HAL Id: jpa-00208812
https://hal.archives-ouvertes.fr/jpa-00208812
Submitted on 1 Jan 1978
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.
On the difference between dielectric and piezoelectric relaxation
K. Nishinari, S. Koide
To cite this version:
K. Nishinari, S. Koide. On the difference between dielectric and piezoelectric relaxation. Journal de
Physique, 1978, 39 (7), pp.771-775. �10.1051/jphys:01978003907077100�. �jpa-00208812�
ON THE DIFFERENCE BETWEEN DIELECTRIC
AND PIEZOELECTRIC RELAXATION
K. NISHINARI
National Food Research
Institute, Ministry
ofAgriculture
andForestry, Shiohama, Koto-ku, Tokyo, Japan 135
and S. KOIDE
Institute of
Physics, College
of GeneralEducation, University
ofTokyo, Komaba, Meguro-ku, Tokyo, Japan
153(Reçu
le 19 décembre1977,
révisé le 21février 1978, accepté
le 9 mars1978 )
Resume. 2014 Les relaxations dielectrique et
piezoelectrique
sont traitées sur la base du modèle à deux états séparés par une barrière de potentiel pour clarifier leur différence. On suppose que le mécanisme moléculaire responsable du phénomène est la réorientation des molécules ayant des dipôles électriques permanents. On suggère que la réponse instantanée au stimulus est essentiellementimportante dans le cas
piézoélectrique.
La dépendance en fréquence et en température est discutéepour un film de polymère. On peut caractériser les relaxations au moyen de ce modèle simple, en supposant un changement approprié de
potentiel
induit par l’application (ou la coupure) du champ électrique ou de la déformation externe.Abstract. 2014 Dielectric and
piezoelectric
relaxations are treated in the same way by the two-site model in order to clarify the difference. The molecular mechanism for these phenomena is assumedto be the reorientation of molecules
having
permanent dipole moments. It is suggested that theinstantaneous response
(polarization)
to the stimulus is essentiallyimportant
in thepiezoelectric
case. The
frequency
and temperaturedependence
is discussed for a polymer film. It is shown that many characteristic features of the relaxationphenomena
can be well explainedqualitatively
by this simple model if one assumes an appropriate potential change for thedipole
caused by theapplication
(or removal) of the external electric field or strain.Classification Physics Abstracts 36.20 - 77.40 - 77.60
1. Introduction. - In recent years there has been a
growing
interest inpiezoelectricity
ofpolymers, especially
in its relaxational behaviour[l, 2]. Although phenomenological
theories such as a mechanicaland
electrical model treatment[3]
and atwo-phase
treat-ment
[4]
have beenproposed,
no moleculartheory
forthe relaxation
phenomenon
has beenproposed
sofar. As concerns dielectric relaxation, on the other
hand,
manyinvestigations
have been made on amolecular basis
[5-8]
and the methods areapplicable
to
piezoelectric
relaxation with minor alterations.In
solids,
thefollowing
two cases are considered to beimportant
as mechanisms in which thedisplacement
of electric
charges
contributes to the mean electric moment[9] : (i)
When the electriccharge
is constrained to theequilibrium position by
an elasticrestoring
force as in ionic
crystals,
its behaviour isgoverned mainly by
therestoring
force and electric field. Born[10]
calculated thepiezoelectric
constant in this case.This case
gives
a resonance-typeabsorption. (ii)
Inmany
polymeric
solids,on , the
contrary, there are severalequilibrium positions
for electriccharges
andeach
equilibrium position
isoccupied
with a certainstatistical
probability
whichdepends
on the localfield
acting
on thecharges.
Thesimplest
model of thispicture
was introduced firstby Debye [11]
for dielectric relaxation. He treated the transition between theseequilibrium positions,
which leads to amacroscopic change
in dielectricpolarization by
apurely
stochasticprocess.
In the present paper, we willadopt
the latterpoint
of view to handle thepiezoelectric
and dielectricrelaxation on a common basis.
