50

ELECTROSTATICS IN THE
PRESENCE OF MATTER
As we have remarked before, for our purposes, matter is to be regarded as a collection
of the positive and negative charges of its constituent nuclei and electrons. We assume
that an ordinary piece of matter is made up of atoms and molecules that have equal
amounts of positive and negative charges, that is, that have zero monopole moments.
In Chapter 6 we considered one class of matter—conductors—which we assumed to
have charges that were free to move about under the influence of fields. In this chapter
we assume that we are dealing with matter that is not a conductor. These latter types of
material are commonly called dielectrics and we refer to their charges as bound charges.
It is not always possible to make a clear-cut characterization of a gjven substance as a
conductor or dielectric (nonconductor), but it is still useful to have these two categories
available for a first rough classification scheme. Later we will be able to combine these
two properties in our overall description.

10-1 POLARIZATION
Although our primary aim is to develop a macroscopic empirical description of
electromagnetism, for a while we will use the microscopic picture of matter in a
qualitative manner in order to get an idea of how we should proceed. There are several
possibilities that come to mind when we consider the effect of an electric field on an
atom or molecule.
In the absence of a field, it may well happen that the molecule has its negative
electronic charge distributed symmetrically about the positively charged nuclei. In this
case, the molecule will have a zero dipole moment as can be easily seen from (8-22). If,
now, there is an electric field present, there will be forces exerted on these charges. The
positive charges will tend to be moved in the direction of the field, and the negative
charges will tend to be moved in the direction opposite to the field. Eventually, the
internal forces within the molecule will produce a new state of equilibrium, but the
molecular charge distribution will have been distorted from its originally spherically
symmetric form. Thus, the previously “coincident” positive and negative charges will
have had their overall “centers of gravity” shifted with respect to each other, and the
net result will be a new charge distribuuon with roughly the character of that shown in
Figure 8-3. But, as we know, this means that the molecule now will have a nonzero
dipole moment p, and probably other higher-order multipoles as well. We say that the
molecule now has an induced dipole moment and that it has become polarized.
Another possibility is that, because of their internal structure, some molecules
already have their positive and negative charge distributions separated so that they
have a dipole moment even in the absence of an electric field. Such molecules are called
polan molecules and the dipole moment is referred to as a permanent dipole moment, the
monopole moment is still zero, of course. An example of this type is the water
molecule, H 20. Here the negative charges tend to cluster more around the oxygen
leaving the hydrogens positively charged; the net result is a charge distribution roughly
 10—J POL A R IZAT ION 141

like that of Figure 8-4 (with the signs of the charges reversed) and with q approximately
equal to e, the magnitude of the electronic charge. In the absence of an electric field,
these permanent dipole moments will generally be randomly oriented so that the total
dipole moment of the whole piece of matter will still be zero. In the presence of a field,
however, there will be a torque on the dipole that will tend to rotate the dipole into the
direction of the field as we found in (8-75). This tendency can be expected to be
opposed by other internal forces of the system and, especially in the cases of gases and
liquids, by the temperature agitation of the molecules that will lead to collisions with
their tendency to make things random. Nevertheless, when equilibrium has been finally
attained in the presence of the field, we can expect that the permanent dipoles will have
been rotated on the average into some degree of alignment with the direction of the
field. The overall effect is that the matter will have a net dipole moment in the direction
of the field. Thus we have concluded again that the material will be polarized.
A final possibility is that the permanent dipole moments of some materials are
aligned to some extent even in the absence of an electric field. Such materials are called
electrets and are said to be permanently polarized. They are comparatively rare, and
not of as great commercial and technical importance as their magnetic analogues—per-
manent magnets.
As we noted above, we can reasonably expect the molecules. especially when
distorted by the field, to have higher-order multipoles, such as quadrupole moments,
associated with them in addition to their dipole moments. On the other hand, we know
from Chapter 8 that the contributions of these other multipoles to the potential and
field fall off more rapidly with distance from the molecule, and, in addition, their
contributions generally vary with angle in a more complex manner. Since we are trying
to get a macroscopic description in terms of the average of the microscopic behavior of
the constituents of matter, we will assume that, on the average, the dominant features
of matter that are of interest to us are simply those associated with the electric dipole
moments. Thus, all of our considerations lead us to the following:

Hypothesis
As far as its electrical properties are concerned, neutral matter is equivalent to an
assemblage of electric dipoles. •

Although this is obviously an hypothesis, it clearly must have worked very well in the
development of this subject. Our next problem is to try to formulate this hypothesis in
a quantitative manner.
For this purpose, we define the polarization P as the electric dipolc moment pcr unit
volume, so that the total dipole moment d p in a small volume d i at r will be
d p - P(i) d i (10-1)
Thus the total dipole moment of a volume K of material will be

(10-2)

From its definition and (8-19), it follows that P will be measured in coulombs/ (meter) 2.
As implied by (10-1) and (10-2), P is generally expected to be a function of position
within the material. As in Section 2-4 when we introduced charge density, we are
assuming di to be small on a macroscopic scale but large on a microscopic atomic
scale so that di actually contains many molecules. This makes P obtainable as an
average over volumes large enough so that small fluctuations in the number of
molecules within a given volume element do not prevent us from treating P as a
 142 £ L CCT RWSTATICS IN THE PR ESF NC £ OF MATT ER

smoothly varying function of position. For example, if there are n molecules per unit
volume each of which has a dipole moment p, then P = up provided that all of the
dipoles are in the same direction.
Because of the way in which we were led to it, we expect there will be a functional
relation between P and E and eventually we shall have to consider how we can
determine P. For the time being, however, we simply take it as giving us a macroscopic
description of the material and we want to investigate the consequences of its existence.

0-2 BOUND CHARGE DENSITIES

Let us assume that we have a polarized object and we want to calculate the potential it
will produce at a field point r located outside the body as shown in Figure 10-1. The
dipole moment of the volume element di’ as given by (10-1) is d yt —— P(r’) di' and its
contribution to the potential at r as obtained from (8-21) is

d (10-3)

where, as usual, R = r — r’ and corresponds to the r of (8-2J). In order to find the total
potential, we integrate this over the volume K’ of the material and we get

r’)-R di’ 1 r’) - W p d (10-4)

4 A
2 ' 4ac
xc
with the use of (1-141). This can be put into a better form if we use (1-115) to write the
integrand as

(10-5)

Figure 10-1. Calculation of the

potential outside a polarized body.
10 2 BOU N D CHA R G E D E N SIT I E S
343
If we insert this into (10-4) and use (1-59) and (1-52), we obtain

