Tetrahedral onsager crosses for solubility improvement and crystallization bypass

Article abstract of DOI:10.1021/ja010019h

Pure organic molecules exhibiting a suitable concave rigid shape are expected to give porous glasses in the solid state. Such a feature opens new opportunities to avoid crystallization and to improve molecular solubility in relation to the high internal energy of these solid phases. To quantitatively explore the latter strategy, a series of rigid tetrahedral conjugated molecules nC and the corresponding models nR have been synthesized. Related to the present purpose, several properties have been investigated using UV absorption, steady-state fluorescence emission, differential scanning calorimetry, ¹H NMR translational self-diffusion, magic angle spinning ¹³C NMR, and multiple-beam interferometry experiments. The present tetrahedral crosses are up to 8 orders of magnitude more soluble than the corresponding model compounds after normalization to the same molecular length. In addition, they give concentrated monomeric solutions that can be used to cover surfaces with homogeneous films whose thickness goes down to the nanometer range. Such attractive features make cross-like molecular architectures promising for many applications.

Full text of DOI:10.1021/ja010019h

J. Am. Chem. Soc. 2001, 123, 8177-8188
8177
Tetrahedral Onsager Crosses for Solubility Improvement and
Crystallization Bypass
Isabelle Aujard,† Jean-Pierre Baltaze,‡ Jean-Bernard Baudin,† Emmanuelle Cogne´,†
Fabien Ferrage,‡ Ludovic Jullien,*,† EÄ ric Perez,§ Vale´ry Pre´vost,† Lin Mao Qian,§ and
Odile Ruel†
Contribution from the Department of Chemistry and the Department of Physics,
EÄ cole Normale Supe´rieure, 24, rue Lhomond, F-75231 Paris Cedex 05, France
ReceiVed January 2, 2001
Abstract: Pure organic molecules exhibiting a suitable concave rigid shape are expected to give porous glasses
in the solid state. Such a feature opens new opportunities to avoid crystallization and to improve molecular
solubility in relation to the high internal energy of these solid phases. To quantitatively explore the latter
strategy, a series of rigid tetrahedral conjugated molecules nC and the corresponding models nR have been
synthesized. Related to the present purpose, several properties have been investigated using UV absorption,
steady-state fluorescence emission, differential scanning calorimetry, 1H NMR translational self-diffusion, magic
angle spinning 13C NMR, and multiple-beam interferometry experiments. The present tetrahedral crosses are
up to 8 orders of magnitude more soluble than the corresponding model compounds after normalization to the
same molecular length. In addition, they give concentrated monomeric solutions that can be used to cover
surfaces with homogeneous films whose thickness goes down to the nanometer range. Such attractive features
make cross-like molecular architectures promising for many applications.
Introduction
one may formally decompose the latter process in two steps,
sublimation and solvation, involving M(g) in the gas phase:
Chemists from different fields often face problems for
solubilizing organic molecules or for avoiding their crystalliza-
tion. In synthesis, poor solubility constrains to use large amounts
of solvent. In addition such a feature may prevent the application
of efficient purification procedures such as column chromatog-
raphy, for instance. For drug applications, solubility is also an
essential aspect for access to biological targets. To obtain
amorphous films free from inclusion of solid particles or
aggregates for optical uses, the manufacturing of organic layers
by solution processing requires avoiding phase separation during
solvent evaporation. Hence, many approaches have been empiri-
cally developed to improve solubility at the molecular level.1,2
Nevertheless, the diversity of situations encountered by chemists
makes always desirable the design of new strategies for favoring
dissolution of solid organic phases.
M(s) ) M(g)
M(g) ) M(s)
(2)
(3)
In the first class of approaches to promote dissolution, the
standard3 Gibbs free energy of dissolution ∆1G°(T) ) ∆1H°(T)
- T∆1S°(T) is dominated by the enthalpic term.4 An appropriate
solvent exhibiting a polarizability rather similar to that of the
solute is first chosen to avoid phase separation between the
solute and the solvent (the first aspect of the rule-of-thumb “like
dissolves like”).5 Even in such a case, the dissolution enthalpy
∆1H°(T) is expected to remain positive in the absence of specific
interactions. In fact, a solid phase of an organic compound is
more compact than the corresponding liquid phase; thus,
stabilizing van der Waals interactions are expected to be stronger
in the solid phase (∆2H°(T) > -∆3H°(T) S ∆1H°(T) > 0).
Hence dissolution is generally determined along this strategy
by introducing specific groups devoted to strongly interact with
the solvent (charged groups in polar solvents or hydrogen bonds
in protic solvents, for instance; the second aspect of the rule-
of-thumb, “like dissolves like”). Such an approach gives rise
to a large negative contribution to the standard solvation
Most existing approaches belong to two different categories
that are more easily analyzed in reference to a process leading
to the dissolution of a molecular compound M such as:
M(s) ) M(s)
(1)
where M(s) and M(s) respectively stand for M in solid phase
and in solution. For the convenience of the subsequent analysis,
† De´partement de Chimie (CNRS UMR 8640), Ecole Normale Su-
pe´rieure, 24, rue Lhomond, F-75231 Paris cedex 05 (France). Fax: (+33)
1 44 32 33 25. E-mail: Ludovic.Jullien@ens.fr.
(3) As commonly encountered, standard states are defined in the text as
(i) the pure solid in the solid phase; (ii) the ideal gas at 1 bar in the gas
phase; (iii) the ideal solute at 1 M for the diluted solution at T.
(4) In the following, the subscripts to the operator ∆ refer to the
considered process: (1) for dissolution, (2) for sublimation, and (3) for
solvation.
(5) (a) Hildebrand, J. H.; Prausnitz, J. M.; Scott, R. L. Regular and
Related Solutions; Van Nostrand Reinhold: Princeton, 1970. (b) Reichardt,
C. SolVent and SolVent Effects in Organic Chemistry, 2nd ed; VCH:
Weinheim, Basel, Cambridge, New York, 1988. (c) Israelachvili, J. N.
Intermolecular and Surface Forces, 2nd ed; Academic Press: London, San
Diego, New York, Boston, Sydney, Tokyo, Toronto, 1992.
‡ De´partement de Chimie (CNRS UMR 8642), Ecole Normale Su-
pe´rieure, 24, rue Lhomond, F-75231 Paris cedex 05 (France).
§ De´partement de Physique (CNRS UMR 8550), Ecole Normale Su-
pe´rieure, 24, rue Lhomond, F-75231 Paris cedex 05 (France).
(1) Techniques of Organic Chemistry; Weissberger, A., Ed.; Separation
and Purification; Interscience Publishers: New York, 1956; Vol. III, Part
I, p 549.
(2) Vogel’s Textbook of Practical Organic Chemistry, 4th ed; Long-
man: London and New York, 1978.
10.1021/ja010019h CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/07/2001

8178 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
Figure 1. Schematic picture illustrating the phenomena expected to occur during the evaporation of a solution of tetrahedral Onsager crosses under
quasistatic conditions. (a) diluted solution, isolated crosses; (b) above a critical concentration C*; (c) without solvent, disordered porous solid
phase.
enthalpy ∆3H°(T) that is expected to counterbalance the above-
mentioned density effect. In the second category of approaches
to promote dissolution, enthalpy plays a minor role, and entropy
dominates the Gibbs free energy of dissolution. The strategy is
based on introducing molecular degrees of freedom that are
frozen in the solid state but active in solution.6 This is most
conveniently achieved by playing with internal rotations. Thus,
flexible substituents such as alkyl, polyoxyethylenic, or siloxane
chains are often grafted to a solute to improve its solubility.
Hence, most dissolution strategies rely on the features of the
solute M(s). Nevertheless, consideration of the processes 1, 2,
and 3 shows that the Gibbs free energy of dissolution of an
organic compound is governed by its chemical potential in both
phases: solid and solution. Thus, one can also destabilize the
solid phase to improve solubility. Then the reasoning exposed
above suggests that decreasing the density of the solid phase
could provide an efficient way to increase the standard enthalpy
of the solid phase so as to decrease the standard Gibbs free
energy of dissolution. A possible approach to obtain poorly
compact solid phases consists of designing molecules of suitable
shapes.
porous glass after solvent removal (Figure 1). It is the latter
porosity that is anticipated to ultimately drive the solubility
increase. In the present manuscript, solubility and differential
scanning calorimetry experiments have been performed to
quantitatively analyze the solubility of the nC molecules. In
light of the behavior of the model compounds nR, the results
are analyzed with a thermodynamic model developed to extract
the true significance of the molecular architecture on solubility
properties. In addition to the preceding favorable feature for
dissolution, the large anisotropy for molecular interaction
between Onsager crosses may also keep low enough the solution
viscosity during the solvent evaporation up to the critical
concentration C*. Together with solution homogeneity, such a
feature could be useful for applications in which a homogeneous
coating is expected from the spreading and the evaporation of
a solution (spin-coating for instance). Although not yielding a
regular lattice, the latter approach to obtain organic nanoporous
materials can be also considered as an interesting alternative to
engineer a crystal network by design from molecular structure.9
In fact, the rigidity of Onsager crosses should prevent some
drawbacks that are commonly encountered when using nonco-
valent strategies to build up nanoporous media since crystal
collapse usually occurs upon removal of interstitial solvent.10
In the present paper, we also report on complementary NMR
and interferometry experiments aiming to illustrate the latter
aspects.
The present paper is conceived to explore the latter dissolution
strategy based on engineering the destabilization of the solid
phase of a molecular organic compound. Despite existing
publications evoking related facts during the course of diverse
studies,7 this report seems to be the first to quantitatively
evaluate such an approach for improving solubility and to draw
its limits. Among possible molecular structures for the present
purpose, a series of molecules nC that can be reasonably
considered as Onsager crosses has been selected. Onsager
crosses are attractive hard nonconvex bodies formed by rigidly
connecting elongated rods that were named as a tribute to
Onsager’s seminal contributions to the theory of lyotropic
crystals.8 Under quasi-static conditions, solvent evaporation from
a diluted solution of isolated Onsager crosses is expected to
give (i) a homogeneous ordered network of connected crosses
at a critical concentration C* related to the rod length,8 (ii) a
Results
Molecular Design and Synthesis. Many chemical com-
pounds obey the generic structure of Onsager crosses. In fact,
several have already been reported in the litterature.7,11 For
the present purpose, we sought for a homologous series of
crosses made of rigid rods grafted onto a central node and for
the series of corresponding model compounds. Different syn-
thetic methodologies lead to rodlike structures exhibiting a
reduced conformational flexibility. Among them, we favored
the p-phenylyne building strategy via palladium-catalyzed
coupling. The latter has been already extensively used in
supramolecular chemistry. It is particularly suitable for obtaining
a series of molecules of increasing length.12 Instead of adopting
an octahedral node as it has been done in theoretical studies,8
the tetrahedral tetraphenylmethane core was chosen.7d,13 In view
(6) See for instance: (a) Hildebrandt, J. H.; Scott, R. L. The Solubility of
Nonelectrolytes, 3rd ed; Reinhold Publishing Corporation, New York, 1950;
(b) Leffler, J. E.; Grunwald, E. Rates and Equilibria of Organic Reactions
as Treated by Statistical, Thermodynamic, and Extrathermodynamic
Methods; Dover: New York, 1963. (c) Benson, S. W. Thermochemical
Kinetics Methods for Estimation of Thermochemical Data and Rate
Parameters, 2nd ed; John Wiley & Sons: New York, London, Sydney,
Toronto, 1976.
(7) (a) Wilson, L. M.; A. C. Griffin, J. Mater. Chem. 1993, 3, 991-
994. (b) Mongin, O.; Gossaeur, A. Tetrahedron 1997, 53, 6835-6846. (c)
Johansson, N.; Salbeck, J.; Bauer, J.; Weisso¨rtel, F.; Bro¨ms, P.; Andersson,
A.; Salaneck, W. R. AdV. Mat. 1998, 10, 1136-1141 and references therein.
(d) Wang, S.; Oldham, W. J., Jr.; Hudack, R. A., Jr.; Bazan, G. C. J. Am.
Chem. Soc. 2000, 122, 5695-5709 and references therein.
(9) Zaworotko, M. J. Angew. Chem., Int. Ed. 2000, 39, 3052-3054.
(10) Batten, S. R.; Robson, R. Angew. Chem., Int. Ed. 1998, 37, 1460-
1494.
(11) Yao, Y.; Tour, J. M. J. Org. Chem. 1999, 64, 1968-1971.
(12) (a) Tour, J. M. Acc. Chem. Res. 2000, 33, 791-804. (b) Lavastre,
O.; Ollivier, L.; Dixneuf, P. H. Tetrahedron 1996, 52, 5495-5504. (c)
Hsung, R. P.; Chidsey, C. E. D.; Sita, L. R. Organometallics 1995, 14,
4808-4815.
(13) (a) Simard, M.; Su, D.; Wuest, J. D. J. Am. Chem. Soc. 1991, 113,
4696-4698. (b) Su, D.; Menger, F. M. Tetrahedron Lett. 1997, 38, 1485-
1488.
(8) (a) Frenkel, D. In Liquids, Freezing and Glass Transition; Hansen,
J. P., Levesque, D., Zinn-Justin, J., Eds.; Elsevier: North-Holland, Am-
sterdam, 1991. (b) Blaak, R.; Mulder, B. M. Mol. Phys. 1998, 94, 401-
405. (c) Blaak, R.; Mulder, B. M. Phys. ReV. E 1998, 58, 5873-5884.

