Nanosized Molecular Propellers by Cyclodehydrogenation of Polyphenylene Dendrimers

Article abstract of DOI:10.1021/ja036732j

In a polymer analogous approach, large dendritic oligophenylenes containing benzene and tetraphenylmethane cores are transformed via oxidative cyclodehydrogenation to novel propeller-shaped molecules with large polycyclic aromatic hydrocarbon units as "bl

Full text of DOI:10.1021/ja036732j

Published on Web 02/21/2004
Nanosized Molecular Propellers by Cyclodehydrogenation of
Polyphenylene Dendrimers
Christopher D. Simpson, Gunter Mattersteig, Kai Martin, Lileta Gherghel,
Roland E. Bauer, Hans Joachim Ra¨der, and Klaus Mu¨llen*
Contribution from the Max-Planck-Institute for Polymer Research,
Ackermannweg 10, 55128 Mainz, Germany
Received June 17, 2003; E-mail: muellen@mpip-mainz.mpg.de
Abstract: In a polymer analogous approach, large dendritic oligophenylenes containing benzene and
tetraphenylmethane cores are transformed via oxidative cyclodehydrogenation to novel propeller-shaped
molecules with large polycyclic aromatic hydrocarbon units as “blades”. Structure analysis is performed by
a combination of MALDI-TOF mass spectrometry, UV/vis, fluorescence, and Raman spectroscopy using
solid-state sample preparation methods. These methods are also utilized to determine the degree of the
cyclodehydrogenation reaction.
Scheme 1. Chemical Structures of the PAHs C42 (HBC) (4),
C132 (5), and C222 (6) and Their Respective Precursor Molecules
Introduction
1, 2, and 3
One of the major goals of polycyclic aromatic hydrocarbon
(PAH) chemistry has been the synthesis of constantly improved,
molecularly defined graphite model compounds.1-3 Significant
advances toward this aim could be achieved by applying a
method recently developed by us.4,5 This approach is based on
the planarization of three-dimensional, dendritic oligophenyl-
enes, which can be drawn in their 2D-projection without overlap
of any phenyl rings, by oxidative cyclodehydrogenation. It
enables not only the synthesis of hexa-peri-hexabenzocoronene
(HBC) (4) and its derivatives but also much larger all-benzoid
PAHs with up to 222 carbons (6) (Scheme 1).6 For an even
larger PAH and thus improved graphite model, the 474 carbon-
containing polyphenylene dendrimer 7 was a suitable precursor
candidate (Scheme 2).
As it turned out in the course of this study, the geometries of
the precursor molecule, and especially of its core, are of
paramount importance. Not only do they influence the efficiency
of the cyclodehydrogenation, but they also play an important
role in the geometrical result of the planarization reaction. In
fact, as described in the following, the outcome of the cyclo-
dehydrogenation can lead to a completely different domain of
chemistry than was originally intended; instead of to planar 2D-
graphite model compounds, it leads to the field of three-
dimensional nanostructures, such as propeller-shaped mol-
ecules,7,8 highly nonplanar PAHs,9 and 3D-graphitic materials.10
This finding prompted us to apply our cyclodehydrogenation
conditions also to a dendrimer with a geometry different from
that of 7, the polyphenylene 11. It is built around a tetrahedral
core, thus separating the molecule into four dendritic arms,
intentionally prohibiting total planarization (Scheme 3).
To transform the dendrimers to the corresponding target PAH
structures, already 136 hydrogens for molecule 11 and 252
hydrogens for molecule 7 have to be removed in one single
reaction. This sheer number sets the demands for the cyclo-
dehydrogenation and takes it from the field of organic synthesis
more into the domain of polymer analogous reactions. The
increasing number of bonds to be closed results in an exponential
(1) Clar, E. The Aromatic Sextet; Wiley: London, 1972.
(2) Zander, M. Polycyclische Aromaten; Teubner: Stuttgart, 1995.
(3) Goddard, R.; Haenel, M. W.; Herndon, W. C.; Kru¨ger, C.; Zander, M. J.
Am. Chem. Soc. 1995, 117, 30-41.
(4) Watson, M. D.; Fechtenko¨tter, A.; Mu¨llen, K. Chem. ReV. 2001, 101, 1267-
1300.
