Lamellar liquid single crystal hydrogels: Synthesis and investigation of anisotropic water diffusion and swelling

Article abstract of DOI:10.1021/ma051479o

The synthesis and characterization of anisotropic liquid single crystal hydrogels (LSCHs) via photoinduced radical polymerization of magnetically aligned samples in the lyotropic mesophase are reported, which are stable against tensile stresses in all thr

Full text of DOI:10.1021/ma051479o

9772
Macromolecules 2005, 38, 9772-9782
Lamellar Liquid Single Crystal Hydrogels: Synthesis and Investigation
of Anisotropic Water Diffusion and Swelling
Felix Kleinschmidt,† Markus Hickl,† Kay Saalwa1chter,† Claudia Schmidt,*,‡ and
Heino Finkelmann*,†
Institut fu¨r Makromolekulare Chemie, Universita¨t Freiburg, Stefan-Meier-Str. 31, D-79104 Freiburg,
Germany, and Department Chemie, Universita¨t Paderborn, Warburger Strasse 100,
D-33098 Paderborn, Germany
Received July 7, 2005
ABSTRACT: The synthesis and characterization of anisotropic liquid single crystal hydrogels (LSCHs)
via photoinduced radical polymerization of magnetically aligned samples in the lyotropic mesophase are
reported, which are stable against tensile stresses in all three dimensions. The hydrogels exhibit a lamellar
phase (LR) in the swollen state. The high mechanical stability is achieved by using a new type of cross-
linker. Monomer and cross-linker molecules have almost the same chemical constitution, varying only in
the number of polymerizable groups. The cross-linker is incorporated perfectly into the liquid-crystalline
phase structure of the lyotropic liquid-crystalline monomers, yielding a network which has covalent bridges
between the lamellae. The anisotropic hydrogels are characterized by a variety of methods on micro- and
macroscopic length scales. Swelling with the nonselective solvent toluene shows that cross-linking within
the liquid-crystalline phase causes an anisotropic topology of the network, which shows a memory effect
even in the isotropic phase. Time- and temperature-dependent pulsed field gradient diffusion NMR
measurements yield a ratio of the order of 10:1 for the self-diffusion coefficients of D2O perpendicular
and parallel to the layer normal. A step in the diffusivity across the lamellae at 312-314 K is interpreted
by a disruption of the lamellae caused by elastic forces due to anisotropic network deformation as a function
of temperature or by an increased porosity of the membranes in analogy to a lamellar-to-sponge
transformation. Hygroelastic measurements, in which the length and width of the hydrogels are measured
as a function of their controlled water sorption, show an anisotropic swelling behavior consistent with
the structure of a lamellar phase. The isotropic-to-lamellar phase transformation upon increasing water
concentration leads to a flattening of the polymer coil along the layer normal. Swelling with water in the
lamellar phase is anisotropic.
I. Introduction
Several methods to obtain LSCHs were recently
reported.3,5-8 The first approach by Lo¨ffler proved that
cross-linked polyamphiphiles indeed exhibit lyotropic
liquid-crystalline phases when swollen with water.7 The
mechanical stress caused by the uniaxial swelling of a
lyotropic hexagonal phase in a glass tube led to a macro-
scopic alignment of the liquid crystal. Fischer succeeded
in obtaining the first liquid-crystalline hydrogel with
permanent uniform orientation of the director of the
liquid-crystalline phase.5 Those hydrogels were realized
in a two-step reaction. In the first step, a conventional
nonionic surfactant is polymerized, and the polymer is
weakly cross-linked in the same batch. In the second
step, a mechanical stress is applied, which forces the
polymeric chain into a nonspherical conformation, and
the cross-linking process is completed under strain. The
deformation of the macromolecules fixed in this way
causes an orientation of the micelles when a lyotropic
phase is formed by water absorption. For systems
forming a lamellar phase, uniaxial extension of the
elastomer results in an alignment of the layer normal
vectors (directors) perpendicular to the axis of extension,
whereas uniaxial compression leads to a lamellar mono-
domain with the director parallel to the axis of com-
pression.5,6 Fischer also reported first results on the
hygroelastic effect, that is, the anisotropic macroscopic
swelling of a substance upon water sorption, which is
caused by elastic stresses due to the coupling between
the network and the liquid crystalline order.5
Macroscopic anisotropy, leading for example to bire-
fringence and anisotropic swelling, is a remarkable
property of lyotropic liquid single crystalline hydrogels
(LSCHs). LSCHs are water-swollen cross-linked am-
phiphilic polymers forming spontaneously a macroscopi-
cally aligned lyotropic liquid crystal phase. In LSCHs
the properties of different classes of materials are
combined. On one hand, LSCHs show the typical
behavior of elastomers such as form stability and rubber
elasticity, and on the other hand, they feature the
properties of liquid crystals such as self-organization
and physical anisotropy. Such liquid-crystalline elas-
tomers on the basis of lyotropic liquid crystals are of
particular interest because they allow a comparison
with the corresponding thermotropic liquid-crystalline
elastomers which have been an object of research for
two decades.1,2 In particular, questions regarding the
network topology in the two types of liquid-crystalline
networks are not answered yet. Besides this important
academic question, LSCHs are also expected to have
interesting applications. A prototype of a bifocal contact
lens making use of the birefringence which causes two
indices of refraction has already been developed.3 LSCH
systems are also discussed as models of artificial cell
membranes.4
† Universita¨t Freiburg.
‡ Universita¨t Paderborn.
A different approach toward oriented hydrogels was
taken by Amigo´-Melchior.3 On the basis of the knowl-
* Corresponding authors. E-mail: Claudia.Schmidt@uni-pad-
erborn.de; Heino.Finkelmann@makro.uni-freiburg.de.
10.1021/ma051479o CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/21/2005

Macromolecules, Vol. 38, No. 23, 2005
Lamellar Liquid Single Crystal Hydrogels 9773
edge that rigid anisometric surfactant molecules9-13
lead to the formation of micelles with a specific geom-
etry14 due to packing restraints, a polymerizable am-
phiphile was developed. The molecule had to form a
lyotropic liquid-crystalline phase, preferably a lamellar
one (LR), over a broad temperature and concentration
range. Additionally, a high anisotropy of the diamag-
netic susceptibility was needed. This should allow for
the uniform orientation of the liquid-crystalline phase
structure by a magnetic field in order to form a mon-
odomain before polymerization. The desired molecule
was found by subsequent alteration of the spacer length
and the hydrophilic/hydrophobic balance (HLB)3,15 of a
rodlike amphiphile with an aromatic core.16 The rodlike
shape causes a lamellar structure, and the aromatic
rings lead to the positive diamagnetic anisotropy that
is required for a uniform layer orientation in a magnetic
field. For the first hydrogels based on this monomer a
hydrophobic cross-linker molecule was used. In this
system the density of cross-linking within the hydro-
phobic part of the smectic layers was higher than the
density of interlayer cross-linking, which caused a lack
of mechanical stability. Similar effects are known from
thermotropic liquid-crystalline elastomers, where the
type of cross-linking influences the mechanical behavior.
Interlayer and intralayer cross-linking result in differ-
ent moduli of elasticity.17-19 It was also shown that the
coupling between liquid-crystalline phase and polymer
network differs.20,21
In the present work we follow the approach of Amigo´-
Melchior but introduce a new type of cross-linker
molecule, which is adapted to the amphiphilic structure
of the monomer and can therefore be incorporated
perfectly into the lamellar phase. This cross-linker has
polymerizable groups in both its hydrophilic and hy-
drophobic parts. Therefore, cross-linking should occur
not only within the layers but also between adjacent
layers, and a three-dimensionally stable network is
expected. The polymerization and the cross-linking
process are carried out in one step in the magnetically
aligned aqueous liquid-crystalline phase.