Crosslike Molecules for Solubility ImproVement
J. Am. Chem. Soc., Vol. 123, No. 34, 2001 8179
Scheme 1
Scheme 2a
deprotected species (nH); (ii) Palladium(0)-mediated coupling
of nH with 3 to give (n + 1)P (Scheme 3). Eventually,
palladium(0)-mediated coupling of the deprotected rods nH (n
) 1, 2, 3) with tetrakis[4-(iodo)phenyl]methane (8)13,16 or with
iodobenzene (9), respectively, yielded the final crosses nC and
the corresponding model compounds nR (n ) 1, 2, 3; Schemes
1 and 4). All of the molecules in the nR series and 1C easily
gave crystals, whereas 2C and 3C were only obtained as solid
powders. The final targets were characterized by 1H NMR and
13C NMR, microanalyses, and mass spectrometry. In the case
of 2C and 3C, 1H NMR and microanalyses revealed the presence
of residual solvent despite extensive drying in a high vacuum
(See Figure 1S).
a
(a) Pd(PPh3)4 (2.5 mol %), CuI (1 mol %), piperidine, 0 °C,
trimethylsilylacetylene (1.2 equiv), rt, 3 h; (b) 6 M HCl, NaNO2, 0 °C,
KI; (c) SOCl2, EtOH.
of previous investigations, the latter can be easily incorporated
in synthetic sequences involving palladium coupling. Additional
constraints related to purifications and characterizations led to
choosing the crosses nC and models nR (n ) 1, 2, 3) (pictured
in Scheme 1) as targets. 1-Iodo-4-(trimethylsilylethynyl)benzene
(3) plays a central role in the present synthesis since it is
repetitively used for extending the rods (Scheme 2). It was
prepared in two steps according to classical procedures.14
Palladium(0)-mediated coupling of 4-iodoaniline (1) with tri-
methylsilylacetylene yielded 4-(trimethylsilylethynyl)aniline (2)
that was subsequently transformed into 3 by iodination of the
diazonium salt. The rods were built from the diester terminal
ring. The protected alkyne 1P was easily prepared from
5-(amino)isophthalic acid (4) in three steps with good yields
(Scheme 2). Esterification of 4 gave diethyl-5-(amino)isoph-
thaloate (5)15 that was transformed into diethyl-5-(iodo)-
isophthaloate (6) by diazotation followed by iodination. Palladium-
(0)-mediated coupling of 6 with trimethylsilylacetylene affoarded
1P. The protected rods (nP) were prepared in excellent yields
upon repeating an extension sequence: (i) removal of the
trimethylsilyl-protecting group in (nP) (n ) 1, 2) by cesium
carbonate in ethanol/dichloromethane to give the corresponding
Compared Absorption and Steady-State Emission Proper-
ties of nC and nR. UV absorption and fluorescence emission
properties have been investigated at room temperature in
dichloromethane to subsequently determine the nC and nR
solubilities. Figure 2 and Table 1 summarize the results. All of
the species strongly absorb in the 300-350 nm range. Except
for the smallest species 1C and 1R, nC/nR absorption bands
compare well in the investigated wavelength range. The
absorption bands of 2C and 3C are only slightly red-shifted
with regard to the corresponding band in 2R and 3R. After
normalization to the same number of chromophores, the molar
absorption coefficients at the wavelength of maximal absorption
are similar in the nC and nR series (n ) 2, 3). In contrast, the
observed 1C absorption band is significantly altered with regard
to that of 1R; 1C UV absorption is red-shifted and more intense.
These results are in agreement with the expected trend based
on varying the distance between interacting chromophores.17
In the present system, the increase of the interchromophoric
distance when n is increased in the nC series is not counterbal-
anced by the corresponding increase of the chromophore
oscillator strength. Consequently, alterations of the absorption
(16) (a) Merkushev, E. B.; Simakhina, N. D.; Koveshnikova, G. M.
Synthesis 1980, 486-487. (b) Neugebauer, F. A.; Fisher, H.; Bernhardt, R.
Chem. Ber. 1976, 109, 2389-2394.
(17) Cantor, C. R.; Schimmel, P. R. Biophysical Chemistry; Freeman:
New York, 1980; Part II.
(14) Takahashi, S.; Kuryama, Y.; Sonogashira, K.; Hagihara, N. Synthesis
1980, 627-630.
(15) Hosangadi, B. D.; Dave, R. H. Tetrahedron Lett. 1996, 37, 6375-
6378.

8180 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
Figure 2. Absorption (thick lines) and emission (thin lines) spectra of nR (dotted lines) and nC (solid lines) in methylene chloride at 298 K; (a)
n ) 1, (b) n ) 2, (c) n ) 3. For a purpose of comparison, molar absorption coefficients have been normalized; thus ꢀnC are compared to 4ꢀnR.
Steady-state emission spectra have been normalized at the wavelength of maximal emission. Excitation wavelength: 260 nm for 1R and 1C, 310
nm for the other molecules.
Scheme 3a
a
(a) Cs2CO3 (1.1 equiv), EtOH, CH2Cl2; (b) Pd(PPh3)4 (2 mol %), CuI (1 mol %), 3 (1 equiv), piperidine, rt, 2 h.
Table 2. Thermodynamic Data Associated with Dissolution as
Extracted from the Analysis of the Solubility of the Molecular Rods
(nR) and Crosses (nC) as a Function of Temperature in
Cyclohexane (n ) 1-3) (See Text)
Scheme 4
∆1G°(311 K) ( 5% ∆1H° ( 1b ∆1S° ( 10b
molecule K°(311 K)a ( 5%
(kJ mol-1
)
(kJ mol-1
)
(J K-1 mol-1
)
1R
2R
3R
1C
2C
3C
3.0 10-2
1.1 10-2
2.4 10-4
3.3 10-4
6.2 10-5
3.7 10-5
9.1
11.6
21.7
20.7
25.1
26.4
12
35
44
11
35
52
5
75
70
-35
30
80
a Solubility at 311 K (standard state: ideal solute at 1 M for the
infinitely diluted solution). See Experimental Section. b Assumed to be
constant in the range [290-330 K].
Table 1. Photophysical Features of the Molecular Rods (nR) and
Crosses (nC) in Methylene Chloride at 298 K (n ) 1-3)
molecule
λ
max,abs (nm)a [ꢀ (M-1 cm-1)]
λ
max,em (nm)b [Φ]c
1R
2R
3R
1C
2C
3C
284 [2.3 × 104]
322 [5.4 × 104]
341 [7.3 × 104]
296 [15 × 104]
329 [25 × 104]
346 [27 × 104]
357 [0.15]
356 [0.75]
375 [0.80]
361 [0.25]
367 [0.75]
381 [0.90]
present study except methanol. In particular, the crosses 2C and
3C can be solubilized at concentrations as large as 98% (w/w)
in good solvents such as dioxane! The corresponding solutions
are optically homogeneous and do not scatter light. To
quantitatively analyze the solubility in relation to molecular
structure, a poorer solvent was more suitable, and cyclohexane
was chosen. Table 2 reports the thermodynamic constants K°-
(311 K) of the nC and of the nR species (n ) 1, 2, 3) in
cyclohexane at 311 K, so as the corresponding standard Gibbs
free energy at 311 K ∆1G°(311 K) associated to dissolution. In
both nC and nR series, K°(311 K) is a decreasing function of
the rod length. Nevertheless, the corresponding decrease is less
pronounced in the nC series than in the nR one. The study of
K° in cyclohexane as a function of temperature was used to
extract the standard enthalpy ∆1H° and the standard entropy
∆1S° upon assuming the latter to remain constant in the
temperature range investigated.19 ∆1H° and ∆1S° increase as a
function of the rod length in both nC and nR series.
a Wavelength of maximum absorption in the 250-500 nm range.
b Wavelength of maximum emission in the 300-750 nm range;
excitation wavelength: 260 nm for 1R and 1C, 310 nm for the other
molecules. c Quantum yield of fluorescence at 298 K.
features are more pronounced for 1C than for 2C and for 3C.
nC and nR molecules are also strongly fluorescent in the 350-
500 nm wavelength range. Nevertheless, in contrast to the
absorption features, the steady-state emission spectra of nR and
nC are not significantly different.18
Compared Solubilities of nC and nR. As anticipated, the
crosses nC are soluble in most organic solvents used in the
(18) Bazan, Mukamel, and co-workers recently investigated the photo-
physical properties of a series of compounds related to the present molecules
(Bazan, G. C.; Oldham, W. J., Jr.; Lachicotte, R. J.; Tretiak, S.; Chernyak,
V.; Mukamel, S. J. Am. Chem. Soc. 1998, 120, 9188-9204; Wang, S.;
Bazan, G. C.; Tretiak, S.; Mukamel, S. J. Am. Chem. Soc. 2000, 122, 1289-
1297. See also ref 7d. With regard to their system, the present observation
points out the significance of the core geometry on the photophysical features
of the conjugated rod assembly.
Thermodynamic Data Related to the Fusion Process. For
the present purpose, it was interesting to collect the thermody-
namic features associated with the fusion process for nC and
(19) Lewis, G. N.; Brewer, L. Thermodynamics; McGraw-Hill: New
York, Toronto, London, 1961.

