The fold-in approach to bowl-shaped aromatic compounds: Synthesis of chrysaoroles

Article abstract of DOI:10.1002/anie.201208547

Molecular jellyfish: A new family of bowl-shaped aromatic compounds swims into view (see picture). Quite unlike true jellyfish, chrysaoroles possess a rigid skeleton, which is assembled from fused carbazole units. Their synthesis involves a fold-in step to convert a macrocyclic precursor into the bowl-shaped target molecule. Copyright

Full text of DOI:10.1002/anie.201208547

Angewandte
Chemie
DOI: 10.1002/anie.201208547
Geodesic Polyarenes
The Fold-In Approach to Bowl-Shaped Aromatic Compounds:
Synthesis of Chrysaoroles**
´
´
Damian Mysliwiec and Marcin Ste˛pien*
The chemistry of bowl-shaped aromatic compounds, known
as buckybowls and geodesic polyarenes, is an area of
significant theoretical and practical interest.[1] The current
extensive research on these systems has been largely moti-
vated by their structural relationship to fullerenes and carbon
nanotubes. Indeed, all-carbon buckybowls, such as corannu-
lene[1b,c,e,2] and sumanene,[1d,3] can be viewed as nanotube end
caps or fullerene sections, and have been explored as
templates for controlled syntheses of well-defined molecular
forms of elemental carbon.[4] However, the interest in bowl-
shaped aromatic compounds extends beyond the quest for
new carbon-rich materials and encompasses aspects of their
internal strain,[5] aromaticity,[6] metal coordination,[1d,7] and
supramolecular chemistry.[1e] All these facets of research
create the need for new structural motifs and synthetic
methodologies.[1f,8]
The crucial points in every synthesis targeting a bowl-
shaped aromatic structure are the choice of chemical reac-
tivity capable of incorporating strain into the p-electron
system being constructed, and the placement of the strain-
inducing step in the overall synthetic plan. In the synthesis of
corannulene by Lawton and Barth,[2] the bowl was con-
structed by multiple annulations around the central five-
membered ring. In the final step, responsible for the
introduction of strain, a partly saturated corannulene pre-
cursor was subjected to catalytic dehydrogenation to yield the
target molecule. Strain buildup was similarly postponed until
the final synthetic step in the first successful preparation of
sumanene.[3] However, in many contemporary syntheses of
corannulene and higher geodesic hydrocarbons,[1,4,8] unsatu-
rated rings have been closed efficiently with the concomitant
introduction of strain. This approach has been successful with
a number of specialized reaction types, most notably high-
temperature pyrolysis and metal-mediated coupling.[1b,c,e]
Interestingly, all of the above approaches rely on a common
“stitching”[1f] tactic: the bowl is elaborated from the center
Figure 1. Synthetic approaches to bowl-shaped aromatic compounds.
(“hub”) towards the rim (Figure 1). One can envisage
a complementary strategy, which begins with a macrocyclic
precursor containing the complete rim of the bowl. The
precursor will consist of a number of aromatic subunits
(shown as yellow trapezoids in Figure 1), which can be
“folded in” and coupled so as to complete the central part of
the bowl. A fold-in synthesis is potentially difficult to design,
because its outcome will depend not only on the geometrical
matching of subunits but also on the extent of conformational
flexibility of the macrocycle, and this conformational flexi-
bility may change in the course of the folding process.
