HAL Id: jpa-00208515
https://hal.archives-ouvertes.fr/jpa-00208515
Submitted on 1 Jan 1976
HAL
is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire
HAL, estdestinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
A molecular theory of surface tension in nematic liquid crystals
J.D. Parsons
To cite this version:
J.D. Parsons. A molecular theory of surface tension in nematic liquid crystals. Journal de Physique,
1976, 37 (10), pp.1187-1195. �10.1051/jphys:0197600370100118700�. �jpa-00208515�
A MOLECULAR THEORY OF SURFACE TENSION
IN NEMATIC LIQUID CRYSTALS
J. D. PARSONS
Departamento
de Fisica Universidade Federal de Santa CatarinaFlorianópolis, Trindade,
Brasil 88000(Reçu
le2 fevrier 1976, accepte
le 10 mai1976)
Résumé. - On calcule la tension superficielle à la surface libre d’un cristal liquide nématique en
utilisant
l’approximation
de Fowler, Kirkwood et Buff pour l’interface et une approximation de champ moyen pour la fonction de distribution de deux molécules. En prenant une interaction de Van der Waals entre molécules, on trouve que : a) les molécules préfèrent toujours être orientées dans le plan de la surface; b) il y a un saut de la tension de surface à la transitionnématique-isotrope,
mais elle est inférieure dans la phase nématique ; c) on peut observer un excès d’ordre en surface qui persiste dans la masse de la phase isotrope. On calcule les expressions générales du paramètre d’ordre
en surface à
partir
d’arguments d’énergie libre. On discute l’effet d’interactions entre momentsdipolaires permanents et l’on trouve que des molécules uniaxiales doivent être parallèles à la surface.
Abstract. - The surface tension at the free surface of a nematic liquid crystal is computed using
the Fowler-Kirkwood-Buff
approximation
to the interface, and a mean field approximation to themolecular pair distribution function. It is found for a Van der Waals interaction between molecules that : a) the molecules always
prefer
an orientation in theplane
of the surface; b) there is a gap at thenematic-isotropic phase transition, but the surface tension is less in the nematic
phase;
c) there may be observable excess surface order which persists into the bulkisotropic
phase. General expressionsare given for the surface order parameter using free energy arguments. The effect of a permanent dipole interaction on the surface tension is considered, and it is found for uniaxial molecules, that this interaction also leads to the condition that the molecules be parallel to the surface.
Classification
Physics Abstracts
7.130
1. Introduction. - The surface
region
and surface tension in nematicliquid crystals
is at present ratherpoorly
understood.Experiments
so farperformed
show several
interesting
effects :a)
Orientation at the free surface : The molecules ofsome nematics
(e.g. PAA)
tend to lie in theplane
of thenematic-air free surface
[1-3]. However,
the orien- tation of MBBA has been measured to lie at a finiteangle
which isslightly temperature dependent [3].
Several nematics which have a low temperature smectic A
phase
seem to have aperpendicular
orien-tation in the nematic
phase just
above the transition[4].
b)
Behaviour of surface tension at the nematic-isotropic
transition temperature : Theearly
measu-rements
[5, 6]
show that the surface tension is discon- tinuous at the transition. Recent work[7, 8]
hasshown that in certain cases the
discontinuity
is nega-tive,
and the surface tension is less in the nematicphase.
c)
Free surface disclinations : In several nematicshaving
a low temperature smectic Cphase,
discli-nations in the orientation appear
just
before the nematic-smectic C transition[4].
Molecular theories of surface tension in nematics
are difficult because of the
anisotropic pair
interac-tions between molecules.
Recently,
ageneral
expres- sion for the surface tension of apolyatomic
fluidhas been derived and
applied
to calculate thedipole- dipole
contribution to the surface tension of low vapor pressure water[9].
