Inr. J. Engng Sci. Vol. 9, pp. 587-604.

Pergamon Press 197 I.

Printed in Great Britain

ON THE NONLINEAR
EQUATIONS
ELE~TROELASTI~ITY

OF THERMO-

H. F. TIERSTEN
Rensselaer Polytechnic Institute, Troy, N.Y. 12181
Abstract-The
di~erential equations and boundary conditions describing the behavior of an electricaIly
polarizable, finitely deformable, heat conducting continuum in interaction with the electric field are derived by
means of a systematic application of the laws of continuum physics to a macroscopic model consisting of an
electronic charge continuum coupled to a lattice continuum. The resulting rotationally invariant description of
thermoeiectroelasticity
consists of five differential equations in five dependent variables and, when thermal
cons~de~tions are omitted, reduces to four differential equations in four dependent variables. These same
four differential equations of electroelasticity in the same four dependent variables along with the associated
boundary conditions can readily be obtained, by means of a Legendre transformation, from existing consistent variational treatments of electroelasticity, which yield a system of seven equations in seven dependent
variables,
I. INTRODUCTION

nonlinear treatment of an electrically polarized elastic continuum

in interaction with the electrostatic field appeared in a monumental paper [ 1I by Toupin
in 1956. S~bseqnently several other workers 12-71 investigated the behavior of deformable continua in interaction with the electric and/or magnetic fields. One among these,
Eringen [2,7], treated electroelasticity.
Both Toupin [ 1I and Eringen [2,71 derive the
equations of static electroelasticity by means of variational principles. Neither one obtains the equations by systematically applying the basic conservation equations of continuum physics to a well-defined macroscopic model, although Toupin discusses such a
model qualitatively. In addition, in both papers [I, 21 the rotationally invariant electric
quantities, upon which the internal stored energy function depends, lead to a system of 7
nonlinear equations in 7 dependent variables. Such a system of equations is an unnecessarily complicated description of the situation because the electrostatic field is
expressible in terms of the electric scalar potential. In this paper a different thermodynamic function is employed, in place of the internal stored energy function, and taken
to depend on a set of rotationally invariant electric quantities fundamentally different
than the sets employed by either Toupin [ 11 or Eringen[2]. The equations of electroelasticity that result consist of a system of 4 nonlinear equations in 4 dependent
variables, in place of the aforementioned 7 equations in 7 dependent variables. Moreover, the field equations for the coupled system are derived by means of a systematic
application of the laws of continuum physics to a well-defined macroscopic model
consisting of an electronic-charge
continuum coupled to a lattice continuum. As a
consequence the nonlinear equations that result are those of the~oelectroelasticity~
which consist of 5 equations in 5 dependent variables. When thermal considerations
are eliminated, the equations reduce to the aforementioned 4 equations of electroelasticity. It should be noted that rotationally invariant electrical variables identicai with the
ones employed here were employed previously by Jordan and Eringen 181 in a different
context. In addition, it should be noted that a somewhat similar particle model was
employed by Dixon and Eringen[91, who used averaging techniques to obtain continuum equations in the electrodynamic case, and employed a different set of electrical
variables in the constitutive equations than the set used here. The variables we employ
THE FIRST consistent

H. F. TIERSTEN

588

result in a more convenient system of equations for the particular case treated in this
paper. In the derivation of the equations presented in this paper the notion of the
quasi-static[lO] electric field is used so that mechanical inertia can be included. The
resulting equations, which are Galilean invariant, encompass such phenomena as
dynamic piezoelectricity, thermoelasticity, and electrostriction, and are valid as long as
frequency-wavelength
combinations are far outside the range associated with electromagnetic propagation. The jump (boundary) conditions across moving non-material
surfaces of discontinuity are obtained from the integral forms of the field equations of
thermoelectroelasticity.
Finally, it should be noted that the variational principle leading
to the aforementioned
4 equations of electroelasticity
and the associated boundary
conditions at material surfaces of discontinuity can be obtained from the variational
principles of Toupin[ l] and Eringen [2] simply by including the mechanical kinetic
energy in the Lagrange densities and employing the temperature independent version
of the Legendre transformation appearing in equation (4.2) of this paper.
2. THE

INTERACTING

CONTINUA

As stated in the Introduction the macroscopic model, which we consider, essentially consists of an electronic-charge continuum coupled to an ordinary mechanical
(lattice) continuum. Initially, both continua occupy the same region of space and, hence,
have the same material coordinates. The deformation of the body is described by the
mapping [ 111
Yi=Yi(XL3

t)v

Y=Y(xt),

(2.1)

which is one-to-one and differentiable as often as required. In (2.1) yi denote the spatial
coordinates and XL the material coordinates. We consistently use the convention that
capital indices denote the Cartesian components of X and lower case indices the Cartesian components of y. In this paper both dyadic and Cartesian tensor notations are used
interchangeably. The one employed in a specific instance is determined by convenience
and common practice. A comma followed by an index denotes partial differentiation
with respect to a coordinate, i.e.,
XKj

Y~,L = aYi/aX,,

= dXKIdyj.

