Synthesis and properties of water-soluble asterisk molecules

Article abstract of DOI:10.1021/ja0206238

An asterisk is comprised of six semirigid arms projecting from a benzene nucleus. In the case at hand, asterisks were synthesized with one, two, or three aromatic rings (connected by sulfur atoms) in each of the six arms. A phosphomonoester at the termini of each arm solubilized the asterisks in water. The colloidal properties of these amphiphilic molecules were investigated by UV-vis and fluorescence spectroscopy, calorimetry, light scattering, surface tensiometry, and pulse-gradient spin-echo NMR. Solubility, solubilization, metal binding, and micelle "seeding" experiments were also carried out. Chain-conformation and supramolecular assembly into remarkable molecular "scrolls" were investigated by X-ray analysis and electron microscopy, respectively. One of the more interesting properties of the asterisks is that they remain monomeric in water despite having as many as 19 hydrophobic aromatic rings exposed to the water. The reasons for this behavior, and the possibility of exploiting it for constructing enzyme models free from aggregation equilibria, are discussed.

Full text of DOI:10.1021/ja0206238

Published on Web 08/23/2002
Synthesis and Properties of Water-Soluble Asterisk Molecules
Fredric M. Menger* and Vladimir A. Azov
Contribution from the Department of Chemistry, Emory UniVersity, 1515 Pierce DriVe,
Atlanta, Georgia 30322
Received April 30, 2002
Abstract: An asterisk is comprised of six semirigid arms projecting from a benzene nucleus. In the case
at hand, asterisks were synthesized with one, two, or three aromatic rings (connected by sulfur atoms) in
each of the six arms. A phosphomonoester at the termini of each arm solubilized the asterisks in water.
The colloidal properties of these amphiphilic molecules were investigated by UV-vis and fluorescence
spectroscopy, calorimetry, light scattering, surface tensiometry, and pulse-gradient spin-echo NMR.
Solubility, solubilization, metal binding, and micelle “seeding” experiments were also carried out. Chain-
conformation and supramolecular assembly into remarkable molecular “scrolls” were investigated by X-ray
analysis and electron microscopy, respectively. One of the more interesting properties of the asterisks is
that they remain monomeric in water despite having as many as 19 hydrophobic aromatic rings exposed
to the water. The reasons for this behavior, and the possibility of exploiting it for constructing enzyme
models free from aggregation equilibria, are discussed.
Introduction
Would our asterisks behave similarly to anionic micelles or
dendrimers to which they have a vague resemblance? Would
This paper deals with molecules that have six semirigid arms
projecting from a benzene-ring core. Each of the arms terminates
in an anionic group (R ) phosphate in the structure below),
thereby imparting water solubility to otherwise highly water-
insoluble molecules. The compounds were prepared in three
different “generations” according to whether the arms each
possess one, two, or three aromatic rings. For example,
generation-3 molecules have a total of 18 aromatic rings within
the six arms. Previously reported compounds, where R ) i-Pr,
called asterisks, inspired the design of our systems.1 Recent
additional examples of various multiarmed aromatics are also
available.2-6 But syntheses of water-soluble asterisks, and
examination of their colloidal properties, have no counterpart
in past work.
the molecules interlock their arms in water? Would an asterisk
solubilize small hydrophobic molecules or metal ions between
its arms? Answering such questions was a prime concern of
ours. Yet, admittedly, pursuit of the project also appealed to us
aesthetically. It was pleasurable to imagine asterisk molecules
darting through waterslike six-sided snow-flakes bandying
about in the wind.
Characterizing the solution properties of the asterisks utilized
the following: UV-vis and fluorescence spectroscopies, cal-
orimetry, light scattering, surface tensiometry, and pulse-gradient
spin-echo NMR in addition to studies of solubility, solubili-
zation effects, and metal binding. Solid-state properties were
also examined via X-ray analysis and electron microscopy. As
stated previously, a balanced perspective in colloid chemistry
can be best achieved by multiple methods, each of which reveals
only a fragment of reality.7
Synthesis
The synthetic strategy used to obtain the water-soluble
asterisks is shown in Scheme l. In general terms, the hydroxyl
of 4-thiomethylphenol was first protected with R, after which
the sulfur was demethylated to give thiol-1. The latter was used
to displace the six chlorines of hexachlorobenzene to produce
a first-generation asterisk system A-l. Similar chemistry with
preformed arms 2 and 3 (made from 1) gave second and third
generation asterisks A-2 and A-3. Removing the protecting
To whom correspondence should be addressed. E-mail: menger@
emory.edu.
(1) Gringas, M.; Pinchart, A.; Dallaire, C. Angew Chem., Int. Ed. 1998, 37,
3149.
(2) Dominique, R.; Liu, B.; Das, S. K.; Roy, R. Synthesis 2000, 862.
