Microwave Devices and Circuits Samuel Liao
Microwave Devices and Circuits Samuel Liao
Microwave Devices and Circuits Samuel Liao
Devices
and Circuits
Third Edition
SAMUEL Y. LIAO
Professor of Electrical Engineering
California State University, Fresno
Iii
Contents
PREFACE
Chapter O
Chapter 1
xv
INTRODUCTION
0-1
Microwave Frequencies
0-2
Microwave Devices
0-3
Microwave Systems
0-4
Introduction
1-1
1-2
10
1-3
Problems
Suggested Readings
Chapter 2
12
15
15
Introduction
2-1
16
16
17
Contents
vi
2-2
2-3
Poynting Theorem
2-4
2-5-3
2-6
32
References
57
Suggested Readings
Problems
Chapter 3
21
2-5
19
57
58
Introduction
Reflection Coefficient, 67
Transmission Coefficient, 69
3-4
61
Transmission-Line Equations. 61
Solutions of Transmission-Line Equations, 64
3-3
61
3-2
61
Standing Wave, 71
Standing-Wave Ratio, 74
Line Impedance, 76
Line Admittance, 81
76
71
67
vii
Contents
3-5
3-6
Smith Chart
Impedance Matching
3-6-1
3-6-2
3-7
82
Problems
96
98
Suggested Readings
Chapter 4
89
Single-Stub Matching, 90
Double-Stub Matching, 92
98
98
4-0
4-1
Introduction
4-2
102
Rectangular Waveguides
4-1-1
4-1-2
4-1-3
4-1-4
4-1-5
4-1-6
4- 1-7
102
103
Circular Waveguides
119
4-2-1
4-2-2
4-2-3
4-2-4
4-2-5
4-3
Microwave Cavities
4-3-1
4-3-2
4-3-3
4-4
4-5
135
Directional Couplers
4-5-1
4-5-2
4-5-3
141
149
Contents
viii
4-6
References
161
Suggested Readings
Problems
Chapter 5
161
161
Introduction
5-2
5-3
169
193
References
204
Suggested Readings
Problems
205
205
208
Introduction
208
6-2
166
166
Chapter 6
156
209
ix
Contents
6-3
6-4
6-5
6-6
245
254
References
261
Suggested Readings
Problems
Chapter 7
262
263
269
Introduction
7-1
7-2
269
Background, 270
Gunn Effect, 271
7-3
230
Modes of Operation
7-3-1
7-3-2
7-3-3
7-3-4
273
284
7-4
LSA Diodes
7-5
InP Diodes
291
293
Contents
x
7-6
7-7
CdTe Diodes
296
7-7-2
References
300
Suggested Readings
Problems
Chapter 8
301
302
Introduction
303
Read Diode
304
8-1-1
8-1-2
8-1-3
8-1-4
8-2
8-3
8-4
314
BARITT Diodes
8-4-1
8-4-2
8-4-3
8-5
309
TRAPATT Diodes
8-3-1
8-3-2
8-3-3
317
Parametric Devices
8-5-1
8-5-2
8-5-3
8-5-4
320
References
331
Suggested Readings
Problems
332
333
Introduction
335
335
303
IMPATI Diodes
8-2-1
8-2-2
8-2-3
Chapter 9
296
338
xi
Contents
9-1-2
9-1-3
9-2
Klystrons
9-2-1
9-2-2
9-2-3
9-2-4
9-2-5
9-3
9-3-3
362
Reflex Klystrons
9-4-1
9-4-2
9-4-3
9-5
341
9-4
373
382
9-6
9-7
404
References
417
Suggested Readings
Problems
Chapter 10
398
418
419
I0-0
Introduction
I0-1
Magnetron Oscillators
10-1-1
10-1-2
425
427
425
xii
Contents
10-1-3
10-1-4
10-1-5
10-1-6
10-2
10-3
10-4
457
References 466
Problems 467
Chapter 11
472
STRIP LINES
11-0
Introduction
11-1
Microstrip Lines
11-1-1
11-1-2
11-1-3
11-2
473
485
l l-3
488
11-4
489
491
References
Problems
Chapter 12
472
492
12-0
Introduction
12-1
Materials
12-1-1
12-1-2
12-1-3
12-1-4
12-2
495
497
495
501
--xiii
Contents
12-3
MOSFET Fabrication
504
12-4
Thin-Film Formation
12-4-1
12-4-2
12-4-3
12-5
514
521
Suggested Readings
Problems
519
522
522
APPENDIX A
523
APPENDIX B
529
INDEX
535
Preface
This third revision has been designed, as have the first two editions, for use in a first
course in microwave devices and circuits at the senior or beginning graduate level in
electrical engineering. The objectives of this book are to present the basic principles,
characteristics, and applications of commonly used microwave devices and to explain the techniques for designing microwave circuits. It is assumed that readers of
this text have had previous courses in electromagnetics and solid-state electronics.
Because this book is self-contained to a large extent, it also serves as a convenient
reference for electronics engineers working in the microwave field.
The format of this edition remains the same, but there are additions and expansions as well as some corrections and deletions. The problems section has been enlarged and includes new and very practical problems. The book is reorganized into
twelve chapters.
Chapter 1 discusses the interactions between electrons and fields.
Chapter 2 deals with plane-wave propagation in different media.
Chapter 3 treats transmission lines.
Chapter 4 analyzes microwave waveguides and components.
Chapter 5 describes microwave transistors and tunnel diodes, and includes heterojunction bipolar transistors (HBTs).
Chapter 6 treats microwave field-effect transistors such as JFETs, MESFETs,
HEMTs, MOSFETs and the NMOS, CMOS, and the charged-coupled devices
(CCDs).
Chapter 7 discusses transferred electron devices (TEDs), including the Gunn,
LSA, InP, and CdTe diodes.
Chapter 8 describes avalanche transit-time devices such as the IMP ATT,
TRAPATT, and BARITT diodes and the parametric devices.
xv
xvi
Preface
Samuel Y. Liao
Chapter O
Introduction
The central theme of this book concerns the basic principles and applications of microwave devices and circuits. Microwave techniques have been increasingly adopted
in such diverse applications as radio astronomy, long-distance communications,
space navigation, radar systems, medical equipment, and missile electronic systems.
As a result of the accelerating rate of growth of microwave technology in research
and industry, students who are preparing themselves for, and electronics engineers
who are working in, the microwave area are faced with the need to understand the
theoretical and experimental design and analysis of microwave devices and circuits.
01 MICROWAVE FREQUENCIES
The term microwave frequencies is generally used for those wavelengths measured in
centimeters, roughly from 30 cm to 1 mm (1 to 300 GHz). However, microwave really indicates the wavelengths in the micron ranges. This means microwave frequencies are up to infrared and visible-light regions. In this revision, microwave frequencies refer to those from 1 GHz up to 106 GHz. The microwave band designation that
derived from World War II radar security considerations has never been officially
sanctioned by any industrial, professional, or government organization. In August
1969 the United States Department of Defense, Office of Joint Chiefs of Staff, by
message to all services, directed the use of a new frequency band breakdown as
shown in Table 0-1. On May 24, 1970, the Department of Defense adopted another
band designation for microwave frequencies as listed in Table 0-2. The Institute of
Electrical and Electronics Engineers (IEEE) recommended new microwave band
designations as shown in Table 0-3 for comparison.
1
Introduction
TABLE 01
Designation
0.225- 0.390
0.390- 1.550
1.550- 3.900
3.900- 6.200
6.200- 10.900
10.900- 36.000
36.000- 46.000
46.000- 56.000
56.000-100.000
P
L
S
C
X
K
band
band
band
band
band
band
Q band
V band
W band
TABLE 02
Designation
A
B
C
D
E
F
G
band
band
band
band
band
band
band
Chap. 0
Designation
0.100-0.250
0.250-0.500
0.500-1.000
1.000-2.000
2.000-3.000
3.000-4.000
4.000-6.000
H band
I band
J band
K band
L band
M band
6.000- 8.000
8.000- 10.000
10.000- 20.000
20.000- 40.000
40.000- 60.000
60.000-100.000
TABLE 03
Designation
HF
VHF
UHF
L band
S band
C band
X band
Ku band
K band
Ka band
Millimeter
Submillimeter
0.003- 0.030
0.030- 0.300
0.300- 1.000
1.000- 2.000
2.000- 4.000
4.000- 8.000
8.000- 12.000
12.000- 18.000
18.000- 27.000
27.000- 40.000
40.000-300.000
>300.000
02 MICROWAVE DEVICES
In the late 1930s it became evident that as the wavelength approached the physical
dimensions of the vacuum tubes, the electron transit angle, interelectrode capacitance, and lead inductance appeared to limit the operation of vacuum tubes in microwave frequencies. In 1935 A. A. Heil and 0. Heil suggested that microwave
voltages be generated by using transit-time effects together with lumped tuned cir-
Sec. 0.3
Microwave Systems
cuits. In 1939 W. C. Hahn and G. F. Metcalf proposed a theory of velocity modulation for microwave tubes. Four months later R. H. Varian and S. F. Varian described
a two-cavity klystron amplifier and oscillator by using velocity modulation. In 1944
R. Kompfner invented the helix-type traveling-wave tube (TWT). Ever since then the
concept of microwave tubes has deviated from that of conventional vacuum tubes as
a result of the application of new principles in the amplification and generation of
microwave energy.
Historically microwave generation and amplification were accomplished by
means of velocity-modulation theory. In the past two decades, however, microwave
solid-state devices-such as tunnel diodes, Gunn diodes, transferred electron
devices (TEDs), and avalanche transit-time devices have been developed to perform
these functions. The conception and subsequent development of TEDs and avalanche
transit-time devices were among the outstanding technical achievements. B. K. Ridley and T. B. Watkins in 1961 and C. Hilsum in 1962 independently predicted that
the transferred electron effect would occur in GaAs (gallium arsenide). In 1963 J. B.
Gunn reported his "Gunn effect." The common characteristic of all microwave solidstate devices is the negative resistance that can be used for microwave oscillation and
amplification. The progress of TEDs and avalanche transit-time devices has been so
swift that today they are firmly established as one of the most important classes of
microwave solid-state devices.
03 MICROWAVE SYSTEMS
A microwave system normally consists of a transmitter subsystem, including a microwave oscillator, waveguides, and a transmitting antenna, and a receiver subsystem that includes a receiving antenna, transmission line or waveguide, a microwave
amplifier, and a receiver. Figure 0-1 shows a typical microwave system.
In order to design a microwave system and conduct a proper test of it, an adequate knowledge of the components involved is essential. Besides microwave
devices, the text therefore describes microwave components, such as resonators, cavities, microstrip lines, hybrids, and microwave integrated circuits.
Microwave
source
Transmitting
horn antenna
Receiving
horn antenna
Waveguide
termination
Crystal
mount
Figure 0-1
Microwave system.
Introduction
Chap. 0
Microwave measures can be expressed in different units, such as the CGS (centimeter-gram-second) unit, MKS (meter-kilogram-second) unit, or another unit. The
meter-kilogram-second units (the International System of Units) are used throughout
unless otherwise indicated. Table 0-4 lists the most commonly used MKS units.
The prefixes tabulated in Table 0-5 are those recommended by the InternaTABLE 04
MKS UNITS
Quantity
Unit
Capacitance
Charge
Conductance
Current
Energy
Field
Flux linkage
Frequency
Inductance
Length
Power
Resistance
Time
Velocity
Voltage
Note:
Symbol
3 x 10- 6 ESU
TABLE 0-5
PREFIXES
Prefix
Factor
Symbol
exa
peta
tera
giga
mega
kilo
hecto
deka
deci
centi
mi Iii
micro
nano
pico
fem to
atto
101
10"
1012
10
106
103
102
10
10-1
10-2
10- 3
10-6
10-
10-12
10-15
10-1
p
T
G
M
k
h
da
d
c
m
,
n
p
f
a
F
Q
u
A
J
E
.,,
Hz
H
n
v
Sec. 0.4
tional Committee on Weights and Measures. They have been adopted by the National Bureau of Standards and are used by the Institute of Electrical and Electronics
Engineers, Inc.
The physical constants commonly used in the text are listed in Table 0-6.
The temperature scales commonly used in scientific work, engineering design,
and everyday life are shown in Table 0-7. Many engineering computations use the
absolute temperature in degrees Kelvin, and therefore a temperature conversion is
necessary to convert the temperatures from either centigrade or Fahrenheit to Kelvin
scale.
TABLE 06
PHYSICAL CONSTANTS
Constant
Symbol
Boltzmann constant
Electronvolt
Electron charge
Electron mass
Ratio of charge to mass of an electron
Permeability of free space
Permittivity of free space
Planck's constant
Velocity of light in vacuum
k
eV
q
m
elm
TABLE 07
Value
1.381
1.602
1.602
9.109
1.759
1.257
8.854
6.626
2.998
f.J.o
0
h
c
x
x
x
x
x
x
x
x
x
10-23 JloK
10-19 J
10-19 c
10-31 kg
10 11 Clkg
10- 6 Him or 47T x 10- 7 Him
10- 12 Flm
10-3Js
108 mis
TEMPERATURE SCALES
Rankine
OR
Fahrenheit
OF
Centigrade
Kelvin
OK
o('
Boiling point
671.4
212
100
373
Ambient point
540
80.6
27
300
Ice point
491.4
32
273
00
Absolute Lero
00
-459.4
-273
00
Chapter 1
Interactions Between
Electrons and Fields
1-q.}NTRODUCTION
In this chapter we are concerned with electron-field interactions. The motion of the
electron beam is assumed to be in a uniform electric field, or a uniform magnetic
field, or a uniform electromagnetic field because the inhomogeneous differential
equations governing the motion of an electron beam in a field involve three dimensions and their solutions in a nonuniform field are, in most cases, extremely difficult
to obtain and usually cannot be determined exactly. On the other hand, fortunately,
all present microwave devices employ a uniform field for the electron-field interactions.
Our primary purpose here is to provide the reader with a background for understanding the electron-field interactions in microwave devices that will be discussed in later chapters.
11 ELECTRON MOTION IN AN ELECTRIC FIELD
In describing fields and electron-field interactions, certain experimental laws of electricity and magnetism are covered first. The fundamental force law of charges is
Coulomb's law, which states that between two charges there exists either an attractive or a repulsive force, depending on whether the charges are of opposite or like
sign. That is,
newtons
6
(1-1-1)
where Q
Sec. 1.1
= charge in coulombs
8.854 x 10- 12 =
9
7T x 10- F/m is the permittivity of free space
36
R = range between the charges in meters
u = unit vector
eo =
It should be noted that since the MKS system is used throughout this text, a factor of
47T
E =
Qm
2 llRm
m=l 47TEoRm
(1-1-3)
In order to determine the path of an electron in an electric field, the force must be
related to the mass and acceleration of the electron by Newton's second law of motion. So
dv
(1-1-4)
F = -eE =ma= mdt
where m = 9.109 x 10- 31 kg, mass of electron
a = acceleration in meters per second squared
v = velocity of electron in meters per second
e = 1.602 x 10- 19 C, charge of electron that is negative
It can be seen that the force is in the opposite direction of the field because the elec-
tron has a negative charge. Thus when an electron moves in an electric field E, it experiences a force -e E newtons. The differential equations of motion for an electron
in an electric field in rectangular coordinates are given by
d 2x
e
= --Ex
dt 2
m
d2y
(1-1-Sa)
dt 2
--E
m Y
(1-1-Sb)
d 2z
dt
e
m
(1-I-5c)
-=
--E
2
where e/m = 1.759 x 10 11 C/kg is the ratio of charge to mass of electron and Ex,
Ey, Ez are the components of E in rectangular coordinates.
In many cases, the equations of motion for electrons in an electric field in
cylindrical coordinates are useful. The cylindrical coordinates (r, cf>, z) are defined
as in Fig. 1-1-1.
It can be seen that
Chap. 1
z
u,
u<t>
Figure 1-1-1
Cylindrical coordinates.
r cos
<f>
(l-l-6a)
= r sin
z=z
<f>
(l-l-6b)
(l-l-6c)
and, conversely,
r = (x2 + y2)1;2
A,.
'f'
= an
(l-l-7a)
(y) -- sm - (x2 +y
I
_
-I
X
y2)1;2 - cos (x2 + y2)1/2
z=z
(l-l-7b)
(l-l-7c)
A system of unit vectors, Ur, U<t>, Dz, in the directions of increasing r, <f>, z, respectively, is also shown in the same diagram. While Dz is constant, Ur and D<t> are functions of <f>; that is,
(l-l-8a)
Ur = COS <f> Ux + sin <f> Uy
D<t> = -
sin
<f> Ux
+ cos
<f>
(l-l-8b)
<f> Uy
yields
dur
def>
-=u<t>
(l-l-9a)
du<t>
def>
(l-l-9b)
= -ur
= rUr + ZUz
(l-l-9c)
Differentiation of Eq. (l- l -9c) with respect to t once for velocity and twice for acceleration yields
dp
dr
dur
dz
v = - = -u + r - + -u
dt
dt r
dt
dt z
dr
dt
= -ur
def>
dt
+ r-u<t> +
dz
dt
-uz
dr
dt r
def>
dt
dur
dz
+ -u
def>
dt z
-u + r - -
(1-1-10)
Sec. 1.1
= dv =
dt
[d2r dt 2
[d2r dt 2
r(d<!>)2]D
dt
+
r
dt dt
dt
r dt
dt
(1-1-11)
dt
Therefore the equations of motion for electrons in an electric field in cylindrical coordinates are given by
d2r - r (d</>)2
dt 2
dt
=-
!!_ E,
m
(1-1-12a)
(1-1-12b)
r dt
dt
d 2z
e
m
-=
--E
2
dt
(1-1-12c)
BE
7f"B v dv
de=
(1-1-13)
VA
J
8
= -
E de
(1-1-14)
Substitution of Eq. ( 1-1-14) in Eq. (1-1-13) and integration of the resultant yield
eV
= 2m(v~ - vl)
(1-1-15)
The left side of Eq. (1-1-15) is the potential energy, and the right side represents the
change in kinetic energy. The unit of work or energy is called the electron volt (eV),
which means that if an electron falls through a potential of one volt, its kinetic
energy will increase 1 eV. That is,
1 eV
= (1.60
x 10- 19 C)(lV)
1.60 x 10- 19 J
(1-1-16)
If an electron starts from rest and is accelerated through a potential rise of V volts,
its final velocity is
v
2 eV)
= ( -;;;-
112
= 0.593
x 106 Vv
m/s
(1-1-17)
ldVI
=Ede
(1-1-18)
10
Chap. 1
In vector notation it is
E
= -VV
(1-1-19)
where the symbol V is the vector operator in three coordinate systems. The minus
sign implies that the field is directed from regions of higher potential to those of
lower potential. Equation (1-1-19) is valid in regions in which there is space charge
as well as in regions that are free of charge.
QV-X B
(1-2-1)
= -ev x
(1-2-2)
= _ .!__
(s
dz _ B dx)
dt
'dt
(l-2-3b)
= _!_(Bvdx _ B,dy)
(l-2-3c)
dt
dt
Since
(l-2-4)
the equations of motion for electrons in magnetic field for cylindrical coordinates
can be given by
dzr - r(d<f>)2
dt 2
dt
~~
r dt
(rz def>)
=
dt
dzz
dt 2
dt
dt
=-
.!__
dt
dt
(s4' drdt -
Brr def>)
dt
(l-2-Sa)
(l-2-5b)
(l-2-Sc)
Sec. 1.2
11
"x
"x
Consider next an electron moving with a velocity of Vx to enter a constant uniform magnetic field that is perpendicular to Vx as shown in Fig. 1-2-1. The velocity
of the electron is assumed to be
( 1-2-6)
V = Vxllx
(1-2-7)
Bzllz
Then the magnetic force at the instant when the electron just enters the magnetic
field is given by
F = -ev x B = evBuy
(1-2-8)
This means that the force remains constant in magnitude but changes the direction
of motion because the electron is pulled by the magnetic force in a circular path.
This type of magnetic force is analogous to the problem of a mass tied to a rope and
twirled around with constant velocity. The force in the rope remains constant in
magnitude and is always directed toward the center of the circle and thus is perpendicular to the motion. At any point on the circle the outward centrifugal force is
equal to the pulling force. That is,
mv 2
R
-- =
evB
(1-2-9)
mv
eB
meters
(1-2-10)
eB
w =-=-
rad/s
(1-2-11)
12
Chap. 1
21Tm
eB
T=-=--
seconds
(1-2-12)
It should be noted that the radius of the path is directly proportional to the velocity
of the electron but that the angular frequency and the period are independent of velocity or radius. This means that faster-moving electrons or particles traverse larger
circles in the same time that a slower-moving particle moves in a smaller circle.
This very important result is the operating basis of such microwave devices as
magnetic-focusing apparatus.
-e(E
dv
dt
+ v x B) = m-
(1-3-1)
The equations of motion for electrons in a crossed field are expressed in rectangular
coordinates and cylindrical coordinates, respectively, as
d2x
dt 2
= - .!_(Ex + Bz dy
- B dz)
Y dt
(l-3-2a)
dz- - Bdx)
-- E +B
m _v
x dt
z dt
(1-3-2b)
e ( E + B -dx - B -dy)
-.v dt
x dt
m z
(1-3-2c)
dt
e(
(1-3-3a)
Sec. 1.3
13
_!_
r dt
(r
d</>)
dt
2
d z
dt 2
= _ !!_
m
dt
(, +
dt
(l-3-3b)
(l-3-3c)
where
d<f>
e
- =We =-B
dt
m
It is, of course, difficult to solve these equations for solutions in three dimensions. In
microwave devices and circuits, however, only one dimension is involved in most
cases. So the equations of motion become simple and can easily be solved. An example may show how to solve some of the preceding equations.
Example 1-3-1:
The inner cylinder of radius a is the cathode and the outer shell with radius b is the
anode. A de voltage V0 is applied between the anode and cathode, and a magnetic flux
density B is into the page as shown in Fig. 1-3-1. The problem is to adjust the applied
voltage Vo and the magnetic flux density B to such levels that the electrons emitted
from the cathode will just graze the anode and travel in space between the cathode and
the anode only.
E,
Anode
V= V0
Solution
a.
b.
2. From (b)
I~ (r2d<f>)
r dt
dt
= !_Bz dr
dt
14
Chap. 1
def>
1
a 2 - = -w.a 2 +constant
dt
2 <
def>= 0
dt
21 Wea
constant = -
Hence
def>
1
rzdt = Zwc(rZ - az)
4. The magnetic field does no work on the electrons:
1
-mv 2 = eV
2
2
v2
= -2e
V = v 2 + v 2 = (dr)
-dt +
m
r
4>
d<f>)z
r-
dt
r = b
V =Vo
b2
(d<f>)
= 2e Vo
dt
m
b
and
b 2def>
dt
dr
dt
= ~w
2
(b 2 - a 2)
c
This means that if Vo < Vae for a given Bo, the electrons will not reach the anode. Conversely, the cutoff magnetic field can be expressed in terms of V0 :
(8Vom/ e) 112
Bae= b(l - a2/b2)
(1-3-4)
This implies that if Bo > Bae for a given Vo, the electrons will not reach the anode.
SUGGESTED READINGS
J. W., and H. A. WATSON, Principles of Electron Tubes. D. Van Nostrand
Company, Princeton, N.J., 1965.
GEWAIUOWSKI,
Problems
15
KRAUS, J. D., and K. R. CARVER, Electromagnetics, 3rd ed. McGraw-Hill Book Company,
New York, 1984.
SKITEK, G. G., and S. V. MARSHALL, Electromagnetic Concepts and Applications. PrenticeHall, Inc., Englewood Cliffs, N.J., 1982.
PROBLEMS
1-1.
1-2.
At time t = to an electron is emitted from a planar diode with zero initial velocity and
the anode voltage is + 15 V. At time t = ti the electron is midway between the plates
and the anode voltage changes discontinuously to - 30 V.
a. Determine which electrode the electron will strike.
b. Compute the kinetic energy of the electron in electronvolts (eV) when it strikes the
electrode.
A circular cavity is constructed of a center conductor and an outer conductor. The inner center conductor has a radius of l cm and is grounded. The outer conductor has a
radius of 10 cm and is connected to a power supply of + 10 kV. The electrons are emitted from the cathode at the center conductor and move toward the anode at the outer
conductor.
a. Determine the magnetic flux density B in webers per square meter in the axial direction so that the electrons just graze the anode and return to the cathode again.
b. If the magnetic flux density B is fixed at 4 mWb/m 2 , find the required supply
voltage so that the electrons just graze the anode and return to the cathode again.
Chapter 2
Electromagnetic Plane
Waves
20 INTRODUCTION
Since Maxwell's fundamental concepts of electromagnetic wave theory have been
established, the electric and magnetic wave equations can readily be derived from
Faraday's electromotive force law, Ampere's circuital law, and Gauss's law for the
electric and magnetic fields. In Chapter 1 interactions between electron and field
were discussed. Here electromagnetic plane waves are described in detail. Many
topics associated with electromagnetic waves, such as Poynting theory, reflection
theory, attenuation concepts, and plane-wave propagation in metallic-film coating on
plastic substrates, are also analyzed, for these basic principles are used frequently in
later chapters.
The principles of electromagnetic plane waves are based on the relationships
between electricity and magnetism. A changing magnetic field will induce an electric field, and a changing electric field will induce a magnetic field. Also, the induced fields are not confined but ordinarily extend outward into space. The sinusoidal form of the wave causes energy to be interchanged between the magnetic and
electric fields in the direction of the wave propagation.
A plane wave has a plane front, a cylindrical wave has a cylindrical front, and
a spherical wave has a spherical front. The front of a wave is sometimes referred to
as an equiphase surface. In the far field of free space, electric and magnetic waves
are always perpendicular to each other, and both are normal to the direction of propagation of the wave. This type of wave is known as the transverse electromagnetic
(TEM) wave. If only the transverse electric wave exists, the wave is called TE-mode
wave. That means there is no component of the electric wave in the direction of
propagation. In TM modes only the transverse magnetic wave exists.
16
17
Sec. 2.1
The electric and magnetic wave equations can be basically derived from Maxwell's
equations, which in time domain are expressed as
VxE=
aB
(2-1-1)
at
aD
VxH=J+-
(2-1-2)
at
V D = Pv
(2-1-3)
VB=0
(2-1-4)
+ -u
+ -u
ay y
az z
(cartesian)
a
a
a
V = -ur + --uq, + -uz
ar
ra<f>
az
a
ar
a
r ao
l
a
r sin() a<f>
V=-u +-uB+---uq,
r
(cylindrical)
(spherical)
(2-1-4a)
(2-l-4b)
(2-1-4c)
It should be noted that boldface roman letters indicate vector quantities or complex
quantities. The units of these field variables are
E = electric field intensity in volts per meter
H = magnetic field intensity in amperes per meter
D = electric flux density in coulombs per square meter
B = magnetic flux density in webers per square meter or in tesla
(1 tesla = 1 weber/m 2 = 104 gausses = 3 x 10-6 ESU)
J = electric current density in amperes per square meter
Pv = electric charge density in coulombs per cubic meter
J =Jc+ Jo
where Jc
Jo
(2-1-5)
18
Chap. 2
= t:E
B = H
Jc = CTE
D
(2-1-6)
(2-1-7)
(2-1-8)
(2-1-9)
/L =/Lr/LO
where t:
(2-1-10)
=
=
=
V x E = - jwH
V
H = (CT + jwt:)E
(2-1-11)
(2-1-12)
V D = Pv
(2-1-13)
VB=0
(2-1-14)
V x V x E = -jwV x H
(2-1-15)
Substitution of Eq. (2-1-12) for the right-hand side of Eq. (2-1-15) gives
V XV
=-
jw(CT + jwt:)E
(2-1-16)
The vector identity for the curl of the curl of a vector quantity E is expressed as
V x V x E = -V 2 E + V(V E)
(2-1-17)
In free space the space-charge density is zero, and in a perfect conductor timevarying or static fields do not exist. So
= Pv = 0
VE = O
V D
(2-1-18)
(2-1-19)
Substitution of Eq. (2-1-17) for the left-hand side of Eq. (2-1-16) and replacement
of Eq. (2-1-19) yield the electric wave equation as
V2 E = y 2 E
(2-1-20)
Sec. 2.2
19
Poynting Theorem
where 'Y
= Yjw (CT +
V2 H
= y 2H
(2-1-21)
It should be noted that the "double del" or "de! squared" is a scalar operator-that
is,
VV = V 2
(2-1-22)
which is a second-order operator in three different coordinate systems. In rectangular (cartesian) coordinates,
a1
a1
ax2
ay
az
v2=-+2 + -2
(2-1-23)
1 a1
a1
!r !._(r!._)
+ - -2+ -2
r2 a<1>
az
ar ar
(2-1-24)
In spherical coordinates,
a( . a)
(2-1-24a)
Also, the solutions of Eqs. (2-1-1) and (2-1-2) solved simultaneously yield the electric and magnetic wave equations in the time domain as
aE
v2E = ,CTat +
V 2 ff
a.a POYNTING
a1 E
E at2
au + E
a2 Hat
at 2
CT-
(2-1-25)
(2-1-26)
THEOREM
At what rate will electromagnetic energy be transmitted through free space or any
medium, be stored in the electric and magnetic fields, and be dissipated as heat?