(2.2)

The lattice continuum is assumed to have a positive charge density p1 and the electronic
continuum to have a negative charge density I*.~,where initially
@1+&e=

0,

(2.3)

since the net charge at any material point vanishes in the reference position. In a (finite)
motion the electronic continuum is permitted to displace with respect to the lattice
continuum by an injkitesimal displacement field r) = I,J( y, t) , which, by virtue of the
charge densities of the two continua, accounts for the polarization. Although the two
continua can displace with respect to each other, elements of each with the same
material coordinates are constrained to have equal volumes at all times and, hence, r)
must satisfy [ 121
7)k,k

0.

(2.4)

On the nonlinear equations of thermoelectroelasticity

589

As a consequence of (2.3) and (2.4) the net charge density for regions with the same
material coordinates must vanish at all times, and we may write
P[YitxL~

t)T tl +Pe[(Yi(XL5

t) +Ti(Yj,

t)), t] =O,

(2.5)

where the same material - and not the same spatial-domain

is understood.
The electronic continuum interacts with the lattice continuum by means of a local
electric field EL, which causes a force per unit volume peEL to be exerted by the lattice
continuum on the charge continuum. Clearly then, since linear and angular momentum
must be conserved, an equal and opposite force -peEL = pEL, along the same line of
action, must be exerted by the charge continuum on the lattice continuum. The quasistatic [ lo] Maxwellian electric field, EM, exerts the usual electric field force on each of
the two continua. The electronic charge continuum experiences no other forces of
action or interaction. The lattice continuum interacts with neighboring elements of the
same continuum by means of the usual traction [ 131 force per unit area acting across the
surface of separation. Since the electronic mass is very small compared to that of nuclei
and we are confining ourselves to the quasi-static electric field, we may neglect the mass
and, hence, the inertia of the electronic continuum compared to that of the lattice continuum. Schematic diagrams showing the relative displacements of the two continua,
the forces in the electronic-charge continuum, the linear momentum, forces and couples
in the lattice continuum, are shown in Figs. I-3, respectively.
Before we proceed with our treatment of the model, it should be carefully noted that
there is not necessarily a direct one-to-one correspondence
between the elements of
our macroscopic model and the atomic structure. Indeed, for this quasi-static macroscopic treatment, the microscopic nature of the polarization-ionic
or electronic - does
not matter.

Fig. 1. Schematic diagram showing the relative displacement for the two continua.

590

H. F. TIERSTEN

Fig. 2. Schematic diagram showing the force vectors

acting in the electronic charge continuum.

Fig. 3. Schematic diagram showing the linear momentum and force and couple vectors acting in the
lattice continuum.
3. BASIC

EQUATIONS

Before we proceed with the determination

the EM shown in Fig. 2 is at (y + r)) , i.e.,
E=

of the equations, it should be noted that

E[(y+q),

t]

(3.1)

while the EMshown in Fig. 3 is at y, i.e.,

E = E(y, t)

(3.2)

c = r) x plEL,

(3.3)

and the couple c in Fig. 3 is given by

On the nonlinear equations of thermoelectroelasticity

591

where the EL is always at ( y + 11)) i.e.,

EL= W(y+r)),

t],

(3.4)

and,ofcourse,~l=E.L(y,t)
while@=pe[(y+r)),f].
The equations of the conservation of linear and angular momentum for the electronic
charge continuum are, respectively,
s @[E(y+r))+EL]dV=O,
j. [y;v]
I;

(3.5)

X@[E(y+r))+EL]dV=O.

(3.6)

The equations of the conservation of mass, linear momentum

for the lattice continuum are, respectively,

and angular momentum

-$ pdV=O,
fV

(3.7)

Jt(n)dS+Ip[E~(y)+EL]dV=~lpvdV,

yxt(n)dS+
I
S

(3.8)

[y~$(E~(y)+E~)f~x$E~]dV=$

y X pvdV,

(3.9)

where d/dt denotes the material derivative [ 141 and v = dyldt is the velocity and p the
density of the lattice continuum. Since q is an injkzitesimal displacement field, we expand E(y + q) in a Taylor series about y and retain the first term to obtain
E(y + r)) = EM(y) + r) * V,E,

(3.10)

where V, = eia/ayi and ei is a unit base vector in the ith Cartesian direction. Substituting
from (3.10) into (3.5) and (3.6) and neglecting terms containing products of Q and adding
(3.5) to (3.8) and (3.6) to (3.9), employing (2.5) and defining the dipole density per unit
volume [ 151 (or polarization) P by
(3.11)
we obtain, respectively,
ft(n)dS+fP-V,EdV=$(pvdV,
s
V
f
S

yXt(n)dS+

[yXP.V,E+PXEM]dV=-$

(3.12)
1

yxpvdv,

(3.13)

as the integral forms of the equations of the conservation of linear and angular momentum, respectively,
of the combined-lattice
plus electronic-continuum,
which are of
ultimate interest to us.

592

H. F. TIERSTEN

Equations (3.12), (3.13), and (3.7) are just the integral forms of the ordinary equations of balance for a mechanical continuum with the body force f and body couple c,
respectively, given by
f=P.V,E,

c=PxE.