Crosslike Molecules for Solubility ImproVement
J. Am. Chem. Soc., Vol. 123, No. 34, 2001 8181
Figure 3. (a) Dependence of the decimal logarithm of the translational self-diffusion coefficients of the nR (circles) and the nC (squares) molecules
on their characteristic length L at 2 mM in dioxane-d8 at 298 K (n ) 1-3) (see text). The dotted straight line with a slope equal to -1 is displayed
as a guideline for eyes. (b) Dependence of the translational self-diffusion coefficients of 2C (squares), tetratrimethylsilane (circles), the residual
protonated dioxane (triangles) and water (diamonds) as a function of the 2C molar concentration in dioxane-d8 at 298 K. The lines are given as
guidelines for the eyes.
Table 3. Thermodynamic Data Extracted from DSC Measurements
for the Molecular Rods (nR) and Crosses (nC) (n ) 1-3)
Table 4. Translational Self-Diffusion Coefficients Dt Measured by
1H NMR and Characteristic Lengths Rsphere, Lrod, and L Resulting
from Hydrodynamic Analysis and Molecular Modeling for the
Molecular Rods nR and Crosses nC (n ) 1-3)a
∆rH* ( 1
∆rS* ( 10
c
molecule
ttransition (K)
(kJ mol-1
)
(J-1 K mol-1
)
Dt
Rsphere ( 0.05 Lrod ( 0.05
1R
2R
3R
1C
2C
3C
360.5 ( 0.5a
389.5 ( 0.5a
477.5 ( 0.5a
482.5 ( 0.5a
363 ( 5b
38
39
58
54
-
110
100
120
110
-
molecule (10-10 m2 s-1
)
(nm)b
(nm)c
L (nm) ( 0.1d
1R
2R
3R
1C
2C
3C
5.4 ( 0.2
4.7 ( 0.2
2.2 ( 0.1
2.7 ( 0.1
2.6 ( 0.1
0.9 ( 0.1
0.30
0.35
0.75
0.60
0.65
1.80
1.5
2.1
8.00
-
1.4
2.0
2.8
0.9
1.7
2.1
390 ( 5b
-
-
-
a Melting point. b Shoulder. c ∆rS* ) ∆rH*/TFusion
.
-
a Experimental conditions: 2 mM solutions in dioxane-d8 at 298 K.
See text and Experimental Section. b As extracted from eq 12 upon
assuming the molecule to behave as a sphere; see Experimental Section.
c As extracted from eq 13 upon assuming the molecule to behave as a
rigid rod of diameter D ) 0.4 nm (average value based on consideration
of the Hyperchem molecular models); see Experimental Section. d L
corresponds either to the distance from the cross core to the arm
extremity for nC crosses or to the length of the nR rods as evaluated
from the Hyperchem molecular models.
nR. In fact, the ideal solubility of a pure solid compound in a
solvent is related to its standard enthalpy of fusion and to its
fusion temperature.19,20 In addition, some information about the
nature of condensed phases can be derived from calorimetry
measurements. The results extracted from DSC records between
300 and 500 K are given in Table 3. The nR species exhibit a
first-order transition ranging from 360.5 to 477.5 K that can be
interpreted as a fusion. In contrast, only 1C displays a melting
point at 482.5 K among the investigated crosses. In fact, 2C
and 3C thermograms only exhibit small reproducible shoulders
at 363 and 390 K, respectively, that may be tentatively
interpreted as glass transitions. As expected from solubility datas
in cyclohexane, both fusion temperature Tf and standard enthalpy
of fusion ∆rH*(Tf) are increasing functions of n in the nR series.
In contrast, the corresponding standard entropies ∆rS*(Tf) remain
rather similar.
series of experiments, the structure of 2C concentrated solutions
in dioxane-d8 has been more especially investigated to evaluate
the significance of the cross shape on reducing the trend for
crystallization or aggregation. 1H NMR translational self-
diffusion experiments were used to detect the possible formation
of any 2C aggregates upon increasing the solution concentration.
Attention has been paid on the NMR line shape, on the signal
integration and on the translational self-diffusion coefficient of
2C in the 0.2-35 mM range. In fact, aggregation is expected
to decrease at the same time the molecular correlation time and
mobility. During the present study, the residual protonated
dioxane and tetratrimethylsilane (TMS) were used as internal
references. Indeed, both species are small enough to see their
integration remaining essentially unaffected even in the presence
of an intricated network such as pictured in Figure 1. In addition,
the diffusion coefficients of the corresponding species could
be used in principle to probe any percolation phenomenon
associated to viscosity changes or excluded volumes. With
regard to protonated dioxane and TMS, neither significant
change of the 2C NMR line shape nor anomaly of 2C integration
have been ever detected in the investigated range of concentra-
tions (0.2-35 mM). To confirm this observation suggesting the
absence of aggregates even at the largest concentrations, a magic
angle spinning (MAS) 13C NMR spectrum was recorded on the
most 2C concentrated sample. The corresponding sequence is
NMR Investigation of nR and nC Solutions. nR and nC
solutions in dioxane-d8 have been examined by NMR to analyze
the aggregation state and the freedom for molecular motions of
the different solutes as a function of their concentration. In a
1
first set of H NMR translational self-diffusion experiments,
the diffusion coefficients Dt of all the nR and nC species have
been collected at 298 K at 2 mM in dioxane-d8; at such a low
concentration, monomeric species are anticipated so as to give
reference values for the following series of NMR experiments.
Figure 3a and Table 4 summarize the results. The measured
diffusion coefficients are decreasing function of the characteristic
molecular lengths L (rod length in the nR series and arm length
in the nC series). More precisely, Dt roughly linearly depend
on 1/L in both series as suggested from Figure 3a. In a second
(20) Lemarchand, H.; Guyot, F.; Jousset, L.; Jullien, L. Thermodynamique
de la Chimie; Hermann: Paris, 1999.

8182 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
Figure 4. Histograms displaying the number n of samples with a given thickness h, resulting from evaporation of diluted nC DME: water 9:1 (v/v)
solutions on mica surfaces as evaluated by multiple beam interferometry. (a) 1C; (b) 2C; (c) 3C. I, II, and III respectively define the clusters that
have been considered to correspond to films involving mono-, bi-, and trilayers. See text and Experimental Section.
known to reveal any species exhibiting structural features that
forbid detection using liquid-state NMR.21 The absence of any
significant difference of integration ratios between the two NMR
spectra recorded led to the first evidence that no large 2C
aggregates formed even at concentrations as high as 98% (w/
w) in dioxane-d8. Figure 3b displays the evolution of the
Discussion
Thermodynamic Analysis of the Solubility in Relation to
Molecular Structure. A thermodynamic model has to be
developed to examine the true significance of molecular shape
on solubility. In fact, solubility is expected to be governed by
both the geometry and the characteristic lengths of the consid-
ered molecule. Thus, it is necessary to appropriately normalize
lengths to extract the role of molecular structure only. For the
latter purpose, we assume in the present analysis (i) the
molecules nC and nR to be strictly analogous in each series.
Then the nR (respectively nC) molecules are only characterized
by their length L (respectively the length L of the four arms
grafted on the central node); (ii) the standard enthalpy associated
to the dissolution process 1 to only depend on L in each series.24
In the following, ∆1H°(T, M, length unit) and ∆1S°(T, M, length
unit) are respectively defined as the common standard enthalpy
and entropy changes associated to the dissolution process 1 per
length unit in a given series nM (M ) C or R). Upon assuming
(i) and (ii), one obtains:
1
diffusion coefficients obtained during the second series of H
NMR translational self-diffusion experiments. In the investigated
concentration range, the self-diffusion coefficients of all the
species but water remain essentially constant upon varying
2C concentration in dioxane-d8. In fact, the self-diffusion co-
efficient of water strongly increases at the largest 2C concentra-
tions.
Structure of Thin nC Films on Mica Surfaces. In view of
their flatness, mica plates have been chosen to examine whether
it was possible to obtain homogeneous thin films of the nC
crosses on surfaces. In addition, multiple beam interferometry
can be fruitfully used to determine the thickness of films
deposited between two mica sheets at angstro¨m resolution.5c
After preliminary investigation, 9:1 ethylene glycol dimethyl
ether (DME):water (v/v) was selected as the nC organic solvent
for deposition due to its good wetting properties toward mica.
Eventually, 10 µM was chosen for nC concentrations to lead
less than a unimolecular layer after solvent evaporation upon
assuming the formation of a uniform film.22 Under the present
experimental conditions, evaporation of 5 µL of 10 µM nC
solution in DME-water (9/1 (v/v)) deposited on mica did not
give a homogeneous film at the scale of the whole spread drop.
In particular, the fringe observation showed the presence of thick
films in the zones corresponding to the drop rim.23 In contrast,
the observation of bumps in the fringes in the middle of the
evaporated drop revealed the presence of very thin islands of
diameter lying in the micrometer range. Figure 4, a-c, displays
the histograms picturing the number of nC islands of a given
thickness as measured by multiple beam interferometry. The
major feature of these histograms is the existence of rather
regularly spaced clusters ranging at multiples of molecular
heights.
∆1H°(T, nM) ) iL∆1H°(T, M, length unit)
(4)
with i ) 1 when M ) R, and i ) 4 when M ) C. Then, one
easily derives:
∆1H°(T, nM) ) n∆1H°(T, 1M)
(5)
Figure 5a displays the derived experimental relationship
between ∆1H°(311 K, nM) and n. Although being based on
three points, satisfactory linear fits of the datas were obtained:
∆1H°(311 K, nR) (kJ mol-1) ) -8.6 + 20.5n and ∆1H°(311
K, nC) (kJ mol-1) ) -1.7 + 16.0n. In contrast to the strict
expectations from the relation 5, one notices in both cases the
non-zero intercept with the y-axis so as the difference between
(24) Assumption (i) could be somehow questioned in view of the presence
of the ethyl ester substituents on the terminal phenyl groups, and of the
central node in the nC series. Nevertheless, it is expected to become
satisfactory when n increases. Along the assumption (ii), the nM standard
molar enthalpies in both liquid and solid phases, h°(T, nM, solute) and
h°(T, nM, solid), are expected to only depend on L in each series (M ) C
or R). Such an assumption is natural for the former phase. For a short-
range interaction such as van der Waals at this scale (see ref 5c), the
solvent-solute interaction is governed by the surface contact scaling as
LD where D designates the rod diameter. In contrast, the same assumption
could be questioned for the solid phase. In fact, it implies that the structure
of the solid does not considerably evolve upon changing n within each
series. The common elongated shape of the nR molecules together with
the expected large compacity of their solid phase (the pure nR solids are
crystals; vide supra) make the assumption reasonable in the nR series. In
contrast, the same assumption could a priori appear more doubtfull for the
nC molecules (see the differences of the DSC results in the nC series;
vide supra) but will be ultimately validated by the experimental results (vide
infra).
(21) Virlet, J. Line Narrowing Methods in Solids. In Encyclopedia of
Magnetic Resonance; Grant, D. M., Harris, R. K., Eds,; Wiley: New York,
1996; p 2694. On a natural abundance carbon-13 sample, the use of MAS
combined with high-power proton decoupling enables one to observe the
signal due to any species in a solid or almost solid phase. A direct 13C
excitation has also been used instead of a cross polarization (CP)-MAS
enhancement transfer to detect the molecules in solution.
(22) The latter estimate is obtained by considering the formation of a
disc of 17 mm diameter on mica from evaporation of a 5 µL drop of 10
µM solution (see Experimental Section). Then the average surface per
molecule is about 80 nm2, whereas the evaluation from molecular models
is 10 nm2/molecule at the most (see Table 5).
(23) The present morphology probably results from mass and heat
transfers occurring during solvent evaporation.