4 r O 4wc K*(

1 “i w - P) di +
—
4 f
 e a’

fi
(i 0-6)
where S’ is the surface bounding K’ and iY is the outer normal to the surface as shown
in the figure. Upon comparing this with (5-7) and (5-8), we see that (10-6) is exactly the
potential g that would be produced by a volume charge density p5 distributed
throughout the volume and a surface charge density •b on the bounding surface where
pz —— — -P (10-7)
(10-8)
for then we would have
d oz
'"’ ' 4 wi c' ” + da’ (10-9)

as we would expect. [In (10-8), # is the normal component of P evaluated on the

surface.]
What we have discovered, therefore, is that, as far as its effects outside itself are
concerned, the dielectric can be replaced by a distribution of volume and surface
charge densities that are related to the polarization P by means of (10-7) and (10-8).
The various steps that have led to our present conceptual scheme are summarized and
outlined in Figure 10-2. The total potential at the field point will then be given by
(10-9) plus the potential due to any other charges which might be present.
It is common practice to omit the prime in (10-7) and simply write

with the understanding that the differentiations are made with respect to the source
point coordinates.
The subscript b appearing in (10-7), (10-8), and (10-10) reflects the fact that these
charge densities arise from the bound charges of the dielectric. Consequently, they are
usually referred to as the bound charge densities or the polarization charge Jensities.
Although we identified these quantities by means of a formal comparison of our
expression for the potential with its general form, it is possible to understand and
calculate the origin of these densities in a more direct “physical” manner; we will,
however, do this only qualitatively. As an extreme example, let us consider a piece of
material with uniform polarization produced by dipoles of the same magnitude all
pointing the same direction as shown in Figure 10-3. As in Figure 8-3, we can associate
a positive charge with an arrow head, and a negative charge of the same magnitude
with the tail of an arrow. Then, in a small volume within the material, as indicated by
the dashed curve, we will have, on the avcrage, as many positive as negative charges so

Assemblage
of dipoles Equivalent volume
Dielectric and surface
described
by B charge densities
g¿ and a¿

Figure 10-2. Conceptual scheme of replacing a dielectric by equivalent charge densities.

 1 44 ELECTROSTATICS IN THE PRESENCE OF MATTER

Figure 10-3. Origin of bound surface charges for a uniformly polarized dielectric.

Figure 10-4. Origin of bound volume charges in a

dielectric with nonuniform polarization.

that pp —— 0 in agreement with (10-10). However, there will be no such cancellation on

the faces and this will produce a surface charge density of the signs shown; these signs
are in agreement with (10-8) since P is positive for the right side of thc figure and
negative for the left side. Now let us suppose that P is not uniform, so that the
distribution of dipoles may be something like that of Figure 10-4 since the dipoles may
differ in both magnitude and direction. Let us consider a fixed volume of the material
as indicated by the dashed line. Before the matter was polarized, this volume will
contain many neutral molecules so that p = 0. Now when the material is polarized
and the dipoles are formed, some charges will be displaced out of the volume while
others will be displaced into it. If P is not uniform, we see that it may well happen that
the volume may end up with more charge of one sign than of the other and there will be
a net bound charge density. This is exactly the effect described by (10-10), and, in fact,
considerations like these can be used as an alternative way of deriving p# —— —p -
P. Suppose we have two polarized dielectrics
meeting at a common boundary as illustrated in Figure 10-5. Since each of them has
a normal component at the surface, they will produce a surface charge there given by
(10-8); the net surface charge density
at „, will be the sum of these two terms so that

where if2 f nd it’s are the outward normals to the respective media. If we now introduce
the unit vector ñ pointing from 1 to 2 as defined in (9-2), we see that ñ’z = — ñ and
 10 3 THE E L ECTRIC FIELD WITH I N A DIELECTRIC 145

Figure 10-5. A boundary between two

regions of different polarizations.

n = ñ so that (10-11) becomes

ñ (P 2 — Pt) - — a „, (10-12)
This result is exactly what we get by applying our previous results (9-6) and (9-24) to
this case, which is described by (10-10); thus (10-7) and (10-8) are consistent with each
other.
As a further illustration of the internal consistency of our results, let us find the total
bound charge Qb of a polarized dielectric of finite extent. According to (2-14) and (2-
16), this will be given by

(10-13)
with the use of (10-7), (10-8), and (1-59). Thus the total bound charge is zero. But this is
exactly what we would expect from our ideas of polarization as arising from the
separation of charge in our originally neutral medium rather than from the creation of
new charge. (In the case of permanent dipoles, the origin of polarization is due to
reorientation of already separated charge, but this does not involve charge creation
either.)

10-3 THE ELECTRIC FIELD WITHIN A DIELECTRIC

Up to this point all of our results have been obtained by considering the potential, and
its corresponding electric field, at a field point in the vacuum outside the body of the
dielectric. The reason is that then there is no problem about what we mean by the
electric field in the sense of (3-14) since we can put a test charge there and measure
the force on it without any difficulty. However, if we look at the situation inside a
dielectric from this point of view, we cannot measure the force on a test charge without,
say, drilling a hole in the material in order to be able to insert the charge. But if we do
this, we can anticipate that we will alter the preexisting situation if for no other reason
than that we will remove some bound volume charges and introduce new surface
charges because we have created some new bounding surface. There are several ways in
which one can proceed to answer the question of what we mean by the electric field
inside the dielectric.
The simplest thing to do, and probably the best, is just to say that we calculate the
potential, and its associated electric field, inside the dielectric in exactly the same way
0 4& ELECT ROST AT ICS I N TH E PRESENCE OF MATTER

in which we do it for a point outside. In other words, we say that (10-6) or (10-9) can be
used to calculate Q anywhere by definition. There is certainly nothing wrong with this
and it is a perfectly reasonable thing to do; it is completely consistent with our
conceptual replacement of the actual dielectric by a set of equivalent volume and
surface distributions of bound charge as shown in Figure 10-2. We have discussed this
approach before in Section 3-4, and we can point out again that our basic aim is to
develop a macroscopic description of electromagnetism that agrees with experiment. We
come back to this point below.
Another approach that can be used is to start with a microscopic picture and try to
determine the macroscopic equations as appropriate averages. This is quite a com-
plicated program to carry out accurately for all of electromagnetism, and we shall have
to content ourselves with a brief outline of what is involved for electrostatics. On this
scale, most of the interior region will be vacuum and the electric field will be
determined by all of the nuclear charges and the atomic electrons. This electric field will
have enormous variations in it as we go from points very near these charges to points
which are comparatively far from them. At the same time, it will vary quite rapidly
with time because of the motion of the constituent charges. Thus, when we define our
macroscopic electric field, we will want to do so as an average both over the time and
over a volume large enough to contain a reasonable number of molecules but not so
large that we cannot consider it as infinitesimal on a laboratory scale. (Tins is similar to
the problem of defining average charge density as we discussed in Section 2-4; thcre we
saw that if we take too fine a subdivision, p can have very large fluctuations.) If we let e
be the microscopic electric field, we want it to have the same basic vacuum properties
that we have already found in (4-10) and (5-4), that is,

(10-14)

where p„, is the charge density defined on a microscopic scale. Our macroscopic field E
would then be defined as
E = (e) (10-15)
where (e) is an average over both time and space. Since the differential operators are
constant as far as this averaging process is concerned, we get from (10-14) and (10-15):
W X (e) = w X E = 0 (10-16)

(10-17)
Now pq itself can be expected to have large spatial and temporal variations, but when it
is averaged we expect it to reduce to the average density of the displaced bound
charges; that is, pz) — p#, so that

(10-18)

But (10-16) and (10-18) are equivalent to simply using (10-9) and (5-4) in the first place.
Thus this approach leads us to exactly the same conclusions we reached in the previous
paragraph.
This microscopic to macroscopic method of determining E in the material is
somewhat different from the question of what is the actual electric field seen by the
molecule, which is what will determine the actual molecular dipole moment produced.
This field will not necessarily be the same as the macroscopic E defined above as the
molecule itself contributes to the resultant field. This question is discussed in Appendix
 0-3 TH E ELECT RIC FIEL D WITHIN A DIELECTRIC 147

