PH 4011 Electromagnetic Fields II

Semester 2 - 2017

45 hours 3 credits

Mid semester (MCQ-Open book) 30%

Year End examination 70%

Suggested readings Classical Electrodynamics, 3rd Edition, Jackson

 Electrostatics
The theory that describe the physical phenomenon related to the interaction between stationary
electric charge or charge distribution in space.

Coulombs Law
Experimentally it has been observed that in classical electrostatics the interaction between two charges
can be given by a mechanical force.

q'

According to the Coulombs law this force is given by

( ) 1
= 3 = (

) 1.1
| | | |
So is along the line of charges
1
, , 2
| |
is attractive if are of opposite polarity and repulsive if and are of same polarity.
depends on the system of units
2
In SI units has units is .
2

1
=
40
2
where 0 = 8.851012 So
.2
 1 ( )
= 3 1.2
40 | |

Linear Superposition
Force on a charge at due to charges 1 , 2 , 3 located at 1 , 2 , 3 . is the
vector sum of

1 ( )
= 3 1.3
40 | |
=1
We can write this as
= () 1.4
Where

1 ( )
() = 3 1.5
40 | |
=1
() is the electrostatic field at the field point (known as the observation point), due to the field
producing charges at the source point .
From (1.1) we can see that for a single charge,
= ()
1 1
() = ( ) 1.6
40 | |

( ) 1
3 = ( )
| | | |

If the discrete charge are numerous enough, we can introduce the term charge density located at
.
Then we can rewrite the equation (1.5) as
1 ( ) 3
() = ( ) 3 1.7
40 | |
Note that charge of a volume element 3 is ( ) 3

( ) 3

We can rewrite equation (1.7) as

1 1
() = ( )
( ) 3 1.8
40 | |
To treat discrete and continuous charge on same footing we can introduce Dirac Delta Function .

Gausss Law

da

Electric field due to at is

() = 1.9
40 2

da

Then
.
(). = = 1.10
40 2 40 2

(). = 1.11
40

( = 2
) total element of solid angle subtended by da
 da

Then

(). = = . 4 = 1.12
40 40 0

In general, if there is a charge density () in space,

1
(). = () 3 1.13
0

Gausss Law is given in equation (1.13). But according to the divergence theorem,

1
(). = (
. ) 3 = ( ) 3 1.14
0

Since this is valid for any volume element 3 , integrands should be equal

( )
. =
1.15
0

This is called the differential form of the Gausss Law

Scalar Potential

From equation (1.5), we have

1 ( )
() = 3
40 | |
=1

That we can write in a general form as

1 ( )
() = 3 1.16
40 | |
=1

We know that
( ) 1
(
= ) 1.17
3
| | | |

Then,

1 1
() = ( )
( ) 3 1.18
40 | |

Since the gradient involves only not , we can write this as

1 ( ) 3
() = . 1.19
40 | |

Since Curl of any gradient of any well-behaved scalar vanishes, we get immediately,

= 0
1.20

As we know (1.15) is not enough to find the 3 components of , we now have equation (1.20).

In equation (1.19), we can see that a vector field is derived from a scalar.

Using (1.19), we can define a potential called a scalar potential (),which is given by,

() =
() 1.21

where,

1 ( ) 3
() = 1.22
40 | |

Application of scalar potential

Work to move charge from A to B in presence of an electric field

= .
= . ).
= (

From Fundamental theorem of gradient

= ( ) 1.23

W is i.d. of path


Proof

Consider two paths (1) and (2) from A to B

2
(2)

1
(1)

, 2 = .
1 = .
(1) (2)

1 2 = .
+ .
= .

(1) (2)

Using Stokes theorem

).
1 2 = (

S is the area surrounded by C.

Since in static condition,

1 2 = 0 1 = 2

So, the work is path independent.

 Poisson and Laplace equations

. E
= , = 0 =
0

. = 2
Therefore,

2 = 1.24
0

1.24 is called the Poisson equation

In charge-free region of space (i.e. = 0)

2 = 0 1.25

1.25 is called the Laplace equation

From (1.22) we have

1 ( ) 3
() =
40 | |

Does this satisfy the Poisson equation?