(3.14)

Since we are considering an electrically polarized mechanical continuum, we could

have written (3.12), (3.13), and (3.7) down directly from our knowledge of mechanical
forces [ 161 and couples [ 171 in electrostatics. However, it is both instructive and enlightening to derive these equations from our macroscopic model of the matter and,
more importantly, as we shall see later on in this section, we need to consider the
macroscopic model in detail in order to determine the rate at which EMdoes work on the
matter and, hence, to write the equation of the conservation of energy.
Since we are concerned with insulators under circumstances such that the quasistatic electric field approximation is valid, we have the integral forms
ss n-DdS=O,

(3.15)

(3.16)
9c EM. dy = 0,
where S denotes an arbitrary closed surface and C an arbitrary closed circuit, both of
which are stationary with respect to an inertial reference frame, and the electric displacement D is given by
D=E+&P

(3.17)

in Gaussian units.
The application of (3.12) to an elementary tetrahedron in the usual manner [ 181
yields the definition of the ordinary mechanical stress tensor T:
t=n-7,

tj = Izi?iJ.

(3.18)

The integral forms of the equations of the conservation of linear and angular
momentum, which have been derived for the combined continuum, i.e., (3.12) and
(3.13), are meaningful only if EM is differentiable; and EM usually is not differentiable
across a surface of discontinuity. However, if we introduce the Maxwell electrostatic
stress tensor [ 193
Tg = & [DiEy -&EfEfMs,] 3

(3.19)

where
(3.20)

and substitute from (3.20) into (3.12) and (3.13), respectively,

theorem and employ (3.19) we obtain, respectively, the forms

apply the divergence

593

On the nonlinear equations of thermoelectroelasticity

n. (r+TE)

dS =z

pvdV,

(3.21)

yxpvdV.

(3.22)

I
s

yx [n* (r+TE)]

dS =-$

The integral forms in (3.21) and (3.22) are taken to be valid across moving surfaces of
discontinuity even when (3.12) and (3.13) are not meaningful and reduce to (3.12)
and (3.13), respectively, when P and EMare appropriately differentiable.
Since equations (3.7), (3.21), (3.22) and (3.15) hold for arbitrary regions and (3.16)
for arbitrary circuits, when the variables are appropriately differentiable, we obtain,
with the aid of the divergence theorem and (3.19) and (3.20), the differential equations
PiEfi = p 2

(3.23)

TG=i(EyPj-PiEjM)y

(3.24)

Tu,i

pJ = POT

(3.25)

Di,i = 0

(3.26)

Ei = -q,i

(3.27)

where rA is the antisymmetric portion of T, p. is the mass density in the reference

configuration, cpis the electric potential and
J = det yi,L.
Equations (3.23)-(3.28) along with (3.17) constitute
also have, from (3.5), the equation

(3.28)
our system of field equations. We

EL=-EM,

(3.29)

which is not needed since in this theory EL does not appear in the equations for the
combined continuum. As usual in any continuum theory, at this stage there are more
variables than equations-in
this instance 9 more-and
we need to obtain constitutive
equations in order to have a determinate theory. To this end we consider the conservation of energy.
The equation of the conservation of energy for the combined continuum may be
written in the form
v*v+pr)
V

dV=

I
s

(tev-n-q)

dS+

adV,

(3.30)

where jv * v is the kinetic energy per unit mass, E is the internal stored energy per unit
mass, t. v is the rate of working per unit area of the mechanical surface tractions, n * q
is the rate of efflux of thermal energy per unit area and (T is the rate of supply of energy
to the material from the quasi-static electric field. In order to obtain the expression for

594

H. F. TIERSTEN

o, we must return to our model consisting of the electronic and lattice continua, and
determine the rate at which the Maxwell electric field, EM, does work on the matter,
i.e., on the two continua. Clearly, the rate at which EM does work on the electronic
continuum is

*[v+ckWl,

dW/dt = [j.~E*(y+rl)]

(3.3 1)

and the rate at which EMdoes work on the lattice continuum is

dWl/dt = [tiE(y)]

. v.

(3.32)

Then the rate of supply of energy from the quasi-static electric field, EM, to the matter
must be given by

O=@$+F

[wlE(y)]

*v-t [/.~~E~(y+rl)]

. [v+#

(3.33)

Substituting from (3.10) into (3.33) neglecting terms containing products of r) and using
(2.5) we obtain
o = [ /.L~v. V,E] . v + EM . pe dr)/dt.

(3.34)

Now, we may write

E.L~
dr)ldt = d($a)ldt

- r) d$ldt,

which, by virtue of (3.1 l), may be written

/.L~dr//dt = dP/dt - (P//J.) (d@/dt) .

(3.35)

If now we define the polarization per unit mass, v, by

77 =

and recall from the conservation

that

P/P,

(3.36)

of mass and charge, i.e., from (2.4), (2.5), and (3.7),

1 dp
--=-_

p.dt

1 d/Y
pe dt

(3.37)

we find, from (3.35H3.37)

/.L~dqldt = p drrldt.

(3.38)

Substituting from (3.11) and (3.38) into (3.34) we obtain

(T= (PaV,E)

.v+E.pd7r/dt.

