Chapter 2 Boundary-Value Problems in Electrostatics I

The correct Green function is not necessarily easy to be found.

3 techniques to electrostatic boundary value problems:
(1) the method of images
(2) expansion in orthogonal functions
(3) finite element analysis (numerical method)

2.1 Method of Images

The method deals with the problem of
point charges in the presence of boundary
surfaces, eg, conductors either grounded
or held at fixed potentials.
Infer from the geometry of the situation
that some suitably placed charges,

external

to the region of interest, can simulate the

boundary conditions. The charges are called

image charges.

2.2 Point Charge in the Presence of

a Grounded Conducting Sphere
A point charge is outside a grounded conducting
sphere. Find the potential (
x=a = 0 ).
By symmetry the image charge q' will lie on the ray
from the origin to the charge q, then

 x =
=

4  0 x  y x  y '
1


q'

4  0 x n  y n ' x n  y ' n '

4  0
 x = a =

q
a

q'

=

q'
y'

y
a

q
a n  y n ' / a

a
y'

y'=


q'

y ' n '  a n / y '

=0

a
, q ' = q
y
y

As q is closer to the sphere, the q' grows and moves out from the center of the
sphere.

q' is equal and opposite in and

The actual charge density induced on the surface of

the sphere can be calculated from the normal
derivative of
at the surface:

=0
=



x

x=a
2

a a  y 

4  a 2  y 2  2 a y cos  a 2 3

It is easy to show by direct

integration that the total
induced charge on the
sphere is equal to the
magnitude of the image
charge according to Gauss's law.
For the force on the charge q,
write down the force between
and q'

y  y ' = y 1
F=

q2

a2
2

y
ay

4  0  y 2  a2 2

The alternative method for the force is to calculate the total force acting on the
surface of the sphere

F=

2

d F=

d a=

q 2 y 2  a 2 2
2

32  0

cos

  y  a  2 a y cos 
2

The whole discussion has been based on

q is outside the sphere.

Actually, the results apply equally for q
the understanding that
inside the sphere.

d =

q2

ay
2

2 2

4  0  y  a 

2.3 Point Charge in the Presence of a Charged, Insulated,

Conducting Sphere
Consider an insulated conducting sphere with total charge

Q in the presence of

a point charge q.
Start with the grounded conducting sphere (with its charge q' distributed over
its surface). Then disconnect the ground wire and add to the sphere an amount of
charge (Q - q'). This brings the total charge on the sphere up to
The added charge (Q - q') will distribute
itself uniformly over the surface. Then

q

4  0
x =
x  y

x

Q

y2

x

The force acting on the charge q can be

written down from Coulomb's law

F=

4  0 y

Q

q a 3 2 y 2  a 2  y
y  y 2  a 2 2

Q.

In the limit of y  a, the force reduces to the usual Coulomb's law for two small
charged bodies. But close to the sphere the force is modified because of the
induced charge distribution on the surface of the sphere.
If the sphere is charged oppositely to q, or is uncharged, the force is attractive
at all distances.
Even if the charge
close distances.

Q is the same sign as q, the force becomes attractive at very

In the limit of Q  q, the point of zero force (unstable equilibrium point) is very
close to the sphere, at

 

y  a 1

1
2

This example explains why an excess of charge on the surface does not leave the
surface because of mutual repulsion of the individual charges.
As soon as an element of charge is removed from the surface, the image force
tends to attract it back.

2.4 Point Charge Near a Conducting Sphere at Fixed Potential

The potential is the same as for the charged sphere, except that the charge
(Q - q') at the center is replaced by a charge 40Va

 x =

4  0 x  y

aq

y x

a2
y

a
x

V F=

q
y2

aV

4  0  y 2  a 2 2 y

For 40aV  q, the unstable equilibrium point: y  a 1

q a y3

1
2

q
4  0 a V

2.5 Conducting Sphere in a Uniform Electric Field by Method of

Images
Consider a conducting sphere of radius a in a uniform electric field. A uniform
field can be thought of as being produced by appropriate positive and negative
charges at infinity.
The electric field near
the origin: E 0 

In the limit as

2  0 R 2

R, Q  

with Q/R2 constant, this

approximation becomes exact.
A conducting sphere is
placed at the origin, the
potential will be that due
to the charges and their
images