(5) Iyer, V. S.; Wehmeier, M.; Brand, J. D.; Keegstra, M. A.; Mu¨llen, K. Angew.
Chem., Int. Ed. Engl. 1997, 36, 1604-1607.
(6) Simpson, C. D.; Brand, J. D.; Berresheim, A. J.; Przybilla, L.; Ra¨der, H.
J.; Mu¨llen, K. Chem.-Eur. J. 2002, 8, 1424-1429.
(7) Barnett, L.; Ho, D. M.; Baldridge, K. K.; Pascal, R. A. J. Am. Chem. Soc.
1999, 121, 727-733.
(9) Scott, L. T.; Boorum, M. M.; McMahon, B. J.; Hagen, S.; Mack, J.; Blank,
J.; Wegner, H.; de Meijere, A. Science 2002, 295, 1500-1503.
(10) Shen, X. F.; Ho, D. M.; Pascal, R. A. Org. Lett. 2003, 5, 369-371.
(8) Vacek, J.; Michl, J. New J. Chem. 1997, 21, 1259-1268.
9
10.1021/ja036732j CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 3139-3147
3139

A R T I C L E S
Simpson et al.
Scheme 2. Chemical Structures of the Polyphenylene Dendrimer 7 (C474H318, M ) 6013 Da), the Cyclodehydrogenation Products 8
(C474H66, M ) 5760 Da), 9a (C474H132, M ) 5826 Da), and 9b (C474H120, M ) 5814 Da), and the Respective “Propeller Arms” 10a (C150H40
M ) 1842 Da) and 10b (C156H40, M ) 1914 Da)
,
Scheme 3. Chemical Structures of the Polyphenylene Dendrimer 11 (C385H260, M ) 4886 Da) and the Cyclodehydrogenation Products 12a
(C385H132, M ) 4757 Da), 12b (C385H128, M ) 4753 Da), and 12c (C385H124, M ) 4749 Da)
9
3140 J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004

Nanosized Molecular Propellers
A R T I C L E S
Scheme 4. Synthesis of Dendrimer 7a
Figure 1. MALDI-TOF mass spectrum of C474-dendrimer 7.
methanol/THF affording 15. Diels-Alder cycloaddition of the
core 15 with the cyclopentadienone 16, a protected dendrimer
branching unit, is employed under extrusion of carbon monoxide
in boiling o-xylene to yield the protected first-generation
dendrimer.13,14 The cyclopentadienone branching units are
designated by their number of reactive sites; in the case of 16,
the molecule can react with one alkyne and thereby introduces
four new protected alkynes. It is therefore called an A4B unit.
The triisopropylsilyl protective groups of the first generation
are removed with fluoride salts under more drastic conditions
than for trimethylsilyl groups to obtain the deprotected first-
generation dendrimer 17 in 87% yield. Finally, in a further
Diels-Alder cycloaddition, 17 is reacted with tetracyclone (18)
to yield the three-dimensional, shape-persistent15,16 dendrimer
7 as a readily soluble and colorless compound with a molecular
mass of 6013 Da, which can be completely characterized by
solution NMR spectroscopy and MALDI-TOF spectrometry
techniques (Figure 1).
a Reagents, conditions, and yields: (a) SiCl4, dry ethanol, 60%; (b)
TMSA, CuI, Pd[PPh3]2Cl2, PPh3, NEt3/THF (1:1), 97%; (c) K2CO3, MeOH/
THF, 85%; (d) o-xylene, 175 °C, 14 h, 95%; (e) NH4F/Bu4NF, THF, 87%;
(f) o-xylene, 175 °C, 14 h, 76%.
The synthesis of the tetrahedral dendrimer 11 has been
reported previously.14 It is prepared in a way similar to 7 by
reaction of A2B branching units, cyclopentadienones with two
protected alkynes, with a tetraphenylmethane core. After depro-
tection, a further cycloaddition of tetracyclone (18) yields the
second-generation dendrimer 11.
Oxidative Cyclodehydrogenation and Characterization.
