Maxwell Equations Fundamentals

1
FUNDAMENTALS
1.1 Maxwell Equations

A. Spatial Frequency k
B. Vector Analysis and Boundary Conditions
1.2 Polarization
Topic 1.2A Stokes Parameters and Poincare Sphere
1.3 Lorentz Force Law
A. Lenz Law and Electromotive Force (EMF)
B. Poyntings Theorem and Poynting Vector
C. Momentum Conservation Theorem
1.4 Hertzian Waves
Topic 1.4A Electric Field Pattern
1.5 Reection and Transmission
A. Reection and Transmission of TE Waves
B. Reection and Transmission of TM Waves
C. Brewster Angle and Zero Reection
1.6 Phase Matching
A. Critical Angle and Total Reection
B. Guided Waves in a Dielectric Waveguide
1.7 Wave Guidance
A. Guidance by Conducting Parallel Plates
B. Guidance by Rectangular Waveguides
1
1. Fundamentals
C. Rectangular Cavity Resonators

1.8 Constitutive Relations
A. Anisotropic and Bianisotropic Media
Topic 1.8A Constitutive Matrices
1.9 Boundary Conditions
Topic 1.9A Derivation of Boundary Conditions
Topic 1.9B Boundary Conditions for Moving Boundaries
Answers
1.1
Maxwell Equations
The laws of electricity and magnetism were established in 1873 by James

Clerk Maxwell (18311879). In three-dimensional vector notation, the Maxwell equations are
D (r, t) + J (r, t)
t
E (r, t) = B (r, t)
t
H (r, t) =
(1.1.1)
(1.1.2)
D (r, t) = (r, t)
(1.1.3)
B (r, t) = 0
(1.1.4)
where E, B, H, D, J, and are real functions of position and time.

E (r, t) = electric eld strength
(volts/m)
B (r, t) = magnetic ux density
(webers/m2 )
H (r, t) = magnetic eld strength
(amperes/m)
D (r, t) = electric displacement
(coulombs/m2 )
J (r, t) = electric current density
(amperes/m2 )
(r, t) = electric charge density
(coulombs/m3 )
Equation (1.1.1) is Ampères law or the generalized Ampère circuit law.

Equation (1.1.2) is Faradays law or Faradays magnetic induction law.
Equation (1.1.3) is Coulombs law or Gauss law for electric elds.
Equation (1.1.4) is Gauss law or Gauss law for magnetic elds.
Maxwells contribution to the laws of electricity and magnetism is the addition of the displacement term D/t in Ampères law (1.1.1).
The electric current density J (r, t) and the electric charge density
(r, t) are governed by the continuity law
J (r, t) =
(r, t)
t
(1.1.5)
which states that the electric current and charge densities at r are conserved.
The divergence of current J from an innitesimal volume surrounding r is
equal to the decreasing of electric charge density with time t .
1. Fundamentals
Constitutive Relations for Free Space

The Maxwell equations are fundamental laws governing the behavior of
electromagnetic elds in free space and in media. Free space is characterized
by the following constitutive relations:
D = o E
B = o H
where
(1.1.6a)
(1.1.6b)
o 8.85 1012
farad/meter
o = 4 107
henry/meter
are, respectively, the permittivity and the permeability of free space. Giving
the velocity of light in free space being c = 3 108 m/s, the permittivity
o = 1/(o c2 ) , which follows from the dispersion relation as derived below.
Wave Equation
The Maxwell equations in dierential form are valid at all times for
every point in space. First we shall investigate solutions to the Maxwell
equations in regions devoid of source, namely in regions where J = 0 and
= 0 . This of course does not mean that there is no source anywhere in all
space. Sources must exist outside the regions of interest in order to produce
elds in these regions. Thus in source-free regions in free space, the Maxwell
equations become
E
t
E = o H
t
E =0
H =0
H = o
(1.1.7)
(1.1.8)
(1.1.9)
(1.1.10)
In the form of scalar partial dierential equations, we have
Hz
Hy = o Ex
y
z
t
Hx
Hz = o Ey
z
x
t
Hy
Hx = o Ez
x
y
t
(1.1.11a)
(1.1.11b)
(1.1.11c)
Ez
Ey = o Hx
y
z
t
Ex
Ez = o Hy
z
x
t
Ey
Ex = o Hz
x
y
t
Ex +
Ey +
Ez = 0
x
y
z
Hx +
Hy +
Hz = 0
x
y
z
(1.1.12a)
(1.1.12b)
(1.1.12c)
(1.1.13a)
(1.1.13b)
A wave equation for E can be derived by eliminating H from (1.1.11) and

(1.1.12). Taking time derivatives of (1.1.11a) and substituting (1.1.12c) and
(1.1.12b), we have

Ey
Ex +
Ex
Ez
o o 2 Ex =
t
y x
y
z z
x

2
2
2
Ex
=
+
+
x2 y 2 z 2
where use is made of (1.1.13).
the three components of E :
2
2
+
x2 y 2
2
2
+
x2 y 2
2
2
+
x2 y 2
Thus we obtain the following equations for

2
2
+ 2 o o 2 Ex = 0
z
t

2
2
+ 2 o o 2 Ey = 0
z
t

2
2
+ 2 o o 2 Ez = 0
z
t
(1.1.14a)
(1.1.14b)
(1.1.14c)
Making use of the Laplacian operator 2 in rectangular coordinate system

2 =
2
2
2
+
+
x2 y 2 z 2
(1.1.15)
we have
2 E o o
2
E=0
t2
This is known as the Helmholtz wave equation.
(1.1.16)
1. Fundamentals
Wave Solution
Solutions to the wave equation (1.1.16) that satisfy all Maxwell equations are electromagnetic waves. We shall now study a solution to (1.1.14a)
assuming Ey = Ez = 0 . Let Ex be a function only of z and t and independent of x and y . The electric eld vector can be written as
E=x
Ex (z, t)
The wave equation it satises follows from (1.1.16) which becomes
2
2
E

Ex = 0
x
o
o
z 2
t2
(1.1.17)
The simplest solution to (1.1.17) takes the form

E=x
Ex (z, t) = x
E0 cos(kz t)
(1.1.18)
Substituting (1.1.18) in (1.1.17) we nd that the following equation,

called the dispersion relation, must be satised:
k 2 = 2 o o
(1.1.19)
The dispersion relation provides an important connection between the spatial

frequency k and the temporal frequency .
There are two points of view useful in the study of a space-time varying
quantity such as Ex (z, t) . The temporal view point is to examine the time
variation at xed points in space. The spatial view point is to examine spatial
variation at xed times, a process that amounts to taking a series of pictures.
From the temporal view point, we rst x our attention on one particular point in space, say z = 0 . We then have the electric eld Ex (z =
0, t) = E0 cos t . Plotted as a function of time in Fig. 1.1.1, we nd that
the waveform repeats itself in time as t = 2m for any integer m . The
period is dened as the time T for which T = 2 . The number of periods
in a time of one second is the frequency f dened as f = 1/T , which gives
f=
(1.1.20)
The unit for frequency f is Hertz (Hz) with 1 Hz = 1 s1 , which is equal to

the number of cycles per second. Since = 2f , is the angular frequency
of the wave.
T = 2
Ex (z = 0, t) = E0 cos t
Figure 1.1.1 Electric eld strength as a function of t at z = 0.
1 sec
1 sec
1 sec
Ex = E0 cos t
Ex = E0 cos 2t
Ex = E0 cos 3t
a. = o = 2 Hz
b. = 2o = 4 Hz
c. = 3o = 6 Hz
Figure 1.1.2 Electric eld strength vs. t for dierent frequencies .
In this book, we often refer to as the frequency, simply because

is more commonly encountered than f . The temporal frequency characterizes the variation of the wave in time. We plot in Fig. 1.1.2a Ex (z = 0, t)
as a function of t instead of t . Let there be one period within the time
interval of 1 second. Thus f = fo = 1 Hz , and we let = o = 2 rad/s.
In Fig. 1.1.2b, we plot = 2o ; there are two periods in a time interval of
one second and the period in time is 0.5 seconds. In Fig. 1.1.2c, = 3o
and there are three periods in one second.
1. Fundamentals
A. Spatial Frequency k
To examine wave behavior from the spatial view point, we let t = 0 and
plot Ex (z, t = 0) in Fig. 1.1.3. The waveform repeats itself in space when
kz = 2m for integer values of m . The spatial frequency k characterizes
the variation of the wave in space. The wavelength is dened as the
distance for which k = 2 . Thus = 2/k , or
k=
(1.1.21)
We call k the spatial frequency or the wavenumber which is equal to the

number of wavelengths in a distance of 2 and has the dimension of inverse
length.
kz
k = 2
Ez (z, t = 0) = E0 cos kz
Figure 1.1.3 Electric eld strength as a function of kz at t = 0.
To further understand the meaning of k as a spatial frequency, we plot

in Fig. 1.1.4a Ex (z, t = 0) as a function of z instead of kz . Let there be
one period within the wavelength of 1 meter. We dene Ko = 2 rad/m .
Thus k = 1Ko = 2 rad/m . In Fig. 1.1.4b, we plot k = 2 Ko ; there are two
periods in a spatial distance of one meter and the wavelength is 0.5 meters.
In Fig. 1.1.4c, k = 3 Ko and there are three periods in one meter.
We dene for the spatial frequency k a fundamental unit Ko :
1Ko = 2 rad/m
(1.1.22)
Similar to the unit Hz which is cycles per second in temporal variation, Ko is

cycles per meter in spatial variation. For a wave that has a spatial frequency
1m
1m
Ex = E0 cos Ko z
Ex = E0 cos 2Ko z
a. k = Ko
b. k = 2Ko
1m
Ex = E0 cos 3Ko z
c.
k = 3Ko
Figure 1.1.4 Electric eld strength vs. distance z with dierent spatial frequency k .
of one period in one meter distance, we have k = 1 Ko . An electromagnetic

wave in free space with k = 5 Ko has ve spatial periods in a distance of one
meter. From the dispersion relation for electromagnetic waves, the spatial
frequency and the temporal frequency are related by the velocity of light.
In free space, the conversion factor is c = (o o )1/2 = 3 108 m/s . Thus
for a spatial frequency of 1 Ko , the corresponding temporal frequency is
f = 300 MHz .
Within the spatial frequency range of 0.01 Ko to 100 Ko electromagnetic waves are used for microwave heating, radar, navigation, and carrying
signals from radio, television, and satellite communications. The visible light
has a spatial frequency band between 1.4 106 2.6 106 Ko . In Fig. 1.1.5
we illustrate the electromagnetic wave spectrum according to the spatial frequency in Ko and corresponding wavelength in meters, frequency in Hz, and
energy in electron-volts (eV).
In this book we shall place great emphasis on the use of k , which
is of more fundamental importance in electromagnetic wave theory than
both of the more popular concepts of wavelength and frequency f .
The corresponding values of wavelength and frequency are, for k = A Ko ,
= 2/k = 2/(AKo ) = 1/A m and f = ck/2 = cAKo /2 = 3 108 A Hz .
The photon energy in electron-volts (eV) is calculated from h
Joule =
34
Joule-second is Plancks constant dihck/q eV , where h = 1.05 10
vided by 2 and q = 1.6 1019 coulombs is the electron charge. Thus
h = (
hcAKo /q) eV 1.24 106 A eV .
10
1. Fundamentals
k
1012
1012
1010
10
MKo
Ko = 2 m
k = A Ko
= 1/A m
f = 300A MHz
h = 1.24 106 A eV
10
MeV
-Ray
f
1020
X-Ray
1018
keV
Ultraviolet
1016
Visible (1.4 2.6 MKo ; 417 789 THz )

Near Infrared (0.77 1.4 MKo ; 231 417 THz )
eV
1014
Thermal Infrared (0.067 0.14 MKo ; 20 43 THz )

104
104
THz
mm
10
Ko
EHF
Millimeter Wave
cm
SHF
Radar C, X, Ku , K bands (13.3 90 Ko ; 4 27 GHz )
UHF
Microwave Oven (8.17 Ko ; 2.45 GHz )

Television (1.57 2.97 Ko ; 470 890 MHz )
VHF
Television (.18 .72 Ko

FM Radio (.29 .36 Ko
HF
Short Wave Radio (.01 .001 K o
MF
AM Radio (.00178 .00535 Ko
535 1605 kHz)
LF
(104 103 K o ; 30 300 kHz)
102
km
104
; 54 216
; 88 108
GHz
MHz )
MHz )
3 30 MHz )
MHz
VLF (105 104 Ko ; 3 30 kHz)
106
106
108
10 8
m
ULF (106 105 Ko ; 0.3 3 kHz)
kHz
SLF (107 106 Ko ; 30 300 Hz)

ELF (108 107 Ko ; 3 30 Hz)
eV
Spatial
)
Ko ( Frequency
Hz
Temporal ) Hz
( Frequency
Figure 1.1.5 Electromagnetic wave spectrum.
11
Phase Velocity and Phase Delay

In Figs. 1.1.6b and 1.1.6c we plot Ex (z, t) at two progressive times
t = /2 and t = . We observe that the electric eld vector at A
appears to be propagating along the z direction as time progresses. The
velocity of propagation Vp is determined from kz t = constant which
gives
dz
Vp =
=
(1.1.23)
dt
k
We call Vp the phase velocity. By virtue of the dispersion relation (1.1.19),
we see that Vp = (o o )1/2 , which is equal to the velocity of light in free
space c .
kz
kz
kz
3
2
a. t = 0
Ex = E0 cos kz
A
Ex
Ex
b. t =
Ex = E0 sin kz
Ex
c. t =
Ex = E0 cos kz
Figure 1.1.6 Electric eld strength vs. kz at dierent times.
The spatial frequency k is, according to the dispersion relation, directly

related to the temporal frequency by the phase delay
p =
= o o
(1.1.24)
which determines how much time it takes for the wave to propagate a unit
distance. In free space p = 108 /3 s/m or it takes 3.33 nanoseconds for an
electromagnetic wave to travel the distance of one meter.
12
Example 1.1.1
1. Fundamentals
Operating frequencies of common devices:
Device
AM Radio
Shortwave Radio
FM Radio
Airport ILS
Commercial Television
Channels 2-4
Channels 5-6
Channels 7-13
Channels 14-83
Microwave Oven
Communication Satellite
Downlink
Uplink
Temporal frequency (Hz)

535 1605 kHz
3 30 MHz
88 108 MHz
108 112 MHz
Spatial frequency (Ko )

0.00178 0.00535 Ko
0.01 0.1 Ko
0.293 0.36 Ko
0.35 0.373 Ko
54 72 MHz
76 88 MHz
174 216 MHz
470 890 MHz
2.45 GHz
0.18 0.24 Ko
0.253 0.293 Ko
0.58 0.72 Ko
1.57 2.97 Ko
8.17 Ko
3.70 4.20 GHz

5.925 6.425 GHz
12.3 14 Ko
19.75 21.4 Ko
End of Example 1.1.1
Electric Field Vector E and Magnetic Field Vector H

For the wave solution in (1.1.18) for the vector electric eld E ,
E=x
Ex (z, t) = x
E0 cos(kz t)
(1.1.25)
the vector magnetic eld H can be determined from (1.1.8). We nd

x

o H = E = x y z = y kEo sin(kz t)
t

E
0
0
x
The magnetic eld vector H is then
H = y
k
E0 cos(kz t)
o
(1.1.26)
Equations (1.1.25) and (1.1.26) are seen to satisfy all Maxwell equations
E
t
E = o H
t
E =0
H = o
H =0
(1.1.27)
(1.1.28)
(1.1.29)
(1.1.30)
13
eld
n
ag
eti
z
M
Ele
ctr
ic
eld
Figure 1.1.7 Electric and magnetic eld vectors of an electromagnetic wave.
Write the amplitude of the magnetic eld vector H as H0

H = yHy (z, t) = yH0 cos(kz t)
(1.1.31)

where H0 = E0 / and = o / o is called the free-space impedance. The
electromagnetic wave is propagating in the positive z direction because as
time t increases, z must increase in order to maintain a constant phase
kz t . The eld vectors of the electromagnetic wave are transversal to the
direction of propagation and lie in the xy -plane, on which the phase kz t
of the wave is a constant. Since the phase front of the wave is the xy -plane,
we call the electromagnetic wave as represented by (1.1.25) and (1.1.31) a
plane wave.
Exercise 1.1.1
An electromagnetic wave propagating in the negative z direction
E=x
Ex (z, t) = x
E0 cos(kz + t)
(Ex1.1.1.1)
because as time t increases, z must decrease in order to maintain kz + t a

constant. The associated magnetic eld vector H takes the form
H = yHy (z, t) = yH0 cos(kz + t)
where H0 = E0 / and =
o / o is the free-space impedance.
(Ex1.1.1.2)
14
1. Fundamentals
Personalities in Electromagnetics
James Clerk Maxwell (13 June 1831 5 November 1879)
James Clerk Maxwell attended University of Edinburgh (18471850),
and studied under William Hopkins at Cambridge University (18501854).
He was a fellow of Trinity (18551856), Professor of Natural Philosophy at
Marischal College of the University of Aberdeen (18561860), and at Kings
College (18601865). He was the rst Cavendish Professor of Experimental
Physics at Cambridge University to build and direct the Cavendish Laboratory (18711879). He published four books and about 100 papers starting
age 14. Inspired by Faradays concept of lines of force, Maxwell published
his papers On Faradays Lines of Forces in 1855, On Physical Lines of
Force in 1861, and A Dynamical Theory of the Electromagnetic Field in
December 1864. In 1865, at age 33, he retired to his country home estate and
spent six years to write his monumental book A Treatise of Electricity and
Magnetism (Constable and Company, London, 1873; Dopver Publications,
New York, 1006 pages, 1954).
Originally written in Cartesian component form, the Maxwell equations
were cast in the vector form by Oliver Heaviside (18 May 1850 3 February
1925). Heinrich Rudolf Hertz (22 February 1857 1 January 1894) experimentally veried Maxwells theory in 1888. Since then, electromagnetic theory has played a central role in the development of radio, television, wireless
communications, radar, microwave heating, remote sensing, and numerous
other practical applications. The special theory of relativity developed by Albert Einstein (14 March 1879 18 April 1955) in 1905 further asserted the
rigorousness and elegance of Maxwells theory. As a well-established scientic
discipline, this sophisticated theoretical structure embodies many principles
and concepts which serve as fundamental rules of nature and vital links to
other scientic disciplines.
Michael Faraday (22 September 1791 25 August 1867)
Faraday became an assistant to Sir Humphry Davy at the Royal Institution on 1 March 1813. In September 1821, his experimentation demonstrated electro-magnetic rotation, initiated the concept of electric motor. In
August 1831, he discovered electro-magnetic induction, and that magnetism
produced electricity through movement, the principle behind the electric
transformer and generator. He became professor of chemistry in 1833. One
of his most important contributions to physics was his development of the
concept of lines of force leading to the development of the concept of elds
by Maxwell. Faraday published many of his results in the three-volume Experimental Researches in Electricity (18391855).
15
Johann Carl Friedrich Gauss (30 April 1777 23 February 1855)

Gauss studied mathematics at the University of G
ottingen from 1795 to
1798, and received his doctoral degree from the University of Helmstedt in
1799. In 1807 he took the position of director of the G
ottingen Observatory.
In 1832 he presented a systematic use of absolute units (length, mass, time)
to measure nonmechanical quantities. From 1831 to 1837 he worked closely
with Weber on terrestrial magnetism and organized a system of stations for
systematic observations.
Andr
e-Marie Amp`
ere ( 20 January 1775 10 June 1836)
Ampère was appointed professor at Bourg Ecole Centrale in 1802, at
the Ecole Polytechnique in 1809, and at Universite de France in 1826. In
September 1820, Ampère showed that two parallel conductors attract each
other if they carry currents that ow in the same direction and repel if the
currents ow in opposite directions. In 18231826, he completed his Memoir
on the Mathematical Theory of Electrodynamic Phenomena, Uniquely Deduced from Experience, which contains the description of his experiments
and a mathematical formulation of the laws that govern the interaction of
electric currents with magnetic elds.
Charles-Augustin de Coulomb (14 June 1736 23 August 1806)
Coulomb worked in the Corps du Genie until he retired in 1791. In
1777 he invented the torsion balance, which enabled him to establish the
fundamental laws of electricity by measuring the force between two small
spheres charged with electricity. Between 1785 and 1791, he published seven
treatises on electricity and magnetism. Using the torsion balance method,
he established laws of attraction and repulsion, the electric point charges,
magnetic poles, distribution of electricity on the surface of charged bodies
and others.
Hermann Ludwig Ferdinand von Helmholtz (31 August 1821
8 September 1894)
Hermann von Helmholtz was a professor of anatomy and physiology at
the University of Bonn in 1858, then became a professor of physics at the
University of Berlin in 1871, and the rst director of the Physico-Technical
Institute of Berlin in 1888. His 3-volume Handbook of Physiological Optics
appeared between 1856 and 1867. Helmholtz, who was born ten years earlier
than Maxwell and died eight months later than his student Hertz death, thus
witnessed the whole development and triumphant verication of Maxwells
electromagnetic wave theory.
16
1. Fundamentals
B. Vector Analysis and Boundary Conditions

A vector has a magnitude and a direction. We write vector A as
A=a
A
is a dimensionless vector with a unit
where A is the magnitude of A and a
magnitude pointing in the direction of A . The vector A can be represented
graphically by a directed straight-line element of length A and pointing in
the direction of a
.
Two vectors A and B , when they are not in the same direction or in
opposite directions, determine a plane. The addition and subtraction of A
and B are illustrated graphically in Fig. 1.1.8.
B
A+B
AB
AB
B
(a)
(b)
Figure 1.1.8 Addition and subtraction of A and B .
The scalar or dot product of A and B , denoted by A B , is a scalar

number:
A B = AB cos AB
where AB is the angle between A and B .
AB
AB
Figure 1.1.9 Cross product A B .
17
z
Az
A
z
y
x
Ay
Ax
Figure 1.1.10 Projection of A in rectangular coordinate system.
The vector or cross product of two vectors A and B , denoted by AB ,

is a vector perpendicular to the plane containing A and B . Thus A B
is perpendicular to both A and B [Fig. 1.1.9]. The magnitude of A B
is |AB sin AB | , which is equal to the area of the parallelogram formed by
A and B . Its direction follows the right-hand rule, i.e., when the ngers of
the right hand rotate from A to B , the thumb of the right hand points in
the direction of A B . Division by a vector is not dened; thus B/A , 1/A
are meaningless expressions.
If none of the operations of addition, subtraction, dot product, or cross
product is imposed on A and B , the entity A B is called a dyad. In the
language of tensor analysis, a dyad is a tensor of second rank, while all
vectors are tensors of rst rank.
Any vector can be represented by three components projected onto a
Cartesian coordinate system (also called the rectangular coordinate system)
with three orthogonal unit vectors x
, y, and z where x
= y z, y =
z x
, z = x
y, x
x
= y y = z z = 1, and x
y = y z = z x
= 0. We
, y, z axes [Fig. 1.1.10].
write Ax , Ay , Az as projections of A onto the x
Position Vector r = x
x + yy + zz , whose three components are (x, y, z) .
18
1. Fundamentals
For the two vectors A and B , we write

A=x
Ax + yAy + zAz
B=x
Bx + yBy + zBz
Thus,
A B = Ax Bx + Ay By + Az Bz
AB =x
(Ay Bz Az By ) + y(Az Bx Ax Bz ) + z(Ax By Ay Bx )

x
y
z

= Ax Ay Az
Bx By Bz
C (A B) = Cx (Ay Bz Az By ) + Cy (Az Bx Ax Bz )
+ Cz (Ax By Ay Bx )

Cx Cy Cz

= Ax Ay Az
Bx By Bz
It is useful to prove the vector identities
C (A B) = A (B C) = B (C A)
(1.1.32)
C (A B) = A(C B) (C A)B
(1.1.33)
Both identities will be used frequently later on.
Gradient of a Scalar
In the Cartesian coordinate system, the del operator is a vector
dierential operator expressed as
=x
+ y
+ z
x
y
z
When operating on a scalar function , the result is a vector

= x
called the gradient of .
+ y + z
x
y
z
(1.1.34)
19
Example 1.1.2
Consider the function = x + y . The gradient of the function is
= x
+ y
For 2 = x2 + y2 > 1 = x1 + y1 , we see that is pointing in the direction of
increasing .
Example 1.1.3
The function = x2 + 2y 2 describes an ellipse. Its gradient is
= x
x + y2y
For the ellipse with equal to a constant, d = dr = 0 , where dr is tangent
to the ellipse. Thus the gradient is normal to the ellipse and pointing in the
directions of an expanding ellipse.
Example 1.1.4 Electric eld vector as gradient of a potential function.

When there is no time variation, we may write the electric eld vector E as
E =
(E1.1.4.1)
and call a potential function. The picture is that E points from high potential
towards low potential, similar to water owing from a high altitude to lower ground.
Divergence of a Vector
The divergence of a vector function is a scalar, dened as
D = x
+ y
+ z
(
xDx + yDy + zDz )
x
y
z
=
Dx +
Dy +
Dz
x
y
z
(1.1.35)
20
1. Fundamentals
z
(x0 , y0 , z0 )
z
x
y
y
Figure 1.1.11 Dierential volume xyz .
Consider a dierential volume of sides x, y, z centered around a

point (x0 , y0 , z0 ) [Fig. 1.1.11]. The divergence as dened states that
D = lim

x0
y0
z0
xyz

x
x
yz Dx (x0 +
, y0 , z0 ) Dx (x0
, y0 , z0 )
2
2

y
y
+ zx Dy (x0 , y0 +
, z0 ) Dy (x0 , y0
, z0 )
z
z

z
z
+ xy Dz (x0 , y0 , z0 +
) Dz (x0 , y0 , z0
)
(1.1.36)
2
2
The rst term in the braces is equal to the eld component Dx at the
surface at x = x0 + x
2 multiplied by the surface area yz . We dene a
surface normal vector dS pointing outward of the volume such that at the
yz and at the surface at x = x0 x
surface at x = x0 + x
2 , dS = x
2 ,
dS =
xyz . Then the negative sign in the second term is due to D
dot multiplied by dS . All six terms account for the six dierential areas
bounding the dierential volume V = xyz with a surface normal
dS . We thus express the divergence of D as
dS D
D = lim
(1.1.37)
V 0
V
21
Derivation of Boundary Conditions for D and B

When there is a plane boundary at z = z0 , and the D eld values
are nite above and below the plane boundary, we can derive the boundary
condition for D by using (1.1.36) with a small pill-box [Fig. 1.1.12] and
letting z go the zero rst. We nd

1
z
z
Dz (x0 , y0 , z0 +
D = lim
) Dz (x0 , y0 , z0
)
z0 z
2
2
(1.1.38)
Making use of (1.1.3), we nd

z
z
) Dz (x0 , y0 , z0
) = lim z
lim Dz (x0 , y0 , z0 +
z0
z0
2
2
(1.1.39)
z
Since Dz (x0 , y0 , z0 + z
2 ) is in region 1 and Dz (x0 , y0 , z0 2 ) in region 2,
z
z
we write Dz (x0 , y0 , z0 + 2 ) = D1z and Dz (x0 , y0 , z0 2 ) = D2z . The right
hand side of (1.1.39) becomes zero when the volume charge density with
unit coulomb/m3 is nite. However, if we assume is innite contained in a
zero thickness, we may dene a surface charge density s = lim z which
z0
is nite and has dimension coulombs/m2 . The concept of surface charge

density will have very practical usefulness. Equation (1.1.39) becomes
Region 1
dS = sdS
1
z
2
Region 2
Figure 1.1.12 Small pill-box volume.
D1z D2z = s
(1.1.40)
Letting the surface element dS = sxy = sdS we can write (1.1.40) as

s D1 D2 = s
(1.1.41)
22
1. Fundamentals
Thus the dierence between the D eld components normal to the boundary
surface is equal to the surface charge density at the boundary surface.
When there is no surface charge density at the boundary surface, we
have
(1.1.42)
D1z D2z = 0
or

s D1 D2 = 0
(1.1.43)
Thus the normal D component is continuous across the boundary. By the

same token, we conclude from (1.1.4) that
or
B1z B2z = 0
(1.1.44)

s B 1 B 2 = 0
(1.1.45)
Thus the B eld component normal to the boundary surface is continuous

since there is no surface charge density at the boundary surface.
Divergence Theorem
Applying the above result to a large volume V containing an innite
number of such innitesimal dierential volumes [Fig. 1.1.13], we note that
when integrating the divergence over the volume surfaces shared by adjacent
dierential volumes will have no contribution because the surface normals
point in opposite directions and thus cancel. The result is the divergence or
Gauss theorem
dV D = dS D
(1.1.46)
V
The divergence theorem states that the volume integral of the divergence of
the vector eld D is equal to the total outward ux D through the surface
S enclosing the volume.
Example 1.1.5 Interpretation of Gauss law for electric elds.