However, by reversing the sequence of bond formation in the
fold-in synthesis, different reactivity patterns that could lead
to otherwise inaccessible systems may be explored. To test the
viability of the fold-in approach, we selected carbazole-based
bowls as initial targets. Our choice was motivated by geo-
metrical and reactivity considerations and, in part, by the
scarcity of bowl-shaped heteroaromatic compounds.[9]
Our reaction sequence starts with two carbazole deriva-
tives (Scheme 1): the dialdehyde 1 and bis(phosphonium) salt
2, which were prepared from 2,7-dibromocarbazole as
described in the Supporting Information. The treatment of 2
with dialdehyde 1 (4 equiv) under Wittig conditions yielded
tricarbazole 3 in 70% yield. This reaction proceeded with
good Z selectivity owing to the beneficial effect of ortho
halogen substituents.[10] However, the Z,Z isomer of 3 is
highly photosensitive and undergoes rapid isomerization to
the insoluble E,E form in ambient light. The use of excess
´
´
[*] D. Mysliwiec, Dr. M. Ste˛pien
Wydział Chemii, Uniwersytet Wrocławski
ul. F. Joliot-Curie 14, 50-383 Wrocław (Poland)
E-mail: marcin.stepien@chem.uni.wroc.pl
Homepage: http://www.mstepien.edu.pl
[**] Financial support from the National Science Center (grant N N204
199340) is gratefully acknowledged. Quantum-chemical calcula-
tions were performed in the Wrocław Center for Networking and
Supercomputing. We thank Prof. Tadeusz Lis for solving the X-ray
crystal structure of 4 and for helpful discussions. We thank Dr. Piotr
Stefanowicz and Dr. Piotr Jakimowicz for their assistance with mass
spectrometry.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201208547.
Angew. Chem. Int. Ed. 2013, 52, 1713 –1717
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1713

.
Angewandte
Communications
Scheme 1. Synthesis of chrysaoroles 5 and 7 (R=nBu): a) aqueous NaOH, CH2Cl2, 1 h, Z,Z/Z,Eꢀ4:1; b) TiCl4, Zn, CuI, THF, reflux, 12 h, slow
addition; c) [Ni(cod)2], 1,5-cyclooctadiene, 2,2’-bipyridyl, DMF/toluene, 858C, 18 h; d) H2, PtO2, CH2Cl2, 208C. cod=1,5-cyclooctadiene,
DMF=N,N-dimethylformamide.
dialdehyde 1 in the synthesis of 3 is helpful in minimizing the
extent of polymerization. The macrocyclization of 3 by
McMurry coupling under pseudo-high-dilution conditions
provided carbazolophane 4 in 87% yield. The direct cyclo-
trimerization of 1 under McMurry conditions was also
explored as a potentially simpler route to 4, but this
alternative procedure suffered from lower yields (ca. 8%)
and purification difficulties.
The structure of 4, the first macrocycle en route to the
bowl-shaped target, was of significant interest. The inspection
of molecular models had shown that multiple substitution of
the cyclophane (bromine atoms and butyl chains) might
hinder the rotation of the carbazole rings in the macrocyclic
frame and thus result in the formation of atropisomers. With
three equivalent subunits in a cyclic arrangement, two
atropisomers are possible: aaa, with three N-butyl chains
on one side of the macrocyclic plane, and aab, with one of the
chains on the opposite side. X-ray crystallographic analysis of
4[11] revealed that aab is the preferred arrangement in the
solid state (Figure 2). In fact, the three carbazole units are
quite sharply tilted relative to the mean plane of the vinylene
bridges, with interplanar angles of 49, 91, and À818. The
carbazole moieties are themselves nearly planar, which
indicates that in spite of the internal crowding, the structure
is relatively unstrained. The 1H NMR spectrum of 4, recorded
in CDCl3 at 298 K, reveals threefold effective molecular
symmetry, which is consistent with either the prevalence of
the aaa structure in solution or rapid conformational
exchange involving the aab form. Preliminary DFT calcu-
lations on the two conformers indicated that the aab form is
more stable than aaa by approximately 11 kcalmolÀ1 and
thus lend support to the latter hypothesis. Variable-temper-
ature 1H NMR spectra of 4 (600 MHz, CD2Cl2) showed
gradual line broadening below 190 K; however, the coales-
cence point could not be reached above 160 K. Although
inconclusive, this observation is also consistent with the
presence of the aab form in solution and its rapid conforma-
tional exchange involving the aaa form.