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01978003907077100
772
Treating
a distorted helicalpolymer
chain with effectivecharges
ordipole
moments on the main chain,Kasai
[121
has shown that thepiezoelectric
effectsoccur
only
in anoptical
shear mode of lattice vibra- tionalong
the helical axis. But the relaxationalphenomenon
is not in the scope of his work.We confine our attention to the
piezoelectric
directeffect
only
and exclude the inverse effect, i.e. theproduction
of a mechanical response due to an electric stimulus[13]. Among
the fourpiezoelectric
constants,the
piezoelectric
stress constant e(i.e.
the constantfor electric response to an external
strain)
will betreated in the
following together
with the dielectric constant a, because motion of thedipoles
willdepend
on strain
directly
rather than on stress. We assumethat the
microscopic
entitiesresponsible
for therelaxation in
question
are permanentdipoles
asso-ciated with molecular chains and that
they
aremaking strongly
hindered rotations or rather vibrations around several favourable directionsaccompanied by
occasional transitions from one to the other. Such a
situation will be most
suitably
describedby
the so-called site model.
According
to linear responsetheory, the frequency dependent f,*(w)
= F’ - iF" ande*(w)
= e’ - ie" can both be derived from thestochastic behaviour of the
dipoles (reorientation)
after
switching
off the constant external electricfield or mechanical strain. We can, therefore,
develop
our molecular
theory
of F* and e* in a unified scheme,and, by doing
so,clarify
the difference between these twoquantities.
2. Site model. - Dielectric and
piezoelectric
dis-persions
are both thefrequency dependence
of thepolarization
M as a response to an external force F, i.e. electric field E, and, mechanical strain Srespecti- vely.
Since the response M is common to bothphe-
nomena, the difference must be ascribed to the nature of the stimulus. It must be remembered that we are
considering
thepiezoelectric
stress constante*,
i.e. aresponse to strain rather than to stress. The
dipole
is forced to follow the
change
in strainpromptly
because it causes some
change
in the favourable directions ofdipoles.
In the dielectric case, this effect may benegligible
and the instantaneous response isessentially
due to thechange
in the electronic motions, which are not considered in thepresent
paper. If the response iscompletely instantaneous,
there is nopossibility
of relaxation. Thedipole
reorientation with finitevelocity
in a solid isappropriately
describedby
the site model. Wewill, therefore,
treat thesedispersions by
thesimplest
sitemodel,
i.e, two-siteor bistable
model,
in order to see the difference between the dielectric andpiezoelectric dispersions.
We will first consider the case where there is no instan- taneous response.
We will assume that 0
symbolically
stands for theangular
variablesdescribing
the motion of thedipole.
In the two-site
model,
there are twopotential
energyminima
V(Ol)
andV(02)
for thedipole
rotation asshown in
figure
1. Forsimplicity,
we will confineourselves to the
symmetric
two-site model in which U = 0 under no externalfield,
since the extension to theasymmetric
case is easy. The site model isappli-
cable to the case where the
potential
barrier between the minima is muchhigher
than kT and most of thedipoles
possess insufficient energy to rotateby getting
over the barriers.
FIG. 1. - Potential energy curve for the dipole orientation. The
angular variable 6 represents symbolically the direction of the dipole
orientation in the three dimensional space.
Imagine
now that the external force, which isapplied
at a constant value(unity)
for along
time, issuddenly
reduced to zero t = 0. Thebehaviour
of the system is describedby
the staticsusceptibility
fort 0 and
approximately by
a relaxation time for t > 0. As is well known, the staticsusceptibility
a, isproportional
to1 IkT
for the site model. The relaxa- tion time i isequal
to(k12
+k21)-I.,
wherekl2
andk21
are the transitionprobabilities
for the rotationof a
dipole
from site 1 to site 2 and vice versa. Theprobabilities w;(t) (i
= 1,2)
that thedipole
stays in site i at time t > 0 areexpressed
asFor a small force,
w;(0) - n,,(oo.)
isproportional
tothe static
susceptibility
a,. Let thedipole
moment of adipole
at site i be denotedby
pi. The wholepolari-
zation
is
given by
use ofequation (1).
Then, as the response for a sinusoidal forceF(t)
=Fo exp(iwt),
thecomplex
electric
susceptibility x*( w)
orcomplex piezoelectric
constant
e*(m)
isgiven by
The
complex
dielectric constant F* is’related tox*
asF* = 1 + 4
Jtx*.
