HAL Id: jpa-00207295
https://hal.archives-ouvertes.fr/jpa-00207295
Submitted on 1 Jan 1972
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.
Static and dynamic behavior of a nematic liquid crystal in a magnetic field - Part I : static results
P. Pieranski, F. Brochard, E. Guyon
To cite this version:
P. Pieranski, F. Brochard, E. Guyon. Static and dynamic behavior of a nematic liquid crys- tal in a magnetic field - Part I : static results. Journal de Physique, 1972, 33 (7), pp.681-689.
�10.1051/jphys:01972003307068100�. �jpa-00207295�
STATIC AND DYNAMIC BEHAVIOR
OF A NEMATIC LIQUID CRYSTAL IN A MAGNETIC FIELD
PART I : STATIC RESULTS
P.
PIERANSKI,
F. BROCHARD and E. GUYON Laboratoire dePhysique
des Solidesd’Orsay
associé au CNRS(Reçu
le 18février 1972)
Résumé. 2014 Nous discutons théoriquement et expérimentalement les propriétés statiques et dynamiques de la transition de Freedericks dans du MBBA à l’état nématique. Les résultats expéri-
mentaux ont été obtenus, dans la géométrie planaire et homéotrope, à partir de l’anisotropie de
conductivité thermique à travers le film. Les expériences permettent d’estimer les valeurs de deux constantes élastiques (flexion et éventail).
Nos experiences conduisent aussi à la première estimation non ambigue de l’anisotropie de
conductivité thermique
(k~
2014k)/k
= 0,64 ± 0,04.Abstract. 2014 We give an extensive theoretical and experimental discussion of the statics and
dynamics of the Freedericks transition of nematic MBBA films in a magnetic field. The experi-
mental data are obtained from measurements of the anisotropy of the thermal conductivity across
the film both in the « planar » and « homeotropic » configurations. Estimated values of the bend and splay elastic constants are obtained.
Our measurements also give the first unambiguous value of the anisotropy of the heat conducti-
vity
(k~
-k)/k
= 0.64 ± 0.04.Classification Physics abstracts :
14-82
1. Introduction. - A nematic
liquid crystal (LC)
thin film of thickness d
kept
between twoparallel glass plates
can be orientedby
the solid boundaries.By rubbing along
one direction theglass
surfaces[1],
one can obtain
single
domainliquid crystals having
the director axis n in the
plane
of the film(planar case). Using
surfaces which are very clean or chemi-cally
treatedby
anappropriate
surfactant agent, ncan be made
perpendicular
to theplane
of theplates (homeotropic case).
If amagnetic
field isapplied
atright angle
to thedirector,
the nematicordering
ismodified above a critical field
H c’
the transitionbeing
second order.
This effect was first studied
optically by
Freede-ricks
[2]
for films in thehomeotropic configuration (geometry
3 ofFig. 1).
Zocher[3] interpreted
thesestatic
properties using
the continuum Frank-Oseen[4], [5] theory.
The critical fieldH,,,
isgiven by
where the index i
(1, 2, 3)
refers to the threegeometries
of
figure
1. This relation expresses theequilibrium
between the
magnetic
torque due to theanisotropic
part of the
magnetic susceptibility
xa and therestoring
torque due to the elastic constantKu.
The critical field has been measured for severalgeometries
indifferent
liquid crystals,
inparticular
from conosco-pic [6]
and dielectric constant[7]
measurements.FIG. 1. - The three geometries of the Freedericks transition.
The alignment of the molecules close to the glass plate
(z
=d/2)
is « planar » in case 1 and 2 and « homeotropic » in case 3.
A large enough magnetic field (H > He) at right angle with the initial alignment creates a « twist » distortion (geometry 2) or
a mixture of « bend » and « splay » distortion (geometries 1
and 3).
In this
article,
weanalyse
the Freedericks transition from measurements of the thermalconductivity anisotropy.
