Anatomy of a gel. Amino acid derivatives that rigidify water at submillimolar concentrations

Article abstract of DOI:10.1021/ja0016811

On the basis of suggestive X-ray data, 14 aroyl L-cystine derivatives were designed, synthesized, and examined for their ability to gelate water. Several members of this amino acid family are remarkably effective aqueous gelators (the best being one that can rigidify aqueous solutions at 0.25 mM, ca. 0.01%, in less than 30 s!). A few of the analogues separate from water as crystals, indicating a close relationship between gelation and crystallization. All effective gelators self-assemble into fibrous structures that entrain the solvent in the capillary spaces among them. Hydrogen-bonding sites on the compounds that might stabilize the fibers were identified from specific substitutions that replace a hydrogen donor with a methyl group, enhance the hydrogen-accepting ability of a carbonyl oxygen, or promote the hydrogen-donating ability of an amide proton. The structural variations were characterized via minimal gelation concentrations and times, X-ray crystallography, light and electron microscopy, rheology, and calorimetry. The multiple techniques, applied to the diverse compounds, allowed an extensive search into the basis of gelation. It was learned, for example, that the compound with the lowest minimum gelator concentration and time also has one of the weakest gels (i.e., it has a low elastic modulus). This is attributed to kinetic effects that perturb the length of the fibers. It was also argued that π/π stacking, the carboxyl carbonyl (but not the carboxyl proton), and solubility factors all contribute to the stability of a fiber. Polymorphism also plays a role. Rheological studies at different temperatures show that certain gels are stable to a 1-Hz, 3-Pa oscillating shear stress at temperatures as high as 90 °C. Other gels have a 'catastrophic' break at lower temperatures. Calorimetric data indicate a smooth transition from gel to sol as the temperature is increased. These and other issues are discussed in this 'anatomy' of a gel.

Full text of DOI:10.1021/ja0016811

J. Am. Chem. Soc. 2000, 122, 11679-11691
11679
Anatomy of a Gel. Amino Acid Derivatives That Rigidify Water at
Submillimolar Concentrations
Fredric M. Menger* and Kevin L. Caran
Contribution from the Department of Chemistry, Emory UniVersity, 1515 Pierce DriVe,
Atlanta, Georgia 30322
ReceiVed May 15, 2000. ReVised Manuscript ReceiVed September 20, 2000
Abstract: On the basis of suggestive X-ray data, 14 aroyl L-cystine derivatives were designed, synthesized,
and examined for their ability to gelate water. Several members of this amino acid family are remarkably
effective aqueous gelators (the best being one that can rigidify aqueous solutions at 0.25 mM, ca. 0.01%, in
less than 30 s!). A few of the analogues separate from water as crystals, indicating a close relationship between
gelation and crystallization. All effective gelators self-assemble into fibrous structures that entrain the solvent
in the capillary spaces among them. Hydrogen-bonding sites on the compounds that might stabilize the fibers
were identified from specific substitutions that replace a hydrogen donor with a methyl group, enhance the
hydrogen-accepting ability of a carbonyl oxygen, or promote the hydrogen-donating ability of an amide proton.
The structural variations were characterized via minimal gelation concentrations and times, X-ray crystallography,
light and electron microscopy, rheology, and calorimetry. The multiple techniques, applied to the diverse
compounds, allowed an extensive search into the basis of gelation. It was learned, for example, that the compound
with the lowest minimum gelator concentration and time also has one of the weakest gels (i.e., it has a low
elastic modulus). This is attributed to kinetic effects that perturb the length of the fibers. It was also argued
that π/π stacking, the carboxyl carbonyl (but not the carboxyl proton), and solubility factors all contribute to
the stability of a fiber. Polymorphism also plays a role. Rheological studies at different temperatures show
that certain gels are stable to a 1-Hz, 3-Pa oscillating shear stress at temperatures as high as 90 °C. Other gels
have a “catastrophic” break at lower temperatures. Calorimetric data indicate a smooth transition from gel to
sol as the temperature is increased. These and other issues are discussed in this “anatomy” of a gel.
Introduction
A wide variety of nonpolymeric compounds have been
encountered that create colloidal gels in organic media.1-4 The
great majority have been discovered by chance and include
diverse structural types such as long-chain n-alkanes,5 steroid/
aromatic conjugates,6-9 carbohydrate derivatives,10-13 dendrim-
ers,14 and two-component systems.15-17 Nonpolymeric com-
pounds capable of gelating water include arborols18,19 and
amphiphiles;20-22 they are less common because 55 M water
competes for hydrogen-bonding sites, a major associative
element. One particularly interesting gelator, a diammonium
gemini surfactant,23 is able to form gels in both water and
Any child who has eaten a gelatin desert, or experienced
“sheer-thinning” by squirting it through his teeth, is familiar
with the gel state. Actually, gels are found not only in food but
throughout Nature. Protoplasm, for example, is a gel. How
strange it is that, despite gels being so commonplace, hard data
on their molecular structure is in limited supply. There are
several reasons for this: Gelators of water are usually large
molecules (i.e., proteins and polymers) whose complicated
intermolecular associations are difficult to define. Moreover,
the interactions are generally not static but can change (often
irreversibly) with time, heat, or stress. And, of course, gels do
not lend themselves to atomic resolution X-ray analysis, the
main source of our most precise structural information.
We have learned how to prepare clear gels in less than 30 s
via extremely dilute (0.25 mM!) aqueous solutions of amino
acid derivatives. The story of how we came to design such
simple gelators is the subject of the present paper. Although
we do not know the structure of our gels with X-ray-like detail,
we now have a basic understanding of how the compounds
rigidify water. But before beginning the story (which will take
us into the realm of microscopy and rheology), let us first give
a succinct overview of gel chemistry.
(5) Abdallah, D. J.; Weiss, R. G. Langmuir 2000, 16, 352.
(6) Geiger, C.; Stanescu, M.; Chen, L.; Whitten, D. G. Langmuir 1999,
15, 2241.
(7) Terech, P.; Ostuni, E.; Weiss, R. G. J. Phys. Chem. 1996, 100, 3759.
(8) Lin, Y.-c.; Kachar, B.; Weiss, R. G. J. Am. Chem. Soc. 1989, 111,
5542.
(9) Lin, Y.-c.; Weiss, R. G. Macromolecules 1987, 20, 414.
(10) Watase, M.; Nakatani, Y.; Itagaki, H. J. Phys. Chem. B 1999, 103,
2366.
(11) Hafkamp, R. J. H.; Feiters, M. C.; Nolte, R. J. M. J. Org. Chem.
1999, 64, 412.
(12) Amanokura, N.; Yoza, K.; Shinmori, H.; Shinkai, S.; Reinhoudt,
D. N. J. Chem. Soc., Perkin Trans. 2 1998, 2585.
(13) Yamasaki, S.; Ohashi, Y.; Tsutsumi, H.; Tsujii, K. Bull. Chem. Soc.
Jpn. 1995, 68, 146.
(14) Jang, W.-D.; Jiang, D.-L.; Aida, T. J. Am. Chem. Soc. 2000, 122,
3232.
(1) For a recent review of low-molecular-mass organic gelators, see:
Terech, P.; Weiss, R. G. Chem. ReV. 1997, 97, 3133.
(15) Inoue, K.; Ono, Y.; Kanekiyo, Y.; Ishi-I. T.; Yoshihara, K.; Shinkai,
S. J. Org. Chem. 1999, 64, 2933.
(16) Maitra, U.; Kumar, P. V.; Chandra, N.; D’Souza, L. J.; Prasanna,
M. D.; Raju, A. R. Chem. Commun. 1999, 595.
(17) Hanabusa, K.; Miki, T.; Taguchi, Y.; Koyama, T.; Shirai, H. J.
Chem. Soc., Chem. Commun. 1993, 1382.
(2) Bhattacharya, S.; Acharya, S. N. G. Chem. Mater. 1999, 11, 3121.
(3) Aoki, M.; Nakashima, K.; Kawabata, H.; Tsutsui, S.; Shinkai, S. J.
Chem. Soc., Perkin Trans. 2 1993, 3, 347.
(4) Terech, P.; Ostuni, E.; Weiss, R. G. J. Phys. Chem. 1996, 100, 3759.
10.1021/ja0016811 CCC: $19.00 © 2000 American Chemical Society
Published on Web 11/10/2000

