PHYSICAL REVIEW E VOLUME 55, NUMBER 2 FEBRUARY 1997

Interactions between particulate inclusions in a smectic-A liquid crystal

M. S. Turner 1,2, * and P. Sens 3

1
Department of Chemical Engineering, University of California at Santa Barbara, Santa Barbara, California 93106
2
TCM Group, Cavendish Laboratory, Madingley Road, Cambridge, CB3 0HE, United Kingdom
3
Departments of Physics and Materials, University of California at Santa Barbara, Santa Barbara, California 93106
~Received 1 November 1996!
We study theoretically the smectic mediated interactions between pointlike inclusions in a lamellar phase.
The two-body interaction is derived and we discuss the virial expansion in the limits of weak and strong
interactions. The leading order finite density corrections to the interaction potential are calculated using a mean
field theory, which allows the particle density, potential, and effective smectic moduli to be calculated self-
consistently. The effective bending modulus of the membranes is found to increase linearly with the particle
density. We also give the structure factor S(q) for small but finite particle density. @S1063-651X~97!50602-7#

PACS number~s!: 61.30.Cz, 82.70.Dd

INTRODUCTION clusions in Sec. V. Throughout, we employ units in which

k B T51.
It is well known that the physical properties of many sys-
tems are highly sensitive to the presence of impurities or I. THEORY FOR PARTICLE-SMECTIC COUPLING
defects. A dramatic example is the increased conductivity
observed in impure semiconductors @1#. Liquid crystalline In the smectic phase the layers are on average flat and
systems have been of interest to physicists for many decades equidistant with layer spacing d and normal in the z direc-
@2# and it seems natural to seek to understand the role of tion. Deformation of the average layer position away from
impurities in these systems too. We might also be motivated this state may conveniently be parametrized by the continu-
by the numerous industrial applications of liquid crystals or a ous scalar displacement field u, which represents the normal
marked similarity with certain biological systems, e.g., cell displacement of the layers in the z direction. Such a descrip-
membranes, which are known to contain many impurities tion leads to the so-called Landau–de Gennes Hamiltonian
@3,4#. Some theoretical @5# and experimental @6# studies of @2,11,12#. We model the effect of pointlike particles in the
the interaction between particles embedded in a single ~or smectic phase by including a term ; r] z u in the energy den-
pair of! membrane~s! have been carried out. In addition we sity that represents the lowest order coupling between the
recently @7# presented a description of aggregation processes particle density r (r) and the local layer compression ~or ex-
involving particles incorporated into a bulk (3d) lamellar pansion! ] z u. ~The reader may be surprised to learn that
phase in which the aggregates were treated as noninteracting. coupling terms like ; r ¹ 2i u lead to identical results. This
Our interest in such bulk systems has been fueled by recent subtle point will be discussed in a future article.! We omit
experimental studies @8,9#. It is our aim here to study the terms scaling like ; r 2 , which corresponds to direct interpar-
interactions between particulate inclusions residing in such a ticle interactions, as these depend on microscopic details of
bulk smectic phase. the particles involved. It is our aim here to study only the
This paper is organized as follows: In Sec. I we briefly membrane-mediated interparticle interactions. Our coupling
discuss the background theory and introduce our model for term is formally the simplest in that it is the only term qua-
the particle-layer interactions. We obtain the interaction dratic in $ r ,u % involving only a first derivative that is accept-
potential for two particles embedded in a bulk smectic-A able on symmetry grounds. The Hamiltonian H now reads
phase. We employ these results in Sec. II to construct a
self-consistent mean field theory for the two-particle interac-
tions at finite density. We calculate the linear correction to
H5
B
2
E d 3 r@~ ] z u ! 2 1l 2 ~ ¹ 2i u ! 2 1 b r] z u # . ~1!
the interaction potential and find that this can be interpreted
in terms of a stiffening of the membranes, with their com- In this expression ¹ i is the gradient operator in the x-y
pressibility unchanged. This is in contrast to one earlier plane, l5 AK/B is a length characteristic of the smectic
study, albeit on a somewhat different system, which pre- ~typically of the order of the layer spacing d), and B and
dicted a decrease in the bending modulus @10#. In Sec. III K are respectively the compressional and bending moduli of
we discuss the static structure factor for scattering from the smectic. The coupling constant b , with dimensions of
impurities embedded in a smectic phase. Finally we briefly volume, controls the amplitude of the local deformation. Its
discuss the far-field contribution to the second virial co- value depends on the microscopic details of the particle-layer
efficient for finite particle densities. We present brief con- interactions. The particles either provide a local inwards
pinch ( b .0) or outwards push ( b ,0) to the neighboring
layer. Certain proteins @3,9# are known to bind to aqueous
*Present address: Department of Physics, Warwick University, surfactant membranes more or less fixing the layer separation
Coventry CV4 7AL, United Kingdom very locally. In such systems b could be chosen so that a

