Electromagnetism 1

Chapter 3 Special techniques

2021 Fall
郭華丞
 Outlines

Laplace’s Equation
The Method of Images
Separation of Variables
Multipole Expansion
 Integral eqns v.s differential eqns.
1 r ( r¢ ) rˆ
E=
4pe 0 ò r 2
dt ¢

Integral equations
1 r ( r¢ )
V (r ) = ò dt ¢
4pe 0 r

Differential form of the same equations

r
ÑV =-
2

e0

If r =0 Ñ 2V = 0 Laplace’s equation

¶ 2V ¶ 2V ¶ 2V
In Cartesian coordinate : + 2 + 2 =0
¶x 2
¶y ¶z
 One-dimensional cases
If the potential only depends on x d 2V
2
=0
dx
The general solution is
V ( x ) = mx + b

Coefficients m and b can be determined by using boundary conditions

Important features of the solutions

1. V(x) is the average of V(x-a) and V(x+a)

2. No local maxima and local minima

Proof: if V(b) is a local maxima, then there exist a such that
V(b)>V(b-a) and V(b)>V(b+a), which contradicts to statement 1.
 Two-dimensional cases
¶ 2V ¶ 2V
+ 2 =0
¶x 2
¶y

Physical analogy: rubber membrane on a fixed stand

Important features of the solutions

1. V(x,y) is the average of the potentials around the point

1
V ( x, y ) = ò Vdl
2p R circle
Remark: method of relaxation for calculating solutions.

2. No local maxima and local minima

Remark : a ball on a rubber membrane will roll over one side and fall off
 Three-dimensional cases
1. V at point r is the average of the potentials on a spherical surface centered at r
1
V (r ) = ò Vda
4p R 2 sphere

Remark: method of relaxation for calculating solutions.

2. No local maxima and local minima

Proof of statement 1

Consider the average potential on a sphere surface due to point

charge q located outside the sphere
1 q
The potential of the area element is V=
4pe 0 r

r = z 2 + R 2 - 2 zR cos q z>R
 Three-dimensional cases
1. V at point r is the average of the potentials on a spherical surface centered at r
1
V (r ) = ò Vda
4p R 2 sphere

Proof of statement 1

Consider the average potential on a sphere surface due to point

charge q located outside the sphere

1 q
The potential of the area element is V=
4pe 0 r

r = z 2 + R 2 - 2 zR cos q z>R
 V ( R 2 sin q df dq )
p 2p
ò
sphere
Vda = ò
0 ò0

q
ò ò (z + R - 2 zR cos q )
p 2p -1 2
= R 2 2 2
( sin q df dq ) q
4pe 0 0 0

q 2 p 2
R ò ( z + R - 2 zR cos q ) d ( cos q ) r
-1 2
=- 2
z
2e 0 0

q R 2 12 p
=
2e 0 z
( z + R - 2 zR cos q )
2
q
0 R
q R q R2
= éë z + R - ( z - R ) ùû =
2e 0 z e0 z

1 1 æ q R2 ö q
ò Vda = 2 ç ÷ = = V ( 0)
4p R 2 sphere
4p R è e 0 z ø 4pe 0 z

By the superposition principle, for a collection of point charges outside the

sphere, the average potential over a sphere surface equals to the potential
at the sphere center.
Boundary conditions and uniqueness theorem

Laplace’s equation
solutions
Boundary conditions

Question: What are the appropriate boundary conditions, sufficient to

determine the solution but not so strong as to generate inconsistencies?

Answer: the uniqueness theorem

 First uniqueness theorem
First uniqueness theorem
The solution of Laplace’s equation in some volume V is uniquely
determined if the value on boundary surface S is specified.