Important questions concern the microscopic proper-
ties of these new hydrogels, such as membrane ordering
and permeability, and the macroscopic anisotropy and
order of the material on different length scales. The
investigation of its anisotropic swelling with water
should provide information about the influence of the
phase transformation on the chain conformation of the
polymer. We were particularly interested in how this
new type of cross-linking influences the topology of the
network and whether this network, which was formed
in the liquid-crystalline phase, differs from earlier ones
that were cross-linked in isotropic solution in a nonse-
lective solvent.
To answer these questions, we carried out comple-
mentary physical measurements on different length
scales in order to obtain a picture as complete as
possible of the properties of these new materials. The
hydrogels were characterized on a macroscopic level
with optical microscopy, stress-strain, swelling, and
hygroelastic measurements and on a micro- and meso-
scopic scale by X-ray diffraction, deuterium NMR spec-
troscopy, and diffusion measurements. In the following,
after the Experimental Section, we will show that our
synthetic concept yields mechanically stable LSCHs. In
section IV, the results of pulsed field gradient diffusion
experiments will be presented that not only confirm the
Scheme 1. Structure of the Amphiphilic Monomer
and Cross-Linker
anisotropy of the hydrogels and establish differences
arising from different preparation conditions but also
show an unexpected temperature dependence of diffu-
sion. In section V the results of the hygroelastic mea-
surements are discussed which reflect the chain con-
formation. Section VI summarizes the major conclusions.
II. Experimental Section
Synthesis. All reactions were carried out in an argon
atmosphere to prevent oxidation of the ethylene glycols. The
reactions with molecules containing methacrylic groups were
carried out in darkness by using glass apparatuses wrapped
in aluminum foil to prevent spontaneous polymerization.
Synthesis of the Monomer and the Cross-Linker. The lyo-
tropic hydrogels were synthesized from the amphiphilic mono-
mer 1 and the adapted cross-linker molecule 2 shown in
Scheme 1. The monomer has a side-on methacrylate function
that can undergo radical polymerization in aqueous solution.
Its lyotropic phase behavior and that of its homopolymer have
already been investigated;3 the binary phase diagrams are
shown in Figure 1. The cross-linker molecule is a modification
of the monomer and has almost the same chemical structure.
It has three methacrylate groups: one attached side-on to the
hydrophobic part as for the monomer and the other two
attached end-on to the hydrophilic moieties.
2,5-Di-(4-{2-[2-(2-{2-[2-(2-Methoxyethoxy)ethoxy]ethoxy}-
ethoxy)ethoxy]ethoxy}benzoic acid ester)benzoic acid 2-(2-me-
thylacryloyloxy)ethyl ester (1) was prepared by using the same
procedures as described by Amigo´-Melchior.3 The purification
of 1 was improved by using column chromatography with
Sephadex LH 20, a modified dextran gel liquid chromatogra-
phy medium, as stationary phase and methanol as mobile
phase. To prevent spontaneous polymerization, the substance
was stored in cryo tubes using liquid nitrogen as cooling
medium.
The route for the synthesis of the cross-linker is presented
in Scheme 2. It starts with the preparation of the hydrophilic
tails of the molecule. The complete cross-linker is gained in
the very last step by connecting the tails 8 with the core
molecule 3. In this step the rigid hydrophobic part is formed.
The sequence begins with the protection of one of the two
alcoholic functions of tetraethylene glycol. This step is followed
by the esterification of the glycol with p-hydroxybenzoic acid
methyl ester using the Mitsunobu procedure.22 In the next two
steps the protection groups of the glycol 4 and benzoic acid
are removed using standard procedures. The polymerizable
group is introduced by an azeotropic esterification of the
deprotected alcoholic function of glycolbenzoic acid (7) with
methacrylic acid. In the final step the side parts 8 and the
core of the molecule 3 are connected via an esterification with
N,N′-dicyclohexylcarbodiimide.
2-(2-{2-[2-(Tetrahydropyran-2-yloxy)ethoxy]ethoxy}ethoxy)-
ethanol (4) was obtained by a monoprotection of tetraethylene

9774 Kleinschmidt et al.
Macromolecules, Vol. 38, No. 23, 2005
Figure 1. Phase diagram of the monomer 1 (left) and the corresponding un-cross-linked polymer (right) in D2O.3 The arrows in
(a) indicate the temperature pathway during the alignment procedure in the magnetic field. The gray bars in (b) show the regions
where diffusion was measured.
Scheme 2. Synthesis of the Cross-Linker Molecule
glycol with a tetrahydropyran group. 116.5 g (0.6 mol) tetra-
ethylene glycol and 10 g of p-toluenesulfonic acid (PTSA) were
dissolved in 200 mL of tetrahydrofuran (THF). 25.2 g (0.3 mol)
of dihydropyran were added dropwise. The mixture was stirred
at room temperature for 2 h to form 4. THF was removed by
distillation. The crude product was dissolved in chloroform and
washed with water. After drying with MgSO4 and removal of
the solvent, the product was used without further purification
[yield: 50 g of a clear liquid containing 70% monoprotected
and 30% diprotected glycol according to gas chromatography].
4-[2-(2-{2-[2-(Tetrahydropyran-2-yloxy)ethoxy]ethoxy}ethoxy)-
ethoxy]benzoic acid methyl ester (5) was prepared by esterifi-
cation of 4 with p-hydroxybenzoic acid methyl ester using the
Mitsunobu procedure. 50 g of the monoprotected glycol (0.125
mol contained in the mixture described above), 16.2 g (0.125
mol) of p-hydroxybenzoic acid methyl ester, and 33 g of
triphenylphosphine were dissolved in 300 mL of THF. 25.5 g
(0.125 mol) of azodicarbonic acid diisopropyl ester was added
dropwise. The mixture was stirred overnight. After removal
of the solvent diethyl ether was added, and a colorless
precipitate of byproducts was formed. It was filtered off, and
the ether was removed by distillation.
4-(2-{2-[2-(2-Hydroxyethoxy)ethoxy]ethoxy}ethoxy)benzoic acid
methyl ester (6) results after deprotection of the glycol. The

Macromolecules, Vol. 38, No. 23, 2005
Lamellar Liquid Single Crystal Hydrogels 9775
crude product of the preceding reaction was dissolved in
ethanol. 3 g of pyridinium p-toluenesulfonic acid was added.
The pH of the solution was 3. The mixture was refluxed for 3
h, and then the ethanol was removed in a vacuum. The crude
product was dissolved in chloroform and washed with water.
After drying with MgSO4 and removal of the solvent, the
product was used in the next step without further purification.
4-(2-{2-[2-(2-Hydroxyethoxy)ethoxy]ethoxy}ethoxy)benzoic acid
(7) was obtained by deprotection of the carbonic acid. The crude
product of the preceding reaction was dissolved in 600 mL of
a water/methanol mixture (2:1). 10 g (0.41 mol) of lithium
hydroxide was added. The mixture was refluxed for 3 h. The
methanol was removed by vacuum distillation. Hydrochloric
acid was added, and the product was extracted with chloro-
form. The solution was dried with MgSO4 and the chloroform
removed. The product was purified by multiple recrystalliza-
tion from ethyl acetate/cyclohexane (1:1) [yield: 10 g of a
colorless wax, 25% with respect to 4 after three reaction steps].