4  0
=


 R  2 r R cos 
Q
2

 R  2 r R cos 




aQ

R 2  a 4  2 a2 r R cos 
aQ

R2  a 4  2 a2 r R cos 

For

R  r: 4  0
=

2Q
R

 
r

a3

cos  

 

To the limit 2Q/40R2 becomes the applied uniform field:
=E 0 r 

cos 

The 1st term (-E0z) is the potential of a uniform field. The 2nd is the potential due
to the induced surface-charge density or, equivalently, the image charges.
The image charges form a dipole of strength D=

a
R

The induced surface-charge density

=0



r

Q2

a2
R

= 4  0 E 0 a 3

= 3  0 E 0 cos 

r =a

the surface integral of this charge density vanishes, so that there is no

difference between a grounded and an insulated sphere.

2.6 Green Function for the Sphere;

General Solution for the Potential
The potential due to a unit source and its image,
chosen to satisfy homogeneous boundary conditions,
is the Green function appropriate for Dirichlet
or Neumann boundary conditions.
For Dirichlet boundary conditions on the
sphere, the Green function for a unit
source and its image is

G  x , x ' =
=


ax'
 2
2
x  x ' x ' x  a x '
1

 x '  2 x x ' cos

a
x ' 2  a 4 2 x x ' a 2 cos

The symmetry in the variables is obvious, as is the condition that

x or x' is on the surface of the sphere.
For solutions of a Poisson equation we need not only

G
n'

x '=a

G, but also G/n'

a2  x 2

G = 0 if either

a  x 2  a 2  2 a x cos 3

The condition is essentially the induced surface-charge density.

The solution of the Laplace equation
specified on its surface is

 x =

outside a sphere with the potential

a  x2  a 2

a ,  ' ,  ' 

4
 x  a  2 a x cos 
2

d'

where cos = cos  cos  '  sin  sin  ' cos  ' 

For the

interior problem,

x=

G
n'

x ' =a

x 2  a2

a  x 2  a2  2 a x cos 3
2

a a  x 

a ,  ' ,  ' 

4
 x  a  2 a x cos 
2

d'

For a problem with a charge distribution, we must add to the potential integral
the appropriate charge density with the Green function.

2.7 Conducting Sphere with Hemispheres at Different Potentials

The solution for the potential is

 x ,  , =

2


4

4

2

a  x 2  a 2  d cos  '

a  x 2  a 2  d cos  '

  x  a  2 a x cos    x  a  2 a x cos 

d'

V a x a 

1

d '

d cos  ' [ x 2  a2  2 a x cos 3/ 2  x 2  a2  2 a x cos 3/ 2 ]

Because of the complicated dependence of angle,

the integral cannot in general be integrated
in closed form.
As a special case we consider the potential on
the positive

z axis

cos = cos  '

z = V 1

= 0
z2  a 2
z z a
2

=V

z =a

 3 V a2/ 2 z2 z  

In the absence of a closed expression for the integral, we can expand the
denominator in power series and integrate term by term
=

=

V a x 2 a 2 

2


4   x  a 
2

2 3

d '

ax
2

x a

d cos  ' [1 2  cos 3/ 2 1  2  cos 3 / 2 ]

in the expansion of the radicals only odd powers of cos will appear

1 2  cos 3 / 2 1 2  cos 3 / 2 =6  cos  35  3 cos 3 

2

0 d  ' 0 cos d cos  ' =  cos 

2
1

3
d

'
cos

d
cos

'
=
cos  3  cos 2 
0
0
4

 x ,  , =

3 a  x 2  a2  V
2  x  a 
2

2 3

[


 cos  1 
2

35 2
 3  cos2 
24

 ]

3 a
7 a 5
3
3
=
V
cos


cos


cos  
2
2
2 x
12 x 2
2




in 
in

x2

only odd powers of cos appear, as required by the symmetry of the problem.
For large values of x/a this expansion converges rapidly and so is a useful
representation for the potential.
It is easily verified that, for cos=1, this expression agrees with the expansion of
the expression for the potential on the axis.

