Tetrabenzo[ab,f,jk,o][18]annulene: Synthesis, crystal structure analysis, ab initio quantum-chemical studies of intermolecular interactions in the gas phase and in crystalline states, photodimerization and photopolymerization

Article abstract of DOI:10.1039/a903451j

The synthesized title compound 8 crystallizes in three different modifications, which were investigated by electron crystallography, high resolution electron microscopy and X-ray structural analysis. The formation of molecular pairs was studied by an ab initio quantum-chemical approach. The calculated "dimers" were compared to the mutual orientations of nearest neighboring molecules in the crystal structures and to covalently bonded dimers obtained by a topochemical photodimerization.

Full text of DOI:10.1039/a903451j

View Article Online / Journal Homepage / Table of Contents for this issue
Tetrabenzo[ab, f,jk,o][18]annulene: synthesis, crystal structure
analysis, ab initio quantum-chemical studies of intermolecular
interactions in the gas phase and in crystalline states,
photodimerization and photopolymerization
Richeng Yu,a Alexander Yakimansky,†a Ingrid G. Voigt-Martin,*a Michael Fetten,b
C. Schnorpfeil,b Dieter Schollmeyerb and Herbert Meierb
a Johannes-Gutenberg-Universität, Institut für Physikalische Chemie, Welder Weg 11,
D-55099 Mainz, Germany
b Johannes-Gutenberg-Universität, Institut für Organische Chemie, Duesbergweg 10-14,
D-55099 Mainz, Germany
Received (in Cambridge, UK) 2nd March 1999, Accepted 25th June 1999
The synthesized title compound 8 crystallizes in three different modifications, which were investigated by electron
crystallography, high resolution electron microscopy and X-ray structural analysis. The formation of molecular
pairs was studied by an ab initio quantum-chemical approach. The calculated “dimers” were compared to the
mutual orientations of nearest neighboring molecules in the crystal structures and to covalently bonded dimers
obtained by a topochemical photodimerization.
yield to the N-phenylimine 7 which showed in a strongly
1. Introduction
alkaline medium a cyclocondensation reaction to the title com-
pound 8 (Scheme 1). Linear condensation products, that still
contain polar imino groups, could be easily separated due to
their better solubility in polar solvents. The generation of
the two olefinic double bonds in the final step is highly stereo-
selective; no trace of a (Z)-configured isomer could be detected
in the 1H NMR spectrum.
Areno-condensed annulenes represent a class of compounds
with interesting photophysical and photochemical properties.1–5
Due to their high thermal and chemical stability these com-
pounds can be applied in materials science, particularly in the
field of liquid crystals and photoconductivity.
Because of the dependence of the physical and chemical
behavior on the molecular and supramolecular structure we
synthesized and investigated tetrabenzo[ab, f,jk,o]cycloocta-
decene (8) as a model compound. It is a formal [18]annulene,
which is condensed with four benzene rings—two are linked in
ortho and two in meta positions. Thus it is closely related to the
naphtho-, phenanthro- and chryseno-condensed [18]annulenes
described earlier.1–5
Compound 8, a formal derivative of [18]annulene, has to be
considered—like related areno-condensed annulenes1–5—as a
system of aromatic ring islands combined by olefinic bridges. A
macrocyclic aromaticity can be excluded.3 The NOE measure-
1
ments permit the exact correlation of the H NMR signals to
certain protons and thus reveal that the inner protons 5-H
(δ = 7.83) and 23-H (δ = 8.29) generate signals at somewhat
lower field than the outer protons 4-H (δ = 7.08); a macrocyclic
aromaticity can be excluded. The fixing of the olefinic
double bonds is also documented by the vicinal coupling con-
In order to understand the aggregation tendency2 and the
photochemistry, the crystal structures of 8 were investigated
using electron crystallography, high resolution electron micro-
scopy and X-ray analysis. The possibility of dimerization of
these molecules in the gas-phase or in solution was studied by
an ab initio quantum-chemical approach. The calculated geom-
etries of the most stable gas-phase dimers were compared with
the mutual orientations of the nearest neighboring molecules in
the crystal structures and related to the observed photodimers
which are covalently bonded.
3
stant J (4-H, 5-H) of 16.1 Hz which is characteristic of the
trans configuration.
2.2 Electron diffraction
For electron diffraction extremely small crystals are required.
These were grown by dissolving 8 in toluene at 80 ЊC, crystal-
lizing for 1 hour and subsequently quenching. The crystal
modification gained by this procedure is called A.
The samples were investigated in a Philips EM-300 electron
microscope with a eucentric goniometer stage. The individual
crystals can be rotated and tilted up to 60Њ. Details of the
experimental and theoretical procedures have been published in
several papers.6–12 A short resume is given below.
