HAL Id: jpa-00247165
https://hal.archives-ouvertes.fr/jpa-00247165
Submitted on 1 Jan 1995
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, est destinée 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.
On Dipole Association and Short-Range Order in Nematic Discotic Liquid Crystals
T. Phillips
To cite this version:
T. Phillips. On Dipole Association and Short-Range Order in Nematic Discotic Liquid Crystals. Jour-
nal de Physique I, EDP Sciences, 1995, 5 (12), pp.1667-1680. �10.1051/jp1:1995222�. �jpa-00247165�
Classification
Physics
Abstracts61.30Cz 61.30Gd 64.70Md
On Dipole Association and Short-Range Order in Nematic Discotic Liquid Crystals
T.J.
Phillips
Defence Research
Agency,
St. AndrewsRoad,
Malvern, Worcestershire, WR143PS,
UK(Received
30 March 1995, revised andaccepted
8August 1995)
Abstract. The Kirkwood
dipole
correlation factor gi of a nematic discoticliquid crystal
is measured as a function of temperature. A
large degree
ofanti-parallel dipole
associationis
found,
which persists into the isotropicphase.
This is in conflict with previous mean-field mortels fordipole
correlation, whichpredict parallel
association mdiscogens.
A cluster mortel ofordering, including short-range dipolar
orientational order,quantified by
a parameterT,
andinduding
theexpenmeutal
fact of anti-parallel dipole association,
is used to attempt a morereasonable treatment of
dipole
correlation m these systems. The theory is found topredict
thedipole
association rather well, with a greatimprovement
over mean-field results, and shows other improvements over mean-field mortels,including
a reduction in transitionenthalpy
anddrop
in the meanpermittivity,
I, when the system is cooled into the nematicphase.
1. Introduction
The first
theory
of the formation of the nematicphase,
basedpurely
ondipolar interactions,
was
proposed by
Born in 1916[ii.
Thistheory
was discreditedby
measurements that showed no surface freecharges [2],
and thediscovery
ofquinquephenyl [3],
which has no permanentdipole
moment but still shows a nematic
phase. Nevertheless,
the vastmajority
of mesogens containpolar
groups, andyet
thedipolar
contribution to the formationof,
andordering
m,liquid crystalline phases
isusually
assumed to be small[e.g. 4,5].
The area m which permanentdipoles
are most
routinely
considered is for the dielectricproperties
ofmesogens-dipolar
interactions then have asignificant
effect [6]. Various models ofdipole
association inanisotropic
media have beenproposed [7-9], mcluding
a mean-fieldtheory
for nematicsby
Dunmur andPalffy- Muhoray
[10]Recently [11],
some nematicdiscogens
have been studied. These were found to have alarge
permanent moleculardipole
momentparallel
to the short(1.e. unique)
axis, but to haveonly
a smallpositive permittivity anisotropy,
due toanti-parallel dipole
association. This is a counter-example
to the mean-fieldtheory [10],
whichpredicts parallel
association for suchmaterials,
due to theassumption
that thedipoles
are ~buried' within the molecule[11]. Although
theuniaxial
long-range quadrupolar
orientational orderparameter
was found [12] to have a similar©
Les Editions de Physique 19951668 JOURNAL DE PHYSIQUE I N°12
~M
~
_~) il °CioH~i
DB126
~
~~~7~5
HET7
Fig. 1. Molecular structures of the materials.
value to calamitic
(rod- like)
materials(and
theMaier-Saupe
mean-fieldprediction
[13]),
short- range columnar order was also found to be present. Thedipole
association and columnar order both influenced thephysical properties
and were found to belinked,
with the columnarordering favouring
correlation betweendipoles [11].
In the past,
discrepancies
have been found between theequations
for thepermittivities
ofa nematic derived
by
àllaier and Meier [14](based
on the mean-fieldtheory
of order[13])
andexperimental
results. Anexample
is thedrop
m the meanpermittivity,
z, seen as somematerials are cooled into their nematic
phase [e.g. 15].
This has beenexplained by
Madhusu- dana and Chandrasekhar [16] asbeing
due to achange
m theanti-parallel dipole association, using
atheory
where the usual mean-field isperturbed by
ashort-range
interaction betweena small nttmber of
partiales. However,
forsimple nematics,
such as thecyano-biphenyls,
thediscrepancies
haveusually
beensmall,
and thetheory largely ignored.