(3.39)

On the nonlinear equations of thermoelectroelasticity

595

Taking the material time derivative of the 1.h.s. of (3.30), using (3.7), substituting from
(3.18) and (3.39) into (3.30), applying the divergence theorem and employing (3.23) and
the arbitrariness of V, we obtain
dni
de
p - = 7uUj.i+ Eyp dt - qi,i 3

(3.40)

dt

which is the first law of thermodynamics for the combined continuum.

Since we are considering a heat conducting electroelastic continuum with a first
law of thermodynamics of the form shown in (3.40) the mathematical expression of the
second law of thermodynamics may be written in the form [20-221
de
dri
dq
P z - 7ijv,,t.-Ez?p - dt = PO--,
dt

(3.41)

where 13is the positive absolute temperature and 77is the entropy per unit mass. From
(3.40) and (3.41), we have the dissipation equation
(3.42)

- qi,i = p8 dq/dt ;
and the entropy inequality may be written in the form

dq
dt+

0qi

=--_=

S.i

Sie,i
py

(3.43)

0,

82

where y is the (positive) rate of entropy production.

At this point it should be noted that this thermoelectroelastic
theory can readily be
gener&ed[23,241
to account for arbitrary functional (viscous) constitutive response
by taking the entropy inequality, (3.43) to be valid even when (3.41) and (3.42) are not
valid. Under such general circumstances, equations (3.41) and (3.42) are not valid
because the variables such as, e.g., E and r) are no longer state functions. Under such
general circumstances the entropy inequality - called the Clausius-Duhem
inequality takes the form
dq

p de

pdr-;z+B+

TijVj

Eyp dni

Bdt-T=

Qi0,i

PY 2 0.

(3.44)

The general functional constitutive equations can then be determined by a systematic

application of the principle of material objectivity[25]
and the Clausius-Duhem
inequality, (3.44), which is taken to be valid for all possible processes, in the manner of
Coleman and No11[23,24]. However, when the thermodynamic treatment is of such a
general nature, external supply (source) terms are (thought to be) required in the
appropriate equations, which in this instance consist of (3.12), (3.1.5), and (3.30),
and in all equations derived from them. But, since we are not interested in this level of
generality, we may determine the heat conducting constitutive equation from the
entropy inequality[261, (3.43), and all the other constitutive equations from (3.41) as
we do in the next section.

596

H. F. TIERSTEN
4. CONSTITUTIVE

EQUATIONS

As we have already stated, in this section we determine the heat conducting constitutive equation from (3.43) and the remaining constitutive equations from (3.41),
which, by virtue of the relation
vj,i = X,,i d (Yj,M)ldt,
may be written in the form
de
d
dni
P & = riJ&f,i z (Yj,M) + EiMp dr+

dq

Pe 5

(4.1)

Since the electric field, EM, is expressible in terms of the electric scalar potential, cp, by
(3.27) and the entropy inequality is of the form shown in (3.43) it turns out to be
particularly convenient to define the thermodynamic function x by the Legendre transformation
x = e-@-

E$ri.

(4.2)

The substitution of the material time derivative of (4.2) into (4.1) yields

(4.3)
Motivated by (4.3), we assume
x = X(Yj,M;

~3;01,

(4.4)

whence
(4.5)
and substituting from (4.5) into (4.3), we obtain

TijXM,i

(4.6)

Since all the material time derivatives appearing in (4.6) are independent and (4.6) holds
for arbitrary d (yj,M)ldt, dE?ldt and dO/dt, we have
XM,iTij = P 8X/a (Yj,M 1 2

(4.7)

ri = - axlap,

(4.8)

q=-axlae.
Solving (4.7) for T, with the aid of the chain rule of differentiation,
Tii =

PYi,hdxIa(Yj,M) .

(4.9)
we find
(4.10)

On the nonlinear equations of thermoelectroelasticity

597

Clearly, x cannot be an arbitrary function of y k,L,Ef and 8 because, in order to satisfy

the principle of material objectivity[25,27],
E and, hence, x must be a scalar invariant
under rigid rotations of the deformed and polarized body, and any arbitrary function of
the 13 assumed variables (4 vectors and a scalar at the point yk) will not be so invariant.
However, owing to a theorem on invariant functions of several vectors due to Cauchy
[28], we know that the invariance of x in a rigid rotation is assured when x is an arbitrary function of, at most, the 9 scalar products
CLN

(4.11)

yk,Lyk.N,

(4.12)

WI, = Yk,&kM,
and the 4 determinants

of their components

taken 3 at a time, i.e.,

J = ~eijkELMh7Yi,LYj,,~Yk,,,I,
KN = Le..
E ,y. y. EM
2 uk LMh 1.1,f&f k

(4.13)
(4.14)

where eijk and ELM,,,represent the alternating tensor[29], and 8. We may eliminate the
4 determinants in (4.13) and (4.14) from our list of variables because they are expressible in terms of the 9 quantities in (4.11) and (4.12). Specifically, we have [30]
J* = det CLN,

(4.15)

with J > 0 and [31]

KL = JC,;: W,.