(1) The basic zone is established by tilting about significant
axes and determining the positions corresponding to the largest
d-spacing. (2) An axis with strong reflections is established and
chosen as the tilt axis. Upon tilting, the diffraction pattern first
disappears and then reappears when a new zonal projection
appears, the angle is noted and the process is repeated. The
tilting is always performed both to the right and to the left
direction. (3) A second axis is chosen as the tilt axis and the
procedure repeated.
2. Methods
2.1 Preparation and characterization of tetrabenzo[ab, f,jk,o]-
[18]annulene by 1H NMR spectroscopy
3-Bromobenzaldehyde (1) was protected as acetal 2 and then
transformed by a Bouveault reaction in high yield to the alde-
hyde 3. The (E)-configured stilbene 6 was formed by a Wittig–
Horner reaction of 3 and the phosphonic ester 5 which could be
obtained from 1-chloromethyl-2-methylbenzene (4) and triethyl
phosphite. The reaction of 6 with aniline led in quantitative
† Permanent address: Institute of Macromolecular Compounds of
Russian Academy of Sciences, Bolshoi pr. 31, 199004 St. Petersburg,
Russia.
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890
1881

View Article Online
Table 1 Experimental details of X-ray scattering for modification B of
compound 8
Chemical Formula
Formula Weight
Crystal System
C32H24
229.79
Trigonal
¯
Space Group
R3
Absorption Coefficient
Final R indices [I > 2σ(I)]
R indices (all data)
Unit Cell Dimensions
0.53 mmϪ1
R1 = 0.052, wR2 = 0.124
R1 = 0.089, wR2 = 0.140
a = 28.1031(12) Å
b = 28.1031(12) Å
c = 7.1509(4) Å
4891.0(4) Å3
Cell Volume
Temperature of Data Collection
Z
298(2) K
9
Measured/Independent Reflections
[R (int)]
2783/2234 [R (int) = 0.0431]
trast function for transfer of the required spatial frequencies on
an adjacent area of the sample as described.6–11 Although lattice
planes cannot be directly viewed on the monitor, a fast Fourier
transform of the image enables a quick decision to be made
regarding the suitability of the area in question for imaging and
whether the correct transfer function has been chosen or not.
Speed and low dose facilities are essential to prevent degrad-
ation of the sample in the electron beam.
2.4 Generation of a molecular model
Ab initio Hartree–Fock calculations at the 3-21G level were
performed.14 The TURBOMOLE ab initio program package15
was used. The geometry of an isolated molecule 8 was com-
pletely optimized and then used to build up different dimers of
8 in which only intermolecular geometrical parameters were
optimized.
2.5 Simulation of electron diffraction patterns
To proceed with the simulations, the crystal unit cell and space
group are required. This information is obtained from the elec-
tron diffraction data combined with X-ray powder results. Gen-
erally several space groups can be postulated on the basis of the
observed extinctions. The number can normally be reduced if
the symmetry requirements of the molecule and the unit cell
are considered. For simulation of the diffraction pattern,
the TURBOMOLE calculated molecule is placed into the
unit cell with the required symmetry using CERIUS. Initial
qualitative agreement and a good value of the density can
generally be achieved within reasonable time. Further refine-
ment is required to produce a negative packing energy while
retaining the correct symmetry. In general an adjustment
of intramolecular rotations and intermolecular distances is
necessary.
Scheme 1
The diffraction patterns are scanned into a PC where they are
evaluated using the program ELD.6–12
From the diffraction patterns the unit cell dimensions can be
determined. On the basis of extinctions observed in different
projections, it is generally possible to reduce the number of
possible space groups to two or three. By subsequent molecular
modeling, the possible number of space groups can normally be
further reduced.
The recent successful structure determinations using direct
methods,13 maximum entropy statistics and/or simulations6–12
based on electron diffraction data demonstrate that the
methods all work under appropriate experimental conditions,
despite the fact that quantitative intensity values are acknow-
ledged to be poor. For maximum entropy methods correct
phase predictions are possible for reasons described in our pre-
vious papers.6,7,11 In the case of the simulation method, model
structures can be predicted because there are so many restric-
tions on (a) possible molecular conformations due to restric-
tions in bond lengths, torsion angles, and symmetry and (b)
possible crystallographic packing due to restrictions in van der
Waals distances, densities and crystal symmetry.
2.6 X-Ray structural analysis
For X-ray analysis relatively large single crystals are required
and these could not be obtained by the procedure described for
electron diffraction. Instead, compound 8 was dissolved in
toluene as before but was crystallized for one month at 30 ЊC.
The modification obtained by this procedure is called B; later
on it turned out that a third modification C could be
obtained by a very rapid crystallization. The experimental
details of the structural analysis of the crystals B are sum-
marized in Table 1; for the tiny crystals of C only the cell
parameters could be determined (Table 2).‡ The automatic
solution of the crystal structure was performed by the appli-
cation of the program SIR 9216 and the refinement by the
program SHELXL 97.17
To obtain a negative packing energy the number of possible
conformations is strongly restricted. The procedure has been
previously described in detail.6–12
2.3 High resolution imaging
High resolution images were obtained with a Philips 420 elec-
tron microscope. The samples are relatively beam sensitive so
that imaging is best achieved by first adjusting the phase con-
‡ CCDC reference number 188/172. See http://www.rsc.org/suppdata/
p2/1999/1881 for crystallographic files in .cif format.