1H NMR (300 MHz, CDCl3): δ [ppm] ) 3.7 (t, 3 H, CH2-
CH2-OH), 3.8 (m, 10 H, glycol), 4.0 (t, 2 H, COO-CH2-CH2),
4.2 (t, 2 H, COO-CH2), 7.0 (d, 2 H, Ar-H), 8.1 (d, 2 H, Ar-
H).
4-[2-(2-{2-[2-(2-Methylacryloyloxy)ethoxy]ethoxy}ethoxy)ethoxy]-
benzoic acid (8) was prepared by functionalization of 7 with
methacrylic acid. A mixture of 7.5 g (0.024 mol) of 7, 10 g (0.12
mol) of methacrylic acid, and 0.2 g of p-toluenesulfonic acid in
chloroform was refluxed in a Dean-Stark apparatus for 72 h.
The chloroform and the excess methacrylic acid were removed
by vacuum distillation. The crude product was purified by
column chromatography (silica gel F60, ethyl acetate) [yield:
6.3 g (68%) of a colorless wax].
types of samples were aligned in an NMR magnet (11 T for
the spherical samples and 7 T for the stripes) to obtain a
uniform director orientation. For the stripes of length z and
width x the layer normal (director) is parallel to z. The lamellar
phase (LR) was reached by slowly heating or cooling from the
isotropic phases (L1 or L2, cf. Figure 1) and subsequent
annealing in the biphasic region. The process of orientation
was followed by 2H NMR. At the starting temperature a
narrow single peak is observed in the isotropic phase. In the
two-phase region the quadrupolar doublet grows at the
expense of the central isotropic peak. The lamellar phase
shows a pure doublet of narrow lines. Within the highly viscous
mesophase, the orientation of the molecules is stable, and the
sample can be removed from the magnetic field for the
polymerization, which was carried out in a Kulzer Dentacolor
XS UV lamp equipped with two xenon lamps. The irradiation
wavelength was 320-520 nm. The exposure time was 120 s.
A home-built cooling device which uses a liquid nitrogen cooled
gas stream allows us to control the temperature within the
polymerization chamber of the UV-lamp. The temperature was
kept at 293 K during the polymerization process.
The stripe-shaped hydrogels were stored in the swollen state
in bidistillated water. The stripes are stable against slight
mechanical deformation in all directions of space. The water
can be removed from the stripes without damaging them. If
they are penetrated with water again, they regain their initial
appearance. The dry elastomers exhibit common rubber
elasticity.
Measurements. X-ray Measurements. The hydrogel samples
were placed in water in a 2 mm diameter glass capillary.
Diffraction studies were performed with a rotating anode
generator (6.4 kW) in a point-collimated Kiessig camera with
Ni-filtered Cu KR radiation (wavelength λ ) 1.542 Å) at a
distance of 180 mm. Two-dimensional patterns were recorded
with an image plate system (Schneider, Freiburg).
1H NMR (300 MHz, CDCl3): δ [ppm] ) 1.9 (s, 3 H, CH2d
C(CH3)-COO), 3.6 (m, 10 H, glycol), 3.9 (t, 2 H, COO-CH2-
CH2), 4.2 (t, 2 H, COO-CH2), 4.3 (t, 3 H, CH2-CH2-OOC-
C(CH3)dCH2), 5.5 (s, 1 H, CH2dC(CH3)-COO), 6.1 (s, 1 H,
CH2dC(CH3)-COO), 6.9 (d, 2 H, Ar-H), 8.0 (d, 2 H, Ar-H).
2,5-Dihydroxybenzoic acid 2-(2-methyl-acryloyloxy)ethyl ester
(3) was prepared using the procedure described by Amigo´-
Melchior.3
2
NMR Measurements. H NMR spectra were recorded on a
Bruker Avance 500 spectrometer (11.7 T) at a frequency of
76.773 MHz. Orientation-dependent measurements were per-
formed using a goniometer probe with a 5 mm solenoid coil.
The sample can be rotated about an axis perpendicular to the
external magnetic field. Orientation of the monodomains used
for hygroelastic measurements were performed on a Bruker
MSL spectrometer (7 T), operating at a frequency of 46.073
MHz. The polymerization mold was placed in a 15 mm saddle
coil. The sample temperature was kept constant within 1 K
by a Eurotherm thermostat.
2,5-Di-(4-[2-(2-{2-[2-(2-Methylacryloyloxy)ethoxy]ethoxy}-
ethoxy)ethoxy]benzoic acid ester)benzoic acid 2-(2-methyl-acry-
loyloxy)ethyl ester (2) was obtained by linking 8 and 3. All
substances were vacuum-dried. 2.5 g (0.0065 mol) of 8, 0.9 g
(0.0033 mol) of 3, and 80 mg of p-(dimethylamino)pyridine
were dissolved in dry methylene chloride. The mixture was
cooled in an ice bath. 1.3 g (0.0065 mol) of N,N′-dicyclohexyl-
carbodiimide (DCC) was added. The mixture was allowed to
warm to room temperature and stirred for 3 h. The precipitate
which was formed during the reaction was filtered off. The
solution was washed with HCl solution and sodium bicarbon-
ate solution consecutively. The solution was dried with MgSO4
and the solvent removed by distillation. The cross-linker was
purified by column chromatography (Sephadex LH 20, metha-
nol). To prevent spontaneous polymerization, the substance
was stored in cryo-tubes using liquid nitrogen as cooling
medium [yield: 1 g (30%) of a colorless, viscous liquid].
1H NMR (300 MHz, CDCl3): δ [ppm] ) 2.1 (s, 9 H, CH2d
C(CH3)-COO), 3.9 (m, 20 H, glycol), 4.1 (t, 4 H, Ar-O-CH2-
CH2), 4.35 (m, 6 H, Ar-COO-CH2-CH2 and Ar-O-CH2-
CH2), 4.5 (t, 4 H, CH2dC(CH3)-COO-CH2-glycol), 4.6 (t, 2 H,
CH2d(CH3)-COO-CH2-CH2-OOC-Ar), 5.7 (s, 3 H, CH2d
C(CH3)-COO), 6.3 (s, 3 H, CH2dC(CH3)-COO), 7.1 (dd, 2 H,
Ar-H), 7.4 (d, 1 H, Ar-H), 7.6 (dd, 1 H, Ar-H), 8.05 (d, 1 H,
Ar-H), 8.3 (d, 2 H, Ar-H).
Synthesis of LSCH Samples. To produce the hydrogels
monomer, cross-linker, deuterium oxide, and photoinitiator
(Ciba Irgacure 2959 and Ciba Darocure 1173 1:1) were mixed
in a Teflon capsule using a Perkin-Elmer vibrating mill. Two
different types of samples were prepared. For the NMR
measurements the thoroughly homogenized mixture was filled
through a capillary of 0.5 mm inner diameter into a small
sphere (4 mm diameter) of quartz glass. The sample was then
sealed by melting of the capillary. For all other investigations
the mixture was filled in a Teflon mold of a stripe form (20
mm × 5 mm × 0.4 mm), covered by a quartz glass plate. Both
Pulsed Gradient Diffusion NMR. Pulsed-field gradient
measurements of the self-diffusion coefficient D were per-
formed in a Bruker high-resolution probe with a singly tuned
saddle coil (∼5 µs 90° pulses) using the pulsed gradient
stimulated-echo technique (PGSTE). The standard method of
Steijskal and Tanner23,24 and its basic theory are extensively
covered in the literature.25 A matched pair of trapezoidal
ramped gradient pulses, separated by the diffusion time ∆,
was inserted in the evolution delays t2 of the stimulated echo
sequence 90°-t2-90°-t1-90°-t2-aq. The attenuation of the
spin-echo intensity due to diffusion is given by
S
ln
) -4π2q2D(∆ - δ/3)
(1)
(S )
0
where a correction for the finite duration of the gradient
pulses, δ, is included. S0 is the signal intensity with gradient
amplitudes of zero. The “wave vector” q ) γδg/(2π) further
depends on the magnetogyric ratio γ and the gradient ampli-
tude g. During one diffusion experiment all delays are kept
constant, and the gradient amplitude g is incremented.
