Evidence for fully conjugated double-stranded cycles

Article abstract of DOI:10.1002/chem.201100443

A chain of circumstantial evidence for the existence of the first fully conjugated, double-stranded cycles is presented. The products have the structure of the belt-region of fullerene C₈₄ (D₂) and carry either four hexyl chains or four phenyl groups. The unsubstituted parent cycle is also presented. The chain of evidence is mainly based on mass spectrometric analysis and trapping reactions, the latter being supported by quantum mechanical calculations. It is also of importance that the phenyl-substituted and unsubstituted products cannot undergo a [1,5] hydrogen shift, the only reasonable side-reaction that recently could not be excluded for the alkyl-substituted analogue. It is concluded that the fully aromatic targets truly exist in the gas phase. Whether they can be generated in solution under the applied conditions cannot yet be firmly decided; theoretical evidence speaks against.

Full text of DOI:10.1002/chem.201100443

FULL PAPER
DOI: 10.1002/chem.201100443
Evidence for Fully Conjugated Double-Stranded Cycles
Malte Standera,[a] Rolf Hꢀfliger,[b] Renana Gershoni-Poranne,[c] Amnon Stanger,[c]
Gunnar Jeschke,[d] Jacco D. van Beek,[e] Louis Bertschi,[b] and A. Dieter Schlꢁter*[a]
Abstract: A chain of circumstantial evi-
dence for the existence of the first fully
conjugated, double-stranded cycles is
presented. The products have the struc-
ture of the belt-region of fullerene C84
(D2) and carry either four hexyl chains
or four phenyl groups. The unsubstitut-
ed parent cycle is also presented. The
chain of evidence is mainly based on
mass spectrometric analysis and trap-
ping reactions, the latter being support-
ed by quantum mechanical calcula-
tions. It is also of importance that the
phenyl-substituted and unsubstituted
products cannot undergo a [1,5] hydro-
gen shift, the only reasonable side-reac-
tion that recently could not be exclud-
ed for the alkyl-substituted analogue. It
is concluded that the fully aromatic tar-
gets truly exist in the gas phase.
Whether they can be generated in solu-
tion under the applied conditions
cannot yet be firmly decided; theoreti-
cal evidence speaks against.
Keywords: aromaticity
· conjuga-
tion · mass spectrometry · thermo-
chemistry · trapping experiments
Introduction
redox behavior, including the aspect of maximum charge-
AHCTUNGTREGaNNUN bility. They would also serve as models for their open-
Chemists have dreamed of fully unsaturated, double-strand-
ed hydrocarbon cycles like [n]cyclacenes I, buckybelts II, cy-
clo[n]phenacenes III, and Vçgtle-belts IV for several de-
cades (Figure 1).[1] In addition to their aesthetic appeal, they
are also important for the study of certain aspects of host–
guest behavior, including hydrogen bonding between weak
donors and p systems, the formation of polyrotaxanes, the
endo/exo cyclic selectivity of reactions, the dynamics of tran-
sition-metal fragments complexed to the p system, and their
chain linear analogues, the ladder polymers,[2] and some of
them could be developed into novel, shape-persistent con-
stituents for molecular constructions.[3] Finally, such com-
pounds would enrich the world of larger, all-carbon struc-
tures such as fullerenes, carbon nanotubes, and graphene.
Although there has been significant progress in their synthe-
sis, a breakthrough has not yet been achieved. Several
double-stranded cycles with the appropriate carbon skele-
tons have been synthesized and even a few chemical modifi-
cations aimed at an expansion of the conjugated parts have
been accomplished.[4] However, attempts to generate any of
the above fully conjugated targets failed altogether and only
recently has synthesis gotten close, specifically the work of
[a] Dr. M. Standera, Prof. A. D. Schlꢀter
Laboratory of Polymer Chemistry
Department of Materials, ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, HCI 541, 8093 Zꢀrich (Switzerland)
E-mail: dieter.schlueter@mat.ethz.ch
[b] R. Hꢁfliger, L. Bertschi
Mass Spectrometry Unit, Laboratory of Organic Chemistry
Department of Chemistry and Applied Biological Sciences
ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, 8093 Zꢀrich (Switzerland)
[c] R. Gershoni-Poranne, Prof. A. Stanger
Schulich Faculty of Chemistry and
The Lise-Meitner-Minerva Center for
Computational Quantum Chemistry, Technion
Technion City, 32000 Haifa (Israel)
[d] Prof. G. Jeschke
Laboratory of Physical Chemistry, ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, HCI F 231, 8093 Zꢀrich (Switzerland)
[e] Dr. J. D. van Beek
Laboratory of Solid-State NMR Spectroscopy
Department of Chemistry and Applied Biological Sciences
ETH Zꢀrich
Figure 1. Fully conjugated, belt-like compounds, [n]cyclacenes I (n=12)
and buckybelts II (shown with two repeat units), and angularly annulated
congeners, the cyclo[n]phenacenes III (n=12) and the Vçgtle belts IV
(shown with six repeat units). The arrows help identify the repeat units.
Wolfgang-Pauli-Strasse 10, 8093 Zꢀrich (Switzerland)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201100443.
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
12163

Bodwell,[5] Iyoda,[6] and ourselves.[7] The reasons for this
considerable complication are target specific, but, by and
large, also have something to do with the curvature of the
compounds, rendering a fully conjugated structure energeti-
cally less favorable than noncurved, flat analogues. This
energy cost makes the final “aromatization” step more com-
plicated to achieve and also the target structure, once ac-
complished, more prone to subsequent reactions, which may
go as far as to render it transient and nonisolable.
compounds 4a–c. It comprises optimizations and novel syn-
theses, trapping reactions supported by quantum-chemical
calculations as well as NMR, EPR, and UV spectroscopic
and mass spectrometric investigations. As will be shown,
none of the evidence provided represents stand-alone un-
equivocal proof. The combination of the results obtained,
however, condenses into a rather conclusive picture that
suggests the existence of fully conjugated, double-stranded
cycles of type II.
Our laboratory recently provided high-resolution mass
spectrometric evidence for a compound with the same
molar mass as 4a (Figure 2), a tetrahexyl-substituted bucky-
belt derivative of type II.[7a] The product was obtained by
thermolysis of the tetraacetate 3a at 3308C. Although this
precursor has the same carbon skeleton as 4a, it cannot be
excluded that, instead of 4a, a rearranged isomer was ana-
lyzed. This isomer may have formed either by passing
through the intermediate 4a or by surpassing it. Unfortu-
nately, neither the relatively drastic reaction conditions nor
the energetic considerations based on computations[7a] al-
lowed this possibility to be excluded. In particular, isomers
of 4a resulting from [1,5] hydrogen shifts[8] seemed possible.
We therefore initiated a broader study into this matter that
comprised 1) completion of the experiments based on the
hexyl system and 2) usage of precursors with the same
carbon skeleton as target II but with lateral substituents that
would render a hydrogen shift impossible (Ph and H instead
of C6H13). The relevant compounds are presented in
Figure 2. In particular, for the laterally unsubstituted com-
pounds (1c–4c), serious solubility issues had to be expected.
In light of the importance of the goal of this work, this
drawback was nevertheless accepted. Herein we report on
the outcome of this broader study, aimed at the three target
Results
All the compounds shown in Figure 2 were synthesized ac-
cording to the published procedure based on the hexyl
system.[7] For this reason, the syntheses will not be described
in the main text, and the reader is referred to the Support-
ing Information, which provides the relevant schemes, pro-
cedures, and analytical data in detail. The tetrahydrate 1a[7a]
was prepared on a several-gram scale and its acid-catalyzed
dehydration gave 2a, as is already known. The new tetrahy-
drate 1b was synthesized on a smaller scale and analyzed by
spectroscopic and spectrometric means only. The corre-
sponding 1c was not isolated but rather analyzed directly
from the crude product mixture by mass spectrometry only.
The following aspects were investigated and will be re-
ported below:
1) The thermolysis of 3a with specific focus on procedural
optimizations and trapping of the presumed target 4a by
Diels–Alder dienophiles.
2) Computational studies on the expected products of these
trapping reactions with the aim of estimating their exo-
thermicity and obtaining the corresponding adduct struc-
tures.
3) The attempted generation of 4b from 1b as well as from
3b and of 4c from 3c.
4) The formation of radical species under the conditions of
thermolysis and dehydration.
5) The attempted purification, revealing the transient
nature of the thermolysis products.
Thermolysis of tetraacetate 3a and trapping experiments:
Until now, the thermolysis of 3a was performed in a metal
bath preheated to 3308C for 3 min under vacuum and in the
presence of a large excess (approximately 10-fold by mass)
of parent anthracene. In this work, factors such as vacuum,
temperature, time, and additives (e.g., solvents, acid quench-
ers, and trapping agents) were varied. A detailed description
is available in ref. [9]. The optimum conditions were con-
cluded to be 3308C, 5 min, an excess of sodium carbonate, a
nitrogen atmosphere at ambient pressure, and 1,10-phenan-
throline as both solvent and Diels–Alder trapping agent.
These conditions resulted in higher reproducibility and a
cleaner reaction course. For example, the products obtained
from a 3 or 5 min thermolysis were practically identical.
Figure 2. Compounds 1a,b,c–4a,b,c of relevance in this study.
12164
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174