As all the factors except(t
+icor)
in
equation (2)
arew-independent,
théfrequency
dependence
of such f,* and e* becomes the well-knownDebye
typesingle
relaxation as shown for the dielectricdispersion
in many textbooks[5,
6,9].
Let us now consider the instantaneous response in the
piezoelectric
case. The sudden removal of strain will beexpressed by
a suddenchange
in thepotential V(O) itself,
and hence there can be no retardation asmight
beexpected
in the relation betweenV(O)
and anexternal mechanical stress. This
change
inV(0)
will beclassified into two
categories.
One is the shift of thepotential minima 01
and02,
which will be illustratedby
some horizontal movement of thepotential
curvein
figure
2. The moleculardipole
will then be forcedto follow this shift
promptly
because the shiftchanges
the form of the container in which it is confined. This prompt rotation of
dipoles
will result in an instan- taneouschange
in the bulkpolarization.
At the same
time,
the removal of strain willgenerally give
rise to some verticalchange
in thepotential
curve,i.e.
change
in the values of its minima. Thisyields changes
in the Boltzmann factors associated with theseminima,
andbegins
the statistical redistribution(i.e. relaxation)
to theequilibrium corresponding
tothe minimum values for F = 0. This relaxation will be described
by
a finite relaxation time i in the two-site model.
In the dielectric case, the latter effect is
predominant
because the
potential change
isgiven by p.E
and isexpressed by
a verticalchange. Asi mentioned
above,electronic motions are not considered in the present paper. What we want to
emphasize
in the present work is theequal importance
of the above two effectson the rotational motion of the permanent
dipoles
for
piezoelectric
relaxation.Since there is no
general
relation between the direc- tions of thepolarization
and of the strain in thepiezoelectric
case, the cases(b), (c)
and(d)
shown inFIG. 2. - a) 1) Potential curve under the constant external force.
2) Potential curve immediately after the removal of the external force. 3) Potential curve in equilibrium in the absence of the external force. Dots represent the population of the dipoles. b) The change
in the dipole moment due to the removal of the éxternal force.
Dipole moments correspond to the case 1), 2) and 3) respectively.
Sgure 3
may takeplace depending
on the directionof the shift of the
potential
minima. The upperfigures
show the response to the step stimulus :
F(t)
= 1 fort 0 and
F(t)
= 0 for t > 0. The value al represents the static response, while a2 represents the instanta-neous
change following
the sudden removal of the external force F.Corresponding
to these types of response(a) - (d),
we get, as the Fourier transforms of theirnegative
time derivative(i.e.
response to a unitFIG. 3. - The upper figures show the response to the step stimulus : F(t) = 1 for t 0 and F(t) = 0 for t > 0. The value al stands for the static response, while a2 expresses the instantaneous part. The sign of the strain is chosen so that al > 0. a) a2 = 0, b) a2 > 0, c) a2 0, 1 (X2 [ al, d) a2 0, I a2 1 > oc,. The lower figures show the response to the sinusoidally varying stimulus. Dotted and solid curves
represent the real and imaginary part of x* and e* respectively.
774
pulse),
thefrequency dependent
response(a’) "" (d’)
to the
sinusoidally varying
stimulus.Now let us consider the temperature
dependence.
From viscoelastic and dielectric
investigations [14],
it is well known that the
time-temperature
relationin
polymers
is describedby
the W.L.F.equation
rather than
by
the Arrheniusequation.
Since, however, the use of the W.L.F.equation neglecting
the distri-bution of relaxation times does not
improve
thesituation,
we will assume the Arrheniusequation
asthe
simplest expression.
In this case, the relaxation time i is written aswhere the
frequency
factor A may or may notdepend
on temperature
[6].
Some authors(e.g.
