Navin Khaneja

Electromagnetic Waves and

Optics, a Linear System
Approach
– Monograph –

June 27, 2019

Springer Nature
Dedicated to Vijay Laxmi Muntichod
Preface

The origin of this book can be traced back to author’s research in nuclear mag-
netic resonance (NMR) spectroscopy. In 1945, Purcell, Torrey and Pound detected
Nuclear Magnetic Resonance (NMR) in condensed matter. They had a tank (cav-
ity/resonator) they filled with radio-frequency (rf) waves and immersed nuclear
spins in it, all placed in static magnetic field B. When B was tuned, spins absorbed rf
and water level in the tank ( rf level in the cavity actually ) went down. How do you
fill tanks with rf/microwaves. How do we understand all this. Modern NMR/radio
equipment uses tank circuits to amplify weak signals and LC oscillators to generate
rf. Radars use special oscillator/cavities to generate microwaves and special pipes
called waveguides to channel them. In this book we study all this in the framework
of linear systems.
Maxwell equations that govern the evolution of EM waves is a linear system.
To solve for the linear system, we have to calculate its eigenvaues and eigenfunc-
tions (modes). In this book, we take this approach and calculate modes for Maxwell
equations, both in free space (where they are travelling waves) to space bounded by
conductors like cavities and waveguides (where they are standing waves). Central to
theme of the book is the phenomenon of resonance. When the modes of the linear
system have imaginary eigenvalues jω0 , and we drive the system with a harmonic
input exp( jω0t), the mode builds up unbounded. We say, we drive the system on
resonance and pump energy into it. In this book, we show how we study cavities,
resonators and waveguides as systems driven on resonance. We study tank circuits
and resonant antennas. We show resonance is central to how we produce so called
microwaves in cavities. Another important notion is study of linear systems is the
concept of a steady state. When eigenvalues of linear sytem have negative real parts
then system is termed lossy and in response to an harmonic input settles down to
a steady state satisfying an algebraic equation. This steady state analysis forms the
basis of analysis of antenna’s and transmission lines as discussed in the text.
In this book, we study analysis of Maxwell equations, waveguides, cavities,
transmission lines and antenna’s. We study oscillators for producing rf/microwaves
and amplifiers and tank circuits for amplifying these signals. The later part of this
book is devoted to study of light including optical instruments and pheonomenon

vii
viii Preface

of optical resonance in an atom, which explains dielectric response of a medium

and absorbtion of light. In writing the book, I have tried to answer all the questions,
I had as an electrical engineering undergraduate at IIT Kanpur, taking an electro-
magnetism course. I have chosen subjects that I feel are conceptual. I have stressed
system theoretic ideas. In using analysis tools of electromagnetic theory that can
be found elsewhere, I have stressed what system theoretic concept we are invoking.
May that be calculation of modes, or steady state of a linear system, notion of reso-
nance and resonant driving, etc. The book is written as a research monograph, but it
is more an effort on author’s part to see all of electromagnetism as a linear system.
The book is intended for a first course in electromagnetism or elctromagnetic theory.
The book may also be used in a course on rf/microwave engineering and optics.
It is an opportunity to acknowledge numerous people who I believe directly and
indirectly helped with this effort. I would like to thank Professor Steffen Glaser and
Niels Nielsen for numerous years of excellent collaboration in NMR spectroscopy
that ultimately led me to develop this text. I would like to thank Professor Roger
Brockett, who helped me in nurturing my taste in physics. I am grateful to Profes-
sor M. Sachidananda at IIT Kanpur whose course on electromagnetism has kept
my interest in the subject alive and culminated in this text. I will like to thank the
wonderful colleagues and academic environment of SYSCON at IIT Bombay that
provided ample opportunity for self development. Finally I like to acknowledge the
support of my family which made this effort possible.

IIT Bombay, Navin Khaneja

June 2019
Contents

Part I Electromagnetic waves and linear systems

1 Linear Systems and Electromagnetic Waves . . . . . . . . . . . . . . . . . . . . . . . 5

1.1 Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.1 Steady State Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 Maxwell Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Applications of Electromagnetic Waves . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.1 Optical Fibres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.2 Radio and Television Broadcast . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 Satellite Communications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.4 Microwave Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.5 Radar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3.6 Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.7 Radio Astronomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.8 Microwave Heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.3.9 Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2 Electromagnetic Waves and Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2.1 Maxwell Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Wave Equation and Linear Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Wave Equation and Media Boundary Conditions . . . . . . . . . . . . . . . . 22
2.4 Conductor Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.5 Conductor Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3 Waveguides and Resonators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.1 Parallel Plane Waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Rectangular Waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.1 TE mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2.2 TM mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Rectangular Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Cylindrical Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

ix
x Contents

3.4.1 TE Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.4.2 TM modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Quality Factor of Resonator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 Vector Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.1 Dipole Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Radiated Power and Antenna Resistance . . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Antenna Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3.1 2D arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.2 Antennas and finite interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . 49

5 Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.1 Coaxial Transmission Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.2 Impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3 Impedance Matching and Smith Charts . . . . . . . . . . . . . . . . . . . . . . . . 56

6 Rf and Microwave Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6.1 Microwave Oscillators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.1.1 Coaxial Cylindrical Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.1.2 TE Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.2 TM Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.2.1 Magnetron Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.2.2 Klystron and Gyrotron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.2.3 Microwave systems and networks . . . . . . . . . . . . . . . . . . . . . . 66
6.3 Impedance matching waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.3.1 Taper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.3.2 Transmission line models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.3 Iris . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.3.4 Post and Screws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.3.5 Plungers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.3.6 Rf-oscillators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.4 Tank Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.4.1 NMR and radio receiver and tank circuit . . . . . . . . . . . . . . . . . 80
6.4.2 Resonant radio antenna and tank circuit . . . . . . . . . . . . . . . . . 80
6.5 Klystron microwave amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
6.6 Transistor Rf amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.6.1 Current Source and amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.7 Rf and Microwave Filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.8 Cavity Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

7 Geometric Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7.1 Light Around Us . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.2 Pin Hole . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.3 Lens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.4 Microscopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Contents xi

7.5 Telescopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.6 Resolution and Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.7 Lens Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7.8 Revisiting Optical Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

8 Optical resonance and colors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

8.1 Atom as an Antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
8.1.1 Electric Dipoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
8.1.2 Classical theory of optical susceptibility . . . . . . . . . . . . . . . . . 100
8.1.3 Quantum theory of optical susceptibility . . . . . . . . . . . . . . . . . 101
8.2 Spontaneous Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
8.2.1 Spontaneous Emission in a Resonant Cavity and Purcell
effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.2.2 Line width in spontaneous emission . . . . . . . . . . . . . . . . . . . . . 106
8.3 Lasers and Cavities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
8.4 World of Colors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
8.4.1 Raman Effect and Color . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.5 Scattering and Blue Sky . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
 Part I
Electromagnetic waves and linear systems
2

Electromagnetic (EM) waves are all around us. From the world of light and color,
to TV and radio broadcast, to cell phones and dish TV, everywhere. In this book, we
study generation and propagation of EM waves as a linear system. The evolution
of EM waves is governed by Maxwell equations. We treat Maxwell equations as
a linear system, driven by current sources so called antennas. So what is a linear
system. To fix ideas, consider a spring mass system of mass m and spring constant
k, to which force F is applied. The displacement of the system from its equilibrium
denoted by x satisfies
mẍ + kx = F, (0.1)
or
ẍ + ω02 x = F/m, (0.2)
F
with x1 = ω0 x and x2 = ẋ, and u = m we find the above system written as
      
d x1 0 ω0 x1 0
= + u. (0.3)
dt x2 −ω0 0 x2 1
| {z } |{z}
A b
 
x1
If we denote X = , we can write the above system as
x2
dX
= AX + bu. (0.4)
dt
In general, A is a n × n matrix, X, b are n × 1 vector and u is a time varying input. We
call above equation a linear system. Maxwell equations are a linear system, where
input are current sources that act as antenna. To solve for evolution of linear system,
we have to solve for the eigenvales and eigenvectors of the matrix A. The eigenvec-
tors are called modes. This is what we do in this book, we solve for modes of the
Maxwell equations which in free space are travelling waves. We solve for modes
of the Maxwell equations when the space is bounded by conductors as in what are
called waveguides, cavities and resonators. These are called cavity modes. Central
to theme of the book is the phenomenon of resonance. When the modes of the linear
system have imaginary eigenvalues jω0 and we drive the system with a harmonic
input u = exp( jω0t) or u = cos ω0t, the mode builds up unbounded. We say we drive
the system on resonance and pump energy into it. In this book, we show how we
study cavities, resonators and waveguides as systems driven on resonance. We study
tank circuits and resonant antennas. We show resonance is central to how we pro-
duce so called microwaves in cavities. Another important notion is study of linear
systems is the concept of a steady state. When eigenvalues of A have negative real
parts then system is termed lossy and in response to an input u = exp( jω0t) settles
down to a steady state vector X0 exp( jω0t) with X0 satisfying the linear equation,

jω0 X0 = aX0 + b (0.5)

This steady state analysis forms the basis of analysis of antenna’s and transmission
lines as discussed in the text. Even when system is lossless, there is always a small
resistance present making it lossy and we can talk about steady state. While the
earlier part of the book covers EM waves as low frequencies radio waves (3 kHz-3
GHz) and microwaves (3 GHz-300 GHz) , the last chapters of the book treat optical
phenomenon, with EM waves in optical frequencies (1014 − 1015 Hz).
Chapter 1
Linear Systems and Electromagnetic Waves

Suppose a city gets fresh water supply as a fountain or eruption from the ground.
How will we use this water. We will build a tank to collect all this water and then use
pipes to carry this water to houses and then have taps at end of pipes to get water.
Now imagine instead of water, I have electromagnetic (EM) waves, say microwaves
(electromagnetic waves with frequency 3-300 GHz) . I have a supply of these mi-
crowaves as radiation from electron circulating a magnetic field. How do we collect
this radiation. How do we channel this radiation. How do we then use this radiation
in say a RADAR that detect flying objects using microwaves or a microwave oven
that cooks food using microwaves. As we see subsequently, we really collect this
radiation in a tank called cavity, and then use pipes called wave-guides to take it to
an antenna (the analogue of tap). which directs these waves towards a flying object.
How do we fill a cavity with microwaves. How do we model such electromagnetic
waves. What equations govern their evolution is space and time. All this we study in
this book. We model generation, storage and transmission of electromagnetic waves
as a linear system. We show that such a perspective of linear system lends great
mathematical insight is analysis of equations that govern the generation, storage
and transmission of electromagnetic waves. So what do we mean by linear systems.
Lets have a look.

1.1 Linear Systems

Consider a mass M tied by spring of spring constant k to a wall as in Fig. 1.1. The
displacement x of the mass from its equilibrium is related to force F as

M ẍ + kx = F. (1.1)
k F
Using ω02 = M and u = M, we get for x1 = ω0 x and x2 = ẋ,

5
6 1 Linear Systems and Electromagnetic Waves

k
M

x
Fig. 1.1 Fig. shows a mass M tied to wall with a spring with spring constant k being pulled with
force F. x measures displacement from equilibrium.

      
d x1 0 ω0 x1 0
= + u. (1.2)
x
dt 2 − ω0 0 x2 1
| {z } |{z}
A b
 
x
If we denote X = 1 , we can write the above system as
x2
dX
= AX + bu. (1.3)
dt
In general, A is a n × n matrix, X, b are n × 1 vector and u is a time varying input.
We call Eq. ( 1.3) a linear system.
Without an input the system takes the form

Ẋ = AX. (1.4)
Using X(t + ∆ t) = (I + A∆ t)X(t), we get X(n∆ t) = (I + A∆ t)n X(0), with t =
n∆ t, we het X(t) = (I + Atn )n X(0). With ∆ t → 0 and n → ∞, we have X(t) =
exp(At)X(0) where

(At)2 (At)n
exp(At) = I + At + +···+ +.... (1.5)
2! n!
Suppose u Eq. (1.3) is nonzero over an interval τ and τ + ∆ . Then X(τ + ∆ t) =
(I + A∆ t)X(τ ) + bu(τ )∆ t, and continuing we get

X(t) = exp(A(t − τ ))X(τ ) + exp(A(t − τ ))bu(τ )∆ t

= exp(At)X(0) + exp(A(t − τ ))bu(τ )∆ t.

and then in general for non zero u, we have

Z t
X(t) = exp(At)X(0) + exp(A(t − τ ))bu(τ )d τ . (1.6)
0
This is called the variation of constant formula. Although Eq. (1.6) solves Eq. (1.3),
we show how we can further simplify.
1.1 Linear Systems 7

To solve for evolution of the system Eq. (1.4), we observe that if the initial state
X(0) of the system is e an eigenvector of A with eigenvalue λ , then the solution
is simply X(t) = exp(λ t)e. Simply substitute in Eq. (1.4). In general, we can de-
compose X(0) = ∑i αi (0)ei where ei are eigenvectors of A with eigenvalue λi . Then
observe X(t) = ∑i αi (0) exp(λit)ei is the solution as can be seen by substitution.
The solution is understood as simply writing the differential equation in terms of
basis ei with X(t) = ∑i αi (t)ei then the vector equation gets expressed as n scalar
equations

α̇i = λi αi , (1.7)
with solution αi (t) = exp(λit)αi (0). In presence of input as in Eq. (1.3), we can
express b = ∑i bi ei and equations take form

α̇i = λi αi + bi u(t), (1.8)

with solution
Z t
αi (t) = exp(λit)αi (0) + bi exp(λi (t − τ ))u(τ )d τ . (1.9)
0
R
With αi (0) = 0, we simply have αi (t) = bi 0t exp(λi (t − τ ))u(τ )d τ . Of partic-
ular
R
interest is the case bi correspond to imaginary eigenvalues jωi . Then αi (t) =
bi 0t exp( jωi (t − τ ))u(τ )d τ . If the input u(t) = exp( jω t), and we choose ω = ωk ,
then all modes average only the mode k survives giving αk (t) = bk exp( jωk t)t. We
say we excite the system on resonance with mode k, which builds as shown. This is
the main idea of this text. We identify modes and their frequencies and then drive
the system at the frequency of a mode to build that mode. We say we pump energy
into a mode.
For example
  in Eq.(1.2),  we have eigenvalues as jω0 and − jω0 with eigenvec-
1 1 0
tors e1 = , and e2 . We write input b = = − 2i (e1 −e2 ). If input u(t) =
j −j 1
exp( jω0 t)+exp(− jω0 t)
exp( jω0t), the mode e1 builds up. If input u(t) = cos(ω0t) = 2 , the
mode e1 and e2 both build up.

1.1.1 Steady State Solution

Given the linear system

dX
= AX + bu, (1.10)
dt
where A is dissipative, i.e, eigenvalues of A have negative real part. Let u =
exp( jω t). Then we can simply observe x = X0 exp( jω t) is a solution to Eq. (1.10),
where X0 is solution to algebraic equation
8 1 Linear Systems and Electromagnetic Waves

jω X0 = AX0 + b. (1.11)

or
( jω 1 − A) X0 = b. (1.12)
| {z }
A

Since A is dissipative, we have A invertible and we can solve for X0 , We call it

steady state solution because if y is the true solution then z = y − x satisfies

dz
= Az (1.13)
dt
with z(T ) = exp(AT )z(0), since A is dissipative, z(T ) → 0 as T → ∞ and y(T ) →
x(T ). In physical problems, one encounters A that have purely imaginary eigenval-
ues, as in spring mass problem in 1.2. However this is an ideal situation, in practice
there is a small amount of friction present, which gives dissipative A. Therefore,
even in such ideal situtations, it makes sense to talk about steady state solution
keeping in mind negligibly small dissipation.
Consider a scalar linear system

ẋ = −α x + u. (1.14)
For s = jω , and input u = u(s)exp(st), the steady state is x(s) exp(st), where x(s)
satisfies algebraic equation
(s + α )x(s) = u(s), (1.15)
with

x(s) 1
G(s) = = . (1.16)
u(s) s + α
G(s) is called the input-output transfer function which is the frequency response of
the system as shown in Fig. 1.2A.

G(s) G(s)
u(s) x(s) u(s) − x(s)

y(s) H(s)
A
B
Fig. 1.2 Fig. shows linear systems represented by transfer functions

Now lets take x(t) and feed it as input to another system

1.2 Maxwell Equations 9

ẏ = −β y + x(t), (1.17)
then
y(s) 1
H(s) = = . (1.18)
x(s) s + β
Now if we take y and feed it back to x with a negative sign

ẋ = −α x + u − y, (1.19)

as shown in Fig. 1.2B. Then

x(s) = G(s)(u(s) − y(s)) = G(u(s) − Hx(s)), (1.20)

or
x(s) G(s)
C(s) = = . (1.21)
u(s) 1 + G(s)H(s)
C(s) is called the feedback or closed loop transfer function.

1.2 Maxwell Equations

Why do we care about linear systems. Because we want to study electromagnetic

waves [18, 21, 22, 31, 17]. We know electromagnetic waves everyday from radio
and TV broadcast and light and color around us. These waves propagate as oscillat-
ing electric and magnetic field. The equation governing their evolution are the the
Maxwell equations. Electromagnetic waves are space and time varying electric and
magnetic fields related by Maxwell equations as in

∂H
µ = −∇ × E (1.22)
∂t
∂E
ε = ∇ × H − J(r,t), (1.23)
∂t
where each spatial location r indices a vector entry in linear system as in Eq. ( 2.16).
The vector entries are E (Electric) and H (Magnetic) fields. The goal is to calculate
E(r,t) and H(r,t) resulting from an current input J(r,t). That is solve a linear system
for an input u(t). It means we have to find modes of the system and corresponding
eigenvalues. That is we have to solve for Ax = λ x in Eq. ( 1.4), i.e solve for

1
− ∇ × E = λ H, (1.24)
µ
1
∇ × H = λ E. (1.25)
ε
10 1 Linear Systems and Electromagnetic Waves

Of particular interest is when λ = jω , imaginary eigenvalues. In which case, the

qthree vector k = |k|k̂ , with x̂, ŷ, k̂ forming a right
eigenvectors are as follows. For
E0 µ0
handed system. For H0 =η = σ0 , we have

E = E0 exp(− jk · r)x̂ = E0 exp(− jkx x + ky y + kz z)x̂, (1.26)

H = H0 exp(− jk · r)ŷ = H0 exp(− jkx x + ky y + kz z)ŷ. (1.27)

is a eigenvector with c|k| = ω . Thus eigenvectors are the Fourier modes.

Now we want to drive Maxwell equations with current sources J(r,t). This cur-
rent source is an antenna. They can infact be simplified to J(r)u(t). Think of J(r) as
b in linear system. These sources are localized and can be expanded into a Fourier
basis but these Fourier basis are precisely the eigenfunctions of Maxwell equation
with imaginary eigenvalues. Then if we choose u(t) = exp( jω t), then as described
above, we will excite the modes with frequency ω . The antenna puts energy into
the Fourier modes whose frequency is same as the frequency of excitation of an-
tenna. These modes get energy, while other modes don’t and we can detect these
modes very far from the antenna because they are delocalized and hence we can
communicate to someone very far because we pump energy into a mode that is
shared simultaneously between the transmitter and receiver. In summary, we care
about eigefunctions corresponding to imaginary eigenvalues as their natural evolu-
tion constitutes travelling waves and we communicate by pumping energy in these
modes.
Till now we talked about Maxwell equations in free space. The eigenmodes with
imaginary eigenvalues are Fourier modes. The evolution of these modes is

E = E0 exp( j(ω t − k · r))x̂, (1.28)

H = H0 exp( j(ω t − k · r))ŷ. (1.29)

This is an EM wave propagating in the k direction.

Another very interesting situation is when Maxwell equations are confined in
a cavity. A cavity is free space enclosed with conducting walls. EM waves don’t
propagate inside conductor, leading to zero tangential electric fields and zero nor-
mal magnetic fields at boundaries. We can solve the eigenvectors of the Maxwell
equations with these boundary conditions to get what are called modes of the cavi-
ties. These modes say
φ1 , φ2 , . . . , φk , . . . ,
and have eigenvalues
jω1 , jω2 , . . . , jωk , . . . .
When we excite the cavity with a current source, again we can expand the localized
source into cavity eigenfunctions. Fig. (1.4) shows a current source as wire inserted
in the cavity. If the source has current exp( jωk t), we pump energy into mode φk
and the mode φk will build up and all cavity will get filled with mode φk . We show
as told now how to fill cavities with microwaves and channel them through pipes
1.2 Maxwell Equations 11

E
E

H H

a b
Fig. 1.3 Fig. a and b shows EM waves propagating along x and k̂ direction respectively.

a
Fig. 1.4 Fig. a shows how we can excite modes of cavity by putting in a wire carrying oscillating
current.