Nematics that havea
strongly negative permittivity anisotropy might
well beexpected
to be acounter-example
to the
theory
of Dunmur andPalffy- Muhoray,
but here molecularbiaxiality comphcates
the situation[17, 18].
The nematic
discogens
studied here have the benefit ofbeing
an obviouscounter-example, having
a correlationopposite
msigq
to thatpredicted,
whileretaining
thesimplifying
feature of little or no molecularbiaxiality.
In this paper, the Kirkwooddipolar
correlation factor gi~of a discotic is determined using some of the
properties
of these materials that were measuredpreviously [il,12].
A cluster model ofshort-range order,
based on those usedpreviously-,
isdeveloped
toproduce
aprediction
for thetemperature dependence
of the correlation factor.The model is shown to be a
great improvement
over the mean-fieldresult,
and to give more accuratepredictions
of otherphysical properties
of discotics.2.
Experimental
2.1. REFRACTIVE INDEX AND PERMITTIVITY l/IEASUREMENT. The material used is the
mixture Tla
[12], containing
DB126[19] (triphenylene hexa-(2-methyl- 4-n-decyloxy)benzoate)
and
triphenylene hexa-n-heptanoate.
HET7 [20](Fig. 1).
The mixture hasphase
transitions(+2°C)
K
66[40] ND
90 1(°C).
The refractive indices of Tla were
measured,
to a tolerance of0.lsl,
using an Abbé re- fractometer withlight
ofwavelength
ÀNaD" 589.6 nm, as described
previously [12].
Thepermittivity
of Tlaparallel
to thedirector, ell,
tuasmeasured,
to a tolerance of0.25%,
us-mg a HP4192A
impedance analyser,
1v-ith thesample
m a 9 /tm thick cell ~v-ithhomeotropic
ahgnment-inducing
rubbedpolyimide layers [11].
2.2. KIRKWOOD CORRELATION FACTOR DETERMINATION. The I(irkwood
dipolar
corre-lation factor gi~ is a factor
showing
thedegree
andsign
ofdipolar association,
definedby [î]
~~~'~~
J#~
where ~j
=
II,1
is amacroscopic axis,
withrespect
to a director n, and~1~,~ is the
component
of aunique
axis m the i~'~ moleculealong
thej>-axis.
Trie factor can be defined in terms of anyunique direction m the molecule a direction
coinciding
v~ith the electricdipolar
directionis used here as electrical
properties
ai-e to be discussed. The summation is over the moleculesin a
macroscopic spherical
volume V centred ondipole
1. The effective mean-squaredipole
iuoment of a
phase (per molecule)
isgi,-en by (Pi )Phase
"
9~(Pi)mol> (~)
lvhere
(p()moi
is the mean-square moment of asingle
molecule. Hence the factor is less thanunity
foranti-parallel
correlation andgreater
than unity forparallel
correlation. Previousstudy
[11] has shown gill < 1 for DB126. Trie factor gi~ as a function of temperature wascalculated
using
the uniaxial Kirk,vood-Frôhlichequation
[7]~
e~ +
(cf e~)Q(
iii~~~ ~~ ~
e~~(?°°
+2)~ 9eokBT~~~
~~~~~~°~' ~' ~~"~'
~~~where e~ are the
required components
of thepermitti,;ity
tensor of the material ande~i
arethe zero
frequency extrapolation
of the atomic and electronicpei~mittivity contributions,
ap-proximated by e~i
= 1.05 + 0.0.5n( [21].
N is the molecular numberdensity,
calculatedusing
trie
approximation
for triedensity
p= 970 + 30
kg m~~ Q(
is ashape
factordepending
onthe
anisotropy
of c~j, defined in[7],
and the mean-square moleculardipole components parallel
and
perpendicular
to the director are taken as(p)
/)>moi =[t1Î(1
+25)
+PI (i S)]
,
(t1[),moi
=t1Î(1 S)
+
PI (i
+jS)j (4)
respectively
~KThere the in vacuo moleculardipole
components are mdexed s and t forlongitu-
dinal and transverse to the molecular short axis
[22].
Thisneglects
any biaxial nature of trie molecules concerned(1.e.
an average transversedipole
moment isassumed,
and the order pa- rameter D taken to bezero).