From the standpoint of complex power in terms of the complex field vectors, the
time average of any two complex vectors is equal to the real part of the product of
one complex vector multiplied by the complex conjugate of the other vector. Hence
the time average of the instantaneous Poynting vector in steady state is given by
(P)
(E x H)
= ~Re(E
x H*)
(2-2-1)
where the notation ( ) stands for the average and the factor of 1/2 appears in the
20
Chap. 2
equation for complex power when peak values are used for the complex quantities E
and H. Re represents the real part of the complex power, and the asterisk indicates
the complex conjugate.
It is necessary to define a complex Poynting vector as
p
= HE x H*)
(2-2-2)
Maxwell's equations in frequency domain for the electric and magnetic fields are
V
x E = - jw,H
(2-2-3)
J + jwEE
(2-2-4)
(V x E) H* = - jw, H H*
(V x H*) E
= (J* -
(2-2-5)
(2-2-6)
jwE E*) E
,IHl 2)
(2-2-7)
-E J*
jw(EIEl 2
,IHl 2)
(2-2-8)
H = !CTE
E*
+ jw(!,H H* -
!EE E*)
+VP
(2-2-9)
Integration of Eq. (2-2-9) over a volume and application of Gauss's theorem to the
last term on the right-hand side give
!(E Jt)dv
where
!CTIEl dv + j2w
(wm - w,)dv +
~ P ds
(2-2-10)
i
i
Equation (2-2-10) is well known as the complex Poynting theorem or the Poynting
theorem in frequency domain.
Furthermore, let
J!
J
Pin = -
"
(Pd) =
! CTI E 12 dv
Sec. 2.3
(Wm - We)
J (wm -
21
"
P1,=fPds
w,)dv
aE
ax
aE
ay
O or
aH
ax
aH
ay
22
Chap. 2
(2-3-2)
in which the electric intensity is arbitrarily chosen in the x direction and the magnetic intensity in the y direction. With no loss in generality, it can be assumed that
the electric intensity is given by
Ex = Eoef(wt-{3Z) = Eoeiw{t-({3/w)Z)
(2-3-3)
The magnetic intensity can be obtained by inserting Eq. (2-3-3) into the curl equation
V
= - jw,oH
(2-3-4)
)WJ.Lo
(2-3-5)
Differentiation of Eq. (2-3-3) with respect to z and substitution of the result in Eq.
(2-3-5) yield
Hy=
where /}_
W
~Ex
Vt:o
(2-3-6)
vp =
V J.LoEo
a= 0
f3 = w ~ is
phase constant
-Ex =
Hy
T/O
Ji:o
Eo
377 fi
(2-3-7)
a=O
(2-3-8a)
wv;::;
T/= Ji:-=~
v;:
/3 =
1
c
v =--=-p
v;::;
(2-3-8b)
(2-3-8c)
(2-3-8d)
Sec. 2.3
23
1. The tangential components of electric field intensity are continuous across the
boundary.
2. The normal components of electric flux density are discontinuous at the
boundary by an amount equal to the surface-charge density on the boundary.
3. The tangential components of magnetic field intensity are discontinuous at the
boundary by an amount equal to the surface-current density on the boundary.
4. The normal components of magnetic flux density are continuous across the
boundary.
The four statements can be proved by applying Faraday's law, Gauss's law,
Ampere's law, and V B = 0 to the boundaries of Fig. 2-3-2(a) and (b).
It can be seen from the diagrams that
f
f
f
f
=o
E di
= E,1 Ai -
E12 Ai
D ds
= Dn1 As -
Dn2 As
H di
= H,1 Ai - H12 Ai = 1, Ai
B ds
= Bn1 As -
= Ps As
Bn2 As
=0
24
(a)
Chap. 2
(b)
= E12
Dnt = Dn2 + Ps
Hn = H12 + ls
E,1
(2-3-9a)
(2-3-9b)
(2-3-9c)
(2-3-9d)
If medium 2 is a perfect conductor (a- = oo, E, = 1, ,, = 1) and medium 1 is a perfect dielectric (vacuum or free space, a- = 0, Eo, ,0), then
D,1
En= -
Eo
=0
(2-3- lOa)
= EoEnt = Ps
if ls = 0
Hn = 0
Bn1 = 0
Dnt
(2-3-lOb)
(2-3- lOc)
(2-3-lOd)
H~ll =
where /31
T/1
Eo(e-Jf31z
feif31z)
(2-3-11)
Eo (e-Jf31z - feif31z)
T/t
(2-3-12)
= w~
=
r =
T/O
. . .
.
d
f
d'
}
yr::.
~ = ~ r- = mtrms1c wave 1mpe ance o me mm
Et
V Ert
reflection coefficient
Sec. 2.3
25
Medium I
Medium 2
Incident wave
Transmitted wave
Reflected wave
Figure 2-3-3
reflection.
Uniform plane-wave
Ei21 =
Eo Te-jf3zz
H~21 = Eo
(2-3-13)
Te-jf32z
(2-3-14)
7/2
where
/32 = w ~
7/2
=
=
"V~
Er2
transmission coefficient
Zz
ill
= Hol
y
I
z=O
+ r
7/1 1 - f
fl
= H(2J
y
I
z=O
T/2
(2-3-15)
r =
T/2 - T/1
T/2 + T/1
(2-3-16)
From the boundary condition the tangential components of electric field intensity
are continuous across the interface. Then
21
i1l 'z=O = Eo(I + f) = i 'z=O = Eo T
(2-3-17)
+f =
<T
2
T/2
= 0),
IE~!x I
SWR
7/
= p = I~/n I =
(2-3-18)
T/1
r
r
(2-3-19)
26
Chap. 2
Ptr = -I (E x H*)
2
u, = E5
- ( I - f2)
2T/1
z=O
(2-3-20)
Then
Ptr
= Pinc( 1 - f2)
(2-3-21)
E Is in the Plane of Incidence. The plane of incidence is defined by the direction of propagation and the line normal to the boundary. The linearly polarized
uniform plane waves with E lying in and H normal to the plane of incidence are impinging obliquely on a boundary between two lossless dielectric materials as shown
in Fig. 2-3-4.
Whenever a wave is incident obliquely on the boundary surface between two
media, the polarization of the wave is vertical or horizontal if the electric field is
normal to or parallel to the boundary surface. The terms horizontal and vertical polarization refer to the phenomenon of waves from horizontal and vertical antennas,
respectively, producing the corresponding orientations of wave polarization when
Normal
Incident wave
Rctlectcd wave
13x =13,sin8,
ri--
13z = 13,
'I
Figure 2-3-4
COS () r I () r
13,
Transmitted wave
Sec. 2.3
27
the waves strike the surface of the earth. For guided waves in waveguides, the terms
transverse electric (TE) and transverse magnetic (TM) are used to designate the fact
that either the electric or the magnetic field is normal to the direction of propagation. The polarization of a wave is an extremely useful concept for computing electromagnetic power flow. For example, a Poynting vector indicates that the powerflow density is the cross product of an electric and a magnetic field with the specific
direction determined by the polarizations of the two fields.
As Fig. 2-3-4 shows, for a lossless dielectric medium, the phase constants (or
propagation constants) of the two media in the x direction on the interface are equal
as required by the continuity of tangential E and Hon the boundary. Thus
= f3r sin 8r
sin 8; = (3, sin 8,
{3; sin 8;
(2-3-23a)
{3;
(2-3-23b)
= f3r = {31,
{3;
(3,
= v2 =
v,
-V ~
(2-3-25)
where v represents the phase velocity. This is well known as Snell's law. In general,
all low-loss dielectrics have equal permeability-that is, , = 2 = J.Lo. If medium
2 is free space and medium I is a nonmagnetic dielectric, the right-hand side of Eq.
(2-3-25) becomes
which is called the index of refraction of the dielectric.
The components of electric intensity E are
\/'Z,
Ex
Ez
=-
Eo COS 8;e-i/31(xsinlJ;+zcos6;)
(2-3-26)
(2-3-27)
Eo
sin
8;e-J/31(xsine;+zcose;J
(2-3-28)
Hx
Hy
= Eo e-J/31(xsine;+zcos1J;l
(2-3-29)
0
T/1
(2-3-30)
(2-3-3 I)
Hz= 0
The wave impedance in the z direction is given by
Ex
Zz =-=ricos8
Hy
(2-3-32)
It should be noted that the subscripts of T/ and 8 have dropped because the wave
impedances of the two regions in the z direction are the same.
The wave impedance can be expressed in terms of the reflection coefficient of
28
Chap. 2
(2-3-33)
H~I)
= Eo[e-Jl3 1zcos8;
[eil3 1zcos8,]
(2-3-34)
711
Zz
~1 >
= H~l)
z=O
= 711
1+
cos 0; 1 -
r
r
(2-3-35)
The impedance must be equal to the z -directed wave impedance in region 2 at the
boundary. Substitution of Zz = 712 cos O, in Eq. (2-3-35) yields
r =
(2-3-36)
2712 cos 0,
= ----'----112 cos O, + 711 cos 0,
(2-3-37)
The preceding two equations are known as Fresnel's formulas for E in the plane of
incidence.
ponents of H are
Hx
(2-3-38)
(2-3-39)
Hz
(2-3-40)
(2-3-41)
Ex= 0
(2-3-42)
Ez
=0
(2-3-43)
Ey
11
Zz = - - = - - = 71 sec 0
Hx
cos 0
(2-3-44)
It should be noted that the subscripts of 71 and 0 have been dropped for the same reason stated previously.
Fresnel's formulas for H in the plane of incidence are
r =
T
(2-3-45)
2712 sec O,
+ 111 sec 0;
(2-3-46)
112 sec 0,
Sec. 2.4
29
Earth
The ionosphere is that region of the earth's atmosphere in which the constituent gases are ionized by solar radiation. This region extends from about 50 km
above the earth to several earth radii and has different layers designated as C, D, E,
and Flayers in order of height. The electron-density distribution of each layer varies
with the time of day, season, year, and the geographical location. During the day
the electron density N is approximately 10 12 electrons per cubic meter at an altitude
between 90 and 1000 km. The E and Flayers have a permanent existence, but the D
layer is present only during the day. The electron density determines the reflection
and refraction of microwaves. For vertical incidence, the critical frequency is given
by
Fer
9\/Jii::.
Hz
(2-4-1)
This means that a microwave of frequency Fer will be reflected back to the earth if
the electron density is equal to or higher than the required maximum electron density Nmax (electrons per cubic meter).
The sky wave reaches the receiving station after reflection from the ionosphere. Although important in many communication systems, the sky wave need not
be considered in most microwave applications because a wavelength shorter than
about 4 m will not return to the earth from the ionosphere. The reflected wave is
reflected from the earth in such a way as to reach the receiver. Energy radiated from
the nondirectional antenna of Fig. 2-4-1 strikes the earth at all points between the
base of the antenna and the horizon, but only that wave which leaves the antenna in
the direction shown reaches the receiver. The surface wave is a wave diffracted
30
Chap. 2
around the surface of the earth or guided by the ground-air interface. This component is important at broadcast frequencies; at microwave frequencies, however, the
surface wave is rapidly attenuated, and at a distance of 2 km from the antenna it has
an amplitude of only a fraction of 1% of the direct wave. This component must be
considered in blind-landing systems in which ranges of less than 2 km are important.
The direct wave travels a nearly straight path from the transmitting antenna to the receiving station. It is the only wave considered in this book. The termfree space will
be used to denote vacuum or any other medium having essentially the same characteristics as vacuum, such as open air, anechoic chamber, and shielded enclosure.
When power radiates from the transmitting antenna, the power density carried by
the spherical wave will decrease with distance as the energy in the wave spreads out
over an ever-increasing surface area as the wave progresses.
The power density is given by
W/m 2
(2-4-2)
A2
47T
gr
= -A2 gr
47T
""
(1';~2)(~gr)
watts
(2-4-3)
Receiving
antenna
Receiver
Gr
Pr
G,
47TR)
+ G, + Gr - 20 log ( A
dBW
(2-4-4)
where P, is in dBW, G, and Gr are in decibels (dB). The term 20 log (47TR/ A) is well
known as the free-space attenuation in decibels. It can easily be found from the stan-
31
Sec. 2.4
dard nomograph shown in Fig. 2-4-3. For example, if the wavelength of a signal is
0.03 m and the range is 20 m, the free-space attenuation is about 79 dB.
It should be noted that the free-space attenuation is entirely different from the
dissipative attenuation of a medium such as atmosphere that absorbs energy from the
wave. The factor (47TR 2) in Eq. (2-4-3) simply accounts for the fact that the power
density is inversely proportionally decreasing with the squared distance when the energy spreads out over free space. The factor (A 2 / 47T) is the isotropic aperture of a receiving antenna. It does not imply that a higher-frequency wave decreases in magnitude more rapidly than a lower-frequency wave. It is simply a consequence of the
fact that, for a given antenna gain, the aperture of a higher-frequency antenna is
smaller than that of a lower-frequency antenna so that it intercepts a smaller amount
of power from the wave.
f
10,000
300
200
sooo
4000
100
3000
2000
so
40
1000
30
20
500
400
10
300
200
s
4
3
70
100
2
,....._
'E
.:;,
::r:
.s
;.,
u
""
c
"'O"
"'....
"'
>
"'
~
"<;
::>
o.s
"'"' 0.4
ii;'
60
so
40
.g so
"'::>c
30
<
20
10
Figure 2-4-3
"""'c
"'
40
0.3
,....._
.:=
32
Chap. 2
("i:_
'V~
r;:
'V~
377
ohms
f3 =w~
(2-4-5)
f3 is given by
(2-4-6)
(2-4-7)
Vp=~
The lossy media are characterized by <T 0. There are three types of lossy media:
good conductor, poor conductor, and lossy dielectric, which are discussed in this
section. The presence of a loss in the medium introduces wave dispersion by conductivity. Dispersion makes a general solution in the time domain impossible except
by Fourier expansion methods. Thus only solutions for the frequency domain (or
steady state) will be given.
The electric and magnetic wave equations in the frequency domain as shown in
Eqs. (2-1-20) and (2-1-21) are repeated here:
= jw(<r + jwE)E
(2-5-1)
V H = jw(<r + jwE)H
(2-5-2)
V2 E
2
aE
--f
= jw(<T
az
1
1
aaz2
Hv
. (
= JWJJ<T
+ jwE)Ex
+
)WE
)H
y
(2-5-3a)
(2-5-3b)
(2-5-4)
Hy
Eo
= -e-az
COS (wt - {3z)
(2-5-5)
1J
<T
)WE
Sec. 2.5
33
y = YJw(<r +}we)=
jw~ ~1 <r
for-~
j <r
WE
WE
r-.
.,,. ;--(
= J.,,.v;--,,.
w<r v - J = J v w<r -
(1
1 - 1. 1)
-
Vz
+ j)~
V2
(2-5-7)
Hence
a={3=~
(2-5-8)
The exponential factor e-az of the traveling wave becomes e- = 0.368 when
1
z=---
(2-5-9)
5=
___ 1
(2-5-10)
{3
Interestingly, at microwave frequencies the skin depth is extremely short and a piece
of glass with an evaporated silver coating 5.40-m thick is an excellent conductor at
these frequencies. Table 2-5-1 lists the conductivities of materials.
The intrinsic impedance of a good conductor is given as
TJ
<r
]WE
~ /45 = (I
-y~
= (1 +
in which Rs
= \! w/(2<r)
j)
+ j)
~S = (1 +
for <r
r;:;
Vt;;.
WE
(2-5-11)
j)Rs
34
TABLE 251
Chap. 2
TABLE OF CONDUCTIVITIES
Substance
Type
Conductivity u
(mhos/m)
Quartz, fused
Ceresin, wax
Sulfur
Mica
Paraffin
Rubber, hard
Glass
Bakelite
Distilled water
Seawater
Tellurium
Carbon
Graphite
Cast iron
Mercury
Nichrome
Constantan
Silicon steel
German silver
Lead
Tin
Phosphor bronze
Brass
Zinc
Tungsten
Duralumin
Aluminum, hard-drawn
Gold
Copper
Silver
insulator
insulator
insulator
insulator
insulator
insulator
insulator
insulator
insulator
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
conductor
10-17
approx.
10-11
approx.
10-15
approx.
10-15
approx.
10-15
approx.
10-15 approx.
10-12
approx.
10-9
approx.
10-
approx.
4
approx.
5 x 102 approx.
3 x 104 approx.
105
approx.
J06
approx.
106
106
2 x 106
2 x 106
3 x 106
5 x 106
9 x 106
107
I. I x 101
1.7 x 107
1.8 x 101
3 x 101
3.5 x 101
4.1 x 101
5.8 x 101
6.1 x 101
ductor surface resistance. The average power density for a good conductor is given
by
p
= ~IHl
Rs
(2-5-12)
= wS
(2-5-13a)
The reflectivity and transmittance of a good conductor in vertical and horizontal polarizations are usually measured in terms of the grazing angle. The grazing angle I/I
is defined as the angle between the incident ray and the media boundary.
Vertical polarization. From Fig. 2-3-4 it can be seen that I/I = 90 - lJ;;
then sin I/I = cos lJ;, sin lJ; = cos /J, sin 2 (J, + cos 2 (J, = 1, and v1 sin (J, = v2 sin 01
The vertical reflectivity of a good conductor for the tangential components of electric intensity as shown in Eq. (2-3-36) is simplified to
Sec. 2.5
35
(2-5-13b)
For vertical polarization, the normal components of the electric fields are generally
used to determine the reflection coefficient. From Fig. 2-3-4 it can be seen that the
vertical components of the incident and reflected electric fields are in opposite directions. Therefore the reflectivity of a good conductor in vertical polarization is
f v = 1]i Sin l/J - 7]2[ 1 - (vz/ Vi COS l/J )2]i/ 2
(2-5-13c)
1]i sin I/I + 772[1 - (v2/vi cos l/J) 2]i/ 2
Similarly, the vertical transmittance of a good conductor for electric fields as shown
in Eq. (2-3-37) is given by
(2-5-13d)
Horizontal polarization. The reflectivity of a good conductor for electric
fields in horizontal polarization as shown in Eq. (2-3-45) is simplified in terms of l/J
as
f h = 7]2 sin l/J[l - (v2/Vi COS 1/1) 2]-i/ 2 - 7]1
(2-5-13e)
1]2 sin I/I [ 1 - (v2/ Vi cos l/J )2]-i 12 + 77 1
Similarly, the transmittance of a good conductor for electric fields in horizontal polarization as shown in Eq. (2-3-46) can be expressed as
Th
(2-5-13f)
In Fig. 2-3-4 it is assumed that medium 1 is free space or air and that medium 2 is
copper; then
111
vi
=3
377
x 108 m/s
7]2
= (1 + j)
V2
= w5 =
r;;,
-Vt;
{;:;
\j~
The conductivity er of copper is 5.8 x 107 mhos/m and its relative permeability is
unity. The magnitudes of reflectivity of copper for vertical and horizontal polarizations are computed by Eqs. (2-5-13c) and (2-5-13e) against the grazing angle l/J of 0
to 90 at a frequency range of 0.1 to 40 GHz. This result indicates that copper is a
perfect reflector for electromagnetic waves.
36
Chap. 2
conduction current is greater than the displacement current, whereas at some high
frequencies the reverse is true.
In general, the propagation constant and intrinsic impedance for a poor conductor are given by
(2-5-14)
(2-5-15)
~w,
.
CT
The term
CT /(wE)
_
-
f;( .CT)-1/2
]WE
1 - 1-
WE
(2-5-16)
tan()= -
(2-5-17)
WE
() = tan- 1 (~)
I
I
I
Figure 2-5-1
0"-~~~~~~~.-iJc=aE
dielectric.
If the loss tangent is very small-that is, cr/(wE) ~ 1-the propagation constant and intrinsic impedance can be calculated approximately by a binomial expansion. Since
= jw ~ ~ 1 - j
(2-5-18)
CT
WE
then
y
= jw ~ [ 1
j _!!___ +
2wE
.!.8
(!!_)
WE
+ ]
(2-5-19)
38
Chap. 2
where Er= -
Eo
18a
x=-=WEo
I/!
fGHz
T, =
I/!
~-;::::::====================-~~~~~~
I/!+
I/!
(2-5-27)
r _ sin I/! h -
sin
I/! +
(2-5-28)
Similarly, the transmittance of a lossy dielectric for electric fields in horizontal polarization as shown in Eq. (2-3-46) is expressed as
T,,
2 sin
I/!
~~~~--;=============
sin
I/! + Y(Er -
(2-5-29)
The reflections of electromagnetic waves by such lossy dielectric materials as seawater, dry sand, and concrete cement are often of concern to many electronics engi-
Sec. 2.5
39
neers. The conductivities a and relative dielectric constants E, of seawater, dry sand,
and concrete cement are tabulated in Table 2-5-2.
Figures 2-5-2 to 2-5-5 show, respectively, the magnitudes of reflectivity of
seawater, dry sand, and concrete cement for vertical and horizontal polarizations
against the grazing angle ijJ of 0 to 90 at a frequency range of O. l to 40 GHz [2].
u(mhos/m)
Seawater
Dry sand
Concrete cement
4
20
2 x 10-
10- 5
10
0.9
0R
07
06
L.-
0.5
04
;;
0.3
0.1
O OO.___....
I 0_ _2.1...0---'30--4..LO---:S:'::0---'60"----::'70:---8:'::0---:9~0-~
Grazing angle l/; in degrees
40
Chap. 2
..."'
p.,
"'
.s
0.1
0.2
0.3
0.5
0.9
0.8
------2
t.:'
=-----l
]"' 0.7
2
;::::::~~~~~~ro
~
:::;::
20
0.6
40
GHz
10
20
40
30
50
60
90
80
70
1.0
c
"'u...
"'
0.8
c....
0.6
p.,
.s
"'
"O
'2
co 0.4
:::;::
"'
0.2
10
20
30
40
60
50
70
80
in
vertical and
horizontal
90
Sec. 2.6
41
1.0
c.,
.,c.u
=
i:..
.,
0.8
0.6
-0
..
-~
0.4
0.2
*Copyright 1975 by the Institute of Electrical and Electronic Engineers, Inc. Light Transmittance and Microwave Attenuation of a Gold-Film Coating on a Plastic Substrate by S. Y. Liao [3];
reprinted from IEEE Trans. on Microwave Theory and Techniques. MTT-23, No. 10, 846-849, October
1975.
42
Chap. 2
Very thin metallic films have a much higher resistivity than a bulk metal because of
electron scattering from the film surface. If the film thickness is very large compared
to the electron mean-free-path, the resistivity is expected to be nearly the same as
that of a bulk metal. When the film thickness is on the order of the electron meanfree-path, then the role of electron scattering becomes dominant. Fuchs [4] and
Sondheimer [5] considered the general form of the solution of the Boltzmann equation for the case of a conducting film and found the film conductivity <TJ in terms of
the bulk conductivity <T, the film thickness t, and the electron mean-free-path p:
<TJ
~; [en(~)
+ 0.4228]
~p
fort
(2-6-l)
The surface resistance of conducting films is generally quoted in units of ohms per
square because in the equation for resistance
R =
pC
tw
(2-6-2)
when units of length e and width ware chosen to have equal magnitude (that is, resulting in a square), the resistance R in ohms per square is independent of the dimensions of the square and equals
Rs= PI= _1
t
t<TJ
fl/square
(2-6-3)
SUBSTRATE MATERIALS
Substrate material
Refractive index n
Corning Vycor
Crystal quartz
Fused silica
Plexiglass
Polycyclohexyl methacrylate
Polyester glass
Polymethyl methacrylate
Zinc crown glass
1.458
1.540
1.458
1.490
1.504
1.500
1.491
1.508
Sec. 2.6
43
ratio of the phase velocities of light in vacuum and in a medium. The extinction index is related to the exponential decay of the wave as it passes through a medium.
Most optical plastics are suitable as substrate materials for a dome window and for
metallic-film applications. Table 2-6-1 lists the values of the refractive index n of
several nonabsorbing plastic substrate materials in common use [6].
The measured values of the refractive index n and the extinction index k of thin
metallic-film coatings deposited in a vacuum [6] are tabulated in Table 2-6-2.
TABLE 262
REFRACTIVE INDEX
FILMS
Copper film
Wavelength
(A)
2000
2200
2300
2400
2500
2600
2700
2800
2900
3000
3100
3200
3300
3400
3500
3600
3700
3800
4000
4500
4920
5000
5460
5500
6000
6500
7000
7500
8000
8500
9000
9500
10,000
Gold film
Silver film
1.427
1.215
1.13
1.32
1.38
1.37
1.39
1.45
1.51
1.57
1.60
1.67
1.54
1.07
0.30
0.16
0.12
0.09
0.06
1.23
1.29
1.31
1.33
1.34
1.35
1.33
1.27
1.17
0.96
0.54
0.32
0.55
1.14
1.35
1.52
1.70
0.870
2.200
1.400
1.880
0.880
2.420
0.840
1.840
0.756
0.186
0.142
0.150
0.157
0.170
0.182
0.190
0.197
0.197
2.462
2.980
3.570
4.049
4.463
4.840
5.222
5.569
5.900
6.272
0.331
0.200
0.142
0.131
0.140
0.149
0.157
0.166
0.174
0.179
2.324
2.897
3.374
3.842
4.266
4.654
4.993
5.335
5.691
6.044
Source: Adapted from the American Institute of Physics Handbook by the American Institute of
Physics. Copyright 1972 by McGraw-Hill, Inc. (Used with permission of McGraw-Hill Book
Company.)
44
Chap. 2
Attenuation
dB
(2-6-4)
Air
_)G
RF energy
Substrate
R<,._:----.
Figure 2-6-1 Absorption and reflection
of film coating on plastic substrate.
Absorption loss A. As described in Section 2-5-1, the propagation constant y for a uniform plane wave in a good conducting material is given in Eq.
(2-5- 7) as
for
<Tf
~WE
(2-6-5)
20 log 10 ea'
8.686t~
20(at) log 10 e
dB
20(0.4343)(at)
(2-6-6)
Sec. 2.6
45
=
f =
Since the thickness of the coated film is very thin-for example, 100 angstroms at
most (A = 10- 10 m)-the absorption loss A is very small and can be ignored.
Reflection loss R. The reflection loss R due to the multiple boundaries of
the substrate glass coated with a metallic film can be analyzed by means of the
energy-transmission theory (see Eq. 2-3-18 in Section 2-3-3), and it is expressed as
R = -20 log
=
where 1Jr
718
11a
=
=
20 log I11a
dB
(2-6-7)
, 11r, =
lo
+ j)
~I = ~
(2-6-8)
where cr8
E8
Er
=
=
=
~ = 377 = 194 n
v;: v3.7s
for
<Tg ~ WEq
(2-6-9)
Substituting the values of the intrinsic impedances 1Jr, 718 , and 71a in Eq. (2-6- 7)
yields the reflection loss as [ 12]
R
= 20 log
[28.33~1
88
+ 10 log
(7)
dB
(2-6-10)
Correction term C. For very electrically thin film, the value of the absorption loss A is much less than 10 dB and the correction term is given by [8]
C = 20 log j l - p 10-Al 10 (cos () - j sin 0) I
1la)
+ 71a
(2-6-11)
= 1
for 71a
~ 1Jr
() = 3.54t~
Over the frequency range of 100 MHz to 40 GHz, the angle() is much smaller
than 1 so that cos () = 1 and sin () = (). Thus the correction term of Eq. (2-6-11)
46
Chap. 2
can be simplified to
dB
(2-6-12)
= 40 - 20 log (Rs)
dB
(2-6-13)
It is interesting to note that the microwave radiation attenuation due to the coated
metallic film on a glass substrate in the far field is independent of frequency and is
related only to the surface resistance of the coated metallic film [ 12].
N = n - jk
(2-6-14)
"o
.. TI
LIGHT
ABSORBING FILM
SUBSTRATE
Using the multireflection and transmission theory, the reflection loss is expressed by
(2-6-15)
nzk1)}
k~)]}
Sec. 2.6
47
47Tk1t1
Ao
a=--
Ao
=
bie"
l6non2(nf + kT)
b2e-" + b3 cos v + b4 sin v
(2-6-16)
A=l-R-T
(2-6-17)
L=A+R
(2-6-18)
When the concave surface of a plastic dome is uniformly coated with an electromagnetic interference shield of metallic film, however, the light is normally incident on the plastic substrate Nz, transmits through the thin metallic film N1 , and then
emerges into the air no. From the electromagnetic theory of luminous transmission
in transparent media, the light transmittance is the same regardless of whether light
is normally incident on the substrate medium Nz or on the absorbing film N1. Thus
the total attenuation loss is the same in both cases.