1882
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890

View Article Online
Table 2 Cell parameters obtained from electron diffraction patterns
(modification A) and from X-ray measurements (modifications B and
C) of compound 8
Modification
a/Å
b/Å
c/Å
α/Њ
β/Њ
γ/Њ
A
B
C
12.11
18.1031
11.902
17.04
28.1031
5.300
5.19
7.1509
17.312
90
90
90
91.2
90
100.1
90
120
90
Fig. 2 Model structure used to simulate electron diffraction patterns.
¯
¯
Successive zones were found to be [100], [201], [101], [101].
The angles between [100] and the successive zones are ϩ12Њ,
ϩ23Њ and Ϫ24Њ.
3.2 Simulation of electron diffraction patterns
Simulation of the electron diffraction patterns requires the
experimental determination of the unit cell and space group.
The molecule with the quantum mechanically calculated geom-
etry is placed in the unit cell using the molecular modeling
program CERIUS in such a way that the symmetry require-
ments based on the observed extinctions in the electron diffrac-
tion experiments are satisfied.
The packing energy is calculated by the relationship:
E = Es ϩ Eb ϩ Et ϩ EvdW ϩ Ecoul ϩ Ehb
Fig. 1 Experimental electron diffraction patterns (LHS) and corre-
sponding simulated electron diffraction patterns.
bonded
non-bonded atoms
The packing energy was reduced by adjusting bond rotations.
Finally a value of Ϫ33 kcal molϪ1 per molecule was obtained.
Computer simulations of the possible molecular arrange-
ment within the cell led to the crystal structure shown in Fig. 2.
Using this structural model, the electron diffraction patterns
were simulated for the zones which had been obtained experi-
mentally. The results are shown in Fig. 1 together with the
corresponding experimental diffraction patterns. The best
agreement between theory and experiment was obtained
when the final optimized geometry calculated by the ab initio
methods was used.
In this P21/c crystal structure, there are two parallel stacks of
molecules. Within these stacks, the nearest neighboring mole-
cules form a dimer with Ci symmetry, the CC double bonds
of one molecule being placed above the double bond of the
neighboring molecule (Fig. 3). The proximity of the double
bonds is such as to be extremely favorable for crosslinking
3. Results and discussion
3.1 Electron diffraction
The very small single crystals (modification A) were used for
electron diffraction studies. The crystal was tilted about a prom-
inent axis which was later identified as b*. It is clear from the
cell parameters found (see Table 2) that the crystals are mono-
clinic, and that there must be 2 molecules per unit cell to obtain
a reasonable density. From the observed extinctions in the
experimental diffraction patterns (Fig. 1) a P21/c space group is
likely. This implies that the two molecules should occupy special
positions with the molecular inversion center coinciding with
that of the crystal. The appearance of weak forbidden reflec-
tions along the b*-axis is due to dynamical scattering. Similarly,
very weak secondary scattering can be observed along the c*-
axis. Such reflections are easily recognized in organic materials
and have been discussed in detail previously.6–12
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890
1883

View Article Online
Table 3 Results of 3-21 G ab initio calculations for the monomeric
compound 8 and molecular pairs of 8
∆E/kcal molϪ1
(molecular pair
formation)
Energy E
(arbitrary units)
Structure
Symmetry
Monomer
Molecular pairs
C2h
C2h
Ci
C2
C1
Ϫ1218.894551
Ϫ2437.791475
Ϫ2437.791485
Ϫ2437.787947
Ϫ2437.788246
—
Ϫ4.67
Ϫ4.67
Ϫ2.46
Ϫ2.64
striations are oriented at an angle of about 42Њ to the lines and
form an angle of about 84Њ to each other. The structures have a
high degree of perfection.
3.4 Simulation of image
Our model of the bc projection is shown in Fig. 5a. The simu-
lated image using a slice thickness of 3.4 Å and taking account
of the transfer function for a defocus value of Ϫ1000 Å and a
thickness of 121.1 Å is shown in Fig. 5c. It is clear that the
simulations agree well with the HRTEM image b.
Fig. 3 Intrastack dimer with Ci symmetry used in electron diffraction
simulations.
3.5 X-Ray diffraction
The crystal structure determined for the large crystals B (Fig. 6)
does not correspond to that of the small ones. These crystals
¯
belong to the R3 space group (No. 148) with the cell parameters
¯ ¯
shown in Table 2. The [522] projection serves to demonstrate
specific features which can be related to the structure deter-
mined by electron diffraction because it shows the stacking of
the dimers.