The C474 dendrimer 7 was subjected to the typical oxidative
cyclodehydrogenation conditions developed for the synthesis
of large, unsubstituted PAHs.6 After reaction with different
quantities of aluminum(III) chloride and copper triflate in carbon
disulfide, mass spectrometry suggested chlorination and incom-
plete dehydrogenation, the extent of which fell far short of that
needed to give the mass of the desired PAH 8 at 5760 Da.17
For the cyclodehydrogenation of alkyl substituted HBCs,4 the
method of choice is to use iron(III) chloride, both Lewis acid
and oxidizing agent in one. By predissolving the iron(III)
decrease of the overall yield. It should be considered that, for
example, with the formation of 126 new bonds as for 8, a
conversion of 99% in each carbon-carbon bond formation
would lead only to a total yield of 28% of desired product. To
achieve at least 90% overall yield, a bond forming efficiency
of 99.9% is required. This is a challenging value for any given
single organic reaction, let alone for a synthesis with a multitude
of individual steps which is encountered by a severe solubility
problem after a few bond formations.
Results and Discussion
Dendrimer Synthesis. Dendrimer 7 is accessible in a six-
step synthesis, as outlined in Scheme 4. First, 1,3,5-tris-(4-
iodophenyl)-benzene (14) is obtained by regioselective cyclo-
trimerization of 1-(4-iodophenyl)-ethanone (13) with silicon
tetrachloride in 60% yield.11 Through the use of iodide as a
functional group, the Hagihara-Sonogashira coupling of tri-
methylsilylacetylene to the core proceeds smoothly at room
temperature in 97% yield.12 The silyl protecting groups are
removed under mild conditions with potassium carbonate in
(13) Dilthey, W.; Hurtig, G. Ber. Dtsch. Chem. Ges. 1934, 67, 2004-2007.
(14) Wiesler, U. M.; Berresheim, A. J.; Morgenroth, F.; Lieser, G.; Mu¨llen, K.
Macromolecules 2001, 34, 187-199.
(15) Liu, D. J.; Zhang, H.; Grim, P. C. M.; De Feyter, S.; Wiesler, U. M.;
Berresheim, A. J.; Mu¨llen, K.; De Schryver, F. C. Langmuir 2002, 18,
2385-2391.
(11) Plater, M. J.; McKay, M.; Jackson, T. J. Chem. Soc., Perkin Trans. 1 2000,
(16) Wind, M.; Wiesler, U. M.; Saalwa¨chter, K.; Mu¨llen, K.; Spiess, H. W.
AdV. Mater. 2001, 13, 752-756.
2695-2701.
(12) Takahashi, S.; Kuroyama, Y.; Sonogashira, K.; Hagihara, N. Synthesis 1980,
627-630.
(17) Przybilla, L.; Brand, J. D.; Yoshimura, K.; Ra¨der, H. J.; Mu¨llen, K. Anal.
Chem. 2000, 72, 4591-4597.
9
J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004 3141

A R T I C L E S
Simpson et al.
Figure 2. (I) MALDI-TOF mass spectra of cyclodehydrogenation products of dendrimer 7 (expanded region): (a) Experiment 1, 0.75 equiv of FeCl3/H, 0.5
h; (b) Experiment 2, 1.5 equiv of FeCl3/H, 20 h; (c) Experiment 3, 3.0 equiv of FeCl3/H, 20 h. Yield (typically): ∼70-80%. The vertical lines represent
the calculated masses of molecules 7, 8, and 9a. (II) UV/vis spectra of thin films on quartz substrates corresponding to Experiments 1-3. In (e), the
spectrum of model compound 10b is also shown.
chloride in nitromethane, it can be added homogeneously to
the solution of the dendrimer in methylene chloride.
which 200 hydrogens were randomly removed, where a broader
and more bathochromically shifted band would be expected.
This follows from a comparison with UV/vis spectra of other
large PAHs with 42-222 carbons.4,6
One of the most crucial factors in the planarization reaction
is the stoichiometry of cyclodehydrogenation reagent employed.