265 Plane Wave in Gold-Film Coating on Plastic
Glass
Surface resistance.
Conductivity:
Resistivity:
Electron mean-free-path:
<T
It is assumed that the thickness t of the gold film varies from 10 to 100 A. Its surface
48
TABLE 263
Thickness t
(AJ
100
90
80
70
60
50
40
30
20
10
Chap. 2
Resistivity Pr
(!1-m x 10-')
Surface resistance R,
(!11square)
1.17
I. 11
1.03
0.96
0.85
0.77
0.68
0.54
0.42
0.22
0.86
0.90
0.97
1.04
1.17
1.30
1.48
1.86
2.41
4.48
8.60
10.00
12.13
14.86
19.50
26.00
37.00
62.00
120.00
448.00
resistances are computed by using Eqs. (2-6-1) and (2-6-3) and are tabulated in
Table 2-6-3.
Figure 2-6-3 shows surface resistances of gold film in ohms per square against
the thicknesses of gold film from 10 to 100 A. According to the Fuchs-Sondheimer
theory, gold films have a typical surface resistance at about 10 to 30 O/square for a
thickness of about 90 to 45 A. The surface resistance is decreased as the thickness of
the gold film is increased.
~
60
::c
Cl
!:
50
...:Ii
~
:l ... 40
Cl
cc
(:I
ci:
~~30
----
... cc
uw
z ....
ci:
20
...c;;cc
...u
10
In
ci:
~
~
I
I
I
I
I
........~--------...~~.._~......~.......~__...__.--...
o~~
20
30
40
50
60
70
80
THICKNESS OF GOLD FILM IN A
90
100
Sec. 2.6
49
25
20
""
.:::
o----Liao's data
c:
.S!
15
""
:l
c:
';;;
.,
.,>
10
jj
0
5
10
20
30
40
Figure 2-6-4
film.
microwave radiation attenuation is about 19 dB. The data agree with Hawthorne's
conclusion [ 10, 12].
Light transmittance. For the visible-light region, the values of the refractive index n and the extinction index k of a gold-film coating on a plastic glass deposited in a vacuum are taken from Table 2-6-2. The refractive index no of air or
vacuum is unity. The refractive index nz of the nonabsorbing plastic glass is taken as
1.50. Light transmittance T and light reflection loss R of a gold-film coating on a
plastic glass are computed by using Eqs. (2-6-16) and (2-6-15), respectively. Then
from the values of T and R absorption loss A and total attenuation L are calculated.
The results are presented graphically in Fig. 2-6-5. It can be seen that for a light
transmittance of 80% the thickness of the gold-film coating is about 80 A. When the
absorption loss in the substrate material is considered, however, the light transmittance may be a little less than 80%.
Optimum condition. The surface resistance of a metallic film decreases as
the thickness of the film coating increases. However, the luminous transmittance is
decreased as the surface resistance of the metallic film is decreased. This relationship for the visible-light region is shown in Fig. 2-6-6.
Figure 2-6-7 illustrates the relationship of light transmittance versus wavelength for a given surface resistance of gold film. If a power dissipation of
5 W/square is allowed for deicing and defogging or keeping warm by the gold-film
coating on a plastic substrate and if the effective area of the coated film is 13 square
inches in a missile, the surface resistance of the coated film must be 12 !1/square.
50
Chap. 2
FILM
THICKNESS
IN A.
.....
z
90
w
u
ct
I-
!:: 80
::!;
"'ctz
a:
II%
~70.__~~..___.~.__~~--'~~~~-"~~~~-..
_,
10
30
20
40
v2
(28) 2
= -R = 12 x 13 = 5 W/square
Sec. 2.6
51
100~~~~-....~~--......--....~
1-
~VISIBLE LIGHT--1
zI.LI
a:
I.LI
a..
90
Rs
z
I.LI
y
37
80
26
ci::
II-
70
20
Cl)
2
ci::
15
a:
II-
60
12
:z:
(.!)
50
3000
8.6
4000
6000
WAVELENGTH ~ IN
10000
Example 2-6-5:
A gold film of 80
termine:
a. The
b. The
c. The
d. The
Solution
a. From Eq. (2-6-1) the gold-film conductivity is
CTJ =
~;[ln (7)
3
+ 0.4228]
10
De-
52
Chap. 2
l
l
R = - = -------s
tu1 80 x 10- 10 x l.03 x 107
= 12.12 O/squares
Silver
Conductivity:
Resistivity:
Electron mean-free-path:
<T =
=
=
<T
= 5.800 x
Copper
Conductivity:
Resistivity:
Electron mean-free-path:
p
p
107 mhos/m
1.724 x 10-s n-m
420 A
It is assumed that the thickness t of the silver and copper films varies from 10 to
100 A. The surface resistances of silver and copper films are computed by using Eqs.
(2-6-1) and (2-6-3) and are tabulated in Tables 2-6-4 and 2-6-5, respectively.
Figure 2-6-8 plots the surface resistances of silver and copper films, respectively, in ohms per square versus the thickness of the silver and copper films from 10
to 100 A.
*Copyright 1976 by The Institute of Electrical and Electronics Engineers, Inc. RF Shielding
Effectiveness and Light Transmittance of Copper or Silver Coating on Plastic Substrate by S. Y. Liao
[II]; reprinted from IEEE Trans. on Electromagnetic Compatibility, EMC-18, No. 4, 148-153,
November 1976.
Sec. 2.6
TABLE 264
53
Thickness t
(A)
Conductivity a1
(mho/m x 107 )
Resistivity p1
(!1-m x 107)
Surface resistance R,
(!1/square)
100
90
80
70
60
50
40
30
20
1.78
1.66
1.55
1.44
1.31
1.16
0.99
0.81
0.61
0.36
0.571
0.602
0.645
0.695
0.763
0.862
1.010
1.230
1.640
2.760
5.71
6.69
8.06
9.93
12.72
17.24
25.25
41.00
82.00
276.00
IO
TABLE 265
Thickness t
(A)
Conductivity a1
(mho/m x 107 )
Resistivity PJ
(!1-m x 10- 1 )
Surface resistance R,
(!1/square)
100
90
80
70
60
50
40
30
20
10
1.93
1.85
1.73
1.62
1.47
1.33
1.17
0.95
0.73
0.43
0.52
0.54
0.58
0.62
0.68
0.75
0.86
1.05
1.37
2.31
5.20
6.00
7.25
8.86
11.33
15.00
21.50
35.00
68.50
231.00
60
O>
....
"'
;:l
50
....
O>
0.
"'
E
..c::
40
.5
Q::;~
30
O>
(.)
c:
v;
"'
20
O>
(.)
....
10
;:l
Cl)
'--~----'~~-'-~~...L-.~~.L--'-----L~~-'-~~-'-~~"-~__.
10
20
30
40
50
60
70
80
90
100
Film thickness t in A
Figure 2-6-8
54
Chap. 2
RF radiation attenuation. Substitution of the values of the surface resistances of silver or copper films in Eq. (2-6-13) yields the microwave radiation attenuation in decibels by the silver-film or copper-film coating on a plastic substrate. Figure 2-6-9 shows the microwave radiation attenuation versus the surface resistance of
silver-film or copper-film coating, respectively.
25
0--- Copper film
0- - - Silver film
Optimum Condition. The light transmittance is increased as the surface resistance is increased. The relationship is illustrated in Fig. 2-6-12 for silver film and
copper film, respectively. The optimum condition occurs at 18 dB of microwave radiation attenuation and 94% of light transmittance with a surface resistance of about
12 fl/square.
100
90
10
-~jg40
50
60
70
80
80
100
90
100
90
70
I-
oZ
zW
<~
1-W
a.
Wz
60
80
70
~ ....
60
ii=
50
(..)
i-z
1-C
--= --=
=---:~,,
------------- .............., ','
,
soi---
en<
Z::::>
<Z
-''
------------- -.. . . '',,,
.....
40~--------------
30
a:w
I- II- I-
------------- ---
-- --- - ------- -- - - - - - - - - - - - - _
201--
..........,
... ,,,
' ....... ' ',,,
,,
.... ... ',' ' \ \
..., ' \ \, \
- .... ..., ' , ' \ \ \'
'
::c < JO
t:i I- ::c
.... t:i
....
)A
i80
I 00
90
\' '\\I
-- ......... ...',,\\'''
,,,,,,,
... , ' ' ,''~'''
\~..\.\\\
'
\ '\~~'''
---------------
--
101------------------.
-Light attenuation, L (per cent)
...
70
60
,
50
' ,'i'::-.;~~":.-- ... ;;::::
40
'
"'"~..-:;.a-::
'
''-.."""~!: !! !: : -::;.--- 30
' , ' ... '::.======~20
""'---------10
Film thickness, t (A)
2500
3000
3500
Wavelength, A (A)
8:
Figure 2-6-10 Light transmittance T and light attenuation loss L of silver film
versus wavelength A with film thickness t as parameter.
.... :,"'
..... ,,,.,
,,,....
-l''l...Hr
((:,~ "_.. . ,,,<. .
H:.-_------vrENul'>-~ 7_;::
10--..:-.:--.::-::- . . . .'---.
.
. . . . ::: . . . . . . ,,,,"" ,,,,,-"
100-- -
,,,,,,,,,,. ,,,.,,,,,,
20
--... .................
..... ____ _..,.,
50----...
..,...,,,. .,,,,,.. ,,,." /
40-----.:::.....
......
____
-'
......
30-----... . . . . . .:::-..._ ________ ...... _ ......
10
FIL~0;~1~;N~~~~+~~.;:~-=---=------4500 5000
6000
7000
WAVELENGTH A IN A
QL-~~~~~~~~-'-~~~~~~-'--~~~~~~-'-~~~~~
2000
40
_,.~'5S"i
VISIBLE LIGHT
4000
56
Chap. 2
Silver film
.::
"'c
<..)
e
"'c
Copper film
so
;:l
........~~~~~~~~
40
30
70'--~~~~-'---'-~~~~..__~~~~
20
10 12
Example 2-6-6:
A copper film of 60
Compute:
a.
b.
c.
d.
The
The
The
The
A is coated on a plastic
Solution
a. From Eq. (2-6-1) the copper-film conductivity is
CTJ
=~;[Jn(~)
+ 0.4228j
3 x 60 x 10- 10 x 5.8 x 7
10 [ In (420)
6Q
4 x 420 x 10- 10
+ 0.4228J
fCTJ
= - - - - - - - - - -7
60 X 10-io X 1.47 X 10
11 . 34 fl/square
Suggested Readings
57
= 40 - 20 log (11.34)
= 19 dB
REFERENCES
[1] ADLER, R. B., et al., Electromagnetic Energy Transmission and Radiation. P. 8. MIT
Press, Cambridge, Mass., 1969.
[2] LIAO, S. Y., Reftectivities of electromagnetic waves by seawater, dry sand, concrete cement, and dry ground. A report for the Naval Weapons Center, Department of the
Navy, China Lake, Calif., August 1976.
[3] LIAO, S. Y., Light transmittance and microwave attenuation of a gold-film coating on a
plastic substrate. IEEE Trans. on Microwave Theory and Techniques, MTT-23, No. 10,
October 1975.
[4] FUCHS, K., The conductivity of thin metallic films according to the electron theory of
metals. Proc. Camb. Phil. Soc., 30, 100 (1938).
[5] SONDHEIMER, E. H., The mean-free-path of electrons in metals. Advances in Physics, 1,
1 (1952).
[6] American Institute of Physics Handbook, Sec. 6-12, 6-119 to 6-121, and 6-138.
McGraw-Hill Book Company, New York, 1972.
[7] SCHULZ, R. B., et al., Shielding theory and practice. Proc. 9th Tri-Service Conj. on
Electromagnetic Compatibility, October 1963.
[8] VASAKA, C. S., Problems in shielding electrical and electronic equipment. U.S. Naval
Air Development Center. Johnsville, Pa., Rep. No. NACD-EL-N5507, June 1955.
[9] LIAO, S. Y., Design of a gold film on a glass substrate for maximum light transmittance
and RF shielding effectiveness. IEEE Electromagnetic Compatibility Symposium
Records, San Antonio, Texas, October 1975.
[10] HAWTHORNE, E. I., Electromagnetic shielding with transparent coated glass. Proc. IRE.,
42, 548-553, March 1954.
[ 11] LIAO, S. Y., Light transmittance and RF shielding effectiveness of a gold film on a glass
substrate. IEEE Trans. on Electromagnetic Compatibility, EMC-17, No. 4, November
1975.
[ 12] LIAO, S. Y., RF shielding effectiveness and light transmittance of copper or silver coating on plastic substrate. IEEE Trans. on Electromagnetic Compatibility, EMC-18, No.4,
November 1976.
SUGGESTED READINGS
COLLIN, R. E., Foundations for Microwave Engineering, Chapter 2. McGraw-Hill Book Company, New York, 1966.
HAYT, W. H., Engineering Electromagnetics, 4th ed., Chapters 10 and 11. McGraw-Hill
Book Company, New York, 1981.
58
Chap. 2
JORDAN, E. and K. G. BALMAIN, Electromagnetic Waves and Radiating Systems, 2nd ed.,
Chapters 4, 5, and 6. Prentice-Hall, Inc., Englewood Cliffs, N. J., 1968.
SEELY, S., and A. D. POULARIKAS, Electromagnetics, Classical and Modern Theory and Applications. Marcel Dekker, Inc., New York, 1979.
PROBLEMS
2-1. In a certain homogeneous medium the group velocity as measured by the propagation
time of a pulse is found to be proportional to the square root of the frequency
(vg = vAw where A is a constant) over a particular frequency range. It is assumed the
medium is a nonmagnetic insulator.
a. Determine the relationship between the phase and group velocities.
b. Derive an expression for the relative dielectric constant of this medium.
2-2. Show that a/az = {3 8 and a/at = jw for a sinusoidal wave propagating in the z direction.
2-3. The electric field of a plane wave propagating in free space is given in complex notation by
2-4.
2-5.
2-6.
2-7.
2-8.
Where Ux and Uy are unit vectors in the x and y directions of a right-hand coordinate
system.
a. In which direction is the wave propagating?
b. Find the frequency of the propagating signal.
c. Determine the type of wave polarization (linear, circular, or elliptical).
d. Express the magnetic field intensity H of the propagating wave.
e. Calculate the average power flow per square meter in the direction of the propagation.
Determine the permittivity of a slab of dielectric material that will reflect 20% of the
energy in a plane wave. The wave is striking normally to and propagating through the
slab. Neglect the reflections from the back face of the slab.
The reflection and refraction of microwave propagating in the ionosphere are determined by the electron density in the ionosphere. If the electron density is assumed to
be 10 14 electrons per cubic meter, detemine the critical frequency for vertical incidence
so that the signal wave will be reflected back to the earth.
The conductivity r:r of copper is 5.8 x 107 mhos/m and its relative permeability is
unity. Calculate the magnitudes of reflectivity of copper for vertical and horizontal polarizations against the grazing angle t/J of 0 to 90 at a frequency range of 1 to 40 GHz.
The increment of the angle t/J is 10 each step and the increment of the frequency is
10 GHz each step.
At the Brewster angle there is no reflected wave when the incident wave is vertically
polarized. If the incident wave is not entirely vertically polarized there will be some
reflection but the reflected wave will be entirely of horizontal polarization. Verify Eq.
(2-5-24B) for the Brewster angle in terms of the relative dielectrics.
Determine the pseudo-Brewster angle tfJ in terms of v,, vi, 71, , and T/2 for a good conductor. [Hint: Start from Eq. (2-5-13b).]
Problems
59
2-9. Calculate the pseudo-Brewster angles for seawater, dry sand, concrete cement, and dry
ground.
2-10. Determine the pseudo-Brewster angle if! in terms of E, and x for a lossy dielectric.
[Hint: Start from Eq. (2-5-26).]
2-11. Bulk gold has a conductivity of 4.1 x 107 mhos/m, a resistivity of 2.44 x 10- 8 !1-m,
and an electron mean-free-path of 570 A. Calculate the surface conductivity, surface
resistivity, and surface resistance of gold film for thicknesses of 10 to 100 A with an
increment of 10 A for each step.
2-12. Silver has a conductivity of 0.617 x 108 mhos/m, a resistivity of 1.620 x I0- 8 !1-m,
and an electron mean-free-path of 570 A. Repeat Problem 2-11 for silver film.
2-13. Seawater has a conductivity of 4 mhos/m and a relative dielectric constant of 20 at a
frequency of 4 GHz. Compute:
a. The intrinsic impedance
b. The propagation constant
c. The phase velocity
2-14. Repeat Problem 2-13 for dry sand (u = 2 x 10- 4 mho/m and E, = 4) and copper
(u = 5.8 x I0 7 mhos/m).
2-15. A uniform plane wave is incident normally from air onto the surface of seawater. The
electric intensity of the incident wave is 100 x I0- 3 Vim at a frequency of 5 GHz in
the vertical polarization. Calculate:
a. The electric intensity of the reflected wave
b. The electric intensity of the transmitted wave
2-16. Repeat Problem 2-15 for an angle of incidence of 30.
2-17. Dry ground has a conductivity of 5 x I0- 4 mhos/m and a relative dielectric constant
of IO at a frequency of 500 MHz. Compute:
a. The intrinsic impedance
b. The propagation constant
c. The phase velocity
2-18. Copper has a conductivity of 5.8 x 107 mhos/m and is considered an ideal material for
shielding. A shield is made of copper with a thickness of 1 mm. If a uniform plane
wave is normally incident upon the copper shield, compute the absorption loss in decibels by the copper atf = MHz.
2-19. A radar transmitter has an output power of 100 kW average. Calculate the power density in dBW/m 2 at a range of 3000 m and the free-space attenuation in decibels at
f =IO GHz.
2-20. Write a complete FORTRAN program to compute the magnitudes of reflectivity in vertical polarization against a grazing angle of seawater. The frequency varies from 0.1 to
40 GHz with an increment of 0.1 GHz between 0.1 to I GHz, I GHz between 1 to IO
GHz, and 5 GHz between IO to 40 GHz. Use FI0.5 format for numerical outputs and
Hollerith format for character outputs. Print the outputs in three columns such as frequency (GHz), grazing angle (degrees), and gamma (vertical reflectivity).
2-21. Write a complete FORTRAN program to compute the magnitudes of reflectivity in
horizontal polarization against a grazing angle for seawater. (Refer to Problem 2-20 for
specifications.)
2-22. Write a complete FORTRAN program to compute the light transmittance and light
reflection of a gold-film coating on a nonabsorbing plastic glass for thicknesses of l 0 to
JOO A with an increment of IO A each step. The wavelength varies from 2000 to
60
Chap. 2
10,000 A with an increment of 500 A each step. The values of the refractive index n
and the extinction index k of a gold-film coating on a nonabsorbing plastic glass deposited in a vacuum are listed in Table 2-6-2. The refractive index n of the nonabsorbing plastic glass is I. 5. Use F 10. 5 format for numerical outputs, Hollerith format
for character outputs, and Data statements to read in the input values.
2-23. Write a complete FOITTRAN program to compute the light transmittance and light
reflection for an aluminum-film coating on a nonabsorbing plastic substrate for thicknesses of IO to JOO A and print out the results in percentages. Use F 10.5 format for
numerical outputs, Hollerith format for character outputs, and Data statements to read
in the input values. Print the outputs in column form with proper head-letters and
units. The refractive index n of the nonabsorbing plastic glass is 1.50. The refractive
index n and extinction index k for aluminum film are tabulated in Table P2-23. (Refer
to Problem 2- 22 for specifications.)
2-24. Repeat Problem 2-22 for a silver-film coating on a nonabsorbing plastic glass for the
wavelengths from 2000 to 3700 A with an increment of 100 A each step. (Refer to
Table 2-6-2 for the values of n and k.)
TABLE P223
Wavelength
(A.J
0000
2000
2200
2400
2600
2800
3000
3200
3400
3600
3800
4000
Refractive
index n
Extinction
index k
Wavelength
0.00
0.00
(A.)
0000
0.11
0.13
0.16
0.19
0.22
0.25
0.28
0.31
0.34
0.37
0.40
2.20
2.40
2.54
2.85
3.13
3.33
3.56
3.80
4.01
4.25
4.45
4500
5000
5500
6000
6500
7000
7500
8000
8500
9000
9500
Refractive
index n
Extinction
index k
0.00
0.00
0.51
0.64
0.82
1.00
1.30
1.55
1.80
1.99
2.08
1.96
1.75
5.00
5.50
5.99
6.50
7.11
7.00
7.12
7.05
7.15
7.70
8.50
2-25. Repeat Problem 2-22 for a copper-film coating on a nonabsorbing plastic glass for the
wavelengths from 4500 to 10,000 A with an increment of 500 Aeach step.
2-26. Repeat Problems 2-20 and 2-21 for dry sand.
2-27. Repeat Problems 2-20 and 2-21 for concrete cement.
2-28. Repeat Problems 2-20 and 2-21 for dry ground.
2-29. Start from Eqs. (2-3-25) and (2-3-36) and verify Eq. (2-5-24a).
2-30. Derive Eq. (2-5-25) from Eq. (2-5-24a) by assuming the loss tangent to be much less
than unity.
Chapter 3
Microwave Transmission
Lines
30 INTRODUCTION
A transmission line can be analyzed either by the solution of Maxwell's field equations or by the methods of distributed-circuit theory. The solution of Maxwell's
equations involves three space variables in addition to the time variable. The
distributed-circuit method, however, involves only one space variable in addition to
61
62
Chap.3
the time variable. In this ~ection the latter method is used to analyze a transmission
line in terms of the voltage, current, impedance, and power along the line.
Based on uniformly distributed-circuit theory, the schematic circuit of a conventional two-conductor transmission line with constant parameters R, L, G, and C
is shown in Fig. 3-1-1. The parameters are expressed in their respective names per
unit length, and the wave propagation is assumed in the positive z direction.
R
~z
~z
V(,,<)~
Generator
~z
~z
i(z, t)
C~z
V(z
1ic
;GI
;
i(z
+ ~z. t)
Load
+ M, t)
--z
Figure 3-1-1
By Kirchhoff's voltage law, the summation of the voltage drops around the
central loop is given by
v(z, t) = i(z, t)R Az
ai(z, t)
at
+ L A z - - + v(z, t) +
av(z, t)
d
Az
(3-1-1)
Rearranging this equation, dividing it by Az, and then omitting the argument (z, t),
which is understood, we obtain
av
ai
--=Ri+Laz
at
(3-1-2)
Using Kirchhoff's current law, the summation of the currents at point B in Fig.
3-1-1 can be expressed as
i(z, t)
+ C Az
av(z
= v(z + Az,
t)G Az
= [ v(z, t) +
av(z, t)
az
Az G Az
+ Az, t)
at
+ i(z + Az, t)
(3-1-3)
J .( )
A
A
A
+ C u.z
-a [ v (z, t ) + av (z, t) u.z
+ l z, t + ai (z, t) u.z
at
az
az
By rearranging the preceding equation, dividing it by Az, omitting (z, t), and assuming Az equal to zero, we have
ai
-- =
az
Gv
av
+ Cat
(3-1-4)
Then by differentiating Eq. (3-1-2) with respect to z and Eq. (3-1-4) with respect to
t and combining the results, the final transmission-line equation in voltage form is
Sec. 3.1
63
found to be
a2 v
av
a2 v
= RGv + (RC + LG)- + LC-2
az2
at
at
(3-1-5)
Also, by differentiating Eq. (3-1-2) with respect tot and Eq. (3-1-4) with respect to
z and combining the results, the final transmission-line equation in current form is
a2 i
ai
a2 i
az 2 = RGi + (RC + LG) at+ LC at 2
(3-1-6)
All these transmission-line equations are applicable to the general transient solution.
The voltage and current on the line are the functions of both position z and time t.
The instantaneous line voltage and current can be expressed as
Re I (z)ejwt
(3-1-7)
(3-1-8)
where Re stands for "real part of." The factors V(z) aand I(z) are complex quantities
of the sinusoidal functions of position z on the line and are known as phasors. The
phasors give the magnitudes and phases of the sinusoidal function at each position of
z, and they can be expressed as
V(z)
I(z)
y
= V+e-yz + V-eyz
= l+e-yz + Leyz
(3-1-10)
=a+ jf3
(3-1-11)
(3-1-9)
(propagation constant)
-ZI
dz
di= -YV
dz
dzV -
dz 2
d2 I -
dz 2
zy
(3-1-12)
(3-1-13)
(3-1-14)
(3-1-15)
Z
Y
y
= R + jwL
= G + jwC
= \/Zr = a + jf3
(3-1-16)
(3-1-17)
(propagation constant)
(3-1-18)
64
Chap. 3
dV
dz
di
dz
d2 V
dz 2
d2I
dz 2
-jwll
(3-1-19)
-jwCV
(3-1-20)
-w 2 LCV
(3-1-21)
-w 2 LCl
(3-1-22)
It is interesting to note that Eqs. (3-1-14) and (3-1-15) for a transmission line are
similar to equations of the electric and magnetic waves, respectively. The only difference is that the transmission-line equations are one-dimensional.
312 Solutions of Transmission-Line Equations
The one possible solution for Eq. (3-1-14) is
(3-1-23)
The factors V+ and V- represents complex quantities. The term involving e-jf3z
shows a wave traveling in the positive z direction, and the term with the factor ejf3z is
a wave going in the negative z direction. The quantity {3z is called the electrical
length of the line and is measured in radians.
Similarly, the one possible solution for Eq. (3-1-15) is
I
Yo(V +e--yz
v_e-YZ)
= Yo(V+e-aze-Jf3z
V-eazejf3z)
(3-1-24)
fz = ~RG ++ jwL
.
jwC =Ro JXo
\{y
(3~1-25)
The magnitude of both voltage and current waves on the line is shown in Fig. 3-1-2.
0
(a) Voltage wave
Figure 3-12
Sec. 3.1
65
wL
and
wC
(3-1-26)
1+ .!_21wL
~) (I + .!_21wC
~) J
(3-1-27)
= jw\/LC[1 + !(~
+ ~)]
2 jWL ]WC
=
~(R~ +a~)
f3 = w\/LC
(3-1-28)
(3-1-29)
Zo =
V
R
G
+ jwL
+ jwC
{i (1 + _!!_)1/2(1 + __Q_)-1/2
-Ve
=
jwL
jwC
(3-1-30)
~ [ + ~ c:L -j~C)]
=~
=
Vp
= ~ = YLC
(3-1-31)
The product of LC is independent of the size and separation of the conductors and
depends only on the permeability , and permittivity of E of the insulating medium.
If a lossless transmission line used for microwave frequencies has an air dielectric
and contains no ferromagnetic materials, free-space parameters can be assumed.
66
Chap. 3
1
1
v = - - = - - - = c = 3 x 108 m/s
p
vLC
(3-1-32)
When the dielectric of a lossy microwave transmission line is not air, the phase velocity is smaller than the velocity of light in vacuum and is given by
v,
1
c
v
= v;:;,
(3-1-33)
y,
_
. f;
e1oc1ty actor -
v.
v;:;,
v =-=--C
(3-1-34)
A low-loss transmission line filled only with dielectric medium, such as a coaxial
line with solid dielectric between conductors, has a velocity factor on the order of
about 0.65.
Example 3-1-1:
G = 0.5 mmho/m
L = 8 nH/m
C = 0.23 pF
= 1 GHz
Solution
a. From Eq. (3-1-25) the line characteristic impedance is
Zo =
VG+};;;c =
\j 0.5
_ I 50.31~
- \j 15.29 x 10-4/70.910
+ 126.50
+ jwl)(G + jwC)
0.2774/79.31 = 0.051
+ j0.273
Sec. 3.2
67
= V+e-yz + V-e+yz
= l+e-yz + Le+yz
(3-2-1)
(3-2-2)
v
v= _:_e-yz
- -eyz
Zo
(3-2-3)
Zo
(3-2-5)
The ratio of the voltage to the current at the receiving end is the load impedance.
That is,
(3-2-6)
--
+-
z,
V+
Sending
end
Zo
v_
Receiving
end
Q
0
Figure 3-21
Z2
68
Chap. 3
(gamma), is defined as
.
ffi .
reflected voltage or current
Reft ection coe c1ent = -.-.---~~---
mc1dent voltage or current
(3-2-7)
If Eq. (3-2-6) is solved for the ratio of the reflected voltage at the receiving end,
Ze - Zo
V+e-rc,
the re-
(3-2-8)
re = V+e--re = Zc + Zo
If the load impedance and/or the characteristic impedance are complex quantities, as
is usually the case, the reflection coefficient is generally a complex quantity that can
be expressed as
re = Ire Je 10c
(3-2-9)
where Irc I is the magnitude and never greater than unity-that is, Irc I : : : ; l. Note
that (Jc is the phase angle between the incident and reflected voltages at the receiving
end. It is usually called the phase angle of the reflection coefficient.