Gradients were calibrated to the known diffusion coefficient
of D2O.26 Diffusion parallel to the layer normal was measured
with the Bruker Diff30 gradient system (which has a stronger
gradient), whereas diffusion perpendicular was measured with
the Bruker Micro5 imaging system with three orthogonal
gradient coils.

9776 Kleinschmidt et al.
Macromolecules, Vol. 38, No. 23, 2005
Table 1. Composition of the Polymerized Mixtures in
wt %a
Table 2. Relative Changes of the Lengths (λz) and Widths
(λx) of the Hydrogels in the Isotropic State Swollen with
the Nonselective Solvent Toluene
sample name
HG33d HG33h HG65d HG65h
sample
λz
λx
amphiphile
28.5
1.9
32.2
2.6
59.3
3.0
64.1
3.7
cross-linker
HG33h
HG65h
1.64
1.61
1.76
1.74
cross-linker relative to
pure amphiphile
D2O
6.3
7.5
4.8
5.8
68.6
1.0
64.2
1.0
36.7
1.0
31.2
1.0
modulus of 24 kPa was determined. This value is
comparable with the Young modulus of a smectic
elastomer in the isotropic phase.27
photoinitiator
a The suffix of each hydrogel labels the type of experiment: d
) diffusion; h ) hygroelastic.
When studying the macroscopic anisotropy of the
material, the important question arises if the cross-
linking process within the liquid crystalline phase
influences the topology of the network. This can be
tested by swelling measurements in the isotropic phase,
using a nonselective solvent that does not induce the
formation of micellar aggregates and lyotropic meso-
phases. An anisotropy of swelling under these conditions
indicates an intrinsic anisotropy of the network due to
the conditions under which the cross-linking process
occurred. In Table 2 the results of swelling experiments
with the nonselective solvent toluene are summarized.
Actually both samples exhibit anisotropic swelling by
the nonselective solvent. The relative change in the
dimension λx is larger than λz. This means that the
polymerization and cross-linking process which occurred
in the liquid-crystalline phase yields a degree of cross-
linking that is higher for interlayer cross-linking than
for intralayer cross-linking. In other words, the network
strands parallel to the layer normal are shorter than
those perpendicular to it. This is plausible when the
molecular structure of the cross-linker molecule is
considered. The cross-linker is hexafunctional since two
network chains start from each methacrylate group.
Only two of the six functions are located in the hydro-
phobic regions, where most of the methacrylate groups
of the monomers are, but four are in the hydrophilic
region. Because of this structure, it is most likely that
the functional groups of the cross-linker in the hydro-
philic regions react with each other (or with the hydro-
philic ethylene oxide chains in their neighborhood),
resulting in very short strands, whereas all functional
groups in the hydrophobic part can react with each
other, leading to longer chains between branching
points. This structural model is supported by the earlier
results obtained by Amigo-Melchior.3 He observed poor
mechanical stability due to a lack of interlayer cross-
links for hydrogels synthesized with a cross-linker that
had no hydrophilic moieties with functional groups. The
swelling of the dry hydrogels proves that the deswollen
gel remembers the topology of the network formed in
the liquid-crystalline phase; the networks prepared in
the liquid crystalline phase have an intrinsic anisotropy.
The optical quality of the hydrogels, which are
transparent, was investigated by polarization micros-
copy. The gels were studied in the swollen state between
crossed polarizers. If the polarization axis is parallel to
the optical axis of the hydrogel, complete extinction of
the light is expected. However, we obtained the photo-
graph shown in Figure 2 (left). The picture with an
orientation of the director of the liquid-crystalline phase
parallel to the axis of polarization shows a characteristic
defect structure, known as focal conics, instead of
complete extinction of the light. The focal conic structure
was proposed by Friedel and Grandjean.28 In these
defects the layers take the form of uniformly spaced
Dupin cyclides. A cross section through a set of Dupin
cyclides as they appear in a focal conic structure is
Swelling Behavior and Hygroelastic Measurements. The
isotropic swelling with the nonselective solvent toluene was
measured under a Will Stru¨bin reflection microscope. The
dimensions were measured before and after swelling, and the
swelling coefficients λz in length and λx in width were calcu-
lated by dividing the length (z) or width (x) in the swollen state
by the respective initial dimensions. Because of the uniaxiality
of the system, λy was taken equal to λx and the swelling
parameter q, defined as the volume ratio of the swollen and
dry sample, was calculated according to
2
q ) λzλx
(2)
The macroscopic dimensions of the hydrogel as a function of
the water concentration is measured with the method of
hygroelasticity, which allows us to measure the change of
dimensions of hydrogels during the process of swelling by
controlling precisely the content of water in the sample. The
measurements are carried out in a home-built computer-
controlled apparatus, equipped with a hygrostat. This hy-
grostat operates by the isopiestic principle. At a given tem-
perature T2 the sample is equilibrated with the partial
pressure of water, pH O. This partial pressure can be altered
2
if nitrogen gas is saturated with water at a temperature T1,
where T1 < T2. Water is absorbed by the hydrogel depending
on pH O. The resulting change of the polymer concentration in
the s2ample can be measured in situ with a microbalance
(Sartorius MP4040). The relative dimensions of the sample
are quantified via a CCD camera.
III. Results and Discussion
Preparation and Characterization of Lamellar
LSCHs. On the basis of the known phase diagram of
the monomer/D2O system,3 depicted in Figure 1, several
hydrogels were prepared at concentrations at which the
aqueous solutions can be aligned in a magnetic field by
cooling or heating slowly from an isotropic phase into
the lamellar phase as indicated by the arrows in Figure
1. Because of the positive anisotropy of the diamagnetic
susceptibility of the aromatic molecules, this procedure
leads to an aligment of the long axes of the molecules
parallel to the magnetic field, resulting in macroscopi-
cally aligned lamellar samples. The mixtures which
were used to produce the hydrogels are given in
Table 1. Different concentrations of the amphiphile were
chosen in order to investigate the influence of the initial
structure of the lamellar phase on the properties of the
LSCH resulting after polymerization and cross-linking.
HG33d and HG33h were synthesized in the water-rich
concentration range, whereas HG65d and HG65h were
synthesized with a high concentration of amphiphile.
They were found to be macroscopically aligned lamellar
systems as will be shown in the following.
The dry isotropic hydrogels prepared with the new
cross-linker that enables intra- and interlamellar cross-
linking are mechanically stable against tensile stress
in all directions. A stress-strain experiment of elas-
tomer HG33h showed a linear relationship between the
stress and the strain z/z0 up to a strain of 1.5. A Young

Macromolecules, Vol. 38, No. 23, 2005
Lamellar Liquid Single Crystal Hydrogels 9777
The orientation-dependent NMR spectra yield also
information on the degree of orientational order. The
width and symmetry of the deuterium line shape depend
on the natural line width (due to relaxation) and on the
orientational distribution function of the local director.
The estimation of the orientational distribution from
NMR line shapes is a typical ill-posed inverse problem.
