6 Electromagnetic Fields and Waves

James Clerk Maxwells unification of electromagnetic phenomena, published in 1865, is

perhaps the best example of a successful modern scientific theory [Maxwell, 1998]. In
just a few simple equations he was able to show that the apparently distinct phenomena of
electricity and magnetism were actually intimately related through a common theoretical
framework that contained unexpected predictions, such as electromagnetic waves, and led
to significant succeeding discoveries including the theory of special relativity. This chapter
studies these Maxwells equations for static and time-varying electric and magnetic
fields. This theory is called electrodynamics because it describes the time variation of
electromagnetic phenomena, and it will be the foundation of much the rest of the book.

6.1 V E C T O R C A L C U L U S
Working with Maxwells equations will require differentiating and integrating field vectors, and so our first step will be a review of the necessary vector calculus.
6.1.1 Vectors
Let ~x (x, y, z) (x1 , x2 , x3 ) be the coordinates of a point expressed as a vector in
rectangular coordinates (Figure 6.1). The magnitude of a vector is
q
(6.1)
|~x| = x21 + x22 + x23 .
~ and B
~ with an angle between them, the dot product measures their
For two vectors A
overlap:
~ B
~ = |A||B| cos()
A
= A1 B1 + A2 B2 + A3 B3
3
X
=
Ai Bi
i=1

Ai Bi

(6.2)

Because such sums recur frequently in manipulating vectors, the last line introduces the
Einstein summation convention of summing over repeated indices.

56

Electromagnetic Fields and Waves

x
Figure 6.1. Rectangular, cylindrical, and spherical coordinate systems.

The cross product of these vectors is



 x 1 x 2 x 3 


~B
~ =  A1 A2 A3 
A


 B B B 
1
2
3
= (A2 B3 A3 B2 )x 1 + (A3 B1 A1 B3 )x 2 + (A1 B2 A2 B1 )x 3

(6.3)

where x 1 (pronounced x hat) is a unit vector in the x1 direction. The magnitude of

the cross product is equal to the product of the lengths of the vectors times the sine of
the angle between them, and its direction is perpendicular to the plane containing the
vectors. The orientation can be remembered by the right hand rule: if the fingers of
~ towards B,
~ your thumb points in the direction of their cross
your right hand curl from A
product. The ith term of the cross product can be written in terms of the summation
convention as
~ B)
~ i = ijk Aj Bk
(A
(6.4)
by using the antisymmetric tensor

if (ijk) = (123), (231), or (312) (cyclic permutation)

1
ijk =
1 if (ijk) = (132), (321), or (213) (anticyclic permutation)

0
otherwise

(6.5)

Interchanging indices shows that the cross product is anticommutative:

~ B
~ = B
~ A
~
A

(6.6)

A useful expansion for the product of antisymmetric tensors is

ijk klm = il jm im jl

(6.7)

where ij is the Kroenecker delta

ij =

1 if i = j
0 otherwise

(6.8)

6.1 Vector Calculus

57

6.1.2 Differential Operators

Now let (~x) be a scalar function of ~x. The gradient of is defined to be
(~x) =

(~x)
(~x)
(~x)
x 1 +
x 2 +
x 3
x1
x2
x3

(6.9)

The gradient is a vector that points in the direction of the fastest change of , and its
magnitude is equal to the rate of change in that direction. If (~x) is the height of a hill,
then a ball released at ~x would roll down the hill in the direction (~x). The vector
operator is called del.
~ x) = (A1 (~x), A2 (~x), A3 (~x)), the divergence is
For a vector-valued function A(~
~ x) = A1 + A2 + A3
A(~
x1
x2
x3
X Ai
=
xi
i
X

i Ai
i

= i Ai

(6.10)

The divergence is a number that measures the rate at which the vector field is locally
expanding or contracting.
The curl of a vector field in three dimensions is defined to be
~ = (2 A3 3 A2 )x 1 + (3 A1 1 A3 )x 2 + (1 A2 2 A1 )x 3
A

(6.11)

The curl points in the direction of circulation of the vector field. Written in the summation
convention,
~ i = ijk j Ak .
( A)
(6.12)
Plugging in the definitions shows that the curl of a gradient vanishes,
( )i = ijk j k = 0

(6.13)

(6.14)

as does the divergence of a curl,

~ = ijk i j Ak = 0
A
The Laplacian of a scalar quantity is
2 =
2 2 2
+
+
=
x21 x22 x23
= i i .

(6.15)

It measures the curvature at a point. The Laplacian of a vector is a vector-valued quantity

defined to be the Laplacian of each of the components of the vector
~ j = i i Aj
(2 A)

(6.16)

In addition to rectangular coordinates, there are two other common coordinate systems,
cylindrical (r, , z) and spherical (r, , ), also shown in Figure 6.1. When the coordinate

58

Electromagnetic Fields and Waves

system reflects the symmetries of a problem the math is much simpler. Inserting the
trigonometric relationships among the varaibles into the definitions of the differential
operators in rectangular coordinates, and taking the appropriate partial derivatives, shows
that in cylindrical and spherical coordinates a differential volume element is
dV = dx dy dz
= r dr d dz
= r2 sin dr d d

(6.17)

the gradient is

x +
y +
z
x
y
z
1

r +
+
z
=
r
r
z
1
1

+
r +

=
r
r
r sin

(6.18)

the divergence is
~ = Ax + Ay + Az
A
x
y
z
1 A Az
1
=
(rAr ) +
+
r r
r
z
1 2
1
1 A
= 2 (r Ar ) +
(A sin ) +
r r
r sin
r sin

