-triphenyl- (1f), 7-isocyano- 7,8,8-triphenyl- (1g), and 7,7,8-triphenylbenzocyclobutene (1h) are amenable to a variety of thermal rearrangements following initial electrocyclic ring- opening to the corresponding 7,8-diphenyl- (2c,d) and 7,8,8-triphenyl-o- quinodimethanes (2e-h). meso-1c was found to undergo a facile meso/rac isomerization at room temperature, indicating that other processes such as a symmetry-forbidden disrotatory ring-opening or a stepwise reaction compete with the symmetry-allowed conrotatory process. An estimate of the energy profile of the 1c/2c reaction system was made by kinetic simulation in combination with oxygen trapping of the intermediate o-quinodimethanes (2c) and semiempirical PM3 calculations, and revealed that the barrier for the symmetry-forbidden pathway is merely about 4 kJ · mol^-1 higher than that for the symmetry-allowed one. o-Quinodimethanes 2c, 2g, 2e, and 2h underwent further electrocyclic hexatriene-cyclohexadiene ring-closure to give 4a,10- dihydroanthracene derivatives at temperatures between 20 and 80 °C. The 4a,10-dihydroanthracenes were further transformed to 9,10-disubstituted anthracenes by elimination of methanol or HCN, as well as to 9,10-substituted 9,10-dihydroanthracene derivatives. ESR and ENDOR spectroscopic detection of related 9-anthryl radicals lends support to the view that 9,10- dihydroanthracene products are formed by a homolytic hydrogen-transfer reaction (retrodisproportionation). By way of contrast, the aforementioned transformations play only a minor role in the case of methyl-substituted benzocyclobutenes 1d, 1f as here they are overruled by faster 1,5-H shift reactions of the corresponding o-quinodimethanes 2d, 2f, leading to styrene derivatives.

Full text of DOI:10.1002/(SICI)1099-0690(199903)1999:3<551::AID-EJOC551>3.0.CO;2-2

FULL PAPER
Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
and Corresponding o-Quinodimethanes
Thomas Paul,[a] Roland Boese,[b] Ingo Steller,[b] Heinz Bandmann,[a] Georg Gescheidt,[c]
Hans-Gert Korth,*[a] and Reiner Sustmann*[a]
In memoriam Wolfgang R. Roth
Keywords: Benzocyclobutenes / o-Quinodimethanes / Pericyclic reactions / Retrodisproportionation / 9-Anthryl radicals
7,8-Dimethoxy-7,8-diphenyl- (1c), 7,8-dimethyl-7,8-diphen-
yl- (1d), 7-methoxy-7,8,8-triphenyl- (1e), 7-methyl-7,8,8-
triphenyl- (1f), 7-isocyano-7,8,8-triphenyl- (1g), and 7,7,8-
triphenylbenzocyclobutene (1h) are amenable to a variety of
thermal rearrangements following initial electrocyclic ring-
symmetry-allowed one. o-Quinodimethanes 2c, 2g, 2e, and
2h underwent further electrocyclic hexatriene-cyclohexa-
diene ring-closure to give 4a,10-dihydroanthracene
derivatives at temperatures between 20 and 80 °C. The
4a,10-dihydroanthracenes were further transformed to 9,10-
opening to the corresponding 7,8-diphenyl- (2c,d) and 7,8,8- disubstituted anthracenes by elimination of methanol or
triphenyl-o-quinodimethanes (2e–h). meso-1c was found to
undergo facile meso/rac isomerization at room
temperature, indicating that other processes such as
HCN, as well as to 9,10-substituted 9,10-dihydroanthracene
derivatives. ESR and ENDOR spectroscopic detection of
related 9-anthryl radicals lends support to the view that 9,10-
dihydroanthracene products are formed by a homolytic
hydrogen-transfer reaction (retrodisproportionation). By way
of contrast, the aforementioned transformations play only a
minor role in the case of methyl-substituted benzocyclo-
butenes 1d, 1f as here they are overruled by faster 1,5-H shift
reactions of the corresponding o-quinodimethanes 2d, 2f,
leading to styrene derivatives.
a
a
symmetry-forbidden disrotatory ring-opening or a stepwise
reaction compete with the symmetry-allowed conrotatory
process. An estimate of the energy profile of the 1c/2c
reaction system was made by kinetic simulation in
combination with oxygen trapping of the intermediate o-
quinodimethanes (2c) and semiempirical PM3 calculations,
and revealed that the barrier for the symmetry-forbidden
pathway is merely about 4 kJ·mol–1 higher than that for the
Introduction
The mutual, thermal interconversion of benzocyclobu-
tenes (1) and o-quinodimethanes (2),[1] one of the classical
examples of a concerted, electrocyclic process in the frame-
work of the rules of conservation of orbital symmetry,[2] is
generally believed to proceed in a conrotatory fashion.[3]
For symmetrically 7,8-disubstituted benzocyclobutenes,
such as 7,8-diphenylbenzocyclobutene (1a, R ϭ Ph), a con-
rotatory electrocyclic process would result in the preser-
vation of the meso or rac stereochemistry of the starting
material (Scheme 1).[4]
Scheme 1. Conrotatory ring-opening of benzocyclobutenes
Any meso/rac isomerization of such benzocyclobutenes of our knowledge it is only recently that this process has
would, therefore, require either a disrotatory ring-opening been investigated in detail for two prototypal systems, na-
or -closure (if the isomerization were to proceed in a con- mely 7,8-diphenyl- (1a, R ϭ Ph) and 7,8-dimethylbenzo-
certed mode) or rotation about only one of the active bonds cyclobutene (1b, R ϭ Me).[5]
in the benzocyclobutenes 1 or o-quinodimethanes 2.
In a recent paper[6] we reported on the photolytic de-
It is interesting to note that non-stereospecific meso/rac composition of 1,3-diphenyl- and 1,1,3-triphenyl-substi-
isomerization of 7,8-disubstituted benzocyclobutenes was tuted 2-indanones 3 and on the reaction of their photo-
briefly mentioned in some early papers,[3,4] but to the best products, benzocyclobutenes 1 and o-quinodimethanes 2,
with nitric oxide in order to assess their potential to serve
[a]
as nitric oxide cheletropic traps (NOCTs)[7] allowing ESR
Institut für Organische Chemie, Universität Essen,
Universitätsstrasse 5, D-45117 Essen, Germany
detection of the resulting nitroxides 4 (Scheme 2).
Fax: (internat.) ϩ49 (0)201/183-3096
In the course of these studies we observed that some of
our benzocyclobutenes underwent facile ring-opening to o-
quinodimethanes 2, even at ambient temperature (Scheme
3). Subsequently, further rearrangements and hydrogen-
E-mail: sustmann@oc1.orgchem.uni-essen.de
[b]
[c]
Institut für Anorganische Chemie, Universität Essen,
Universitätsstrasse 5, D-45117 Essen, Germany
Institut für Physikalische Chemie, Universität Basel,
Klingelbergstrasse 80, CH-4056 Basel, Switzerland
Eur. J. Org. Chem. 1999, 551Ϫ563
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
1434Ϫ193X/99/0303Ϫ0551 $ 17.50ϩ.50/0
551

H.-G. Korth, R. Sustmann et al.
FULL PAPER
preserved during the photolysis indicates that decar-
bonylation and ring-closure occur very rapidly. From the
photolysis mixture, pure meso-1c was isolated by fractional
crystallization. The stereochemistry of the product was veri-
fied by X-ray structural analysis (Figure 1).
Scheme 2. NO trapping of o-quinodimethanes
transfer reactions took place. These included symmetry-for-
bidden thermal meso/rac isomerization (in the case of the
symmetrically substituted 7,8-diphenylbenzocyclobutene
1c), 1,5-hydrogen shift to produce styrene derivatives 5
(from 1d and 1f), and electrocyclic ring-closure to give
4a,10-dihydroanthracenes 6 (from 1c and 1eϪh). Com-
pounds 6 were further transformed to 9,10-substituted an-
thracene (7) and 9,10-substituted 9,10-dihydroanthracene
(8) derivatives, in the course of which process substituted 9-
anthryl radicals (9) could be detected by ESR. Although
some of these reaction pathways have been known for quite
some time,[1,3,8] detailed kinetic and mechanistic analyses
are scarce.[5] Here, we present a spectroscopic and kinetic
investigation of some of these reactions.
Figure 1. X-ray crystal structure of meso-1c
Whereas solid meso-1c is stable at room temperature, in
solution it undergoes slow conversion to rac-1c, as can be
monitored spectrometrically by 1H- and 13C-NMR in
[D3]acetonitrile solution. In oxygen-free solution, a mesoϪ
rac equilibrium was attained after about 2000 h at 20°C,
with an equilibrium constant of K(rac-1c/meso-1c) ϭ 7.5,
corresponding to a ∆RG° of Ϫ4.9 kJ·molϪ1
.
As mentioned above, two borderline mechanistic schemes
(Scheme 4, framed areas) can be envisaged to account for
the meso/rac interconversion of 1c: First, if ring-opening/
ring-closure follows exclusively the conrotatory pathway,
then the symmetry-forbidden process must result from a
one-bond rotation (E,Z-2c/Z,Z-2c isomerization) in the in-
termediate o-quinodimethanes (Scheme 4a, framed area).
Second, if the latter isomerization does not take place, then
a symmetry-forbidden disrotatory ring-opening or ring-
closure reaction in competition with the conrotatory pro-
cess must be assumed (Scheme 4b, framed area). Schemes
4a and 4b cannot be distinguished kinetically by monitor-
ing the time dependence of meso- and rac-1c. Thus, the ex-
perimental concentration-time data (Figure 2) could be fit-
ted equally well to both simplified reaction schemes with
the aid of a kinetic simulation procedure.[9] In both
schemes, we neglect the fact that true biradical intermedi-
ates[5] might be involved in the isomerization process (see
Discussion). For comparison purposes, the calculated rate
constants for both schemes are collected in Table 1.
Scheme 3. Thermal rearrangements of phenyl-substituted benzo-
cyclobutenes 1 via o-quinodimethanes 2
Results
The significance of the magnitudes of the rate constants
in Table 1 lies in the fact that the simulations predict, in
agreement with the failure to detect 2c by conventional 1H-
7,8-Dimethoxy-7,8-diphenylbenzocyclobutene (1c)
Owing to its favourable thermal stability, this compound NMR spectrometry and our previous UV/vis spectroscopic
was investigated most thoroughly. Compound 1c was ob- and nitric oxide trapping experiments,[6] that the stationary
tained as a 95:5 meso/rac mixture by photolytic decar- concentration of the putative o-quinodimethane intermedi-
bonylation of pure meso-3c in acetonitrile solution.[6] The ates 2c never exceeds 10Ϫ5  over the entire time of 2000 h
fact that the configuration of the starting ketone is largely required for the meso/rac equilibration.