11680 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Menger and Caran
organic solvents.24 All of these systems can be considered
“physical gels” in that noncovalent intermolecular associations
are responsible for gelation.25
Gels are semirigid colloidal systems rich in liquid. Most gels
consist of long fibers that have been self-assembled as a result
of the usual array of supramolecular forces (hydrogen-bonding,
aromatic stacking, electrostatics, etc.). Noncovalent cross-links
among the fibers and/or mechanical entanglements create a
three-dimensional network. Solvent is entrapped within the
interstices, thereby imparting rigidity to the system. Other gel
morphologies, such as densely packed vesicles,26-28 have also
been identified.
Two not entirely distinct theories have been advanced to
explain the mechanism of the sol-to-gel transition that occurs,
for example, upon cooling a hot gelator solution. The first,
championed by Bradford in the 1920s,29 maintains that gelation
is a type of incomplete crystallization (the gel consisting of
microcrystalline forms surrounded by solvent). Alternatively,
a gel may be formed by noncrystalline aggregates which, as
mentioned, entrain the dispersing medium in the capillary spaces
between them.
Hoffman gels remained unexplored by modern instrumentation
until, in 1978, we applied 13C NMR to the system.36 It was
demonstrated from spin-lattice relaxation times (T1) and from
line widths that highly concentrated (0.3-0.6 M) gelated 1 exists
as two distinct molecular types: (a) a species that possesses T1
values and line widths expected for a monomer and (b) an
aggregated species whose 13C resonances are broadened to the
point of unobservability by dipolar interactions at these relatively
high gelator concentrations. The results are best interpreted in
terms of soluble monomer, entrapped within a microcrystalline
network, that neither exchanges with the network (on the NMR
time scale) nor experiences difficulty moving about the small
aqueous domains.
Seventeen years went by, and then, in the course of
synthesizing various aroyl-L-cystines, crystals of X-ray quality
were obtained from di(p-toluoyl)-L-cystine.37 We asked of our
crystals (as did others38) the perennial question of whether solid-
state properties actually reflect solution properties. We were
inclined to answer affirmatively because, for one thing, the
crystal showed a “fibrous” molecular orientation. But there was
another reason for taking the X-ray structure seriously. By
applying Occam’s razor and assuming a relationship between
crystal and gel, we could examine the intermolecular forces
within the crystal and, thereby, design an even better gel by
enhancing those forces. If, on the other hand, we had assumed
that the crystal structure was irrelevant to the gel, then the future
experimental course of action would have been less motivated.
In any event, the strategy succeeded. Without reproducing the
original X-ray picture, let us summarize the thought sequence,
derived therefrom, that led us to the design and synthesis of
our powerful new gelators.
The thermal stability of a gel may be probed by observing
spectroscopic features (NMR,3,12,16 UV,16,20 CD,8,9,12 fluores-
cence8,9), thermal properties (DSC3,5,10,11), and physical state3,5,11,30
over a range of temperatures. Gel-to-sol transition temperatures
(TGS) and the (not necessarily identical) sol-to-gel transition
temperatures (TSG) are obtained by these means. A gel’s
macroscopic physical characteristics are quantified rheologically
from a sample’s response to an oscillating stress.1,21,26-28,31-33
It is now possible to return to our amino acid-based gelators
that rigidify water at remarkably low concentrations. The story
begins in 1921, when Gortner and Hoffman discovered that
dibenzoyl-L-cystine (1) forms a strong aqueous gel.34,35 It was
an amazing discovery. Only 0.2% 1, corresponding to ap-
proximately 12 000 waters/gelator molecule, suffices to create
a hydrogel.
Typically, the gels are prepared by dissolving 1 in 5 mL of
hot ethanol and then diluting to 100 mL with water. Gortner-
The X-ray picture showed a linear stacking of the di(p-
toluoyl)-L-cystine molecules in which each unit is connected
to the one above and below it by two hydrogen bonds each.39
As can be seen in the structure below, amide-NH’s serve as the
hydrogen-bond donors, whereas carboxyl carbonyls serve as the
hydrogen-bond acceptor. Since the aromatic rings in the
“fibrous” array are situated one above another, π/π stacking
might further contribute to the fiber’s stability. In any event, it
was apparent that if a similar packing existed in the gel, then a
stronger gel could be produced by accentuating the hydrogen-
bonding acceptor capabilities of the carbonyl. This could be
readily accomplished by converting the R-carboxyl into a
primary R-carboxamide. Similarly, the amide-NH would become
a better hydrogen-bonding donor via the presence of electron-
withdrawing groups on the aromatic ring.
(18) Newcome, G. R.; Baker, G. R.; Arai, S.; Saunders, M. J.; Russo, P.
S.; Theriot, K. J.; Moorefield, C. N.; Rogers, L. E.; Miller, J. E.; Lieux, T.
R.; Murray, M. E.; Phillips, B.; Pascal, L. J. Am. Chem. Soc. 1990, 112,
8458.
(19) Newcome, G. R.; Baker, G. R.; Saunders, M. J.; Russo, P. S.; Gupta,
V. K.; Yao, Z.-q.; Miller, J. E.; Bouillion, K. J. Chem. Soc., Chem. Commun.
1986, 752.
(20) Kimura, T.; Shinkai, S. Chem. Lett. 1998, 1035.
(21) Clausen, T. M.; Vinson, P. K.; Minter, J. R.; Davis, H. T.; Talmon,
Y.; Miller, W. G. J. Phys. Chem. 1992, 96, 474.
(22) Furhop, J.-H.; Schnieder, P.; Rosenberg, J.; Boekema, E. J. Am.
Chem. Soc. 1987, 109, 3387.
(23) For gemini surfactants, see: Menger, F. M.; Littau, C. J. Am. Chem.
Soc. 1991, 113, 1451. Menger, F. M. Keiper, J. S. Angew. Chem., Int. Ed.
2000, in press.
(24) Oda, R.; Huc, I.; Candau, S. J. Angew. Chem., Int. Ed. 1998, 37,
2689.
(25) A chemical gel, on the other hand, is defined as one in which the
gelator aggregates are held together by covalent cross-links, trapping the
liquid phase within a non-thermoreversible three-dimensional polymeric
structure.
(26) Gradzielski, M.; Bergmeier, M.; Muller, M.; Hoffman, H. J. Phys.
Chem. B 1997, 101, 1719.
(27) Hoffman, H.; Thunig, C.; Schmiedel, P.; Munkert, U. Faraday
Discuss. 1995, 101, 319.
(33) Aggeli, A.; Bell, M.; Boden, N.; Keen, J. N.; McLeish, T. C. B.;
Nyrkova, I.; Radford, S. E.; Semenov, A. J. Mater. Chem. 1997, 7, 1135.
(34) Gortner, R. A.; Hoffman, W. F. J. Am. Chem. Soc. 1921, 43, 2199.
The gelation of dibenzoyl-L-cystine had been noted earlier [Brezinger, Z.
Phisiol. Chem. 1892, 16, 537], but Gortner and Hoffman, who reported
observing this gel before consulting this earlier work, were the first to study
its properties.
(35) Wolfe, C. G. L.; Rideal, E. K. Biochem. J. 1922, 16, 548.
(36) Menger, F. M.; Venkatasubban, K. S. J. Org. Chem. 1978, 43, 3143.
(37) Menger, F. M.; Yamasaki, Y.; Catlin, K. K.; Nishimi, T. Angew.
Chem., Int. Ed. Engl. 1995, 34, 585.
(28) Hoffman, H.; Thunig, C.; Schmiedel, P.; Munkert, U. Langmuir
1994, 10, 3972.
(29) Bradford, S. C. Biochem. J. 1921, 15, 553.
(30) Tan, H.-M.; Moet, A.; Hiltner, A.; Baer, E. Macromolecules 1983,
16, 28-34.
(31) Ikeda, S.; Foegeding, E. A.; Hagiwara, T. Langmuir 1999, 15, 8584.
(32) Aggeli, A.; Bell, M.; Boden, N.; Keen, J. N.; Knowles, P. F.;
McLeish, T. C. B.; Pitkeathly, M.; Radford, S. E. Nature 1997, 386, 259.
(38) Ostuni, E.; Kamaras, P.; Weiss, R. G. Angew. Chem., Int. Ed. Engl.
1996, 35, 1324.
(39) 2 was recrystallized from methanol/ethyl acetate, see ref 37.

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11681
Table 1. Derivatives of L-Cystine
R
R′
R′′
1
2
OH
OH
H
H
benzoyl
p-toluoyl
3
4
5
OH
NH2
NH2
H
H
H
p-nitrobenzoyl
p-nitrobenzoyl
benzoyl
The consequences of these modifications, plus those of several
other modifications, is the focus of the ensuing discussion. Table
1 lists our 14 L-cystine derivatives, and Scheme 1 summarizes
their synthesis. Although all of these compounds are aromatic
derivatives, we have also examined the diacetyl derivative
(which fails to form a gel) and the dioctanoyl derivative (which
does gelate).40
6
7
8
9
10
11
12
13
14
NH2
NH2
NH2
NH2
H
H
H
H
H
H
H
CH3
H
p-toluoyl
p-anisoyl
3,5-dimethoxybenzoyl
3,5-dinitrobenzoyl
2-naphthoyl
benzoyl
benzoyl
benzoyl
benzoyl
NH2
NHCH3
N(CH3)2
OH
OCH3
Results and Discussion
Synthesis of the Gelators. Access to our colloidal systems
was obviously made possible through the intervention of
synthetic organic chemistry. We are indebted to the discipline
because only with the aid of organic synthesis can key
relationships between colloidal activity and molecular structure
ever be established.
purity of all final products were verified by elemental analysis,
1
MS, H and 13C NMR after the solids were dried in a vacuum
oven over P2O5. Full details are given in the Experimental
Section.
General Characteristics. Since the compounds in Table 1
are water-insoluble, a water-miscible cosolvent was required
to prepare the aqueous gels. Ethanol could not be used as a
cosolvent (as it had been in the past with carboxylic acids)
because the corresponding amides in Table 1 do not dissolve,
even in hot ethanol. All the L-cystine derivatives are, however,
soluble in DMSO, which was therefore used throughout our
study as a cosolvent and as a dispersant that allows the gelator
molecules to spread throughout the water prior to fiber forma-
tion. As will be seen, various concentrations of DMSO (5-
25% v/v) were examined in order to assess the effect of
cosolvent upon gelation. Typically, the compounds in Table 1
were dissolved in warm DMSO, hot water was added, and
cooling was allowed to take place from about 90 °C to room
temperature. It may be no accident that this gelation procedure
resembles a crystallization protocol. Cooling lowers the solubil-
ity and promotes self-assembly into a sparingly soluble network
in the gel case and into an insoluble ordered array in the case
of a crystal.
A general scanning of our 14 potential gelators was carried
out prior to our performing more quantitative measurements. It
should be stated at the outset that having 14 compounds on
hand gave us great versatility with regard to understanding the
effects of structure upon gelation. Unfortunately, it also gives
the reader a collection of organic structures which, obviously,
cannot be kept continuously in mind. Thus, the reader will be
forced to refer frequently to Table 1 as we mention one
particular gelator or the other. In any event, Table 2 gives the
gross appearance of the gels made in 95% H2O/5% DMSO.
The “properties” column classifies the gels into such categories
as G (stable to inversion of the container) and J (a jelly unstable
to inversion). The “concentration” column gives the approximate
minimum gelator concentration necessary for gelation. As can
be seen, gelator 10 (the naphthoyl amide) gelates at only 0.25
mM. We know of no other low-molecular-weight gelator that
competes with this number. In the “appearance” column, we
reveal whether a gel is clear, translucent, or opaque at the
minimum gel concentration. Finally, the “gel time” column lists
the approximate time required for the minimum gelator con-
centration to gelate after mixing of the components at about 90
°C. Gelator 10 rigidifies water in less than 30 s, while most of
Preparation of L-cystine derivatives 1,41 2,37 and 3 (Table 1,
Scheme 1) was accomplished by acylation under Schotten-
Bauman conditions. Primary amide derivatives 4-10 were
obtained in two steps. L-Cystine dimethyl ester dihydrochloride
was first reacted with liquid ammonia, yielding intermediate
15 after acidification.42 This was followed by acylation of the
primary amide with an appropriate acyl chloride using the
Schotten-Bauman reaction with sodium acetate as the base for
4 and 5, or with triethylamine in chloroform/DMSO for 6-10.
Secondary amide 11 was prepared in a manner analogous to 5
except that methylamine was substituted for ammonia in the
first step. Despite several attempts, extending the method to
make the tertiary amide with dimethylamine was unsuccessful.
Thus, dimethylamide 12 was synthesized in four steps from
commercial bis(benzyloxycarbonyl)-L-cystine: The diacyl chlo-
ride43 was prepared with PCl5 in ether at 0 °C; the solution was
subsequently condensed with dimethylamine in ether at -78
°C. Deprotection of the resulting dimethylamide 17 with HBr
in acetic acid afforded diamine salt 18, which in turn was
benzoylated to 12. L-Cystine derivative 13 (the only member
of Table 1 with a tertiary R-amido group) was obtained by the
reduction of R-(-)-thiazolidine-4-carboxylic acid with sodium/
liquid ammonia followed first by air oxidation to the disulfide
(19)44 using catalytic iron(II) sulfate and then by Schotten-
Bauman benzoylation.45 Derivative 14, the only ester in the
group, was made by benzoylation of L-cystine dimethyl ester
dihydrochloride.
Except for derivatives 4 (from which X-ray-quality crystals
were obtained after chromatography on a silica column with
ether as the eluant) and 12 (which was recrystallized from
MeOH/H2O), the compounds in Table 1 were purified by
trituration in hot methanol, water, or ether. The identity and
(40) McSorley, K. S. Master’s Thesis, Emory University, Atlanta, GA,
1991.
(41) Martin, T. A. J. Med. Chem. 1969, 12, 950.
(42) Martin, T. A.; Causey, D. H.; Sheffner, A. L.; Wheeler, A. G.;
Corrigan, J. R. J. Med. Chem. 1967, 10, 1172.
(43) Gustus, E. L. J. Org. Chem. 1967, 32, 3425.
(44) Keller-Schierlein, W.; Mihailovie, M. Lj.; Prelog, V. HelV. Chim.
Acta 1959, 26, 305.
(45) Bloch, K.; Clarke, H. T. J. Biol. Chem. 1938, 125, 275.