Crosslike Molecules for Solubility ImproVement
J. Am. Chem. Soc., Vol. 123, No. 34, 2001 8183
is introduced with R, the gas constant, and log the decimal
logarithm, to facilitate the linear analysis when plotting log-
([nM]eq)(n). Eventually it becomes now possible to extract the
significance of the molecular geometry from the datas collected
in the nC and nR series upon using the appropriate length
normalization. For the present purpose, the rod nRef of length
4L is introduced as the reference compound for the cross nC
with arm length L. Thus, the corresponding standard enthalpy
change associated to the dissolution process 1 and the nRef
solubility features are obtained from the preceding results
derived in the nR series. Thus, one has:
Figure 5. Relationships between thermodynamic datas associated to
the dissolution process 1 for the nM molecules in cyclohexane at 311
K (M ) R, circles; M ) C, squares). (a) relationships between ∆1H°-
(311 K, nM) and n (filled markers), and between ∆1S°(311 K, nM)
and ∆1H°(311 K, nM) (empty markers). The thin and thick lines
respectively picture the best linear fits of the experimental points ∆1H°-
(311 K, nM)(n) and ∆1S°(311 K, nM)[∆1H°(311 K, nM)] (M ) R,
dotted line; M ) C, solid line). Note that ∆1H°(311 K, 2C) ) ∆1H°-
(311 K, 2R). (b) Evolution of the decimal logarithm of the nM
solubilities in cyclohexane at 311 K. Markers: experimental points;
dotted lines: best linear fits of the experimental points; solid line:
nRef expected behavior according to eq 10.
∆1H°(T, nRef) ) 4∆1H°(T, nR)
log([nRef]eq) )
(9)
4n log{R(T, 1R)[1R]eq} - log R(T, 1R) (10)
The latter equations allow to interpret the evolution of the
experimental solubilities displayed in Figure 5b. The linear
dependence of log([nM]eq) on n expected from eq 7 is
satisfactorily verified in both nM series. Moreover, it can be
used to directly extract R(T, 1M) with more reliability than when
using eqs 6a and 8, and the â(T, 1M) value derived from the
analysis of the preceding extrathermodynamic relationship.
Hence, one finds log R(T, 1R) ) 0.3 and log R(T, 1C) ) 3.1.
The experimentally observed decrease of the nM solubility when
n rises can also be interpreted from eqs 7 and 8. In fact, the
â(T, M) values derived from analysis of Figure 5a are such
that {R(T, 1M)[1M]eq} < 1 around room temperature so as to
make [nM]eq a decreasing function of n from eq 7.
The comparison between the datas collected in the nR and
the nC series is now performed. Equation 9 is first discussed
in relation with Figure 5a showing that ∆1H°(T, nRef) . ∆1H°-
(T, nC) ≈ ∆1H°(T, nR). Since h°(T, nC, solute) ≈ h°(T, nRef,
solute) is anticipated,25 one deduces h°(T, nC, solid) . h°(T,
nRef, solid) which suggests that molecular interactions are much
stronger in the nRef solid phase than in the nC one. Such an
analysis is in line with the expectations based on molecular
structures. In the solid state, the nR molecules are enough
“compact” to give crystals whereas the more open structures
of the largest crosses forbid crystal formation and lead to glasses
such as pictured in Figure 1c. In the latter case, significant
molecular interactions would only occur at the level of the few
intermolecular contact points leading to low values of h°(T, nC,
solid). Sample observation, microanalyses and DSC results are
in line with the latter conclusion. All of the nR species are
crystalline, give correct microanalyses, and exhibit a melting
point. In contrast, 2C and 3C samples are powders containing
some residual solvent and do not exhibit any first-order transition
in the DSC scans. Equation 10 is now addressed. Figure 5b
compares the solubility of nRef and nC at 311 K in cyclohex-
ane. [nC]eq always exceeds [nRef]eq by several orders of
magnitude. In fact, [nC]eq appears only poorly sensitive to n,
whereas [nRef]eq is a decreasing function of n. Thus 3C is
already more than 108 times more soluble than nRef in
cyclohexane under ambient conditions, and the trend is antici-
pated to be more pronounced for larger n values! The latter
observation is most conveniently analyzed in relation with eq
the slopes snM of the ∆1H°(311 K, nM)(n) linear fits, and
∆1H°(311 K, 1M) (see Table 2). In fact, end-groups (ethyl ester
groups and central core in the nC series) probably weaken the
van der Waals interaction involving the conjugated parts in the
smallest terms of the series,24 so as to introduce some curvature
in the ∆1H°(311 K, nM)(n) dependence. As anticipated by
neglecting the negative term of second order when performing
the linear fit, one thus coherently observes: (i) negative intercepts
of the linear fits with the y-axis. Nevertheless, these intercepts
can be considered as small (using the slope snM of the ∆1H°-
(311 K, nM)(n) for scaling: 8.6 , snR ) 20.5 in the R series
and 1.7 , snC ) 16.0 in the C series); (ii) snR > ∆1H°(311
K, 1R) and snC > ∆1H°(311 K, 1C). Despite the latter small
deviations, only the linear relation 5 was retained in the
following to facilitate the subsequent calculations conceived to
derive orders of magnitude. To progress further in the theoretical
treatment of experimental datas, some extrathermodynamic
relationship was sought for between the standard enthalpy and
entropy changes associated to the dissolution process 1 per
length unit in each series nM (M ) C or R). Such relations
are common in organic chemistry.6b Figure 5a does exhibit some
reasonable trend for a linear relationship between ∆1S°(T, nM)
and ∆1H°(T, nM) in each series. Introducing the slope â(T, M)
(â(T, R) ) 2.3 10-3 K-1 and â(T, C) ) 2.8 10-3 K-1), one
writes:
∆1S°(T, nM) - ∆1S°(T, 1M) ) â(T, M)[∆1H°(T, nM) -
∆1H°(T, 1M)] (6a)
or similarly:
∆1S°(T, M, length unit) ) â(T, M)∆1H°(T, M, length unit)
(6b)
From eqs 5 and 6a, it is then possible to derive the dependence
of the solubility [nM]eq of the nM species on n:
log([nM]eq) ) n log{R(T, 1M)[1M]eq} - log R(T, 1M) (7)
where the parameter R defined as:
(25) As explained in ref 24, h°(T, nM, solute) is reasonably expected to
depend only on the characteristic length L. In addition, h°(T, nC, solute,
length unit) is anticipated to marginally depend on the presence of the central
node since the small cyclohexane molecules can easily access to the whole
arm length. Thus, one can reasonably assume h°(T, nC, solute, length unit)
≈ h°(T, nR, solute, length unit) so as to eventually write h°(T, nC, solute)
≈ h°(T, nRef, solute).
RT log{R(T, 1M)[1M]eq} ) [â(T, M)T - 1]∆1H°(T, 1M)
(8)