(3) Reitzel, N.; Hassenkam, T.; Balashev, K.; Jensen, T. R.; Howes, P. B.;
Kjaer, K.; Fechtenko¨tter, A.; Tchebotareva, N.; Ito, S.; Mu¨llen, K.;
Bjornholm, T. Chem. Eur. J. 2001, 7, 4894.
(4) Ko¨hn, F.; Hofkens, J.; Wiesler, U. M.; Cotlet, M.; van der Auweraer, M.;
Mu¨llen, K.; De Schryver, F. C. Chem. Eur. J. 2001, 7, 4126.
(5) Tucker, J. H. R.; Gingras, M.; Brand, H.; Lehn, J.-M. J. Chem. Soc., Perkin
Trans. 2 1997, 1303.
(6) Mayor, M.; Lehn, J.-M.; Fromm, K. M.; Fenske, D. Angew. Chem., Int.
Ed. Engl. 1997, 36, 2370.
(7) Menger, F. M.; Mbadugha, B. N. A. J. Am. Chem. Soc. 2001, 123, 875.
9
10.1021/ja0206238 CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 11159-11166
11159

A R T I C L E S
Menger and Azov
Scheme 1. Synthetic Strategy for Making Water-Soluble
Asterisks, Where R ) Protecting Group
Simple extension of the synthetic procedure used for genera-
tion-2 arms (eq 2) to make generation-3 arms was complicated
by formation of dimeric and tetrameric byproducts that could
not be removed chromatographically. An approach, combining
both Cu and Pd chemistry (Scheme 2), was therefore required.1
With the three thiols of eqs 1 and 2 and Scheme 2 in hand, we
were ready to prepare the asterisks.
Asterisks were synthesized according to the MacNicol
procedure1,14 in which the aryl thiolates (1.2-fold/Cl excess) were
reacted with hexachlorobenzene in 1,3-dimethyl-2-imidazoli-
dinone (DMI) at room temperature for 1-3 days (eq 3). (It was
fortunate that our thiolates were protected with a tert-butyl
protecting group because later attempts to use 4-hydroxy-
thiophenol in the MacNichol procedure gave only inseparable
mixtures despite reported successes to the contrary in the
literature).15 The coupling products were purified by column
chromatography and then deprotected with trifluoroacetic acid.
All three generations of hexaphenols form clathrates with
group R, and endowing the resulting terminal hydroxyls with
phosphates, led to the water-soluble asterisks. The apparent
simplicity of the scheme belies the many difficulties encountered
in its pursuit. Impaired solubility and slow reaction rates, relative
to those of low molecular weight counterparts, were common
with the asterisks. Poor yields were unacceptable since we
needed sufficient compound for multiple tests. The following
brief discussion of how these problems were overcome seems,
generally speaking, relevant to the design of supramolecular
systems in which p-phenylene sulfide “wires” are a desired
structural element.
The choice of protecting group R on the oxygen was critical.8
It needed to be stable in the presence of strong bases and
nucleophiles. And it had to be easily removed in high yield (a
stipulation made more acute by the presence of six such groups
per molecule). The triisopropylsilyl group failed us because
subsequent demethylation of the sulfur by two different routes
(t-BuSNa in dimethylformamide (DMF) and Na in NH3)9,10 gave
low yields due to decomposition or byproducts. We ended up
using the tert-butyl group despite the fact that its installation
(eq 1) took place with only a moderate yield of 58%.11 On the
other hand, the demethylation of sulfur was now almost
quantitative.
ether; even a high vacuum for several days does not destroy
them. The hexaphenols are soluble in ether, acetone, and
methanol, but only generation-l readily dissolves in basic water.
Air-oxidization is rapid, as might be expected. Treatment of
the generation-l hexaphenol with acetyl chloride or benzoyl
chloride gave crystalline hexaacyl compounds suitable for X-ray
analysis (reported later in the paper). Phosphorylation was
effected quantitatively using a 2.5-fold excess/(hydroxyl of
POCl3) (eq 4).16 Intermediates were subsequently hydrolyzed
with water at room temperature (generation-l) or 40-50 °C
(generations-2 and -3). Acidification of aqueous solutions with
concentrated HCl caused the fully protonated hexamonophos-
phates, designated A-l, A-2, and A-3, to precipitate.
Interconnecting the aromatic rings within the arms required
the coupling of aromatic thiolates with aromatic halides. Of two
previously described methods for accomplishing this feat (Cu2O
and Pd catalysis),12,13 the latter was far more effective (eq 2).
Rather than again using Na/NH3 to liberate the thiol (which
risked Birch-type interference), we carried out a high-yield
demethylation with sodium alkanethiolate. The resulting thiol
could be purified satisfactorily by vacuum distillation on a
Kugelrohr.
Physical Studies
To confine the paper to a reasonable length, the data are
presented in a concise format that retains the principal informa-
tion derived from the various physical methods.
(8) Greene, T. W.; Wuts, P. G. M. ProtectiVe Groups in Organic Synthesis,
3rd ed.; Wiley: New York, 1999.
(9) Testaferri, L.; Tiecco, M.; Tingoli, M.; Chianelli, D.; Montanucci, M.