Proof: If V1 and V2 are solutions of Laplace’s equation satisfying the

same boundary condition, we have

Ñ 2V1 = Ñ 2V2 = 0
Expressing their difference as V3=V1-V2, we have V3 also
obeys Laplace’s equation Ñ 2V3 = 0

Since V1 and V2 have the same value on the surface (same

boundary condition), we have V3 = 0 on S

Notice that the solutions of Laplace’s equation do not have local max. or
min. Therefore, V3 = 0 in V
V1 = V2
uniqueness theorem for Poison equations
If V1 and V2 are solutions of Poisson’s equation due to the
same charge distribution satisfying the same boundary
condition, we have
r
Ñ V1 = Ñ V2 =
2 2

e0
their difference V3=V1-V2, should obey Laplace’s equation

Ñ 2V3 = Ñ 2 (V1 - V2 ) = 0

Since V1 and V2 have the same value on the surface (same

boundary condition), we have V3 = 0 on S, and moreover in
V V3 = 0

Once the charge distribution and the potential on the

boundary are specified, the solution is uniquely determined.
 Second uniqueness theorem
Second uniqueness theorem
in a volume V surrounded by some conductors and containing a
specific charge density r, the electric field is uniquely determined if the
total charge on each conductor is given.

Outer boundary
The boundary S consist of outer
boundary and conducting surfaces.
S

Qi

i-th conducting surface

Proof If E1 and E2 are electric fields satisfying the condition of the
problem. They both obey the Gauss’s law
r
Ñ × E1 = Ñ × E2 =
e0
For surfaces bounding a specific conductor, we have

𝑄2
! 𝐄0 # 𝑑𝐚 = ! 𝐄1 # 𝑑𝐚 =
𝜀3
!"#$ %&'()%#!'* !"#$ %&'()%#!'*
+),-.%/ +),-.%/

For the outer boundary (bounding V)

𝑄676
! 𝐄0 # 𝑑𝐚 = ! 𝐄1 # 𝑑𝐚 =
𝜀3
&)#/, 4&)'(.,5 &)#/, 4&)'(.,5

Examine the difference E3 = E1 - E2

We have Ñ × E3 = 0 ! 𝐄8 # 𝑑𝐚 = ! 𝐄8 # 𝑑𝐚 = 0
&)#/, 4&)'(.,5 !"#$ %&'()%#!'*
+),-.%/
Consider potentials on the conductors as V1 and V2 respectively. Then V3=V1-V2 is
constant over the conductor surface

By applying the vector analysis rule

Ñ × (V3E3 ) = ( ÑV3 ) × E3 + V3 ( Ñ × E3 ) = - E32 E3 = -ÑV3

ò 3 3
Ñ × (V E ) dt = ò 3 3
V E × d a = - ò 3 dt
E 2

V S V

ò 3 dt = -V3 ò E3 × da = - ò V3E3 × da = 0
2
E
V S outer
boundary

For the outer surface, we may choose infinity so that V1=V2=0

E3 = 0 Everywhere in V
 Purcell’s example
+ − + − 0 0

− + − + 0 0

This is a stable What if we connect

This is also a stable
charge configuration them in pairs with
charge configuration
conducting wires?
Will it be stable?

By the uniqueness theorem, the charges will move through the

wires and cancel itself off
Laplace’s Equation
The Method of Images
Separation of Variables
Multipole Expansion
 Image problems z

The system of a point charge and an infinite q

grounded conducting plane
d

V=0 y
Reasoning:
The point charge will induce some negative charge
on the conducting plane, then the potential is due x
to the point charge and the induced surface charges

Goal:
To solve Poisson’s equation in the region z>0, with a point charge q at
(0,0,d) and the boundary conditions that V=0 at z=0 and at infinity.
 Trick:
Consider another electrostatic problem that two point charges q at (0,0,d)
and –q at (0,0, –d) without any conducting plane. The potential is easily
found as
é ù
1 ê q q ú
V ( x, y , z ) = -
4pe 0 ê x 2 + y 2 + ( z - d )2 x 2
+ y 2
+ ( z + d )
2 ú
ë û

We may check that

V=0 at z=0 and at infinity.