In the final synthetic step, we subjected 4 to an Ullmann-
type coupling in the presence of zerovalent nickel[12] with the
À
intention of forming three new C C bonds and completing
the hub of the molecule. This fold-in transformation proved
successful, and 9,18,45-tributylchrysaorole (5) was isolated in
86% yield after chromatographic workup. Chrysaorole[13,14] 5,
a yellow compound with yellow-green fluorescence, has
limited stability in solution, but solid samples can be stored
for weeks in the refrigerator. Compound 5 shows a character-
1
istically simple H NMR spectrum (Figure 3), which reflects
its high molecular symmetry (C3v). The spectrum contains
a signal at d = 9.42 ppm assigned to the hub hydrogen atoms,
two signals at d = 7.59 and 7.33 ppm due to the hydrogen
atoms at the rim, and a set of resonances corresponding to the
three equivalent butyl chains. Although the downfield signals
appear as singlets, a weak remote coupling was identified in
the COSY spectrum between the lines at d = 9.42 and
7.33 ppm (5J < 1 Hz). To complete the assignment, we noted
Figure 2. Observed molecular structure of 4 in the solid state. Solvent
molecules, disordered butyl chains, and hydrogen atoms (in the stick
model) are omitted for clarity.
1714
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1713 –1717

Angewandte
Chemie
Figure 3. 1H NMR spectra of chrysaorole 5 (500 MHz, CDCl3, 300 K, bottom) and hexahydrochrysaorole 7 (500 MHz, CD2Cl2, 300 K, top). The
inset shows the simulated multiplet structure of endo and exo signals as derived from the DFT calculation of coupling constants. sol. =solvent.
in the ROESY map that the signal at d = 7.59 ppm yields
a dipolar coupling not only with the other rim signal but also
with the a-CH2 resonance. The 13C NMR spectrum of 5
contains seven signals in the region of sp2 carbon atoms, in
agreement with the proposed structure. HSQC and HMBC
spectra enabled the complete assignment of the 13C reso-
nances of 5 (see the Supporting Information).
Further insight into the three-dimensional structure of 5
was obtained from DFT calculations performed at the
B3LYP/6-31G** level of theory.[15] In the optimized model,
the fused chrysaorole skeleton shows the expected bowl-like
curvature distributed quite evenly over the constituent rings
(Figure 4). Because of the nonplanarity of the system, each of
the butyl chains can adopt two nonequivalent orientations
À
that differ by rotation by 1808 about the N C bond:
“pendant” (shown in Figure 4) and “outstretched”. These
two orientations are predicted to be effectively isoenergetic
(see the Supporting Information). Rotamer-averaged 1H and
13C nuclear-shielding values calculated by the GIAO method
are in excellent agreement with experimental data (Figure 5).
This good correlation indicates that the optimized DFT
Figure 4. DFT-optimized structures of 9,18,45-tributylchrysaorole (5;
C3v conformer with pendant chains) and 9,18,45-tributyl hexahydrochry-
saorole (7; C3 conformer with pendant chains). The labeling of ethylene
hydrogen atoms is shown for 7.
structure is in quantitative agreement with the actual
molecular geometry.[16]
Chrysaorole differs from typical buckybowls in that its
hub consists of a uniquely large, 18-membered ring.[17]
However, the presence of six hydrogen atoms in the hub
makes the remaining opening quite small. The depth of the
bowl, as measured from the level of hub carbon atoms to the
rim, is 1.96 ꢀ, and thus larger than the depth of corannulene
(0.87 ꢀ[18]). The pyramidalization of sp2 centers in chrysaor-
ole, as measured on the basis of p-orbital-axis-vector (POAV)
inclination angles (q),[19] is smaller than that observed for
most geodesic polyarenes, but the distribution of values over
specific positions in the structure follows a similar trend. In
particular, small q values (1–28) are observed for tertiary
centers, whereas the largest inclinations (q = 4–58) occur at
the quaternary carbon atoms that form the hub. Interestingly,
even though they are located on the rim, the nitrogen atoms
show a fair degree of pyramidalization, which is dependent on
the assumed orientation of the butyl chains (q = 3.7–5.88). A
strain energy of 53.4 kcalmolÀ1 was estimated for the
unsubstituted chrysaorole (5a, R = H, B3LYP/6-31G**) on
the basis of the homodesmotic reaction shown in Scheme 2.