2.8 Orthogonal Functions and Expansions

The orthogonal set chosen depends on the symmetries or near symmetries
involved.
Consider an interval (a, b) in a variable  with a set of real or complex

orthonormal functions Un(). The orthogonality condition is

b

a U n  U m  d  = m n
An arbitrary function can be expanded in a series of the orthonormal functions


f =  a n U n 
n =1


=

n =1

[a U
b


n

completeness of the function set

 '  f  '  d  ' U n 

a n =a U n  f  d 

 U n  '  U n =   '  

closure relation or completeness

n =1

The most famous orthogonal functions are the sines and cosines, an expansion
in terms of them being a

Fourier series.

The orthonormal functions are

the constant function is

sin

2 m x

a
a
for m = 0

2
a

cos

2 m x
a

a a
for x [ , ]
2 2

A function is customarily written in the form:



f  x =

2 m x
2m x
1
A0 
 Bm sin
Am cos
2
a
a
m =1

where

Am =

a /2


a

a / 2

f x  cos

2m x

dx,

Bm =

a /2


a

a / 2

f  x  sin

2m x

dx

Suppose that the space is 2d, and the variable  ranges over the interval (a, b)
while the variable  has the interval (c,d). The orthonormal functions in each
dimension are

Un() and Vm(). Then the expansion of an arbitrary function is

f  , =   a n m U n  V m 

n

For

a U


n

a n m =a d  c d  U n  V m  f  , 

U m  x  U  m , x 

 x Um  x d x 

 U  n , x  U m , x  d x = m  n

Fourier integral, start with: U m  x =

f  x =

a , b   , 
b


a

m =

Am e

2m x
a

Am =

a
1

2 m x
a

a / 2

m =0, 1, 2, 
x a / 2, a /2

a/2


a

i

m n

2m x'
a

f x ' d x '

a  

 k

a


f  x =
1

2 m

dm 

A k 



Am 



2 



2


2



A k  e

ikx

dk

d x = k  k '  ,

orthogonality condition 
The 2 continuous variables


2



dk


i k k ' x

A k =
1


2 



ei k x f  x  d x


2



ei k  x  x '  d k =   x  x ' 

completeness relation 

x and k are complete equivalent.

2.9 Separation of Variables;

Laplace Equation in Rectangular Coordinates
Equations involving the three-dimensional Laplacian operator are known to be
separable in 11 different coordinate systems.
discuss only 3 of these rectangular, spherical, and cylindrical.
The Laplace equation in rectangular coordinates:
assume
 x , y , z= X  x Y y  Z  z

1 d2 X

X dx

2

x


2

y

1 d2 Y
Y dy

2

z

1 d2 Z
Z dz

=0
=0

hold for arbitrary values of the independent coordinates, each of the 3 terms
must be separately constant:

1 d2 X
X dx

= ,

1 d2 Y
Y dy

=! ,

1 d2 Z
Z dz

=  !

i  x

= e

i ! y

  2  ! 2 z

By linear superposition, the solution can construct a very large class of solutions
to the Laplace equation.
To determine  and ! it is necessary to impose specific boundary conditions on
the potential.

To find the potential everywhere inside the box,

X = sin  x

 x = 0, y , z  = 0

 x , y = 0, z  = 0 Y = sin ! y

 x , y , z = 0= 0
Z = sinh   2  !2 z 
n
n =

 x = a , y , z=0
a

 x , y = b , z = 0
m
!m =
b

n m = sin  n x  sin  !m y sinh  2n  !2m z 



 x , y , z=

An m
n m

n , m=1


A n m sin  n x sin  !m y sinh  2n  !2m z 

n , m=1


 x , y , z = c = V  x , y

V  x , y=


n , m =1

An m =

a b sinh c  n  !m 

A n m sin  n x  sin ! m y sinh   2n  !2m c

0 d x 0 d y V  x , y sin n x  sin ! m y

If the rectangular box has potentials different from zero on all 6 sides, the
required solution for the potential inside the box can be obtained by a linear
superposition of six solutions, one for each side.
The problem of the solution of the Poisson equation, i.e., the potential inside the
box with a charge distribution inside, as well as prescribed boundary conditions
on the surface, requires the construction of the appropriate Green function.