The general solution of the reflection coefficient at any point on the line, then,
corresponds to the incident and reflected waves at that point, each attenuated in the
direction of its own progress along the line. The generalized reflection coefficient is
defined as
(3-2-10)
From Fig. 3-2-l let z = C - d. Then the reflection coefficient at some point located
a distance d from the receiving end is
rd =
V-ey(f-d)
V+e--r<t-dl
v_eyC
= ---e-2yd = r
V+e--rc
e-2yd
e
(3-2-11)
Next, the reflection coefficient at that point can be expressed m terms of the
reflection coefficient at the receiving end as
(3-2-12)
This is a very useful equation for determining the reflection coefficient at any point
along the line. For a lossy line, both the magnitude and phase of the reflection
coefficient are changing in an inward-spiral way as shown in Fig. 3-2-2. For a lossless line, a = 0, the magnitude of the reflection coefficient remains constant, and
only the phase of r is changing circularly toward the generator with an angle of
-2{3d as shown in Fig. 3-2-3.
It is evident that re will be zero and there will be no reflection from the receiving end when the terminating impedance is equal to the characteristic impedance
Sec. 3.2
69
I fl= I
I
I
I
oo I
\.
'\.
"
'-..
'--
---
--------
of the line. Thus a terminating impedance that differs from the characteristic
impedance will create a reflected wave traveling toward the source from the termination. The reflection, upon reaching the sending end, will itself be reflected if the
source impedance is different from the line characteristic impedance at the sending end.
earlier, there is a reflection coefficient r at any point along an improperly terminated line. According to the principle of conservation of energy, the incident power
minus the reflected power must be equal to the power transmitted to the load. This
can be expressed as
1- f ~
= Zo T1
Ze
(3-2-13)
Equation (3-2-13) will be verified later. The letter T represents the transmission
coefficient, which is defined as
70
V1, = ~
Vine
Chap. 3
(3-2-14)
line
Figure 3-2-4 shows the transmission of power along a transmission line where Pinc is
the incident power, P,er the reflected power, and P1r the transmitted power.
_P,
P,,
Load
Figure 3-2-4
V+e-yc + V-eyc
= V1,e-yc
(3-2-15)
V+ e-yc - V- eye
Zo
Zo
= V1r e-yc
(3-2-16)
Zc
Zc - Zo
V-eyc
re = V+e-yc
= Zc + Zo
(3-2-17)
(3-2-18)
The power carried by the two waves in the side of the incident and reflected waves is
(3-2-19)
The power carried to the load by the transmitted waves is
P1r
By setting Pinr
(V,,e-af)2
2Zc
(3-2-20)
T1
= Zc (1
Zo
rn
(3-2-21)
This relation verifies the previous statement that the transmitted power is equal to
the difference of the incident power and reflected power.
71
Sec. 3.3
Example 3-2-1:
A certain transmission line has a characteristic impedance of 75 + j0.01 0 and is terminated in a load impedance of 70 + j50 0. Compute (a) the reflection coefficient;
(b) the transmission coefficient. Verify: (c) the relationship shown in Eq. (3-2-21);
(d) the transmission coefficient equals the algebraic sum of 1 plus the reflection
coefficient as shown in Eq. (2-3-18).
Solution
2(70 + j50)
70 + j50 + (75 + jO.Ol)
2Ze
Ze
Zo
172.05~ _
/Jh, <;JO_
,
153.38/19.03 - 1.l 2illl.L - l.08 + 10 32
c.
T2
= (l.12/16.51) 2 = 1.25/33.02
Ze (l - r2) = 70 + / 50 [l - (0.33)/76.680)2]
Zo
75 + 10.0l
=
86
~ x l . 10/-2 .6 = l .25/33
75&_
The general solutions of the transmission-line equation consist of two waves traveling
in opposite directions with unequal amplitude as shown in Eqs. (3-1-23) and
(3-1-24). Equation (3-1-23) can be written
(3-3-1)
72
Chap. 3
With no loss in generality it can be assumed that V+e-az and V-e"' are real. Then
the voltage-wave equation can be expressed as
(3-3-2)
This is called the equation of the voltage standing wave, where
Vo= [(V+e-"z
2
-
(3-3-3)
which is called the standing-wave pattern of the voltage wave or the amplitude of
the standing wave, and
<P
V+e-az - V-e"z
)
arctan ( V _
V
tan (/3z)
+e az + -e"z
(3-3-4)
which is called the phase pattern of the standing wave. The maximum and minimum
values of Eq. (3-3-3) can be found as usual by differentiating the equation with respect to f3z and equating the result to zero. By doing so and substituting the proper
values of f3z in the equation, we find that
v +e-az + V-e"
= V+e-"1 (1
+ Ir\)
0, 1, 2, ....
V-e"z = V+e-" (1 2
(3-3-5)
\f\)
(3-3-6)
n1T
n1T
n1T
/3
21T I A
A
2
z = - = - - = n-
(n
= 0,
1, 2, ... )
Then
A
Zt
=-
(3-3-7)
+ Le" = l+e-"z(l +
2
Ir\)
Ir\)
(3-3-8)
(3-3-9)
The standing-wave patterns of two oppositely traveling waves with unequal amplitude
in lossy or lossless line are shown in Figs. 3-3-1 and 3-3-2.
A further study of Eq. (3-3-3) reveals that
1. When V +
Sec. 3.3
73
..
Voltage
Max ....._
"O
Ci.
Min
<
Figure 3-3-1
2. When V+ = 0 and V-
(3-3-11)
V-e'"
3. When the positive wave and the negative wave have equal amplitudes (that is,
IV + e-az I = IV- e az I) or the magnitude of the reflection coefficient is unity, the
standing-wave pattern with a zero phase is given by
Vs = 2V+e-az
COS
({3z)
(3-3-12)
(3-3-13)
Equations (3-3-12) and (3-3-13) show that the voltage and current standing waves
are 90 out of phase along the line. The points of zero current are called the current
nodes. The voltage nodes and current nodes are interlaced a quarter wavelength
apart.
The voltage and current may be expressed as real functions of time and space:
= (z,
is = (z,
Vs
t)
t)
(3-3-14)
(3-3-15)
The amplitudes of Eqs. (3-3-14) and (3-3-15) vary sinusoidally with time; the
74
Chap. 3
voltage is a maximum at the instant when the current is zero and vice versa. Figure
3-3-3 shows the pure-standing-wave patterns of the phasor of Eqs. (3-3-12) and
(3-3-13) for an open-terminal line.
.
.
Stan d mg-wave rat10
(3-3-16)
The standing-wave ratio results from the fact that the two traveling-wave components
of Eq. (3-3- l) add in phase at some points and subtract at other points. The distance
between two successive maxima or minima is A/2. The standing-wave ratio of a pure
traveling wave is unity and that of a pure standing wave is infinite. It should be noted
that since the standing-wave ratios of voltage and current are identical, no distinctions are made between VSWR and ISWR.
When the standing-wave ratio is unity, there is no reflected wave and the line is
called a fiat line. The standing-wave ratio cannot be defined on a lossy line because
the standing-wave pattern changes markedly from one position to another. On a lowloss line the ratio remains fairly constant, and it may be defined for some region. For
a lossless line, the ratio stays the same throughout the line.
Since the reflected wave is defined as the product of an incident wave and its
reflection coefficient, the standing-wave ratio p is related to the reflection coefficient
r by
p=
i +
1-
1r1
1r1
(3-3-17)
Sec. 3.3
75
0.7
t:cu
"CJ
0.6
!.::
._
0.5
cu
0
s:
.g
u
cu
cu
!;::
ci::
0.4
0.3
0.2
0.1
0.0
I
10
VSWRp
Figure 3-3-4
As a result of Eq. (3-3-17), since Ir I ::; 1, the standing-wave ratio is a positive real number and never less than unity, p ~ 1. From Eq. (3-3-18) the magnitude of the reflection coefficient is never greater than unity.
Example 3-3-1:
Standing-Wave Ratio
Solution
a. From Eq. (3-2-8) the reflection coefficient is
r
1+
1-
Ir I 1 + o.377
Ir I = 1 - o.311 = 2 21
76
Chap. 3
=V(z)
(3-4-1)
I(z)
-- Ix
I,
- I,
0---
IQ
rv
z
Zo
Figure 3-4-1
Receiving end
v,,
Zo
ZQ
z,
In general, the voltage or current along a line is the sum of the respective incident wave and reflected wave-that is,
(3-4-2)
(3-4-3)
At the sending end z
lsZo
V-
(3-4-4)
V+ - V-
(3-4-5)
lsZs = V+
V-
Substitution of V + and
V
v_ in Eqs.
=~[(Zs +
ls (
2
Zs+
= 1(Zs
2
Zo)
(3-4-6)
- Zo)
(3-4-7)
2
]
(3-4-8)
(3-4-9)
Sec. 3.4
Then the line impedance at any point z from the sending end in terms of Zs and Zo is
expressed as
Z
At z
Z (Z,
(3-4-10)
Z (Zs
0
( _ _l l)
34
= V +e-ye + V-eye
leZo = V +e-ye - V-eye
leZe
(3-4-12)
(3-4-13)
le (Ze + Zo )e Ye
=2
(3-4-14)
V-
= -le (Ze -
(3-4-15)
Z o)e - Y e
Then substituting these results in Eqs. (3-4-2) and (3-4-3) and letting z =
we obtain
V = i[(Ze
le
2
e-
d,
(3-4-16)
(3-4-17)
Next, the line impedance at any point from the receiving end in terms of Ze and Zo
is expressed as
z=z
0
(3-4-18)
The line impedance at the sending end can also be found from Eq. (3-4-18) by letting d = :
=
s
z
0
(3-4-19)
eyz
= cosh (yz)
sinh (yz)
(3-4-20)
Substitution of the hyperbolic functions in Eq. (3-4-10) yields the line impedance at
78
Chap. 3
any point from the sending end in terms of the hyperbolic functions:
(3-4-21)
Similarly, substitution of the hyperbolic functions in Eq. (3-4-18) yields the line
impedance from the receiving end in terms of the hyperbolic function:
= Zo Ze +
Zo tanh (yd)
Zo + Ze tanh (yd)
( _ _ )
3 4 22
For a lossless line, y = }{3; and by using the following relationships between hyperbolic and circular functions
=
(j{3z) =
sinh (j{3z)
j sin ({3z)
(3-4-23)
cosh
cos ({3z)
(3-4-24)
Ro - j Zs tan ({3z)
( _ _ )
3 4 25
and
z = R0 Ze cos ({3d)
( _ _ )
3 4 26
+r
z = z0 1 -
e-Zyd
f
ree-Zyd
(3-4-27)
r e -_ Ze
- Zo
Ze + Zo
(3-4-28)
From Eq. (3-2-12) the reflection coefficient at a distanced from the receiving end is
given by
(3-4-29)
Then the simple equation for the line impedance at a distance d from the load is expressed by
Z=Zl+r
0
1-
(3-4-30)
r = lrleN'
(3-4-31)
Sec. 3.4
79
1e-
where Ir I = I re 2"d
cf> = Oe - 2{3d
The impedance variation along a lossless line can be found as follows:
Z(d)
+ Ir le 14>
1 - lrle 1<1>
= Zo 1
= R (d) +
where I Z(d) I = Ro
R (d)
jX (d)
= IZ(d) lej0d
(3-4-32))
1 + 21 r I cos cf> + I r 12
1 - 21 r I cos <!> + Ir l2
(3-4-33)
1 - 1r1 2
Ro 1 - 21 r I cos cf> + Irj 2
x (d) = Ro 1 ()d
= Ro 1 +
arctan
21 r I sin <1>
21 r I cos cf> +
(~)
(3-4-34)
(3-4-35)
Ir 1
arctan (
~ r~ 1i: I~)
= 7T.
(3-4-36)
However, cos(Oe - 27T)
(3-4-37)
It is concluded that the impedance along a lossless line will be repeated for every interval at a half-wavelength distance.
Furthermore, the magnitude of a reflection coefficient Ir I is related to the
standing-wave ratio p by
lrl
p-1
=p+I
and
= _1+1r1
__,___,_
1-1r1
(3-4-38)
The line impedance at any location from the receiving end can be written
Z=
(p + 1) + (p - l)ej<I>
(p + 1) - (p - l)ej<I>
0 -'-'--~--'--~.:..:...._~--'---
(3-4-39)
This is a very useful equation for determining the line impedance in terms of standing-wave ratio p, since p can easily be measured by a detector or a standing-wave
meter.
Determination of characteristic impedance. A common procedure for
determining the characteristic impedance and propagation constant of a given transmission line is to take two measurements:
1. Measure the sending-end impedance with the receiving end short-circuited and
record the result:
(3-4-40)
80
Chap. 3
2. Measure the sending-end impedance with the receiving end open-circuited and
record the result:
(3-4-41)
Zo = ~
(3-4-42)
=a +
Normalized impedance.
is defined as
.
jf3
e1 arctanh -vZoc
(3-4-43)
z =1+ r
-=r
1- r
z=Zo
jx
(3-4-44)
It should be noted that the lowercase letters are commonly designated for normalized quantities in describing the distributed transmission-line circuits.
An examination of Eqs. (3-4-39), (3-4-40), and (3-4-44) shows that the normalized impedance for a lossless line has the following significant features:
Zmax
IVmax I
(3-4-45)
= Ro = Roi /min I
Here Zmax is a positive real value and it is equal to the standing-wave ratio p at
the location of any maximum voltage on the line.
2. The minimum normalized impedance is
Zmin
z=-=
mm
Ro
IVmin I
Ir I =-1
Roi I max I 1 + Ir I p
=
1-
(3-4-46)
Here Zmin is a positive real number also but equals the reciprocal of the
standing-wave ratio at the location of any minimum voltage on the line.
3. For every interval of a half-wavelength distance along the line, Zmax or Zmin is
repeated:
Zmax(z)
Zmin(z) =
Zmax(z
~)
(3-4-47)
Zmin(Z
~)
(3-4-48)
~)
Zmax
is equal to the
(3-4-49)
Zmin(z)
Sec. 3.4
81
When a transmission line is branched, it is better to solve the line equations for the
line voltage, current, and transmitted power in terms of admittance rather than
impedance. The characteristic admittance and the generalized admittance are
defined as
Yo
= _!__ = Go
jBo
(3-4-50)
= _!_ = G
jB
(3-4-51)
Zo
y=Yo
Example 3-4-1:
Zo
=- =- = g
jb
(3-4-52)
Line Impedance
A lossless line has a characteristic impedance of 50 11 and is terminated in a load resistance of 75 n. The line is energized by a generator which has an output impedance of
50 n and an open-circuit output voltage of 30 V (rms). The line is assumed to be 2.25
wavelengths long. Determine:
a. The input impedance
b. The magnitude of the instantaneous load voltage
c. The instantaneous power delivered to the load
Solution
a. From Eq. (3-4-26) the line that is 2.25 wavelengths long looks like a quarterwave line. Then
27T ,\
1T
{3d = - . - = ,\ 4
2
From Eq. (3-4-26) the input impedance is
Z 10 =
Ra =
(50) = 33.33 11
Re
75
= Re - Ro = 75 - 50 = O 20
Re + Ro 75 + 50
Pc= ( ) = 17.28 W
75
82
Chap. 3
- Zo
\ \ .
r c = Zc
Zc + Zo = r c e
181
.
+ Jr
(3-5-1)
Since Ifc I cs: 1, the value of fe must lie on or within the unity circle with a radius of
1. The reflection coefficient at any other location along a line as shown in Eq.
(3-2-12) is
(3-5-2)
which is also on or within the unity circle. Figure 3-5-1 shows circles for a constant
reflection coefficient r and constant electrical-length radials {3d.
From Eqs. (3-4-29) and (3-4-44) the normalized impedance along a line is
given by
+ f ce- 2-rd
Zo
1 - f ce- 2-rd
z =-=
1 - fc
Zc
Zo
= O;
Re+ }Xe
Zo
(3-5-3)
then
=r +
.
JX
(3-5-4)
and
fc
z- l
z+ 1
= -- =
f r + 1T
(3-5-5)
n
n
(3-5-6)
Sec. 3.5
83
Smith Chart
Ir i=
r,
and
x
2f;
(1 - r,)2
(3-5-7)
r )1 + r2 =
( r r _ _l+r
I
and
(f, - 1)2
(-1)2
l+r
(c - ~r = Gr
(3-5-8)
(3-5-9)
Equation (3-5-8) represents a family of circles in which each circle has a constant resistance r. The radius of any circle is 1/(1 + r), and the center of any circle
is r/(1 + r) along the real axis in the unity circle, where r varies from zero to
infinity. All constant resistance circles are plotted in Fig. 3-5-2 according to Eq.
(3-5-8).
Equation (3-5-9) also describes a family of circles, but each of these circles
specifies a constant reactance x. The radius of any circle is (1/x), and the center of
any circle is at
r,
= 1
f;
=-
All constant reactance circles are plotted in Fig. 3-5-3 according to Eq. (3-5-9).
84
r,
0.4
Chap. 3
Sec. 3.5
Smith Chart
85
There are relative distance scales in wavelength along the circumference of the
Smith chart. Also, there is a phase scale specifying the angle of the reflection
coefficient. When a normalized impedance z is located on the chart, the normalized
impedance of any other location along the line can be found by use of Eq. (3-5-3):
1+
z= 1 -
fee-zyd
(3-5-10)
fee-lyd
where
(3-5-11)
The Smith chart may also be used for normalized admittance. This is evident
since
1
Yo = Zo = Go + }Bo
and
Y = _!_ = G + jB
(3-5-12)
Y
Zo
1
.
= - = - = g + jb
Yo
Z
z
y = -
(3-5-13)
Figure 3-5-4 shows a Smith chart which superimposes Figs. 3-5-2 and 3-5-3
into one chart. The characteristics of the Smith chart are summarized as follows:
1. The constant r and constant x loci form two families of orthogonal circles in
the chart.
2. The constant r and constant x circles all pass through the point (L = 1,
f; = O).
3. The upper half of the diagram represents + jx.
4. The lower half of the diagram represents - jx.
5. For admittance the constant r circles become constant g circles, and the constant x circles become constant susceptance b circles.
6. The distance around the Smith chart once is one-half wavelength (A/2).
7. At a point of Zmin = 1/ p, there is a Vmin on the line.
8. At a point of Zmax = p' there is a Vmax on the line.
9. The horizontal radius to the right of the chart center corresponds to Vmax, /min,
Zmax, and p (SWR).
10. The horizontal radius to the left of the chart center corresponds to Vmin, I max,
Zmin, and l/p.
11. Since the normalized admittance y is a reciprocal of the normalized impedance
z, the corresponding quantities in the admittance chart are 180 out of phase
with those in the impedance chart.
12. The normalized impedance or admittance is repeated for every half wavelength
of distance.
13. The distances are given in wavelengths toward the generator and also toward
the load.
86
l~llANCt:
OR ADMITTANCE COORDINATES
fllAOIALLY
!I
,~
I If
....
ti
SCAl[O PAfllAlllf:T[lllS
_,_.,~OH
uo.-
"
..
0
2 Iii
t:
:=11
- -
Chap. 3
..
~~~o=~,;:
..
Figure 3-5-4
Smith chart.
p - l
p + 1
(3-5-14)
A Smith chart or slotted line can be used to measure a standing-wave pattern directly
and then the magnitudes of the reflection coefficient, reflected power, transmitted
power, and the load impedance can be calculated from it. The use of the Smith chart
is illustrated in the following examples.
Example 3-5-1:
Load
Sec. 3.5
87
Smith Chart
Figure 3-5-5
,\ = 5 cm, determine the first Vmax, first Vmin from the load, and the VSWR
pas shown
in Fig. 3-5-5.
Solution
1. Enter zc = I + j 1 on the chart as shown in Fig. 3-5-6.
2. Read 0.162,\ on the distance scale by drawing a dashed-straight line from the
center of the chart through the load point and intersecting the distance scale.
3. Move a distance from the point at 0 .162,\ toward the generator and first stop at
the voltage maximum on the right-hand real axis at 0.25,\. Then
di (Vmax)
4. Similarly, move a distance from the point of 0.162A toward the generator and
first stop at the voltage minimum on the left-hand real axis at 0.5A. Then
d2(Vmin)
88
Chap. 3
S. Make a standing-wave circle with the center at (1, O) and pass the circle through
the point of I + j I. The location intersected by the circle at the right portion of
the real axis indicates the SWR. This is p = 2.6.
Example 3-5-2:
\/\/"'\!
-1
r- Loaded
s
=son
R0
Figure 3-5-7
0. I SA.
Shmtod
Solution
1. When the line is shorted, the first voltage minimum occurs at the place of the
load as shown in Fig. 3-5-8.
0.125
SWR
circle
Sec. 3.6
89
Impedance Matching
2. When the line is loaded, the first voltage minimum shifts 0.15A from the load.
The distance between two successive minima is one-half wavelength.
3. Plot a SWR circle for p = 2.
4. Move a distance of 0.15A from the minimum point along the distance scale
toward the load and stop at 0.15A.
5. Draw a line from this point to the center of the chart.
6. The intersection between the line and the SWR circle is
Ze
1 - }0.65
(1 - j0.65)(50)
= 50 -
}32.5 0
36 IMPEDANCE MATCHING
Impedance matching is very desirable with radio frequency (RF) transmission lines.
Standing waves lead to increased losses and frequently cause the transmitter to malfunction. A line terminated in its characteristic impedance has a standing-wave ratio
of unity and transmits a given power without reflection. Also, transmission
efficiency is optimum where there is no reflected power. A "flat" line is nonresonant; that is, its input impedance always remains at the same value Zo when the frequency changes.
Matching a transmission line has a special meaning, one differing from that
used in circuit theory to indicate equal impedance seen looking both directions from
a given terminal pair for maximum power transfer. In circuit theory, maximum
power transfer requires the load impedance to be equal to the complex conjugate of
the generator. This condition is sometimes referred to as a conjugate match. In
transmission-line problems matching means simply terminating the line in its characteristic impedance.
A common application of RF transmission lines is the one in which there is a
feeder connection between a transmitter and an antenna. Usually the input impedance to the antenna itself is not equal to the characteristic impedance of the line.
Furthermore, the output impedance of the transmitter may not be equal to the Zo of
the line. Matching devices are necessary to flatten the line. A complete matched
transmission-line system is shown in Fig. 3-6-1.
z,
r----1
r------,
0~------.....,o
~---1 Matching
Matching 1
I device I
I device I
I
I
I
I
1
1111 :111
z 1z;
Zo1Zo
1
I
I
'-----+a
oI
L ____ J
Figure 3-6-1
zo
1:1 1lr
I
*I
Z0 1Z0
Z2 1Z2
I
I
Io
o+------'
L ____ _J
90
Chap. 3
For a low-loss or lossless transmission line at radio frequency, the characteristic impedance Zo of the line is resistive. At every point the impedances looking in
opposite directions are conjugate. If Zo is real, it is its own conjugate. Matching can
be tried first on the load side to flatten the line; then adjustment may be made on the
transmitter side to provide maximum power transfer. At audio frequencies an ironcored transformer is almost universally used as an impedance-matching device.
Occasionally an iron-cored transformer is also used at radio frequencies. In a practical transmission-line system, the transmitter is ordinarily matched to the coaxial
cable for maximum power transfer. Because of the variable loads, however, an
impedance-matching technique is often required at the load side.
Since the matching problems involve parallel connections on the transmission
line, it is necessary to work out the problems with admittances rather than impedances. The Smith chart itself can be used as a computer to convert the normalized
impedance to admittance by a rotation of 180, as described earlier.
R0
50
Figure 3-6-2
(3-6-1)
z - ---- --50---Q -
2 + j(2 + fj)
Sec. 3.6
Impedance Matching
91
if the stub has the same characteristic impedance as that of the line. Otherwise
(3-6-2)
The stub length is then adjusted so that its susceptance just cancels out the susceptance of the line at the junction.
Example 3-6-1:
Single-Stub Matching
Solution
1. Compute the normalized load admittance and enter it on the Smith chart (see
Fig. 3-6-3).
l
Ye= -
ze
Ro
= 2
Ze
= -
r::
+ 1(2 + v 3)
= 2
+ 13.732
2. Draw a SWR circle through the point of ye so that the circle intersects the unity
circle at the point Yd
Yd= l - j2.6
Note that there are an infinite number of Yd Take the one that allows the stub to
be attached as closely as possible to the load.
+ j5.2
0.031
0.031
-jS.2
92
Chap. 3
3. Since the characteristic impedance of the stub is different from that of the line,
the condition for impedance matching at the junction requires
Y11 =Yd+ Y,
where Y, is the susceptance that the stub will contribute.
It is clear that the stub and the portion of the line from the load to the junction are in parallel, as seen by the main line extending to the generator. The admittances must be converted to normalized values for matching on the Smith
chart. Then Eq. (3-6-2) becomes
= (Y11 -
yd) ( -Yo)
Yos
.
100
(1 - 12.6)]50
= [l -
.
= + 15.20
4. The distance between the load and the stub position can be calculated from the
distance scale as
d = (0.302 - 0.215)A = 0.087A
5. Since the stub contributes a susceptance of+ j5.20, enter + j5.20 on the chart
and determine the required distance C from the short-circuited end (z = 0,
y = oo), which corresponds to the right side of the real axis on the chart, by
transversing the chart toward the generator until the point of+ j5.20 is reached.
Then
C = (0.50 - 0.03l)A = 0.469A
When a line is matched at the junction, there will be no standing wave in the line
from the stub to the generator.
6. If an inductive stub is required,
y~ =
j2.6
y;
= - j5.2
C' = 0.031A
Sec. 3.6
93
Impedance Matching
Y22
= Yd'l Ys2 = 1
= yd'}. Ys2 = Yo
(3-6-3)
(3-6-4)
In these two equations it is assumed that the stubs and the main line have the same
characteristic admittance. If the positions and lengths of the stubs are chosen properly, there will be no standing wave on the line to the left of the second stub measured from the load.
R0
=son ---
ZQ = 100
Y22
+ jlOO
Example 3-6-2:
Double-Stub Matching
The terminating impedance Zeis 100 + j 100 !1, and the characteristic impedance Z0
of the line and stub is 50 !1. The first stub is placed at 0.40A away from the load. The
spacing between the two stubs is ~A. Determine the length of the short-circuited stubs
when the match is achieved. What terminations are forbidden for matching the line by
the double-stub device?
Solution
I. Compute the normalized load impedance ze and enter it on the chart as shown in
Fig. 3-6-5:
Ze
lOO+j!OO
50
= - -- - - = 2
j2
2. Plot a SWR p circle and read the normalized load admittance 180 out of phase
with ze on the SWR circle:
Ye = 0.25 - j0.25
94
Chap. 3
Forbidden
area
Spacing
circle
(g = I) through a phase angle of 2{3d = 2{3~A = ~1T toward the load. Now Y11
must be on this spacing circle, since yd2 will be on the g = l circle (y11 and yd2
are ~A apart).
4. Move ye for a distance of 0.40A from 0.458 to 0.358 along the SWR p circle
toward the generator and read Yd1 on the chart:
Yd1
= 0.55 -
j 1.08
5. There are two possible solutions for y 11 They can be found by carrying yd 1 along
the constant-conductance (g = 0.55) circle that intersects the spacing circle at
two points:
Y11
= 0.55 - JO.II
Yd1
Ysi
Sec. 3.6
Impedance Matching
95
Then
Ys1
= Y11
- YJ1
j 1.08)
= + j0.97
Similarly,
y;,
= - j.080
+ 0.123)A
= (0.25
e;
= 0.373A
Y11
Y<12 = 1 - j0.61
ydi
= 1
+ j2.60
+ j0.61
y;2 = - j2.60
10. The lengths of stub 2 are found as
+ 0.087)A
= 0.337A
2 = (0.308 - 0.25)A
= 0.058A
= (0.25
11. It can be seen from Fig. 3-6-5 that a normalized admittance ye located inside the
hatched area cannot be brought to lie on the locus of y,, or y;, for a possible
match by the parallel connection of any short-circuited stub because the spacing
circle and g = 2 circle are mutually tangent. Thus the area of a g = 2 circle is
called the forbidden region of the normalized load admittance for possible match.
Normally the solution of a double-stub-matching problem can be worked out
backward from the load toward the generator, since the load is known and the distance
of the first stub away from the load can be arbitrarily chosen. In quite a few practical
matching problems, however, some stubs have a different Zo from that of the line, the
length of a stub may be fixed, and so on. So it is hard to describe a definite procedure
for solving the double-matching problems.