Synthesis 1983, 751.
(10) McKinnon, D. M.; Lee, K. R. Can. J. Chem. 1988, 66, 1405.
(11) Holcombe, J. L.; Livinghouse, T. J. Org. Chem. 1986, 51, 111.
(12) Adams, R.; Ferretti, A. J. Am. Chem. Soc. 1959, 81, 4927.
(13) Kosugi, M.; Shimizu, T.; Migita, T. Chem. Lett. 1978, 13.
(14) MacNicol, D. D.; Mallinson, P. R.; Murphy, A.; Sym, G. J. Tetrahedron
Lett. 1982, 23, 4131.
(15) Stack, T. D. P.; Holm, R. H. J. Am. Chem. Soc. 1988, 110, 2484.
(16) Bourne, N.; Williams, A. J. Org. Chem. 1984, 49, 1200.
9
11160 J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002

Water-Soluble Asterisk Molecules
A R T I C L E S
Scheme 2. Synthesis of the Generation-3 Arm
Figure 1. Diffusion coefficients k obtained from FT-PGSE NMR for A-1
(9) and for A-2 (2). The lines are eye guides.
phospholipid assemblies.17 Self-assembly, driven by hydropho-
bic association of the asterisks’ aromatic arms, either fails to
occur or, if it does take place, the corresponding phase transitions
must fall outside our experimental temperature range.
Dynamic Light Scattering. Dynamic light scattering (DLS)
measurements, performed on a Coulter N4 sub-micrometer
particle sizer, were carried out to determine if any aggregates
existed at 1-3 mM concentrations in basic water (0.l0 M
NaOH). Experiments were performed at different scattering
angles of 30, 63, and 90° on optically clear solutions. The data
did not reveal particles larger than 5-7 nm, the resolution
threshold of the instrument. According to the DLS data,
therefore, aggregates of A-1, A-2, and A-3, if they form at all,
must be micelle size or smaller
Fourier Transform Pulsed-Gradient Spin-Echo NMR.
Fourier transform pulsed-gradient spin-echo NMR (FT-PGSE)
was used to obtain the diffusion rates of the asterisks in
water.18,19 If micelle-sized aggregates are formed, then past
experience would have us expect the diffusion rates to decrease
3-4-fold.20 Measurements were performed using Shigami
special precision NMR tubes and a 600 MHz INOVA 600
spectrometer: 0.l0 M NaOD/D2O, 0.3-30 mM A-l, and 0.3-
10 mM A-2. (Unfortunately, broadened signals for A-3 are too
weak to allow FT PGSE NMR studies on our largest asterisk.)
Results of the study are shown in Figure 1. The diffusion
rates for A-l and A-2 are seen not to differ substantially, nor is
there the marked concentration dependence that one generally
finds when a conventional surfactant passes through its critical
micelle concentration. Once again, no aggregation of the
asterisks in water is indicated.
The diffusion data allowed the calculation of the particle
diameters from the Stokes-Einstein equation and the diffusion
coefficient D (cm2/s) according to D ) kBT/3πηd, where kB is
the Boltzmann constant (erg/K), T is the temperature (K), η is
the viscosity (P) (assumed to be that of water), and d is the
diameter of the particle (cm). Both A-1 and A-2 asterisks have
apparent hydrodynamic diameters of 2.3 ( 0.l nm, a value
consistent with the monomolecular state. It is surprising that
A-2 manifests the same size as A-1, an observation whose
origins may be related to folding of the asterisks in water (to
minimize hydrocarbon-water contact) and to the inherent
assumptions of the Stokes-Einstein equation (i.e. that the
General Properties. Compounds A-l, A-2, and A-3 are bright
yellow compounds that are insoluble in most organic solvents
(hexane, CH2Cl2, CHCl3, ether, and THF) but highly soluble
in dimethyl sulfoxide (DMSO). Water-solubility, determined
semiquantitatively using a spectrophotometric assay, is pH-
dependent, as indicated by the following: (a) A-l is highly
soluble from pH ) 2 to pH ) 10 (only in concentrated HCl
does A-l fail to dissolve); (b) A-2 becomes highly soluble above
pH ) 8; and (c) the solubility of A-3 is low (ca. 4 × l0-4 M)
at pH ) 8 but increases more than l0-fold at pH ) 10.
Obviously, two opposing factors dictate solubility: the degree
of phosphate ionization and the length of the arms. Water-
solubility of A-3, for example, requires the formation of the
phosphate dianion, whereas the two smaller asterisks dissolve
in the monoanionic state.
UV-Vis Spectrophotometry. Departures from Beer-Lam-
bert linearity in plots of absorbance vs concentration would
imply self-assembly of the asterisks. Thus, UV-vis absorbance
measurements were carried out at 25.0 °C in 0.l0 M degassed
aqueous KOH using the following concentration ranges: (a)
A-l, 0.0l-l0 mM; (b) A-2, 0.0l-3 mM; and (c) A-3, 0.0l-3
mM. Such large concentration ranges were made possible by
using, and interrelating, eight wavelengths of widely differing
absorptivities between 390 and 460 nm (part of a long “tail”
ranging from 320 to 460 nm and accounting for the compounds’
yellow color). Linear plots over the indicated concentration
ranges (not shown) were an indication (although not a direct
proof) that none of the asterisks aggregates in water.