8184 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
6b by noticing that â(T, R) < â(T, C); for the same standard
entropy change associated to the dissolution process 1 per length
unit, the corresponding standard enthalpy change is lower in
the nC series than in the nR one. Hence, the favorable enthalpic
term is not totally counterbalanced by an unfavorable entropic
one in the nC series, and the crosses are comparatively much
more soluble than the rods.
Thus, the analysis of thermodynamic datas confirms that a
suitable molecular shape can considerably improve molecular
solubility; decreasing the lattice energy of a solid phase is indeed
an efficient strategy to promote dissolution, even in poor solvents
such as cyclohexane.
coefficients of all the species but water underlines that no major
modification of molecular mobility occurs upon concentrating;
the solution seems to remain in a diluted regime even at the
98% (w/w) concentration. To evaluate the latter result, the
average distance r between two 2C molecules was derived
upon considering that the solute is distributed over a regular
cubic lattice. At the 98% (w/w) concentration, r is about 3.6
nm which is only twice the length of the lateral 2C arms (see
Table 4). Under such conditions, the observation of a diluted
regime may appear surprising for 2C but probably reveals some
intrisic geometrical feature of the open and anisotropic shape
of the 2C cross. The diffusion behavior of water in concentrated
2C solutions is singular; its self-diffusion coefficient consider-
ably increases at the largest concentrations. The solvent effect
associated to the change of water environment at the largest
2C concentrations cannot account for such a major increase. In
fact, the viscosity of an aromatic solvent such as benzene or
toluene is about half that of dioxane at room temperature. Thus
a solvent effect only could increase the diffusion coefficient of
water by a factor 2 at the most (see eq 12). In view of the value
of the self-diffusion coefficient of water at low 2C concentration
which is in reasonable agreement with a monomeric state, no
aggregate breaking can be invoked to explain the mobility rise
of the water molecules. Upon considering hydrophobic forces,
we tentatively propose the translational motion of the water
molecules to be restricted in a percolated medium of low
dimension in the 2C network at the largest concentrations. Then
the usual analysis based on a free motion occurring in three
dimensions would lead to too large an apparent self-diffusion
coefficient.
Structure of Solutions of the nC Crosses. The analyses of
the integration or line shape of the signals obtained during the
1H NMR experiments showed that large 2C aggregates did not
form in dioxane-d8, even at the highest concentration investi-
gated (33 mM; 98% (w/w)). In addition, 1H NMR translational
self-diffusion experiments emphasized no significant evolution
of the 2C self-diffusion coefficient to take place in the con-
centration range 0.2-35 mM. To prove that the monomeric form
was the major species present in 2C solutions even at the largest
concentrations, the experimental self-diffusion coefficients of
the nR and nC molecules were analyzed with a hydrodynamic
model. Upon assuming the molecules to behave as rigid bodies
of given shapes in a continuous medium, Stokes-Einstein
relations can be used to derive molecular dimensions from the
self-diffusion coefficients and the solvent viscosity.17 Empirical
relations are available for the right cylinder geometry that is
appropriate for describing the nR molecules (see Experimental
Section). In contrast, no such relation seems to be available for
tetrahedral stars. In a purpose of comparison, the crosses were
consequently modeled as spheres since a simple relation is
available for the latter geometry (see Experimental Section).
Table 4 summarizes the results of the present hydrodynamic
analysis for 2 mM solutions of nR and nC. In both series of
molecules, the agreement is satisfactory as demonstrated by (i)
the experimentally observed linear dependence of Dt on 1/L
conforming to eqs 12 and 13; (ii) the comparison between the
extracted characteristic lengths and the monomer dimensions.26
In connection to the absence of change of the 2C self-diffusion
coefficient in the 0.2-35 mM concentration range, this analy-
sis: (i) suggests that the solutions of crosses contain only
monomeric species even at the largest concentrations. Such a
conclusion is in line with the low trend for crystallization
exhibited by the largest nC molecules; (ii) validates a posteriori
the interpretation of the measured diffusion coefficients in the
second series of NMR experiments. Indeed it was legitimate to
analyze the latter at the molecular level without involving
differently moving species in a regime of fast exchange.
Thus 1H NMR experiments emphasized that both monomeric
states and translational degrees of freedom are essentially kept
even in very concentrated solutions of crosses. From a practical
point of view, the latter features are distinct advantages to obtain
homogeneous thin films resulting from the spreading and the
evaporation of solutions as it will be demonstrated in the next
paragraph.
Structure of Thin Films Made of nC Crosses. To conclude
on the structure of the films resulting from evaporation of diluted
nC solutions on surfaces, two major issues have to be addressed
from the interferometry experiments performed on films: (i)
the presence of discrete clusters in the histograms obtained from
data analysis of very thin nC films; (ii) the characteristic heights
associated to the average values over each observed cluster
(Figure 4).
The first observation is reminiscent of the oscillatory force
that is observed when molecules with not too irregular shapes
are confined between two hard walls.5c In fact, the high
symmetry of the nC crosses is especially favorable to observe
such a phenomenon; only the orientation making contact of three
nC arm ends with the mica surface is expected at the same
time (i) to minimize molecular height, (ii) to maximize nC:
mica interaction. By analogy with the explanation given for
spherical molecules or linear alkanes, such a behavior may be
reasonably interpreted as resulting from some periodical order
whose repeat distance is closely related to the molecular size.
An organized structure such as displayed in Figure 1b could be
prone to exhibit such a feature. Then the presence of discrete
clusters in Figure 4 could be considered as a clue supporting
the theoretically predicted cubatic order at large nC concentra-
tions.8
1
In the second series of H NMR translational self-diffusion
experiments, the absence of evolution of the self-diffusion
(26) At a better level of analysis, one notices that the extracted length
rod in the nR series is much too large for 3R, whereas it is much satisfactory
L
for 1R and 2R. This discrepancy is tentatively interpreted upon considering
that the characteristic molecular length of 3R cannot be neglected in front
of the average intermolecular distance at 2 mM which lies around 0.9 nm.
The solution cannot be anymore considered as diluted. Then two conse-
quences are anticipated: (i) a possible drop of the self-diffusion coefficient
due to collisions between 3R molecules; (ii) some nematic order among
the rodlike molecules leading to a dimensional restriction of the 3R molecule
translational motion. Under such conditions a data treatment based on free
motion in three dimensions will underestimate the self-diffusion coefficient
by a factor equal to d/3 where d represents the actual dimension in which
molecular motion is restricted. The same analysis cannot be easily performed
in the nC series. Nevertheless, the sphere model applies surprisingly rather
well to describe the hydrodynamic of the crosses upon considering the
similarity between the extracted radii and the arm lengths L.
Along with the preceding interpretation, Table 5 and Figure
6 display the average film thicknesses h(i) extracted from the
histograms of Figure 4 by assuming the clusters as respectively

Crosslike Molecules for Solubility ImproVement
J. Am. Chem. Soc., Vol. 123, No. 34, 2001 8185
taking 5 kJ mol-1 per carbon atom as a reasonable order of
magnitude.5c,28,29 The comparable estimates obtained in making
these evaluations suggest that rather large molecular deforma-
tions could be envisaged to take place in the present system. In
particular one could imagine the nC molecular tetrahedrons to
tend to adopt the shape of a trigonal pyramide in order to
maximize their interaction with the mica surfaces. In such a
case, the molecular heights should be effectively close to the
arm lengths as experimentally observed.
Consequently, the present investigation demonstrates that nC
molecular crosses do lead to the formation of homogeneous thin
films without building crystals at the micrometer scale. In
contrast to nanoporous solids involving noncovalent interac-
tions,10 solvent removal does not promote the internal collapse
of the present nanoporous structure. In addition, this study also
suggests that some shape alteration of even the most “rigid”
organic molecules could occur close to surfaces due van der
Waals dispersive forces. In relation with the proposed strategy
for favoring dissolution, the latter observation more generally
emphasizes on the significance of the rigidity of the molecular
motif that would be responsible of a low packing in the solid
state. Indeed attractive forces are hudge at the molecular level
so as to deform molecules to fill in any vacuum if they are not
rigid enough.
Table 5. Theoretical Molecular Height and Surface, and
Experimental Thicknesses of Layers of the Molecular Crosses nC (n
) 1-3) as Evaluated from Molecular Models and Interferometry
Experiments
hTh (1) ( 0.1
S
h (1) ( 0.1
Sn
b
molecule
(nm)a
(nm2 molecule-1
)
(nm)c
(nm/layer)d
1C
2C
3C
1.2
2.3
2.8
2
7
10
1.1
1.7
2.0
0.9
1.7
2.1
a As evaluated with the Hyperchem software from hTh ) 4L/3 where
L represents the total length of the cross arm (see text). b As evaluated
from S ) 4 cos(π/6) sin2(θ/2)L2 with θ ) Arccos(-1/3). c As
experimentally determined from the average position of the data cluster
at the lowest values of heights. d Slope extracted from the linear fit of
the h(i) experimental data.
Conclusions
The present study on rigid tetrahedral molecules considered
as models of Onsager molecular crosses demonstrates that
suitably designing molecular shape to avoid compact packing
in the solid state is an efficient strategy to bypass crystallization
and to promote dissolution of organic compounds in many
solvents. Even in the absence of solubilizing substituents such
as lateral chains or groups strongly interacting with the solvent,
it has been possible to obtain homogeneous concentrated
solution of monomeric rigid species that are traditionnally more
prone to insolubility. In addition to the latter strong solubility
improvement, an extremely concave molecular architecture was
also shown to bring distinct advantages such as low viscosity
and low trend for crystallization to obtain homogeneous pure
nanoporous organic glasses after solvent removal even in
ultrathin films. Despite this battery of attractive features, the
present report also points on molecular rigidity as a limit that
could be significant for several applications. In fact, even the
traditionally envisioned most rigid organic backbones remain
rather soft, and strong molecular deformation can be anticipated
when such concave rigid molecules are located close to surfaces.
Figure 6. Evolution of the thickness h(i) (empty markers) and of the
ratio h(i)/hTh(i) (filled markers) of thin films measured by interferometry
as a function of the number i of nC layers in the film. (Diamonds) 1C;
(squares) 2C; (circles) 3C. See text.
corresponding to mono-, bi, and trilayers of nC molecules (i )
1-3). Whatever the nC cross, the h(i) values are linearly related
to the number i of layers in the film. The latter observation
probably indicates that the film density perpendicularly to the
mica surface is reasonably independent of i for i ) 1-3. In
addition, both h(1) and the slope Sn of the straight line h(i) fitting
as a function of i are significantly lower than the molecular
height evaluated from molecular models. A first explanation
could be some molecular entanglement taking place in the nC
solid phase.10 Nevertheless, it is hard to imagine this phenom-
enon to already occur for i ) 1. Another explanation was
envisaged from the favorable comparison between Sn and the
arm lengths L of the nC crosses. To address the latter fact, we
derived some orders of magnitude of energy to evaluate whether
some molecular deformation could take place under the
experimental conditions used during the interferometry experi-
ments. In a first step, typical estimates of bending energies Wb
) kδθNAvogadro associated to 10° bends were sought for. Upon
assuming the bending motion to occur at the C-CtC level,
one obtains Wb ≈ 30 kJ mol-1 whereas the corresponding value
is about 100 kJ mol-1 if the bending takes place at the Arm-
C-Arm central node of the tetraphenylmethane core.27 In a
second step, we evaluated the energies associated to possible
van der Waals interactions occurring in the system with special
emphasis on the dispersion interaction between a cross arm and
the mica surface at contact. Hence, one obtains for the latter
more than 100 kJ mol-1 for every members of the nC series by
Experimental Section
Synthesis. General Procedures. Microanalyses were performed by
the Service de Microanalyses de l′Universite´ Pierre et Marie Curie
(Paris). Melting points were determined with a Bu¨chi 510 capillary
apparatus or using DSC 7 Perkin-Elmer. 1H and 13C NMR spectra were
(28) The present value is probably underestimated. In fact, the corre-
sponding value is about 7 kJ mol-1 in the alkane series that exhibits a much
less pronounced polarizability. See ref 5c.
(29) Other sources of van der Waals interaction are much less important.
The energy WvdW associated to the van der Waals attractive interaction
between the mica sheets can be evaluated from the expression WvdW
)
-A/(12πh2) where A designates the Hamaker constant of the system, and
h, the distance between the two mica sheets.5c Taking A ) 10-20 J as a
reasonable estimate for the Hamaker constant of a mica/organic medium/
mica system,5c one obtains WvdW ≈ 0.3 mJm-2 for h ) 1 nm and WvdW
≈
(27) Respectively using bending constants kδ ≈ 3 ×10-19 N mrad-1 and
kδ ≈ 10-18 N mrad-1. See Herzberg, G. Molecular Spectra and Molecular
Structure II. Infrared and Raman Spectra of Polyatomic Molecules; Krieger
Publishing Company: Malabar, Florida, Original edition 1945: Reprint
edition 1991.
6 µJm-2 for h ) 7 nm. Thus, the pressure exerted by the mica surfaces on
the thin films of nC can be neglected. Similarly, the direct attractive van
der Waals interaction between cross arms in the solid state is expected to
be rather small since molecular contact can occur only at a few sites due to
the low compacity.