For molecules withpair
interactionsdepending only
on the center of mass vector R andon the Euler
angles [10] (Q, 0, 03C8) describing rigid body
rotations between the molecules relative to their centers of mass, the result for the surface tension is :where R = r2 - ri is the center of mass vector, and where the
pair potential U12 depends
on R and theinternal states
e1, 62
which denote the Eulerangles :
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0197600370100118700
1188
d3ei
=sin oi doi dcpi dgii. p(2)
is thepair,
or doubletdistribution function defined so that
gives
the average number ofpairs
ofmolecules,
oneof internal state
61
and withind3r,
of rl, the other of internal statee2
and withind3r2
of r2’ We define the distribution functiong (2) by
the relationwhere p is the
singlet
distributionfunction,
or numberdensity
of the fluid. In(1),
the z axis is chosen to be normal to the freesurface,
which we shallconveniently
fix at z=0.
From
(1)
and(2)
we see that at thenematic-isotropic transition,
y willchange abruptly
because of the two factors in thepair
distribution function :a)
Thenematic-isotropic
transition isalways
firstorder with a discontinuous
change
in thedensity
p.Since the
density
islarger
in the nematicphase,
thiswill lead to a gap in y with y also
larger
in the nematicphase. However,
thedensity discontinuity
at thetransition is very
small,
0.3% being
atypical
value.Since y oc
p2,
thisimplies
that the gapAny
in the surface tension at the transition will be no more than about 0.6%
due to thedensity discontinuity.
b)
There is also asharp discontinuity
in the distri- bution functiong (2)
at thetransition,
duemainly
to thediscontinuity
in the orientational order parameter.Using
a Van der Waals interaction forU12,
and amean field
approximation,
we will show that the gapin y
due to thisdiscontinuity
should belarger
thanthat due to the
density discontinuity.
The surface tension will also
depend
on the orien-tation of the molecules at the free
surface,
since the interactionpotential
isanisotropic,
anddepends
onthe direction of the center of mass vector R as well
as its
magnitude.
In the bulkphase
we know that the nematic state is characterizedby long
range orienta-tional order
along
a certain directiongiven by
a unitvector n. In the absence of boundaries and external volume torques, the direction of n is
completely arbitrary :
the free energy of the nematic does notdepend
on n butonly
on thedegree
of order Salong
n.However,
the free surface breaks the translational symmetry, so that the surface free energy willdepend
on
n . k,
where k is the normal to the surface. It isimportant
to realise that this isessentially
a geome- tricaleffect,
and not due to anyspecial
interaction between molecules at the surface. Therefore weexpect
an
anisotropic
surface energy even in the case of anisotropic
surface(e.g. glass,
anisotropic liquid, etc.) and,
inparticular,
even in the case of anematic-vapor
surface. Since the bulk free energy is still
independent
of n, the total free energy
(bulk plus surface)
willbe minimized for the value of
n.k
which minimizes y.Below we show that for a Van der Waals interaction between
anisotropic molecules,
y is minimized whenn . k
= 0 i.e. when the molecules lie in theplane
ofthe free surface.
2. Interaction
potential.
- In this section we consi- der the form ofU12 in (1),
when the molecules interactthrough
the induceddipole-dipole
term of the Cou-lomb interaction
(Van
der Waalsforces).
This inte- raction was usedby
Maier andSaupe [11]
in theirtreatment of the
nematic-isotropic phase
transition.In this section we assume that the molecules have no
permanent
dipole
moments. This interaction will be discussed in section 6.The
starting point
is the interaction energy between two molecules asgiven
in the second order pertur- bationtheory :
where 1 0(1) >, 1 0(2) >
are theground
states of mole-cules 1 and
2, and ,u(1) >, v(2) >
are excited states ofenergy
E,
andEv. Euv = Ell + Ev
and we assume allmolecules are in their electronic
ground
states.Finally,
H’ isgiven by
adipole-dipole
interaction :ql, qk are
charges
atpositions
xi, xk in themolecule,
and R is the center of massseparation
between molecules.We assume that the molecules are uniaxial and
rodlike,
and that thedipole
moments are inducedonly along
thelong
axes of eachmolecule,
all off-axisdipole
momentscancelling
out. This latterassumption
is unnecessary ifwe are
only
interested in the orientational part of theinteraction,
but it effects thedependence
on the center ofmass vector R. Write
xi(1)
=ni(1) il, XK(2)
=nK(2) e2,
whereê1
ande2
are unit vectorsalong
thelong
axes of the twomolecules.