(4.16)

where C-l denotes the reciprocal of C. Since CLMdoes not vanish in the natural undeformed state, it is convenient to replace C L,Mby the material strain tensor ELlw,where
(4.17)

ELM=HCLM-~LM),

and does vanish in the undeformed

state. Thus, at this point we have

X=X(ELM,

WL,~')

(4.18)

in place of the form shown in (4.4), and in (4.18) x may be an arbitrary function of the
variables shown. From (4.8)-(4.12), (4.17) and (4.18), we obtain
(4.19)
ri

-yi,LaxIaW~,

v = -axlae,

(4.20)
(4.2 1)

in which we have introduced the convention ax/aE KL= axIdELK, and it is to be assumed that aERNIaENR= 0 in differentiating x. From (4.19), (4.20) and (3.36) note that

598

H. F. TIERSTEN

(4.22)
which is identical with (3.24), which was obtained from the conservation of angular
momentum. Thus, as originally observed by Toupin[l] using a variational principle
instead of a macroscopic model and different functions and variables than the ones
employed here, the antisymmetric portion of the mechanical stress tensor is derivable
from a thermodynamic state function and has just the value required by the conservation of angular momentum.
This brings us to a consideration of the heat-conducting constitutive equation, which
proceeds from the entropy inequality (3.43) or
-qje,j

(4.23)

0,

since 8 > 0. Clearly, we must have

Yi =

c/i(e,,)

(4.24)

but since the other constitutive equations (4.19)-(4.2 1) depend on the Y~,~,EE and 8,
there is no logical reason to exclude them from this one[32]. Thus, by virtue of the chain
rule of differentiation, we may write
(4.25)
for the general functional dependence of the heat flux vector in thermoelectroelasticity.
Now, qi cannot be an arbitrary function of the variables shown in (4.25) because an
arbitrary function of the 16 variables (4 vectors and 4 scalars) will not satisfy objectivity[25], i.e., will not be independent of the (possibly moving but not deforming)
frame of reference of the observer. However, if we express qi in the form
(4.26)
then it can be shown readily, using established methods[26,33,23],
that the principle
of material objectivity is satisfied if LK is a vector invariant in a rigid motion. Then the
previous application of Cauchys theorem [28] on invariant functions of vectors shows
that the required invariance of L, is assured if LK is of the form
(4.27)
where the functional dependence on the 13 variables shown in (4.27) may be arbitrary.
However, LK must be an odd function[33,23]
of e,.,, (or 0,) because of (4.23). On the
other hand, it is easy to show, with the aid of (4.3) that x cannot depend on O.M.Thus,
the constitutive equation for the heat flux vector in the general case is given by
qi = Yi,KLK,

(4.28)

with LK as given in (4.27).

Equations (4.19)-(4.21) and (4.28) determine the constitutive equations for our
thermoelectroelastic
continuum. Thus, all that remains in the determination of explicit

599

On the nonlinear equations of thermoelectroelasticity

const~tutive equations is the selection of specific functional forms for x and LK. Once
the constitutive equations have been determined, we have a determinate thermoelectroelastic theory, which by appropriate substitution can readily be reduced to 5
equations in the 5 dependent variables yj, cpand 8. The 5 equations are the 3 of (3.23),
(3.26), and (3.42). In order to have a complete field theory, the boundary (or jump) conditions at surfaces of discontinuity have to be adjoined to the aforementioned 5 field
equations. This is done in section 6, after a brief digression on the equation of the
balance of energy in section 5.

5. THE

INTEGRAL

FORM

OF THE

BALANCE

OF ENERGY

Jump (or boundary) conditions across surfaces of discontinuity are determined from
the integral forms such as, e.g., 13.2 l), which are valid even when the differential forms
such as, e.g., (3.23), are not. Although the jump conditions on energy are not needed
in the solution of boundary-value problems, such conditions are interesting in themselves and can be used, at times, for purposes of approximation and for determining
certain information when the solution is not available. In order to determine the jump
conditions associated with the conservation of energy, (3.30), we must first obtain the
integral form of the energy equation, which does not have a volume source term. In
addition, the integral form, of which we are speaking, is both interesting and enlightening in itself. To find this form we begin with Poyntings theorem for the quasi-static
electric field, which takes the degenerate form[34]

(5.1)
where (4F)-@Dil8t
is the form taken by the Poynting vector[35] cEX H/4rr (H =
magnetic field, c = speed of light) in a quasi-static electric field in insulators, and the
term on the 1.h.s. represents the rate of flow of quasi-static electric field energy out
of the closed surface S.
The degenerate form of Poyntings theorem is written in terms of ~~ufia~ elements
of volume V and surface S, which are stationary with respect to an inertial reference
frame, while the energy equation, (3.30), is written in terms of material elements of
volume V and surface S, which are moving with respect to the inertial reference frame.
With the aid of (3.17) and the well-known relations [36]
alat = dldt-v,(

),K,

(dldt) dV = 2)k,h_
dV,

which are used in obtaining all forms of the transport

from (5. I), we obtain

theorem[37~,

(5.2)

in the usual way,

which is the quasi-static form of Poyntings theorem when written in terms of material
regions and surfaces. From (5.3), (3.30) and (3.39), with the aid of (3,7), (3.36), (5.2)