Inaddition,
the use of theseequations implies
trieneglect
of trieanisotropy
of inducedpolarisation [21],
1-e- that lia « ii. This is a rathergood approximation
for thisexample,
as shownpre,>iously [12].
The values for the moleculardipole components (taken
to be thetime-average
of the molecular resultanthere)
are measured using the methodadopted
for DB126previously [11].
These ai-e ps= 20.7 +1.0 x
10~~°
cm, /tt " 0 due to
the rotational
symmetry
of the molecule. The measured value of eII ruas
used,
and I and ei calculatedby extrapolating
e;sot<opte into the nematicphase, giving
asystematic uncertainty
of~-
2Sl.
Thisapproximation
isjustified by
thelarge
errors m p and ps that dommate the finaluncertainty,
calculated as~-
6%.
with a furtherpossible systematic
errer of~-
5%
due toe~i.
The results for the
dipole
correlation factor, gi~, ai-e shown mFigure
2 versus reduced temperature. The factor isisotropic
with a value of about 0.32 at temperaturesgreater
than TNT. This shows that there is a considerabledegree
ofanti-parallel
coi-relation in theisotropic
1670 JOURNAL DE
PHYSIQUE
I N°12~Î
O a
lÎ
°
~i g °O~a
C
i
°°Oa
£
°°~
ÎÎ
~~O
1
°O~
~
°Oa
o
°O
U
°q
© O O
~
é g
f~~
g aaoooooooooooooooooooooooq
0.80 0.85 0 90 1.00 1.05
Reduced temperature,
T/T~
Fig.
2. Kirkwood correlation factor gi~ of T1a versus reduced temperature.phase
asexpected
for these materials. When thesample
is cooled into the nematicphase,
the factor becomesanisotropic.
Theparallel component
has a value of about 0.24just
below thetransition;
this value continues todrap
as thetemperature
is lowered.Comparison
calamitic values [15] are forSPCH,
which has giIIdrap
from 0.70 to 0.61 at thetransition,
andSCB,
which sholv-s adi~op
from 0.49 to 0.43. We see that T1a has asignificantly higher degree
ofassociation, parallel
to thedirector,
than these calamitic materials.The
perpendicular
correlationfactor,
gii, has a ratherlarger
value than gi / /,showing
that there is less correlation in this direction. Thereis, however,
asignificant degree
ofperpendicular association,
which is found to decrease as thetemperature
is lo~KTered; this isexpected
when the associationparallel
to the director isincreasing.
3. Theoretical
3.1. CLUSTER MODEL OF SHORT-RANGE ORDER. In this
section,
asimple
model of short- rangeorder,
derived from apicture
of thephase
as small clusters ofmolecules,
is introduced.Each cluster consists of a central molecule surrounded
by
~i nearestneighbours,
none of whichare nearest
neighbours
of each other(this
is the Betheapproximation [23],
whichobviously
becomes less reahstic as ~i
mcreases).
Arepresentation
of this is shown inFigure
3. Each of the pairs is taken to mteract with apotential
E(cos ô~j)
= e
(ÀPI(cos ô~j) P2(cosd~j))
,
(5)
where
Pi
are theLegendre polynomials, d~j
is theangle
between the two molecularunique
axes,e is the
strength
of thepotential,
and is a parameter of the modelgiving
theproportion
ofdipolar
andquadrupolar
interaction(usually dipolar
interaction isneglected
in models of thistype,
e.g. that ofYpma
andVertogen [24],
and so =0).
Notice thatanti-parallel dipole
correlation is introduced
by
using apositive
value of and is not fundamental to the model.Parallel coi-relation could be included
by
usmgnegative
values of À. A considerablesimplifica-
tion has been made to arrive at this
potential, namely
that the manydipoles
andquadrupoles
that can occur on
real,
flexible molecules are taken as one,fixed, resultant,
so that thepotential
is
only
considered to bedependent
on theangle
between the molecules. Thisassumption
has been madeby
most otherauthors,
and is necessary in this case topreclude
thelarge
number ofa)
b)
~
',
l
,Î
1"
Fig.