552
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
Table 1. Rate constants at 20°C from the kinetic simulation of
Schemes 4a, 4b
Rate
constants
Scheme 4a[a] Scheme 4b[a] Scheme 4a[b] Scheme 4b[b]
kmeso-1c,EZ-2c[c] 1.4·10Ϫ6
kmeso-1c,ZZ-2c[c] Ϫ
1.4·10Ϫ6
2.8·10Ϫ7
4.2·10Ϫ3
3.9·10Ϫ3
2.5·10Ϫ7
6.9·10Ϫ8
1.1·10Ϫ2
2.8·10Ϫ3
Ϫ
Ϫ
Ϫ
Ϫ
Ϫ
1.8·10Ϫ6
Ϫ
1.4·10Ϫ6
1.1·10Ϫ6
5.3·10Ϫ3
3.9·10Ϫ3
2.5·10Ϫ7
9.2·10Ϫ8
8.1·10Ϫ3
2.8·10Ϫ3
Ϫ
kEZ-2c,meso-1c[c] 2.8·10Ϫ3
3.3·10Ϫ3
Ϫ
[c]
kZZ-2c,meso-1c
Ϫ
[c]
krac-1c,ZZ-2c 4.7·10Ϫ7
1.4·10Ϫ6
Ϫ
[c]
krac-1c,EZ-2c
Ϫ
[c]
kZZ-2c,rac-1c 1.1·10Ϫ2
1.9·10Ϫ2
Ϫ
[c]
kEZ-2c,rac-1c
Ϫ
[c]
kEZ-2c,ZZ-2c 8.3·10Ϫ3
8.3·10Ϫ3
5.6·10Ϫ3
2.3·10Ϫ1
2.3·10Ϫ1
4.2·10Ϫ2
3.3·10Ϫ2
[c]
kZZ-2c,EZ-2c 5.6·10Ϫ3
Ϫ
[d]
kEZ-2c,rac-10
Ϫ
4.4·10Ϫ1
4.3·10Ϫ1
6.1·10Ϫ2
6.1·10
kZZ-2c,meso-10[d]Ϫ
Figure 2. meso/rac Isomerization of 1c (0.05 ) at 20°C in [D3]ace-
tonitrile; continuous curves: simulation according to the framed
part of Scheme 4b
[d]
kEZ-2c,others
Ϫ
[d]
kZZ-2c,others
Ϫ
Ϫ
[a]
[b]
[c]
In oxygen-free solution. Ϫ
At 8.1·10Ϫ3  O2. Ϫ
[sϪ1]. Ϫ
[d]
[Ϫ1 Ϫ1].
s
tion mixture, where two adjacent spots giving a positive
peroxide test[11] were detected. Unfortunately, peroxides 10
turned out to be too unstable to be purified by preparative
TLC or column chromatography. Decomposition seemingly
occurred using either silica or alumina as the stationary
phase, since significant amounts of rac-1c (up to 70%) were
invariably recovered. The tentative assignment of the princi-
pal component of 10 as the rac-diastereomer is based on the
reasonable assumption that the conrotatory ring-opening of
meso-1c should be somewhat faster than a possible disrota-
tory one (see Discussion), and thus (E,Z)-2c should have
been predominantly formed and trapped by O2.[5]
Scheme 4. Borderline kinetic schemes for the meso/rac isomeriza-
tion of 1c in the absence (framed areas) and presence of oxygen
Scheme 5. Oxygen trapping of o-quinodimethanes 2c
Evidence for the formation of short-lived o-quinodimeth-
anes can be obtained by the facile reaction of these species
10-Methoxy-10-phenylanthrone (11)[12] and methanol
with nitric oxide[6,7] (Scheme 2) or, alternatively, by trap- were identified as by-products in yields of 16 and 14 mol%,
ping with molecular oxygen, as has been demonstrated by respectively, after 100 d.
Roth and co-workers.[5,10] Therefore, the isomerization of
The time course of the meso-to-rac conversion of 1c in
meso-1c was also carried out in the presence of molecular the presence of oxygen was followed by monitoring the 1H-
oxygen. In oxygen-saturated [D3]acetonitrile solution, a var- NMR signals of the methoxy groups in the δ ϭ 2.5Ϫ4 re-
iety of new H- and 13C-NMR resonances appeared over a gion (Figure 3), with the sum of their integrals accounting
1
period of days at 20°C. The principal (non-aromatic) sig- for the mass balance (there was no indication that the meth-
nals were consistent with the predicted formation of the oxy groups themselves were subject to any chemical trans-
diastereomeric, cyclic peroxides meso- and rac-10 (61 mol% formation).
combined yield) (Scheme 5), in a molar ratio varying from
The shape of the kinetic traces implied that 11 and meth-
about 2.7 to 1.1 over the reaction period (100 d). The for- anol originated from unspecified reactions of the intermedi-
mation of two peroxides was confirmed by TLC of the reac- ates 2c with oxygen rather than from further decomposition
Eur. J. Org. Chem. 1999, 551Ϫ563
553

H.-G. Korth, R. Sustmann et al.
FULL PAPER
the tentative assignment of the stereoisomers is based on
PM3-calculated heats of formation, see Discussion). In ad-
dition, 9-methoxy-10-phenylanthracene (7a) (16 mol%) was
1
identified from its known H-NMR resonances.[12a,12b] In
accordance with the proposed formation of 7a by methanol
elimination from 6a, a comparable amount of methanol (12
1
mol%) was detected by “spiking” of the corresponding H-
NMR signal. The time-concentration profile is displayed in
Figure 4.
Figure 3. Decay of meso-1c (0.05 ) at 20°C in oxygen-saturated
[D3]acetonitrile solution; (᭺): meso-1c, (ț): rac-1c, (᭜): meso-10,
(᭡): rac-10, (ᮀ): other products. Lines: simulation according to
Scheme 4a (dashed) and Scheme 4b (solid)
of the peroxides 10. Though the latter decay path cannot be
ruled out, the concentration-time profile shown in Figure 3
was analyzed in terms of the reactions depicted in Schemes
4a and 4b, respectively, by kinetic simulation.[9] Here,
“other products” stands for the sum of anthrone 11, meth-
anol, and further minor, non-identified by-products. Again,
the possibility of biradical intermediates was neglected.
Furthermore, possible reversibility of the 2c Ǟ 10 reactions,
non-specific decay of peroxides 10 (to form “other” prod-
ucts), and/or possible meso/rac isomerization of 10 were not
taken into account.
Scheme 6. Thermolysis of 1c at 80°C
At a stationary O2 concentration of 8.1·10Ϫ3 ,[13] the
fitting procedure yielded the rate constants shown in Table
1. A better fit with the experimental data was achieved
using Scheme 4b as a basis (see Figure 3). Decay of meso-
1c in air-equilibrated [D3]acetonitrile solution gave a similar
concentration-time profile. Here, in accordance with the
lower steady-state level of oxygen (1.9·10Ϫ3

[13,14]), forma-
tion of 10 was about four times slower, with the amount of
rac-1c going through a maximum of about 60 mol% after
42 d (data not shown).
Whereas in the absence of oxygen at room temperature
the meso/rac isomerization of 1c proceeded without ac-
cumulation of any 1H-NMR-detectable (Ն 0.1%) intermedi-
ates or by-products, heating of the samples to temperatures
Figure 4. Thermolysis of the meso/rac-1c equilibrium mixture at
80°C in oxygen-free [D3]acetonitrile; (᭺): rac-1c, (ț): syn-8a, (᭿):
meso-1c, (᭡): 6a, (ᮀ): 7a, (᭞): anti-8a, (*): methanol. Continuous
lines: simulation according to Scheme 7.
above 50°C resulted in the slow build-up of a variety of
As will be reported below, during decay of some of our
1
new H-NMR resonances. Several weak, complex line pat- benzocyclobutenes we observed ESR spectra of long-lived
terns in the δ ϭ 5.0Ϫ6.5 region pointed to the formation 9-anthryl radicals (9) (Scheme 3), which implied that the
of diastereomeric 4a,10-dihydroanthracenes 6a (Scheme 6) conversion of 6a to 8a might have been initiated by a bimo-
in an approximate 2.6:1 molar ratio. These signals disap- lecular hydrogen-transfer reaction of 6a, a process termed
peared upon prolonged (> 1 h) heating at 80°C. The inter- “retrodisproportionation” by Rüchardt and co-workers.[16]
mediacy of 6a was further supported by similar changes in The likelihood of such a process in the present case was
the UV/vis absorptions in the 350Ϫ410 nm range, typical corroborated by the observation of a multi-line ESR spec-
for 4a,10-dihydroanthracenes.[6,15] After 30 h at 80°C, 1c trum when a sample of 1c in diphenyl ether was heated to
and 6a had almost completely decayed, leaving only several temperatures above 150°C. Though the low signal-to-noise
singlet lines in the non-aromatic region of the 1H-NMR ratio prohibited a detailed spectral analysis, the ESR spec-
spectrum, the majority of which were identified as being trum was most probably due to the 9,10-dimethoxy-10-
due to the syn- and anti-diastereomers of the 9,10-dihy- phenyl-9-anthryl radical (9a) (Scheme 6). Its overall width
droanthracene derivative 8a (75 and 8 mol%, respectively; and the major hyperfine splittings (Table 3) resembled those
554
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
of the related spectra described below. Radical 12a is an was paralleled by the development of a UV/vis absorption
expected intermediate of the retrodisproportionation pro- at λmax ϭ 384 nm, typical for dihydroanthracenes of type
cess (see Discussion, Scheme 9).