Introduction of 0.75 equiv of iron(III) chloride per hydrogen to
be removed results in a partially cyclized product with a peak
maximum at 5890 Da (Figure 2a). The optimum amount for
cyclodehydrogenation in this case turned out to be 1.5 equiva-
lents, especially when employed in a 1:1 solvent mixture of
methylene chloride and carbon disulfide as in Experiment 2
(Figure 2b). These conditions reproducibly result in a compara-
tively sharp main peak at 5830 Da, corresponding to the removal
of about 200 hydrogens. A small peak at 5900 Da can be
assigned to a 2-fold chlorination of this product, and a further
small peak at 5760 Da indicated the formation of the fully
planarized C474 disk 8. Due to the different desorption
behaviors of fully planarized disk and partially cyclized
products, an exact quantification of the product distribution by
mass spectrometry is not possible. Attempts to push the reaction
further toward the completely planarized disk by employing up
to 3-6 equiv of iron(III) chloride result only in higher molecular
weight products with broader distribution, due to chlorination
(Figure 2c). We see only minor differences between reactions
quenched after 1 and 20 h.
We therefore propose that cyclodehydrogenation takes place
independently in the three different arms of the dendrimer as
suggested by the conformation of the dendritic precursor 7
(Figure 3a), leading to a partitioning and thus creating three
large PAH subunits connected in the center with a 1,3,5-
triphenylbenzene (9a) or benzene (9b) core. Such a molecule
is propeller-shaped due to the steric hindrance of the “propeller
blades” and conveniently explains why the cyclodehydrogena-
tion is not proceeding further to the fully planarized disk 8:
the remaining hydrogens are simply too far apart to participate
in a new carbon-carbon bond formation (Figure 3b,c). The close
distance and favorable geometric arrangement of two carbon
atoms are decisive for the bond closure, which also explains
why there are no intermolecular reactions taking place. Thus,
the cyclodehydrogenation is probably mainly progressing from
the “outside” periphery toward the “inside” of the dendrimer.
The aromatic “blades” of these propeller structures contain 150
and 156 carbons, respectively. The existence of these moieties
satisfactorily explains the observed UV/vis and also the mass
spectrum.
While the insolubility of large PAHs severely limits classical
solution characterization, solid-state methods are revealing. UV/
vis spectra recorded after smearing a thin film of a sample from
Experiment 2 on a quartz substrate show a surprisingly well-
defined band at 565 nm (Figure 2e), sharper than that for a
C222-PAH, which exhibits a very broad band at around 770
nm.6 The obtained spectrum corresponded to a relatively defined
PAH containing about 150 carbons, not to a C474 molecule in
Part of the problem of exactly determining the composition
and structure of the extremely large and insoluble cyclodehy-
drogenation product is the lack of isotopic resolution of the mass
spectra. In principle, it is possible to obtain isotopic resolution
in the investigated mass range with the used MALDI-TOF mass
spectrometer (e.g., PMMA 6.000 Da standard or the dendrimer
7). Due to the laser excitation wavelength of 337 nm and the
UV/vis absorption of the investigated product, ionization does
9
3142 J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004

Nanosized Molecular Propellers
A R T I C L E S
Figure 3. Computer-generated three-dimensional structures of (a) dendrimer 7, (b) propeller with triphenylbenzene core 9a, and (c) propeller with benzene
core 9b.20
Scheme 5. Synthesis of Model Compound 10ba
a Reagents, conditions, and yields: (a) o-xylene, 175 °C, 2 h, 98%; (b) NH4F/Bu4NF, THF, 10 min, 98%; (c) o-xylene, 175 °C, 14 h, 81%; (d) 3 equiv
of AlCl3/Cu(OTf)2/H, CS2, 24 h, 95%.
not take place by a conventional MALDI mechanism but by
photoionization, leading to radical cations. We assume that the
lack of high resolution is due to the higher internal energy of
such molecules because of partial laser excitation, which,
combined with the required high desorption energy, causes a
peak broadening.
dienone 16 was reacted with an excess of phenylacetylene in a
Diels-Alder cycloaddition until the purple color of 16 had
disappeared. After subsequent cleavage of the silyl protective
groups, the tetraacetylene compound 20 was obtained in 98%
yield. A further [2+4] cycloaddition with 18 afforded the
dendrimer 21, which was cyclodehydrogenated to the 156-
carbon PAH 10b.
