Chapter 5

IMAGES AND GREENS

FUNCTIONS
5.1 Introduction
The potential due to a point charge placed above a grounded conducting plane can be found as
a sum of two potentials, one due to the charge itself, and the other due to an image charge, ,
placed at the mirror point with respect to the plane. This simple example indicates a possibility
of greatly simplifying boundary value problems by introducing appropriate images for a given
boundary shape.
In this Chapter, image charges for simple boundary shapes will be discussed. Finding an
appropriate image enables us to derive a formula for the potential due to a prescribed boundary
potential on the surface without necessarily going through the methods developed in Chapter 3. As
will be shown, solving a boundary value problem is essentially identical to nding an appropriate
scalar function (Greens function) for a given boundary shape.
5.2 Image for Flat Surfaces
Consider the conguration shown in Figure 5.1. The grounded, wide, conducting plane can be
replaced by an image placed at the mirror point of the real charge + as long as the potential
and electric eld in the space above the plane are concerned. The potential above the plane is thus
given by
=
+
+

=
1
4c
0
_

r
+


r

_
(5.1)
where r
+
and r

are the distances from the observing point to the respective charges. In Eq. (5.1),
the rst term in RHS is the contribution from the charge . The potential
+
satises the singular
Poissons equation
r
2

+
=

c
0
c (r r
0
) (5.2)
1
image charge
conductor
q
-q
d
d
r
r
+
-

Figure 5-1: Image charge for a at conducting plate.

where r
0
= de
z
is the location of the charge . The image charge is below the plane. Therefore,
in the space above the plane, the potential

due to the image charge should satisfy the Laplace

equaion without any singularities,
r
2

= 0 (. 0) (5.3)
In other words, the potential

can be regarded as a general solution to the singular Poissons

equation in Eq. (5.2) which is added in order that the total potential satisfy the boundary condition,
= 0 on the plane (5.4)
Obviously, the solution
+
is the particular solution to the Poissons equation. Note that we have a
freedom to add general solutions satisfying the Laplace equation to a particular solution satisfying
the Poissons equation.
Images for a grounded wedge can be similarly found. For example, if the wedge angle is 90

,
three images appear as shown in Figure 5.2. The potential is given by the sum of four contributions
from each charge. The positive image charge is the image of the negative image charges.
The potential due to a charge placed above a dielectric body can also be worked out by images.
In this case, we may assume the potential in the air

>
=
1
4c
0
_

r
1


0
r
2
_
(5.5)
and the potential in the dielectric

<
=
1
4c
0

00
r
3
(5.6)
where the distances r
1
, r
2
and r
3
are indicated in Fig. 5.3,
0
is the image charge in the dielectric,
and
00
is another image located at the same position as the real charge . Note that the potential
in the dielectric remains regular (no singularity). To determine
0
and
00
, we rst impose the
2
q
q
-q
-q
90 degree wedge
Figure 5-2: Images for 90 degree conductor wedge.
continuity of the potential at the boundary, r
1
= r
2
= r
3
. This yields
1
c
0
(
0
) =
1
c

00
(5.7)
The second boundary condition is the continuity in the displacement vector, 1
z
(normal compo-
nent). Since
r
1
=
_
r
2
+ (. d)
2
(5.8)
we have
0
0.
_
1
r
1
_
=
. d
[r
2
+ (. d)]
3=2
(5.9)
Similarly,
0
0.
_
1
r
2
_
=
. +d
[r
2
+ (. +d)
2
]
3=2
0
0.
_
1
r
3
_

. d
[r
2
+ (d .)
2
]
3=2
(. < 0) (5.10)
Therefore, the displacement vector just above the boundary surface becomes
1
z1
=
d
4(r
2
+d
2
)
3=2
(
0
) (5.11)
and that just below the surface is
1
z2
=
d
00
4(r
2
+d
2
)
3=2
(5.12)
From 1
z1
= 1
z2
, we obtain

00
=
0
+ (5.13)
3
q
q'
q''
r
r
r
1
2
3

(z < 0)
(z > 0)
Figure 5-3: Images
0
and
00
for a at dielectric boundary.
Solving Eqs. (5.7) and (5.13) for
0
and
00
, we nd

0
=
c c
0
c +c
0
(5.14)

00
=
2c
c +c
0
(5.15)
5.3 Image for Cylindrical Surfaces (Two Dimensional)
The potential due to a long line charge ` (C/m) placed parallel to a grounded conducting cylinder
can be found by the method of image. In Chapter 2, we saw that two opposite line charges of equal
magnitudes, +` and ` create a family of equipotential cylindrical surfaces,
(r
+
, r