This value is significantly higher than the estimate given for
corannulene (24.2 kcalmolÀ1), but the amount of strain per
Angew. Chem. Int. Ed. 2013, 52, 1713 –1717
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1715

.
Angewandte
Communications
chrysaorole nucleus is indeed nonplanar: a crucial piece of
information that was lacking in the spectrum of 5. The endo
and exo positions are differentiated by the convex shape of
the chrysaorole, and they would be exchanged by a bowl-
inversion process. However, no signs of chemical exchange,
either in the form of line broadening or EXSY peaks, were
observed in the 1H NMR spectra of 7, even in those recorded
at 410 K in [D10]p-xylene. Thus, unlike corannulene,[1b,20]
hexahydrochrysaorole (and most likely chrysaorole itself) is
a conformationally rigid molecule.
DFT calculations performed for 7a (R = H) predict
a strain energy very similar to that of the parent system 5a
(54 kcalmolÀ1; see the Supporting Information). DFT models
obtained for the substituted system 7 show that, apart from
the mobility of chains as described above for 5, the reduced
system possesses additional degrees of freedom associated
with the tetrahedral centers on the rim (Figure 4). The CH2–
CH2 bridges adopt a gauche conformation with a CCCC
torsion angle of approximately 458. As a result, the surface of
the bowl in 7 becomes slightly ruffled, and the molecular
symmetry is lowered to C3 or C1, depending on the relative
configurations of the ethylene bridges. The gauche arrange-
ment of the ethylene bridges causes further differentiation of
the endo and exo hydrogen atoms into pseudoaxial and
pseudoequatorial positions (Figure 4). However, the corre-
Figure 5. Correlation between the rotamer-averaged 1H and 13C chem-
ical shifts calculated for 5 and 7 (GIAO/B3LYP/6-31G(d,p), dcalc) and
the corresponding experimental values (dexptl). See the Supporting
Information for details.
1
sponding differentiation (or broadening) of H NMR signals
was not observed in CD2Cl2 even at 170 K; we can therefore
conclude that the process of bridge inversion must be very
rapid.[21] We were nevertheless able to find a link between the
DFT structure and the fast-exchange NMR spectroscopic
data by calculating not only chemical shifts but also the 1H–1H
coupling constants for hexahydrochrysaorole. The resulting
values were averaged to account for the assumed bridge-
inversion process and used to simulate the 1H NMR spectrum
of 7 (Figure 3). The characteristic AA’BB’ spin system of the
endo–exo pair was very accurately reproduced by the
calculation.
Scheme 2. Homodesmotic reaction used to estimate the strain energy
The fold-in synthesis of chrysaoroles described herein
offers a potentially general route to other bowl-shaped
molecules. Further investigation is needed to test its useful-
ness, which will depend on the identity of the coupled
subunits, the size of the macrocycle, and the substitution
pattern. We hope that, with further refinement, the present
methodology will provide access to nonplanar aromatic
compounds with unique physical characteristics.
in chrysaorole.
sp2 center in these two systems is nearly identical (1.2 kcal
molÀ1).