Flourescence. All three asterisks fluoresce in basic aqueous
solutions (0.l0 M KOH in degassed Milli-Q water). The
following excitation and emission maxima at l.0 mM were
recorded for characterization purposes: (a) A-l, 432 nm, 536
nm; (b) A-2, 446 nm, 538-539 nm; (c) A-3: 448 nm, 539-
540 nm. Relative fluorescence intensities at 0.30 mM increase
from 1 to 3 to 12 upon going from A-l to A-2 to A-3.
Differential Scanning Calorimetry. Calorimetric studies
were performed on near-saturated basic solutions of the three
asterisks in order to determine if they undergo thermally induced
phase transitions Thus, heating and cooling aqueous samples
in a Hart Scientific differential scanning calorimeter (DSC) at
a rate of 10 ïC/h between 10 and 90 °C gave no peak indicative
of a cooperative phase transition such as that found with
(17) Menger, F. M.; Wood, M. G., Jr.; Zhou, Q. Z.; Hopkins, H. P.; Fumero, J.
J. Am. Chem. Soc. 1988, 110, 6804.
(18) Menger, F. M.; Peresypkin, A. V.; Wu, S. J. Phys. Org. Chem. 2001, 14,
392.
(19) So¨derman, O.; Stilbs, P. Prog. Nucl. Magn. Reson. Spectrosc. 1994, 26,
445.
(20) Leaist, D. G. J. Colloid Interface Sci. 1986, 111, 230.
9
J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002 11161

A R T I C L E S
Menger and Azov
Figure 3. Dependence of mesitylene solubility on A-2 (2) and on A-3
(b) concentrations. Graphs represent averaged data from three consecutive
series of experiments.
low intrinsic solubility in water of about 0.4 mM (much lower
than that of benzene or toluene). (b) Mesitylene, having nine
identical methyl protons, can be analyzed by NMR with
relatively high sensitivity. (c) A high boiling point (162-164
°C) minimizes errors caused by volatilization. Thus, 1 mL of
asterisk solution (0.3-3 mM in 0.l M NaOD/D2O) was stirred
with excess (100 µL) mesitylene for 30 min after which the
layers were allowed to separate overnight. The aqueous layers
were then analyzed by proton NMR using a known quantity of
methanol as an internal reference to quantify the mesitylene
content.
An NMR spectrum of the water layer without asterisk showed
two mesitylene methyl signals: (a) a 1.76 ppm peak corre-
sponding to tiny amounts of suspended mesitylene which
persisted despite the long period allowed for phase separation;
(b) a 2.23 ppm peak corresponding to dissolved (aqueous)
mesitylene. As A-2 or A-3 was added to the water, the 2.23
ppm peak gradually shifted to 1.96 or 2.09 ppm, respectively.
Presumably, bound mesitylene “senses” the presence of both
water and aromatic host. Although A-l has no solubilization
capacity, A-2 and A-3 can solubilize 0.15 and 0.50 mesitylene
molecules/(asterisk molecule) (Figure 3). Stated in another way,
at 3 mM A-3 the mesitylene solubility in water increases almost
5-fold over and above its “natural” value.
How does the solubilization capacity of the asterisks compare
with that of micelles? Answering this question is complicated
by a sensitivity of micellar solublization to surfactant and
additive structure. Typically, ca. 20-40 ethylbenzene molecules,
for example, are solublized per 100-molecule micelle composed
of 12-carbon ionic surfactants.22 On a per-surfactant-molecule
basis, therefore, A-3 is equivalent to a conventional surfactant.
If consideration is given to the high molecular weight of A-3,
and to its restricted solubility as a monomer, then A-3’s overall
solubilization capacity is not competitive with micelles. Aster-
isks are thus best viewed as water-soluble compounds that
weakly complex aromatic guests.
Cooperative Effects. The question arose whether the anionic
asterisks, might “seed” micelle formation of a cationic surfactant
(e.g. n-dodecyltrimethylammonium bromide or DTAB). The
study of mixed micelles, a venerable area of colloid chemistry,23
has shown large reductions in critical micelle concentration
(CMC) when two surfactants of opposite charge are mixed.24
Such systems are different, however, from an asterisk/DTAB
Figure 2. Surface tension/concentration plots for A-1 (9), A-2 (2), and
A-3 (b). The lines are eye guides.
molecules are spherical). In a subsequent section, X-ray data
will disclose more detailed information on asterisk conformers.