B; there we consider the question of calculating the electromagnetic properties of

matter as the resultant response of its constituent atoms and molecules, but it is not at
all necessary to worry about that now.
It is helpful at this point to verify that our definitions of Q and E, which we have just
adopted, are in agreement with early experiments on dielectrics done by Faraday. We
do this only qualitatively now, but we will be able to give a quantitative discussion in
Section 10-7. We consider two simple experiments. We charge a capacitor to some
charge @, and measure the potential difference fiQ when there is a vacuum between the
plates. The capacitance will be

(10-19)

according to (6-28). Now we take this same capacitor, completely fill the region
between the plates with some “reasonable” dielectric like wax or oil, while keeping Q
constant, and then measure the potential difference fig. The capacitance is now given
by

(10-20)
The experimental result is that AQ < b , and therefore
C> C (10-21)
How can this result that the potential difference decreased be understood? Suppose, in
order to be specific, we consider a parallel plate capacitor. Then we can assume the
electric field to be directed from the positively charged plate to the negatively charged
plate as in Figure 6-9, and when we calculate fiQ by means of (6-38), we get
fiQ - + Ed de —— E od (10-22)

where Eg is the electric field with a vacuum present and d is the plate separation.
Similarly, if E is the electric field with the dielectric between the plates, we get
k Q —— Ed (10-23)
and, since ñ$ < fig , while d is constant, we conclude that
E< E (10-24)
so that the electric field has been decreased by the presence of the dielectric. How could
this have come about? In Figure 10-6, we show the capacitor with dielectric between its
plates; the surfaces of the dielectric are shown slightly separated from the conductors
for clarity, but we really assume that the region is completely filled. If the dielectric is
uniformly polarized so that P = const., then p p — 0 by (10-10). Thus the only bound
charges will be surface charges given by (10-8), and they will have the indicated signs.
These surface charges will produce a field Eb with the direction shown; since the charge

Figure 10-6. Charges and fields

for a parallel plate capacitor.
 14& ELECTROSTATICS IN THE PRESENCE OF MATT£R

Figure 10-7. Cavity used to measure E in a

dielectric.

Q will still produce the field Ed, the resultant field E is seen to be
E —— E — Ez (10-25)
so that E <r 0 in agreement with (10-24) and experiment. Thus, this method of
evaluating the field inside the material by the use of equivalent bound charge agrees, in
this case at least, with experiment and our general ideas about the relation between
electric field and potential as expressed by (6-38).
Although we briefly alluded to the difficulty of actually measuring the electric field in
the dielectric by means of a test charge, one can still ask if there is a way of doing
it — obviously we will have to cut a hole in order to insert the charge. It is possible to
devise such a scheme by means of a judicious use of the boundary conditions satisfied
by E; the relevant one is (9-21), which says that E, is continuous. The reasoning is as
follows: (1) somehow determine the direction of E in the dielectric (as we will see in
Section 10-6, this is easily done for a large class of dielectrics); (2) make a long narrow
holc in the material parallel to E as shown in Figure 10-7—this cavity now has a
vacuum inside (such a cavity is usually called “needlelike”); (3) any bound surface
charges that might alter the fields are on the ends of the cavity since that is the only
place P has a normal component, and, since the ends are so far away and of such a
small area, we can neglect their effects; (4) since only tangential components are
involved, E,. = E where E is the field in the cavity; (5) now insert a test charge 6q into
the cavity and measure the force 6F on it. Then, by (3-14),

E = E (10-26)

and the electric field iii the dielectric can thus be found, in principle, from measure-
ments made in the cauity. This scheme is known as the contra definition of E.

10-4 UNIFORMLY POLARIZED SPHERE

As an example of the use of bound charges, let us consider a sphere of radius a that has
a constant polarization P. We choose the z axis to be in the direction of P and the
origin at the center of the sphere as shown in Figure 10-8; thus, P = Pz. Since P is
constant, pb —— 0 by (10-10). We see from the figure that the outer normal to the
dielcctric ñ’ = é, and therefore there is a bound surface charge density which is found
from (10-8) and (1-94) to be
a (8’) = P co$ 8 (10-27)
As we see, o is not constant but varies in magnitude and sign with angle as indicated
in Figure 10-9. For simplicity, we will find the potential and electric field only for
points on the z axis and for positive z; in the next chapter, we solve this problem
completely by a quite different method.
We see from Figure 10-10 that
2
A = ( z 2 + o 2 — 2zci cos 6’)” (10-2g)
and, therefore, if we use (10-27), (1-100) (with r -— u), and (2-22), we find that as
 1 0-4 UNIFORMLY POLA RIZ ED SPHERE 149

Figure 10-9. Equivalent rharge

Figure 10-8. A uniformly distribution of a uniformly
polarized sphere. polarized sphere.

Figure 10-10. Calculation nf the potential at a point nn

the axis.

given by (10-9) is
P cos a 2 sin 8’ d8’ dip’
8-’
4UE o ‘o ‘0
(z 2
+ a 2 — 2zo cos 6’ )" 2

Pa 2 'i p dp
(10-29)
The integral can he found from tables to be
( 2+ 2+ 2 2
Z ay)( z' + a — 2zoy “ ‘
2 2
3z A

and there are two cases to be considered.

150 ELECTROSTATICS IN IHE PRE$ENCE OF /vtATTER

1. Outside the sphere. Here z o and |r — «| = r — a; also |r -F o| = z -F a since

both z and a are positive. Using these values, we find that (10-30) becomes 2a/3z 2 and
when this is put into (10-29) we find the potential outside the sphere to be
Pa’

,.0.2
(10-31)

and therefore
6Q, 2 Pa’
E p( z ) = (10-32)
8z ‘ 3e z '
These results become more understandable if we express them in terms of the total
dipole moment p of the sphere which, as found from (10-2), is
(10-33)
so that (10-31) and (10-32) can also be written as
P
(10-34)
4a< z 2
2p
(10-35)

Upon comparing these results with (8-48) and (8-50), and remembering that f = 0 and
r — z for a field point on the r axis, we see that they are exactly those of a point dipole
of total moment p. This makes us suspect that the field everywhere outside is a dipole
field corresponding to this total dipole moment; this turns out to be correct as we will
see in the next chapter.
2. Inside the sphere. Here z < a, so that |z — a| = a — r; again, |z + a| = z + a.
In this case, (10-30) becomes 2 r,Z3n 2, which, when put into (10-29), gives the potential
inside the sphere as

(1n-36)

and therefore

E, z)= — (J0-37)
âz 3r
We note that the electric held is constant; this may make us suspect that E is constant
throughout the whole interior of the sphere, and this turns out to be the case.
It will be left as an exercise to show that these same results also hold for negative
values of z, that is, E p is always in the positive direction, and E„ is constant and
given by (10-37) for all z. In Figure 10-11, we illustrate the directions for E that we
have found; these directions can be understood from the source charge distribution
shown in Figure 10-9.
It is a worthwhile exercise to verify that these results are in agreement with the
general boundary conditions we previously obtained. We find from (10-31) and (10-36)
that the potential is indeed continuous at the surface of the sphere as we found in (9-
29), that is, when z = o, Q = $, = Pa,Z3• . Along the z axis, E, is a normal component,
and therefore we find that
2P
0 ’* 0 0
 J 0-5 THED FIELD 151

° Figure 10-11. Electric fields onz axis due to a uniformly

polarized sphere.

with the use of (10-32), (10-37), and (10-27). This value for the difference in the normal
components is exactly what it should be according to (9-26).