8186 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
recorded at room temperature on a Bruker AM 200 SY, an Avance
Bruker DRX 400 or an Avance Bruker DRX 600 spectrometers;
chemical shifts are reported in ppm with protonated solvent as internal
reference (1H, CHCl3 in CDCl3 7.27 ppm, protonated dioxane in
dioxane-d8 3.43 ppm; 13C, 13CDCl3 in CDCl3 77.0 ppm); coupling
constants J are given in Hz. Mass spectra (chemical ionization with
NH3 or FAB positive) were performed by the Service de Spectrome´trie
de Masse de l’ENS (Paris). Column chromatography was performed
on silica gel 60 (0.040-0.063 mm) Merck. Analytical or preparative
thin-layer chromatography (TLC) was conducted on Merck silica gel
60 F254 precoated plates. Commercially available reagents were used
as obtained.
mmol, 1.2 equiv); 4 (3.7 g, 10.7 mmol, 1 equiv); tetrakis(triphenylphos-
phine)palladium(0) (308 mg, 0.27 mmol, 2.5 mol %); copper iodide
(20 mg, 0.11 mmol, 1 mol %); piperidine (20 mL). Column chroma-
tography on silica gel with dichloromethane as eluent to give 1P as a
pale yellow solid (3.0 g, 88% yield): mp 66.2 °C; 1H NMR (200 MHz,
CDCl3): δ ) 8.60 (t, J ) 1.7 Hz, 1H), 8.28 (d, J ) 1.7 Hz, 2H), 4.40
(q, J ) 7.1 Hz, 4H), 1.41 (t, J ) 7.1 Hz, 6H), 0.27 (s, 9H); 13C NMR
(50 MHz, CDCl3): δ ) 164.9, 136.5, 130.9, 130.0, 123.8, 102.7, 96.3,
61.3, 14.1, -0.4; anal. calcd for C17H22O4Si (318.44): C 64.12, H 6.96;
found: C 64.13, H 6.87.
Diethyl-5-(ethynyl)isophthaloate (1H). Cesium carbonate (3.4 g,
10 mmol, 1.1 equiv) was added to a solution of 1P (3.0 g, 9.4 mmol,
1 equiv) dissolved in an ethanol/dichloromethane mixture (10 mL/10
mL). The solution was vigorously stirred at room temperature for 1 h.
After removal of the solvents under reduced pressure, the residue was
dissolved in a water/dichloromethane mixture. The aqueous layer was
extracted twice with dichloromethane, and the combined organic layers
were dried over magnesium sulfate. After solvent evaporation, the crude
residue was purified by column chromatography on silica gel with
dichloromethane as eluent to yield 1H as pale yellow crystals (2.15 g;
4-(Trimethylsilylethynyl)aniline (2).14 Trimethylsilylacetylene (3.9
mL, 27.4 mmol, 1.2 equiv) was added to a solution of 4-iodoaniline
(1) (5.0 g, 23 mmol, 1 equiv) in deoxygenated distilled piperidine (20
mL) with tetrakis(triphenylphosphine)palladium(0) (658 mg, 0.57 mmol,
2.5 mol %) and copper iodide (43 mg, 0.23 mmol, 1 mol %) at 0 °C.
After stirring for 3 h at room temperature, the reaction was quenched
with a saturated aqueous solution of ammonium chloride, and dichlo-
romethane was added. The aqueous layer was extracted twice with
dichloromethane. The combined organic layers were washed with brine
and dried over sodium sulfate. After evaporation of the solvent, the
brown crude residue was purified by column chromatography on silica
gel with dichloromethane/cyclohexane (70/30) as eluent to give 2 as a
pale yellow solid (3.7 g, 85% yield): mp 93.5 °C (methanol, lit.14 95-
1
93% yield): mp 90.5 °C; H NMR (200 MHz, CDCl3): δ ) 8.60 (t,
J ) 1.6 Hz, 1H), 8.28 (d, J ) 1.6 Hz, 2H), 4.38 (q, J ) 7.1 Hz, 4H),
3.18 (s, 1H), 1.39 (t, J ) 7.1 Hz, 6H); 13C NMR (50 MHz, CDCl3):
δ ) 164.8, 136.7, 131.1, 130.4, 122.8, 81.5, 78.9, 61.4, 14.1; anal.
calcd for C14H14O4 (246.26): C 68.28, H 5.73; found: C 68.16, H 5.72.
Diethyl-5-[(4-[trimethylsilylethynyl]phenyl)ethynyl]-
isophthaloate (2P). The same procedure was used as for 2 except that
copper iodide was added at last. Copper iodide (10 mg, 0.05 mmol, 1
mol %); 1H (1.3 g, 5.3 mmol, 1 equiv); 3 (1.6 g, 5.3 mmol, 1 equiv);
tetrakis(triphenylphosphine)palladium(0) (122 mg, 0.1 mmol, 2 mol
%); piperidine (20 mL). Column chromatography on silica gel with
dichloromethane as eluent to give 2P as a pale yellow solid (1.94 g,
88% yield): mp 74.2 °C (ethanol); 1H NMR (200 MHz, CDCl3): δ )
8.60 (t, J ) 1.7 Hz, 1H), 8.28 (d, J ) 1.7 Hz, 2H), 7.47 (s, 4H), 4.42
(q, J ) 7.1 Hz, 4H), 1.42 (t, J ) 7.1 Hz, 6H), 0.26 (s, 9H); 13C NMR
(50 MHz, CDCl3): δ ) 164.9, 136.2, 131.8, 131.4, 131.1, 130.0, 123.8,
123.4, 122.4, 104.3, 96.5, 90.5, 89.1, 61.4, 14.1, -0.3; anal. calcd for
C25H26O4Si (418.56): C 71.74, H 6.26; found C 71.82, H 6.44.
Diethyl-5-[(4-ethynylphenyl)ethynyl]isophthaloate (2H). The same
procedure was used as for 1H. 2P (1.90 g, 4.54 mmol); cesium
carbonate (1.63 g, 5 mmol, 1.1 equiv); ethanol/dichloromethane (1/1;
10 mL). Column chromatography on silica gel with dichloromethane
as eluent to yield 2H as pale yellow crystals (1.35 g; 87% yield): mp
1
96 °C); H NMR (200 MHz, CDCl3): δ ) 7.28 (apparent d, J ) 8.6
Hz, 2H), 6.55 (apparent d, J ) 8.6 Hz, 2H), 3.82 (bs, 2H), 0.27 (s,
9H); 13C NMR (50 MHz, CDCl3): δ ) 146.7, 133.1, 114.3, 112.1,
106.0, 91.2, 0.0.
1-Iodo-4-(trimethylsilylethynyl)benzene (3).14 A solution of sodium
nitrite (870 mg, 12.7 mmol, 1.2 equiv) in 2 mL of water was added to
a solution of 2 (2.0 g, 10.6 mmol, 1 equiv) in 170 mL of hydrochloric
acid (6 M) cooled below 5 °C. After stirring for 45 min at 0-5 °C, an
ice-cold solution of potassium iodide (2.6 g, 16 mmol, 1.5 equiv) in
10 mL of water was poured dropwise, and 100 mL of dichloromethane
was added. The resulting mixture was allowed to reach room temper-
ature and was stirred for 4 h. The aqueous layer was extracted twice
with dichloromethane, and the combined organic layers were washed
with brine and dried over magnesium sulfate. After evaporation of the
solvent, the crude residue was purified by column chromatography on
silica gel with cyclohexane as eluent to give 3 as a pale yellow solid
(2.1 g, 65% yield): mp 52.4 °C (methanol; lit.12c 56-58 °C); 1H NMR
(200 MHz, CDCl3): δ ) 7.63 (apparent d, J ) 8.6 Hz, 2H), 7.18
(apparent d, J ) 8.6 Hz, 2H), 0.24 (s, 9H); 13C NMR (50 MHz,
CDCl3): δ ) 137.2, 133.3, 122.6, 103.9, 95.7, 94.4, -0.2.
1
87.8 °C; H NMR (200 MHz, CDCl3): δ ) 8.61 (t, J ) 1.7 Hz, 1H),
8.33 (d, J ) 1.7 Hz, 2H), 7.48 (s, 4H), 4.41 (q, J ) 7.1 Hz, 4H), 3.20
(s, 1H), 1.42 (t, J ) 7.1 Hz, 6H); 13C NMR (50 MHz, CDCl3): δ )
164.9, 136.1, 131.9, 131.4, 131.1, 130.0, 123.7, 122.7, 122.3, 90.3,
89.2, 82.9, 79.1, 61.4, 14.1; anal. calcd for C22H18O4 (346.38): C 76.29,
H 5.24; found: C 76.21, H 5.32.
Diethyl-5-(amino)isophthaloate (5).15 Thionyl chloride (12 mL, 165
mmol, 3 equiv) was added dropwise at 0 °C to a stirred solution of
5-(amino)isophthalic acid (4) (10 g, 55 mmol) in 100 mL of absolute
ethanol. After stirring under reflux for 5 h, ethanol was distilled out.
The crude residue was dissolved in ethyl acetate, and the solution was
washed with a saturated aqueous solution of sodium bicarbonate. After
drying over sodium sulfate, the solvent was evaporated to give the
diester 5 as a white solid (11.6 g, 89%): mp 117.6 °C (ethanol, lit.30
118 °C); 1H NMR (200 MHz, CDCl3): δ ) 8.06 (t, J ) 1.4 Hz, 1H),
7.53 (d, J ) 1.4 Hz, 2H), 4.37 (q, J ) 7.1 Hz, 4H), 3.82 (bs, 2H), 1.39
(t, J ) 7.1 Hz, 6H); 13C NMR (50 MHz, CDCl3): δ ) 166.0, 146.5,
131.7, 120.5, 119.6, 61.3, 14.2.
Diethyl-5-[(4-[(4-[trimethylsilylethynyl]phenyl)ethynyl]phenyl)-
ethynyl]isophthaloate (3P). The same procedure was used as for 2P.
2H (1.1 g, 3.15 mmol); 3 (1.0 g, 3.15 mmol); tetrakis(triphenylphos-
phine)palladium(0) (73 mg, 0.06 mmol, 2 mol %); copper iodide (9
mg, 0.05 mmol, 1.5 mol %); piperidine (12 mL). Column chromatog-
raphy on silica gel with dichloromethane/cyclohexane: 9/1 as eluent
to give 3P as pale yellow crystals (1.57 g; 96% yield): MS (CI, NH3);
m/z 535.0 (calcd av mass for C33H30O4Si + NH3: 535.71), 519.0 (calcd
Diethyl-5-(iodo)isophthaloate (6). The same procedure was used
as for 3. Sodium nitrite (3.5 g, 50 mmol, 1.2 equiv) in 30 mL of water;
5 (10 g, 42 mmol) in 11 mL of hydrochloric acid (2 M); potassium
iodide (10.5 g, 63 mmol, 1.5 equiv) in 100 mL of water. Column
chromatography on silica gel with dichloromethane as eluent to give 6
as a pale yellow solid (12.4 g, 85% yield): mp 76 °C (ethanol, lit.31 76
°C); 1H NMR (200 MHz, CDCl3): δ ) 8.62 (t, J ) 1.4 Hz, 1H), 8.53
(d, J ) 1.4 Hz, 2H), 4.40 (q, J ) 7.1 Hz, 4H), 1.41 (t, J ) 7.1 Hz,
6H); 13C NMR (50 MHz, CDCl3): δ ) 164.2, 142.1, 132.3, 129.6,
93.2, 31.6, 14.1.
1
av mass for C33H30O4Si: 518.6); mp 142 °C (iPrOH); H NMR (200
MHz, CDCl3): δ ) 8.64 (t, J ) 1.6 Hz, 1H), 8.36 (d, J ) 1.6 Hz, 2H),
7.53 (s, 4H), 7.46 (s, 4H), 4.44 (q, J ) 7.1 Hz, 4H), 1.44 (t, J ) 7.1
Hz, 6H), 0.26 (s, 9H); 13C NMR (50 MHz, CDCl3): δ ) 165.2, 136.4,
132.0, 131.8, 131.7, 131.5, 131.4, 130.2, 124.0, 123.5, 123.3, 123.0,
122.6, 104.6, 96.6, 91.2, 90.9, 90.8, 89.4,, 61.7, 14.4, -0.1; anal. calcd
for [90% C33H30O4Si (518.6), 10% iPrOH]: C 75.53, H 5.93; found:
C 75.31; H 5.80.
Diethyl-5-[(4-[(4-ethynylphenyl)ethynyl]phenyl)ethynyl]-
isophthaloate (3H). The same procedure was used as for 1H. 3P (1.5
g, 2.9 mmol); cesium carbonate (1.1 g 3.4 mmol, 1.1 equiv); ethanol/
dichloromethane (1/2; 15 mL). Column chromatography on silica gel
with dichloromethane as eluent to give 3H as pale yellow crystals (0.8
g; 61% yield): MS (CI, NH3); m/z 464.0 (calcd av mass for C30H22O4
Diethyl-5-(trimethylsilylethynyl)isophthaloate (1P).12b The same
procedure was used as for 2. Trimethylsilylacetylene (1.8 mL, 12.8
(30) Beyer, B. J. Prakt. Chem. 1882, 25, 465-517.
(31) Burton, H.; Kenner, J. J. Chem. Soc. 1923, 123, 1043-1045.