The boundary conditions for z>0 region is the same as the previous problem, and the
charge distribution for z>0 is the same as the previous problem

By using the uniqueness theorem, we deduce that the solution of the later
problem at z>0 is the solution of the previous problem
 Induced surface charge
¶V
The surface charge can be computed using s = -e 0
¶n
The normal direction is z-direction
é ù
¶V 1 ê -q ( z - d ) q(z + d ) ú
= +
¶z 4pe 0 ê é x 2 + y 2 + ( z - d )2 ù 3 2 é x 2 + y 2 + ( z - d )2 ù 3 2 ú
êë ë û ë û úû

1 -qd
s ( x, y ) = (z=0)
2p ( x 2 + y 2 + d 2 )3 2

The total induced charge is Q = ò s da

¥
2p ¥ 1 -qd -qd
Q=ò ò rdrd f = = -q
0 0 2p ( r 2 + d 2 )3 2 ( r 2
+ d )
2 12
0
 Force and energy
Recall the problem of two charges q and –q. the force between them is
1 q2
F=- zˆ
4pe 0 ( 2d ) 2

The energy of the problem is

q2 1
W =-
4pe 0 2d
The force of the charge-conducting plane problem is the same, whereas the
work is half of that. This is because the energy can be expressed as
e0
W= ò dt
2
E
2
In the second problem, only z>0 region contributes the energy
q2 1
W =-
4pe 0 4d
Consider the work done by moving the charge q from infinity, the work is
d 1 d q2 1 q2
W = ò F × dl = ò dz = -
¥ 4pe 0 ¥ 4z 2
4pe 0 4d
 Image charges of a grounded
conducting sphere
A point charge q is placed at a distance a from
the center of a grounded conducting sphere of
R
radius R
q
a
Examine another configuration that a negative V=0
charge q¢ = - R q is placed a distance of
a
b=
R2 (b < R )
r¢ r
a ( q¢ < q ) R
q
q q¢ a
The potential of this configuration is b

1 æ q q¢ ö
V (r ) = ç - ¢÷
4pe 0 è r r ø
We may check the potential on the sphere by using the spherical coordinate in
which the origin is the sphere center.

1 æ q q¢ ö r 2 = a 2 + R 2 - 2aR cos q
V (r ) = ç - ¢÷
4pe 0 è r r ø r ¢2 = b 2 + R 2 - 2bR cos q
On the sphere surface
æ ö
V ( R, q ) =
1 ç q
-
(a R) q ÷
ç
4pe 0 ç a + R - 2aR cos q ÷
( R a ) + R - 2 ( R a ) R cos q ÷ø
2 2 2 2 2 2
è
1 æ q ( a R) q ö
= ç - ÷
4pe 0 ç a 2
+ R 2
- 2 aR cos q ( a R ) R 2
+ a 2
- 2aR cos q ÷
è ø
=0

The attracting force is

1 qq¢ 1 æRö q2 1 Raq 2

F= = ç ÷ =
4pe 0 ( a - b )2 4pe 0 è a ø ( a - R a )
2 2
4pe 0 ( a 2 - R 2 )2
Laplace’s Equation
The Method of Images
Separation of Variables
Multipole Expansion
 Cartesian coordinate
y
The boundary conditions are V=0
V =0 when y = 0 V0(y) d V=0
V =0 when y = a
V = V0 ( y ) when x = 0 x
V =0 as x ® ¥ z

The system is independent of z, and is a two-dimensional problem:

¶ 2V ¶ 2V
+ 2 =0
¶x 2
¶y

Begin with a trial solution V ( x, y ) = X ( x ) Y ( y )

Insert the trial solution into Laplace’s equation, one has ¶2 X ¶ 2Y
Y 2 +X 2 =0
¶x ¶y
 The next step is separating variables
1 ¶ 2 X 1 ¶ 2Y
+ =0
X ¶x 2
Y ¶y 2