The unit cell contains 9 molecules in special positions with
inversion center symmetry. Thus, an individual molecule from
the crystal structure is of Ci symmetry (Fig. 7a,b). The torsion
angles at the o-phenylene–vinylene and m-phenylene–vinylene
building blocks are 23.1Њ and 16.9Њ, respectively.
The neighboring molecules along the c-axis form a pair of Ci
symmetry shown in Fig. 8. Within this dimer, the distance
between the atoms of CC double bonds of different molecules
is about 3.6 Å which also makes this crystal structure a good
candidate for displaying solid-state photodimerization and
photopolymerization properties (Fig. 9).
Along the c-axis, there are 3 parallel stacks of molecules, the
neighboring molecules from different stacks forming a C1 pair
(Fig. 10). The angle between the two molecules in these pairs is
80Њ.
As expected, simulated electron diffraction patterns from
this structure do not agree with the experimental electron dif-
fraction patterns, confirming the result that the crystals form
different polymorphs.
Fig. 4 Interstack dimer with C1 symmetry obtained from electron
diffraction simulations.
3.6 Ab initio calculations of 8
On the basis of the molecular geometry and packing deter-
mined by simulation of the electron diffraction patterns, a
starting geometry for the quantum mechanical calculations
was available.
(ca. 3.5 Å), a prediction which was subsequently corroborated
by the experiment. The torsion angles about the o-phenylene–
vinylene and m-phenylene–vinylene bonds are 41.6Њ and 34.1Њ
respectively.
Two neighboring molecules in different stacks form a
molecular pair of C1 symmetry (Fig. 4). The angle between the
two molecules in these interstack pairs is 72Њ.
The completely optimized geometry of the molecule 8 is
shown in Fig. 11. It has C2h symmetry and is non-planar (the
total energy value is given in Table 3). The comparison of the
ab initio optimized geometrical parameters with those found by
X-ray single crystal analysis is presented in Table 4. The agree-
ment between calculated and experimental values is very good
for bond lengths and bond angles. The calculated torsion angles
at the o-phenylene–vinylene and m-phenylene–vinylene build-
ing blocks are Ϫ41.6Њ and 35.9Њ, respectively. In the crystal
structure determined by X-ray analysis, these angles are closer
to zero (Ϫ20.9Њ and 11.1Њ, respectively). However, when the
3.3 High resolution imaging
The high resolution images of the [100] projection are shown in
Fig. 5b. The Figure reveals extensive areas consisting of straight
lines with a separation of 17.2 Å/two lines. Inside these struc-
tures there are striations on a much smaller scale. These
1884
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890

View Article Online
Fig. 5 Model structure (a) with high resolution image (b) and simulated image (c).
Table 4 Selected geometrical parameters of 8
Bond lengths/Å
Bond angles/Њ
Torsion angles/Њ
2-3
3-4
4-5
1-2-3
2-3-4
3-4-5
4-5-6
1-2-3-4
3-4-5-6
Ab initio calculated
Found by X-ray analysis
1.483
1.477
1.324
1.322
1.481
1.477
120.7
120.5
124.4
126.2
124.8
125.4
121.3
121.5
Ϫ41.6
Ϫ20.9
35.9
11.1
molecules in the ab initio optimized geometry are placed into
the unit cell determined by electron diffraction, the agreement
between experimental and simulated electron diffraction
patterns is very good.
Dimerization was indicated by the structures determined
both by electron diffraction and by X-ray diffraction. There-
fore the appropriate quantum mechanical calculations were
performed for molecular pairs of 8.
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890
1885

View Article Online
Fig. 8 Intrastack dimer with Ci symmetry from X-ray structure.
Fig. 9 Distance between molecules within stacks from X-ray structure.
¯ ¯
Fig. 6 Structure obtained by X-ray analysis showing [522] projection.
Fig. 10 Interstack dimer with C1 symmetry from X-ray structure.
Gas-phase calculations of possible dimeric structures led to
four most stable molecular pairs of C2h (Fig. 12), Ci (Fig. 13),
C2 (Fig. 14) and C1 (Fig. 15) symmetries. It can be seen from the
data given in Table 3, the C2h and Ci molecular pairs are of
comparable stability and are more stable than the C1 and C2
aggregates. The energy of intermolecular attraction within the
C2h and Ci pairs is about 4.7 kcal molϪ1, which is a sufficiently
high value to be able to predict definitely the stability of such
dimers in solution or in the gas phase.
Within the C2h molecular pair, the molecules are shifted with
respect to each other so as to keep their common mirror plane
(Fig. 12a). In the Ci pair, the shift is realized mainly in a direc-
tion perpendicular to that found for the C2h dimer (Fig. 13a).
It should be noted that in both C2h and Ci aggregates, a pair
of CC double bonds of one molecule is situated above the
corresponding pair of the other at a distance of about 4 Å
(Figs. 12b, 13b).
Fig. 7 Individual molecule from X-ray structure showing Ci sym-
metry, (a) top view; (b) side view.
1886
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890

View Article Online
Fig. 13 Intrastack dimer in Ci symmetry (ab initio calculations),
(a) top view; (b) side view.