6.[1,15] These dihydroanthracenes appeared to be rather per-
Kinetic simulation of the concentration-time profile in sistent; after 24 h (94% conversion of 1e) the yields of 6b,
terms of the simplified reaction scheme (Scheme 7) pro- of the two diastereomers of 6c, and of methanol amounted
vided the rate constants listed in Table 2. In agreement with to 40, 19, 18, and 16 mol%, respectively. The “18%-isomer”
the 1H-NMR observations, the concentration of 4a,10-dihy- of 6c underwent further decay within 8 days, which was
droanthracene 6a is predicted to pass through a minute accompanied by an equivalent increase in the methanol
maximum within 1 h.
content and the formation of 9,10-diphenylanthracene (7b)
(identified by HPLC and UV/vis, λmax ϭ 338, 354, 373, and
393 nm,[17] by comparison with authentic material). NMR
peaks that would have indicated the production of dihy-
droanthracenes 8b and/or 8c at levels of Ն 1% could not be
detected. By way of contrast, when the decay of 1e was
followed at the same temperature in [D6]DMSO solution
1
[k(1e)303 ϭ 4.5·10Ϫ5
s
Ϫ1], only very weak H-NMR signals
of dihydroanthracenes 6 were detected, but in this case di-
hydroanthracenes 8b and 8c (two diastereomers, ratio ca.
8.5:1) were identified as the major products, in yields of 40
and 39 mol%, respectively, after 27 h (99% conversion of
1e). The material balance was accounted for by several
minor, unidentified by-products.
Scheme 7. Kinetic scheme for the thermolysis of 1c at 80°C
When the decomposition of 1e in acetonitrile solution
was performed in the cavity of the ESR spectrometer at
20°C, the slow evolution of a highly persistent, multi-line
ESR spectrum was observed. The hyperfine splittings (hfs)
were determined by ENDOR spectroscopy in diphenyl
ether solution (Figure 5). The ENDOR data allowed the
simulation of the ESR spectrum (Figure 6), where inclusion
of one large 13C-isotope splitting greatly improved the fit
with the experimental spectrum. The ENDOR/ESR hfs
data (Table 3) unambiguously identified the spectrum as be-
ing due to the 9-methoxy-10,10-diphenyl-9-anthryl radical
(9b). The ESR signal intensity of 9b increased reversibly
with increasing temperature, thus indicating the onset of the
equilibrium with its dimer.
7,8-Dimethyl-7,8-diphenylbenzocyclobutene (1d)
This compound was obtained as a 1:1.2 meso/rac mixture
upon photolysis of 3d in acetonitrile solution.[6] At ambient
temperature, no change in the diastereomeric ratio was indi-
cated by 1H-NMR in [D3]acetonitrile solution over a period
of several days, nor were other products detected. On heat-
ing the solution to 60°C, 1d underwent a smooth trans-
formation to α-phenyl-o-(1-phenylethyl)styrene (5a)
(Scheme 3), indicating that ring-opening to 2d was followed
by a rapid 1,5-H shift. After 27 h at 60°C, about 45% of
the 1d had been converted to 5a [k(1d)353 ϭ 8.2·10Ϫ6 Ϫ1],
s
and the meso/rac ratio had changed to about 1:3. 1H-NMR
signals that might have been expected if an electrocyclic
ring-closure to a corresponding 4a,10-dihydroanthracene
derivative had occurred, could not be detected.
7-Methoxy-7,8,8-triphenylbenzocyclobutene (1e)
Due to the absence of a second chiral center in 1e, the
probable symmetry-forbidden electrocyclic ring-opening
or -closure cannot be probed by meso/rac interconversion.
However, rapid ring-opening of 1e to o-quinodimethane 2e
in acetonitrile solution at 30°C was evident from the devel-
1
opment of the H-NMR signals of dihydroanthracenes 6b
and 6c (two diastereomers), and those of methanol. Com-
pound 1e decayed with a rate constant of k(1e)303
ϭ
Figure 5. ENDOR spectrum of the 9-methoxy-10,10-diphenyl-9-
3.1·10Ϫ5 sϪ1, and the disappearance of the 1H-NMR signals anthryl radical (9b) in diphenyl ether at 20°C
Table 2. Rate constants at 80°C from the kinetic simulation of Scheme 7
kmeso-1c,2c
[sϪ1
k2c,meso-1c
k2c,rac-1c
krac-1c,2c
[sϪ1
k2c,6a
[sϪ1
k6a,9a
k9a,anti-8a
k9a,syn-8a
k6a,7a
[sϪ1
]
[sϪ1
]
[sϪ1
]
]
]
[Ϫ1sϪ1
]
[Ϫ1sϪ1
]
[Ϫ1sϪ1
]
]
1.7·10Ϫ3
4.0
3.0
2.6·10Ϫ4
0.75
1.6
7.0·108
8.0·107
3.0·10Ϫ4
Eur. J. Org. Chem. 1999, 551Ϫ563
555

H.-G. Korth, R. Sustmann et al.
FULL PAPER
minor, additional decay path for 1f. This was corroborated
by the observation of the development, at the same rate, of
a UV/vis absorption at λmax ϭ 388 nm (Figure 7), attribu-
table to 9,10-diphenyl-10-methyl-4a,10-dihydroanthracene
(6d) (α-phenylstyrenes can be ruled out from being respon-
sible for this absorption[17]). The assignment was supported
1
by the detection of corresponding, transient H-NMR sig-
nals, all exhibiting the same time dependence as the UV/vis
absorption [k(6d)296 ϭ (4.9 ± 0.2) ϫ10Ϫ4 sϪ1 ]. Two singlets
at δ ϭ 1.66 and 1.75, assigned to the diastereomeric 10-
methyl groups, pointed to a maximum concentration of 6d
of ca. 11 mol% (relative to 1f) and a diastereomeric ratio of
about 2.6:1, similar to the value found for 6a (see above).
A further transformation of 6d was not observed within
three days.
Figure 6. ESR spectrum of the 9-methoxy-10,10-diphenyl-9-anthryl
radical (9b) in acetonitrile at 20°C. Top: experimental; bottom: si-
mulated
Table 3. ESR parameters of substituted 9-anthryl radicals
9a[a]
9b[b]
9c[b]
9e[c]
T [°C]
150
20
16
72
g factor
2.0025
2.00255
2.00255
2.00250
Figure 7. Time-dependent UV/vis spectra of the photolysate of 3f
at 23°C
hfs [mT] a(H1) 0.34 (2 H) 0.341 (2 H) 0.345 (2 H) 0.334 (2 H)
a(H2)
a(H3)
a(H4)
a(H5)
a(H6)
a(H7)
a(H8)
a(other)
0.12 (2 H) 0.117 (2 H) 0.120 (2 H) 0.124 (2 H)
0.39 (2 H) 0.398 (2 H) 0.400 (2 H) 0.401 (2 H)
0.12 (2 H) 0.114 (2 H) 0.114 (2 H) 0.111 (2 H)
n. a.[d]
n. a.
n. a.
n. a.
Ϫ
0.057 (4 H) 0.060 (4 H) 0.060 (4 H)
0.006 (4 H) 0.006 (4 H) 0.006 (4 H)
0.014 (2 H) 0.007 (2 H) 0.012 (2 H)
7-Isocyano-7,8,8-triphenylbenzocyclobutene (1g)
0.074 (3 H)
Ϫ
1.362 (1 H)
Ϫ
300-nm photolysis of ketone 3g in [D3]acetonitrile for 24
h at Ϫ30°C afforded benzocyclobutene 1g in high yield
(Figures 8a, b). Heating of the photolysate to 20°C for 30
min resulted in almost complete conversion of 1g to 10-
1.040 (1 13C) 0.034 (1 N)
[a]
[d]
[b]
[c]
In diphenyl ether. Ϫ
Not analyzed.
In acetonitrile. Ϫ
In benzene. Ϫ
isocyano-9,10-diphenyl-4a,10-dihydroanthracene
(Scheme 3), as indicated by characteristic H-NMR reson-
(6e)
1
7-Methyl-7,8,8-triphenylbenzocyclobutene (1f)
ances in the δ ϭ 4.5Ϫ6.5 region (Figure 8c). The assign-
Photolysis of 1-methyl-1,3,3-triphenylindan-2-one (3f) in ment of these resonances to 6e and not to the possible re-
benzene solution at 10°C yielded a mixture of benzocyclo- gioisomer 6f follows from the observation that at 20°C the
butene 1f and the styrene derivative 5b, in slightly variable olefinic signals underwent further decay within 3 d to yield
1
relative yields depending on the duration of the photoly- an NMR spectrum exhibiting only minor non-aromatic H
sis.[6] For instance, the 1f/5b ratio changed from about 57:43 signals (Figure 8d). 9,10-Diphenylanthracene (7b) was
after 2 h (14% conversion of 3f) to 68:32 after 8 h of pho- identified as the major product (68% isolated yield). Its for-
tolysis (72% conversion). Clearly, some degree of photo- mation can reasonably be explained in terms of the facile
chemically-induced 1,5-hydrogen shift occurs in this case, in elimination of hydrocyanic acid from 6e, as confirmed by a
contrast to 3d, which gave 1d as the only detectable photo- peak corresponding to HCN at δ ϭ 4.47 (Figure 8d).
product. By monitoring the process spectrometrically by
Obviously, the alternative cyclization 2g Ǟ 6f also took
1H-NMR, 1f was found to decay with a rate constant of place, albeit to a much lesser extent, as during the decay of
k(1f)293 ϭ (4.2 ± 0.1) ϫ10Ϫ4 sϪ1 at 20°C, to give mainly 5b. 1g a fairly persistent ESR spectrum (Figure 9) was observed
The slightly lower rate of formation of the latter, k(5b)295
ϭ
which could be satisfactorily simulated only with a set of
(3.8 ± 0.6) ϫ10Ϫ4
s
Ϫ1, indicates a contribution from a hyperfine splittings that matched the 9-isocyano-10,10-di-
556
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
the characteristic α-H doublet splitting of 1.36 mT (Table
3).
1
Figure 8. 300-MHz H-NMR spectra in [D3]acetonitrile of (a) ke-
tone 3g prior to photolysis, (b) after 24 h of photolysis at Ϫ30°C,
(c) after 30 min. at 20°C, (d) after 3 d at 20°C; (ț): unknown
photolysis product
Figure 10. ESR spectrum (benzene, 72°C) of 9e
phenyl-9-anthryl radical (9c) (Table 1). The location of the
1
radical center in 9c requires 6f as its precursor. A tiny H-
Discussion
NMR peak at δ ϭ 5.55 (Figures 8c, d) is indeed consistent
with the 9-hydrogen of the expected 9,10-dihydroanthracene
product 8d, although no definite identification was possible.