Then, putting (4)
into(3)
we have :Since
Eoo E/-Lv,
we have a sum ofnegative
terms in(5).
Hence the interaction can be written as :where T is the
symmetric
tensorn
with 1 a unit tensor.
We now average the interaction
(6)
over a distribution of molecules. Introduce aspherical
coordinatesystem with
polar
axis n and express61
ande2
in this coordinate system. Since the molecules areuniaxial,
all azimuthalangles
Q 1, (P2 areequally probable.
The average W over theseangles
then works out to bewhere the
Tij
are the cartesiancomponents
of T.Next,
we average(8)
over thepolar angles 0,, 02-
From the symmetry of the nematic state, the distribution functionf(0)
is related to the order parameter Sby
In
doing
the average in(8)
we assume thatf(o1,’ o2)
=f (o1) f (o2)
withf(O) given by (9).
This is a mean fieldapproximation
where local correlations in orientation areignored.
With thisapproximation,
the interactionaveraged
over all orientations of the twomolecules,
becomes :Finally,
we must average(10)
over all orientations of the center of mass vector R. We express R in the samespherical
coordinate system as before(with
n thepolar vector).
Let 0 and 0 be the azimuthal andpolar angles
of R. Since the nematic state is
cylindrically symmetric,
all values of 0 areequally probable,
and an average of(10)
over this
angle gives,
with thehelp
of(7) :
In their treatment of the bulk
phase,
Maier andSaupe [11]
assume that allangles
0 areequally
pro-bable,
i.e. the centers of mass of the molecules arecompletely random,
even at short range. Thisapproxi-
mation is in the same
spirit
as the one that invokes theseparability
off(01, 02);
i.e.local,
short range correlations in bothposition
and orientation of molecules areignored. However,
note that the inte- ractionpotential (6)
is a strong function of the direc- tionof R,
therefore this isonly
arough approximation.
With this
assumption cos2 0 >
=1/3,
so that thelinear term in S
drops
out, and(11)
reduces to :which is the well known result that the mean field is
proportional
toS2.
In
(11)
we seeif cos’ 0 )
>1 /3 (molecules
tendto be lined up end to end with
R // n)
then the termin S is
positive. If cos’ 0 > 1 /3 (molecules
tendto
align
inlayers
with n 1R)
then this term isnegative.
On the other
hand,
the term inS2
isalways positive
and rather insensitive to the direction of R. The interaction
(11)
has an extremum which is a localminimum when 0 =
n/2,
which favors thetendency
to order in
layers.
If we havecomplete
translational symmetry, the linear term in(11) dissapears,
butif this symmetry is broken in some way, this term will determine the translational
symmetry
of the struc-ture.
One way to break the translational symmetry is
through
anequilibrium phase
transition. The inte- raction(11)
suggests that apossible
lowtemperature phase
of a nematic would be characterizedby
astructure where the molecules tend to form
layers
with nematic
ordering
in eachlayer
and with theplanes
of thelayers perpendicular
to n. This isexactly
1190
the symmetry of the low temperature smectic A
phase
observed in many materials that form nematics
[12].
Of course, one cannot say whether a
given
nematicwill form such a smectic state
just by looking
at theinteraction
(11). Nevertheless,
thetendency
toalign
in
layers,
characteristic of the smecticphase,
isalready
present in the nematicphase.