(wave-guides) for many applications. Electromagnetic waves in 30 kHz to 3 GHz

are called radio waves. 3 − 300 GHz are called microwaves. 1012 − 1014 Hz EM
waves are called infrared light and 1014 − 1015 Hz, visible light and so on. Each
frequency type has its own applications. Lets look at some of these applications
before we jump into main content of this book.
12 1 Linear Systems and Electromagnetic Waves

1.3 Applications of Electromagnetic Waves

1.3.1 Optical Fibres

These days long range communication between cities and countries is done using
fibre optic communication. The light used is Infrared frequency 180−330 THz. This
light is carried by a optical fibre [26] which is a cylindrical dielectric rod made of
silica, surrounded by a dielectric shell called cladding as in 1.5A. Light moves in the
core. The refractive index n1 of core is greater the refractive index n2 of surrounding
air. When light tries to leave the core as in 1.5B, it gets totally internally reflected
for small θ . This way fibre confines light and light travels far.

CLADDING

CORE
A n1

n2

θ
B Incident light Reflected Light

Fig. 1.5 Fig. A shows an optical fibre with core and cladding. Fig. B shows how light gets totally
internally reflected when it tries to leave the fibre.

1.3.2 Radio and Television Broadcast

The most traditional application of electromagnetic waves is Radio and Television

broadcast. Radio transmission is in low radio frequency 1 − 2 MHz in AM radio
(amplitude modulated) or in VHF (30 − 300 MHz) for FM radio (frequency modu-
lated). AM broadcast over long distance is done over short wave (HF (3 − 30 MHz).
Radio waves in the shortwave band can be reflected or refracted from a layer of
1.3 Applications of Electromagnetic Waves 13

electrically charged atoms in the atmosphere called the ionosphere. Therefore, short
waves directed at an angle into the sky can be reflected back to Earth at great dis-
tances, beyond the horizon. This is called skywave or ”skip” propagation [11, 31].
Thus shortwave radio can be used for very long distance communication, in contrast
to radio waves of higher frequency which travel in straight lines (line-of-sight prop-
agation) and are limited by the visual horizon, about 64 km (40 miles). Television
broadcast is done VHF frequency. Cell-phones we use every day, work in UHF (300
MHz to 3 GHz). This band is also used for television broadcasting, satellite com-
munication including GPS, personal radio services including Wi-Fi and Bluetooth,
walkie-talkies, cordless phones, and numerous other applications [28, 25].

1.3.3 Satellite Communications

We all have Dish TV at home where signal is received from a satellite [15]. The
signal is first transmitted to satellite using an uplink which operates in Ku microwave
band (12 − 18 GHz) frequency. The is is downconverted and broadcast in C-Band
(4 − 8 GHz).The signals are received via an outdoor parabolic antenna commonly
referred to as a satellite dish.

1.3.4 Microwave Links

Pre-fibre optic days, line of sight microwave links provided telephone connectivity
between cities. Typical frequencies used were less than 10 GHz. These days terres-
trial microwave relay links are used in in telecommunications networks including
cellular networks linking base transceiver station (BTS) and base station controller
(BSC) and Mobile switching center (MSC). Frequencies from 10−80 GHz are used.

1.3.5 Radar

One of the most prominent application of microwaves is the microwave radar

[13, 14] . Radar is a detection system that uses microwaves to determine the range,
angle, or velocity of objects. It can be used to detect aircraft, ships, spacecraft,
guided missiles, motor vehicles, weather formations, and terrain. A radar system
consists of a transmitter, producing electromagnetic waves in the microwave do-
main, a transmitting antenna, a receiving antenna (often the same antenna is used for
transmitting and receiving) and a receiver and processor to determine properties of
the object(s). Microwaves (pulsed or continuous) from the transmitter reflect off the
object and return to the receiver, giving information about the object’s location and
speed. Fig. (1.6) shows a schematic of radar where a oscillator cavity (magnetron,
14 1 Linear Systems and Electromagnetic Waves

klystron) labelled a produces microwaves which are channeled by wave-guides, la-

belled b to Horn antenna, labelled c.

a
Fig. 1.6 Fig. shows schematic of a RADAR including oscillator cavity, wave-guide and Horn
antenna.

The modern uses of radar are highly diverse, including air and terrestrial traffic
control, air-defense systems, antimissile systems, marine radars to locate landmarks
and other ships, aircraft anticollision systems, ocean surveillance systems, outer
space surveillance and rendezvous systems, meteorological precipitation monitor-
ing, altimetry and flight control systems, guided missile target locating systems, and
ground-penetrating radar for geological observations. Radar frequency ranges from
3-30 GHz.

1.3.6 Remote Sensing

Electromagnetic radiation in the microwave wavelength region is used in remote

sensing to provide useful information about the Earth’s atmosphere, land and ocean.
A radar mounted on a satellite transmits microwave pulses towards the earth sur-
face. The microwave energy scattered back to the spacecraft is measured. Remote
sensing makes use of the radar principle to form an image by utilizing the time delay
of the back-scattered signals. In real aperture radar imaging, the ground resolution
is limited by the size of the microwave beam sent out from the antenna. Finer de-
tails on the ground can be resolved by using a narrower beam. The beam width is
inversely proportional to the size of the antenna, i.e. the longer the antenna, the nar-
rower the beam. The microwave beam sent out by the antenna illuminates an area
on the ground (known as the antenna’s ”footprint”). In radar imaging, the recorded
1.3 Applications of Electromagnetic Waves 15

signal strength depends on the microwave energy back-scattered from the ground
targets inside this footprint. Increasing the length of the antenna will decrease the
width of the footprint. It is not feasible for a spacecraft to carry a very long antenna
which is required for high resolution imaging of the earth surface. To overcome this
limitation, principle of synthetic aperture radar (SAR) is used. SAR capitalizes on
the motion of the space craft to emulate a large antenna from the small antenna it
actually carries on board. When microwaves strike a surface, the proportion of en-
ergy scattered back to the sensor depends on many factors: Physical factors such
as the dielectric constant of the surface materials which also depends strongly on
the moisture content; Geometric factors such as surface roughness, slopes, orien-
tation of the objects relative to the radar beam direction; The types of land cover
(soil, vegetation or man-made objects). Microwave frequency, polarization and in-
cident angle. The ability of microwave to penetrate clouds, precipitation, or land
surface cover depends on its frequency. Generally, the penetration power increases
for longer wavelength (lower frequency). The SAR back-scattered intensity gener-
ally increases with the surface roughness. However, ”roughness” is a relative quan-
tity. Whether a surface is considered rough or not depends on the length scale of the
measuring instrument. If a meter-rule is used to measure surface roughness, then
any surface fluctuation of the order of 1 cm or less will be considered smooth. On
the other hand, if a surface is examined under a microscope, then a fluctuation of the
order of a fraction of a millimeter is considered very rough. In SAR imaging, the
reference length scale for surface roughness is the wavelength of the microwave.
If the surface fluctuation is less than the microwave wavelength, then the surface
is considered smooth. For example, little radiation is back-scattered from a surface
with a fluctuation of the order of 5 cm if a L-band (15 to 30 cm wavelength, 1 GHz
frequency) SAR is used and the surface will appear dark. However, the same surface
will appear bright due to increased back-scattering in a X-band (2.4 to 3.8 cm wave-
length, 10 GHz frequency ) SAR image. The C band is useful for imaging ocean and
ice features. However, it also finds numerous land applications. The L band has a
longer wavelength and is more penetrating than the C band. Hence, it is more useful
in forest and vegetation study as it is able to penetrate deeper into the vegetation
canopy.

1.3.7 Radio Astronomy

Radio astronomy is a sub-field of astronomy that studies celestial objects at radio

frequencies [?, 30]. For example radiation at 21 cm, 1428 MHz frequency is emitted
from galactic hydrogen due to what is called hyperfine splitting of its orbital. We can
detect this radiation with high angular precision, which gives the angular location
of the distant galaxy. The Doppler shift in the frequency gives estimate of how fast
λ
Galaxy is receding from us. The angular precision is measured by the ratio D where
D is the diameter of the antenna. With large wavelength, we need large diameter D.
16 1 Linear Systems and Electromagnetic Waves

This is made feasible by having a array of such antennas a technique called Synthetic
Aperture Radar similar to as in remote sensing.

1.3.8 Microwave Heating

Water in the food absorbs microwaves around 2.450 GHz. At this frequency, transi-
tions between rotational states are induced in the water molecule. The molecule in
excited rotational state dissipates energy into heat. This forms the basis of what is
called a microwave heating in microwave oven. Like Radar in a microwave oven,
we have a cavity to produce microwaves, which are carried by a short wave-guide
to the main cavity which is filled with microwaves to heat the food.

1.3.9 Spectroscopy

One of big applications of electromagnetic waves is spectroscopy. Using electro-

magnetic waves, we can probe structure of atoms and molecules. Different type
of spectroscopies correspond to different frequency electromagnetic waves. Radio-
frequency electromagnetic waves can induce transition between the spin states of
atomic nucleus placed in a magnetic field B0 . The nuclear spin is like a charged
spinning top which carries a magnetic moment µ . The difference in the energy of
spin (magnetic moment) oriented along and opposite to magnetic field is E = 2µ B0 .
Writing this energy in frequency E = h̄ω0 , we find ω0 for hydrogen at field of 10 T
corresponds to around 400 MHz. Thus radio frequency EM can probe nuclear spins
and this field of spectroscopy is called Nuclear Magnetic Resonance (NMR) [10]
spectroscopy. Electron spin has µ around 103 times higher than that of hydrogen
nucleus and we can probe electron spin at around same B0 field strength by GHz
radiation. This field is called Electron Paramagnetic Resonance (EPR). The energy
difference in rotational states of molecules correspond to GHz frequency and result-
ing spectroscopy goes by name of microwave spectroscopy. The energy difference
in the vibrational states of molecules correspond to THz frequency and resulting
spectroscopy goes by name of vibrational spectroscopy. The transition of electrons
between different orbitals correspond to optical 1015 Hz frequency and resulting
spectroscopy goes by name of optical spectroscopy.
Chapter 2
Electromagnetic Waves and Propagation

Electromagneic (EM) waves are oscillating electric and magnetic fields. In this
chapter, we develop the mathematics to describe their propagation. We begin with
Maxwell equations. We show how Maxwell equations constitute a linear system.
We show how we can solve for the modes of the linear system. These turn out to be
plane waves. We describe the evlotion of these plane waves. We then consider how
waves travel across interfaces [11]. We consider interfaces between two dielectrics
and interface between a dielectric and a conductor. In the former case, we see how
waves change course when crossing an interface and we recover the famous laws
of refraction. In the later case, we show how conductor relects EM waves. This
reflection forms the basis of mirrors in optics [?] and waveguides and cavities in
microwave engineering [6, 27] as we see in the subsequent chapters.

2.1 Maxwell Equations

Our starting point is the electric field due to a point charge q0 which is
1 q0
E= r̂. (2.1)
4πε0 r2
1
where ε0 is the permitivity of free space with 4πε 0
= 9 × 109 SI units. By use of the
Stoke’s theorem the above equation can be written as
ρ
∇E = . (2.2)
ε0
where ∇E is the divergence of E and ρ charge density. Eq. (2.2) above constitutes
our first Maxwell equation.
If we bring a test charge q, in presence of electric field E, it feels a force F =
qE. In addition to electric fields we also have magnetic fields. They arise due to
moving charges. When charge q0 moves it produces a magnetic field. Consider a

17
18 2 Electromagnetic Waves and Propagation

wire carrying current i0 , then the magnetic field at location O due to infinitesimal
current element is given by Biot-Savart law [1, 2] as shown in Fig. 2.1a

0 B
θ
r
dl
E

b
a
Fig. 2.1 Fig. a shows magnetic field at point O goint inside the page caused due to infinitesimal
current element. Fig. b shown how change of magnetic flux through a loop induces tangential
electric field.

µ0 i0 dl × r µ0 i0 dl sin θ
B= = φ̂ . (2.3)
4π r3 4π r2
where φ is the unit vector at O going inside the page.
As can be seen

∇ · B = 0. (2.4)
If we integrate over the whole wire, we get the net field at location O as
µ 0 i0
B= φ̂ . (2.5)
2π r
If we have current density J at location x and we come infinitely close to this then
the density looks infinitely long and we can write B as above in Eq. (2.5). If A is
the area of the current density we can write B(2π r) = (∇ × B)A and i0 = JA, which
from Eq. (2.5) gives
∇ × B = µ0 J, (2.6)
or

∇ × H = J. (2.7)
2.2 Wave Equation and Linear Systems 19

Electric field E induces a dipole per unit volume given by the displacement D = ε E.
Then ∂∂Dt tantamounts to a current density and above Eq. 2.7 generalizes to

∂D
∇×H = J + . (2.8)
∂t
Finally consider a loop of wire (of radius r and area A) with magnetic field B
passing through it as shown in Fig. 2.1b. This results in flux Φ = BA through the
loop. Change of flux induces a tangential electric field in the loop whose line integral
is given by

∂φ
E(2π r) = − . (2.9)
∂t
As we shrink the loop we get for infinitesimal loop E(2π r) = (∇ × E)A, which gives
the famous Faraday’s law

∂B
∇×E = − . (2.10)
∂t
Putting equations 2.2, 2.4, 2.8 and 2.10 together we get the famous Maxwell
qquations.

2.2 Wave Equation and Linear Systems

The Maxwell equations are

ρ (x,t)
∇·E = , (2.11)
ε0
∇ · B = 0, (2.12)
∂B
= −∇ × E, (2.13)
∂t
∂D
= ∇ × H − J(x,t). (2.14)
∂t
Suppose we start with E and B satisfying Eq. (2.11 and 2.12). Then evolution
under Eq. (2.13 and 2.14) ensures they are staisfied for all t. All we have to take
notice is the fact and divergence of curl is zero and the equation of continuity

∂ ρ (x,t)
+ ∇ · J = 0. (2.15)
∂t
So we focus on Eq. (2.13 and 2.14). These equations constitute what we call a
linear system. In a linear system, we call the state of the system X, which evolves
according to the equation
20 2 Electromagnetic Waves and Propagation

Ẋ = AX + bu(t), (2.16)
where X is say an n × 1 vector and A an n × n matrix and b an n × 1 input vector and
u(t) input. Without an input the system takes the form

Ẋ = AX. (2.17)
To solve for evolution of this system, we observe if the initial state X(0) of the
system is e, an eigenvector of A, with eigenvalue λ , then the solution is simply
X(t) = exp(λ t)e. Simply substute in Eq. (2.17). In general, we can decompose
X(0) = ∑i αi (0)ei , where ei are eigenvectors of A with eigenvalue λi . Then observe
X(t) = ∑i αi (0) exp(λit)ei is the solution as can be seen by substitution.
The solution is understood as simply writing the differential equation in terms of
basis ei with X(t) = ∑i αi (t)ei , then the vector equation gets expressed as n scalar
equations

α̇i = λi αi , (2.18)
with solution αi (t) = exp(λit)αi (0). In presence of input as in Eq. (2.16), we can
express b = ∑i bi ei and equations take the form

α̇i = λi αi + bi u(t), (2.19)

with solution
Z t
αi (t) = exp(λit)αi (0) + bi exp(λi (t − τ ))u(τ )d τ . (2.20)
0
R
With αi (0) = 0, we simply have αi (t) = bi 0t exp(λi (t − τ ))u(τ )d τ . Of particular
interest
R
is the case when λi correspond to imaginary eigenvalues jωi . Then αi (t) =
bi 0t exp( jωi (t − τ ))u(τ )d τ . If the input u(t) = exp( jω t) and we choose ω = ωk ,
then all modes average only the mode k survives giving αk (t) = bk exp( jωk t)t. We
say we excite the system on resonance with mode k which builds as shown. This is
the main idea of this text. We identify modes and their frequencies and then drive
the system at the frequency of a mode to build that mode. We say we pump energy
into a mode. Lets now go to Maxwell equations in

∂H
µ = −∇ × E, (2.21)
∂t
∂E
ε = ∇ × H − J(r,t), (2.22)
∂t
where each spatial location r indices a vector entry in linear system as in Eq. ( 2.16).
The vector entries are E and B fields. The goal is to calculate E(r,t) and B(r,t)
resulting from an current input J(r,t). That is solve a linear system for an input u(t).
It means we have to find modes of the system and corresponding eigenvalues. That
is we have to solve for Ax = λ x in Eq. ( 2.17), i.e solve for
2.2 Wave Equation and Linear Systems 21

1
− ∇ × E = λ H, (2.23)
µ
1
∇ × H = λ E, (2.24)
ε
with E = Eo exp(−kx) ŷ and H = Ho exp(−kx) ẑ , where x̂, ŷ, ẑ are standard unit
vectors, we get

k
Eo = λ Ho , (2.25)
µ
k
Ho = λ Eo , (2.26)
ε
q
1 Eo µ
which gives ck = ±λ , where c2 = µε and H o
= ± η , where η = ε . Of par-
ticular interest is when λ = jω is imaginary, then E = Eo exp(− jkz) x̂ and H =
Ho exp(− jkz) ŷ with ck = ω . The modes evolve as E = Eo exp(− j(kx − ω t)) ŷ, and
H = Ho exp(− j(kx − ω t)) ẑ. This constitues a travelling wave along x direction, as
shown in Fig. 2.2a. There is another eigenvector corresponding to ck = −ω , which
evolves as E = Eo exp( j(kx + ω t)) ŷ, and H = −Ho exp( j(kx + ω t)) ẑ and constitues
a travelling wave along −x direction.
Let x̂, ŷ, ẑ be some unit vectors forming a right handed coordinate system such
k
that x̂ = |k| with k = (kx , ky , kz ) and E = Eo exp(− j kx x + ky y + kz z) ŷ also written as
E = Eo exp(− jk · r) ŷ and H = Ho exp(− jk · r) ẑ are eigenvectors corresponding to
eigenvalue jω and E = Eo exp(− j(k · r − ω t)) ŷ and H = Ho exp(− j(k · r − ω t)) ẑ is
a travelling wave along k direction as shown in Fig. 2.2b and there is a corresponding
wave E = Eo exp( j(k · r + ω t)) x̂ and H = Ho exp( j(k · r + ω t)) ŷ travelling along
−k direction.
What have we learnt? The eigefunctions corresponding to imaginary eigenval-
ues are oscillatory in space and their evolution constitute travelling waves. Even-
tually we want to drive Maxwell equations with current sources J(r,t). These are
antenna’s. They can infact be simpified to J(r)u(t). Think of J(r) as b in linear sys-
tem. These sources are localized and can be expnaded into a Fourier basis but these
Fourier basis are precisely the eigenfunctions of Maxwell’s equation with imagi-
nary eigenvalues. Then if we choose u(t) = exp( jω t) then as described above we
will excite the modes with frequency ω . The antenna puts energy into travelling
wave modes whose frequency is same as the frequency of excitation of Antenna.
These modes get energy and become much bigger than other modes and we can de-
tect these modes very far from the antenna beacuse they are delocalized and hence
we can communicate to someone very far because we pump energy into a mode that
is shared simultaneously between the transmitter and receiver. In subsequent chap-
ters, we present other elegant techniques for solving for electric and magnetic fields
arising due to antenna excitation. In summary, we care about eigefunctions corre-
22 2 Electromagnetic Waves and Propagation

E
E

H H

a b
Fig. 2.2 Fig. a and b shows EM waves propagating along x and k̂ direction respectively.

sponding to imaginary eigenvalues as their natural evolution constitutes travelling

waves and we communicate by pumping energy in these modes.

2.3 Wave Equation and Media Boundary Conditions

We have talked about travelling waves q E = E0 exp(− j(k · r − ω t)x̂, and H =

E0 µ
H0 exp(− j(k · r − ω t)ŷ, with H 0
= η = ε . This is transmission in free space.
How about transmission across a media interface. Consider the scenario as depicted
in figure 2.3. We have a transmitted wave in region I that arrives at media interface,
which is say the y − z plane. Region I is x < 0 and region 2 is x > 0. The permitivity
if I is ε1 and II is ε2 . The incident wave cannot continue unaltered into region II as
η for region II is different. We will need to alter E0 and H0 . We do this through a
reflected wave and transmitted wave. We say there is a reflected wave that goes back
in region I and transmitted wave that proceeds in region II. This is depicted in figure
2.3.
If we denote the spatial dependence of incident, refelcted and transmiited wave
as exp(− jkI · r), exp(− jkr · r) and exp(− jkt · r) , then our first claim is that kI , kr , kt
are in the same plain. W.L.O.G, assume kI in the x, z plain. Then first note that
tangential and normal components of E and H is same at the interface. This follows
from figure 2.4. In figure 2.4a ,

E2 − E1 = (∇ × E)⊥ ∆ x, (2.27)

where ∇ × E is the curl and (∇ × E)⊥ its component perpendicular to the plane. The
wave arrangement of incident, transmitted and reflected wave is an eigenfunction of
the Maxwell equation satisfying
2.3 Wave Equation and Media Boundary Conditions 23

z
y

x
E

t
H

I i H
II

E
H

E
Fig. 2.3 Figure shows incident, transmitted and reflected waves across a media interface between
media I and II.