Applying (1.1.46) to Gauss or Coulombs law for electric elds (1.1.3), we nd
dS D =
S
dV D =
V
dV = q
(E1.1.5.1)
Thus the divergence of the vector eld D , also called electric ux, out of an enclosed
surface S , is equal to the total charge q in the volume V enclosed by the surface.
23
Figure 1.1.13 Derivation of divergence theorem.
Curl of a Vector
The curl of a vector eld H is a vector, dened as

H =
+ y
+ z
x
y
z

H
(1.1.47)
Consider a dierential volume of sides x, y, z centered around a point

(x0 , y0 , z0 ) . In the Cartesian coordinate system, the curl of a vector H as
dened states that
H = lim
x0
y0
z0
xyz

x
x
yz x
H(x0 +
, y0 , z0 ) H(x0
, y0 , z0 )
2
2

y
y
+ zx y H(x0 , y0 +
, z0 ) H(x0 , y0
, z0 )
2
2

z
z
+ xy z H(x0 , y0 , z0 +
) H(x0 , y0 , z0
)
2
2
(1.1.48)
The six terms in the above equation are associated with the six dierential
surfaces bounding (x0 , y0 , z0 ) . For the rst term, the surface normal is in the
yz . For the second term dS =
xyz .
x
direction; we write dS = x
24
1. Fundamentals
For the third term dS = yzx , etc. Thus we write (1.1.48) as
H = lim
dS s H
V
V 0
(1.1.49)
Applying the above result to a large V containing an innite number of

such dierential volumes, we nd the curl theorem
dV H = dS s H
(1.1.50)
This is similar to the divergence theorem except that now the result is in
vector form.
Derivation of Boundary Conditions for E and H

When there is a plane boundary at z = z0 , and the E eld values
are nite above and below the plane boundary, we can derive the boundary
condition for E by using (1.1.48) and letting z go to zero rst [Fig. 1.1.12].
We nd

1
z
z
x Hy (x0 , y0 , z0 +
) Hy (x0 , y0 , z0
)
H = lim
z0 z
2
2

z
z
) Hx (x0 , y0 , z0
)
+ y Hx (x0 , y0 , z0 +
2
2
From (1.1.1), we nd

z
z
Dx
) + Hy (x0 , y0 , z0
) = lim z
+ Jx
lim Hy (x0 , y0 , z0 +
z0
z0
2
2
t

Dy
z
z
lim Hx (x0 , y0 , z0 +
) Hx (x0 , y0 , z0
) = lim z
+ Jy
z0
z0
2
2
t
(1.1.51)
On the right hand sides of the above two equations, the time derivatives
Dx /t and Dy /t are nite but we may assume Jx and Jy to be innite
to create a surface current density J s when z 0 :
J s = lim J z
z0
J
(1.1.52)
25
z
Notice that Hy (x0 , y0 , z0 + z
2 ) and Hx (x0 , y0 , z0 + 2 ) are in region 1, and
z
Hy (x0 , y0 , z0 z
2 ) and Hx (x0 , y0 , z0 2 ) are in region 2, we obtain from
(1.1.51)
H1y + H2y = Jsx

H1x H2x = Jsy
(1.1.53)
Since the dierential surface dS = sdS = zxy we can write (1.1.53) in

vector form as
(1.1.54)
s (H 1 H 2 ) = J s
Thus the discontinuity in the tangential components of H is equal to the
surface current at the boundary surface.
When there is no surface current density at the surface boundary, we
have
H1y = H2y
H1x = H2x
(1.1.55)
s (H 1 H 2 ) = 0
(1.1.56)
or
Thus the tangential H components are continuous across the boundary
surface. By the same token, we conclude from (1.1.2) that
E1y = E2y
E1x = E2x
(1.1.57)
s (E 1 E 2 ) = 0
(1.1.58)
or
Thus the tangential components of E are continuous across the boundary
surface.
Stokes Theorem
The curl of H is dened in the Cartesian coordinate system as the
vector

y
z
x

H = x
+ y
+ z
H =
x y z
x
y
z

Hx Hy Hz

=x
Hz
Hy + y
Hx
Hz + z
Hy
Hx
y
z
z
x
x
y
(1.1.59)
26
1. Fundamentals
Writing in dierential form, we have

H = lim
x0
y0
z0
xyz

y
y
, z0 ) Hz (x0 , y0
, z0 )
x
xz Hz (x0 , y0 +
2
2

z
z
xy Hy (x0 , y0 , z0 +
) Hy (x0 , y0 , z0
)
2
2

z
z
) Hx (x0 , y0 , z0
)
+ y xy Hx (x0 , y0 , z0 +
2
2

x
x
, y0 , z0 ) Hz (x0
, y0 , z0 )
yz Hz (x0 +
2
2

x
x
+ z yz Hy (x0 +
, y0 , z0 ) Hy (x0
, y0 , z0 )
2
2

y
y
xz Hx (x0 , y0 +
, z0 ) Hx (x0 , y0
, z0 )
2
2
(1.1.60)
The 12 terms in the above equation are associated with the 6 dierential
surfaces bounding (x0 , y0 , z0 ) .
Applying to open surfaces, the curl theorem becomes the well-known
Stokes theorem in vector calculus. We have, for the z component,
( H)z = lim
x0
y0
xy

x
x
, y0 , z0 ) Hy (x0
, y0 , z0 )
y Hy (x0 +
2
2

y
y
x Hx (x0 , y0 +
, z0 ) Hx (x0 , y0
, z0 )
2
2
The rst term in the bracket is equal to the component Hy at x = x0 + x

2
multiplied by the dierential length y . We dene a vector dierential
dy ;
length dl [Fig. 1.1.14] such that for the side y at x = x0 + x
2 , dl = y
x
,
dl
=
x
dx
;
for
the
side
y
at
x
=
x
for the side x at y0 + y

0
2
2 ,
y
y dy ; and for the side x at y = y0 2 , dl = x
dx . If we use
dl =
27
z
dl =
xdx
y
dl = ydy
(x0 , y0 )
C
Figure 1.1.14 Derivation of z-component of the curl of a vector eld.
the ngers of the right hand to trace the direction of dl along the loop, the
right-hand thumb points in the surface normal direction z . Thus

1
dl H
z ( H) = lim
x0 S C
y0
where C denotes the contour circulating the area S = xy . Similar

results are derivable for the x
and y components of H . Consequently
for a dierential area Sj bounded by a contour Cj and with a surface
n Sj and
normal a
n , we have S j = a

dl H
S j ( H)j =
Cj
For an open surface S , we can subdivide it into N such as dierential areas

[Fig. 1.1.15]. Adding the contributions of all N dierential areas, we have
lim
N

Sj 0
N j=1

S j ( H)j =
dl H
C
Since the common part of the contours in two adjacent elements is traversed
in opposite direction by the two contours, the net contribution of all the
common parts in the interior sums to zero and only the contribution from
the external contour C bounding the open surface S remains in the line
integral on the right-hand side. The left-hand side becomes a surface integral,
and the result is Stokes theorem:
28
1. Fundamentals
S
Figure 1.1.15 Derivation of Stokes theorem.

dS ( H) =
dl H
(1.1.61)
Stokes theorem states that the surface integral of the curl of the vector eld
H over an open surface S is equal to the closed line integral of the vector
along the contour enclosing the open surface.
Example 1.1.6 Electromotive force (EMF) and Lenz law.

Applying Stokes theorem (1.1.61) to Faradays law (1.1.2), we have
dl E =
t
C
dS B
(E1.1.6.1)
The line integral of E is dened as the electromotive force (EMF):

dl E =
EMF =
C
(E1.1.6.2)
where
dS B
(E1.1.6.3)
is the magnetic ux linking a loop with area A bounded by the closed contour C .
Equation (E1.1.6.2) states that the EMF is equal to the negative time derivative of
the magnetic ux linking the loop. Thus the EMF always produces a ux in the loop
29
to oppose the direction of change of the ux linking the loop; if is increasing,

the EMF decreases the ux, and vice versa. This is known as the Lenz law.
The integration of E along a line segment of l from point a to point b is
dened as the voltage drop between a and b :
dl E
Vab =
(E1.1.6.4)
Note that the EMF has unit of voltage (Volt) and not unit of force. The voltage
Vab between points a and b is the dierence of potential at the two points. For
positive Vab , the electric eld vector points from a to b . Thus point a is at a higher
potential a and point b is at a lower potential b < a and Vab = a b .
Consider a closed loop C . According to (E1.1.6.1), the sum of the voltage
around the closed loop is equal to the magnetic ux linking the area formed by the
loop. Kirchhos voltage law (KVL) states that the voltage around a closed loop is
equal to zero. Thus KVL is correct only when E = 0 . It is important to note
that if there is magnetic eld linking the loop, then KVL is incorrect.
Example 1.1.7
Electromotive force (EMF).
Consider the loop in Fig. E1.1.7.1 consisting of two resistors with resistances
R1 = 2.5 ohm and R2 = 7.5 ohm . Let the magnetic ux linking the loop be increasing at the rate of 10 Wb/s. According to (E1.1.6.1), an EMF of 10 V is
induced to counter the increase. The direction of the induced current is as shown so
as to produce a magnetic eld in the opposite direction of the increasing magnetic
eld. The voltage across R1 is V1 = 2.5 V , which can be obtained by taking the
closed loop consisting of the voltmeter and R1 yielding 0 = 2.5 V1 , or by taking
the loop consisting of the voltmeter and R2 which includes the time varying magnetic eld and yielding 10 = 7.5 + V1 . Likewise, the voltage readings for the other
two voltmeters are V2 = 7.5 V and V3 = 2.5 V . It is noted that although the
voltmeters for V2 and V3 are connected to the same two nodes, the two readings
are drastically dierent, a clear violation of Kirchhos voltage law (KVL). This
demonstrates that KVL is correct only when there is no magnetic ux linking the
loop where it is applied, namely E = 0 .
Consider another case as shown in Fig. E1.1.7.2 where the induced counter EMF
is 20 V. Following the same analysis, we nd V1 = 5 V , V2 = 15 V , V3 = 5 V ,
V4 = 10 V , and V5 = 20 V .
30
1. Fundamentals
V2
R2
7.5V
R1
2.5V
V3
V1
Figure E1.1.7.1 EMF of the loop is 10 volts.
V2
5V
15 V
+
V5
R1
R2
V4
V1
V3
Figure E1.1.7.2 EMF of the loop is 20 volts.

Example 1.1.8 Derive boundary conditions for E and H with Stokes theorem.
Consider a ribbon-like surface as shown in Fig. E1.1.8.1. Integrating over the
surface of the ribbon area, Faradays law and Ampères law become
d
dl E =
dS s B
dt

d
dl H =
dS s D +
dS s J
dt
Let the ribbon area approach zero in such a manner that goes to zero rst and
the terms involving are discarded. To relate E 1 , H 1 in region 1 to E 2 , H 2 in
region 2, we proceed as follows.
The integral forms of Faradays law and Ampères law as applied to the ribbon
31
n
dl
dl
dS = s dS
region 1
region 2
Figure E1.1.8.1
area in Fig. E1.1.8.1 yield, as 0,
dS s B = 0
dt
d
dS s D = 0
dt
d
because dt
s B and
Therefore
dt s
D remain nite while the ribbon area approaches zero.

dl (E 1 E 2 ) = 0
dl (H 1 H 2 ) = s J dl
The electric eld E in the dl direction is tangential to the surface and can be
written as dl E = dl
sn
E = dl
sn
E for all dl
s along the interface and
similarly for H . We thus have
n
(E 1 E 2 ) = 0
n
(H 1 H 2 ) = lim J J s
0
Useful Vector Identities

Several vector identities that are used later on are listed below:
( E) = ( E) 2 E
(1.1.62)
32
1. Fundamentals
(E H) = H ( E) E ( H)
( A) = 0
() = 0
(1.1.63)
(1.1.64)
(1.1.65)
The proofs can be carried out in the Cartesian coordinate system or in

general vector notation.
Exercise 1.1.2 To prove (1.1.62), we use the vector identity C (A B) =
A(C B) (C A)B and identify C and A with and B with E . We can also
use the denition for curl to show that

x
y

x
y
( E) =
Ez Ey Ex Ez

y
z
z
x
2
2
2

Ey
Ex

x
y

2
z
Ey
Ez
Ex
Ex
+
xy xz
y 2
z 2
=x

y

+ z

=x
2 Ey
2 Ez
2 Ez
2 Ex
+
2
xz
zy
x
y 2
( E) 2 Ex
x

+ z
2 Ey
2 Ey
2 Ex
2 Ez
+
2
2
x
z
xy
yz
( E) 2 Ez
z
+ y

( E) 2 Ey
y
= ( E) 2 E
where
2 = =
2
2
2
+
+
x2
y 2
z 2
is called the Laplacian operator, and in the x

component we added and subtracted
2 Ex
the term x2 and similarly for the y and z components.
To prove (1.1.63), we write
(E H)
=
(Ey Hz Ez Hy ) +
(Ez Hx Ex Hz ) +
(Ex Hy Ey Hx )
x
y
z
33
H z + H z Ey Ez Hy H y Ez
x
x
x
x
+ Ez H x + Hx Ez Ex Hz Hz Ex
y
y
y
y
+ Ex Hy + H y Ex Ey Hx H x Ey
z
z
z
z

= Hx
Ez
Ey + Hy
Ex
Ez + Hz
Ey
Ex
y
z
z
x
x
y
= Ey
Ex
Hz
Hy
y
z
Ey
Hx
Hz
z
x
Ez
Hy
Hx
x
y
= H ( E) E ( H)
To prove (1.1.64), we write
( A) =
x
+
z
Az
Ay
y
z
+
y
Ay
Ax
x
y
Ax
Az
z
x
=0
To prove (1.1.65), we nd

x

() = x

x
z

z = 0

z
Example 1.1.9 Poisson equation and Laplace equation.

In (E1.1.4.1), we wrote the electric eld vector as the gradient of a potential
function :
E =
(E1.1.9.1)
By virtue of (1.1.65), we see that E = 0 . Thus the above denition for the
electric eld is true only when the term B/t in Faradays law can be neglected,
i.e., when there is no time variation. We may refer to the above electric eld as the
static electric eld.
The Coulomb law (or Gauss law for electricity) in free space is E = / o .
In terms of the potential function, we have
2 = / o
(E1.1.9.2)
34
1. Fundamentals
This is known as the Poisson equation. In places where there is no charge density,
= 0 , we have
2 = 0
(E1.1.9.3)
This is known as the Laplace equation.

Show that the potential function
Exercise 1.1.3
=
x2
+ y2 + z2
where C is a constant, satises the Laplace equation (E1.1.9.3).
Example 1.1.10 Static electric eld vector

as gradient of a potential function.
With the potential function = C/ x2 + y 2 + z 2 , the associated static electric eld vector is
x
y
z
E = = C x
2
+ y 2
+ z 2
(x + y 2 + z 2 )3/2
(x + y 2 + z 2 )3/2
(x + y 2 + z 2 )3/2
In terms of
the position vector r = x
x + yy + zz , and the length of the position
vector r = x2 + y 2 + z 2 , the electric eld vector takes the form
E=r
C
C
= r 2
r3
r
where r is pointing in the direction of r with unit length. Thus the electric eld
vector is pointing away from the origin along the direction of the position vector
r , and its magnitude decreases as the squared inverse distance r .
Index Notation
A vector in the Cartesian coordinate system can be represented by its
three components. Thus Aj with j = 1, 2, 3 represents A1 , A2 , A3 of the
vector A . The dot product A B is written as Aj Bj where the repeated
index j implies summation over j from 1 to 3:
Aj Bj =
3

j=1
Aj Bj = A1 B1 + A2 B2 + A3 B3
35
To express cross products in index notation we need to dene a Levi-Cevita

symbol ijk where i, j, k take values from 1 to 3. When any of the two
indices are equal the Levi-Cevita symbol is zero. Otherwise, ijk is either
+1 or 1 . It is +1 if ijk is an even permutation of 1,2,3; 1 if ijk is an
odd permutation of 1,2,3. Thus 123 = 231 = 312 = 1 and 213 = 132 =
321 = 1 and all others equal to 0. Let C = A B . In index notation, we
write Ci = ijk Aj Bk . Thus, C1 = 123 A2 B3 + 132 A3 B2 = A2 B3 A3 B2 .
The dyad A B is then Aj Bk , no summation implied because no index is
repeated. The identities (1.1.32) and (1.1.33) are
Ci ijk Aj Bk = Aj jki Bk Ci = Bk kij Ci Aj
and
ijk Cj klm Al Bm = (ijk klm )Cj Al Bm = (il jm im lj )Cj Al Bm
= Ai Cm Bm Cl Al Bi
where ij = 1 when i = j and ij = 0 when i = j .
In index notation, divergence of Dj , Dj , is j Dj .
In index notation, is represented by i and by i .
In index notation, curl of Hi , Hi , is written as ijk j Hk .
The identities (1.1.62)(1.1.65) are, in index notation
ijk j klm l Em = (il jm im jl )j l Em
= m i Em j j Ei
i (ijk Ej Hk ) = ijk Hk i Ej + ijk Ej i Hk
= Hk kij i Ej Ej jik i Hk
i ijk j Ak = jik i j Ak = 0
ijk j k = ikj j k = ikj k j = 0
Cylindrical and Spherical Coordinate Systems

Although we have emphasized the use of the rectangular Cartesian coordinate system, expressions written in vector notation are not coordinate
dependent. In addition to the coordinates with unit vectors x
, y, z , the cylin z , and the spherical coordidrical coordinate system with unit vectors
, ,
are often used in this book.
nate system with unit vectors
, ,
36
1. Fundamentals
z
dd
dz
ddz
ddz
y
x
Figure 1.1.16 Cylindrical coordinate system.
In the cylindrical coordinate system [Fig. 1.1.16], the vector dierential

length is
dl =
d + d
+ zdz
the dierential area is
ddz + ddz
+ zdd
dS =
and the dierential volume is
dV = dddz
In the spherical coordinate system [Fig. 1.1.17], the vector dierential
length is
sin d
dl = rdr + rd + r
the dierential area is
dS = rr2 sin dd + r sin drd + rdrd

the dierential volume is
dV = r2 sin drdd
37
dr
d
y
d
r sin
Figure 1.1.17 Spherical coordinate system.
In a general orthogonal coordinate system, we use u

i (i = 1, 2, 3) to
denote the three basis vectors, dli = hi dui to denote a dierential length,
where hi is called a metric coecient. The basis vectors are perpendicular
j = 0 for i = j but they are not necessarily of unit
to one another u
i u
length or even with the dimension of length. In Table 1.1.1 we summarize the
basis vectors and the metric coecients for the rectangular (or Cartesian),
cylindrical, and spherical coordinate systems.
Orthogonal
Coordinate System
Base Vectors
2, u
3)
(
u1 , u
Metric Coecients
(h1 , h2 , h3 )
Dierential Volume
(h1 h2 h3 du1 du2 du3 )
Rectangular
Coordinates
(x, y, z)
Cylindrical
Coordinates
(, , z)
Spherical
Coordinates
(r, , )
x
, y, z
, ,
r, ,
1, 1, 1
1, , 1
1, r, r sin
dxdydz
dddz
r2 sin drdd
Table 1.1.1 Orthogonal coordinate systems.
In terms of the general orthogonal coordinate system, the gradient, the
38
1. Fundamentals
divergence, and the curl are dened as
+u
2
+u
3
h1 u1
h2 u2
h3 u3

D =
(h2 h3 D1 ) +
(h3 h1 D2 ) +
(h1 h2 D3 )
h1 h2 h3 u1
u2
u3

h1 u
1 h2 u
2 h3 u
3

1

H =

u2
u3
h1 h2 h3 u1
h1 H 1 h2 H 2 h3 H 3
= u
1
The Laplacian operator is

1
h2 h3
= =
h1 h2 h3 u1
h1 u1

+
h3 h1
+
h1 h2
u2
h2 u2 u3
h3 u3
2
Identifying the metrics h1 , h2 , h3 with those as listed in Table 1.1.1, we

readily obtain the expressions in cylindrical and spherical coordinates.
Example 1.1.11 Static electric eld due to a charged particle.
In the spherical coordinate system, the static electric eld associated with the
potential function = C/r is
E = =
r
C
= r 2
r
r
The corresponding electric displacement eld vector is D = o E . Applying the

Coulomb law D = to the divergence theorem (1.1.46)
dV D = dS D
we nd
(E1.1.11.1)
dV = dS D
(E1.1.11.2)
Assuming the electric eld is due to a charged particle q situated at the origin,
we can integrate over a small spherical volume with radius r = surrounding
the origin. The left hand side, integrating over the charge density over the volume
containing the charged particle, gives rise to the total charge q . In the spherical
39
coordinate system, the surface integration element at a small distance r = is

rr2 sin dd . We thus have
q = dS D =
S
dd 2 sin
0
o C
= 4 o C
2
Thus we have determined the constant C = q/4 o and obtained the static electric
eld
E = r
q
4 o r2
(E1.1.11.3)
due to a charged particle situated at the origin.

Problems
P1.1.1
Maxwells equations were originally written in the form of scalar partial dierential equations. For all eld components, write (1.1.1), (1.1.2), (1.1.3), (1.1.4), and
(1.1.5) in terms of partial derivatives. Derive the continuity equation from (1.1.1)
to (1.1.4). Prove that given the continuity law, Coulombs law can be derived from
Ampères law. Likewise, show that Gauss law can be derived from Faradays law,
and that (1.1.3) and (1.1.4) are not independent scalar equations.
P1.1.2
An electromagnetic wave has spatial frequency ko = 100 Ko . Determine the

wavelength in meters and the temporal frequency in GHz.
Determine the spatial frequency in units of Ko for a laser light at wavelength
= 0.6328m .
Determine the spatial frequency in units of Ko for a microwave oven at frequency 2.4 GHz.
P1.1.3
Electromagnetic waves satisfy all of the Maxwell equations. Consider, in free

space, the following electric eld vectors:
E1 = x
cos(t kz)
E 2 = z cos(t kz)

E 3 = (
x + z) cos t + k|x z|/ 2

E 4 = (
x + z) cos t + k|x + z|/ 2
E 5 = (
x + z) cos(t + ky)
40
1. Fundamentals
Do these electric eld vectors satisfy the wave equation and all Maxwell equations?
Which of the ve elds qualify as electromagnetic waves? For those not qualied
as electromagnetic waves, you should state which of the Maxwell equations are
violated.
P1.1.4
The known spectrum of electromagnetic waves covers a wide range of frequencies. Electromagnetic phenomena are all described by Maxwells equations and, by
convention, are generally classied according to wavelengths or frequencies. Radio
waves, television signals, radar beams, visible light, X rays, and gamma rays are
examples of electromagnetic waves.
(a) Give in meters the wavelengths corresponding to the following frequencies:
(i) 60 Hz
(ii) AM radio (5351605 kHz)
(iii) FM radio (88108 MHz)
(iv) L- band (12 GHz)
(v) Visible light ( 1014 Hz)
(vi) X-rays ( 1018 Hz)
(b) Give in Hertz the temporal frequencies corresponding to the following wavelengths:
(i) 1 km, (ii) 1 m, (iii) 1 mm, (iv) 1 m , (v) 1
A.
(c) Give in unit Ko the spatial frequencies corresponding to the wavelengths in
(b).
P1.1.5
Prove the following

identities:

E = E 2 E

EH =H E E H

A =0
() = 0
P1.1.6
Three vectors A, B , and C drawn in a head-to-tail fashion form the three

sides of a triangle. What is A + B + C and what is A + B C ?
P1.1.7
A position vector r = x
2+ y + z2 . Determine its spherical components r, ,
and its cylindrical components , , z .
P1.1.8
Find a unit vector c that is perpendicular to both A = x

4 + y5 z3 and
B=x
2 y7 z1.5 .
P1.1.9
For the vector A = 2 + z2z , verify the divergence theorem for the circular
cylindrical region enclosed by = 5, z = 0 , and z = 3 .
41
P1.1.10
Derive the boundary conditions for E by applying the curl theorem to a small
pill-box volume on the x-y plane which has an area A and an innitesimal thickness
z .
P1.1.11
What is the result if the surface integral of H is carried out over a closed
surface? Compare with the curl theorem we obtained for the curl integrated over a
volume V enclosed by a surface S in (1.1.50).
42
1.2
1. Fundamentals
Polarization
The polarization of a wave is conventionally dened by the time variation of

the tip of the electric eld vector E at a xed point in space. If the tip moves
along a straight line, the wave is linearly polarized. When the locus of the
tip is a circle, the wave is circularly polarized. For an elliptically polarized
wave, the tip of E describes an ellipse. If the right-hand thumb points in
the direction of propagation while the ngers point in the direction of the tip
motion, the wave is dened as right-hand polarized. The wave is left-hand
polarized when it is described by the left-hand thumb and ngers.
Consider the following wave solution:
E(z, t) = x
Ex + yEy
=x
cos(kz t) + yA cos(kz t + )
(1.2.1)
The wave propagates in the +

z direction. From the temporal view point,
E(t) = x
cos(t) + yA cos(t )
We now study polarization for the following special cases:
Case 1) = 2m , where m = 0, 1, 2, ... is an integer. We have
E(t) = x
cos(t) + yA cos(t)
The tip of the electric eld vector moves along a line as shown in Fig. 1.2.1a.
The wave is linearly polarized.
Case 2) = (2m + 1) , we have
E(t) = x
cos(t) yA cos(t)
The tip of the electric eld vector moves along a line as shown in Fig. 1.2.1b.
The wave is linearly polarized.
Case 3) = /2 and A = 1 , we have
cos(t) + y sin(t)
E(t) = h
(1.2.2)
It can be seen that while the x component is at its maximum the y component is zero. As time progresses, the y component increases and the x
component decreases. The tip of E rotates from the positive Ex axis to the
positive Ey axis [Fig. 1.2.1c]. Elimination of t from the x and y components in (1.2.2) yields a circle of radius 1 , Ex2 + Ey2 = 1 . Thus the wave is
right-hand circularly polarized.
1.2 Polarization
43
Ey
Ey
A
Ex
Ex
a. Linear polarization
b. Linear polarization
Ey
Ey
Ex
c. Right-hand
circular polarization
Ex
d. Left-hand
circular polarization
Ey
Ey
A
A
Ex
e. Right-hand
elliptical polarization
Ex
f. Left-hand
elliptical polarization
Figure 1.2.1 Polarizations.
44
1. Fundamentals
Case 4) = /2 and A = 1 , we have

E(t) = x
cos(t) y sin(t)
(1.2.3)
As time progresses, the y component increases and the x component decreases. The tip of E rotates from the positive Ex axis to the negative Ey
axis. Thus the wave is left-hand circularly polarized [Fig. 1.2.1d].
Case 5) = /2 , we have
E(t) = x
cos(t) yA sin(t)
(1.2.4)
The wave is right-hand elliptically polarized for = /2 [Fig. 1.2.1e] and

left-hand elliptically polarized for = /2 [Fig. 1.2.1f].
1
2
Figure 1.2.2 Polarizations for various values of and A.
The above discussion can be summarized in Fig. 1.2.2 where we illustrate

the polarization for dierent values of A and . On the horizontal axis,
= 0, or , the wave is linearly polarized. If A = 1 and = /2, the
wave is right-hand circularly polarized. For A = 1 and = /2, the wave
is left-hand circularly polarized. Otherwise the wave is elliptically polarized.
The polarization is right-handed if the phase dierence is between zero and
, and left-handed if is between and 2.
1.2 Polarization
45
Example 1.2.1 Polarization from the spatial view point.