This method is also well suited forstudying
the
dynamics
of the transition when the characte- ristic times of the transition i, uponvarying
the fieldH,
are
large compared
to the thermal relaxation time zth[8] expressed by
(k
is the thermalconductivity,
p thespecific
mass,Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01972003307068100
682
c is the heat
capacity).
Thedynamic behavior,
whichwill be discussed in part II is characterized
by
a timeconstant 1"
given by [9] :
K is an elastic constant, yi is a
viscosity
coefhcient.With
typical
values of k = 4 x10-4 cal/s
cmOK,
c =
0,4 cal/g,
p = 1g/cc,
K = 10-6dynes,
y - 10-1poise,
we getIn
chapter II,
we describe thesample preparation
and the
general
results : an estimate of the absolute thermalconductivity k,
and characteristic times of the transition.In
chapter III,
we calculate the static thermal transportproperties
in the presence of a constantmagnetic
field in the threeconfigurations of figure
1.Our
experiments
were done in the twocomplemen-
tarygeometries
1 and 3. The measurements of the effective thermalconductivity along
oz- ke
- in thepresence of distortion
give
two sets of valuesK11
and
K33,
and two determinations ofk.Ikil,
We callkjj
andkl
the thermalconductivity
when the heat flow isparallel
orperpendicular
to the axis of a welloriented
liquid crystal.
Theanisotropy
of the thermalconductivity
iska
=kil - k,.
The second part of this article is devoted to the
dynamics
of the Freedericks transition based on thehydrodynamic
Leslie-Ericksen[10] theory
and onthermal as well as
optical
results.II.
Expérimental.
-A)
MEASUREMENT TECHNI- QUES. - The aim of theexperiments
was to measuresmall
changes
in thermalproperties
rather than to obtain absolute thermalconductivity
data. Nospecial
care was taken to suppress heat leaks
(radiation, convection,
thermal conductionthrough
other heatpaths).
The cell used in the thermal measurements wasdesigned
to allow directoptical microscope
obser-vation
during
theexperiments.
It was surroundedby
a vertical solenoid and a
pair
of horizontal axis Helmholtz coils. Fields up to 600G,
in anydirection,
could be obtained.
A schematic
representation
of theoriginal
set-up isgiven
infigure
2. The LC film[1]
issqueezed
bet-ween two horizontal
(2
cm x 3cm) glass plates [2],
with a
thin, low-thermal-conductivity ring [3]
as aspacer. The
sensing
elements are metallic filmsdeposited
on theglass plates, using
standard evapo- rationtechniques.
The heater is asemi-transparent gold
film[4] evaporated
on the lower face of the1/10
mm thin bottomglass plate.
The heat conductionthrough
theglass
islarge compared
to thatthrough
the LC. The
temperature
of the upperglass plate
was
regulated by circulating
water. The latterplate
FIG. 2. - Schematic description of the thermal measurements on a L C film (in 1).
was
kept
thick(1 mm)
to avoid effects due to water0
pressure. A dc heat
input Q
up to 5 W can be obtained.0
An unknown fraction of
Q (of
the order of 60%)
flows
through
theliquid crystal
film. In thefollowing,
we assume that this fraction does not
depend
on theamount of heat
input although
none of themajor
conclusions of this work
depend strongly
on thisassumption.
Two identical sets of nickel
[5]-copper [6]
thermo-couple films,
Xshaped,
andfacing
eachother,
wereevaporated
on theglass plates.
The films were 1 000A
thick and not transparent but their width was small
(2 mm)
andthey
coveredonly
a small area of thevisible field. The two nickel films
[5]
were connectedby
a thin nickel wire.The temperature difference between the thermo-
couples
- AT - could be read from thevoltage
difference between two copper wires soldered at the ends of the two copper films
[6].
The thermoelectric power of thesethermocouples
was found to beequal
to 25pV/°C
at 25 °C with agood reproduci- bility
betweensamples,
and withoutaging.
Thedifferential
voltage
as well as the absolutereading
from each
thermocouple,
was measured on aKeithley
148 nanovoltmeter and recorded on an
X,
tplotter.