10-5 THE D FIELD

We recall that when we defined E in Section 3-1, we pointed out that E is determined,
from its definition, by all charges of whatever origin or type. In Section 10-2, we
encountered a particular class of charge—the bound charge whose density is given by
(10-10) as pb —— — p- P. It is conventional and convenient to divide charge into the two
broad classes of bound charge and free charge with corresponding densities p p and p/.
As we have seen, we regard the bound charges as arising from the constituents of
maiter, and we generally have no control over their distribution. Free charge»
essentially the rest of the charges. The name “free” is given to them because wc Can
control their distribution to a large extent by physically moving them about, by
spraying them onto or into material by electron beams, for example, and the like. This
class is also generally taken to include the mobile (free) charges of conductors. This
division does not always give us a clear-cut classification, but it is useful nevertheless.
Thus, we can write the total charge density as the sum of these two:

and when this is substituted into w- E = p/c , as we found in (4-10), we obtain

W - ( e E + P) = p/ (10-39)
The form of this equation, in which only the free charge density appears on the
right-hand side, suggests that it may be useful to define a vector field D(r) as follows:
D = c BE + P (10-40)
152 r L ECTROSTATICS I N THE PR£SE NCE OF MASTER

for then
w- D= p (10-41)
The vector D is often called the electric displacement or simply the displacement, or,
even more simply, the D field. The principal characteristic of D, and the primary reason
for its definition, is its property that its divergence depends only on the free charge
density. The dimensions of D are the same as those of P, and thus D will be measured
in coulombs/(meter) 2. We can think of (10-41) as an expression of Coulomb’s law for
the force between point charges plus the electrical effects of matter.
Now that we have (10-41), we can quickly find some of the properties of D. The
boundary condition satisfied by its normal component can be found from (10-41), (9-6),
(9-7), and (9-24) to be
n’ (D2 i) = Dz» D i»' % (10-42)
where o/ is the surface density of free charge. Thus, the normal component of D will be
discontinuous only if there iS a free surface charge density; this is in contrast to E
whose normal component is discontinuous if there is a surface density of any kind of
charge.
Gauss’ law for D is easily found from (10-41) and (1-59) to be

U D- da = qp/ di —— Q/ „ (10-43)
S
where @ „ is the net free charge contained within the volume K surrounded by the
closed surface S. We note both the analogy to and the contrast with the corre.sponding
result for E given by (4-1). Equation 10-43 can often be advantageously used for the
calculation of D for problems of sufficient symmetry, in much the same manner as we
did for E in Chapter 4.
Even though (10-41) contains only the free charge density, this does not mean that
free charges are the only source for D, since, according to the Helmholtz theorem of
Section 1-20, we have found only one of the source equations. The remaining one is
Q X D. This is easily found by taking the curl of the definition (10-40) and using (5-4);
the result is
p x D = x P (10-44)
so that D can have sources in bound charges as well as in free ones. The boundary
condition satisfied by the tangential component of D can be found most easily from (9-
21) and (10-40) to be

I t is also possible to devise a cavity definition for D in a manner analogous to that

for E as illustrated in Figure 10-7. The definition is based on the continuity of the
normal components of D in the absence of free surface charge, as stated in (10-42). For
this purpose, we imagine a cavity in the form of a small right cylinder whose height is
very small compared to the radius of its base. ft is cut out of the dielectric so that its
base is perpendicular to D in the dielectric as shown in Figure 10-12. If we consider a
point near the center of the cavity, the edges will be too far away to affect the fields so
that D,. (in the cavity) will be parallel to D (in the dielectric), and since, by construc-
tion, they have only normal components, D = D. Then E,. = D /c = D/e since
P = 0 in the vacuum of the cavity. If we now imagine putting a small test charge bq in
the cavity and measuring the force 6 F on it, we will have 6 F, = 6q E = 6q D/e , so
that D = e (fi F„Zfiq). Thus, in principle, we can find D iii the dfelectric from measure-
ments made in thy cauit y.
 10 5 THE D FIELD 153

Figure 10-1 2. Cavity used to measure D in a

dielectric.

° Figure 10-13. D fields onz axis due to a uniformly

polarized sphere.

In spite of their apparent simplicity, (10-41) and (10-44) clearly will not be too useful
until we can relate E, P, and D better; we will do this in the next two sections.
Nevertheless, we can still briefly look at an example.

Example
Uniformly polarized sphere. This is the system we discussed in the last section for which
the results are summarized in Figures 10-8, 10-9, and 10-11. We would like to ob tain
similar things for D; we will continue to consider only field points on the positive z
axis. Outside the sphere where there is no matter, P - 0, and we find at once from
(10-40) and (10-32) that
2Pc3
(10-46)
The value of D inside the sphere as obtained from (10-40) and (10-37) is
_ _ i P (10-47)
which is seen to be independent of z and in the same direction as I . These results are
illustrated in Figure 10-13. Since there is no free charge on the surface, we expect the
 154 r r ECTROSTATICS IM THE PRESENCE OF MATTER

normal components of D to be continuous there according to (10-42). We can verify

this and check our results at the same time. We get, as expected, D2 — D
—— D (a) — D„(a) - (2 P 3) — (2f•/3) = 0. n

10.6 CLASSIFICATION OF DIELECTRICS

As we mentioned at the end of Section 10-1, we generally expect that there will be a
functional relation between the polarization and the electric field, that is, P = P(E) or
P —— P ( E , E , E ) and so on. Macroscopic descriptive electromagnetic theory does
not predict the form of these functions, but takes them as external information. From
this point of view, these relations are left to be determined from experiment or to be
calculated theoretically from the microscopic properties of matter in other branches of
physics such as statistical mechanics and solid state physics. This does not leave us in a
hopeless position, however, because a combination of experiment and general theory
shows us that most materials fall into easily classifiable groups, and this result can be
used to simplify our theory and make it more useful. It is desirable to do this step by
step in order to aid our understanding of the limitations of our final form.

1. Permanent Polarization
If E = 0, there are two possibilities for the value of P(0). If P(0) 0, then the material
is polarized even in the absence of a field and, as we noted before, it is said to have a
permanent polarization and is called an electret. Although electrets do occur, we will not
consider them further in this section. The situation for which P(0) - 0 is more typical
and the sort of thing we expect when we think of the polarization as being produced by
the field; we shall generally use the term dielectric only for this latter case.