600

H. F. TIERSTEN

and the divergence theorems we obtain

U,JI~+PE+~
I

87r

EkEk dV

fkVk-nkqk-nkGF+nkVk

aDk

E?fEM
z gn

+ nkVkE?Pi

dS,

(5.4)

s
which is the integral form of the equation of the conservation of energy we have been
after. As with (3.21), equation (5.4) is taken to be valid across moving surfaces of discontinuity even when (3.30) is not meaningful, and (5.4) reduces to (3.30) when P and
EM are appropriately differentiable. Clearly, equation (5.4) can be used to determine
jump conditions on the energy. In addition, equation (5.4) says that the material time
rate of change of kinetic plus stored internal plus quasi-static electric field energy is
equal to the rate at which work is done by the surface tractions acting across S less the
flux of thermal and quasi-static electric Poynting energy outward across S plus a convective flux of quasi-static electric field energy and electric field-electric dipole interaction energy. Obviously, the first law of thermodynamics,
(3.40) could have been
determined as readily from (5.4) as from (3.30) with (3.39). However, it should be
remembered that the integral form in (5.4) is now taken to be valid even when (3.30)
is not meaningful, such as across moving (non-material) surfaces of discontinuity, and
reduces to (3.30) when p, r, and EMare appropriately differentiable.
6. THE

BOUNDARY

CONDITIONS

In this section we determine boundary conditions which must be adjoined to the

system of differential equations, as mentioned at the end of section 4, in order to
formulate boundary-value
problems. These boundary (or jump) conditions may be
determined by applying the integral forms of the appropriate field equations to a pill-box
region surrounding the moving (not necessarily material) surface of discontinuity[38]
with normal velocity u,, and assuming that the pertinent variables remain bounded. The
appropriate integral forms are (3.7) (3.21) (3.15) and the integral form of (3.43) which
takes the form

(6.1)
and the jump conditions resulting from them are

ni

4JPl-

(6.2)

ni[vipl = 0,

(D,EJ-&EfEfG,)

+u,[Pu~]

-ni[qpvj]

ni[Di] = 0,

ni[qil4 - &t[pVl+ ni[QP71 2 0,

= 0,

(6.3)
(6.4)

(6.5)

where we have introduced the conventional notation [CJ for Cc - C; and ni denotes the
components of the unit normal directed from the - to the + side of the surface of
discontinuity. The jump condition on EM may be determined from (3.16) just as in

On the nonlinear equations of thermoelectroelasticity

electrostatics

601

[39] and is given by

e&Zj[E~] = 0.

If the surface of discontinuity

(6.6)

is material,
U, =

niVi

+--

nit+- ,

and (6.2) reduces to nothing and (6.3) and (6.5), respectively,

(6.7)
reduce to

(6.8)
ni[qJO] 2 0.

If, further, 0 is continuous,

(6.9)

i.e.,

101= 0,
across the surface of discontinuity,
have

(6.10)

y is bounded and, from (6.1), in place of (6.9) we

ni[qi] = 0.

(6.1 I)

This latter situation, consisting of the jump conditions (6.4), (6.6), (6.8), (6.10), and
(6.1 l), is the one in which we are primarily interested. Note that the condition in (6.6)
may be replaced by the equivalent
LPI = 0.
If a body does not abut another
ditions are fully defined by (6.4),
another solid body and the full
satisfied in both regions, we must

(6.12)

body but abuts, say, air instead, the boundary con(6.8) and (6.10)-(6.12). However, if a body does abut
field equations (3.23) (3.26) and (3.42) have to be
have ageometric condition on [yi], usually,
[Yil = O,

(6.13)

at the surface of discontinuity in addition to (6.4), (6.8), and (6.10)-(6.12). This is the
case of primary interest to us. Frequently the electrical and/or thermal conditions at a
surface are such that we may eliminate either (6.4) or (6.12) and/or either (6.10) or
(6.11). Clearly, all boundary expressions, which are not prescribed, may be expressed
in terms of the same 5 field variables as the 5 equations (3.23), (3.26) and (3.42) by
making the appropriate straight-forward substitutions.
Thus we have obtained a properly invariant nonlinear field theory of thermoelectroelasticity consisting of the aforementioned 5 equations and 10 boundary conditions, all
expressed in terms of the 5 field variables yi, cp, and 8 in each region. All that remains in
the determination of explicit equations is the selection of a specific form for x and LK. A
sensible polynomial approximation for x might be of the form

IJES Vol. Y.No.

7-B

602

x=

H. F. TIERSTEN

-!CKLM&KLE.U.V
- &
2Po

XKLWKWL
+ hC0-

$- EKLMWKELM
-+-

hmvWKW~,E~,v

where the material coefficients CKLMN,xxL, C, eKLM,bKLMN,aKI,, and hK are called the
elastic, electric susceptibility, thermal, piezoelectric, electrostrictive, thermoelastic and
pyroelectric constants, respectively, and may be functions of the absolute temperature
8 (and perhaps even the electric field through WK in some instances). For linear heat
conduction the form for LK would be
LK

(6.15)

KKN~,.v,

where the linear thermal conductivity tensor K may be a function of 8, ELM and W,.
As noted earlier we can determine an energetic jump condition from the integral
form (5.4), which condition is not needed in the determination of the full solution of a
boundary value problem, but can be useful for obtaining information when the full
solution is not available. This jump condition is obtained by applying (5.4) to the
aforementioned pill-box region surrounding the (non-material) surface of discontinuity
and assuming that p, UjyE, and Ef remain bounded, with the result
50

TjkV*-qj---+Vj

aDi

4~ at
+ u,

ipvkvk

EkEk

+ PE +

8lr
&

+ UjEfPk
E$Ef

1
3pv,vk

+ pi +&

EfEf

)I

= 0.