Wave polarization can be viewed by either taking a series of still pictures at
several xed times, called the spatial view point or by making observations at a xed
point in space, called the temporal view point. The denition of polarization so far
has been discussed from the temporal view point. Let us now look at polarization
from the spatial view point.
E|t=t+
z = zo
E|t=to
y
x
Figure E1.2.1.1 Spatial view of polarization.
Consider the right-hand circularly polarized wave with = /2 and A = 1

in case 3), setting t = 0 in wave solution (1.2.1), we have
E(z = 0, t) = x
cos(kz) y sin(kz) = x
Ex (z) yEy (z)
This is a left-handed helix as shown below.
Ex = E0 cos
2
Ey = E0 sin
The parametric equation of a helix is
x = R cos
2
z
p

y = R sin
2
2
z
p

r
where p is the pitch of the helix. Thus the locus of the tip point of the electric eld
vector measured along the z axis is a left-handed helix with the pitch p = . The
helix advances along +
z without rotating. At z = z0 = 3/4 , electric eld vector
is at E|t=to when to = 0 , it is shown as E|t=t+ when t+ = /4 .
46
1. Fundamentals
Topic 1.2A Stokes Parameters and Poincar

e Sphere
To facilitate a mathematical discussion of polarization, we decompose the
E vector of a wave into two components perpendicular to the direction of
propagation. For a specic point in space, we write
h + vEv = he
h cos(t h ) + vev cos(t v )
E(t) = hE
(1.2A.1)
, v , and the direction of propagation are mutually perpendicular

where h
and thus form an orthogonal system. We assume the amplitudes ev and eh
are both positive. The locus of the tip E(t) is determined by eliminating
the time t dependence between the two components Eh and Ev .
When h and v dier by an integral multiple of 2 , v h = 2n ,
the two components are in phase. We have Eh /eh = Ev /ev . The wave is linearly polarized, and the straight-line locus traverses the rst and third quadrants [Fig. 1.2.1a]. When h and v dier by an odd integral multiple of ,
the two components are 180 out of phase. We have Eh /eh = Ev /ev , and
the straight-line locus traverses the second and fourth quadrants [Fig. 1.2.1b].
When the magnitudes of the two components are equal, eh = ev = e0 , and
v h = /2 , the wave is right-hand circularly polarized [Fig. 1.2.1c].
When eh = ev = e0 , and v h = /2 , the wave is left-hand circularly
polarized [Fig. 1.2.1d].
Ev
e2
e1
Eh
Figure 1.2A.1 Elliptical polarization.
In general, a polarized wave has elliptical polarization; that is, when

time is eliminated from the two components of E , the resultant equation
describes an ellipse. Consider the case h = 0 , v h = /2 in (1.2A.1)
and let eh = e1 > e2 = ev . We have
1 cos(t 0 ) ve2 sin(t 0 )
E(t) = he
(1.2A.2)
1.2 Polarization
47
Elimination of time yields an ellipse

Eh
e1
2

+
Ev
e2
2
=1
We see that e1 is the major axis of the ellipse and e2 the minor axis. With
the plus sign in (1.2A.2), we have a right-hand elliptically polarized wave
and with the minus sign a left-hand elliptically polarized wave. The shape of
the ellipse can be specied by an ellipticity angle . Referring to (1.2A.2),
with e1 denoting the major axis and e2 the minor axis, we write
tan =
e2
e1
(1.2A.3)
where the plus sign corresponds to right-hand polarization for which 0

/4 and the negative sign to left-hand polarization for which /4 0.
Ev
e2
Eh
e1
Figure 1.2A.2 Elliptical polarization.
The general polarization states are more popularly described with the
Poincare sphere as discussed below. Consider the elliptical polarization as
given by (1.2A.1), which describes a tilted ellipse as plotted in Fig. 1.2A.2.
The major axis of the ellipse described in (1.2A.2) is rotated and makes the
angle with the Eh axis with 0 . We call the orientation angle.
48
1. Fundamentals
Example 1.2A.1 Coordinate transformation.

Consider two rectangular coordinate systems represented by unit vectors (
x, y)
and (
x , y ) . In terms of the unprimed coordinates, the unit vectors for the primed
coordinate system are represented as
x
= x
cos + y sin
y =
x sin + y cos
A position vector r can be represented as r = x
x + y y = x
x + yy. The transformation between coordinates (x , y ) and (x, y) is obtained as follows:
x = x
r = (
x x
)x + (
x y)y = x cos + y sin
y = y r = (
y x
)x + (
y y)y = x sin + y cos
End of Example 1.2A.1
In view of (1.2A.2) and Fig. 1.2A.2, we have from coordinate transformation

e1 cos(t 0 ) = Eh cos + Ev sin
e2 sin(t 0 ) = Eh sin + Ev cos
(1.2A.4a)
(1.2A.4b)
Substituting the components Eh and Ev of (1.2A.1) in (1.2A.4) and comparing the coecients of cos t and sin t , we obtain
e1 cos 0
e1 sin 0
e2 cos 0
e2 sin 0
= eh cos h cos + ev cos v sin

= eh sin h cos + ev sin v sin
= eh sin h sin + ev sin v cos
= eh cos h sin ev cos v cos
(1.2A.5a)
(1.2A.5b)
(1.2A.5c)
(1.2A.5d)
Eliminating 0 from (1.2A.5a) and (1.2A.5b) by squaring and adding, we

nd
(1.2A.6a)
e21 = e2h cos2 + e2v sin2 + eh ev sin 2 cos
Similarly from (1.2A.5c) and (1.2A.5d) , we have
e22 = e2h sin2 + e2v cos2 eh ev sin 2 cos
(1.2A.6b)
Multiplying (1.2A.5a) by (1.2A.5c) , (1.2A.5b) by (1.2A.5d) and adding,

we again eliminate 0 and obtain
e1 e2 = eh ev sin
(1.2A.6c)
1.2 Polarization
49
Finally we multiply (1.2A.5a) by (1.2A.5d) and subtract from (1.2A.5b)

multiplied by (1.2A.5c) , which yields
2eh ev cos = (e2h e2v ) tan 2
(1.2A.6d)
Equation (1.2A.6) will be used in the following discussion on Stokes parameters and the Poincare sphere.
To facilitate the discussion of polarization states of electromagnetic
waves, the four Stokes parameters pertaining to E(t) given in (1.2A.1) are
dened as follows :

1 2
eh + e2v

1 2
eh e2v
Q=
2
U = eh ev cos
2
V = eh ev sin
I=
(1.2A.7a)
(1.2A.7b)
(1.2A.7c)
(1.2A.7d)
Notice that I 2 = Q2 + U 2 + V 2 .
Adding (1.2A.6a) and (1.2A.6b) yields e21 +e22 = e2h +e2v = I . Making
use of (1.2A.3), we have
(1.2A.8)
e21 = I cos2
Subtracting (1.2A.6b) from (1.2A.6a) and making use of (1.2A.6d) , we
nd e21 e22 = (e2h e2v )/ cos 2 . Making use of (1.2A.3) and (1.2A.8) , we
nd
1
Q = (e2h e2v ) = I cos 2 cos 2
(1.2A.9a)
In terms of I , we nd from (1.2A.7c) , (1.2A.6d) and (1.2A.9a)

U = I cos 2 sin 2
(1.2A.9b)
and from (1.2A.7d) , (1.2A.6c) and (1.2A.8)

V = I sin 2
(1.2A.9c)
Again we see from (1.2A.9) that I 2 = Q2 + U 2 + V 2 .

Equation (1.2A.9) suggests a simple geometrical representation of all
states of polarization by recognizing that Q, U, and V can be regarded as
50
1. Fundamentals
Figure 1.2A.3 Poincar

e sphere.
the rectangular components of a point on a sphere with radius I [Fig. 1.2A.3].

We dene, in the spherical coordinate system, = /2 2 and = 2.
As seen from (1.2A.3), positive is for right-hand polarization which is
represented by points on the upper hemisphere. On the lower hemisphere, the
points correspond to left-hand polarization. The north pole represents righthand circular polarization and the south pole represents left-hand circular
polarization. The sphere is called the Poincare sphere. Fig. 1.2.2 is seen to
be a planar projection of the Poincare sphere with the plane and the sphere
touching each other at Q = I. The equator is mapped into the horizontal
axis.
For polarized waves,
I = Ih + Iv
Q = Ih Iv = I cos 2 cos 2
U = I cos 2 sin 2
V = I sin 2
When the wave is right-hand circularly polarized Q = U = 0 , V = I , as
= /4 . When the wave is left-hand circularly polarized, Q = U = 0 ,
V = I , as = /4 . When the wave is linearly polarized, V = 0 , as
= 0.
1.2 Polarization
51
Partial Polarization
Radiation from many natural and man-made sources consists of eld
components that uctuate with time. We write

Eh = eh (t) cos t h (t)

Ev = ev (t) cos t v (t)
The wave is quasi-monochromatic when eh (t), ev (t), h (t), and v (t) are
slowly varying compared with cos t. The Stokes parameters are dened
by a time-average procedure over a large time interval T , denoted with the
brackets <> :
1 T
2
dt [Eh (t)]2
< Eh (t) >=
T 0
The Stokes parameters are

1
< Eh2 > + < Ev2 >

1
Q = Ih Iv =
< Eh2 > < Ev2 > = I < cos 2 cos 2 >
2
U = < Eh Ev cos > = I < cos 2 sin 2 >
2
V = < Eh Ev sin > = I < sin 2 >
I = Ih + Iv =
For completely unpolarized waves, Eh and Ev are uncorrelated and we have

I = total Poynting power and Q = U = V = 0. For completely polarized
waves we have I 2 = Q2 + U 2 + V 2 . For partially polarized waves it can be
shown that I 2 Q2 + U 2 + V 2 [Example 1.2A.2]. With the Poincare sphere
of radius I , the partially polarized waves correspond to points inside the
sphere.
In concluding this section on wave polarization, we remark that the
polarization is dened according to the time variations of the E vector. As
we shall see in Chapter 3, it is imperative that we dene polarization in
terms of D when anisotropic and bianisotropic media are involved. This is
because in isotropic media E is perpendicular to k , k E = 0, while in
non-isotropic media k D = 0. This also suggests that wave polarization can
be dened in terms of the eld vector B .
52
1. Fundamentals
Example 1.2A.2
(a) Assume ev (t), eh (t) and (t) remain constant for a fractional time interval tn
and let T be subdivided into t1 , t2 , . . . tN . The time-averages can be written
in a summation form such that
Ih =
1
tn e2hn
T
n
1
Iv =
tn e2vn
T
n
2
U=
tn e2vn An cos n
T
n
2
V =
tn e2vn An sin n
T
n
where
An =
ehn
evn
denotes the ratio of eh and ev in the time interval tn . Assume that both eh
and ev are positive.
Show that
4Iv Ih U 2 + V 2
or
I 2 = (Ih + Iv )2 Q2 + U 2 + V 2
The equal sign holds when n = m and An = Am for all n and m ,
which means that the amplitude ratio and phase dierence of Ev and Eh
stay constant. This is the case for elliptical polarization.
(b) For polarized waves
I = Ih + Iv
Q = Ih Iv = I cos 2 cos 2
U = I cos 2 sin 2
V = I sin 2
Show that when the wave is right-handed circularly polarized Q = U = 0 and
V = I, when it is left-hand circularly polarized, Q = U = 0 and V = I,
and when the wave is linearly polarized, V = 0.
1.2 Polarization
53
Answer:
(a)

4
4Ih Iv = 2 2
tn tm A2n e2vn e2vm
T

4
= 2 2
T

t2n A2n e4vn +
tn tm A2n e2vn e2vm
n=m

4
= 2 2
t2n A2n e4vn +
tn tm (A2n + A2m )e2vn e2vm
T
n>m

4
U = 2 2
tn tm An Am e2vn e2vm cos n cos m
T
4
= 2 2
T
4
V2 = 2 2
T

t2n A2n e4vn cos2 n + 2
tn tm An Am e2vn e2vm cos n cos m
n>m

n
tn tm An Am e2vn e2vm sin n sin m

4
= 2 2
t2n A2n e4vn sin2 n + 2
tn tm An Am e2vn e2vm sin n sin m
T
n
n>m
Thus

4
U2 + V 2 = 2 2
t2n A2n e4vn + 2
tn tm e2vn e2vm An Am cos(n m )
T
n
n>m
We nd
4Iv Ih (U 2 + V 2 )
4
= 2 2
tn tm e2vn e2vm [A2m + A2n 2Am An cos(n m )]
T
n>m
4
2 2
tn tm e2vn e2vm [A2m + A2n 2Am An ]
T
n>m
4
= 2 2
tn tm e2vn e2vm (Am An )2
T
n>m
The right-hand side is always non-negative. Hence 4Iv Ih U 2 + V 2 .
54
1. Fundamentals
(b) For a right-handed circularly polarized wave = /4 , then

Q = I cos(2/4) cos(2) = 0
U = I cos(2/4) sin(2) = 0
V = I sin(2/4) = I
For a left-handed circularly polarized wave = /4 , then
Q = I cos(2/4) cos(2) = 0
U = I cos(2/4) sin(2) = 0
V = I sin(2/4) = I
For linearly polarized wave = 0 , then
V = I sin 0 = 0
Problems
P1.2.1
Consider an electromagnetic wave propagating in the z-direction with

E=x
ex cos(kz t + x ) + yey cos(kz t + y )
where ex , ey , x , and y are all real numbers.
(a) Let ex = 2, ey = 1, x = /2, y = /4. What is the polarization?
(b) Let ex = 1, ey = x = 0. This is a linearly polarized wave. Prove that it can
be expressed as the superposition of a right-hand circularly polarized wave and
a left-hand circularly polarized wave.
(c) Let ex = 1, x = /4, y = /4, ey = 1. This is a circularly polarized wave.
Prove that it can be decomposed into two linearly polarized waves.
P1.2.2
h + vEv = he
h cos(t h ) + vev cos(t v ) show that
From E(t) = hE
Eh
Ev
sin v
sin h = cos t sin
eh
ev
Eh
Ev
cos v
cos h = sin t sin
eh
ev
Eliminating the time dependence t to obtain the equation
Eh
eh
2

+
Ev
ev
2
2
Eh Ev
cos = sin2
eh ev
1.2 Polarization
55
show that this is a tilted ellipse.

P1.2.3
Wave polarization can be viewed by either taking a series of still pictures at

several xed times, called the spatial view point or by making observations at a
xed point in space, called the temporal view point. We dene polarization from
the temporal view point. Let us now look at polarization from the spatial view
point.
Consider an electromagnetic wave with k = 100 Ko propagating in the z
direction.
E(r, t) = E0 [
x cos(kz t) y sin(kz t)]
What are the wavelength and the polarization of this wave?
From the spatial point of view, by taking a picture at t = 0 , the tips of the
electric eld vectors form a helix. Is the helix right-handed or left-handed? What is
the pitch of this helix?
Observing at a xed point in space, show that the tip of the electric eld describes the same polarization as in the temporal view point when the helix advances
without turning.
P1.2.4
Sun navigation was rst observed in 1911. It was found that some species of
ants, horseshoe crabs, honeybees, etc., are sensitive to polarized light. These creatures can navigate as long as there is a small patch of blue sky. The sky polarization
depends upon the angle between the suns rays to a particular point in the sky
and an observers line of sight to the same point. The sunlight, which is unpolarized,
or randomly polarized, excites air molecules which behave like small dipole antennas
when irradiated by the incident electric elds of the sunlight. The scattered electric
eld E s for each excited dipole antenna is linearly polarized in planes perpendicular to the sunlight path; and looking along the sun ray path the scattered wave is
unpolarized, or randomly polarized.
At sunset, if an ant looks directly at the sun ( = 0 ), what is the polarization?
What is the polarization if the ant looks at the zenith ( = 90 ) perpendicular to
the sun ray path? Show that the sky light appears to be partially linearly polarized
when it looks at other parts of the sky [Scientic American, July 1955].
56
1.3
1. Fundamentals
Lorentz Force Law
The interaction of the electric and magnetic elds with the current and
charge densities are governed by the Lorentz force law
f = E + J B
(1.3.1)
where f is the force density. The Lorentz force law relates electromagnetism
to mechanics. The manifestation of the electric eld vector E and the magnetic eld vector B can be demonstrated with the forces exerted on the
charge density and the current density J . It can thus be used to dene
the elds E and B .
Example 1.3.1 Coulombs law.

For static electric elds in the absence of magnetic elds, the Lorentz force law
becomes f = E. Acting on a charged particle q , the total force is F = qE .
Assuming the electric eld E is generated by another charged particle Q situated
at the origin, we have from (E1.1.11.3)
E = r
Q
4 o r2
Thus the total force acting on the charged particle q is

F = r
qQ
4 o r2
which is proportional to the squared inverse distance. This is known as Coulombs

law.
Example 1.3.2 Cyclotron frequency.

Consider a particle with charge q and mass m moving with velocity v in a
uniform static magnetic eld in the
z direction, B =
z B0 . In the absence of
electric elds, if the velocity v has no component in the z direction, the Lorentz
force is perpendicular to the direction of the velocity and the charge particle moves
in the x-y plane. Let v = x
vx + yvy , we have
F = qv B =
xqvy B0 + yqvx B0
Equating to Newtons law
F =m
mdvy
mdvx
dv
=x
+ y
dt
dt
dt
we nd
57
dvx
= c vy
dt
dvy
= c vx
dt
(E1.3.2.1a)
(E1.3.2.1b)
where
qB0
(E1.3.2.2)
m
is called the cyclotron frequency, which is proportional to the magnitude of the
magnetic eld and is independent of the velocity of the particle.
c =
v
Figure E1.3.2.1 Cyclotron frequency.
The solution to (E1.3.2.1) can be written as

dx
(E1.3.2.3a)
= v cos c t
dt
dy
vy =
(E1.3.2.3b)
= v sin c t
dt
To nd the trajectory of the particle, we write the solution of (E1.3.2.3) as
v
x=
sin c t
(E1.3.2.4a)
c
v
y = cos c t
(E1.3.2.4b)
c
vx =
The trajectory of the particle is thus a circle with radius

v
mv
R = (x2 + y 2 )1/2 =
=
c
qB0
(E1.3.2.5)
in view of (E1.3.2.2). It is seen that the larger the magnetic eld, the smaller the
radius. If the charged particle has a velocity component in the z direction, the
trajectory of the particle will follow a helical path.
58
1. Fundamentals
Example 1.3.3 Cyclotron.

A cyclotron [Fig. E1.3.3.1] is an accelerator for charged particles. The a.c.
source provides an alternating voltages at the cyclotron frequency and a charged
particle is repeatedly accelerated every time it passes through the voltage drop.
Uniform B eld
a.c. source
Figure E1.3.3.1 Cyclotron.

Example 1.3.4 Isotope separation.

To separate the isotope Uranium 235 from Uranium 238, the isotopes are rst
vaporized and then ionized by electric discharge. Accelerated through a voltage drop
V , they acquire a kinetic energy qV = mv 2 /2 . Passing through [Fig. E1.3.4.1] a
uniform magnetic eld, the isotopes move along circular paths of dierent radii.
Uniform B eld
V
m238 m235
Figure E1.3.4.1 Isotope separation.
R235
m235 v235
m235
=
=
R238
m238 v238
m238
m238
=
m235
m235
m238
Thus Uranium 235 can be obtained in a collector with a smaller radius.

59
Example 1.3.5
The two rods attract each other when their currents are in the same direction
and are repulsive when their currents are in the opposite directions.
I1
I1
I2
I2
Figure E1.3.5.1 Attractive and repulsive forces.

Example 1.3.6 Linear motor.

In Fig. E1.3.6.1, we show a sliding bar with length l moving perpendicular to
a DC magnetic eld B = zB0 in the z direction. According to the Lorentz force
law, a force
F m = yIl zB0 = x
IlB0
is produced that moves the sliding bar in the x
direction.
sliding bar
Fm
y
x
Figure E1.3.6.1 Linear motor.
If a force is applied to move the sliding bar with velocity v =

x , an induced
voltage V = vlB0 will be generated across the resistor.
60
1. Fundamentals
Example 1.3.7 Magnetic moment and magnetic torque.

A rectangular loop [Figure E1.3.7.1] carrying a static current I is placed in
a static magnetic eld B = x
B0 . The magnetic moment of the current loop is
M = mM
. Its direction m
follows from the right-hand rule: with the ngers
pointing in the direction of the current, the thumb of the right hand is pointing in
the direction of m
. Its magnitude M is equal to the area of the loop A times the
current I , M = AI . If the rectangular loop has lengths lb and ly , the area of the
loop is A = lb ly .
x
B=x
B0
ly
B
lx
Figure E1.3.7.1 Torque on a loop.
The loop is on the x-z plane with two sides aligned with the x-axis and two
sides aligned with the y-axis. Since the static magnetic eld is in the x
direction,
there is no force acting on the two sides with length lb aligned with the x-axis.
The forces acting on the two sides with length ly aligned with the y axis are in
the positive and negative y directions. Thus the loop is rotating around the yaxis following the right-hand rule; with the ngers pointing in the direction of the
rotation, the thumb of the right hand is pointing in the y direction.
The torque acting on the loop is calculated as
T =
1
1
(
yx
Ilz B0 ) lb x
(
yx
Ilz B0 ) = yIAB0
lb x
2
2
For the current conguration, M = zIA and B = x

B0 . In general, the magnetic
torque is
T =M B
(E1.3.7.1)
Thus there is no torque acting on the component of M in the direction of the

magnetic eld.
61
Example 1.3.8
A simple DC motor [Fig. E1.3.8.1] consists of a loop of area A with N turns,
called an armature, which is immersed in a uniform magnetic eld, either produced
by a permanent magnet or an electromagnet. The armature is connected to a commutator which is a divided slip ring. A DC current I is supplied through a pair of
brushes resting against the commutator such that the torque
T = N Bo IA sin
produced by the current on the armature always acts in the same direction.
B = Bo
z
F
I
x
N
S
F
Brush
Armature
Commutator
Figure E1.3.8.1a DC motor.
Brush
Figure E1.3.8.1b Side view of a DC motor.
62
1. Fundamentals
A. Lenz Law and Electromotive Force (EMF)

We apply Stokes theorem to Faradays law and dene the line integral of E
as the electromotive force (EMF):

dl E =
EMF =
C
dS B
(1.3.2)
where
dS B
(1.3.3)
is the magnetic ux linking a loop with area A bounded by a closed contour

C [Fig. 1.3.1]. Equation (1.3.2) states that the EMF is equal to the negative
time derivative of the magnetic ux linking the loop. Thus the EMF always
produces a ux in the loop to oppose the direction of change of the ux
linking the loop; if is increasing, the EMF decreases the ux, and vice
versa. This is known as Lenz law.
V
+
Figure 1.3.1 Flux linking a loop.
Notice that the EMF has unit of voltage (Volt) and not unit of force. The
voltage drop across the loop V is equal to the negative of the induced EMF.
V = EMF =
dt
(1.3.4)
Thus in the presence of a time varying magnetic eld linking a loop, a voltage
is generated to oppose the time change of the magnetic eld. The voltage
generated across the loop V is equal to the negative of the induced EMF.
63
The voltage Vab is dened as the integration of E along a line segment

of ? from point a to point b .
b
Vab =
d? E
(1.3.5)
a
Thus Vab is the potential dierence between points a and b . For positive
Vab , the electric eld vector points from a to b . Point a is at a higher
potential a than b at point b , b < a and Vab = a b .
LeChateliers Principleis the chemists version of Lenz law, which states that
when an external stress (pressure, concentration, or temperature change) is
applied to a chemical system that is in a state of equilibrium, the system will
automatically respond so as to undo the stress applied externally. In Physics,
this same phenomenon is embodied in the Third Law of Motion,that is, for
every action there is an equal and opposite reaction. In biology, a condition
in an organism known as homeostasis means that when a stress is applied
to an organism, the organisms bodily functions automatically respond so as
to remove the stress.
Example 1.3.9 Linear generator.

If a force is applied to move the sliding bar with velocity v = dx/dt as
shown in Fig. E1.3.9.1, the total magnetic eld = xlBo linking the loop will be
decreasing at the rate of vlB0 . According to Lenz law, a current in the bar must
be produced to oppose the decreasing of the magnetic ux. Thus an induced voltage
V = vlB0 is generated across the resistor.
sliding bar
sliding bar
EMF
y
x
Figure E1.3.9.1 Linear generator.

64
1. Fundamentals
Example 1.3.10 AC generator.

An AC generator can be made of the DC motor by replacing the DC current
source with a load resistance R and providing an external rotatory force on the
armature. Applying a torque that makes the loop turn in the direction as shown in
Fig. E1.3.10.1, a motional EMF
V =
dl E =
dl F /q =
dl v B = AB sin
x
slip rings
Figure E1.3.10.1 AC generator.
is produced. For the armature rotating with an angular frequency , we have

v B = lBA sin and = t .
The same result can be derived by using Lenz law
EMF = d/dt
where
dS B = AB cos
(E1.3.10.1)
We nd the generated AC voltage

V = EM F = AB sin t
65
B. Poyntings Theorem and Poynting Vector

Energy conservation immediately follows from the Maxwell equations. Dotmultiply Faradays law (1.1.2) by H , Amperes law (1.1.1) by E and subtract. By making use of the vector identity (EH) = H EEH ,
we obtain Poyntings theorem
(E H) + H
B
D
+E
= E J
t
t
(1.3.6)
The Poynting vector

S =EH
(1.3.7)
is interpreted as the power ow density with the dimension of watts/m2 ,

and H (B/t) + E (D/t) represents the time rate of change of the
stored electric and magnetic energy density. On the right-hand side of (1.3.6),
E J is the power supplied by the current J .
Example 1.3.11
Consider the simple wave solution
E=x
E0 cos(kz t)
(E1.3.11.1a)
H = yH0 cos(kz t)
(E1.3.11.1b)
where H0 = E0 /o and o = o / o is called the characteristic impedance of free

space. Substituting (E1.3.11.1) in (1.3.6) we see that Poyntings theorem is satised.
The Poyntings vector is calculated to be

S = E H = z
o 2
E cos2 (kz t)
o 0
(E1.3.11.2)
In free space, we nd
H
(o H) =
o H H = Wm
t
t 2
t
and
E
( o E) =
o E E = We
t
t 2
t
In the source-free region we also have J = 0 . Poyntings theorem becomes

(E H) +
(We + Wm ) = 0
t
(E1.3.11.3)
66
1. Fundamentals
where
We =
1 2
1
o E = o E02 cos2 (kz t)
2
2
(E1.3.11.4)
is the stored electric energy density and

Wm =
2 1
1
o H = o H02 cos2 (kz t)
2
2
(E1.3.11.5)
is the stored magnetic energy density. It is seen that the stored electric energy is
equal to the stored magnetic energy, We = Wm .
Example 1.3.12 Power, energy, force, and radiation pressure.