We also used the temperature
dependence
of theresistance of the
gold
film to cross check ourresults,
with a verysatisfactory
agreement.In the later
experiments
thethermocouples
wereseparated
form the LCby
thinglass plates glued against
the initialglass plates
withoptical
epoxy resin. The thermal time constant wasconsequently increased,
but it was easier to carry out « efficient » surface treatments discussed below. The LC thickness could also be variedcontinuously by
a micrometricdisplacement
of the upperplate.
The accuracy on the thickness measurement was rather poor(5 %),
due in a
large
part to the deformation of theglass plate (effect
of the externalstresses).
Inparticular,
the
temperature
differencedepended
on the pressure of thecooling
water which wasregulated by
a constantpressure
supply
and a fine flow-rateregulator
at the output. Because of theuncertainty
ond,
we will characterizeit,
wheneverpossible, by
the value ofH,,.
B)
MATERIAL USED. - We have studied metho-xybenzylidène butyl
anilin MBBA. The critical tem-perature,
Te,
for thenematic-isotropic
transitionwas around 42,OC. In the initial
experiments [11],
the material used had a
Tc
of 36 °C. The effect ofimpurities [6]
may haveexplained
thesignificantly
different values of the elastic constants obtained in this report
(see
IIIC).
C)
THERMAL CONTROL EXPERIMENTS. - Severalexperiments
were carried to check thereliability
ofthe
techniques
used.1)
The heatconductivity
of MBBA was determinedfrom a
comparison
with several materials of compa- rable known heatconductivity.
If the measurements aredone in the same conditions
(same applied
power and averagethickness),
thechanges
of temperature AT fora
given
material areexpected
to varyproportionally
to the
changes
of thickness Ad :(AdIAT)Q,d
= oc.kwhere a is
independent
of the material. This has been checkedexperimentally
on severalexperiments.
The results of a
typical
runusing glycerol, amyl
acetate and
planar
andhomeotropic
MBBA aregiven
on
figure
3 andgive :
FiG. 3. - Several materials have been studied thermally in similar conditions. The variation of the temperature drop due to
a small change of thickness is proportional to k for known materials and provides a measurement of the heat conductivity
of MBBA in the two orientations. (Note straight line depen- dence through origin.)
We will see in III that the
anisotropy
is in excellentagreement with measurements in a
magnetic
field.The values are of the same order of
magnitude
asthose for PAA
[12], [13] where kil
= 2.8 x10-4 cal/
cm . s . deg.
Theconductivity
does not varyappreciably
with T up to 40oC. In this paper, we restrict our
attention to measurements around 20 OC.
2)
In a secondexperiment,
we have studied the thermal time constant -rthby witching
on(off)
thepower at t = 0 and
watching
the rise(decay)
of AT.The results in the
planar
geometry aregiven
infigure
4.For
large
valuesof t,
the behavior isexponential
FIG. 4. - A measurement of the thermal diffusivity is obtained
from the thermal response when the heat power is applied. In
a planar L C, the thermal response is slower in zero field where the heat conductivity is smaller. The initial slopes are the same
in the two cases because the saturated t5T value is larger in zero field. (The anisotropy observed at saturation is smaller than the maximum anisotropy of k as H was not very large compared to Hc and as this recorded AT induded also the temperature drop
across the limiting planes.)
with a
single
time constant which reflects the dominant conduction mechanism : Tlh - 3 s. Suchdynamic experiments
should inprinciple give
a measurement of the thermaldiffusivity
of the system(= K/C,
where Cis the heat
capacitance
and K the thermal conduc-tance),
b is controlledby
the slowest heat process(the
heatdiffusivity through
theliquid crystal
is10 to 50 times smaller than that
through
theglass).
We can see
that,
when alarge enough
field isapplied
so that the thermalconductivity
of the LCgoes
from kl
tokll,
the time scale of the curveAT(t)
is decreased
by
acorresponding
amount.(The
satu-rated AT value is
larger
in theplanar
case and theinitial
slopes
areequal.)