(6.19)

the Laplacian is
2 2 2
+
+
x2
y 2
z 2


1

1 2 2
=
r
+ 2
+
r r
r
r 2
z 2





1
1
1
2
2
r
sin
+
+ 2
= 2
2
r r
r
r sin

r2 sin 2
|
{z
}
1 2
(r)
r r2

2 =

(6.20)

and the curl is








Az
Ax
Ay
Ay
Az
Ax
x +
y +
z

y
z
z
x
x
y






Ar
1 (rA ) Ar
A
Az
1 Az
r +
+
z

=
r
z
z
r
r
r





1
1
(A sin ) A
1 Ar
(rA )
=
r +

r sin

r sin
r


1 (rA ) Ar
.
(6.21)

+
r
r

~=
A

6.1 Vector Calculus

59

Finally, Table 6.1 lists some vector and trigonometric relationships that will be needed
later.
Table 6.1. Vector and trigonometric identities.
= 0
~ =0
( A)
~
~
~
(A) = A + A
~
~
~
(A) = A + A
~ (B
~ C)
~ = B(
~ A
~ C)
~ C(
~ A
~ B)
~
A
~ B)
~ =B
~ ( A)
~ A
~ ( B)
~
(A
cos(A B) = cos A cos B sin A sin B
sin(A B) = sin A cos B cos A sin B
cos(2A) = cos2 A sin2 A
sin(2A) = 2 sin A cos A
sin2 A + cos2 A = 1

6.1.3 Integral Relationships

The differential operators introduced in the last section measure the local properties of
scalar and vector fields. Not surprisingly, there are intimate relationships between these
local properties and the global properties of the fields. These will be very useful for
relating fields to their sources. Without proof, here are two important cases of general
theorems relating local and global properties:
Divergence Theorem (or Gauss Theorem)
Z
Z
~ dV =
~ dA
~
E
E
V

(6.22)

The volume integral of the divergence is equal to the surface integral of the normal
component of the field. V is an arbitrary volume, S is its surface, dV is a volume
~ is a surface area element. dA
~ is the same as n dA, where n is an
element, and dA
outward-pointing unit vector that is perpendicular to the patch dA. Adding up the
net flux into or out of the volume is equivalent to adding up all the local sources
and sinks.
Stokes Theorem
Z
I
~ dA
~= E
~ d~l .
E
(6.23)
S

The line integral around a closed path is equal to the surface integral of the curl
over any arbitrary surface bounded by the path (Figure 6.2). If your fingers curl in
the direction of the line integral, your thumb points in the direction of the surface
normal. The total circulation around the path is equal to the sum of all the local
circulations.

60

Electromagnetic Fields and Waves

d A
S

Figure 6.2. Definition of Stokes Theorem.

6.2 S T A T I C S
This section reviews the governing equations for time-independent electromagnetic phenomena; the following one turns on time dependence to arrive at Maxwells equations.
Together these will serve as the foundation for much of the rest of the book. As surprising
as the remarkable phenomenology contained in these apparently simple relationships is
the sophistication of the techniques needed to reveal it.
6.2.1 Electrostatics
The force in a vacuum between two charges is given by Coulombs Law:
F~ =

1 q1 q2
r (N)
40 r2

(6.24)

0 = 107 /(4c2 ) = 8.8541012 F/m is a constant called the permittivity of free space,
q1 and q2 are the size of the charges in coulombs, r2 is the distance between them in
meters, and r is a unit vector pointing between them. This relationship was determined
experimentally by Charles Augustin Coulomb in 1785.
The force on one charge due to an applied electric field is
~
F~ = q E

(6.25)

and so the electric field due to a single charge is

 
q
V
~ =

E
r
2
40 r
m

(6.26)

With this definition the electric field points from positive charge towards negative charge.
The field diverges as you approach the charge; very close to a charge the expression is
no longer valid and quantum electrodynamics is needed to describe the field.
We will shortly see that the curl of the electric field vanishes if there are no timevarying magnetic fields, which according to equation (6.13) means that the electric field
can be written as the gradient of a potential
~ =
E

Given this definition, the potential from a point charge is

q
(V) .
=
40 r

(6.27)

(6.28)

6.2 Statics

61

Since the electric field is the gradient of the potential, the potential difference between
two points ~x and ~y can be found by integrating the electric field along an arbitrary path
between the points
V (~x, ~y ) = (~y ) (~x)
Z y~
~ d~l (V)
=
E

(6.29)

~
x

It is convenient to view the electric field in terms of fictitious field lines, which are
perpendicular to the lines of constant potential and have an areal density proportional to
the field strength.
An electric field inside a material is modified by the response of the material to the
field. In a dielectric, the charge is bound so that it is not free to move, but an applied
electric field will polarize the bound charge. Because of this polarization, the strength of
the field generated by a free charge in the material will be reduced by a factor called the
permittivity
~ = q r .
(6.30)
E
4r2
The permittivity equals the relative permittivity of the material r times the permittivity of free space
= 0 r