The ESR spectrum revealed a slight asymmetry, which
might be due to a contribution from a second, similar rad-
ical, most reasonably 9d derived from retrodisproportion-
ation of 6e.
meso/rac Isomerization of 1c
To the best of our knowledge, 1c is the first benzocyclo-
butene for which a meso/rac isomerization process has been
investigated at ambient temperature. The considerable steric
strain in the cyclobutene moiety, which should be the pri-
mary reason for the low barrier to isomerization, is indi-
cated by the experimental C7ϪC8 bond length (Figure 1)
of d(C7ϪC8)exp ϭ 164.7 pm, which is about 7.1 pm longer
than that in unsubstituted benzocyclobutene[18a] and 6.3
pm shorter than the corresponding distance in the more
sterically crowded 3,8-dichloro-1,1,2,2-tetraphenylcyclo-
buta[b]naphthalene.[18b] The orientation of the phenyl and
methoxy substituents in the solid state is largely reproduced
by the PM3-calculated[19] structure, with the one exception
that in the calculated structure the methoxy group at C8
has a similar orientation as that at C7. The PM3-calculated
bond length of d(C7ϪC8)calcd ϭ 166.3 pm also compares
well with the X-ray value.
The preference for rac-1c by 4.9 kJ·molϪ1 (see above)
over the meso isomer is in agreement with the PM3 calcu-
Figure 9. ESR spectrum ([D3]acetonitrile, 16°C) of 9c
lations, which favor rac-1c [∆fH°(rac-1c)
ϭ 167.8,
∆fH°(meso-1c) ϭ 173.2 kJ·molϪ1] by about 5.4 kJ·molϪ1
.
The PM3 structures of the three configurational isomers of
o-quinodimethane 2c reveal a pronounced distortion of the
7,7,8-Triphenylbenzocyclobutene (1h)
The rearrangement of this compound has been investi- π-system, with out-of-plane torsional angles of the ex-
gated previously.[8h] In accordance with the expected lower ocyclic double bonds of 50.8, 53.9, and 57.9° for (E,E)-,
steric strain in the cyclobutene moiety of 1h, no change of (E,Z)-, and (Z,Z)-2c, respectively. The calculated heats of
its 1H-NMR spectrum in [D6]benzene solution could be de- formation [∆fH°(E,E-2c) ϭ 181.2, ∆fH°(E,Z-2c) ϭ 187.9,
tected at temperatures up to 50°C. Above this temperature, ∆fH°(Z,Z-2c) ϭ 192.9 kJ·molϪ1] increase in the same order.
a small increase in UV/vis absorption in the λ ϭ 340Ϫ395
The data presented here and in our previous paper[6] sug-
nm range was observed, attributable to the formation of the gest that only two of the three possible configurational iso-
corresponding 4a,10-dihydroanthracene derivatives 6g and/ mers of 2c are likely to be involved in the thermal isomeri-
or 6h.[8h] Retrodisproportionation of the former was indi- zation process. Two transients, with lifetimes of τ ϭ 0.16
cated by the observation of a weak ESR spectrum (Figure and 24 µs, observed in the LFP experiments,[6] were orig-
10), unequivocally identified as being due to radical 9e by inally attributed to (E,Z)- and (Z,Z)-2c. A third, long-lived
Eur. J. Org. Chem. 1999, 551Ϫ563
557

H.-G. Korth, R. Sustmann et al.
FULL PAPER
(τ ϭ 30 h at 40°C) o-quinodimethane species that was pro- mediate of the (E,Z)-2 Ǟ (Z,Z)-2 and meso-1 Ǟ rac-1 reac-
duced in the photochemical decomposition of ketone 3c, tions. In conclusion, in terms of kinetic representation,
but not in the thermal reaction, was reasonably attributed Scheme 4b rather than Scheme 4a (framed areas) should
to (E,E)-2c. Thus, in the discussion of the energetics of the hold.
thermal isomerization processes, the involvement of (E,E)-
For the 1b/2b couple, a difference of 20.5 kJ·molϪ1 be-
2c can be disregarded.
tween the activation barriers for the “forbidden” and the
Like the benzocyclobutenes 1a, 1b,[5] compounds 1c rep- conrotatory isomerization was estimated.[5] This difference
resent another test case for probing the energetics of “sym- dropped to just 4.2 kJ·molϪ1 in the case of 7,8-diphenyl-
metry-forbidden” electrocyclic ring-opening reactions. The benzocyclobutene (1a), an effect attributed to the higher
barrier for the symmetry-allowed reaction may be evaluated (benzyl-type) stabilization in the orthogonal biradical tran-
from the racemization kinetics of enantiomerically pure 1c, sition state/intermediate 13b. Following this argument, we
e.g. by the GLC method of Roth and co-workers.[5] Unfor- can assume that for our 1c/2c system the energy difference
tunately, our synthetic procedure[6] yielded mainly meso-1c; between the conrotatory and the “forbidden” process
attempts to separate the enantiomers from the small frac- should be of the same order, probably even more in favor
tion of racemic 1c were not successful.
of the “forbidden” one due to the additional stabilizing ef-
For the “symmetry-forbidden” meso/rac isomerization fect of the methoxy groups. Thus, the energetic difference
process, three pathways may be envisaged (Scheme 8): (i) a between the “allowed” and “forbidden” reaction pathways
true disrotatory (concerted) one; (ii) a non-concerted, sin- might have virtually vanished.
gle-bond rotation involving either a biradicaloid transition
According to our kinetic simulations (Scheme 4b), the
state or an orthogonal biradical intermediate of type 13, or rate constants at 20°C for ring-opening of 1c (Table 1) ap-
(iii) isomerization of the intermediate o-quinodimethanes pear to be only slightly higher than those for diphenyl-o-
by double-bond rotation.
quinodimethane 1a. For the con- and disrotatory reactions
we obtained rate constants of kmeso-1c,EZ-2c ϭ 1.4·10Ϫ6 sϪ1
and kmeso-1c,ZZ-2c ϭ 2.8·10Ϫ7 Ϫ1, respectively. These data
s
have to be compared with the values for 1b, kmeso-1b,EZ-2b ϭ
9.0·10Ϫ7 sϪ1 and kmeso-1b,ZZ-2b ϭ 1.0·10Ϫ7 Ϫ1, as calculated
s
from HuisgenЈs[3a] and RothЈs[5] activation data. Hence, re-
placement of the two hydrogen atoms in 1a by methoxy
groups does not seem to exert a noticeable effect on the
barriers for ring-opening.
Because the experimental barrier for the conrotatory
ring-opening reaction was not available, we attempted to
estimate the energy profile of the meso/rac isomerization
processes by semiempirical PM3 calculations. The calcu-
lations predict the o-quinodimethanes to be 13Ϫ25
kJ·molϪ1 higher in energy than the benzocyclobutenes (Fig-
ure 11, bold bars), in line with the experimental data for
1b.[5] In accord with the experimental evidence, the (E,E)-
isomer represents the lowest-energy structure of 2c. Unfor-
tunately, attempts to locate possible transition states as well
as singlet biradical intermediates 13 failed due to conver-
gence problems.
Estimates of the activation barriers for ring-opening/ring-
closure were therefore sought as follows: By assuming a
Scheme 8. Possible pathways of meso/rac isomerization of benzo-
cyclobutenes
reasonable A-factor of 1013 Ϫ1 for ring-opening/-closure of
s
Roth[5] provided evidence that for 1a and 1b double-bond o-quinodimethanes,[3a,5] from the lifetimes of the two transi-
isomerization of the corresponding o-quinodimethanes is ent species observed by LFP (see above) we calculated acti-
unlikely. It was suggested that the system avoids the ener- vation barriers of EA(Z,Z-2c Ǟ 1c) ϭ 34.8 and EA(E,Z-
getically unfavourable disrotatory pathway by following the 2c Ǟ 1c) ϭ 47.0 kJ·molϪ1. However, the LFP-derived rate
biradical one (which does not rely on orbital symmetry con- constants were much too high to reproduce the concen-
siderations). A difficulty then arises in that 13 would thus tration/time profiles shown in Figures 2, 3, and 4 by kinetic
represent a common transition state for two independent simulation. Therefore, the energy barriers displayed in Fig-
processes, (E,Z)-2 Ǟ (Z,Z)-2 and meso-1 Ǟ rac-1, an ob- ure 11 were calculated from the rate constants obtained
vious contradiction to the definition of the transition state. from the simulation of Scheme 4b (Table 1) using the same
Roth et al. circumvented this problem by arguing that the A-factor. Specifically, barriers for the “allowed” and “for-
biradical structure derived from the (E,Z)-2 Ǟ (Z,Z)-2 pro- bidden” ring-opening of meso-1c, EA(meso-1c Ǟ E,Z-2c) ϭ
cess is not necessarily a transition state, but rather could be 103 and EA(meso-1c Ǟ Z,Z-2c) ϭ 107 kJmolϪ1, were ob-
an intermediate. 13 would then represent a common inter- tained. The corresponding barriers for ring-closure were es-
558
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
release of ring strain in 1c overcompensates for the loss of
aromaticity in one phenyl group.
For the 1e/2e couple, the facile formation of 6b and 6c at
30°C may reflect both a higher steric strain in benzocyclo-
butene 1e, as well as an increased tendency of 2e to undergo
the hexatriene ring-closure. From its rate of decay
[k(1e)303 ϭ 3.1·10Ϫ5 Ϫ1], a barrier of 101.7 kJ·molϪ1 was
s
estimated for the ring-opening of 1e, i.e. about 4Ϫ8
kJ·molϪ1 lower than that for 1c. PM3 calculations predict
a negligibly small energy difference between 1e [∆fH°(1e) ϭ
446.0 kJ·molϪ1] and the two configurations of the o-quino-
dimethanes 2e [∆fH°(E,Z-2e) ϭ 447.3; ∆fH°(Z,Z-2e) ϭ
451.9 kJ·molϪ1]. As no noticeable build-up of 2e could be
detected, the further electrocyclic ring-closure appears to be
accelerated [k(6b ϩ 6c)303 ϭ 3.1·10Ϫ5 sϪ1] compared to that
in the case of 1c. This is in line with increased reaction
enthalpies for the 1e Ǟ 6b and 1e Ǟ 6c conversions, calcu-
lated to be about Ϫ39 and Ϫ58 kJ·molϪ1, respectively.