Fully Conjugated Double-Stranded Cycles
FULL PAPER
high amounts of carbonate, the
intensity increased. Whether
these signals are due to the
presence of an oxygen atom in
the cycle or a “water” molecule
could not be clarified with cer-
tainty. Also, it could not be
clarified whether this increased
mass is due to residual frag-
ments or due to addition reac-
tions, the latter considered to
be the more likely case. Similar
effects have been reported pre-
viously.[12]
Computational studies: The en-
thalpy and free energy of the
trapping reactions were calcu-
lated by the following ap-
proach: The model reaction [re-
action (1)] of addition of phe-
Figure 3. MALDI-FT mass spectrum of the crude mixture resulting from the thermolysis of 3a in the presence
of 1,10-phenanthroline showing the known isotopic patterns at m/z 932 [M]+, 934 [M+2H]+, and 936 [M+
4H]+ and the signal groups centered at m/z 1116 [M+phen]+ and 1294 [M+2phen]+.
Figure 3 shows a typical and representative MALDI-FT
mass spectrum of the crude product obtained from thermol-
ysis under such conditions. In contrast to the previous condi-
tions, in which anthracene was used as a solvent, no signals
were observed that stem from cycles still containing acetate
functions. Thus, the conversion under the optimized condi-
tions was clearly higher. In addition to the signal group at
m/z 932 [M]+, 934 [M+2H]+, and 936 [M+4H]+, the spec-
trum shows relatively complex signal groups centered at m/z
1116 and 1294, with the latter being somewhat simpler. The
masses of these two groups correspond to protonated and
nonprotonated mono- and diphenanthroline adducts of the
compounds with m/z 932 and 934.[10] If 2,9-dimethyl-1,10-
phenanthroline was used instead, the corresponding mono-
and diadducts were observed at m/z 1144 [M+phen]+ and
1350 [M+2phen]+. If this trapping agent was replaced by
4,4’-dimethyl-2,2’-bipyridyl, which lacks a dienophile, no
adduct was detected. This suggests that Diels–Alder addi-
tion reactions take place with the phenanthrolines,[11] howev-
er, firm structural conclusions cannot be drawn. Note that
the observed adducts did not form in the mass spectrometer
but rather during thermolysis. This was confirmed by record-
ing a mass spectrum of a blend consisting of a sample pre-
pared from a thermolysis product generated in the absence
of a trapping agent and 2,9-dimethyl-1,10-phenanthroline.
This spectrum did not show any indication of adducts. Final-
ly, note that on the basis of ball-and-stick model considera-
tions, a host–guest complex formed between the presumed
fully conjugated target 4a and the “trapping” agents could
not have formed and thus cannot explain the signal at m/z
1116. There is a final aspect relating to the mass spectrum
shown in Figure 3 that is worth mentioning. This regards the
low intensity signals at m/z 950, 1138, and 1316. In the hun-
dreds of mass spectra recorded of the thermolysis products,
these signal groups appeared in each experiment. Their in-
tensity depended upon reaction time and the amount of car-
bonate present. For long times (longer than 10 min) and
nanthroline (A) to anthracene (B) was calculated at the G3
level of theory to obtain reliable thermochemical data. The
AHCTUNGTRENNUNG
thermochemistry of the addition of one and two phenan-
throline molecules to 4c was estimated by using isodesmic
reactions, the energies of which were obtained at the
B3LYP/6-311G* level of theory. Isodesmic reactions are
much less sensitive to the computational level and therefore
their results can be considered accurate. The DA reaction of
A and B is endothermic by 2.3 kcalmolꢀ1 (at the G3 level).
Reactions (2) and (3) compare the addition of phenan-
throline to 4c, this unsubstituted compound serving as a
model for 4a. These reactions are highly exothermic:
ꢀ15.3 kcalmolꢀ1 for the first addition and ꢀ19.6 kcalmolꢀ1
for the second. Considering reaction (1), this means that the
addition of the first and second phenanthroline molecules to
4c are exothermic by 13.0 and 17.3 kcalmolꢀ1, respectively.
4 c þ C ! 4 c-phen þ B
ð2Þ
DHꢁ ¼ ꢀ15:3 kcal molꢀ1; DGꢁ ¼ ꢀ28:7 kcal molꢀ1
4 c-phen þ C ! 4 c-diphen þ B
ð3Þ
DHꢁ ¼ ꢀ19:6 kcal molꢀ1; DGꢁ ¼ ꢀ32:2 kcal molꢀ1
As in other reactions of 4,[7a] these results can be ex-
plained by the release of strain following the DA reactions
(Figure 4).
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12165