(6.31)

r , also called the dielectric constant, is 1 in a vacuum, between 2 and 5 for typical
plastics, and can be over 100 in a material such as SrTiO3 . Depending on the symmetry
of the material the permittivity can be a tensor that depends on direction, and for strong
fields (such as those generated by a laser, Problem 6.5) it will depend nonlinearly on the
field. This latter property is very useful for mixing and generating harmonics of incident
beams of light.
Lets take the divergence of the field due to a charge and integrate over an infinitesimal
spherical volume of radius r around the charge. According to Gauss Theorem,
Z
Z
~ dV =
~ dA
~
E
E
V
S
Z
q
=
r r dA
2
S 4r
q
4r2
=
4r2
q
=
Z

dV .
(6.32)
=
V
In the last line weve introduced the charge density , which for this point charge is just
a delta function. Since the left side must equal the right side independent of the volume,
the integrands must be equal, and we see that
~ =
E

0 r

(6.33)

62

Electromagnetic Fields and Waves

~ = E,
~ and if is a constant, this reads
If we define D
~ =
D

(6.34)

~ is called the displacement field, and this is the differential form of Gauss Law. E
~ is
D
~
the real physical field that exerts a force on charges; D is the effective field that results
from source charges.

+
P

r
i

Figure 6.3. Relationship between an electric field with a gradient, the resulting spatial
variation of the polarization, and the net charge induced by the local charge imbalance.

~ field in the material is uniform, the induced polarization will be constant.

If the E
~ field varies in space (Figure 6.3) then there will be a spatially varying
However, if the E
induced polarization, leading to an average induced charge density. To understand this,
lets return to equation (6.33). The electric field in the material can be viewed as being
the sum of the field due to any free charge free that would be there if the material was
not present (r = 1), and the field due to the induced charge induced . The induced charge
is conventionally defined in terms of a polarization vector P~
induced P~

(6.35)

and so
~ = free + induced
0 E
= free P~

(6.36)

~ + P~ ) = free
(0 E
| {z }
~
D

(6.37)

or
.

~ and this relationship

If the field is not too strong then P~ will be linearly related to E,
defines the electric susceptibility e
~ 0 E
~ + P~ = 0 (1 + e )E
~ = 0 r E
~ = E
~
D

(6.38)

The dipole moment

of a charge distribution is defined as the integral of the charge
R
times the position ~x(~x) dV . To relate this to P~ , first note that by differentiating xP~
and writing out the terms,
(xP~ ) = x P~ + Px
Therefore,
Z

xinduced dV =

~ dV
x P

(6.39)

6.2 Statics

=
=

Px dV
Px dV

63

(xP~ ) dV
~
xP~ dA

(6.40)

In the limit that the volume goes to zero, P~ will be uniform and so the second term will
vanish. Dropping it and repeating the calculation for the y and z components gives
Z
Z
(6.41)
~xinduced dV = P~ dV .
Since this must be true for any volume, we see that the polarization vector is equal to
~ field, the dipole moment
the local density of the dipole moment. Note that unlike the E
is defined to point from negative charge to positive charge.
Substituting the definition of the potential into Gauss Law and assuming a homogeneous polarizibility gives Poissons equation
2 (~x) =

(~x)

(6.42)

For a point charge located at ~x0 , (~x) = q (|~x ~x0 |), and the potential is given by
equation (6.28). Plugging these into Poissons equation,
(~x)

q
q
2

= (|~x ~x0 |)
4|~x ~x0 |

1
= 4 (|~x ~x0 |)
2
|~x ~x0 |
2 (~x) =

(6.43)

This relationship can then be used to show that

Z
1
(~x )
(~x) =
d3 x
4
|~x ~x |

(6.44)

solves Poissons equation. 1/|~x ~x | is a Greens function for this problem, relating the
~ by using
field to an integral over its source. A similar solution can be found for E
~x ~x
1
=

|~x ~x |
|~x ~x |3
~ x) = (~x)
E(~
Z
1
~x ~x 3
=
(~x )
dx
4
|~x ~x |3

(6.45)

In free space, Poissons equation reduces to

2 (~x) = 0 .

(6.46)

This is called Laplaces equation, and governs many other phenomena including the
profile of a membrane such as a drumhead stretched around a boundary. One of the
properties of its solution is that it can take on an extremum only on the boundary, and
so electromagnetic particle traps require time-varying fields. The solution of Laplaces
equation requires the specification these boundary conditions, usually given by either

64

Electromagnetic Fields and Waves

the distribution of the potential on the boundary (Dirichlet boundary conditions) or its
normal derivative (Neumann boundary conditions).
Note that we found these equations by integrating Coulombs Law, which involved
canceling the r2 dependence of a three-dimensional surface area with the r2 dependence
of the field. In anything other than three dimensions this would not work. Laplaces
equation is routinely solved numerically, and to make it tractable it is frequently solved
in two dimensions, but it is very important to remember that solving Laplaces equation
in two dimensions corresponds to taking Coulombs law to have a r1 form because
the length of a two-dimensional perimeter is proportional to r. This may effectively be
correct if the problem has two-dimensional symmetry so that each point corresponds
to a line of charge, but it will be incorrect for an arbitrary two-dimensional slice of a
three-dimensional problem.
The capacitance between two electrodes relates the charge Q on each of them to the
potential difference V between them:
C=

Q
V

(6.47)

The relationship to current is found by differentiating:

C

dV
dQ
=
=I
dt
dt

(6.48)