According to PM3 calculations, for the triphenyl-methyl
system 1f/2f the relative energies of the benzocyclobutene
1f [∆fH°(1f) ϭ 560.2 kJ·molϪ1] and the o-quinodimethanes
Figure 11. Estimated energy profile of the 1c/2c system. Bold bars:
PM3 data; small bars: data estimated from kinetic simulation, assu-
ming A ϭ 1013 sϪ1 (rounded to next decimal number)
2f [∆fH°(E,Z-2f) ϭ 559.7; ∆fH°(Z,Z-2f) ϭ 569.9 kJ·molϪ1
]
timated to be in the range 81Ϫ86 kJ·molϪ1. Thus, there is
an obvious discrepancy between the latter and the LFP-
derived barriers. Because the values from the meso/rac iso-
merization experiments are in much better accord with the
literature data for 1b,[3a,5] this discrepancy may indicate that
our previous assignments of the two transient LFP absorp-
tions were in error, i.e. other species might actually have
been the source of the detected absorptions.
are very similar to those of the 1e/2e couple. Compound 2f
appears to be more reactive in the [π2s ϩ π2s ϩ π2s] electro-
cyclic process than the triphenyl-methoxy analogue 2e, as
indicated by the faster rate of development [k(6d)296 ϭ (4.9
± 0.2) ϫ10Ϫ4 sϪ1 ] of the corresponding UV/vis absorption
(Figure 7). Only methyl group resonances due to the two
1
diastereomers of 6d were detected in the H-NMR spectra,
hence ring-closure of 2f seems to occur almost exclusively
across the (Z)-oriented phenyl group at the diphenyl-substi-
tuted double bond, and does not involve the one at the
phenyl-methyl site. This is in contrast to the situation in
2e, where ring-closure proceeds into both directions. Rapid
decay upon reaction with NO[6] identified the additional
weak absorption in the 450Ϫ550 nm range of the UV/vis
spectrum of the photolysate of 3f (Figure 7) as being due
to a rather long-lived o-quinodimethane. However, as no
temporal change of this absorption was monitored in the
In conclusion, the energy profile of the dimethoxy-di-
phenyl system 1c/2c appears to be very similar to that of
the 1b/2b system. In particular, the difference between the
“allowed” and “forbidden” electrocyclic processes is very
small, of the order of 4 kJ·molϪ1
.
Electrocyclic Hexatriene Ǟ Cyclohexadiene
Ring-Closure
The formation of dihydroanthracenes 6 in the thermal time window of the decay of 1f, it must be attributed to a
decay of phenyl-substituted benzocyclobutenes has been re- configuration of 2f, most reasonably (Z,Z)-2f, that is only
ported previously.[8d,8h,20] This reaction can easily be under- photochemically accessible and is not involved in the ther-
stood as a symmetry-allowed [π2s ϩ 2s ϩ 2s] electrocyclic mal processes.
π
π
ring-closure of the stereoisomers of the intermediate o-qui-
nodimethanes 2 that have at least one of the phenyl sub- π2s] ring-closure, the most reactive system among those in-
stituents in (Z)-orientation (Scheme 3). vestigated here is the 1g/2g couple. The NMR spectra (Fig-
In terms of cyclobutene ring-opening and [π2s ϩ 2s ϩ
π
For the 1c/2c system, the kinetic simulation according to ures 8b, 8c) reveal a preferred, by at least a factor of ten,
Scheme 7 (Figure 4, Table 2) predicts that at 80°C the cycli- formation of 6e over 6f, with a rate constant Ն 10Ϫ3 sϪ1
zation of 2c is about 440 times faster than the ring-opening at ambient temperature. Since no noticeable build-up of o-
of 1c. Assuming again identical A-factors of 1013 Ϫ1 for all quinodimethanes 2g could be monitored, this rate constant
s
the processes, we arrive at a barrier of about 93.7 kJ·molϪ1 should reflect the ring-opening of 1g. Also, the 2g Ǟ 6e
for the 2c Ǟ 6a transformation, i.e. 8Ϫ13 kJ·molϪ1 higher ring-closure should be faster. As 6e can only be formed
than the estimated barriers for the 2c Ǟ 1c ring-closure from the (E,Z)-configuration of 2g, one may conclude that
reactions (Figure 11). This result is in line with the fact that ring-opening of 1g proceeds preferably by an “inward” ro-
at room temperature no formation of 6a could be detected tation of the isocyano group. Intuitively, such a rotation can
by conventional 1H-NMR spectrometry. According to PM3 be expected to be stereochemically favoured, though PM3
calculations, the 1c Ǟ 6a reaction [∆fH°(6a) ϭ 31.9 calculations predict similar thermal stabilities for (E,Z)-2g
kJ·molϪ1] is exothermic by about 33Ϫ42 kJ·molϪ1, i.e. the (∆fH° ϭ 865.7 kJ·mol-1) and (Z,Z)-2g (∆fH° ϭ 867.1
Eur. J. Org. Chem. 1999, 551Ϫ563
559

H.-G. Korth, R. Sustmann et al.
FULL PAPER
kJ·molϪ1). The heat of formation of the parent benzocyclo- and co-workers provided evidence that the so-called “retro-
butene 1g (∆fH° ϭ 869.3 kJ·molϪ1) is predicted to be disproportionation” reaction involves homolytic CϪH
slightly higher, in agreement with the “fast” rate of decay bond cleavage, thereby producing a pair of carbon-centered
of this species. It should be noted that only one dia- radicals, rather than following concerted hydrogen-transfer
1
stereomer of 6e could be detected in the H-NMR spectra pathways. Obviously, such a reaction is also responsible for
(Figure 8c). We explain this fact in terms of a faster elimin- the conversion of 4a,10-dihydroanthracenes 6 to the corre-
ation of HCN from one of the diastereomers rather than sponding 9,10-dihydroanthracenes 8 (a monomolecular 6 Ǟ
by invoking a highly diastereoselective ring-closure of 8 transformation involving a concerted 1,3-H shift would
(E,Z)-2g, since a significant amount of 9,10-diphenylan- be a symmetry-forbidden reaction). The 6 Ǟ 8 conversions
thracene (7b) was already present after a reaction period of (Scheme 9) represent examples of a variant of the retrodis-
just 30 min (Figure 8c).
proportionation process, where a single substrate acts as
both a hydrogen donor and as a hydrogen acceptor. The
driving force for the overall process is most certainly the
gain in aromatic stabilization. ESR detection of the 9-
anthryl radicals 9 provides sound evidence that the reaction
indeed involves initial homolytic cleavage of the 4a-CϪH
bond (Scheme 9, reaction 1). This would result in cyclo-
hexadienyl- and/or benzyl-type radicals 12 and/or 14 being
generated concomitantly. Whereas radicals 12 are certainly
too short-lived to be observed under the conditions em-
ployed (their dimerization is sterically unhindered), radicals
9 and 14 are expected (proven for 9b) to be in equilibrium
with their dimers and co-dimers. The high yields of 9,10-
dihydroanthracenes 8a؊c implies that further back-transfer
of hydrogen (disproportionation) occurs between radicals 9
and 12 and/or 14, ultimately leading to the conversion of
all 6 into 8 (Scheme 9, reactions 2Ϫ4). Alternatively, forma-
tion of 8 may also be explained in terms of direct hydrogen
atom transfer from 6 to 9. As radicals 9 are regenerated in
this step, a chain reaction would be initiated (Scheme 9,
reaction 5).
1,5-H Shift Reactions
Similar to the situation seen with 1c, the PM3 calcu-
lations predict the thermal stability of rac-1d (∆fH° ϭ 401.2
kJ·molϪ1) to be 4.6 kJ·molϪ1 higher than that of its meso
isomer (∆fH° ϭ 405.8 kJ·molϪ1). The three configurational
isomers of the o-quinodimethanes 2d [∆fH°(E,E-2d) ϭ
408.8; ∆fH°(E,Z-2d) ϭ 412.5; ∆fH°(Z,Z-2d) ϭ 420.5
kJ·molϪ1] are calculated to be higher in energy by 7.6, 11.3,
and 19.3 kJ·molϪ1, respectively, i.e. following the same or-
der as in the case of 2c. On steric grounds, 1d was expected
to exhibit reactivity similar to that of 1c. However, the lack
of any product formation at room temperature points to a
somewhat higher barrier for ring-opening in this case. At
elevated temperature, ring-opening is indicated by forma-
tion1 of the 1,5-H shift product 5a, with the steady-state
1
level of o-quinodimethanes 2d below the H-NMR detec-
tion limit. To what extent the observed change of the meso/-
rac ratio from 1:1.2 to 1:3 over the reaction period of 26 h
at 60°C reflects the setting-up of a possible meso/rac equi-
librium and/or the relative reactivity of the (E,E)- and
(E,Z)-diastereomers of 2d in the 1,5-H shift reaction cannot
be determined. In any case, the failure to detect dihydro-
anthracene or anthracene products implies that for (E,Z)-
2d the rate of the 1,5-H shift reaction overrules that of a
possible hexatriene Ǟ cyclohexadiene ring-closure by at le-
ast a factor of ten, and/or that (Z,Z)-2d is too high in en-
ergy to play a significant role as an intermediate.
For the triphenyl-methyl system 1f/2f, the 1,5-H shift re-
action 2f Ǟ 5b is statistically less favoured, though still the
dominant reaction, overruling the 2f Ǟ 6d ring-closure re-
action (see above) by about a factor of nine, as deduced
from the relative intensities of the appropriate 1H-NMR
resonances. As a 1,5-H shift can only occur from the (E,Z)-
configuration, the extensive conversion of 1f to 5b indicates
that ring-opening of 1f produces primarily (E,Z)-2f. This is
in agreement with the PM3 results (see above), which place
(Z,Z)-2f about 10.2 kJ·molϪ1 above (E,Z)-2f.
Scheme 9. 9,10-Dihydroanthracenes via retrodisproportionation of
4a,10-dihydroanthracenes
Our analytical and kinetic data do not allow an unam-
biguous discrimination between the two possibilities. How-
ever, trial kinetic simulations of the concentration-time de-
Retrodisproportionation
Rüchardt et al.[16] have recently reviewed a class of uncat- pendencies in Figure 4 reveal that in the case of such a
alyzed, bimolecular hydrogen-transfer reactions that are in- chain process the maximum concentration of the intermedi-
itiated by a hydrogen atom transfer from a σ-type CϪH ate dihydroanthracenes 6a should be of the order of just
1
hydrogen donor to a π-type hydrogen acceptor. Rüchardt 10Ϫ5 . The much higher, H-NMR-detectable yield of 6a
560
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
nesium sulfate and evaporation of the solvent, the residue was puri-
fied by column chromatography on silica gel (30 ϫ 4.5 cm; eluent
CHCl3) to give 0.72 g 3i (1.87 mmol, 28%), colorless solid, m.p.
is only satisfactorily reproduced if at least one of the dispro-
portionation reactions 9 ϩ 12 or 9 ϩ 14 is explicitly con-
sidered. By-products, which may have resulted from pos-
sible coupling reactions of radicals 8, 9, 12, and 14, were
sometimes detected in the 1H-NMR spectra, but their yields
were far too low for reliable identification.
