Buckling Instability in Liquid Crystalline Physical Gels

Rafael Verduzco,
1
Guangnan Meng,
2
Julia A. Korneld,
1
and Robert B. Meyer
2
1
California Institute of Technology, MC 210-41,Pasadena, California 91125, USA
2
The Martin Fisher School of Physics, Brandeis University, Waltham, Massachusetts 02454-9110, USA
(Received 1 December 2005; revised manuscript received 28 February 2006; published 13 April 2006)
In a nematic gel we observe a low-energy buckling deformation arising from soft and semisoft elastic
modes. We prepare the self-assembled gel by dissolving a coil side-group liquid-crystalline polymer
coil copolymer in a nematic liquid crystal. The gel has long network strands and a precisely tailored
structure, making it ideal for studying nematic rubber elasticity. Under polarized optical microscopy we
observe a striped texture that forms when gels uniformly aligned at 35

C are cooled to room temperature.

We model the instability using the molecular theory of nematic rubber elasticity, and the theory correctly
captures the change in pitch length with sample thickness and polymer concentration. This buckling
instability is a clear example of a low-energy deformation that arises in materials where polymer network
strains are coupled to the director orientation.
DOI: 10.1103/PhysRevLett.96.147802 PACS numbers: 61.30.v, 61.41.+e, 82.70.Gg
Liquid-crystalline elastomers and gels combine the ori-
entational order of liquid crystals and the translational
order of a polymer network. The coupling of these two
effects in liquid-crystalline elastomers and gels gives rise
to unique phenomena such as soft elasticity in which
certain director rotations and network strains cost no free
energy in the ideal case. Soft elastic modes are predicted
and well understood theoretically by a molecular theory of
nematic rubber elasticity [13]. However, a signicant
limitation in testing existing theories is the difculty in
preparing well-characterized networks. Many approaches
have been undertaken to prepare liquid crystal elastomers,
including free-radical polymerization of bulk liquid crystal
acrylates and diacrylates [4], free-radical polymerization
of diacrylates in the presence of liquid crystals [5,6], and
side-coupling reactions with poly(methylsiloxane) [7].
Unfortunately, these approaches result in nematic elasto-
mers and gels with heterogeneities in the polymer network
and, in the case of many gels, phase separation of the
polymer network from the liquid crystal host [8,9]. More
recently, self-assembly of block copolymers has been used
as an approach to create model liquid crystal networks
[1012]. We have utilized self-assembly of a triblock
copolymer that has a narrow molar mass distribution and
a very long midblock to produce well-dened nematic gels
[10,11]. This gel facilitates comparison to present theories
which assume the material is lightly cross-linked and
single phase.
In this Letter, we present a study of the effects of
nematic rubber elasticity in thin layers of uniformly
aligned gels contained between rigid parallel plates.
Using polarized optical microscopy, we observe a banded
texture that arises in response to small temperature
changes. We propose that this texture represents a buckling
deformation made possible by the presence of soft and
semisoft elastic modes. We formulate a description based
on the molecular theory of nematic elastomers that incor-
porates soft and semisoft elasticity as well as liquid crystal
elasticity, also known as Frank elasticity. The theory cor-
rectly accounts for the observed dependence of the pitch
and threshold temperature on the sample thickness and
polymer concentration.
The self-assembled gels are prepared by dissolving a
high molecular weight (1200
kg
mol
) polystyreneside-
group liquid-crystalline polymer polystyrene triblock co-
polymer in the nematic liquid crystal 4-cyano-4-n-
pentylbiphenyl (5CB) (Fig. 1) [10,11]. The polystyrene
block is considerably less soluble than the liquid-
crystalline block in 5CB, and it aggregates to form physical
cross-links. The resulting self-assembled network has a
uniform molecular weight between cross-links.
In this work, we chose to study gels with polymer
concentrations of 10 wt. % or less where the polymer
rigidity is important but not dominant. Samples were
loaded into cells with well-dened gap spacings and align-
ment layers. For gaps smaller than 10 m, we used cells
purchased from LC Vision. The cells consist of indium-tin-
oxide coated glass plates with a layer of SiO
2
to induce
planar alignment. For gaps larger than 10 m, we used
glass slides with teon spacers and a rubbed polyimide
layer for alignment. The cell thickness was determined
FIG. 1. Chemical structure of liquid crystal 5CB and the end-
associating triblock copolymer used in this study. The midblock
is a random copolymer of liquid crystal side-group and butadiene
monomer. The polydispersity (
M
v
M
n
) of the triblock is approxi-
mately 1.6 as measured by gel-permeation chromatography.
PRL 96, 147802 (2006)
P HYS I CAL RE VI E W L E T T E RS
week ending
14 APRIL 2006
0031-90070696(14)147802(4)$23.00 147802-1 2006 The American Physical Society
with a uv-visible spectrophotometer by interferometry for
gaps smaller than 10 m [13] or by measuring the thick-
ness with a micrometer for larger gaps. The nematic gels
have a sol-gel transition temperature that is approximately
equal to the nematic-isotropic phase transition temperature
(T
NI
) of 5CB, 35.2