6.2.2 Magnetostatics
Electric fields are produced by stationary charges; magnetic fields are produced by moving
~ due to an infinitesimal section of a wire
charges. The strength of the magnetic field H
carrying a current (Figure 6.4) was found experimentally in 1820 to be governed by the
BiotSavart Law
~
~ = I dl r ,
(6.49)
dH
4r2
or integrated over space
I d~l r
4r2
Z

1
~ x ) ~x ~x d3 x (A/m)
=
J(~
4
|~x ~x |3

~ =
H

(6.50)

where J~ is the current density. Using the right hand rule, if your thumb points in the
direction of current flow then you fingers will curl in the direction of the field.
~ and J~ can be written more simply by taking the curl,
The relationship between H
Z

~ x ) ~x ~x d3 x
~ x) = 1
J(~
H(~
4
|~x ~x |3
Z
1
1
~ x ) ~x
J(~
d3 x
=
4
|~x ~x |
Z
1
~ x ) ~x
~ x) = 1
~x J(~
d3 x
~x H(~
4
|~x ~x |

6.2 Statics

65

I
dl
r
r
dH
Figure 6.4. The magnetic field due to a differential current element.

1
=
4

~ = J~
H

1
1
~ x ) 2
~ x ) d3 x
~x
J(~
~x J(~
~
x
|~x ~x |
|~x ~x | | {z }
{z
}
|
0
(~x ~x )

(6.51)

~ (B
~ C)
~ = B(
~ A
~ C)
~ C(
~ A
~ B),
~ Problem 6.1). We
(using the BACCAB rule A
will soon see that a term must be added to this equation if the fields are time-varying.
The force on an infinitesimal current element is given in terms of the magnetic flux
~ by
density B
~ ,
dF~ = Id~l B
(6.52)
or for a single moving charge
~
F~ = q~v B
The continuum version is
~ =
F

~ dV
J~ B

(6.53)

(6.54)

~ and E,
~ in a material H
~ and B
~ are related
Analogous to the relationship between D
by the permeability :
~ = 0 (H
~ +M
~ ) = 0 (1 + m )H
~ = 0 r H
~ = H
~ (T)
B

(6.55)

~ is the magnetization, the

0 = 4107 H/m is the permeability of free space, M
~
relative permeability r = 1 for a vacuum, and m is the magnetic susceptibility. H
~
is the effective field that results from source currents, and B is the physical field that
~ = E,
~ the linear relationship B
~ = H
~ applies only
exerts a force on charges. As with D
to weak fields; magnetic recording depends on the nonlinear hysteresis in . Chapter
13 will explain the origin of this, as well as the reason why m < 0 for diamagnetic
materials but m > 0 for paramagnetic, ferromagnetic, and ferrimagnetic materials. For
iron m 103 ; for non-magnetic materials m 105 and so r 1 for them. In a
high-permeability alloy such as mumetal (Fe18 Ni75 Cu5 Cr2 ) the relative permeability is
105 .
~
Taking the curl of H,
~ = J~
H

66

Electromagnetic Fields and Waves

1 ~
~
BM
0
= Jfree Jinduced .
=

(6.56)

As with the electrostatic case, the effective current in the material can be decomposed
into free and induced currents. The induced current can be understood in terms of the
magnetic moment of a current distribution, defined to be
Z
1
~ x) dV .
~x J(~
(6.57)
m
~ =
2
For example, for a circular current loop for radius r,
1
r I 2r
2
= Ir2

|m|
~ =

= I area

(6.58)

The magnetic moment associated with the induced current is equal to

Z
1
~x J~induced (~x) dV
m
~ induced =
2
Z
1
~ ) dV ,
=
~x ( M
2

(6.59)

or for the ith component

mi,induced

1
=
2

ijk xj klm l Mm dV

(6.60)

Since

this can be rewritten as

l (xj Mm ) = xj l Mm + Mm l xj
|{z}
jl

mi,induced =

1
2



ijk klm l (xj Mm ) Mm jl dV

(6.61)

(6.62)

The integral over all space of the first term in the brackets will vanish. This is because the
integral over all space of the magnetization M is bounded, therefore asymptotically M
must fall off faster than 1/x. Hence limx xM (x) = 0, so the integral of the derivative
of this quantity is just its value at , which is 0. That leaves the remaining term
Z
1
ijk klm Mm jl dV
mi,induced =
2
Z


1
=
il jm im jl Mm jl dV
2
Z
1
=
il jm jl Mm im jl jl Mm dV
2
Z
1
il ml Mm im 3Mm dV
=
2

6.2 Statics

or for all components

Z
1
Mi 3Mi dV
=
2
Z
= Mi dV ,

m
~ induced =

~ dV
M

67

(6.63)

(6.64)

The magnetic dipole moment is equal to the integral of the magnetization, therefore the
magnetization is the local density of the dipole moment.
As far as we know there is no such thing as a magnetic charge, called a magnetic
~ = 0. This means that B
~ can be written as the curl of a vector field
monopole, so B
~ (equation 6.14):
A
~ =A
~ ,
B
(6.65)
~ is related to source currents by
called the vector potential. A
Z ~
J(~x ) 3