Since the first term only depends on x, while the second only on y, the two
terms must be constant.
1 ¶2 X 1 ¶ 2Y
=C =-
X ¶x 2
Y ¶y 2

d2X
2
= k 2
X
A partial differential equation becomes dx Here C=k2 is chosen
two ordinary differential equations d 2Y
2
= - k 2
Y (Why not choose C=-k2?)
dy
The solutions are X = Ae kx + Be - kx
Y = C sin ky + D cos ky

V ( x, y ) = X ( x ) Y ( y ) = ( Ae kx + Be - kx ) ( C sin ky + D cos ky )
To determine all to coefficients and k, we need to impose the boundary conditions

Firstly, V =0 as x ® ¥ , therefore A=0

V ( x, y ) = e - kx ( C ¢ sin ky + D¢ cos ky )

V =0 when y = 0 D¢ = 0
V =0 when y = a e - kx C ¢ sin ka = 0
np
k= , n = 1, 2,3!
a
And the potential is in the form V ( x, y ) = C ¢e - kx sin ky

By choosing any single n value function, we cannot fit it to the last boundary
condition. Here we construct a more general function by using the series:
¥
np y
V ( x, y ) = å Cn e - np x a
sin
n =1 a
It is easily to find that
æ ¥ np y ö ¥ 2 æ - np x a np y ö
Ñ V ( x , y ) = Ñ ç å Cn e
2 2 - np x a
sin ÷ = å Cn Ñ ç e sin ÷=0
è n =1 a ø n =1 è a ø
The last boundary condition yields
¥
np y
å Cn sin
n =1 a
= V0 ( y )

Fourier series

n¢p y
Multiply by sin , and integrate y from 0 to a,
a
¥
np y n¢p y n¢p y
å
a a
Cn ò sin sin dy = ò V0 ( y ) sin dy
n =1
0 a a 0 a
ì0 if n¢ ¹ n
a np y ¢
np y ï
ò0 sin
a
sin
a
dy = ía
if n¢ = n
ïî 2
2 a np y
ò 0( )
Therefore, we have Cn = V y sin dy
a 0 a
If V0 ( y ) = V0
ì0 for even n
2V anp y 2V ï
Cn = 0 ò sin dy = 0 (1 - cos np ) = í 4V0
a 0 a np ïî np for odd n

4V0 ¥
1 - np x a np y
V ( x, y ) = å
p n =1,3,5! n
e sin
a
 Orthogonal functions
Completeness Any function can be expressed as a linear combination
of some set of functions, fn
¥
f ( x ) = å Cn f n ( x )
n =1

Orthogonality The integral of the product of any two different

functions in the set is zero
a
ò f n ( x ) f n¢ ( x ) dx = 0 for n = n¢
0
 A 3D example
y
A metal pipe of side a and b is grounded, with an V=0
end maintained at a potential V0(y,z)
V0(y,z)
V =0 when y = 0
V=0
V =0 when y = a x
V =0 when z = 0 z
V =0 when z = b
d
V =0 as x ® ¥
V = V0 (x, y ) when x = 0

We look for solutions of the form V ( x, y , z ) = X ( x ) Y ( y ) Z ( z )

¶ 2V ¶ 2V ¶ 2V 1 ¶ 2 X 1 ¶ 2Y 1 ¶ 2 Z
+ 2 + 2 =0 + + =0
¶x 2
¶y ¶z X ¶x 2
Y ¶y 2
Z ¶z 2
 1 ¶2 X
= C1
X ¶x 2