1
− ∇ × E = jω H (2.28)
µ
1
∇ × H = jω E (2.29)
ε
hence curl is finite and therefore in Eq. (2.27) as ∆ x → 0 we get E2 = E1 . Further-
more from Eq. (2.28 and 2.29), we find that ∇ · E = 0 and ∇ · H = 0. Therefore from
figure 2.4b we get for ∆ x → we have

E2 − E1 = (∇ · E)∆ z, (2.30)
and since ∇ · E is finite we have E1 = E2 . Same is true for H and we get across
the boundary the tangential and normal componets of E and H are the same. Lets
compare y component of E, we get
24 2 Electromagnetic Waves and Propagation

∆z ∆x
E E1 E
E1 2 2

∆x
a b
Fig. 2.4 Fig. a. and b. shows tangential and normal components of the electric field across a media
interface.

Eio exp(− j(kix x+kiz z))+Ero exp(− j(krx x+kry y+krz z)) = Eto exp(− j(ktx x+kty y+ktz z)).
(2.31)
Lets only vary y to get

Eio + Ero exp(− jkry y) = Eto exp(− jkty y). (2.32)

Differentiaiting we get kty = kry , which further implies kty = kry = 0. Which means
transmitted and refelcted wave is in same plane. Then writing Eq. (2.31), we get
from fig. 2.5.

Eio exp(− jki (sin θi z+cos θi x))+Ero exp(− jki (sin θr z−cos θr x)) = Eto exp(− jkt (sin θt z+cos θt x)).
q (2.33)
sin θ t ki ε1
Again only vary z to conclude sin θi = sin θr and sin θ i = kt = ε2 . That is angle of
incidence is same as angle of relflection. The ratio of sine of transmitted and incident
angle obeys what is knows as snell’s law of refraction. The transmitted wave bends
inwards as it moves into a dense (larger ε ) media.
Now equating tangential components of E and H at the origin gives

Eio + Ero = Eto , (2.34)

cosθi cos θt
(Eio − Ero ) = Eto , (2.35)
η1 η2
2.3 Wave Equation and Media Boundary Conditions 25

z
y E
H

x
H
E

θr
I θ t
θi II

E
Fig. 2.5 Fig. shows incident, transmitted and reflected wave at the media interface of medium I
and II. E field is parallel to interface

which gives

(Eio + Ero ) = Eto , (2.36)

η1 cos θt
(Eio − Ero ) = Eto , (2.37)
η2 cos θi
η1 cos θt 2Eio 1−ab
with a = η2 , and b = cos θi , we have Eto = 1+ab and Ero = 1+ab Eio , where b is
sin θt −1
calculated from sin θi = a .
For normal incidence, with θi = θt = 0, we get b = 1 and Ero = 1−a 1+a Eio .
We discussed the case when incident wave has E tangential to interface while H
is incident at an angle. The second scenario is when incident wave has H tangential
to interface while E is incident at an angle. This is as shown in fig. 2.6.
For this case we get

η1
(Eio + Ero ) = Eto , (2.38)
η2
(Eio − Ero ) cos θi = Eto cos θt , (2.39)
26 2 Electromagnetic Waves and Propagation

z
y
E
x

H H

E
θr
I θ t
θi II

Fig. 2.6 Fig. shows incident, transmitted and reflected wave at the media interface of medium I
and II. H field is parallel to interface

(a−b)Ei0
which gives Eto = 2E
a+b and Ero =
io
a+b .
a−1
For normal incidence with θi = θt = 0, we get b = 1 and Ero = a+1 Eio .

2.4 Conductor Interfaces

−∇ × E = jω µ H, (2.40)
∇ × H = ( jωε + σ )E, (2.41)

with E = Eo exp(−γ x) ŷ amd H = Ho exp(−γ x) ẑ , where x̂, ŷ, ẑ are standard unit
vectors, we get
2.4 Conductor Interfaces 27

γ Eo = jω µ Ho , (2.42)
γ Ho = ( jωε + σ )Eo , (2.43)

giving
p
γ= jω µ ( jωε + σ ) = α + jβ (2.44)
q
E0 jω µ
and H 0
= η = jωε +σ . The wave evolves then E = Eo exp(−(α + j β )x + j ω t) ŷ
and H = Ho exp(−(α + jβ )x + jω t) ẑ. This is travelling wave in x direction that
decays with x with exponent α .
−10
For a typical metal σ = 107 SI units, ε = 104π SI units and µ = 4π × 10−7
SI units and for ω = 1011 SI units we get α ∼ 10−6 SI units. Therefore within a
distance of a micron, the wave attenutates to e−1 of its value.
Now consider a media interface where region I is a dielectric with permitivity
ε1 and region II is a conductor with permitivity ε2 and conductivity σ . For normal
incidence, there is transmitted wave in the conductor that decays rapidly inside the
conductor as seen above.qFor E tangential we get from previous section, we get
σ + jωε2
Ero = 1−a
1+a Eio . Note a = jωε1 . From numbers quoted above |a| ≫ 1 and we
get Ero =∼ Eio and Eto ∼ 0. In limit a → ∞, we say we have perfect conductor
which gives Eto = 0 and Er0 = −Eio , i.e, from a perfect conductor we have perfect
reflection. The total electric field goes to zero at the interface, i.e. Eio + Ero = 0 and
Eto = 0 and Hio − Hro = 2Hio and Hto = 2Hio .
At a rate α this Hto decays to zero such that within a small width ∆ x ∼ α −1 we
drop from 2Hio to 0. But observe ∇ × H = ( jωε2 + σ )E . This gives J ∆ x = σ E ∆ x =
2Hio . We call the quantity Js = J ∆ x as the surface current density at the interface,
JS = J ∆ x = H where H = 2Hio , the magnetic field at the boundary.
Now consider the case when angle of incidence is not normal. Then we have two
cases ,
a) As in Fig. 2.5 E is tangential and we get Eio = −Ero and the normal componets
of Hio and Hro cancell and H at interface is 2Hio cos θ . The surface current density
is then JS = 2Hio cos θ .

b) As in Fig. 2.6 E is tangential, and we get Eio = Ero . The tangential componets
of Eio and Ero cancell while H = 2Hio . The surface current density is then JS = 2Hio
along y direction. Furthermore normal componets of Eio and Ero add to give us a
net normal field of 2Eio sin θ at interface. But there is no normal field inside the
conductor. If we allude to fig. 2.4b, this means there is surface charge density of
magnitude ρs = 2Eio sinthetaε2 at the interface.
In summary at conductor interface, the tangential component of E field is zero.
The normal component of H-field is zero. The tangential component of H field leads
to a surface current sensity and normal component of E field leads to a surface
charge density. In both cases the fields inside the conductor decay rapidly. It is
28 2 Electromagnetic Waves and Propagation

worthwhile writing the fields after reflection in the region 1. They arise from in-
terference of incident and reflected wave and take the form
a)

E = − j2Eio exp(− jkz sin θ ) sin(kx cos θ ) ŷ, (2.45)

H = 2Hio exp(− jkz sin θ )(cos θ cos(kx cos θ ) ẑ (2.46)
+ j sin θ sin(kx cos θ ) x̂).

b)

E = 2Eio exp(− jkz sin θ )(sin θ cos(kx cos θ ) x̂ (2.47)

+ j cos θ sin(kx cos θ ) ẑ),
H = 2Hio exp(− jkz sin θ ) cos(kx cos θ ) ŷ. (2.48)
(2.49)

2.5 Conductor Losses

∆y
∆x
∆z

z
Fig. 2.7 Fig. shows how electric field decays exponentially inside a conductor

Recall from eq. 2.44 inside a conductor, when σ ≫ ωε

2.5 Conductor Losses 29
r r
p ωµ ωµ
γ= jω µ ( jωε + σ ) = α + jβ = +j (2.50)
σ σ
Inside the conductor, electric field is E(z) = E0 exp(−γ z) as shown in Fig. 2.7.
Then current density J(z) = σ E(z) and integrating it for z from 0 to ∞ , we get
R σ E(0)
Js = J(z)dz = J(0)γ = γ = H(0). The RMS power loss is a same cubical as
shown in Fig. 2.7 is
1 1 ρ∆ x
P(z) = |I|2 R = |J(z)|2 (∆ z∆ y)2 ( ). (2.51)
2 2| {z } ∆ z∆ y
|I|2
| {z }
R
R q
ω µ0
Then P(z)dz = 21 Js2 Rs , where Rs = 2σ is termed surface resistance. Alterna-
tively, Js = J ∆ z and RMS power is

1 1 ρ∆ x 1 Js2 ∆ x∆ y
P = |I|2 R = (J ∆ z∆ y)2 ( )= . (2.52)
2 2 | {z } ∆ z∆ y 2 ∆ zσ
|I|2
| {z }
R
q
ωµ
But note 1
σ∆z = σ −1 α = 2σ = Rs . As an example if H at surface of conductor is
2 SI units and ω is 2π × 10 , then for σ = 107 SI units, we get P = .56W /m2 .
9
Chapter 3
Waveguides and Resonators

We have to calculate the eigenfunctions of Maxwell equation for eigenvalue λ = jω .

In last chapter, we did it for free space and media interface. In this chapter, we
look at free space bounded by conductors. This geometry forms what are called,
waveguides, cavities and resonators.

3.1 Parallel Plane Waveguides

Consider two parallel conductor planes as shown in Fig. 3.1 . Fig. a shows the EM
mode confined between parallel conductors with E parallel to the conductor plane.
Fig. b shows the EM mode confined between parallel conductors with H parallel to
the conductor plane. From the last chapter, after reflection from the interface I, the
E and H field in two cases take the form.
a)

E = − j2Eio exp(− jkz sin θ ) sin(kx cos θ ) ŷ, (3.1)

H = 2Hio exp(− jkz sin θ )(cos θ cos(kx cos θ ) ẑ (3.2)
+ j sin θ sin(kx cos θ ) x̂).

b)

E = 2Eio exp(− jkz sin θ )(sin θ cos(kx cos θ ) x̂ (3.3)

+ j cos θ sin(kx cos θ ) ẑ),
H = 2Hio exp(− jkz sin θ ) cos(kx cos θ ) ŷ. (3.4)
(3.5)

From 3.1 a we have, for tangential E and normal H at the interface II to be zero,
mπ
cos θ d = , (3.6)
kd

31
32 3 Waveguides and Resonators

z E
E
y

H H
x

θ θ

II θ θ
I II I
H
H

E E

a b
Fig. 3.1 Fig. a shows EM mode confined between parallel conductors with E parallel to the con-
ductor plane. Fig. b shows EM mode confined between parallel conductors with H parallel to the
conductor plane.

and β = k sin θ satisfies, we have c2 (β 2 + ( mdπ )2 ) = ω 2 from which we get

r
ω mπ 2
β = ( )2 − ( ) . (3.7)
c d
We have a cut-off frequency when β = 0. Integer m defines modes and determines
θ . Frequency ω defines how high m can be before we reach cutoff.
From 3.1 b we have, for tangential E and normal H at the interface II to be zero,
mπ
k cos θ d = , (3.8)
kd
and β = k sin θ satisfies, we have c2 (β 2 + ( mdπ )2 ) = ω 2 from which we get
r
ω mπ 2
β = ( )2 − ( ) . (3.9)
c d
We have a cut-off frequency when β = 0. Integer m defines modes and determine
θ . Frequency ω defines how high m can be before we reach cutoff.
3.2 Rectangular Waveguides 33

3.2 Rectangular Waveguides

We now consider a geometry of a hollow rectangular pipe as shown in 3.2. It is

called a rectangular waveguide. The x − y directions are called transverse, while z
direction longitudinal. When electric field is all transverse, we call it T E mode and
when magnetic field is all transverse we call it T M mode.

a
Fig. 3.2 Fig. shows a hollow rectangular pipe called a rectangular waveguide.

3.2.1 TE mode

We propose a TE mode propagating along z direction, where we have Hz and

Ex , Ey , Hx , Hy , all with a exp(− jβ z) dependence, such that using −∇E = jω µ H
and ∇H = jωε E, we get

∂ Ey
jω µ Hx = , (3.10)
∂z
∂ Ex
jω µ Hy = − , (3.11)
∂z
∂ Hz ∂ Hy
jωε Ex = − , (3.12)
∂y ∂z
∂ Hz ∂ Hx
jωε Ey = − + . (3.13)
∂x ∂z
34 3 Waveguides and Resonators

Substituting Eq. 3.10 and 3.11 in Eq. 3.12 and 3.13, we get

β2 ∂ Hz
j ω (ε − )Ex = , (3.14)
ω2µ ∂y
β2 ∂ Hz
jω (ε − 2 )Ey = − . (3.15)
ω µ ∂x
Substituting above in Eq. 3.10 and 3.11, we get

µ 2 β2 ∂ Hz
jω (ε − 2 )Hx = , (3.16)
β ω µ ∂x
µ 2 β2 ∂ Hz
jω (ε − 2 )Hy = . (3.17)
β ω µ ∂y
Observe transverse fields Ex , Ey , Hx , Hy all depend on longitudinal component Hz .
Observe

mπ x nπ y
Hz ∝ cos( ) cos( ), (3.18)
a a
mπ x nπ y
Hx ∝ sin( ) cos( ), (3.19)
a a
mπ x nπ y
Hy ∝ cos( ) sin( ), (3.20)
a a
mπ x nπ y
Ex ∝ cos( ) sin( ), (3.21)
a a
mπ x nπ y
Ey ∝ sin( ) cos( ). (3.22)
a a
(3.23)

That’s all, at boundary tangential E and normal H are zero all boundary conditions
are satisfied. The mode is written as T Emn .
From −∇E = jω µ H and ∇H =qjωε E, we get Using the fact that jω µ Hz =
∂ Ex ∂ Ey
∂y − from which we get β = ( ωc )2 − ( maπ )2 − ( nbπ )2 . Therefore mode m, n
∂z ,
q
exists only if ω ≥ c ( maπ )2 + ( nbπ )2 .

3.2.2 TM mode

We propose a TM mode propagating along z direction, where we have Ez and

Ex , Ey , Hx , Hy , all with a exp(− jβ z) dependence, such that using −∇E = jω µ H
and ∇H = jωε E, we get
3.2 Rectangular Waveguides 35

∂ Ey ∂ Ez
jω µ Hx = − , (3.24)
∂z ∂y
∂ Ex ∂ Ez
jω µ Hy = − + , (3.25)
∂z ∂x
∂ Hy
jωε Ex = − , (3.26)
∂z
∂ Hx
jωε Ey = . (3.27)
∂z
Substituting Eq. 3.26 and 3.27 in Eq. 3.24 and 3.25, we get

β2 ∂ Ez
jω (µ − )Hx = − , (3.28)
ω 2ε ∂y
β2 ∂ Ez
jω (µ − 2 )Hy = + . (3.29)
ω ε ∂x
Substituting above in Eq. 3.26 and 3.27, we get

ε 2 β2 ∂ Ez
jω (µ − 2 )Ex = , (3.30)
β ω ε ∂x
ε 2 β2 ∂ Ez
jω (µ − 2 )Ey = . (3.31)
β ω ε ∂y
Observe that transverse fields Ex , Ey , Hx , Hy all depend on longitudinal component
Ez . Observe if

mπ x nπ y
Hz ∝ sin( ) sin( ), (3.32)
a a
mπ x nπ y
Hx ∝ sin( ) cos( ), (3.33)
a a
mπ x nπ y
Hy ∝ cos( ) sin( ), (3.34)
a a
mπ x nπ y
Ex ∝ cos( ) sin( ), (3.35)
a a
mπ x nπ y
Ey ∝ sin( ) cos( ). (3.36)
a a
(3.37)

That’s all, at the boundary tangential E and normal H are zero, allq
boundary condi-
tions are satisfied. The mode is written as T Mmn . Again we get β = ( ωc )2 − ( maπ )2 − ( nbπ )2 .
q
Therefore mode m, n exists only if ω ≥ c ( maπ )2 + ( nbπ )2 .
36 3 Waveguides and Resonators

3.3 Rectangular Cavity

We considered a rectangular waveguide, which is open and indefinite in z direction.

If we close its ends and length z length be z. Then the exp(− jβ z) wave gets reflected
an the end and we have two waves ∝ exp(− jβ z) and ∝ − exp(− jβ z), adding the
two we get something ∝ sin(β z). For electric fields to go to zero at the z ends, we
have β = pcπ . The resulting mode is called T Emnp or T Mmnp depending on if electric
or magnetic field is transverse, i.e, if we have Hz or Ez . This gives mode frequency
r
mπ 2 nπ pπ
ω= ( ) + ( )2 + ( )2 . (3.38)
a b c
Writing explictly the modes for T Emnp

mπ x nπ y pπ z
Hz ∝ cos( ) cos( ) sin( ), (3.39)
a a c
mπ x nπ y pπ z
Hx ∝ sin( ) cos( ) cos( ), (3.40)
a a c
mπ x nπ y pπ z
Hy ∝ cos( ) sin( ) cos( ), (3.41)
a a c
mπ x nπ y pπ z
Ex ∝ cos( ) sin( ) sin( ), (3.42)
a a c
mπ x nπ y pπ z
Ey ∝ sin( ) cos( ) sin( ). (3.43)
a a c
Writing explicitly the modes for T Mmnp ,

mπ x nπ y pπ z
Ez ∝ sin( ) sin( ) cos( ), (3.44)
a a c
mπ x nπ y pπ z
Hx ∝ sin( ) cos( ) cos( ), (3.45)
a a c
mπ x nπ y pπ z
Hy ∝ cos( ) sin( ) cos( ), (3.46)
a a c
mπ x nπ y pπ z
Ex ∝ cos( ) sin( ) sin( ), (3.47)
a a c
mπ x nπ y pπ z
Ey ∝ sin( ) cos( ) sin( ). (3.48)
a a c

3.4 Cylindrical Cavity

Consider a cylindrical cavity of radius a and height b, where b direction is longitu-

dinal z direction.
3.4 Cylindrical Cavity 37

a
Hz E
z

Ηr b
Eθ Hθ Εr

A B
Fig. 3.3 Fig. shows a cylindrical cavity of radius a and height b.

3.4.1 TE Modes

As shown in Fig. 3.3A we consider T E mode with Hz . In cylindrical coordinates (r,

θ , z), we have

∂ ∂ sin θ ∂
= cos θ − , (3.49)
∂x ∂r r ∂θ
∂ ∂ cos θ ∂
= sin θ + . (3.50)
∂y ∂r r ∂θ
For TE modes from Eq. (3.14 and 3.15), we can write

β2 1 ∂ Hz
j ω (ε − 2
)Er = , (3.51)
µω r ∂θ
β2 ∂ Hz
j ω (ε − 2
)Eθ = − . (3.52)
µω ∂r
from Eq. (3.16 and 3.17), we get

µ 2 β2 ∂ Hz
jω (ε − 2 )Hr = , (3.53)
β ω µ ∂r
µ 2 β2 1 ∂ Hz
jω (ε − 2 )Hθ = . (3.54)
β ω µ r ∂θ
38 3 Waveguides and Resonators

By choosing Hz as function of r we get Eθ and Hr that vanishes at the boundary.

Using the fact that

∂ Ex ∂ Ey ∂ 2 Hz ∂ 2 Hz
jω µ Hz = − ∝ + ,
∂y ∂z ∂ x2 ∂ y2
which gives
From
ω2 ∂2 1 ∂
( 2 − β 2 ) Hz = ( 2 + )Hz .
c
| {z } ∂ r r ∂r
l2

The solution is J0 (lr). If Z1 , Z2 , Zn are zeros of the J0′ (x) (other than trivial zeros
at 0)Then modes are defined as la = Zi . This ensures that Eθ and Hr that vanishes
at the boundary. q
2
This defines β 2 = ( ωc2 − ( Zai )2 . We have a wave along z with dependence
exp(− jβ x). Superimposing on it a reflected wave exp(− jβ x) we get electric field
has z dependence sin(β z) and magnetic field cos(β z). By choosing β = pbπ , we get
a standing wave mode indexed by p and Zi , with
r
pπ Zi
ω = c ( )2 + ( )2 , (3.55)
b a
The resulting fields are shown in Fig. 3.3A.
We choose Hz as function of r only. This is not the only solution. We can choose

Hz = f (r) cos nθ . (3.56)

Then we get

ω2 ∂2 1 ∂ 1 ∂2
− ( 2 − β 2 ) Hz = ( 2 + + 2 )Hz .
| c {z } ∂r r ∂r r ∂θ2
l2

n2 ∂2 1 ∂
(−l 2 + 2
) f (r) = ( 2
+ ) f (r).
r ∂r r ∂r
The solution is Jn (lr). If Z1 , Z2 , . . . , Zn are zeros of the Jn′ (x) (other than trivial
zeros at 0)Then modes are defined as la = Zi . This ensures that Eθ and Hr that
vanishes at the boundary.
q
2
This defines β 2 = ( ωc2 − ( Zai )2 . We have a wave along z with dependence
exp(− jβ z). Superimposing on it a reflected wave exp(− jβ z) we get electric field
has z dependence sin(β z) and radial magnetic field cos(β z). By choosing β = pbπ ,
we get a standing wave mode indexed by p and Zi , with
r
pπ Zi
ω = c ( )2 + ( )2 . (3.57)
b a
3.4 Cylindrical Cavity 39

The resulting fields are shown in Fig. 3.3A.