(6.16)

If the surface is material, we have (6.7), and equation (6.16) reduces to

TjTjRVk-qj---+Vj

EpEk
cpa@ -+
47~ at
87r

UjEfPk

= 0s

(6.17)

Acknowledgement-This
work was supported in part by the Office of Naval Research under Contract
No. NO00 14-67-A-0 117-0007and the National Science Foundation under Grant No. GK- 1 I 195.
REFERENCES

[II
PI
l31
t:;
l61
I71
t81
l91
t101

R. A. T0UPIN.J.
rution. Mech.Analysis 5,849 (1956).
A. C. ERINGEN, Int. J. Engng Sci. 1,127 (1963), and many subsequent papers in the same journal.
H. F. TIERSTEN, J. marh. Phys. 5, 1298 (1964).
W. F. BROWN, Jr.,J. appl. Phys. 36,994 (1965).
D. E. EASTMAN, Phys. Rev. 148,530 (1966).
E. S. SUHUBI, I&. J. Engng Sci. 7,993 (1969).
A. C. ERINGEN, Nonlinear Theory ofContinuous Media, Sections 110 and 111. McGraw-Hill, New
York (1962).
N. F. JORDAN and A. C. ERINGEN, In!. J. Engng Sci. 2,59 (1964).
R. C. DIXON and A. C. ERINGEN, Int. J. Engng Sci. 3,359 (1965).
H. F. TIERSTEN, Linear Piezoelectric Plafe Vibrurions, Chap. 4, Section 4. Plenum Press, New York
(1969).
C. TRUESDELL and R. A. TOUPIN, The classical field theories In Encyclopedia ofPhysics, edited
by S. FLUGGE, Vol. III, Sections 13.15 and 16. Springer, Berlin (1960).

On the nonlinear equations of thermoelectroelasticity

603

[121Ref. [lo], Chap. 3, Section 3.

u31 Ref. [ 111, Sections 199-200.
[I41 Ref. [ll], Section 17.
[I51 M. MASON and W. WEAVER, The Electromagnetic Field, Eq. (lo), Section 10. Dover Publications,
New York (1929).
1161 Ref. [IS], Eq. (52).
[I71 Ref. [15], Eq. (53).
[181 Ref. [ll], Section 203.
[I91 Ref. [l], Eq. (9.1).
WI Ref. [7], Sections 36-40.
PII B. A. BOLEY and J. H. WEINER, Theory of Thermal Srresses, Sections 1.8-1.11. Wiley, New York
(1960).
Thermodynamics.
North Holland Publishing
WI S. R. DeGROOT and P. MAZUR, Non-Equilibrium
Company, Amsterdam (1962).
P31 B. D. COLEMAN and W. NOLL,Arch. r&on. Mech. Analysis 13, 167 (1963).
[241 B. D. COLEMAN, Arch. ration. Mech. Analysis 17, 1 (1964).
1251 Ref. [7], Sections 27 and 44; Ref. [I I]. Sections 293 and 296.
WI A. E. GREEN and J. E. ADKINS, Large Elastic Deformations and Nonlinear Continuum Mechanics,
Chap. 8. Oxford University Press, London (1960).
1271 Ref. [ 11, Section 11.
1281 A. L. CAUCHY, Mem. Acud. Sci. XXII, 615 (1850) (Oeures (I) 2,351).
[291 H. JEFFREYS, Cartesian Tensors, p. 12. Cambridge University Press, New York (193 1), reprinted
(1952).
[301 Ref. [7], Section 12.
[311 See Appendix A of Ref. [3].
[321 This procedure was give the name Principle of Equipresence by TRUESDELL. See Ref. [I 11, Section
293.
[331 A. C. PIPKIN and R. S. RIVLIN, Technical Report DS 4531/4 to the Department of the Army,
Ordnance Corps (1958).
[341 Ref. 1101,Chap. 4, Sections 3 and 4.
[351 Ref. [15], Eqs. ( 182) and (183).
1361 Ref. [7], Eqs. (17.5) and (19.7).
[371 Ref. [ll], Section 81.
1381 Ref. [ll], Sections 192 and 193.
Theory, Section 1.13. McGraw-Hill Book Company (1941).
[391 J. A. STRATTON, Electromagnetic
(Received

24 February

1970)

Resume- Les equations differentielles et les conditions aux limites decrivant le comportement dun milieu
continu, polarisable Clectriquement, deformable de facon finie et conducteur thermique, en interaction avec
le champs electrique, sont deduites par le moyen de lapplication systematique des lois de la physique des
milieux continus a un modtle macroscopique consistant en un continuum de charge Clectronique couple avec
tin continuum en treillis. La description resultante. invariante du point de vue rotationnel, de la thermoelectro-elasticite consiste en cinq equations ditferentielles a cinq variables dependantes et, en faisant abstraction des considerations thermiques, se reduit a quatre equations ditferentielles a quatre variables dependantes.
Ces quatre mCmes equations differentielles de lelectroelasticite avec les quatre m&mes variables dependantes,
en mime temps que les conditions aux limites qui leur sont associees, peuvent ftre obtenues facilement, par
le moyen dune transformation de Legendre, a partir de letude variationnelle compatible deja existante de
lelectro-tlasticite, qui aboutit a un systeme de sept equations a sept variables dependantes.
Zusammenfassung-