The time-average Poynting vector power density is given by
1
< S >=
T
dt S = z
0
E02
1
= z o H02 = zP
2o
2
(E1.3.12.1)
where
P =
E02
1
= o H02
2o
2
is the power density of the wave with units of Watts/m2 . The total time-average
electromagnetic energy density (with units J/m3 ) is equal to the sum of the electric
energy density and the magnetic energy density,
W =< We > + < Wm >=
1
1
o E02 = o H02
2
2
(E1.3.12.2)
We may dene an energy velocityve equal to the ratio of power density to energy
density. We nd P/W = ve = 1/ o o which is the velocity of light.
Radiation pressure is force per unit area F = P/ve (with unit N/m2 ). Thus
the radiation pressure of the wave p is
p = F = P/ve = W =
1
1
o E02 = o H02
2
2
(E1.3.12.3)
which is equal to the time-average total energy density in the wave and acts in the
direction of propagation of the wave. The radiation pressure, although generally
very small, can lead to large scale eects. For example, comet tails are forced to
point away from the Sun due to the radiation pressure from the Sun.
Example 1.3.13
67
Consider an electromagnetic wave with

Ei = x
E0 cos(kz t)
(E1.3.13.1a)
H i = yH0 cos(kz t)
(E1.3.13.1b)
impinging upon the surface of a perfectly conducting surface [Fig. E1.3.13.1]. The
boundary condition at the surface of the boundary requires that
n
(E 1 E 2 ) = 0
(E1.3.13.2a)
n
(H 1 H 2 ) = J s
(E1.3.13.2b)
where n
=
z is the normal to the surface. A perfect conductor is dened to have
elds zero inside, thus E 1 = H 1 = 0 .
y
x
Hi
Hr
Js
Ei
incident
reected
Er
n
=
z
z0
F
E=H=0
Perfect conductor
Figure E1.3.13.1 Reection by a perfect conductor.
The reected wave that satises the boundary conditions (E1.3.13.2) is

E r =
xE0 cos(kz + t)
(E1.3.13.3a)
H r = yH0 cos(kz + t)
(E1.3.13.3b)
which is propagating in the

z direction. The surface current J s at z = 0 is
found to be
Js = n
[(H i + H r ) 0]z=0 = x
2H0 cos t
From the Lorentz force law, the force density acting on J s is
1
F = J s B = 2
z o H02 cos2 t
2
The factor 1/2 is due to the fact that there is magnetic eld only on one side of
the current sheet. The time-average value is thus
F = o H02
which is twice the value of that in Example 1.3.12. This is because the reected
wave exerts a recoil force on the conductor when it launches the reected wave.
68
1. Fundamentals
Example 1.3.14
For an electromagnetic wave with electric eld with E0 = 3 106 V/m (which
is the breakdown electric eld strength for air), nd the power density and radiation
pressure. What is the area required in order to supply the electric power of 2.4
1011 W for use by a nation?
Answer:
The power density is P = 1.2 1010 W/m2 .
The radiation pressure is p = 40 N/m2 .
The area required is 20 m2 .
Applying the divergence theorem to Poyntings theorem (1.3.6), we write

1
dV
dV E J (1.3.8)
o E 2 + o H 2
dS E H =
t
2
2
S
V
V
The left-hand side represents power ow out of the surface enclosing the
volume V . The rst term on the right-hand side represents the depletion
of the electric energy and the magnetic energy inside the volume in order
to supply the outow of the Poynting power. The last term represents the
power generated by the source J inside the volume V .
C. Momentum Conservation Theorem
Substituting the Maxwell equations for and J in the Lorentz force law
f = E + J B
(1.3.9)
we nd that

1
f = (D B) (D E + B H)I D E B H
t
2
(1.3.10)
where I is a unit dyad with diagonal elements equal to 1 and all o-diagonal
elements equal to zero.
The interpretation of the terms is
G = D B = momentum density vector
1
T = (D E + B H)I D E B H
2
= Maxwell stress tensor
(1.3.11)
(1.3.12)
69
Thus we have the theorem

G
(1.3.13)
= f
t
which expresses conservation of momentum. This is in a form similar to
Poyntings theorem in (1.3.6) except that it is now a vector equation. In
fact, (1.3.6) and (1.3.13) combine to become a four-dimensional conservation
theorem in relativity.
T +
Problems
P1.3.1
According to the Bohr model of an atom, electrons revolve around the nucleus
in quantized orbits with radii R = n
h/mv where n is an integer, m is the electron
mass and v is the electron velocity. Letting the nucleus be a positive charge of Ze ,
calculate R by equating the centrifugal force with the Lorentz force. Estimate the
radius for a hydrogen atom with Z = 1 .
P1.3.2
The Earth receives over all frequency bands about 1.5 kW/m2 of power from
the Sun.
(a) The Earth-Sun distance is 150 109 m . How long does it take the sunlight to
reach the Earth?
(b) The Earth radius is 6400 km. What is the total power received by the Earth?
(c) The Sun radiates 1020 W m2 Hz1 at 3 GHz. Assuming constant power
level over 1 GHz bandwidth, what is the Poynting power density and the corresponding electric eld amplitude?
P1.3.3
The interpretation of S = E H as power ow per unit area at a point in space

is a very useful and experimentally veried concept in electromagnetic wave theory.
However, strict adherence to such interpretation in total disregard of Poyntings
Theorem may lead to paradoxical results.
(a) Consider a charged particle placed next to a permanent magnet. Is there power
ow due to the electric eld E generated by the charged particle and the
magnetic eld H generated by the magnet? Show that the net power ow is
zero for static electric and magnetic elds in the absence of currents.
(b) Consider a current density J owing through a resistor of conductivity ,
length l, and circular cross-section with radius a. The electric eld is E =
J/, and the magnetic eld H = aJ/2. Show that the total Poynting power
ow into the resistor is a2 lJ 2 /.
P1.3.4
Use Maxwells equations to show that for J = 0 and = 0 ,

(D B) + W I D E B H = 0
t
70
1. Fundamentals
where the total stored energy density W = D E + B H /2 . Consider D = o E

and B = o H and use index notation.
P1.3.5
The solar wind is a high-conductivity plasma which is emitted radially from

the surface of the Sun. Let us calculate the ux of electromagnetic energy in the
solar wind at the orbit of the Earth.
In the plane of the Earths orbit, the magnetic eld of the Sun is approximately radial, pointing outward in certain regions and inwards in others. This eld
is frozen in the high-conductivity plasma. Since the Sun rotates (with a period of
27 days), and the plasma has a radial velocity, the lines of B are in fact Archimedes
spirals (r = a in polar coordinates) and, at the Earth, they form an angle of about
45 with the Sun-Earth direction. This is the so-called garden hose eect.
At the orbit of the Earth the solar wind has a density of about 107
proton-masses/m3 and a velocity of about 4 105 m/sec , while the magnetic eld
2
of the Sun is about 5 109 (webers/m ).
(a) First show that, in an electrically neutral ( = 0) and nonmagnetic uid of
conductivity and velocity v , the Maxwell equations become
D =0
B = 0
B =0
E
(E + v B) + 0
t
E =
B
t
the polarization currents being negligibly small compared to the conduction

currents.
Note that, for an innite conductivity,
E = v B
This is a satisfactory approximation for the solar wind.
(b) Show that the component of v which is normal to B is
vn =
1
B (v B)
B2
(c) Show that the Poynting vector of the solar wind is

S=
B2
vn
o
Numerically it is approximately equal to 4 109 times the average value of

the Poynting vector of the solar radiation, which is about 1.4 kW/m2 . The
Poynting vector of the solar wind is normal to the local B and it points at an
angle of 45 away from the Sun-Earth direction.
71
(d) Compare the relative magnitudes of the kinetic, electric, and magnetic energy
densities. Which is largest?
P1.3.6
The magnetic moment M is analogous to the expression for the mechnical

angular momentum L in terms of the velocity of mass distributions instead of
the charge distributions. We set M = L and called the gyromagnetic ratio.
In mechanics, the classical equations of motion are dL/dt = T . Applying to the
magnetic moment, we have dM /dt = dL/dt = T = M B .
(a) The magnetic moment of a particle with charge q at position r with velocity
v is dened as
M=
1
qr v
2
If the charged particle has mass m , the mechanical angular momentum is

L = mr v
Determine the gyromagnetic ratio for the charged particle. For a complicated
structure of charged distributions, the gyromagnetic ratio is written as =
gq/2m , where the g-factor g describes the magnetic structure.
Show that the magnetic moment of a plane loop with area A carrying current
I is M = n
IA , where n
is the normal to the plane loop following the righthand rule: with the ngers following the direction of the current, the thumb of
the right haven is pointing in the direction of n
.
(b) Consider a nucleus with magnetic moment M placed in a dc magentic eld
in the z -direction, B = zB0 . The nucleus is precessing about the z axis.
Determine the frequency of precession.
P1.3.7
Joules law, Pd = J E , determines power dissipation per volume due to Ohmic

loss. Derive Joules law by using the Lorentz force law, f = E , and assuming an
average constant drifting velocity v due to collision of the conduction electrons.
72
1.4
1. Fundamentals
Hertzian Waves
A Hertzian dipole is made of two opposite charges q separated by an

innitesimally small distance ? . The dipole moment p = q? has an angular
frequency such that each point charge changes from +q to q and vice
versa in a period of 2/ . Mathematically, p is dened as the product of
? 0 and q such that p is a constant. Assume that the two charges
are situated at z = ?/2 on the z-axis [Fig. 1.4.1]. Hertz solved for all
the electromagnetic elds with the use of a potential function known as the
Hertzian potential , which is a solution of the wave equation
2
o o 2
t
=0
(1.4.1)
+ +q
?
- q
Figure 1.4.1 Hertzian dipole.
To derive the electromagnetic elds E and H , we write = z and

dene a scalar potential such that
=
1
1
=
o
o z
(1.4.2)
and a vector potential A such that

A = o
= zo
t
t
(1.4.3)
1.4 Hertzian Waves
73
It is seen that
=0
(1.4.4)
t
which is known as the Lorenz gauge condition relating the scalar and vector
potentials. Since satises the wave equation, and A also satisfy the
wave equation.
In terms of and A , the magnetic eld H and the electric eld E
are written as
1
2
2
H=
A=x
y
(1.4.5)
o
yt
xt

2
A
1
E =
=
zo o 2
(1.4.6)
t
o
z
t
A + o o
Since satises the wave equation, H and E also satisfy the wave equation. It is seen that H = 0 and E = 0 . It is also straightforward to
show that Amperes law and Faradays law are satised.
From Gauss law for magnetism, B = 0 , we see how the vector potential
A is dened as a result of the vector identity A = 0 . Likewise, from
Faradays law, we have
E =
= A
t
t
We see that
A) = () = 0
t
A) = a scalar potential can be dened.

Thus from (E + t
(E +
The solution to the wave equation for that Hertz studied for his
Hertzian dipole assumes spherical symmetry. We write
q?
cos(kr t)
(1.4.7)
=
4r
where r = (x2 + y 2 + z 2 )1/2 . Substituting into the wave equation, we obtain
the dispersion relation
(1.4.8)
k 2 = 2 o o
relating angular frequency to spatial frequency k as derived in the following example.
74
1. Fundamentals
Derivation of the dispersion relation.
Example 1.4.1

q? k x
=
x
4
2

q?
=
2
x
4
r3
x 2
k
2

3 x 2
2 =
sin(kr t) +
x
r
cos(kr t) 2
k
cos(kr t) 2
r
1
r2
+ 2+ 2
2
x
y
z
x 2
r
cos(kr t)
2
k x
r2 r
sin(kr t) +
k
sin(kr t)
r2
1
sin(kr t) + 3 cos(kr t)
r
k 2 q?
cos(kr t)
4r
(E1.4.1.1)
2
2 q?
=
cos(kr t)
2
t
4r
Substituting 2 and 2 /t2 in (1.4.1), we obtain the dispersion relation (1.4.8).
Example 1.4.2 Derive the wave equation in the spherical coordinate system for
the Hertzian potential and obtain the dispersion relation.
In spherical coordinates, where is a function of r and independent of
and , we nd
2
=
x2 x
=
x
r
x r

x 1
(r)
r r r
x
1
1
(r)
2 (r) +
r
r
r r
1
1
1
x2
1
1
= 2 (r) +
3 (r) + 2 (r)
(r) +
r
r
r r
r r
r
r r
1
1
3x2
3x2
x2 2
(r)
+
(r)
(r)
(r)
+
(r)
+
r3
r2 r
r5
r4 r
r3 r2
Similar results hold for 2 /y 2 , and 2 /z 2 . Substituting in the Laplacian
operator 2 and noticing that r2 = x2 + y 2 + z 2 , we obtain
=
2 =
2 2 2
1 2
+
+
=
(r)
x2
y 2
z 2
r r2
(E1.4.2.1)
The dispersion relation (1.4.8) follows from the wave equation

1 2
2
(r)

=0
o
o
r r2
t2
by substituting the Hertzian dipole solution (1.4.7).
(E1.4.2.2)
1.4 Hertzian Waves
75
Example 1.4.3
We now derive the electric and magnetic eld vectors in the spherical coordinate
system.
The unit vectors for the spherical coordinate system are [Fig. E1.4.3.1]
r = x
sin cos + y sin sin + z cos
=x
cos cos + y cos sin z sin
=
x sin + y cos
z
r
r
y
x
Figure E1.4.3.1 Unit vectors in spherical coordinates.
Writing the position vector r = rr = x

x + yy + zz , we see that x = x
r =
r sin cos , y = r sin sin , and z = r cos . The magnetic eld H is found to
be
2
2
y
yt
xt

q?
r
r
= x
y
sin(kr t)
y r
x r 4r
H=x
y
x q? k
1
y
= x
cos(kr t) 2 sin(kr t)
r
r 4 r
r

kq?
1
=
sin cos(kr t) +
sin(kr t)
4r
kr
(E1.4.3.1)
(E1.4.3.2)
76
1. Fundamentals
where we use the fact that x

yr y xr = sin (
x sin + y cos ) = sin . The
electric eld is found to be, using r = (
xx + yy + zz)/r and z = r cos sin ,
E=
zo o 2
o
z
t

z

1 z
sin(kr t) 2
cos(kr t)
r
r
r

1
+
z 2 o o cos(kr t)
2 r

xz
q?
k
k xz
=
x

cos(kr
t)
+
3
sin(kr t)
4 o
r r2
r2 r2

1 xz
+3 3
cos(kr t)
2
r2 r

yz
k
k yz
+ y
cos(kr
t)
+
3
sin(kr t)
r r2
r2 r2

1 yz
+3 3
cos(kr
t)
2
r2 r2

z
k
k z 2
+ z
cos(kr t) + 3 2
sin(kr t)
r r
r
r
q?
=
4 o
k

r

1 z 2
k
cos(kr t) 2 sin(kr t)
3
r
r
r

k2
1
3 cos(kr t) +
z cos(kr t)
r
r
+3
(E1.4.3.3)
q?
k2
1
k
r cos cos(kr t) + 3 2 sin(kr t) + 3 3 cos(kr t)
4 o
r
r
r
k
1
k2
+ z 2 sin(kr t) 3 cos(kr t)+ cos(kr t)
r
r
r
q?
k
1
r cos 2 2 sin(kr t) + 2 3 cos(kr t)
4 o
r
r
+ sin
k
1
k2
sin(kr
t)
+
cos(kr
t)
cos(kr t)
r2
r3
r
1
k 2 q?
1
=
r2 cos
sin(kr t) + 2 2 cos(kr t)
4 o r
kr
k r
+ sin
1
1
sin(kr t) + 2 2 cos(kr t) cos(kr t)
kr
k r
(E1.4.3.4)
1.4 Hertzian Waves
77
Thus in the spherical coordinate system, the magnetic and electric eld
vectors are

kq?
1
H=
sin
sin(kr t) cos(kr t)
(1.4.9)
4r
kr

1
k 2 q?
1
sin + r2 cos
E=
cos(kr t) +
sin(kr t)
4 o r
k2 r2
kr

sin cos(kr t)
(1.4.10)
The solution of E and H for the Hertzian dipole is of fundamental importance as it is a closed-form analytical expression valid for all distances r
from the dipole.
Consider the following special cases:
Case A) Along the z-axis, = 0, . We nd that there is no magnetic
eld and the electric eld vector takes the form

1
kq?
E = z
cos(kr t) + sin(kr t)
2 o r2 kr
which always points in the z direction.
Case B) In the x-y plane, = /2 . We nd

1
kq?
cos(kr t) +
H=
sin(kr t)
4r
kr

k 2 q?
1
1
E =
1 + 2 2 cos(kr t) +
sin(kr t)
4 o r
k r
kr
where =
z and the electric eld is always perpendicular to the x-y
plane.
Case C) In the immediate neighborhood of the dipole, kr 0 . Keeping
terms of the orders 1/r2 , the magnetic eld vector is
q? sin sin t =
d(q cos t) ? sin
H =
4r2
dt
4r2
I?
sin
=
4r2
(1.4.11)
78
1. Fundamentals
This corresponds to the eld produced by an element of length ? carrying

current I along the z axis, and is known as the Biot-Savart law.
Keeping terms of the orders 1/r3 , the electric eld vector is
E=
q?
(r2 cos + sin ) cos t
4 o r3
(1.4.12)
Let = 0 . We nd H = 0 and
E=
q?
(r2 cos + sin )
4 o r3
This is the electric eld of a static dipole.

Case D) At very great distances, kr 1 , and we only keep terms of the
order of 1/r . The eld vectors are
kq? sin cos(kr t)
H =
4r
2 q?
k
E =
sin cos(kr t)
4 o r
(1.4.13)
(1.4.14)
It is seen that both H and E are tangent to the surface of a large sphere
with radius r . The eld vectors H and E are perpendicular to each other
and their magnitudes are related by (o / o )1/2 .
Case E) To investigate the power and energy issues, Hertz invoked Poyntings theorem. Poyntings power density vector S for elds at very great
distances kr 1 is
k 3
S = E H = r
o
q?
4r
2
sin2 cos2 (kr t)
which is seen to be pointing in the r -direction away from the large sphere.
The time-average power density radiating from the large sphere is obtained in the following manner:
1
< S >=
2
1 k 3
d(t)E H = r
2 o
The radiation pattern is shown in Fig. 1.4.2.
q?
4r
2
sin2
1.4 Hertzian Waves
79
Figure 1.4.2 Radiation power pattern.
Example 1.4.4
We now calculate the time-average power density from the exact solutions for
H in (1.4.9) and E in (1.4.10)
1
< S >=
2
1
=
2
d(t)E H
0
k 3
d(t)
o
2
q?
1
cos(kr t) +
sin(kr t)
4r
kr
0

1
1
r sin2 1 + 2 2 cos(kr t) +
sin(kr t)
k r
kr
1

1
2 sin cos
sin(kr t) + 2 2 cos(kr t)
kr
k r
Notice that the time average of sin(kr t) cos(kr t) is zero, and the time
average of either sin2 (kr t) or cos2 (kr t) is 1/2 . The above expression,
after integration, is equal to the time-average power density at any point r .
Integrating the r directed power over the surface of a sphere of radius

r gives [Fig. 1.4.3]
P = dS r < S >

k 3
q? 2 2
4k 3 q? 2
2
=
d 2r sin
sin =
2 o 4r
3 o
4
0
where 2r2 sin is the ribbon-like surface element to be integrated from
= 0 to = . Notice that the total time-average power leaving the dipole
source obtained by calculating with a spherical surface with any radius r is
the same.
80
1. Fundamentals
2r2 sin d
r sin
rd
x
Figure 1.4.3 Integration geometry for time-average power density.
Topic 1.4A Electric Field Pattern

To study and sketch the electric and magnetic eld lines, Hertz introduced
a parameter Q in terms of the cylindrical coordinate system [Fig. 1.4A.1]
with
x
y
+ y
y
x
=
x sin + y cos =
x + y
=x
cos + y sin = x
noticing that x = cos , y = sin , r2 = 2 + z 2 , and = r sin . Hertz

dened

q?
r
2
=
= r sin
cos(kr t)
Q=
r
r 4r

1
kq?
2
sin sin(kr t)
cos(kr t)
=
4
kr
which is the product of two factors, one depends only on , and the other
1.4 Hertzian Waves
81
x
Figure 1.4A.1 Cylindrical coordinate system.
on r and t . From (1.4.5), we nd

H=
y
t
y
x

1

x
=
y
t
y
x
1
1 Q
= (
x sin y cos )
Q =
t
t
From (1.4.6), we have

2
A
1
E =
=
o zo o 2
t
o
t

2
2
2
1

z 2 + z
o o
o
z
z
z 2
t2

2
2
2
2

1

=
+
+ y
z
x
o
xz
yz
x2
y 2
(1.4A.1)
82
1. Fundamentals
t = 0
t = /2
t =
t = 3/2
Figure 1.4A.2 Electric eld patterns.
1.4 Hertzian Waves
83
where the last term follows from the wave equation for . Noticing that

x
=
=
x
x

x
2
x
=
x2
x
x x2 + y 2 p

2x2
x2
= 2
2
+ 2
x + y 2
(x + y 2 )2
(x + y 2 )
we obtain

1
1
1
E=
Q z
Q
o z

Q
Q
1
1
Q
z
=
=
o
z
The electric eld lines on any -z plane are seen to follow the intersection of Q = constant surfaces with the -z plane. In Fig. 1.4A.2, we plot
the electric eld lines at dierent times.
Example 1.4A.1
Consider the radiation eld zone when kr 1 and
Q=
kq?
sin2 sin(kr t)
4
Construct three constant Q surfaces at t = /2 (or 3/2) and indicate the

electric eld line directions.
Answer: Consider
sin2 cos(kr) = c
We sketch the three cases of c = 0,
1
2,
1 in Fig. E1.4A.1.1.
For c = 0, kr = 2m 2
For c = 12 , kr = 2m
for = 4
kr = 2m 3 for = 2
For c = 1, = 2 and kr = 2m.
84
1. Fundamentals
kz
2m
c = 1/2
c=0
= /4
c=0
c=1
2m /2
2m
kx
2m + /2
Figure E1.4A.1.1 Radiation eld plot.

Example 1.4.5 Determine the static electric eld for a Hertzian dipole oriented
px + ypy + zpz , with dipole moment p = q? .
in a general direction p = x
Answer:
For the pz component, the electric eld vector in the rectangular coordinate
system is obtained from (E1.4.3.3):
E pz =
pz
3
x
3
4 o
r
xz
r2
+ y
3
r3
yz
r2
+ z
3
r3
z 2
r
1
r3
The total electric eld due to all three components is therefore

E = [3
r(
r p) p]
1
1
= [(p r) r + 2
r(
r p)]
4 o r3
4 o r3
Example 1.4.6 For a dipole p = q? , the magnetic eld can be determined from
(E1.4.3.1), and the electric eld can be determined from (E1.4.3.3). The results are
1
k
(p r)
sin(kr t) cos(kr t)
H=
4r
kr
(E1.4.6.1)
1.4 Hertzian Waves
85
1
k2
1
E=
[(p r) r + 2
r(
r p)] 2 2 cos(kr t) +
sin(kr t)
4 o r
k r
kr
[(p r) r] cos(kr t)
(E1.4.6.2)
Example 1.4.7
Applying the Biot-Savart law to derive the magnetic eld of an innitely long
wire, we rst make use of the rst term in (E1.4.6.1) with the same approximation
as for (1.4.11) to obtain
1
d(q cos t)
k
1
(p r) sin(t) =
H
? r
2
4r
kr
4r
dt
1
1
=
(I? r) =
(I? r)
2
4r
4r3
where I is the current and ? denotes the direction and length of the current
element. The vector r = + zz points from the source element to the observation
point. We then place the observation point at the origin and let ? = zdz to integrate
over z from to + as shown in Fig. E1.4.7.1. We thus nd
1
1
I z r
I

H=
dz 2
=
dz
4
4
(z + 2 )3/2
(z 2 + 2 )3/2
I
=
2
The integration was performed by making the substitution z = sec .
z
z
r=

z 2 + 2
dz
Figure E1.4.7.1 Integration of current elements in an innitely long wire.
86
1. Fundamentals
The above result can also be obtained by applying Stokes theorem to Ampères
law H = J . The integration path for the line integral is a circle of radius
around the line source whose area integral gives rise to the current I . The result
is 2H = I .
Problems
P1.4.1
The magnetic eld H and electric eld E of a Hertzian dipole at very large
distances ( kr 1 ) are
kq?
H =
sin cos(kr t)
4r
k 2 q?
E =
sin cos(kr t)
4 o r
(a) Find the Poyntings power density vector
S as a function of time. What is the
time-averaged power density vector S ?
(b) By integrating the Poynting vector over the surface of a sphere of radius r ,
nd the time-averaged power P radiated by the Hertzian dipole.
(c) The amplitude of the current in the Hertzian dipole is Io = q . By using
P = 12 Io2 Rrad , nd the radiation resistance Rrad of the Hertzian dipole.
(d) A radio station is 15 km away from a city. The transmitting antenna tower may
be modeled as a Hertzian dipole antenna of dipole moment q? . To maintain
the FCC standard of 25 mV/m eld strength in the city, how much radiation
power P must be provided?
P1.4.2
Find H and E in terms of the vector Hertzian potential = z .

P1.4.3
The scalar potential is dened from E = . Show from (1.4.12) that

=
q?
cos t
4 o
z
1
r
This corresponds to an electric dipole, with dipole axis along the z axis and oscillating with the period T = 2/ .
P1.4.4
From a time-evolution point of view, examine the electric eld lines at the
spatial point r = /2 and = /4 . The slope of the eld line dr/rd is obtained
from
Q=
kq?
1
sin2 sin(kr t)
cos(kr t)
4
kr
1.4 Hertzian Waves
87
Show that
1
cos(kr t)
sin(kr t) + kr
dr
2 cos

=
1
1
d
k sin 1 k2 r2 cos(kr t) kr
sin(kr t)
Substituting kr = and = /4 , the evolution of dr/d with time can be

determined. Show that the line segment rotates a full circle as t spans 2 radian.
In particular, let dr/rd be equal to tan 1 at t = 0 , and tan 2 at t = /2 .
What is 1 2 and is the line segment at t = 0 perpendicular to the line segment
at t = /2 ? Sketch the rotation of the line segment at several discrete times from
t = 0 to t = .
P1.4.5
A turnstile antenna consists of two Hertzian dipoles positioned at right angles

to each other with Hertzian potential
=x
q?
q?
cos(kr t) y
sin(kr t)
4r
4r
Find the electric and magnetic eld vectors and show that on the x-y plane as
k 1 , the electric eld is in the -direction with space-time dependence of
sin(k t ) .
P1.4.6
The Biot-Savart law states that the magnetic eld at (r, , ) produced by an
element of length ? at the origin carrying current I along the z axis is
o I?
B =
sin
4r2
Consider a wire with innite length carrying current I in the direction of z .
Use the Biot-Savart law to show that the magnetic eld produced by the wire is
B =
o I
2
where is the distance from the wire. Apply Stokes theorem to Ampères law
without the displacement term, nd B and conrm the above result.
For a high-voltage transmission line carrying current I = 1 kA , nd the magnetic eld strength 10 meters away from the wire, and compare with the earth
magnetic eld strength which is approximately 5 105 Tesla.
P1.4.7
The solution to the wave equation for that Hertz studied for his Hertzian
dipole takes the following form:
=
q?
sin(kr t)
4r
88
1. Fundamentals
Derive the magnetic and electric eld vectors and show that
kq?
1
H =
sin sin(kr t) +
cos(kr t)
4r
kr

1
k
=
(p r)
4r
kr

k 2 q?
1
1
E=
sin 1 2 2 sin(kr t) +
cos(kr t)
4 o r
k r
kr
1

1
cos(kr t) 2 2 sin(kr t)
+
r2 cos
kr
k r

1

k2
1
=
[(p r) r + 2
r(
r p)] 2 2 sin(kr t) +
cos(kr t)
4 o r
k r
kr
+ [(p r) r] sin(kr t)
Study the limiting cases of kr 0 and kr 1 . What is the electric eld when
= 0?
1.5
89
Reection and Transmission
When a linearly polarized plane wave is propagating in a general x

and z
direction, we write
E(r, t) = E cos (k r t) = E cos (kx x + kz z t)
(1.5.1)
H(r, t) = H cos (k r t) = H cos (kx x + kz z t)
(1.5.2)
where E and H are constant vectors and the dispersion relation gives
kx2 + kz2 = 2 = k 2
(1.5.3)
We dene a wave vector or k vector

k=x
kx + zkz
(1.5.4)
x+
y y+
zz
The scalar product of the wave vector and the position vector r = x
gives k r = kx x + kz z . A constant phase front is determined by k r =
constant, which indicates that the front is a plane perpendicular to the k
vector. The electromagnetic wave is called a plane wave.
Substituting (1.5.1) and (1.5.2) into source-free Maxwell equations, we
nd, for the plane wave solution,
k E = H
(1.5.5)
k H = E
(1.5.6)
kE =0
(1.5.7)
kH =0
(1.5.8)
It is seen from (1.5.7) and (1.5.8) that E and H lie on the constant phase
plane perpendicular to k . Eqs. (1.5.5)(1.5.8) can also be derived for generally polarized plane waves.
The vector Poyntings power density is
1
S(r, t) = E H =
E (k E) cos2 (k r t)
1
k |E|2 cos2 (k r t)
=
The time-average vector power density is
2
1
1
1
d(t) k |E|2 cos2 (k r t) =
k |E|2
< S(r, t) > =
2 0
2
which shows that the time-average Poyntings power density is also propagating in the direction of k .
90
1. Fundamentals
A. Reection and Transmission of TE Waves

Consider a plane wave incident from a medium with permittivity 0 and
permeability 0 upon another medium with permittivity t and permeability t . The boundary surface of the two media is situated at x = 0 . Let
the incident plane wave be linearly polarized with the electric eld vector in
the y direction [Fig. 1.5.1].
z
Ht
Er
Et
Hr
r
i
Hi
0 , 0
t , t
Ei
Region 0
Region t
Figure 1.5.1 Reection and transmission of TE waves at a plane boundary

separating Regions 0 and t .
We call the x-z plane the plane of incidence, which is formally dened
as the plane formed by the normal to the boundary surface and the incident
wave vector k . The incident electric eld vector E i is perpendicular to the
plane of incidence and the magnetic eld vector H i is parallel to the plane
of incidence. We call the incident wave a transverse electric (TE) wave. The
TE wave is also called perpendicularly polarized, horizontally polarized, or
simply the E wave or s wave.
An incident wave of general polarization can be decomposed into two linearly polarized waves; one with the electric eld vector perpendicular to the
plane of incidence which is the TE wave, and one with the electric eld vector parallel to the plane of incidence which is called the transverse magnetic
(TM) wave. The TM wave will have the magnetic eld vector perpendicular
to the plane of incidence and is also called parallelly polarized, vertically
91
polarized, or simply the H wave or p wave. We shall rst study the case of
TE wave incidence.
The incident electric eld vector is assumed to have unit amplitude and
is written as
E i (r, t) = yEiy
Eiy = cos(k r t) = cos(kx x + kz z t)
(1.5.9a)
with the wave vector

k=x
kx + zkz
The magnetic eld vector can be obtained from (1.5.5)
H i (r, t) =
1
1
k E(r, t) =
(
xkz + zkx )Eiy
0
0
(1.5.9b)
The Poynting vector power density for the incident plane wave is
S i (r, t) = E i (r, t) H i (r, t) = k
=k
1
E2
0 iy
1
cos2 (kx x + kz z t)
0
(1.5.9c)
which is in the direction of the wave vector k .