A. D. Schlꢀter et al.
Figure 4. Calculated structures of compounds “4c-phen” (left) and “4c-
diphen” (right), the products of a DA trapping reaction involving the
central ring of the anthracenes of 4c.
Interestingly, the DG8 values for reactions (2) and (3) are
ꢀ28.7 and ꢀ32.2 kcalmolꢀ1, respectively. Unlike the compar-
ison between hydrated anthracenes and hydrated cycles, for
which DH8 and DG8 are almost identical,[7a] the respective
differences for the DA reactions are ꢀ13.4 [reaction (2)]
and ꢀ12.6 kcalmolꢀ1 [reaction (3)]. These numbers result
from the highly positive entropy of the reactions, approxi-
mately 45 and 42 entropy units [calmolꢀ1 Kꢀ1]. This can be
understood in terms of flexibility. In the addition of phenan-
throline to anthracene, two rigid molecules produce a rigid
system. However, owing to the release of the strain, the
products of the reaction of 1c with phenanthroline are
“softer” than the starting materials, that is, the bonds have
smaller force constants and therefore higher entropies of
formation. Thus, the DA reactions are not only enthalpy-
driven (strain release) but also entropy-driven, making them
far more facile than the corresponding reactions of anthra-
cene.
Figure 5. Room-temperature 1H NMR spectra of a solution of 1b in
[D5]nitrobenzene without acid (top) and after the addition of trifluoro-
methanesulfonic acid at room temperature and subsequent heating at
1008C for 8 min (bottom). Note that the signals at approximately d=6.0
(exo-CHO) and 6.5 ppm (endo-CHO) and the triplets at d=7.5 and
7.8 ppm in the top spectrum are absent from the bottom spectrum. The
same applies to the signals of 1b in the aromatic range. The signal at d=
4.5 ppm was not identified.
Generation of 4b and 4c: The generation of 4b was at-
tempted by the enforced dehydration of 1b and the ther-
molysis of 3b, whereas the generation of 4c was attempted
by the thermolysis of 3c. The conditions in which 1a was
smoothly dehydrated to 2a (nitrobenzene, methanesulfonic
acid, 1308C, 20 min) could not be successfully applied to 1b;
no reaction was observed and the starting material remained
completely unchanged (by 1H NMR spectroscopy). If 1b
was treated with the stronger trifluoromethanesulfonic acid,
an instantaneous color change of the initial yellow-orange
solution to red and finally almost black was observed and its
1H NMR signals disappeared (Figure 5). At no point were
we able to collect evidence (e.g., mass spectrometric) for
the intermediary of the dihydrate 2b. The near-black
powder product was insoluble in common organic solvents
including chloroform, DCM, THF, toluene, DMF, and meth-
anol. It was therefore analyzed by CP MAS 13C NMR spec-
troscopy, for which it was necessary to allow the reaction to
run for 2 h (instead of minutes) and at 1608C (instead of
1308C) to ensure complete conversion.[13] As can be seen by
comparison with the starting material 1b, the signals of all
the saturated (benzylic) carbon atoms disappeared com-
pletely, whereas aromatic signals could still be observed
(Figure 6) although the shift range of the latter narrowed.
This comparison suggests a complete dehydration of the
starting cycle, although it is not unequivocal proof (see
below). Note, it was not possible to obtain any meaningful
mass spectrum of this material. Although the reason for this
Figure 6. CP MAS 13C NMR spectra of the tetrahydrate 1b and its dehy-
dration product obtained under enforced conditions.
is unclear, it cannot be excluded that cross-linking took
place.
To perform the acetate thermolysis also on the phenyl
series, the dihydrate 2b had to be generated first. However,
dihydrates were never observed as intermediates in the
above dehydration. To gain access to the key compound 2b,
highly dilute nitrobenzene solutions of 1b (<0.1 mm) were
treated with traces of trifluoromethanesulfonic acid. The
crude product of this reaction was analyzed by mass spec-
trometry and, in fact, the spectrum showed the molecular
ion of 2b at m/z 936. Without isolation, this material was
subjected to acetic anhydride/zinc chloride to convert the
two ether bridges into four acetate substituents. The product
mixture of this conversion was directly subjected to a
MALDI-FT mass spectrometric analysis, which gave the
spectrum shown in Figure 7. This spectrum not only shows
the expected molecular ion of 3b at m/z 1140, but, most in-
terestingly, also the whole series of acetic acid elimination
products of 3b, including a signal that would fit the molecu-
lar ion of the final target 4b at m/z 900! The signals at m/z
12166
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174

Fully Conjugated Double-Stranded Cycles
FULL PAPER
below. Further characterization of this mixture or even iso-
lation of the tetrahydrate 1c was not attempted because of
solubility issues. It was therefore decided to subject this
crude product directly to the sequence consisting of 1) meth-
anesulfonic acid catalyzed dehydration to the dihydrate 2c,
2) tetraacetate generation starting from 2c to give 3c, and
3) in situ generation of target 4c from 3c. The full spectrum
of the product obtained after this sequence is available in
Figure 8. As can be seen, the molecular ion of 4c at m/z 596
is observed, as are the signals of [M+2H]+ and [M+4H]+,
known from 4a and 4b. In addition, diverse protonation
products are observed. All relevant masses were confirmed
by high-resolution studies. Thus, the target molecular ion
can be obtained for the unsubstituted case!
Figure 7. MALDI-FT mass spectrum of the acetylation crude product of
3b showing the whole sequence of acetic acid elimination reactions start-
ing from the tetraacetate 3b to the final target “4b”. The signal groups at
m/z 902, 962, and 1022 are accompanied by low intensity signal groups at
m/z 920, 980, and 1039, which are a result of the addition of oxygen or
“water”.
900, 1081, and 1140 were confirmed by high-resolution tech-
niques and established the following formulae: C72H36,
C78H48O6 +H, and C80H52O8, respectively. These formulae
represent the target cycle and the triacetate and tetraacetate
precursors, respectively. If the same analysis was performed
at a higher laser energy, the intensities of the signals due to
the precursor molecules decreased whereas that of the
target cycle 4b increased (Figure S1). Clearly, the laser
beam affects thermolysis, a phenomenon that is less pro-
nounced within the hexyl series and may point towards a
higher tendency of the phenyl-based tetraacetate to suffer
elimination of acetic acid than the hexyl-based analogue.
MALDI-TOF mass spectrometry with a lower laser energy
was also applied. Although the spectra still showed the
signal at m/z 900, they looked more complex and thus this
technique was not further applied. Note that the molecular
ion peaks of the cycles under investigation appear in com-
plex signals groups that are a result of the removal and/or
up-take of hydrogen atoms, oxygen atoms, and water mole-
cules. It may therefore be that the most intense signal does
not correspond to the molecular ion of the target com-
pound.
After this encouraging finding, we attempted to obtain
similar evidence for the unsubstituted parent cycle 4c. The
last fully characterized compound of its synthesis was the
open-chain precursor 15 of the tetrahydrate 1c (Scheme S1).
This precursor was cyclized as normal and the crude product
mixture of this reaction was analyzed by MALDI-FT mass
spectrometry. The full spectrum with an expanded critical
range is presented in Figure S2. Despite the use of the crude
product, this mass spectrum is relatively simple and exhibits
the expected signals for 1c at m/z 691.1880 and 707.1907 for
[M+Na]+ and [M+K]+, respectively, in high resolution.
Thus, assuming a cyclic structure, there is no reasonable
doubt that the tetrahydrate was contained in the product
mixture. The possibility of ring-opening will be addressed
Figure 8. MALDI-FT mass spectrum of the crude product obtained after
the reaction sequence from 15 to 4c. The inset shows the enlarged critical
region at high-resolution of the molecular ion of 4c.
EPR spectroscopic studies of the crude products: As already
encountered in previous studies aimed at cycle II,[7a] in this
study it was also impossible to obtain reasonable solution
NMR spectra of the compounds even though mass spec-
trometry showed the buckybelt target molecular ion. Apart
from solvent signals, the NMR spectra only gave broad lines
and those of the aromatic hydrogen atoms even “disap-
peared” altogether. Figure 9 shows a typical spectrum. The
spectra recorded in [D5]nitrobenzene and [D6]DMSO at
high temperature (1008C) had the same appearance. Room
temperature spectra were not recorded.[14]
This not only hampered proof of the final structure but
also raised concerns as to the possible formation of radicals
during thermolysis and dehydration. The crude products ob-
tained from the thermolysis of tetraacetate 3a and the en-
forced dehydration of 1b were therefore investigated by
EPR spectroscopy. Figure 10 shows the EPR spectra at
room temperature, both with an absorption at g=2.0026.
This g value indicates a carbon-centered radical that is in-
volved in an extended p-conjugated system.[15] The forma-
tion of a triplet state could be excluded from the shapes and
widths of the signals. In particular, there is no zero-field
splitting. Although the radical concentrations could not be
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12167