Surface Tensiometry. The ability to decrease the surface
tension of water ranks as one of the most commonly reported
properties of amphiphilic substances.21 A conventional surfactant
adsorbs at the air/water interface with the polar headgroup
embedded in the water and the hydrophobic tail projecting into
the air. It was unclear how the asterisks would behave since
they have six ion-terminated arms radially disposed about a
central benzene ring. Surface tension measurements were made
with a Kibron µTrough (0.l mM KOH in Milli-Q water; 25.0
°C; 300 µL samples; three to four repeats per concentration
ranging from 0.01 to 10 mM), and the data are given in Figure
2. It is seen that all three asterisks are mildly surface-active,
reducing the surface tension from 72 to near 50 mN/m (or dyn/
cm) as compared to 30-40 mN/m for many conventional
surfactants.21 The decline in surface tension occurs over a l0-
fold concentration range (as opposed to more precipitous
declines with most conventional surfactants), suggesting that
adsorption at the air/water interface is unaccompanied by
discrete micelle formation. When the plots in Figure 2 level
off, the air/water interface is saturated with loosely packed
asterisks having, in all likelihood, the central core and an
unknown portion of the six arms exposed to the air. As would
be expected, saturation is reached first with A-3, then with A-2,
and last with A-l.
Solubilization. Most amphiphilic molecules self-assemble in
water into micelles, vesicles, etc., which are capable of
solubilizing hydrophobic molecules in water. The fact that the
asterisks fail to aggregate allowed a welcome opportunity to
study hydrophobic association to a single multiionic molecule
unburdened by aggregation equilibria. Since A-3, for example,
possesses 19 aromatic rings, there would seem to be ample
“exposed” surface area for the asterisk to hydrophobically bind,
and therefore to solubilize, nonpolar compounds that would not
ordinarily dissolve in water. The question, in other words, was
whether the asterisks would behave like “unimolecular mi-
celles”.
Mesitylene (1,3,5-trimethylbenzene) was selected as a test
compound for the following reasons: (a) It is a liquid with a
(22) Myers, D. Surfactant Science and Technology; VCH: New York, 1988; p
164.
(21) Jo¨nsson, B.; Lindman, B.; Holmberg, K.; Kronberg, B. Surfactants and
(23) Bergstro¨m, M.; Eriksson, J. C. Langmuir 2000, 16, 7173.
(24) Villeneuve, M.; Kaneshina, S.; Imae, T.; Aratono, M. Langmuir 1999, 15,
2029.
Polymers in Aqueous Solutions; Wiley: Chichester, U.K., 1998; pp 253-
255.
9
11162 J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002

Water-Soluble Asterisk Molecules
A R T I C L E S
and hexabenzoate differ only in their terminal para-substituents,
space-filling requirements, as opposed to intramolecular repul-
sions near the benzene core, seem to determine the solid-state
conformations. Neither the hexaacetate nor the hexabenzoate
crystal lattices show lattice porosity favorable to formation of
inclusion compounds as found with certain other hexasubstituted
benzenes.26 One speculates that the aabbab conformation would
be favored over the ababab conformation in water due to the
former’s greater intramolecular contact among the apolar arms,
but there is no evidence supporting this contention.
Molecular Scrolls. Upon carrying out the phosphorylation
given in eq 4, we observed that long needles of A-l precipitate
from the acidic solution. Optical and electron microscopy
revealed that these fibers (several millimeters long and up to
0.5 mm wide) had a scroll-like appearance (Figure 7). Molecular
scrolls have been reported previously,27 but the ones in Figure
7 are particularly beautiful. When the scrolls were washed with
acetone, they began to unroll and wrinkle (Figure 8). Scrolling
demonstrates the strong tendency of A-1 to self-assemble into
uniform molecular sheets that are ca. 4 µm thick. It is possible
that the molecules fill space (like hexagonal kitchen tile) with
intermolecular hydrogen-bonding among the phosphomonoesters
serving as an adhesive force at the six “corners”.
Figure 4. Surface tension/concentration dependence for DTAB in 0.5 mM
NaOH/Milli-Q water (() and in the same solvent in the presence of 1.0 ×
10-6 mol/L of asterisks: A-1 (9), A-2 (2), and A-3 (b). The lines are eye
guides.
pair where one of the components, namely, the asterisk, is not
in and of itself a micelle-forming molecule. Any enhanced
assembly of DTAB caused by an asterisk can be attributed to
an effect at the molecular level.
As seen in Figure 4, all three asterisks, at a remarkably low
concentration of l.0 µM in 0.5 mM NaOH, sharply reduce the
surface tension of aqueous DTAB solutions below DTAB’s
CMC of ca. 5 mM. For example, l.0 µM A-3 causes the surface
tension of l mM DTAB to decrease from 70 to 42 mN/m.
Clearly, 1 asterisk molecule/(1000 DTAB molecules) efficiently
promotes DTAB adsorption at, and saturation of, the air/water
interface. Although the data do not supply proof that assembly
into micelles is likewise favored by the asterisks, it seems likely
that DTAB micellization should respond similarly to the
interfacial assembly with regard to the presence of these
additives. Micelle-like NMR peak-broadening of DTAB below
its CMC (along with reduced diffusion rates measured by PGSE
NMR) brought on by micromolar levels of A-l (data not shown)
supports this speculation.