The flexible coaxial lines are available in different types. Their diameters vary
from 0.635 cm (0.25 in.) to about 2.54 cm (1 in.), depending on the power requirement. In some coaxial cables, the inner conductor is stranded or a solid wire, but the
outer conductor is a single braid or double. The dielectric material used in these
coaxial lines is polyethylene, which has low loss at radio frequencies. Particularly
for the RG series, the dielectric is either solid or foam polyethylene. The loss per
unit length for foam polyethylene is even appreciably less than the equivalent solid
polyethylene.
96
Chap. 3
Sec. 3.7
Microwave Co
ax1at Connectors
97
BNC female
APC-3.S male
(a) APC-3.S
(b) APC-7
BNc female
BNCmale
(c)
BNC
SMA female
SMA male
(d) SMA
APC-3.S
male
SMc ma/e
(plug)
TypeNma/e
Type N female
(e) SMC
Figure 3-7- J
(f) Type
Microwa
ve coaxial
connectors.
98
Chap. 3
for the frequency range of 1 to 18 GHz. It is 50- or 75-ohm connector and its
VSWR is extremely low, less than 1.0?
REFERENCES
[l] SMITH, P.H.
Transmission line calculator. Electronics, 12, 29-31 (1939).
An improved transmission line calculator. Electronics, 17, 130-133 and 318-325
(1944).
Smith Charts-Their Development and Use. A series published at intervals by the Kay
Electric Co.; No. 1 is dated March 1962, and No. 9 is dated December 1966.
SUGGESTED READINGS
BROWN, R. G., et al., Lines, Waves, and Antennas, 2nd ed. John Wiley & Sons, New York,
1970.
DwoRSKY, L. N., Modern Transmission Line Theory and Applications. John Wiley & Sons,
New York, 1979.
LIAO, s. 'i.' Engineering Applications of Electromagnetic Theory. \\Jest ~ub\isbing eompany'
St. Paul, Minn., 1988.
RIZZI, PETER A., Microwave Engineering: Passive Circuits. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1988.
SINNEMA, W., Electronic Transmission Technology: Lines, Waves, and Applications. PrenticeHall, Inc. , Englewood Cliffs, N. J. , 1979.
PROBLEMS
3-1. A transmission line has a characteristic impedance of 300 0 and is terminated in a load
of 300 - j300 0. The propagation constant of the line is 0.054 + j3.53 per meter.
Determine:
a. The reflection coefficient at the load
b. The transition coefficient at the load
c. The reflection coefficient at a point 2 m away from the load
3-2. A lossless transmission line has a characteristic impedance of 50 0 and is terminated in
a load of 100 0. The magnitude of a voltage wave incident to the line is 20 V (rms).
Determine:
a. The VSWR on the line
b. The maximum voltage Vmax and minimum voltage Vmin on the line
c. The maximum current lmax and minimum current /min on the line
d. The power transmitted by the line
3-3. A lossless line has a characteristic impedance of 75 0 and is terminated in a load of
Problems
99
300 0. The line is energized by a generator which has an open-circuit output voltage of
20 V (rms) and output impedance of 75 0. The line is assumed to be 2i wavelengths
long.
a. Find the sending-end impedance.
b. Determine the magnitude of the receiving-end voltage.
c. Calculate the receiving-end power at the load.
3-4. A lossless transmission line has a characteristic impedance of l 00 0 and is terminated
in a load of 75 0. The line is 0.75 wavelength long. Determine:
a. The sending-end impedance
b. The reactance which, if connected across the sending end of the line, will make the
input impedance a pure resistance
3-5. A coaxial line with a solid polyethylene dielectric is to be used at a frequency of
3 GHz. Its characteristic impedance Zo is 50 0 and its attenuation constant a is
0.0156 Np/m. The velocity factor which is defined as the ratio of phase velocity over
the velocity of light in free space is 0.660. The line is 100 m long and is terminated in
its characteristic impedance. A generator, which has an open-circuit voltage of 50 V
(rms) and an internal impedance of 50 0, is connected to the sending end of the line.
The frequency is tuned at 3 GHz. Compute:
a. The magnitude of the sending-end voltage and of the receiving-end voltage
b. The sending-end power and the receiving-end power
c. The wavelengths of the line
3-6. An open-wire transmission line has R = 5 Olm, L = 5.2 x 10- 8 H/m, G =
6.2 x 10- 3 mho/m, and C = 2.13 X 10- 10 F/m. The signal frequency is 4 GHz. Calculate:
a. The characteristic impedance of the line in both rectangular form and polar form
b. The propagation constant of the wave along the line
c. The normalized impedance of a load 100 + j 100
d. The reflection coefficient at the load
e. The sending-end impedance if the line is assumed a quarter-wavelength long
3-7. A quarter-wave lossless line has a characteristic impedance of 50 0 and is terminated
in a load of 100 0. The line is energized by a generator of 20 V (rms) with an internal
resistance of 50 0. Calculate:
a. The sending-end impedance
b. The magnitude of the receiving-end voltage
c. The power delivered to the load.
3-8. A lossless transmission line is terminated in an open circuit. The sending end is energized by a generator which has an open-circuit output voltage of Vg (rms) and an interval impedance equal to the characteristic impedance of the line. Show that the sendingend voltage is equal to the output voltage of the generator.
3-9. A lossless transmission line has a characteristic impedance of 300 0 and is operated at
a frequency of 10 GHz. The observed standing-wave ratio on the line is 5.0. It is proposed to use a short-circuited stub to match a pure resistor load to the line.
a. Determine the distance in centimeters from the load to the place where the stub
should be located. (Two possible solutions.)
b. Find the length of the stub in centimeters. (Two possible solutions.)
3-10. A lossless line has a characteristic impedance of 50 0 and is loaded by 60 - j60 0.
One stub is at the load, and the other is 3A/8 distance away from the first.
a. Determine the lengths in wavelength of the two short-circuited stubs when a match
is achieved.
100
3-11.
3-12.
3-13.
3-14.
3-15.
Chap. 3
b. Locate and crosshatch the forbidden region of the normalized admittance for possible match.
A lossless transmission line has a characteristic impedance of 300 fl and is terminated
by an impedance Z1. The observed standing-wave ratio on the line is 6, and the distance of the first voltage minimum from the load is 0.166A.
a. Determine the load Ze.
b. Find the lengths in A of two shorted stubs, one at the load and one at A/4 from the
load, which are required to match the load to the line.
A single-stub tuner is to match a lossless line of 400 fl to a load of 800 - j300 fl.
The frequency is 3 GHz.
a. Find the distance in meters from the load to the tuning stub.
b. Determine the length in meters of the short-circuited stub.
A half-wave-dipole antenna has a driving-point impedance of 73 + j42.5 fl. A lossless transmission line connected to a TV set has a characteristic impedance of 300 fl.
The problem is to design a shorted stub with the same characteristic impedance to
match the antenna to the line. The stub may be placed at a location closest to the antenna. The reception is assumed to be Channel 83 at a frequency of 0.88525 GHz.
a. Determine the susceptance contributed by the stub.
b. Calculate the length in centimeters of the stub.
c. Find the distance in centimeters between the antenna and the point where the stub
is placed. [Note: There are two sets of solutions.]
A lossless transmission line has a characteristic impedance of 100 fl and is loaded by
100 + j 100 fl. A single shorted stub with the same characteristic impedance is inserted at A/4 from the load to match the line. The load current is measured to be 2 A.
The length of the stub is A/8.
a. Determine the magnitude and the phase of the voltage across the stub location.
b. Find the magnitude and the phase of the current flowing through the end of the
stub.
A double-stub matching line is shown in Fig. P3-15. The characteristic resistances of
the lossless line and stubs are 100 fl, respectively. The spacing between the two stubs
is A/8. The load is 100 + j 100. One stub is located at the load. Determine:
a. The reactances contributed by the stub
b. The lengths of the two shorted double-stub tuners [Note: There are two sets of solutions.]
R 0 = 50 .11
Figure P3-15
Problems
101
3-16. A lossless transmission line has a characteristic impedance Zo of 100 0 and is loaded
by an unknown impedance. Its voltage standing-wave ratio is 4 and the first voltage
maximum is A/8 from the load.
a. Find the load impedance.
b. To match the load to the line, a quarter section of a different line with a characteristic impedance Zo1 < Zo is to be inserted somewhere between (in cascade with) the
load and the original line. Determine the minimum distance between the load and
matching section, and calculate the characteristic impedance Zo1 in terms of Z0 .
3-17. A lossless transmission line has a characteristic impedance of 100 0 and is loaded by
an unknown impedance. The standing-wave ratio along the line is 2. The first two
voltage minima are located at z = -10 and - 35 cm from the load where z = 0.
Determine the load impedance.
3-18. A matched transmission line is shown in Fig. P3-18.
a. Find l' 1 and d which provide a proper match.
b. With the line and load properly matched determine the VSWR on the section of
line between the stubs.
z, = 100 + jlOO
R 0 = 100 Q
Figure P3-18
Chapter 4
Microwave Waveguides
and Components
4-0 INTRODUCTION
In general, a waveguide consists of a hollow metallic tube of a rectangular or circular shape used to guide an electromagnetic wave. Waveguides are used principally at
frequencies in the microwave range; inconveniently large guides would be required
to transmit radio-frequency power at longer wavelengths. At frequency range X band
from 8.00 to 12.0 GHz, for example, the U.S. standard rectangular waveguide WR90 has an inner width of 2.286 cm (0.9 in.) and an inner height of 1.016 cm
(0.4 in.); but its outside dimensions are 2.54 cm (1 in.) wide and 1.27 cm (0.5 in.)
high [l].
In waveguides the electric and magnetic fields are confined to the space within
the guides. Thus no power is lost through radiation, and even the dielectric loss is
negligible, since the guides are normally air-filled. However, there is some power
loss as heat in the walls of the guides, but the loss is very small.
It is possible to propagate several modes of electromagnetic waves within a
waveguide. These modes correspond to solutions of Maxwell's equations for particular waveguides. A given waveguide has a definite cutoff frequency for each allowed
mode. If the frequency of the impressed signal is above the cutoff frequency for a
given mode, the electromagnetic energy can be transmitted through the guide for
that particular mode without attenuation. Otherwise the electromagnetic energy
with a frequency below the cutoff frequency for that particular mode will be attenuated to a negligible value in a relatively short distance. The dominant mode in a particular guide is the mode having the lowest cutofffrequency. It is advisable to choose
the dimensions of a guide in such a way that, for a given input signal, only the energy of the dominant mode can be transmitted through the guide.
The process of solving the waveguide problems may involve three steps:
102
Sec. 4.1
103
Rectangular Waveguides
1. The desired wave equations are written in the form of either rectangular or
cylindrical coordinate systems suitable to the problem at hand.
2. The boundary conditions are then applied to the wave equations set up in
step 1.
3. The resultant equations usually are in the form of partial differential equations
in either time or frequency domain. They can be solved by using the proper
method.
41 RECTANGULAR WAVEGUIDES
'
(4-1-1)
A =-A-
(4-1-2)
cos ()
sin ()
104
Chap. 4
denotes the number of half waves of electric or magnetic intensity in the x direction,
and n is the number of half waves in the y direction if the propagation of the wave is
assumed in the positive z direction.
;1-z
I /
I/
IL ___ _
Figure 4-1-2
Rectangular coordinates.
= "f 2 E
(4-1-3)
v2 u = '"Y u
(4-1-4)
V2 E
jwe) =a
tions.
Rectangular coordinates are the usual right-hand system. The rectangular components of E or H satisfy the complex scalar wave equation or Helmholtz equation
v2 ljl = 1 21/1
(4-1-5)
a2 lfJ
ax2
a2 lfJ
iJy2
a2 lfJ
iJz2
'}'21/1
(4-1-6)
This is a linear and inhomogeneous partial differential equation in three dimensions. By the method of separation of variables, the solution is assumed in the
form of
ljJ = X(x)Y(y)Z(z)
(4-1-7)
where X (x) = a function of the x coordinate only
Y (y) = a function of the y coordinate only
Z(z) = a function of the z coordinate only
Sec. 4.1
105
Rectangular Waveguides
Substitution of Eq. (4-1-7) in Eq. (4-1-6) and division of the resultant by Eq.
(4-1-7) yield
1 d 2X
1 d 2Y
1 d 2Z
+- - 2 +- - 2
2
X dx
Y dy
Z dz
- -
= ')'2
(4-1-8)
Since the sum of the three terms on the left-hand side is a constant and each term is
independently variable, it follows that each term must be equal to a constant.
Let the three terms be k;, k;, and k;, respectively; then the separation equation
is given by
-k; - k; - k; = ')' 2
(4-1-9)
dx 2
-k;x
(4-1-10)
(4-1-11)
-k;z
(4-1-12)
(4-1-13)
(4-1-14)
(4-1-15)
l/J
(4-1-16)
The propagation of the wave in the guide is conventionally assumed in the positive z
direction. It should be noted that the propagation constant y 8 in the guide differs
from the intrinsic propagation constant 'Y of the dielectric. Let
y; = y 2 + k; + k; = ')' 2 + k~
(4-1-17)
where kc = v'k; + k; is usually called the cutoff wave number. For a lossless
dielectric, y 2 = -w 2 . Then
')'g
Vw
- k~
(4-1-18)
There are three cases for the propagation constant y 8 in the waveguide.
Case I. There will be no wave propagation (evanescence) in the guide if
and y 8 = 0. This is the critical condition for cutoff propagation. The
cutoff frequency is expressed as
w~ = k~
106
Chap. 4
(4-1-19)
Case II.
>
k~ and
(4-1-20)
This means that the operating frequency must be above the cutoff frequency in order
for a wave to propagate in the guide.
Case III.
<
k~ and
w~~~Y -1
(4-1-21)
This means that if the operating frequency is below the cutoff frequency, the wave
will decay exponentially with respect to a factor of - a 8 z and there will be no wave
propagation because the propagation constant is a real quantity. Therefore the solution to the Helmholtz equation in rectangular coordinates is given by
I/!
= [A
sin (kxx)
(4-1-22)
'1
f ,..___/_/,._______,,
1 "---/____
b
//
_y_ __
Figure 4-1-3
gular guide.
o~--a---
Coordinates of a rectan-
y 2 Hz
(4-1-23)
Sec. 4.1
Rectangular Waveguides
107
+ Dn cos
(n:y) ]e-j~,z
(4-1-24)
= -jwH
Vx E
(4-1-25)
V x H = jwE
(4-1-26)
= _ jwHx
(4-1-27)
---=
. H
-1w
(4-1-28)
-aEy - -aEx =
.
-1wHz
aEz _ aEy
ay
az
aEx
az
ax
aEz
ax
ay
(4-1-29)
(4-1-30)
aHx
.
a;- aHz
ax = JwEy
(4-1-31)
= jwEz
(4-1-32)
aHy _ aHx
ax
ay
With the substitution a/az
plified to
= - jf3g
and Ez
{38 Ey
{38 Ex
aEy
ax
aEz
ay
- -
= -wHx
= wHy
.
= -1wHz
:a H
.
-aHz
+ }tJg
y = )WEx
ay
-1:a
tJg Hx - aHz
ax
aHy _ aHx
ax
ay
(4-1-33)
(4-1-34)
(4-1-35)
(4-1-36)
= 1wE
(4-1-37)
=O
(4-1-38)
Solving these six equations for Ex, Ey, Hx, and Hy in terms of Hz will give the TE-
108
Chap. 4
E = - jw iJHz
x
k~
ay
= jw
k~
(4-1-39)
iJHz
(4-1-40)
ax
Ez = 0
= -
(4-1-41)
jf3g aHz
(4-1-42)
k! ax
= - jf3g aHz
k~
Hz
= Eq.
(4-1-43)
ay
(4-1-44)
(4-1-24)
aHz = O
an
(4-1-45)
at the guide walls. Therefore the magnetic field in the positive z direction is given by
(mTTx)
(nTTy)
b e-ll.iZ
3
(4-1-46)
Ex
= Eox cos
(4-1-47)
(4-1-48)
Ez = 0
(4-1-49)
(4-1-50)
(4-1-51)
HY = H0y cos (
Sec. 4.1
109
Rectangular Waveguides
(4-1-52)
= 0, 1, 2, .. .
n = 0, 1, 2, .. .
m = n = 0 excepted
The cutoff wave number kc. as defined by Eq. (4-1-17) for the TEmn modes, is given
by
(4-1-53)
where a and b are in meters. The cutoff frequency, as defined in Eq. (4-1-19) for the
TEmn modes, is
1~
fc
= 2~ 'J-;;z +
(4-1-54)
The propagation constant (or the phase constant here) {3g, as defined in Eq. (4-1-18),
is expressed by
(4-1-55)
The phase velocity in the positive z direction for the TEmn modes is shown as
W
Vg
Vp
(4-1-56)
YI -
(Jc/ f) 2
(4-1-57)
where T/ = ~is the intrinsic impedance in an unbounded dielectric. The wavelength Ag in the guide for the TEmn modes is given by
A
g
YI - (fc/ f)
(4-1-58)
110
TABLE 411
:2:
Chap. 4
Modes
f I !10
TEo1
TE11
TM11
1
1.5
2
3
1
1.5
2
3
1.414
1.803
2.236
3.162
00
00
00
TE 10
a/b
TE20
TEo2
TE21
TM21
TE12
TM12
TE22
TM22
2
2
2
2
2
2
3
4
6
2.236
2.500
2.828
3.606
2.236
3.162
4.123
6.083
2.828
3.606
4.472
6.325
00
00
00
00
TE3o
3
3
3
3
3
guide is called the dominant mode. The dominant mode in a rectangular guide with
a > bis the TE10 mode. Each mode has a specific mode pattern (or field pattern).
It is normal for all modes to exist simultaneously in a given waveguide. The situation is not very serious, however. Actually, only the dominant mode propagates,
and the higher modes near the sources or discontinuities decay very fast.
Example 4-1-1:
,11-z
yl
I /
~---/
t ,____/,,,_____,,
/
3.5 cm
//
.l.._/~~~~~-~~0 1 - - 7 cm ---1
Solution
3 X 108
a. fc = 2a = 2 x 7 x 10-2 = 2.14 GHz
C
b.
Vg
c.
VI -
3 x 108
c
(fc/f}2 =
Ao
Vl -
Sec. 4.1
Rectangular Waveguides
111
Ez =[Am sin
which must be determined according to the given boundary conditions. The procedures for doing so are similar to those used in finding the TE-mode wave.
The boundary conditions on Ez require that the field vanishes at the waveguide
walls, since the tangent component of the electric field Ez is zero on the conducting
surface. This requirement is that for Ez = 0 at x = 0, a, then Bm = 0, and for
Ez = 0 at y = 0, b, then Dn = 0. Thus the solution as shown in Eq. (4-1-60) reduces to
Ez
sin
(n:y)e-jf3gZ
(4-1-61)
where m = I , 2, 3, . . .
n = I, 2, 3, ...
If either m = 0 or n = 0, the field intensities all vanish. So there is no TM01 or
TM10 mode in a rectangular waveguide, which means that TE10 is the dominant
mode in a rectangular waveguide for a > b. For Hz = 0, the field equations, after
expanding V x H = jweE, are given by
iJEz
aJ
+
:a
}pg
Ex
:a
}pg
= - )WJL
. Hy
+ -iJE, = )WJL
iJx
(4-1-62)
(4-1-63)
=O
(4-1-64)
= WEEx
(4-1-65)
-{3gHx = WEEy
(4-1-66)
= jweE,
(4-1-67)
iJEy _ iJEx
iJx
iJy
{3gHy
iJHy - iJHx
iJx
iJy
112
Chap. 4
E = - }{38 aEz
=-
}{38 aEz
Ez
= jwE aEz
k; ay
Eq. (4-1-61)
=-
(4-1-69)
ay
k~
(4-1-68)
ax
k~
(4-1-71)
jwE aEz
(4-1-72)
ax
k~
(4-1-73)
Hz= 0
where f3i
(4-1-70)
= -k'i is replaced.
Ex
. (m7Tx)
Ez
Eq. (4-1-61)
Rx = H0x sin (
Hy
(mry)
.
b e-;13,z
= H0y
cos (
(4-1-74)
(4-1-75)
(4-1-76)
(4-1- 77)
(4-1-78)
Hz= 0
(4-1-79)
fc
/3 8 =
r;;;;-;;;.
2~ y~
w~ ~1 -
,\
=--:====
,\
v1 -
v
g
V1 -
(fc/ /) 2
vp
(fc I !)2
i;z
0r
(4-1-80)
(4-1-81)
(4-1-82)
(4-1-83)
Sec. 4.1
Rectangular Waveguides
113
z = WE
(3g = 11~1
g
(f)f
C4-1-84)
f p ds = f ~CE x H*) ds
C4-1-85)
For a lossless dielectric, the time-average power flow through a rectangular guide is
given by
C4-1-86)
For TEmn modes, the average power transmitted through a rectangular waveguide is
given by
P1r =
V l - ~fc/ J)
ff Cl
2
2
Ex 1
+ IBy 1
) dx dy
C4-1-87)
For TMmn modes, the average power transmitted through a rectangular waveguide is
given by
P1r =
211
1
1-
lbla Cl Ex 1 + IEy j
2
(fc/ !)2
dx dy
C4-1-88)
114
Chap. 4
loss dielectric (that is, CT ~ e), the propagation constant for a plane wave traveling
in an unbounded lossy dielectric is given in Eq. (2-5-20) by
a =
~ [ ; = ri;
(4-1-89)
The attenuation caused by the low-loss dielectric in the rectangular waveguide for
the TEmn or TMmn modes is given by
CTT/
2v' 1 -
ag =
Uc! f) 2
for TE mode
for TM mode
(4-1-90)
(4-1-90a)
As f P fc , the attenuation constant in the guide approaches that for the unbounded
dielectric given by Eq. (4-1-89). However, if the operating frequency is way below
the cutoff frequency, f ~ fc , the attenuation constant becomes very large and nonpropagation occurs.
Now we shall consider power losses caused by the guide walls. When the electric and magnetic intensities propagate through a lossy waveguide, their magnitudes
may be written
\E\ = \Eo,\e-agz
1H1
(4-1-91)
1Hoz )e-ng
l4-l-C>'l)
where Eo, and Hoz are the field intensities at z = 0. It is interesting to note that, for a
=
2agz.
(4-1-93)
For Pioss
l,
(4-1-94)
Finally,
(4-1-95)
where PL is the power loss per unit length. Consequently, the attenuation constant of
the guide walls is equal to the ratio of the power loss per unit length to twice the
power transmitted through the guide.
Since the electric and magnetic field intensities established at the surface of a
low-loss guide wall decay exponentially with respect to the skin depth while the
waves progress into the walls, it is better to define a surface resistance of the guide
walls as
{}/square
(4-1-96)
Sec. 4.1
Rectangular Waveguides
115
(4-1-97)
(4-1-98)
where
I
H 12 = I H, 12
2
Example 4-1-2:
H, 1
Hy 12
(4-1-99)
H,..,12 + I Hty 1
(4-1-100)
Ex= 0
Z8
(7TX)
e- 13
a
1 g'
Hy= 0
E, = 0
where Z8 = wo/{3 8
i ;1-z
I
/
I/
')
~-----
//
/
TF---/-//..,.____--<
I cm
j_L/-/-----~----;~
Of---2
cm
--j
Chap. 4
116
~
=w2J-1-0E0
Tr
- ~ =
~(2f) Tr7
a2 =
Tr ~4
x 9 x
9
20
10
X 1016
X 10-4
=Re[~
ff H*)l
(E x
dx dy u,
dx dy
f3g
= -E~-ab
E~ =
WJLo
53.87 kV/m
In general, the field intensities of the desired mode in a waveguide can be established
by means of a probe or loop-coupling device. The probe may be called a monopole
antenna; the coupling loop, the loop antenna. A probe should be located so as to excite the electric field intensity of the mode, and a coupling loop in such a way as to
generate the magnetic field intensity for the desired mode. If two or more probes or
loops are to be used, care must be taken to ensure the proper phase relationship between the currents in the various antennas. This factor can be achieved by inserting
additional lengths of transmission line in one or more of the antenna feeders.
Impedance matching can be accomplished by varying the position and depth of the
antenna in the guide or by using impedance-matching stubs on the coaxial line feeding the waveguide. A device that excites a given mode in the guide can also serve
reciprocally as a receiver or collector of energy for that mode. The methods of excitation for various modes in rectangular waveguides are shown in Fig. 4-1-6.
In order to excite a TE10 mode in one direction of the guide, the two exciting
antennas should be arranged in such a way that the field intensities cancel each other
in one direction and reinforce in the other. Figure 4-1-7 shows an arrangement for
launching a TE 10 mode in one direction only. The two antennas are placed a quarterwavelength apart and their phases are in time quadrature. Phasing is compensated by
use of an additional quarter-wavelength section of line connected to the antenna
Sec. 4.1
117
Rectangular Waveguides
Coaxial line
Antenna
probe
TE 10 mode
TE 20 mode
Short-circuited-----__,(_
end
Antenna probe
TM 11 mode
Figure 4-1-6
Outof
i
phase 2 -
TM 21 mode
Methods of exciting various modes in rectangular waveguides.
-1
-2
In phase
Waveguide
feeders. The field intensities radiated by the two antennas are in phase opposition to
the left of the antennas and cancel each other, whereas in the region to the right of
the antennas the field intensities are in time phase and reinforce each other. The resulting wave thus propagates to the right in the guide.
Some higher modes are generated by discontinuities of the waveguide such as
obstacles, bends, and loads. However, the higher-order modes are, in general, more
highly attenuated than the corresponding dominant mode. On the other hand, the
dominant mode tends to remain as a dominant wave even when the guide is large
enough to support the higher modes.
417 Characteristics of Standard Rectangular
Waveguides
118
TABLE 417
Physical dimensions
EIA"
designation
WRb( )
2300
2100
1800
1500
1150
975
770
650
510
430
340
284
229
187
159
137
112
90
75
62
51
42
Chap.4
Inside, in cm (in.)
Outside, in cm (in.)
Width
Height
Width
Height
58.420
(23.000)
53.340
(21.000)
45.720
(18.000)
38.100
(15.000)
29.210
(11.500)
24.765
(9.750)
19.550
(7.700)
16.510
(6.500)
12.954
(5.100)
10.922
(4.300)
8.636
(3.400)
7.214
(2.840)
5.817
(2.290)
4.755
29.210
(11.500)
26.670
(10.500)
22.860
(9.000)
19.050
(7 .500)
14.605
(5.750)
12.383
(4.875)
9.779
(3.850)
8.255
(3.250)
6.477
(2.500)
5.461
(2.150)
4.318
(1.700)
3.404
(1.340)
2.908
(1.145)
2.215
59.055
(23.250)
53.973
(21.250)
46.350
(18.250)
38.735
(15.250)
29.845
(11.750)
25.400
(I0.000)
20.244
(7.970)
16. 916
(6.660)
13.360
(5.260)
11.328
(4.460)
9.042
(3.560)
7.620
(3.000)
6.142
(2.418)
5.080
(1.872)
(0.872)
(2.000)
4.039
(1.590)
3.485
(1.372)
2.850
(1.122)
2.286
(0.900)
1.905
(0.750)
1.580
(0.622)
1.295
(0.510)
1.067
(0.420)
2.019
(0.795)
1.580
(0.622)
1.262
(0.497)
1.016
(0.400)
0.953
(0.375)
0.790
(0.311)
0.648
(0.255)
0.432
(0.170)
4.364
(1.718)
3.810
(1.500)
3.175
(1.250)
2.540
(1.000)
2.159
(0.850)
1.783
(0.702)
1.499
(0.590)
1.270
(0.500)
29.845
(11.750)
27.305
(10.750)
23.495
(9.250)
19.685
(7. 750)
15.240
(6.000)
13.018
(5.125)
10.414
(4.100)
8.661
(3.410)
6.883
(2.710)
5.867
(2.310)
4.724
(1.860)
3.810
(1.500)
3.233
(1.273)
2.540
(J.000)
2.344
(0.923)
1.905
(0.750)
1.588
(0.625)
1.270
(0.500)
1.207
(0.475)
0.993
(0.391)
0.851
(0.335)
0.635
(0.250)
Cutoff
frequency for
air-filled
waveguide
in GHz
Recommended
frequency
range for
TE10 mode
in GHZ
0.257
0.32-0.49
0.281
0.35--0.53
0.328
0.41-0.62
0.394
0.49-0.75
0.514
0.64-0.98
0.606
0.76-1.15
0.767
0.96-1.46
0.909
1.14-1.73
1.158
1.45--2.20
1.373
1.72-2.61
1.737
2.17-3.30
2.079
2.60--3.95
2.579
3.22-4.90
3.155
3.94-5.99
3.714
4.64-7.05
4.304
5.38-8.17
5.263
6.57-9.99
6.562
8.20--12.50
7.874
9.84-15.00
9.494
11.90-18.00
11.583
14.50--22.00
14.058
17.60--26.70
Sec. 4.2
TABLE 417
119
Circular Waveguides
Physical dimensions
EIA"
designation
WRh( )
34
28
22
19
15
12
10
8
7
5
4
3
Inside, in cm (in.)