A Tikhonov regularization procedure employing a self-
consistent method to find the regularization parameter
has proven particularly reliable in finding the solutions
to such problems.35,36 We used the procedure developed
by Winterhalter et al.37 to obtain the orientational
distribution function of the director by Tikhonov regu-
larization from two NMR signals, recorded at different
angles â of the symmetry axis of the sample with respect
to the static magnetic field. The orientational distribu-
tion function found in this way has one peak with a
maximum at an angle of zero. Its width and shape vary
somewhat depending on the values of â chosen for the
two NMR signals, probably due to angular dependent
transverse relaxation times. Averaging the results of the
orientational distribution functions obtained for differ-
ent sets of data yields a value for the full width at half-
maximum of 13.5°. Calculating a domain order param-
eter from the orientational distribution function according
to Mitchell results in SNMR ) 0.89. This value is
significantly larger than the value of SX-ray. The differ-
ence is attributed to the diffusion of the water molecules,
whereby the NMR spectrum represents a dynamic
average over regions of the sample spanning several
micrometers. Disorder on an even smaller length scale,
which is detected by X-ray scattering, is not detected
in the NMR experiment.
The different techniques of characterization show that
the synthetic concept has been successful for the prepa-
ration of mechanically stable lamellar hydrogels with
a macroscopically uniform director orientation. Further
information about the phase structure is provided by
the investigation of the diffusion of water as described
in the following section.
Diffusion Measurements. Diffusion was measured
for each composition in the direction parallel to the
director (layer normal) and perpendicular to it, resulting
in self-diffusion coefficients D| and D , respectively. The
echo decays due to diffusion gained from the pulsed
gradient experiments of HG65d and HG33d are pre-
sented in Figure 3 for varying diffusion times ∆ between
30 and 200 ms.
Figure 2. Optical microscopy picture of the swollen hydrogel
between crossed polarizers with the orientation of the director
parallel to the axis of polarization (left) and cross-section
through a set of Dupin cyclides as they appear in the focal
conic texture (right).
drawn in Figure 2 (right). These defects are well-known
from smectic liquid crystals29 but have also been found
in lyotropic liquid crystals containing surfactants.30,31
The process by which these arrays form is not entirely
clear. It is known that the planar regions which develop
between the oily streaks during the annealing process
ultimately transform into focal conic arrays. Interest-
ingly, the orientations of the defects are not distributed
randomly over the sample, which would indicate a
polydomain. They form perfectly aligned arrays with the
apex of the hyperbola showing in one direction. Fur-
thermore, the apices are found on more or less straight
lines. This indicates that the lamellar hydrogel is
macroscopically aligned but not free of defects.
The proof of the macroscopic alignment of the sample
is provided by two-dimensional X-ray diffraction using
an image plate geometry. The swollen hydrogel at
295 K exhibited a diffraction pattern with one sharp
single-point Bragg reflection in the small-angle region,
characteristic of an aligned lamellar phase. The lamellar
period is 63 Å. For comparison, the monomer in its all-
trans conformation has a length of 65 Å. The analysis
of the azimuthal width of the Bragg peaks according to
the model of Mitchell32,33 yields a domain order param-
eter SX-ray ) 0.65. Assuming an exponential loss of
interlayer correlation, the peak shape as a function of
the polar scattering angle can be fitted with a Lorent-
zian function. From the inverse of its full width at half-
maximum,34 a smectic correlation length ês is obtained
as 420 ( 8 Å, which corresponds to 6.7 layers.
The existence of a macroscopically oriented lamellar
phase is confirmed by angular dependent 2H NMR
spectroscopy. The quadrupolar splitting of a uniaxial
phase is related to the angle â between director and
magnetic field by
The linear dependence of the natural logarithm of the
normalized echo amplitude S/S0 on the square of the
parameter q (cf. eq 1) signifies that the echo decay has
a Gaussian shape, which means that there are no
restrictions to diffusion on the time scale of the experi-
mental diffusion times ∆. The diffusion time ∆ was
varied from experiment to experiment while the gradi-
ent duration δ and the maximum gradient amplitude
gmax were chosen adequately to reach an echo intensity
close to zero at gmax. Since all gradient parameters in
the pulse sequence were varied independently, technical
artifacts can be excluded. The diffusion coefficients were
derived by fitting the raw data according to eq 1. No
tendential dependence on ∆ was observed for any
temperature, but the scattering of the data is bigger
than their individual fitting error. Therefore, the mean
value of the diffusion coefficients Dmean was taken, and
the error interval was calculated from the different
measurements with varying ∆.
1
2
∆ν
(3 cos2 â - 1)
(3)
If the system has a uniformly aligned lamellar phase,
the quadrupolar splitting of a sample whose director is
perpendicular to the magnetic field is exactly half of the
quadrupolar splitting of a sample whose director is
parallel to the magnetic field. The splitting vanishes at
the magic angle of 54.7°. The observed splitting ∆ν of
the hydrogels (830 Hz for HG65d and 720 Hz for HG33d
at 303 K) as a function of the sample orientation follows
perfectly a Legendre polynomial of second order, as
given by eq 3. This proves that the lamellar structure
of the monomer mixture is maintained during the
process of polymerization.

9778 Kleinschmidt et al.
Macromolecules, Vol. 38, No. 23, 2005
The upper limit for the length scale is restricted by ∆max
,
the maximal experimentally accessible diffusion time,
which is limited to 200 ms due to relaxation. Residual
mobile components (2.56 ppm, 30% of the total signal),
whether originating from un-cross-linked mesogens or
remaining photoinitiator is not clear to us, were mea-
sured in the proton spectrum of the hydrogels. Their
particularly long T1 relaxation time expanded the dif-
fusion time to 1 s, thereby giving access to even longer
time scales. Proton diffusion coefficients were also
obtained by fitting the echo intensity decay curves, but
including a constant term in eq 1 to account for the
immobile network. As for D2O no dependence on ∆ was
observed. From the data in Table 3, a lower estimate of
the size of a homogeneously ordered domain can be
given as 7.3 µm × 26.9 µm × 26.9 µm for HG65d and
12.0 µm × 29.0 µm × 29.0 µm for HG33d, where the
smaller values refer to the extension of the domains
parallel to the layer normal. With a layer distance of
63 Å this means that homogeneity parallel to the layer
normal is given over 1160 layers for HG65d and 1900
layers for HG33d. It should be emphasized that this
does not mean that HG33d is ordered more perfectly;
it just gives the lower limit of homogeneity. The differ-
ence in the calculated numbers is owed to the difference
in the diffusion coefficients. A direct comparison of the
diffusion coefficients gives additional information about
the state of order in the two hydrogels. All diffusion
coefficients of HG65d are smaller than the ones of
HG33d, with a larger difference for D|. Although the
differences are small, this indicates that HG65d has less
defective membranes and is better ordered.
The temperature dependence of the diffusion coef-
ficients is analyzed in terms of the obstruction factor
Q ) Dmeasured/Dfree. Here, Dfree is the diffusion coefficient
of bulk water and Dmeasured is the coefficient of obstructed
diffusion in the lamellar system. Hence, Q is expected
to be less than one and smaller values of Q correspond
to more strongly hindered diffusion. We calculated Q
using the temperature-dependent D2O diffusion coef-
ficients, Dfree, measured by Mills.26 The temperature-
dependent obstruction factors Q| and Q for diffusion
across and parallel to the lamellae are shown in Figure
4 for HG33d and HG65d. The values of Q for HG33d
are higher, that is, diffusion is less obstructed, as
expected for the gel with higher content of water. Q| of
HG65d shows a step in the diffusivity at around 312 K,
a similar step can be seen for HG33d at around 314 K.
The hindrance of the water diffusion perpendicular to
the layers is reduced above these temperatures. The
temperature dependence of Q shows only a continuous
increase with temperature, as shown in Figure 4b.