Kauzmann(1 5]) apply
the ratetheory
ofEyring [16],
whichyields
A rkT/h
in thesimplest
case, and assumes a lineardependence
on the temperature. Hoffman and Pfeif- fer[17]
assume it to be constant, without any dis-cussion,
while Lauritzen[18] simply
states that it isonly slightly dependent
upon temperature. Aftera rather detailed consideration of the motion of a
dipole
rotator, Bauer[19]
obtained A oc T1/2. All that we can say at present is, therefore, that it is someslowly increasing
function of temperature T. Then, in the case offigure
3a’, the temperaturedependence
of the real and the
imaginary
parts ofx*
and e* aredescribed
by
These relations may be
expressed graphically.
Indrawing
such curves, however, the temperaturedepen-
dence of A does not
yield
anappreciable
effects,because the
exponential
factorexp(W/kT)
dominatesthe temperature
dependence overwhelmingly
as in thecase of Richardson’s formula for the thermionic emission. The effect of the instantaneous response
simply
adds a constant a2 to the real partsx’
and e’.3.
Application
to thepiezoelectric
relaxation inpolymer
films. - When a sinusoidal shear strain withan
angular frequency
isapplied
touniaxially
drawnfilms as shown in
figure
4, thepolarization
of the samefrequency
is observed on the film surface. In- such adynamic
case, thepiezoelectric
stress constant isusually
denoted ase*(o»
= e’ - ie". In the caseshown in
figure
4, thepolymer
film has the symmetryD oo( 00 22)
and theonly non-vanishing
component of thepiezoelectric
tensor ise* - - e* -
ep].
Recently Hayakawa
et al.[20]
and Furukawa and Fukada[211 ]
have made measurements of thepiezo- electricity
ofpoly-y-methyl-L-glutamate (PMLG)
indetail.
Hayakawa
and Wada[4, 22]
suggest that theFIG. 4. - A film sample and rectangular coordinates. The film is drawn along the z-axis and the strain is applied along the Z-axis.
E : electrode, -P : the angle between the z-axis and the Z-axis.
piezoelectric
temperaturedispersion
near room tempe-rature is attributed to the
quasi-crystallite
which has arelaxing piezoelectricity.
Auniaxially
drawn film of PMLG is considered to be an orientedone-phase
system. Further, a conformationalanalysis [23]
seemsto show that U = 0 in PMLG. We will,
therefore,
tryto
apply
ourtheory
to PMLG.The
frequency dependence
of F’ and F" observedby
many authors
[21,
23,24]
seems to beroughly
of theDebye
type.Strictly speaking,
of course, it should be différent from the pureDebye
type because of the distribution of the relaxation times. Thefrequency dependence
of’ e’ and e" observedby Hayakawa et
al.[20]
is shown infigure
5. Theimaginary part e"
shows aFIG. 5. - Frequency dependence of the piezoelectric constants e’
and e" of PMLG at 12 °C observed by Hayakawa et al. [20].
behaviour
quite
similar to that of f,"., while the curveof e’ seems to be obtained
by shifting
that of e,’ down- wardsalong
the ordinate. Thiscorresponds
to case(d’)
in
figure
3. As stated in section 2, thepositions
of thepotential
minima can shift in thepiezoelectric
casebecause the
dipole
is forced to follow thechange
ofthe strain
instantaneously.
Thus thefrequency depen-
dence of the dielectric and
piezoelectric
constants inPMLG can be well
explained qualitatively by
the two-site model.
The temperature
dependence
of(e’, e")
and(B’., a")
observedby
Furukawa and Fukada[2l],
and thecurves calculated
by
means ofequation (3) adopting
the value W = 23
kcal/mol [24, 25]
andassuming
A = 8.0 x 1019 are shown
together
infigure
6.FIG. 6. - Temperature dependence of dielectric and piezoelectric
constants of PMLG. Solid curve : observed by Furukawa and Fukada [21] at 10 Hz. Dotted curve : calculated by eq. (3). The calculated curve of e’ in this figure is shifted downwards by
1.8 x 101.
Again,
while the temperaturedependence
of a" and e" shows almost the sametendency
in theexperiment,
the curve of e’ seems to be obtained
by shifting
thatof e,’ downwards
along
the ordinate. Since the instan- taneouschange
in theposition
of thepotential
minimain the
piezoelectric
case isexpected
to takeplace
tothe same extent at any temperature, this result is
quite
reasonable. The calculated curves are too steepby comparison
with theexperimental
ones. This may be attributed to oursimplified assumption
of asingle
relaxation time.