Fig. 11 Molecular conformation for C2h symmetry obtained from
ab initio calculation, (a) top view; (b) side view.
Fig. 12 Intrastack dimer in C2h symmetry (ab initio calculations), (a)
top view; (b) side view.
Fig. 14 Intrastack dimer in C2 symmetry (ab initio calculations),
(a) top view; (b) side view.
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890
1887

View Article Online
Fig. 16 Reaction spectra of the monochromatic irradiation (λ = 366
nm) of 8 in CHCl3: the maximum of the absorbance A at 292 nm
decreases almost completely within 1 h.
Fig. 15 Interstack dimer in C1 symmetry (ab initio calculations).
Within the C2 pair the molecules are rotated with respect to
each other by about 38Њ (Fig. 14a). The intermolecular attrac-
tion energy is almost two times lower than in the C2h and Ci
pairs (Table 3). For one of the intermolecular pairs of CC
double bonds the corresponding intermolecular distances
between the C-atoms are also about 4 Å. However, the angle
between the planes of these double bonds amounts to about 53Њ
(Fig. 14b), making an overlap between π-orbitals less favorable.
In the C1 molecular pair (Fig. 15), ab initio calculations indi-
cate that the molecules are arranged so as to form an angle of
95Њ to one another.
Scheme 2
AM 400 spectrometer. Mass spectra were recorded in EIϩ made
on a Finnigan MAT 95 spectrometer. The starting compound
3-bromobenzaldehyde (3) is commercially available.
3.7 Photochemistry
Irradiation of 8 in the polymorphous crystalline state led in a
topochemical photodimerization to dimer 9 and to polymers.
The short distance between the olefinic centers, i.e. the strong
π–π-interaction enables the cycloaddition to occur (topo-
chemical control). As soon as the first 4-membered ring is
closed, the distance of the remaining double bonds is too far
1-Bromo-3-(dimethoxymethyl)benzene (2). Dowex 50W-X8
resin (2.0 g) was added to 3-bromobenzaldehyde (11.10 g, 60.0
mmol) in 100 cm3 dry methanol and 50 cm3 (459 mmol) tri-
methyl orthoformate. After refluxing for 2 h, the reaction mix-
ture was cooled to ambient temperature, Na2CO3 (2.0 g) was
added under stirring for 10 min and the volatile compounds
were removed at reduced pressure. The crude product was dis-
solved in 100 cm3 of hexane and filtered in order to remove the
impurities. Evaporation of the hexane led to a colorless oil
1
for further selective [2π ϩ 2π] cycloaddition reactions. The H
NMR spectrum of 9 shows the A2X2 spin pattern (δ = 3.51 and
4.42) for the cyclobutane protons (C2 symmetry) and three AB
systems for the remaining olefinic protons. The coupling con-
stants of 16 Hz prove that the original (E)-configurations were
preserved. The unambiguous structure determination of 9 is
based on the fact that only the shift of one molecule 8 with the
carbon atoms C-4 and C-5 over the carbon atoms C-10 and
C-11 of a second molecule 8 leads to a 4-membered ring with an
A2X2 spin pattern. All other overlaps, like for example C-4/5
with C-22/21 of C-4/5 with C-16/15 (Scheme 1) would generate
AAЈXXЈ spin patterns. Dimer 9 (Scheme 2) can be rationalized
as the photoproduct of modification A and/or C; modification
B as well as prolonged irradiation of all crystals led to the same
result as the irradiation in solution, namely to polymerization
reactions. Broad resonance signals for the methine protons in
the polymer prove the CC bond formation to (different)
neighboring molecules. The reaction spectra (Fig. 16) of the
monochromatic irradiation in chloroform showed the almost
complete disappearance of the stilbenoid ππ* band. An isos-
bestic point could be found at λ = 253 nm.
1
(12.20 g, 88%): H NMR δ 3.30 (s, 6 H, OCH3), 5.34 (s, 1 H,
CH), 7.21 (m, 1 H, 5-H), 7.40 (m, 2 H, 4-H, 6-H), 7.60 (m,
1 H, 2-H); 13C NMR δ 52.6 (OCH3), 102.0 (CH), 122.4 (C-1),
125.3, 129.8, 129.9, 131.5 (C-2, C-4, C-5, C-6), 140.4 (C-3);
MS (70 eV) m/z 232/230 (Mϩ, 7, Br pattern), 201 (98), 199
(100); C9H11BrO2 (231.1) calculated C, 46.78; H, 4.80; Br, 34.58;
found C, 46.69; H, 4.79; Br, 34.06%.