The extent to which retrodisproportionation can occur in
our systems depends on the competition with the possible
elimination of R1H from the 4a,10-dihydroanthracenes to
1
159Ϫ160°C. Ϫ H NMR (300 MHz, CDCl3): δ ϭ 7.70Ϫ6.97 (m).
Ϫ
13C NMR (75 MHz, BB, CDCl3): δ ϭ 67.4, 72.2, 125.8, 126.6,
127.5 (2 C), 127.6, 128.4, 128.5 (3 C), 128.8, 129.0, 129.6, 131.0 (s,
tert. aromatic C), 135.5, 136.9, 1139.1, 143.0, 144.1 (s, quat. aro-
matic C), 159.0 (s, NC), 204.8 (s, CϭO). Ϫ IR (KBr): ν˜ ϭ 3059
cmϪ1, 3034, 2128 (NC), 1756 (CϭO), 770, 750, 738, 695. Ϫ UV/
vis (CH3CN): λmax (log ε) ϭ 314 nm (2.55), 325 (2.54). Ϫ MS (70
give 9,10-disubstituted anthracenes 7. Thus, whereas ben- ev); m/z (%): 385 (2) [Mϩ], 356 (100), 331 (24), 280 (52). Ϫ HR-
MS (70 eV): C28H19NO: calcd. 385.1467; found 385.1453.
zene elimination from 6b is highly unfavourable and meth-
anol elimination from 6a and 6c can only compete to some
Photolytic Generation of o-Quinodimethanes from Ketones 3aϪj:
degree, the retrodisproportionation decay path plays only a The general procedure for photolytic decarbonylation of 2-in-
danones and the spectroscopic properties of the products not men-
minor role in the case of 6e due to the rapid elimination
of HCN. The mechanism of elimination of methanol and
tioned in the following have been reported in ref.[6]
hydrocyanic acid from 6a/6c and 6e, respectively, was not Kinetic Measurements: The kinetics of the thermal rearrangements
were measured spectrometrically by integration of the non-aro-
matic 1H-NMR signals, using 1,2,4,5-tetramethylbenzene as in-
ternal standard and cross-checking the mass balance against the
sum of the methoxy group integrals. Generally, the raw photolys-
ates from the photolytic decarbonylation of deoxygenated 0.05 
solutions of ketones 3 in the above mentioned deuterated solvents
were directly employed in sealed NMR tubes, except in the case of
1c, where purified material was used. Product yields were corrected
for conversion of the starting material and photochemically pro-
duced by-products. Short-term kinetics were measured by thermo-
statting (± 0.3°C) the sample in the probe head of the spectrometer;
for long-term kinetics the samples were kept in a thermostatted
bath (± 0.1°C) in the dark and were only briefly removed for re-
cording the spectra. For the decomposition of 1c in the presence
of oxygen the samples were equilibrated by bubbling with oxygen
(5.0) or air, respectively, for 30 min in septum-capped NMR tubes
by means of hypodermic needles. The oxygen level in the tubes was
maintained by daily flushing the head space with oxygen or air,
respectively. Data analysis was performed with the CKS simulation
software[9] using the rate constants from single exponential least-
squares fits to the respective concentration-time data as starting
values.
further explored. It might involve ionic, proton-catalyzed
reactions as traces of water were invariably present in our
solutions.
Experimental Section
1
General Remarks: H- and 13C-NMR spectra: Varian Gemini-200
and Bruker AMX-300. Internal standard TMS. Ϫ IR: Perkin-
Elmer 1600 series FT-IR. Ϫ UV/vis: Varian Cary 219, modified for
use with light pipes and thermostatted sample holders. Ϫ GC/MS:
Hewlett-Packard HP 5971A mass-sensitive detector (70 eV) and
HP5890 GC, capillary column HP 1; 50 m ϫ 0.2 mm; SF 0.33 µm.
Ϫ High-resolution MS: Fisons Instruments VG Pro Spec 300 (70
eV). Ϫ HPLC: Varian LC 5000 and Hewlett-Packard 1040 diode-
array UV detector, Varian MCH-10 column (10 µm; 4 ϫ 300 mm).
Ϫ ESR: Bruker ER-420 X-band spectrometer (9.4 GHz) equipped
with a double cavity and a variable temperature unit. Ϫ ESR data
acquisition: Pentium PC equipped with a Microstar DAP 1200/4
data acquisition board and DigiS (GfS Aachen, Germany)
software. Ϫ ENDOR: Bruker ESP-300 Ϫ Melting points (uncor-
rected): Büchi 510.
9,10-Dimethoxy-10-phenyl-4a,10-dihydroanthracene (6a) (2 isomers,
ratio ca. 2.6:1): Major isomer: 1H NMR (300 MHz, CD3CN): δ ϭ
3.05 (s, 3 H, OCH3), 3.73 (s, 3 H, OCH3), 4.42 (br m, 1 H),
5.37Ϫ5.41 (m, 1 H), 5.85Ϫ5.91 (m, 2 H), 6.59Ϫ6.64 (dm, J ϭ 10
Hz, 1 H), 6.75 (dm, J ϭ 7.9 Hz, 1 H), 6.9Ϫ7.7 (m, 8 H). Ϫ Minor
Materials: The synthesis of ketones 3cϪh and 3j has been described
previously.[6] 1-Isocyano-1,3,3-triphenylindan-2-one (3i) was syn-
thesized from 1,3,3-triphenylindan-2-one (3j) via 1-bromo-1,3,3-tri-
phenylindan-2-one.
1-Bromo-1,3,3-triphenylindan-2-one: To 3.0 g (8.32 mmol) of 3i in
tetrachloromethane (60 mL) was added 0.46 mL (8.95 mmol) of
bromine and the solution was stirred in the dark for 24 h. The
solvent was then evaporated under reduced pressure (1 Pa) to leave
3.55 g (8.08 mmol, 97%) of an orange solid, m.p. 184°C. Ϫ 1H
NMR (300 MHz, CDCl3): δ ϭ 7.70Ϫ6.96 (m). Ϫ 13C NMR (75
MHz, BB, CDCl3): δ ϭ 65.2, 67.1, 127.2, 127.3, 127.4, 127.6, 128.2,
128.3, 128.4, 128.8, 128.9, 129.0, 130.4 (s, tert. aromatic C), 137.7,
141.1, 141.5, 142.9, 143.1 (s, quat. aromatic C), 208.4 (s, CϭO). Ϫ
IR (KBr): ν˜ ϭ 3050 cmϪ1, 3032, 1756 (CϭO), 1121, 1056, 762,
744, 722, 700.
1
isomer: H NMR (300 MHz, CD3CN): δ ϭ 3.00 (s, 3 H, OCH3),
3.67 (s, 3 H, OCH3), 4.85 (br m, 1 H), 5.40Ϫ5.42 (m, 1 H),
5.85Ϫ5.98 (m, 2 H), 6.50Ϫ6.56 (dm, J ϭ 10 Hz, 1 H), 6.69 (dm,
J ϭ 7.9 Hz, 1 H), 6.9Ϫ7.7 (m, 8 H).
9-Methoxy-10,10-diphenyl-4a,10-dihydroanthracene (6b): 1H NMR
(300 MHz, CD3CN): δ ϭ 3.53 (s, 3 H, OCH3), 4.95 (m, 1 H), 5.71
(ddm, J ϭ 10, 4 Hz, 1 H), 5.85Ϫ5.95 (m, 1 H), 6.08 (ddt, J ϭ 10.4,
4.0, 1.2 Hz, 1 H), 6.69 (dm, J ϭ 7.8 Hz, 1 H), 6.95Ϫ7.7 (m, 13 H).
10-Methoxy-9,10-diphenyl-4a,10-dihydroanthracene (6c) (2 iso-
1
mers): Major isomer: H NMR (300 MHz, CD3CN): δ ϭ 3.34 (s,
3 H, OCH3), 4.55 (br m, 1 H), 5.43 (dm, J ϭ 10 Hz, 1 H),
5.84Ϫ5.95 (m, 2 H), 6.54 (dm, J ϭ 10.4 Hz, 1 H), 6.82 (dm, J ϭ
7.5 Hz, 1 H), 6.95Ϫ7.7 (m, 13 H). Ϫ Minor isomer: 1H NMR (300
MHz, CD3CN): δ ϭ 3.11 (s, 3 H, OCH3), 4.88 (br m, 1 H), 5.37
(dm, J ϭ 10 Hz, 1 H), 5.73Ϫ5.98 (m, 2 H), 6.71 (dm, J ϭ 10.4 Hz,
1 H), 6.75 (dm, J ϭ 7.5 Hz, 1 H), 6.95Ϫ7.7 (m, 13 H).
Isocyano-1,3,3-triphenylindan-2-one (3i): Under nitrogen, 1.25 g
(9.34 mmol) silver(I) cyanide was added to a solution of 3.0 g (6.83
mmol) 1-bromo-1,3,3-triphenylindan-2-one in DMF (120 mL).
After stirring for 24 h in the dark, the solvent was removed in
vacuo (1 Pa) and the residue was redissolved in chloroform (100
mL). The resulting solution was washed with sodium cyanide solu-
tion (10%; 3 ϫ 50 mL), water (3 ϫ 40 mL), and finally with satu-
rated sodium chloride solution (40 mL). After drying over mag-
9,10-Diphenyl-10-methyl-4a,10-dihydroanthracene (6d) (2 isomers,
ratio ca. 2.6:1): Major isomer: 1H NMR (300 MHz, C6D6): δ ϭ
Eur. J. Org. Chem. 1999, 551Ϫ563
561

H.-G. Korth, R. Sustmann et al.