11682 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Scheme 1. Synthesis of l-Cystine Derivatives
Menger and Caran
Table 2. Behavior of L-Cystine Derivatives in Water/DMSOa
compd propertiesb concn (mM)c appearanced
gel time
1
2f
3
4
5
6
7
8
9
G
R
R
3.0
C
2-3 h
G,Jg
G
G
G
Ph
J
2.0
0.5
0.5
2.0
T
C
C
O
5-10 min
3-5 min
1-2 min
5-10 min
10i
11
12
13
14
G
G
R
P
G
0.25
2.0
C
T
<30 s
20-30 min
2.0
O
3-5 min
Figure 1. Three gels composed of 5 (0.5, 1.0, and 3.0 mM, left to
right) which are stable when inverted as shown (90/10 DMSO/water).
Note the increased opacity of the 3.0 mM gel.
a
The gelation was tested in 95% H2O/5% DMSO unless otherwise
noted. b G: forms a gel which is stable to inversion. J: forms a gel-
like solid (“jelly”) which is unstable to inversion. P: precipitates. R:
recrystallizes. Minimum concentration required for gelation. d Ap-
c
Returning now to Table 2, we summarize below the structure/
activity relationships derived therefrom. For ease of discussion,
the terms “acid” or “amide” will be used according to the
derivitization state of the L-cystine R-carboxyls (R in Table 1).
All our compounds of course possess, additionally, two aromatic
amides (R′′ in Table 1).
(a) Since ester 14 can form gels, the carboxyl proton in the
parent compound 1 is not an essential intermolecular hydrogen-
bonding site within the gel network.
(b) All other successful gelators (4-7, 10, and 11) are amides
of the carboxyl group (R ) amino in Table 1). They gelate at
lower concentrations (0.25-2.0 mM) and at faster times (from
<30 s to ∼30 min) than the parent carboxylic acid 1. Since an
amide carbonyl is a stronger hydrogen-bonding carbonyl than
a carboxyl carbonyl, we surmise (as discussed in the Introduc-
tion) that the carbonyl oxygen serves as a key hydrogen-bonding
pearance of gel at minimum gel concentration. C: clear; little or no
appearance of a solid phase. T: translucent; some small crystallites or
slightly opaque regions. O: opaque. e Approximate time required for
gelation at minimum concentration after mixing components at ∼90
°C (allowing them to cool to room temperature in air). f From ref 37.
g Compound 4 displayed inconsistent behavior. h Precipitated as fibrous
balls (see text). i 25% DMSO was required to prepare gels from 10
(see text).
the other amides do so in minutes. Increasing the gelator
concentration can substantially reduce these gelation times even
further.
Figure 1 offers a visual respite from the tabular information
in Table 2. Three gels of 5 (0.5, 1.0, and 3.0 mM in 90% H2O/
10% DMSO) are all seen to be stable to inversion. The gel
becomes opaque only at the highest concentration.

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11683
unit in the self-assembly of the gelating fibers. One must also
recognize that the amides tend to be less water-soluble than
the parent acid. This will promote both the rate and extent of
gelation among the amides.
(c) Comparison of 5, 11, and 12 shows that having two
methyls (but not one) on the R-carboxamide group nitrogen
destroys gelation. Unless an R-carboxamide proton is necessary
for hydrogen-bonding (and there is no independent evidence
for this), the two methyl groups may simply be exerting a
deleterious steric effect.
(d) Compound 10, with primary amides and with naphthoyls
on the R-amino group, is the best gelator in the series. In fact,
10 gelates so rapidly (even before all the hot water could be
added) that it was necessary to prepare the gel in 25% rather
than 5% DMSO (the only gelator so handled in Table 2). Even
with this modification, 10 forms a clear gel in a matter of
seconds. One is tempted to conclude that π/π stacking of the
large naphthalene groups stabilizes the gel fibers. Low water
solubility of 10 no doubt also plays a role.
(e) Placing methyl or nitro groups on 1’s aromatic rings
converts 1 from a gelator into crystal-forming compounds 2
and 3. Even a seemingly innocuous p-methyl group is seen to
affect the gel/crystal relationship. The nondependence of the
gel/crystal relationship on the electron-withdrawing ability of
the aromatic substituent, and thus the NH hydrogen-bonding
acidity, demonstrates the complexity of factors dictating crystal-
linity.
(f) In contrast to the acids, altering the aromatic rings of amide
5 did not necessarily destroy their gelating ability. Substituents
on the aroyl groups of the primary amides (i.e., p-NO2 in 4;
p-CH3 in 6; p-OCH3 in 7; 3,5-di-OCH3 in 8; and 3,5-di-NO2 in
9) show no obvious trend compared to that of unsubstituted 5.
Thus, as seen in Table 2, neither the highly electron-rich 8 nor
the electron-poor 9 is a good gelator (forming a precipitate and
a fluid “jelly”, respectively). This was surprising since we had
expected from our X-ray data (see Introduction) that the aroyl
amide NH protons were engaged in hydrogen-bonding and that,
therefore, the nitro groups should be gel-strengthening. However,
it must be borne in mind that there is a delicate balance between
gelation and crystallization in accord with Bradford’s “incom-
plete crystallization” theory. Apparently, a relatively minor
structural change can tip the balance in an unpredictable manner.
albeit nondefinitive, insight into how the L-cystine derivatives
might self-assemble. X-ray-quality crystals from a water/DMSO
system, which would perhaps have given a greater insight into
gel fiber structure, could not be obtained. X-ray structures of
two polymorphic forms of 3 are given in Figure 2. These two
crystal forms were found side-by-side in chromatography
columns and could be separated physically. The structure in
Figure 2a is similar to that obtained for 2 in an earlier paper.37
Thus, the molecules stack linearly with the aid of hydrogen
bonds between the carboxyl carbonyl and the aromatic amide
NH. A donor/acceptor combination on one side of the S-S
linkage hydrogen-bonds to a lower molecule, while the pair on
the other side hydrogen-bonds to an upper molecule. Carboxylic
acid protons are involved in interfiber hydrogen-bonding with
aromatic amide carbonyls (see Supporting Information for
details). The aromatic rings of two adjacent molecules are
parallel to each other but oriented with the nitro groups displaced
60° from one another. It was this general structure that led us
to convert the R-carbonyls into R-carboxamides in order to
beneficially enhance the accepting ability of the carbonyl.
One motivating reason for testing the nitro compound was
that nitros should enhance the hydrogen-bonding acidity of the
critical NH protons. This turned out to be the case. The average
N(H)‚‚‚O distance in the solid state, according to the X-rays, is
3.085 Å for toluoyl derivative 2 but only 2.936 Å for
nitrobenzoyl derivative 3. As it turned out, however, the nitros
do not improve the gels but, instead, lead to crystallization. A
major problem in gel design, therefore, is to make structural
alterations that favor gelation without inadvertently shifting the
propensity to self-assemble into the crystalline domain.
A polymorph of 3 in Figure 2b has a rather different packing.
In this case, both aromatic amides of a molecule provide
carbonyls to hydrogen-bond with NH protons in the unit directly
below it. Concurrently, the same aromatic amides of the
molecule donate NH protons to amide carbonyls in the unit
directly above it. Unlike the situation in Figure 2a, the
R-carboxyl carbonyls are engaged only in weak hydrogen bonds
to R-CH protons. A single ether molecule per molecule of 3
(not shown) joins two fibers lying side-by-side. Polymorphism
in the crystalline state emphasizes the ever-present possibility
that the gels themselves are composed of polymorphic struc-
tures.46
It is possible, of course, that a third packing mode character-
izes the gel fibers of the compounds in Table 2. But assuming
for simplicity that this is not the case, then both morphologies
in Figure 2 demand the presence of the aromatic amide NH
proton. One would predict, therefore, that substitution of this
proton by a methyl would be fatal to gel formations. It turns
out that such a compound (13) does indeed fail to form gels.
Unfortunately, the results are ambiguous because, according to
the NMR, 13 exists in two s-cis/s-trans isomers of roughly equal
concentration (see below). Judging from the fact that DL-cystine
derivatives do not gelate (unpublished observations),40 one
cannot exclude the possibility that the presence of a mixture
suffices in and of itself to prevent gelation.
To summarize: Gelation is undoubtedly the product of
intermolecular forces, particularly hydrogen-bonding. Since it
is known from previous work that replacing -S-S- with
-CH2CH2- destroys gelation,37 we presume that an imposed
R-S-S-R dihedral angle of 90° favors hydrogen-bonding
contacts (a presumption borne out by X-ray data to follow).
Two factors impede gelation: water solubility and crystallinity.
Thus, as mentioned, the highly water-soluble diacetyl-L-cystine
does not gelate. Water insolubility is promoted by aroyl groups,
which might also contribute π/π stacking to the stability of the
fibers. That the latter is not essential is shown by the fact that
the dioctanoyl-L-cystine is a good gelator. Hydrophobic forces
may, in addition to the water-insolubilizing effect of the octanoyl
groups, be playing a role here. Crystallinity, the second enemy
of gelation, depends on complex solid-state forces that are
difficult to define. The best gelators are compounds which,
although water-insoluble, are not so insoluble that they pre-
cipitate or crystallize rapidly. Under nonequilibrium constraints,
the gelators are able to grow into long, linear arrays that remain
dispersed within the water.
Microscopy. We now turn to the microscopic properties of
the gels. Gel fibers (dried by graded dehydration into ethanol
followed by critical point drying from liquid CO2) were coated
X-ray Crystallography. The ability to recrystallize 3 from
ether, and to thereby obtain X-ray structures, provides a valuable,
(46) Furman, I.; Weiss, R. G. Langmuir 1993, 9, 2084.