1063-651X/97/55~2!/1275~4!/$10.00 55 R1275 © 1997 The American Physical Society

R1276 M. S. TURNER AND P. SENS 55

single particle fixes the local layer separation at d * . Such a ized version of the resulting equations to obtain corrections
condition may be shown to imply @7# b .ld(d2d * ). to the two-body interaction potential.
Minimization of Eq. ~1! and subsequent calculations are We calculate the potential field f (r) due to a particle at
most easily performed in Fourier space. The Fourier trans- the origin and its associated correlated neighbors. Thus the
form is defined by f q5 * d 3 rf (ri ,z)e i(qi •ri 1q z z) with ri and energy required to move an infinitesimal number of inclu-
qi vectors in the x-y plane. Minimizing Eq. ~1! with respect sions d n from infinity to r is d n f (r).
to u q the energy is given by @7# Invoking Boltzmann statistics the density field near the
particle is of the form
H5 E d 3q
G r r 5
~ 2 p ! 3 q q 2q
E 8E
d 3r d 3 rG ~ r2r8 ! r ~ r! r ~ r8 ! , r ~ r! 5 r̄ e 2 f ~ r! 1 d ~ r! , ~6!
~2! where r̄ is the density at infinity and the term d (r) fixes one
where particle at the origin.
We may linearize Eq. ~6! whenever f !1, which is satis-
q 2z fied whenever the energy scale E 0 !1. The linearized form
G q52E 0 ld 2
, ~3! of Eq. ~6! is
q 2z 1l 2 q 4i
r ~ r! 5 r̄ @ 12 f ~ r!# 1 d ~ r! ⇒ r q5 r̄ @~ 2 p ! 3 d ~ q! 2 f q# 11.
with ~7!
AKB b 2 Our second equation relates the potential f to the density via
E 05 , ~4! the bare two-body Green’s function G. From Eq. ~2! we have
8l 2 d 2

a number characterizing the energy scale of the interactions

and, in real space
f ~ r! 5 E d 3 r8 G ~ r2r8 ! r ~ r8 ! ⇒ f q5G qr q . ~8!

G ~ r! 5 S DS
E0 d
8p z
2
12
r 2i
4l u z u
D exp2
r 2i
4l u z u
. ~5!
Solving Eqs. ~7! and ~8! for f q and substituting Eq. ~3! we
find @14#

E 0 ld 2 q 2z
The fact that Eq. ~5! diverges in the limit z→0 does not f q52 , ~9!
12 r̄ E 0 ld 2 q z 1l 8 q i
2 2 4
reflect an underyling problem in the theory since the energy
is well behaved for densities r (r) that are smoothly varying
@13#. The two-body interaction potential for z@d is approxi- where
mately U 12(r)52G(r) @the factor 2 arises because of the l 8 5 ~ 12 r̄ E 0 ld 2 ! 21/2l, ~10!
intrinsic double counting in Eq. ~2!#.
From Eq. ~5! we see that the interaction between two a result that is valid for q z !1/d. In some sense Eq. ~10!
particles is long ranged in the z direction and is anisotropic indicates that the range of the interactions is increasing with
and radially nonmonotonic. This anisotropy is characterized the particle density. Transforming Eq. ~9! we find that the
by the paraboloid r 2i 54l u z u , which also appears in the clas- real space potential is well approximated by the following
sical problem of the interaction between two dislocations in form for z@d:
smectic A @2#, although the precise functional form is rather
different in this case. One can also show that the energy scale
for the interaction potential E 0 is equal to the ~self-! energy
of a single isolated particle a E 0 , to within a numerical pref-
f ~ r! 5
E0
8 p ~ 12 r̄ E 0 ld 2 ! 1/2
S DS
d
z
2
12
r 2i
4l 8 u z u
D exp2
r 2i
4l 8 u z u
.
~11!
actor a of order unity @7#. A quantitative estimate of this
prefactor formally requires an improvement in Eq. ~1! and We may calculate the correction to the zero density interac-
also depends on microscopic details. tion potential either by expanding Eq. ~11! directly in powers
of r̄ or, equivalently, by expanding Eq. ~9! before transform-
II. FINITE DENSITY CORRECTIONS TO THE TWO- ing to real space. Writing the real space potential
BODY INTERACTION: A SELF-CONSISTENT f 5G1 d f we find