We feel thatprevious
molecular theories of the smectic A
phase [13, 14]
have not taken
enough
account of theanisotropy
ofthe interaction with respect to the center of mass
vector
R, corresponding
to the linear term in S in(11).
Translational symmetry is also broken at the free surface in the nematic
phase.
Even if all orientations of R areequally probable
in thebulk,
this will cer-tainly
not be true close to the free surface. We willsee below that the linear term in S in
(11)
is crucialfor
determining
both the surfacetension,
and the orientation of molecules at the free surface. So the orientation at a freesurface,
and thetendency
of thefluid to form a
layered,
smectic structure areclosely
related. A
study
of the free surfacegives
informationon molecular interactions which are
averaged
out inthe
translationally symmetric,
bulk nematicphase.
This is also the case in some types of
polar
and asso-ciated
liquids
which do not have a bulkliquid crystal phase [15] :
the fluid isisotropic
far from thesurface,
but there is ananisotropy
closeenough
to thesurface, corresponding
to apreferential ordering
of the mole-cules.
3. The FKB
approximation
for surface tension. - Thegeneral
formula(1)
for the surface tension is very difficult to use without furtherapproximation
since thedensity profile
andpair
correlation function areunknown in the surface
region
where the fluid isnonhomogeneous. Indeed,
in an exact treatment, thedensity profile
cannot bespecified beforehand,
but must be determined as part of the solution[16].
At lov vapor pressure, the Fowler-Kirkwood- Buff
(FKB) approximation
leads to afairly
accurateestimate of y for
simple
fluids[17].
We will use thesame
approximation
intreating
thenematic-vapor
free surface. The FKB
assumptions
are that :a)
thevapor
density
isnegligible,
andb)
thedensity
of theliquid phase
has the bulk value up to the(Gibbs’) dividing surface,
which we take at z = 0. These twoassumptions imply :
where Z = zl - Z2’ In addition we assume that
where
1(8)
isgiven by (9).
Thisapproximation
involvesthe
following assumptions : a)
the centers of massvectors are
isotropic
in the bulkphase, b)
localcorrelations between the orientations of the molecules
are
neglected (a
mean fieldapproximation),
andc)
thez,
dependence
of the distribution functionf(0)
isignored.
Thatis,
we assume that the orderparameter S
has the bulk value up to the free surface. The first two
assumptions
areobviously
in the samespirit
as theMaier-Saupe theory
for the bulkphase,
and the lastassumption
is in the samespirit
as the FKBapproxi-
mation
(13), although
it is much lessobvious,
and infact,
as we shall see, contradicts theboundary
condition on S at the free surface which can be derived from free energy arguments. As with the
density profile,
the detailed form ofS(z) through
the surfaceregion
must be obtainedself-consistently
in thesolution,
and cannot bespecified
in advance.Recently,
the
S(z) profile
has been discussed[18],
but since itdepends
on the unknowndensity profile p(z),
thediscussion is
only qualitative.
Theonly
other casethat can be treated within the FKB
approximation
is one where
S(z) decays
to the bulk value S in acharacteristic distance
large compared
to the mole-cular interaction range. This case is discussed in an
appendix
where it is shown that theparameter S appearing
in y must be taken to be the value of S at thesurface,
and not the bulk value of S.With these
approximations (1)
reduces to :with
d2e
= sin 0 dOd(p,
thesuperfluous angle 0 having
beenintegrated
out. The interactionU12
is taken to be
(6)
for Van der Waals forces. Put(6)
into
(16)
andintegrate
overd 2e, d2e2 first, keeping
Rconstant. We get
where ( W >
isgiven by (10).
The surface tension canbe broken up into three terms
in W > :
First assume that n lies
along
the surface normal(z axis).