Crosslike Molecules for Solubility ImproVement
J. Am. Chem. Soc., Vol. 123, No. 34, 2001 8187
+ NH3: 463.5), 447.0 (calcd av mass for C30H22O4: 446.5); mp 166
°C (dioxane/ethanol); H NMR (200 MHz, CDCl3): δ ) 8.64 (t, J )
89.3, 64.8, 61.5, 14.2; anal. calcd for [1.00 mol of C145H100O16
(2098.39), 0.22 mol of dioxane]: C 82.33, H 4.84; found: C 82.31, H
4.78.
1
1.6 Hz, 1H), 8.36 (d, J ) 1.6 Hz, 2H), 7.54 (s, 4H), 7.49 (s, 4H), 4.44
(q, J ) 7.1 Hz, 4H), 3.20 (s, 1H), 1.44 (t, J ) 7.1 Hz, 6H); 13C NMR
(50 MHz, CDCl3): δ ) 165.2, 136.4, 132.2, 131.8, 131.7, 131.5, 131.4,
130.2, 124.0, 123.5, 123.4, 122.6, 122.2, 91.0, 90.9, 90.8, 89.5, 83.2,
79.2, 61.7, 14.4; anal. calcd for C30H22O4 (446.50): C 80.69, H 4.96;
found: C, 80.09; H, 5.01.
Tetrakis[4-(iodo)phenyl]methane (8).13b A suspension of tetraphe-
nylmethane (2.5 g, 5.5 mmol), bis(trifluoroacetoxy)iodobenzene (5.9
g, 13.7 mmol, 2.5 equiv), and iodine (4.2 g, 16.5 mmol, 3 equiv) in 35
mL of carbon tetrachloride was stirred at reflux for 16 h. After cooling
at 40-35 °C, the suspension was filtered, and the solid residue was
washed twice with 10 mL of dichloromethane. After drying, 8 was
obtained as a pink powder (2.7 g; 60% yield): MS (CI, NH3); m/z
824.0 (calcd av mass for C25H16I4: 824.0); 1H NMR (200 MHz,
CDCl3): δ ) 7.60 (apparent d, J ) 8.7 Hz, 2H), 6.89 (apparent d, J )
8.7 Hz, 2H).
Diethyl-5-[(phenyl)ethynyl)]isophthaloate (1R). The same proce-
dure was used as for 2P. 1H (183 mg, 0.74 mmol); (9) (227 mg, 1.1
mmol, 1.5 equiv); tetrakis(triphenylphosphine)palladium(0) (43 mg, 0.04
mmol, 5 mol %); copper iodide (3 mg, 0.02 mmol, 2 mol %); piperidine
(3 mL). Column chromatography on silica gel with dichloromethane
as eluent to give 1R as white crystals (227 mg, 95% yield): MS (CI,
NH3); m/z 322.99 (calcd av mass for C20H18O4: 322.36); mp 87.3 °C
(ethanol); 1H NMR (200 MHz, CDCl3): δ ) 8.64 (t, J ) 1.7 Hz, 1H),
8.37 (d, J ) 1.7 Hz, 2H), 7.62-7.57 (m, 2H), 7.42-7.38 (m, 3H),
4.43 (q, J ) 7.1 Hz, 4H), 1.43 (t, J ) 7.1 Hz, 6H); 13C NMR (50
MHz, CDCl3): δ ) 164.8, 136.0, 131.5, 131.0, 129.7, 128.6, 128.2,
124.0, 122.3, 90.9, 87.3, 61.3, 14.1; anal. calcd for C20H18O4 (322.36):
C 74.52, H 5.63; found: C 74.55, H 5.82.
Diethyl-5-[(4-[(phenyl)ethynyl]phenyl)ethynyl]isophthaloate (2R).
The same procedure was used as for 2P. 2H (197 mg, 0.57 mmol); (9)
(174 mg, 0.85 mmol, 1.5 equiv); tetrakis(triphenylphosphine)palladium-
(0) (33 mg, 0.03 mmol, 5 mol %); copper iodide (2 mg, 0.01 mmol, 2
mol %); piperidine (3 mL). Column chromatography on silica gel with
dichloromethane as eluent to give 2R as white crystals (220 mg, 91%
yield): MS (CI, NH3); m/z 422.0 (calcd av mass for C28H22O4: 422.4);
mp 116.3 °C (ethanol); 1H NMR (200 MHz, CDCl3): δ ) 8.64 (t, J )
1.7 Hz, 1H), 8.37 (d, J ) 1.7 Hz, 2H), 7.57-7.53 (m, 6H), 7.40-7.35
(m, 3H), 4.42 (q, J ) 7.1 Hz, 4H), 1.44 (t, J ) 7.1 Hz, 6H); 13C NMR
(50 MHz, CDCl3): δ ) 165.1, 136.3, 131.7, 131.6, 131.4, 130.2, 128.5,
128.4, 124.0, 123.8, 123.0, 122.3, 91.6, 90.8, 89.2, 89.0, 87.3, 61.6,
14.3; anal. calcd for C28H22O4 (422.48): C 79.60, H 5.25, found: C
79.65, H 5.12.
Tetrakis[4-(diethyl-5-[ethynyl]isophthaloate)phenyl]methane (1C).
The same procedure was used as for 2P. 8 (160 mg, 0.19 mmol, 1
equiv); 1H (224 mg, 0.9 mmol, 4.8 equiv); tetrakis(triphenylphosphine)-
palladium(0) (22 mg, 0.02 mmol, 10 mol %); copper iodide (2 mg,
0.01 mmol, 5 mol %); piperidine/tetrahydrofuran (4/1; 5 mL). Column
chromatography on silica gel with dichloromethane as eluent to give
1C as white crystals (200 mg, 82% yield); MS (CI, NH3); m/z 1314.4
(calcd av mass for C81H68O16 + NH3: 1314.4); mp 209.1 °C (ethanol);
1H NMR (200 MHz, CDCl3): δ ) 8.63 (t, J ) 1.6 Hz, 1H), 8.36 (d,
J ) 1.6 Hz, 2H), 7.50 (apparent d, J ) 8.2 Hz, 2H), 7.25 (apparent d,
J ) 8.2 Hz, 2H), 4.43 (q, J ) 7.1 Hz, 4H), 1.43 (t, J ) 7.1 Hz, 6H);
1H NMR (600 MHz, dioxane-d8): δ ) 8.54 (t, J ) 1.7 Hz, 1H), 8.35
(d, J ) 1.7 Hz, 2H), 7.58-7.57 (dd, Jortho ) 8.5 Hz, Jmeta ) 1.8 Hz,
2H), 7.37-7.36 (dd, Jortho ) 8.5 Hz, Jmeta ) 1.8 Hz, 2H), 4.41 (q, J )
7.1 Hz, 4H), 1.41 (t, J ) 7.1 Hz, 6H); 13C NMR (50 MHz, CDCl3): δ
) 165.1, 146.2, 136.3, 131.3, 131.2, 130.8, 130.0, 124.0, 120.7, 90.6,
87.9, 64.9, 61.5, 14.2; anal. calcd for C81H68O16 (1297.42): C 74.98,
H 5.28; found C 75.09, H 5.26.
Diethyl-5-[(4-[(4-[(phenyl)ethynyl]phenyl)ethynyl]phenyl)ethynyl]-
isophthaloate (3R). The same procedure was used as for 2P. 3H (165
mg, 0.37 mmol); (9) (112 mg, 0.55 mmol, 1.5 equiv); tetrakis-
(triphenylphosphine)palladium(0) (22 mg, 0.02 mmol, 5 mol %); copper
iodide (1.5 mg, 0.01 mmol, 2 mol %); piperidine (2 mL). Column
chromatography on silica gel with dichloromethane as eluent to give
3R as a white powder (170 mg, 88% yield): MS (CI, NH3); m/z 540.0
(calcd av mass for C36H26O4 + NH3: 539.6); mp 204.5 °C (ethanol/
Tetrakis[4-(diethyl-5-[(4-[ethynyl]phenyl)ethynyl]isophthaloate)-
phenyl]methane (2C). The same procedure was used as for 2P. 8 (160
mg, 0.19 mmol, 1 equiv); 2H (315 mg, 0.9 mmol, 4.8 equiv), tetrakis-
(triphenylphosphine)palladium(0) (22 mg, 0.02 mmol, 10 mol %);
copper iodide (2 mg, 0.01 mmol, 5 mol %); piperidine/tetrahydrofuran
(4/1; 5 mL). Column chromatography on silica gel with dichloromethane
as eluent to yield 2C as a pale yellow powder (260 mg; 81% yield).
The sample for DSC was precipitated from dioxane/iPrOH: MS (FAB,
1
tetrahydrofuran); H NMR (250 MHz, CDCl3): δ ) 8.65 (t, J ) 1.6
Hz, 1H), 8.37 (d, J ) 1.6 Hz, 2H), 7.57-7.36 (m, 13H), 4.44 (q, J )
7.1 Hz, 4H), 1.44 (t, J ) 7.1 Hz, 6H); 13C NMR (100 MHz, CDCl3):
δ ) 165.1, 136.3, 131.6, 131.6, 131.5, 131.2, 130.1, 128.5, 128.3, 123.9,
123.4, 123.3, 122.9, 122.6, 122.4, 91.4, 91.2, 90.7, 90.7, 89.3, 89.0,
61.6, 14.3; anal. calcd for C36H26O4 (522.60): C 82.74, H 5.01, found:
C 82.62; H 5.21.
1
TFA); m/z 1697.7 (calcd av mass for C113H84O16: 1697.8); H NMR
UV/Vis Spectroscopic Measurements. All experiments were per-
formed at 298 K in spectroscopic grade methylene chloride. The UV/
Vis absorption spectra were recorded on a Kontron Uvikon-930
spectrophotometer. Corrected fluorescence spectra were obtained with
a Photon Technology International LPS 220 spectrofluorometer.
Solutions for fluorescence measurements were adjusted to a specific
concentration so that the maximum absorbance was e0.15 at the
excitation wavelength. The overall fluorescence quantum yields of nR
and nC were calculated from the relation:32
(200 MHz, CDCl3): δ ) 8.65 (t, J ) 1.6 Hz, 1H), 8.37 (d, J ) 1.6
Hz, 2H), 7.54 (s, 4H), 7.49 (apparent d, J ) 8.4 Hz, 2H), 7.24 (apparent
d, J ) 8.4 Hz, 2H), 4.44 (q, J ) 7.1 Hz, 4H), 1.44 (t, J ) 7.1 Hz, 6H);
1H NMR (600 MHz, dioxane-d8): δ ) 8.56 (t, J ) 1.7 Hz, 1H), 8.38
(d, J ) 1.7 Hz, 2H), 7.65 (apparent d, J ) 8.3 Hz, 2H), 7.60 (apparent
d, J ) 8.3 Hz, 2H), 7.52 (apparent d, J ) 8.4 Hz, 2H), 7.34 (apparent
d, J ) 8.4 Hz, 2H), 4.42 (q, J ) 7.1 Hz, 4H), 1.41 (t, J ) 7.1 Hz, 6H);
13C NMR (100 MHz, CDCl3): δ ) 165.1, 146.0, 136.3, 131.6, 131.6,
131.2, 131.1, 130.8, 130.1, 123.9, 123.5, 122.3, 121.0, 91.0, 90.7, 89.4,
89.2, 64.9, 61.6, 14.2; anal. calcd for [1.00 mol of C113H84O16 (1697.8),
0.60 mol of dioxane]: C 77.81, H 5.11; found: C 77.84, H 5.02.
A(λ)sFun2u
A(λ)uFsn2s
Φu ) Φs
(11)
Tetrakis[4-(diethyl-5-[4-(4-[ethynyl]phenyl)ethyny]phenyl)ethynyl]-
isophthaloate) phenyl]methane (3C). The same procedure was used
as for 2P. (8) (160 mg, 0.19 mmol, 1 equiv); 3H (405 mg, 0.9 mmol,
4.8 equiv); tetrakis(triphenylphosphine)palladium(0) (22 mg, 0.02 mmol,
10 mol %); copper iodide (2 mg, 0.01 mmol, 5 mol %); piperidine/
tetrahydrofuran (4/1; 5 mL). Column chromatography on silica gel with
dichloromethane to give 3C as a pale yellow powder (374 mg; 94%
yield): MS (FAB, NBA); m/z 2097.8 (calcd av mass for C145H100O16:
where the subscripts s and u indicate the standard and unknown sample,
A(λ) corresponds to the absorbance of the solution at the excitation
wavelength λ, F is the integrated emission spectrum, and n is the
refractive index for the solvent carrying the unknown and the standard.
The standard fluorophore for solution measurements was quinine sulfate
in 0.1 M H2SO4 with Φs ) 0.50.32
Solubility Measurements. Enough of each compound nM (M )
R or C) (1-10 mg) was added to 1 mL of analytical grade cyclohexane
to keep some solid remaining in equilibrium with the solution in the
whole temperature range investigated. After centrifugation and tem-
perature reequilibration, an aliquot of 25 µL of supernatant was poured
1
2098.3); H NMR (400 MHz, CDCl3): δ ) 8.64 (t, J ) 1.5 Hz, 1H),
8.36 (d, J ) 1.5 Hz, 2H), 7.54 (s, 4H), 7.52 (s, 4H), 7.47 (apparent d,
J ) 8.4 Hz, 2H), 7.23 (apparent d, J ) 8.4 Hz, 2H), 4.43 (q, J ) 7.1
Hz, 4H), 1.44 (t, J ) 7.1 Hz, 6H); 13C NMR (100 MHz, CDCl3): δ )
165.1, 146.0, 136.3, 131.6, 131.5, 131.5, 131.2, 131.1, 130.8, 130.1,
123.9, 123.3, 123.2, 122.7, 122.4, 121.1, 91.2, 90.9, 90.8, 90.7, 89.5,
(32) Demas, J. N.; Crosby, G. A. J. Phys. Chem. 1971, 75, 991-1024.