1 ¶ 2Y C1 + C2 + C3 = 0
= C2
Y ¶y 2

1 ¶2Z
= C3 Since for y and z the boundary condition are V=0
Z ¶z 2

C2 < 0
C3 < 0
¶2 X
Setting C2 = - k 2 ,we have ¶x 2
= ( k 2
+ l 2
) X

C3 = -l 2 ¶ 2Y
= - k 2
Y
¶y 2

¶2Z
= -l 2
Z
¶z 2
The solutions can be written as

X ( x ) = Ae k 2 +l 2 x
+ Be - k 2 +l 2 x

Y ( y ) = C sin ky + D cos ky
Z ( z ) = E sin lz + F cos lz
np mp
From the first 5 boundary conditions, we have A= D= F =0 k= , l=
a b
( n a )2 + ( m a )2 x np y mp z
V ( x, y, z ) = Ce
-p
sin sin
a b

A general solution is
¥ ¥
np y mp z
V ( x, y, z ) = åå Cn ,m e
-p ( n a )2 + ( m a )2 x
sin sin
n =1 m =1 a b
To determine the coefficients
4 a b np y mp z
Cn , m = ò ò V0( y , z ) sin sin dzdy
ab 0 0 a b

Here we have used

b ¥ ¥
np y mp z n¢p y m¢p z ab
åå sin
a
òò
0 0
n =1 m =1 a
sin
b
sin
a
sin
b
dzdy = d nn¢d mm¢
4

V0 ( y, z ) = V0 4V0 a b np y mp z
If Cn , m =
ab ò0 ò0
sin
a
sin
b
dzdy

ì 0 if n or m is even
ï
= í 16V0
ïî p 2 nm if n and m are odd

16V0 ¥
1 -p np y mp z
V ( x, y , z ) =
p2
å
n , m =1,3,5! nm
e
( n a )2 + ( m a )2 x
sin
a
sin
b
 Spherical coordinates
1 ¶ æ 2 ¶V ö 1 ¶ æ ¶V ö 1 ¶ 2V
Ñ V = 2 çr ÷+ 2 ç sin q ÷+ 2 2 =0
2

r ¶r è ¶r ø r sin q ¶q è ¶q ø r sin q ¶f 2

Suppose the problem has azimuthal symmetry, the potential is independent of f

So that 1 ¶ æ 2 ¶V ö 1 ¶ æ ¶V ö
çr +
÷ 2 ç sin q ÷=0
r ¶r è ¶r ø r sin q ¶q è ¶q
2
ø
Assuming a solution is product V ( r , q ) = R ( r ) Q (q ) we have

1 d æ 2 dR ö 1 d æ dQ ö
ç r ÷ + ç sin q ÷=0
R dr è dr ø Q sin q dq è dq ø

Since the variables are separated, it follows that each term must be constant
1 d æ 2 dR ö 1 d æ dQ ö
çr ÷ = l ( l + 1) ç sin q ÷ = -l ( l + 1)
R dr è dr ø Q sin q dq è dq ø
 Legendre polynomials
d æ 2 dR ö
÷ = l ( l + 1) R
The radial equation has the general solution
çr
dr è dr ø
1
R ( r ) = Ar l + B
r l +1
d æ dQ ö
The angular equation ç sin q ÷ = -l ( l + 1) sin qQ has the solutions what are call
dq è dq ø
Legendre polynomials Q (q ) = Pl ( cos q )
l
1 æd ö 2
Pl ( x ) = l ç ÷ ( x - 1)
l
The polynomials can be defined by Rodrigues formula
2 l ! è dx ø
in which l must be nonnegative integers
P0 ( x ) = 1
Only x even powers
We may list the first few Pl(x) P1 ( x ) = x
P2 ( x ) = ( 3x 2 - 1) 2 Only x odd powers
P3 ( x ) = ( 5 x 3 - 3x ) 2
 Is there the other set of solutions? The answer is that the other set solutions are
unphysical since it diverges at q=0 and p

In the end, we obtained the solution for the problems having azimuthal symmetry with l

æ B ö
V ( r , q ) = ç Ar l + l +1 ÷ Pl ( cos q )
è r ø
We may write the general solution as a linear combination of these functions