3-(Dimethoxymethyl)benzaldehyde (3). n-Butyllithium (40
cm3 of a 1.6 M solution in n-hexane, 64.0 mmol) was added
dropwise under argon to a solution of 2 (12.02 g, 52.0 mmol) in
200 cm3 of dry diethyl ether which was cooled to Ϫ78 ЊC. The
mixture was stirred for 0.5 h and then warmed to 0 ЊC. Dry
DMF (10.0 cm3, 130 mmol) was added, the mixture stirred for
1 h and then quenched with 50 cm3 water. The ether layer was
washed with brine, dried over MgSO4, filtered and the solvent
removed to yield a pale yellow oil (9.05 g, 97%): 1H NMR δ 3.30
(s, 6 H, OCH3), 5.42 (s, 1 H, CH), 7.55 (m, 1 H, 5-H), 7.68 (m,
1 H, 4-H), 7.81 (m, 1 H, 6-H), 7.94 (m, 1 H, 5-H), 9.99 (s, 1 H,
CHO); 13C NMR δ 52.6 (OCH3), 102.1 (CH), 128.3, 128.9,
129.4, 132.7 (C-2, C-4, C-5, C-6), 136.1 (C-1), 139.3 (C-3), 192.1
(CHO). The crude product was directly used for the next step.
Experimental
Synthesis and spectroscopic characterization
Melting points are uncorrected. 1H and 13C NMR spectra were
measured in solution (CDCl3 or solvent indicated) on a Bruker
1888
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890

View Article Online
3
Diethyl (2-methylbenzyl)phosphonate (5). 1-Chloromethyl-2-
methylbenzene (4) (27.04 g, 192.3 mmol) and triethyl phosphite
(35.0 g, 211 mmol) were refluxed for 2 h. The excess phosphite
was removed at 1 Torr (1.33 × 102 Pa) and the residue distilled
between 85 and 86 ЊC at 0.001 Torr (1.33 × 10Ϫ1 Pa) to yield
δ 7.08 (d, J 16.1 Hz, 4 H, 4-H), 7.20 (m, 4 H, 1-H), 7.27 (m,
4 H, 7-H), 7.32 (t, 2 H, 2-H), 7.70 (m, 4 H, 6-H), 7.83 (d, 3J 16.1
Hz, 4 H, 5-H), 8.29 (d, 2 H, 23-H); the correlation of the signals
was achieved by NOE measurements. The low solubility of 8 in
usual NMR solvents did not permit the measurement of a 13C
NMR spectrum. MS (70 eV) m/z 408 (Mϩ, 100), 204 (M2ϩ, 14);
C32H24 (408.5) calculated C, 94.08; H, 5.92; found C, 93.80; H,
6.04%.
1
30.0 g (64%) of a colorless oil: H NMR δ 1.17 (t, 6 H, CH3),
5
2
2.33 (d, J (H,P) 1.6 Hz, 3 H, CH3), 3.11 (d, J (H,P) 21.9 Hz,
2 H, CH2), 3.92 (m, 4 H, OCH2), 7.09 (m, 3 H), 7.20 (m, 1 H)
3
(3-H, 4-H, 5-H, 6-H); 13C NMR δ 15.8 (d, J (C,P) 5.9 Hz,
1
Photodimerization of 8
CH3), 19.3 (2-CH3), 30.5 (d, J (C,P) 138.2 Hz, CH2), 61.4 (d,
4
2J (C,P) 6.7 Hz, OCH2), 125.4 (d, J (C,P) 3.0 Hz, C-3), 126.4
A suspension of 16.0 mg (0.04 mmol) 8 in 500 ml water
was circulated through a photoreactor equipped with a 150 W
mercury lamp (Hanau TQ 150 Z 3) with a main emission
between 300 and 400 nm. After 1 h the irradiation was stopped
and the suspension filtered. The dry residue was dissolved in
(d, 4J (C,P) 3.5 Hz, C-5), 129.5 (d, 2J (C,P) 9.4 Hz, C-1), 129.8
(d, 5J (C,P) 2.7 Hz, C-4), 130.0 (d, 3J (C,P) 5.5 Hz, C-6), 136.3
(d, 3J (C,P) 6.5 Hz, C-2); MS (70 eV) m/z 242 (Mϩ, 48), 186 (33),
105 (100); C12H19O3P (242.3), calculated C, 59.50; H, 7.91;
found C, 59.38; H, 7.79%.