~
dx ,
A(~x) =
4
|~x ~x |

(6.66)

verified by taking the curl to obtain equation (6.50). This relation holds for static current
distributions; Section 8.1 will extend the scalar and vector potentials to time-dependent
sources.
In quantum mechanics the vector potential takes on a deep significance beyond this
formal definition. The AharonovBohm effect considers a particle moving outside an
infinite solenoid; the magnetic field vanishes there but the vector potential does not,
and this leads to observable quantum interference effects [Sakurai, 1967]. This effect
demonstrates the physical reality of the vector potential.
6.2.3 Multipoles
The theory of multipoles provides a systematic way to approximate the fields produced by
more complex charge and current distributions. One way to understand it is by expanding
the inverse distance
1
~r ~x
1
=
+
+
|~r ~x| |~r|
|~r|3
1 ~r ~x
= + 3 + ,
(6.67)
r
r
where ~r is the distance from the source to where the field is being evaluated, ~x is the
location within the source relative to its origin, and ~x is assumed to be much smaller than
~r. Substituting this series into the potential rewrites it as
Z
(~x)
1
dV
(~r) =
4
|~r ~x|
 Z

Z
1
1
~r
=
(~x) dV + 3 (~x)~x dV +
4 r
r


q p~ ~r
1

+ 3 +
.
(6.68)
4 r
r

68

Electromagnetic Fields and Waves

q is the monopole term, p~ is the dipole moment, and the next term would be the
quadrupole moment. The corresponding electric field can be found by taking the gradient
~ =
E
p r)
p~
3r(~
q r
+
+
=
2
4r
4r3
2p cos
p sin
q
+
r +
r +
=
2
3
4r
4r
4r3

(6.69)

and the energy associated with the charge distribution is

Z
U = (~x)(~x) dV
Z


= (~x) (0) + ~x (~x)|~x=0 + dV

(6.70)

~ +
= q(0) p~ E(0)

The same expansion can be used with the vector potential,

Z
Ji (~x)

dV
Ai (~r) =
4
|~r ~x|
 Z

Z
1
~r
=
Ji (~x) dV + 3 Ji (~x)~x dV +
4 r
r

(6.71)

Because of the vectorial character, finding this field is a bit more tricky. To start, notice
that for arbitrary functions f and g, integration by parts shows that
Z
Z
~ dV
f J~ g dV = 0 g (f J)
Z
Z
= g J~ f dV f g J~ dV ,
(6.72)
where the integrals are over all space, and the only assumption thats been made is that
J vanishes at infinity. Rearranging terms,
Z 

f J~ g + g J~ f + f g J~ dV = 0 .
(6.73)
If we plug in J~ = 0, and take f = 1, g = xi , then
Z
Ji dV = 0

(6.74)

and so the monopole vector potential term vanishes (there are no free magnetic charges).
Now taking f = xi and g = xj ,
Z


xi Jj + xj Ji dV = 0
(6.75)

or

1
2

xi Jj xj Ji dV =

xi Jj dV

(6.76)

6.2 Statics

This can be substituted into the dipole term to relate it to the vector potential:
Z
~r

~xJi (~x) dV
Ai (~r) =
4 r3
Z
1
=
rj xj Ji dV
4 r3
Z
1 1
(rj xi Jj rj xj Ji ) dV
=
2 4 r3
Z
~ x) dV
~ r ) = 1 ~r ~x J(~
A(~
2 4 r3
~ ~r
m
=
.
4 r3

69

(6.77)

The magnetic field is found from the curl

~ =A
~
B

m
~ ~r
=

4
r3
1

ijk j 3 klm ml rm
Bi =
4
r

rm
=
ijk klm ml j 3
4
r



3rj rm
jm
=
il jm im jl ml

4
r3
r5


3mi mi 3mi 3ri mj rj
3 3 +
=
4
r3
r
r
r5
m
m
~ r)
~
~ = 3r(
B
.
4
r3

(6.78)

This is exactly the same as the electrostatic dipole field (equation 6.69).
The force on a magnetic dipole can be derived by applying the substitution used to
find the vector potential:
Z
~ dV
~
F = J~ B
Z
~ i dV
Fi = (J~ B)
Z
= ijk Jj Bk dV
Z
= ijk Jj [Bk (0) + ~x Bk (~x)|~x=0 ] dV
Z
= 0 + ijk Jj ~x Bk dV
Z
= ijk Bk Jj ~x dV
= ijk (m
~ Bk )j

= ijk jlm ml m Bk
= (im kl il km ) ml m Bk

70

Electromagnetic Fields and Waves

= mk i Bk mi k Bk
~
~ m
~
F = (m
~ B)
~ ( B)
| {z }
0

(6.79)

Since a conservative force is the gradient of the potential energy, the energy of a magnetic
dipole in a field is
~
U = m
~ B
= mB cos

(6.80)

where is the angle between the dipole and the local field. There is an angular dependence
to this that will seek to align the dipole with the field,
U
= mB sin ,

i.e., there will be a torque about the axis perpendicular to them of

~
~ = m
~ B

(6.81)

(6.82)

Note that all of these calculations have assumed that the distance to the point where
the field is being evaluated is large compared to the special extent of this source. If the
fields are needed closer to the source its necessary to either use the full distribution or
carry the multipole approximation out to a high order.