Outside, in cm (in.)
Width
Height
Width
Height
0.864
(0.340)
0.711
(0.280)
0.569
(0.224)
0.478
(0.188)
0.376
(0.148)
0.310
(0.122)
0.254
(0.100)
0.203
(0.080)
0.165
(0.065)
0.130
(0.051)
0.109
(0.043)
0.086
(0.034)
0.432
(0.170)
0.356
(0.140)
0.284
(0.112)
0.239
(0.094)
0.188
(0.074)
0.155
(0.061)
0.127
(0.050)
0.102
(0.040)
0.084
(0.033)
0.066
(0.026)
0.056
(0.022)
0.043
(0.017)
1.067
(0.420)
0.914
(0.360)
0.772
(0.304)
0.681
(0.268)
0.579
(0.228)
0.513
(0.202)
0.457
(0.180)
0.406
(0.160)
0.343
(0.135)
0.257
(0.101)
0.211
(0.083)
0.163
(0.064)
0.635
(0.250)
0.559
(0.220)
0.488
(0.192)
0.442
(0.174)
0.391
(0.154)
0.358
(0.141)
0.330
(0.130)
0.305
(0.120)
0.262
(0.103)
0.193
(0.076)
0.157
(0.062)
0.119
(0.047)
Cutoff
frequency for
air-filled
waveguide
in GHz
Recommended
frequency
range for
TE10 mode
in GHZ
17.361
21.70--33.00
21.097
26.40--40.00
26.362
32.90--50.10
31.381
39.20--59.60
39.894
49.80--75.80
48.387
60.50--91.90
59.055
73.80--112.00
73.892
92.20--140.00
90.909
114.00--173.00
115.385
145.00--220.00
137.615
172.00--261.00
174.419
217.00--333.00
high, and its inside dimensions are 2.286 cm (0.90 in.) wide and 1.016 cm (0.40
in.) high. Table 4-1-7 tabulates the characteristics of the standard rectangular
waveguides.
42 CIRCULAR WAVEGUIDES
A circular waveguide is a tubular, circular conductor. A plane wave propagating
through a circular waveguide results in a transverse electric (TE) or transerse magnetic (TM) mode. Several other types of waveguides, such as elliptical and reentrant
guides, also propagate electromagnetic waves.
421 Solutions of Wave Equations in Cylindrical
Coordinates
As described in Section 4-1 for rectangular waveguides, only a sinusoidal steadystate or frequency-domain solution will be attempted for circular waveguides. A
cylindrical coordinate system is shown in Fig. 4-2-1.
120
Chap. 4
Figure 4-2-1
Cylindrical coordinates.
I~
T ar
T2 aq,2
az2
"'2,,,
I
(4-2-1)
'f'
Using the method of separation of variables, the solution is assumed in the form of
'I'= R(r)<P(</>)Z(z)
where R (r)
ct>(</>)
Z (z)
(4-2-2)
Substitution of Eq. (4-2-2) in (4-2-1) and division of the resultant by (4-2-2) yield
1 d ( dR)
rR dr \r dr
1 d <P
r <P d</> 2
2
1d Z
+ Z dz 2
= y
4 _2_3)
Since the sum of the three independent terms is a constant, each of the three terms
must be a constant. The third term may be set equal to a constant y~:
d 2Z
dz 2
- '}'g2z
(4-2-4)
(r dR)
dr
r2 = 0
(4-2-6)
The second term is a function of </> only; hence equating the second term to a con-
Sec. 4.2
121
Circular Waveguides
stant ( -n 2) yields
d2<1>
d</>2 -- - n2<1>
(4-2-7)
<I>
(4-2-8)
d ( dR) + [(kcr) 2 -
2
n ]R
r dr r dr
=0
(4-2-9)
+ 'Y 2 = 'Yi
(4-2-10)
This equation is called the characteristic equation of Bessel's equation. For a lossless
guide, the characteristic equation reduces to
~---
{3g =
Yw 2.E
k~
(4-2-11)
:::,
....,~
~
0.4
0.2
:0
>
-0.2
-;;;
-0.4
-0.6
10
12
Figure 4-2-2
14
16
Chap. 4
122
0.6
0.4 ~No
0.2
'o
-o.4
-0.6
'
-0.8
-1.0
-1.2
N
2
x)\./ \
>
:
N3
I ~ lX~t-- !'\.
1/
N,I
t"'-
II
i,.-
\ IJ\
' 7-
,..,,
1-
l\tr"\ ) K l7
v
v
/ )c
\.. ~
"'"-- """"'
... i"--o..
/v'-
)
-~
I /,
0
10
12
14
16
Figure 4-2-3
cos (n</J)
(4-2-14)
'I' = 'l'oJ.(kcr)
COS
(n</J)e-jJ3gz
(4-2-15)
Figure 4-24
waveguide.
Coordinates of a circular
Sec. 4.2
Circular Waveguides
123
that the z component of the magnetic field Hz must exist in the guide in order to have
electromagnetic energy transmission. A Helmholtz equation for Hz in a circular
guide is given by
V 2 Hz
= y 2 Hz
(4-2-16)
Hz
= Hozln(kcr)
COS
(n</> )e-if3gz
(4-2-17)
= -jwH
H = jweE
V x E
(4-2-18)
(4-2-19)
1 aEz aEq,
.
- - = - 1wH,
r a<f>
az
--
aE,
az
aEz
ar
= - 1wHq,
1 a
1 aE,
- -(rEq,) - - r ar
r a<f>
= - JWJLHz
- -
.
.
az
- }{38 H, -
aHz
.
ar
= ]WEEq,
1 a
1 aH,
- - (rHq,) - - r ar
r a<f>
.
= ]WEE,
(4-2-20)
(4-2-21)
(4-2-22)
(4-2-23)
(4-2-24)
(4-2-25)
When the differentiation a/ az is replaced by ( - }{38 ) and the z component of electric field Ez by zero, the TE-mode equations in terms of Hz in a circular waveguide
are expressed as
E = _jw _! aH,
r
k~ r a<f>
jw aHz
(4-2-26)
Eq,
= k~ ar
(4-2-27)
Ez
=0
(4-2-28)
H = - }{38 aHz
r
k~
ar
(4-2-29)
124
Hi =
H _ - jf3 8 _! aHz
"' k~ r aq,
Hozln(kcr) cos (n<f>)e-jJ3gz
Chap. 4
(4-2-30)
(4-2-31)
aa~z Ir~a = 0
E"'
= 0 at r = a
:.
H,
= 0 at r =a
:. a;z
or
lr~a = 0
aHz
ar
(4-2-32)
r=a
Hence
(4-2-33)
where J ~ indicates the derivative of J" .
Since the ln are oscillatory functions, the J~ (kca) are also oscillatory functions. An infinite sequence of values of (kca) satisfies Eq. (4-2-32). These points,
the roots of Eq. (4-2-32), correspond to the maxima and minima of the curves
J~(kca), as shown in Fig. 4-2-2. Table 4-2-1 tabulates a few roots of J~(kca) for
some lower-order n.
TABLE 421
p
1
2
3
4
n=
pth ZEROS OF
J~(Kcal
0
3.832
7.016
10.173
13.324
1.841
5.331
8.536
11.706
3.054
6.706
9.969
13.170
4.201
8.015
11.346
5.317
9.282
12.682
6.416
10.520
13.987
kc=
x:.,,
a
(4-2-34)
Substitution of Eq. (4-2-17) in Eqs. (4-2- 26) through (4-2-31) yields the complete
field equations of the
modes in circular waveguides:
TEnp
E, = Eo,ln(X;r) sin (n<f>)e-jJ3gZ
(4-2-35)
Sec. 4.2
Circular Waveguides
125
Eq, = Eoq,l~(x;r)
cos (ncf>)e-Jf3gz
Ez = 0
(4-2-36)
(4-2-37)
H, = - ~: 1.(x;r)
cos (ncf>)e-Jf3gz
(4-2-38)
(4-2-39)
Hz
cos (ncf>)e-Jf3gz
(4-2-40)
Z8
Hozl.(X;r)
where Z 8 = E,/ Hq, = - Eq,/ H, has been replaced for the wave impedance in the
guide and where n = 0, 1, 2, 3, ... andp = 1, 2, 3, 4, ....
The first subscript n represents the number of full cycles of field variation in
one revolution through 21T rad of cf>. The second subscript p indicates the number of
zeros of Eq,-that is, J~(X~Pr/a) along the radial of a guide, but the zero on the axis
is excluded if it exists.
The mode propagation constant is determined by Eqs. (4-2-26) through
(4-2-31) and Eq. (4-2-34):
(4-2-41)
The cutoff wave number of a mode is that for which the mode propagation constant
vanishes. Hence
(4-2-42)
The cutoff frequency for TE modes in a circular guide is then given by
j;
'
X~p
21Ta~
(4-2-43)
w
Vg
{3g
Yl - ~fcl /)2
(4-2-44)
\11 - (fc/ /) 2
and
Z = WJL =
g
f3g
V1 -
T/
(j;/ /) 2
(4-2-46)
126
where A
T/
7=
Chap. 4
Example 4-2-1:
Figure 4-2-5
x
4-2-1.
a. From Table 4-2-1 for TE11 mode, n = 1, p = l, and X; 1 = l.841 = kca. The
cutoff wave number is
k
= l.841 = l.841
a
5 x 10- 2
= 36 82
(36.82)(3 x 108 )
k,
~~
f=
l. 758 x 109 Hz
27T V f-loEo
b. The phase constant in the guide is
{3 8 =
Yw
,ot:o -
k~
50.9 rads/m
Ag
= 27T = 6.28 = 12 3
{3g
50.9
. cm
(36.82) 2
Sec. 4.2
Circular Waveguides
127
= WJ.Lo =
(21T x 3 x 10 )(41T x
io- 1) = 465 n
50.9
f3R
v2,
y2E,
(4-2-47)
(4-2-48)
1
2
3
4
n=
5.136
8.417
11.620
14.796
6.380
9.761
13.015
7.588
11.065
14.372
8.771
12.339
0
2.405
5.520
8.645
11.792
3.832
7.106
10.173
13.324
For H, = 0 and ajaz = - jf3g, the field equations in the circular guide, after
expanding V x E = - jwH and V X H = jwE E, are given by
E
r
= -
jf3g aEz
ar
k~
- jf3g 1 aE,
(4-2-50)
(4-2-51)
Eq,
= 77-
E,
= Eq. (4-2-48)
(4-2-52)
= jwE !
(4-2-53)
k~
iJ<f>
aE,
r a<f>
128
Chap. 4
(4-2-54)
H,
=0
(4-2-55)
Er=
Eorl~(X;r)
COS
(n<f>)e-if3gZ
(4-2-56)
Eq,
= Eoq,ln(X;r)
sin (ncf>)e-if3gZ
(4-2-57)
E,
= Eo,ln(X;r)
COS
(ncf>)e-Jf3gZ
(4-2-58)
Eoq, (Xnpr)
. (ncf> )e-1.13gz
Hr = -z;ln
-;_;- sm
(4-2-59)
(4-2-60)
zli
H,
=0
(4-2-61)
where Z8 = Er/Hq, = - Eq,/Hr = {3 8 /(we) and kc= Xnp/a have been replaced and
where n = 0, 1, 2, 3, ... and p = l, 2, 3, 4, ....
Some of the TM-mode characteristic equations in the circular guide are identical to those of the TE mode, but some are different. For convenience, all are shown
here:
(4-2-62)
(4-2-63)
f.
Xnp
21TayP;
W
(4-2-64)
Vp
V1 -
(fc/ f) 2
Vg
= {3g =
,\Ii
= -Yr=l=-=(=fc=/J=)2
Zg
~: = ~ ~ 1 -
(4-2-65)
(4-2-66)
Gr
(4-2-67)
It should be noted that the dominant mode, or the mode of lowest cutoff frequency
Sec. 4.2
Circular Waveguides
129
in a circular waveguide, is the mode of TE,, that has the smallest value of the
product, kca = 1. 841, as shown in Tables 4-2-1 and 4-2-2.
Example 4-2-2:
An air-filled circular waveguide has a radius of 2 cm and is to carry energy at a frequency of IO GHz. Find all the TEnp and TMnp modes for which energy transmission is
possible.
Solution Since the physical dimension of the guide and the frequency of the wave remain constant, the product of (kca) is also constant. Thus
Any mode having a product of (kca) less than or equal to 4.18 will propagate the wave
with a frequency of IO GHz. This is
TM01(2.405)
TE21(3.054)
TM11(3.832)
TEo,(3.832)
= H, = 0
This means that the electric and magnetic fields are completely transverse to the direction of wave propagation. This mode cannot exist in hollow waveguides, since it
requires two conductors, such as the coaxial transmission line and two-open-wire
line. Analysis of the TEM mode illustrates an excellent analogous relationship between the method of circuit theory and that of the field theory. Figure 4~2-6 shows
a coaxial line.
Maxwell's curl equations in cylindrical coordinates
Vx E
(4-2-68)
= -jwH
H = jweE
(4-2-69)
= wH<I>
= wH,
(4-2-70)
become
BgEr
BgE<f>
(4-2-71)
130
Chap. 4
y
Figure 4-2-6
line.
d
dEr
dr (rEc1>) - de/> = 0
{3 8 H,
{3gHc/>
(4-2- 72)
= -wEEcJ>
= WEE,
-d (rHcJ>) - -dH, = 0
dr
Coordinates of a coaxial
(4-2- 73)
(4-2-74)
(4-2-75)
de/>
{3g
= w~
(4-2-76)
which is the phase constant of the wave in a lossless transmission line with a dielectric.
In comparing the preceding equation with the characteristic equation of the
Helmholtz equation in cylindrical coordinates as given in Eq. (4-2-11) by
{3 8
V W 2 f.LE
k~
(4-2-77)
it is evident that
(4-2-78)
This means that the cutoff frequency of the TEM mode in a coaxial line is zero,
which is the same as in ordinary transmission lines.
The phase velocity of the TEM mode can be expressed from Eq. (4-2- 76) as
Vp
= {3g =
1
~
(4-2-79)
Sec. 4.2
Circular Waveguides
131
The wave impedance of the TEM mode is found from either Eqs. (4-2-70) and
(4-2-73) or Eqs. (4-2-71) and (4-2-74) as
TJ (TEM) = [ ;
(4-2-80)
H . de
Ioe-jf3gZ
= 2TTrHq,
(4-2-81)
where I is the complex current that must be supported by the center conductor of a
coaxial line. This clearly demonstrates that the TEM mode can only exist in the twoconductor system-not in the hollow waveguide because the center conductor does
not exist.
In summary, the properties of TEM modes in a lossless medium are as follows:
1. Its cutoff frequency is zero.
2. Its transmission line is a two-conductor system.
3. Its wave impedance is the impedance in an unbounded dielectric.
4. Its propagation constant is the constant in an unbounded dielectric.
5. Its phase velocity is the velocity of light in an unbounded dielectric.
P,r
= 22
Pir
= ~g
where Z8
a
(4-2-82)
2,,.fa
(4-2-83)
Er
.
d
. h
.d
= Hq,
= - Eq,
Hr = wave 1mpe ance mt e gm e
= radius of the circular guide
Substitution of Z 8 for a particular mode in Eq. (4-2-82) yields the power transmitted
by that mode through the guide.
132
Chap. 4
For TEnp modes, the average power transmitted through a circular guide is
given by
(4-2-84)
where T/ = ~is the intrinsic impedance in an unbounded dielectric.
For TMnp modes, the average power transmitted through a circular guide is
given by
P
tr
= 2YJ
(4-2-85)
For TEM modes in coaxial lines, the average power transmitted through a
coaxial line or two-open-wire line is given by
Pir
2
= 21 i21Tia [IE, 1
+ IEq, 2 ]r dr def>
(4-2-86)
T/
(4-2-87)
the magnetic intensity induced by the current around the center conductor is given
by Ampere's law as
lo
Hq, = - e
2TTr
_113 z
(4-2-88)
The potential rise from the outer conductor to the center conductor is given by
V,
=-
1f3gz
(4-2-89)
= ~ = .!]_ ln
2TT
(~)
a
(4-2-90)
Pir
Substitution of
l f21Tlb YJHq,
= 2YJ
2
1
r dr def>
(b)
16
= ~'TT In ;;
(4-2-91)
t Volo
(4-2-92)
This shows that the power transmission derived from the Poynting theory is the
same as from the circuit theory for an ordinary transmission line.
Sec. 4.2
Circular Waveguides
133
PL= 2aP1r
a=~(R~+G~)
(4-2-94)
TM 01 mode
TE 11 mode
134
Single stub
____.,..__---1.__ /_
Chap. 4
Single stub
Waveguide
slot
~II
I) )))(~ 11 w.,oguido
_coaxial-line
Coaxial line
(a)
(b)
EIA"
designation
WCh( )
992
847
724
618
528
451
385
329
281
240
205
175
150
128
109
94
80
69
59
50
44
38
33
28
25
22
19
17
14
13
II
Inside diameter
in cm (in.)
25.184
21.514
18.377
15.700
13.411
11.458
9.787
8.362
7.142
6.104
5.199
4.445
3.810
3.254
2.779
2.383
2.024
1.748
1.509
1.270
1.113
0.953
0.833
0.714
0.635
0.556
0.478
0.437
0.358
0.318
0.277
0.239
(9.915)
(8.470)
(7.235)
(6.181)
(5.280)
(4.511)
(3.853)
(3.292)
(2.812)
(2.403)
(2.047)
(1.750)
(1.500)
(1.281)
(1.094)
(0.938)
(0.797)
(0.688)
(0.594)
(0.500)
(0.438)
(0.375)
(0.328)
(0.281)
(0.250)
(0.219)
(0.188)
(0.172)
(0.141)
(0.125)
(0.109)
(0.094)
Cutoff frequency
for air-filled
waveguide in GHz
Recommended
frequency range
for TE11 mode in GHz
0.698
0.817
0.957
1.120
1.311
1.534
1.796
2.102
2.461
2.880
3.381
3.955
4.614
5.402
6.326
7.377
8.685
10.057
11.649
13.842
15.794
18.446
21.103
24.620
27.683
31.617
36.776
40.227
49.103
55.280
63.462
73.552
0.80-1.10
0.94-1.29
1.10-1.51
1.29-1.76
1.51-2.07
1. 76-2.42
2.07-2.83
2.42-3.31
2.83-3.88
3.31-4.54
3.89-5.33
4.54-6.23
5.30-7.27
6.21-8.51
7.27-9.97
8.49-11.60
9.97-13.70
11.60-15.90
13.40-18.40
15.90-21.80
18.20-24.90
21.20-29.10
24.30-33.20
28.30-38.80
31.80-43.60
36.40-49.80
42.40-58.10
46.30-63.50
56.60-77.50
63.50-87 .20
72.70-99.70
84.80-116.00
Sec. 4.3
Microwave Cavities
135
The inner diameter of a circular waveguide is regulated by the frequency of the signal being transmitted. For example: at X-band frequencies from 8 to 12 GHz, the inner diameter of a circular waveguide designated as EIA WC(94) by the Electronic
Industry Association is 2.383 cm (0.938 in.). Table 4-2-8 tabulates the characteristics of the standard circular waveguides.
43 MICROWAVE CAVITIES
In general, a cavity resonator is a metallic enclosure that confines the electromagnetic energy. The stored electric and magnetic energies inside the cavity determine
its equivalent inductance and capacitance. The energy dissipated by the finite conductivity of the cavity walls determines its equivalent resistance. In practice, the
rectangular-cavity resonator, circular-cavity resonator, and reentrant-cavity resonator are commonly used in many microwave applications.
Theoretically a given resonator has an infinite number of resonant modes, and
each mode corresponds to a definite resonant frequency. When the frequency of an
impressed signal is equal to a resonant frequency, a maximum amplitude of the
standing wave occurs, and the peak energies stored in the electric and magnetic
fields are equal. The mode having the lowest resonant frequency is known as the
dominant mode.
The electromagnetic field inside the cavity should satisfy Maxwell's equations, subject to the boundary conditions that the electric field tangential to and the magnetic
field normal to the metal walls must vanish. The geometry of a rectangular cavity is
illustrated in Fig. 4-3-1.
The wave equations in the rectangular resonator should satisfy the boundary
condition of the zero tangential E at four of the walls. It is merely necessary to
choose the harmonic functions in z to satisfy this condition at the remaining two end
I /
I/
,//!'- --
r//
fl
_J/~
of.-a~
Figure 4-3-1
gular cavity.
Coordinates of a rectan-
136
Chap. 4
m7T x)
(mr
. (p1Tz)
b y) sm
d
(TEmnp)
(4-3-1}
x) . (nTT
b x) cos (p1Tz)
d
. (m7T
Ez = Eoz sm
-a- sm
(TMmnp)
(4-3-2)
where m = 1, 2, 3, 4, ..
n = I, 2, 3, 4, ..
p = 0, I, 2, 3, ..
The separation equation for both TE and TM modes is given by
(4-3-3)
For a lossless dielectric, k 2
by
= w 21u::; therefore,
Sec. 4.3
137
Microwave Cavities
z
Figure 4-3-3
resonator.
Coordinates of a circular
boundary conditions described for a rectangular-cavity resonator. It is merely necessary to choose the harmonic functions in z to satisfy the boundary conditions at the
remaining two end walls. These can be achieved if
(X~pr)
( ) . (q1Tz)
d
(4-3-5)
Xnpr)
= Eozln (-;;-
COS
(n</>)
COS
(q1TZ)
d
(TMnpq)
(4-3-6)
= 0, l, 2, 3, .. .
= 1, 2, 3, 4, .. .
q = 0, l, 2, 3, .. .
wheren
p
(x;pr + (q;r
k2 = (X;pr + (q:r
k2 =
(TE mode)
(4-3- 7)
(TM mode)
(4-3-8)
Substitution of k2 = w 2 E in Eqs. (4-3-7) and (4-3-8) yields the resonant frequencies for TE and TM modes, respectively, as
fr
21T~
~(X~p)z + (q1T)2
a
(TE)
(4-3-9)
138
f,
(4-3-10)
(TM)
Chap.4
X~pr)
= Hozln (-;;-
.
cos (ncp ) sm
(qTTz)
d
(TE mode)
(4-3-11)
where n = 0, I, 2, 3, .. .
p = 1, 2, 3, 4, .. .
q = 1, 2, 3, 4, .. .
a = radius of the semicircular-cavity resonator
1
d
l_
y
The wave function of the TMnpq mode in the semicircular-cavity resonator can be
written
(TM mode)
(4-3-12)
where n = 1, 2, 3, 4, ..
p = 1, 2, 3, 4, ..
q = 0, 1, 2, 3, ..
With the separation equations given in Eqs. (4-3- 7) and (4-3-8), the equations of
resonant frequency for TE and TM modes in a semicircular-cavity resonator are the
same as in the circular-cavity resonator. They are repeated as follows:
f, =
/(X~p) 2 + (q1Ta)
2TTa~\j
d
l
(TEnpq mode)
(4-3-13)
Sec. 4.3
139
Microwave Cavities
fr
/ (Xnp) 2
27Ta~\j
(q-rra)
d
(TMnpq mode)
(4-3-14)
However, the values of the subscripts n, p, and q differ from those for the circularcavity resonator. Also, it must be emphasized that the TE111 mode is dominant if
d > a and that the TM110 mode is dominant if d < a.
4.3.3 Q Factor of a Cavity Resonator
The quality factor Q is a measure of the frequency selectivity of a resonant or antiresonant circuit, and it is defined as
Q =
7T
wW
(4-3-15)
where Wis the maximum stored energy and Pis the average power loss.
At resonant frequency, the electric and magnetic energies are equal and in
time quadrature. When the electric energy is maximum, the magnetic energy is zero
and vice versa. The total energy stored in the resonator is obtained by integrating
the energy density over the volume of the resonator:
W,
J ~ 11 dv =Wm= J ~ IHl
2
dv
=W
(4-3-16)
where IE I and IHI are the peak values of the field intensities.
The average power loss in the resonator can be evaluated by integrating the
power density as given in Eq. (2-5-12) over the inner surface of the resonator.
Hence
(4-3-17)
where H, is the peak value of the tangential magnetic intensity and Rs is the surface
resistance of the resonator.
Substitution of Eqs. (4-3-16) and (4-3-17) in Eq. (4-3-15) yields
2
Q = wfvlHl dv
Rs f. J H, 12 da
(4 _3 _18)
Since the peak value of the magnetic intensity is related to its tangential and normal
components by
where Hn is the peak value of the normal magnetic intensity, the value of I H, 12 at the
resonator walls is approximately twice the value of I H 12 averaged over the volume.
So the Q of a cavity resonator as shown in Eq. (4-3-18) can be expressed approximately by
w(volume)
Q = 2Rs(surface areas)
(4-3-19)
140
Chap. 4
= 2-rrVlc
fo
Qo = WoL
(4-3-21)
(a)
Figure 4-3-5
circuit.
(b)
woL
N 2Z g
(4-3-22)
Nizg
R
(4-3-23)
Qe
= R(l +
Qo
K)
= -1-+-K
(4-3-24)
1
Qo
-=-+-
(4-3-25)
Qext
(4-3-26)
Sec. 4.4
141
K=p
Qe
Qo
I
+p
(4-3-29)
K =p
(4-3-30)
Qc - p + I Qo
(4-3-31)
7
Cl.
0
.....
;:
1U
~
5
4
Undercoupled
.:0
:ac::
c::
..
Coupling coefficient K
Chap.4
142
plane tee, H -plane tee, magic tee, hybrid ring (rat-race circuit), directional coupler,
and the circulator. This section describes these microwave hybrids, which are shown
in Fig. 4-4-1.
A two-port network is shown in Fig. 4-4-2. From network theory a two-port
device can be described by a number of parameter sets, such as the H, Y, Z, and
ABCD parameters.
H parameters: [
~:]
[h11
hz1
h12J[/']
hz2 Vi
Y parameters:
[~:]
[Y11
Y21
Y12]
Y22 Vi
[Vi]
Vi = h1il1 + h12 Vi
/z = hz1 I, + hz2 Vi
(4- 4-1)
(4- 4- 2)
= Y11 V1 + Y12Vz
/z = Y21 Vi + Y22 Vi
(4-4-3)
(4-4-4)
11
Collinear arms
H
Port 1
Port 1
Port 3
Harm
Collinear arms
4Zo
2Z0
Zo
(d) Hybrid ring
(c) Magic tee
(2n + I)
Ag
4 .. ,
0--~-----~--
0--1'.:s,---~}~
Port I
-0~p-
Port 3
Coupling holes
(e) Directional coupler
(f) Circulator
Figure 4-4-1 Microwave hybrids. (a) -plane tee. (b) H-plane tee. (c) Magic tee.
(d) Hybrid ring. (e) Directional coupler. (f) Circulator.
Sec. 4.4
--
143
Ii
12
-- +
v2
--
Two-port
device
V1
--
Figure 4-4-2
Z parameters:
ABCD parameters:
[VVi]2
=
_ [Z11
Vi = Z11 /1 + Z12/2
Z12J [/1]
Z22 /2
Z21
Two-port network.
Vz = Z21f1 + Z22/z
Vi = AVi - B/2
11
= CVi -
D/2
(4-4-5)
(4-4-6)
(4-4-7)
(4-4-8)
All these network parameters relate total voltages and total currents at each of the
two ports. For instance,
h11
Vi I
=I;
(short circuit)
(4-4-9)
(open circuit)
(4-4-10)
V2=0
h12
= -Vi I
Vi
11=0
If the frequencies are in the microwave range, however, the H, Y, and Z parameters
cannot be measured for the following reasons:
1. Equipment is not readily available to measure total voltage and total current at
the ports of the network.
2. Short and open circuits are difficult to achieve over a broad band of frequencies.
3. Active devices, such as power transistors and tunnel diodes, frequently will not
have stability for a short or open circuit.
Consequently, some new method of characterization is needed to overcome these
problems. The logical variables to use at the microwave frequencies are traveling
waves rather than total voltages and total currents. These are the S parameters,
which are expressed as
(4-4-1 la)
(4-4-1 lb)
Figure 4-4-3 shows the S parameters of a two-port network.
a2~
VV'-'- a1
Su
S12
Z2
v,
~
b1
S21
S22
b2
Figure 4.4.3
Two-port network.
144
Chap. 4
1. A short circuit may always be placed in one of the arms of a three-port junction in such a way that no power can be transferred through the other two
arms.
2. If the junction is symmetric about one of its arms, a short circuit can always be
placed in that arm so that no reflections occur in power transmission between
the other two arms. (That is, the arms present matched impedances.)