1
CDCl2–CDCl2. The H NMR spectrum at 323 K shows the
signals of 8 and its dimer 9 in a ratio of 30:70 and a small
amount of oligomers. Prolonged irradiation yielded higher
amounts of oligomeric material. 9: 1H NMR (C2C2Cl4):
δ = 3.51, 4.42 (A2X2, 4 H, cyclobutane ring); 6.10, 6.42, 6.46
(3 d, 3J = 16 Hz, 6 H, A parts of three olefin. AB systems), 6.70–
7.80 (m, 34 H, arom. H and B parts of AB systems), 7.89 (“s”,
2 H, arom. H), 8.22 (“s”, 2 H, arom. H). MS (FD): m/z (%) =
(E)-3-[2-(2-Methylphenyl)ethenyl]benzaldehyde (6). To
a
solution of 5 (9.69 g, 40.0 mmol) in 100 cm3 1,2-dimethoxy-
ethane (DME) NaH (3.84 g, 160.0 mmol) was added under
argon. The mixture was refluxed for 20 min followed by drop-
wise addition of 3 (8.73 g, 48.4 mmol) dissolved in 50 cm3 of
dry DME. After further refluxing for 1 h, the reaction mixture
was cooled, quenched carefully with 20 cm3 of methanol and
poured into a stirred mixture of 200 cm3 ether and 100 cm3 2 M
HCl. After 2 h the organic layer was separated, washed with 50
cm3 saturated NaHCO3 solution, dried over MgSO4 and fil-
tered. Removal of the solvent gave an oil which was chromato-
graphed on silica gel using dichloromethane as eluant. A pale
816 (31) [Mϩ ], 652 (85), 408 (100).
ؒ
Conclusions and outlook
Disc-like areno-condensed annulenes exhibit a high aggregation
tendency in solution, in mesophases and in the solid state.2 We
studied this phenomenon in the solid state for the model com-
pound 8 which crystallizes in three different modifications. The
results discussed above show that gas phase ab initio calcu-
lations predict reasonable dimeric structures of 8 which agree,
in most of the essential features, with those found in crystals.
However, some important intermolecular parameters, such as
the precise shift of molecules in specific directions, are influ-
enced by the crystal field. Therefore a combination of both
methods is essential in order to understand solid state
properties. Subsequent gas-phase ab initio quantum-chemical
calculations of the intermolecular interaction within different
molecular pairs are important in order to obtain inform-
ation about possible physical effects such as arrangement and
order in liquid crystalline (LC) phases, photoconductivity or
photopolymerization and photocrosslinking. Irrespective of a
topochemical photodimerization 8→9, all crystal modifications
of 8 enter a polymerization on irradiation. The arrangement
of neighboring molecules in the crystal structures found is, in
principle, suitable for photoconductivity as well as for photo-
crosslinking properties. However, photocrosslinking in the
crystal state is rather probable due to the very close distance (ca.
3.5 Å) of double bonds of neighboring molecules. This photo-
crosslinking would destroy the π-conjugation in a part of the
molecule and can, thus, prevent or reduce the photoconductiv-
ity. However, in the liquid crystalline state, photocrosslinking is
much less probable during the singlet excited state lifetime. The
fluorescence lifetimes of similar [18]annulenes condensed with
aromatic ring systems were measured to be less than 10 ns.2
In that case, photoconductivity may become a dominating
physical property if a stacking (columnar) arrangement of
molecules survives.
1
yellow oil (4.79 g, 54%) was isolated: H NMR δ 2.44 (s, 3 H,
CH3), 7.02, 7.43 (AB, 3J 16.2 Hz, 2 H, olefin. H), 7.20 (m, 3 H,
arom. H), 7.51 (m, 2 H, arom. H), 7.75 (m, 2 H, arom. H), 8.01
(m, 1 H, 2-H), 10.04 (s, 1 H, CHO); 13C NMR δ 19.9 (CH3),
125.4, 126.2, 127.1, 128.0, 128.2, 128.3, 128.8, 129.3, 130.4,
132.3 (arom. and olefin. CH), 135.7, 135.9, 136.7, 138.5 (arom.
Cq), 192.2 (CHO); MS (70 eV): m/z 222 (Mϩ, 100), 193 (26), 178
(59); C16H14O (222.3) calculated C, 86.45; H, 6.35; found C,
86.50; H, 6.43%.
(E,E)-3-[2-(2-Methylphenyl)ethenyl]-N-phenylbenzaldimine
(7). Aniline (2.33 g, 25.0 mmol) and 6 (4.58 g, 20.6 mmol) were
heated with stirring for 2 h at 60 ЊC. During the reaction the
generated water was several times removed at 15 Torr (2.0 × 103
Pa). After the condensation had come to an end, excess aniline
was evaporated at 0.01 Torr (1.33 Pa) to leave 6.13 g (100%) of a
yellow viscous oil which was analytically pure. 1H NMR δ 2.45
3
(s, 3 H, CH3), 7.05, 7.41 (AB, J 16.2 Hz, olefin. H), 7.21 (m,
6 H, arom. H), 7.41 (m, 2 H, arom. H), 7.49 (m, 1 H, arom. H),
7.62 (m, 2 H, arom. H), 7.77 (m, 1 H, arom. H), 8.07 (m, 1 H,
2-H), 8.49 (s, 1 H, CHN); 13C NMR δ 19.9 (CH3), 120.8, 125.4,
126.0, 126.2, 126.5, 127.4, 127.7, 128.1, 129.0, 129.1, 129.2,
129.3, 130.4 (arom. and olefin. H), 135.9, 136.1, 136.5, 138.2
(arom. C ), 151.9 (C -N), 160.2 (CH᎐N); MS (70 eV) m/z 297
᎐
q
q
(Mϩ, 14), 110 (100), 93 (64); C22H19N (297.4) calculated C,
88.85; H, 6.44; N, 4.71; found C, 88.71; H, 6.42; N, 4.77%.