A. D. Schlꢀter et al.
ditions and in the absence or presence of oxygen. They
always afforded a red spot that moved close to the solvent
front for solvents such as chloroform, methylene chloride, or
toluene. This spot was followed by an orange band, which in
turn was followed by a yellow band. The red spot decolor-
ized both upon standing (under nitrogen) and upon moving
further on the plate (under nitrogen). In this process, the
red color changed first to orange and then to yellow. If per-
formed under nitrogen, the process took longer. This indi-
cates that the bands following the red spot at the forefront
were decomposition products. Several attempts to isolate
and analyze any of the colored spots failed altogether. Even
attempts in which the silica gel with the red spot was
scratched off the TLC plate and analyzed directly by
MALDI-FT mass spectrometry did not lead to any conclu-
sive results. Either no signals were observed or the entire
collection that had already been obtained from the crude
mixture was seen. In line with this disappointing finding, all
conventional separations, including column chromatography
(using reversed-phase and Buckyprepꢄ columns) and recy-
cling gel permeation chromatography (GPC), failed to give
single fractions.
Figure 9. 1H NMR spectrum of a solution of the thermolysis crude prod-
uct of 3a in [D2]tetrachloroethane. Although there are a few low intensi-
ty signals that could not be assigned, the general appearance of the spec-
trum is in sharp contrast to the complexity of the mass spectrum obtained
from the identical mixture (see Figure 3). Signal assignment: *: phenan-
throline, **: solvent, and #: residual acetoxy groups.
Discussion
The main motivation behind the work reported herein was
to investigate structural alternatives to 4a that would not be
capable of undergoing hydrogen shifts. The work was also
encouraged by the hope that the generation of structures 4b
and 4c would possibly not be accompanied by the formation
of radicals, which so effectively prevented an NMR spectro-
scopic structure proof of 4a. How far did we get?
Figure 10. EPR spectra of the thermolysis products obtained from 3a
(left) and 3b (right).
The mass spectrometric analysis of the acetylation crude
product aiming at 3b gave the expected molecular ion of 4b
at m/z 900, which is the mass of the fully unsaturated target.
In addition, the entire fragmentation cascade, m/z 1140,
1080, 1020, and 960, was detected. These data point to a
“well-behaved” mechanism that leads to 4b. The matching
high-resolution mass spectrum of [M]+ at m/z 900.2812
(calcd. 900.2811) removes any lingering doubts regarding
the elemental composition of the molecular ion. Despite the
rather crude nature of the experiments aimed at the unsub-
stituted belt 4c, the fact that for both its precursor 3c and
the belt itself high-resolution mass spectrometric evidence
could be obtained suggests that we have found a method to
generate compounds of the general structure 4, at least
under mass spectrometric conditions. In contrast to the tet-
rahexyl congener 4a, the tetraphenyl belt and the unsubsti-
tuted belt have no option to escape their strain by rear-
rangements involving hydrogen shifts. Rearrangements of
the carbon skeleton in the applied temperature range are
also unlikely.[16] The three [M]+ peaks of 4a, 4b, and 4c are
therefore considered the first strong evidence for the target
belt. Clearly a molecular ion peak, even if highly resolved,
does not prove structure. Because of the unfortunate finding
that 4b is somehow associated with seemingly rather persis-
rigorously quantified, crude estimations based on the mass
of the starting material and an external tetramethylpiperi-
din-1-yloxyl (TEMPO) standard suggested trace amounts of
radicals in the hexyl case and a few percent in the phenyl
case (for details, see the Supporting Information). A temper-
ature-dependent investigation of line intensity revealed a
proportionality of signal amplitude to inverse temperature.
Such behavior is expected from Curieꢃs law if the concentra-
tion of the paramagnetic species is constant. The low con-
centration suggests that the major fraction of the buckybelts
4a and 4b at least are unaffected by radicals. Also, it rules
out any eventual disproportionation of the target cycle into
a radical cation and a radical anion as a possible mechanism
to reduce the strain energy. An equilibrium state of the radi-
cal species is ruled out by the constant concentration of the
radical. We consider it likely that the EPR signals are
caused by an unknown side-product.
Attempted purification revealing the transient nature of
thermolysis products: Silica gel TLC separations of the ther-
molysis (3a) and dehydration (2b) crude products of the
hexyl and phenyl series were performed under ambient con-
12168
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174

Fully Conjugated Double-Stranded Cycles
FULL PAPER
tent radicals (for 4c this aspect
was not investigated), NMR
spectroscopic investigations did
not lead anywhere. This is the
point at which the trapping ex-
periments with the presumed
cycle 4a come in. For a com-
pound with anthracene units
one would expect it to be able
to undergo DA reactions. As
described above, if 4a was gen-
erated in the presence of excess
2,9-dimethyl-1,10-phenanthro-
line, the mass spectrometric
analysis of the crude product
mixture showed, in addition to
unreacted cycle 4a, only two
groups of adduct signals at m/z
1140 and 1348. The simplest,
but not necessarily the only, ex-
planation is that 4a is trapped
once and twice. Again, note
that the adducts did not form in
the spectrometer but during
thermolysis. The reactivity of
phenanthrolines as dienophiles
is known and the computations
Scheme 1. Retro-DA reactions of 2b and 3 (see text for discussion); a) R=C6H13, b) Ph, c) H.
clearly show that DA reactions
are energetically attractive.[11]
In addition to these mass spectrometric considerations,
the solid-state NMR analysis of the dehydration product of
1b is of importance. The spectra in Figure 6 prove that the
four “water” molecules contained in 1b can be removed vir-
tually quantitatively. Knowing that 1a easily dehydrates to
2a, it is proposed that the dehydration of 1b to 4b passes
through the same intermediate, namely 2b.[17] It is thus
tempting to propose that the product spectrum in Figure 6
belongs to 4b. However, the proposed intermediate 2b has
the potential option to undergo a ring-opening reaction fol-
lowing a (acid-catalyzed) retro-DA path.[18] Such a path
would result in the formation of compound 5b (Scheme 1),
which is unlikely to undergo a subsequent retro-DA reaction
because of the already liberated ring strain. Under the ap-
plied conditions, compound 5b would undoubtedly be dehy-
drated to give 6b. The existence of 6b, however, cannot be
excluded because both its end groups should have chemical
shifts well within the observed range. They would, therefore,
be obscured by the signals of the aromatic carbon atoms. Al-
though the mass spectrometric evidence for the retro-DA
reaction has no bearing on the behavior of 2b in acidic solu-
tion, such an unwanted reaction cannot be rigorously dis-
proved in the present state of experimentation.
dienes and olefins are known, we were not able to find the
corresponding retro-reactions. This is in line with mass spec-
trometric observations; in contrast to all buckybelt di- and
tetrahydrates, the tetraacetates 3 do not show any related
signals at half mass (for example, see Figure 7). Apart from
the question as to whether or not tetraacetate cycles can
open up, the most probable initial products of such an open-
ing, compounds 7, do not have a realistic route for losing
the two terminal acetic acids. Such a loss would require a
carbene and/or other high-energy intermediary that are un-
reasonable to propose. It is therefore concluded that the
molecular ions for compounds 4 refer, in fact, to cyclic spe-
cies.
Finally, the UV/Vis spectrum of the fraction of 4a, which
had been analyzed by mass spectrometry and exhibited the
pattern m/z 932, 934, and 936,[7a] is shown in Figure 11.[19] It
has a longest wavelength absorption at approximately lmax
=
550 nm, which is where one would intuitively expect it on
[20]
the basis of UV/Vis data for C60 and C70 as well as differ-
ent C84 isomers[21] and an open-chain buckyribbon (lmax
=
611 nm, dichloromethane).[22] We consider it reasonable that
this spectrum is not due to the products of two-fold [1,5]-hy-
drogen-shifted 4a.
It is worthwhile, in this context, to also have a look at the
acetate route and to assess whether open-chain products
could possibly be involved. The most likely retro-DA prod-
uct 7 of any of the tetraacetates 3 is shown in Scheme 1. Al-
though DA addition reactions between 1,4-diacetoxybuta-
The synthetic endeavors aimed at fully unsaturated
double-stranded cycles have thus reached a rather exciting
level. During the last few years different laboratories have
got so much closer to the goal than in all the preceding
years together. Especially noteworthy is the achievement of
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12169