Distribution of the relaxation times can be intro-
duced,
forexample, by
the multi-site model treatment.The two-site model used so far can be
easily
extendedto the multi-site model as in the case of dielectric relaxation
[7,
17,18].
In the multi-site model, thedecay
ofpolarization
after the removal at time t = 0of the external stimulus is
given by
where in is the relaxation time of the n-th mode of inter-site transitions. The
coefficient cn
are determinedby
the initial condition at t = 0, whichdepends
onthe nature of the stimulus under consideration. Thus the time
dependence
ofM(t)
in thepiezoelectric
casecan be different from that in the dielectric case. Note
that,
apart from the constant factor and the additive constant in the real part, there can be no such diffe-rence in the two-site model.
Acknowledgment.
- The authors wish to thank Dr. E.Fukada,
Dr. M. Date, Dr. T. Furukawa at the Institute ofPhysical
and Chemical Research, andDr. R.
Hayakawa
at theUniversity
ofTokyo
for theirvaluable advice.
References
[1] FUKADA, E., Adv. Biophys. 6 (1974) 121.
[2] WADA, Y. and HAYAKAWA, R., Jpn. J. Appl. Phys. 15 (1976)
2041.
[3] FUKADA, E. and DATE, M., Polymer J. 1 (1970) 410.
[4] HAYAKAWA, R., KUSUHARA, J. and WADA, Y., J. Macromol.
Sci. Phys. B 8
(1973)
483.[5] HILL, N. E., VAUGHAN, W. E., PRICE, A. H. and DAVIES, M., Dielectric Properties and Molecular Behaviour (Van Nos- trand Reinhold Co., London) 1969.
[6] DANIEL, V. V., Dielectric Relaxation (Academic Press, London and New York) 1967.
[7] WYLLIE, G., Dielectric and Related Molecular Processes, I (Davies, M. ed., The Chemical Society, Burlington House, London) 1972.
[8] WILLIAMS, G. and WATTS, D. C., Dielectric Properties of Polymers (Karasz, F. E. ed., Plenum Press, New York
and London) 1972.
[9] FROHLICH, H., Theory of Dielectrics (Oxford Univ. Press, London) 1949.
[10].BORN, M. and HUANG, K., Dynamical Theory of Crystal
Lattices (Oxford Univ. Press, London) 1948.
[11] DEBYE, P., Polar Molecules (Reinhold Publ. Corp., New York)
1929.
[12] KASAI, K., J. Phys. Soc. Jpn. 27 (1969) 1268.
[13] CADY, W. G., Piezoelectricity (Dover Publ. Inc., New York)
1964.
[14] MCCRUM, N. G., READ, B. E. and WILLIAMS, G., Anelastic and Dielectric Effects in Polymeric Solids (Wiley, London) 1967.
[15] KAUZMANN, W., Rev. Mod. Phys. 14 (1942) 12.
[16] GLASSTONE, S., LAIDLER, K. J. and EYRING, H., The Theory of
Rate Processes (McGraw-Hill, New York and London) 1941.
[17] HOFFMAN, J. D. and PFEIFFER, H. G., J. Chem. Phys. 22 (1954)
132.
[18] LAURITZEN, J. I., J. Chem. Phys. 28 (1958) 118.
[19] BAUER, E., Cah. Phys. 20 (1944) 1.
[20] HAYAKAWA, R., NAMIKI, K., SAKURAI, T. and WADA, Y., Rep. Prog. Polymer Phys. Jpn 19 (1976) 317.
[21] FURUKAWA, T. and FUKADA, E., To be published.
[22] HAYAKAWA, R. and WADA, Y., Adv. Polymer Sci. 11 (1973) 1.
[23] TANAKA, A. and ISHIDA, Y., J. Polymer Sci. Polymer Phys. Ed.
11 ( 1973) 1117.
[24] MAKINO, S., KAMASHIMA, K., KUBOTA, S. and SUGAI, S., Jpn.
J. Appl. Phys. 3 (1964) 55.
[25] WATANABE, J. and UEMATSU, I., Polymer J. 9 (1977) 195.