3.a)
Schematicrepresentation
of the way m which the actual molecules arearrangea
in the mixture;b)
the way in whichthey
arerepresented
in the cluster mortel, with a coordination number of three. The dotted fines mdicate thepair-wise
interactions.unknown parameters that would otherwise be introduced. Various
simple
molecules are rather well describedby
thisapproximation,
such as thecyanobiphenyls. Unfortunately,
the discotic molecule whoseexperimental
results werepresented
earlier isconsiderably
morecomplicated.
However,
we shall see later that the results may well beapplicable despite
this.The outer molecules are taken to interact with the rest of the medium via the
Maier-Saupe
pseudo-potential [13]
V(cosdj)
=
-uP2(cosdj), (6)
where
dj
is theangle
between the normal to thej~~
molecule and thedirector,
and u is thestrength
of the effectivepotential acting
on the outer molecules m trie cluster. Thisapproximation neglects
the effect of adipolar
term m thepseudo-potential,
which is believed to bejustified
as thedipolar
correlation between nearestneighbours
isanti-parallel,
and sotends to cancel over
larger
distances.Retaining
adipolar
term m thepseudo-potential
also tends togive
a non-zero value for thedipolar
order parameter,(Pi (cos dj )),
which is undesirable for a nematicphase.
Theangle d~j
is found from thespherical
harmonic addition theorem [25]m=+1
n
(COSiY~j"
~j (~i)'~'Cl,m(Ùl> i~i)Cl -m(iYj
i~J)>(7)
m=-1
where çJ~,j are the azimuthal
angles
of the molecular short-axes about thedirector,
andCi m(d,
çJ~) are modifiedspherical harmonics,
giirenby (for
1=1)
Cl,Ù(Ùi>§~1)
~ COSÙz,Ci,+i(Ùi, ç7i)
=+2~i
sm
d~e~J~' (8)
In this case, we have
Pi (cas ô~j
= cas d~ cas
dj
+ sind~ smdj
cas çJ~j,
(9)
16î2 JOURNAL DE
PHYSIQUE
I N°12where çJ~j = ~~ pj. ~ie can then construct
(un-normalised)
distribution functionsf(Qo, Qj)
=exp(-E(cosdoj)fl), g(Qj)
=
exp(-17(cosçJj)fl) (10)
where
fl
=
III.BT
andQj
=([, çJj),
fi.om ~vhich we obtain the statistical~veight
of agiven
cluster of molecules
~
P(Qo; Qi, Q~,
,
Q~ =
j ~j f(Qo. Qj)g(Qj), (ii)
j=i
~vhere molecule '0' is the central one and Z is the
partition
coefficient.The
potential,
u, is then found from theYpma
andVertogen
condition[24], namely
that the condition of translational invariance(the
model netdepending
on which molecule is taken as the centralone)
isinterpreted by requiring
that the average orientation of the central molecule and itsneighbours
be the same. This is a weaker condition than that ofKrieger
and James[26],
whorequired
that the~veight
P of apair
of molecules be the sameregardless
of which moleculeis chosen as the central one, i e.
P(Qo, Qj)
=P(Qj, no)- (12)
The
Ypma
andifertogen
condition can be written/dQoP2(Qo)F(Qo)~
"/ dQoF(Qo)~~~ / dQiP2(Qi)f(Qo, Ri )g(Qi), (13)
where
F(fia)
"
/ dQJf(Qo, Qj)§(QJ). (14)
The external field
strength
is therefore foundby solving
/d(cas do)P2(cas do)F(cas ôo)~
=d(cas ôo)F(cas do)~
~~d(cas ôi)P2(cas di)
xÎ Î
xG(cas ôo,casai)g(cas ôi), (là)
for
flv,
as a fonction oftemperature (used
in the reduced inverse formfie here),
withG(cas ôo,casai)
=
dçJoif(cosôo,
casôi,ç7oi), (16)
and
F(cosdo)
=
/ d(cas ôi)G(cosdo, casai)g(casai). (lî)
This is found to be the method that reduces
computation
time to a minimum.The entropy of the cluster is given
by
-kB / doo / dei /
dQ~ P(Qo Qi,
Q~ InP(Qo; Qi,
,
Q~
=
-kB
In Z~ikBfle(a ÀT) j,kBflvs. (18)
(19)
where
~ç =
p~jcos doj))>
is a
short-range dipolar
orderparameter,
a =
(P2(cas ôoj)), (20)
is a
short-range quadrupolar
orderparameter (the angle
bracketsindicating
the ensembleaverage),
and S is the usuallong-range quadrupolar
orderparameter
S =
jP2(cosÙo)). j21)
The internal energy of.the cluster is
U =
-~ie(a ÀT) ~ivS, (22)
where the last term
represents
the contribution of the effectivefield, giving
for the Helmholtz free energy~ ~~ ~ ~
Î~'~~'
~~~~This function is found to exhibit a first-order transition for all values of ~i
greater
than two.This is as
expected,
since a system with two nearestneighbours
isone-dimensional,
and it is well known that one-dimensionalsystems
exhibit no transitions at finitetemperatures [27j.