This shows indeed that the thermal conductionthrough
theliquid crystal
isresponsible
for themajor
part of the heat flow.However,
a detailedquantitative analysis
for thismultilayer
system would be difficult[8].
D)
SURFACE TREATMENT AND CONTROL. -1)
Pre-paration.
-a)
Planar case. A LC can be orientedparallel
to aglass
surfaceby polishing
theglass plate along
one direction with paper or abrasive[1].
It appears
(see 2b)
that a rather strongpolishing
isrequired
to get agood ordering.
In the initial expe-riments,
thethermocouples
were in direct contact with the LC andonly
a weakpolishing
wasapplied
in order to avoid to scratch the metallic films the
alignment
remainedimperfect (n
in theplane
of symmetry ofpolishing
but finite tiltangle
at thesurface).
b) Homeotropic
case. On the clean surfaces of theglass plates,
a thin film ofhexadecyltrimethylammo-
684
nium bromide was
applied starting
with a dilutesolution in toluene. The thickness of this film corres-
ponded probably
to asingle
molecularlayer [14].
The LC was then inserted between the
parallel glass plates.
The initialalignment
of the LC takes along
time
(several
hours for a400 J.1 film) :
when theliquid
is inserted between the
plates,
the molecules arestrongly
orientedby
the flowparallel
to theplates.
The
large
time relates to the exclusion of the structural defects(disclination loops)
which is followedoptically.
2)
Controlof
orientation. -a)
Useof polarized light.
The search for theprincipal
axis between crossedpolaroids
does not lead to the value of the tiltangle
in a poor
planar sample (see la).
It is also insufficient for agood testing
of thehomeotropic alignment.
We have characterized the orientation more accu-
rately, using conoscopic techniques.
In recent expe-riments,
we also used them tostudy
the static[15]
and
dynamic [16]
aspects of themagnetic
transitionquantitatively.
b) Optical images.
- Thestudy
of the textures inthe presence of
magnetic
fields and thermal effects has been used to control the orientation but isbeyond
the scope of the present
study.
Thebirefringence
canalso
give
asimple
test ofalignment :
If a small dustparticle
is located close to the lower face of theLC,
itgives
asingle image only
when one of theprincipal
axes is
perpendicular
to the LCplane.
This wasusually
obtained for the films used in thisstudy.
When a field
larger
thanH,,
isapplied,
twoimages
are seen. The
separation
goesthrough
a maximumfor an intermediate field value. In
large enough fields,
where most of the LC is orientedalong
thefield,
the twoimages
get very closeagain.
c) Finally,
the thermal measurementsgive
alsogood
tests ofalignment.
Let usanticipate
on nextchapter
andfollow,
onfigure 5,
thechange
of Tacross a
homeotropic
film when its thickness issuddenly
decreased(point A).
Along enough
timeFIG. 5. - The thickness of a 500 p homeotropically aligned L C is decreased suddenly (point A) to 400 p. Right after this change (point B), the temperature drop across the sample increased due to the decrease of the effective heat conductivity in the disor- dered state created by the flow (the same process is reproduced
from 400 to 300 p).
after the
change
of thickness(point C),
the tempera-ture difference across the film is smaller than at the initial
point
A. However,right
after thechange of d,
the temperature difference increases : the disordered stateobtained,
due to thischange,
has an effectiveheat
conductivity ke
smaller than the ordered state onekll (as kll > ke
>kl). However,
the oriented state is restored much faster than when the LC was insertedinitially
between theplates.
This time becomes evenshorter for smaller thicknesses.
A last thermal test uses the measurement of the
dynamics
of the distortion when amagnetic
field issuddenly applied (see
partII).
It turned out to bethe most sensitive to small
misalignments.