A possible solution to this problem was to chemically
fragment the propeller molecule and to analyze the lower
molecular weight fragments, which should be in the range of
one-third of the original mass if the assumed structure above
was true. Therefore, the cyclodehydrogenated product from
Experiment 2 was heated for 30 min at 450 °C in an
aluminum(III) chloride/sodium chloride melt, harsh conditions
under which single aryl-aryl bonds are expected to break, while
fused PAH segments should remain stable. Indeed, the resulting
mass spectrum showed peaks at 1842 and 1914 Da, correspond-
ing to the molecules 10a and 10b. At six mass units higher
from each peak, there were additional peaks, obviously corre-
sponding to defect structures with roughly three carbon-carbon
bonds unclosed. This result indicates strongly that the 150 and
156 carbon-containing PAHs are present as subunits in the
original molecule before fragmentation, therefore clearly proving
the proposed partitioning of the cyclodehydrogenated molecule.
For a better comparison of the electronic properties of the
propeller molecule, the C156 model compound 10b was
prepared in a separate approach (Scheme 5). The cyclopenta-
The solid-state UV/vis spectra of 9 and the model compound
10b show a high degree of agreement, with 10b also having a
peak maximum at about 560 nm (Figure 2e). This indicates that
both compounds contain the same or a very similar number of
fused benzene rings within their conjugated system. The
cyclodehydrogenation product 9 possesses an altogether broader
spectrum and an additional shoulder, the cause of which is not
quite clear. As expected, the partially closed product resulting
from 0.75 equiv FeCl3 has an absorption maximum at shorter
wavelength (525 nm, Figure 2d), and the chlorinated product
from Experiment 3 (3 equiv of FeCl3, Figure 2f) has a
bathochromic shifted maximum at 590 nm. It does seem possible
to close even more bonds with higher equivalents, but the
chlorinating side-reactions prevail.
Complementary to UV/vis spectroscopy, solid-state Raman
spectroscopy is a further tool for the characterization of insoluble
PAHs.18 The profile of the Raman spectrum of 9, characterized
by two strong bands at 1603 and 1320 cm-1, is very similar to
that of 10b (Figure 4). These two bands, usually referred to as
9
J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004 3143

A R T I C L E S
Simpson et al.
chlorination or partial cyclization defects could be avoided
(Figure 5). The yellow compound was soluble under gentle
warming in organic solvents such as chloroform, THF, and
toluene. Solution NMR studies, however, resulted, even under
heating, only in broad, unresolved spectra.
Due to the steric repulsion of the alkyl chain substituents,
the trimer molecule is very unlikely to assume a planar
conformation. The twisting of the three HBC-moieties out of
plane leads to the formation of a propeller-shaped object, as
illustrated by the calculated 3D-visualization in Figure 7. The
three blades can be expected to possess a certain degree of
rotational freedom around the bond connecting the core, thus
explaining the good observed solubility. Excluding the flexible
alkyl chains, such a rotor possesses a diameter of about 3 nm.
The out-of-plane twist of the HBC-sections is also confirmed
by the absorption spectrum of the propeller molecule, which is
depicted in Figure 6. The positions of the main bands are almost
unchanged as compared to the hexaalkyl HBC-“monomer” 4a.
This proves that the aromatic chromophores in 26 are HBC-
units, which are not in conjugation with each other. This is a
result of the torsion angle with respect to the benzene linker
and prevents an efficient electronic communication by overlap
of the pz orbitals. The peak width of the bands of 26 is, however,
increased. The comparison of the absorption spectra of the HBC-
“monomer” and its propeller-shaped trimer bears a striking
resemblance to the situation of the model compound C156 and
the C474 propeller as seen in Figure 2e. This serves as a
confirmation that an aromatic chromophore that is the same as
or very similar to that of C156 is present in the C474 propeller.
Figure 4. Raman spectra of propeller 9 (a) and the model compound 10b
(b).
the G- and D-bands, respectively, are typical features of PAHs,
carbonaceous materials, and disordered graphite.19 Slight dif-
ferences between the two spectra (i.e., relative intensities and
positions of the bands) appear in the spectral region at around
1300 cm-1 because of the higher sensitivity of the D-peak
frequency to molecular size and geometry than the G-peak.