11684 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Menger and Caran
Figure 2. ORTEP diagrams (30% probability) of recrystallized 3 (from diethyl ether), demonstrating the polymorphic nature of the structure. (a)
Neat crystals, similar to those of 2,37 arranged with a cross-hydrogen-bonding pattern between the acid carbonyl and the amide proton. N(H)‚‚‚O
distances (and NsHsO angles) alternate between 2.987 (161.1°) and 2.885 Å (149.2°). Aromatic rings are parallel, with nitro groups on adjacent
molecules displaced 60° from one another. Acid proton-amide carbonyl hyrdogen-bonding occurs between fibers (see Supporting Information).
(b) Alternative crystalline molecular organization, incorporating one equivalent of ether (H-bonded to acid protons, not shown). Note intermolecular
hydrogen-bonding between aromatic amide carbonyls and aromatic amide NH. N(H)‚‚‚O distance is 3.093 Å (NsHsO angle ) 154.7°); C(H)‚‚‚O
distance is 3.259 Å (CsHsO angle ) 158.2°).
with Au/Pd and observed by conventional scanning electron
microscopy (SEM). Some of the gels contracted slightly (∼10%)
during dehydration. All of the gel-forming derivatives showed
fibrous aggregates of varying morphology ranging from ∼50
to 300 nm in width (Figure 3). Filaments from a dried 1 mM
sample of 5 are shown in Figure 3a,b. The fibers tend to be
roughly linear over moderate distances and randomly oriented.
Fibers of 7, on the other hand, are less defined and more
interconnected (Figure 3c). Gels of 7 also contain fibrous
clusters, 300 nm wide and 2-4 µm long, which may account
for the relative opacity of this gel (Figure 3d). Since clusters
are not as effective in entraining solvent as are individual fibers,
one also would expect 7 to have a higher minimum gelation
concentration, as is the case (Table 2).
Cryo-high-resolution SEM permits a glimpse at this aqueous
colloidal system without the prior removal of the liquid phase.47
Thus, gels of 5 (3 mM, 10% DMSO) were frozen in liquid
ethane, fractured, sputter-coated with 1 nm of chromium, and
observed in the upper stage of a dual-stage field-emission SEM.
The resulting fibers are quite similar in appearance and size to
those recorded from the dried gels. Occasional regions of long-
range order, where fibers are roughly parallel, are evident
(Figure 4).
(Figure 5b). Thus, 8 retains the propensity to self-assemble into
fibers; its failure to gelate stems mainly from a collective
insolubility of non-independent fibers. Fine-tuning of gel
preparations (including varying H2O/DMSO ratios) could not
coerce 8 to form a gel.
Simple light microscopy of 5 (75/25 H2O/DMSO) in Figure
6 reveals another important characteristic of the gels: two
different fiber morphologies are evident. One type of fiber is
thin and linear, whereas the other type of fiber is thick and
curved. Both seem to radiate from central points.8,9 In view of
the two crystal forms (Figure 2), polymorphic aggregation is
perhaps not surprising.
Rheology.48 Let us summarize up to this point. Fourteen
potential L-cystine-based gelators were synthesized. The com-
pounds were assessed according to their gelation ability,
minimum gelation concentration, and gelation time. Structure/
activity relationships among the diverse compounds, coupled
with suggestive X-ray data, led to reasonable speculations about
the molecular associations accompanying the self-assembly into
fibers. That fibers do, in fact, form was shown clearly by
electron microscopy.
Now, the present paper is entitled “Anatomy of a Gel”. If
we are to live up to the broad sweep of this title, then the
behavior of the fibers must be more fully described. What
happens to fibers when they are exposed to mechanical stress?
What does one see by electron microscopy with derivative
8sa compound that does not gelate water, but instead forms
insoluble clumps (Figure 5a)? At higher magnification, these
clumps show fine fibers emanating from central nucleation sites
(48) Barnes, H. A.; Hutton, J. F.; Walters, K. An Introduction to
Rheology; Elsevier Science B.V.: Amsterdam, 1989; pp 37-54.
(49) Shusterman, A. J.; McDougal, P. G.; Glasfield, A. J. Chem. Educ.
1997, 74, 1222.
(47) Apkarian, R. P.; Caran, K. L.; Robinson, K. A. Microsc. Microanal.
1999, 5, 197.

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11685
Figure 3. SEM images of fibers obtained from the dehydration and critical point drying of gels of H2O/DMSO (95/5) gels. (a) Relatively straight
fibers from a 1 mM gel of 5 (bar ) 2 µm). (b) Higher magnification of fibers from the same gel (bar ) 300 nm). (c) Less ordered fibers from a
2 mM gel of 7 (bar ) 2 µm). (d) A cluster of fibers from the sample in (c). These aggregates may account for the opacity of gels of 7 (bar ) 1
µm).
specified torque. The torque is converted directly into “stress”
(expressed in pascal units). A position sensor on the oscillating
shaft measures the amplitude of the gel’s deformation to give
the unitless “strain”. Gel deformation under the applied stress
is assumed to be free from turbulence.
When a sinusoidal shear stress is applied to an ideal Hookean
solid, the resulting strain will be in phase (i.e., the phase angle
) 0°). For ideal Newtonian liquids, stress and strain will be
90° out of phase. All real materials are “viscoelastic”, with phase
angles between 0° and 90°.
The complex modulus G* is defined as the ratio of the
amplitudes of stress/strain in an oscillatory experiment. G* is
comprised of two useful components: (a) G′, the “storage” (or
“elastic”) modulus which represents the ability of the deformed
Figure 4. Cryo-HRSEM image of a frozen hydrated 3 mM gel of 5
material to “snap back” to its original geometry, and (b) G′′,
(10% DMSO). Fibers range in size from ∼50 to 300 nm (bar ) 1
the “loss” (or “viscous”) modulus which represents the tendency
µm).
of a material to flow under stress. The rheometer automatically
gives G′ and G′′ in units of pascals as calculated from the
measured phase angle (δ) according to the equations below.
Clearly, G′ ) G* for an ideal solid (δ ) 0°), and G′′ ) G* for
an ideal liquid (δ ) 90°). Since δ < 10° for most of our gels,
they have large G′ values and can be considered, therefore, rather
“solid-like”.
How does gelator concentration affect gel rigidity? How do the
various gels compare with regard to their “yield stress”? What
are the effects of temperature and cosolvent upon fiber dynam-
ics? Such questions fall into the province of rheology, a subject
which we now confront. Although textbooks have warned that
“rheology is not an easy branch of science”, we felt that useful
rheological information was within reach.
Our work used a Bohlin controlled-stress rheometer employ-
ing a cone-and-plate configuration. A thin layer of gel is placed
between a round flat plate at the bottom and a round conical
plate (40 mm diameter), fixed to a rotatable shaft, at the top.
The cone rotates back and forth at 0.1-10 Hz and at a constant
G′ ) G* cos δ
G′′ ) G* sin δ
Gels were subjected to a nondestructive frequency sweep in
which an initial stress of 3 Pa was allowed to adjust to a constant

11686 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Menger and Caran
Figure 5. SEM images of a dendritic, fibrous precipitate obtained from an attempted gelation of 8. (a) A cluster (bar ) 100 µm). (b) Higher
magnification of one member of the cluster reveals its fibrous nature (bar ) 25 µm).
Figure 6. Light micrograph of a 1.0 mM gel of 5 (75/25 H2O/DMSO).
Note two different fiber morphologies: (1) thick and curved (ar-
rowheads) and (2) thin and linear (arrows), suggesting polymorphic
aggregation.
strain of 0.001. As can be seen in Figure 7a, a gel of 1 mM 6
(10% DMSO) has G′ and G′′ parameters that are virtually
independent of the oscillation frequency. This is true of the other
gels as well. Note that G′ is an order of magnitude greater than
G′′ (as is the case for all the gels), demonstrating the dominant
elastic behavior of the systems. Figure 7b plots G′ and G′′ vs
Figure 7. Viscoelastic properties of a 1 mM gel of 6 (10% DMSO).
the imposed stress (σ) for the same gel using a 1-Hz oscillation.
(a) Frequency sweep using a constant target strain of 0.001. Note that
At a so-called “yield stress” (designated σy) of 135 Pa, the gel
G′ (() is an order of magnitude greater than G′′ (9), both values
breaks under the applied force and begins to flow. Each gel
remaining essentially constant from 0.1 to 10 Hz. (b) The amplitude
has its own particular σy according to its strength, as will be
sweep (f ) 1 Hz) shows only a small dependence of G′ and G′′ on the
discussed later.
stress amplitude until the yield stress (σy, 135 Pa) is reached. Above
Plots of G′ vs [gelator] for gels in 75/25 H2O/DMSO (Figure
8a) and in 90/10 H2O/DMSO (Figure 8b) contain a wealth of
information. (Such plots were obtained for all our gelators, but
for the sake of digestibility we are focusing on only a few
representative compounds.) (a) Comparison of 5 in the two
solvent systems shows that the elastic modulus at a constant
concentration increases with diminishing cosolvent. This is
intuitively reasonable. Thus, since the gelator is soluble and
nongelating in DMSO, the presence of DMSO must adversely
affect the integrity of the fibers. Whether the fibers are fewer
in number, shorter, or thinner is not known. (b) Rather minor
structural variations can have a large impact on gel rigidity.
For example, amide 5 is a far better gelator than the corre-
sponding acid 1 (as we had surmised qualitatively in Table 2
and now show quantitatively in Figure 8b). Indeed, compound
1, the classical gelator, fails to form a viable gel at concentra-
the yield stress, gels break and begin to flow.
tions up to 20 mM in 75/25 H2O/DMSO. Figure 8b supports a
gelation model, such as Figure 2a, in which a carbonyl oxygen
plays a prominent role in fiber formation. (c) The case of
compound 10 is instructive. As can be seen in Figure 8a, its G′
values, which vary from 51 to 1600 Pa, fall on the baseline
when a scale adequate for the other gelators is used. Yet,
according to Table 2, 10 is the best gelator with regard to
minimum gel concentration (0.25 mM) and gelation time (<30
s). What is the origin of this apparent disparity? Why is it that
although 0.25 mM suffices for 10 to form a physically stable
gel (i.e., one that will not flow out of an inverted beaker), the
gel has, in fact, limited elastic strength when investigated
rheologically?
The answers may lie in the kinetics. Thus, gels were prepared