The
equilibrium
conditions at thetemperature
of interest can then be found from the minimum of F withrespect
to the orderparameters,
which are calculated from trieconsistency
relationsd(cosdo)P2(cosdo)F(cosdo)~
S =
,
(24)
Z
d(cas do) d(cas di) dçJoi P2(cas ôoi)~b(cas do,cas dl, ç7oi)
~ =
(~~)
Z
and
/ d(cas do) / d(cas di) / dçJoiPi(cas dei)ù(cas do,cas di, ç7oi)
~f
=
(~6)
Z where
Z =
/d(cas ôo)F(cas ôo)~, (27)
and
#(cas ôo,cas di, ç7oi)
"
f(cas do,
casdi, ç7oi)g(cas ôi)F(cas ôo)~~~ (28)
Given the
parameters
~i and À, the above model isuniversal,
if the volumedependence
ofe is
neglected (it
~vill tend todepend
on the inter-molecular distance to the inverse power of six[24j).
The calculated order parameters ~s a function of reduced
temperature
are shown inFigures
4 to
7,
with ~i=
3,
for various values of À. Transition data are shown in Table I. The first-ordertransition
temperature, (fie)NI,
is found from the conditionthat,
at thetransition,
the free energies of the nematic andisotropic phases
ai-eequal.
The reducedtemperature
is thenequal
to
(fie)NI Ifle.
The transitionenthalpies
in Table I show a considerable reduction in comparison with the mean fieldtheory,
however the calculatedIiHNI
for Tla of 850 Jmol~~
is stilllarger
than the measured value of around 200 Jmol~~
Notice that theparameter
T isnegative
at16î4 JOURNAL DE
PHYSIQUE
I N°12à=o,y=3
~
~ ""...-...
Q~
~
É E
~
~
0(
~O
07 08 09 10 1-1
Reduced temperature,
T/T~
Fig.
4. Calculated order parameters for ~ = 3, à= 0. The shaded fine shows the
34aier-Saupe
mean-field S order parameter for companson.
= 0.25, y= 3
$
""""1,,_
Î$
É E
~
~
_,_~,-.~.,-
(
~ T _~___~_,,,.-~.-"j~j_~__,,,~---.O ~-."jzziZIlCI1°Z~~~~~'~~~~~~"'1'
0.7 10
Reduced
temperature, T/T~
Fig.
5. Calculated order parameters for ~ = 3, à= 0.25. The shaded fine shows the
Maier-Saupe
mean-field S order parameter for
comparison.
all
temperatures,
asexpected
for a parameterdescribing anti-parallel dipole
correlation. Itis also non-zero in the isotropic
phase,
as is a,although
a(small)
discontinuousdrop
in itsmagnitude
is seen,showing that,
asexpected, short-range
orderpersists
even whenlong-range
order has vanished. While the calculated form for T is
qualitatively
similar to that for a similar orderparameter predicted by
~fadhusudana and Chandrasekhar[16j,
themagnitude
of T israther
larger.
Madhusudana and Chandrasekhar used anapproximate compliance
with theKrieger-James condition,
which has been found togive
incorrect results[24j). They
also do not state the coordination number in their calculation.One important
point
about this model is that it makes no transparent distinction between discotic and calamitic nematics.However,
differences between the two types of molecules can be covered in several ways, not least because themagnitude
and volumedependence
of thepotential
e(and
therefore the transitiontemperature)
may well be different for the differentaspect
ratios. The main distinction between discotics and calamitics is that alarger
number ofÀ=1.0, y=3
i
oo io i-i
Reduced temperature,
T/T~
Fig.