All
samples
used in this report weresatisfactorily
tested with the different
alignment
methods.III. Static behaviour. -
A)
THEORY. - Westudy successively
the effect of amagnetic
field on the ther-mal
conductivity
in thegeometries
offigure
1.1) Geometry
1. - Amagnetic
field isapplied along
the z axis
perpendicular
to theliquid crystal
directortaken
along
ox. We callO(z)
the average orientation of the molecule with respect to the x axis. The effective heatconductivity ke
isgiven by :
where
A;[0]
=kl + ka sin’ 0 ; ka
is theanisotropy
of the heat
conductivity.
The
equilibrium configuration
can be obtainedfrom the minimization of the Frank free energy
density (per
unit area of thewall)
The first term describes the elastic contribution of the
splay
distortion(K11)
and the second one that of thebending (K33).
The distortion is causedby
theanisotropy
of themagnetic susceptibility (xa).
Thethird term represents this
magnetic
energy contri- bution. For small values of thefield,
the stableconfiguration corresponds
to0(z)
= 0. Above acritical field H =
H,,,,
theliquid crystal
becomes dis-torted.
HCI
is obtainedby looking
for a minimum of Fkeeping
the lowest order term in 0This limit
equation depends only
on one «splay »
elastic constant. The distortion is of the form 0 = a cos
z/çl(H).
The
length
gives
the range of effect of the wall in the presence of H and is called coherencelength (although
it is differentin essence from the coherence
length
insuperfluids,
for
example,
where the coherencelength
is fieldindependent).
By applying
theboundary condition, 0 [z
= +d/2]
=0,
one gets :
or
When H is
larger
thanHC1’
the solutionO(z) depends
on the two parameters
K11
andK33.
A firstintegral
of the functional derivative of F with respect to 0
can be
easily
obtained :where n
=(K33 - K11)/K11
and the maximum dis-tortion
angle 0.
= 0(z
=0).
The
integration
of eq.(III.S) using
theboundary
conditions
gives Om
as a function of I-I.By using
sin u = sin
0/ sin 0 m’
one findsThe
integral
I islarger
thann/2
and one recoversthe fact that
0.
> 0 is obtainedonly
when H >HCI.
Using (111.5),
the eq.(III.1)
becomes :The eq.
(III.6)
and(III.7) give
animplicit
expres- sion of the effectiveconductivity ke
=f(h),
wherewe write h =
HIH,,.
In the limit where h - 1 is
small,
theseequations
can be
easily
solvedThe
anisotropy
of the thermalconductivity ka
isobtained
by measuring ke
forlarge
fields wherek,r --> k il .
From the initialslope
of theke(h)
curve,one gets a value of the ratio
K11IK33.
The
shape
of thecomplete
curvek.(h)
iseasily expressed
in thelimiting
case of a small thermalanisotropy k.1k, « 1
and when the Frank coefficientsare
nearly equal.
Theintegral
of(6)
and(7)
can beexpressed
then as anelliptical integral (we
will usethe notations of reference
[17])
andke
isgiven
fromthe
coupled equations :
The
corresponding
calculated curves aregiven
asa function
of 11
onfigure
6. One seesclearly
that thevalue
of 11
is ratherunambiguously
determined from the initialslope
alone.FIG. 6. - The decrease of temperature Jf in the presence of
a field H is calculated in the planar case in the limit conditions of formula (III.9) for several values of the anisotropy of the
elastic constants Ki, 1 and K33. The normalized curve is rather unambiguously characterized by the initial slope.
2) Geometry
2. - In this case, one has a pure twist deformation. The free energy can be written as :In the case of small fields
(B small),
the results canbe deduced from
(1) by having tl
= 0 and K =K22-
The critical field is :
In this geometry, the
anisotropy
of k would haveto be measured in the
plane
of the film. An unknown fraction of the heat flow istaking place through
theend
plates.
For this reason, we will not report here anyexperimental
results in this geometry.In the limit where
(ka/kl) sin’ Bm
«1, ke
isgiven
from the
following equations :
686
3) Geometry
3. - The free energy is :This case is identical to case 1 if one
interchanges Ki l
and
K33, kli
andk1-.