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11687
Figure 9. Example of transition temperature (TGS) as determined by
temperature-sweep oscillation rheology (1 mM gel of 5, 25% DMSO).
TGS is given by the temperature at which the elastic modulus, G′ ([),
drops below the viscous modulus, G′′(9), i.e. the gel breaks, under the
applied stress of 3 Pa.
intermediate between the two. It seems unlikely that the p-OCH3
and p-CH3 groups destabilize the gel by adversely affecting the
π/π overlap within the fibers. More plausibly, the substituents
sterically impede the side-by-side alignment of the fibers or,
alternatively, lower the monomer solubility and therefore
accelerate fiber formation and reduce the average fiber length.
Whatever the correct explanation, the remarkable sensitivity of
gelation to small structural changes is a noteworthy feature of
our physical gels and, perhaps, physical gels in general.
No “anatomy of a gel” is even semicomplete without a
discussion of thermal stability, and we will complete our survey
with this topic. The gel-to-sol transition temperatures (TGS) were
measured rheologically by subjecting an equilibrated gel to a
small, discontinuous (2 s on, 10 s off) oscillating stress (1 Hz,
3 Pa) and slowly increasing the temperature from 25 to 90 °C.
The point at which the loss modulus, G′′, exceeded the storage
modulus, G′ (i.e., where the gel had broken), was recorded as
Figure 8. Elastic modulus (G′) vs gelator concentration for selected
gels from (a) 75/25 H2O/DMSO and (b) 90/10 H2O/DMSO. G′ generally
increases with increasing gelator concentration up to a certain point
(see text). Note that gels of 10 (25% DMSO) are stable even at 0.25
mM concentration.
T
GS. Since TGS depends on the experimental design (the greater
by adding a DMSO solution of gelator to hot water and allowing
the system to cool. With compound 10, gelation is almost
immediate, even at higher temperatures. The speed of fiber
growth may determine the fiber length just as the speed of
crystallization determines crystal size. Compound 10 forms
gelating fibers almost instantly, but the gelating fibers are
probably short, and, as a consequence, the gel is not rheologi-
cally robust. In other words, the propensity to gelate does not
necessarily reflect the quality of the ensuing gel. The fact is
driven home clearly by the behavior of 5 at higher concentra-
tions, where G′ actually decreases (Figure 8a). High concentra-
tions, above a certain limit, tend to accelerate gelation but
diminish the elastic modulus. Gels from the various L-cystine
derivatives can be legitimately compared only in the sense that
they were all prepared under conditions as similar as possible.
The yield stress (σy), where the stress finally becomes
sufficient to break the gel, was mentioned earlier (Figure 7b).
In general, the yield stress increases with concentration (although
the function can be complicated owing, we presume, to factors
mentioned in the previous paragraph). In any event, at 3 mM
gelator (75/25 H2O/DMSO), σy has the following values, in
decreasing order for several derivatives: 5 (>600 Pa, the
rheometer maximum); 6 (380 Pa); 14, 11, and 7 (<40 Pa).
Perhaps the comparison between amides 5, 6, and 7, which differ
only in the para-substituent on the aromatic ring, is the most
illuminating. Gel 7 succumbs to the applied stress at a σy an
order of magnitude smaller than that for gel 5; gel 6 is
the stress, the lower the TGS), TGS values reported here are useful
for comparison purposes only. In any event, a transition is
observed at TGS as both G′ and G′′ decrease sharply and the
gel becomes more “liquid-like” than “solid-like” (Figure 9). It
is at this temperature that the system is no longer sufficiently
robust to resist the applied stress.
Values of TGS for several gelators are plotted as a function
of gelator concentration in Figure 10. Gels 5, 6, 10, and 11, at
only 3 mM concentration, are stable at 90 °C (our highest
temperature). Gels of 10 are stable at 90 °C at an astounding
0.25 mM. For comparison purposes, we have included the
classical dibenzoyl-L-cystine 1 (dotted line), which required 10%
DMSO (rather than the 25% DMSO used for all the others) to
form a gel. Even with this special treatment, gel 1 could not
reach a TGS ) 80 °C at 10 mM. Once again, the dramatic effect
of the R-carboxylic acid-to-R-carboxamide conversion is evi-
dent.
One might think that the sharp breaks in G′ and G′′ vs
temperature plots (Figure 9) provide evidence for sudden phase
changes. Differential scanning calorimetry (DSC) tells us
otherwise. Thus, DSC scans in both directions gave no clear
endothermic or exothermic peaks. These data point strongly to
a continuous gel-to-sol conversion upon heating and a sol-to-
gel conversion upon cooling. The reason that Figure 9 plots
have such sharp breaks is that, upon heating a gel, the gel
suddenly loses its ability to withstand an applied torque,

11688 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Menger and Caran
reaction was allowed to warm to room temperature and run for an
allotted amount of time, after which the filtered precipitate (often
gelatinous) was washed with water. (C) L-Cystine diamide dihydro-
chloride (15), DMSO, CHCl3, and triethylamine were combined and
cooled over an ice bath. The aromatic acyl chloride (neat liquid or a
CHCl3 solution of the solid) was added dropwise to the stirring solution.
The reaction was allowed to warm to room temperature.
N,N′-Dibenzoyl-L-cystine (1).41 Benzoylation Method A. L-Cystine
(Acros, 2.4 g, 0.01 mol), diethyl ether (10 mL), 3% EtOH (31 mL), 2
N NaOH (10.5 + 12.5 mL), benzoyl chloride (Aldrich, 2.5 mL, 0.022
mol), reaction time ∼20 h. The filtered precipitate was washed with
hot water, and dried yielding 1.275 g (28% yield) of white crystalline
flaky solid. mp 178-180 °C (ref 180-181 °C). A commercial
compound (Fluka) was used in some gel preparations.
N,N′-Di(p-nitrobenzoyl)-L-cystine (3). Benzoylation Method A.
L-Cystine (4.94 g, 0.025 mol), Et2O (20 mL + 60 mL), 3% EtOH (60
mL), 2 N NaOH (22 + 28 mL), and p-nitrobenzoyl chloride (Aldrich,
10.2 g, 0.055 mol) were reacted for ∼24 h. The filtered solid product
was extracted with ethyl acetate and dried with Na2SO4. The solvent
was removed to yield 9.754 g of crude dry product (slightly yellow).
The solid was purified by dry-column flash chromatography49 using
ether as the eluant. Two distinct crystal types of the product had formed
(long colorless needles and light yellow compact crystals) which were
found to be randomly distributed throughout the fractions after ∼2 d.
X-ray structures revealed their distinct crystal morphologies (see Results
and Discussion). The colorless needles incorporated one ether of
crystallization, while the yellow crystals were neat: 1.686 g (13% yield);
mp ) 208-209 °C. FAB-HRMS (M - H)-: 537.04061 (calculated:
537.03864). 1H NMR (400 MHz, acetone-d6): δ CH2 (2H, 3.295 ppm,
Figure 10. Average TGS values of several gels, determined rheologi-
cally as described in Figure 9, plotted against gelator concentration.
Arrows indicate that the TGS values lie above 90 °C. Note that gels of
10 are stable at 90 °C, even at an amazingly low 0.25 mM. In contrast,
gel 1 fails to reach a TGS ) 80 °C even at 10 mM and 10% DMSO.
whereupon a “catastrophe” occurs. Nature is full of such events.
When pressure is applied on a paper stapler, nothing happens
until the pressure suddenly exceeds the ability of a staple to
maintain its geometry.
Thus concludes our anatomy. We have used the multiple tools
of organic synthesis, X-rays, electron microscopy, light micros-
copy, rheology, and calorimetry, and, nonetheless, the precise
anatomical features of gels are still elusive. Yet we know a great
deal more about our gels, and gels in general, than when we
began, and this alone justifies the effort. If our gelators, which
function at amazingly low concentrations and high temperatures,
prove to have commercial applications (as may well be the case),
all the better.
3
2
3
dd, J ) 9.6 Hz, J ) 14.0 Hz), δ CH2 (2H, 3.493 ppm, dd, J ) 4.4
2
Hz, J ) 14.0 Hz), δ CH (2H, 5.048 ppm, m), δ Ar-H (4H, 8.118
3
3
ppm, J ) 8.7 Hz), δ Ar-H (4H, 8.273 ppm, J ) 8.85 Hz), δ NH
(2H, 8.467 ppm, 3J ) 8.0 Hz). 13C NMR (acetone-d6): δ 40.499, 53.317,
124.397, 129.699, 131.787, 150.568, 166.313, 171.861 ppm. IR
(KBr): 3700-2800 (br), 1738, 1731, 1704, 1693, 1644, 1599, 1535,
1520, 1486, 1422, 1348, 1306, 1291, 1246, 1216, 1107, 1013, 867,
845, 799, 713 cm-1. Anal. Calcd for C20H18N4O10S2: C, 44.61; H, 3.37;
N, 10.40; S, 11.91. Found: C, 43.74; H, 3.46; N, 9.62; S, 11.74.
Experimental Section
L-Cystine diamide dihydrochloride (15) was prepared according
to the method of Martin et al.42 with minor modifications. NH3 (∼150
mL) was condensed in a 1-L round-bottom flask at -78 °C. L-Cystine
dimethyl ester dihydrochloride (Sigma, 10 g, 0.0293 mol) was added
with stirring. The reaction was allowed to slowly warm to -33 °C.
The NH3 was refluxed on a coldfinger cooled to -78 °C, protected
with a KOH drying tube. After ∼4 h, the NH3 was allowed to evaporate,
yielding a crude yellow solid which was heated to 50 °C in vacuo.
Methanol was added, and the solution was warmed to 60 °C, which
dissolved all but a trace of the crude solid, which was removed by
filtration through cotton. The product was precipitated by acidification
with excess HCl in MeOH. The solid was filtered and washed with
cold MeOH to yield 7.48 g (82% yield) of slightly off-white powder:
mp ) 222-224 °C (lit. mp ) 226.5-227.5 °C). The solid was
General Considerations. Melting points were conducted on a
Thomas-Hoover capillary melting point apparatus and are uncorrected.
Solvents were reagent grade. Reagents were purchased from Aldrich,
Sigma, or Fluka (as noted) and used without further purification except
for benzoyl chloride, which was distilled prior to use. No attempt was
made to optimize yields. Mass spectra measurements were performed
by the Mass Spectroscopy Center (Emory University), electron
microscopy imaging at the Integrated Microscopy and Microanalytical
Facility (Emory University), and NMR spectra measurements (recorded
on a Mercury 300, Inova 400 or Omega 600 instrument as indicated)
at the NMR Center (Emoy University). Residual solvent peaks were
used as NMR reference. X-ray data were collected at room temperature
using Cu KR graphite-monochromated radiation (1.54178 Å) on a
Bruker P4/RA single-crystal X-ray diffractometer (Emory University).
Atlantic Microlab, Inc. (Norcross, GA) performed elemental analyses.
All final products were dried in a vacuum oven (50 °C) over P2O5.
General Methods Used for Benzoylation Reactions. (A) L-Cystine,
2 N NaOH, 3% EtOH, and Et2O were combined in a round-bottom
flask and cooled in an ice bath. The aromatic acyl chloride (neat liquid
or an Et2O solution of the solid) and additional 2 N NaOH were added
dropwise to the rapidly stirring mixture, after which it was allowed to
warm to room temperature. Once the reaction was complete, H2O was
added to dissolve any formed precipitate, and the aqueous layer was
washed several times with Et2O. The product was precipitated by
acidification (1 N HCl) of the warm (50 °C) solution. (B) The
dihydrohalide salt of L-cystine derivative (methyl ester or primary,
secondary, or tertiary amide), anhydrous sodium acetate, water, and
Et2O were combined in a round-bottom flask and cooled in an ice bath.
The aromatic acyl chloride (neat liquid or an Et2O solution of the solid)
was added dropwise to the rapidly stirring mixture. Additional water
or Et2O was sometimes needed to break up the formed precipitate. The
1
recrystallized from 66% aqueous MeOH. H NMR (400 MHz, D2O,
presaturated): δ CH2 (2H, 3.326 ppm, dd, 3J ) 8.4 Hz, 2J ) 15.2 Hz),
3
2
δ CH2 (2H, 3.446 ppm, dd, J ) 4.8 Hz, J ) 15.2 Hz), δ CH (2H,
3
4.414 ppm, dd, J ) 8.4, 4.8 Hz), δ NH2 (7.486, 8.067 ppm).
N,N′-Di(p-nitrobenzoyl)-L-cystine Diamide (4). Benzoylation
Method B. L-Cystine diamide dihydrochloride (15, 1.00 g, 3.21 mmol),
NaOAc (1.053 g, 12.84 mmol), H2O (30 mL), Et2O (5 + 10 mL), and
p-nitrobenzoyl chloride (Acros, 1.49 g, 8.03 mmol) were reacted for
∼12 h. The solid was washed with MeOH, purified by trituration in
hot MeOH, and dried, yielding 0.778 g of a light green solid (45%
yield): mp ) 218-219 °C. FAB-HRMS (M + H)+: 537.0857
(calculated: 537.0862). 1H NMR (400 MHz, DMSO-d6): δ CH2 (2H,
3
2
3.002 ppm, dd, J ) 10.23 Hz, J ) 13.42 Hz), δ CH2 (2H, 3.289
ppm, dd, 3J ) 4.42 Hz, 2J ) 13.58 Hz), δ CH (2H, 4.702 ppm, m), δ
NH2 (4H, 7.286, 7.621 ppm), δ Ar-H (4H, 8.062 ppm, d, J ) 8.86
3
3
Hz), δ Ar-H (4H, 8.274 ppm, d, J ) 8.85 Hz), δ NH (2H, 9.987
ppm, d, J ) 8.09 Hz). 13C NMR (DMSO-d6): δ 171.740, 164.867,
3
149.043, 139.583, 129.054, 123.410, 52.596 ppm (methylene signal