8188 J. Am. Chem. Soc., Vol. 123, No. 34, 2001
Aujard et al.
into 5 mL of dichloromethane. Then the corresponding solution was
used for determining the saturated concentration in nM, [nM]eq,T, from
UV absorption experiments. Eventually the thermodynamic constant
K°(T) associated with the dissolution process at T was evaluated from
K°(T) ) [nM]eq,T/C° with C° ) 1 mol L-1).
DSC Experiments. Differential scanning calorimetry experiments
were carried out using a DSC 7 Perkin-Elmer. The following procedure
was applied for the nR and nC compounds (1-2 mg samples): first
heating cycle (20-230 °C; 5 °C/min but 20 °C/min for 2C and 3C),
cooling cycle (20-230 °C; 5 °C/min but 20 °C/min for 2C and 3C);
second heating cycle (20-230 °C; 5 °C/min but 20 °C/min for 2C and
3C). TLC using reference compounds nR and nC was used to check
that no degradation occurred during the DSC experiments.
NMR Studies. The measurements of translational diffusion coef-
ficients were carried out on an Avance Bruker DRX 600 spectrometer
equipped with a TBI probe with three orthogonal pulsed field gradients
(PFG). The pulse sequence used was a double stimulated echo (DSTE)
which suppresses the effects of fluid velocity (convection).33 Eddy
current effects were minimized using sine shaped PFG and introducing
a longitudinal eddy current delay (LED) of 5 ms before acquisition.
The total diffusion delay length was 200 ms. Encoding PFG lengths
were of 2 ms for all experiments. Their maximum amplitudes were
varied from 1.1 G‚cm-1 to 52.9 G‚cm-1 in 32 steps. Each experiment
was acquired with 32 transients, except for 1R, 2R, and 1C where
eight transients led to good enough signal-to-noise ratios. Two
dimension diffusion-ordered spectroscopy (DOSY) spectra were ob-
tained after monoexponential inverse Laplace transformation and
diffusion coefficients were measured, corresponding to the maximum
intensity of the projection onto the diffusion axis (see Figure 2S).
Additional peaks in the diffusion dimension were observed for each
nucleus in each molecule. They arise from artifacts of the monoexpo-
nential inverse Laplace transform. In fact, the decrease of the intensity
as a function of the squared gradient strength is not perfectly
exponential.34 Moreover, a biexponential inverse Laplace transform
modifies the measured diffusion coefficient by less than 2%, so that
the error introduced by these artifacts may be neglected.
sent with normal incidence on the resulting interferometer. The
transmitted beam collected with a lens consists of fringes of equal
chromatic order (FECO) arising from constructive interference of the
light reflected at the various interfaces of the interferometer.5c After
analysis with a calibrated spectroscope, the FECO wavelengths
independently provide both the film thickness at 0.1 nm accuracy and
its refractive index. Upon scanning on a line by steps equal to 2 µm,
one obtains interferometric pictures of the film cross section at a 4 µm
lateral resolution. Under the present conditions, solvent evaporation
produced islandlike homogeneous zones (characteristic size: at least
several micrometers). At least 10 islands from each film were used to
evaluate its thickness.
Derivation of Molecular Characteristic Lengths from Hydrody-
namic Analysis of the Diffusion Coefficients Dt. Under considering
the medium as a continuum and assuming stick boundary conditions
to apply in the present case,17 the following expressions were used to
derive molecular characteristic lengths from the diffusion coefficients:
kT
Body assimilated to a sphere Dtsphere
)
(12)
6πηRsphere
where k is the Boltzmann constant, T the absolute temperature, η the
medium viscosity and Rsphere the sphere radius.
Body Assimilated to a right Circular Cylinder (Length Lrod
,
Diameter D). A semiempirical formula is available in the range 2 e
p ) Lrod/D e 30.35
kT(ln p + ν)
3πηLrod
Dt )
with ν ) 0.312 + 0.565/p - 0.100/p2 (13)
For the calculations, the viscosity of dioxane - d8 was taken equal to
1.295 10-3 Pas at 298 K.36
Acknowledgment. This paper is dedicated to Jean Jacques.
We acknowledge Dr. Jose´e Brienne for her essential assistance
in recording DSC scans. We are also indebted to Professor
Geoffrey Bodenhausen for interesting discussions.
Interferometry Experiments. Five microliters of 10 µM nC solution
in DME-water (9/1 (v/v)) were spread on the clean side of a 2-5 µm-
thin mica sheet silvered on the other side (transmittivity about 1% in
green visible light) to yield discoidal films typically lying in 16-19
mm for the diameter. After evaporation in a vacuum for 30 min, a
similar second piece of mica was put on. Collimated white light was
Supporting Information Available: Additional figures
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA010019H
(35) Tirado, M. M.; Martinez, C. L.; Garcia De La Torre, J. J. Chem.
Phys. 1984, 81, 2047-2052.
(36) Holz, M.; Mao, X.-A.; Seiferling, D.; Sacco, A. J. Chem. Phys.
1996, 104, 669-679.
(33) Jerschow, A.; Muller, N. J. Magn. Reson. 1997, 125, 372-375.
(34) Johnson, C. S. Prog. Nucl.Magn. Reson. Spectrosc. 1999, 34, 203-
256.