1 ( ) 1 1
2 () = 2 ( ) 3 = 2 ( ) ( ) 3
40
| | 4 0
| |

1 4 =
2 ( )={
| | 0

So

1
2 ( ) = 4( ) 1.26
| |

When we substitute back

1
2 ( ) 3 = 4 ( ) 3 = {4 =
| | 0

Thus if

1 ( ) 3
() =
40 | |

1 1 1 1
2 () = 2 ( ) ( ) 3 = ( )( ) 3 = 4()
40 | | 40 40
 ()
2 () = 1.27
0

So, the equation 1.22 satisfy the Poisson equation for discrete or continuous charge distribution.

Surface distribution of charges and dipoles & discontinuities in the electric field and potential

Let us find the electric field or potential due to a given surface charge distribution

Gausss Law tells us

(2 1 ). = 1.28
0

(2)
2

( )
(1)

Equation (1.28) not enough to determine the fields 1 and 2 .

Replacing ( ) of equation (1.22) with ( ), we can write on expression for potential,

1 ( ) 2
() = (1.29)
40 | |

For volume or surface charge distribution, the potential is everywhere continuous, even within the charge
distribution.

Another problem of interest is the potential due to dipole layer distribution on a surface S

Dipole layer can be considered as a two surfaces S and S with opposite charge densities separated by a
distance ().

Dipole layer of strength () (surface dipole moment density) is formed by letting S approach
infinitesimally close to S, while the surface charge density () becomes infinite in such a manner that,
 lim ()() = ()
)0
(

() ()

-()

( )

( )

( )

The potential due to two surface Charges at can be given by

1 ( ) 1 ( )
() = 1.30
40 | | 40 | + |

1
For small d we can expand | + | as

1 1 1
| + | = + (nd). ( ) + 1.31

| | | |

As 0, we can neglect the higher order terms. Substituting (1.31) to (1.30), we get

1 ( ) 1 ( ) 1 1
() = ( ). . ( )
40 | | 40 | | 40
| |

1 1 1 1
() = ( ). . ( ) = ( ). [ . ( )]
40 | | 40 | |
So,

1 1
() = ( ). . ( ) 1.32
40 | |

Integrand of equation (1.32) is the potential of a point dipole with a dipole moment = Dda. The
potential caused by a dipole at is

1 1 1
() = . . ( )= . 1.33
40 | | 40 | |3

We can see that

1 [( )] | | cos
. ( ) = .
. = 3 cos =
2 =
| | | |
| |
| |

| |

Then the potential can be written as

1
() = ( ) 1.34
40

Does not depend on the shape of the surface

If the observation point views the inner side of the dipole layer, is positive.

S

| |

There is a discontinuity in potential crossing a double layer. From (1.34), it is clear that the potential has

a discontinuity in crossing the inner to the outer side.
0

2 1 = 1.35
0

20

20

Equation (1.35) agrees with the equation (1.28), which explain the discontinuity of electric field at
boundary.
 Greens Theorem
Equation (1.22) gives a straightforward convenient solution for electrostatic problems, which involve
continuous or discrete charge distribution without any boundaries.

However, most of problems associated with electrostatic comes with boundary conditions.

George Green (1824), introduce a new mathematical procedure to solve these problems. This follows a
simple application of Divergence theorem.

Divergence theorem says

. 3 = . 2
1.36

is a vector field defined in the volume V bounded by the surface S.

Let and are two arbitrary scalar fields and

=
1.37

Substituting (1.37) to (1.36),

. (
) 3 = . 2 =
. 2

But

. (
) = 2 + .

and

. =

Then we have,
 2
. 3 + 2 3 = 1.38

This is called the Greens first identity.