We can then express R in terms ofspherical
coordinates with
k
= n as thepolar
vector, andintegrate
over the twoangles e,
0. The first term yo in(18) easily
works out to be :The second and third terms in
(18)
can beexpressed
in terms of the
angular integrals
After a rather tedious
calculation,
one finds thatFrom
(10), (17),
and(21)
we get :hence the surface tension for
perpendicular
orien-tation is
with Yo
given by (19).
4. Surface tension for
parallel alignment.
- Nextconsider the surface tension when the molecules lie in the
plane
of thesurface,
i.e. when n isalong
the xaxis. Since R is
expressed
in terms of0, 0
in a coordi-nate system with n the
polar axis,
we can calculate yby simply interchanging x
and z in the coefficients(22) (i.e. Ixx
->Izz, Ixy
->Izy, etc.). Then, using
thegeneral
results
(17)
and(10),
one finds :We see
that y
II V’ 1. for all values of S. Further- more, the symmetry of the interaction(10)
rules out asmaller value
of y
for any other orientation. If there are no external torquesapplied,
and the influence of other boundaries isignored,
the total free energy(volume plus surface)
will be minimized when the orientation isparallel
to the free surface. We concludethat,
for moleculesinteracting
with Van der Waalsforces,
the orientation will tend to beparallel
to the free surface.The
result, (25)
alsopredicts
a gap in the surface tension vs. temperature curve at thenematic-isotropic phase
transitionT,,
due to the discontinuouschange
in S there :
where AS =
S(T Tk) - S(T > Tk).
If we assumethat the S in
(25)
is the bulkvalue,
then we know thatS(T > Tk)
= 0 andS(T Tk) N 1/3.
Thisgives
a gap of about10 %
in the surface tension.Moreover,
thegap is
negative
- the surface tension is less in the nematicphase.
We can relate this to molecular interactions as follows : the surface tension is essen-tially proportional
to thepair interaction ( W >
averaged
over all orientations. From(11)
one can seethat W )
will be less in the nematicphase
when thelinear term in S becomes
negative,
i.e. when the dis- tribution of the centers of mass vectors favors alayered
like structure close to the free surface. Thedecrease
in ( W >
with Simplies
there is lessrestoring
force on s,urface molecules in the nematic
phase,
than in the
isotropic phase.
Thisimplies, ignoring
thesmall
density effect,
that y will be less in the nematicphase.
Note that this canonly happen
when theorientation is
parallel
to the freesurface;
for a perpen- dicular orientation the surface tension gap would bepositive,
as can be seen from(24).
We can make some further remarks on the
y(T)
curve away from
Tk.
Since y ocpe
we expect y toslowly
increase withdecreasing
temperature due to thermalexpansion.
In fact this is the main temperaturedependence of y in
normalorganic liquids [19].
For atypical
value of10-3/deg
for the volumeexpansion
coefficient we expect a value
Ayfy = 0.2 %
perdegree.
Now for T
Tk,
we know thatS
increases withdecreasing temperature,
with the rate of increasebeing largest
nearTk.
Since forparallel orientation,
anincrease in S leads to a decrease in y, it is
possible
that close to
Tk
one may see ydecreasing
with decreas-ing temperature.
Far belowTk,
where the order parameter is almost constant, the thermalexpansion
effect will
probably
win out, and y should start to increaseagain.
The situation is furthercomplicated by
the fact that nearTk
the thermalexpansion
coef-ficient shows
pretransitional
anomalies. A careful measurement ofy(T)
aboveTk
couldgive
informationon the temperature
dependence
andmagnitude
ofthe order parameter at the surface
(see
section5).
The
dependence
of y on anapplied magnetic
fieldhas received some attention in the literature
[20].
If the field is
applied parallel
to the free surface there should be no bulkdistortions,
and one can see from(25)
that y should decrease with H because of the increase in S with H. However this effect is very small in nematics because of the small value of theanisotropy
of themagnetic susceptibility
and it is notexpected
to be observable.We have calculated y for
parallel
and forperpendi-
cular orientation and shown that y || yl. For certain distortion
problems
near the free surface it is of interest to calculatey(03C8 ) ; where 0
is theangle
that nmakes with the surface normal
k :
:cos 0
=n.k.