Figure 3. Diffusion time dependent spin-echo decay for
HG65d and HG33d at 298 K. Half-filled and open symbols
correspond to diffusion parallel and perpendicular to the layer
normal, respectively.
Table 3. Diffusion Coefficients of D2O and Mobile
Protons at 298 K, Parallel and Perpendicular to the
Layer Normal
Dmean(D2O) [10-10 m2/s]
parallel perpendicular
Dmean(1H) [10-10 m2/s]
parallel perpendicular
sample
name
HG33d 0.84 ( 0.02 7.09 ( 0.07 0.72 ( 0.02 4.20 ( 0.12
HG65d 0.40 ( 0.01 5.35 ( 0.04 0.27 ( 0.01 3.61 ( 0.05
The average diffusion coefficients of D2O and the
mobile protonated components are summarized in
Table 3. The measured values of D| and D indicate that,
on one hand, diffusion across the layers is possible and,
on the other hand, that the motion along the layers is
not free. The diffusivity along the layers is reduced by
about 50% compared to free water, which can be
explained by the strong interaction with the ethylene
oxide chains. The anisotropy of D2O diffusion, D /D|,
was in the range of 13 for HG65d and a bit lower for
HG33d. This anisotropy is lower than for low molecular
weight lyotropic systems, where anisotropies of 17 and
more were found.38,39 For a lamellar lyotropic mixture
of 40 wt % C10E3/D2O we found anisotropies in the
diffusivity of D2O of more than 100.40 Probably, the
structure of the lamellar phase of the hydrogel is less
perfect than that of low molecular weight lamellar
systems due to the restrictions of the amphiphiles
tethered to the polymer network. Another origin of
defective lamellae could be the energy transfer to the
sample during UV cross-linking, which might heat the
sample locally and disturb the mesophase.
The steps in Q| at 312 K for HG65d and at 314 K for
HG33d indicate a formation of defects in the lamellae.
They are observed in a temperature range where the
un-cross-linked polymer is still in the liquid crystalline
phase, but where the corresponding low molecular
weight monomer mixture is already phase separated.
Two factors may contribute to the breaking up of the
lamellae. One is an elastic force from the network. The
network has a higher cross-linking density between
neighboring lamellae as shown above by the swelling
experiments with toluene. Because of this anisotropy
of the network, thermal expansion or contraction (due
to rubber elasticity) may be anisotropic. This can
generate an internal stress that leads to the partial
disruption of the mesophase above a certain threshold.
From the diffusion time dependent data an estimation
of the minimal lateral and longitudinal extension of a
“perfectly” ordered domain can be made. The length
scale over which information is accessible is given by
the square root of the mean squared displacement
z2 and can be calculated according to z2 ) 2D∆max
.
x

Macromolecules, Vol. 38, No. 23, 2005
Lamellar Liquid Single Crystal Hydrogels 9779
Figure 5. Macroscopic diffusion of (a) methylene blue and
(b) methyl red, started from a crystal of the dye placed in the
center of hydrogel HG33h, after 360 min. The layer normals
are oriented vertically.
organic phases over diffusing along each of these,
presumably because of the low mobility and small
extension of the organic phase in which it preferably
resides.
Hygroelastic Measurements. Detailed information
about the topology of the network and the coupling
between the liquid crystalline phase and the polymer
network can be gained from swelling measurements.
The preceding discussion of the swelling with a nonse-
lective solvent has shown that the network has an
intrinsic anisotropy. In the following, the influence on
the phase transformation induced by swelling of the
dried network with water, a solvent selective for the
hydrophilic regions, is investigated. Therefore, we car-
ried out hygroelastic experiments; that is, we measured
the macroscopic changes of the dimensions of the
hydrogels depending on their water content. These
measurements allow us to draw conclusions on the
microscopic structure of the hydrogels, in particular, on
the conformations of the polymer chains and on the
topology of the networks.
The main focus of this study concerns the influence
of the micellar aggregation on the chain conformation
of the polymer network. A macroscopically anisotropic
behavior at the phase transformation and within the
liquid-crystalline phase as shown schematically in
Figure 6 is expected. The swelling factors of the length
and width, λz and λx, have been calculated assuming
constant partial molar volumes of water and network.
Starting from the dry network with an isotropic polymer
coil on the right-hand side of the diagram, the swelling
in the isotropic L2 phase leads to a linear increase of
λ ) λx ) λz. A spontaneous increase of the width and a
simultaneous decrease of the length marks the phase
transformation to the LR phase. It is caused by the
formation of planar micelles stacked along the z axis,
and the corresponding oblate deformation of the coil that
has to be accommodated in the hydrophobic layers. If
no discontinuity of the volume occurs at the phase
transformation, the discontinuities ∆λx and ∆λz are
related by
Figure 4. Results of the temperature-dependent PFG diffu-
sion measurements given as D2O obstruction factors Q| and
Q . The dotted lines for Q| serve as a guide to the eye.
The other factor are the temperature-dependent proper-
ties of the amphiphile membranes. Similar to most oligo-
(ethylene oxide) ether surfactants, the phase diagram
of the monomer (cf. Figure 1) shows a sponge phase (L3)
above the lamellar phase. This sponge phase results
from the LR phase by partial fusion of adjacent mem-
branes that lead to passages between the water layers
and eventually to the bicontinuous structure of the L3
phase.41 The L3 phase is isotropic and has a different
topology than the LR phase, but on a local scale the
membrane structure is preserved. It is conceivable that
a similar transition in membrane topology occurs in the
network (and also in the linear polymer) but that the
structure above the transition remains anisotropic
because the mobility of the amphiphiles is restricted by
the polymer backbone. The creation of passages between
neighboring water layers occurring at such a masked
LR-L3 transformation could explain the increased dif-
fusivity perpendicular to the layers.
The anisotropy of diffusion on a macroscopic length
scale can be visualized by the direct observation of the
diffusion of a dye. Crystals of methylene blue and
methyl red were put into the center of hydrogel HG33h
at equilibrium degree of swelling. The diffusion of the
dye was followed by taking pictures with a digital
camera. An anisotropy of the diffusion is clearly visible
in Figure 5a. By simply taking the squared ratio of the
extensions of the stain perpendicular and parallel to the
layers, an approximate D /D| ) 5.3 is obtained. An
interpretation of this value in terms of diffusion ani-
sotropy and large-scale defects is not straightforward,
as it depends on the solubilities in the aqueous and
organic parts of the sample. In this regard, the example
of methyl red (Figure 5b) is particularly interesting, as
it exhibits a reversed anisotropy. This more lipophilic
dye appears to prefer crossing subsequent aqueous and
(1 + ∆λx)2(1 - ∆λz) ) 1
(4)
Notice that this behavior is just the opposite of what is
found for hydrogels with a hexagonal phase, for which

9780 Kleinschmidt et al.
Macromolecules, Vol. 38, No. 23, 2005
Figure 6. Schematic plot of the expected ideal hygroelastic
behavior of a lamellar sample with the relative swelling
coefficients parallel (λz) and perpendicular (λx) to the layer
normal.
the length increases and the width decreases.8 Upon
further swelling with water in the supposedly ideal
lamellar phase, that is on the left side of the phase
transformation shown in Figure 6, ∆λx will remain
constant, but ∆λz will continue to grow with a slope that
is steeper than in the isotropic phase. Therefore, λx and
λz will eventually cross each other at higher water
concentration.