3.4.2 TM modes

As shown in Fig. 3.3B we consider T M mode with Ez . For TM modes from Eq.
(3.30 and 3.31), we can write

β2 1 ∂ Ez
jω (µ − )Hr = , (3.58)
εω 2 r ∂θ
β2 ∂ Ez
jω (µ − )Hθ = . (3.59)
εω 2 ∂r
from Eq. (3.28 and 3.29), we get

ε 2 β2 ∂ Ez
jω (µ − 2 )Er = , (3.60)
β ω ε ∂r
ε 2 β2 1 ∂ Ez
jω (µ − 2 )Eθ = . (3.61)
β ω ε r ∂θ

By choosing Ez as function of r we get Eθ and Hr that vanishes at the boundary.

Using the fact that

∂ Ey ∂ Ex ∂ 2 Ez ∂ 2 Ez
jωε Ez = − ∝ + ,
∂x ∂y ∂ x2 ∂ y2
which gives
From
ω2 ∂2 1 ∂
( 2 − β 2 ) Ez = ( 2 + )Hz .
c
| {z } ∂ r r ∂r
l2

The solution is J0 (lr). If Y1 ,Y2 ,Yn are zeros of the J0 (x). Then modes are defined
as la = Yi . This ensuresqthat Ez vanishes at the boundary.
2
This defines β 2 = ( ωc2 − ( Yai )2 . We have a wave along z with dependence
exp(− jβ x). Superimposing on it a reflected wave exp(− jβ x) we get radial elec-
tric field has z dependence sin(β z), z electric field has dependence cos(β z) and
magnetic field cos(β z). By choosing β = pbπ , we get a standing wave mode indexed
by p and Yi , with
r
pπ Yi
ω = c ( )2 + ( )2 . (3.62)
b a
The resulting fields are shown in Fig. 3.3B. We choose Ez as function of r only.
This is not the only solution. We can choose
40 3 Waveguides and Resonators

We choose
Ez = f (r) cos nθ . (3.63)

∂ Ey ∂ Ex ∂ 2 Ez ∂ 2 Ez
jωε Ez = − ∝ + ,
∂x ∂y ∂ x2 ∂ y2
which gives
From

ω2 ∂2 1 ∂ 1 ∂2
( 2 − β 2 ) Ez = ( 2 + ++ 2 )Ez .
| c {z } ∂r r ∂r r ∂θ2
l2

n2 ∂2 1 ∂
(−l 2 + ) f (r) = ( + ) f (r).
r2 ∂ r2 r ∂ r
The solution in the annulus is Jn (lr). If Y1 ,Y2 , . . . ,Yn are zeros of the Jn (x) (other
than trivial zeros at 0)Then modes are defined as la = Yi . This ensures that Eθ and
Hr that vanishes at the boundary.
q
2
This defines β 2 = ( ωc2 − ( Yai )2 . We have a wave along z with dependnece
exp(− jβ z). Superimposing on it a reflected wave exp(− jβ z) we get transverse
electric field has z dependence sin(β z) and transverse magnetic field cos(β z). By
choosing β = pbπ , we get a standing wave mode indexed by p and Yi , with
r
pπ Yi
ω = c ( )2 + ( )2 . (3.64)
b a

3.5 Quality Factor of Resonator

We talked about cavities and their modes, say mode φ with eigenvalue λ = jω . The
mode will just evolve as exp( jω t)φ (r). It means the mode will live in the cavity for
ever. That is not true. The mode corresponds to J on the walls which means resistive
losses. This means, we really have λ = −α + jω , where λ is a decay constant. How
do we calculate α . We can calculate how much power P is dissipated on the walls
using the formula P per unit area as 21 Js2 Rs and we know the energy E of the mode
decreases by ∆ E = P∆ t in unit time ∆ t. If A is amplitude of mode then in time ∆ t
we have
∆A
= α∆ t. (3.65)
A
Energy E ∝ A2 and change of energy ∆ E = P∆ t ∝ 2A∆ A ,which gives 2α = EP .
We define the quality factor Q of the cavity as
α
Q= . (3.66)
ω
Chapter 4
Antennas

4.1 Vector Potential

Fig. 4.1 Fig. shows a point current source called Hertz dipole.

In this chapter, we study Maxwell equations, when they are driven by a current
source also called an antenna [16, 19, 20, 24, ?]. Recall in the presence of current
source, the Maxwell equations take the form

∂H
µ = −∇ × E, (4.1)
∂t
∂E
ε = ∇ × H − J(r,t), (4.2)
∂t
where each spatial location r, indices a vector entry in the linear system as in Eq.
( 2.16). The vector entries are E and B fields. The goal is to calculate E(r,t) and
B(r,t) resulting from an current input J(r,t). That is solve a linear system (2.16) for
an input u(t). For instance an input of the type u(t) = exp( jω t). Earlier we said one

41
42 4 Antennas

way to approach this is to find modes of the system and corresponding eigenvalues.
That is we have to solve for Ax = λ x in Eq. ( 2.17), i.e solve for

1
− ∇ × E = λ H, (4.3)
µ
1
∇ × H = λ E. (4.4)
ε
In this section, we adopt a more direct approach to solve for the Maxwell equa-
tions given antenna excitation. The antenna we consider is pint current source, with
current density J0 located at origin with current in z direction as shown in fig. 4.1
. It has x-y area Ae and z length dl such that Ae J = i0 . We first observe that since
∇ · B = 0, means it can be written as B = ∇ × A. From 4.1, we get

∂A
∇×( + E) = 0, (4.5)
∂t
which implies that

∂A
+ E = −∇V, (4.6)
∂t
∂A
E=− − ∇V. (4.7)
∂t
We still have some independence in the choice of V and A that result in same E
and B V → V − ∂∂φt and A → A + ∇φ doesnot change E and B. We call these Gauge
degrees of freedom. We use these degree of freedom to choose A,V such that

∂ V ∂ Ax ∂ Ay ∂ Az
+ + + = 0. (4.8)
∂t ∂x ∂y ∂z

This is called a Lorentz gauge. This gives from 4.2 that for i ∈ {x, y, z}.

∂ 2 Ai
∇2 Ai − = µ0 Ji (x,t). (4.9)
c2 ∂ t 2
Recall we assume current density is in the z direction and we take a point current
source stationed at the origin such that Jz (x) is constant at J(t) over a infinitesimal
volume we take it as a J(t) = exp( jω t)J0 , as shown in fig. 4.2a. The current source
is also called a Hertz dipole.

∂ 2 Az
− ∇2 Az − = µ0 Jz (x,t). (4.10)
c2 ∂ t 2
We first solve the above equations when current source is stationary and solve
for
4.1 Vector Potential 43

−∇2 Az = µ0 Jz (x). (4.11)

C
We can see that Az (r) = satisfies the equation ∇2 Az = 0. How do we determine
r
C, we get ∇2 Az = ∇ · (∇Az ), where ∇Az = − Cr2r̂ . Using stokes theorem, we get
−∇Az (4π r2 ) = µ0 J0V0 , which gives C = µ04Jπ0V0 and we get
µ0 J0V0
Az (r) =
ẑ. (4.12)
4π r
Now we claim that in presence of time varying J(t), we have

µ0 J(t − cr )V0
Az (r,t) = ẑ. (4.13)
4π r
It is just verified that Eq. (4.13) satisfies (4.10). Writing J0V0 = idl where i is the
current and dl the length of the current element we can write [3]

µ0 i(t − cr )dl
Az (r,t) = ẑ. (4.14)
4π r

Az (r)
Az (r)
r

+
−

a b
Fig. 4.2 Fig. a and b show vector potential A due to Hertz dipole.

We can calculate B by using B = ∇ × A this gives using i(t − cr ) = i0 exp( jω (t −

r
where k = ωc .
c )) = i0 exp( j ω (t − kr)),
44 4 Antennas
 
µ0 dl i(t − cr ) i′ (t − cr )
B(r,t) = + φ̂ , (4.15)
4π r2 cr
 
µ0 i0 dl exp(− jkr) sin θ 1
B(r,t) = exp( jω t) + jk φ̂ . (4.16)
4π r r
| {z }
B(r)

Thus we get steady state value B(r). To get steady state value E(r) observe

jω µ H(r) = −∇ × E(r), (4.17)

jωε E(r) = ∇ × H(r) − J(r). (4.18)

From (4.18) we have for r away from origin

j
ω E = ∇ × B, (4.19)
c2
which gives

   
c2 µ0 i0 dl exp(− jkr) 3k 3 j 2k 2 j
E(r,t) = exp( jω t) { − jk2 + − 2 sin θ θ̂ + − 2 ẑ} .
4π rω r r r r
| {z }
E(r)
(4.20)
At large distances r ≫ λ , we only keep terms of order r−1 giving,

j k µ0 i0 dl exp(− jkr) sin θ

B(r) = φ̂ , (4.21)
4π r
− j k2 i0 dl exp(− jkr) sin θ
E(r) = θ̂ . (4.22)
4π rωε

4.1.1 Dipole Radiation

We considered a current source and radiation due to that. Now consider an oscillat-
ing dipole, where two opposite charges with charge q go back and forth along say z
axis with a maximum separation of d. The dipole moment

D(t) = qd cos(ω t). (4.23)

dD
This is as shown in Fig. (4.2b), where dt constitutes a current. For simplicity take

D(t) = qd exp( jω t). (4.24)

4.2 Radiated Power and Antenna Resistance 45

Then everything we did for point current source carries over and we make the cor-
respondence i0 dl = jqd ω . Then in the expression for Az (r,t), B(r,t), E(r,t) in Eq.
(4.14, 4.21, 4.22), we just substitute

i0 dl = jqd ω . (4.25)
This gives us radiation due to an oscillating dipole.

4.2 Radiated Power and Antenna Resistance

Fig. 4.3 Figure shows power outflux from a enclosed volume V and surface S.

Consider an enclosed volume V as in 4.3. The volume contains electric and mag-
netic fields and the enclosed energy is
Z
1
E = ε0 |E|2 + µ0 |H|2 . (4.26)
2 V
Then the change of energy with time
46 4 Antennas
Z
∂E ∂E ∗ ∂H
= Re(ε0 · E + µ0 · H ∗ ), (4.27)
∂t ∂t ∂t
ZV
= Re(∇ × H · E ∗ − ∇ × E · H ∗ ), (4.28)
V
Z
1
= Re(∇ × (H ∗ × E)), (4.29)
2 V
Z
1
= Re(H ∗ × E).dA, (4.30)
2 S

where we use Maxwell equations to substitute for ∂∂Et , ∂∂Ht . We use the identity a ·
(b × c) = 21 (c · (a × b) − b · (a × c)). The last equation uses Stoke’s theorem. The
quantity
1
P = Re(E × H ∗ ), (4.31)
2
denotes power leaving the surface per unit area. It is also called Poynting vector.
If we integrate P for E and B as in Eq. (4.21 and 4.22) for a Hertz dipole over a
large sphere, we get

1 (ki0 dl)2
P= = i20 R. (4.32)
2 6π cε
where

(kdl)2
R= . (4.33)
6π cε
q
If dl = λ /2, then R ∼ µε0o ∼ 50 SI units. It is called radiation resistance of antenna.
The power loss due to radiation is seen as a resistance by antenna circuit.

4.3 Antenna Arrays

Now consider an array of Hertz dipoles as shown in Fig. (4.4). Let total be N + 1.
Let d be the separation between them. Then the Az (r,t), B(r,t), E(r,t) due to each
dipole carries a factor ∝ exp(ikr). Where r between successive dipoles differs by
∆ r = d cos φ as shown in Fig. (4.4). This phase factor between successive dipoles
differs by a factor Z = exp(ikθ ), where θ = dcosφ = d sin υ , where υ = π2 − φ . Then
adding all the phase factors for a point far than size of array we get for p = kd cos φ ,
N Np
(N − 1)p sin 2
η= ∑ Z k = exp( j 2
)
sin 2p
. (4.34)
k=0

sin N p
The factor f (p) = sin 2p gives the dependency of |η | as function of p. It peaks at
2
p = 0 where its value is N. Then
4.3 Antenna Arrays 47

∆r
φ
d

Fig. 4.4 Fig. shows a 1D array of Hertz dipoles with a separation d between them.

|η | sin N2p Np
=| p | ∼ | sinc( )|, (4.35)
N N sin 2 2
where | sinc(p)| is as in fig. 4.5. Few lobes down from origin the function sig-
nificantly diminishes. Lets say the array is 1000 wavelengths long which means
N p = 1000 cos φ . If we are ten lobes down means N2p = 10π or cos φ = 50 π
or
π
sin υ = 50 a small angle. The antenna array fires vertically up as in A in fig 4.6.
Now suppose we introduce a phase difference in the current in the succes-
sive elements of array the form exp(iδ ), then writing δ = −kd sin υ0 we get
p = kd(sin υ − sin υ0 ). Now p = 0 when υ = υ0 . Therefore this array will not fire
vertically but rather at an angle υ0 . When υ0 = π2 , we get an array that fires at the
end called an endfire array as as in B in fig 4.6.

4.3.1 2D arrays

Fig. 4.7 A shows a 2D array of Hertz dipoles. Fig. 4.7 B shows the array laid on
x − y plane and beam direction making angle θ with z axis and φ with x axis. Let d
be spacing between dipoles. let
48 4 Antennas

f(x)

Fig. 4.5 Fig. depicts the function f (x) = | sinc(x)|

Fig. 4.6 Fig. depicts how antenna array fires vertically A or horizontally B depending on phase
difference of the current in array elements.

p1 = d sin θ cos φ , (4.36)

p2 = d sin θ sin φ . (4.37)

Z1 = exp( j p1 ) and Z2 = exp( j p2 ). Then adding phases over antennas we get

4.3 Antenna Arrays 49

A B
Fig. 4.7 Fig. A shows a 2D array of Hertz dipoles. Fig. B shows the array laid on x − y plane and
beam direction making angle θ with z axis and φ with x axis.

Np Np
(N − 1)p1 + (N − 1)p2 sin 2 1 sin 2 2
η = ∑ Z1k Z2l = exp( j ) . (4.38)
k,l 2 sin p21 sin p22

Lets say the array is 1000 × 1000 wavelengths in area which means N p1 =
π
1000 sin θ cos φ and N p2 = 1000 sin θ sin φ . If we have sin θ = 50 , then one of the
N p1 N p1
2 or 2 is ∼ 10π or ten lobes down and η diminishes significantly. Therefore all
intensity is concentrated in θ very small, antenna array fires vertically up.

4.3.2 Antennas and finite interfaces

In previous chapters, we studied how EM waves reflect of conducting interface. For

ease of calculation, we took the conductor plane to be infinite. How about finite
conductor interfaces as in mirrors. To see how mirror reflects EM waves say light
(optical frequency 1015 Mhz), which is sub-micrometer wavelength. Imagine an
infinite plane. Incident light induces current density JS on the plane, which acts as
an antenna array. If we partition the plane into say rectangles of say 1000 × 1000
wavelengths in area (this means mm2 sub-blocks for light) then these rectangles will
fire vertically discussed in previous section and shown in fig. 4.8A. Then suppose the
real interface we are given is say region II in fig. 4.8A. Then if we remove remaining
regions, one by one, we find that since their respective antenna did not interfere (at
moderate distances upto meter) with the radiation of region II, the radiation in the
vertical line of sight of region II is uninterrupted. All we are saying is that pretend
the plane is infinite, calculate, the reflected wave and only take the reflected wave in
vertical line of sight of region II. That is to say we can draw ray diagrams as 4.8B,
which is starting point of geometric optics. Instead of light if we had radio waves
50 4 Antennas

then a mm2 size reflector will act as a point antenna and send waves in all directions.
Then we cannot follow rules of geometric optics.

II

III

IV

A
Fig. 4.8 Fig. A shows reflection of a conducting plane as finite size antenna arrays firing vertically.
Fig. B shows ray diagram for incident and reflected light.
Chapter 5
Transmission Lines

5.1 Coaxial Transmission Lines

B
a E

B
A

a
E B
+
E
−
C D
Fig. 5.1 Fig. A shows a coaxial cable. Fig. B shows a cross-section with radial E and circular B
fields. Fig. C shows how radial E gives capacitance. Fig. D shows how circular B gives inductance.

In the previous chapter we taked about a special kind of waveguide, a hollow

pipe. In this chapter we talk of a hollow pipe containing a conductor inside it as
shown in Fig. 5.1A. This is called a coaxial cable [11, 12] , or just coax. Such a
structure supports a novel EM mode called the TEM mode, which was absent in

51
52 5 Transmission Lines

a hollow pipe. TEM means transverse electric and magnetic. The electric field is
radial and magnetic field circular as shown in Fig. 5.1B, which shows cross section
of coax.
The radial electric field produces electric charge Q of opposite polarity in the
inner and outer conductors such that resuting voltage V (x) = E(x)a can be related
to the charge Q. Then we can relate V and Q as V (x)C∆ x = Q(x) where C is the
capacitance per unit length.
Further more (E(x + ∆ x) − E(x))a = − ∂∂Bt Ar , where Ar = a∆ x is the area. But
BAr = L∆ xI, where L is the inductance per unit length and I(x) is the current in the
inner conductor. This is written as

i(x)
V(x) V(x +∆ x)

B v Z
l

Fig. 5.2 Fig. A shows electrical equivalent of a transmission line with distributed inductors and
capacitors. Fig. B shows voltage source at one end of the line (called generator end) and load at
other end.

dI
V (x + ∆ x) −V (x) = L∆ x , (5.1)
dt
∂V dI(x)
− =L . (5.2)
∂x dt
Similarly
5.1 Coaxial Transmission Lines 53

dQ(x)
I(x) − I(x + ∆ x) = , (5.3)
dt
dQ(x) dV (x)
= C∆ x , (5.4)
dt dt
∂I dV (x)
− =C . (5.5)
∂x dt
Thus we get two equations of the transmission line

∂V dI(x)
− =L , (5.6)
∂x dt
∂I dV (x)
− =C . (5.7)
∂x dt
Now we drive the transmission line with a voltage source V = V0 exp( jω t), see
fig. 5.2B. Then that gives steady state values V (x) exp( jω t) and I(x) exp( jω t) which
satisfy,

∂V
− = jω LI(x), (5.8)
∂x
∂I
− = jω CV (x). (5.9)
∂x
See fig. 5.2B. Where we take load at right and voltage generator at left. Lets take
x going from right to left. Then above equations read,

∂V
= jω LI(x), (5.10)
∂x
∂I
= jω CV (x). (5.11)
∂x

∂ 2V
= −ω 2 LCV (x) = −γ 2V (x), (5.12)
∂ x2
√ q
where γ = jβ where β = ω LC and let Z0 = CL .

V (x) = V+ exp( jβ x) +V− exp(− jβ x), (5.13)

1
I(x) = (V+ exp( jβ x) −V− exp(− jβ x)) (5.14)
Z0
1
= (I+ exp( jβ x) − I− exp(− jβ x)). (5.15)
Z0
54 5 Transmission Lines

5.2 Impedance
V−
Then for Γ = V+ ,

V (0) 1+Γ
Z(0) = ZL = = Z0 , (5.16)
I(0) 1−Γ
V (x) 1 + Γ exp(−2 jβ x)
Z(x) = = Z0 . (5.17)
I(x) 1 − Γ exp(−2 jβ x)
Zl −Z0
Using Γ = Zl +Z0 , we get

ZL cos β x + jZ0 sin β x

Z(x) = Z0 . (5.18)
Z0 cos β x + jZL sin β x
Until now we assumed perfect conductors and the dielectric seprating them as
perfect insulator. In practice conductor has small resistance R per unit length and
dielectric small admittance G per unit length.
Thus we get two equations of the transmission line change to

∂V dI(x)
− =L + RI, (5.19)
∂x dt
∂I dV (x)
− =C + GV. (5.20)
∂x dt
With steady state,

∂V
− = ( jω L + R)I(x), (5.21)
∂x
∂I
− = ( jω C + G)V (x). (5.22)
∂x
See fig. 5.2B. Where we take load at right and voltage generator at left. Lets take
x going from right to left. Then above equations read,

∂V
= ( jω L + R)I(x), (5.23)
∂x
∂I
= ( jω C + G)V (x). (5.24)
∂x

∂ 2V
= γ 2V (x), (5.25)
∂ x2
p q
jω L+R
where γ = ( jω L + R)( jω C + G) = α + jβ and let Z0 = jω C+G .
5.2 Impedance 55

To fix some typical numbers, R = .01Ω /m, L = .01µ H/m, C = 100pF/m and
G = .01℧/m giving at 2 GHz, Z0 = 10 + j.04Ω .