The formal reduction of double bonds at the rim was
envisaged as a spectroscopically relevant modification of the
chrysaorole ring system. Accordingly, hexahydrochrysaorole
7 was prepared by the catalytic hydrogenation of 4, followed
by the fold-in procedure (18% yield, Scheme 1). As in the
case of 5, the complete assignment of 1H and 13C NMR
spectroscopic signals was possible for 7 by means of two-
dimensional spectroscopy, and excellent correlation with the
GIAO shifts was observed (Figure 5). As expected, the low-
Received: October 23, 2012
Revised: November 20, 2012
Published online: December 18, 2012
Keywords: density functional calculations · fused-ring systems ·
.
macrocycles · nitrogen heterocycles · NMR spectroscopy
1
field region of the H NMR spectrum of 7 features only two
signals, at d = 8.67 and 6.95 ppm, which correspond to the hub
and rim carbazole hydrogen atoms, respectively (Figure 3).
The alkyl region contains the expected four signals of the
butyl substituent and two higher-order multiplets at d = 4.00
and 3.38 ppm, which correspond to the saturated peripheral
bridge. The presence of these two resonances, marked exo and
endo in Figure 3, is direct spectroscopic proof that the
[1] a) G. Mehta, H. S. P. Rao, Tetrahedron 1998, 54, 13325 – 13370;
b) Y.-T. Wu, J. S. Siegel, Chem. Rev. 2006, 106, 4843 – 4867;
c) V. M. Tsefrikas, L. T. Scott, Chem. Rev. 2006, 106, 4868 – 4884;
d) T. Amaya, T. Hirao, Chem. Commun. 2011, 47, 10524 – 10535;
e) A. Sygula, Eur. J. Org. Chem. 2011, 1611 – 1625; f) C. Thilgen,
1716
www.angewandte.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2013, 52, 1713 –1717

Angewandte
Chemie
Angew. Chem. 2012, 124, 7190 – 7192; Angew. Chem. Int. Ed.
2012, 51, 7082 – 7084.
[2] a) W. E. Barth, R. G. Lawton, J. Am. Chem. Soc. 1966, 88, 380 –
381; b) R. G. Lawton, W. E. Barth, J. Am. Chem. Soc. 1971, 93,
1730 – 1745.
[3] H. Sakurai, T. Daiko, T. Hirao, Science 2003, 301, 1878.
[4] a) L. T. Scott, E. A. Jackson, Q. Zhang, B. D. Steinberg, M.
Bancu, B. Li, J. Am. Chem. Soc. 2012, 134, 107 – 110; b) T.
Amaya, T. Nakata, T. Hirao, J. Am. Chem. Soc. 2009, 131,
10810 – 10811.
[12] T. Yamamoto, A. Morita, Y. Miyazaki, T. Maruyama, H.
Wakayama, Z. H. Zhou, Y. Nakamura, T. Kanbara, S. Sasaki,
K. Kubota, Macromolecules 1992, 25, 1214 – 1223.
[13] The name proposed for 5 reflects its resemblance to a jellyfish
and the yellow color of the compound. Chrysaora is a genus of
jellyfish from the Pelagiidae family. The genus was named after
´
Chrysaor, Medusaꢃs son (C1usaw1, “golden armament”): F.
Pꢄron, C. A. Lesueur, Ann. Mus. Hist. Nat. 1809, 14, 325 – 366.
[14] The von Baeyer name of 5 is 9,18,45-tributyl-9,18,45-triazatri-
decacyclo[24.18.1.03,42.06,41.08,39.010,38.012,36.015,35.017,33.019,32.021,30
024,29.027,44]pentatetraconta-1(44),2,4,6,8(39),10(38),11,13,15,
17(33),19(32),20,22,24,26,28,30,34,36,40,42-heneicosaene.
.
[5] Q. Zhang, K. Kawasumi, Y. Segawa, K. Itami, L. T. Scott, J. Am.
Chem. Soc. 2012, 134, 15664 – 15667.
[6] a) M. Bꢁhl, Chem. Eur. J. 1998, 4, 734 – 739; b) E. Steiner, P. W.
Fowler, L. W. Jenneskens, Angew. Chem. 2001, 113, 375 – 379;
Angew. Chem. Int. Ed. 2001, 40, 362 – 366; c) M. A. Dobrowol-
[15] For a performance analysis of computational methods with
respect to geodesic polyarenes, see: M. A. Petrukhina, K. W.