C, and the gel loads easily by capillary

action when heated above this temperature.
A magnetic eld or shear are used to uniformly align the
gel. In the rst case, the gel is placed in an 8.8 T magnetic
eld oriented parallel to the desired director orientation,
heated above its T
NI
, and allowed to cool to room tempera-
ture. After removing the cell from the magnetic eld, the
cell is heated to 34

C and then cooled back down to room

temperature at 1

C min to remove any residual irregu-

larities. In some cases it is easier to use shear to align the
sample. This is because alignment by magnetic eld is
more sensitive to bubbles and defects in the sample. In
order to shear align the sample, the sample is loaded into a
Mettler FP 82 hot stage and the temperature is set to
34.5

C while the top slide is carefully moved back and

forth across the sample.
Gels that are uniform monodomains at 35

C develop a
striped texture when cooled to room temperature (Fig. 2).
The texture reversibly disappears and reappears when the
gel is heated above and then cooled below a threshold
temperature for stripe formation. Below this threshold
temperature, the pitch length is independent of tempera-
ture. The stripes run perpendicular to the initial director
orientation n
o
. Rotating the crossed polarizers to approxi-
mately 15

maximizes contrast between dark and bright

bands, and the stripes that appear dark at +15

appear
bright at 15

and vice versa. The stripes are less evident

but still visible when the polarizers are aligned with n
o
.
This indicates that the director rotates within the plane and
in opposite directions in adjacent stripes but does not
preclude an out-of-plane rotation of the director coupled
with this in-plane rotation.
The characteristics of the striped texture depend on the
sample thickness and polymer concentration. The pitch is
larger in thicker samples and more dilute gels (Fig. 3).
Stripes do not appear when the gap is too thin or the
polymer concentration too dilute. In the case of a 5 wt. %
gel, stripes do not appear in 4 and 9 m thick cells but they
do appear at larger gap thicknesses. For 10 wt. % gels,
stripes appear even in thin cells. Stripe formation is thus
favored in larger gaps and more concentrated gels.
A periodic modulation of the director is unexpected
because it introduces a free energy contribution from
Frank elasticity. [A periodic striped texture has been ob-
served in very thin (<1 m) nematic lms due to surface
gradient terms in the elastic free energy expression [14,15].
Our samples are thick enough that the surface gradient
terms are not signicant.] The appearance of the stripes
can be accounted for, however, by including the rubber
elastic energy of the polymer network. A side-chain liquid-
crystalline polymer dissolved in a nematic solvent adopts
an anisotropic chain conformation coupled to the nematic
order of the solvent [16]. Based on prior literature on
nematic elastomers [1] we expect that an unconstrained
freestanding gel will change shape: in the case of a strongly
prolate alignment tendency, the gel will expand in the
direction of director orientation with decreased tempera-
ture (our gel is too soft to carry out the same experiment).
When constrained between glass plates, however, expan-
sion of the gel in the direction of increasing anisotropy is
prevented by the boundary conditions, and the network
becomes increasingly frustrated as the temperature de-
creases. The stripe instability represents a low-energy de-
formation incorporating a periodic network strain eld
coupled to a periodic rotation of the director. The defor-
mation reduces the rubber elastic energy relative to the
undeformed state at the cost of a Frank elastic penalty.
We formulate a molecular model of the instability by
treating the gel as a nematic elastomer, as has been pre-
viously done successfully [1,17]. We start with a macro-
scopically uniform nematic gel in which the director n is
aligned in a unique direction parallel to the substrates of a
dielectric cell. The polymer strands in the gel have an
anisotropic conformation described by a step length tensor
l
|]
l
J
u
|]
+(l
L
l
J
)n
|
n
]
where u
|]
is the identity tensor.
The network anisotropy r is dened as the ratio
l
L
l
J
. The
anisotropy r is a temperature dependent property that in-
FIG. 2. A 5 wt. % gel in a 25 m gap at 20.0