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11689
obscured by solvent peaks). Anal. Calcd for C20H22N6O8S2: C, 44.77;
H, 3.76; N, 15.66; S, 11.95. Found: C, 44.58; H, 3.79; N, 15.50; S,
12.08.
160.217, 166.035, 172.013 ppm. Anal. Calcd for C24H30N4O8S2: C,
50.87; H, 5.34; N, 9.89; S, 11.32. Found: C, 50.77; H, 5.33; N, 9.80;
S, 11.43.
N,N′-Dibenzoyl-L-cystine Diamide (5). Benzoylation Method B.
L-Cystine diamide dihydrochloride (15, 1.00 g, 3.21 mmol), NaOAc
(1.053 g, 12.84 mmol), H2O (20 mL), Et2O (10 mL), and benzoyl
chloride (0.84 mL, 7.23 mmol) were reacted overnight. The product
was purified by trituration in hot MeOH, yielding 750 mg of a white
solid (1.68 mmol, 52% yield): mp ) 239-240 °C. FAB-HRMS (M
+ Li)+: 453.12392 (calculated: 453.12427). 1H NMR (300 MHz,
N,N′-Di(3,5-dinitrobenzoyl)-L-cystine Diamide (9). Benzoylation
Method C. L-Cystine diamide dihydrochloride (15, 0.500 g, 1.61
mmol), DMSO (10 mL), CHCl3 (10 + 5 mL), NEt3 (1.2 mL, 8.66
mmol), and 3,5-dinitrobenzoyl chloride (Aldrich, 0.814 mg, 3.53 mmol)
were reacted for ∼4 h. The solution turned pink upon addition of the
acyl chloride. After 3 h at room temperature (the solution appears
tannish-pink), the CHCl3 and excess NEt3 were removed by rotary
evaporation, after which HNEt3 + Cl- was filtered out. DMSO was
removed by Kugelrohr distillation, to reveal a dark brown solid material.
The solid was washed with CHCl3, yielding a tan solid, which was
purified by trituration in MeOH. The crystalline off-white solid was
dried overnight (0.198 g, 0.31 mmol, 19% yield): mp ) 217-219 °C.
3
2
DMSO-d6): δ CH2 (2H, 3.055 ppm, dd, J ) 10.35 Hz, J ) 13.35
Hz), δ CH2 (2H, 3.264 ppm, dd, 3J ) 13.35 Hz, 2J ) 4.35 Hz), δ CH
(2H, 4.713 ppm, m), δ NH2 (2H, 7.244, s), δ NH2, Ar-H, Ar-H (8H,
3
7.423-7.530 ppm, m), δ Ar-H (4H, 7.868, J ) 7.2 Hz, d), δ NH
(2H, 8.589, 3J ) 8.4 Hz, d). 13C NMR (DMSO-d6): δ 172.038, 166.470,
133.986, 131.344, 128.163, 127.499, 52.460, 33.224 ppm. Anal. Calcd
for C20H22N4O4S2: C, 53.80; H, 4.97; N, 12.55; S, 14.36. Found: C,
53.76; H, 5.01; N, 12.45; S, 14.43.
1
FAB-HRMS (M + Li)+: 633.0654 (calculated: 633.0646). H NMR
3
2
(400 MHz, DMSO-d6): δ CH2 (2H, 3.011 ppm, dd, J ) 13.3 Hz, J
2
) 4.4 Hz), δ CH2 (2H, 3.3 ppm, dd, J ) 4.4 Hz), δ CH (2H, 4.741
ppm, m), δ NH2 (2H, 7.312 ppm, s), δ NH2 (2H, 7.713 ppm, s), δ
N,N′-Di(p-toluoyl)-L-cystine Diamide (6). Benzoylation Method
C. L-Cystine diamide dihydrochloride (15, 0.500 g, 1.61 mmol), DMSO
(10 mL), chloroform (10 mL), NEt3 (1.2 mL, 8.66 mmol), and p-toluoyl
chloride (Aldrich, 0.468 mL, 2.59 mmol) were reacted for 8 h. The
volume of the milky white reaction mixture was increased with CHCl3
(∼80 mL). The suspension was filtered, washed with CHCl3, and
allowed to dry overnight. The solid was purified by trituration in hot
MeOH. After drying, it yielded 0.643 g (1.35 mmol, 84% yield): mp
) 227-231 °C. FAB-HRMS (M + Li)+: 481.15722 (calculated:
481.15555). 1H NMR (400 MHz, DMSO-d6): δ CH3 (6H, 2.341 ppm),
4
Ar-H (2H, 8.930 ppm, t, J ) 2.1 Hz), δ Ar-H (4H, 9.035 ppm, d,
4J ) 2.1 Hz), δ NH (2H, 9.435 ppm, d, 3J ) 8.2 Hz). 13C NMR
(DMSO-d6): δ 52.687, 120.922, 127.780, 136.519, 148.019, 162.364,
171.383 ppm (methylene signal obscured by solvent peaks). Anal. Calcd
for C20H18N8O12S2: C, 50.37; H, 5.38; N, 10.68; S, 12.22. Found: C,
50.42; H, 5.00; N, 10.62; S, 12.36.
N,N′-Di(2-naphthoyl)-L-cystine Diamide (10). Benzoylation Method
C. L-Cystine diamide dihydrochloride (15, 0.500 g, 1.61 mmol), DMSO
(10 mL), CHCl3 (10 + 5 mL), NEt3 (1.2 mmol, 8.66 mmol), and
2-naphthoyl chloride (Aldrich, 0.673 g, 3.54 mmol) were reacted for
∼1.5 h. The precipitate was filtered, washed with CHCl3, and dried
overnight, yielding a slightly yellow solid. Trituration in hot MeOH
followed by drying yielded 0.6871 g (1.26 mmol, 78% yield) of white
solid: mp ) 236-237 °C. FAB-HRMS (M + Li)+: 553.15686
(calculated: 553.15558). 1H NMR (400 MHz, DMSO-d6): δ CH2 (2H,
3
2
δ CH2 (2H, 3.049 ppm, dd, J ) 10.4 Hz, J ) 13.3 Hz), δ CH2 (2H,
3.244 ppm, dd, 3J ) 4.3 Hz, 2J ) 13.4 Hz), δ CH (2H, 4.698 ppm, m),
δ Ar-H, NH2 (6H, 7.245 ppm, m), δ NH2 (2H, 7.521 ppm, s), δ Ar-H
3
3
(4H, 7.769 ppm, d, J ) 7.9 Hz), δ NH (2H, 8.502 ppm, d, J ) 8.1
Hz). 13C NMR (DMSO-d6): δ 21.016, 39.99 (somewhat obscured by
solvent peaks), 52.436, 127.560, 128.713, 131.201, 141.252, 166.354,
172.157 ppm. Anal. Calcd for C22H26N4O4S2: C, 55.68; H, 5.52; N,
11.81; S, 13.51. Found: C, 55.50; H, 5.54; N, 11.65; S, 13.59.
N,N′-Di(p-anisoyl)-L-cystine Diamide (7). Benzoylation Method
C. L-Cystine diamide dihydrochloride (15, 0.500 g, 1.61 mmol), DMSO
(8 mL), CHCl3 (8 + 5 mL), NEt3 (1.2 mL, 8.66 mmol), and p-anisoyl
chloride (Aldrich, 0.603 g, 3.53 mmol) were reacted for 5.5 h. The
reaction volume was increased with CHCl3 (25 mL), filtered, and
washed with a large volume of CHCl3. The product was triturated in
methanol and dried, yielding a white solid (760 mg, 1.50 mmol, 93%
yield): mp ) 216-218 °C. FAB-HRMS (M + Li)+: 513.1459
(calculated: 513.1454). 1H NMR (400 MHz, DMSO-d6): δ CH2 (2H,
3
2
3.121 ppm, dd, J ) 10.23 Hz, J ) 13.43 Hz), δ CH2 (2H, 3.329
3
2
ppm, dd, J ) 4.6 Hz, J ) 13.5 Hz), δ CH (2H, 4.798 ppm, m), δ
NH2 (2H, 7.300 ppm, s), δ Ar-H, NH2 (6H, 7.583-7.619 ppm, m), δ
Ar-H (8H, 7.919-8.002 ppm, m), δ Ar-H (2H, 8.483 ppm, s), δ NH
(2H, 8.795 ppm, d, J ) 8.2 Hz). 13C NMR (DMSO-d6): δ 52.588,
3
124.419, 126.801, 127.560, 127.666, 127.757, 128.811, 128.895,
131.353, 132.050, 134.197, 166.536, 172.112 ppm (methylene signal
obscured by solvent peaks). Anal. Calcd for C28H26N4O4S2: C, 61.52;
H, 4.79; N, 10.25; S, 11.73. Found: C, 61.41; H, 4.85; N, 10.15; S,
11.78.
L-Cystine Dimethylamide Dihydrochloride (16). This compound
was prepared using the same method as for 15, substituting NH2Me
for NH3. L-Cystine dimethyl ester dihydrochloride (Sigma, 1.0 g, 2.93
mmol) was dissolved in ∼20 mL of anhydrous methylamine at -78
°C (protected with a CaCl2 drying tube). The methylamine solution
was allowed to warm to its boiling poing (condensed on -78 °C
coldfinger) and react for 3 h, after which the bulk of the excess solvent
was allowed to evaporate. Further evaporation at 50 °C under vacuum
(aspirator) yielded a thick, clear, colorless oil which was dissolved in
methanol and acidified with HCl. Precipitation or crystallization of the
di-HCl salt proved unsuccessful, so the excess solvent was removed,
and the crude yellow oil was carried on to the next step. (Impurity by
1H NMR: NH2Me‚HCl.)
3
2
3.044 ppm, dd, J ) 10.5 Hz, J ) 13.3 Hz), δ CH2 (2H, 3.235 ppm,
dd, 3J ) 4.2 Hz, 2J ) 13.3 Hz), δ CH3 (6H, 3.796 ppm, s), δ CH (2H,
4.684 ppm, m), δ Ar-H (4H, 6.972 ppm, d, 3J ) 8.7 Hz), δ NH2 (2H,
7.223 ppm, s), δ NH2 (2H, 7.503 ppm, s), δ Ar-H (4H, 7.843 ppm, d,
3J ) 8.7 Hz), δ NH (2H, 8.431 ppm, d, 3J ) 8.1 Hz). 13C NMR
(DMSO-d6): δ 52.452, 55.346, 113.367, 126.178, 129.373, 161.678,
165.933, 172.203 ppm. Anal. Calcd for C22H26N4O6S2‚H2O: C, 50.37;
H, 5.38; N, 10.68; S, 12.22. Found: C, 50.42; H, 5.00; N, 10.62; S,
12.36.
N,N′-Di(3,5-dimethoxybenzoyl)-L-cystine Diamide (8). Benzoy-
lation Method C. L-Cystine diamide dihydrochloride (15, 0.500 g, 1.61
mmol), DMSO (10 mL), CHCl3 (10 + 4 mL), NEt3 (1.2 mL, 8.66
mmol), and 3,5-dimethoxybenzoyl chloride (Aldrich, 0.711 g, 3.54
mmol) were reacted overnight. The CHCl3 and excess NEt3 were
removed by rotary evaporation, followed by a Kugelrohr distillation,
which removed the DMSO, revealing a light brown solid. The crude
material was washed with chloroform, triturated in hot MeOH, and
dried to yield 0.703 g (1.24 mmol, 77% yield) of pure white solid:
mp ) 224-225 °C. FAB-HRMS (M + Li)+: 573.1682 (calculated:
N,N′-Dibenzoyl-L-cystine Di(methylamide) (11). Benzoylation
Method B. Crude L-cystine dimethylamide dihydrochloride (16, 2.93
mmol, assuming 100% yield), NaOAc (1.06 g, 12.9 mmol), Et2O (15
mL), and benzoyl chloride (1.50 mL, 12.9 mmol) were reacted for 2.5
h. The crude dry white solid, after being washed with Et2O, was purified
by trituration in hot MeOH. After drying, a pure white solid was
obtained (1.01 g, 72% yield after two steps): mp ) 275-276 °C dec.
1
1
573.1665). H NMR (400 MHz, DMSO-d6): δ CH2 (2H, 3.032 ppm,
FAB-HRMS (M + Li)+: 481.1579 (calculated: 481.1556). H NMR
dd, 3J ) 10.3 Hz, 2J ) 13.2 Hz), δ CH2 (2H, 3.242 ppm, dd, 3J ) 4.3
Hz, 2J ) 13.5 Hz), δ CH3 (12H, 3,774 ppm, s), δ CH (2H, 4.693 ppm,
m), δ Ar-H (2H, 6.639 ppm, s), δ Ar-H (4H, 7.030 ppm, d, 4J ) 2.1
Hz), δ NH2 (2H, 7.242 ppm, s), δ NH2 (2H, 7.521 ppm, s), δ NH (2H,
8.568 ppm, d, 3J ) 8.1 Hz). 13C NMR (DMSO-d6): δ 39.81 (somewhat
obscured by solvent peaks), 52.459, 55.433, 103.307, 105.454, 136.041,
(400 MHz, DMSO-d6): δ CH3 (6H, 2.601 ppm, d, J ) 4.6 Hz), δ
3
3
2
CH2 (2H, 3.034 ppm, dd, J ) 10.1 Hz, J ) 13.4 Hz), δ CH2 (2H,
3.231 ppm, dd, 3J ) 4.6 Hz, 2J ) 13.4 Hz), δ CH (2H, 4.707 ppm, m),
δ Ar-H (4H, 7.448 ppm, t, 3J ) 7.8 Hz), δ Ar-H (2H, 7.530 ppm, tt,
3J ) 7.3 Hz, 4J ) 1.6 Hz), δ Ar-H (4H, 7.870 ppm, dd, 3J ) 7.8 Hz,
3
4J ) 1.5 Hz), δ NH (2H, 8.031 ppm, q, J ) 4.6 Hz), δ NH2 (2H,