V (x) = V+ exp(γ x) +V− exp(−γ x), (5.26)

1
I(x) = (V+ exp(γ x) −V− exp(−γ x)) (5.27)
Z0
= (I+ exp(γ x) − I− exp(−γ x)). (5.28)
V−
Then for Γ = V+ ,

V (0) 1+Γ
Z(0) = ZL = = Z0 , (5.29)
I(0) 1−Γ
V (x) 1 + Γ exp(−2γ x)
Z(x) = = Z0 . (5.30)
I(x) 1 − Γ exp(−2γ x)
Zl −Z0
Using Γ = Zl +Z0 , we get
ZL cosh γ x + Z0 sinh γ x
Z(x) = Z0 . (5.31)
Z0 cos γ x + ZL sinh γ x
For low loss transmission line where γ = jβ , we get Z(x) as in Eq. (5.18). There
is no loss in the transmission line then the power transferred to the load is as follows.

Z0
Z0

v ZL
v ZL

A B
Fig. 5.3 Fig. A shows a voltage source V with impedance Z0 , a transmission line and a load ZL .
Fig. B shows load as seen from generator end of transmission line.
56 5 Transmission Lines

5.3 Impedance Matching and Smith Charts

Consider a voltage source V with source impedance Z0 . It is attached to transmission

line with characteristic impedance Z0 and a load at the end of transmission line ZL .
We want to know what should be the value of load impedance ZL so that there is
maximum power transferred to load.
Then voltage Ve at the end of the transmission line is

Ve = V+ +V− = V+ (1 + Γ ). (5.32)
Then voltage Vb and current ib at the beginning of the transmission line is

Vb = V+ exp( jβ l) +V− exp(− jβ l), (5.33)

and
1
ib = (V+ exp( jβ l) −V− exp(− jβ l)). (5.34)
Z0

Vb = V0 − ib Z0 , (5.35)
giving
V0
V+ = exp(− jβ l). (5.36)
2
The power transferred is P = Re(Ve Ie∗ ), which gives

|V+ |2
P= (1 − |ΓL |2 ), (5.37)
Z0
where

1 − ZZL0
Γ= . (5.38)
1 + ZZL0
For maximum transfer Γ = 0 or ZL = Z0 .
what if ZL 6= Z0 , what should we do. We can move back from the load towards
the voltage source. What we show is after some distance x the resistance Z(x)/Z0
becomes 1 + jv . We can arrange to cancel the reactive part v by using an additional
piece of transmission line leaving the resulting load Z(x) = Z0 which allows for
maximum power transfer.
We can graphically find this distance x using what is called a smith chart. Given

Z(x) 1 + Γ
r + jx = = . (5.39)
Z0 1−Γ
Then for Γ = u + jv, we can plot the locus of Γ for fixed r and fixed x. These are
called constant resistance and reactance circles. These are given by
5.3 Impedance Matching and Smith Charts 57

r 2
(u − ) + v2 = (r + 1)−2 (5.40)
r+1
r 1
a circle centered at ( r+1 , 0) with radius r+1 . This is called constant resistance circle.
These are depicted in 5.4A, with smaller circle corresponding to larger r.
Similarly constant reactance circle is
1 1
(u − 1)2 + (v − )2 = 2 , (5.41)
x x
which is centered at (1, 1x ) with radius 1
|x| . These are depicted in 5.4B, with smaller
circle corresponding to larger x.

A B
Fig. 5.4 Fig. shows constant resistance A and reactance circle B.

We start with a given load corresponding to given (r, x). The intersection of con-
stant r and x circles uniquesly determine Γ . As we move back on transmission line
Γ (x) → Γ exp(− j2β x) that is we rotate Γ counterclockwise keeping its magnitude
constant, till it hits the constant resistance r = 1 circle. We stop and note the reac-
tance value there and this is the reactance we need to annul. All this is depicted in
5.5. Where we located the normalized load r + jx as intesection of constant resis-
tance circle r (shown in solid) and constant reactance circle x (shown in dashed) as
point A. This point is rotated on cirlce centered at origin (shown in dotted) till we hit
constant unit resistance circle (shown in bold) at point B. We can read the reactance
of point B. The angle θ of rotation of A to B is given as Θ = 2β x.
58 5 Transmission Lines
x
A

r
1

Fig. 5.5 Fig. shows how to calculate using a Smith chart the distance travelled up a transmission
line to reach unity normalized resistance.

How do we cancel the reactance. One way is to add a segment transmission line
in parallel (shunt) as shown in 5.7A.
The admittance of the shunt adds to admittance of the load at x. With Y (x) =
Z −1 (x) and Y0 = Z0−1 we find for Γ ′ = −Γ , we have

Y (x) 1 + Γ ′
g + jy = = . (5.42)
Y0 1−Γ ′
Now everything is same proceduraly as before. We start with a given load corre-
sponding to admittance g + jy. The intersection of constant g and y circles uniquesly
determine Γ ′ . As we move back on transmission line Γ ′ (x) → Γ ′ exp(− j2β x) that
is we rotate Γ ′ counterclockwise keeping its magnitude constant, till it hits the con-
stant conductance g = 1 circle. We stop and note the susceptance value there and
this is the susceptance we need to annul. All this is same as depicted in 5.5. Except
now our starting point A is normalized admittance which is different from normal-
ized impedance. The susceptance at point B is removed by adding a transmission
line in shunt.
What is the impedance and admitance of a transmission line pf length l. If it is
closed at other end as in fig. 5.7B, then it means ZL = 0 and from Eq. (5.18), Z(l) =
j tan β l or Y (l) = − j cot β l. If it is open at other end as in fig. 5.7C, then it means
ZL = ∞ and from Eq. (5.18), Z(l) = − j cot β l or Y (l) = j tan β l. Therefore to cancel
a positive susceptance, we should use closed-ended transmission line of 0 ≤ l ≤ λ4
and to cancel a negative susceptance, we should use open-ended transmission line
of 0 ≤ l ≤ λ4 .
Observe Γ (x) repeats every 2β x = 2π i.e., every x = λ2 . Therefore load resistance
is the same when we step back x = λ2 . Infact the maximum voltage on transmission
5.3 Impedance Matching and Smith Charts 59

Fig. 5.6 Fig. shows a smith chart

60 5 Transmission Lines

ZL B

A C

Fig. 5.7 Fig. A shows how impedance matching is done by adding admittance in parallel to trans-
mission line also called stubs. Fig. B shows a short circuited transmission line. Fig. C shows a open
circuited transmission line.

Vmax = V+ (1 + |Γ |), (5.43)

and

Vmin = V+ (1 − |Γ |). (5.44)

1+|Γ |
and the ratio VVmax
min
= is called VSWR, voltage standing wave ratio. When
1−|Γ |
VSWR = 1, implies Γ = 0 and there is no reflected wave. To check if we have
matched the load perfectly we measure VSWR to find if VSWR = 1.
Chapter 6
Rf and Microwave Engineering

6.1 Microwave Oscillators

We first discuss how to produce microwaves (3 − 300 GHz) frequencies. These are
produced in cavities called microwave oscillators. The basic principle is to excite
the modes of a cavity resonantly by motion of an electron. It is discusses in the
following.

6.1.1 Coaxial Cylindrical Cavity

H
r

Eφ
Hz

A B
Fig. 6.1 Fig. A shows a cylindrical cavity with inner and outer conductors. Fig. B shows its cross-
section containing TE mode.

61
62 6 Rf and Microwave Engineering

Consider a cylindrical cavity as shown in fig. 6.1A. Fig. 6.1A shows cross sec-
tion. The cavity has an outer conductor of radius r2 and an inner conductor of radius
r1 with r2 − r1 = a. The cavity supports both T E and T M modes. We discuss both.

6.1.2 TE Modes

In cylindrical coordinates (r, θ , z), we have

∂ ∂ sin θ ∂
= cos θ − , (6.1)
∂x ∂r r ∂θ
∂ ∂ cos θ ∂
= sin θ + . (6.2)
∂y ∂r r ∂θ
For TE modes from Eq. (3.14 and 3.15), we can write

β2 1 ∂ Hz
j ω (ε − 2
)Er = , (6.3)
µω r ∂θ
β2 ∂ Hz
j ω (ε − 2
)Eθ = − , (6.4)
µω ∂r
from Eq. (3.16 and 3.17), we get

µ 2 β2 ∂ Hz
jω (ε − 2 )Hr = , (6.5)
β ω µ ∂r
µ 2 β2 1 ∂ Hz
jω (ε − 2 )Hθ = . (6.6)
β ω µ r ∂θ

We choose
Hz = f (r) cos nθ (6.7)
Then we get

ω2 ∂2 1 ∂ 1 ∂2
− ( 2 − β 2 ) Hz = ( 2 + + 2 )Hz .
| c {z } ∂r r ∂r r ∂θ2
l2

n2 ∂2 1 ∂
(−l 2 + 2
) f (r) = ( 2 + ) f (r).
r ∂r r ∂r
The solution in the annulus is Jn (lr + o). If Z1 , Z2 , . . . , Zn are zeros of the Jn′ (x)
(other than trivial zeros at 0)Then mode lm is defined as
6.2 TM Modes 63

lr1 + o = Zl , (6.8)
lr2 + o = Zm . (6.9)

This ensures that Eθ and Hr that vanishes at the boundary. We solve for l = Zm a−Zl .
q
This defines β 2 = ( ωc2 − ( Zm a−Zl )2 . We have a wave along z with dependence
2

exp(− jβ z). Superimposing on it a reflected wave exp(− jβ z) we get transverse

electric field has z dependence sin(β z) and transverse magnetic field cos(β z). By
choosing β = pbπ , we get a standing wave mode indexed by p and Zm , Zl , with
r
pπ Zm − Zl 2
ω = c ( )2 + ( ) . (6.10)
b a

6.2 TM Modes

For TM modes from Eq. (3.30 and 3.31), we can write

β2 1 ∂ Ez
jω (µ − 2
)Hr = , (6.11)
εω r ∂θ
β2 ∂ Ez
jω (µ − )Hθ = , (6.12)
εω 2 ∂r
from Eq. (3.28 and 3.29), we get

ε 2 β2 ∂ Ez
jω (µ − 2 )Er = , (6.13)
β ω ε ∂r
ε 2 β2 1 ∂ Ez
jω (µ − 2 )Eθ = . (6.14)
β ω ε r ∂θ

We choose
Ez = f (r) cos nθ . (6.15)

∂ Ey ∂ Ex ∂ 2 Ez ∂ 2 Ez
jωε Ez = − ∝ + ,
∂x ∂y ∂ x2 ∂ y2
which gives
From

ω2 ∂2 1 ∂ 1 ∂2
( 2 − β 2 ) Ez = ( 2 + ++ 2 )Ez .
| c {z } ∂r r ∂r r ∂θ2
l2

n2 ∂2 1 ∂
(−l 2 + 2
) f (r) = ( 2 + ) f (r).
r ∂r r ∂r
64 6 Rf and Microwave Engineering

The solution in the annulus is Jn (lr + o). If Y1 ,Y2 , . . . ,Yn are zeros of the Jn (x)
(other than trivial zeros at 0)Then mode lm is defined as

lr1 + o = Yl , (6.16)
lr2 + o = Ym (6.17)

This ensures that Eθ and Hr that vanishes at the boundary. We solve for l = Ym a−Yl .
q
This defines β 2 = ( ωc2 − ( Ym a−Yl )2 . We have a wave along z with dependence
2

exp(− jβ z). Superimposing on it a reflected wave exp(− jβ z) we get transverse

electric field has z dependence sin(β z) and transverse magnetic field cos(β z). By
choosing β = pbπ , we get a standing wave mode indexed by p and Ym ,Yl , with
r
pπ Ym −Yl 2
ω = c ( )2 + ( ) . (6.18)
b a

6.2.1 Magnetron Oscillator

A B

Fig. 6.2 Fig. A shows the cross-section of a cylindrical cavity with electron circulating the space
between inner and outer conductor. Fig. B shows how inner cylindrical cavity is aperture coupled
to outer cavities with same resonant frequency.

In fig. 6.2A, we have a coaxial cylindrical cavity as shown. Consider T E mode

φ that has angular dependence cos θ and frequency ω0 . If we introduce an elec-
tron in the cavity and a magnetic field B0 along the z direction. Then this elec-
tron will presses around the magnetic field at the angular velocity ω = eBm . If we
choose B0 such that ω = ω0 , then the electron orbits in the cavity at angular fre-
6.2 TM Modes 65

quency ω0 . This comprises a circulating current density that is nonzero at angu-

lar position θ = ω t and hence its overlap with the T E mode (Eφ part of it) is of
the form cos ω0t. Thus we have a mode at frequency ω0 and we are exciting it as
cos ω0t = exp( jω0 t)+exp(−
2
jω0 t)
. Thus we excite a linear system at resonance, we will
build up this mode φ . Rapidly the cavity will be filled up with this mode and if ω0
is in the GHz range we produce microwaves.
We talked about the T E mode. But we also have T M modes. Let φ1 be a TM
mode that has angular dependence Eφ ∝ sin θ and some frequency ω0 . If we intro-
duce an electron in the cavity and a magnetic field B0 along Z direction. Then this
electron will presses around the magnetic field at the angular velocity ω = eB m . If
we choose B0 such that ω = ω0 , then the electron orbits in the cavity at angular
frequency ω0 . This comprises a circulating current density that is nonzero at an-
gular position θ = ω t and hence its overlap with the T M mode (Eφ part of it) is
of the form sin ω0t. Thus we have a mode at frequency ω0 and we are exciting as
sin ω0t = exp( jω0 t)−exp(−
2j
jω0 t)
. Thus we excite a linear system at resonance, we will
build up this mode φ . Rapidly the cavity will be filled up this mode and if ω0 is in
GHZ range we produce microwaves.
Thus we are saying that if the precession frequency of electron matches the fre-
quency of either the T E and T M mode with cos θ dependence we will excite these
modes. We also have modes with angular dependence cos nθ and frequency ω0 .
Then it is excited with electron with precession frequency ωn0 . The electron is in-
jected into cavity from inner conductor that acts as cathode. The electrons are in-
jected radially outwards and under the Lorentz force of the magnetic field acquire a
circular orbit as shown in fig. (6.2)A. The strength of magnetic field determines the
angular velocity ω as ω = eB m.
Fig. (6.2)B shows cartoon of a microwave oscillator called Magnetron. The inner
conductor is a cathode that injects electrons. The Central conductor is anode that
attracts these electrons radially and makes them acquire a velocity which under
magnetic field gets turned into a circular orbit. Between center conductor and outer
conductor are circular chambers/cavities with resonate at frequency ω0 and can store
microwaves. Without these cavities, the volume of central cavity is small, and as we
fill the cavity with microwaves it offers increased resistance to motion of electron.
Observe the circulating electron creates mode −φ and resulting Eφ opposes electron
motion. By adding more room, in terms of outer cavities, we store same energy a
lower Eφ value and reduced resistance. Thus we have an arrangement where we
have central cavity as in Fig. (6.2)A aperture coupled to outer cavities all of which
resonate at same frequency. We excite the central cavity by cyclotron motion of
electron, the outer cavities provide more storage room.

6.2.2 Klystron and Gyrotron

In magnetron, we have a cylindrical mode with costheta dependence and frequency

ω0 . We traverse the mode with an electron rotating circularly at angular frequency
66 6 Rf and Microwave Engineering

B
B

Fig. 6.3 Fig. a and B. show cylindrical cavity being resonantly excited by linear and gyrating
motion of electron respectively.

ω0 , which excites the mode resonantly as cos ω0t. Now consider cylindrical cavity
again (without central conductor). We have TM modes in the cavity with Ez having
cos kz dependence. If we make an electron travel along the z direction with velocity
υ as shown in Fig. 6.3. Then we are exciting this mode as sin kυ t with ω = kυ ,
we can write the excitation as sin ω t. Now if ω matches ω0 the resonant frequency
of the mode then we have excited the mode with an rectilinear motion. This is the
principal of a Klystron a device used to produce microwaves especially for the Radar
application.
In magnetron, we made the electron move in circles which talked to cos θ de-
pendence of the mode. In Klystron, we made the electron move in z direction and
it talked to cos kz dependence of the mode. Now we can do both. We can make
the electron gyrate as shown in Fig. 6.3B, where we travel vertically along z and
also in loops. The loopy motion comes from an applied magnetic field. If ω1 is
the angular (loopy) velocity of the electron and electron travels along the z direc-
tion with velocity υ , such that kυ = ω2 . Then we have excitation of the mode as
cos ω1t cos ω2t, i.e. we excite the mode at frequency ω1 ± ω2 . If the frequency of
mode is say ω0 = ω1 + ω2 , we obtain resonant excitation of the mode. Furthermore
as we we add two frequencies ω1 + ω2 , we can get high resulting frequency ω0 .
This mechanism of producing microwaves by gyrating motion is realized in device
called Gyrotron.

6.2.3 Microwave systems and networks

We have talked about microwave cavities. We have talked about waveguides. Now
we talk about microwave networks made out of these systems. See Fig. 6.4, which
shows a cylindrical cavity (source cavity) acting as a microwave source (a) coupled
to a waveguide (b), which is terminated in an microwave cavity (load cavity) (c),
6.2 TM Modes 67

Fig. 6.4 Fig. shows a microwave network with source and load cavity connected by a waveguide.

which could be a cavity as in microwave oven where we heat our food. The modes
of the source are appropriately excited and we build these modes.The EM radiation
is then channelled through a waveguide to resonator. How to analyze such networks.

B
Fig. 6.5 Fig. A shows how an aperture is modelled as mode current with no aperture and a fictitious
current that opposes this current. Fig. B shows magnetic field in infinitesimal volume in proximity
of wall current.

Lets say source cavity has a modeR φ0 at frequency ω0 . Lets normalize the mode
such that hφ0 , φ0 i = 1, (hφ0 , φ0 i = ε |E|2 + µ |H|2 ). When we excite the source
cavity mode resonantly its amplitude x builds up linearly as time as ẋ = u. Now
suppose we make a small aperture of area Ae in the wall of the source cavity to let
microwaves out. We can analyze the aperture by assuming that there is no aperture,
in which case the mode will cause certain current on the walls of the waveguide and
68 6 Rf and Microwave Engineering

now we subtract this current by assuming a fictitious current i0 , opposite to mode

current on the wall. This is as shown in 6.5A. This fictitious current i0 radiates inside
and outside the cavity. The radiation inside produces H in proximity to current that
opposes the H due to mode current and hence the fictitious current radiates inwards
to de-excite the mode. Since fictitious mode is proportional to mode amplitude, we
can write

ẋ = −α02 x + u. (6.19)
How much is α 2.
Let HA be the magnetic field value of φ0 at the aperture. Then
in time ∆ t, the fictitious current i0 will produce magnetic field near aperture of value
HA
2 , that fills a cylindrical volume V = Ae c∆ t. The overlap of this field with φ0 is
µ 2
2 HAV . Thus the mode decrements in value as

ẋ = −α02 x + u, (6.20)
µ 2
where α02 = 2 HA cAe . If V0 is cavity volume, we have
µ cAe cAe
α02 = ( HA2V0 ) =γ , (6.21)
2 V0 V0
µ 2
2 HA
where γ = R
ε |E|2 +µ |H|2
. The quantity γ is ratio of energy near aperture to average
V0
energy of the mode in the cavity.
So we are exciting the mode but it leaks out of aperture and decays at rate α02 .
We can solve the above equation and we get steady amplitude of mode in the cavity
as x = αu2 .
0
Forget loss for a moment. What if we excited the cavity away from resonance
frequency ω0 at ω . Then we can describe the the mode amplitude at frequency
1 1
s = jω as x(s) = s−a u(s) where a = jω0 . The quantity G(s) = s−a is called the
cavity transfer function. At s = jω0 it becomes infinite as on resonance the cavity
amplitude grows unbounded. This is as shown in Fig. 6.6A. In presence of loss now
The transfer function is modified as feedback shown in Fig. 6.6B and the the new
transfer function G1 (s) is

G(s) 1
G1 (s) = = . (6.22)
1 + α 2 G(s) s − a + α 2

Now we connect source cavity to the waveguide which is coupled to load cavity.
How do we analyze this interconnect. Lets simplify and say we have two resonant
cavities connected by a small aperture as shown in Fig. 6.7a.
Let φA be mode of cavity A and φB be mode of cavity B both with frequency
ω0 before we create an aperture. Both modes will create currents on cavity wall. If
modes are say identical then φA and φB will create same current where aperture is. If
we consider φ = φA − φB , the currents due to φA and φB will cancel φ has no current
at the boundary where aperture is and we get a mode of the cavity with aperture.
6.2 TM Modes 69

u(s) G(s) x(s) A

− α2

B
u(s) G(s) x(s)

− α2

u(s) Ga(s) x(s)

− αβ C
− αβ Gb(s)

− β2

Fig. 6.6 Fig. A shows how off-resonant excitation to a cavity is modelled as a transfer function.
Fig. B shows how loss through aperture is modelled as negative feedback. Fig. C shows intercon-
nect of two cavities through an aperture modelled as a feedback loop.