2. Nonlinear Dielectrics
Even with P(0) = 0, it is still possible that the relation between P and E can be quite
complicated. For most materials, however, this usually requires quite exceptional
conditions such as very large fields, or low temperatures, or both. Thus, it is found that
it is often sufhcient to write P as a power series expansion in the components of E, that
is, we can write

(10-46)
j j k

where the indices i, y, and k take on the values x, y, and r; we note that this form
satisfies our assumption that P(0) = 0. The specific values of the coefficients a„, Q„t,
will depend upon the particular dielectric involved. If the second-order or higher terms
in the components of E are required to describe the material adequately, the dielectric
is called nonlinear. lt requires experiment to determine whether (10-48) is necessary in a
given case; for example, some ceramics fall into this category. We will not consider
nonlinear dielectrics further, but now restrict ourselves to cases for which only the first
term in (10-48) is required; such materials are called linear dielectrics.
 0-6 CLASSIFICATION OF DI ELECTR I CS 155

3. Linear Dielectrics
In this case, the general expression relating the components of P to the components of
E can be written in the form

(10-49)

where the proportionality factors x:, are called components of the electric susceptibflity
tensor. We have introduced the factor eg so that the y„ will be dimensionless as can be
seen from (10-40). In general, the y„ need not be constants but may be a function of
position within the material. The y„ cannot depend on E because that would bring us
back to the nonlinear case of (10-48). We see from the form of these relations that P
will not be parallel to E even in linear dielectrics, nor will D generally be parallel to E.
This situation is quite common in crystals and leads to such phenomena as double
refraction. be can now proceed to our next simplifying assumption.

4. Linear Isotropic Dielectrics

We now assume in addition that at a giaen point the electrical properties of the
dielectric are independent of the direction of E; such a condition is known as isotropy.
Since one direction is completely equivalent to any other, P must necessarily be parallel
to E the ;t = 0 if i4 y, and x - x = y ., so that (10-49) can be written in terms
of a ’single ’factor of proportionality as” ’“
' X e* o (10-50)
where y, is called the electric susceptibility. Combining (10-50) with (10-40), we find
that
(10-51)
where
x = 1 + x. = dielectric constant — relative permittivity (10-52)
i = x c = (absolute) permittivity (10-5a)
The quantities z „ x „ and r wiJ1 characterize the electrical properties of the material
and are to be found by experiment; their values can be found in many tables. For all
known substances, y is positive for static fields and therefore a, > 1. We see from (10-
51) that D and E are parallel in this situation. The equation D = e E is called a constitutive
equation; it is not a fundamental equation of electromagnetism but is applicable only
where it turns out to be applicable, so to speak.
In this case of linear isotropic dielectrics, we can also find the differential equation
satisfied by the scalar potential. Using (5-3), we can write (10-51) as D = — e 4Q, and if
we substitute this into (10-41), and use (1-115) and (1-45), we obtain
(10-54)
as the equation we would have to solve for ‹p since we still must allow for the possibility
that c is a function of position, and we can go no further ufltil this dependence is
known.
Our next simplification leads to such an important situation that we will allot it its
own section.
 156 EtECIROSTATlCS lN THE PRESENCE OF MATTER

10.7 LINEAR ISOTROPIC HOMOGENEOUS (I.i.h.) DIELECTRICS

We now assume in addition that the electrical properties are independent of position;
such materials are called electrically homogeneous. Generally gases and liquids, as well
as many solids, fall into this category so that it is not such a special situation as one
might think. Then the quantities x., • „ and c are covszanii; they are still characteristic
of the material, however. Equations 10-50 through 10-53 are still applicable and, in
addition, (10-54) simplifies to
'2 — (10-55)
since now be = 0. By comparison with (5-15), we see that for 1.i.h. dielectrics the
potential Q again satisfies Poisson’s equation but with ‹ replacing c and p replacing
the total charge density p. (The bound charges of the material haven’t disappeared, of
course, but their effect is now summarized in the factor e.) The fact that $ satisfies the
equation (10-55) also means that for 1.i.h. dielectrics we can, with care, bodily take over
solutions we have previously found for the vacuum case and simply replace c by e ; we
will do little of this, however. Also, if p —- O, then Q will again satisfy Laplace's
equation W 2$ = 0 in this region.
The boundary conditions at a surface of discontinuity can now be expressed
completely in terms of E. If we insert (10-51) into (10-42), and remember that (9-21) is
still applicable, we obtain

(10-J6)

But now we see that, even if a/ = 0, the normal components of E will not generally be
continuous across the bounding surface separating two dielectrics, so that, as shown in
Figure 10-14, the direction of E can change at the boundary. Thus, the lines of E can be
refracted even in the absence of a free surface charge density and a 2 will be different

I n a l.i.h. dielectric, the bound and free charge densities are related in a simple way,
as are the polarization and the displacement. If we eliminate E between (10-50) and (10-
51), and use (10-32), we find that
X‹ (x— 1)
P = D= D (IU-57)

Figure 10-14. Electric fields at a

boundary between two dielectrics.
 10-7 LINEAR ISOTROPIC HOMOGENEOUS (I.i.h.) DIELECTRICS 157

which shows us that P and D are also parallel and that |P| < |D|. Now if we take the
divergence of (10-57) and use (10-10) and (10-41), we get
(«,—1)
pt — — (10-58)
so that p p < pdf . If we insert this result into (10-38), we find that the total charge
density in a l.i.h. dielectric can always be written as

(10-59)

which shows us that the total charge density is always less than the free charge density
since x > 1. As a special case, we see that if p/ = 0, then pp — 0, so that at any point
in a l.i.h. dielectric where there is no free charge density, there is no bound charge
density either.
From now on, we will assume that we are dealing with l.i.h. dielectrics in our
examples and exercises unless we explicitly specify otherwise. At this point, we are able
to discuss some examples quantitatively; we begin with the capacitor, which we first
considered qualitatively in Section 10-3.

Example
Parallel plate capacitor ivf?h changes constant. In Figure 10-15, we illustrate a capacitor
of total free charge Q/ w'ith a vacuum between the plates in (a) and with a dielectric
completely filling the region between the plates in b1. The directions of the various
field vectors are also shown. The vacuum value of the electric field was discussed in the
text immediately preceding (6-40) and was found to be E = o//c where ay = Q /A is
the free surface charge density and A is the plate area. Since Pg = 0, we find from
(10-40) that the displacement D is
D —— c z Fz = « (l 0-60)
Now @¿ and e/ are kept constant when the dielectric is put between the plates, so that
D will not be changed and will equal the vacuum value:
D — D —— o (10-61)
This result is also consistent with (10-42) since the fields are zero inside the conducting
plate and therefore D2. — D, —— D — 0 = «y. The electric field has changed, however,
since we find from (10-51), (10-61), and (10-60) that
D D r o
E= —= = (10-62)

e,

Figure 10-1 S. Parallel plate capacitor of constant charge. (a) Vacuum between the plates.
(6) Dielectric between the plates.
158 E L ECTROSTATICS IN THE PRESENCE OF MATTER

and therefore E < r 0 in agreement with (10-24). We see now that the factor by which
the electric field is reduced is exactly equal to the relative permittivity. The potential
difference now is

(10-63)

with the use of (10-62) and (10-22). Thus, the potential difference is less than the
vacuum value by the same factor < so that AQ < fig in agreement with the
experimental result. We can find the capacitance by expressing (10-63) in terms
of the total charge:

and therefore
A
- •,C (TO-64)
d
where we have used (10-20) and (6-41) to identify the vacuum value of the capacitance
Cl. Thus, we see that the presence of the dielectric has increased the capacitance, in
agreement with (10-21), and the ratio of the capacitance with and without the dielectric
is exactly equal to the dielectric constant, that is, CCCI = < .
The polarization as found from (10-50) and (10-62) is