Die Differentialgleichungen und Grenzbedingungen, die das Verhalten eines elektrisch

polarisierbaren, endlich defonnierbaren,
wlrmeleitenden
Kontinuums in Wechselwirkung mit einem elektrischen Feld beschreiben, werden mit Hilfe einer systematischen Anwendung der Gesetze der Kontinuumphysik auf ein makroskopisches Model1 abgeleitet, das aus einem elektronischen Ladungskontinuum gekoppelt zu einem Gitterkontinuum besteht. Die resultierende rotationsinvariante
Beschreibung von Thermoelektroelastizitat besteht aus funf Differentialgleichungen in funf abhiingigen Veranderlichen und wird, wenn
thermische Betrachtungen weggelassen werden, auf vier Differentialgleichungen in vier abhangigen Veranderlichen reduziert. Dieselben vier Differentialgleichtmgen
von Elektroelastizitat in denselben vier abhangigen Veranderlichen, zusammen mit den assozierten Grenzbedingungen, konnen leicht durch eine Legendre
Transformation von bestehenden konsistenten Variationsbehandlungen
der ElektroelastizitEt erhalten werden, die ein System von sieben Gleichungen in sieben abhangigen Veranderlichen ergeben.

604

H. F. TIERSTEN

Sommario-Si
ricavano le equazioni differenziah e le condizioni limite the descrivono il comportamento di
un continua termoconducente
finitamente deformabile e polarizzabile elettricamente in interazione con il
campo elettrico mediante Iapplicazione sistematicadelle leggi di fisica de1 continua a un modello macroscopico
consistente in un continua a carica elettronica accoppiato a un continua a traliccio. La descrizione rotativamente invariante the ne resulta per la termoelettroelasticita
consiste in cinque equazioni differenziali in
cinque variabili dipendenti e, se si tralasciano le considerazioni termiche, si riduce a quattro equazioni
differenziali in quattro variabili dipendenti. Le stresse equazioni differenziali di elettroelasticita nelle stesse
variabili dipendenti (quattro in ambi i casi) con le relative condizioni limite sono facilmente ottenibili mediante
una trasformazione di Legendre partendo da esistenti trattamenti variazionali consistenti the danno un sistema di sette equazioni in sette variabili dipendenti.
A&TpaKT-%JBOAflTCSi
3JIeICTpEiYeCKEt
A&CTBHIl

C 3,ICKTpHYeCKHM

CKOIIHYeCKOii

MOACJIC,

KOHTHHYYMOM.
113 IIRTb
WCKHZ

AH+$CpCHWiUIbHbIe

lIOJI~ptl3kipyeMO~0,

PC3yJlbTIipylO~ee

coo6paxemi~

3itBIICHMbIX

OH0

Ha

BWIHYHH.

OTHOCUTeAbHO

IIOAeM,

COCTOflIl@

AkI+j,ZpeH~Wi.bHbIX

CaMbIX

IIO,IJYaTCH

,IWYe

ynpy,-OCTH,

KOTOPbIt2

CACTCMaTHYCCK%I

CBOAHTCII

K YeTbIpZ-

OCHOBC IIpeO6pa30BaHEix
3tlBHCUMbIX

CFfCTeMy

OtIliCaHlie

IIRTb

&SiZHApa
BWIHYHH

,PaBHeHUii

BUIUYHH,

3T1( YeTbIpe

K MaKpO-

PeIIIeTYaTHbIM

MeTOAOB
CeMb

COCTOHT

II He YYliTbIBaR

TCPMH-

OTHOCEiTWIbHO

YeTbIpC

Aki~~epeHlWJIbHbIX

C CBH3aHHbIMR

OTHOCHTCJIbHO

B BSakiMO-

@iSEiKK

CBrt3aHHOi

YPaBHeHUnM

BapHWHOHHbIX

nOBeAeHHC

TepMO3~CKTpOynpyrOCrA

3aBUCHMbIX

BMeCTC

KOHTEiHYyMa

KOHTAHYYMHOfi

KOHTHHYYMa,

AH~@peHUWIbHbIM

COBMeCTHbIX
113 CCMb

3aKOHbI

3apSlAHOrO

OTHOCHTeAbHO

YCJIOBkiSI, OIIHCbIBalOIWie

TeIlSIOIIpOBOAHOrO

IIpAMeHSIX

pOTaWiOHHO-tiHBapEiaHTHO2
)paBHCHHii

CyIQCCTBjIOUHX

AWJT

Ii rpaHHYHbIC

AC+OpMHpyCMOrO,

3JIeKTPOHHOTO

X02 YeTbIpe

113

YpaBHeHHSl

KOHWHO

YPaBHeHMii

rPaHRYHbIMH

YCJIOBHnM

paCCMOTpCHliR

WIeKTPO-

3aBEiCUMbIX

BWIIIYHH.