The reected elds for the incident TE wave are
k r =
xkrx + zkrz
(1.5.10a)
E r (r, t) = yEry
= yR cos(krx x + krz z t)
H r (r, t) =
(1.5.10b)
1
1
k r E r (r, t) =
(
xkrz + zkrx )REry (1.5.10c)
0
0
The Poynting vector power density for the reected plane wave is
S r (r, t) = k r
R2
0
(1.5.10d)
where R is the reection coecient for the electric eld component Eiy .
92
1. Fundamentals
The incident wave vector k = x

kx + zkz and the reected wave vector
k r =
xkrx + zkrz are governed by the dispersion relations
kx2 + kz2 = 2 0 0 = k 2
(1.5.11)
2
2
krx
+ krz
= 2 0 0 = kr2
(1.5.12)
This is seen by substituting (1.5.9a) and (1.5.10a) in the Helmholtz wave

equations for Eiy and Ery .
In Region t , we write the transmitted TE wave solution in the following
form
kt = x
ktx + zktz
E t (r, t) = yEty
(1.5.13a)
= yT cos(ktx x + ktz z t)
(1.5.13b)
1
1
k t E t (r, t) =
(
xktz + zktx )T Ety (1.5.10c)
t
t
T2
S t (r, t) = k t
(1.5.10d)
t
H t (r, t) =
where T is the transmission coecient, and the dispersion relation

2
2
+ ktz
= 2 t t = kt2
ktx
(1.5.14)
ktx + zktz .
governs the magnitude kt for the transmitted wave vector k t = x
Let the boundary surface be at x = 0 where the tangential components
of E and H are continuous for all z and t . We obtain
cos(kz z t) + R cos(krz z t) = T cos(ktz z t)
(1.5.15)
kx
krx
ktx
cos(kz z t)
R cos(krz z t) =
T cos(ktz z t)
0
0
t
(1.5.16)
Since (1.5.15) and (1.5.16) must hold for all z and t , it follows that
kz = krz = ktz
This is called the phase matching condition.
(1.5.17)
93
From the dispersion relations (1.5.11) and (1.5.12), we nd krx = kx .

Equations (1.5.15) and (1.5.16) then reduce to
1+R=T
0 ktx
T
1R=
t kx
(1.5.18)
(1.5.19)
Note that the boundary conditions of normal D and normal B components

continuous at x = 0 are satised since the condition of continuous normal B
yields the same equation as (1.5.18) and there is no normal D component.
The reection and transmission coecients R and T are determined
from (1.5.18) and (1.5.19), giving
TE
R = R0t
=
1 pT0tE
1 + pT0tE
(1.5.20)
TE
T = T0t
=
2
1 + pT0tE
(1.5.21)
and
where,
pT0tE =
0 ktx
t kx
(1.5.22)
T E in (1.5.20) is called
With pT0tE for the TE waves dened in (1.5.22), R0t
the Fresnel reection coecient for a TE wave incident from Region 0 and
T E is
reected at the boundary separating Regions 0 and t . In (1.5.21), T0t
the transmission coecient from Region 0 to Region t .
Equation (1.5.17), the phase matching condition, is a very important
formula arising from the boundary conditions. In terms of the angle of incidence i , the angle of reection r , and the angle of transmission t , and
the relation kr = k as seen from (1.5.11) and (1.5.12), the phase matching
condition (1.5.17) gives
k sin i = kr sin r = kt sin t

Thus the angle of reection is equal to the angle of incidence r = i , and
0 0
sin t
k
n0
=
=
=
(1.5.23)
sin i
kt
t t
nt
where n0 = c 0 0 is called the refractive index for Region 0 and nt =
c t t is the refractive index for Region t . Equation (1.5.23) is known as

Snells law.
94
1. Fundamentals
z
Srz
Siz
Stz
Srx
Stx
Six
Siz
Stz
Srz
Figure 1.5.2 Power conservation at a plane boundary.
Power Conservation
The time-average Poynting vectors for the incident, the reected, and
the transmitted waves are calculated to be
< Si > =
1
1
k=
(
xkx + zkz )
20
20
(1.5.24)
< Sr > =
|R|2
|R|2
kr =
(
xkx + zkz )
20
20
(1.5.25)
< St > =
|T |2
|T |2
kt =
(
xktx + zkz )
2t
2t
(1.5.26)
Power conservation is observed by considering a control volume across the

boundary surface [Fig. 1.5.2]. We must prove that the x components of all
the Poynting vectors entering and exiting the control volume are equal. We
dene the power reection coecient or the reectivity to be
r=
x < S r >
= |R|2
x
< S i >
(1.5.27)
95
By the same token, we dene the power transmission coecient or the transmissivity to be
x
< S t >
(1.5.28)
= p0t |T |2
t=
x
< S i >
By virtue of (1.5.20)(1.5.21), we see that
r+t=1
This demonstrates power conservation for reection and transmission at a
plane boundary surface.
Exercise 1.5.1
Notice that
< Six > < Srx > =< Stx >
< Siz > < Srz > =< Stz >
B. Reection and Transmission of TM Waves

The reection and transmission of TM waves [Fig. 1.5.3] by a plane
boundary can be carried out in a manner similar to the treatment of TE
waves. The incident magnetic eld vector H i = yHiy is assumed to have
unit amplitude and the magnetic and electric eld components are written
as
Hiy = cos(kx x + kz z t)
kz
Eix =
cos(kx x + kz z t)
0
kx
Eiz =
cos(kx x + kz z t)
0
(1.5.29a)
(1.5.29b)
(1.5.29c)
The reected eld components for the incident TM wave are

Hry = RT M cos(krx x + krz z t)
krz T M
Erx =
R
cos(krx x + krz z t)
0
krx T M
Erz =
R
cos(krx x + krz z t)
0
(1.5.30a)
(1.5.30b)
(1.5.30c)
96
1. Fundamentals
z
Er
Ht
Hr
kr
r
kt
Et
i
k
Hi
Ei
Figure 1.5.3 Reection and transmission of TM waves.
where RT M is the reection coecient for the magnetic eld component

Hiy . In Region t , the transmitted TM eld components are
Hty = T T M cos(ktx x + ktz z t)
ktz T M
Etx =
T
cos(ktx x + ktz z t)
t
ktx T M
Etz =
T
cos(ktx x + ktz z t)
t
(1.5.31a)
(1.5.31b)
(1.5.31c)
where T T M is the transmission coecient for the magnetic eld component

Hiy .
kx + zkz , the reected wave vector k r =
The incident wave vector k = x
xkrx + zkrz , and the transmitted wave vector satisfy the same dispersion
relations (1.5.11), (1.5.12), and (1.5.14) as for the TE wave case. Matching
the boundary conditions of tangential components of E and H continuous
at x = 0 , we obtain the same phase matching condition (1.5.17) and the
reection and transmission coecients RT M and T T M
TM
RT M = R0t
=
1 pT0tM
1 + pT0tM
(1.5.32)
TM
T T M = T0t
=
2
1 + pT0tM
(1.5.33)
and
97
where
pT0tM =
0 ktx
t kx
(1.5.34)
Note that the Fresnel reection coecient for TM waves is now representing
the ratio of the reected and incident magnetic elds.
Exercise 1.5.2 At the surface of a perfect conductor, we may calculate the

reection coecients by letting t . We nd that for TE waves pT0tE
TE
TM
and R0t
1 while for TM waves pT0tM 0 and R0t
1 . Thus the tangential
electric eld vanishes at the boundary and the tangential magnetic eld doubles its
strength in order to support the induced surface currents.
C. Brewster Angle and Zero Reection

We now discuss the Brewster angle for nonmagnetic media where t =
0 . The Brewster angle b is the incident angle at which there is no reected
power. Setting R = 0 or p0t = 1 we nd, from (1.5.22), for TE waves
ktx = kx or
(1.5.35)
kt cos t = k cos i
To solve for the incident angle, we make use of Snells law condition
kt sin t = k sin i
(1.5.36)
It follows from (1.5.35) and (1.5.36) that t = i and t = 0 . Thus there

is zero reection since there is no boundary.
For TM waves, we obtain from (1.5.34), 0 ktx = t kx or
0 kt cos t = t k cos i
(1.5.37)
Since kt = 0 t and kt = 0 t , we obtain from (1.5.37)

k cos t = kt cos i
(1.5.38)
Multiplying (1.5.36) and (1.5.38), and call the incident angle for R = 0 the
Brewster angle b , we obtain
sin 2b = sin 2t
98
1. Fundamentals
z
0 , 0
t , t
kr
kt
b
b
ki
Region 0
Region t
Figure 1.5.4 Incidence at the Brewster angle.
The solution t = b again implies identical media and is of no interest. In

view of the fact that both b and t are larger than zero and less than 90 ,
we obtain as the other solution
b + t =
(1.5.39)
Since the reected direction is perpendicular to the transmitted direction, the

reected wave vector k r is perpendicular to the transmitted wave vector k t
[Fig. 1.5.4].
Physically we can explain this by visualizing the dielectric media as
consisting of dipoles that are excited by the transmitted wave and radiating
at the same frequency. Each individual dipole has a radiation pattern that
is maximum in a direction perpendicular to the dipole axis and null along
the dipole axis. For a TM wave excitation, all dipoles oscillate parallel to the
plane of incidence along the E-eld lines. At the Brewster angle of incidence,
the reected k r vector is in the same direction as the dipole oscillation in
the transmitted medium. Thus, no TM wave is reected.
Substituting (1.5.39) in (1.5.38), we obtain the Brewster angle

kt
t
(1.5.40)
= tan1
b = tan1
k
0
In Fig. 1.5.5, we plot the reectivities as functions of the incident angle. In
general, on a solid dielectric surface, the TE waves reect more than the
99
TE
TM
0
Incident angle
Figure 1.5.5 Reectivity of TE and TM waves.
TM waves. When an unpolarized wave is incident at b upon an isotropic

medium, the reected wave becomes linearly polarized perpendicular to the
plane of incidence. Thus the Brewster angle is also referred to as the polarization angle.
Problems
P1.5.1
A TE plane wave is incident from a medium with permittivity and permeability 1 upon a second medium with the same permittivity but dierent
permeability 2 . The incident plane wave has an angular frequency and an
electric eld vector.
(a) Find the Brewster angle i = B for zero reection for an incident TE wave
and express your result in terms of 1 and 2 .
(b) Let 1 = 42 ; nd the critical angle i = C for total reection and give its
numerical value in degrees.
P1.5.2
When the incident k vector is normal to a plane boundary, a TE wave becomes

a TEM wave; a TM wave also becomes a TEM wave. Compare the reection and
transmission coecients for TE and TM waves at normal incidence. Do both TE
and TM results reduce to the same numbers? If not, why? Do the reectivities
and transmissivities for TE and TM waves at normal incidence reduce to the same
result?
P1.5.3
A uniform plane wave is incident on a planar boundary at an incident angle

from a dielectric medium to a half space as shown in Figure P1.5.3.1, where > o .
100
1. Fundamentals
The electric eld vector of the incident wave is given by

E i = E0 (
x cos + y + z sin ) cos (kx x kz z t) ,
where the amplitude E0 is a constant.
ki
kr
, o
o , o
Figure P1.5.3.1 Conguration of the problem.
(a)
(b)
(c)
(d)
(e)
Show that this incident electric eld satises Gauss law.

Decompose the incident electric eld into its TE and TM components.
What is the polarization of the incident wave?
Determine the expression for the reected electric eld.
For incident angles less than the critical angle, < c , what are the polarizations of the reected wave and the transmitted wave?
(f) For incident angles greater than the critical angle, > c , show that the phase
dierence between the TE and TM components of the reected electric eld is
given by

1
= T M T E = 2 tan
cos
where n is the refractive index dened by

You may nd the following identity useful:
tan(A + B) =

sin
2
n2

2
/sin
/ o o .
tan A + tan B
1 tan A tan B
(g) If = 6 o , for what angle is the reected wave circularly polarized?

P1.5.4
The gas laser depicted in Fig. P1.5.4.1 uses Brewster angle quartz windows
on the gas discharge tube in order to minimize reection losses. Determine the angle
if the index of refraction for quartz at the wavelength of interest is n = 1.46 .
Because of these windows, the laser output is almost completely linearly polarized.
101
Part of laser light

Mirror
Mirror
Tube
Brewster window
Part of laser light
Figure P1.5.4.1 A gas laser with Brewster windows.
What is the direction of polarization, i.e., is E parallel or perpendicular to the

paper? Why?
P1.5.5
Sun light glares caused by reections from plane surfaces are partially linearly
polarized.
(a) Determine the Brewster angle for t = 9 o . The Brewster angle, B , is also
called the polarization angle because at B the reected wave is entirely TE
polarized.
(b) Your polaroid glasses absorb one linear component of incident light. To minimize sun glare, what component, TE or TM, reaches your eyes after passing
through the glasses? Explain why.
102
1.6
1. Fundamentals
Phase Matching
The phase matching condition in equation (1.5.17)

kz = krz = ktz
(1.6.1)
states that the tangential components of the incident, the reected, and
the transmitted wave vectors are continuous. The incident, reected, and
transmitted wave vectors all lie in the plane of incidence. The magnitudes
of the wave vectors are determined from the dispersion relations
k 2 = kr2 = 2 0 0
(1.6.2)
kt2 = 2 t t
(1.6.3)
and
We shall now illustrate the phase-matching condition with a simple gure

[Fig. 1.6.1]. Let n0 < nt . We draw a k surface for Region 0 which has a
shorter radius than the k surface for Region t . Requiring that the tangential
kz components be equal, we construct the k vectors for the reected and
transmitted waves as shown in Fig. 1.6.1.
Suppose that the medium in Region 0 is denser than the medium in
Region t such that n0 > nt . Then the radius of the k surface in Region t is
shorter than that in Region 0 [Fig. 1.6.2]. By the phase-matching condition,
we see that as kz of the incident wave becomes larger than kt , there is
no intersection with the smaller semi-circle. The k surface in Region t is
described by
2
+ kz2 = kt2
(1.6.4)
ktx
Since kz > kt , ktx = iktxI now becomes a purely imaginary parameter,
where ktxI is real. The dispersion relation in (1.6.4) becomes
2
ktxI
+ kz2 = kt2
(1.6.5)
As we shall see later, imaginary wave vector components suggest exponential

attenuation of wave amplitudes. We shall show in the following that the
transmitted wave is evanescent in the x
direction, a phenomenon known as
total reection.
1.6 Phase Matching
103
kz
kt
kr
r
i
kx
k
k surface for medium 0
k surface for medium t

Figure 1.6.1 Phase matching using k surfaces.
kz
kr
kt
c
c
kx
k
k surface for medium t
k surface for medium 0
Figure 1.6.2 k surface for medium 0 is larger than that for medium t.
104
1. Fundamentals
A. Critical Angle and Total Reection

Total reection occurs as kz > kt . Let kt = kz = k sin c , the angle c
is called the critical angle of incidence. When the angle of incidence is larger
than the critical angle, we have total reection. We illustrate with the case of
TM wave incidence. The development applies equally well to the case of TE
wave incidence. In Region t , we derive the transmitted TM wave solution
as follows.
2 = k 2 , the dispersion relation (1.6.5)
Since kz > kt , ktx
txI
2
+ kz2 = kt2
ktxI
(1.6.6)
suggests that solutions to the wave equation

2
2
2
+
2 Hty = 0
x2 z 2
t
have an exponential x dependence and sinusoidal z and t dependence. We
write the solution for Hty and its associated electric eld as
H t (r, t) = yHty = y T ektxI x cos(kz z t + )
E t (r, t) =
(1.6.7)
T ektxI x
[
xkz cos(kz z t + ) + zktxI sin(kz z t + )]
t
(1.6.8)
which satises Ampères law,

E =
x Hy + z Hy
t
z
x
The solution can be regarded as a plane wave with constant phase fronts
perpendicular to the boundary surface. The transmitted wave propagates
along the z direction with the phase velocity /kz . Its amplitude is maximum at the surface and decays exponentially away from the surface. The
wave is called a surface wave.
The total eld in the incident region 0 must have the same dependence
cos(kz z t + ) , which can be written as
H 0 (r, t) = y 2 cos(kx x ) cos(kz z t + )
(1.6.9)
= Hi + Hr
= y cos(kx x + kz z t) + y cos(kx x + kz z t + 2) (1.6.10)
1.6 Phase Matching
105
The rst term is the incident wave and the second the reected wave. The
associated total electric eld in Region 0 is, from (1.6.9)
E 0 (r, t) = x
2kz
cos(kx x ) cos(kz z t + )
0
2kx
+ z
sin(kx x ) sin(kz z t + )
0
(1.6.11)
The boundary condition of continuity of tangential E and H at x = 0

is satised by equating (1.6.7) with (1.6.9) and the z -component of (1.6.8)
with the z -component of (1.6.11).
T = 2 cos
0 k1xI
T = 2 sin
t kx
It follows that
tan =
0 k1xI
= p0tI
t kx
The second term in (1.6.10) is the reected wave, which gained a phase shift
of 2 , known as the Goos-H
anchen shift.
The Poynting power vector of the transmitted wave is
S t (r, t) =
T 2 e2ktxI x
[
z kz cos2 (kz z t + )
t
x
ktxI sin(kz z t + ) cos(kz z t + )] (1.6.12)
The time-average Poynting power density is

< S t (r, t) > = z
T 2 e2ktxI x
kz
2 t
(1.6.13)
Consequently, the time-average Poynting power for the reected wave in the
x direction is equal to the time-average incident Poynting power. Thus

the incident power is totally reected and the evanescent transmitted wave
carries no time-average power into the transmitted region in the x
direction.
For this reason, the phenomenon is considered to be total reection.
At total reection, the transmitted wave is an inhomogeneous plane
wave, and its amplitude exponentially attenuates away from the boundary
surface. The transmitted wave is an inhomogeneous plane wave propagating
106
1. Fundamentals
along the surface and has maximum amplitude at the surface, which is also
called a surface wave. The surface wave is evanescent in the x
direction.
At total reection, we nd the reected wave from (1.6.11) by using the
trigonometric relation of 2 cos A cos B = cos(A + B) + cos(A + B) and
2 sin A sin B = cos(A + B) + cos(A + B) ,
2kz
cos(kx x ) cos(kz z t + )
0
2kx
+ z
sin(kx x ) sin(kz z t + )
0
kz
kx
= (
x
z
) cos(kx x + kz z t)
0
0
kz
kx
+ (
x
+ z
) cos(kx x + kz z t + 2) (1.6.14)
0
0
E 0 (r, t) = x
The reected eld is seen from (1.6.14) and (1.6.10) as

H r (r, t) = y cos(kx x + kz z t + 2)
kz
kx
E r (r, t) = (
x
+ z
) cos(kx x + kz z t + 2)
0
0
(1.6.15)
(1.6.16)
The reected wave is seen to have gained a phase shift of 2 = 2 tan1 p0tI .
Example 1.6.1
For TE incidence, we have
E t (r, t) = yEty = y T ektxI x cos(kz z t + )
ktxI x
(E1.6.1.1)
Te
[
xkz cos(kz z t + ) + zktxI sin(kz z t + )] (E1.6.1.2)
t
E 0 (r, t) = y 2 cos(kx x ) cos(kz z t + )
= y cos(kx x + kz z t) + y cos(kx x + kz z t + 2)
(E1.6.1.3)
2
H 0 (r, t) =
[
xkz cos(kx x ) cos(kz z t + )
0
+ zkx sin(kx x ) sin(kz z t + )] (E1.6.1.4)
H t (r, t) =
which satises Faradays law,
H=x
Ey z Ey
t
z
x
1.6 Phase Matching
107
The boundary condition of continuity of tangential E and H at x = 0 gives

T = 2 cos
0 k1xI
T = 2 sin
t kx
It follows that
tan =
0 k1xI
= pTE
0tI
t kx
The reected eld is

E r (r, t) = y cos(kx x + kz z t + 2)
kz
kx
H r (r, t) = (
x
z
0
0
(E1.6.1.5)
(E1.6.1.6)
The reected wave is seen to have gained a phase shift of 2 = 2 tan1 pTE
0tI .
Example 1.6.2
To explore the implication of the Goos-H
anschen shift for the reected wave,
consider reection of a TM wave from a perfect conductor situated at x = x0 . The
incident and the reected waves will be
H i (r, t) = y cos(kx x + kz z t)
kz
kx
E i (r, t) = (
x
z
) cos(kx x + kz z t)
0
0
H r (r, t) = y cos(kx x + kz z t + 2)
kz
kx
E r (r, t) = (
x
+ z
0
0
(E1.6.2.1)
(E1.6.2.2)
(E1.6.2.3)
(E1.6.2.4)
At the surface of the conductor of x = x0 , the z components of the electric elds

cancel each other. We thus have 2kx x0 = 2 = 2 tan1 p0tI .
For TE wave reection from a boundary surface at x = x0 , the incident and
the reected electric elds are
E i (r, t) = y cos(kx x + kz z t)
(E1.6.2.5)
E r (r, t) =
y cos(kx x + kz z t + 2)
(E1.6.2.6)
At the surface of the conductor of x = x0 , the z components of the electric elds

cancel each other. We thus have 2kx x0 = 2 + = 2 tan1 pTE
0tI .
108
1. Fundamentals
B. Guided Waves in a Dielectric Waveguide

Consider a dielectric waveguide with boundaries at x = d and x =
0 [Fig. 1.6.3]. In Region t , H t and E t are the same as (E1.6.1.1) and
(E1.6.1.2)
H t (r, t) = yHty = y 2 cos t ektxI x cos(kz z t + t )
2 cos t ktxI x
E t (r, t) =
e
[
xkz cos(kz z t + t )
t
+ zktxI sin(kz z t + t )]
Region s
Region 0
Region t
s, s
0 , 0
t , t
x = d
(1.6.17)
(1.6.18)
x=0
Figure 1.6.3 Asymmetric dielectric waveguide.
In Region 0 the electric and magnetic eld vectors are

H 0 = y 2 cos(kx x t ) cos(kz z t + t )
2kz
E0 = x
cos(kx x t ) cos(kz z t + t )
0
2kx
+ z
sin(kx x t ) sin(kz z t + t )
0
(1.6.19)
(1.6.20)
In Region s , the eld components must have the dependence cos(kz z t +

x-direction, we write
t ) and exponentially attenuate in the
H s = ySekbxI x cos(kz z t + t )
(1.6.21)
S ksxI x
Es =
e
[
xkz cos(kz z t + t ) zksxI sin(kz z t + t )] (1.6.22)
s
1.6 Phase Matching
109
The boundary condition of continuity of tangential E and H at x = d

gives
SeksxI d = 2 cos(kx d + t )
(1.6.23)
0 ksxI ksxI d
Se
= 2 sin(kx d + t )
s kx
(1.6.24)
Eliminating S from (1.6.23) and (1.6.24), we nd

tan(kx d + t ) =
0 ksxI
= tan s = tan(m s )
s kx
(1.6.25)
with tan s = 0 ksxI / t kx . It follows that kx d + t = m s , or

2kx d + 2t + 2s = 2m
(1.6.26)
which is referred to as the guidance condition. The phase gained by the

round trip 2kx d , plus the phases gained at the boundary surfaces at x = 0
and x = d add up to 2m . The elds in Region 0 are called guided waves.
Problems
P1.6.1
Compare the phenomena of total reection for > C and total transmission for
= B at an isotropic dielectric interface.
(a) Total reection occurs at a range of incident angles larger than the critical
angle C ; total transmission of TM waves occurs only at the Brewster angle
B .
(b) Total reection occurs only when the incident medium is denser than the transmitted medium. The Brewster angle occurs for any two media.
(c) When an unpolarized wave is totally reected, the reected wave is still unpolarized. When the TM wave components of an unpolarized wave are totally
transmitted, the reected wave contains only TE waves.
Suppose a TM wave is incident at an angle such that = B > C . Then the
wave is totally transmitted and at the same time it is totally reected. Explain.
P1.6.2
Consider a plane wave incident on a planar boundary at z = 0 from a dielectric medium with = 9 o as shown in Figure P1.6.2.1. The right-hand circularly
polarized incident electric eld is
110
1. Fundamentals
region 1
o , o
region 2
t , t
Reflected wave
Ei
Figure P1.6.2.1
E i = E0 ( 3
x + z) cos(kx x kz z t) + 2
y sin(kx x kz z t)
where E0 is a real constant. The reected eld is

E r = E0 2RT E y sin(kx x + kz z t) + RT M ( 3
x + z) cos(kx x + kz z t)
(a) Show that the incident angle is 30 .
(b) For kx = 1 Ko , nd the frequency (Hz) and the wavelength (m) in region 1.
(c) Find the value of t (0 < t / o < ) for which the incident angle is equal to
the critical angle. Find the polarization of the reected eld.
(d) Find the value of t (0 < t / o < ) for which the reected wave is linearly
polarized.
P1.6.3
A laser beam in free space with the polarization of electric eld parallel to the
paper is incident normally upon a glass surface. There is 16% power of the incident
wave being reected and the rest being transmitted. Neglect the reection on the
bottom surface. The reection coecient of TE and TM incident waves are given
by, respectively,
R
TE
cos i
cos i +
R
where n =
TM
n2 cos i
n2 cos i +
n2 sin2 i
n2 sin2 i
n2 sin2 i
n2 sin2 i
/ o is the refraction index and i is the incident angle.
(a) What is the amplitude of the reected electric eld E r in terms of the amplitude of the incident electric eld E
i?
(b) What is the refraction index (n = / o ) of the glass?
(c) Let the surface of the glass rotate by = sin1 (2/3) about an axis perpendicular to the paper. How much of the incident power is reected?
1.6 Phase Matching
111
(d) Let the surface of the glass rotate by about an axis perpendicular to the
paper, so that the laser beam is totally transmitted without reection. What
is the rotation angle in radians?
P1.6.4
It is interesting to compare the following two cases: (i) at total reection, the
incident power is totally reected; and (ii) when the transmitting medium is a
perfect conductor, the incident power is also totally reected. As far as the reected
TE wave is concerned, it makes no dierence whether the wave is reected by a
perfect conductor or by a dielectric boundary if both boundaries render the reected
wave with identical phases.
(a) Consider a perfectly conducting boundary at x = d . Show that the wave is
phase-shifted by at the perfectly conducting boundary.
(b) The path length from x = 0 to x = d provides another phase shift, 2kx d ,
for the wave. Find d such that the reected TE wave at x > 0 experiences
the same amount of phase shift in both cases.
Experiments by Goos and H
anchen have demonstrated that a beam of light
is laterally shifted when totally reected at a dielectric boundary. We must note,
however, that although this argument can be used to display a lateral shift for the
reected k vector, the analysis is true only for plane waves, and there is no way to
observe a lateral shift experimentally for plane waves.
P1.6.5
Consider a plane wave incident from a dielectric region with permittivity =

3 o upon a halfspace with = o as shown in Figure P1.6.5.1.
z
o , o
z
3 o, o
r
i
o , o
o , o
Region 2
Region 1
kr
3 o, o
kt
t
ki
ki
Figure P1.6.5.1
(a) Find the Brewster angle for Region 1.

(b) Suppose the transmitted electric eld is given by
z x
3
E0
E t = y cos(ktx x + ktz z t) + E0
sin(ktx x + ktz z t)
2
2 2
112
1. Fundamentals
(i) Determine the incident and transmitted angles, i and t .