A. D. Schlꢀter et al.
0.5 kcalmolꢀ1. Thus, it was decided to obtain the thermo-
chemical data for the dehydration reactions of 1–3 from ho-
modesmic/isodesmic-type equations.
Figure 11. UV/Vis spectrum in chloroform of a GPC fraction obtained
from the thermolysis of tetraacetate 3a.
Water is eliminated from 1c to yield dehydro-1c and D
[reaction (7)]: Two products are possible, dehydro-1c-1 and
dehydro-1c-2 (Figure 12).
a conjugated halfcycle with its enormous curvature by Bod-
well and co-workers.[5a] This shows that strong bending of
conjugated compounds is not an insurmountable obstacle
per se and thus supports the credo of this publication. We
have presented circumstantial evidence, each piece of which
can be argued, the sum of which, however, proves beyond
reasonable doubt that at least for the phenyl- and hydrogen-
substituted cases, fully conjugated double-stranded cycles of
the general structure II have been obtained in the gas phase
and may have been obtained in the solution phase. Regard-
ing the latter, the reader is referred to the section below in
which theoretical considerations of the dehydration raise
questions as to its feasibility. However, substantial effort is
still required to find ways to isolate the compounds and ide-
ally grow single crystals for structure analysis.
1 c þ E ! dehydro-1 c þ D
ð7Þ
Figure 12. Computed structure of dehydro-1c-1 (left) and dehydro-1c-2
(right).
Reaction (7) is exothermic by 4.2 kcalmolꢀ1 when leading
to dehydro-1c-1 and endothermic by 60.3 kcalmolꢀ1 when
yielding dehydro-1c-2. The free energies for these reactions
are similar to the enthalpies of the reactions: ꢀ4.9 and
+59.5 kcalmolꢀ1, respectively. Because the elimination of
water from D is exothermic by 19.4 kcalmolꢀ1 [reaction (4)],
it can be concluded that the elimination of water from 1c to
yield dehydro-1c-1 is exothermic by 23.6 kcalmolꢀ1 whereas
the elimination to yield dehydro-1c-2 is endothermic by
40.9 kcalmolꢀ1. The reason for this difference becomes evi-
dent when the structures of the two dehydro-1c isomers are
studied. As was shown in previous studies,[7] the chemistry
of the system is governed by
Theoretical aspects on the sequential elimination of water
from 1: Sequential dehydration of 1 leads to systems that
contain zero, one, or two D-, E-, and F-like (anthracene)
fragments. Thus, as model reactions in nonstrained systems,
we studied the dehydration of D!E!F [reactions (4)–(6)]
at the G3 and B3LYP/6-311G(d) computational levels. The
results, summarized in Table 1, indicate that the B3LYP/6-
311G(d) results are less exothermic than the G3 results by
an almost constant value of around 2 kcalmolꢀ1 for reac-
tions (4) and (5). However, when a homodesmic reaction
was devised [reaction (6)], this difference is reduced to 0.1–
the build-up or release of strain
following the reactions. Because
only a little strain (or none at
all) is released in the transition
from 1c to dehydro-1c-1 (the
structure of 1c is discussed
below), whereas the conjuga-
tion gained is larger than in the
model reaction, the elimination
Table 1. Results of thermochemical calculations of reactions (4)–(6) at the G3 and B3LYP/6-311G(d) levels of
theory.
Reaction
G3
B3LYP
DH8
DH
(0 K)
DH8
DG8
DH
(0 K)
DG8
[kcalmolꢀ1
]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
[kcalmolꢀ1
]
4
5
6
ꢀ21.4
ꢀ16.0
ꢀ5.4
ꢀ19.4
ꢀ14.0
ꢀ5.4
ꢀ32.2
ꢀ26.5
ꢀ5.8
ꢀ19.6
ꢀ13.7
ꢀ5.9
ꢀ17.7
ꢀ11.8
ꢀ5.9
ꢀ30.1
ꢀ24.2
ꢀ5.9
12170
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174