Metal Binding. Addition of Hg2+, an ion with a known
affinity for sulfur, to aqueous solutions of A-1 caused the
solution’s yellow color to intensify. (Corresponding experiments
with A-2 and A-3 could not be carried out because Hg2+
precipitates in the basic solutions necessary to dissolve these
asterisks.) UV-vis spectrophotometry showed absorbance
increases at wavelengths above 350 nm. Eight wavelengths
between 400 and 490 nm yielded consistent Job plots25 with
maxima at 0.5 (not shown), indicative of a 1:1 complex. No
further metal binding studies were pursued.
X-ray Studies. Compounds A-l, A-2, and A-3 are all
amorphous solids that decompose above 160 °C. The corre-
sponding generation-l hexaphenol could, however, be completely
acetylated to give a high-melting (199-202 °C) solid from
which X-ray-quality crystals were obtained by crystallization
(chloroform/hexane). Figure 5 shows the X-ray structure. As
was previously reported for other asterisks, the arms on the
central benzene ring alternate in a 1,3,5-up/2,4,6-down (or
ababab) arrangement.14,15 In contrast, the generation-l hexa-
benzoate has a 1,2,5 up/3,4,6-down (or aabbab) configuration
(Figure 6). The majority of known hexasubstituted benzenes
assume ababab conformations, although two other instances of
aabbab variations (with six â-naphthylthio or six 2-methylphe-
nylthio groups) have been reported.14,15 Since the hexaacetate
Discussion
There are but few instances in which the colloidal properties
of amphiphilic aromatics in water have been defined. Compared
to the aliphatic chains of conventional surfactants, aromatic rings
possess a reduced hydrophobicity28 and a greater rigidity, both
of which disfavor the assembly process in water. Micelles, after
all, are a disordered array of molecules where the hydrocarbon
chains prefer to twist and turn in a tangled brush-heap not unlike
that found in hydrocarbon solvents themselves.29 Even a single
cis double bond in a surfactant’s chain elevates the CMC by
3-4-fold because the permanent chain bend reduces the ability
to pack freely into a disorganized, multimolecular assembly.30
A benzene ring within an aliphatic chain does lower the CMC
but only by an amount equivalent to that of 3.5 methylenes.31
Moreover, since aromatics do not “freeze” water molecules as
effectively as do aliphatic chains,32 solvent release upon
aggregation (with its attendant entropic benefits) is less capable
of counteracting the concurrent electrostatic repulsion among
the assembled headgroups. In summary, packing difficulties and
an impaired hydrophobicity of aromatic rings explain a key
feature of water-soluble asterisks: Asterisks prefer the mono-
meric state in water despite having as many as 19 aromatic rings.
UV-Vis, DSC, DLS, FT-PGSE NMR, and tensiometry affirm
this conclusion.
Several years ago we explored the ability of an amphiphilic
polycyclic aromatic, namely, 1-pyrenesulfonic acid, to self-
assemble in water.32 1-Pyrenesulfonic acid is a 16-carbon
hydrocarbon with four coplanar aromatic rings. Although the
compound has many fewer aromatic rings than the asterisks, it
(26) Freer, A.; Gilmore, C. J.; MacNicol, D. D.; Wilson, D. R. Tetrahedron
Lett. 1980, 21, 1159.
(27) Elemans, J. A. A. W.; de Gelder, R.; Rowan, A. E.; Nolte, R. J. M. Chem.
Commun. 1998, 1553.
(28) Tanford, C. The Hydrophobic Effect; Wiley-Interscience: New York, 1973;
p 8.
(29) Menger, F. M. Acc. Chem. Res. 1979, 12, 111.
(30) Reference 22, pp 114-115.
(31) Shinoda, K. J. Phys. Chem. 1977, 81, 1300.
(32) Menger, F. M.; Whitesell, L. G. J. Org. Chem. 1987, 52, 3793.
(25) MacCarthy, P. Anal. Chem. 1978, 50, 2165.
9
J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002 11163

A R T I C L E S
Menger and Azov
Figure 5. X-ray picture of the first-generation asterisk terminating at each arm with an acetoxy group and sharing an ababab configuration.