Usually D-bands of PAHs show structured features composed
of a very strong band and a few low intensity lines. Minor
differences are observed in the positions of these low intensity
peaks of the two compounds. The direct comparison of the two
spectra also shows an increased intensity and broadening of the
D-peak of 9 in relation to the D-peak of the model compound
10b. For PAHs generally, the intensity of the D-peak with
respect to the G-band is dependent on the degree of π-electron
delocalization. The broadening of this band corresponds to the
convolution of Raman transitions from a distribution of aromatic
“domains” with different structures and dimensions. Therefore,
two conclusions can be drawn from the Raman measurements:
the subunits of the C474-propeller 9 and the C156 model
compound 10b are generally very similar; there are, however,
some defect structures present in the “propeller blades”.
A propeller-shaped model compound in which the partitioned
aromatic domains were unambiguous should further assist in
understanding the analytical data obtained for the propeller 9.
A suitable molecule can be prepared starting also from
1,3,5-tris-(4-iodophenyl)-benzene (14). By coupling 14 under
Hagihara-Sonogashira conditions with a phenylacetylene and
subsequent Diels-Alder reaction, an oligophenylene (25) is
accessible, bearing three separate hexaphenylbenzene moieties,
which cannot further interconnect. These can in turn be
planarized individually by cyclodehydrogenation, so that a trimer
of HBC-units (26) is formed, which is tethered in the center
via a trigonal benzene connector (Scheme 6). To enable
characterization by solution-based methods, alkyl substitution
in the form of 3,7-dimethyloctyl chains was possible by
employing suitably derivatized building blocks. Using 3 equiv
of iron(III) chloride as reagent, 26 was successfully formed as
a monodisperse compound by removing 36 hydrogens. Accord-
ing to the isotopically resolved MALDI-TOF mass spectrum,
Having clearly reached the limits of synthesis and spectro-
scopical characterization, it was desirable to further characterize
the C474 propeller objects also by visualization, for example,
by TEM. This attempt failed essentially because it was not
possible to prepare separated single molecules from the insoluble
compound on TEM grids. Only large particles with a very low
degree of order down to about 100 nm were observed,
corresponding to large clusters of the molecule.
Nevertheless, the available spectroscopical structure analysis
data were sufficient to be confident of the proposed three-
dimensional structure of 9 and to move on to apply the
cyclodehydrogenation conditions to a different geometric den-
drimer structure.
The dendrimer 11 possesses a tetraphenylmethane core
forcing the oligophenylene branches to adopt a tetrahedral
geometry (Scheme 3 and Figure 8a) which thus prevents a
complete planarization. Ideally, the removal of 136 hydrogens
should afford four planarized wings consisting of rigid PAHs,
which originate from the central core and point to the four
corners of a tetrahedron. The resulting molecule resembles a
tetra-bladed propeller (Figure 8c). The conditions for the
oxidative cyclodehydrogenation of dendrimer 11 with iron(III)
chloride were similar to those applied to the dendrimer 7. A
detailed analysis of the mass spectra became possible because
isotopic resolution of the individual peaks could be achieved
in this mass region. Using 0.75 equiv of iron(III) chloride per
intendendly removed hydrogen led to a partially cyclodehydro-
genated product with a peak maximum at 4800 Da. Employing
1.5 equiv afforded a product with the main peak at 4760 Da,
much closer to the desired mass of 4749 Da for the target
molecule 12c (Figure 9). Even better results were obtained with
(18) Shifrina, Z. B.; Averina, M. S.; Rusanov, A. L.; Wagner, M.; Mu¨llen, K.
Macromolecules 2000, 33, 3525-3529.
(19) Negri, F.; Castiglioni, C.; Tommasini, M.; Zerbi, G. J. Phys. Chem. A 2002,
106, 3306-3317.
(20) Using the molecular mechanics force field of Wavefunction Spartan Pro
1.0.5.
9
3144 J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004

Nanosized Molecular Propellers
A R T I C L E S
Scheme 6. Synthesis of HBC-Trimer 26a
a Reagents, conditions, and yields: (a) CuI, Pd[PPh3]2Cl2, PPh3, NEt3/THF (1:1), 83%; (b) Ph2O, 260 °C, 3 d, 76%; (c) 3 equiv of FeCl3/H, CH3NO2,
CH2Cl2, 71%.
Figure 7. Computer-generated three-dimensional structure of the HBC-
trimer 26 (alkyl chains omitted for clarity).20
Figure 5. MALDI-TOF mass spectrum of the HBC-trimer 26.