Fully Conjugated Double-Stranded Cycles
FULL PAPER
has almost the same exothermicity as the elimination of
water from D. However, due to the large curvature of the
aromatic moiety that is formed during the elimination of
water to yield dehydro-1c-2, this reaction becomes highly
endothermic. Clearly, as is also evident from the structure of
2c, which cannot be formed from dehydro-1c-2, the latter is
not an intermediate in the reaction that forms 2 from 1 and
therefore is not considered further.
Elimination of a second water molecule may lead to three
different isomers. This can happen either from the D-like
moiety or from the E-like moiety in dehydro-1c-1 (hence,
dehydro-1c). Thus, the elimination reactions of water from
dehydro-1c to D or E are represented by the isodesmic re-
actions (8)–(10). The dehydration of dehydro-1c to 2c-1 is
more exothermic than the dehydration of E by 4.5 kcal
molꢀ1. This means that the absolute exothermicity of this re-
action is approximately 24 kcalmolꢀ1, which is in accord
with the fact that dehydro-1c was never isolated. The first
and second water eliminations have similar exothermicities
(the second elimination is somewhat more exothermic) and,
assuming that the TSs are similar, both occur under the
same conditions. The two other possibilities for elimination
(leading to 2c-2 and 2c-3, reactions (9) and (10), respective-
ly) are more endothermic by around 52 and 68 kcalmolꢀ1
relative to D and E, respectively, which explains why 2c-1 is
the only isomer that is isolated from the reaction.
dro-1c such that almost no strain is released or introduced
on dehydration and at the same time the conjugation that is
gained is greater than in the model compound. However,
2c-2 and 2c-3 are much more curved than dehydro-1c, a
fact that is manifested by additional strain energy, which
causes the overall dehydration reaction to be endothermic.
The third elimination [reaction (11)] is endothermic by
58 kcalmolꢀ1 relative to E. This, too, can be explained by
the shape of the molecule (Figure 13) because a large
degree of curvature is introduced into the structure upon de-
hydration. The absolute enthalpy of this reaction is approxi-
mately +44 kcalmolꢀ1. This result explains the need for a
different elimination procedure, as is described in the exper-
imental procedure for 3a.
2 c-3 þ F ! 4 c-hydrate þ E
ð11Þ
DH ¼ 58:0 kcal molꢀ1; DGꢁ ¼ 57:6 kcal molꢀ1
The elimination of the last water molecule is endothermic
by around 29 kcalmolꢀ1 [reaction (12)] relative to E due to
the additional strain that is inherent in the resulting struc-
ture (Figure 3). Altogether, this renders the process endo-
thermic by 15 kcalmolꢀ1, once again requiring a synthetic
route different from simple water elimination.
4 c-hydrate þ F ! 4 c þ E
ð12Þ
DH ¼ 29:4 kcal molꢀ1; DGꢁ ¼ 30:3 kcal molꢀ1
dehydro-1 c þ E ! 2 c-1 þ D
ð8Þ
DH ¼ ꢀ4:5 kcal molꢀ1; DGꢁ ¼ ꢀ4:6 kcal molꢀ1
It is noted that the last elimination is less endothermic
than the elimination of the third water molecule [reac-
tions (12) and (11), respectively] and therefore if the elimi-
nation from 3 to give 4 was to involve sequential aromatiza-
tion (i.e., elimination of two HOAc from one ring and only
then elimination from the other ring), then the reaction
would easily yield 4. The fact that this does not occur sug-
gests that the process proceeds by elimination of the first
two HOAc from different rings and aromatization only
occurs in the following steps.
There are several isomers of 1c that can lead to 2c. We
have considered in the discussion above the experimentally
isolated isomer 1c-1. However, the more symmetric isomer,
1c-2 (Figure 14), in which all the (eliminated) hydrogen
atoms are syn with respect to the oxygen atoms, was consid-
ered as well. The isomer 1c-2 is less stable than 1c-1 by
9.4 kcalmolꢀ1. This can be rationalized by the flatness (thus
dehydro-1 c þ E ! 2 c-2 þ D
ð9Þ
DH ¼ 51:8 kcal molꢀ1; DGꢁ ¼ 51:6 kcal molꢀ1
dehydro-1 c þ F ! 2 c-3 þ E
ð10Þ
DH ¼ 67:9 kcal molꢀ1; DGꢁ ¼ 67:6 kcal molꢀ1
The structures of the products (Figure 13) explain the re-
sults on the basis of acquired and released strain. The curva-
ture in 2c-1 is minimal and similar to the curvature in dehy-
Figure 13. Calculated structures of compounds 2c-1 (top left), 2c-2 (top
center), 2c-3 (top right), 4c-hydrate (bottom left), and 4c (bottom right).
Figure 14. Side-view of 1c-1 (left) and 1c-2 (right).
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12171

A. D. Schlꢀter et al.
pounds were carried out under nitrogen using standard Schlenk tech-
niques and dry solvents. All commercial reagents were used without fur-
ther purification. Solvents were purified and dried by standard proce-
dures.
strainless) naphthalenic moieties in 1c-1, whereas the same
moieties are somewhat curved in 1c-2. This curvature causes
each subunit to be slightly strained (i.e., less than 2.4 kcal
molꢀ1), but this adds up to almost 10 kcalmolꢀ1 for the four
naphthalene subunits and probably accounts for the experi-
mental isolation of only 1c-1.
Analytical methods
Chromatography: Reactions and columns were monitored by thin-layer
chromatography (TLC) using TLC silica gel coated aluminium plates 60
UV254 (Merck “Kieselgel 60” or Macherey–Nagel with fluorescence indi-
cator F254) and visualized by ultraviolet light (l=254 and 366 nm).
As mentioned in the results section, 1b is completely
stable under the conditions used for the conversion of 1a
into 2a. It is rather unlikely that the thermodynamics of the
elimination changes dramatically with the substituents (i.e.,
a vs. b). However, the kinetics may change due to stabiliza-
tion or destabilization of reactive intermediates (probably
cations, or, less likely but still possible, radicals), which may
open up different reaction channels. Under more forceful
conditions the elimination proceeded to completion but the
product could not be identified. The computational results
suggest an endothermicity of 44 kcalmolꢀ1 for this elimina-
tion. Assuming a value of DS of 30–40 entropy units for the
reaction, this means that the elimination of the third water
molecule from 2 will have a negative free energy of reaction
at temperatures of 830–11908C. The calculations were car-
ried out in the gas phase and solvent may change the entro-
py of the reaction. However, because the reaction was car-
ried out at 1608C, obtaining the product would suggest that
DS for the elimination reaction is +102 entropy units, an
unreasonable number for such a reaction. Alternatively, if
30–40 entropy units for the elimination are assumed, it im-
plies that the DH of the reaction is not larger than around
13–17 kcalmolꢀ1. It is unlikely that the error in the calcula-
tions is so large. Therefore, the calculations of the dehydra-
tion presented above do not support the formation of 4b
under the forceful elimination conditions. Rather, the option
that is described in Scheme 1 (in which 6b may lose a water
molecule in the MS) or a (cationic or radical) polymeri-
zation of the cyclic skeleton under these conditions seem to
be more reasonable option.
1
NMR spectroscopy: H and 13C NMR spectra were recorded on a Bruker
AVANCE 300 (1H: 300 MHz; 13C: 75 MHz) or AVANCE 500 (1H:
500 MHz; 13C: 125 MHz) spectrometer at normal probe temperatures
using [D]chloroform, 1,1,2,2-[D2]tetrachloroethane, or [D5]nitrobenzene.
Signals of residual nondeuterated solvent molecules or TMS were used
as internal standard for the calibration of the 1H and 13C NMR spectra.
Solid-state NMR measurements were performed on a Varian InfinityPlus
console using a Chemagnetics 4 mm double-resonance MAS probehead
and vortex tube cooling. A standard variable-amplitude cross polariza-
tion sequence with XIX proton decoupling (120 kHz rf amplitude) during
the acquisition was applied. Saturation pulses were applied before each
scan to ensure the same initial condition. A contact time of 2 and 4 ms
was chosen. A high-order baseline correction was applied to remove the
background signal from the probehead. This was found to be more reli-
AHCTUNGERTGaNNUN ble than a background subtraction. All data were processed by using the
matNMR processing toolbox.
Mass spectrometry: High-resolution mass spectra were recorded by the
MS service of the Laboratorium fꢀr Organische Chemie at ETH Zꢀrich
on a Varian IonSpec Ultima MALDI-FT-ICR, a Bruker Daltonics Ultra-
Flex II MALDI-TOF, or a Waters Micromass Autospec Ultima EI-EBE-
triSector instrument. trans-2-[3-(4-tert-Butylphenyl)-2-methyl-2-propenyli-
dene]malononitrile (DCTB) and 3-hydroxypyridine-2-carboxylic acid (3-
HPA) served as matrices for MALDI mass spectrometry.
UV/Vis spectroscopy: UV/Vis spectra were recorded at room temperature
on a Perkin–Elmer Lambda 19 instrument using standard 1 cm quartz
glass cuvettes.
EPR spectroscopy: The EPR spectra were measured with an ElexSys
E680 X-/W-band dual frequency spectrometer (Bruker Biospin GmbH)
in X-band mode at typical microwave frequencies of 9.7 GHz. An elec-
tron-nuclear double resonator EN 4118X-MD-4 (Bruker Biospin GmbH)
was used for both continuous wave (CW) EPR and Mims ENDOR mea-
AHCTUNGERTGsNNUN urements. For CW EPR the resonator was critically coupled, whereas it
was overcoupled to Qꢂ100 to obtain sufficiently short dead times in
Mims ENDOR. For Mims ENDOR a model 250A250A radiofrequency
amplifier with a maximum power of 250 W (Amplifier Research, Inc.)
was used. Measurement temperatures of 160 K for CW EPR and 80 K
for Mims ENDOR were maintained with a CF935 cryostat and ITC 503S
temperature controller (Oxford Instruments). Saturation behavior was
checked for all samples in CW EPR experiments. An attenuation of
30 dB (microwave power of 0.2 mW) was found to be sufficient to avoid
separation. Determination of spin concentration was double checked at
40 dB attenuation (0.02 mW). Glassy frozen solutions of 2,2,6,6-tetrame-
thylpiperidin-1-yloxyl (TEMPO) were used as concentration standards
and for g value reference. Modulation amplitudes of 0.1 or 0.2 mT peak-
to-peak were used according to line width. In Mims ENDOR, the pulse
lengths were 20 ns for microwave p/2 pulses and 12 ms for the radiofre-
quency pulse, the range from 7–22 MHz was scanned to detect all proton
ENDOR signals.
Conclusion
High-resolution mass spectra that match fully conjugated
double-stranded cycles 4b and 4c were obtained. In contrast
with their hexyl-substituted congener 4a, these two cycles
(phenyl-substituted and parent) do not have the option to
undergo a [1,5] hydrogen shift that would lead to a disrup-
tion of conjugation. It is, therefore, concluded that these in-
triguing, long-sought-for aromatic belt-type compounds
were generated for the first time, at least in the gas phase.
Although cycle 4a could be trapped with phenanthroline, at-
tempts to isolate products 4a–4c failed.
Syntheses
Tetraphenyl-buckybelt (4b): A suspension of the tetraphenyl-buckybelt
tetrahydrate 1b (20 mg, 0.02 mmol) in nitrobenzene (10 mL) was heated
to 1608C. Upon addition of trifluoromethanesulfonic acid (25 mL,
0.29 mmol) the mixture turned black instantaneously. After 2 h, the sol-
vent and acid were removed by using a rotary evaporator and the residu-
al black material was washed with toluene (5ꢅ2 mL) and dried in high
vacuum.
Experimental Section
The procedures for the synthesis of compounds 1b and 1c are described
in the Supporting Information. All compounds of the hexyl route have
previously been described.[7a] All reactions involving air-sensitive com-
CPMAS 13C NMR (220 MHz): d=140–110 (peaks at d=138, 128, and
120 ppm).
12172
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174