Figure 6. X-ray picture of the first-generation asterisk terminating at each arm with a benzoyloxy group and sharing an aabbab configuration.
is flat and, it would seem, well set up for hydrophobically driven
π-π stacking. In actual fact, 1-pyrenesulfonic acid merely
dimerizes in water. Even at 50 mM (a concentration 100-fold
greater than the CMC of sodium hexadecyl sulfate), dimerization
of 1-pyrenesulfonic acid is only 50% complete. It is clear that
aromatic/aromatic association in water is far less efficient than
aliphatic/aliphatic association. But why, one may ask, does
simple pyrene dimerization overshadow its multimolecular
stacking? As seen in the next paragraph, an attempt to address
this question discloses some important properties of aromatics
relevant to the asterisks.
the presence of a benzene molecule. Theory also tells us that
the first hydration shell of benzene consists of 23 water
molecules.34 Of these, 21 comprise a hydrophobic “in-plane”
hydration. The other two water molecules solvate the benzene
on either side of the ring at the center via weak hydrogen-
bonding between the water protons and the π-cloud. Computer
simulations were also carried out on a benzene dimer embedded
in 510 water molecules.35 Two distinct energy minima were
detected. One of them corresponds to a contact interaction in
which two benzenes reside back-to-back only 4.3 Å apart. The
other represents a solvent-separated dimer with a 5.l Å separa-
tion. Although six water molecules are situated at the edges of
the complex, no water lies directly between the two benzene
rings. In summary, mildly structured water gathers primarily at
Monte Carlo studies of benzene in water suggest that water
molecules “close to the benzene molecule show only small
deviations from what is found in pure water”.33 Expressed in
another way: Water structure is only modestly perturbed by
(34) Ravishanker, G.; Mehrotra, P. K.; Mezei, M.; Beveridge, D. L. J. Am. Chem.
Soc. 1984, 106, 4102.
(35) Ravishanker, G.; Beveridge, D. L. J. Am. Chem. Soc. 1985, 107, 2565.
(33) Linse, P.; Karlstro¨m, G.; Jo¨nsson, B. J. Am. Chem. Soc. 1984, 106, 4096.
9
11164 J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002

Water-Soluble Asterisk Molecules
A R T I C L E S
Figure 7. SEM images of A-1 scrolls that precipitate from acidic water. Bars ) 20 µm (left) and 97 µm (right).
Figure 8. SEM images of A-1 scrolls washed with water and dried in air. Bars ) 242 µm (left) and 31.5 µm (right).
the periphery of the benzene rings rather than above or below
the rings where the actual hydrocarbon-hydrocarbon contact
takes place. One can only speculate that multimolecular stacking
is not observed because this creates internal benzene rings whose
edges, where most of the hydration of monomeric aromatics
normally occurs, are not optimally solvated. The entropic
benefits of releasing weakly bound water during stacking are
seemingly insufficient to compensate for a sterically impaired
“edge” solvation.
The upshot of the pyrene behavior and the theoretical
computations is that the asterisks, even A-3 with its six arms
containing three aromatic rings each, would not necessarily be
expected to self-assemble in water. Strong solvation of the ionic
groups in water, coupled to weak solvation of the aromatic rings,
may succeed in dragging these molecules into water, but there
the asterisks remainsunassembledsuntil the solubility limit is
reached. The principles involved here are useful because they
reveal a strategy for constructing enzyme models that are at
once (a) water soluble, (b) possessive of hydrophobic sites that
can bind substrates, and (c) free from complicating self-assembly
equilibria. The strategy, of course, is to construct the models
from solubilized aromatics bearing catalytic groups. By employ-
ing aromatic instead of aliphatic binding sites, one sacrifices a
degree of hydrophobic substrate association but one gains in
ensuring the monomeric state of the enzyme model.
more-or-less equivalent to the capabilities of a micellized
conventional surfactant. But when consideration is given to the
molecular weight of A-3, it is seen that in fact A-3 is not a
particular efficient solubilizer. Perhaps the same solvation factors
that discourage aggregation of A-3 also discourage binding of
small aromatics. A much bigger effect was observed with
asterisks “seeding” the assembly of DTAB, a conventional
cationic surfactant. For example, only 1.0 µM of A-l, A-2, or
A-3 (it does not seem to make much difference which) causes
a massive decrease of the surface tension from 70 to 41-45
mN/m in the presence of 1.0 mM “submicellar” DTAB (Figure
4). Apparently, a single asterisk molecule can provide a nucleus
around which a surfactant of opposite charge can self-assemble.
Although the efficiency by which the asterisks accomplish this
feat is remarkable, the phenomenon itself is not new. For
example, it is known that low concentrations of dyes can
sometimes lower the CMC of surfactants.36 Micelle seeding is
a worthy of a more systematic investigation.
Finally, we should mention the solid-state properties of
generation-l asterisks. The conformation in the crystal lattice
(i.e. the relative disposition of the six arms) is highly dependent
upon a substituent (acetyl or benzoyl) at the terminal carbon of
the arms (Figures 5 and 6). Difficult-to-predict packing factors,
as opposed to intramolecular steric effects near the core, must
control the configuration. Compound A-l itself precipitates from
water as uniform sheets that coil into scrolls (Figure 7). Although
the molecular basis of this supramolecular assembly is not yet
understood, its great beauty should be appreciated. And one
The asterisks display two properties of particular interest:
They solubilize mesitylene in water, and they trigger micelle
formation of a conventional surfactant. With regard to solubi-
lization, it was pointed out that A-3 can solubilize up to 0.5
molecule of mesitylene/(asterisk molecule) (Figure 3). This is
(36) Mukerjee, P.; Mysels, K. J. J. Am. Chem. Soc. 1955, 77, 2937.