2.10 A 2d Potential Problem; Summation of a Fourier Series

consider the solution by
separation of variables of the 2d
Laplace equation in Cartesian
coordinates, in which the
potential is independent of z,

~ ei  x e y

 " or #

Boundary conditions:

 x = 0, y=
 x = a , y= 0

 x , y = = 0

 x , y = 0= V
n

~ e y sin  x   =
a


=  An e
n=1

ny
a

sin

n x
a

n x
2 a
dx
 A n = 0
 x , 0 sin
a
a
4 V 1 for n odd
=
n  0 for n even


 x , y= 

n odd

4V 
e




y
a

4V 
e
n

n y
a

sin

n x
a

x
a
sin
for y &

a

x of the
asymptotic solution sets in for y
a,
regardless of the complexities of
(x,0).
the smooth behavior in

i

sin  = $ e 

= $  

n odd

4V
e
n

i n
xi y
a

4V
Z
$
  Z %e

n
n odd

i
xi y
a

Z2 Z3 Z 4
ln 1 Z = Z    
2
3
4


n odd

1 1 Z
Z
1 Z
= ln
= ln
n 2 1 Z
1 Z

1 Z
2V

 x , y=
$ ln

1 Z

Since the imaginary part of a logarithm is equal to the phase of its argument,

1 Z
1 Z

1Z2  2 i $  Z 
1 Z2

 x , y=

2V


 
sin

tan 1

the phase of the argument of ln

x
a

sinh

y
a

 
sin

0 ' tan 1

1 Z

= tan1

2 $Z
1Z2

x
a

sinh

1 Z

y

'


2

The real or the imaginary part of an analytic function satisfies the Laplace
equation in two dimensions as a result of the Cauchy-Riemann equations.

2.11 Fields and Charge Densities in 2d Corners and Along Edges

The geometry suggests use of polar
rather than Cartesian coordinates.
The Laplace equation:

1 
( (

 
(



(

1 2

2

( 

=0

Separation of variables:

( , = R ( ) 

2

( d
dR
1 d )
(

=0
R d(
d(
) d

( d
dR
1 d2 )
2,
(
= v
=v 2
R d(
d(
) d

v

R (= a (  b (
but for v = 0
) = A cos v  B sin v 

R (= a 0  b 0 ln (
) = A0  B0 

If there is no restriction on , it is necessary that v be a positive or negative

integer or zero to ensure that the potential is single-valued B 0 = 0
The general solution:

( ,  = a 0  b 0 ln (  [ a n ( n sin  n    n   bn (n sin n   ! n ]

n =1

If there is no charge at the origin, all the bn are zero. If the origin is excluded,
the

bn can be different from zero.

The logarithmic term is equivalent to a line charge on the axis with charge
density per unit length * = -20b0 .

( , 0=
 ( , ! = V b0 = B0 = b = A= 0, sin + != 0  +=

m

( , = V   a m (

sin

, m = 1,2, 

m
!

m =1

The undetermined coefficients

m

am depend on the potential remote from ( = 0.

For small enough ( only the 1st term in the series will be important.

!

( ,  V  a1 ( sin


!

E ( =
E  =

( , 0=
( , != 0 E  ( , 0 



(



1 

( 

0 a 1 
!

a1 
!




1
!

a1 
!


1

sin



!

1

cos


!

surface charge density

The components of the field and the surface-charge density near ( = 0 all vary
with distance as (/!-1 .

For a very deep corner (small !) the power of ( becomes very large, and no
charge accumulates in such a corner.
When ! > , the 2d corner becomes an edge and the field and the surfacecharge density become singular as (  0.
The 2d electrostatic considerations apply to many 3d situations, even with timevarying fields.
The singular behavior of the fields near sharp edges is the reason for the
effectiveness of lightning rods.

2.12 Introduction to Finite Element Analysis for Electrostatics

FEA encompasses a variety of numerical approaches for the solution of
boundary-value problems in physics and engineering.
Use Galerkin's method for 2d electrostatics as an illustration.
Consider the Poisson equation, . 2 ,=g in a 2d region
boundary conditions on the boundary curve
The vanishing integral

R [ . 2 ,  g]  d x d y = 0

R [. ,/.   g ] d x d y = 0

R, with Dirichlet

  x , y : test function
  x C , yC  = 0

Green's 1st identity on the 1st term

Approximate , by a finite expansion of a set of localized, linearly independent

functions, ij(x,y), with support only in a finite neighborhood of x=xi, y=yj.
Imagine the region
choice for ij is,