(4E,10E,15E,21E)-Tetrabenzo[ab, f,jk,o]cyclooctadecene
(tetrabenzo[ab, f,jk,o][18]annulene) (8). Dry DMF (150 cm3)
was degassed several times and heated under argon to 80–90 ЊC.
Potassium tert-butoxide (9.32 g, 83.05 mmol) was slowly added
with vigorous stirring followed by a dropwise addition of 7
(4.94 g, 16.61 mmol) in 200 cm3 of dry DMF. The mixture
turned violet and was stirred for 1.5 h at 90 ЊC. After cooling to
ambient temperature, 100 cm3 of water and 100 cm3 of 2 M HCl
were added. The formed yellowish–brown precipitate was
sucked off, washed first with 200 cm3 water, then with 50 cm3
methanol and 20 cm3 acetone and finally with 20 cm3 pentane
to yield 1.24 g (37%) of a beige solid which could be recrystal-
The crystal structures found by electron diffraction and
X-ray analysis are different polymorphs with strongly favorable
negative crystal packing energies. The total packing energies of
the small crystals whose structure was determined by electron
diffraction are Ϫ33 and Ϫ38 kcal molϪ1 per molecule for the
structures simulated with the ab initio gas-phase conformation
and conformation taken out from the X-ray structure. The van
der Waals energy of the large crystals is Ϫ47 kcal molϪ1 per
molecule. These values are sufficiently close to one another to
explain the polymorphism.
lized from toluene, mp >300 ЊC: IR (KBr) ν
˜
/cmϪ1 3050, 1590,
1485, 1315, 960, 780, 755, 690; H NMR (C2D2Cl4, 400 MHz)
1
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890
1889

View Article Online
9 I. G. Voigt-Martin, Z. X. Zhang, D. H. Yan, A. Yakimanski,
R. Wortmann, R. Matschiner, P. Krämer, C. Glania and N. Detzer,
Colloid Polym. Sci., 1997, 275, 18.
10 A. V. Yakimanski, I. G. Voigt-Martin, U. Kolb, G. N. Matveeva and
Z. X. Zhang, Acta Crystallogr., Sect. A, 1997, 53, 603.
11 I. G. Voigt-Martin, Z. X. Zhang, U. Kolb and C. Gilmore, Ultra-
microscopy, 1997, 68, 43.
Acknowledgements
The authors gratefully acknowledge financial support by the
Deutsche Forschungsgemeinschaft and by the Fonds der
Chemischen Industrie.
12 I. G. Voigt-Martin, Gao Li, A. Yakimanski, J. J. Wolff and H. Gross,
J. Phys. Chem., 1997, A 101, 7265.
13 D. Dorset, Structural Electron Microscopy, Plenum Press, New York
and London, 1995.
14 J. S. Binkley, J. A. Pople and W. J. Hehre, J. Am. Chem. Soc., 1980,
102, 939.
15 R. Ahlrichs, M. Bär, M. Häser, H. Horn and C. Kölmel, Chem.
Phys. Lett., 1989, 162, 165.
16 A. Altomare, G. Cascarano, C. Giacovazzo, A. Guagliardi, M. C.
Burla, G. Polidori and M. Camalli, J. Appl. Crystallogr., 1994, 27,
435.
17 G. M. Sheldrick, SHELXL 97, Program for crystal structure
refinement 1997, University of Göttingen, Germany.
References
1 H. Meier and K. Müller, Angew. Chem., 1995, 107, 1598; Angew.
Chem., Int. Ed. Engl., 1995, 34, 1003.
2 K. Müller, H. Meier, H. Bouas-Laurent and J. P. Desvergne, J. Org.
Chem., 1996, 61, 5474.
3 H. Kretzschmann, K. Müller, H. Kolshorn, D. Schollmeyer and
H. Meier, Chem. Ber., 1994, 127, 1735.
4 H. Meier, K. Müller and M. Fetten, J. Inf. Rec., 1996, 22, 421.
5 H. Meier, Th. Lifka and K. Müller, J. Inf. Rec. Mater., 1994, 21,
457.
6 I. G. Voigt-Martin, D. H. Yan, C. Gilmore, K. Shankland and
G. Bricogne, Ultramicroscopy, 1994, 56, 271.
7 I. G. Voigt-Martin, D. H. Yan, A. Yakimansky, D. Schollmeyer,
C. J. Gilmore and G. Bricogne, Acta Crystallogr., Sect. A, 1995, 51,
849.
8 I. G. Voigt-Martin, Li Gao, A. Yakimanski, G. Schulz and
J. J. Wolff, J. Am. Chem. Soc., 1996, 118, 12830.
Paper 9/03451J
1890
J. Chem. Soc., Perkin Trans. 2, 1999, 1881–1890