(10-65)

With the further use of (10-62), the third form can be rewritten as
P —— ‹ (E - E) (10-66)
so that

E —— Ez — (10-67)

which also follows from (10-40) and (10-61), that is D —— ‹ E + P - D = e ñ . This

latter result has exactly the form we previously deduced in (10-25) and now we have a
quantitative expression for Ep. Since P is constant,q - P - 0 and there are no bound
volume charges in agreement with (10-58) since the only free charges are on the
surfaces of the conducting plates. However, there are bound surface charges on the
dielectric surfaces. Their magnitude as found from (10-8), (10-65), and (10-60) is

(10-68)

and the signs are exactly those already illustrated in Figure 10-6. We see from the figure
that the sign of at is always opposite to that of the immediately adjacent at; taking this
into account we can write (10-68) as
c — 1
(10-69)

which is co r letely analogous to the result previously found for volume charge
densities in (10-58). These bound surface charges, acting as two plane infinite sheets,
will produce the field Ep. We can calculate the magnitude of E by using our previous
 0-7 LINEAR ISOTROPIC HOMOGENEOUS (I. i. h.) DIEL ECTRICS 159

result (3-12) along with (10-68), and we find that

Ez (10-70)
2( o °o ^o ^o
This agrees exactly with (10-67), which we found by other means, and verifies our
previous analysis which led to the form E —— rd — Ez given in (10-25). This field E is
often called the local field and one says that the resultant field E is the sum of the
vacuum field Ed, produced by the free charges as if there were no matter present, plus
the local field E;q = Et produced by the bound charges arising from the polarization
of the dielectric. •

Example
Capacitance in general. Although we obtained (10-64) by considering the specific case of
the parallel plate capacitor, the simplicity of this result, and the fact that C = x Cl no
longer contains any characteristics of the parallel plate capacitor, suggests that this may
actually be a general relation that holds for any capacitor. This turns out to be the case.
If we combine (10-41) and (10-51), we get

(10-71)
since c is constant for a 1.i.h. dielectric. We also have w x E = 0 since E is still
conservative. Suppose we are given a certain free charge distribution described by p/. If
there is no matter in the region of interest, we solve the pair of source equations to find
an electric field Ed. Now if we fil1 all of the region with a dielectric described byr „
and if we keep p/ unchanged, then we see from (10-71) that the problem is exactly the
same as the vacuum case we solved, except that the source charges are smaller
everywhere by the factor x . Then, as we can see from (3-3) for instance, the field E we
now get will also be smaller by this same factor, that is,
E
E=

is a general result. The potential difference fiQ between the capacitor plates as given by
(6-38) will be

(10-72)

as in the special result (10-63). We note that this result is consistent with our comments
following (10-55).
Since Q/ is the same in both cases, the capacitance will be

" e’0 (10-73)

thus showing in the general case that the capacitance of any capacitor is increased by
the factor x, when all the space between its plates is filled with a dielectric. The use of
this result provides a convenient way of measuring c .
If the dielectric is not homogeneous, or if not all of the space is filled with dielectric, (10-
73) is not generally true. One can often treat such problems by writing the potential
difference as

(10-74)
160 E L ECT ROST ATICS I N TH E PRESE NCE OF MATTER

Then it is usually possible to express D in terms of the total free charge Q by means of
(10-41), or by (l 0-43) if the problem has enough symmetry, so that when the integral is
evaluated, the capacitance can be found from (6-38). In such a case, the boundary
condition (10-42) often is of great help. Many of the exercises involve these considera-
tions. ■

Eaample
Point charge in an irifinite dielectric. Suppose a point charge q is embedded in a
dielectric as shown in Figure 10-16. The field of this charge will polarize the dielectric.
II the dielectric has a finite size, the bound charges on the surface will contribute to the
resultant field and the problem of calculating the field everywhere could be quite
complicated. However, if the dielectric is of infinite extent, the effect of any bound
surface charges on the outer surface can be neglected and we can assume spherical
symmetry. Then we can write D - D(2t)R and use Gauss’ law for D as given by (10-
43). If we integrate over the sphere of radius fi shown dashed in the figure, we get, in
the by-now-familiar way,

so that
gR
D and E- (10-75)
' 4
2
and, as expected, the electric field is just that of a point charge reduced by the factor
e e = x,. If we now imagine a point charge q’ placed at R, the force on it will be
qq’R
F*=g*E= (10-76)
which is simply Coulomb’s inverse square law with ‹ replacing e .
This result (10-76) is the basis for a common statement that the presence of a
dielectric decreases the forces between charges by the factor c/e = x,. We see now,
however, that this is true only for a l.i.h. dielectric of infinite extent, or one so large that
the bound surface charges will not affect the field and we can safely assume spherical
symmetry. In fact, it is possible for the force on q’ to actually increase under the right
circumstances.
The polarization of the dielectric can be found from (10-57) and (10-75) to be

P —' (J 0-77)

Figure 10-16. Point rharge in an infinite

’ dielectric.
 10—8 E NERVY 161

and is directed radially from q. According to (10-58), the bound charge density should
vanish. It follows from (10-10), (10-77), and (1-145) that this is indeed the case. •

10.8 EMERGY
We recall our result (7-10) for the energy of a system of charges:

(10-78)
When we obtained this expression, we calculated it as the amount of reversible work
required to assemble a gjven configuration of charges, and we did not make any
distinction between free and bound charges, nor was it necessary to do so. The
implication was that this energy could be retrieved from the system as reversible work,
and in that sense it is appropriate to regard it as potential energy as we did. Now when
we consider charges in the presence of matter, we have to ask ourselves: what is the
useful definition of energy? Again, this must be the work required for us to put together
an assembly of charges ouer which we haae control and thus energy that can, in
principle, be retrieved from the system. But the only charges over which we have any
sort of control are the free charges and it is only these in a broad sense that we can use
to store or extract energy, as in the process of charging or discharging a capacitor in a
reversible manner. Hence, the energy of interest to us is that of the free charge
distribution in the presence of matter. Failure to remember the restrictive nature of this
definition leads to, and has led to, a con.siderable amount of confusion and argument
about energy relations when matter is involved.
In order to appreciate the limitations of our final result, it will be useful to begin
again and (5-48), which gives the energy of a point charge in terms of the potential,
provides a convenient starting point. We assume that we have some preexisting
situation and increase the free charge density in a volume element di by the amount
hps. Then bp d r is small enough to be treated as a point charge and, by (5-48), the
energy of the free charge distribution will change by
bU —— hps d i (10-79)
The bound charges of matter are included, of course, since Q is the potential resulting
from all charges; for example, it could be evaluated by substituting (i 0-38) into (5-7).
As usual, we are assuming rigid mechanical supports or constraints on the charge
positions so that no purely mechanical work is done. If we now integrate over all space,
we get the total change in the electrostatic energy:

(10-80)
During this process, D will have changed and by (10-41), we initially have w - D = p
and, finally, w ' (D + 6 D) = pt + hps, so that by using (1-114), we getq - 6 D = b p/.
Then (10-50) becomes
(10-8J)

If we now compare this expression wi th (7-22), we see that they are of exactly the same
form with fi D replacing 1 i E. The arguments that led us from (7-22) to (7-28) can be
applied in a completely analogous manner, so that (10-81) can also be written as

(10-82)
162 r LECTROSTATICS IN THE PRESENCE OF MATTER

Finally, if we take the zero of energy to correspond to D = 0, we can obtain the total
energy U by integrating this from the initial state of D = 0 to the final value:
U
- «ii s «c‹ o‘ E’ 6D d (10-83)

In general, (10-83) cannot be evaluated further until the dependence of E upon D is

known, and this could be quite complicated. However, for the important case of linear
isotropic dielectrics, we can use (10-51) to write E 6 D = D- 6D e = 6(D 2,Z2 r ) so
that
D’ D 2
2c = 2c - 'D- E
and thus (10-83) becomes
. E di
°-—— a. „.gg}D (10-84)

and gives the total reversible work done on the free charges. We can interpret (10-84)
just as we did (7-28) by introducing an energy density u given by
2
D
u, — -D E = 2e E (10-85)
2 ' 2e

for linear isotropic dielectrics.