By interchanging and , we get

=
and equation (1.38) changed in to

2
. 3 + 2 3 = 1.39

Taking (1.38) (1.39),

2
. . ) 3 + (2 2 ) 3 = (
( )

2
(2 2 ) 3 = ( ) 1.40

Equation (1.40) is called the Second Greens identity or the Greens Theorem

The Poisson differential equation for potential can be converted in to integral by selecting,
1 1
= = |
and scalar potential as .
|

Also, we have

()
2 () =
0

and

1
2 () = 2 ( ) = 4( )
| |

Applying above relations to (1.40), we get

( ) 3 1 1 2
(4( ) + ) = ( ( ) )
0 R

If the point lies within the volume V,

4( ) = 4()

Then

( ) 3 1 1 2
4() = + ( ( ) )
0 R


1 ( ) 3 1 1 1
() = + ( ( )) 2 1.41
40 0 4 R

The first term gives the potential due to the charge distribution inside the volume

The second term gives the B.C.

If the point lies outside the surface S the LHS of the (1.41) is zero.

if V is all space so that S is at infinity, then the surface integral will vanish and we recover the volume
integral over all space. Then equation (1.41) drops back to (1.22)

Note also that if (x) = 0 for all x in V, then the potential is found simply from the surface integral of
and its normal derivative.

1 ( ) 3
() = 1.22
40 | |

For a charge-free region (i.e. for ( ) = 0 for all )equation (1.41) change into a surface integral in
terms of potential and its normal derivatives at the surface.

If the specification of potential on a closed surface defines a unique potential, that is called Dirichlet
problem or a Dirichlet B.C. can be achieved by conductors

If the electric field (normal component of the potential) everywhere on the surface is defined a unique
problem, that is called Neumann B.C. not so common in electrostatics

specification Boundary Condition

() specified on S Dirichlet

()/ specified on S Neumann

 () specified on part of S ()/ specified Mixed
on remainder

Uniqueness of Solutions to Poisson equation with Dirichlet or Neumann B.C

V
S

Suppose 2 = in V and is defined on S or is defined on S.
0

In 1st case is uniquely determined, in 2nd case is uniquely determined within a constant.

Proof of Dirichlet case

Let 1 satisfy 2 1 = in V and 1 = () specified on S
0

Let 2 satisfy 2 2 = in V and 2 = ( ) specified on S
0

So 1 and 2 are two solutions of 2 = satisfying the same B.C.
0

Let = 2 1 1.42

Then

2 = 2 2 2 1 = 0 in V, and = 0 on S

Let us apply the Greens first identity with = = . Then

3 U 3
.
3 + 2 3 = U.
U 3 + U2 U 3 = U

U
U 3 = 0 since U = 0 on the surface

Also U2 U 3 = 0 since 2 = 2 2 2 1 = 0 in V. Then,

2
U.
U 3 = 0 |
| 3 = 0
= 0

i.e., = in V. But = 0 on S for Dirichlet B.C.

Therefore, = 0

i.e. = 0 in V and 2 = 1 . So the solution is unique.

For Neumann BC
2 1
=0 =0

So is constant on the surface S.

From Greens first identity as we saw on previous case

2
| 3 = 0 = 0
|

Therefore = in V

But we cant say that this constant is zero.

Therefore, for Neumann BC, is determined only to within a constant

The preceding proof is also valid for the case of mixed boundary conditions because the surface integral
vanishes in this case also. Finally, one cannot in general specify both () and / everywhere on S
(Cauchy boundary conditions); either one alone is sufficient to determine a unique solution and the two
solutions so determined are not necessarily the same. However, if Cauchy boundary conditions are given
on just an appropriate part of S, that can be sufficient to give a unique solution for the potential.

For Dirichlet BC, we saw that 2 = 1 and the solution is unique. In Neumann BC, still the solution is
unique apart from an unimportant constant.

Solutions of electrostatic boundary value problems with Greens function

The solutions of the Poisson or Laplace equation in a finite volume V with Dirichlet or Neumann BC on
the boundary surface S can be obtained by Greens theorem (1.40) and so caaled Greens function.