Table 4. Bond lengths [A] for meso-1c
FULL PAPER
˚
1.67 (s, 3 H, CH3), 4.02Ϫ4.10 (m), 4.43 (m, 1 H), 5.2Ϫ5.3 (m),
5.42Ϫ5.52 (m), 5.97Ϫ6.25 (m). Ϫ Minor isomer: 1H NMR (300
MHz, C6D6): δ ϭ 1.75 (s, 3 H, CH3), 4.02Ϫ4.10 (m), 4.88 (m, 1
H), 5.2Ϫ5.3 (m), 5.42Ϫ5.52 (m), 5.97Ϫ6.25 (m).
bond
bond length
bond
bond length
10-Isocyano-9,10-diphenyl-4a,10-dihydroanthracene (6e): 1H NMR
(300 MHz, CD3CN): δ ϭ 4.75 (br m, 1 H), 5.65 (dm, J ϭ 10 Hz,
1 H), 5.75 (dm, J ϭ 10 Hz, 1 H), 6.05 (dm, J ϭ 10.4 Hz, 1 H),
6.35 (dm, J ϭ 7.5 Hz, 1 H), 6.7 (dm, J ϭ 7.5 Hz, 1 H), 6.9Ϫ7.7
(m, 13 H).
9-Methoxy-10-phenylanthracene (7a):[12a,12b] 1H NMR (300 MHz,
CD3CN): δ ϭ 4.16 (s, 3 H, OCH3), 7.05Ϫ7.65 (m, 11 H), 8.37 (dd,
J ϭ 8.8, 1 Hz, 2 H). Ϫ 13C NMR (75 MHz, DEPT, CD3CN): δ ϭ
63.9 (s, OCH3), 123.1, 126.2, 126.7, 127.3 (?), 127.7, 129.5, 132.2
(s, tert. aromatic C), 133.9, 139.9, 149.0 (s, quat. aromatic C).
O(1)ϪC(7)
O(1)ϪC(9)
O(2)ϪC(8)
O(2)ϪC(10)
C(1)ϪC(6)
C(1)ϪC(2)
C(1)ϪC(7)
C(2)ϪC(3)
C(3)ϪC(4)
C(4)ϪC(5)
C(5)ϪC(6)
C(6)ϪC(8)
C(7)ϪC(11)
C(7)ϪC(8)
1.418(2)
1.423(2)
1.416(2)
1.430(2)
1.376(2)
1.393(2)
1.533(2)
1.390(3)
1.399(3)
1.391(2)
1.382(2)
1.525(2)
1.510(2)
1.647(2)
C(8)ϪC(17)
C(11)ϪC(12)
C(11)ϪC(16)
C(12)ϪC(13)
C(13)ϪC(14)
C(14)ϪC(15)
C(15)ϪC(16)
C(17)ϪC(18)
C(17)ϪC(22)
C(18)ϪC(19)
C(19)ϪC(20)
C(20)ϪC(21)
C(21)ϪC(22)
1.504(2)
1.384(3)
1.391(2)
1.386(3)
1.383(3)
1.377(3)
1.384(3)
1.388(2)
1.391(2)
1.386(3)
1.379(3)
1.383(3)
1.376(3)
9,10-Dimethoxy-10-phenyl-9,10-dihydroanthracene (8a) (2 isomers,
1
ratio ca. 9.4:1): Major isomer: H NMR (300 MHz, CD3CN): δ ϭ
2.95 (s, 3 H, OCH3), 3.00 (s, 3 H, OCH3), 5.82 (s, 1 H), 7.12Ϫ7.25
(m, 4 H), 7.32Ϫ7.45 (m, 5 H), 7.50Ϫ7.62 (m, 2 H), 7.68Ϫ7.73 (dd,
J ϭ 7.7, 0.8 Hz, 2 H). Ϫ 13C NMR (75 MHz, DEPT-135, CD3CN):
δ ϭ 51.9 (s, OCH3), 52.2 (s, OCH3), 73.1 (s, tert. aliphatic C), 79.9
(s, quat. aliphatic C), 126.3, 127.3, 128.4, 128.6, 128.7, 129.0, 129.3
(s, tert. aromatic C), 135.5, 135.9, 150.1 (s, quat. aromatic C). Ϫ
Minor isomer: 1H NMR (300 MHz, CD3CN): δ ϭ 2.74 (s, 3 H,
OCH3), 3.37 (s, 3 H, OCH3), 5.39 (s, 1 H), 7.12Ϫ7.25 (m, 4 H),
7.32-7.45 (m, 5 H), 7.50Ϫ7.62 (m, 2 H), 7.68Ϫ7.73 (dd, J ϭ 7.7,
0.8 Hz, 2 H). Ϫ 13C NMR (75 MHz, DEPT-135, CD3CN): δ ϭ
51.4 (s, OCH3), 56.1 (s, OCH3), 77.4 (s, tert. aliphatic C), 80.6 (s,
quat. aliphatic C), 127.2, 127.3, 128.5, 128.5, 128.9, 129.8, 130.4 (s,
tert. aromatic C), 141.0, 136.1, 153.0 (s, quat. aromatic C).
Table 5. Bond angles [°] for meso-1c
bond bond angle bond
bond angle
C(7)ϪO(1)ϪC(9) 113.75(12) C(17)ϪC(8)ϪC(6)
C(8)ϪO(2)ϪC(10) 117.25(12) O(2)ϪC(8)ϪC(7)
117.74(14)
115.45(13)
114.93(12)
85.07(12)
C(6)ϪC(1)ϪC(2) 122.2(2)
C(6)ϪC(1)ϪC(7) 94.93(13)
C(2)ϪC(1)ϪC(7) 142.8(2)
C(3)ϪC(2)ϪC(1) 115.3(2)
C(2)ϪC(3)ϪC(4) 122.5(2)
C(5)ϪC(4)ϪC(3) 121.2(2)
C(6)ϪC(5)ϪC(4) 116.1(2)
C(1)ϪC(6)ϪC(5) 122.7(2)
C(1)ϪC(6)ϪC(8) 95.23(13)
C(5)ϪC(6)ϪC(8) 142.1(2)
C(17)ϪC(8)ϪC(7)
C(6)ϪC(8)ϪC(7)
C(12)ϪC(11)ϪC(16) 118.4(2)
C(12)ϪC(11)ϪC(7) 120.4(2)
C(16)ϪC(11)ϪC(7) 121.2(2)
C(11)ϪC(12)ϪC(13) 120.8(2)
C(14)ϪC(13)ϪC(12) 120.2(2)
C(15)ϪC(14)ϪC(13) 119.6(2)
C(14)ϪC(15)ϪC(16) 120.1(2)
C(15)ϪC(16)ϪC(11) 120.9(2)
9-Methoxy-10,10-diphenyl-9,10-dihydroanthracene (8b): 1H NMR
(300 MHz, [D6]DMSO): δ ϭ 2.74 (s, 3 H, OCH3), 5.59 (s, 1 H),
7.1Ϫ7.7 (m, 18 H).
O(1)ϪC(7)ϪC(11) 111.40(13) C(18)ϪC(17)ϪC(22) 118.7(2)
O(1)ϪC(7)ϪC(1) 116.66(12) C(18)ϪC(17)ϪC(8) 122.1(2)
C(11)ϪC(7)ϪC(1) 116.90(13) C(22)ϪC(17)ϪC(8) 119.2(2)
O(1)ϪC(7)ϪC(8) 107.84(12) C(19)ϪC(18)ϪC(17) 120.2(2)
C(11)ϪC(7)ϪC(8) 116.59(13) C(20)ϪC(19)ϪC(18) 120.4(2)
10-Methoxy-9,10-diphenyl-9,10-dihydroanthracene (8c) (2 isomers,
1
ratio ca. 8.5:1): Major isomer: H NMR (300 MHz, [D6]DMSO):
C(1)ϪC(7)ϪC(8) 84.75(12)
C(19)ϪC(20)ϪC(21) 119.7(2)
δ ϭ 3.07 (tentative) (s, 3 H, OCH3), 5.53 (s, 1 H), 7.1Ϫ7.7 (m, 18
O(2)ϪC(8)ϪC(17) 105.96(12) C(22)ϪC(21)ϪC(20) 120.0(2)
1
H). Ϫ Minor isomer: H NMR (300 MHz, [D6]DMSO): δ ϭ 3.04 O(2)ϪC(8)ϪC(6) 117.14(13) C(21)ϪC(22)ϪC(17) 121.0(2)
(tentative) (s, 3 H, OCH3), 5.78 (s, 1 H), 7.1Ϫ7.7 (m, 18 H).
rac-3,6-Dimethoxy-3,6-diphenyl-4-benzo-1,2-dioxane (rac-10) (stere-
ochemical assignment tentative): 1H NMR (300 MHz, CD3CN):
δ ϭ 3.57 (s, 3 H, OCH3), 7.16 (dd, J ϭ 6.0, 3.2 Hz, 2 H), 7.32Ϫ7.74
(m, 10 H). Ϫ 13C NMR (75 MHz, DEPT, CD3CN): δ ϭ 53.04
(OCH3), 104.19 (s, quat. aliphatic C), 128.06, 128.20, 129.22,
129.22, 129.98 (s, tert. aromatic C), 135.12, 139.0 (s, quat. aro-
matic C).
phenyl ether. Microwave power: 8 mW, radiofrequency power: 1
dB, modulation depth: 34 kHz.[22]
Quantum Chemical Calculations: Semiempirical PM3 calcu-
lations[19] were carried out with the SPARTAN 4.0.3 program pack-
age (Wavefunction Inc., Irvine) on a Silicon Graphics Indy workst-
ation. Structures were fully optimized with respect to all geometri-
cal coordinates.
meso-3,6-Dimethoxy-3,6-diphenyl-4-benzo-1,2-dioxane
(meso-10)
(stereochemical assignment tentative): 1H NMR (300 MHz,
CD3CN): δ ϭ 3.38 (s, 3 H, OCH3), 6.95 (dd, J ϭ 6.0, 3.6 Hz, 2
H), 7.32Ϫ7.74 (m, 10 H). Ϫ 13C NMR (75 MHz, DEPT, CD3CN):
δ ϭ 53.18 (s, OCH3), 105.31 (s, quat. aliphatic C), 128.10, 128.24,
129.60, 129.51, 130.01 (s, tert. aromatic C), 136.08, 140.67 (s, quat.
aromatic C).