Fully Conjugated Double-Stranded Cycles
FULL PAPER
Tetraphenyl-buckybelt dihydrate (2b) and buckybelt dihydrate (2c): A so-
lution of tetrahydrate 1b (100 mg, 0.10 mmol) or 1c (80 mg of crude ma-
terial) in nitrobenzene (1b: 1500 mL; 1c: 50 mL) was heated to 1608C
(1c: 1408C) and treated with trifluoromethanesulfonic acid (1b: 20 mL,
0.23 mmol; 1c: methanesulfonic acid, 20 mL, 0.31 mmol) for 2 h (1c:
20 min). The solvent was removed by using a rotary evaporator. The re-
sidual black solid was washed with toluene (1b: 5ꢅ10 mL; 1c: benzene,
5ꢅ5 mL) and dried under high vacuum.
Engl. 1987, 26, 892–894; F. H. Kohnke, J. F. Stoddart, Pure Appl.
Chem. 1989, 61, 1581–1586.
[5] a) B. L. Merner, L. N. Dawe, G. J. Bodwell, Angew. Chem. 2009, 121,
5595–5599; Angew. Chem. Int. Ed. 2009, 48, 5487–5491. Also, see:
b) G. J. Bodwell, J. N. Bridson, M. Cyranꢃski, J. W. J. Kennedy, T. M.
Krygowski, M. R. Mannion, D. O. Miller, J. Org. Chem. 2003, 68,
2089–2098; c) G. J. Bodwell, D. O. Miller, R. J. Vermeij, Org.Lett.
2001, 3, 2093–2096; d) G. J. Bodwell, J. J. Fleming, D. O. Miller, Tet-
rahedron 2001, 57, 3577–3585; e) G. J. Bodwell, J. J. Fleming, M. R.
Mannion, D. O. Miller, J. Org. Chem. 2000, 65, 5360–5370; f) G. J.
Bodwell, J. N. Bridson, T. J. Houghton, J. W. J. Kennedy, M. R. Man-
nion, Chem. Eur. J. 1999, 5, 1823–1827; g) G. J. Bodwell, J. N. Brid-
son, T. J. Houghton, J. W. J. Kennedy, M. R. Mannion, Angew.
Chem. 1996, 108, 1418–1420; Angew. Chem. Int. Ed. Engl. 1996, 35,
1320–1321.
[6] a) Personal communication with M. Iyoda; b) Y. Kuwatani, J.-I. Igar-
ashi, M. Iyoda, Tetrahedron Lett. 2004, 45, 359–362; c) T. Yoshida,
Y. Kuwatani, K. Hara, M. Yoshida, H. Matsuyama, M. Iyoda, S.
Nagase, Tetrahedron Lett. 2001, 42, 53–56; d) Y. Kuwatani, T. Yoshi-
da, K. Hara, M. Yoshida, H. Matsuyama, M. Iyoda, Org. Lett. 2000,
2, 4017–4020.
[7] a) C. Denekamp, A. Etinger, W. Amrein, A. Stanger, M. Stuparu,
A. D. Schlꢀter, Chem. Eur. J. 2008, 14, 1628–1637; b) M. Stuparu, V.
Gramlich, A. Stanger, A. D. Schlꢀter, J. Org. Chem. 2007, 72, 424–
430; c) M. Stuparu, D. Lentz, H. Rꢀegger, A. D. Schlꢀter, Eur. J.
Org. Chem. 2007, 88–100; d) W. D. Neudorff, D. Lentz, M. Anibar-
ro, A. D. Schlꢀter, Chem. Eur. J. 2003, 9, 2745–2757.
[8] a) J. E. Norton, B. H. Northrop, C. Nuckolls, K. N. Houk, Org. Lett.
2006, 8, 4915–4918; b) E. Clar, J. W. Wright, Nature 1949, 163, 921–
922; c) E. Clar, Chem. Ber. 1949, 82, 495–514.
[9] Malte Standera, PhD Thesis, ETH Zurich (Switzerland), 2010.
[10] Whether there are phenanthroline adducts derived from a reaction
with the compound with m/z 936 cannot be decided because of
signal complexity. If the perception is correct (see ref. [7a]), that the
signals at m/z 934 and 936 are due to the addition of two and four
hydrogen atoms to the 9,10-positions of the one and two anthra-
cenes of 4a, respectively, the compound with m/z 936 would not
have an appropriate site to undergo a Diels–Alder reaction.
[11] For phenanthrolines as dienophiles see the work presented at the
third Brisbane biological and organic chemistry symposium: D. Mar-
getic, R. N. Warrener, D. N. Butler, 1,3-Dipolar Cycloadditions to
Polycyclic Aromatic Hydrocarbons, the Queensland University of
Technology, Brisbane, 2002, .
2b: MALDI-FT HRMS: calcd. for [M]+: 936.3023; found: 936.3023.
2c: The MALDI-FT mass spectrum did not show any conclusive signals.
Tetraphenyl-buckybelt-tetraacetate (3b) and buckybelt-tetraacetate (3c):
The crude buckybelt-dihydrate 2b (20 mg) or 2c (30 mg) were suspended
in acetic anhydride (2b: 10 mL; 2c: 5 mL) at 608C. A crystal of zinc
chloride was added and the reaction mixture was stirred for 6 h (2c:
overnight) before the solids were filtered off by suction. After washing
with ethanol (2b: 3ꢅ5 mL; 2c: 5ꢅ3 mL), the crude product was dried
under high vacuum. Because crude products were used, the amounts of
products in mmol cannot be given.
3b: MS (MALDI-FT): m/z: 1140 [M]+, 1080 [Mꢀ60]+, 1020 [Mꢀ120]+,
960 [Mꢀ180]+, 904 [Mꢀ236]+, 902 [Mꢀ238]+, 900 [Mꢀ240]+. HRMS
(MALDI-FT): calcd. for [M]+: 1140.3657; found: 1140.3657; calcd. for
[MꢀAcOH+H]+: 1081.3523; found: 1081.3524; calcd. for [Mꢀ4AcOH]+
: 900.2811; found: 900.2812.
3c: MS (MALDI-FT): m/z: 596 [M]+, 536 [Mꢀ60]+, 476 [Mꢀ120]+, 416
[Mꢀ180]+, 360 [Mꢀ236]+, 358 [Mꢀ238]+, 356 [Mꢀ240]+. HRMS
(MALDI-FT): calcd. for [Mꢀ4AcOH]+: 596.1565; found: 596.1560.
Computations: The Gaussian 03[23,24] software package was used for the
computations. All the systems underwent full geometry optimization at
the B3LYP/6-311G(d) computational level. Frequency (i.e., analytical
second derivatives of the energy with respect to 3Nꢀ6 degrees of free-