In above we assumed identical cavities. Suppose they are different. Consider a

resonant cavity connected to a waveguide as shown in Fig. 6.7b. Let φA be mode
of cavity and φB be mode of waveguide both with frequency ω0 (say both modes
normalized) . The aperture current distribution due to φA and φB are different in
general. Lets close cavity A by introducing fictitious current iA and close cavity B
by introducing fictitious current iB . If x and y are the amplitudes of modes φA and
φB we can write
70 6 Rf and Microwave Engineering

A B

b A
B

Fig. 6.7 Fig. A and B shows aperture coupling of two identical and different cavities respectively.

      
d x −α02 −αβ x u
= + . (6.23)
dt y −αβ −β 2 y y 0
Since the two cavities have different volumes, we find α 2 6= β 2 . from Eq. (6.21),
we find smaller cavity has larger loss. Furthermore β φA − αφB is the joint mode
that doesnot dissipate. When we excite, this mode builds up. Now suppose the two
cavities donot share same resonance. Let resonances be a and b. Then the the joint
transfer function as in fig. 6.6C is

1 1
G(s) ∝ ∝ (6.24)
α2β 2 (s − a)(s − b) + β 2 (s − a) + α 2 (s − b)
1− (s−a+α 2 )(s−b+β 2 )

1
Now see two limits, when α ≫ β then G(s) ∝ (s−a)(s−b)+α 2 (s−b)
i.e, we have reso-
1
nance at a. When α ≪ β , we have G(s) ∝ (s−a)(s−b)+β 2 (s−a)
i.e, we have resonance
at b. For intermediate values it is a weighted sum. If we assume cavity volume is
much larger than the waveguide volume then the final resonance is more weighted
towards cavity resonance. If both waveguide and cavity resonance a then as expected
the joint system gas has resonance at a irrespective of α , β values.
Now return to Fig. 6.4. Suppose source and cavity have resonance at a and sup-
pose waveguide is off, may be its length is not correct. The length is decided between
6.3 Impedance matching waveguides 71

where source and destination are and not much can be done then. Then the joint res-
onance of waveguide and load cavity will be different and we donot do a good job of
exciting this system by source. How can we change the waveguide resonance with-
out changing its length so that everything is on resonance, i.e., impedance matched.
This is called impedance matching.
Other scenario is that the load cavity resonance goes off a bit because we put
something in it. Then we can change the waveguide resonance so that jointly they
come back to the source resonance. In either case, we have to study how to vary the
resonance of the waveguide. There are many ways to do it and we study them under
impedance matching.

6.3 Impedance matching waveguides

6.3.1 Taper

Recall of a waveguide with T Emn mode we have

r
ω2 mπ 2 nπ
β= −( ) − ( )2 . (6.25)
c2 a b
Lets focus on T E11 mode with
r
ω2 π2 π2
β= − 2 − 2. (6.26)
c2 a b
If l is the length of the waveguide then if β l = pπ , then waveguide is resonant
with ω . If (p − 1)π < β l < pπ , then we can increase the length and bring waveguide
into resonance with ω . l is fixed so we can change the dimension of waveguide and
make say a and b as functions of z as shown in fug. 6.8, where in 6.8a we change b

b
a
b a

Fig. 6.8 Fig. a shows how we change b dimension of waveguide. Fig. b shows how we change a
dimension of waveguide

as function of z and in 6.8b we change a as function of z. Then

72 6 Rf and Microwave Engineering
s
ω2 π2 π2
β (z) = 2
− 2 − 2 . (6.27)
c a (z) b (z)
R
As we increase say a(z), β increases and now 0l β (z)dz = pπ and we bring waveg-
uide into resonance. ofcourse this requires changing structure of waveguide perma-
nently. We prefer instead to not do permanent change to waveguide structure and
instead introduce alterations, that change the impedance as described below. For
this we first develop a transmission line model of waveguide.

6.3.2 Transmission line models

a
+ + + +
− +
− +
y b − + y
− +
− +
− − − −
C1 x C2 L x

A B C

Le
y
D

x z
Fig. 6.9 Fig. shows capacitors and inductors in TE mode.

To fix ideas consider a waveguide with T E11 mode. Cross section of the waveg-
uide shows electric field Ey in the y and Ex in the x direction as in Fig. 6.9A and B
respectively. The electric field deposits charges on the boundary and the cross sec-
tion acts as capacitor. Lets see the field Ey , it has the form E0 sin( πax ) cos( πby ) and it
is simply E0 sin( πax ) at the top and bottom boundary, which gives a charge density
σ
2ε = Ey . We can integrate the charge density to get total charge Q and then using the
R 2
fact that energy stored in electric field ε0 Ey2 = QC , we get the capacitance as ε0 aπ∆bx
where ∆ x is the small length in the z direction, or capacitance per unit length is
C1 = ε0 πab . Similarly Ex gives C2 = ε0 πba . We remark if we have T Emn mode instead
a
of T E11 then the capacitance C1 = ε0 π mb and C2 = ε0 πbna .
The energy oscillates between electric field Ex , Ey and magnetic fields. The mag-
netic field Hz corresponds to circular currents as shown in 6.9C, which gives induc-
tance L, the value of the inductance is just the area of the loop µ0 ab. For thickness
6.3 Impedance matching waveguides 73

∆ x, we get L → ∆Lx . Hz is not the only magnetic field, we also have magnetic field
Hx , HY , which correspond to longitudinal currents as shown in Fig. 6.9D. Lets see
the field Hy , it has the form H0 sin( πby ) cos( πzx ), and it is simply H0 sin( πby ) at the side
wall, which gives a current density 2j = Hy . We can integrate the current density to
Rget total current I, and then using the fact that energy stored in the magnetic field
µ0 Hy2 = LI 2 , we get the inductance as µ0 aπ∆bx where ∆ x is the small length in the z
direction, or inductance per unit length is L1 = µ0 πba . Similarly Hx gives L2 = µ0 πab .
We remark, if we have T Emn mode instead of T E11 then the inductance L1 = µ0 πban
and L2 = µ0 π bm a . Let

Le = L1 + L2 . (6.28)
Then we have a transmission line model as shown in Fig. 6.10, with line length
signifying longitudinal z direction. Then we can write transmission line equations

Le L
e

L C1
C2

Fig. 6.10 Fig. shows transmission line model for TE mode.

− jβ V (z) = jω Le I(z), (6.29)

1
− jβ I(z) = ( jω Ce + )V (z), (6.30)
jω L

where Ce−1 = C1−1 +C2−1 . On multiplying the above two equations, this gives with

LeCe = c−2 . (6.31)

1 1 1 1
ω 2 = c2 β 2 + = c2 β 2 + ( + ), (6.32)
LCe L C1 C2
ω2 π π
= β 2 + ( )2 + ( )2 . (6.33)
c2 a b
or
74 6 Rf and Microwave Engineering
r
ω2 π π
β= − ( )2 − ( )2 . (6.34)
c2 a b
We now show we can change the capacitance of the cross section and get

r
1 1 1 1
β (z) = ω 2 − ( + ). (6.35)
c L C1 C2
There are various ways to change the capacitance of cross-sections, we study
them below.
Before we move forward, we develop a transmission line model for T M mode.

L
2
C

L Ce
1

Fig. 6.11 Fig. shows transmission line model for TM mode.

To fix ideas, consider a waveguide with T M11 mode. Cross section of the waveg-
uide shows magnetic field field Hy in the y and Hx in the x direction. These fields
give longitudinal currents on side wall and top wall of the waveguide. The energy
oscillates between magnetic field Hx , Hy and the electric fields. The electric field Ez
corresponds to charge along longitudinal direction. These currents and charge along
longitudinal direction are represented by two inductors L1 and L2 (for side and top
wall) in parallel and a series capacitor C as shown in Fig. 6.11. L1 and L2 are same
as in TE mode. The capacitance C = ε0∆ab . The transverse electric fields deposits
charge on top and side wall of waveguide cross section and represented by capacitor

Ce = C1 +C2 , (6.36)

where C1 and C2 are same as in TE mode.

Then we have a transmission line model as shown in Fig. 6.11, with line length
signifying longitudinal z direction. Then we can write transmission line equations

1
− jβ V (z) = ( jω Le + )I(z), (6.37)
jωC
− jβ I(z) = jω CeV (z), (6.38)

where Le−1 = L1−1 + L2−1 . This gives with Ce Le = c−2 .

6.3 Impedance matching waveguides 75

1 1 1 1
ω 2 = c2 β 2 + = c2 β 2 + ( + ), (6.39)
CLe C L1 L2
ω2 π π
= β 2 + ( )2 + ( )2 , (6.40)
c2 a b
or
r
ω2 π π
β= − ( )2 − ( )2 . (6.41)
c2 a b

6.3.3 Iris

Now focus on TE modes.

∆ ∆
111
000 111
000 ∆ 111111111111
000000000000
b 000
111
000
111 000
111
000
111 000000000000
111111111111 b
000
111
000
111 000
111
000
111 000000000000
111111111111
000
111 000
111 ∆ 000000000000
111111111111
A 000000000000
111111111111
a a B

Fig. 6.12 Fig. shows how we change cross-section of a waveguide using metal strip called iris.

We change the cross section as in fig. 6.12A. The two capacitance C1 and C2
change from ε0ba and ε0ab to ε0 (a−
b
∆)
and (a− ε0 b
∆ ) respectively and we get
r
ω2 π π ∆ π π
β= − ( )2 − ( )2 + (( )2 − ( )2 ). (6.42)
c2 a b a a b
Since a > b we find that β decreases.
We change the cross section as in fig. 6.12B. The two capacitance C1 and C2
ε0 (b−∆ )
change from ε0ba and ε0ab to b−
ε0 a
∆ and a respectively and we get
r
ω2 π π ∆ π π
β= − ( )2 − ( )2 − (( )2 − ( )2 ). (6.43)
c2 a b b a b
76 6 Rf and Microwave Engineering

Since a > b we find that β increases.

Thus using iris as above we can increase or decrease β for a cross-section and
impedance match.

6.3.4 Post and Screws

Focus on TE modes.

∆
111
000 ∆
11
00
000
111 00
11
b 000
111
000
111 00
11
00
11 b
000
111 00
11
00
11
A 00
11
a a B

Fig. 6.13 Fig. shows how we change cross section of a wave guide using a tuning screw.

In addition to using a waveguide iris, post or screw can also be used to give
a similar effect and thereby provide waveguide impedance matching. Fig. 6.13A
shows a screw coming down partly. Its effect is to increase both C1 and C2 and
hence β increases.
To decrease β screw or post should extend through the waveguide completely
making contact with both top and bottom walls as in Fig. 6.13B. The effect is same
as iris in Fig. 6.12A and hence β decreases.
Screw comes from top to bottom. Posts go bottom to up.

6.3.5 Plungers

As discussed before in discussion of tapered transmission lines we can change a and

b locally and change β (z) by the equation.
6.3 Impedance matching waveguides 77

b b

a a

A B
Fig. 6.14 Fig. shows how cross section of a waveguide is changed using plungers.

s
ω2 π 2 π 2
β (z) = −( ) −( ) . (6.44)
c2 a(z) b(z)
This can be achieved by aid of plungers as shown in Fig. 6.14A and B.

6.3.6 Rf-oscillators

Rf oscillators can be made with inductors and capacitors in series or parallel. Fig.
6.15a shows the series version where v0 is the applied voltage and v1 voltage across
the capacitor. The current through the capacitor is c dv 1
dt and hence the voltage across
d dv1
the inductor is L dt (C dt ) giving

d 2 v1
v0 − v1 = LC , (6.45)
dt 2
with ω0 = √1 ,
LC

d 2 v1
ω02 v0 = + ω02 v1 . (6.46)
dt 2
1 dv1
let x = v1 and y = ω0 dt .
Then
      
d x 0 ω0 x 0
= + . (6.47)
dt y −ω0 0 y ω0 v0
Using variation of constant formula,
78 6 Rf and Microwave Engineering

v
v0 1
a

i1
b i0

V
D

c i 0

Fig. 6.15 Fig. a shows an oscillator that uses a voltage source. Fig. b shows an oscillator that uses
a current source. Fig. c shows a transistor implementation of Fig. b.

  Z t   
x(t) sin ω0 σ 1 − cos ω0t
= ω0 v0 d σ = v0 . (6.48)
y(t) 0 cos ω0 σ sin ω0t
We find the voltage v1 is oscillating.
Fig. 6.15b shows the parallel version of rf-oscillator where we use a current
source. Fig. 6.15c shows implementation of a current source using a transistor cir-
cuit. The current through inductor is i2 and therefore voltage across it is L didt1 which
is also the voltage across capacitor and hence current through capacitor is C dtd (L didt1 ),
which gives

d 2 i1
i0 − i1 = LC , (6.49)
dt 2
with ω0 = √1 ,
LC
6.4 Tank Circuits 79

d 2 i1
ω02 i0 = + ω02 i1 . (6.50)
dt 2
1 di1
Let x = i1 and y = ω0 dt .
Then
      
d x 0 ω0 x 0
= + . (6.51)
dt y −ω0 0 y ω0 i0
Using variation of constant formula,
  Z t   
x(t) sin ω0 σ 1 − cos ω0t
= ω0 i0 d σ = i0 . (6.52)
y(t) 0 cos ω0 σ sin ω0t
We find the voltage across inductor y is oscillating.

6.4 Tank Circuits

Fig. 6.16 Fig. a shows a tank circuit. Fig. b shows a tank circuit in NMR detection. Fig. c shows a
tank circuit in a radio receiver.
80 6 Rf and Microwave Engineering

Fig. 6.16 shows a tank circuit to amplify small oscillating voltage v0 = A cos ω0t
. Let v1 be voltage across the capacitor. The current through the capacitor is c dvdt
1

dv
and hence the voltage across the inductor is L dtd (C dt1 ) giving

d 2 v1
v0 − v1 = LC , (6.53)
dt 2
with ω0 = √1 ,
LC

d 2 v1
ω02 v0 = + ω02 v1 . (6.54)
dt 2
1 dv1
Let x = v1 and y = ω0 dt .
Then
      
d x 0 ω0 x 0
= + . (6.55)
dt y −ω0 0 y ω0 v0
Using variation of constant formula,

  Z t  
x(t) sin ω0 σ
= ω0 A cos ω0 (t + σ ) dσ (6.56)
y(t) 0 cos ω0 σ
 
Aω0t − sin ω0t
= . (6.57)
2 cos ω0t

We find the voltage v1 grows linearly, i.e, amplified.

6.4.1 NMR and radio receiver and tank circuit

Tank circuits are used in field of nuclear magnetic resonance (NMR) to detect
nuclear spins. The magnetic moments of nuclear spins constitutes a flux passing
through an inductor coil as shown in 6.16b. The precession of the spins in a mag-
netic field cause this flux to sinusoidally oscillate which induces a very small EMF
in the coil. This can then be amplified using a tank circuit.
Tank circuits find use in radio receiver circuits as shown in 6.16c where antenna
voltage is amplified using a tank circuit.

6.4.2 Resonant radio antenna and tank circuit

Shown in 6.17a is a Hertz dipole antenna. The applied voltage pulls charges towards
it, making ends of the antenna charged. This charge opposes the pull. If we bend the
ends as shown in 6.17b, it begins to look like a capacitor, where charges on capacitor
6.4 Tank Circuits 81
−
−−
l L
− −−
V V C
++ + V
a
R
+ +
+ b c
L

d V C

Fig. 6.17 Fig. shows how resonant antenna acts a tank circuit and has gain inbuilt in it.

plate oppose the pull. Therefore the antenna has a capacitance and of-course it has
inductance as current flowing in antenna produces a magnetic field and change in
this current or resulting magnetic field produces an EMF that opposes the change
and finally the antenna has some resistance. Therefore antenna is like an RLC circuit
as shown in 6.17c. Neglecting resistance for a moment, the resonant frequency of
this circuit is ω0 = √1LC . If driving voltage V is at this frequency, the impedance of
the antenna vanishes and we see a large current flow in it and antenna efficiently
transmits. Ofcourse in fig. 6.17a charge is not just concentrated at the ends but there
is a charge density along the length. Hence the capacitance is distributed in the form
of a transmission line as shown in 6.17d. If we choose the length of line (antenna) as
l such that β l = π2 , then the impedance of the antenna is − jZ0 cot β l which becomes
0 at β l = π2 and huge current flows (of course we are neglecting resistance) and
antennna transmits efficiently.
We talked about antenna in the transmit mode. In receive mode, the incident elec-
tric field acts as a voltage source and the received voltage is the voltage across the
capacitor as in 6.17c. As shown in our discussion of tank circuits, once the driving
frequency is on resonance this voltage builds like crazy and is only limited by the
resistance of the circuit. Once again, we can be more realistic and treat antenna as a
transmission line. We excite it at one end with voltage V0 and find the open circuit
voltage at the other end. Since there is no current at the open end, V+ = V− and
voltage at the driving end is

V0 = V+ exp(− jβ l) +V− exp( jβ l) = V+ exp(− jβ l)(1 + exp( j2β l)). (6.58)

When 2β l = π , we get V+ → ∞. Thus transmission line amplifies the driving voltage

V0 . Of course there is a resistance and infinity only means large gain as in a tank
circuit.
82 6 Rf and Microwave Engineering

6.5 Klystron microwave amplifier

Output

Input

Buncher cavity Catcher cavity

C
B

Fig. 6.18 Fig. shows schematic of a microwave amplifier called Klystron.

Consider a microwave cavity as in fig. 6.18A. Suppose we pass electrons through

it as shown. These will excite cavity mode as discussed in the Klystron oscillator.
The electrons come one by one and trigger cavity excitation at different times. These
modes are not in sync and cavity excitation doesnot build up. But suppose, we can
bunch these electrons as in fig. 6.18B, so that excitation times are in resonance with
the cavity, then the excitations due to electron will coherently add and we will see
cavity fill up with microwaves. Now how do we bunch these electrons. We pass
the electrons through a input cavity containing input microwaves that we want to
amplify. The electric field due to these input signal will oscillate and oppose and
accelerate the incoming electrons in a cycle. The result is in a cycle, it ends up
bunching them. The bunching is more, if input signal is more, i.e, electric field is
stronger. Then these bunched electrons go and fill the output cavity. When bunch-
ing is stronger output is stronger. Thus we get a linear amplifier called Klystron
microwave amplifier.
6.6 Transistor Rf amplifiers 83

6.6 Transistor Rf amplifiers

Rf signals can be amplified by a solid state device called transistor [7]. A transistor
is made of semiconductor material like silicon. Silicon has 4 electrons in its outer
shell. In a solid crystal, neighboring silicon atoms talk and these electrons of the
silicon lattice organize themselves into waves states whose energies lie in a band.
Wavestates are indexed by their wavenumber and each wavestate can carry two elec-
trons. When all wavestates in a band are full as in silicon, we cannot conduct as
conduction means increasing momentum or wavenumber which is not possible as
the band is full. When we replace some of the silicon atoms with say phosphorus
then we get extra electrons (phosphorus has 5 valence electrons instead of 4) which
go into next available energy band called conduction band which is not full and can
conduct. We say we have n-doped silicon. Similarly when you replace some of the
silicon atoms with say aluminium (aluminium has 3 valence electrons instead of 4)
then we get less electrons than native silicon which means some of wavestates of the
filled band (also called valence band) become available and it can conduct. Valence
band has lower energy than the conduction band. The energy gap is called the band
gap. Fig. 6.19A shows bands for silicon, B shows bands for n-doped silicon and C
shows bands for p-doped silicon.

silicon N−doped P−doped

Conduction band Conduction band Conduction band

E E E

Valence band Valence band Valence band

A B C

Fig. 6.19

When we bring a n-doped silicon next to a p-doped silicon, we form a junction

called a pn junction. Some conduction electrons will flow from the n-region go the
valence electrons in p-region to minimize energy. This will result in build up of
+ and − charge in the n and p region at the junction, which creates an electric
field that eventually halts this flow. This interface filed corresponds to a voltage of
.7 V. When we apply positive voltage to the p-region compared to n region, we
pull valence electrons from p-region which are substituted by flow of conduction
electrons from n region and current flows. For this to happen the applied voltage
should be larger to overcome the interface field which is roughly .7 V. We say we
have forward biased the pn junction. if instead we apply positive pull on n electrons
we cannot pull anything as the valence electrons in n-region are immobile and the
84 6 Rf and Microwave Engineering

conduction electrons can only be substituted by valence lectrons from p region but
they have much lower energy. Hence current cannot flow and we say we have reverse
biased the pn junction. pn junction only conducts in one direction.

E B C E B C

n p n
n p n

A B

E B

C
C

Fig. 6.20 The fig A shows schematic ofa transistor. Fig. B shows a cartoon of how electrons and
bands are in various regions without taking into account interface electric fields. Fig C shows the
energy level diagram of a transistor.