) =
`
2c
0
ln
_
r

r
+
_
(5.16)
where r
+
and r

are the distances to the respective line charges, +` and `. Consider a line
charge at a distance j
0
from the axis of a cylinder having a radius a. A negative line charge `
at a distance j
00
= a
2
,j
1
and the positive line charge ` make the cylinder surface an equipotential
surface

s
=
`
2c
0
ln
_
a
j
0
_
(5.17)
(Check this by considering the potentials at and 1 in Figure 5.4.) In order to make the cylinder
surface at zero potential, we have to subtract this potential from the potential given in Eq. (5.16)
which is based on the choice = 0 on the midplane, r
+
= r

. Therefore, the solution to the

4
potential becomes
(r) =
`
2c
0
_
ln
_
r

r
+
_
ln
_
a
j
1
__
(5.18)
But
r
+
=
_
j
2
+j
02
2jj
0
cos c

1=2
(5.19)
r

=
_
j
2
+
_
a
2
j
0
_
2
2j
a
2
j
0
cos c
_
1=2
(5.20)
Substituting these into Eq. (5.18), we nally obtain
(j, c) =
`
2c
0
ln
_

_
_
j
2
j
02
a
2
+a
2
2jj
0
cos c
_
1=2
(j
2
+j
02
2jj
0
cos c)
1=2
_

_
(5.21)
When j = a, this indeed vanishes.
The surface charge density induced on the cylinder surface is given by
o = c
0
0
0j

=a
=
`
2
j
2
1
a
a
j
02
+a
2
2aj
0
cos c
(C/m
2
) (5.22)
Therefore, the total charge (per unit length) on the cylinder surface is

|
=
`
2
_
j
02
a
2
_
_
2
0
dc
j
02
+a
2
2aj
0
cos c
(5.23)
The integral reduces to
2
j
02
a
2
(5.24)
when j
1
a. Therefore, the charge per unit length of the cylinder is simply `, as expected.
5.4 Image for a Sphere
We consider a charge at a distance r
0
from the center of a grounded conducting sphere of radius
a. The potential is symmetric about the line connecting the center and the charge. Therefore, an
image charge should be on the same line. We assume a negative charge
0
placed at a distance r
from the center. The potential at the point is

A
=
1
4c
0
_

r
0
a


0
a r
_
(5.25)
5
a

'
''
r+
r-

Figure 5-4: A line charge ` at a distance j

0
from the center of a conducting cylinder of radius a
and its image line charge ` at j
00
= a
2
,j
0
make the cylinder surface an equipotential surface at

s
=

2"
0
ln(a,j
0
) .
This should vanish if the image is to replace the sphere without aecting the potential outside the
sphere. Therefore,

r
0
a
=

0
a r
(5.26)
Similarly, the potential at 1 is given by

B
=
1
4c
0
_

1 +a


0
a +r
_
= 0 (5.27)

r
0
+a
=

0
a +r
(5.28)
Solving Eqs. (5.26) and (5.28) for r and
0
, we nd

0
=
a
r
0
(5.29)
r =
a
2
r
0
(5.30)
Therefore, a charge a,r
0
placed at a distance a
2
,r
0
from the center can be used to replace the
sphere as far as the potential and electric eld outside the sphere are concerned. Outside the sphere,
the potential is given by
=

4c
0
_
1
r
+

a,r
0
r

_
(5.31)
6
a
r
r''
r'
q -q
r+
r-
A
B
Figure 5-5: Image for a sphere. Charge at distance r
0
from the center of a sphere of radius a and
its image at r
00
= a
2
,r
0
make the potential vanish on the sphere surface r = a.
where r
1
and r
2
are the distances between the observing point and respective charges, and given,
in terms of the coordinate (r, 0) at the observing point, by
r
+
=
_
r
2
+r
02
2rr
0
cos 0
_
1=2
r