Crystal Structure Determination of meso-1c: C22H20O2, formula
weight 316.38, crystal colour pale yellow, plate-shaped, approxi-
mate dimensions 0.67 ϫ 0.52 ϫ 0.29 mm, measured on a Nicolet
R3 diffractometer with Mo-Kα radiation. T ϭ 120 K. Cell dimen-
sions (from 50 centered reflections): a ϭ 8.965(3), b ϭ 11.903(3),
3
˚
˚
c ϭ 15.874(3) A, V ϭ 1693.8(7) A , orthorhombic crystal system,
Z ϭ 4, dcalcd. ϭ 1.241 g·cmϪ3, µ ϭ 0.078 mmϪ1, space group
P212121, data collection of 4413 intensities (2Θmax ϭ 50°), 2909
independent (Rmerg ϭ 0.0323), no absorption or extinction correc-
tion, structure solution with direct methods (SHELXS-86) and re-
finement on F2 (Siemens SHELXTL Vers. 5.03), 217 parameters,
the hydrogen atom positions were calculated and refined as riding
groups with the 1.2 fold (1.5 fold for methyl groups) isotropic U
value of the corresponding C atoms. R1 ϭ 0.0406 [2762 observed
10-Methoxy-10-phenylanthrone (11): 1H NMR (300 MHz,
CD3CN): δ ϭ 2.98 (s, 3 H, OCH3), 7.2Ϫ7.7 (m, 11 H), 8.27 (ddd,
J ϭ 8.1, 1.1, 1.0 Hz, 2 H). Ϫ IR (KBr): ν˜ ϭ 1670 cmϪ1 (CϭO)[12]
.
ESR Measurements: ESR experiments were carried out in deoxy-
genated solvents in sealed 4-mm o.d. quartz tubes. ESR parameters
were refined by computer simulation using the LMB simulation
program.[21]
ENDOR Measurements: ENDOR spectra were recorded at 20°C.
The samples contained 10Ϫ2  solutions of 1c in deoxygenated di-
data with Fo Ն 4σ(F)], wR2 (all data) ϭ 0.0982, GoF ϭ 1.066,
2
wϪ1 ϭ σ2(Fo2) ϩ (0.0656·P)2 ϩ (0.2334·P), where P ϭ [(max Fo
)
562
Eur. J. Org. Chem. 1999, 551Ϫ563

Thermal Rearrangements of Di- and Triphenyl-Substituted Benzocyclobutenes
FULL PAPER
K. K. de Fonseka, J. J.
[8e]
2
ϩ (2 Fc )]/3, maximum residual electron density 0.176 eAϪ3, abso-
˚
Ber. 1968, 101, 2302Ϫ2325. Ϫ
McCullough, A. J. Yarwood, J. Am. Chem. Soc. 1979, 101,
lute structure parameter 0.2(12). Crystallographic data (excluding
structure factors) for the structure reported in this paper have been
deposited with the Cambridge Crystallographic Data Centre as
supplementary publication no. CCDC-101797. Copies of the data
can be obtained free of charge on application to CCDC, 12 Union
Road, Cambridge CB2 1EZ, U.K. [Fax: (internat.) ϩ44 (0)1223/
336033; E-mail: deposit@ccdc.cam.ac.uk].
[8f]
3277Ϫ3282. Ϫ
V. Wintgens, J. C. Netto-Ferreira, H. L. Ca-
sal, J. C. Scaiano, J. Am. Chem. Soc. 1990, 112, 2363Ϫ2367. Ϫ
[8g]
R. W. Redmond, C. W. Harwig, J. C. Scaiano, Photochem.
[8h]
Photobiol. A 1992, 68, 255Ϫ259. Ϫ
J. C. Netto-Ferreira, V.
Wintgens, J. C. Scaiano, Tetrahedron Lett. 1989, 30, 6851Ϫ6854.
Ϫ
[8i] J. C. Scaiano, V. Wintgens, J. C. Netto-Ferreira, Pure Appl.
[8j]
Chem. 1990, 62, 1557Ϫ1564. Ϫ
Chem. Soc., Perkin Trans. 1 1975, 171Ϫ174. Ϫ
D. W. Jones, G. Kneen, J.
[8k]
J. M. Hol-
land, D. W. Jones, J. Chem. Soc. (C) 1971, 608Ϫ612.
CKS, Chemical Kinetics Simulator, v. 1.02, IBM Almaden Re-
search Center, Almaden, 1995.
[9]
[10]
[11]
W. R. Roth, T. Ebbrecht, A. Beitat, Chem. Ber. 1988, 121,
1357Ϫ1358.
Acknowledgments
H. Kropf, S. Munke, in Methoden der Organischen Chemie
(Houben-Weyl), vol. E13/2, 4th Ed., Thieme, Stuttgart, 1988,
p. 1388.
This work was supported by the Deutsche Forschungsgemein-
schaft.
[12] [12a]
J. Rigaudy, G. Chelu, N. K. Cuong, J. Org. Chem. 1985,
[12b]
50, 4474Ϫ4478. Ϫ
F. Gobert, S. Combrisson, Tetrahedron
[1] [1a]
[1b]
J. J. McCullogh, Acc. Chem. Res. 1980, 13, 270Ϫ276. Ϫ
1974, 30, 2919Ϫ2924. Ϫ [12c] M. Oyama, K. Nozaki, S. Okazagi,
J. Electroanal. Chem. 1991, 304, 61Ϫ73.
C. W. G. Fishwick, D. W. Jones in The Chemistry of Quinoid
Compounds, Vol. 2 (Eds.: S. Patai, Z. Rappoport), Wiley, Chich-
ester, 1988, pp. 403Ϫ453. Ϫ [1c] J. L. Charlton, M. M. Alauddin,
Tetrahedron 1987, 43, 2873Ϫ2889.
[13]
[14]
J. Wisniak, M. Herskowitz, Solubility of Gases and Solids, Part
B, Physical Sciences Data 18, Elsevier, Amsterdam, 1984.
F. Wilkinson, Pure Appl. Chem. 1997, 69, 851Ϫ856.
[2]
R. B. Woodward, R. Hoffmann, Die Erhaltung der Orbitalsym-
metrie, Akademische Verlagsgesellschaft Geest & Portig, Le-
ipzig, 1970, p. 50; cf. Angew. Chem. 1969, 81, 797Ϫ869; Angew.
Chem. Int. Ed. Engl. 1969, 8, 781.
[15] [15a] G. Porter, M. F. Tchir, J. Chem. Soc. (A) 1970, 1372Ϫ1373.
[15b]
Ϫ
G. Porter, M. F. Tchir, J. Chem. Soc., Chem. Commun.
[15c]
1971, 3772Ϫ3777. Ϫ
Wiersdorff, Chem. Ber. 1969, 102, 1597Ϫ1605.
G. Quinkert, H.-P. Lorenz, W. W.
[3] [3a]
[16]
[17]
R. Huisgen, H. Seidel, Tetrahedron Lett. 1964, 3381Ϫ3384.
C. Rüchardt, M. Gerst, J. Ebenhoch, Angew. Chem. 1997, 109,
1474Ϫ1498; Angew. Chem. Int. Ed. Engl. 1997, 36, 1406Ϫ1430.
H. H. Perkampus, UV-Vis Atlas of Organic Compounds, Part 2,
2nd ed., VCH, Weinheim, 1992, p. 622.
Ϫ
[3b] G. Quinkert, K. Opitz, W. W. Wiersdorff, M. Finke, Tetra-
hedron Lett. 1965, 3009Ϫ3013.
[4]
[5]
L. A. Carpino, J. Am. Chem. Soc. 1962, 84, 2196Ϫ2201.
W. R. Roth, V. Rekowski, S. Börner, M. Quast, Liebigs Ann.
1996, 409Ϫ430.
[18] [18a]
R. Boese, D. Bläser, Angew. Chem. 1988, 100, 293Ϫ295;
Angew. Chem. Int. Ed. Engl. 1988, 27, 304Ϫ306. Ϫ [18b] F. Toda,
K. Tanaka, Z. Stein, I. Goldberg, Acta Crystallogr. Sec. C 1996,
52, 177Ϫ180.
[19]
[6]
T. Paul, M. A. Hassan,. D. V. Avila, H.-G. Korth, R. Sustmann,
J. Org. Chem. 1996, 61, 6835Ϫ6848.
[7] [7a]
H.-G. Korth, K. U. Ingold, R. Sustmann, H. de Groot, H.
J. J. P. Stewart, J. Comput. Chem. 1989, 10, 109Ϫ220.
[20] [20a]
Sies, Angew. Chem. 1992, 107, 915Ϫ917; Angew. Chem. Int. Ed.
E. F. Ullman, K. R. Huffman, Tetrahedron Lett. 1965, 23,
[7b]
[20b]
Engl. 1992, 31, 891Ϫ893. Ϫ
H.-G. Korth, R. Sustmann, P.
1863Ϫ1867. Ϫ
R. M. Wilson, K. Hannemann, K. Peters,
Lommes, T. Paul, A. Ernst, H. de Groot, L. Hughes, K. U.
E.-M. Peters, J. Am. Chem. Soc. 1987, 109, 4741Ϫ4743.
WINSIM, Author: D. R. Duling, Laboratory of Molecular Bio-
physics, NIEHS, NIH, Research Triangle Park, North Carolina;
see: D. R. Duling, J. Magn. Reson. B. 1994, 104, 105Ϫ110.
H. Kurrek, B. Kirste, W. Lubitz, Electron Nuclear Double Res-
onance Spectroscopy of Radicals in Solution, VCH, Weinheim,
1988.
[21]
[22]
Ingold, J. Am. Chem. Soc. 1994, 116, 2767Ϫ2777.
[8] [8a]
[8b]
G. Quinkert, Pure Appl. Chem. 1964, 9, 607Ϫ621. Ϫ
K.
H. Grellmann, J. Palmowski, G. Quinkert, Angew. Chem. 1971,
83, 209Ϫ210; Angew. Chem. Int. Ed. Engl. 1971, 10, 196Ϫ197.
[8c]
Ϫ
G. Quinkert, J. Palmowski, H. P. Lorenz, W. W.
Wiersdorff, M. Finke, Angew. Chem. 1971, 83, 210Ϫ212; Angew.
Chem. Int. Ed. Engl. 1971, 10, 197Ϫ198. Ϫ
[8d]
G. Quinkert, W.
Received September 18 1998
W. Wiersdorff, M. Finke, K. Opitz, F.-G. von der Haar, Chem.
[O98423]
Eur. J. Org. Chem. 1999, 551Ϫ563
563