R spanned by a square lattice with spacing h, then a possible

i j  x , y=

1x  x i/ h  1y  y j/ h  for x  x i' h ,y  y j' h

0
N0

 i j  x , y= 1
i, j

otherwise

,  x , y-  ) i j  x , y   N 0 = number of lattice sites

i, j
including the boundary
i j

the constant coefficients )ij may be thought of as the approximate values of

,(xi,yj). If the lattice spacing h is small enough, the expansion will be a good
approximate to the true ,, provided the coefficients are chosen properly.
choose the test function  to be the (i,j)th function on the expansion set, for
0

i, j  N (number of internal sites). Assume g(x,y) varies slowly, then

N0

 ) k 0 R . i j  x , y/. k 0  x , y d x d y = g x i , y j  R i j  x , y d x d y

1

k ,0

N coupled inhomogeneous linear algebraic equations for the N unknowns, )k0.

The coupling among )k0 is confined to a small number
of sites near (xi,yj) for the localized function i j.

R i j  x , y d x d y = h2

8 /3 for

R . i j/. k 0 d x d y = 1 /3

k =i , 0= j

k = i 1, 0= j
for k = i ,
0= j 1
k = i 1, 0= j 1

Eq (1) can be written in matrix form:

K : N N matrix

K )=G

) and G : N -column vectors

Since K is a sparse matrix,

the solution can be rapidly
found by special numerical techniques.
A triangle is more optimal as a basic unit in 2d than a square.
The triangular element shoud be small enough that the
field variable change little such that

,  x , y - ,e  x , y = A  B x  C y.
then use the value ,1 , ,2 , ,3  at nodes or
vertices to determine (A,B,C).
To be systematic, define 3 shape
functions

Nje(x,y) at the vertices such that N ej  x = x j , y = y j = 1 only.

a 1  b 1 x 1  c1 y 1 = 1

Consider

N 1 = a 1  b 1 x  c1 y

a 1  b 1 x 2  c 1 y2 = 0
a 1  b 1 x 3  c 1 y3 = 0

 
1 x1 y1

12 1
13
The determinant D= 1 x y = x 2  x 1   y 3  y 1  x 3  x 1  y 2  y 1 = 1
2
2
1 x 3 y3

D is invariant under rotations of the triangle, and

a1 =
D= 2 S e  S e : area of the triangle

c1 =

The other

N
i =1

2 Se

b1 =

y2  y3
2 Se

x3  x2
2 Se

Nj can be known by cyclic permutation of indices.

3
e
i

x 2 y3  x3 y 2

 x , y= 1,

 ai = 1,  bi = 0,  ci = 0
i= 1

i =1

i =1

x1  x2  x 3
y 1  y2  y3
1
, ye =
a j  b j x e  c j ye =
 j = 1, 2, 3 where x e =
3
3
3
The field variable ,(x,y) has the expansion: ,  x , ywhere f: triangle number

 ) fj N jf  x , y
f ,j

Because of the linearity of the shape functions, the value along the common
side of the two triangles from either representation is the same weighted
average of the values at each end.

choose the test function (x,y)=Nj (x,y) for some element

e and vertex i, then

 ) ej e . N ei/. N ej d x d y =e g N ie d x d y- g  x e , y e  e N ie d x d y
j =1

e N ie d x d y = S e ai  bi x e  c1 y e = S e / 3
 N ie
x

= bi ,

 N ei
y

= c i k ei j % e . N ie/. N ej d x d y = S e bi b j  ci c j 

Kei j's only depend on the shape of triangle, not in the orientation
or size.
3

2

k

e
i j

e
j

) =

j =1

S e ge
3

i =1,2,3  ge % g  x e , y e 

K e ) e = G e in matrix form
Enlarge the matrix to include all triangles: K ) = G
e
e
where K =2k 2 ,  k =
k
,
k
=
k
for i 3 j
i j
ii
ii
i j
i j

1
Gi =
3

N0

 Se ge  
T

j=N 1

k ie j ) ej   boundary terms

T: all the triangles connected to the node

E: all the triangles with a side from node i to node j

2

The obvious generalization of the triangle to 3d FEA

is to add another vertex out of the plane to
make a tetrahedron the basic element of volume.
Selected problems:
2.5, 2.7, 2.13, 2.15, 2.20, 2.26