9
J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002 11165

A R T I C L E S
Menger and Azov
in ethyl ether, the ether was washed with water, dried over sodium
sulfate, and removed. The resulting solid was chromatographed on silica
(ether/methanol with 0-6% gradient; ether/methanol with 0-4%
gradient; ether/methanol with 0-4% gradient for generations-1, -2, and
-3, respectively. Yields of purified material ranged from 74 to 93%.
All three products gave the expected 1H and 13 C NMR and FAB HRMS
(for generations-1 and -2) or FAB MS (for generation-3). C,H,S
elemental analyses agreed only to within 0.7 absolute % due to the
strong adhesion to ether that even vacuum pumping could not fully
remove. An X-ray structure of A-1-OH, and its acetyl and benzoyl
derivatives, affirmed the structure assignment.
Compounds A-l, A-2, and A-3. A solution of A-n-OH (0.41 mmol)
in pyridine (5 mL) was cooled in an ice bath and mixed with phosphorus
oxychloride (0.55 mL, 6 mmol). Stirring at 0 °C was continued for 2
h, after which the solution was allowed to warm slowly to room
temperature and diluted with 25 mL of dry ether. The resulting
precipitate was removed by filtration, leaving a filtrate that was taken
to dryness under reduced pressure. After 40 mL of water was added to
the solid, the mixture was stirred for 1 h. During that period, a yellow
syrup on the walls dissolved (creating a viscous turbid solution), and
a yellow material began to precipitate. Addition of 5-10 mL of
concentrated HCl caused further precipitation of a solid that was
removed by filtration and dried under reduced pressure over phosphorus
pentoxide. Yields ranged from 78 to 82%. Suitable 1H and 13 C spectra
were obtained, although the presence of bound water complicated the
C,H,S elemental analyses. Purity of the compounds was affirmed by
single 31P NMR signals (DMSO-d6) of -3.49, -3.70, and -3.35 ppm
for A-1, A-2, and A-3, respectively. Structural identity was affirmed
by mass spectral data in two cases. FAB HRMS. Calcd for A-1:
1302.8425. Obsd: 1302.8400. FAB MS. Calcd for A-2: 1949.8549.
Obsd: 1949.9.
wonders if the phenomenon is more generalsif other multi-
nuclear aromatics can self-assemble into sheets and scrolls when
hydrogen-bonding units are suitably positioned at critical contact
points.
Experimental Section
Physical Methods. Certain details and references pertaining to the
physical methods are given in the text; additional details are available
from the Supporting Information.
Synthesis. Yields, references, etc., pertaining to the procedures for
synthesizing the asterisk arms (eqs 1 and 2; Scheme 2) are given in
the text. Additional details are available from the Supporting Informa-
tion. Procedures for assembling the asterisks themselves are given
below, where A-n-X refers to an asterisk of generation n and terminal
group X.
A-n-OtBu. An arylthiophenol (7.2 mmol) and hexachlorobenzene
(2.85 g, 1 mmol) were dissolved in 1,3-dimethyl-2-imidazolidinone (25
mL), and the solution was degassed with argon for 1 h and cooled in
an ice bath after which sodium hydride (0.35 g, 14.4 mmol) was added
with vigorous stirring. Hydrogen began to evolve, the mixture became
a yellowish-brown, and a bright yellow solid began to precipitate. The
mixture was allowed to warm to room temperature and stirred for 12
(generation-l), 24 (generation-2), or 48 h (generation-3). The suspension
was then cooled to 0 °C, 50 mL of ice-cold 1 M HCl was added, and
the resulting mixture was poured in 200 mL of distilled water.
Precipitate was removed by filtration, washed with water several times,
and dissolved in chloroform. The chloroform solution was shaken with
water and brine and then dried over sodium sulfate. Removal of solvent
on a rotary evaporator produced a bright yellow solid that was purified
by column chromatography on silica (96:4 CHCl3/EtOAc; benzene/
EtOAc with 0-2% gradient; and hexane/benzene with 66-100%
gradient plus benzene/EtOAc with 0-2% gradient for generations-1,
-2, and -3, respectively). Yields of purified material ranged from 74 to
Acknowledgment. This work was supported by the Army
Research Office.
1
84%. Products from all three generations gave a satisfactory H and
13C NMR, FAB HRMS, and C,H,S elemental analysis.
Supporting Information Available: Experimental details,
analytical data for all new compounds, and crystallographic data
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
A-n-OH. A tert-butylated asterisk derivative (1 mmol) was
dissolved in 20 mL of dichloromethane, degassed for 1 h with argon,
and mixed with 20 mL of trifluoracetic acid. The solution was stirred
for 1 h and then evaporated to dryness to give a yellow solid residue
that was dried under reduced pressure. After the product was dissolved
JA0206238
9
11166 J. AM. CHEM. SOC. VOL. 124, NO. 37, 2002