6. Calculated order parameters for ~= 3, à
= 1.0. The shaded line shows the
Maier-Saupe
mean-field S order parameter for comparison.À=2.5, y= 3
0.7 0.8 0 9 10 1 1
Reduced temperature,
T/T~
Fig.
7. Calculated order parameters for ~ = 3, à= 2.5. The shaded fine shows the
Maier-Saupe
mean-field S order parameter for
comparison.
nearest
neighbours
may be tolerated in calamiticexamples
without the Betheapproximation
breaking
downseverely.
One wouldexpect
~i in a calamitic version of the model to have values of around 6 to8,
whereas the minimumintegral
value of 3 is used here for discotic calculations.This bas the desired effect of
making
the deviation from mean-field less incalamitics,
which does seemsupported by experimental
observations. Forexample
theenthalpy
of transition isan
increasing
function of j, and transitionenthalpies
are found to be lower in discotics than in calamitics. Results with anexample parameterisation
for apolar calamitic,
~i =8,
=o-à,
are shown in
Figure 8,
and in Table I.The ttse of a low coordination number in the discotic calculations is
supported by
computersimulations of a
polar discogen [28j,
which find the molecules associated in shortcolumns,
andby
NMR measurements ofbenzene-hexa-n-heptanoate [29j,
which find the molecules associ-ated in pairs. It is also consistent with the distribution function results from the computer
1676 JOURNAL DE
PHYSIQUE
I N°12Table I. Transition
temperatures (fie)Ni,
transition orderparameters SNI
andenthaipies liHNi/TNT for
avariety of
valuesof
the modelparameters
~i and À. Note thatliHNi
"~IUNI,
as the volume
dependence of
the model has beenneglected.
~i
(fie)Ni SNI liHNi/TNT
J
I<~l mol~~
3
3 0.25 2.202 0.404 2.387
3 0.50 2.011 0.404 2.414
3 1.00 1.660 0.404 2.444
3 1.50 1.406 0.404 2.422
3 1.80 1.288 0.403 2.402
3 2.50 1.078 0.401 2.349
8 0.50 0.5954 0.424 3.344
oJ 0.00 4.452 0.429 3.473
S ~~~'~'~~~
0.7 0.8 0 9 1-o 1 1
Reduced temperature,
T/T~
Fig.
8. Calculated order parameters for ~= 8, à
= o-à- The shaded fine shows the
Maier-Saupe
mean-field S order parameter for comparison.simulations of Emerson et ai.
[30]
and de Luca et ai.[31].
The paper of Biscarini et ai.[28],
which uses two-site cluster
theory
as well as Monte-Carlosimulation, provides
values for the short- andlong-range
orderparameters
that arecomparable
with thosepresented
in this pa- per. The results for the second-ranklong-range
order parameterpresented
here are well magreement
withprevious experiment [12].
In contrast, the results from Emerson et ai.[30]
donot
correspond
withexperiment, being
much toolarge
at the neiuatic toisotropic
transition.Of course the
parameter
allows additional iTariation to beintroduced,
and acomplete
theory
of intermolecular interactions should allow this parameter to be calculated forparticular
molecules. For now, we can estimateby attempting
to calculate the ratio ofdipolar
andquadrupolar
interactions fortypical
molecules. There is no reason, either mathematical orphysical, why
the parameter i need beintegral,
aslong
as it isthought
of as the average coordination nttmber in theliquid~
This willprovide
further freedom fordescription
of realisticliquid crystal phases,
and allow transitionenthalpies
to be reduced still further.~Î
Iso~opic(
E~
é
~
Î
Nematic~a~
O
T/T~,=0.85
00 0 2.0 3.0
Model parameter, 1
Fig.
9. Values of the parameter T' for vanous values of the mortel parameter à at diiferent reduced temperatures, I.o(in
theisotropic phase),
I.o(in
the nematicphase)
and 0.85.3.2. USE OF SHORT-RANGE ORDER PARAMETER IN DIPOLE CORRELATION CALCULATION.
Using
the definition ofgill
given inequation (2) above,
we can express the correlation factor in terms of the cluster model asgill
" 1 +(cosdo cosdj)
= 1 +T', (29)
with
d(cos do) d(cos di) dçJoi
cosdo
cosdiifi(cos do,
cosôi, ç7oi)
T'=
(30)
Z
In
principle
we can now use thisequation
to fit trieexperimental data,
with one free parameter, À. The range ofpossible
values for T' is shown inFigure 9,
and the calculated correlation factorshown in
Figure 10,
for= o-à and
2.5,
wîth theexperimental
values forcomparison.