Example
Spherical capacitor. We imagine all space between the plates of the capacitor shown in
Figure 10-17 to be filled with a dielectric of permittivity e. The total free charge on the
capacitor is @. As usual, D will be radial because of the spherical symmetry, so that } D}
will be constant on the surface of the dashed sphere of radius r. Then, if we apply (10-
43) to this sphere, we get, as in the last section:

U D- dv —— 4wr 2D —— Q , —— Q
S
2
so that D —— O,Z4 or and (10-85) gives

(10-86)
*° ' 32 u 2cr 4

Figure 10-17. Spherical capacitor with dielec-

tric between the plates.
 4 0 8 £ N fi P Y J63

This applies to the region between the spheres; u = 0 elsewhere since the fields are
different from zero only in the region occupied by the dielectric. Inserting (10-86) into
(10-84), and using (1-99), we find the total energy of the capacitor to be
2 " b Q 2
1
2 r 2 Sin 8 dr
d8 d g —— — — — (10-87)
’' 32 £ 0 ’0 ’o T
8uc a b
If there were a vacuum between the plates, the energy t/h would be given by (10-
87) with e replaced by e . Therefore, since cue - c„ we see that

U, —— t/„
(10-88)

so that U < t/, and the total energy is decreased by the presence of the dielectric in
this case in which Q is kept the same.
In (7-21), we found that the energy of a capacitor is equal to O 2 2C. This result was
obtained by using very general properties of the charge distributions on conductors,
and thus is generally true whether there are dielectrics present or not; this same result
was also obtained by using only general properties of work and potential difference in
Exercise 7-3. Thus, we can still apply energy methods to the calculation of capacitance,
and this is sometimes more convenient. In this example, if we equate (10-87) and (7-21),
we find the capacitance to be
9 rrcab

with the use of (10-53), and where C is the vacuum value as given by (6-37). As it
should, this specific result agrees with our general conclusion about the effect of a
dielectric on capacitance as expressed in (10-73). ■

In general, and as we just saw in the last example, the presence of a dielectric can
alter the values of D and E everywhere so that the total energy as given by (10-84) can
be expected to change. The exact amount of this change will normally depend on the
particular manner in which the process is carried out; for example, charges kept
constant, potentials kept constant, only part of space filled with the dielectric, more
than one dielectric used and so on. As a result, the general problem of energy involving
dielectrics can be quite complex, and it is not always possible to make an unambiguous
assignment of the energy change to specific parts of the system. There is one compara-
tively simple case, however, in which it is possible to ascribe the energy change to the
dielectric itself, and we shall consider it as an illustration.
Let us assume that initially we have vacuum everywhere and some distribution of
charges which result in the fields Ed and D = c Ed throughout all space. The initial
energy U can then be evaluated from (10-84) as

U ' ' a D’ ’ E' d (10-90)

I space
Now let us assume that we keep the source charges fixed in values and locations and
introduce a dielectric of volume P" into this preexisting field Ed. (This is in contrast to
the previous example where all of the region containing the already existing field was
filled with a dielectric.) As we know, the presence of a dielectric will generally change
the values of E and D everywhere and the new energy U can be found from (10-84) by
using these new values of the fields. The change in the energy, U — It q, in this case,
can be ascribed entirely to the presence of the dielectric, and if we call it U z, we havc
164 t r ECT ROSTATICS I N THE PRESE NC E OF MA7TER

It can now be shown that under these conditions, (10-91) can be written as an integral
over the volume K of the dielectric only. Since this is a rather lengthy calculation, we
have to leave it to an exercise and simply quote the final result, which turns out to be

= —' (10-92)

Since this involves only the volume of the dielectric, it is reasonable to consider it to be
localized irt the dielectric and to represent the energy o/ the dielectric. Thus we can
introduce an energy density for these bound charges, u,t, given by

where we also used (10-50). These expressions (10-92) and (10-93) are thus appropriate
for a situation in which the polarization is thought of as being produced by the field.
In all of our discussion so far, we have assumed that <, is constant during these
processes. It turns out that many dielectrics have a <, that depends on the temperature,
as we will see in Appendix B, so that here we are effectively assuming isothermal
processes. This is consistent with our emphasis on U as being related to the reversible
work, so that it is really more analogous to the Helmholtz function or free energy of
thermodynamics, but for systems at constant temperature, there is no distinction
between their changes. With this in mind, we can look on (10-92) as a contribution to
the internal energy of the dielectric system.
We should also remind ourselves of the distinction between these energies and the
interaction energy of a charge distribution and an external field, which we discussed in
Section 8-4. In particular, we obtained (8-64) for the energy of a dipole. If we want to
apply this to a polarized material, we will have to assume that the polarization is either
a permanent one or that the external field is small enough so that it does not affect P
appreciably. Then the dipole moment of a small volume will be given by (10-1) and the
external interaction energy as obtained from (8-64) can be written as
dU , e „ - — P- E„, dz (10-94)
where we write the external field as E„t, rather than Eg that we have been using
recently to represent vacuum values. Thc total intcraction energy will then be obtained
by integrating (10-94) over the dielectric to give

(10-95)
For example, if Ee„ does not vary much over the volume, we can take it out of the
integral and use (10-2) to get U e„ = — p- E„, in agreement with (8-64).

Example
Energy of a capacitor in general. We have already seen the cffect of a dielectric on a
capacitor in the special case by which we obtained (10-88). Let us now briefly look at
the general case. The energy is general is given by (7-21). In addition, we leave ob tained
the general relation (10-73) for the effect on the capacitance. Therefore, if the charge Q
is kept constant, the energy U with the dielectric between the plates will be U —
O 2 ,r2 C = O 2,Z2a C, = U,/<, where Uz is the vacuum energy, so that the decreased
energy given by
U’
U — ( Q —— const.) (10-96)

is a completely general result, in agreement with (10-88).