The results of the hygroelastic measurements on
samples HG33h and HG65h are shown in Figures 7 and
8. For each sample three types of curves are shown: the
relative swelling coefficients λx and λz (Figures 7a and
2 1/3
8a), the isotropic swelling coefficient λiso ) (λzλx )
(Figures 7b and 8b), and the reduced swelling coef-
ficients ꢀx ) λx/λiso and ꢀz ) λz/λiso (Figures 7c and 8c).
For hydrogel HG33h the three regimes for the iso-
tropic phase, the phase transformation, and the lamellar
phase can be clearly recognized in Figure 7a. At the
phase transformation, which has a certain width be-
cause of the coexistence of isotropic and lamellar phase,
the hydrogel becomes shorter and broader, exactly as
expected (cf. Figure 6). The micellar aggregation forces
the polymer into a flattened, oblate conformation. Once
the phase transformation is complete, the slopes of the
curves change again. There are, however, some signifi-
cant deviations from the predicted behavior. First,
swelling in the L2 phase is not isotropic. Swelling with
water has the same type of anisotropy, namely stronger
swelling along x, as swelling with toluene (see above).
Therefore, the same reason for this effect must be
assumed, that is, a swelling anisotropy caused by the
intrinsic network anisotropy due to stronger cross-
linking between adjacent layers than within a single
layer. Second, in the lamellar phase, λx does not remain
constant but continues to grow. This indicates substan-
tial swelling perpendicular to the layer normal, probably
due to a defective porous lamellar structure, which is
also evident from the diffusion studies. However, swell-
ing is still stronger along z as expected, which is inferred
from the steeper slope of λz compared to λx.
Figure 7. Hygroelastic results for HG33h: (a) swelling factors
of the length (λz) and the width (λx) of the network; (b) isotropic
swelling factor λiso; (c) reduced swelling coefficients ꢀz and ꢀx.
Three consecutive measurements (first to third cycle) are
shown, each starting from the completely dried network.
volume stays almost the same for all cycles, which
makes evident that the measurements are carried out
correctly.
The slope of λiso as a function of the water concentra-
tion shows two discontinuities: one when reaching the
two-phase region (coming from the isotropic L2 phase)
and the other when reaching the pure lamellar phase.
The concentrations at which these discontinuities are
observed are in almost exact agreement with the phase
diagram of the linear polymer, shown in Figure 1. The
discontinuities of the slope, not taken into account in
Figure 6, correspond to the discontinuity of the volume
at the phase transformation and are consistent with a
first-order transition. The change of slope within the
isotropic phase furthermore shows that the partial
molar volumes are not independent of concentration, a
feature that can be explained by the specific interactions
(hydrogen bonds) between the amphiphiles and water
and is also not considered in Figure 6.
The volume degree of swelling q, calculated according
to eq 2 from the experimental data of Figure 7a, serves
as a check of the experimental accuracy. The cubic root
of q, denoted by λiso, is a pseudo-isotropic swelling
parameter describing the mere volume effect. In Figure
7b the calculated values of λiso are depicted as a function
of the water concentration. The small differences in λiso
observed upon repeated measurements shows that the
While Figure 7b illustrates the volume effect upon
swelling with water, a plot of the reduced swelling

Macromolecules, Vol. 38, No. 23, 2005
Lamellar Liquid Single Crystal Hydrogels 9781
lyotropic systems is just a few percent. One explanation
for this difference may be the fact that two opposing
effects contribute to the hygroelastic effect in lyotropics.
On one hand, there is the change of the coil conforma-
tion caused by micellization. In lamellar systems this
leads to a flattening of the coil and a shrinking parallel
to the director (z), while in hexagonal systems an
elongation of the coil and a growth along the director is
caused. In both cases, the second effect, due to the
anisotropic swelling of the hydrophilic regions, is op-
posed to the first one. Swelling of a lamellar structure
leads to an increase along the director, but swelling of
a hexagonal phase leads to an increase perpendicular
to it. As a result, the two contributions partly cancel
each other and the remaining effect is small.
The hydrogel HG65h was synthesized at a different
water concentrations as shown in Table 1. The precursor
mixture has a higher amphiphile concentration than in
sample HG33h. The lamellar phase exhibits its highest
stability in this concentration range (cf. the phase
diagram shown in Figure 1). We therefore expected a
lamellar phase with less defects. The hygroelastic
behavior of HG65h, shown again both in terms of the
directly measured relative swelling coefficients λx and
λy and in terms of the isotropic swelling coefficient λiso
and the anisotropy ꢀ, is plotted in Figure 8. The
characteristics of the swelling curves can be compared
with HG33h. At the first glance, the shrinking along
the z axis at the phase transformation is missing in
Figure 8a. It is obviously completely compensated by
the swelling along z. Figure 8c, however, reveals that
the anisotropic behavior is essentially the same as that
of HG33h. The behavior within the lamellar phase is
noteworthy, since λx stays almost constant as theoreti-
cally expected for an ideal lamellar phase. This indicates
that the lamellar structure is more perfectly ordered
compared to HG33d, a result in agreement with the
diffusion experiments discussed above. Also observed is
the predicted crossing of λx and λz.
Figure 8. Hygroelastic results for HG65h: (a) swelling factors
of the length (λz) and the width (λx) of the network; (b) isotropic
swelling factor λiso; (c) reduced swelling coefficients ꢀz and ꢀx.
IV. Summary and Conclusion
Liquid single crystal hydrogels (LSCH) were synthe-
sized via radical polymerization by incorporating an
amphiphilic cross-linker molecule with reactive groups
in both the hydrophobic and hydrophilic moieties di-
rectly into the lamellar liquid-crystalline phase of a
rigid-rod-like amphiphile that is aligned by a magnetic
field. Hydrogels with a macroscopically uniform director
orientation were obtained that are stable against tensile
forces in all three dimensions. Cross-linking within the
liquid-crystalline phase results in an anisotropy of the
network, which becomes evident by swelling experi-
ments with a nonselective solvent in the isotropic phase.
The structure and ordering of these anisotropic hy-
drogels were investigated by a variety of methods on
micro- and macroscopic length scales. The X-ray cor-
relation length of about seven layers showed a rather
low order on very small scales. The director orientation
averaged over a mesoscopic scale of several micrometers,
which was investigated by deuterium line shape analy-
sis, was found to have a narrow distribution with its
maximum parallel to the magnetic field and a full width
at half-maximum of 13°. Diffusion NMR proved the
existence of defects or pores in the lamellae on a similar
length scale of several micrometers, but based on the
absence of diffusion time-dependent diffusion coef-
ficients gave also an estimation of the minimal extension
coefficients ꢀx ) λx/λiso and ꢀz ) λz/λiso, as shown in Figure
7b makes the anisotropy of swelling evident. Plotting
both ꢀx and ꢀz is redundant as uniaxiality requires that
2
ꢀz ) 1/ꢀx . Deviations of ꢀ from unity directly indicate
an anisotropic deformation of the original network in
the dry state. Thus, the curves in Figure 7c show the
deviation from ideal isotropic swelling behavior and can
therefore be related to an anisotropy of the chain
deformation.
Comparing the different measurement cycles on the
same sample an increase of anisotropy is observed, both
in the isotropic and the lamellar phase. The disappear-
ance of entanglements which are formed during the
process of polymerization by repeated swelling and
deswelling could provide a reasonable explanation of
this effect. A similar behavior was observed in all
samples.
The hygroelastic effect at the phase transformation
is small compared to the corresponding thermoelastic
effect observed in thermotropic nematics. For nematic
side-chain liquid-crystal polymers an increase of their
length up to 40% and more can be measured,42 whereas
the hygroelastic effect observed for the nonionic hydro-
gels of this study and other lamellar5 or hexagonal8

9782 Kleinschmidt et al.
Macromolecules, Vol. 38, No. 23, 2005
(3) Amigo´-Melchior, A.; Finkelmann, H. Polym. Adv. Technol.