6.3 D Y N A M I C S
6.3.1 Maxwells Equations
Were now ready for Maxwells contribution. The statics equations tell us that the di~ and the curl of E
~ vanish, and relate the divergence of D
~ and the curl of
vergence of B
~ to their sources. Faraday had found that a varying magnetic field induces a current in
H
a wire, and Amp`ere that a current produces a magnetic field; Maxwell realized that for
~
these equations to be consistent (Problem 6.2) there must also be a time derivative of D:
~ = (~x)
D
~ =0
B
~
~ = B
E
t
~
D
.
(6.83)
t
These are now called Maxwells equations; they show that electric and magnetic fields
are connected through a more general theory of electromagnetic phenomena.
~ = E
~ and B
~ = H.
~ In
Material properties appear in Maxwells equations through D
~
~
~
addition, the current J is related to the electric field by J = E. The coefficient is
the conductivity, equal to the inverse of the resistivity . In real materials , , and
can become tensors that depend on direction, and can be complex quantities because of
loss mechanisms.
~ = J(~
~ x) +
H

6.3 Dynamics

71

Weve already seen the first of Maxwells equations; integrating it over a volume gives
the integral form of Gauss Law
Z
Z
~
~
D dA =
dV = Q .
(6.84)
S

~ is equal to the charge Q enclosed.

The surface integral of the normal component of D
The second equation lacks a source term and the third one lacks a current term because
there are no (known) magnetic monopoles. Integrating the last equation over a surface
gives Stokes Law
!
Z
I
~

D
~ .
~ d~l =
dA
(6.85)
J~ +
H
t
S
The line integral of the magnetic field around a path is equal to the current crossing an
arbitrary surface bounded by the path. The first term on the right hand side of equation
~
(6.85) is the conventional current, and the second term D/t
is called the displacement
current. This acts just like a real current, but instead of charge moving it is associated
with an electric field changing. The current that flows into and out of a capacitor appears
to travel through the space between the capacitor plates; Problem 6.2 shows that this is
accounted for by the displacement current. Anyone who has been shocked by a charged
capacitor can attest to the reality of this current.

I
r
H

Figure 6.5. The magnetic field around a current-carrying wire.

It can be possible to use Stokes Law to find magnetic fields without direct integration.
For example, if a wire is carrying a current I, by symmetry according to the BiotSavart
Law the magnetic field must be directed circumferentially around the wire (Figure 6.5).
This means that the line integral is just the field strength times the circumference, and
the surface integral of the current density is equal to the current flowing through the
wire, therefore
I
.
(6.86)
2rH = I H =
2r
6.3.2 Boundary Conditions
The integral forms of Maxwells equations can be used to find the conditions that the
fields must satisfy at interfaces between materials, as specified by the dielectric constant
, the conductivity , and the permeability . Start by integrating Gauss Law over the
volume V in Figure 6.6:
Z
Z
Z
~ dA
~ = dV .
~ dV = D
(6.87)
D

72

Electromagnetic Fields and Waves

In the limit that the height of the box h 0, the only contributions to the surface
integral will come from the top and bottom. If the box if infinitesimal then the fields can
be taken to be constant, and so the surface integral is just the normal component of the
field times the area:
Z
~1 D
~ 2 ) nA
=
(D

dV

(6.88)

where n is a unit vector normal to the interface. There is a sign change between the
integrals over the top and the bottom because the surface normal changes directions. If
there is charge at the interface with an areal density s , then
Z
~ = s A .
dV
(6.89)
V

Therefore,
~1 D
~ 2 ) n = s
(D

(6.90)

~ across the interface is equal to the charge

The change in the normal component of D
density at the interface. An applied field will create such surface charge to match the
boundary conditions.

e1 s1 m1

l
h
S

e2 s2 m2
Figure 6.6. Loop and volume used for evaluating boundary conditions.

~ over the surface S in Figure 6.6:

Next, integrate the curl of E
Z
Z
I
~
B
~
~ .
~
~
~
dA
E dA = E dl =
S t

(6.91)

As before, if the height h of the loop goes to zero then the only contribution to the line
integral comes from the top and bottom, and the integral over the surface on the right
hand side vanishes. Therefore
~1 E
~ 2 ) n l = 0
(E

(6.92)

or
~1 E
~ 2 ) n = 0
(E

(6.93)

73

6.3 Dynamics

~ is continuous across the interface.

The tangential component of E
~
~
~ field in an ideal
Since J = E, and in a perfect conductor = , there can be no E
conductor otherwise there would be an infinite current. According to equation (6.93), this
means that there can be no tangential component on either side of the interface. Equation
(6.90) does permit a normal component outside the interface, but it must be screened
by a surface charge s = D1 . We see that the electric field must be perpendicular to the
surface of a perfect conductor, and that a surface charge is induced to screen the interior
from the field.
~ over the volume as we did for the divergence of D
~
Integrating the divergence of B
shows that
~1 B
~ 2 ) n = 0
(B

(6.94)

~ field is continuous across the interface. Similarly, inteThe normal component of the B
~
~ gives
grating the curl of H over the surface instead of the curl of E
~1 H
~ 2 ) n = J~s
(H

(6.95)

~ changes
where J~s is the density of current at the surface. The tangential component of H
by the surface current density across the interface.
The solution to Laplaces equation is unique. This means that any solution that satisfies
the boundary conditions, no matter how it is found, is the solution. This observation leads
to the useful method of images. Consider a point charge about an infinite ground plane,
shown in Figure 6.7. A fictitious image charge is shown equidistant below the plane.
By symmetry, the electric field lines are perdendicular to the surface. This is exactly the
boundary condition that must be satisfied for a perfect conductor, hence to find the field
above the plane the continuous surface charge distribution can be replaced by the single
image charge. There are some geometries like this for which its possible to use an image
charge to easily solve for the fields; in more complex geometries image charges are still
useful as an expansion technique based on iterating the induced image charge produced
by the source image charge.