=
_
r
2
+
_
a
2
r
0
_
2
2
a
2
r
0
r cos 0
_
1=2
(5.32)
The surface charge density on the sphere surface can be evaluated from
o(0) = c
0
0
0r

r=a
=
1
4
_
a
r
02
a
_

(a
2
+r
02
2ar
0
cos 0)
3=2
(5.33)
Therefore, the total charge residing on the outer surface of the sphere is given by the integral,

s
=
_

0
o(0)2a
2
sin0d0
=
_
a
r
02
a
_
a
2
_
1
1
1
(a
2
+r
02
2ar
0
j)
3=2
dj (j = cos 0)
=
a
r
0
=
0
This result is expected from the Gauss law because the
0
is the only charge enclosed by the
sphere surface.
Example
7
A charge is placed at a distance 1 from the center of an isolated (oating) conducting sphere of
a radius a. When the sphere carries no charges, what will the sphere potential be?
We rst make the sphere potential zero by placing a charge
0
= a,1 at r
0
= a
2
,1 as in
the preceding discussion. Since the sphere carries no net charge, if a charge +
0
= +a,1 is placed
at the center of the sphere, the sphere potential will be raised (when 0) to

s
=
1
4c
0
a,1
a
=
1
4c
0

1
(5.34)
Note that this is independent of the sphere radius a. When the sphere carries a net charge Q, the
sphere potential becomes

s
=
1
4c
0
_

1
+
Q
a
_
(5.35)
5.5 Greens Theorem and Greens Function
The method of image charges has a more important application in potential boundary value prob-
lems. When a potential is prescribed on a closed surface, it uniquely determines the potential in the
space surrounding the surface and also in the space surrounded by the surface. The most general
formulation for the potential is given in terms of a scalar function called Greens function, which
is closely related to the potential due to a charge and its image for a given surface shape.
For arbitrary scalar functions c and c, the following identity immediately follows from Gauss
theorem,
_
r (crc)d\ =
_
crc dS (5.36)
The LHS is equal to
_
_
rc rc +cr
2
c

d\ (5.37)
Therefore,
_
_
rc rc +cr
2
c
_
d\ =
_
crc dS (5.38)
Exchanging c and c, we obtain
_
_
rc rc +cr
2
c
_
d\ =
_
crc dS (5.39)
Substracting Eq. (5.39) from Eq. (5.38), yields
_
_
cr
2
c cr
2
c
_
d\ =
_
(crc crc) dS (5.40)
This is known as the Greens theorem. Its usefulness in potential problems can be appreciated as
follows.
8
In Eq. (5.40), we identify c as the desired scalar potential (r), and c as a solution to the
following singular Poissons equation
r
2
G = c
_
r r
0
_
(5.41)
Since the potential (r) satises Poissons equation,
r
2
=
1
c
0
j (5.42)
and
_
(r
0
)c(r r
0
)d\
0
= (r) (5.43)
Eq. (5.40) yields,
(r) =
1
c
0
_
G(r, r
0
)j(r
0
)d\
0

_
s
(
s
r
0
GGr
0

s
) dS
0
(5.44)
where
s
(r
0
) is the surface potential prescribed on o, and r
0
indicates derivative with respect
to the surface coordinates r
0
. When there are no surfaces, and the potential is entirely due to a
prescribed charge density distribution j, the Greens function is
G(r, r
0
) =
1
4
1
jr r
0
j
(5.45)
and the potential is given by the familiar form,
(r) =
1
4c
0
_
j(r
0
)
jr r
0
j
d\
0
(5.46)
In the absence of charges (j = 0), the potential is determined by the surface potential
s
and
its derivative, r
s
, which is of course related to the electric eld on the surface.
The Greens function in Eq. (5.45) is essentially the potential due to a charge located at r
0
.
However, in addition to this particular solution, any general solutions satisfying
r
2
G
g
= 0 (5.47)
can be added to the Greens function so that it vanishes on a given surface. In this case (G = 0 on
o), the potential due to a prescribed surface potential distribution becomes,
(r) =
_

s
(r
0
)
0G
0:
0
do
0
(5.48)
where n
0
is the normal vector on the surface, which is directed away from the region in which we
wish to evaluate the potential. (Fig. 5.6) This is known as the Dirichlets formulation for potential
boundary value problems.
9
Solving a boundary value potential problem is now reduced to nding a suitable Greens function
for a given surface shape, on which G = 0. This is where the method of image will be very useful.
Greens Function for a Sphere
Consider a spherical surface having radius a. We place a charge at r
0
. An image charge

0
= a,r
0
at
r
00
=
a
2
r
02
r
0
(5.49)
and the original charge at r
0
make the surface potential vanish. Therefore, the Greens function
for a sphere is immediately written as
G(r, r
0
) =
1
4
_
1
jr r
0
j

a,r
0
jr r
00
j
_
(5.50)
Noting,

r r
0

=
_
r
2
+r
02
2rr
0
cos
_
1=2
(5.51)

r r
00

=
_
r
2
+
_
a
2
r
0
_
2
2
a
2
r
0
r cos
_
1=2
(5.52)
Fig. 5.6 Geometry for the Dirichlet boundary value problems,
Fig. 5.7 Greeens function for a spherical boundary is essentially equivalent to the potential due
to a charge near a grounded conducting sphere of the same radius.
we nd
G(r, r
0
) =
1
4
_
_
_
_
_
1
(r
2
+r
02
2rr
0
cos )
1=2

a
r
0
_
r
2
+
_
a
2
r
0
_
2
2
a
2
r
0
r cos
_
1=2
_
_
_
_
_
=
1
4
_
_
_
1
(r
2
+r
02
2rr
0
cos )
1=2