(ii) What is the polarization of the transmitted eld? Be sure to specify the
handedness (left or right) if necessary.
(iii) What is the polarization of the reected wave? Be sure to specify the
handedness (left or right) if necessary.
(iv) Give an expression for the incident electric eld, E i , and the reected
electric eld, E r .
(c) For this part, assume that the incident wave is a TE wave with magnitude,
|E| = E0 and i = c where c is the critical angle.
(i) Find the Goos-H
anchen phase shift of the reected wave.
(ii) Make a sketch of the electric eld amplitude, Ey , for x > 0 .
(iii) Suppose a second boundary is placed at the position x = d to form a
dielectric waveguide as shown in Figure P1.6.5.1. Determine the possible
values of kx .
P1.6.6
Total reection occurs as kz > kt . The transmitted magnetic eld Hty and
its associated electric eld components can be written as
Hty = ektxI x [T1 cos(kz z t) + T2 sin(kz z t)]
kz ktxI x
Etx =
e
[T1 cos(kz z t) + T2 sin(kz z t)]
t
ktxI ktxI x
Etz =
e
[T1 sin(kz z t) T2 cos(kz z t)]
t
(P1.6.6.1)
(P1.6.6.2)
(P1.6.6.3)
We write the reected magnetic and electric eld components as

Hry = R1 cos(kx x + kz z t) + R2 sin(kx x + kz z t)
kz
Erx =
[R1 cos(kx x+kz zt)+ R2 sin(kx x +kz z t)]
0
kx
Erz =
0
(P1.6.6.4)
(P1.6.6.5)
(P1.6.6.6)
The transmitted and reected eld amplitudes are due to the incident wave
Hiy = cos(kx x + kz z t)
kz
Eix =
cos(kx x + kz z t)
0
kx
cos(kx x + kz z t)
Eiz =
0
Determine R1 , R2 , T1 , and T2 .
(P1.6.6.7)
(P1.6.6.8)
(P1.6.6.9)
1.6 Phase Matching
113
P1.6.7
Total reection occurs as kz > kt . The transmitted TE wave solution

Ety = ektxI x [T1 cos(kz z t) + T2 sin(kz z t)]
kz ktxI x
Htx =
e
[T1 cos(kz z t) + T2 sin(kz z t)]
t
ktxI ktxI x
Htz =
e
[T1 sin(kz z t) + T2 cos(kz z t)]
t
(P1.6.7.1)
(P1.6.7.2)
(P1.6.7.3)
where T1 and T2 denote the transmission coecients for the electric eld. We
write the reected electric and magnetic eld vectors as
Ery = R1 cos(kx x + kz z t) + R2 sin(kx x + kz z t)
kz
Hrx =
0
kx
Hrz =
0
(P1.6.7.4)
(P1.6.7.5)
(P1.6.7.6)
The transmitted and reected eld amplitudes are due to the incident TE wave
Eiy = cos(kx x + kz z t)
kz
Hix =
cos(kx x + kz z t)
0
kx
Hiz =
cos(kx x + kz z t)
0
Determine R1 , R2 , T1 , T2 , and all eld components.
(P1.6.7.7)
(P1.6.7.8)
(P1.6.7.9)
114
1.7
1. Fundamentals
Wave Guidance
A. Guidance by Conducting Parallel Plates

Consider the guidance of electromagnetic waves by a pair of perfectly conducting plates at x = 0 and x = a [Fig. 1.7.1]. For TM waves, the Maxwell
equations are

2
2
2
+
2 Hy = 0
(1.7.1a)
x2 z 2
t
Ex = H y
(1.7.1b)
t
z
(1.7.1c)
Ez =
Hy
t
x
co
s(
kx
x+
kz
x+
kx
os(
z
z
z
t
x
z
x=0
,
x=d
Figure 1.7.1 Parallel-plate waveguide.
In the parallel-plate waveguide, the wave is guided along the

z directions.
The two wave solutions with wave vectors k and k r in the guided region
are
H i = y cos(kx x + kz z t)
1
E i = [
xkz zkx ] cos(kx x + kz z t)

H r = y R cos(kx x + kz z t + r )
R
E r = [
xkz + zkx ] cos(kx x+kz zt + r )

(1.7.2)
(1.7.3)
(1.7.4)
(1.7.5)
1.7 Wave Guidance
115
The boundary conditions at the parallel plates require that the tangential
electric eld be zero at x = 0 and x = d .
cos(kz z t) + R cos(kz z t + r ) = 0
cos(kx d + kz z t) + R cos(kx d + kz z t + r ) = 0
(1.7.6a)
(1.7.6b)
Solution to the above equations yields R = 1 , r = 2m , and

(1.7.7)
2kx a = 2m
which is known as the guidance condition. It states that in the x

direction
the bouncing waves must interfere constructively with 2kx a = 2m in order
for the wave to be guided [Fig. 1.7.2]. The set of discrete kx values admissible
inside the guide is
kx =
m
m
m1 =
Ko = kcm
a
2a
(1.7.8)
where m is any integer. We name the guided waves TMm modes.

kz
( )1/2
kx
0
/d
2/d
3/d
Figure 1.7.2 Interpretation of the guidance condition.
Thus as a result of the boundary condition at x = 0 and x = a , the spatial

variation along the x
direction of a guided wave must be an integer number
in a distance of 2a . The magnetic and electric vector elds are
H = yA cos kx x cos(kz z t)
(1.7.9)
kz
kx
E=x
A cos kx x cos(kz z t) + z A sin kx x sin(kz z t) (1.7.10)

116
1. Fundamentals
Hy = cos(
mx
)
a
TM 1
m=1
TM 2
m=2
TM 3
m=3
Figure 1.7.3 Field amplitudes for TM1 , TM2 , and TM3 modes.
In Fig. 1.7.3, we plot Hy for m = 1, 2, 3 . They are standing waves in the

transversal x direction and propagate in the z direction. We see that there
are more spatial variations in the waveguide with separation of d , when the
x component of the spatial frequency, kx = m/2a Ko , is higher with larger
m . The velocity of the TMm mode in the z direction is determined from
the dispersion relation
2
(1.7.11)
kz2 = k 2 kcm
kz
TE 1 TM 1
TM 2
TE 2
TM 0
TM 3 TE 3
kc1
kc2
kc3
Figure 1.7.4 kz diagram.
1.7 Wave Guidance
117
The phase and group velocities are

vp = /kz = ck/kz
vg = d/dkz = cdk/dkz = ckz /k
(1.7.12)
(1.7.13)
vp t
ct
vg t
x
Figure 1.7.5 Distances traveled with phase and group velocities.
where c = 1/ . The phase velocity vp is larger than c , as is seen from

Fig. 1.7.5. Let sin m = kcm /k = m/ka = m/2a . We see that vp =
c/ cos m and vg = c cos m , thus vp vg = c2 . In Figure 1.7.6 we show
that for a propagating TMm mode, as frequency increases, the angle m
decreases, and the group velocity vg = ck/kz increases.
kz
k3
k2
k1
m
kx
kcm
Figure 1.7.6 Guidance with increasing frequency.
118
1. Fundamentals
2 k 2 ) = k 2 ,
It is seen from (1.7.11) that as k < kcm , kz2 = (kcm
zI
suggesting that the guided wave will attenuate in the z direction. The elds
satisfying the Maxwell equations and the boundary conditions become
H = yA cos kx x ekzI z cos t

kzI
kx
E=x
A cos kx x ekzI z sin t z A sin kx x ekzI z sin t

(1.7.14)
(1.7.15)
The time-average power in the z direction is zero, and the guided modes
for k < kcm are evanescent.
The spatial frequency at which kz = 0 is called the cuto spatial frequency kcm
m
m
kcm =
Ko =
(1.7.16)
2a
cm
corresponding to cuto wavelength cm = 2a/m . In order for the mth order
TM mode to propagate, the spatial frequency k must be larger than kcm or
the wavelength must be smaller than cm . Notice that if the TMm mode
is propagating, then all TMl modes with l < m can also propagate. Thus
for a given spatial frequency k such that kcm < k < kc(m+1) , there will
be m + 1 TM modes admissible inside the waveguide. The lowest-order TM
mode is TM0 whose kc0 = 0 .
The electric and magnetic elds for the TM0 mode are, since kx = 0
and kz = k ,
Hy = A cos(kz t)
kA
Ex =
cos(kz t)

(1.7.17a)
(1.7.17b)
which is equivalent to a plane wave propagating in the z direction. The

TM0 mode is also called the fundamental mode or the TEM mode in the
parallel-plate waveguide.
Exercise 1.7.1 Find the cuto wavelength cm and the cuto angular frequency
cm corresponding to the cuto spatial frequency kcm = m/d .
Answer:
The corresponding cuto wavelength is cm = 2d/m and the corresponding
cuto angular frequency is cm = m/d( )1/2 .
1.7 Wave Guidance
119
Example 1.7.1 TE modes in parallel-plate waveguide.

We write the solution for TE waves as
Ey = (A cos kx x + B sin kx x) cos(kz z t)
(E1.7.1.1)
The boundary conditions at x = 0, d require Ey = 0 which gives A = 0 and the

same guidance condition (1.7.7). We thus obtain the electric eld in the parallelplate waveguide
Ey = B sin kx x cos(kz z t)
Ey
(E1.7.1.2)
TE 1 mode
x
d
0
Ey
TE 2 mode
x
0
Ey
TE 3 mode
x
0
Figure E1.7.1.1 Field amplitudes for TE1 , TE2 , and TE3 modes.
The above result can be interpreted in terms of plane waves reecting from the
conducting plates in the same way as for the TM waves. One important dierence
is that TE0 does not exist and the lowest-order TE mode is TE1 .
120
1. Fundamentals
Example 1.7.2
Mathematically, the Helmholtz wave equation can be solved by the method of
separation of variables. In the method of separation of variables, we let
Hz (x, y, z, t) = X(x)Y (y)Z(z)T (t)
Substituting into the wave equation, we obtain
1 d2 X(x)
1 d2 Y (y)
1 d2 Z(z)
1 d2 T (t)
+
+

=0
X(x) dx2
Y (y) dy 2
Z(z) dz 2
T (t) dt2
We then let
1 d2 X(x)
X(x) dx2
1 d2 Y (y)
Y (y) dy 2
1 d2 Z(z)
Z(z) dz 2
1 d2 T (t)
T (t) dt2
= kx2
= ky2
= kz2
= 2
where kx , ky , kz , and are separation parameters related by

kx2 + ky2 + kz2 = 2
There are two independent solutions to each of the four ordinary dierential equations. Out of the possible sixteen combinations, many are absent owing to the constraints imposed by physical requirements and boundary conditions. For instance,
cos kx x cos t + sin kx x sin t = cos(kx x t)
represents a plane wave propagating in the +
x direction, and
cos kx x cos kz z cos t + cos kx x sin kz z sin t = cos kx x cos(kz z t)
is a wave propagating in the z direction in a parallel-plate waveguide. If one of the
separation parameters, say, kz = ikzI , then the two independent solutions will be
cosh kz z and sinh kz z or ekzI z and ekzI z representing growing and attenuating
waves in the z direction.
1.7 Wave Guidance
121
B. Guidance by Rectangular Waveguides

Consider a metallic rectangular waveguide having dimensions a along the
x axis and b along the y axis [Fig. 1.7.7]. The TE wave elds inside the
guided region can be written as
E =
y B sin kx x sin(kz z t)
kz
H=x
B sin kx x sin(kz z t)
kx
+ z
B cos kx x cos(kz z t)
(1.7.18)
(1.7.19)
The boundary conditions at x = 0, a require Ey = Ez = 0 and at y = 0, b

require Ex = Ez = 0 which give rise to the same guidance condition (1.7.7).
Since there is no variation in the y directions, the elds are for TEm0 modes.
a
b
z
x
Figure 1.7.7 Metallic rectangular waveguide.
In treating guided waves along the z direction, the z t dependence

of all eld vectors is written as cos(kz z t) and sin(kz z t) , where
kz is the propagation constant and the signs indicate propagation along
positive and negative z directions. With this dependence, we can replace
2 /z 2 by kz2 and 2 /t2 by 2 . Due to its unique position in guided
wave theory, the propagation direction z is used to classify guided modes.
122
1. Fundamentals
From the Maxwell equations written in Cartesian component form,
Hz
H y = Ex
y
z
t
Hx
H z = Ey
z
x
t
Hy
H x = Ez
x
y
t
Ez
Ey = Hx
y
z
t
Ex
Ez = Hy
z
x
t
Ey
Ex = Hz
x
y
t
Ex +
Ey +
Ez = 0
x
y
z
Hx +
Hy +
Hz = 0
x
y
z
(1.7.20a)
(1.7.20b)
(1.7.20c)
(1.7.21a)
(1.7.21b)
(1.7.21c)
(1.7.22)
(1.7.23)
we can express the x and y eld components in terms of the z components. Taking the time derivative of (1.7.20a) and substituting (1.7.21b) in
(1.7.20a), we obtain

Hz +
Ex
Ez = 2 Ex
t y
z z
x
from which we can express Ex in terms of Ez and Hz . Following a similar
procedure, we nd for all eld components transversal to the z direction:
2
1
2
(1.7.24a)
Ez
Hz
Ex = 2
kz2 xz
yt
2
1
2
Ey = 2
Ez +
Hz
(1.7.24b)
kz2 yz
xt
2
1
2
(1.7.25a)
+

H
Ez
Hx = 2
z
kz2 xz
yt
2
1
2
(1.7.25b)
Hy = 2
Hz
Ez
kz2 yz
xt
To nd the equation governing the longitudinal component Ez , we substitute (1.7.21a-b) in (1.7.20c) and make use of (1.7.22). To nd the equation governing the longitudinal component Hz , we substitute (1.7.20a-b) in
1.7 Wave Guidance
123
(1.7.21c) and make use of (1.7.23). The result is

2
2
2
2
+
kz + Ez = 0
x2 y 2

2
2
2
2
+
k
+

Hz = 0
z
x2 y 2
(1.7.24c)
(1.7.25c)
Equations (1.7.24) and (1.7.25) are the fundamental formulas for the study
of guided waves in cylindrical waveguides. We shall now study cylindrical
waveguides with rectangular cross-section.
Consider a metallic rectangular waveguide having dimensions a along
the x axis and b along the y axis [Fig. 1.7.7]. We shall study transverse
electric (TE) elds where Ez = 0 and all electric eld components are
transversal to the direction of propagation z. Consider a guided wave propagating in the +
z direction with Hz = H0 cos kx x cos ky y cos(kz z t) .
The following eld components are solutions of (1.7.24) and (1.7.25):
Ez = 0
Ex
Ey
Hx
Hy
Hz
(1.7.26)
ky
= 2
H0 cos kx x sin ky y sin(kz z t)
kz2
kx
= 2
H0 sin kx x cos ky y sin(kz z t)
kz2
kx kz
= 2
H0 sin kx x cos ky y sin(kz z t)
kz2
ky kz
= 2
H0 cos kx x sin ky y sin(kz z t)
kz2
= H0 cos kx x cos ky y cos(kz z t)
(1.7.27)
(1.7.28)
(1.7.29)
(1.7.30)
(1.7.31)
They also satisfy the boundary conditions that Ez and Ey vanish at x = 0 ,

and that Ez and Ex vanish at y = 0 . The rest of the boundary conditions
require that Ez and Ey vanish at x = a , and that Ez and Ex vanish at
y = b . We thus obtain the guidance condition for the rectangular waveguide
as
kx a = m
ky b = n
where m and n are integer numbers.
(1.7.32a)
(1.7.32b)
124
1. Fundamentals
Substituting (1.7.31) in (1.7.25c), we nd the dispersion relation

kx2 + ky2 + kz2 = 2 = k 2
(1.7.33)
Combining the dispersion relation (1.7.33) with the guidance conditions

(1.7.32), we nd the propagation constant
kz =
k 2 (m/2a)2 (n/2b)2 Ko
(1.7.34)
According to particular values of m and n, the TE waves inside the rectangular waveguide are classied into TEmn modes. The rst index m is
associated with the number of variations along the x
direction and the second index with the number of variations along the y direction. Substituting
the guidance conditions (1.7.32) in the eld expressions, we see that for larger
m there will be more variations for the elds as a function of x , and for
larger n there will be more eld variations along the y direction.
Cuto occurs when kz becomes imaginary such that the wave attenuates
exponentially along the direction of propagation. For a TEmn mode, the
cuto spatial frequency is
kcmn =
(m/2a)2 + (n/2b)2 Ko
(1.7.35)
The lowest order TE mode is TE10 with kc10 = /a m1 = 1/2a Ko , which

is the fundamental mode. As an example, if we let a = 2 cm, b = 1 cm, we
nd that the cuto spatial frequency for the TE10 mode is kc10 = 25 Ko ,
and the cuto temporal frequency is f = 7.5 GHz.
Example 1.7.3
For the TE00 mode, kx = ky = 0 , and kc00 = 0 . It has a zero cuto frequency,
and we shall show that it is not a guided wave mode. We see from (1.7.31) that
Hz = H0 cos(kz z t)
Since kz = , Equations (1.7.24) and (1.7.25) are not useful in obtaining the
transverse eld components. We nd from (1.7.23) that
kz H0 cos(kz z t) = 0
(E1.7.3.1)
Thus k = kz = 0 and we only have a static magnetic eld in the z direction.

1.7 Wave Guidance
125
Example 1.7.4 TM modes in rectangular waveguide.

For TM elds, all magnetic elds are transverse to the direction of propagation z. We have Hz = 0 , and all eld components can be derived from a single
longitudinal component Ez . We write
Ez = E0 sin kx x sin ky y cos(kz z t)
(E1.7.4.1)
We nd for all eld components

Ex =
kx kz
E0
2
kz2
cos kx x sin ky y sin(kz z t)
ky kz
E0 sin kx x cos ky y sin(kz z t)
2 kz2
= E0 sin kx x sin ky y cos(kz z t)
ky
= 2
E0 sin kx x cos ky y sin(kz z t)
kz2
kx
= 2
E0 cos kx x sin ky y sin(kz z t)
kz2
=0
(E1.7.4.2)
Ey =
(E1.7.4.3)
Ez
(E1.7.4.4)
Hx
Hy
Hz
(E1.7.4.5)
(E1.7.4.6)
(E1.7.4.7)
The dispersion relation is the same as (1.7.33). We see that at x = 0 and a , Ez

and Ey vanish, and at y = 0 and b, Ez and Ex vanish, provided that
kx a = m
ky b = n
(E1.7.4.8a)
(E1.7.4.8b)
which is the guidance condition identical to (1.7.32). The propagation constant kz

is the same as (1.7.34), and we see that for a TMmn mode, the cuto wavenumber
is the same as (1.7.35). There is, however, a very signicant dierence between the
TE and TM modes. We note that for TMmn modes neither m nor n can be
zero. For the TEmn modes it is possible to have either m or n equal to zero.
The lowest order TM mode is seen to be the TM11 mode. As an example, if we
let a = 3 cm, b = 1.5 cm, we nd the cuto wavenumber for the TM11 mode to be
kc11 = 234 m1 = 37.3 Ko .
Example 1.7.5 Power in rectangular waveguide.

For TE modes, we have
Ez = 0
Hz = H0 cos kx x cos ky y cos(kz z t)
Hz
2
kz yt
(E1.7.5.1)
Ex =
(E1.7.5.2a)
126
1. Fundamentals
Hz
2
kz xt
2
1
Hx = 2
Hz
2
kz xz
2
1
Hy = 2
Hz
2
kz yz
The Poynting power density is
Ey =
(E1.7.5.2b)
(E1.7.5.3a)
(E1.7.5.3b)
S = z(Ex Hy Ey Hx ) yEx Hz + x
Ey Hz
< S > = z < Ex Hy Ey Hx >
2 2

kz
2
2
2
2
2
= z
H
k
sin
k
x
cos
k
y
+
k
cos
k
x
sin
k
y
x
y
x
y
o
x
y
2( 2 kz2 )2
For TM modes, we have
Hz = 0
Ez = E0 sin kx x sin ky y cos(kz z t)
(E1.7.5.4)
1
Ez
2
kz xz
2
1
Ey = 2
Ez
kz2 yz
2

Hx = 2
Ez
kz2 yt
2

Hy = 2
Ez
kz2 xt
Ex =
(E1.7.5.5a)
(E1.7.5.5b)
(E1.7.5.6a)
(E1.7.5.6b)
< S > = z < Ex Hy Ey Hx >

kz
= z
E02 kx2 cos2 kx x sin2 ky y + ky2 sin2 kx x cos2 ky y
2
2
2
2( kz )
Thus for both TE and TM modes, the wave power is guided along the z direction.
2
When k < kcmn , kz2 = kzI
, and the guided modes are evanescent. For TE
modes,
Hz = H0 cos kx x cos ky y ekzI z cos t
(E1.7.5.7)
Ez = E0 sin kx x sin ky y ekzI z cos t
(E1.7.5.8)
and for TM modes,

The time-average Poynting power density is seen to be zero for both TE and TM
modes.
1.7 Wave Guidance
127
C. Rectangular Cavity Resonators

Consider a metallic rectangular cavity as shown in Fig. 1.7.8. It is a waveguide closed with metallic walls at z = 0 and z = d. The elds for the TE
waves (1.7.18) and (1.7.19) inside the resonator can be written as
E =
y B sin kx x sin kz z sin t
kz
H=x
Bsin kx x cos kz z cos t
kx
+ z
B cos kx x sin kz z sin t
(1.7.36)
(1.7.37)
a
b
z
d
y
x
Figure 1.7.8 Rectangular cavity.
The boundary conditions at x = 0, a require Ey = Ey = 0 , at x = 0, a

require Ey = Ey = 0 , and at z = 0, d require Ex = Ey = 0 , which give
rise to the resonance condition
kr =
(m/2a)2 + (n/2b)2 + (p/2d)2 Ko
(1.7.38)
and the resonance mode is called TEm0p modes.

A resonator with uniform cross-section in the z direction can be viewed
as a waveguide with both ends closed. Instead of guided waves propagating
along the z axis, the waves are standing in the z direction. The standing
wave can be viewed as a superposition of a guided wave in the +
z direction
and a guided wave in the
z direction. The formulation for waveguides is
also applicable to resonators.
128
1. Fundamentals
Consider the metallic rectangular cavity as shown in Fig. 1.7.8. It is a

waveguide closed with metallic walls at z = 0 and z = d. We nd for TM
modes
Ez = E0 sin kx x sin ky y cos kz z cos t

kx kz
Ex = 2
E0 cos kx x sin ky y sin kz z cos t
kz2
ky kz
E0 sin kx x cos ky y sin kz z cos t
Ey = 2
kz2
ky
E0 sin kx x cos ky y cos kz z sin t
Hx = 2
kz2
kx
Hy = 2
E0 cos kx x sin ky y cos kz z sin t
kz2
(1.7.39a)
(1.7.39b)
(1.7.39c)
(1.7.39d)
(1.7.39e)
and for TE modes

Hz = H0 cos kx x cos ky y sin kz z cos t
kx kz
Hx = 2
H0 sin kx x cos ky y cos kz z cos t
kz2
ky kz
H0 cos kx x sin ky y cos kz z cos t
Hy = 2
kz2
ky
H0 cos kx x sin ky y sin kz z sin t
Ex = 2
kz2
kx
Ey = 2
H0 sin kx x cos ky yb sin kz z sin t
kz2
(1.7.40a)
(1.7.40b)
(1.7.40c)
(1.7.40d)
(1.7.40e)
To satisfy the boundary conditions, we must have

kx a = m
ky b = n
kz d = p
which is called the resonance condition for the resonator.
The dispersion relation for both the TM and TE modes is
kr2 = (m/a)2 + (n/b)2 + (p/d)2
This gives the resonant spatial frequency
(1.7.41a)
(1.7.41b)
(1.7.41c)
1.7 Wave Guidance
129
kr =
(m/2a)2 + (n/2b)2 + (p/2d)2 Ko
(1.7.42)
The resonant spatial frequencies for TMmnp modes and TEmnp modes are
identical. It is interesting to observe that TMmn0 modes correspond to waveguide modes at cuto, where kz = 0.
Example 1.7.6
When the resonator dimensions are such that a > b > d , the lowest resonant
spatial frequency is found to be
kr =
(1/2a)2 + (1/2b)2 Ko
(E1.7.6.1)
with m = n = 1 and p = 0. The mode inside the resonator is TM110 . The

non-zero eld components for the TM110 mode are
x
y
sin
cos t
a
b
x
y
Hx =
E0 sin
cos
sin t
b
a
b
x
y
Hy =
E0 cos
sin
sin t
a
a
b
Ez = E0 sin
(E1.7.6.2a)
(E1.7.6.2b)
(E1.7.6.2c)
y
H
x
a
Figure E1.7.6.1 TM110 mode in rectangular cavity.
The eld distribution is illustrated in Fig. E1.7.6.1. We see that the electric elds
are perpendicular to the plate boundaries at z = 0 and z = d and concentrate at
the center of the cavity so that the tangential E eld vanishes at the boundaries
x = 0, a and y = 0, b. As an example, if we let a = 4 cm, b = 3 cm , and d = 2 cm ,
we nd the resonant spatial frequency for the TM110 mode to be kr = 21 Ko .
130
1. Fundamentals
Problems
P1.7.1
Consider a perfectly conducting parallel-plate waveguide lled with a dielectric

medium for z > 0 as shown in Fig. P1.7.1.1. The dielectric medium has permittivity
1 . The operating frequency is 30/2 GHz. The guided wave propagates in the z
direction.
Dielectric
o , 1
o , o
z
z=0
Figure P1.7.1.1 Parallel-plate waveguide.
(a) Let d = 2 3 cm and consider the empty waveguide with 1 = 0 (in the
absence of the dielectric). Which TEm and TMm modes can propagate in
this waveguide.
(b) Find expressions for the E and H elds for the TM2 mode in the absence
of the dielectric.
(c) What are the phase and group velocities for the TM2 mode at this operating
frequency in the absenceof the dielectric?
(d) Let 1 = 3 0 and d = 2 3 cm. For waves propagating in the +
z direction,
for which values of m will the TMm modes be totally reected at the dielectric
boundary? Why?
(e) For which value of m will a propagating TMm mode be totally transmitted
(no reection) and why?
P1.7.2
Consider a perfectly conducting parallel-plate waveguide, which is lled with

air for z < 0 and lled with a dielectric medium of permittivity 1 for z > 0 .
Let d = 1 cm and 1 = 4 o . The waveguide is excited at f = 20 GHz . Waves are
guided in
z directions.
(a) List all possible propagating modes of TE and TM waves for z < 0 and z > 0 .
(b) If the incident wave comes from z < 0 , what is the reection coecient for
TE1 mode at the boundary z = 0 ?
(c) If the dielectric medium ( 1 ) is replaced by a perfectly conducting medium,
nd the total eld E for TE1 mode for z < 0 .
1.7 Wave Guidance
131
P1.7.3
Consider a perfectly conducting parallel-plate waveguide with the plates (separated by d ) parallel to the y-z plane as shown in Figure P1.7.3.1. The waveguide
is lled with a plasma medium with the plasma frequency
p and permeability

o . The permittivity of the plasma medium is = o 1 p2 / 2 . Let a Hertzian
dipole be placed on the x-y plane along the z -axis. Assume that the Hertzian
dipole can excite all the modes in the waveguide.
x
Parallel-plate waveguide
r
d
p = p
z
Plasma medium
Hertzian dipole
Figure P1.7.3.1
(a) In the absence of the parallel-plate waveguide, sketch the radiation pattern of
the Hertzian dipole on the x-y plane in free space.
(b) What is the
polarization of the electric eld on the x-y plane in the waveguide?
(c) Let = 2p . Find the range of the separation
d in terms of the wavelength
in free space o , where o = 2/ko = 2/ o o , so that there is ONLY
ONE mode propagating in the parallel-plate waveguide. What is this mode
(indicate TE or TM, and the mode number m )?
P1.7.4
exhaust air duct
4.19m
6.55m
fresh air duct
Figure P1.7.4.1 Tunnel modeled as rectangular waveguide.
An AM radio in an automobile cannot receive any signal when the car is inside
a tunnel. Consider, for example, the Lincoln Tunnel under the Hudson River, which
132
1. Fundamentals
was built in 1939. A cross-section of the tunnel as shown in Figure P1.7.4.1. Ignore
the air ducts; assume they are closed. Model the tunnel as a rectangular waveguide
of dimension 6.55m 4.19m .
(a) Give the range of frequencies for which only the dominant mode, T E10 , may
propagate.
(b) Explain why AM signals cannot received.
(c) Can FM signals be received? Above what frequencies?
P1.7.5
In an air-lled rectangular waveguide with dimensions a = 3 2 cm and b =

a/2 , the guided wave is given by
E = yEo sin
H=x
Ho sin

a
a
x sin
x sin
z t

z t + zHo cos

a
x cos

a
z t
where Eo and Ho are real constants.

(a) What is the mode for this wave? Indicate TEmn or TMmn and the mode
number m and n .
(b) Show that the frequency is f = 5 GHz .
(c) What is the phase velocity in z direction in terms of the speed of light c ?
(d) What is the cuto frequency of this mode?
(e) If the waveguide is used as a rectangular cavity resonator for frequency f =
5 GHz by closing the ends at z = 0 and z = d using perfectly conducting
plates, what is the value of d for the lowest mode? Indicate this lowest mode
( TEmnp or TMmnp ) and the mode number m, n and p . (Hint: d can be
larger than a and b .)
P1.7.6
Consider a rectangular waveguide with dimensions 1 cm 0.5 cm.

(a) What are the cuto frequencies for the rst ve modes?
(b) If the waveguide is excited at 20 GHz, what are the propagation constants kz
for the rst ve modes?
(c) If the waveguide is excited at 50 GHz, how many modes will propagate?
P1.7.7
A metallic rectangular waveguide has dimension 1.5 cm 0.75 cm .

(a) Determine the cuto frequencies for the rst 5 modes.
(b) Write down the electric and magnetic elds for the TM11 mode.
(c) Sketch the electric eld distribution of the TM11 mode at cuto in the x-y
plane.
(d) Closing the waveguide at z = 0 and z = 3 cm , the TM11 mode at cuto is
the fundamental resonator eld TM110 . What is the resonant frequency?
(e) If the resonator is indented at the center of the front wall at z = 0 or the back
wall at z = 3 cm , will the resonant frequency for TM110 mode become lower
or higher?
1.7 Wave Guidance
133
P1.7.8
Consider an air-lled metallic rectangular waveguide operating at frequency

f = 7.5 GHz ( k = 50 m1 ) with dimensions a = 3 cm and b < a .
b
x
Figure P1.7.8.1
(a) For the fundamental mode ( TE10 mode), determine

(i) the cuto frequency
(ii) the propagation constant kz
(iii) the phase velocity
(b) Consider a propagating mode inside the waveguide operating at the same frequency f = 7.5 GHz (k = 50 m1 ) . The plots of E versus time t at z = 0
and z = 1 cm are shown below.
E
At z = 0
At z = 1 cm
to = 2 1011 s
Figure P1.7.8.2 Electric eld as function of time.
(i)
(ii)
(iii)
(iv)
Determine the period T.