The resultant calculated value for the
dipole
correlation factor is found to be rather toolarge,
1-e- themagnitude
of T' is still too small togive
agood
fit to the data.Figure
9suggests
thatlarger
values cannot be obtained without iismgunrealistically large
values of À. Indeed the value of 2.5 is still ratherlarger
than islikely
inreality.
However thetheory
isclearly
farbetter than
previous
mean- fieldresults,
as these wouldgive
a correlation factor muchgreater
thanunity.
3.3. PERMITTIVITIES IN THE SHORT-RANGE ORDER À~ODEL. À~adhusudana and Chan-
drasekhar
[16]
give expressions for the averagedipole
momentparallel
andperpendicular
to thedirector,
with /tt"
0,
as(in
thisnotation)
l~ll/)ajuster
= ~IÎ(~~
+ (COSlYo COSYJ)ajuster) 131)
and
l~llcluster
"
~Î l~ j
~ +jslll do
SllliJj
COSÇ7o COS7j)cj~~t~~.) (32)
1678 JOURNAL DE
PHYSIQUE
I N°12bù
fl
<E
(
_____,,.Q À~_§;j~__-,,----~-""~"'
~'~""""' À=25
~ jf
ù
ooooooooooooooooooooooooooooo
o.ss o 9o o
Reduced temperatare,
T/T~
Fig.
10. Calculated values for gill versus reduced temperature, for two values ofà,
with the measured values(open circles)
forcomparison.
=0.5, y=3
42 ~,
)
~ Em'@
i _,__....:...
à
.'"'"'Ei
0.8 0 9 1-Ù 11
Reduced temperature,
T/T~
Fig.
Il. Calculatedpermittivities
for ~= 3, à
= o-à, and other parameters calculated using the measured
properties
of T1a.Substitution of these results into the Maier-Meier
equations [14]
8~ -1 =
@ (a~)
+
)1J1j)j
, ~t =
II,1, (33)
gives results for the
permittivities
in theshort-range
ordermodel,
with the meanpermittivity given by
1 1
=
~~~
a +
~~)
l +(cos do
cos dj +
2(sin do
smdj
cos çJo cos çJj(34)
eo
3kB
~
We can then calculate these
permittivities
usmg values fordipole
moments andpolarisabilities
calculated
earlier,
andtaking
both h and F asapproximately 1.3,
as calculated from themean
permittitivity
andpolarisability [14].
The calculatedpermittivities
for the case ~i=
3,
À= 2.5, y=3
3
""1'
E~ ~
~
'@ _...""'
)@ _,,:""'
Î
__:..."' E
~ ,,:."'
0.8 0 9 1-Ù 1 1
Reduced temperature,
T/T~
Fig.
12. Calculatedpermittivities
for ~= 3, à
= 2.5, and other parameters calculated
using
the measuredproperties
of Tia.= o-à are shown in
Figure
11. Thelarge drop
in the meanpermittivity
isclearly
seen, due to thechange
indipole
correlation at the transition. The results are m fact rather similar to thosepresented
for DB126[11],
andprovide
further evidence for theefficacy
of this model.For
comparison,
the mean-field result for the same valuespredicts
apermittivity anisotropy
of around +7, which is much too
large. Very large
values of theparameter give
different results(Fig. 12),
as thelarge
amount ofanti-parallel
correlation removes the effectiveness of theparallel dipole
moment,leaving
thepolarisability amsotropy,
which isnegative,
as theonly
contributor to the
permittivity anisotropy. Density
variations areignored
in thepermittivity calculations,
as ~v-ell as thetemperature dependencies
of h andF,
which would make thediscontinuities at the transition
greater,
and trietemperature dependencies
somewhat different.4. Conclusions
The Kirkwood
dipole
association factor gi~ bas been measured as a function of temperature for a nematic discotic mixture. Thesystem
showed a considerabledegree
ofanti-parallel dipole
correlation,
assuggested by previous study [11],
whichpersisted
into theisotropic phase.