B)
EXPERIMENTS. -1) Homeotropic
case. - Infigure
7, we present atypical
set of recorded thermalFiG. 7. - Homeotropic film : A magnetic field is applied in the plane of the film (point A) and the temperature drop across
the sample increases. When the field is decreased by steps, the decrease of ô T is exponential.
results
AT(t).
The400 Jl
thick LC film was orientedperpendicularly
to theglass plates.
In theexperiment,
we first
applied,
in zerofield,
agiven
power to pro- duce a temperature differencefllll
of the order of onedegree.
The nanovoltmeter is zeroed so that this AT is taken as the base line. Additionalchanges,
ôT = OT -
AT,,,
are read with alarger sensitivity.
At the initial time t = 0
(point A),
a dcmagnetic
field is
applied perpendicularly
to the LC axis andAT is found to increase. At
points B,
the field issuddenly
decreasedby steps. Finally,
for zerofield,
AT returns to its
original value,
withusually only
small drift effects. The
original
value isalready
reco-vered as soon as the field is below a certain threshold which
corresponds
to the Freedericks critical field.Above this
limit,
thesign
of thetemperature change
in
field,
ô T >0, corresponds
to a decrease of thethermal
conductivity.
The time constant, which will be studied in partII,
arelarge compared
to the thermal ones,especially
in thick films. Astudy
of thedynamics
of the transition is essential to get theequilibrium
valueô T(H),
whenthe temperaturechanges
are small
(H N jHc).
a)
Criticalfields.
- In lowfield,
H ;:HC3’
we findthat ô T varies
linearly
with H, inagreement
with III. A.We get
Hr ,,,
with a5 %
accuracy from a linear extra-polation
of ôT to zero. Infigure 8,
wegive
the varia-tion of
IIH,,
versus d for three film thicknesses. FromFIG. 8. - The critical field varies as the inverse of the thickness,
in the two geometries studied. The slopes can be used to get values of Kl and K33.
the
slope
of the linearvariation,
we can deduce a value of the bend elastic constantK33
around 260 :b)
Thermalequilibrium
behaviorfor
H >Hr.
-Our results for
samples
of different thicknesses aregiven
in the normalized units :FIG. 9. - The relative increase of the temperature drop across homeotropic L C films of different thicknesses is plotted as a
function of the normalized field.
They
should define aunique
curve, function of thesplay
elastic constantKl i
as well as onK33.
In par-ticular,
when h is close to1,
the initialslope
is obtainedfrom :
This
equation
can be deducedeasily
from eq. 111.8 for theplanar
case.- The observed
shape
of the curve obtained from the different datapoints
offigure
9 is similar tothat of the limit calculation
leading
tofigure
6.However,
there is a ratherlarge
andsystematic smearing
of the resultsleading
to alarger anisotropy
for thinner films. This was observed
reproducibly
fordifferent
samples
of the same thickness and wasfound to be
relatively
insensitive to the thermalgradient
across thesample. (In fact, H,,,
andK11
should
only
varyslowly
with temperature close toTc [6]).
- Close to
Hc,
the extremeslopes
1 and 2give
avalue of the ratio
K331K, 1 :
- In
large
fields(h » 1), ke
is close fromkl.
Only
the molecules close to theglass surface,
withina coherence
distance (
ocIIH,
contribute tokll.
This
simplified analysis
as well as the use of thelimiting expression
111.9predict
anasymptotic
behaviorke(h)
= 1 -alh.
This form contradicts theprediction ke(h)
= 1 -P/h2
based on the « swarm » model inthe thermal
experiments
of reference[18].
The presentexperiments
do not extend to fieldslarge enough
to separate between these two forms. In
independent optical experiments
in muchlarger
fields(up
toh =
20) [15],
we have shown that theoptical
bire-fringence
does varylinearly
withIlh.
-
Finally,
our results indicate a value of theanisotropy
of the heatconductivity :
between
2)
Planar case. - Theexperimental
results aresimilar in this case
(geometry
1 ofFig. 1)
to thosein the
homeotropic
geometry. When a field h > 1 isapplied,
the temperature difference is found to decrease. This is consistent with the resultka
> 0already
obtained.a)
Criticalfields.