Calculate the phase velocity.
Show that kz = 30 m1 .
Find the mode designation (i.e., TEmn , TMmn ) for the propagating mode
in the waveguide by assigning numerical values to m and n .
(v) Calculate the length b in cm.
P1.7.9
(a) A rectangular air-lled waveguide is 2 cm 4 cm. List all the modes which
can propagate at a frequency of 15 GHz and give their corresponding cuto
frequencies.
(b) Thin slots may be cut in the walls of waveguides without signicantly perturbing the propagating waves provided that the slots do not transect wall currents,
134
1. Fundamentals
slot at x
b
y
a
Figure P1.7.9.1 Waveguide slot.
i.e., are parallel to the surface currents. For example, a thin slot in the z direction can be placed in the center of the top wall of a waveguide without
perturbing the TE10 , TE30 , or certain other modes (see Fig. P1.7.9.1). For
each of the following waveguide modes, dene all positions and orientations
where thin slots may be cut without signicantly perturbing the propagating
mode: (i) TE10
(ii) TE11
(iii) TM21
P1.7.10
Dissipation occurs when a cavity is lled with conducting media. Since the
resonance conditions restrict the values of kx , ky , and kz , the eld inside the
resonator will attenuate in time. We write the eld components for the TM110
mode as
ny
mx
sin
(R cos R t I sin R t) eI t
a
b
n
mx
ny
Hx = A sin
cos
sin R t eI t
b
a
b
mx
ny
m
A cos
sin
sin R t eI t
Hy =
a
a
b
Ez = 0 A sin
Determine the rate of attenuation in time and how much R is changed from its
value when there is no dissipation due to conductivity.
1.8
135
Constitutive Relations
The Maxwell equations are fundamental laws governing the behavior of electromagnetic elds in free space and in media. They are given by
D (r, t) = J (r, t)
t
E (r, t) +
B (r, t) = 0
t
D (r, t) = (r, t)
H (r, t)
B (r, t) = 0
(1.8.1)
(1.8.2)
(1.8.3)
(1.8.4)
Taking the divergence of (1.8.1) and introducing (1.8.3), we nd that

J (r, t) +
(r, t) = 0
t
(1.8.5)
This is the conservation law for electric charge and current densities. Regarding (1.8.5) as a fundamental equation, we can use it to derive (1.8.3)
by taking the divergence of (1.8.1). Equation (1.8.4) can also be derived
by taking the divergence of (1.8.2) which gives ( B (r, t))/t = 0 or
that B (r, t) is a constant independent of time. Such a constant, if not
zero, then implies the existence of magnetic monopoles similar to free electric charges. Since magnetic monopoles have not been found to exist, this
constant must be zero and we arrive at (1.8.4).
We have so far made no reference to the various material properties that
provide connections to other disciplines of physics, such as plasma physics,
continuum mechanics, solid-state physics, uid dynamics, statistical physics,
thermodynamics, biophysics, etc., all of which interact in one way or another
with electromagnetic elds. It is time to state how we are going to account
for this vast outside world. From the electromagnetic wave point of view,
we shall be interested in how electromagnetic elds behave in the presence
of media, whether the wave is diracted, refracted, or scattered. Whatever
happens to a medium, whether it is moved or deformed, is of secondary
interest. Thus we shall characterize material media by the so-called constitutive relations that can be classied according to the various properties of
the media.
The necessity of using constitutive relations to supplement the Maxwell equations is clear from the following mathematical observations. In
most problems we shall assume that sources of electromagnetic elds are
136
1. Fundamentals
given. Thus J and are known and they satisfy the conservation equation
(1.8.5). Let us examine the Maxwell equations and see if there are enough
equations for the number of unknown quantities. There are a total of 12
scalar unknowns for the four eld vectors E, H, B, and D . As we have
learned, (1.8.3) and (1.8.4) are not independent equations; they can be derived from (1.8.1), (1.8.2), and (1.8.5). The independent equations are (1.8.1)
and (1.8.2), which constitute six scalar equations. Thus we need six more
scalar equations. These are the constitutive relations.
The constitutive relations for an isotropic medium can be written simply
as
D = E
B = H
where = permittivity
where = permeability
(1.8.6a)
(1.8.6b)
By isotropy we mean that the eld vector E is parallel to D and the eld
vector H is parallel to B. In free space void of any matter, = o and
= o ,
henry/meter
o = 4 107
o 8.85 1012
farad/meter
Inside a material medium, the permittivity is determined by the electrical

properties of the medium and the permeability by the magnetic properties
of the medium.
A dielectric material can be described by a free-space part and a part
that is due to the material alone. The material part can be characterized by
a polarization vector P such that
D = E = o E + P
(1.8.7)
The polarization P symbolizes the electric dipole moment per unit volume
of the dielectric material. In the presence of an external electric eld, the
polarization vector may be caused by induced dipole moments, alignment of
the permanent dipole moments of the medium, or migration of ionic charges.
A magnetic material can also be described by a free-space part and a
part characterized by a magnetization vector M such that
B = H = o H + o M
(1.8.8)
A medium is diamagnetic if < o and paramagnetic if > o . Diamagnetism is caused by induced magnetic moments that tend to oppose
137
the externally applied magnetic eld. Paramagnetism is due to alignment

of magnetic moments. When placed in an inhomogeneous magnetic eld,
a diamagnetic material tends to move toward regions of weaker magnetic
eld, and a paramagnetic material toward regions of stronger magnetic eld.
Ferromagnetism and antiferromagnetism are highly nonlinear eects. Ferromagnetic substances are characterized by spontaneous magnetization below
the Curie temperature. The medium also depends on the history of applied
elds, and in many instances the magnetization curve forms a hysteresis
loop. In an antiferromagnetic material, the spins form sub-lattices that become spontaneously magnetized in an antiparallel arrangement below the
Neel temperature.
A conducting material is characterized by Ohms law
J c = E
(1.8.9)
which relates conduction current J c to the electric eld E by the conductivity .

A. Anisotropic and Bianisotropic Media
The constitutive relations for anisotropic media are usually written as
D = E
B =H
where = permittivity tensor

where = permeability tensor
(1.8.10a)
(1.8.10b)
The eld vector E is no longer parallel to D, and the eld vector H is no

longer parallel to B. A medium is electrically anisotropic if it is described
by the permittivity tensor and a scalar permeability , and magnetically
anisotropic if it is described by the permeability tensor and a scalar permittivity . Note that a medium can be both electrically and magnetically
anisotropic as described by both and in (1.8.10).
Crystals are described in general by symmetric permittivity tensors.
There always exists a coordinate transformation that transforms a symmetric
matrix into a diagonal matrix. In this coordinate system, called the principal
system,
0
x 0
(1.8.11)
y 0
=0
0
0 z
The three coordinate axes are referred to as the principal axes of the crystal.
For cubic crystals, x = y = z and they are isotropic. In tetragonal,
138
1. Fundamentals
hexagonal, and rhombohedral crystals, two of the three parameters are equal.
Such crystals are uniaxial. Here there is a two-dimensional degeneracy; the
principal axis that exhibits this anisotropy is called the optic axis. For a
uniaxial crystal with
0 0
(1.8.12)
= 0 0
0 0 z
the z axis is the optic axis. The crystal is positive uniaxial if z > ; it
is negative uniaxial if z < . In orthorhombic, monoclinic, and triclinic
crystals, all three crystallographic axes are unequal. We have x = y = z ,
and the medium is biaxial.
For isotropic or anisotropic media, the constitutive relations relate the
two electric eld vectors and the two magnetic eld vectors by either a scalar
or a tensor. Such media become polarized when placed in an electric eld
and become magnetized when placed in a magnetic eld. A bianisotropic
medium provides the cross-coupling between the electric and magnetic elds.
The constitutive relations for a bianisotropic medium can be written as
D = E+H
(1.8.13a)
B = E+H
(1.8.13b)
When placed in an electric or a magnetic eld, a bianisotropic medium becomes both polarized and magnetized.
Magnetoelectric materials, theoretically predicted by Dzyaloshinskii
[1959] and by Landau and Lifshitz [1957], were observed experimentally in
1960 by Astrov, in antiferromagnetic chromium oxide. The constitutive relations that Dzyaloshinskii proposed for chromium oxide have the following
form:
0 0
0 0
D = 0 0 E + 0 0 H
(1.8.14a)
0 0 z
0 0 z
0 0
0 0
B = 0 0 E + 0 0 H
(1.8.14b)
0 0 z
0 0 z
It was then shown by Indenbom [1960] and by Birss [1963] that 58 magnetic
crystal classes can exhibit the magnetoelectric eect. Rado [1964] proved
that the eect is not restricted to antiferromagnetics; ferromagnetic gallium
iron oxide is also magnetoelectric.
139
In 1948, the gyrator was introduced by Tellegen as a new element, in addition to the resistor, the capacitor, the inductor, and the ideal transformer,
for describing a network. To realize his new network element, Tellegen conceived of a medium possessing constitutive relations of the form
D = E + H
B = E + H
(1.8.15a)
(1.8.15b)
where 2 / is nearly equal to 1. Tellegen considered that the model of the

medium had elements possessing permanent electric and magnetic dipoles
parallel or antiparallel to each other, so that an applied electric eld that
aligns the electric dipoles simultaneously aligns the magnetic dipoles; and
a magnetic eld that aligns the magnetic dipoles simultaneously aligns the
electric dipoles. Tellegen also wrote general constitutive relations (1.8.13)
and examined the symmetry properties by energy conservation.
Chiral media, which include many classes of sugar solutions, amino acids,
DNA, and natural substances have the following constitutive relations
H
(1.8.16a)
t
E
B = H +
(1.8.16b)
t
where is the chiral parameter. Media characterized by the constitutive
relations (1.8.15) and (1.8.16) are biisotropic media.
Media in motion were the rst bianisotropic media to receive attention
in electromagnetic theory. In 1888, R
oentgen discovered that a moving dielectric becomes magnetized when it is placed in an electric eld. In 1905,
Wilson showed that a moving dielectric in a uniform magnetic eld becomes
electrically polarized. Almost any medium becomes bianisotropic when it is
in motion. In Chapter 7, we shall derive constitutive relations for uniformly
moving media using the Lorentz transformation of eld vectors.
The bianisotropic description of material has fundamental importance
from the point of view of relativity. The principle of relativity postulates that
all physical laws of nature must be characterized by mathematical equations
that are form-invariant from one observer to the other. For electromagnetic
theory, the Maxwell equations are form-invariant with respect to all observers, although the numerical values of the eld quantities may vary from
one observer to another. The constitutive relations are form-invariant when
they are written in bianisotropic form. In Chapter 7 we shall treat special
relativity and electromagnetic theory in detail.
D = E
140
1. Fundamentals
Topic 1.8A Constitutive Matrices

Constitutive relations in the most general form can be written as
cD = P E + L cB
H = M E + Q cB
(1.8A.1a)
(1.8A.1b)
where c = 3 108 m/s is the velocity of light in vacuum, and P , Q, L, and

M are all 3 3 matrices. Their elements are called constitutive parameters.
In the denition of the constitutive relations, the constitutive matrices L
and M relate electric and magnetic elds. When L and M are not identically zero, the medium is bianisotropic. When there is no coupling between
electric and magnetic elds, L = M = 0 and the medium is anisotropic. For
an anisotropic medium, if P = c I and Q = (1/c)I with I denoting the
33 unit matrix, the medium is isotropic. The reason that we write constitutive relations in the present form is based on relativistic considerations. First,
the elds E and cB form a single tensor in four-dimensional space, and so
do cD and H. Second, constitutive relations written in the form (1.8A.1)
are Lorentz-covariant. These aspects will be discussed in Chapter 7.
Equation (1.8A.1) can be rewritten in the form

E
cD
=C
cB
H
(1.8A.2a)
and C is a 6 6 constitutive matrix:

P
C=
M
L
Q

(1.8A.2b)
which has the dimension of admittance.

The constitutive matrix C may be a function of space-time coordinates,
thermodynamical and continuum-mechanical variables, or electromagnetic
eld strengths. According to the functional dependence of C, we can classify
the various media as (i) inhomogeneous if C is a function of space coordinates, (ii) nonstationary if C is a function of time, (iii) time-dispersive if
C contains time derivatives, (iv) spatially dispersive if C contains spatial
derivatives, (v) nonlinear if C depends on the electromagnetic eld, and so
141
forth. In the general case C may be a function of integro-dierential operators and coupled to fundamental equations of other physical disciplines.
We have dened constitutive relations by expressing D and H in terms
of E and B. We may also express constitutive relations in the form of D
and B as a function of E and H :

D
E
= C EH
(1.8A.3a)
B
H
where in view of (1.8.13) and (1.8A.1),

1
1 P LQ M

=
C EH =
1
c

Q M
1
LQ
(1.8A.3b)
Here C EH is the constitutive matrix under E H representation. To express

E and H in terms of B and D, we write

E
D
= C DB
(1.8A.4a)
H
B
where

C DB

1
P
=c
1
M P
QM P
(1.8A.4b)
Here C DB is the constitutive matrix under D B representation. The other

possible construction for expressing E and B in terms of H and D is not
shown because it will not be needed in later developments.
We can show that for linear bianisotropic media with symmetric constitutive matrix C EH , Poyntings theorem becomes
W
+ S = p
t
(1.8A.5)
where
1
W = (D E + B H) = total stored energy density
(1.8A.6)
2
S = E H = Poyntings power density vector (1.8A.7)
p = J E = externally supplied power density (1.8A.8)
This is a general form expressing conservation of energy.
142
1. Fundamentals
Exercise 1.8A.1 Derive (1.3.10) for linear, homogeneous, bianisotropic media

with symmetric constitutive matrix C EH .
Problems
P1.8.1
The constitutive relation D = E can also be expressed in terms of a freespace part o E and a polarization vector P characterizing the properties of the
material. We write
D = o E + P
In the case of induced dipole moments, the polarization P is proportional to the
polarizability per unit volume N , where N is the number of dipoles per unit
volume, and is the polarizability for each dipole
P = N E loc
The local electric eld E loc at the place of the induced dipole comprises the applied
eld E and the eld caused by the surrounding dipoles. Under the quasi-static
approximation, the local electrical eld is shown to be
E loc = E +
P
3 o
Prove that

1 + (2N /3 o )
=
o
1 (N /3 o )
This is the well-known Clausius-Mossotti (or Lorentz-Lorenz formula).
P1.8.2
Similar to the expression of the constitutive relation D = E = o E + P , the

constitutive relation B = H can also be represented in terms of a free-space
part o H and a magnetization vector M such that
B = o H + o M
Notice that while P has the same dimension as D , M has the same dimension
as H .
In the case of media possessing permanent moments, the polarization P and
the magnetization M are given classically by the Langevin equation
L(x) = coth x
1
x
143
For a paramagnetic material with magnetic moments N m,
M = N mL
mH
kT
where k = 1.381023 joule/kelvin is Boltzmanns constant, and T is the absolute

temperature in kelvins. Show that in the low-eld limit, since mH kT , the
medium is linear.
P1.8.3
For each of the following constitutive relations, state whether the given medium
is
(1)
(2)
(3)
(4)
Isotropic/anisotropic/bianisotropic,
Linear/nonlinear,
Spatially/temporally dispersive,
Homogeneous/inhomogeneous.
(a) Cholesteric liquid crystals can be modeled by a spiral structure with constitutive relations given by
(1 + cos Kz)
sin Kz
D=

sin
Kz
(1
cos Kz)
E
z
where the spiral direction is along the z axis.

(b) In view of the optical activities in quartz crystals, the constitutive relation for
a quartz crystal is phenomenologically described as
Ej = ij Di +
Hj =
1
Gij Bi
o o
t
1
1
Bj
Gij Di
o
o o
t
(c) When a magnetic eld B 0 is applied to a conductor carrying a current, an electric eld E is developed. This is called the Hall eect. Assuming the conduction
carrier drifts with a mean velocity v proportional to RE, the constitutive
relation that takes care of the Hall eect is given by
J = E + RE B 0
where is the conductivity and R is the Hall coecient. For copper,

6.7 107 mho/m and R 5.5 1011 m3 /coul .
(d) The phenomenon of natural optical activity can be explained with the use of
the constitutive relation [Landau and Lifshitz, 1957]
Di = ij Ej + ijk
Ej
xk
144
1. Fundamentals
where ij and ijk are functions of frequency and ijk = jik .

(e) The phenomenon of pyroelectricity in a crystal is observed when it is heated.
The constitutive relation for a pyroelectric material can be written as
D = D0 + E
where a spontaneous term D0 exists even in the absence of an external eld.
(f) The phenomenon in which dipole moments are induced in a crystal by mechanical stress is called piezoelectricity. A piezoelectric material is characterized by
a piezoelectric tensor i,kl = i,lk such that
Di = D0i + ik Ek + i,kl skl
where skl is the stress tensor to second order in electric elds. All pyroelectric
media are also piezoelectric.
(g) An isotropic dielectric can exhibit the Kerr eect when placed in an electric
eld. In this case the permittivity can be written as
ij = ij + Ei Ej
where is the unperturbed permittivity. The principal axis of ij coincides
with the electric eld.
(h) In an electrooptical material that exhibits Pockels eect the constitutive relation can be written as
Di = ij Ej + ijk Ej Ek
where ijk = jik is a third-rank tensor symmetrical in i and j, and therefore
has 18 independent elements.
P1.8.4
The constitutive relation for superconductors in weak magnetic elds can be

macroscopically characterized by the rst London equation
J sup
= E
t
and the second London equation
J sup = 1 B
where J sup stands for the superconducting current, = ns q 2 /m and 1 with

ns , m and q denoting, respectively, the number density, the eective mass, and the
charge of the Cooper pairs responsible for the superconductivity in a charged Boson
uid model.
145
(a) From the rst London equation, derive an equation for B = B/t by using
the static Maxwell equation H = J sup without the displacement current.
Show that
2 B = o B
and deduce that the penetration depth of the time-varying magnetic eld is
(o )1/2 .
(b) From the second London equation, derive an equation for B from H =
J sup and show that the penetration depth of the stationary magnetic eld is
(o 1 )1/2 . Thus for static elds, both the current and the magnetic eld are
conned to a thin layer of the order of the penetration depth which is very
small. The exclusion of a static magnetic eld from a superconductor is known
as the Meissner eect experimentally discovered in 1933.
146
1.9
1. Fundamentals
Boundary Conditions
The Maxwell equations have been written in dierential form. They must
be supplemented with boundary conditions and initial conditions wherever
derivatives do not exist. The boundary conditions can be derived from the
integral form of the Maxwell equations. The eld vectors E, B, D , and
H are assumed to be nite but may be discontinuous across the boundary.
The volume current and charge densities J and , however, may be innite,
such as on the surface of a perfect conductor, where we can dene the surface
current density J s = J in the limit as 0 and J , and dene
the surface charge density s = in the limit as 0 and .
The surface current density has dimension ampere/m and the surface charge
density has dimension coulomb/m2.
For a stationary boundary separating regions 1 and 2, we let the surface
normal n
point from region 2 to region 1. The boundary conditions are as
follows:
n
(E 1 E 2 ) = 0
n
(H 1 H 2 ) = J s
n
(B 1 B 2 ) = 0
n
(D1 D2 ) = s
(1.9.1)
(1.9.2)
(1.9.3)
(1.9.4)
where subscripts 1 and 2 denote elds in regions 1 and 2, respectively. Essentially the boundary conditions state that the tangential components of
E and the normal components of B are continuous across the boundary;
the discontinuity of the tangential components of H is equal to the surface
current density J s ; and the discontinuity of the normal components of D
is equal to the surface charge density s .
Topic 1.9A Derivation of Boundary Conditions

First we consider the integration of a vector eld A over a volume V enclosed by a surface S with surface normal s . The following formulas are
useful:
(1.9A.1a)
dV A = dS s A
(1.9A.1b)
dV A = dS s A
147
s = n
region 1
region 2
s
Figure 1.9A.1 Pillbox for derivation of boundary conditions.
where (1.9A.1a) is the familiar Gauss theorem which relates integration of

the divergence of the vector eld A over the volume V to the integration of
the eld over the surface S enclosing V . Equation (1.9A.1b) is derived from
(1.9A.1a) by noting that (C A) = C A where C is a constant
vector independent of position. Applying the Gauss theorem (1.9A.1a) to
(C A) , we obtain
dV A = dS s C A = C dS s A
This is seen to be (1.9A.1b) dot-multiplied by C on both sides. Letting C

be an arbitrary vector, the result is then (1.9A.1b) .
Now consider an interface separating regions 1 and 2 [Fig. 1.9A.1]. Assume a small pillbox volume across the interface. Integrating (1.1.1)(1.1.4)
over the volume and applying (1.9A.1), we obtain
dS s E +
dS s H
B=0
t
dV D =
dV J
t
dV
dS s B = 0
dV
dS s D =
(1.9A.2)
(1.9A.3)
(1.9A.4)
(1.9A.5)
148
1. Fundamentals
These are the Maxwell equations in integral form, which will be used to
derive boundary conditions for both stationary and moving boundaries.
If we assume that the boundary surface is not in motion, then for the
terms involving partial derivatives with time, /t can be moved to the
outside of the integral. Since the integration is over the volume, the result is
a function of time only, and the partial derivatives become ordinary derivatives. Therefore, for stationary boundary surfaces, the Maxwell equations in
integral form become
d
(1.9A.6)
dV B
dS s E =
dt
d
dS s H =
dV J
(1.9A.7)
dV D +
dt
(1.9A.8)
dS s B = 0
dV
(1.9A.9)
dS s D =
Now we let the volume of the pillbox approach zero in such a manner that
the thickness of the ribbon side, , goes to zero before the top and bottom
areas a shrink to a point. We dispose of terms of the order of .
First we see that the terms involving time derivatives in (1.9A.6) and
(1.9A.7) drop out because they are proportional to . We then consider
the right-hand sides of (1.9A.7) and (1.9A.9) which become aJ and a ,
respectively. If J and are nite, both terms will be zero because they
are proportional to . When there are surface charges and currents at the
boundary, the right-hand sides of (1.9A.7) and (1.9A.9) become aJ s and
as . We then see that the surface integral terms involving cross and dot
products will be dropped except when s is in the directions n
or
n.
After canceling a on both sides of the equations, we obtain from (1.9A.6)
(1.9A.9) the boundary conditions (1.9.1)(1.9.4).
Topic 1.9B Boundary Conditions for Moving Boundaries

The boundary conditions as stated in (1.9.1)(1.9.4) are for stationary
boundaries. When the boundary surface is moving, the partial time derivatives no longer commute with the volume integrals in (1.9A.2)(1.9A.3). To
derive the boundary conditions at moving boundaries, we let the pillbox
149
move with the boundary surface. In accordance with kinematic theory, for a
moving volume with velocity v ,

1
dV A(t + t)
dV A(t)
dV A = lim
t0 t
t+t
t

1
A
= lim
dV + dS (s vt) A(t) +
dV A(t)
t
t0 t
t
t
t
A + dS (s v) A
(1.9B.1)
=
dV
t
d
dt
where A denotes any vector eld. The surface integration term accounts for
the motion of the boundary. We see that for moving boundaries, the Maxwell
equations in integral form (1.9A.2)(1.9A.5) become

d
dV B
dS s E (s v)B =
dt
V

d
dS s H + (s v)D =
dV D +
dV J
dt
V
V

dS s B = 0

dV
dS s D =
(1.9B.2)
(1.9B.3)
(1.9B.4)
(1.9B.5)
We shrink the pillbox in the same manner as before such that terms of the
order of are disposed of. The boundary conditions now become

n v) B 1 B 2 = 0
n
E 1 E 2 (

n
H 1 H 2 + (
n v) D1 D2 = J s

n
B1 B2 = 0

n
D1 D2 = s
(1.9B.6)
(1.9B.7)
(1.9B.8)
(1.9B.9)
They reduce to (1.9.1)(1.9.4) when v = 0 . It is important to note that they

also reduce to (1.9.1)(1.9.4) when n
v = 0 . Therefore, when the velocity is
parallel to the interface, the boundary conditions of such a moving boundary
are identical to those of a stationary boundary.
150
1. Fundamentals
Problems
P1.9.1
Consider at time t a surface S1 bounded by the contour C1 [Fig. P1.9.1.1].

Let v be the instantaneous velocity of the element dS of the surface. The surface
S1 together with the contour C1 may change shape with time, as v need not be a
constant for all elements of S1 . At time t + t, S1 and C1 become S2 and C2
[Fig. P1.9.1.1].
C2
dl
vt
S1
S2
C1
Figure P1.9.1.1
(a) Applying the divergence theorem to the volume bounded by S 1 , S 2 , and the
dierential ribbon area dl vt between C1 and C2
(dS vt)( A) =
dS 2 A
S2

dS 1 A +
(dl vt) A
S1
show that the total time derivative of a vector eld A integrated over the
surface is
d
dt
1
dS A = lim
t0 t
S
dS
dS
=
S
dS 2 A(r, t + t)
S2
A(r, t) +
t
dS 1 A(r, t)
S1
dS v A
S
A
+ (A v) + v A
t
(dl v) A
The rate at which the ux of A through S changes is seen to depend on

three processes. The rst term is due to the time rate of change of A for a
151
stationary contour. The second term accounts for the contribution due to the
ux crossing the surface generated by the motion of the contour C . The last
term arises when the surface moves through the regions of the sources for the
ux A. When A is identied as the magnetic induction B, the last term will
be zero.
(b) Identifying A with B and D eld vectors and employing Faradays law and
Ampères law to obtain
dl E + v B =
d
dt
d
dt
dl H v D =
dS s B
dS s D +
dS s J v
Applying the integrals to a ribbon-like area as shown in Fig. E1.1.8.1, which

is now moving with the boundary surface, determine the relationship between
E 1 , H 1 in region 1 and E 2 , H 2 in region 2.
P1.9.2
The jump conditions for moving boundaries have been derived by assuming that
the pillbox moves with the boundary. If the pillbox is stationary while the boundary
is moving, then v = 0 in (1.9B.2) and (1.9B.3) as v refers to the velocity of the
pillbox:
dS
sE =
d
dt
d
dS
sH =
dt
t + t
t
dV B
dV D +
dV J
vt
0
Figure P1.9.2.1
Refer to Fig. P1.9.2.1. The boundary surface has been moved a distance vt
from time t to time t + t. Show that for a vector eld A
dV A = a
dz A = A2 a0 + A1 a( 0 )
0
dV A = a
t+t
dz A + a
0
and that
0 +vt
dz A + a
0
dz A
0 +vt
= A2 a(0 + vt) + A1 a( 0 vt)
152
1. Fundamentals
dV A
t+t
dV A = A2 avt A1 avt
t
where a is the area of the top side or the bottom side of the pillbox and A1 and
A2 denote the A eld in medium 1 and medium 2, respectively. Identifying A with
B and D , determine the relationship between E 1 , H 1 in region 1 and E 2 , H 2 in
region 2.
Answers
153
Answers
P1.1.1
For all eld components, we have
Hz
Hy = Dx + Jx
y
z
t
Hx
Hz = Dy + Jy
z
x
t
Hy
Hx = Dz + Jz
x
y
t
(A1.1.1.1a)
(A1.1.1.1b)
(A1.1.1.1c)
for Ampères law,
Ez
Ey = Bx
y
z
t
Ex
Ez = By
z
x
t
Ey
Ex = Bz
x
y
t
(A1.1.1.2a)
(A1.1.1.2b)
(A1.1.1.2c)
for Faradays law,
Dx +
Dy +
Dz =
x
y
z
(A1.1.1.3)
Bx +
By +
Bz = 0
x
y
z
(A1.1.1.4)
Jx +
Jy +
Jz =
x
y
z
t
(A1.1.1.5)
for Coulombs law
for Gauss law, and
for the continuity law. The continuity equation (A1.1.1.5) can be derived from the
sum of (A1.1.1.1a)/x , (A1.1.1.1b)/y , (A1.1.1.1c)/z , and making use of
(A1.1.1.3). We see that giving the continuity law, Coulombs law can be derived
from Ampères law. Likewise, Gauss law can be derived from Faradays law. Thus
(A1.1.1.3) and (A1.1.1.4) are not independent scalar equations, they can be derived
from (A1.1.1.1) and (A1.1.1.2).
P1.1.2
k0 = 100 K0 = 200 m1 .
Wavelength = 2
k0 = 0.01 m.
154
1. Fundamentals
Frequency f = c = 30 GHz.
For = 632.8 nm, k = 1 = 1.58 106 Ko .
For f = 2.4 GHz, k =
f
c
2.4109 Hz
3108 m/s
= 8 Ko .
P1.1.3
E 1 and E 5 qualify as electromagnetic waves.