¥
æ B ö
V ( r , q ) = å ç Al r l + l +l1 ÷ Pl ( cos q )
l =0 è r ø
Orthogonalily of Legendre polynomials
1 p
ò Pl ( x ) Pl ¢ ( x ) dx = ò Pl ( cos q ) Pl ¢ ( cos q ) sin q dq
-1 0

ì 0 if l ¹ l ¢
ï
=í 2
ïî 2l + 1 if l = l ¢

You may want to prove this by using Rodrigues formula

Potential inside a spherical cavity
The potential V0 (q ) on a sphere of radius R is specified.

V0 (q )
The potential will not diverge at r=0, it follows that
Bl are zero
¥
V ( r , q ) = å Al r l Pl ( cos q )
l =0

On the sphere, we apply the boundary condition

¥
V ( R, q ) = å Al R l Pl ( cos q ) = V0 (q )
l =0

Using the orthogonality, we have

2 p
Al R l
= ò V0 (q ) Pl ( cos q ) sin q dq
2l + 1 0
Suppose V0 (q ) = k sin 2 (q 2 )

k
V0 (q ) = k sin 2 (q 2 ) = (1 - cos q )
2
k
= éë P0 ( cos q ) - P1 ( cos q ) ùû
2
2l + 1 p
l ò0 0 ( ) l (
Because of Al = V q P cos q ) sin q dq
2R

k k
A0 = A1 = -
2 2R

ké r ù
V ( r ,q ) = P ( cos q ) - P ( cos q )
2 êë úû
0 1
R
kæ r ö
= ç1 - cos q ÷
2è R ø
 Potential outside a sphere
The potential V0 (q ) on a sphere of radius R is specified.

The potential will not diverge at infinity, it follows

that Al are zero
¥
Bl
V ( r ,q ) = å l +1 l (
P cos q )
l =0 r

Using the orthogonality, we have

2l + 1 l +1 p
Bl = R ò V0 (q ) Pl ( cos q ) sin q dq
2 0
Induced charges on a metal sphere
A metal sphere is placed in a uniform external electric field,
and charges would induced on the surface. We will find out
the potential due to the induced charges.

In a uniform field, the potential is not zero at infinity. E = E0 zˆ

Assuming the field is in z-direction, then we have the
potential far from the sphere:

V = - E0 z + C
On the surface, the potential is constant, which we can
assume as zero. It follows that C=0

Rewrite the boundary conditions in spherical coordinates,

we get

V = 0 when r = R
V ® - E0 r cos q when r ! R
The general solution for potential is
¥
æ B ö
V ( r , q ) = å ç Al r l + l +l1 ÷ Pl ( cos q )
l =0 è r ø
From the first boundary condition, we get
Bl
Al R l + l +1
=0
R
Bl = - Al R 2l +1
¥
æ l R 2l +1 ö
V ( r , q ) = å Al ç r + l +1 ÷ Pl ( cos q )
l =0 è r ø

For r>>R, the second term is negligible, so the second boundary condition is
¥
V ( r , q ) ® å Al r l Pl ( cos q ) = E0 r cos q
l =0

A1 = - E0 is the only none-zero coefficient

In conclusion, we have

æ R3 ö
V ( r , q ) = - E0 ç r - 2 ÷ cos q
è r ø

The first term is due to the external field, so the second term contributed from the
induced charges
R3
E0 2 cos q
r Note: this is a dipole field

The induced surface charge can be obtained by

¶V
s (q ) = -e 0 = 3e 0 E0 cos q
¶r r=R
Potential due to surface charge distribution
In some cases, we do not have potential but the surface charge specified. This is
related to the so-called Neumann boundary condition problems

It has been showed that inside the sphere, the potential

V> ( r , q )
can be expressed as
¥
s (q )
V< ( r , q ) = å Al r l Pl ( cos q )
V< ( r , q )
l =0