Transistor has two junctions as shown in 6.20A. A p region called base sand-
witched between two n regions (called emitter and collector) as shown in fig 6.20A.
One pn junction is forward biased (emitter and base) and other reverse biased (base
and collector). Forward biased pn junctions suck valence elctrons from p region and
hurls electrons from the n region in the p-region as shown in 6.20B. These electron
are immediately sucked by the reverse biased junction and on their fraction of them
α recombine and contribute to the p junction current the remaining 1 − α fraction
just fly by to other n region as shown in 6.20B. If I is the base current 1−αα I = β I is
the collector current and (1 + β )I the emitter current. α is small so β is large and is
called the current gain of the transistor. Fig. 6.20C shows energy level diagram for
the transistor.

6.6.1 Current Source and amplifier

Transistor acts a current source. Fig. 6.21A shows by use of a forward bias and
resistance R1 , we establish a current in emitter, simply given by Ie = V1R−.7
1
. This
β
gives a collector current β +1 Ie ∼ Ie and the potential drop across resistance R2 as
R2 (V1 −.7) R2 R2
Ie R2 = R1 . Thus the voltage V1 is amplified to R1 V1 , where R1 is the amplifier
6.6 Transistor Rf amplifiers 85

E B C

n p n R2

R1 R2
V1 V2
V0
Vin
R1
A
B

Fig. 6.21 The fig A shows schematic of a transistor. Fig. B shows a cartoon of how electrons and
bands are in various regions without taking into account interface electric fields. Fig C shows the
energy level diagram of a transistor.

gain. V1 can have oscillating parts as shown in figure 6.21B, which get amplified,
the DC part of V1 is used to bias the base emitter circuit as shown in figure 6.21B.
A good amplifier should have large input impedence as all voltage from a source
appears across it then. Similarly a good amplifier should have low output impe-
dence as all amplified voltage appears across the amplifier. Lets compute the input
and output impedence of the amplifier in 6.21B. Changing input voltage by ∆ V
gives ∆ Ie = ∆RV1 and ∆ Ib = β∆RV , then input impedence is ∆∆ VI = β R1 . Similarly, if
1 b
we measure the voltage across R2 with a voltmeter with impdence R′ then R2 gets
R′ R′ R2
modified to RR22+R ′ and hence the voltage is R +R′ R V1 , hence the voltage is reduced
2 1
′
by a factor R2R+R′ , this means output impdence is just R2 .
Until now we only amplified input voltage V . In many applications we want to
amplify difference of voltage. Such an aplifier is called a differential amplifier or an
opamp. It has two inputs V1 and V2 and output V0 which is a G(V1 − V2 ), where G
is the amplification factor. The differential amplifier circuit is shown in 6.22A. The
current through the emitter of transistor 1 is V1R−V
1
2
and hence voltage across output
resistor R0 is (V1 −V
R1
2 )R0
. The input impedence at lead 1 is simply β R1 while at lead 2
we change the input by ∆ V2 , then current in emitter 2 changes by ( R11 + R12 )∆ V2 and
hence the input impedence is β R′ where R1′ = R11 + R12 , i.e. R1 , R2 in parallel. The
opamp is shown as a schematic in 6.22B, with two input leads and a output lead.
Now we show using opamp, how to build circuits with tunable gain. We do this
by using concept of negative feedback. We assume that our transistors have very
high gain β , so that input impedences are very high so that there is negligible current
going in the leads. Further the gain is so high and any difference of the input voltages
will lead to large current in the output transistor throwing it onto saturation. So we
assume two leads have the same voltage. Now consider circuit 6.22C. The + and
−1 lead have same voltage 0 and current RV1 goes directly to output instead of in the
input lead and hence output voltage is −V RR21 . We choose R2 ≫ R1 to have good gain,
86 6 Rf and Microwave Engineering

this is called inverting amplifier. The gain can be tuned by changing R1 , R2 . There
is alternate arrangement. See circuit 6.22D. Here V = V0 R1R+R1
2
or V0 = (1 + RR21 )V ,
this is called noninverting amplifier. The gain can be tuned by changing R1 , R2 [8].

A R B
0
V0

V1 V1
V +
2 V0
V2 −

R1

R
2

R1 V
V R2 +

+ − V0
− V0

R R1
2

C D
Fig. 6.22 The fig A shows schematic of a differential amplifier. fig B shows symbolic represen-
tation of a differential amplifier. fig. C shows an inverting amplifier. fig. D shows a noninverting
amplifier.

6.7 Rf and Microwave Filters

We talked about Rf and Microwave oscillators. Q of the oscillator limits the purity
of resulting frequency source. Source signal may contain higher harmonics. We can
improve signal purity by passing it through a band pass filter. We mention here
cavity filters. We talked about cavity with transfer function
1
G(s) = (6.59)
s − jω0 + α 2
6.8 Cavity Coupling 87

The factor Q = αω02 measures the quality factor of the band-pass filter which can
be high as 106 . The input port to the cavity couples the input signal. The signal
at cavity resonance has large gain while other frequencies get attenuated. Filter also
find natural application when a signal with noise needs to be amplified, we first filter
out noise.

6.8 Cavity Coupling

Probe Loop

A B

Cavity Cavity
Waveguide slot
Waveguide Aperture

C D
Fig. 6.23 Fig. A shows a cavity coupled to a coax capacitively (probe coupling). Fig. B shows
a cavity coupled to a coax inductively (loop coupling). Fig. C shows waveguide/cavity coupled
through aperture. Fig. D shows waveguide/cavity coupled through slot

The power between the cavity resonator and signal source or load can be coupled
by means of a coaxial line, whose center conductor is extended inside the cavity in
form of a probe 6.23A or loop 6.23B. The cavity may be coupled to a waveguide by
means of an aperture or a slot in the main wall as in 6.23C and 6.23D [6].
Chapter 7
Geometric Optics

C
P1 P0

Fig. 7.1 Fig. A shows a pin-hole camera. Fig. B shows a diffuse image if we increase the size of
pin-hole. Fig. C shows how light can be focused with a lens.

89
90 7 Geometric Optics

7.1 Light Around Us

In past chapters, we talked about radio (3 kHz-3 GHz) and microwave (3-300 GHz)
frequency EM waves. In this chapter, we focus on light or optical frequency (4-8 ×
1014 Hz). This is all the world around us. Blue sky, green grass, light from the sun,
all the world of beautiful colors. How do we see this world. Light reflects of objects
and enter our eyes to form an image at the back (called retina) of the eye. Lets look
at this image formation.

7.2 Pin Hole

We can model our eye as a pin hole camera as in 7.1A. The light rays from object
after passing through pin-hole form an image on the screen as shown in 7.1A. This
does not let very much light in. How can we gather all the light coming from the
object. We can increase the size of the pin hole as shown in 7.1B. This produces a
very diffuse image as light from a point is spread all over the screen as shown in
7.1B. How can we collect more light coming from the object and yet have a sharp
image. We need a lens as shown in 7.1C [?]. The light emanating from each point
on the object gets focused as shown in 7.1C. How does the lens work ?

7.3 Lens

Fig. 7.2 shows a dense spherical medium like glass. r is the radius of the sphere.
Light rays parallel to equator move into a rare medium like air. Consider a light ray
making an angle α with radial line as shown in 7.2.
From Snell’s laws of refraction

sin α + β
= C. (7.1)
sin α

r sin α = d tan β , (7.2)

for small α , β , we have

α +β
= C, (7.3)
α
α
= (C − 1)−1 = C′ . (7.4)
β

r
d= . (7.5)
C′
7.3 Lens 91

α
α β

r d
O

Fig. 7.2 Fig. shows light moving from dense spherical medium to rare medium.

We find irrespective of α , we focus at same d as above. We call d2 = f , the focal

point and f focal length. We call d, 2F point. If we reverse light direction, we will
start from the focal point and when light rays enter the glass medium, they become
parallel. We assume all action is near equator, so that α , β are small. If we cut this
sphere at the periphery, as shown by dotted line QP in 7.2, we get a half lens and
we join two such half lenses back to back to form a lens as shown in 7.3. Light
rays emanating from 2F point become parallel inside the lens and focus on the 2F
point on other side. The two lenses need not have same focal length. We can have
two focal lengths f1 and f2 . This is how the eye lens works. The focal length of the
half lens that sees the world is changed to focus objects at different distances while
the half lens facing retina focuses everything on retina and its focal length is not
changed that much.

P0 P1

Fig. 7.3 Fig. shows light rays emanating from 2F point become parallel inside the lens and focus
on the 2F point on other side.
92 7 Geometric Optics

φ Q1
φ
P0

P1
Q0

Fig. 7.4 Fig. shows how light from point Q0 shifted from 2F point converges to Q1 on the other
side and is shifted the other way.

We say how light rays emerging from a 2F point become parallel and focused
again on the 2F point other side of lens. How about a point that is displaced vertically
from the 2F point. The light rays tilt by an angle φ as shown in fig. 7.4. Fig. 7.4
shows that the point Q0 converges to Q1 on the other side that is shifted the other
way. If focal lengths are different then if Q0 is displaced by x , Q1 displaced by y
then for φ as shown in fig. 7.4,
x y
φ= = , (7.6)
f1 f2
f2 x
which gives Q1 is displaced by f1 .

7.4 Microscopes

This forms the basis of what is called a microscope. If f2 ≫ f1 , the object gets
magnified. This is shown in fig. 7.5.

7.5 Telescopes

Light coming from a distant object can be first focused with a lens (objective lens)
and then magnified by another lens (eye piece) as in microscope, to form what is
called a telescope that helps us see far. This is shown in fig. 7.6. With bare eyes, the
image on retina is negligibly small. In telescope, the microscope action first enlarges
the image P1, before it is seen.
7.6 Resolution and Diffraction 93

f1

Q0
f2
P1
P0

Q1

Fig. 7.5 Fig. shows the magnification principle of a microscope with P0Q0 magnified to P1Q1.

P1

P2 P0

Fig. 7.6 Fig. shows the magnification principle of a telescope, with P magnified to P2.

7.6 Resolution and Diffraction

∆x
α
θ
0 2f

Fig. 7.7 Fig. shows how diffraction limits resolution of magnification.

94 7 Geometric Optics

Microscopes can magnify. Is there any limit ? Can we magnify arbitrary small
objects ? We find the small size is limited by the wavelength of light used. Consider
a small slit in the wall in fig. 7.7 of aperture size D. Light coming from left passes
through the slit and goes to the right but not in straight line. The slit acts as an
antenna array and light spreads out, a phenomenon called diffraction. The spread θ
of light in fig. 7.7 is as described in treatment of antenna arrays and given by

kDθ = π , (7.7)
2π λ
where k = λ or θ = 2D . To resolve O and P, we have to make sure θ < α or

λ x
< , (7.8)
2D 2f
fλ
x> . (7.9)
D

The limiting size is x = fDλ , which becomes smaller if we use smaller wavelength
light or larger D. We can now simply replace slit with a lens with D as aperture of
lens and above analysis say how small an object can be for it to be magnified.

7.7 Lens Equation

Until now we showed design of optical instruments with lenses whose half lenses
were of different focal lengths. However it is possible to do everything with lenses
with half lenses of same focal length. When we put object at 2 f light rays converge
other side at 2 f . They turn around by angle 2φ , where φ = 2xf , where x is vertical
distance from lens center as shown in fig. 7.8A. If light rays started parallel when
they turn around by 2φ , they will pass through the focal point as shown in fig. 7.8B.
In general when they emanate from point o as in fig. 7.8C, the angle φ1 = ox and
they converge at point i as in fig. 7.8C, the angle φ2 = xi . They turn by φ + φ2 = 2φ ,
which gives the famous lens equation
1 1 1
+ = . (7.10)
i o f

7.8 Revisiting Optical Instruments

Now we revisit design of optical instruments using lens equation. When we move
object to be magnified close to f . The image point i moves far and object size x gets
magnified to ixo . This is shown in fig. 7.9. This forms the basis of a microscope.
7.8 Revisiting Optical Instruments 95

φ/2 φ/2

A 2f 2f

B
f
φ1 + φ 2 = φ

φ1 φ2

C
o i

Fig. 7.8 Fig. A shows for object at 2 f , light rays converge other side at 2 f . They turn around by
angle 2φ . Fig. B shows parallel light rays pass through focal point. Fig. C shows object and image
distances related by lens equation.

2f f f 2f

Fig. 7.9 Fig. shows design of a microscope using lens equation.

Fig. 7.10 shows design of telescope using lens equation. The object is further than
2 f for the objective lens whose image is formed between f and 2 f of the objective.
This image is between f and 2 f of eye piece and is then magnified.
96 7 Geometric Optics

2f f f 2f f 2f

Fig. 7.10 Fig. shows design of a telescope using lens equation.

Chapter 8
Optical resonance and colors

In this book, we talked about dielectric constant ε of a media. We can write

ε = ε0 (1 + χ ), where ε0 is dielectric constant of vacuum and χ is called electric
susceptibility of the medium. What is the source of χ . Why do radio waves go
through the wall as in mobile telephones while light does not. What is the source of
all beautiful colors around us. In this chapter we will try to answer some of these
questions.

8.1 Atom as an Antenna

Atom has electrons organized in orbitals. An electron from higher energy state (ex-
cited orbital) can jump to a lower energy state (ground state orbital) giving radiation
with frequency ω0 , where h̄ω0 is the energy difference between the two states (h̄ is
the Planck’s constant) . Similarly, when light is shone on an atom, with frequency
ω0 it can absorb light, and jump from lower energy state to higher energy state.
We show that in both cases, the atom acts as an antenna and always radiates. We
commonly use the phrase, atom absorbs light. What we say absorption is in fact ra-
diation from atom that cancels part of incident radiation and hence subtracts from
the energy of the incident light. Lets understand how this antenna works. This an-
tenna is a dipole that we have studied in the previous chapters.
Consider an EM wave incident on an atom with electron in ground state. We
take the EM field as plane wave travelling along x direction and electric field
E0 exp( jω0t)ẑ. Lets us denote ground state orbital by φ0 and the excited state by
φ1 . Electron can be in the superposition aφ0 + bφ1 . In this superposition state, it has
a dipole moment along x axis as

−ehaφ0 + bφ1 |x|aφ0 + bφ1 i = d(ab∗ + a∗ b), (8.1)

97
98 8 Optical resonance and colors
R
where d = −ehφ1 |x|φ0 i = −e φ1∗ φ0 x, which we take real. We will show in next
section that incident EM field will take an electron in ground state and promote it to
excited state making it go through a trajectory
ω0t ω1t ω0t ω1t
s(t) = exp(− j ) cos φ0 − j exp( j ) sin φ1 , (8.2)
2 2 2 2
E0 d
where ω1 = h̄ . This gives the dipole moments d(t) as

d(t) = −d sin ω0t sin ω1t. (8.3)

This oscillating dipole radiates. We have studied this before. Recall antennas
from chapter 4. an oscillating dipole, is seen as where two opposite charges with
charge q go back and forth along say z axis with a maximum separation of d. The
dipole moment
D(t) = d sin(ω0t) sin ω1t, (8.4)
dD
Then dt constitutes a current i0 and we make the correspondence

i0 dl = −d ω0 sin ω1t cos ω0t. (8.5)

where sin ω1t changes much slowly (adiabatically) compared to sin(ω0t).

Let us see how this radiating current element adds to the travelling wave Rmode
E . We normalize the mode such that it has unit energy (energy given by ε0 E2 ).
Then using ω1 = Eh̄0 d , the projection of the current element onto the mode is

Z
E
i0 dl E0 exp( jω0 (t − τ ))d τ , (8.6)
2
Z
h̄ω0 E
= − exp( jω0t) ω1 sin ω1t , (8.7)
2 2
E
= −h̄ω0 exp( jω0t) . (8.8)
2
1
Then the mode E → E − E h̄ω0 . How does the energy of the mode change. The
2
| {z }
Z ∆E
energy changes as −h̄ω0 ε0 E2 . Thus the field energy has reduced by h̄ω0 and we
| {z }
=1
say we have absorbed a photon.
If we start from the excited state first then the incident wave makes the electron
go through the trajectory
ω0t ω1t ω0t ω1t
s(t) = exp(− j ) cos φ0 + j exp( j ) sin φ1 . (8.9)
2 2 2 2
This gives the dipole moments d(t) as
8.1 Atom as an Antenna 99

d(t) = d sin ω0t0 sin ω1t. (8.10)

Then the mode E → E+ ∆ E. The energy changes as h̄ω0 . We say we have emitted
a photon also called stimulated emission as it is emitted in the mode that excites the
system.
Let us return to the trajectory s(t) part. Lets φ0 and φ1 denote two electronic
states (orbitals) of an atom. Let φ0 be ground state and φ1 excited state. When an
electric field is applied say in the x direction it produces a potential

V (x) = −eEx. (8.11)

at the atomic site. This potential scatters φ0 to φ and we have dipole matrix element

d = −ehφ1 |x|φ0 i. (8.12)

 of the system be given by ψ = a(t)φ0 + b(t)φ1 which we write as ψ (t) =

Let state

a(t)
. Then if hω0 is the difference in energies of φ0 and φ1 and E = E0 exp( jω0t)
b(t)
then ψ (t) evolves as
" #
d ∗ E0
i ω0 exp(− jω0t)
ψ̇ = − h̄ ψ. (8.13)
2 dEh̄ 0 exp( jω0t) −ω0
dE0
Let h̄ = ω1 ,
 
i ω0 0
Ψ (t) = exp( t)Ψ (t), (8.14)
2 −ω0
then
 
i 0 ω1
Ψ̇ = − Ψ, (8.15)
2 ω1 0
then
   
1 cos( ω21 t )
Ψ (0) = → . (8.16)
0 − j sin( ω21 t )
   
1 exp(− j ω20t ) cos( ω21 t )
ψ (0) = → . (8.17)
0 − j exp( j ω20 t ) sin( ω21 t )
This explains s(t) in Eq. (8.2).

8.1.1 Electric Dipoles

Recall Maxwell equation in a medium

100 8 Optical resonance and colors

∂B
∇×E = − , (8.18)
∂t
∂D
∇×H = . (8.19)
∂t

∇ × E = − j µω H, (8.20)
∇ × H = jεω E. (8.21)

In a material medium, we have

ε = ε0 (1 + χ (ω )), (8.22)
χ (ω ) = χ ′ (ω ) + j χ ′′ (ω ). (8.23)

χ (ω ) is called optical susceptibility and depends on the frequency. It has a real part
χ ′ (ω ) and a imaginary part χ ′′ (ω ).
r
ε p
= 1 + χ ′ (ω ) + j χ ′′ (ω ) = n + jα , (8.24)
ε0
n is the index of refraction and α absorption coefficient. Let us understand the source
of χ (ω ), the optical susceptibility.