With the assumption of a similar behavior in desorption and
ionization for all products, the molecular weight distribution
ascertains an average degree of cyclodehydrogenation of 96%.
The total peak distribution covers the range from about 90% to
100% cyclodehydrogenation. The signal distribution within the
MALDI-TOF mass spectrum can therefore be mainly attributed
to a mixture of cyclodehydrogenation products covering the
molecular formulas C385H124 to C385H134. The highest signal
intensity can be found for compound C385H128 (12b), which is
the most dominant species in the mixture. More drastic
conditions to force the reaction to complete the removal of all
hydrogens, however, favored the formation of higher molecular
weight products attributable to chlorination. The spectrum
resulting from the use of 6 equiv of iron(III) chloride contains
further peaks assignable to products ranging from mono-
chlorinated up to 6-fold chlorinated compounds in addition to
the main signal at 4753 Da.
Figure 6. UV/vis spectra of chloroform solutions of (a) the HBC-trimer
26 and (b) the “monomer”: HBC-(n-dodecyl)6 (4a).
To further characterize the insoluble cyclodehydrogenated
products 12, they were also studied by solid-state spectroscopy.
The UV/vis spectra of the “smeared” thin films of the samples
showed, dependent on the quantities of iron(III) chloride, an
absorption maximum successively red-shifted from 466 to 482
nm, correlated to the increasing size of the aromatic systems
(Figure 10). The optical absorptions of the cyclodehydrogenated
products using 2-3 equiv are very similar, and the well-defined
band at 482 nm is comparable to that observed for a PAH with
96 carbon atoms (C96H30).4 This resemblance indicates that the
investigated molecules are composed of four separate wings
2-3 equiv of iron(III) chloride. In the case of 2.5 equiv, there
is a weak chlorination peak at 4787 Da; however, the main
product distribution is narrower than before, but still covers
different degrees of cyclodehydrogenation. The optimum condi-
tions were reached when 3 equiv was used, resulting in an even
sharper main peak at 4753 Da. This peak corresponds to the
removal of about 132 hydrogens from the precursor and supports
the desired structure, however, with two carbon-carbon bonds
missing to complete cyclodehydrogenation. A small peak at
4787 Da indicates monochlorination of this product.
9
J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004 3145

A R T I C L E S
Simpson et al.
Figure 8. Computer-generated three-dimensional structures of (a) dendrimer 11 (b) propeller with tetraphenylmethane core 12a, and (c) fully cyclized
propeller 12c.20
Figure 10. UV/vis spectra of the cyclodehydrogenation products of
dendrimer 11 using (a) 0.75 equiv of FeCl3/H; (b) 1.5 equiv of FeCl3/H;
(c) 2.5 equiv of FeCl3/H.
Figure 9. MALDI-TOF mass spectra of cyclodehydrogenation products
of dendrimer 11 (expanded region): (a) 1.5 equiv of FeCl3/H, 20 h; (b) 2.5
equiv of FeCl3/H, 20 h; (c) 3.0 equiv of FeCl3/H, 20 h. Yield (typically):
∼80-90%. The vertical lines represent the calculated masses of molecules
12a and 12c. The colored graphs (red, blue, and green) represent the
simulated mass spectra for 12a, 12b, and 12c, respectively. Inset: Full range
of spectrum c.
consisting of conjugated systems with approximately 96 carbons
Figure 11. Raman spectra of the cyclodehydrogenation products of
dendrimer 11 using (a) 0.75 equiv of FeCl3/H; (b) 1.5 equiv of FeCl3/H;
(c) 2.0 equiv of FeCl3/H; (d) 2.5 equiv of FeCl3/H; (e) 3.0 equiv of FeCl3/H
(λex ) 514.5 nm).
each.
By applying Raman spectroscopy, relevant information about
the degree of cyclodehydrogenation in relation to the number
of employed iron(III) chloride equivalents could be obtained.