11690 J. Am. Chem. Soc., Vol. 122, No. 47, 2000
Menger and Caran
3
8.643 ppm, d, J ) 7.9 Hz). 13CNMR (DMSO-d6): δ 25.867, 39.9
of absolute EtOH precipitated a thick white solid which was removed
by centrifugation (4 °C, 20 000 rpm, 30 min) and washed twice with
50% EtOH. The solid was dissolved in water and lyophilized yielding
1.924 g of fluffy off-white solid. FAB-LRMS (M + H)+: 269.12
(somewhat obscured by solvent peaks), 52.567, 127.591, 128.175,
131.407, 133.910, 166.472, 170.394 ppm. Anal. Calcd for C22H26-
N4O4S2: C, 55.68; H, 5.52; N, 11.81; S, 13.51; O, 13.48. Found: C,
55.68; H, 5.48; N, 11.69; S, 13.62; O, 13.56.
1
(calculated: 269.06). H NMR (D2O, 400 MHz): δ CH3 (6H, 2.738
N,N′-Di-Z-L-cystine di(dimethylamide) (17) (Z ) Cbz) was made
according to the method of Gustus for the preparation of the analogous
primary amide.43 Di-Z-L-cystine (Fluka, 2.0 g, 3.93 mmol) was ground
to a fine powder, suspended in anhydrous Et2O in a 50-mL Erlenmeyer,
flask and cooled to 0 °C. PCl5 (1.8 g, 8.65 mmol, ground to a fine
powder) was added, and the suspension was alternately shaken
vigorously and cooled (∼2-min cycles) for 30 min. The resulting diacyl
chloride (white solid) was filtered, washed with cold anhydrous Et2O,
and then suspended in 50 mL of anhydrous Et2O at -78 °C under N2.
Cold dimethylamine (∼2 mL) in Et2O (10 mL) was added to the
suspension under N2, and the mixture was stirred at -78 °C for 1 h,
after which the reaction was allowed to warm to room temperature in
a dimethylamine atmosphere. Et2O and excess NHMe2 were removed
under vacuum to reveal a fluffy white solid. Repeated attempts at
recrystallization resulted only in an oil, which was carried on to the
next step.
ppm, s), δ CH2 (4H, 3.325 ppm, d, 3J ) 4 Hz), δ CH (2H, 3.937 ppm,
3J ) 4 Hz).
N,N′-Dimethyl-N,N′-dibenzoyl-L-cystine (13). This compound was
made according to the method of Bloch and Clark45 with modifications.
N,N′-Dimethyl-L-cystine (19, 1.00 g, 3.73 mmol) and potassium
bicarbonate (5.33 g, 53.2 mmol) were dissolved in 10 mL of water to
which 10 mL of Et2O was added. The solution was cooled to 0 °C,
and benzoyl chloride (Aldrich, 3.0 mL, 25.8 mmol) was added dropwise
with rapid stirring. After the solution was allowed to react overnight
and was passed through filter paper, it was acidified with 1 N HCl to
pH 3, causing a thick yellow mass to precipitate. The water was
decanted off, and the solid was dissolved in acetone. The solvent was
removed to yield a slightly sticky yellow solid. Trituration in Et2O
yielded followed by trituration in water and drying yielded 0.671 g of
a pure off-white solid containing a ∼1:1 mixture of cis and trans amide
isomers: mp ) 99-119 °C. FAB-HRMS (M + H)+: 477.1161
(calculated: 477.1154). 1H NMR (400 MHz, DMSO-d6): δ Ar-H
(10H, 7.387 ppm, m), δ CH (1H, 5.097, m), δ CH (1H, 4.493 ppm,
m), δ CH2 (4H, 3.242 ppm, m), δ CH3 (3H, 2.859 ppm, d), δ CH3
(3H, 2.785 ppm, d). The CH3 peaks coalesced to a singlet (δ 2.87 ppm)
by 75 °C and the CH peaks to a single wide peak (δ 4.92 ppm) by 100
°C in a VT 1H NMR experiment (600 MHz) as exchange between the
rotamers became more rapid. Anal. Calcd for C22H24N2O6S2: C, 55.45;
H, 5.08; N, 5.88; O, 20.14; S, 13.46. Found: C, 55.15; H, 5.04; N,
5.93; O, 20.34; S, 13.40.
L-Cystine Di(dimethylamide) Dihydrobromide (18). The crude
N,N′-di-Z-L-cystine di(dimethylamide) (17) oil was deprotected in 20%
HBr in AcOH (10 mL). The clear yellow solution was stirred at room
temperature for 1 h, after which the bulk of the AcOH was removed
under vacuum, revealing an orange-brown oil. Repeated trituration in
room temperature acetone resulted in a crude white solid that was
1
filtered from the colored solution and dried (0.82 g). H NMR (D2O,
400 MHz): δ CH3 (6H, 2.975 ppm, s), δ CH3 (6H, 3.151 ppm, s), δ
CH2 (2H, 3.172 ppm, m), δ CH2 (2H, 3.358 ppm, dd, 3J ) 4 Hz, 2J )
15 Hz), δ CH (2H, 4.840 ppm, m). 13CNMR (D2O with acetone for
reference): δ 35.656, 36.319, 36.684, 49.438, 166.896 ppm.
N,N′-Dibenzoyl-L-cystine Dimethyl Ester (14). Benzoylation
Method B. L-Cystine dimethyl ester dihydrochloride (Sigma, 10.5 g,
30.8 mmol), NaOAc (10.1 g, 123.2 mmol), H2O (100 mL), Et2O (100
mL), and benzoyl chloride (7.87 mL, 67.8 mmol) were reacted for 4
h. Trituration in Et2O and drying yielded 8.181 g (56% yield) of pure
dry white solid: mp ) 169-172 °C. FAB-HRMS (M + H)+: 477.1151
(calculated: 477.1154). 1H NMR (300 MHz, DMSO-d6): δ CH2 (2H,
N,N′-Dibenzoyl-L-cystine Di(dimethylamide) (12). Benzoylation
Method B. Crude L-cystine di(dimethylamide) dihydrobromide (18,
0.82 g), NaOAc (0.65 g, 7.92 mmol), H2O (20 mL), Et2O (30 mL),
and benzoyl chloride (0.63 mL, 5.4 mmol) were reacted for 6 h. Since
the 1H NMR showed a mixture of mono- and dibenzoylated products,
the crude solid (0.33 g) was dissolved in 10 mL of CHCl3, to which
were added benzoyl chloride (58 µL, 0.5 mmol) and triethylamine (55
µL, 0.4 mmol). After reaction overnight, the solvent was removed, and
the crude tan solid was suspended in Et2O and filtered (0.335 g). The
product was recrystallized from MeOH/H2O, yielding 0.188 g of pure
needle-shaped crystals after drying (10% overall yield after four
steps): mp ) 199-200 °C. FAB-HRMS (M + Li)+: 509.1872
(calculated: 509.1869). 1H NMR (300 MHz, CDCl3): δ CH3 (6H, 2.957
ppm, s), δ CH3, CH2 (10H, 3.120-3.275 ppm, m), δ CH (2H, 5.502-
5.434 ppm, m), δ Ar-H, δ NH (8H, 7.288-7.497 ppm, m), δ Ar-H
(4H, 7.783 ppm, d, 3J ) 7.3 Hz). 13CNMR (CDCl3): δ 36.247, 37.715,
41.843, 48.983, 127.300, 128.613, 131.894, 133.646, 166.784, 170.074
ppm. Anal. Calcd for C24H30N4O4S2: C, 57.35; H, 6.02; N, 11.15; O,
12.73; S, 12.76. Found: C, 57.24; H, 6.09; N, 11.02; O, 12.86; S, 12.79.
N,N′-Dimethyl-L-cystine (19) was made according to the method
of Keller-Schierlein et al.44 (R)-(-)-Thiazolidine-4-carboxylic acid50
(Aldrich, 6.66 g, 50 mmol) was dissolved in ammonia at -78 °C, and
to this was added 0.9 mL of H2O. Solid Na was added until the solution
remained blue (∼3.7 g). The reaction was quenched with NH4Cl and
allowed to warm to room temperature and evaporate overnight. After
drying under vacuum, the crude white solid was dissolved in 50 mL
of water and acidified to pH 1 with 6 N HCl. The solvent and excess
acid were removed under vacuum to reveal a sticky off-white solid.
The solid was extracted with absolute EtOH (5 × 50 mL), which yielded
a sticky yellow material (7.261 g) after removal of the solvent. The
methylated amino acid salt was air-oxidized to the disulfide by
dissolving it in 250 mL of water, bringing it to pH 9 with ammonium
hydroxide, adding one crystal of iron sulfate (FeIISO4‚6H2O), and
bubbling air through the solution overnight. After 13 h, the solution
tested negative for thiolates using the nitroprusside reaction.51 The
solution was acidified to pH 6 with 25% AcOH. Addition of 200 mL
3
2
3.150 ppm, dd, J ) 9.9 Hz, J ) 13.8 Hz), δ CH2 (2H, 3.295 ppm,
dd, 3J ) 4.5 Hz, 2J ) 13.8 Hz), δ CH3 (6H, 3.664 ppm, s), δ CH (2H,
4.776 ppm, m), δ Ar-H (6H, 7.491 ppm, m), δ Ar-H (4H, 7.848
3
3
ppm, d, J ) 7.3 Hz), δ NH (2H, 8.955 ppm, d, J ) 7.6 Hz). 13C
NMR (DMSO-d6): δ 38.580 (somewhat obscured by solvent peaks),
51.814, 52.258, 127.250, 128.179, 131.477, 133.275, 166.284, 170.860
ppm. Anal. Calcd for C22H24N2O6S2: C, 55.45; H, 5.08; N, 5.88; O,
20.14; S, 13.46. Found: C, 55.44; H, 5.03; N, 5.88; O, 20.23; S, 13.32.
Physical Studies on Gels. (a) Gelation Tests. Each of the
compounds tested was dissolved in an appropriate amount of warm
DMSO in a glass vial. The solution was heated to >90 °C in a water
bath. Hot (>90 °C) Milli-Q-purified water (18 MΩ‚cm) was added to
the vial, bringing the final volume to 5 mL. The vial was sealed (screw
cap), removed from the water bath, and allowed to cool slowly
(undisturbed) to room temperature.
(b) Rheology. Rheological measurements were carried out on a
Bohlin CSR-10 constant stress rheometer, using a cone-and-plate
geometry (truncated 4/40 cone, 4° cone angle, 40-mm diameter). A
Neslab circulating water bath or a Peltier device controlled the
temperature of the bottom plate. The gap distance was fixed at 150
µm. Gels (1.25-mL total volume) were prepared by mixing a DMSO
solution of the gelator with hot Milli-Q water on the heated (90 °C)
plate, lowering the cone, and cooling the system to 25 °C at ∼3 °C/
min. A low-viscosity oil (viscosity standard S3, 3.494 mPa‚s at 25 °C,
Cannon Instrument Co.) was used around the edges of the gel, which
acted as a moisture barrier (impeding evaporation). An oscillatory shear
stress (2 s on, 10 s off) was applied to the gel typically after it remained
at 25 °C for 1 h. A constant frequency of 1 Hz and a uniform stress
(3-50 Pa, depending on the strength of the gel) was employed. The
level of stress used (g3 Pa, instrument minimum) was approximately
the minimum value able to generate a strain of at least 4 × 10-4
(approximately the minimum measurable strain for 4/40 cone) for each
gel. The shear strain (γ), complex modulus (G*), storage modulus (G′),
loss modulus (G′′), complex viscosity (η*), and phase angle (δ) were
(50) Ratner, S.; Clarke, H. T. J. Am. Chem. Soc. 1937, 59, 200.
(51) Szaciłowski, K.; Stochel, G.; Stasicka, Z. New J. Chem. 1997, 21,
893.