 Now let us suppose, instead, that the potential difference fiQ is kept constant.
Combining (7-21) with (10-73), we find that t/, = 2
that
U —— •,U ( fiQ = const.) (10-97)
and the energy of the capacitor has increased by the factor x „ in contrast to (10-96).
The reason for this is that, with the increase in capacitance, the charge has also been
increased and the battery had to do work on these free charges in order to separate
them. In addition, it had to polarize the dielectric and the net effect of all of these
changes is just that described by (10-97). u

40-9
FORCES
When a dielectric is polarized, the resultant bound charge densities will have forces on
them due to the electric field. This general subject of forces on dielectrics, and on
conductors in the presence of dielectrics, is really quite complex and it is easy to get an
incorrect answer. Generally speaking, the only satisfactory way of handling these and
similar problems is to use “energy methods,” that is, by comparing the energy of the
initial configuration of the system with that of the final configuration. Usually such
problems fall into two classes: (1) the system of interest is completely isolated and thus
its total energy is conserved or (2) the system is not isolated and one has to consider the
possibility of energy transfer between it and external energy sources such as batteries;
thus the system energy will not be conserved, although that of the system plus battery
will be. Ordinarily, these two classes will correspond to our previous cases of constant
free charge or constant potential difference.
Because of these considerations, we will content ourselves here with discussing only
two of the simplest situations, although other common examples will be found in the
exercises. Furthermore, we will not discuss the balance between electrical and mechani-
cal forces that is required to maintain mechanical equilibrium within thc system, or to
produce a new one when electric fields are applied. If the dielectric is not rigid, it will
generally deform under the influence of these electrical forces. This phenomenon is
called electrostrictiori,- it is generally a small effect, but can be of interest and
importance under some circumstances.

Example
Average surface force on a dfelectric. Let us consider the particular case of a capacitor.
As we saw in (10-96), the energy of the capacitor for constant charge will be decreased
when the dielectric is present. Since the general tendency of systems is to reduce their
energy, we see that the capacitor will “want” to have the dielectric in place. In other
words, there should be a force on the dielectric whose direction is such as to pull it into
the region between the plates.
In order to be specific, we consider a parallel plate capacitor with square plates of
side L so that A —— L 2. We also assume we have a solid slab of dielectric of the correct
size to just fit between the plates. We neglect edge effects and assume the field to be
different from zero only in the region between the plates. In Figure 10-18, we show a
side view in which the dielectric is partway between the plates. With the dielectric
completely in place, the total change in energy can be found from (10-96) and is

ITU —— U — t/0 = (10-98)

166 EN ECTROSTATICS IN THE PRESENCE OF MATTER

Figure 10-18. Force on a

dielectric.

We let ( I) be the average force on the dielectric; the total displacement of the
dielectric is L, so that using (7-37) we get

(10-99)

which is positive, showing that the dielectric will be pulled into the regjon between the
plates. If we express this in terms of the original constant energy density u , we find
from (10-84) and (10-85) that tr0 = u 0 (volume) = u L 2d so that

Now M is the area of the face of the dielectric upon which F) is acting, so that if
(/ ) is the average force per unit area we get

(10-100)

as expressed in terms of the initial vacuum situation.

Example
Capacitor immerseJ in a fiuid dielectrfC. As an example of the force on a conductor in
the presence of a dielectric, we consider a plate of a parallel plate capacitor for which
all of the region where the field is different from zero is occupied by a dielectric. In
Section 7-4, we evaluated this by looking at the energy changes produced by and the
force required to change the plate separation by a small amount. Our treatment was
general enough so that we can use the same results and need only supplement them
with our knowledge of the effect of a dielectric on capacitance as given by (10-73), that
is, C = c C . If the charge is kept constant, the total force on the plates in the presence
of a dielectric will still be given by (7-39), and we get

( cOHSt.) (10-101)

in terms of the forceI q when there is a vacuum between the plates. Thus the force on
the plates has been reduced by the factor c . In using C’ —r C0 in this way, it is
essential that the dielectric completely fill the region between the plates, as will be
illustrated by some of the exercises. This means that we are in effect assuming the
dielectric to be a fluid and that the capacitor is immersed in the fluid. Then, if the
separation is decreased, the fluid can move out from between the plates, while if the
separation of the plates is increased, the fluid can move in to fill the regjon and there is
also fluid available for this purpose.
 EX ER SISES J67

If the potential difference is kept constant, the middle form of (7-46) is appropriate
to use, and we get

F 2 = c,F ( fi$ - const.) (10-102)

dx
and, in this case, the force on the plate is increased by the factor x .
Even though we realize that there are different processes involved, it still may seem
somewhat paradoxical that the presence of the dielectric decreases the total force in one
case while increasing it in the other. If we recall (7-37) and (7-45), we see that we should
be able to understand these results somewhat more by looking at the total field energy.
We found in (10-85) that the energy density is given by u = *D- E. If the free
charge Q is kept constant, then D will be constant and we find that u, = D 2 2e =
u c . The use of (10-84) then gives U —— U Cfc and the force will decrease by the
factor ce as in (10-101). Similarly, if the potential difference AQ is kept constant, then E
will be constant, and we are led to u = }c E 2 = x,u . Thus, in this case, we will find
that U —— c U 0 which corresponds to the increase in the force described by (10-102).
[We note that these results for the capacitor energy are exactly those we found by a
somewhat different approach as given by (10-96) and (10-97).]
Further study as to exactly how these mechanical forces are affected by the presence
of a dielectric shows that the pressure distribution within the fluid is altered in the
presence of an electric field. The resulting pressure changes can be shown to account in
detail for the changes in the force on the conductor. ■

EXERCISES
101 The permanent dipole moment of the water Similarly, find the net gain due to motion of
molecule is about 6.2 x 10 ’° coulomb-meter. What is the negative charge of density — pq (Why7) and aver-
maximum polarization possible for water vapor at 100° C age displacement R_ . Combine the results for
and atmospheric pressure? all faces found in this way and thus show that
102 Electrostatics began with the observation the net gain of charge per unit volume is p -
that charged objects attracted small pieces of —W - P.
matter. Describe qualitatively how a charged body
can exert a net force on neutral matter and show
IT-5 Find the potential Q and E on the axis
produced by the uniformly polarized sphere dis-
that the force will be attractive as observed.
cussed in Section 10-4 for negative values of z.
103 A slab of material has parallel faces. One Show that your answers are consistent with the
coincides with the xy plane while tire other is results found forr » 0 and with Figure 10-11.
given by z —— t. The material has a nonuniform
polarization P - P(I + or)1 where P and o are 10-6 Find the total posi0ve bound charge of the
constants. Find the volume and surface densities uniformly polarized sphere of Figure 10-8.
of bound charge. Find the total bound charge in a IT-7 A sphere of radius o has a radial polariza-
cylinder of the material of cross section A and tion given by P = or"é where a and n are con-
sides parallel to the z axis and thus verify directly stants and n z 0. Find the volume and surface
that (10-13) holds for this case. densities of bound charge. Find E outside and
10-4 Consider a parallelepiped of volume inside the sphere. Verify that your results for E
k x b y b z fixed within a dielectric. If p is the sati.sfy the appropriate boundary conditions. Find
average positive bound charge density, and R, is Q outside and inside the sphere. Sketch your
the average displacement of these charges when results for E and Q.
the material is polarized, show that the net gain of 1lL8 Repeat Exercise 10-7 for the case n -- — 1.
positive charge through the faces parallel to the Also show that n = — 2 is not a possibility be-
yz plane will be cause of (10-13).
8
10-9 A cube of edge 2 « has its faces perpendic-
ular to the xyz axes and the origjn at its center. It