8.1.2 Classical theory of optical susceptibility

Electrons are bound to atoms by positive charge of the nucleus. Electric field pulls
the electrons away and we can describe the displacement as

mẍ + γ ẋ + kx = −eE. (8.25)

γ −eE
ẍ + ẋ + ω02 x = , (8.26)
m m
where ω02 = mk
with E = E0 exp( jω t), we get
−eE0 1
x(ω ) = (8.27)
m (ω0 − ω 2 ) + jω mγ
2

Susceptibility is proportional to x. Fig. 8.1a and 8.1b gives the imaginary and real
part of susceptibility as function of ω . The imaginary part peaks as ω = ω0 .
8.1 Atom as an Antenna 101

Susceptibility vs frequency

Fig. 8.1

8.1.3 Quantum theory of optical susceptibility

Lets φ0 and φ1 denote two electronic states (orbitals) of an atom. Let φ0 be ground
state and φ1 excited state. When an electric field is applied say in the x direction, it
produces a potential

V (x) = −eEx. (8.28)

at the atomic site. This potential scatters φ0 to φ1 and we have dipole matrix element

d = −ehφ1 |x|φ0 i. (8.29)

 the system be given by ψ = a(t)φ0 + b(t)φ1 , which we write as

Let the state of
a(t)
ψ (t) = . Then if hω0 is the difference in energies of φ0 and φ1 and E =
b(t)
2E0 cos ω t, then ψ (t) for ω1 = 2dE 0
h̄ evolves as
 
i ω0 ω1 cos(ω t)
ψ̇ = − ψ. (8.30)
2 ω1 cos(ω t) −ω0
Let
102 8 Optical resonance and colors
 
i ω 0
Ψ (t) = exp( t)ψ (t), (8.31)
2 −ω
then
 ω1 
i (ω0 − ω ) 2 (1 + exp(2 j ω t)) Ψ .
Ψ̇ = − ω1 (8.32)
2 2 ((1 + exp(−2 jω t)) −(ω0 − ω )
We neglect the fast oscillating term exp(2 jω t) to get
 
i (ω0 − ω ) ωx
Ψ̇ = − Ψ, (8.33)
2 ωx −(ω0 − ω )
dE0
where ωx = h̄ . Then we can write
1
ρ = ΨΨ † = + sx σx + sy σy + sz σz . (8.34)
2
where,
     
1 01 1 0 −i 1 1 0
σx = ; σy = ; σz = . (8.35)
2 10 2 i 0 2 0 −1
(sx , sy , sx ) have special meaning. sz corresponds to population in ground and excited
state. When atom is in ground state sz = 1 and in excited state sz = −1. sx , sy cor-
respond to real and imaginary parts of susceptibility as shown below. Then observe
the resulting dipole is

haφ0 + bφ1 |x|aφ0 + bφ1 i, (8.36)

jω0t jω0t jω0t jω0t
= hΨ0 exp(− )φ0 + Ψ1 exp( )φ1 |x|Ψ0 exp(− )φ0 + Ψ1 exp( φ1 )i,
2 2 2 2
= (Ψ1∗Ψ0 + Ψ1Ψ0∗ ) cos ω t − i(Ψ1∗Ψ0 − Ψ1Ψ0∗ ) sin ω t, (8.37)
= sx cos ω t − sy sin ω t. (8.38)

From Eq. (8.33), we find that (sx , sy , sz ) evolves as

    
sx 0 ω − ω0 0 sx
d   
sy = ω0 − ω 0 −ωx   sy  . (8.39)
dt
sz ωx 0 0 sz
Due to dissipative processes like spontaneous emission (which we describe soon),
the above equations have two decay constants Γ1 and Γ2 , and the above equation
takes the form
      
s −Γ2 ∆ ω 0 sx 0
d  x 
sy = −∆ ω −Γ2 −ωx   sy  +  0  . (8.40)
dt
sz 0 ωx −Γ1 sz Γ1
We can solve for the steady state of above equations to find [9],
8.2 Spontaneous Emission 103

1
(sx , sy , sz )∞ = (∆ ω ωx , ωxΓ2 , Γ22 + ∆ ω 2 ). (8.41)
Γ22 + ∆ ω 2 + ωx ΓΓ21
Here sx , sy corresponds to real and the imaginary part of susceptibility and if we plot
them as function of ∆ ω , they take the form as in Fig. 8.1b and 8.1a respectively.
Observe imaginary part decays as function of ∆ ω . Also recall imaginary part is
responsible for decay in a medium as in an conductor. Now optical resonance ω0
in in energy range h̄ω0 ∼ eV or ω0 ∼ 1014 − 1015 Hz. If we use radio waves, ∆ ω
is very large, and hence χ ′′ (ω ) is negligible, loss is negligible and radio waves go
through walls. Light doesn’t, because ∆ ω is small, hence χ ′′ (ω ) is not negligible
and we have attenuation or loss. Furthermore χ ′′ (ω ) decays much faster as function
of ∆ ω than χ ′ (ω ). hence even for radio or microwaves we have non-zero χ ′ (ω ) a
medium which explains dielectric response ε as different from ε0 .

8.2 Spontaneous Emission

A B
Fig. 8.2 Fig. A shows transition from ground to excited state. Fig. B shows excited state level
broadened.

Consider two level system as shown in 8.2A. Let Ω be the rate of on resonance
excitation
  from ground state to excited state. The dynamics of the system with ψ =
ψ1
is
ψ0
 
i ω0 Ω exp(− jω0t)
ψ̇ = − ψ. (8.42)
2 Ω exp( jω0t) −ω0
 
i ω0 0
In the rotating frame ψ → exp( 2 t)ψ we have
0 −ω0
 
i 0 Ω
ψ̇ = − ψ. (8.43)
2 Ω 0
Now suppose the excited level is detuned by energy ∆ ω , then
104 8 Optical resonance and colors

 
i ω0 + 2∆ ω Ω exp(− jω0t)
ψ̇ = − ψ, (8.44)
2 Ω exp( jω0t) −ω0
 
i ω0 + 2∆ ω 0
then in the rotating frame ψ → exp( 2 t)ψ , we have
0 −ω0
 
i 0 Ω exp( j∆ ω t)
ψ̇ = − ψ. (8.45)
2 Ω exp(− j∆ ω t) 0
R
Then starting from ψ0 (0) = 1, we have ψ1 (∆ t) = 0∆ t Ω exp( j∆ ωτ )d τ . Now
consider many
R
excited levels as in 8.2B, with level ψn detuned by n∆ ω . Then
ψn (∆ t) = 0∆ t Ω exp( jn∆ ωτ )d τ . Then
Z ∆t Z ∆t
ψk∗ (∆ t)ψk (∆ t) = Ω 2 exp( jk∆ ω (t − τ ))d τ dt. (8.46)
0 0
Then Φ = ∑nk=−n ψk∗ (∆ t)ψk (∆ t) =

Z Z Z
Ω 2 ∆t ∆t B
( exp( jω (t − τ ))d ω ) d τ dt, (8.47)
∆ω 0 0 −B
Z Z
Ω 2 ∆ t ∆ t sin(B(t − τ )
= d τ dt. (8.48)
∆ω 0 0 t −τ
Ω2
= ∆ t. (8.49)
∆ω
where we use the approximation B ≫ Ω . There is time ∆ t such that Ω ∆ t ≪ 1 and
B∆ t ≫ 1. Then

Φ Ω2
Γ= = , (8.50)
∆t ∆ω
gives the transition rate out of ground state.
With Eq. 8.50 also known as Fermi Golden Rule, we can study the spontaneous
emission rate of an atom. It is the rate at which atom decays into ground sate. Let
h̄ω0 be the energy difference between the excited and ground state. When atom de-
cays from excited to ground sate it radiates EM wave mode. These modes are plane
waves with defined k = (kx , ky , kz ). We give the mode some width (∆ kx , ∆ ky , ∆ kz ),
∆k ∆k ∆k
which gives the mode a finite volume V −1 = x(2π )y3 z . Fig. 8.3 depicts this width
for a mode with k along x axis. The energy of this mode is h̄ω0 and hence it has
electric field ε0 E 2V = h̄ω0 and hence the transition rate due to this electric field
Ω = Ed h̄ . Infact, transition rate is cos θ Ω , where θ is the vector k vector makes with
z axis as transition is induced by electric field in the equatorial plane. The mode
has width ∆ k along radial direction and hence frequency width of ∆ ω = c∆ k. The
modes are spread radially with energy spacing of ∆ ω . Therefore by Fermi golden
2
rule the decay rate Γ is Γ = Ω∆ ωM , where M is the number of modes on the surface
of sphere of radius k0 . This is
8.2 Spontaneous Emission 105

Z
M = 2π k02 ∆ k cos2 θ sin θ d θ ,

where cos2 θ comes from the factor cos θ in transition rate. When we put everything
together we get [4]

d 2 ω03 d 2 ω03
Γ= 2 3
= . (8.51)
6π c h̄ε0 3π c3 hε0

k−sphere

Fig. 8.3 Fig. shows blowup of a mode depicting it has some width in k-space.

8.2.1 Spontaneous Emission in a Resonant Cavity and Purcell effect

Now consider a resonant cavity, with volume V . Let its natural frequency be ω0 . Let
Q be the quality factor of the cavity and consider a two level atom in the cavity with
natural frequency ω0 . We are interested in the spontaneous emission rate of the atom
in the cavity. This rate is the same rate as exciting the atom from the ground state
with electric field worth h̄ω0 worth of energy, i.e., ε0 E02V = h̄ω0 . The corresponding
transition frequency is Ω = Eh̄0 d . The electric field decays as E0 exp(−α t), with ωα0 =
Q. The transition with decaying field E can be simulated by a broadened transition
with n excited levels, with frequency spacing ∆ ω , such that n∆ ω = α , and transition
frequency √Ωn . Then by the Fermi Golden rule we have

Ω2 Q d2
Γ= = . (8.52)
n∆ ω V ε h̄
106 8 Optical resonance and colors

c3
Observing that ω03 ∝ V , we get

d 2 ω03
Γ =Q . (8.53)
c3 ε h̄
Thus there is huge enhancement of spontaneous emission rate over vacuum if quality
factor Q is high. This enhancement is called Purcell effect.

8.2.2 Line width in spontaneous emission

Refer back to Eq. 8.6, where the emitted light s(t) is of the form s(t) ∝ ω1 sin2ω1 t exp(− jω0t).
Recall the excited state population is p(t) = cos2 ω21 t and hence we can write the
emitted light as s(t) ∝ − ddtp exp(− jω0t). In spontaneous emission the with rate Γ ,
the excited state population is p(t) = exp(−Γ t), and hence s(t) ∝ Γ exp(−Γ t) exp(− jω0t).
If we take Fourier transform of s(t) we get S(ω ) such that

Γ2
|S(ω )|2 = . (8.54)
(ω − ω0 )2 + Γ 2

Linewidth vs Frequency
1
|S( ω)| 2

0.5

0
-10 -5 0 5 10
(ω-ω0 )/ Γ
Fig. 8.4 Fig. plots line-width vs frequency in spontaneous emission.
8.4 World of Colors 107

8.3 Lasers and Cavities

LASER (Light amplification by stimulated emission of radiation) is a device that

amplifies light. Imagine, you have a cavity, with small light in a cavity mode, that
is resonant with two level atoms in the cavity, i.e., the energy difference between
excited and ground states is h̄ω0 , where ω0 is the frequency of the cavity mode. If
atoms are prepared in all excited state, then the small light will induce stimulated
emission in the atoms and as the atoms go from excited to the ground state, they
radiate into the same cavity mode. Suppose , the small light electric field at atom
is of the form E0 exp( jω0t). Then it induces a dipole in the atom as studied earlier
of the form d0 sin(ω0t), with a current equivalent of i0 cos(ω0t), which radiates into
the mode and the mode builds up as t exp( jω0t) as discussed earlier. Everything is
in the same phase as the mode builds up. The small light can simply come from
spontaneous emission from one of the atoms itself. We contrast this situation to the
situation, where atoms spontaneously emit into cavity mode at random times ti In
which case the mode amplitude at time t will look like ∑i exp( jω0 (t − ti )) and light
is incoherent with photons with different phases ω0 (t − ti ). Stimulated emission in
a laser gives the light the same phase. When enough light builds up the cavity mode
it is emitted.

8.4 World of Colors

We are surrounded by beautiful colors. Really beautiful. Our clothes have beauti-
ful colors on them. Plants are green, blood is red. What is the source of this color.
Clothes have dyes containing pigments, like paints have pigments. These pigments
are primarily inorganic in the sense, they have a transition metal element in them.
Transition metal elements are the one found in the center of the periodic table that
have electrons in their d-orbitals. These include, for example, Cobalt (Co) , Cad-
mium (Cd), Chromium (Cr), Manganese (Mn) etc. For example, Cobalt (atomic
number 27), has electronic configuration 1s2 2s2 2p6 3s2 3p6 3d 7 4s2 . The d-orbitals
are five fold degenerate. These orbitals are dz2 , dx2 −y2 , dxy , dyz , dxz . However in a
transition metal compound, binding with other atoms called ligands, this degener-
acy gets broken. We have orbitals dz2 , dx2 −y2 called eg manifold at higher energy
than the orbitals dxy , dyz , dxz called t2g manifold as shown below
The energy difference ∆ = h̄ω0 is sub-eV and corresponds to visible wavelength.
If we treat upper and lower manifolds as a two level atom, then the system has
very high (imaginary) susceptibility χ (ω0 ) (as discussed in optical resonance, the
imaginary susceptibility is high on resonance). Hence ω0 color light is reflected.
This is the primary source of color. There is a small caveat though, if we write
spatial part of wave-functions of d orbitals we find
108 8 Optical resonance and colors

e
g z2 x2 − y2
∆

t 2g
xy xz yz

Fig. 8.5 Fig. shows splitting of energy of d-orbitals in eg manifold and t2g manifold.

dz2 ∝ f (r)(3z2 − r2 ), (8.55)

2 2
dx2 −y2 ∝ f (r)(x − y ), (8.56)
dxy ∝ f (r)xy, (8.57)
dyz ∝ f (r)yz, (8.58)
dxz ∝ f (r)xz. (8.59)

where f (r) simply means r dependence.

There is just no dipole transition between these orbitals. How will we absorb
light ? How do we see colors then. The answer is there is indirect transition between
these orbitals mediated through a higher p orbital. This is called Raman effect. If we
write spatial part of wavefunctions of d orbitals, we find

pz ∝ f (r)z, (8.60)
px ∝ f (r)x, (8.61)
py ∝ f (r)y. (8.62)

Then we can see dipole transition between d and p orbitals.

8.4.1 Raman Effect and Color

We first describe a three level atomic system so called Λ system. In a Λ system

as shown in Fig. 8.6 we have two ground state levels |1i and |3i at energy E1 and
excited level |2i at energy E2 . The transition from |1i to |2i has strength Ω1 and
transition from |2i to |3i has strength Ω2 . In the interaction frame of natural Hamil-
tonian of the system, we get a second order term connecting level |1i to |3i with
strength (EΩ1−E
Ω2
. This term creates an effective coupling between ground state lev-
1 2)
els and drives transition from |1i to |3i. To see this observe,
8.4 World of Colors 109

2
E2

Ω1 Ω2

1 3

E1 E1

Fig. 8.6 Above Fig. shows a three level Λ system with two ground state levels |1i and |3i and an
excited level |2i.

The state of the three level system evolves according to the Schröedinger equa-
tion
 
E Ω∗ 0
−i  1 1 ∗ 
ψ̇ = Ω1 E2 Ω2 ψ . (8.63)
h̄
0 Ω2 E1
We proceed into the interaction frame of the natural Hamiltonian (system energies)
by transformation
 
E 0 0
i 1
φ = exp( 0 E2 0 )ψ . (8.64)
h̄
0 0 E1
This gives for ∆ E = E2 − E1 ,

 
0 exp(− h̄i ∆ E t)Ω1∗ 0
−i 
φ̇ = exp( h̄i ∆ E t)Ω1 0 exp( h̄i ∆ E t)Ω2∗  φ . (8.65)
h̄ i
0 exp(− h̄ ∆ E t)Ω2 0
| {z }
H(t)

2π
H(t) is periodic with period ∆ t = ∆E . After ∆ t, the system evolution is

Z ∆t Z ∆ t Z σ1
φ (∆ t) = (I + H(σ )d σ + H(σ1 )H(σ2 )d σ2 d σ1 + . . . )φ (0). (8.66)
0 0 0

The first integral averages to zero, while the second integral

110 8 Optical resonance and colors

Z ∆ t Z σ1 Z ∆ t Z σ1
1
H(σ1 )H(σ2 )d σ2 d σ1 = [H(σ1 ), H(σ2 )]d σ2 d σ1 . (8.67)
0 0 2 0 0

Evaluating it explicitly, we get for our system that second order integral is
 
Ω1∗ Ω2∗
0 E1 −E2 

−i∆ t  0 0 0 
 
 Ω1 Ω2 . (8.68)
h̄  0 0 
 E1 − E2 
| {z }
M

Thus we have created an effective Hamiltonian

 Ω1∗ Ω2∗

0 E1 −E2
 
 0 0 0 , (8.69)
Ω1 Ω2
E1 −E2 0 0
which couples level |1i and |3i and drives transition between them at rate M =
Ω1 Ω2
(E −E ) .
1 2

p
ε2

ε3
eg
ε1
t 2g

Fig. 8.7 Above Fig. shows transition between t2g and eg manifolds mediated by higher lying p
orbital.

Now lets come back to color. Consider fig. 8.7, where two d levels have energy
ε1 , ε3 and one in top p level at energy ε2 .
Now suppose we start with level 1 and we bring in optical photon of frequency
ω1 , then lets say our initial energy is E1 = h̄ω1 + ε1 , the atomic energy plus the
photon energy. This photon induces a transition to level 2 which has energy E2 = ε2
and finally we can emit a photon of frequency ω2 and make a transition to level
3 . The total energy of the emitted photon plus atomic energy is E3 = h̄ω2 + ε3 .
Observe E1 = E3 when h̄(ω1 − ω2 ) = ε3 − ε1 = ∆ E. Then we have constructed
8.5 Scattering and Blue Sky 111

a three level system from joint atom-photon states. In this three level system that
we have constructed we have a transition amplitude between initial state which is
atom in level 1 and photon in ω1 to final state, atom in level 3 and photon ω2 . The
transition amplitude goes as M = (EΩ1−EΩ2
) , where E1 − E2 = h̄ω1 − (ε2 − ε1 ).
1 2
When h̄ω1 is close to ε2 − ε1 , we get a big enhancement in M = (EΩ1−E
Ω2
. Observe
1 2)
for the emitted photon h̄ω2 = h̄ω1 − ∆ E. Suppose ∆ E is green color then ω2 will be
complementary color of green, i.e., red. This emitted photon is the complementary
color we see.

8.5 Scattering and Blue Sky

Why is sky Blue? The dust particles in atmosphere act like dipoles. The induced
current i ∝ dD
dt , where D is induced dipole (arising due to incident radiation) on
them is proportional to i ∝ ω , the frequency of incident radiation. Then from 4.22
and 4.21, E ∝ iω ∝ ω 2 and E ∝ iω ∝ ω 2 and we find emitted power P ∝ ω 4 . More
power is scattered at higher frequencies and we see blue sky. What do we understand
by scattering? Dust particles are like antenna array. The emission from the antenna
array from Eq. (4.38) has the factor

sin N2p
η∝ (8.70)
sin 2p

where p = kd sin υ . The spacing between dust particles is much bigger than wave-
length and as a result kd ≫ 2π and hence 2p goes through multiples of π as υ is
varied and hence η peaks at may υ . We say antenna array then scatters light. Its
emitted light is maximum not is one direction but multiple directions.
References 113

References

1. R.P. Feynman, R.B. Leighton, and M. Sands, “The Feynaman Lectures on Physics” Vol II,
Narosa Publishing House, (1986).
2. E. M. Purcell and D. Morin, “Electricity and Magnetism”, Cambridge University Press
(2013).
3. D. Griffiths, “Introduction to Electrodynamics”, Prentice Hall (1999).
4. M. O. Scully and M.S. Zubairy, “Quantum Optics” Cambridge University Press (1997).
5. Eugene Hecht, “Optics” Pearson (2014).
6. A. Das and S.K. Das, “Microwave Engineering” McGraw Hill, New Delhi (2015).
7. S. M. Sze, “Semiconductor Devices: Physics and Technology”, Wiley (1985).
8. P. Horowitz, W. Hill, “Art of Electronics”, Cambridge University Press (1989).
9. D. Suter, “ The physics of laser-atom interaction” Cambridge University Press (1997).
10. A. Abragham, “Principles of Nuclear Magnetism” Oxford University Press (2006).
11. R. K. Shevgaonkar, “Electromagnetic Waves” Tata McGraw Hill, New Delhi (2006).
12. V. Ramachandran, K. Shankar and R.K. Shevgaonkar, “Transmission lines and Networks”,
Tata McGraw-Hill (1994).
13. D.R. Wehner, “High Resolution Radar” Artech House, Boston (1995).
14. M.I. Skolnik, “Introduction to Radar Systems” McGraw Hill (1962).
15. T. Pratt and C.W. Bostian, “Satellite Communication” Wiley (1986).
16. C.A., Balanis “Antenna Theory” John Wiley and Sons (ASIA), Singapore (2002).
17. L. Boithias “Radio Wave Propagation” North Oxford Academic Publishers (1987).
18. D.K. Cheng, “Field and Wave Electromagnetics” Pearson Education, Delhi (2002).
19. R.E. Collin, “Antennas and Radio Wave Propagation” McGraw Hill, New York (1985).
20. R.S. Elliott, “Antenna Theory and Design” Prentice-Hall, New Jersey (1981).
21. J. Griffiths, “Radio Wave Propagation and Antennas” Prentice Hall International, London
(1987).
22. E.C. Jordan and K.G. Balmain, “Electromagnetic Waves and Radiating Systems” Prentice
Hall of India, New Delhi (2004).
23. J.D. Kraus, “Radio Astronomy” McGraw Hill, New York (1966).
24. J.D. Kraus, “Antennas” McGraw Hill, Singapore (1988).
25. D.M. Pozar, “Microwave and RF Wireless Systems”, John Wiley and Sons, New Jersey
(2001).
26. A.W. Snyder and J. D. Love, “Optical Waveguide Theory” Chapmann and Hall, London,
1983.
27. D.M. Pozar, “Microwave Engineering”, John Wiley and Sons, Singapore (2003).
28. S. Ramo, J.R. Whinnery, and T. Van Duzer “Fields and Waves in Communication Electronics”
John Wiley and Sons, Singapore (2004).
29. W. L. Stutzman and G. T. Thiele, “Antenna Theory and Design” John Wiley and Sons (1987).
30. W. N. Christiansen and J.A. Hogbom, “Radio Telescopes (Monographs of Physics)” Cam-
bridge University Press (1985).
31. A.R. Harish and M. Sachidananda, “Antennas and Wave Propagation” Oxford University
Press (2007).