dom) calculations were undertaken to verify the energy minima (the
number of imaginary frequencies, Nimag =0) and to allow thermodynamic
calculations. Model systems, namely water, anthracene dihydrate, anthra-
cene hydrate, and anthracene, were also calculated at the G3 level to
obtain accurate energy data according to which the B3LYP/6-311G(d)
calculations could be calibrated.
Acknowledgements
This work was supported by the Swiss National Funds (SNF 200020-
113607 and 200020-126451). We are very thankful to Professor Lawren-
ce T. Scott, Boston College, for sharing with us some of his thoughts on
strain effects in aromatic compounds. We thank Dr. M. C. Stuparu for
the UV spectrum of 4a and C. Szabados for help with some of the ex-
periments.
[12] A. Koshio, M. Inakuma, T. Sugai, H. Shinohara, J. Am. Chem. Soc.
2000, 122, 398–399.
[13] The material obtained under milder conditions still contained resid-
ual water molecules.
[14] R. H. Mitchell, R. V. Williams, T. W. Dingle, J. Am. Chem. Soc.
1982, 104, 2560–2571.
[15] F. Gerson, W. Huber Electronspin Resonance Spectroscopy of Or-
ganic Radicals, Wiley-VCH, Weinheim, 2003.
[1] a) B. D. Steinberg, L. T. Scott, Angew. Chem. 2009, 121, 5504–5507;
Angew. Chem. Int. Ed. 2009, 48, 5400–5402; b) R. Gleiter, B. Esser,
S. C. Kornmeyer, Acc. Chem. Res. 2009, 42, 1108–1116; c) T. Yao,
H. Yu, R. J. Vermeij, G. J. Bodwell, Pure Appl. Chem. 2008, 80,
533–548; d) K. Tahara, Y. Tobe, Chem. Rev. 2006, 106, 5274–5290;
e) R. Gleiter, B. Hellbach, S. Gath, R. J. Schaller, Pure Appl. Chem.
2006, 78, 699–706; f) R. Herges in Modern Cyclophane Chemistry
(Eds.: R. Gleiter, H. Hopf), Wiley-VCH, Weinheim, 2004, pp. 337–
358; g) L. T. Scott, Angew. Chem. 2003, 115, 4265–4267; Angew.
Chem. Int. Ed. 2003, 42, 4133–4135; h) A. Schrçder, H.-B. Mekel-
burger, F. Vçgtle, Top. Curr. Chem. 1994, 172, 179–201; i) F. H.
Kohnke, J. P. Mathias, J. F. Stoddart, Top. Curr. Chem. 1993, 165, 1–
69.
[16] A. J. Stone, D. J. Wales, Chem. Phys. Lett. 1986, 128, 501–503. It
cannot be excluded that Stone–Wales rearrangements take place
under mass spectrometric conditions. They would lead to different
structures than the proposed ones here, but would not affect
double-strandedness.
[17] For 2b a high-resolution mass spectrum shows [M]+ at m/z 936.3023
(calcd: 936.3023).
[18] Note that as for 2a, the mass spectra of 2b and 2c also show signals
due to retro-DA products. For the latter they appear as [M/2+Na]+
and [M/2+K]+.
[19] M. Stuparu, PhD Thesis, ETH Zurich (Switzerland), 2007.
[20] H. Ajie, M. M. Alvarez, S. J. Anz, R. D. Beck, F. Diederich, K. Fos-
tiropoulos, R. Huffman, W. Krꢁtschmer, Y. Rubin, K. E. Schriver, D.
Sensharma, R. L. Whetten, J. Phys. Chem. 1990, 94, 8630–8633.
[21] J. Crassous, J. Rivera, N. S. Fender, L. Shu, L. Echegoyen, C. Thil-
gen, A. Herrmann, F. Diederich, Angew. Chem. 1999, 111, 1716–
1721; Angew. Chem. Int. Ed. 1999, 38, 1613–1617.
[2] A. D. Schlꢀter, Adv. Mater. 1991, 3, 282–291.
[3] J. Sakamoto, J. van Heijst, O. Lukin, A. D. Schlꢀter, Angew. Chem.
2009, 121, 1048–1089; Angew. Chem. Int. Ed. 2009, 48, 1030–1069.
[4] For example, see: F. H. Kohnke, A. M. Z. Slawin, J. F. Stoddart, D. J.
Williams, Angew. Chem. 1987, 99, 941–943; Angew. Chem. Int. Ed.
Chem. Eur. J. 2011, 17, 12163 – 12174
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
12173

A. D. Schlꢀter et al.
[22] B. Schlicke, A. D. Schlꢀter, P. Hauser, J. Heinze, Angew. Chem.
1997, 109, 2091–2093; Angew. Chem. Int. Ed. Engl. 1997, 36, 1996–
1998.
Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari,
J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cio-
slowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaro-
mi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng,
A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W.
Chen, M. W. Wong, C. Gonzalez, J. A. Pople, Gaussian, Inc., Wall-
ingford CT, 2004.
[23] Gaussian 03 Revision D.02, M. J. Frisch, G. W. Trucks, H. B. Schle-
gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgom-
ery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S.
Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani,
N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda,
O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J.
Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
[24] L. A. Curtiss, K. Raghavachari, P. C. Redfern, V. Rassolov, J. A.
Pople, J. Chem. Phys. 1998, 109, 7764–7776.
Received: February 9, 2011
Revised: May 13, 2011
Published online: September 14, 2011
12174
www.chemeurj.org
ꢂ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 12163 – 12174