All of the solid-state Raman spectra showed, with minor
differences, as common features a first strong band located at
1324 cm-1, a weak band at 1234 cm-1, and a second strong
band located at 1602 cm-1 (Figure 11). A significant difference
of these spectra is their decreasing background intensity, with
increasing number of equivalents of iron(III) chloride used for
cyclodehydrogenation. This behavior can be attributed to a
fluorescence effect. The fluorescence appears as a broad spectral
9
3146 J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004

Nanosized Molecular Propellers
A R T I C L E S
indicated by various analysis methods, these imperfections
consist of a few unclosed bonds in the PAH “wings” and a small
content of chlorination. The defects, however, have no effect
on the basic propeller structure of the molecules and the
significant findings of this study. Even without an exact
quantification of the product distribution, the individual bond
formation of the oxidative cyclodehydrogenation reaction can
be considered to have an extremely high efficiency, which is
impressive taking into account the large size and unfavorable
solubility properties of the studied compounds. The optimal
stoichiometry, that is, the number of equivalents of cyclo-
dehydrogenation reagent, is, however, very dependent on the
individual molecule and can vary considerably.
We succeeded in the synthesis of new nanometer scale
structures consisting of three and four PAH wings tethered by
a central core, with diameters ranging from about 3 to 4.5 nm.
The combination of spectroscopic and spectrometric methods
used for their structure elucidation proved highly revealing in
cases where no single technique could provide clear answers
anymore.
While on the single molecule side the appealing propeller
shape is of great interest, the torsion of the PAH-“blades” with
respect to each other also must have implications on the bulk
state of the molecule. It can be expected that the graphene-like
subunits of different molecules are not arranged in large coplanar
domains as in graphite itself, but rather in small microdomains
tilted at different angles. The exact nature of this packing
behavior, however, remains to be studied more closely.
Further investigations toward the visualization, manipulation,
and application in hydrogen storage of these new 3D nano-
propellers are underway. Also, there is sufficient ground to
believe that introduction of alkyl chain substituents on the
periphery will make these large structures soluble and thus
processable.
Figure 12. Solid-state fluorescence spectra of dendrimer 11 (a) and of its
cyclodehydrogenation products using (b) 0.75 equiv of FeCl3/H; (c) 1.5
equiv of FeCl3/H; (d) 2.5 equiv of FeCl3/H (λex ) 514.5 nm).
emission which adds to the background and is an indication of
the amount of partially cyclized structures in the product.
This finding is supported by the solid-state emission spectrum
of the cyclodehydrogenated propeller structures. Using 0.75
equiv for cyclodehydrogenation, the fluorescence is very intense
(Figure 12). Upon the addition of higher iron(III) chloride
equivalents, there is a distinct decrease in fluorescence intensity,
as well as a shift of the emission spectrum to longer wave-
lengths. As in the Raman spectra, this decrease can also be
correlated to an increase of the degree of cyclization in the
propeller molecule.
These solid-state spectroscopical results serve as comple-
mentary confirmations of the results already obtained by mass
spectrometry. They are of particular interest as independent
sources of information concerning the degree of cyclodehydro-
genation for other insoluble, for example, polymeric, carbon-
rich compounds where mass spectrometry is no longer an option.
Conclusions
Acknowledgment. We thank H.-J. Menges for the Raman
and fluorescence measurements and Dr. N. Pschirer for helpful
discussions. This work was supported by the EU-TMR project
SISITOMAS, EU project MAC-MES (GRD2-2000-30242), and
the German “Bundesministerium fu¨r Forschung und Technolo-
gie” as part of the program “Zentrum fu¨r multifunktionelle
Werkstoffe und miniaturisierte Funktionseinheiten” (BMBF
03N6500).
It has been demonstrated that the concept of completely
planarizing large oligophenylenes reaches its limit when the
dendrimer becomes so big that the cyclodehydrogenation
reaction takes place preferentially within the dendritic arms and
not in between. This partitioning results, however, in a new
approach to propeller-like, three-dimensional molecular struc-
tures with large PAH subunits. As originally expected, the
cyclodehydrogenation reaction for such a giant molecule cannot
be 100% efficient. While the cyclodehydrogenation still yields
a monodisperse product for the HBC-trimer 26, with the
increased number of individual bond formation steps as in the
reaction of the larger, second-generation dendrimers 7 and 11,
some degree of defects have to be taken into account. As
Supporting Information Available: Full experimental and
spectroscopic information of the presented compounds (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA036732J
9
J. AM. CHEM. SOC. VOL. 126, NO. 10, 2004 3147