C viewed under
a polarizing optical microscope. The orientation of the director
imprinted during cross-linking n
o
and the orientation of the
crossed polarizers is shown in the top left-hand corner of each
image. Note that the dark lines in the left-hand image become
bright in the right-hand image and vice versa.
FIG. 3. Dependence of pitch on gap in 10 and 5 wt. % polymer
gels.
PRL 96, 147802 (2006)
P HYS I CAL RE VI E W L E T T E RS
week ending
14 APRIL 2006
147802-2
creases with decreasing temperature for our system. The
anisotropy during cross-linking is denoted as r
o
.
Below a critical temperature, r becomes sufciently
large that the gel buckles. The buckling instability repre-
sents a director rotation coupled to a network strain eld.
We assume that the director stays in the x-y plane and that
the director orientation depends on x and z but is indepen-
dent of y (Fig. 4). Coupled to this director rotation is a
network strain eld described by the deformation gradient
tensor
|]
. At the interface with the cell (z
J
2
), the
substrates impose the boundary conditions of no-slip and
that n must be aligned along x.
The free energy density consists of three terms: ideal
polymer network elasticity, nonideal polymer network
elasticity, and Frank elasticity [Eq. (1)]. The ideal nematic
elastic term is weighted by . The nonideal contribution is
weighted by n and describes the semisoft elasticity
observed experimentally [3]. Semisoftness arises from
compositional uctuations and from a memory of the
aligned state imprinted during cross-linking. The last
term is the Frank elastic term. We make the one-constant
approximation [18] with Frank elastic constant K.
E
1
2
(Tr|l
o

T
l
1
|)
+
1
2
nTr|(u n
o
n
o
)
T
n n |
+
1
2
K|(\ n)
2
+(n \n)
2
+(n \n)
2
|. (1)
We analyze the behavior near the transition to the striped
state where we assume small values for the amplitudes of
the director rotation and the displacements. We propose
simple sinusoidal functions for the director rotation w and
the displacement eld [Eqs. (2)], with
|
denoting the
displacement along the | axis. These functions vary sinu-
soidally in the x direction with wave vector k, and the wave
vector q has the value
r
J
so that the director and displace-
ment elds satisfy both the homogeneous boundary con-
dition and the no-slip boundary condition at the substrates
(z
J
2
). The terms and ~ set the amplitudes of the
rotation and displacement elds, respectively.
w(x. z) cos(qz) cos(kx).