Suppose we have

2 () = in a volume
0

() is specified on

Take
1
(, ) = = and () =
| |

Then

2 (, ) = 4( )

Substituting to Greens theorem (1.40), we have

(, ) ( ) 2
[( )2 (, ) (, , )2 ( )] 3 = [( )
, )
( ]

Then we have

1 (, ) ( ) 2
4() + [ (, , )( )] 3 = [( )

( ,
) ]
0

1 1 (, ) ( ) 2
() = [ (, , )( )] 3 [( )

( ,
) ] 1.44
40 4

In Dirichlet BC, we can demand that

(, ) = 0 for on

Then (1.44) becomes

1 1 (, ) 2
() = [ (, , )( )] 3 [( ) ] 1.45
40 4

= ( )

2 () = in a volume
0

Has the solution

1 1 (, ) 2
() = [ (, , )( )] 3 [( ) ]
40 4

Where the Green Function (, )

Satisfies 2 (, ) = 4( ) in and (, ) = 0 for on

 (, ) can be regarded as the potential due to a unit point source.

GD (, ) = 0

GD (, )

In Neumann BC we have to handle (1.44) carefully,

1 1 (, ) ( ) 2
() = [ (, , )( )] 3 [( )

( ,
) ]
40 4

Only BC that we can select is

)
,
(
= 0 for on . This gives the first term of the second integration zero.

But, we know that

2 (, ) = 4( )

This gives us

2 ( , ) 3 = 4 ( ) 3 = 4

Using Divergence theorem

2 ( , ) 3 =
.
(, ) 3 =
(, ). 2 = 4

2
(, ). 2 =
= 4

)
,
(
Therefore, we cannot use the BC = 0, unless the surface is at infinity

The simplest BC that we can use is,

( )
, 4
= for on

where, is the total area of the whole surface.

Then the solution is

 1 1 (, ) 2 1 ( ) 2
() = [ (, , )( )] 3 ( ) + [
, )
( ]
40 4 4

1 1 4 1 ( ) 2
() = [ (, , )( )] 3 ( ) ( ) 2 + [ (, ) ]
40 4 4

1 1 ( ) 2
() = + [ (, , )( )] 3 + [ (, ) ] 1.46
40 4

- average value of the potential over the whole surface

(1.46) is not a very useful relation

Useful relationship

(, ) = ( , )

Proof

Let = ( , ) and = (, )

(1.40) gives that

3
(2 2 ) 3 = ( )

, y) 2 GD(X
GD(X , y) GD (X
, y) 2 GD(X
, ) 3

, y)
GD (X , y)
(X
, y)
= ( (X , y)
GD (X ) 3

, y) = 0 for y on S
(X

, y) = 0 for y on S
(X

So

4 (, ) = 4 ( , )

(, ) = ( , )

Example 1

Use (1.45) to find the potential due to a point charge q at = in infinite space.
Example 2

2 () = 0 with BC ( = ) = (, , ). Find the potential () in the region .

Electrostatic Potential Energy

Start with a charge 1 at 1. Work to be done to make a charge 2 from infinity to 2 is

1
2

1 1
2 = 2 ( ) = 2 (2 ) 1.47
40 |1 2 |

Work to be done to move a charge 3 from infinity to 3 is

1 1 2
3 = 3 [ + ]
40 |1 3 | |2 3 |

2
3
3 = 1.48
40 |2 |
=1

Then the total work to be done to assemble N charges against the charge is
1

= ( ) = 1.49
=1
40 | |

Then the total work to be done to assemble N charges is

1
1
= = 1.50
40 | | 2
=2 =1
40 | |
=1 =1

If you have a continuous charge distribution, the summation can be replaced by an integration

1 ()( ) 3 3
= 1.51
80 | |

But
 1 ( ) 3
() =
40 | |

So, we can rewrite

1
= ()(X
) 3 1.52
2
)
(
We know that 2 =
0

Substituting to 1.52,

1
= 0 2 . () 3
2
. [(
)] = . + 2

1
. [(
= 0 [ )]
.
] 3
2
Applying Divergence theorem to the first term

1 1
)d3 X + 0
= 0 ( .
d3 X
2 2
At large distances the first term goes to zero (depend on 1/ 2 )

Then

1 2 0 ||2
= 0 d3 X = 0
. | ()| 3 = 1.53
2 2 2

1.53 is the energy density including the self-energy term.

At the integration of (1.53) all terms with charges disappeared. Only the electric field all over the space is
available now.