Anatomy of a Gel
J. Am. Chem. Soc., Vol. 122, No. 47, 2000 11691
monitored and recorded as a function of time, and each run was allowed
to continue until these values remained approximately constant. All
gels were subjected to frequency sweeps (0.1-10 Hz, auto stress mode,
target strain ) 0.001), while stress amplitude sweeps (3-600 Pa, 1
Hz) were run with selected samples of each concentration. Temperature
sweep rheological methods are described below.
(c) Temperature-Sweep Oscillation Rheology. Gels formed on the
rheometer and tested as above were used in temperature-sweep
experiments. (Those subjected to an amplitude sweep were not used
for temperature sweep tests.) Gels were subjected to a small oscillating
stress (3 Pa), and the temperature was slowly (3 °C/min) increased
from 25 to 90 °C. The temperature at which the gel could no longer
support the applied stress (when G′′ > G′) was recorded as the transition
temperature (TGS).
(d) Differential Scanning Calorimetry. Gels (500 µL) were
prepared by mixing a DMSO solution of the gelator and Milli-Q water
to each of three hermetically sealed DSC cells at room temperature.
Gels were homogenized at 100° C for 20 min and then cooled to 10
°C prior to each test. DSC runs were recorded from 10 to 100 °C and
from 100 to 10 °C at 10 °C/h. Baselines were run with 500 µL of the
corresponding of water/DMSO mixture.
(e) Light Microscopy. Eighty-microliter gels were prepared by
adding Milli-Q water to DMSO solutions of 5 on microscope slides
with small glass O-rings cemented to them (height ) 1.5 mm, inside
diameter ) 8 mm). Glass coverslips were placed on top, and the gels
were allowed to set. Phase-contrast images were recorded on a Nikon
Diaphot-TMD inverted microscope (40× objective) equipped with an
Optronics DEI-750TD Peltier-cooled 3-CCD color camera.
observed on the lower stage of a DS-130 LaB6 scanning electron
microscope (SEM) using a 10- or 19-kV accelerating voltage. Images
were digitally recorded and processed using Adobe Photoshop.
A detailed procedure for cryo-high-resolution (HR) SEM sample
preparation and imaging of frozen, hydrated samples has been
published47 and is summarized as follows. A small portion of a gel
(∼10 µL cut from the rest of the gel with a razor) was placed onto a
gold planchet (Balzers), frozen in liquid ethane at its melting point
(∼ -183 °C), and stored under liquid nitrogen (LN2). The sample was
loaded into a prechilled cryopreparation chamber (cooled to ∼ -170
°C with LN2), attached to an Oxford CT-3500 cold stage, fractured
with a cold blade, and rinsed with LN2. The sample (with stage shutters
closed to avoid frost contamination) was quickly transferred to a Denton
DV-602 magnetron sputter coater, where it was coated with 1 nm Cr.
Samples were observed in the upper stage of an ISI DS-130F Schottky
field-emission SEM (where they were warmed to -110 °C to allow
any ice condensed on the surface of the Cr to sublime) using a 25-kV
accelerating voltage and processed as above.
Acknowledgment. We thank the staff at Bohlin Instruments
and Dr. Thomas Oomann (Kimberly Clark Corp.) for their
advice on and assistance with the rheology, Dr. Robert P.
Apkarian for imaging the cryo-HRSEM samples, Dr. Nikolai
A. Khanjin and Mr. Bao Do for their determination of and
assistance with the X-ray crystal structures, Dr. David Goldsmith
for photographing the gels, and Dr. Vincent P. Conticello for
helpful discussions. In addition, we thank the National Institutes
of Health (Grant GM21457) for funding.
(f) Electron Microscopy. Gels, jellies, or precipitates (5 mL) were
prepared as explained in the Gelation Tests section above. Samples
were successively dehydrated in a graded ethanol series (30, 50, 70,
90, 100, 100, 100%) for 15-30 min each. The ethanol was slowly
replaced with liquid CO2 in a critical point drying (CPD) apparatus at
0 °C. Once the exchange was complete (∼3 h), the temperature was
increased, bringing CO2 above its critical point, after which the CO2
was slowly released as a gas. The dried gels were mounted onto SEM
stubs with carbon tape, coated with ∼10 nm Au/Pd (60/40), and
Supporting Information Available: Additional ORTEP
diagram of recrystallized 3 (morph from Figure 2a) showing
inter-fiber hydrogen-bonding (PDF), and X-ray crystallographic
files (CIF) for both polymorphs of 3. This material is available
free of charge via the Internet at http://pubs.acs.org.
JA0016811