y
~cos(qz) sin(kx).
x

z
0.
(2)
We calculate the deformation gradient tensor from the
displacement eld (
|]
u
|]
+

|
x
]
). Substitution of and
n into Eq. (1) gives the free energy density of an incom-
pressible gel subject to the periodic deformation and direc-
tor rotation described by Eqs. (2). We Taylor expand this
function in around 0 and keep only terms of O(
2
)
or greater. Next, we integrate the simplied free energy
density over one wavelength of the deformation to obtain
the average free energy density of the gel. Subtracting the
energy of the uniformly aligned gel from this free energy
gives the free energy relative to the monodomain state
[Eq. (3)].
E
1
8
K
2
(k
2
+q
2
) +

1
8
~
2
(q
2
+k
2
r
o
) +
r
o
4
~k

1
r
1

+
1
8

r
o
1+n+
1
r

r
o
r

. (3)
A crossing from a positive to a negative value in the free
energy indicates an onset of the buckling instability. The
rst term in Eq. (3) is the contribution from Frank elasticity
and is always positive, as expected. The contribution from
nonideal network elasticity, the term weighted by n, is also
always positive and increases the threshold to the striped
state. In a semisoft gel (nonzero n) a small but nite
cooling is always required for the appearance of the insta-
bility. The soft elastic part contributes negative terms to the
free energy expression which are responsible for the ap-
pearance of the buckling instability.
In order to compute an anisotropy and wave vector, we
look for the minimum value of r and a critical wave vector
k for which the right combination of innitesimal displace-
ment and rotation costs zero energy, signalling an insta-
bility as r passes through this value. For slightly larger r
values, the amplitude of the instability grows rapidly, with
the wavelength determined at the instability.
We do not know the values of many of the material
parameters in the model, but they can be reasonably ap-
proximated from independent measurements or from
known properties of the liquid crystal solvent 5CB. For
example, the liquid crystal elastic constants have been
measured previously for 5CB at room temperature
[13,19] and have values of about 1 10
11
N. We mea-
sured the rubber modulus in a cone-and-plate rheometer
and obtained a value of approximately 220 Pa for a
10 wt. % gel and 50 Pa for a 5 wt. % gel [10,11]. We set
the initial anisotropy r
o
to the small value of 1.2. In
addition to these parameters, we reasonably approximate
the semisoftness parameter n to have a value of 0.05 based
on previous studies of nematic elastomers [17].
A comparison of the model prediction of the pitch and
experimental results shows good qualitative agreement
(Fig. 3). The model predicts the trends of increasing pitch
with increasing gap and decreasing polymer concentration.
The model also accounts for the disappearance of stripes at
small gaps from a prediction of the threshold anisotropy
(Fig. 5). Neutron scattering measurements show the an-
isotropy of the present midblock at 25

C to be 10 [10].
FIG. 4. Nematic gel director orientation in (a) the initial uni-
formly aligned state and (b) the striped state. The cell substrates
are at z
J
2
.
PRL 96, 147802 (2006)
P HYS I CAL RE VI E W L E T T E RS
week ending
14 APRIL 2006
147802-3
Therefore, the model predicts that bands are suppressed for
J < 3.7 m for the 5 wt. % gel and for J < 1.9 m for the
10 wt. % gel.
In very thick samples (q - 0), the buckling instability is
a soft mode that carries no penalty from nematic rubber
elasticity. This can be seen in the free energy expression
[Eq. (3)]: in the limit q -0, the negative contribution
r
o
~k(
1
r
1) can always compensate the positive con-
tributions for any k by properly selecting ~ and . When
the sample has a nite thickness, however, the boundary
conditions necessitate gradients from the center to the
walls of the gap. The resulting gradient