From
(10)
and(17)
we can see thatonly
even powers ofcos #
can enter y, but theresulting expression
is noteasy to carry out because the interaction
(10)
involves1192
the squares of the tensor components
Tij
rather thanthese
quantities
themselves. Forqualitative
discus-sions it may be sufficient to use the
simple expression :
From
(24)
and(25)
it follows thatIn the nematic
phase
where S >1/3,
it isimportant
to
recognize
that this is a verylarge
difference in surfaceenergies;
at most an order ofmagnitude
lessthan the actual interfacial surface tension of the
liquid.
This means that it will be very difficult to tilt the surface orientation
by
theapplication
of bulk torques. Forexample,
if oneapplies
amagnetic
fieldperpendicular
to the
surface,
the orientation at the surface will begiven by
anangle t/Jo (measured
withrespect
to thenormal)
of[21]
where 3 DS 2 ~
k,
where k is a Frank orientation elastic constant[22].
Now with k ~ 10-6dynes,
xa N 10-’ c.g.s., H ~ 104
G,
and yl -Y II N
1dyne/
cm we find that cos
t/Jo
3 x10-3,
which ishardly
observable. So our
theory
suggests that thegeneral
solutions of the elastic
equations
for the distortions in surface orientationproduced by
volume torques[21, 23]
will be difficult to observe in the nematicphase.
The surface orientation is thensubject
tostrong
anchoring :
inproblems involving
orientation at the freesurface,
instead ofminimizing
the bulkplus
surface freeenergies,
one maysimply
minimizethe bulk free energy and
require
that theboundary
condition
n-k =
0 be satisfied at the free surface.Instead of
using
bulk torques, one may try tochange
the orientation at the free surface
by using
solid bound- aries.Suppose
we introduce a thin nematic film on top of a solid thatrigidly
maintains aperpendicular alignment
of molecules[24].
For a thinenough layer,
one may be able to tilt the
alignment
at the free surface.This
problem
was considered in ref.[21]
where itwas shown that the
alignment
at the surface satisfies :where d is the thickness of the film and
using
the same estimates as before. For d >dc
wehave
approximately parallel alignment
at the surface(tf 0 ~ x/2),
but for ddc,
thealignment
at the freesurface is forced to be
perpendicular
tosatisfy
therigid boundary
condition at the solid surface. It would beinteresting
to try to observe thiseffect,
as the gapin y at the transition temperature is very sensitive to the orientation
tf¡ 0’
and even reversessign (see
eq.
(24))
for00 -+
0. The surface tension itself willdepend
on thelayer thickness d, varying from 01,
Iwhen d >>
dc
to yl when ddc.
Finally,
it should bepointed
out that the surfaceenergy favors
parallel orientation,
but the direction of n within the surface isarbitrary.
Thisdegeneracy
is inherent in the
problem,
since what we have chosento call the X axis is
arbitrary.
Thedegeneracy
will beremoved,
inpractice, by
smallperturbing
forces inthe
bulk,
whoseorigin
may come from other boun- daries in theproblem,
or from an externalmagnetic
field
applied
in agiven
directionparallel
to the surface.5. Excess surface order. - The fact that
Any
0 for thepreferred parallel
orientation alsoimplies
that there will be excess surface
order,
andthat S #
0at the surface even above
Tk.
It has been shown elsewhere[21]
that theboundary
condition that S mustsatisfy
at aplane
interface isgiven by :
where
5(S)
is the volume free energy. To get arough
estimate of the
magnitude
of thiseffect,
we may usea Landau
expansion
forf(S) [25] :
where A =
a(T - Tc*)
withT * slightly
below the(first order) phase
transition temperatureTk,
anda,
B, C,
D arepositive
constants. Since D >0,
note that excess surface order canonly
occur whenwhich is the case for
parallel
orientation. Theequi-
librium
S(z)
mustsatisfy
the EulerLagrange
equa- tion :which for
(32) gives
a nonlinearequation
forS(z) :
with the
boundary
conditions S= Sb, dS/dz
= 0at z = - oo.