1
_
r
2
r
02
a
2
+a
2
2rr
0
cos
_
1=2
_
_
_
When we wish to evaluate the potential outside the sphere, the normal derivative is given by
0G
0:
0
=
0G
0r
0

r
0
=a
=
1
4
_
_
_
_
a r cos
(r
2
+a
2
2ra cos )
3=2

r
2
a
r cos
(r
2
+a
2
2ra cos )
3=2
_
_
_
_
=
1
4
_
a
r
2
a
_
1
(r
2
+a
2
2ra cos )
3=2
(r a) (5.53)
For interior problems, the negative of the above equation is to be employed. Recall that cos in
10
Eq. (5.53) is given by
cos = cos 0 cos 0
0
+ sin0 sin0
0
cos(c c
0
) (5.54)
in terms of the angular locations of the vectors r = (r, 0, c) and r
0
= (r
0
, 0
0
, c
0
). Substituting Eq.
(5.53) into Eq. (5.48), we nally obtain the desired exterior potential,
(r) =
a
2
4
_
r
2
a
a
__

0
sin0
0
d0
0
_
2
0
dc
0

s
(0
0
, c
0
)
(r
2
+a
2
2ra cos )
3=2
(5.55)
Let us apply this formula to the problem of oppositely charged hemisphere which we solved in
Chapter 3. The surface potential
s
(0) is described by,

s
(0) =
_

_
+\ 0 < 0 <

2
\

2
< 0 <
(5.56)
and is independent of the azimuthal angle c. On the axis, 0 = 0, and the potential becomes
(.) =
a
2
4
_
.
2
a
a
_
2
_
_
=2
0
sin0
0
d0
0
_
.
2
+a
2
2a. cos 0
0
_
3=2

_

=2
sin0
0
d0
0
_
.
2
+a
2
2a. cos 0
0
_
3=2
_
= \
_
a
2
.
2
.
p
.
2
+a
2
+ sign .
_
(5.57)
where
sign . =
_
+1 for . 0
1 for . < 0
(5.58)
Note that the potential (.) must be an odd function of .. When . a, Eq. (5.57) may be
expanded as
(.) = \
_
3
2
_
a
.
_
2

7
8
_
a
.
_
4
+
_
(5.59)
Therefore, at an arbitrary location (r a, 0), the potential is given by,
(r, 0) = \
_
3
2
_
a
r
_
2
1
1
(cos 0)
7
8
_
a
r
_
4
1
3
(cos 0) +
_
(5.60)
which agrees with the previous result in Eq. (3.175). The interior potential (r < a) can immediately
be written down by inspection as
(r, 0) = \
_
3
2
r
a
1
1
(cos 0)
7
8
_
r
a
_
3
1
3
(cos 0) +
_
(5.61)
Greens Function for a Long Cylinder (Two Dimensional)
11
The image line charge and corresponding potential due to a long line charge placed parallel
to a grounded conducting cylindrical surface has been worked out in the preceding section. The
potential is given by

_
r, r
0
_
=
`
2c
0
ln
_
a jr r
00
j
j
0
jr r
0
j
_
(5.62)
where
r = (j, c) (observing point)
r
0
= (j
0
, c
0
) (location of the line charge)
r
00
=
_
a
2
j
, c
0
_
(location of the image)
Then, the Greens function for a cylinder can be identied as
G(r, r
0
) =
1
2
ln
_
a jr r
00
j
j
0
jr r
0
j
_
=
1
2
ln
_
j
2
+j
02
2jj
0
cos(c c
0
)

1=2
+
1
2
ln
_
j
2
j
02
a
2
+a
2
2jj
0
cos(c c
0
)
_
1=2
(5.63)
For exterior (j a), the relevant normal derivative is
0G
0:
=
0G
0j
0

0
=a
=
1
2
a
j
2
a
j
2
+a
2
2aj cos(c c
0
)
(j a) (5.64)
For interior, the negative of Eq. (5.64) is to be used.
It is left for an exercise to recover the potential due to a cylinder, its upper half being at +\
and lower half at \ , worked out in Chapter 3.
12