2002, 13, 363.
(4) Bauer, H.; Hermes, R. Pharmazie 1995, 50, 481.
(5) Fischer, P.; Finkelmann, H. Prog. Colloid Polym. Sci. 1998,
111, 127.
(6) Fischer, P.; Schmidt, C.; Finkelmann, H. Macromol. Rapid
Commun. 1995, 16, 435.
(7) Lo¨ffler, R.; Finkelmann, H. Makromol. Chem., Rapid Com-
mun. 1990, 11, 321.
of homogeneous domains of the order of 10 µm (parallel
to the director) by 30 µm (perpendicular to the director).
The temperature-dependent obstruction factors for
diffusion parallel to the layer normal showed a charac-
teristic step at around 312-314 K. Whether this is
caused by inherent stresses due to an anisotropic
thermal change of dimensions of the elastomer or
whether it is triggered by a temperature-induced change
of the lyotropic phase structure analogous to an LR-L3
transformation remains to be further investigated.
The macroscopic thermodynamic behavior evident
from swelling experiments is consistent with established
network theories. The direct comparison of the hy-
groelastic response with the thermoelastic correspon-
dent of thermotropic liquid-crystalline elastomers shows
a similar direct mechanical effect at the formation of
the liquid-crystalline phase. However, the hygroelastic
effect, which reflects the coupling between liquid-
crystalline phase structure and the network chain
conformation, is much weaker than its thermoelastic
analogon. The reason for this behavior is still unknown.
Obviously the formation of ordered, nonionic, micellar
aggregates does not require considerable rearrangement
of the network strands. Here it must be noted that very
recent results on ionic systems reveal a much stronger
hygroelastic response, which can be explained by an
increased Kuhn length of the main-chain segments due
to the ionic charges.
(8) Weiss, F.; Finkelmann, H. Macromolecules 2004, 37, 6587.
(9) Keller-Griffith, R.; Ringsdorf, H.; Vierengel, A. Colloid Polym.
Sci. 1986, 264, 924.
(10) Ringsdorf, H.; Schlarb, B.; Venzmer, J. Angew. Chem. 1988,
100, 117.
(11) Lindner, N.; Ko¨bel, M.; Sauer, C.; Diele, S.; Jokiranta, J.;
Tschierske, C. J. Phys. Chem. B 1998, 102, 5261.
(12) Lu¨hmann, B.; Finkelmann, H. Colloid Polym. Sci. 1987, 265,
506.
(13) Schafheutle, M. A.; Finkelmann, H. Liq. Cryst. 1988, 3, 1369.
(14) Lee, M.; Cho, B.-K.; Kim, H.; Zin, W.-C. Angew. Chem. 1998,
110, 661.
(15) Amigo´-Melchior, A.; Finkelmann, H. Colloid Polym. Sci. 2002,
280, 207.
(16) Schafheutle, M. A.; Finkelmann, H. Liq. Cryst. 1988, 3, 1369.
(17) Brodowsky, H. M.; Boehnke, U.-C.; Kremer, F. Langmuir
1999, 15, 274.
(18) Stannarius, R.; Ko¨hler, R.; Dietrich, U.; Lo¨sche, M.; Tolksdorf,
C.; Zentel, R. Phys. Rev. E 2002, 65, 041707.
(19) Schu¨ring, H.; Stannarius, R.; Tolksdorf, C.; Zentel, R. Mac-
romelecules 2001, 34, 3962.
(20) Gebhard, E.; Zentel, R. Macromol. Chem. Phys. 2000, 201,
902.
(21) Brodowsky, H. M.; Kremer, F.; Gebhard, E.; Zentel, R.
Ferroelectrics 2000, 243, 115.
The comparison of two different hydrogels that were
synthesized with a different content of water showed
that ordering and structure are much better in the
system with the lower content of water whose composi-
tion is close to the one with maximum temperature of
the lamellar phase. Consistent results pointing out the
difference in order were obtained by such different
methods as diffusion NMR and hygroelastic measure-
ments.
(22) Mitsunobu, O. Synthesis 1981, 1.
(23) Stejskal, E. J. Chem. Phys. 1965, 43, 3597.
(24) Stejskal, E.; Tanner, J. J. Chem. Phys. 1965, 42, 288.
(25) Callaghan, P. T. Principles of Nuclear Magnetic Resonance
Microscopy; Clarendon Press: New York, 1993.
(26) Mills, R. J. Phys. Chem. 1973, 77, 685.
(27) Assfalg, N.; Finkelmann, H. Macromol. Chem. Phys. 2001,
202, 794.
(28) Friedel, G.; Grandjean, F. Bull. Soc. Fr. Mineral. 1910, 33,
409.
Future investigations should include the design of
new types of amphiphiles which allow an end-on archi-
tecture of the polysoaps instead of the side-on architec-
ture presented here. The latter is expected to disturb
the order of the amphiphiles, and better ordered phases
are expected for polysoaps with end-on attached am-
phiphilic chains. This should also help to understand if
and how the type of attachment influences the coupling
between liquid-crystalline phase and polymer network.
(29) Gray, G. W.; Goodby, J. W. Smectic Liquid Crystals; Leonard
Hill: Glasgow, 1984.
(30) Benton, W. J.; Fort, T. J.; Miller, C. A. 53rd. Ann. Fall.
Technol. Conf. and Exh. of Soc. Pet. Eng of AIME, Houston
TX, 1978.
(31) Benton, W. J.; Miller, C. A. Prog. Colloid Polym. Sci. 1983,
68, 71.
(32) Lovell, R.; Mitchell, G. R. Acta Crystallogr. A 1981, 37, 135.
(33) Mitchell, G.; Windle, A. Developments in Crystalline Polymers;
Basset, D., Eds.; Elsevier: London, 1984; p 140.
(34) Adhikari, B.; Vora, R. A.; Paul, R. Mol. Cryst. Liq. Cryst. 1996,
287, 129.
Acknowledgment. The authors are greatly in-
debted to the Deutsche Forschungsgemeinschaft (Sonder-
forschungsbereich 428) and the Fonds der Chemischen
Industrie for financial support. They thank Laura
Hartmann for her help with the dye diffusion experi-
ments.
(35) Honerkamp, J.; Weese, J. Continuum Mech. Thermodyn.
1990, 2, 17.
(36) Weese, J. Comput. Phys. Commun. 1992, 69, 99.
(37) Winterhalter, J.; Maier, D.; Grabowski, D. A.; Honerkamp,
J.; Mu¨ller, S.; Schmidt, C. J. Chem. Phys. 1999, 8, 4035.
(38) Gaemers, S.; Bax, A. J. Am. Chem. Soc. 2001, 123, 12343.
(39) Callaghan, P. T. Aust. J. Phys. 1984, 37, 359.
(40) Kleinschmidt, F. Dissertation, University Freiburg, 2005.
(41) Porte, G.; Appell, J.; Bassereau, P.; Marignan, J. J. Phys.
(Paris) 1989, 50, 1335.
References and Notes
(1) de Gennes, P. G. C. R. Sce´ances Acad. Sci., Ser. B 1975, 281,
101.
(2) Finkelmann, H.; Kock, H. J.; Rehage, G. Makromol. Chem.,
Rapid Commun. 1981, 2, 317.
(42) Ku¨pfer, J.; Finkelmann, H. Makromol. Chem., Rapid Com-
mun. 1991, 12, 717.
MA051479O