3. It is impossible for a general three-port junction of arbitrary symmetry to
present matched impedances at all three arms.
The -plane tee and H-plane tee are described below.
E-plane tee {series tee). An -plane tee is a waveguide tee in which the
axis of its side arm is parallel to the E field of the main guide (see Fig. 4-4-4). If the
collinear arms are symmetric about the side arm, there are two different transmission characteristics (see Fig. 4-4-5). It can be seen from Fig. 4-4-4 that if the E plane tee is perfectly matched with the aid of screw tuners or inductive or capacitive
windows at the junction, the diagonal components of the scattering matrix, S11, S22,
and S33, are zero because there will be no reflection. When the waves are fed into
the side arm (port 3), the waves appearing at port l and port 2 of the collinear arm
will be in opposite phase and in the same magnitude. Therefore
(4-4-12)
It should be noted that Eq. (4-4-12) does not mean that S13 is always positive and S23
Port 3
Figure 4-4-4
-plane tee
Sec. 4.4
145
............------'fl Port
l. .
In
. . . . . . . .___.Fl'ort,
Po~~t~t~t________--Jlul_O:~rt
(b)
is always negative. The negative sign merely means that SB and S23 have opposite
signs. For a matched junction, the S matrix is given by
[s~1 S~ ~::]
S31
S32
(4-4-13)
From the symmetry property of S matrix, the symmetric terms in Eq. (4-4-13) are
equal and they are
(4-4-14)
From the zero property of S matrix, the sum of the products of each term of any
column (or row) multiplied by the complex conjugate of the corresponding terms of
any other column (or row) is zero and it is
(4-4-15)
Hence
(4-4-16)
This means that either S13 or Sf3 , or both, should be zero. However, from the unity
property of S matrix, the sum of the products of each term of any one row (or
column) multiplied by its complex conjugate is unity; that is,
(4-4-17)
+ S32Sf2 =
S13Sf3 + S23Sf3 =
(4-4-18)
(4-4-19)
IS23 l2
(4-4-20)
146
Chap. 4
zero and thus Eq. (4-4-19) is false. In a similar fashion, if S23 = 0, then S 13 becomes
zero and therefore Eq. (4-4-20) is not true. This inconsistency proves the statement
that the tee junction cannot be matched to the three arms. In other words, the diagonal elements of the S matrix of a tee junction are not all zeros.
In general, when an -plane tee is constructed of an empty waveguide, it is
poorly matched at the tee junction. Hence SiJ 0 if i = j. However, since the
collinear arm is usually symmetric about the side arm, \ S13 \ = \S23 \ and S11 = S22.
Then the S matrix can be simplified to
s = [~::
(4-4-21)
sl3
H-plane tee (shunt tee). An H-plane tee is a waveguide tee in which the
axis of its side arm is "shunting" the E field or parallel to the H field of the main
guide as shown in Fig. 4-4-6.
Port I
It can be seen that if two input waves are fed into port 1 and port 2 of the
collinear arm, the output wave at port 3 will be in phase and additive. On the other
hand, if the input is fed into port 3, the wave will split equally into port 1 and port 2
in phase and in the same magnitude. Therefore the S matrix of the H-plane tee is
similar to Eqs. (4-4-13) and (4-4-21) except that
(4-4-22)
442 Magic Tees (Hybrid Tees}
A magic tee is a combination of the -plane tee and H -plane tee (refer to Fig.
4-4-7). The magic tee has several characteristics:
Port 3
Port 2
Port I
Figure 4-4-7
Magic tee.
Sec. 4.4
147
1. If two waves of equal magnitude and the same phase are fed into port 1 and
l
I
S=
~ ~ ~:: ~::
S31
S32
S41
S42
0
0
0
0
(4-4-23)
The magic tee is commonly used for mixing, duplexing, and impedance measurements. Suppose, for example, there are two identical radar transmitters in equipment
stock. A particular application requires twice more input power to an antenna than
either transmitter can deliver. A magic tee may be used to couple the two transmitters to the antenna in such a way that the transmitters do not load each other. The
two transmitters should be connected to ports 3 and 4, respectively, as shown in
Fig. 4-4-8. Transmitter 1, connected to port 3, causes a wave to emanate from port
1 and another to emanate from port 2; these waves are equal in magnitude but opposite in phase. Similarly, transmitter 2, connected to port 4, gives rise to a wave at
port 1 and another at port 2, both equal in magnitude and in phase. At port 1 the
two opposite waves cancel each other. At port 2 the two in-phase waves add together; so double output power at port 2 is obtained for the antenna as shown in Fig.
4-4-8.
To antenna
Transmitter 2
Port I
A hybrid ring consists of an annular line of proper electrical length to sustain standing waves, to which four arms are connected at proper intervals by means of series
or parallel junctions. Figure 4-4-9 shows a hybrid ring with series junctions.
148
Chap.4
Port 4
The hybrid ring has characteristics similar to those of the hybrid tee. When a
wave is fed into port 1, it will not appear at port 3 because the difference of phase
shifts for the waves traveling in the clockwise and counterclockwise directions is
180. Thus the waves are canceled at port 3. For the same reason, the waves fed into
port 2 will not emerge at port 4 and so on.
The S matrix for an ideal hybrid ring can be expressed as
= 521
0
541
512 0 514]
0 523 0
532 0 534
0 543 0
(4-4-24)
It should be noted that the phase cancellation ocurs only at a designated frequency
for an ideal hybrid ring. In actual hybrid rings there are small leakage couplings,
and therefore the zero elements in the matrix of Eq. (4-4-24) are not quite equal to
zero.
L=(2n+1)~8
(4-4-25)
= l.5b
for an E bend
(4-4-26)
Sec. 4.5
Directional Couplers
149
R = I.Sa
for an H bend
(4-4-27)
where a and b are the dimensions of the waveguide bend as illustrated in Fig.
4-4-10.
(b)
(a)
(d)
(c)
Figure 4-4-10 Waveguide corner, bend, and twist. (a) -plane corner.
(b) H-plane corner. (c) Bend. (d) Continuous twist.
Port I -
Primary
Port 3
--
waveguide
Coupling
device
Secondary
waveguide
---
Port 2
-- Port 4
-
Figure 4-5-1
Directional coupler.
150
Chap. 4
respectively, by
where P,
p,
p4
10 log10 -
Directivity (dB)
10 log10 -
p4
p3
(4-5-1)
(4-5-2)
acteristic impedances. The coupling factor is a measure of the ratio of power levels
in the primary and secondary lines. Hence if the coupling factor is known, a fraction
of power measured at port 4 may be used to determine the power input at port 1.
This significance is desirable for microwave power measurements because no disturbance, which may be caused by the power measurements, occurs in the primary
line. The directivity is a measure of how well the forward traveling wave in the primary waveguide couples only to a specific port of the secondary waveguide. An
ideal directional coupler should have infinite directivity. In other words, the power
at port 3 must be zero because port 2 and port 4 are perfectly matched. Actually,
well-designed directional couplers have a directivity of only 30 to 35 dB.
Several types of directional couplers exist, such as a two-hole directional
couler, four-hole directional coupler, reverse-coupling directional coupler
(Schwinger coupler), and Bethe-hole directional coupler (refer to Fig. 4-5-2). Only
the very commonly used two-hole directional coupler is described here.
(b)
(a)
Coupling holes
Rotated
(c)
(d)
Sec. 4.5
Directional Couplers
151
L=(2n+l)~g
(4-5-3)
Primary
Port I
Port 3
waveguide
Canceled..._ __ A
_ _ _ _ _ _ _ _ _ _ _ Added
Port 2
Port 4
Figure 4-5-3
coupler.
waveguide
Secondary
Two-hole directional
A fraction of the wave energy entered into port 1 passes through the holes and
is radiated into the secondary guide as the holes act as slot antennas. The forward
waves in the secondary guide are in the same phase, regardless of the hole space,
and are added at port 4. The backward waves in the secondary guide (waves are progressing from right to left) are out of phase by (2L/ A8 )27T rad and are canceled at
port 3.
452 S Matrix of a Directional Coupler
In a directional coupler all four ports are completely matched. Thus the diagonal elements of the S matrix are zeros and
(4-5-4)
As noted, there is no coupling between port 1 and port 3 and between port 2 and
port 4. Thus
SI3
(4-5-5)
S32 0
0 S43
S34
0
(4-5-6)
Equation (4-5-6) can be further reduced by means of the zero property of the S matrix, so we have
(4-5-7)
152
S21 Sf,
+ S41 St3 = 0
Chap. 4
(4-5-8)
(4-5-9)
=
I S21 II S23 I =
[S32[[S34[
(4-5-10)
I s41 II s43 I
(4-5-11)
1s34I
(4-5-12)
I S23 I
(4-5-13)
S12 = S34 = p
(4-5-14)
[S12[[S14[
=
I S14 I =
IS12I
Let
S41)
(4-5-15)
Let
S23
S41
(4-5-16)
jq
p2 + q2 = 1
(4-5-17)
J~Lq 1
j~
~i
o p o
(4-5-18)
Bolometer 2
Figure 4-5-4
Bolometer I
Sec. 4.5
Directional Couplers
153
reads 8 mW and bolometer 2 reads 2 mW, find: (a) the amount of power dissipated in
the load Ze; (b) the VSWR on arm 2.
Solution
2.0
r------------------'-------.
-1
Port 2
mW~~-------~
---------1 mW--------
8mW
SWR = 2.0
Figure 4.5.5
where p+
Ir\i
then
The incident power to port 4 is Pt = 9 mW, and the reflected power from port
4 is Pi = 1 mW.
3. Since port 3 is matched and the bolometer at port 3 reads 2 mW, then 1 mW
must be radiated through the holes.
4. Since 20 dB is equivalent to a power ratio of 100: 1, the power intput at port 1 is
given by
P1 = lOOpt = 900 mW
and the power reflected from the load is
Pi- = 100
x (1 mW) = 100 mW
154
Chap. 4
1r1= Vr>
fr= !loO=!
Y9oo 3
Then the VSWR on arm 2 is
l+lfl
=I
I+t
Ir I = I - t = 2.0
A signal wave incident in port 1 couples equal power into ports 2 and 4, but
none into port 3. There are two basic types of Lange couplers: 180 hybrids and 90
(quadrature) hybrids. The latter are also called 3-dB directional couplers.
Hybrid couplers are frequently used as components in microwave systems or
subsystems such as attenuators, balanced amplifiers, balanced mixers, modulators,
discriminators, and phase shifters. The hybrid has a directivity of over 27 dB, a return loss of over 25 dB, an insertion loss of less than 0.13 dB, and an imbalance of
less than 0.25 dB over a 40% bandwidth.
In modern microwave circuit design, Lange hybrid couplers are commonly
Sec. 4.5
155
Directional Couplers
used in balanced amplifier circuitry for high-power and broad-bandwidth applications, as shown in Fig. 4-5-7.
Single-stage or cascaded double-stage GaAs MESFEf chips are connected in
parallel to two 3-dB and 90-degree Lange hybrid couplers. Their basic relationship
can be expressed by the following three equations:
S11
= 1(S11a
- S11b)
(4-5-19)
(4-5-20)
and
Gain
= IS21 /2 =
~I S21a +
s21b /2
(4-5-21)
where a and b indicate the two GaAs MESFET chips, and 1 and 2 refer to the input
and output ports, respectively. The VSWRs of the balanced amplifier can be expressed as
for the input port
(4-5-22)
(4-5-23)
and
VSWR
l + IS22 /
l - IS22 /
GaAs MESFET a
Input
11
power 1
12
I
-1
I
I
I
I
I
I
I
I
I
I
I
'iOil
I
I
I
I
1_
son
Lange
coupler
14
Lange
coupler
Output
power
GaAs MESFET b
Figure 4-5-7
Theoretically, if the two GaAs MESFEf chips (or four chips in a double-stage
amplifier circuit) are identical, the amplifier is balanced and its VSWR will be unity.
Practically, however, characteristics of the two GaAs MESFEf chips are not actually measured and they may not be the same. When their characteristics are different, the amplifier will not be balanced and manual tuning will be needed to balance
it. Therefore, for mass production it is necessary to characterize the GaAs MESFET
156
Chap. 4
chips in advance before placing them in the microwave integrated circuit in order to
minimize the tuning work, reduce the production cost, and increase the hybrid reproducibility.
Example 4-5-2: Operation of a Balanced Amplifier
A GaAs MESFET balanced amplifier with two Lange couplers has the following
parameters:
S11a
S22.
S11b
S22b
P.n = 200 mW
Gain= 10 dB
S parameters:
Input signal power:
Power gain of each GaAs chip:
=
=
Determine: (a) the input and output VSWRs; (b) the output power in watts; (c) the linear output power gain in dB.
Solution
a. From Eqs. (4-5-22) and (4-5-23), the input and output VSWRs are unity.
b. The output power is
P u, = 200 x 10 x 2 = 4000 mW = 4 W
0
c. Because two GaAs chips are in parallel, the linear output power gain is
Gain = 10 log (2) = 3 dB
JlH
(4-6-1)
where
[X
xm = J~
jK
Xm
0
~]
(4-6-2)
(4-6-3)
Sec. 4.6
157
which is the tensor magnetic susceptibility. Here x is the diagonal susceptibility and
K is the off-diagonal susceptibility.
When a de magnetic field is applied to a ferrite, the unpaired electrons in the
ferrite material tend to line up with the de field because of their magnetic dipole moment. However, the nonreciprocal precession of unpaired electrons in the ferrite
causes their relative permeabilities (,:, ,;) to be unequal and the wave in the ferrite is then circularly polarized. The propagation constant for a linearly polarized
wave inside the ferrite can be expressed as [4]
'}'
=JwYE,o(,
+ K)
(4-6-4)
where
= 1 + xm
,: = , + K
,; = , - K
(4-6-5)
(4-6-6)
(4-6-7)
The relative permeability ,r changes with the applied de magnetic field as given by
,
%Me
i Yel Hdc + W
(4-6-8)
I~
I
I
I
I
+
r
Hctc
158
Chap. 4
the clockwise direction. Consequently, the propagation phase constant 13+ for the
forward direction differs from the propagation phase constant W for the backward
direction. By choosing the length of the ferrite slab and the de magnetic field so that
w
= (J3+ -
W)C
= -1T
(4-6-9)
a differential phase shift of 90 for the two directions of propagation can be obtained.
Figure 4-6-2
Port 2
Coupler I
r- -
Primary guide
Coupler 2
r--...,
__,
I
I
I
I
:I
II
Phase shift er
1
1_1
1W1 = 180 W2 = 901
I
I
180
Port I -----Tlc~-----~~-----~2
I
I 190~
II '9o~I80o--------;'~t
I
II I -I- - - - - - - - 1 I
i
'II L.....------....Li__J
I
iW3 = 0 W4 = 90 I I 270
Port 3
1
!I
I
Secondary guide
1~
lI
~--J
Figure 4-6-3
~ 1L--+-----Port4
1
Phase shifter
I
90 0
L--------~----
L __ J
Sec. 4.6
159
W3
W2 -
W4
= 2n1T
(2m
+ l}rr
rad/s
rad/s
(4-6-10)
(4-6-11)
where m and n are any integers, including zeros. A similar analysis shows that a
wave incident to port 2 emerges at port 3 and so on. As a result, the sequence of
power flow is designated as 1 ~ 2 ~ 3 ~ 4 ~ 1.
Many types of microwave circulators are in use today. However, their principles of operation remain the same. Figure 4-6-4 shows a four-port circulator constructed of two magic tees and a phase shifter. The phase shifter produces a phase
shift of 180. The explanation of how this circulator works is left as an exercise for
the reader.
Phase shifter
Figure 4-6-4
A four-port circulator.
0
S21
S31
S41
S12
0
S32
S42
S13
S23
0
S43
S141
S24
S34
0
(4-6-12)
160
Chap. 4
S=
[~ ~1
0 0
0 0
1 0
0
(4-6-13)
Direction
of rotation
Resistive
vane
E-
z.r,;.
Input
waveguide
-......./
----
Reflected
wave
vector
Figure 4-6-5
Ferrite rod
Faraday-rotation isolator.
Problems
161
end without attenuation at all. On the contrary, a reflected wave from the output end
is similarly rotated clockwise 45 by the ferrite rod. However, since the reflected
wave is parallel to the input resistive card, the wave is thereby absorbed by the input
card. The typical performance of these isolators is about 1-dB insertion loss in forward transmission and about 20- to 30-dB isolation in reverse attenuation.
REFERENCES
[l] SAAD, T., and R. C. HANSEN, Microwave Engineer's Handbook, Vol. 1. Artech House,
Dedham, Mass. , 1971.
[2] SOUTHWORTH, G. E., Principles and Applications of Waveguide Transmission. Chapters 8
and 9. D. Van Nostrand Company, Princeton, N.J., 1950.
[3] LANGE, J. Interdigited stripline quadrature hybrid. IEEE Trans. on Microwave Theory and
Techniques, MTT-17, No. 12, 1150-1151, December 1969.
[4] SOOHOO, R. F., Theory and Applications of Ferrites. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1960.
[5] BowNESS, C., Microwave ferrites and their applications. Microwave J., 1, 13-21, JulyAugust 1958.
SUGGESTED READINGS
LIAO, S. Y., Engineering Applications of Electromagnetic Theory. West Publishing Company, St. Paul, Minn., 1988.
RIZZI, P. A., Microwave Engineering: Passive Circuits. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1988.
PROBLEMS
Rectangular waveguides
162
Chap. 4
~) cos ( ~)
Alm
The phase constant is f3 = 1.0475 rad/cm, the quantities x and y are expressed in centimeters, and a = b =
are also in centimeters. Determine the cutoff frequency fc,
phase velocity v8 , guided wavelength A8 , and the magnetic field intensity in they direction.
4-5. A rectangular waveguide is designed to propagate the dominant mode TE 10 at a frequency of 5 GHz. The cutoff frequency is 0.8 of the signal frequency. The ratio of the
guide height to width is 2. The time-average power flowing through the guide is 1 kW.
Determine the magnitudes of electric and magnetic intensities in the guide and indicate
where these occur in the guide.
v'6
4-7. An impedance of (0.5 - j0.4)Zo is connected to a rectangular waveguide. A capacitive window with a susceptance jB = j0.4Yo is located at a distance of 0.2A from the
load.
= 25 cm. A
dominant mode TE 10 is propagated in the guide, and its VSWR is measured as 2.8 at a
frequency of 8 GHz. The adjacent voltage minima are located at z = 9.46 cm and
z = 12.73 cm.
a. Determine the value of the load impedance in terms of Zo.
b. Find the position closest to the load where an inductive window is placed in order to
obtain a VSWR of unity.
c. Determine the value of the window admittance.
4-11. A rectangular waveguide is filled by dielectric material of E, = 9 and has inside dimensions of 7 x 3.5 cm. It operates in the dominant TE10 mode.
a. Determine the cutoff frequency.
b. Find the phase velocity in the guide at a frequency of 2 GHz.
c. Find the guided wavelength A8 at the same frequency.
4-12. The electric field intensity of the dominant TE1o mode in a lossless rectangular waveguide is
E.v = Eo sm
(7TX)
-13 Z
~ e
1
for f > fc
Problems
163
Circular waveguides
Microwave cavities
164
Chap. 4
VSWR
Directional
coupler
Cavity
P4-24
Hybrid circuits
--~/
Cavity
15 dB
z,
.,.
-:-
P4-25
4-26. A symmetric directional coupler has an infinite directivity and a forward attenuation of
20 dB. The coupler is used to monitor the power delivered to a load Ze as shown in
Fig. P4-26. Bolometer 1 introduces a VSWR of 2.0 on arm I; bolometer 2 is matched
to arm 2. If bolometer 1 reads 9 mW and bolometer 2 reads 3 mW:
a. Find the amount of power dissipated in the load Ze .
b. Determine the VSWR on arm 3.
Problems
165
VSWR = 2.0
Bolometer 2
2:
Bolometer 1
:I
P4-26
[ziJ]
[~
!]
Determine the scattering matrix by using S -parameter theory and indicate the values of
the components:
4-29. A hybrid waveguide is constructed of two identical rectangular waveguides across each
other at the center and works as a four-port device. Write a general scattering matrix
and then simplify it as much as possible by inspection of geometric symmetry and by
use of the known phases of the electric waves.
4-30. A helical slow-wave structure has a pitch P of 2 mm and a diameter of 4 cm. Calculate
the wave velocity in the axial direction of the helix.
4-31. Two 3-dB quadrature Lange couplers are used in a GaAs MESFEf balanced amplifier
circuit with the following parameters:
MESFEf a: Reflection coefficients
S11a
0. 7488/-158.3
Sna = 0.8521/-155.7
Forward transmission
coefficient
MESFEf b: Reflection coefficients
Forward transmission
coefficient
Compute:
a.
b.
c.
d.
The
The
The
The
S21a
1.3500/-8.5
S11b
0.6210/-175.9
s22b
o.1121;-151.4
S21b
1.2200/-19.1
Chapter 5
Microwave Transistors
and Tunnel Diodes
50 INTRODUCTION
Microwave solid-state devices are becoming increasingly important at microwave
frequencies. These devices can be broken down into four groups. In the first group
are the microwave bipolar junction transistor (BJT), the heterojunction bipolar transistor (HBT), and the tunnel diodes. This group is discussed in this chapter. The
second group includes microwave field-effect transistors (FETs) such as the junction field-effect transistors (JFETs), metal-semiconductor field-effect transistors
(MESFETs), high electron mobility transistors (HEMTs), metal-oxide-semiconductor field-effect transistors (MOSFETs), the metal-oxide-semiconductor transistors
and memory devices, and the charge-coupled devices (CCDs). This group is described in Chapter 6. The third group, which is characterized by the bulk effect of
the semiconductor, is called the transferred electron device (TED). These devices
include the Gunn diode, limited space-charge-accumulation diode (LSA diode), indium phosphide diode (InP diode), and cadmium telluride diode (CdTe diode). This
group is analyzed in Chapter 7. The devices of the fourth group, which are operated
by the avalanche effect of the semiconductor, are referred to as avalanche diodes:
the impact ionization avalanche transit-time diodes (IMPATT diodes), the trapped
plasma avalanche triggered transit-time diodes (TRAPATT diodes), and the barrier
injected transit-time diodes (BARITT diodes). The avalanche diodes are studied in
Chapter 8. All those microwave solid-state devices are tabulated in Table 5-0-1.
In studying microwave solid-state devices, the electrical behavior of solids is
the first item to be investigated. In this section it will be seen that the transport of
charge through a semiconductor depends not only on the properties of the electron
but also on the arrangement of atoms in the solids. Semiconductors are a group of
substances having electrical conductivities that are intermediate between metals and
insulators. Since the conductivity of the semiconductors can be varied over wide
166
Sec. 5.0
167
Introduction
TABLE 501
Microwave transistor
Field-effect
transistors
Microwave
solid-state
devices
Transferred electron
devices
-f
Microwave BJT
HBT
Tunnel diode
JFEf
MESFEf, HEMT
MOSFEf
NMOS, PMOS, CMOS
Memories
CCD
Gunn diode
LSA diodes
InP diodes
CdTe diodes
f
~
Read
diode
IMPATT diode
TRAPATT diode
BARITT diode
Semiconductor
Ge
Si
Al Sb
GaSb
GaAs
GaP
In Sb
InAs
InP
CdS
CdSe
ZnO
ZnS
Mobility at 300K
(cm 2 /V s)
0K
300K
Holes
Electrons
5.51
0.89
1.16
1.75
0.80
1.52
2.40
0.26
0.46
1.34
2.56
1.85
5.47
0.803
1.12
1.63
0.67
1.43
2.26
1.80
0.33
1.29
2.42
1.70
3.20
3.60
1600
1900
450
420
1400
400
75
750
460
150
50
1800
3900
1600
200
4000
8500
110
78,000
33,000
4600
300
800
200
165
3.70
Relative
dielectric
constant
5.5
16
11.8
11
15
13. l
IO
17
14.5
14
IO
IO
9
8
168
Chap. 5
the valence band is designated by Ev . The separation between the energy of the lowest conduction band and that of the highest valence band is called the bandgap energy E 8 , which is the most important parameter in semiconductors.
Electron energy is conventionally defined as positive when measured upward
whereas the hole energy is positive when measured downward. A simplified band diagram is shown in Fig. 5-0-1.
;..,
.,~
i:::
~~~""'""~~""'""""""'~""'""~~~~
0
j~
Figure 5-0-1
Energy-band diagram.
In the 1970s, it seemed that microwave transistors would be useful for generating power up to about 5 GHz. Since their inception, avalanche diodes have produced
in excess of 4 W continuous wave (CW) at 5 GHz. Gunn diodes had been considered
only for local oscillators or low-power transmitter applications, but recent results indicate that a single Gunn diode can generate an output power of 1 W at X band. At
higher microwave frequencies, and even well into the millimeter range, limited
space-charge-accumulation diodes (LSAS) can provide the highest peak power of
any solid-state device, up to 250W in C band, 100 Win X band, and 50 Win Ku
band. Since the pulsed Gunn and TRAPATT diodes are essentially transit-time
devices, their operating frequency is approximately determined by the thickness of
the active layer in the diode. An operating frequency of 10 GHz requires an active
layer thickness on the order of 10 m (microns). Thus only a limited voltage can be
applied to such a thin layer because of breakdown limitations. Consequently, the
~
ro
IOO
...
"'0~
0.
ro
""
c..
IO
8 IO
Frequency (GHz)
20
Sec. 5.1
169
peak power capability of both the pulsed Gunn diodes and the TRAPATT diodes is
greatly limited at higher frequencies. On the other hand, the peak power capability
of an LSA diode is approximately proportional to the square of the thickness of the
active layer because its operating frequency is independent of the thickness of the
active layer. Thus the LSA diode is capable of producing higher peak power than either the pulsed Gunn diodes or the TRAPATT diodes. Figure 5-0-2 shows peak
power versus frequency for these three devices.
Solid-state microwave power sources are widely used in radar, communications, navigational and industrial electronics, and medical and biological equipment.
Representative applications for microwave solid-state devices are listed in Table
5-0-3.
TABLE 503 APPLICATIONS OF MICROWAVE SOLID-STATE DEVICES
Devices
Transistor
TED
IMPA TT
TRAPATT
BARITT
Applications
L-band transmitters for telemetry
systems and phased array radar
systems
L- and S-band transmitters for
communications systems
C-, X-, and Ku-band ECM amplifiers
for wideband systems
X- and Ku-band transmitters for radar
systems, such as traffic control
Transmitters for millimeter-wave
communications systems
S-band pulsed transmitters for phased
array radar systems
Local oscillators in communications
and radar receivers
Advantages
Low cost, low power supply, reliable,
high CW power output, light weight
170
Chap. 5
All microwave transistors are now planar in form and almost all are of the silicon
n-p-n type. The geometry can be characterized as follows: (a) interdigitated, (b)
overlay, and (c) matrix (also called mesh or emitter grid) as shown in Fig. 5-1-1.
The interdigitated type is for a small signal and power, but the overlay type and
matrix type are for small power only. The figure of merit for the three surface
geometries shown in Fig. 5-1-1 is listed in Table 5-1-1 [1).
Emitter
metalization
.
E m1tie rs
c:::::J
c:::::J
Base
metalization
c:::::J
(a) Interdigitated
(b) Overlay
-
Emitter metalization
Base metalization
(c) Matrix
Figure 5-1-1 Surface geometries of microwave power transistor. (From H. Sobol
and F. Sterzer [1 ]: reprinted by permission of IEEE. Inc.)
Sec. 5.1
171
TABLE 511
M=-
BA
- - - - -l-~ Overlay
:--+p--i
1:f:I
I
CJ
_l _ -
2( + w)
(w+s)(f+p)
~-f~E.."?.::!._i-
lw I
c:J
Oii 0 \D
Emitter
w +s
'""ro;gitted
2( + w)
f(w + s)
I4
w +s
2( + w)
(w+s)(f+p)
Source: From H. Sobol and F. Sterzer [I]; reprinted by permission of IEEE, Inc.
Base
Collcdor
.\'
!,,
'"
x 10 11
x 10 9
l'p = 3.7
c
11
11"=1.7XI0
x 10 15
11,,=1.7XI0 14
,,,, = 3.7 x 10 12
Figure 5-1-2
172
Chap. 5
ity of the transistor increases. Additional fingers, however, increase the device parasitics and degrade the noise and upper frequency capability of the devices.
Figure 5-1-3 shows two schematic diagrams for a bipolar junction transistor
(BJT): (a) the cross section of a discrete n-p-n planar BJT and (b) the cross section
of a chip-type n-p-n integrated BJT.