And outside the sphere, the potential is

¥
Bl
V> ( r , q ) = å l +1 l (
P cos q )
l =0 r
First, the potential is continuous at r=R, and it turns out
¥ ¥
Bl
å
l =0
Al R Pl ( cos q ) = å l +1 Pl ( cos q )
l

l =0 R

Because of the orthgonality, the coefficients of same l will be equal Bl = Al R 2l +1

Second, the surface charge will produce a discontinuity of radial derivative of V

æ ¶V> ¶V< ö 1
ç - ÷ = - s 0 (q )
è ¶r ¶r ø r = R e0
¥ ¥
Bl 1
-å ( l + 1) l + 2 Pl ( cos q ) - å lAl R l -1 Pl ( cos q ) = - s 0 (q )
l =0 R l =0 e0

Combining the two results, we have

¥
1
å ( 2l + 1) Al Rl -1Pl ( cos q ) =
l =0 e0
s 0 (q )

1 p
Al = l -1 ò0
s 0 (q ) Pl ( cos q ) sin q dq
2e 0 R

One may re-check the previous example by using s (q ) = k cos q

Laplace’s Equation
The Method of Images
Separation of Variables
Multipole Expansion
 Approximate potential at large distance
A systematic expansion for the potential of an arbitrary localized charge
distribution in powers of 1/r
1 r ( r¢ )
V ( P) = ò dt ¢
4pe 0 r
r
Using the law of cosines we have
r
r 2 = r 2 + r ¢2 - 2rr ¢ cos q
r¢ q
æ æ r ¢ ö
2
æ r¢ ö ö
= r ç1 + ç ÷ - 2 ç ÷ cos q ÷
2
ç èrø è r ø ÷
è ø
If P lies well outside the charge distribution,
r¢ and we can expand
!1
r
2
1 1
= (1 + e )
-1 2 æ r¢ ö æ r¢ ö
e = ç ÷ - 2 ç ÷ cos q ! 1
r r èrø èrø
 1 1 1æ 1 3 5 ö
= (1 + e ) = ç1 - e + e 2 - e 3 + ! ÷
-1 2

r r rè 2 8 16 ø

It turns out the expansion can be expressed using Legendre polynomials

n
1 1 ¥ æ r¢ ö
= å ç ÷ Pn ( cos q )
r r n =0 è r ø
¥
1 1
V ( P) =
4pe 0
år
n =0
n +1 ò
r ¢ n
Pn ( cos q ) r ( r¢ ) dt ¢

This is called the multipole expansion

1 é1 1 1 2æ3 1ö ù
V ( P) = ( ) ( ) ( )
4pe 0 êë r ò r2 ò r3 ò è 2
r r ¢ dt ¢ + r ¢ cos qr r ¢ dt ¢ + r ¢ ç cos 2
q - ÷ r r ¢ dt ¢ + !ú
2ø û
monopole dipole quadrupole
 Dipole field
1 1
Vdip ( P ) = ò ( r ¢ cos q ) r ( r¢) dt ¢ cos q = rˆ × rˆ ¢
4pe 0 r 2
1 1
= rˆ × ò r¢r ( r¢ ) dt ¢
4pe 0 r 2

Here we may define the dipole moment as p = ò r¢r ( r¢ ) dt ¢

p × rˆ
Vdip ( P ) =
4pe 0 r 2
n
For a collection of point charges, the dipole moment is p = å qi ri¢
i =1
The electric field of a dipole
p × rˆ p cos q
Vdip ( P ) = =
4pe 0 r 2 4pe 0 r 2

The electric field is

¶V 2 p cos q
Er = - =
¶r 4pe 0 r 3

Eq = -
1 ¶V p sin q
=
r ¶q 4pe 0 r 3
Ε ( r ,q ) =
p
4pe 0 r 3 ( 2 cosq rˆ + sin qqˆ )
1 ¶V
Ef = - =0
r sin q ¶f