S DF S DG
MEAN FIELD THEORY
r̄ ld 2 E 20 d 2
r 2i r 2i 2
r 2i
We now examine the effect of surrounding particles on d f ~ r! 5 11 2 exp2 ,
16p z 4l u z u 4l u z u 4l u z u
the two-body interaction potential. We will employ a mean ~12!
field theory in which the particle density and potential fields
are determined self-consistently. Such an approach has been where we neglect terms of O( r̄ 2 ). This function is plotted in
employed with success in many other physical problems. Fig. 1. This finite density correction to the interactions can
Perhaps the best known example is the Debye-Hückel theory be identified with weakly correlated three-body interactions.
for electrolytes. In the present work we use the smectic po- Finally we can examine the influence of the particles on
tential f ~in k B T units! determined via our modified the effective moduli K 8 and B 8 . These moduli are those
Landau–de Gennes functional ~1!. We assume Boltzmann which mimic the effect of the finite particle density. In the
statistics for the distribution of particles and solve the linear- limit of zero particle density they are merely the bare smectic
55 INTERACTIONS BETWEEN PARTICULATE INCLUSIONS . . . R1277

system is well described by the ; r] z u term in Eq. ~1!. Sec-

ondly the finite density corrections calculated in Sec. II pre-
dict variation of the length l 8 with particle density according
to Eq. ~10!.

IV. AVERAGE PARTICLE INTERACTIONS AND THE

SECOND VIRIAL COEFFICIENT

For dilute systems an expansion of the free energy density

F/V in powers of the particle concentration r̄ can usually be
performed. Provided that such an analytic expansion exists
the second virial coefficient B V is given by @16#
FIG. 1. Plot of the bare potential, or Greens function G(r) ~solid
line! Eq. ~5! and the finite density corrections to the potential d f
~broken line! Eq. ~12! against r̃ i 5r i /(4l u z u ) 1/2. The function
G(r) is shown in units where (E 0 /8 p )(d/z) 2 51 and the correc-
B V5
1
2
E d 3 r~ 12e 2U 12~ r! ! , ~17!

tion d f is in units where ( r̄ ld 2 E 20 /16p )(d/z) 2 51.

where U 12 is the two-body interaction potential for particles
separated by a vector r and the zero of free energy is taken to
moduli K and B introduced in Sec. I. However, identifying be the pure smectic phase. If we consider only the long range
the scaling E 0 } AKB Eq. ~4! and l5 AK/B we see that membrane mediated interactions, neglecting any additional
E 80 5E 0 (12 r̄ E 0 ld 2 ) 21/2 and Eq. ~10! together imply a microscopic short range forces, we can estimate the far field
renormalization of K to contribution to B V , in both the limits E 0 !1 and E 0 @1.
The weak interaction limit E 0 !1 may often be appropri-
K ate for lyotropic phases, in which the characteristic energy
K 85 , ~13!
12 r̄ E 0 ld 2 scale E 0 &1. In this regime the far field contribution B far
V may
be calculated by expanding the exponential factor in Eq.
with B 8 5B unchanged at leading order. Equation ~13! indi- ~17!. The second order term is found to give the leading
cates an increase in the effective bending modulus with par- order far field contribution
ticle density.
B V .2E 20 ld 2 for E 0 !1, ~18!
III. STATIC STRUCTURE FACTOR
which is negative, implying that the net interactions are at-
Static scattering measurements measure the ensemble av- tractive. This result necessarily neglects all additional short
erage of the Fourier transform of the density-density corre- range interactions and involves crudely cutting off the vol-
lation function @15# ume integral in Eq. ~17! at z5 n d ~with n of order unity!.

KE E L
The numerical prefactor depends on the choice of this cutoff.
1 The opposite, strong interactions, limit may often be rel-
S ~ q! 5C a2 d 3 rd 3 r8 r ~ r! r ~ r8 ! e iq~ r2r8 ! , ~14!
N evant for particles incorporated in thermotropic liquid crys-
tals, such as the diblock copolymer lamellar phases. In this
where C a is the scattering amplitude per inclusion and N is limit we can no longer expand the exponential e 2U 12 in Eq.
the total number of particles. We wish to calculate this func- ~17! everywhere but can make a crude estimate of the far
tion for scattering from inclusions incorporated in a lamellar field contribution to B V by ~i! integrating Eq. ~17! by steepest
phase and assume that this signal can be isolated from that descents in the regime where U 12@1 and ~ii! expanding the
scattered by the lamellae. If we treat the inclusions as point- exponential when U 12!1. We proceed by crudely cutting off
like and neglect the details of the short range interactions we the integral at z5 n d, as before, and find
may write S(q) in terms of the two-body potential f (r) by
2 23/2
invoking Boltzmann statistics V .2ld E 0
B far exp@ E 0 / ~ 4 p e 2 n 2 !# for E 0 @1.
~19!
S ~ q! 5C a2 r̄ E 3
d re iq•r2 f ~ r!
. ~15! This represents a large ~exponential! average attraction be-
tween particles.
By assuming that the potential f is small, which is often a We emphasize again that these results neglect any addi-
reasonable approximation for uncharged lyotropic systems, tional short range interactions not described by the linear
we may expand the exponential in Eq. ~15! to obtain S(q) coupling term b r] z u in Eq. ~1!.
for small but finite q.
CONCLUSION
S ~ q! 52C a2 r̄ f q , ~16!
We have studied the effect of particulate inclusions in a
where f q is given by Eq. ~9!. This prediction allows for a bulk smectic-A phase and have derived equations for the
test of the present theory on two grounds. Firstly a structure energy of an arbitrary distribution of particles. These results
factor of the form S(q);q 2z /(q 2z 1l 8 2 q 4i ) would provide are used to develop a self-consistent mean field theory for the
strong evidence that the particle-layer coupling in a given particle distribution, which is used to study the effect of fi-
R1278 M. S. TURNER AND P. SENS 55