Note: It is clear that the energy is always positive regardless of the charges involved. But according to the
(1.50), we can see that the energy can be negative depending on the polarity of charges involved.

This is due to the fact that (1.53) is derived including the self-energy term.

To see this fact clearly, let us consider the following example.

Let us calculate the energy due to two charges 1 and 2 located at 1 and 2 respectively.

The electric field at P is

1 1 ( 1 ) 2 ( 2 )
() = [ + 3 ] 1.54
40 | |3 | |
1 2
 2
2

Substituting to (1.53) we get,

1 12 22 2
= [ +
2 2 0 4 4 + 21 2 ( 1 ). 3 3] 1.55
32 0 | 1 | | 2 | | 1 | | 2 |

We can see that the first two terms are the self-energies and the third term is the interaction energy.

1 2 2
= 1 ).
(
16 2 0 3 3
| 1 | | 2 |

We can show that

1 1 2
=
40 | |3
1 2

Capacitance

We know that

1
= ()() 3
2
Consider a system of conductors in space. There is no charges on space except on conductors. Potential
of each conductor is with total charge

1
2 3 i

The potential of the jth conductor is .


1 1
= () 3 = Q
2 2
=1

Since the potentials linearly depends on , the potential of the ith conductor can be written as

= ( = 1,2, )
=1

depend on the geometry of the conductors. These n equations can be inverted to yield the charge on
the ith conductor in terms of all the potentials.

= ( = 1,2, . . )
=1

1
= 1.56
2
=1 =1

The coefficient are called capacities or capacitance while the , are called coefficients of
induction.

E.g. One sphere of radius R. If it carries a charge Q. Then

= for >
40 2

=
40

So, capacitance

= 40 =

Total energy of a uniformly charged sphere

We know that

0 ||2
=
2

=
40 2

3
= <
40 2 3

= <
40 3
 1
= >
40 2

0
2 2

1 2 2
= [ ( ) 4 + ( ) ]
2 0 40 3 40 2

2 1 1 3 2
= [ + ]=
80 5 20 0

Variational approach to the solution of Laplace and Poisson equations

Variational methods are widely used in classical and quantum physics. Variational method can be used to
obtain an accurate solution.

Consider the functional,

1
[] = . (
) 3 3 1.57
2

Function ()is well behaved inside the volume V and on its surface S. () is a specified source without
singularities within V.

The first order changes in the functional when we change from + (the modification ()
is infinitesimal within V), is

= .
() 3 3 + 1.58

Proof

( + ) = + ()

( + ).
( + ) =
.
+
.
() +
.
() +
().
()

[ + ] []
1 1 1
= .
3 +
.
() 3 +
.
() 3 .
3
2 2 2
3 3 + 3 +..

= .
() 3 3 +

Neglecting higher order terms and using Greens first identity

= , = , yields

(2 + . ) 3 = da

().
(2 + ) 3 = da ().
3 = (2 ) 3 + da

Substituting to (1.58),

= (2 ) 3 + da 3

= ( 2 ) 3 + da

Provided = 0 on (so that the surface integral vanishes) the first order change in [] vanishes if

2 () = 1.60

Neglected terms in (1.58) are semi-positive ( > 0). We see that [] is a stationary minimum if satisfy
Poisson like equation within and vanishes on .

With the and /0 , the minimization of the functional [], yields the equation of motion
of the electrostatic potential in the presence of a charge density and Dirichlet B.C. ( gives on and =
0 there)

Using the variational functional method one can derive the Poisson equation

The stationary nature of the extremum of [] permits a practical approach to an approximate solution
for ()

We choose a flexible trial function () = (, , , . . ) that depends on a normalized constants A and

some number of other parameters ,

() is constructed to satisfy the given b.c. on the surface .

In Neumann b,c,, are specified by | = () on . Then the appropriate functional is,

1
[] = . d3 X 3 1.61
2

Following the same procedure as Dirichlet b.c.

= (2 ) 3 + ( ()) da 1.62

The requirements vanishes i.d. of implies

2 = within V and = () on S 1.63

Again, approximate solution can be found by using a trial function that satisfy the Neumann b.c.