y
z
a shearing
deformationis not a soft mode. Nevertheless, the defor-
mation is semisoft, and the cost to nematic rubber elasticity
is small. The reason that the elastic cost is small is also
shown by Eq. (3): by selecting a large spacial frequency k,
the positive contribution from the ~
2
q
2
term is negligible
and the negative term can compensate for the positive
elastic contribution with a sufciently large amplitude of
director rotation . Thus, in the absence of Frank elasticity,
the gel would favor a small wavelength distortion.
However, Frank elasticity goes inversely with the square
of the wavelength, making very short wavelengths costly
in energy. The nal wavelength is a compromise of ne-
matic rubber elasticity and Frank elasticity. The nal
wavelength is set by the rubber modulus , the Frank
elastic constant K, the initial anisotropy r
o
, and the semi-
softness parameter n.
We have presented a stripe distortion in nematic gels and
shown that it arises due to the coupling of orientational and
translational degrees of freedom in nematic elastomers. A
model based on the molecular theory of nematic rubber
elasticity accounts for the observed experimental features:
the onset of the instability with changes in the network
anisotropy, the increase in the pitch with increasing sample
thickness, the decrease in the pitch with increasing poly-
mer concentration, and the disappearance of the instability
at small sample thicknesses.
This work was supported by the Department of Defense,
the James Irvine Foundation, AFOSR LC-MURI (F49620-
97-1-0014), NSF Grant No. DMI-0523083, the W. M. Keck
Foundation, and at Brandeis NSF Grant No. DMR-
0322530.
[1] M. Warner and E. M. Terentjev, Liquid Crystal Elastomers
(Oxford University Press, Oxford, 2003).
[2] M. Warner, P. Bladon, and E. M. Terentjev, J. Phys. II
(France) 4, 93 (1994).
[3] H. Finkelmann, I. Kundler, E. M. Terentjev, and
M. Warner, J. Phys. II (France) 7, 1059 (1997).
[4] D. J. Broer, H. Finkelmann, and K. Kondo, Makromol.
Chem. 189, 185 (1988).
[5] R. A. M. Hikmet, J. Appl. Phys. 68, 4406 (1990).
[6] C.-C. Chang, L.-C. Chien, and R. B. Meyer, Phys. Rev. E
56, 595 (1997).
[7] J. Kupfer and H. Finkelmann, Makromol. Chem., Rapid
Commun. 12, 717 (1991).
[8] A. Jakli, D. R. Kim, L. C. Chien, and A. Saupe, J. Appl.
Phys. 72, 3161 (1992).
[9] P. Gautier, M. Brunet, J. Grupp, L. Noirez, and
E. Anglaret, Phys. Rev. E 68, 011709 (2003).
[10] M. D. Kempe, N. R. Scruggs, R. Verduzco, J. Lal, and J. A.
Korneld, Nat. Mater. 3, 177 (2004).
[11] M. D. Kempe, R. Verduzco, N. R. Scruggs, and J. A.
Korneld, Soft Matter (to be published).
[12] M.-H. Li, P. Keller, J. Yang, and P.-A. Albuoy, Adv. Mater.
16, 1922 (2004).
[13] I.-C. Khoo and S.-T. Wu, Optics and Nonlinear Optics of
Liquid Crystals (World Scientic, Singapore, 1993).
[14] O. D. Lavrentovich and V. M. Pergamenshchik, Phys. Rev.
Lett. 73, 979 (1994).
[15] V. M. Pergamenshchik, Phys. Rev. E 61, 3936 (2000).
[16] M. D. Kempe, J. A. Korneld, and J. Lal, Macromolecules
37, 8730 (2004).
[17] E. M. Terentjev, M. Warner, R. B. Meyer, and
J. Yamamoto, Phys. Rev. E 60, 1872 (1999).
[18] P. G. DeGennes and J. Prost, The Physics of Liquid
Crystals (Clarendon Press, Oxford, 1993).
[19] K. Skarp, S. T. Lagerwall, and B. Stebler, Mol. Cryst. Liq.
Cryst. 60, 215 (1980).
FIG. 5. Predicted value of the threshold anisotropy r
th
as a
function of gap thickness J and rubber modulus . Bands form if
the polymer has r > r
th
. The values of correspond to those of
the 5 and 10 wt. % gels.
PRL 96, 147802 (2006)
P HYS I CAL RE VI E W L E T T E RS
week ending
14 APRIL 2006
147802-4