Ez

Figure 6.7. Image charge solution for the field of a charge above a ground plane.

74

Electromagnetic Fields and Waves

6.3.3 Electromagnetic Units

There are two common systems of electromagnetic units: MKS that weve used here, and
Gaussian or CGS. MKS uses familiar macroscopic quantities (volts, amps, ohms) and
hence is most suitable for macroscopic phenomena and is commonly used in engineering;
CGS is better matched to microscopic phenomena and is commonly used in physics.
Table 6.2 gives the governing equations in these two systems, and in Table 6.3 the (nonobvious) conversion factors are summarized. In addition to these conventional definitions
there are many other possible systems offering endless opportunities to go astray, ranging
from minor variations of these to the theorists favorite system in which all fundamental
constants are set equal to 1, with units being put back in at the end of a calculation based
on dimensional grounds. [Jackson, 1999] has a thorough discussion of the logic of, and
the relationships among, the systems.
Table 6.2. MKS and CGS governing equations.
MKS

CGS

~ + P~
~ = 0 E
D
~ M
~
~ = 1B
H
0
~ =
D
~ =0
B

~ =E
~ + 4 P~
D

~
~ = B
E
t
~
~ = J~ + D
H
t
~
~
~
F = q E + q~v B

~ =B
~ 4 M
~
H
~ = 4
D
~ =0
B
~
~ = 1 B
E
c t
~
4
1 D
~ =
H
J~ +
c
c t
~v
~
~
~
F = qE + q B
c

6.4 R A D I A T I O N A N D E N E R G Y
6.4.1 Waves
Perhaps the single most remarkable feature of Maxwells equations is that they contain a
wave solution. To see this, start with the equations for free space
~ =0
D
~ =0
B
~
~ = 0 H
E
t
~
~ = 0 E .
H
t

(6.96)

75

6.4 Radiation and Energy

Table 6.3. Conversion between MKS and CGS units. All prefactors of 3 are
actually 2.99792..., from the speed of light.
Quantity
charge
current
potential
polarization
electric field
displacement
resistance
capacitance
magnetic flux
magnetic induction
magnetic field
magnetization
inductance

Symbol
q
I
V
P
E
D
R
C

B
H
M
L

MKS

CGS

1 coulomb
1 ampere
1 volt
1 coulomb/m2
1 volt/m
1 coulomb/m2
1 ohm
1 farad
1 weber
1 tesla
1 ampere/m
1 ampere/m
1 henry

3109 statcoulombs
3109 statamps
1/300 statvolt
3105 dipole moment/cm3
1
4
statvolt/cm
3 10
4 3105 statvolt/cm
1
1011 s/cm
32
2
3 1011 cm
108 gausscm2 (maxwells)
104 gauss
4103 oersted
103 magnetic moment/cm3
1
11
stathenry
32 10

Take the curl of the last two equations:

~ = 0
E

~
H
t

~
~ = 0 E
H
t

(6.97)

~ = ( E)
~ 2 E
~
E

(6.98)

These can be simplified with the identity (Problem 6.1)

to give
~ = 0
2 E
~ = 0
2 H

~
H
t

~
E
t

(6.99)

~ and B.
~ Substituting in the curls from equation
noting that the divergences will be 0 for E
(6.96),
~
2E
t2
2~
~ = 0 0 H .
2 H
t2
~ = 0 0
2 E

(6.100)

These are wave equations, solved by a plane wave for the electric field
~ x, t) = E
~ 0 ei(~k~xt)
E(~

(6.101)

with the wave vector ~k pointing in the direction of propagation with a magnitude |k| =
2/, and the velocity c = /|k| = (0 0 )1/2 . Let there be light.

76

Electromagnetic Fields and Waves

~ =E
~ 0 ei(~k~xt) and H
~ =H
~ 0 ei(~k~xt) ,
If we take E
~
H
t
~
~
~
ik E = i0 H
~ = 0
E

k ~
~
kE =H
0
r
0 ~
~
kE =H
0

(6.102)

Similarly,

0
~ =E
~
kH
0

(6.103)

The electric and magnetic fields of a plane electromagnetic wave are perpendicular to
Their ratio
each other and to the direction of travel k.
r
~
0
|E|
377
(6.104)
=
~
0
|H|
has the units of resistance and defines the impedance of free space. It will return in
Chapter 8 in the effective impedance of antennas.
Since its reasonable to assume that a wave travels in a medium, the recognition of
the wave solution to Maxwells equations led to a search for the ether that supports
the wave. The failure of the MichelsonMorely experiment in 1887 to detect the motion of the Earth through the ether helped plant the seeds for the discovery of quantum
mechanics, and special relativity. The resolution of the paradox is that electromagnetic
waves are carried by photons, which are particles that travel in free space but which also
act like waves. One way to understand electromagnetic propagation is to remember that
information cannot travel faster than the speed of light. If a charge is moved instantaneously, there is a kink in its electric field that travels out at the speed of light: that is
an electromagnetic wave packet. Moving the charge periodically creates a wave.
6.4.2 Electromagnetic Energy
If electromagnetic waves can propagate, and if an electromagnetic field can accelerate a
charge that is initially at rest, then it must be possible to store and transmit energy in the
fields. In this section we will calculate that energy.
~ If the charge moves a distance d~x,
A charge in an electric field feels a force F~ = q E.
~
work dW = q E d~x is done against this force. If the charge is moving at a velocity ~v
~ ~v . If
then the rate at which work is being done, or power is being consumed, is W = q E
~
there is a continuous current density J, the total rate of work is this quantity integrated
over space
Z
~ J~ dV .
W =
E
(6.105)
V