- From the linear relationbetween
Hc!
and1/d (Fig. 8),
one gets :b)
The normalized thermal results aregiven
onfigure
10 for different film thicknesses.The extreme values of the initial
slope give
a valueof the
anisotropy
of the elastic constants :From the behavior in
large fields,
we obtain a newFIG. 10. - The relative decrease of the temperature drop
across planar L C films in field (same situation as Fig. 7).
value of the heat
conductivity anisotropy (between
h = 0 and
3) :
C)
DISCUSSION. - Acomparison
between theplanar
andhomeotropic
resultsgive
correlative infor- mations on theanisotropy
of the heatconductivity
and on the elastic constants
Kl 1
andK33-
1) Anisotropy of
the thermalconductivity.
- Themagnetic
measurements gave two values for thepartial anisotropy
between h = 0 and h = 3 :The direct measurements of II gave a value of the total
anisotropy k.1k,
= 0.64. If oneestimates,
from the calculated results offigure 6,
that thepartial anisotropy
for h = 3 is 15 to 25%
smaller than the saturated one, we can see that the agreement between these results is excellent and that the last value canbe retained with an accuracy of + 5
%.
This consis- tency is agood
confirmation of thereliability
of ourmeasurements.
Although
there has not been anypreviously publish-
ed thermal
conductivity
data on MBBA some resultshave been
given
in other nematic LC. Fisher and Frederickson[12]
have measured an increaseof ke
across a
sample
of PAA atright angle
with a shearflow.
They implied
anegative anisotropy.
A valueof the same
sign
was found from heatdiffusivity
measurements in PAA oriented in an electric field
[19].
This
sign
ofka
is not consistent with recent measure- ments on PAAaligned
with amagnetic
field[13].
Recent heat
diffusivity
measurements in another LC- DBA - would also
give
apositive sign of ka [18].
However,
thealignment
close to the boundarieswas never defined or controlled in any of these expe- riments. In this
respect,
ourexperiments provide
thefirst
unambiguous investigation
of the heat conduc-tivity anisotropy.
Theexperiments
of references[18]
and
[19]
wereinterpreted
in terms of afield-independent,
oriented
boundary layer having
a heatconductivity
688
different from the bulk. We have found no evidence of such a
layer
in our results.2)
Elastic constants. - Our determinations of the elastic constants fromHc
and from the behavior aboveHc
are rather scattered. Beforeanalysing
these
results,
let us consider thereported
values ofelastic constant
K11
andK33
for MBBA. Williamsand Cladis
[6]
obtainedK331X.
= 6.8 ± 0.7 from theconoscopic
measurement ofHC3.
Robert and Labru-nie
[20]
measuredK331X,, =
7.3 ± 0.4 x 10-’ andKll/K33 ’"
0.7 frombirefringence
measurements inhomeotropic
MBBA distortedby
an electric field.Haller
[21]
indicated a valueof K33 -
5.1 x 10-’at 220 and
K, 1 IK3 3 =
0.9. Rondelez and Hulin[7]
obtained K3 3 = 7.4 + 1 X 10-7 and K, 1 = 5.3 + 0.5 x 10-7
from their dielectric measurements. The
published
results
already
indicate afairly large scattering.
Inaddition, they
were obtained on films thinner thanours. In order to
study
apossible
thicknesseffect,
we have
developped,
to alarger
accuracy, the cono-scopic technique
usedby
Williams andCladis,
and obtainedoptically,
the thickness and the value ofHC3
for
homeotropic
films of various thicknesses. Fora
175 g
thickfilm,
we have obtainedK33/Xa
= 7.7:t 0.5and for two
350 Jl
thick onesK331X.
= 7 ± 0.5 and8.3 + 0.5.
If we come back to our
results,
we see that the values ofK33
andK11
deduced from the criticalfield measured
thermally
are too small. This disa- greement should not be taken tooseriously
as thethermal effect is
only
an indication ofH,,
and is an order ofmagnitude
less accuratethan,
forexample,
the
optical
measurements we have used[15].