This is in conflict withprevious
models ofdipole correlation,
whichpredict parallel
associationm
discogens.
A cluster model ofordering, mcluding short-range dipolar
orientationalorder, quantified by
aparameter T,
was used toattempt
a more reasonableprediction
ofdipole
correlation in thesesystems.
Thetheory
was found topredict
thedipole
association ratherwell,
with agreat improvement
over mean~ fieldresults,
and also had other desirablefeatures, including
a reduction m transitionenthalpy
anddrop
m the meanpermittivity,
?, when thesystem
was cooled into the nematicphase.
Acknowledgments
Workers at the School of
Chemistry, University
ofHull,
are thanked forsynthesis
of the ma- terials described in this paper. The invaluable contribution of J.C. Jones to this work isacknowledged.
This work wassupported by
theStrategic
Research Committee of the U-Ii-Ministry
of Defence. British CrownCopyright, 1995/DRA.
Publishedby
permission of Her BrittanicMajesty's Stationery
Office.1680 JOURNAL DE
PHYSIQUE
I N°12References
Ill
Bom M., Sitz.Phys.-Math.
25(1916)
614.[2] Szivessy G., Z.
Phys.
31(1926)
159.[3] Vorlander D., Z.
Phys.
Chem. A126(1927)
449.[4] Cotter
àI.A.,
Phu. Trans. R. Soc. Lond. A309(1983)
127.[Si Gelbart iii-M-, J.
Phys.
Chem. 86(1982)
4298.[6]
Toriyama
K. and DunmurD.A.,
Mol.Cryst. Liq. Cryst.
139(1986)
123.[î]
Bordewijk
P.,Physica
75(1974)
146.[8] Dunmur D.A. and Miller
iii-H-,
Mot.Cryst. Liq. Cryst.
60(1980)
281.[9]
Toriyama
K. and Dunmur D.A.. Molec.Phys.
56(1985)
479.[10] Dunmur D.A. and
Palify~muhoray
P., Motec. Phys. 76(1992)
1015.[Il] Philhps
T.J., Jones J.C. and McDonnell D.G., Liq.Crystatsls (1993)
.203.[12]
Philhps
T.J. and Jones J.C.. Liq. Crystals 16(1994)
805.[13] Maier W. and
Saupe A.,
Z.Naturforsch.
15a(1960)
287.[14] à>laier W. and Meier G., Z.
Naturforsch.
16a(1961)
262; à'laier W. and Meier G., Z.Naturforsch.
16a
(1961)
470.[15] Bradshaw M.J., Ph.D.
Thesis, University
of Exeter(1985).
[16] Madhusudana N.V. and Chandrasekhar S., Proc. Int.
Liq. Cryst.
Conf.Bangalore,
Pramana.Suppl. (1973)
57.[iii Toriyama
K., Dunmur D.A. and Hunt S-E-,Liq. Crystals
5(1989)
1001.[18] Jones
J.C.,
Proc. FLC '93(1993).
[19] Beattie D.R., Ph.D. Thesis,
University
of Hull(1993).
[20] Destrade C., Mondon M.C. and Malthete J., J.
Phys.
France 40(1979)
C3-1î.[21]
Bordewijk
P. and De Jeu iV.H., J. Chem.Phys.
68(1978)
116.[22] De Jeu ~V.H.,
Physical Pioperties
ofLiquid Crystalhne
Materials(Gordon
and Breach. London1980).
[23] Bethe H.A., Froc. R. Soc.149
(1935)
1.[24] Ypma J.G.J. and
Vertogen
G., Sohd State Commun. 18(1976)
475.[25] Bleaney B.I. and
Bleaney
B.,Electricity
and Magnetism, 3rd. edition(Oxford
University Press1976)
p.40.[26]
Krieger
T.G. and -JamesH-M-,
J. Chem.Phys.
22(1954)
796.[27]
Sheng
P. andWojtowicz P.J., Phvs.
Rev. A 14(1976)
1883.[28]
Zarragoicoechea
G-J-.Levesque
D. and uleis J.J., Mot.Phys.
78(1993)
1475.[29] Mabmak A., J. Chem.
Phys.
96(1992)
2306.[30] Emerson A.P.J., Luckhurst G-R- and
Whatling S.G.,
Mot. Phys. 82(1994)
113.[31] De Luca