Sb
is the bulk value(the
value the order parameter would have if the fluid wereunbounded).
It satisfies :
For T
Tk
the solution which minimizes(32)
iswith
For T >
Tk
the solution that minimizes(32)
isSd
= 0.At T =
Tk
there is adiscontinuity
inSb
and thetransition is first order.
An exact solution of
(34)
for all cases would involveelliptic
functions and would be very difficult to carry out. Here weonly
considerapproximate
solutionsfor T >
Tk
and TTk.
First suppose TTk
sowe are in the bulk nematic
phase.
Inparticular,
choose T =
Tc *,
thenA = 0, Sb
=B jC.
Assumethat the effect of the surface is small and write S
= Sb
+S’(z)
and linearize(34)
inS’(z).
Thisgives
asimple exponential equation
forS’(z) :
where
a2
=(3 CSb -
2CSb)/D
=B 2/CD
> 0. The solution for S is thenwhere
So
=S(O)
is the order at the surface. The interfacialboundary
condition(31) along
with(25)
then
gives
For
yo/Da ->
oo,So approaches
thelimiting
valueof
3/4.
Inevaluating (39)
we use thetypical
valuesC N 1
J/cm3,
D ~lO-12 J/cm,
yo N 30ergs/cm’,
Sb
= 0.5. Then(39) gives So N
0.68 which is about 30%
excess surface order. Thedecay
distance isa -1 N
100A
which issmall,
but still muchlarger
than the range of interaction of Van der Waals forces
(only
a fewA).
The case T >
Tk
poses more difficulties sinceSb
= 0. A firstintegral
of(34)
isreadily
obtained :Note that the r.h.s. is
always positive
for A > 0 since this isjust
the free energy of the bulk nematicphase
which has an absolute minimum at S = 0 when T >
Tk.
Thus(40) represents
a nonlineardecay
fromthe surface : oscillations in
S(z)
cannot occur.Using (40)
and theboundary
condition(31)
we can get a condition onSo
withoutknowing
thecomplicated dependence
on z :Again
when yo -> oo,So -> 3/4. Eq. (41)
can besolved
numerically
forSo
at alltemperatures
if thequantities A, B, C, D,
yo are known. For the valuesA/C
=1/10, B/C
=1/2,
and yo and D as estimatedbefore, (41)
has theapproximate
solution ofSo
= 0.5just
aboveTk.
This ratherlarge
value decreasesroughly inversely
to the temperature difference T -Tc*.
A more detailed
qualitative
discussion of the excessorder
parameter, taking
into account the fact that the surfaceregion
isreally
diffuse rather thansharp,
has been
given
in ref.[18].
6. Effect of a
permanent dipole-dipole
interaction. - Interactions between permanent moleculardipoles
are
ignored
in thetheory
of bulk nematics because noferroelectric effects have ever been measured in
nematics;
the states n and - n seem to becompletely equivalent.
Thisimplies
that the molecular distri- bution functionf(O)
has the symmetry propertyf(n - 0)
=f(O),
so that even if the molecules have permanentdipole
moments, as manypoint
up aspoint
down on the average and anylong
range ferro- electric effect cancels out.However,
since a free surface breaks the translationalsymmetry,
it ispossible
tohave cos
0 > :0
0 at thesurface,
where there is an.anisotropy
inthe
distribution of center of mass vectors.Since
cos0 >
= 0 in thebulk,
it mustdecay
to zero in a characteristic
distance
from the free surface.In this section we calculate the contribution to the surface tension
produced by
permanentdipole- dipole
interactions. We assume that the permanentdipoles
arealong
thelong
axes of the molecules. Then the interaction isgiven by :
where T is
given by (7), and p
is thedipole
moment ofthe molecule.