There is no work done by a magnetic field alone on a charge, because the magnetic force
is perpendicular to the velocity:
~ ~v = (q~v B)
~ ~v = 0
F

(6.106)

6.4 Radiation and Energy

77

Since
~
~
~ = J~ + D J~ = H
~ D ,
H
t
t
equation (6.105) can be rewritten as
#
Z "
~

D
~ ( H)
~ E
~
E
dV .
W =
t
V

(6.107)

(6.108)

This in turn can be rewritten by using the vector identity

~ H)
~ =H
~ ( E)
~ E
~ ( H)
~
(E
as
W =

Z "
V

~
~ ( E)
~ (E
~ H)
~ E
~ D
H
t

(6.109)

dV

~ = B/t,
~
Plugging in E
#
Z "
~
~

B
~ H)
~ +E
~
~
W =
(E
+H
dV
t
t
V

(6.110)

(6.111)

~ = E,
~
Since D
~
~
~ ~
E
~ +E
~ D
(E D) =
D
t
t
t
~

D
~
= 2E
t

(6.112)

and similarly
~
~ ~
~ B
(B H) = 2H
t
t

(6.113)

Therefore, if we define
1 ~ ~ ~ ~
D + B H)
U = (E
2

J
m3

(6.114)

then equation (6.111) becomes

W =

Z 
V

~ H)
~ + U
(E
t

dV

(6.115)

with
~
~
U
~ D + H
~ B
=E
t
t
t

(6.116)

Further defining
~ H
~
P~ = E

J
m2 s

the first term can be turned into a surface integral:

Z
Z
U
~
dV
W = P~ dA
S
V t

(6.117)

(6.118)

78

Electromagnetic Fields and Waves

This has a very natural interpretation. The first term represents an energy flux transported
across the boundary of the integration volume by the field, and the second term represents
the change in the energy stored in the volume by the field. P~ is called the Poynting vector,
~ has nothing to do with the
and U is the energy density. Note that the Poynting vector P
~
polarization vector P , they just use the same symbol by convention. Integrating P over
an area gives the energy being carried by an electromagnetic wave; integrating U over a
volume gives the energy stored in a static field.
Since the energy stored in an electric or magnetic field is equal to the volume integral
of the square of the field strength, field lines behave like furry rubber bands in finding
the lowest-energy configuration. They want to be as short as possible to minimize the
volume of the integral, and they want to be as far apart from each other as possible to
minimize the field density and hence the quadratic energy density.

6.5 S E L E C T E D R E F E R E N C E S
[Jackson, 1999] Jackson, John David. (1999). Classical Electrodynamics. 3rd edn. New
York: Wiley.
The definitive electrodynamics reference.
[Heald & Marion, 1995] Heald, Mark A., & Marion, Jerry B. (1995). Classical
Electromagnetic Radiation. 3rd edn. Fort Worth: Saunders.
Less depth than Jackson, but a more accessible introduction to electrodynamics.

6.6 Problems
(6.1) Prove the BACCAB rule
~ (B
~ C)
~ = B(
~ A
~ C)
~ C(
~ A
~ B)
~
A

(6.119)

by writing it out in the summations convention, and use it to show that

~
~ = ( E)
~ 2 E
( E)

(6.120)

(6.2) (a) Use Gauss Law to find the capacitance between two parallel plates of area A
at a potential difference V and with a spacing d. Neglect the fringing fields by
assuming that this is a section of an infinite capacitor.
(b) Show that when a current flows through the capacitor, the integral over the
internal displacement current is equal to the external electrical current.
(c) Integrate the energy density to find the stored energy at a fixed potential. The
answer should be expressed in terms of the capacitance.
(d) Batteries are rated by amp-hours, the current they can supply at the design
voltage for an hour. Consider a 10 V laptop battery that provides 10 A h.
Assuming a plate spacing of 106 m 1 m and a vacuum dielectric, what area
would a capacitor need to be able to store this amount of energy? If such plates
were 10 cm on a side and stacked vertically, how tall would the stack have to
be to provide this total area?

6.6 Problems

79

(6.3) (a) Use Stokes Law to find the magnetic field of an infinite solenoid carrying a
current I with n turns/meter.
(b) Integrate the energy density to find the energy stored in a solenoid of radius r
and length l, once again neglecting fringing fields.
(c) Consider a 10 T MRI magnet (Section 10.4) with a bore diameter of 1 m and
a length of 2 m. What is the outward force on the magnet? Remember force
is the gradient of potential for a conservative force.
(6.4) Calculate the force per meter between two parallel wires one meter apart, each
carrying a current of one ampere (this is the geometry used to define the ampere).
(6.5) (a) Assume that sunlight has a power energy density of 1 kW/m2 (this is a peak
number; the typical average value in the continental USA is 200 W/m2 ).
Estimate the electric field strength associated with this radiation.
(b) If 1 W of power is focused in a laser beam to a square millimeter, what is the
field strength? What about if it is focused to the diffraction limit of 1 m2 ?