Andreini, J. Mack, L. T. Scott, J. Org. Chem. 2005, 70, 5713 –
5716.
´
ski, A. Ciesielski, M. K. Cyranski, Phys. Chem. Chem. Phys.
2011, 13, 20557 – 20563.
[7] A. S. Filatov, A. Y. Rogachev, E. A. Jackson, L. T. Scott, M. A.
Petrukhina, Organometallics 2010, 29, 1231 – 1237.
[16] Accurate geometries are considered a prerequisite for the
determination of reliable GIAO shieldings for protons in
aromatic systems: C. S. Wannere, K. W. Sattelmeyer, H. F.
Schaefer III, P. v. R. Schleyer, Angew. Chem. 2004, 116, 4296 –
4302; Angew. Chem. Int. Ed. 2004, 43, 4200 – 4206. For earlier
GIAO calculations on bowl-shaped systems, see Ref. [8a].
[17] This structural feature creates an analogy between chrysaoroles
and kekulenes: B. Kumar, R. L. Viboh, M. C. Bonifacio, W. B.
Thompson, J. C. Buttrick, B. C. Westlake, M.-S. Kim, R. W.
Zoellner, S. A. Varganov, P. Mçrschel, J. Teteruk, M. U. Schmidt,
B. T. King, Angew. Chem. 2012, DOI: 10.1002/ange.201203266;
Angew. Chem. Int. Ed. 2012, DOI: 10.1002/anie.201203266.
[18] J. C. Hanson, C. E. Nordman, Acta Crystallogr. Sect. B 1976, 32,
1147 – 1153.
[8] See, for example: a) B. D. Steinberg, E. A. Jackson, A. S. Filatov,
A. Wakamiya, M. A. Petrukhina, L. T. Scott, J. Am. Chem. Soc.
2009, 131, 10537 – 10545; b) K. T. Rim, M. Siaj, S. Xiao, M.
Myers, V. D. Carpentier, L. Liu, C. Su, M. L. Steigerwald, M. S.
Hybertsen, P. H. McBreen, G. W. Flynn, C. Nuckolls, Angew.
Chem. 2007, 119, 8037 – 8041; Angew. Chem. Int. Ed. 2007, 46,
7891 – 7895; c) K. Y. Amsharov, M. A. Kabdulov, M. Jansen,
Angew. Chem. 2012, 124, 4672 – 4675; Angew. Chem. Int. Ed.
2012, 51, 4594 – 4597.
[9] a) K. Imamura, K. Takimiya, T. Otsubo, Y. Aso, Chem.
Commun. 1999, 1859 – 1860; b) Q. Tan, S. Higashibayashi, S.
Karanjit, H. Sakurai, Nat. Commun. 2012, 3, 891.
[10] E. C. Dunne, ꢂ. J. Coyne, P. B. Crowley, D. G. Gilheany,
Tetrahedron Lett. 2002, 43, 2449 – 2453.
[19] R. C. Haddon, L. T. Scott, Pure Appl. Chem. 1986, 58, 137 – 142.
[20] L. T. Scott, M. M. Hashemi, M. S. Brachter, J. Am. Chem. Soc.
1992, 114, 1920 – 1921.
[11] CCDC 906691 contains the supplementary crystallographic data
for this paper. These data can be obtained free of charge from
The Cambridge Crystallographic Data Centre via www.ccdc.
cam.ac.uk/data_request/cif.
[21] For an analogous conformational-exchange process in 9,10-
dihydrophenanthrene, the inversion barrier was estimated by
NMR spectroscopy to be lower than 25 kJmolÀ1: R. Cosmo, S.
Sternhell, Aust. J. Chem. 1987, 40, 35 – 47.
Angew. Chem. Int. Ed. 2013, 52, 1713 –1717
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1717