Again
we express the6i
inspherical
coordinates with
polar
vector n and average(42)
over the two azimuthal
angles
wi, Q2. Theonly surviving
term is :An average
over 01 and 02 gives
afactor cos () >2 == G2
in a mean field
approximation. Expressing TZZ
inpolar
coordinates asbefore,
the average interaction turns out to be :As
expected,
for a random distribution of centers of mass,cos’ 0 )
=1/3,
and the interaction vanishes in the bulkphase.
But near thesurface,
the interaction
(44)
will not average out.1194
The contribution of the
permanent dipole
interac-tion to the surface tension is then
where U > = R -3(1 - 3 cos’ 0). Doing
the angu- larintegrals
as before we find forperpendicular alignment
and for the
parallel alignment
So we have yd 11 yal and hence the permanent
dipole forces,
like the Van der Waals’forces,
lead to aparallel alignment
of molecules at the surface. This appears to rule out the ferroelectric interaction between permanentdipoles
in theinterpretation
ofthe
perpendicular
orientation of MBBA in a thin film with two free surfaces[2].
We have assumed the
dipole
moment of a moleculeis
along
thelong
axis. If we havepermanent
off axisdipole
moments which make anangle
a to the directionof easy
polarizability
then the molecules are nolonger
uniaxial but biaxial[26].
In this case, if thedipole-dipole
interaction is theonly interaction,
the above arguments gothrough exactly
as before andthe surface tension is minimized when the
dipole
moments are
parallel
to the surface which means thatthe director makes an
angle
a to the surface. Whenthere are Van der Waals forces present, these will tend
to line up the director
parallel
to thesurface,
so that the actualangle tf
that n makes with the surface will be less than a. Theangle tf
is thus determinedby
twocompeting
forces and will ingeneral
betemperature dependent
since the two order parameters S and e may be different functions of temperature. In MBBA the molecules are tilted at anangle r
750 from thesurface which
depends
upon temperature[3].
The caseof MBBA would be an
interesting
one tostudy
indetail since we have
experimental
information on y at the transition[27],
the tiltangle,
and the electricdipole
moment[28].
Other mechanisms that can lead to a tilt in the molecular orientation at the surface are :
a)
Interactions of surface molecules withimpu-
rities
selectively
adsorbed to the surface which tend toorient the
polar
heads of the molecules andthereby
cause a term in y which favors
perpendicular align-
ment. In this case, the tilt
angle
would vary with the concentration of the surfaceimpurities.
b)
The formation of smectic-likelayers
near thesurface, resulting
in a breakdown of the translational symmetry at finite distances from the surface. If thelayers
tend to lie in theplane
of the freesurface,
asthey
do in
smectics,
this will lead to aperpendicular
orien-tation at the surface. This kind of
ordering,
whereone has a non-zero smectic order parameter close to the surface which
decays
to zero in thebulk,
is ana-logous
to the case of orientational surface order above the bulkclearing
temperature where the bulk order parameter vanishes. A betterunderstanding
of therelationship
between order and molecular interactions in smectics willprobably
be necessary, before such amodel could be worked out in detail.
Appendix. Spatial dependence
of S. - If there isexcess surface
order,
the orderparameter
willdecay
to the bulk value within a characteristic distance from the surface. In other
words,
the order para- meter, and therefore thepair
distributionfunction, depend
on the coordinate z, in(1).
We assume the FKB
approximation
and put(13), (14), (15)
into(1), where f (0)
is allowed todepend
onzl. The
integral
overd3 R
can be broken up into apart with Z > 0 and a part with Z 0 :
Since
p(2)
= 0 for zi >0,
this reduces to :where
Interchange
the orders ofintegration
in(A. 2),
inte-grating
overd3el d3e2
first :where A and B are