E 2 and E 4 violate Gauss theorem E = 0.
H 3 violates H = 0 at x = z.
P1.1.4
(a-i) 60 Hz: = c/f = 5 106 (m)

(a-ii) AM radio (5351605 kHz): = 186.9 560.8 (m)
(a-iii) FM radio (88108 MHz): = 2.778 3.409 (m)
(a-iv) L- band (12 GHz): = 0.15 0.3 (m)
(a-v) Visible light ( 1014 Hz): = 3 106 (m)
(a-vi) X-rays ( 1018 Hz): = 3 1010 (m)
(b-i) 1 km: f = c/ = 3 105 (Hz)
(b-ii) 1 m: f = 3 108 (Hz)
(b-iii) 1 mm: f = 3 1011 (Hz)
(b-iv) 1 m: f = 3 1014 (Hz)
(b-v) 1
A: f = 3 1018 (Hz)
(c-i) 1 km: k = 2/ = Ko / = 103 Ko
(c-ii) 1 m: k = 1Ko
(c-iii) 1 mm: k = 103 Ko (c-iv) 1 m: k = 106 Ko (c-v) 1
A: k = 1010 Ko
P1.1.5
y
z
E = /x /y /z
Ex
Ey
Ez
y
/x
/y
=

Ey
Ez
Ex
Ez
y
z
z x
=
x
+
y

Ey
Ex
Ez
+
+
x
y
z
Ey
Ex
Ez
+
+
x
y
z
Ey
Ez
Ex
+
+
x
y
z
= E 2 E
z
/z
Ey
x
Ex
y
2
2
2
+ 2+ 2
2
x
y
z
2
2
2
+
+
y 2
y 2
z 2
2
2
2
+ 2+ 2
2
y
y
z

Ex x

Ey y

Ez z
Answers
155
y
z
EH =
Ey Ez
Hy Hz
=
(Ey Hz Ez Hy ) +
(Ez Hx Ex Hz ) +
(Ex Hy Ey Hx )
x
y
z

= Hx
Ez
Ey + Hz
Ey
Ex + Hy
Ex
Ez
y
z
x
y
z
x
Ex
Ex
Hx
Hz
Hy
y
z
Ey
=H E E H
A =
x
=0

() =
y
/y
Ay
Ay
Az
y
z
/x
/x

=
/x
Ax
y
/y
/y
2
2
yz
yz
Hx
Hz
z
x
z
/z
Az
+
z
/z
/z
x
+
Ez
Hy
Hx
x
y

Ax
Az
z
x

+
Ay
Ax
x
y
2
2
xz
xz

y +
2
2
xy xy

z
=0
P1.1.6
A + B + C = 0 and A + B C = 2C.
P1.1.7
r=
7, = /3, = /4 ; and =
3, = /4, z = 2.
P1.1.8
c = x
0.6 + z0.8.
P1.1.10
We apply the curl theorem to a small pill-box volume on the x-y plane
[Fig. A1.1.10.1], which has an area A and an innitesimal thickness z . We let
z 0 faster than A 0 , such that terms involving z can be neglected:
dV H A
z (H z>0 H z<0 )
156
1. Fundamentals
z
z
y
A
z
x
Figure A1.1.10.1 Small pill-box volume.
Such results are useful in the derivation of boundary conditions for the Maxwell
equations. Integrating Ampères law H = D/t + J over the pill-box volume,
we have A
z (H z>0 H z<0 ) = Az D
t + AzJ
The rst term on the right-hand side is neglected because physically D/t
is nite. However if J is innite in the pill-box then zJ = J s is nite, where
J s = z (H z>0 H z<0 ) . We call the J s surface current.
P1.1.11
If the surface integral of H is carried out over a closed surface, there will
be no external contour enclosing the surface and the result will be zero.
dS ( H) = 0
(A1.1.11.1)
This scalar equation should not be confused with the curl theorem we obtained
previously for the curl integrated over a volume V enclosed by a surface S in
(1.1.50), which was a vector relation.
P1.2.1
(a) At z = z0 , Ex = 2 sin(kz0 t) , and Ey = 12 cos(kz0 t) 12 sin(kz0 t)
E2
Since 2x 2Ex Ey + 2Ey2 = 1 , the wave is elliptically polarized.
(b) E = 12 [
x cos(kz t) + y sin(kz t)] + 12 [
x cos(kz t) y sin(kz t)]
(c) E = x
cos(kz t + /4) + y cos(kz t /4) . This is the superposition of
two linearly polarized waves.
P1.2.3
The wave has wavelength 1 cm, and is right-hand circularly polarized, the helix
is left-handed, and its pitch is 1 cm.
P1.2.4
Looking at = 0 , the sky is unpolarized, looking at the zenith ( = 90 ) the

sky is linearly polarized, looking at other parts of the sky, it is partially linearly
polarized.
Answers
157
P1.3.1
mv 2 /R = Ze2 /4 R2 gives R = 4 n2 h2 /Zme2 0.52n2 1010 m for

Z = 1.
P1.3.2
(a) T = D/C = 500sec = 8.33 min

2
(b) Pr = 1.5kW/m (6.4 106 )2 m2 = 1.93 1014 kW

(c) S = P
Hz) W (bandwith) = 1011 Wm2
(power density per
5
E = 2S = 8.68 10 volt/m
P1.3.3
(a) When J = 0 and for static elds

(E H) = 0
Thus there is no net power ow at any given point.
(b) The total Poynting power ow into the resistance is
P = 2alEH =
a2 lJ 2
The source of energy for this problem is from the source supplying the current
J.
P1.3.4
For i =
as
xi
, free-space Maxwells equations can be written in index notation
D
t
Di
= ijk j Hk
t
(1.4.4.1)
B
t
Bi
= ijk j Ek
t
(1.4.4.2)
H =
E =
D =0
i Di = 0
(1.4.4.3)
B =0
i Bi = 0
(1.4.4.4)
The ith component of the time derivative of D B is

Dj
Bk
(D B)i = (ijk Dj Bk ) = ijk Dj

+ ijk
Bk
t
t
t
t
(1.4.4.5)
158
1. Fundamentals
Use (1.4.4.1), (1.4.4.2), and the identity ijk = kij in (1.4.4.5) to get
(D B)i = ijk kpq Dj p Eq + ijk jmn Bk m Hn

t
= ijk pqk Dj (p Eq ) + kij mnj Bk m Hn
(1.4.4.6)
Use the identity ijk rsk = ir js is jr , equation (1.4.4.6) becomes
(D B)i = (ip jq iq jp )Dj (p Eq ) + (km in kn im )Bk (m Hn ) (1.4.4.7)

t
Successively sum over p, q, m, and n to obtain
(D B)i = Dj j Ei Dj i Ej + Bk k Hi Bk i Hk
t
(1.4.4.8)
Identify the terms in the right hand side of equation (1.4.4.8):

1
1
Dj i Ej = 0 Ej i Ej = i ( 0 E E) = i ( D E)
2
2
(1.4.4.9)
1
1
Bk i Hk = 0 Hk i Hk = i ( 0 H H) = i ( B H)
2
2
(1.4.4.10)
Dj j Ei = Dj j Ei + Ei j Dj = j (Dj Ei ) = (DEi )
(1.4.4.11)
Bk k Hi = Bk k Hi + Hi k Bk = k (Bk Hi ) = (BHi )
(1.4.4.12)
Equations (1.4.4.3) and (1.4.4.4) have been used in (1.4.4.11) and (1.4.4.12) respectively. Substituting (1.4.4.91.4.4.11) in (1.4.4.8) yields
1
1
(D B)i = i ( D E + B H) + (DEi + BHi )
t
2
2
(1.4.4.13)
Expression (1.4.4.13) is identical to
(D B) + W I DE BH
t
P1.3.6
(a) = q/2m
(b) = B0 is known as the Larmor frequency.
P1.4.1
(a)

S = r k
24o

q5 2
4r
sin2

=0
(1.4.4.14)
Answers
159
k 3
d2r sin
2 o
(b) P =
0
2P
Io2
2k3
34o
(c) Rrad =
(d) P =
8k3
34
q5 o2
=
4

5 2
4
2
3o
q?
4r
2
2
sin =
4k 3
3 o
q?
4
2
(Eo r) = 781.25 W
P1.4.2
H = t
.
1
E = 40 ( ).
P1.4.4
Q
(a) Set
Q kq?
=
equal 0 ,
1
cos(kr t)
kr

k
r
1
r
r
2
+
sin(kr t)
+
+ sin k cos(kr t)
cos(kr t)
kr
kr2
2 sin cos sin(kr t)
=0
we nd
1
2 sin cos sin(kr t) kr
cos(kr t)
r

=
2
k
sin k cos(kr t) + kr sin(kr t) + kr12 cos(kr t)
1
cos(kr t)
sin(kr t) + kr
2 cos
1
1
k sin (1 k2 r2 ) cos(kr t) kr
sin(kr t)
(b) Substitute kr = and = /4 ,

sin(t) 1 cos(t)
tan(t) 1
r
2
2
=
=
1
1
k (1 2 )( cos(t)) sin(t)
k (1 12 ) + 1 tan(t)
(c)
tan 1 =
2
r
2 1
|t=0 =
=
1
r
(1 2 )
1 2
tan 2 =
r
2 1
|t=/2 = 1 = 2
r
tan 1 tan 2 = 1,
they are not perpendicular
1 2 = 76
160
1. Fundamentals
P1.4.5
Let p1 = x
q? and p2 =
y q? , we nd
1
k
[p r]
sin(kr t) cos(kr t)
4r 1
kr
H=
1
k
[p2 r]
4r
kr
1
k2
1
[(p1 r) r) + 2
r(
r p1 )] 2 2 cos(kr wt) +
sin(kr t)
4 o r
k r
kr
E=
[(p1 r) r] cos(kr t)
1
k2
1
+
[(p2 r) r) + 2
r(
r p2 )] 2 2 sin(kr wt) +
cos(kr t)
4 o r
k r
kr
+ [(p2 r) r] sin(kr t)
On the x-y plane as k 1 , we have
1
k
[p1 r]
sin(kr t) cos(kr t)
H=
4r
kr
1
k
[p r]
4r 2
kr
k2
E=
[(p1 r) r] cos(kr t) + [(p2 r) r] sin(kr t)
4 o r
k2
=
sin cos(kr t) + cos sin(kr t)
4 o r
k2
=
sin(kr t )
4 o r
where we used the fact that r = = x
cos + y sin .
P1.4.6
b = 2 105 Tesla.
P1.5.1
(a) tan B =
St
S
(b) C = sin1
%
=
2
1
B = tan1
2
1
= sin1
1
2
= 30
P1.5.2
RT E =
1n
1+n
RT M =
n1
n+1
2
1
Answers
161
P1.5.3
(a) E i = Eo [kx cos + kz sin ] sin (kx x kz z t) = 0

TE
(b) TE wave: E i
= Eo y cos (kx x kz z t) .
TM
(c)
(d)
(e)
(f)
TM wave: E c = Eo (
x cos + z sin ) cos (kx x kz z t) .
Linear polarization.

E r = Eo
xRT M cos + yRT E + zRT M sin cos (kx x + kz z t)
For < c , both the reected and the transmitted waves are linearly polarized.
For > c , both TE and TM waves are totally reected. At total reection,
the total phase dierence between the TE and TM components of the reected
electric eld is
= T M T E = 2 T M T E = 2(tan1
&
tz
o tz
tan1
)
o kz
o kz
sin2 n2
sin2 n2
tan1
cos
cos

n
sin2 n2
sin2 n2
cos sin2 n2
cos
cos
=
=
n2 sin2 n2
sin2 n2
sin2
1+
cos
cos
n2
tan
= tan tan1
2
'
(g) For circularly polarized wave, = /2 , thus cos sin2 n2 /sin2 = 1,

or 12 sin4 7 sin2 + 1 = 0. The solutions are sin = 12 , 13 . Therefore
= 30 , 35.264 .
P1.5.4
B = tan1 n = tan1 1.46 = 55.59 , = 90 B = 34.41 .

P1.5.5
(a) The Brewster angle for t = 9 is B = tan1 t = tan1 9 = 71.57 .

(b) The dominant portion of the sun glares is TE polarized wave. The polaroid
glasses absorb the TE component of the incident light, thus the TM component
reaches the eyes after passing through the polaroid glasses.
P1.6.1
The critical angle C = sin1 44ot . The Brewster angle B = tan1 44ot . The
critical angle C is always larger than the Brewster angle B . It is impossible to
have total transmission and total reection at the same time.
P1.6.2
1
(a) E i k i = 0 3kx kz = 0 i = tan1 (kx /kz ) = tan
(1/ 3) = 30 .
(b) For kx = 1 (Ko ) , we get kz = 3kx = 3 (Ko ) k = kx2 + kz2 = 2 (Ko ),
and k = o 9 o = 3/c. So f = /2 = ck/(3 2) = 2 108 (Hz) and

= 2/k = 0.5 m.

(c) At the critical angle, sin c = sin 30 = t / 9 o t = 94 o .
162
1. Fundamentals
(d) If the totally reected wave is linearly

angle is the Brew polarized, the incident
1
2
t /9 o t = 9 o tan 30 = 3 o .
ster angle, thus i = 30 = tan
P1.6.3
(a)
(b)
(c)
(d)

Pr = 0.16Pi , so E r = 0.16 E i = 0.4 E i .
R = (n 1)/(n + 1) = 0.4 so n = 7/3 .

2
2
This problem is the TM wave case, so Pr /Pi = RTM = (11/38) .
The tilted angle is the Brewster angle, B = tan1 n = tan1 (7/3).
P1.6.4
(a) The reection coecient for a TE wave is

R=
1 p0t
1 + p0t
p0t =
0 ktz
t kz
For a perfect conductor |p0t | , so that R = 1 which corresponds to a

phase shift of .
(b) = 2 tan1 0tkktzI
, the phase shift of a wave reected from a perfect conz
ductor at x = d is + 2kx d . Then, if we want case (i) and (ii) to have an
identical phase shift, we should make
+ 2kx d = + 2n,
1
1
kx d = + (n + ), d =
+ (n + )
2
2
2kx
2 kx
P1.6.5
(a) B = tan1 4421 = tan1 13 = 30

(b)
(i) k is on the x-z plane, and E k = 0 . We need to nd a E vector that
is onthe x-z plane, which is given by the
E t and is in the direction of
z 3
x . Thus k t isin the direction of 3
z+x
. From that calculation,
we have t = tan1 ( 3) , as the angle is measured from the normal of the
surface. Thus, i = 30 , t = 60 .
(ii) To nd the polarization, we look at E t |x,z=0 and nd the magnitude and
phase of the polarization vectors.
E t |x=z=0
z x
3
E0
sin(t)
= y cos(t) + E0
2
2 2
= e1 E1 cos(t) + e2 E2 sin(t)
It is seen that E1 and E2 have the same magnitude and e1 e2 is in

the k direction.
At t = 0 , E t |x,z=0 (t = 0) is in the e1 direction.
Answers
163
At t = t , E t |x,z=0 (t = t) is in the e1
e2 direction.
E t |x,z=0 (t = 0) E t |x,z=0 (t = t) is in the k direction, i.e. LHCP.
So, transmitted eld is LHCP.
(iii) Reected eld is linearly polarized, because the TM component is totally
transmitted already.

0 cos(kx x + kz z t) + E0
(iv) E i = y 32E
2
2 2
2 6
sin(kx x + kz z t)
Er =
cos(kx x + kz z t)
(c)
(i) Normal component of ktI is 0. = 0
(ii) T = 1 + R, R = 1 (with no phase shift) E t is plane wave with amplitude
of 2E0 propagating in the z direction. Thus amplitude is constant in the
y direction
(iii) With no phase shift, kx = m/d
y 3E02
P1.6.6
R1 =
1p2
0tI
1+p2
0tI
T2 = R2 =
= cos 2 , T1 =
2p0tI
1+p2
0tI
2
1+p2
0tI
= 1 + cos 2 ,
= sin 2 .
P1.6.7
We let p0tI = 0 ktxI /t kx = tan , it follows that

R1 = cos 2
T1 = 2 cos2 = 1 + cos 2
T2 = R2 = sin 2
In region t , the transmitted TE eld components are
Ety = 2 cos ektxI x cos(kz z t + )
kz
Htx =
2 cos ektxI x cos(kz z t + )
t
ktxI
Htz =
2 cos ektxI x sin(kz z t + )
t
In region 0, the reected TE eld components are
Ery = cos(kx x + kz z t + 2)
kz
Hrx =
cos(kx x + kz z t + 2)
0
kx
Hrz =
cos(kx x + kz z t + 2)
0
The reected wave at total reection, gained a phase shift of 2 , which is the
Goos-Hanchen shift for TE waves.
164
1. Fundamentals
P1.7.1
d 4
o o
3010 2 310
(a) m <
= d
= 3.46 .
c =
3108
The possible guided modes are TMm ( m = 0, 1, 2, 3 ) and TEm ( m = 1, 2, 3 ).
(b) For TM2 mode,

Ho
2x
2
2x
E=
x
kz cos
cos (kz z t) + z sin
sin (kz z t)
o
d
d
d
9
(c) vp = kz = 3010
= 3.67 108 (m/s).
100
2/3
8 2
)
d
vg = dk
= koz4o = c kz = vcp = (310
= 2.45 108 (m/s).
3.67108
z
(d) Since 1 = 3 o > 0 , there is no total reection for any modes.
(e) TM3 wave can be totally transmitted.
P1.7.2
(a) In region z < 0, the possible modes are TE1 , TM0 TM1 . In region z > 0,
the possible modes
are TE1 TE2 ,
TM
0 TM1 TM2 .

1p0t
tz
755 = 0.4741.
(b) p0t = kk0z
= 55
R
=
=
7
1+p0t
7+ 55
(c) E = yEo sin
d x sin (kz z) , where kz =
2 o o
2
d
P1.7.3
(a) The radiation pattern of the Hertzian dipole on the x-y plane is a circle.
(b) Linearly polarized, perpendicular to the paper, TE wave.

2

2
2
2
(c) For m = 1 ,
2/o (/d) > 0 . For m = 2 ,
2/o (2/d) < 0 .
So o / 2 < d < 2o . The propagating mode is TE1 .

P1.7.4
c10
2
4.19
(a) fc10 =
3108
2
c
2
3108
2
6.55
= 22.9 (MHz) < f < fc01 =
c
2
= 35.8 (MHz)
(b) An AM radio operates in the range of 500 to 1600 (KHz) is below the cuto
frequency of the fundamental mode TE10 . Therefore AM signals can not be
received in the tunnel.
(c) FM signals operate in the range of 88.1 to 107.9 (MHz) can be received in
the tunnel.
P1.7.5
(a)
(b)
(c)
(d)
(e)
Ez = 0 , m = 1 and
n = 0 , the mode
is TE10 .
2
2
2
kx = kz = /a , k = kx + kz = a , f = c/ 2a = 5 GHz .
kz = /a , vp = /kz = 2c .
2
kc2 (/a) = 0 , fc = c/2a = 5/ 2 GHz .

2
2
2
2
k2 =
2/a = (m/a) + (n/b) + (p/d) , m = 1 , n = 0 , p = 1 and
d = a . The mode is TE101 .
P1.7.6
The cuto wavenumber is kc =
(m/a)2 + (n/b)2 and for a = 2b , fc =
Answers
c
2a
165
m2 + 4n2 . Therefore, the rst ve modes are TE 10 , TE 20 , TE 01 , TE 11 , and

TM 11 and the cuto frequencies are 15 GHz, 30 GHz, 30 GHz, 33.5 GHz , and
33.5 GHz respectively.
At a frequency of 20 GHz , k = /c = 400/3. Therefore the propagation
constants are
TE10 : kz = 100
(4/3)2 1 = 100
kz = i100 20/9

kz = i100 20/9

kz = i100 29/9
TE20 :
TE01 :
TE11 :
TM11 : kz = i100
7/9
evanescent modes
29/9
Cuto frequencies are 43 GHz for TE 21 and TM 21 modes, 45 GHz for TE 30

modes and 50 GHz for TE 31 and TM 31 modes.
At a frequency of 50 GHz , there will be 8 propagating modes TE 10 , TE 20 ,
TE 01 , TE 11 , TM 11 , TE 21 , TM 21 , TE 30 .
P1.7.8
(a)
(i) For TE10 mode, the cuto frequency is fc =
(ii) kz =
(iii) vp =
(b)
= 37.3 m
= 4 108 ms1
k2
kz
kx2
c
2 a
c
2(0.03)
= 5 GHz
4
10
s
(i) T = 2
= 3 10
0.01
(ii) vp = k = 21011 = 5 108 ms1
(iii) kz = vp = 30 m1
(iv) TE01mode
(v) b = k 2 kz2 = 2500 2 900 2 = 40 b = 2.5 cm
P1.7.9
c
n 2
n 2
10
2
2
(a) fc = 2
( m
( m
a ) + ( b ) = 1.5 10
4 ) + ( 2 ) Hz
fc < 15 GHz TE10 fc = 3.75 GHz; TE20 , T E01 fc = 7.5 GHz;
TE11 , TM11 fc = 8.4 GHz; TE21 , TM21 fc = 10.6 GHz;
TE30 fc = 11.25 GHz; TE31 , TM31 fc = 13.5 GHz
(b) Surface current can be calculated using J s = n
H , where n
is the unit
vector normal to the surface and points into the waveguide.
Top: J s =
yH
Bottom: J s = y H
Side: @ x = 0 , J s = x
H,
@ x = a , J s =
xH
(i) TE10
a
x
x
H = Ho [
x kz sin( ) cos(kz z t) z cos( ) sin(kz z t)]
a
a
166
1. Fundamentals
Top:
J s =
yH
a
x
x
= Ho [
z kz sin( ) cos(kz z t) + x
cos( ) sin(kz z t)]
a
a
Thus on top or bottom, at x =
fore slots can be placed at x =
a
2
a
2
, there is only current in z -direction. There, oriented in the z direction
Sides:
Js = x
H = yHo sin(kz z t)
There is only currents in y -direction, slots can be placed anywhere on
the side walls, oriented in the y -direction.
(ii) TE11
ky kz
kx kz
sin( x) cos( y) + y 2 cos( x) sin( y) cos(kz z t)

k2
a
b
k
a
b
H =H0 x
kx2 + ky2
cos( x) cos( y) sin(kz z t)

2
k
a
b
Top, y = b :
J s =
yH
=
z
kx2 + ky2
kx kz
x)
cos(k
sin(
z
t)
cos( x) sin(kz z t)
z
2
2
k
a
k
a
Slots can be placed at x = a/2 along the z -direction.

Sides:
Js = x
H
= z
kx2 + ky2
ky kz
sin( y) cos(kz z t) + y
cos( y) sin(kz z t)
2
k
b
k2
b
Slots can be placed at y = b/2 along the z -direction.

(iii) TM21
H = Ho
ky
2
kx
2
x
sin( x) cos( y) y cos( x) sin( y) cos(kz z t)
k
a
b
k
a
b
Top, y = b :
J s =
y H =
z
ky
2
sin( x)Ho cos(kz z t)
k
a
Answers
167
Slots can be placed anywhere along the z -direction.

Sides, x = 0 :
Js = x
H =
z
2
kx
sin( y)Ho cos(kz z t)
k
b
Slots can be placed anywhere along the z -direction.

P1.7.10
The rate of attenuation in time is I = 240 and R is decreased by 0 2 /4 0

from its value when there is no dissipation due to conductivity.
P1.8.1
Let E SD be the eld caused by the surrounding dipoles, we write

E loc = E + E SD
Where E is the eld in the dielectric in the absence of the hole, and E SD can be
calculated from the polarization charges. Let
P = n
P = Polarization surface charge at the spherical surface.
Using spherical coordinates, we have
P =
r P = P cos
Making electroquasistatic assumptions we write the electric eld as
)
E SD =
E0
a 3
r2 cos + sin
r
E0 (
r cos sin )
for r > a
for r < a
The continuity of the tangential E SD can be checked by taking the cross product
of r with E SD . It is seen that the tangential E SD is continues at r = a and the
constant E0 is evaluated using the discontinuity relation for the normal D eld:
0 E SD (r = a+ ) E SD (r = a ) r = P = P cos
We nd E0 =
P
and
3 0
3
P a (2r cos + sin )

E SD =
3 0 r
+ P (r cos sin ) = z P
3 0
3 0
for r > a
for r < a
168
1. Fundamentals
Thus, at the place of the induced dipole, E loc = E +
P
3 0
P = N E loc
P
= N E +
3 0
N
E = ( 0 )E
1 N /3 0

1 + 2N /3 0
=
0
1 N /3 0
P =
P1.8.2
In the low-eld limit M
N m2 H
3kT ,
and the medium is linear.
P1.8.3
(a) This constitutive relation for cholesteric liquid crystals is

(1) Anisotropic
(2) Linear
(4) Inhomogeneous: depends on position.
(b) This constitutive relation for the quartz crystals is
(1) Bianisotropic
(2) Linear
(3) Temporally dispersive
(4) Homogeneous
or
(1) Anisotropic
(2) Linear
(3) Spatial dispersive
(4) Homogeneous
(c) Hall Eect, the constitutive relation is
(1) Anisotropic
(2) Linear
(3) Temporally dispersive: Permittivity depends on .
(4) Homogeneous
(d) Consider the following dispersion relation:
Di = ij Ej + ijk
Ej
xk
The constitutive relation is

(1) Anisotropic: D and E are not related by a scalar factor.
(2) Linear
(3) Spatially dispersive: The constitutive relation has space derivatives of E .
(4) Homogeneous: ij and ijk do not depend on r .
(e) The constitutive relation for pyroelectricity is
(1) Anisotropic
Answers
169
(2) Linear : Variations of D and E are linearly related. D = E .

(4) Homogeneous
(f) The constitutive relation for piezoelectricity is
(1) Anisotropic
(2) Linear : Variations of D and E are linearly related. D = E .
(4) Homogeneous
(g) For the Kerr Eect, the constitutive relation is
(1) Anisotropic
(2) Nonlinear
(4) Homogeneous
(h) For the Pockels Eect, the constitutive relation is
(1) Anisotropic
(2) Nonlinear
(4) Homogeneous
P1.9.1
E1 + v B1 E2 + v B2
,

=0
n
(H1 H2 ) n
v (D1 D2 ) = J s vs

Maxwell Equations Fundamentals

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Maxwell Equations Fundamentals

Загружено:

Авторское право:

Доступные форматы

1

1.1 Maxwell Equations

C. Rectangular Cavity Resonators

1.1 Maxwell Equations

The laws of electricity and magnetism were established in 1873 by James

where E, B, H, D, J, and are real functions of position and time.

B (r, t) = magnetic ux density

H (r, t) = magnetic eld strength

D (r, t) = electric displacement

J (r, t) = electric current density

(r, t) = electric charge density

Equation (1.1.1) is Amp`eres law or the generalized Amp`ere circuit law.

Constitutive Relations for Free Space

In the form of scalar partial dierential equations, we have

1.1 Maxwell Equations

A wave equation for E can be derived by eliminating H from (1.1.11) and

Thus we obtain the following equations for

Making use of the Laplacian operator 2 in rectangular coordinate system

This is known as the Helmholtz wave equation.

The simplest solution to (1.1.17) takes the form

Substituting (1.1.18) in (1.1.17) we nd that the following equation,

The dispersion relation provides an important connection between the spatial

The unit for frequency f is Hertz (Hz) with 1 Hz = 1 s1 , which is equal to

1.1 Maxwell Equations

Figure 1.1.2 Electric eld strength vs. t for dierent frequencies .

In this book, we often refer to as the frequency, simply because

We call k the spatial frequency or the wavenumber which is equal to the

To further understand the meaning of k as a spatial frequency, we plot

Similar to the unit Hz which is cycles per second in temporal variation, Ko is

1.1 Maxwell Equations

of one period in one meter distance, we have k = 1 Ko . An electromagnetic

Visible (1.4 2.6 MKo ; 417 789 THz )

Thermal Infrared (0.067 0.14 MKo ; 20 43 THz )

Radar C, X, Ku , K bands (13.3 90 Ko ; 4 27 GHz )

Microwave Oven (8.17 Ko ; 2.45 GHz )

Television (.18 .72 Ko

Short Wave Radio (.01 .001 K o

AM Radio (.00178 .00535 Ko

535 1605 kHz)

(104 103 K o ; 30 300 kHz)

VLF (105 104 Ko ; 3 30 kHz)

ULF (106 105 Ko ; 0.3 3 kHz)

SLF (107 106 Ko ; 30 300 Hz)

Figure 1.1.5 Electromagnetic wave spectrum.

1.1 Maxwell Equations

Phase Velocity and Phase Delay

Figure 1.1.6 Electric eld strength vs. kz at dierent times.

The spatial frequency k is, according to the dispersion relation, directly

Operating frequencies of common devices:

Temporal frequency (Hz)

Spatial frequency (Ko )

3.70 4.20 GHz

Electric Field Vector E and Magnetic Field Vector H

the vector magnetic eld H can be determined from (1.1.8). We nd

1.1 Maxwell Equations

Figure 1.1.7 Electric and magnetic eld vectors of an electromagnetic wave.

Write the amplitude of the magnetic eld vector H as H0

because as time t increases, z must decrease in order to maintain kz + t a

o / o is the free-space impedance.

1.1 Maxwell Equations

Johann Carl Friedrich Gauss (30 April 1777 23 February 1855)

B. Vector Analysis and Boundary Conditions