Apossible
additional cause of
disagreement
would be the effectof the temperature
gradient
across thesample, although
we have found no
appreciable change
in theHc
value whenchanging
the power across thesample by
a factor of three.If we consider the results on the thinner films
(d
= 300Il),
the different determinations of the ratioK11IK33
are found to becompatible
andgive
a valueof
Kl lI K33 ^-’
0.7 ± 0.15.Remark. - In a
preliminary report
we had obtained valuesK11/Xa
= 1.4 + 0.3 andKll/K33
= 0.25 + 0.05in
striking disagreement
with these latest results. It may be instructive toanalyse
the causes. Apossible
reason is related to the poor
quality
of the material used(Tc = 36 OC).
Williams and Cladis[6]
foundthat, by
the addition of a surfactant agent,K33
decreas-ed
by 14 %
per percent ofimpurity.
Another reasonwas the poor
alignment
of the LC(see
the discussion in partII).
In the presence of a finite
tilt,
9, of the director at theglass surface,
the critical field vanishes.However,
the distortionchanges rapidly
around a value offield which decreases very
rapidly
as lfJ becomeslarger. Typically
for qJ =10°,
the apparent critical field would decreaseby
30% (and
theK33
determi-nation
by
60%).
IV. Conclusion. - Our results
provide
the firstcomplete
andunambiguous study
of theanisotropy
of the thermal
conductivity
of a nematic LC.Yun and Frederickson
[29]
hasreported
someeffects of heat
generation
in LCsubjected
tomagnetic
fields.
However,
aspointed
outby
Kessler andLong- ley-Cook [13],
thepersistent changes
in temperatureare more
probably
due to thechange
of the thermalimpedance
between the LC and theexterior,
due to theanisotropy
of k in a field. Our results support this lastexplanation.
It would be of interest tostudy
now the
coupling
effects between thermalgradients
and other transport
properties,
via thisanisotropy
term in the nematic state.
The static
study
of the Freedericks transition has confirmed the results of the Frank-Oseendescription.
However the thermal method is too insensitive to get
accurate determination of the elastic constants.
Optical experiments
appear to be moreappropriate
but a very fine control of the surface conditions is
required.
Acknowledgments.
- We aregrateful
to P. G. deGennes for several
illuminating
discussionsalong
this work. We thank Mrs. Williams and Mrs. Cladis for
helpful
critics based on theiroptical study
of the Freedericks transition. We have benefited from theexperience
of theGroupe
d’Etude des CristauxLiquides d’Orsay
and from a critical review of this workby
C. Mitescu.One of us
(E. G.)
wants toacknowledge
thePhysics Department
of theUniversity
ofCalifornia,
LosAngeles,
forwelcoming
himduring
a time where part of the work was done. He hadstimulating
dis-cussions with F. J. Kahn and L. B. Kovalenko.
Note added in
proof :
M.Longley-Cook (Thesis, University of Arizona, 1971)
has doneindependently
similar measurements on PAA. His results are dis- cussed in terms of our initial report
[11]
and agree with thosepresented
here.References
[1] CHATELAIN (P.), Bull. Soc. Franç. Mineral, 1943, 66,
105.
[2] FREEDERICKSZ (V.), ZOLINA (V.), Trans. Faraday Soc., 1933, 29, 919.
[3] ZOCHER (H.), Trans. Faraday Soc., 1933, 29, 945.
[4] FRANK (F. C.), Disc. Faraday Soc., 1958, 25, 1.
OSEEN (C. W.), Trans. Faraday Soc., 1933, 29, 883.
[5] RAPINI (A.), PAPOULAR (M.), PINCUS
(P.), C.
R. Acad.Sci., Paris, 1968, 267, 120B.
RAPINI (A.), PAPOULAR (M.), J. Physique, 1969, C 430.
[6] WILLIAMS (C.), CLADIS (P. E.), to be published in Solid
State Communications, 1972, 10, 357.