nite particle density on the effective two-body interaction suggest that direct measurements of this may provide a direct
potential. The correction to this potential is computed and we test of our theory. We also briefly discuss the second virial
find that although the effective compressional modulus of the coefficient for the mixed particle-lamellar system, noting that
phase is unchanged, the effective bending modulus increases the far-field interactions are always attractive on average.
linearly with the particle density according to Eq. ~13!. This
result is in contrast to one, rather different, study of mem-
brane impurities, which predicted a decrease in the bending ACKNOWLEDGMENTS
modulus @10#, but is in qualitative agreement with the trend
observed for membranes densely decorated with PEG lipids The authors wish to thank P. Pincus for his encourage-
@17#. In addition both of these studies show a change in B ~an ment and advice, including a critical reading of the manu-
increase in the first and a decrease at large densities in the script. Stimulating discussions are acknowledged with G. H.
second!. The qualitative differences with the present theory Fredrickson, D. Morse, and C. M. Marques. The authors ac-
are unsurprising in view of the differences between the sys- knowledge support from the Royal Society ~UK! and the
tems. We give the static structure factor for the particles and MRL program of the NSF under DMR-9123048.

@1# C. Kittel, Introduction to Solid State Physics ~Wiley, New @11# P. G. de Gennes, J. Phys. ~France! Colloq. 30, C4-65 ~1969!.
York, 1976!. @12# S. A. Safran, Statistical Thermodynamics of Surfaces, Inter-
@2# P. G. de Gennes and J. Prost, The Physics of Liquid Crystals, faces, and Membranes, Frontiers in Physics ~Addison-Wesley,
2nd ed. ~Oxford University Press, Oxford, 1993!. New York, 1994!.
@3# See, for example, Biological Membranes, edited by D. Chap- @13# The singular nature of point particles means that our treatment
man ~Academic Press, London, 1968!. fails for separations z&d. This failure can be identified with
@4# J. Darnell, H. Lodish, and D. Baltimore, Molecular Cell Biol- the appearance of higher order terms in the expansion ~1! on
ogy ~Scientific American, New York, 1990!. these length scales, where the ratio u ] zz u u / u ] z u u becomes big-
@5# See, e.g., R. Bruinsma, M. Goulian, and P. Pincus, Biophys. J. ger than an inverse microscopic length .1/d. In fact we may
67, 756 ~1994!; N. Dan, P. Pincus, and S. A. Safran, Langmuir argue that the continuum smectic theory is also breaking down
9, 2768 ~1993! and @7# and references therein. at these length scales. This feature is discussed in more detail
@6# Y. Shen, C. R. Safinya, K. S. Liang, A. F. Ruppert, and K. J. elsewhere ~Ref. @7#! and is planned for forthcoming work.
Rothschild, Nature ~London! 366, 48 ~1993!. @14# We neglect the term ; d (q), which corresponds to fixing the
@7# P. Sens, M. S. Turner, and P. Pincus ~unpublished!. zero of energy at infinity.
@8# F. Nallet, D. Roux, C. Quilliet, P. Fabre, and S. Milner, J. @15# P. M. Chaikin and T. C. Lubensky, Principles of Condensed
Phys. ~France! I 4, 1477 ~1994!. Matter Physics ~Cambridge University Press, Cambridge,
@9# C. Nicot, M. Waks, R. Ober, T. Gulik-Krzywicki, and W. Ur- 1995!.
bach, Phys. Rev. Lett. 77, 3485 ~1996!. @16# L. D. Landau and E. M. Lifshitz, Statistical Physics ~Pergamon
@10# S. Leibler and D. Andelman, J. Phys. ~France! 48, 2013 Press, Oxford, 1981!.
~1987!. @17# P. Pincus and C. R. Safinya ~unpublished!.