The p-n-p bipolar junction transistor is a complementary structure of the n-p-n
BJT by interchanging p for n and n for p. The p-n-p BJT is basically fabricated by
first forming an n-type layer in the p-type substrate; then ap+-type region is <levelInsulator layer
Si0 2
Emitter contact
Base-collector junction
n-emitter layer
Collector contact
n-collector layer
(a) Discrete n-p-n planar BJT
Emitter contact
Insulator layer
Si0 2
Collector contact
11
11-p-11
BJT
Sec. 5.1
173
oped in the n layer. Finally, metallic contacts are introduced to the p+ region and
p layer through the windows opened in the oxide layer and to the p region at the
bottom.
5 12 Bipolar Transistor Configurations
In general, there are two types of bipolar transistors: p-n-p and n-p-n. In practical
applications, a transistor can be connected as three different configurations: common
base (CB), common emitter (CE), and common collector (CC), depending on the
polarities of the bias voltages connected to its terminals.
Common-base configuration. The common-base (CB) con.figuration refers
to the one where the emitter (input circuit) and collector (output circuit) terminals
are common to the base as shown in Fig. 5-1-4.
11-11-11
11-1J-11 transi~tor
transistor
(a) CB configuration
Figure 5-1-4
( b I CB configuration
The CB configuration is also called the grounded-base con.figuration. For a pn-p transistor, the largest current components are caused by holes. Holes flow from
the emitter to the collector and down toward ground out of the base terminal. In an
n-p-n transistor all current and voltage polarities are negative to those in a p-n-p
transistor. The CB configuration of a transistor is usually used in amplifier applications. Its input voltage VEB and output current le can be expressed in terms of the output voltage VcB and input current h as
VEB
le
(5-1-1)
(5-1-2)
174
Chap. 5
(5-1-3)
(5-1-4)
Figure 5-1-5
Sec. 5.1
175
/i
JI
/!
JI
Bo------<
Figure 5-1-6
h;e
= OVbe
oib I
(5-1-7)
Vce=constant
h
re
OVbe
oib .
(5~1-8)
'b
fe
= a
oic I
(5-1-9)
lb Vee
hoe
=~1
OVce
(5-1-10)
.
'b
176
Chap. 5
ic
~~---~----<>C
b'
cc
ic
B 0---'l!'\Nv---<t---O---f t - - - < i - - - . . . - - - - - - 0 C
Eo--------<~-,.__
B~
ib
R;
ic
b'
Vb'e
C;
0------~----...._
co
Ro
_ _.__ _.__--0
Figure 5-1-7
model.
Hybrid-pi equivalent
When the dimensions of a bipolar junction transistor become very small, their Z, Y,
or H parameters cannot be measured because the input and output terminals cannot
be openly and shortly realized. Therefore, the S parameters are commonly measured. In transistor design, it is necessary to convert the S parameters into Y
parameters for the network component computations.
An incremental change of the emitter voltage .1"Vb'e at the input terminal will
induce an incremental change of the collector current '1ic at the output terminal.
Then the mutual conductance (or transconductance) of a small-signal transistor is
defined by
(5-1-11)
From the diode junction theory the thermal equilibrium density at the junction is
equal to the minority density times the forward-bias voltage factor. That is,
(5-1-12)
Sec. 5.1
177
and
. / = qADnnp(O)
Ile
L
n
(5-1-13)
Substitution of Eq. (5-1-12) into Eq. (5-1-13) and differentiation of the resultant
yield
_ Ii, I
gm where Vr
(5-1-14)
Vr
dQb
= dVbe
Cb'e
(5-1-15)
Qb
= qnp(O)WbA
(5-1-16)
2
Then the diffusion capacitance is expressed by
Cb'e
qAWbnp(O)
2Vr
W~
= gm 2Dn
(5-1-17)
ib
Vbe
(5-1-18)
= jwC
The small-signal input conductance of the emitter junction looking at the input of
the base is defined as
gb
where
In
Vr
gm
le
hFE Vr
= - = - = -- = Rb
(5-1-19)
hFE
Example 5-1-1:
fc =
hFE
6 mA
120
T = 300 K
= 10-s cm 2
wb
178
Chap. 5
Compute: (a) the mutual conductance gm; (b) the input conductance gb and resistance
R;; (c) the electron diffusion coefficient D.; and (d) the diffusion capacitance Cf,,.
Solution
mho
gm
0.23
gb = - = - - = 1. 92
hFE
120
R; = 521
x 10-
mho
ohms
KT
x 26 x 10- 3
= 41.6 cm 2/s
'
Cbe =
Wi
gm2D. =
0 23
.
10-s
x 2 x 41.6
= 2.76pF
51 3 Principles of Operation
The bipolar junction transistor (BJT) is an active three-terminal device which is
commonly used as an amplifier or switch. Its principles of operation are discussed in
this section.
Modes of operation. A bipolar transistor can operate in four different
modes depending on the voltage polarities across the two junctions: normal (active)
mode, saturation mode, cutoff mode, and inverse (or inverted) mode as shown in
Fig. 5-1-8.
1. Normal Mode. If the emitter junction of an n-p-n transistor is forward-biased
Sec. 5.1
179
II
II
( d) Inverse mode
(c I Cutoff mode
Figure 5-1-8
open circuit. Both the cutoff and saturation modes of a transistor are used as
switching devices for the OFF and ON states. Fig. 5- I-8(c) shows the cutoffmode bias-voltage connection.
4. Inverse Mode. When the emitter is reverse-biased and the collector is
forward-biased, the transistor is operated in the inverse (or inverted) mode,
and its current gain is designated as the inverse alpha a1 . If the transistor is
symmetric, the normal alpha aN is nearly equal to the inverse alpha a1 . The
two current gains, however, are not actually equal because of their unequal
dopings. The inverse mode is shown in Fig. 5-1-8(d). In practice, the inverse
mode is not commonly used except as a multiemitter transistor in TTL
(transistor-transistor logic) logic gate.
Current flow in normal mode. When a transistor is properly biased, the
holes and electrons in the transistor will follow the field direction in motion. Figure
5-1-9 shows the current flow of an n-p-n transistor.
The current flow in an ideal n-p-n bipolar junction transistor is analyzed under
the following assumptions:
1.
2.
3.
4.
The
The
The
The
(5- 1- 20)
180
--+-
Hole flux
1,~.
,,
I
I
))
I
~----_._ _ _ _ I___ I
~--L----r---
IL
!Ill:
: '\
I
I
_____ ____.._
______
/pt
nt
I
I
I
!,,
-xlo
I
I
I
I
-----1
I
I
le
C
-lea
Xe
L
-
1+
I
Figure 5-1-9
1,,e
II
/ ue
B <)
-
t I ./--i
____._
Chap. 5
1+
I
For a common-base n-p-n transistor, the emitter junction is forward-biased and the
collector junction is reverse-biased as shown in Fig. 5-1-9. Consequently, the emitter current h consists of electron current InE crossing from the emitter into the base
and the hole current lpE crossing from the base into the emitter. Since the doping of
the emitter is much larger than the doping of the base, the hole current is negligible.
However, not all the electrons crossing the emitter junction le reach the collector
junction le because some of them combine with the holes in the p-type base. If Inc
is the electron current at the collector junction le, there must be a recombination
current InE - Inc leaving the base. When the emitter is open-circuited, then h = 0
and Inc = 0. As a result, the collector current le is equal to the reverse saturation
current fc because the junction between base and collector is reverse-biased. From
Fig. 5-1-9, we have
0
(5-1-21)
(5-1-22)
and
(5-1-23)
For an n-p-n transistor, lea consists of holes moving across the collector junction le
from right to left (collector to base) and electrons crossing le in the opposite direction. Since the reference direction for lea in Fig. 5-1-9 is assumed from left to right,
then, for an n-p-n transistor, lea is positive for forward-biased le junction and lea is
negative for reverse-biased le junction. The saturation current lea at the le junction
of an n-p-n transistor is given by
2(
- AqDnpo
AqDppno - A
Dn
Dp )
W
+ LE - qn; WNa + LENd
lea -
(5-1-24)
Sec. 5.1
181
Also from Fig. 5-1-9, the sum of the three terminal currents should be zero and it is
le+ le+ IB = 0
(5-1-25)
(5-l-25a)
In= AqDn dx
and
(5-1-26)
where Dn
np
npo
= electron lifetime
11,,101
,,
II
~
II
;, I
-
"
S
'&
!',, 101
I
I
::: I
I
()
~
hnittcr
()
Ba'c
w
Collector
--------
Figure 5-1-10 Minority-carrier densities under normal active bias with negligible recombination.
182
Chap. 5
(5-1-29)
np x
[(W - x)/Ln]
sinh (W / Ln)
( vEfvT _ l)[sinh
npo e
npo
(l _
sinh (x/Ln) )]
sinh (W / Ln)
(5-1-30)
In almost all transistors, the base width is made very narrow (W ~ Ln) so that
the minority-carrier recombination in the base is negligible. As a result, the
boundary conditions specify the two end points of the base carrier concentration
with a straight line as shown in Fig. 5-1-11. (Note: sinh y
y, cosh y
1,
coth y
l/y, and sech y
1 - y 2/2 for y ~ 1.)
0.8
s,,"-
0.6
J-
0.4
0.2
W
Distance _,.
For W
plified to
(5-1-31)
Then the boundary conditions for holes in the emitter and collector depletion regions
can be expressed, respectively, as
at x = -xE
(5-1-32)
Sec. 5.1
183
and
(5-1-33)
atx =Xe
where PED
Substituting Eqs. (5-1-32) and (5-1-33) into Eq. (5-1-27) yields the minority distributions in the emitter and collector regions as
pE(x)
for x :::::; - XE
(5-1-34)
for x ::::: Xe
(5-1-35)
and
pc(x)
Q8
= Aq
AqWnp(O)
2
[np(x) - npo(x)] dx =
(5-1-36)
Tn
is
AQn
AqWnpo v ;v
In 8 = - - =
e E r
Tn
(5-1-37)
2Tn
npc
=0
at le for x
(5-1-38)
This assumption is reasonable because the electric field of the collector junction
sweeps carriers into the collector so that the collector is almost a perfect sink.
The electron current InE which is injected from the emitter into the base at
x = 0 for L ~ W is proportional to the gradient of the minority carrier density and
is expressed as
InE
= AqDn dnp I
dx
FD
v ;v
I) - AqDnnT _,__ - AqDnnT eET
v ;v
-_ -AqDnnT (et.TNaW
NaW
NaW
1
cosh (W!Ln)
(5-1-39)
(5-1-40)
The collector current Inc can also be expressed as
Chap. 5
184
Inc
where LE
~:)
(5-1-41)
= AqDEPEo (
LE
pE
vEfvT _ l)
for x
-xE
(5-1-42)
and
I
_ AqDcpco
Le
pe -
for
Xe
(5-1-43)
where DE and De are the hole diffusion constants in the emitter and collector,
respectively
Le = diffusion length of the collector
The current flow in an n-p-n transistor as described so far is an ideal model,
and its recombination-generation current is not counted. If the recombination current is considered, the current flow is the sum of the drift, the diffusion, and the
recombination-generation currents. That is,
I = Idr (drift)
Idf (diffusion)
lr8 (recombination-generation)
(5-1-44)
where xd
To
= Aqn;xdevEf<2vTJ
TO
(5-1-45)
depletion-layer width
Sec. 5.1
185
Saturation
region -~-----Active region
50
Ir= -40 mA
40
<!'.
.....'-'
-30
30
"'
I::
""
'::;
.-"
-20
20
0
v
-10
10
Cutoff region
-0.25
lea
Figure 5-1-12
3. Cutoff Region: In this region the emitter and collector junctions are both
reverse-biased. Consequently, the emitter current is cut off to zero, as shown
in the lower right side of Fig. 5-1-12.
Example 5-1-2:
w = 10-s cm
LE= 10- 4 cm
Le = 5 X 10- 4 cm
Pn =
PE =
Pc =
VE=
Ve=
A =
0.15 !1-cm
0.006 !1-cm
16 !1-cm
0.5 v
0.6 V
2 x 10- 2 cm 2
Find:
a. The impurity densities in the emitter, base, and collector regions
b. The mobilities in the emitter, base, and collector regions
c. The diffusion lengths in the emitter, base, and collector regions
186
Chap. 5
b. The mobilities
,pE = 80
1-tnE = 105
,pB = 400
1-lnc = 1600
=
=
=
=
,pEVT
,nEVT
,pBVT
JLneVT
=
=
=
=
80 x 26 x 10- 3 = 2.08
105 x 26 x 10- 3 = 2. 73
400 x 26 x 10- 3 = 10.4
1600 X 26 x 10- 3 = 41.6
cm 2/s
cm 2/s
cm 2/s
cm 2/s
= - ---
-0.2946 mA
14.976
0.337 JLA
10- 16
2.248
108
Sec. 5.1
187
-13.104
-1.312 pA
10-n - 14.976
10- 16
From Eq. (5-1-40), the electron current which reaches the collector is
Inc = - AqDnnl
NaW e V/V
E
T =
-1 3 .104 x 10- I'' x 2.24 8 x l os
=
-0.2946 mA
-lpE
lnE
-33.67
10-s - 0.295
10- 3
-0.295 mA
The collector current is
le= -fco - Inc= l.312 X 10- 12
=
(-0.295
10- 3 )
0.295 mA
33.67
0.337 ,A
fco
10-s - (-29.46
10- 5 + 29.46
10- 5 ) + l.312
10- 12
51 4 Amplification Phenomena
Bipolar transistors are usually used for signal amplification. The amplification phenomena can be described from the common-base and common-emitter transistors.
Common-base n-p-n transistor. The ratio of the output current to the input current for a small signal in a bipolar junction transistor is known as the current
gain alpha a, or h1b The current gain of a common-base p-n-p transistor is defined
by the current components crossing the emitter and collector junctions as
a=
le+ lea_ h
fb
(5-1-46)
where lea = collector-junction reverse saturation current with zero emitter current
Since le and le have opposite signs, the alpha a, as defined, as always positive. Typical numerical values of a are between 0.9 and 0.995.
188
Chap. 5
where lnE
/pE
=
=
(5-1-47)
/3 * - --"-----------W2
- 2TnDn
Inc
lnE
(5-1-48)
W2
2L~
where Inc
lnE
(5-1-49)
-!co - Inc
a= -
= {3*y
(5-1-50)
(5-1-51)
The current fco is the current crossing the p-n junction, and it is expressed in Eq.
(5-1-24). Then the complete expression of le for any Ve and his
le= -ah+ fco(l -
where Io
=-
Example 5-1-3:
(5-1-52)
evcfVT)
I co is replaced.
Silicon Bipolar Transistor
A silicon n-p-n bipolar transistor operates in common-base mode at 300 K and has the
following parameters:
Silicon intrinsic density:
Acceptor density in base region:
Donor density in emitter region:
Hole lifetime:
1.5 x 10 10
Na= 5 x 10 16
Nd= 5 x 10 18
'Tp = I
n,
cm- 3
cm - 3
cm - 3
Sec. 5.1
189
Electron lifetime:
Cross section:
Base width:
Emitter length:
.s
cm 2
cm
cm
10- 4
w = 10- 3
LE = 10- 2
A
Determine:
a.
b.
c.
d.
e.
The
The
The
The
The
Solution
+ 5.20 x 5 x 10 18 x 10- 2
0 997
f3 * = I
oo-3)2
0 904
2 x 10- 6 x 5.20 =
Common-emitter n-p-n transistor. In the active region of a commonemitter n-p-n transistor, the emitter junction is forward-biased and the collector
junction is reverse-biased. The base current is
IB
-(le+ h)
(5-1-53)
1-a
= 0, then h = le
1-a
(5-1-54)
-Ico
= -h = - = lcEo
1- a
(5-1-55)
190
Chap. 5
The actual collector current with the collector junction reverse-biased and the base
junction open-circuited is designated by the symbol ICEo. We define the commonemitter current gain f3 or h1e as
{3
Since h
h1e
= le
:Blea
(5-1-56)
+ h + le + lea= 0, we have
f3 =
a
1- a
(5-1-57)
'The question then arises as to whether microwave power transistors have any limitations on their frequency and output power. The answer is yes. Several authors have
discussed this subject. Early [4] first introduced the power-frequency limitations inherent in (1) the limiting velocity of carriers in semiconductors and (2) the maximum fields attainable in semiconductors without the onset of avalanche multiplication. These basic ideas were later developed and discussed in detail by Johnson [5],
v,
m
wherefr = - -
27TT
r=-
Vm
Vs
Em
=
=
=
EmLmin
Voltage-Frequency Limitation:
fr=
EmVs =
27T
(5-1-58}
191
Sec. 5.1
be reduced even further by decreasing the distance L. The lower limit on L can be
reached when the electric field becomes equal to the dielectric breakdown field.
However, the present state of the art of microwave transistor fabrication limits the
emitter-collector length L to about 25 m for overlay and matrix devices and to
nearly 250 m for interdigitated devices. Consequently, there is an upper limit on
cutoff frequency. In practice, the attainable cutoff frequency is considerably less
than the maximum possible frequency indicated by Eq. (5-1-58) because the saturated velocity Vs and the electric field intensity will not be uniform.
Second equation:
Current-Frequency Limitation:
(fmXc ) fT
where Im
EmVs
= 27T
(5-1-59)
Power-Frequency Limitation:
f'_
(p mX c)l/2JT
= Em7TVs
2
(5-1-60)
This equation was obtained by multiplying Eq. (5-1-58) by Eq. (5-1-59) and replacing Vmlm by Pm. It is significant that, for a given device impedance, the power
capacity of a device must be decreased as the device cutoff frequency is increased.
For a given product of Emvs (that is, a given material), the maximum power that can
be delivered to the carriers traversing the transistor is infinite if the cross section of
the transistor can be made as large as possible. In other words, the value of the reactance Xe must approach zero. Thus Eq. (5-1-60) allows the results to be predicted.
Figure 5-1-13 shows a graph of Eq. (5-1-60) and the experimental results reported
from manufacturers [6].
Fourth equation:
where Gm
= EmVs
7T
2
(5-1-61)
192
10
0.01
0.1
10
fr in GHz
T
e
Chap. 5
The maximum available power gain of a transistor was derived by Johnson [5]
as
Gm
= (~)
Zout
Zin
(5-1-62)
where Zout and Zin are the output and input impedances, respectively. If the electrode
series resistances are assumed to be zero, the ratio of the output impedance to the
input impedance can be written
(5-1-63)
where Cn is the input capacitance and Cout is the output (base-collector) capacitance.
When the maximum total carrier charges Qm move to the collector in a carrier-base
transit time Tb and with a thermal voltage \.'th, the input capacitance Cin and the emitter diffusion capacitance Cd are related by
(5-1-64)
The output capacitance is given by
= lmTo
C
out
Vm
(5-1-65)
Substitution of Eqs. (5-1-58), (5-1-64), and (5-1-65) in Eq. (5-1-62) yields Eq.
(5-1-61). The actual performance of a microwave transistor will fall far short of that
predicted by Eq. (5-1-61). At present the high-frequency limit of a 28-V silicon
n-p-n transistor operating at the 1-W level is approximately 10 Ghz. Typical power
gains of microwave transistors lie in the 6- to 10-dB range.
Sec. 5.2
Example 5-1-4:
193
Power-Frequency Limitation
Xe= I fl
fr= 4 GHz
Em = 1.6 X 105 V/cm
v, = 4 X 105 cm/s
Reactance:
Transit-time cutoff frequency:
Maximum electric field:
Saturation drift velocity:
Determine the maximum allowable power that the transistor can carry.
Solution.
_
m -
1
Xcf}
(EmVs)
21T
2
_
- 1
1
(4
x 10
9) 2
(1.6
10
21T
5 2
)
10
= 6 .48 W
1.
2.
3.
4.
Fig. 5-2-1 shows the model diagram of a heterojunction transistor formed by n -Ge,
p-GaAs, and n-GaAs materials.
194
11-Ge
p-GaAs 11-GaAs
Chap. 5
c
Figure 5-2-1 Model diagram of a heterojunction transistor.
(5-2-2)
x = electron affinity in eV
Eg = bandgap energy in eV
Vacuum level
Example 5-2-1:
= 5.646 A
a2 = 5.653 A
eV
X1 = 4.0
xz = 4.07 eV
Eg1 = 0.80 eV
Eg2 = 1.43 eV
a1
Sec. 5.2
195
Determine:
a. The lattice match in percent
b. The conduction-band differential between Ge and GaAs
c. The valence-band differential between Ge and GaAs
Solution
Di.Ee
Xi - X2
= 4.0
- 4.07
-0.07
eV
Ev
0.70 eV
When the two semiconductor materials are jointed together, their Fermi energy levels are aligned and their energy bands are depleted at the junction, as shown in Fig.
5-2-3 [7].
I
I
-.--~v----Ec2
1/102
I
l_
_I
1 :
l
p-GaAs
ti.EC I
I
1/101Ll
Ec1
---------~----------4
n-Ge
A-~Ev2
Evi-------'f'I
~ 1ti.Ev I
I I~
I Xi I
Figure 5-2-3
X2
I
I
When the two materials are jointed together, the electrons in the n-Ge are
injected into the p-GaAs, and the holes in the p -GaAs are transferred to the n-Ge
until their Fermi energy levels are aligned. As a result, the energy bands at the junction are depleted or bent. The bending energy then creates a built-in voltage in both
sides of the junction. The total built-in voltage is expressed by
t/102
(5-2-3)
196
Chap. 5
E0 ErJ\gl
EoE'r2\g2
(5-2-4)
= E1\g1 = E2\g2
where e = permittivity of free space
E = permittivity
e, = relative permittivity or dielectric constant
\g = electric field
0
The space charges on both sides of the junction are equal and it is given by
(5-2-5)
where x1 = depletion width inn-Ge
x1 = depletion width in p-GaAs
Nd = donor density
Na = acceptor density
The electric fields in both sides can be written as
cp
{QJ -
l/io1 - Vi
(5-2-6)
X1
and
cp
(02 -
1/102 -
Vi
(5-2-7)
X2
where Vi
Vi) E,1Nd1
(1/102 -
Vi) Er2Na2
(5-3-8)
For the heterojunction shown in Fig. 5-2-3, the electron current from n-Ge to
p-GaAs is very small because the potential barrier of (%1 + 1/102 + AEclq) across
the junction for electron injection is very high. In contrast, the hole current from the
p-GaAs side to the n-Ge side is dominant because of the low potential barrier l/J02
for hole injection. Therefore, the junction current can be approximated as shown in
Eq. (5-1-42) to be
(5-2-9)
where A = cross section
q = electron charge
DP = hole diffusion constant
Pn = minority or equilibrium hole density in n-Ge
Lp = hole diffusion length
V = bias voltage
VT = voltage equivalent of temperature
0
197
Sec. 5.2
Nd = 5
N. = 6
Tp = 6
Vt;=I
A = 2
x 10 18 cm- 3
x 10 16 cm- 3
x 10-6 sec
V
10- cm 2
2
Compute:
a.
b.
c.
d.
e.
f.
g.
The
The
The
The
The
The
The
Solution
6 x 1016)
= -26 x 10- 3 In ( l .S x 106 = -0.63 V
400 cm 2/V s
Pno
(l.5
x 10!0) 2
x I018
= 45 cm
Chap. 5
198
x 106)2
6 X 10'6 = 5 .4
( l.8
npo
10 -3
cm
-3
Lp =
Y6
10- 2
1.6
10- 19
7.9 x 10- 3
I =
10.4
45
[eI/(z 6 x 1o-3) _
l]
0.958 A
199
Sec. 5.3
~---Open---~
Vo
Distance
(I)
< V <VP
..,,.,.,,.,.,.,7",.,.,.,,.,.,.,'7 EF
---~-
p
N
Ee _ _ _ _ _ _ _ _ _ _ _ _ Conduction
--band
F.B.
"'7,..,..,.,,.,.,.,777777'.>'>EF
Ee
EC
200
Chap. 5
I
I
(2)
- ------------+-I
.......(4)
vu
Forward voltage
I
Forward voltage
Figure 5-3-2
In ordinary diodes the Fermi level exists in the forbidden band. Since the tunnel diode is heavily doped, the Fermi level exists in the valence band in p-type and
in the conduction band inn-type semiconductors. When the tunnel diode is forwardbiased by a voltage between zero and the value that would produce peak tunneling
current lp(O < V < Vp), the energy diagram is shown in part (1) of Fig. 5-3-l(b).
Accordingly, the potential barrier is decreased by the magnitude of the applied
forward-bias voltage. A difference in Fermi levels in both sides is created. Since
there are filled states in the conduction band of the n type at the same energy level as
allowed empty states in the valence band of the p type, the electrons tunnel through
the barrier from the n type to the p type, giving rise to a forward tunneling current
Sec. 5.3
201
from the p type to then type as shown in sector (1) of Fig. 5-3-2(a). As the forward
bias is increased to Vp, the picture of the energy band is as shown in part (2) of Fig.
5-3-l(b). A maximum number of electrons can tunnel through the barrier from the
filled states in the n type to the empty states in the p type, giving rise to the peak
current Ip in Fig. 5-3-2(a). If the bias voltage is further increased, the condition
shown in part (3) of Fig. 5-3-l(b) is reached. The tunneling current decreases as
shown in sector (3) of Fig. 5-3-2(a). Finally, at a very large bias voltage, the band
structure of part (4) of Fig. 5-3-l(b) is obtained. Since there are now no allowed
empty states in the p type at the same energy level as filled states in the n type, no
electrons can tunnel through the barrier and the tunneling current drops to zero as
shown at point (4) of Fig. 5-3-2(a).
When the forward-bias voltage V is increased above the valley voltage Vv , the
ordinary injection current I at the p-n junction starts to flow. This injection current
is increased exponentially with the forward voltage as indicated by the dashed curve
of Fig. 5-3-2(a). The total current, given by the sum of the tunneling current and
the injection current, results in the volt-ampere characteristic of the tunnel diode as
shown in Fig. 5-3-2(b). It can be seen from the figure that the total current reaches
a minimum value Iv (or valley current) somewhere in the region where the tunneldiode characteristic meets the ordinary p-n diode characteristic. The ratio of peak
current to valley current (/pllv) can theoretically reach 50 to 100. In practice, however, this ratio is about 15.
Load line
c
......
Q)
IP
...
...0"'~
"O
u..
Iv
0
VP Vb
vv
Forward voltage
202
Chap. 5
crowave oscillation or amplification generated by the tunnel diode is our major concern in this section. The second load line intersects the 1-V curve at point b only.
This point is stable and shows a dynamic negative conductance that enables the tunnel diode to function as a microwave amplifier or oscillator. The circuit with a load
line crossing point bin the negative-resistance region is called astable. Another load
line crossing point a in the positive-resistance region indicates a monostable circuit.
The negative conductance in Fig. 5-3-3 is given by
- g -ai
-
av
(5-3-1)
-Rn
vb
-Rn= -30 0
R,
L,
Ls= 5 nH
Rs= l 0
C = 20 pF
i-T;;-n~cldi~de-1
I
I
I
I
-R. I
I
I
I
L ________ _J
Figure 5-3-4
nel diode.
The input impedance Zin of the equivalent circuit shown in Fig. 5-3-4 is given
by
z
m
=R +. L + Rn[}/(wC)]
s JW s -Rn - j/(wC)
Z in = Rs - 1
Rn
+ (wRnC) 2 +
.[
J wLs -
wR~C ]
+ (wRnC) 2
(5-3-2)
For the resistive cutoff frequency, the real part of the input impedance Zin must be
zero. Consequently, from Eq. (5-3-2) the resistive cutoff frequency is given by
1~
Jc= 27TRnC
VR. - l
(5-3-3)
For the self-resonance frequency, the imaginary part of the input impedance must be
zero. Thus,
(5-3-4)
Sec. 5.3
203
The tunnel diode can be connected either in parallel or in series with a resistive
load as an amplifier; its equivalent circuits are shown in Fig. 5-3-5.
,-------,
,------,
Tunnel
Tunnel
I
I
I
C
-RI
n1
I
I
I
I
i
L ________ J
diode
diode
L _______ J
Parallel loading. It can be seen from Fig. 5-3-5(a) that the output power in
the load resistance is given by
v2
(5-3-5)
Pout= Re
One part of this output power is generated by the small input power through the tunnel diode amplifier with a gain of A, and this part can be written
v2
(5-3-6)
Pin= ARe
Another part of the output power is generated by the negative resistance, and it is expressed as
v2
P=n Rn
(5-3-7)
Therefore
v2
v2
v2
ARe
Rn
Re
-+-=-
(5-3-8)
A=
Rn
Rn - Re
(5-3-9)
When the negative resistance Rn of the tunnel diode approaches the load resistance
Re, the gain A approaches infinity and the system goes into oscillation.
Series loading.
A is given by
204
A=
Re
Re - Rn
Chap. 5
(5-3-10)
= -Rn - Ro
(5-3-11)
-Rn+ Ro
r---'funnel----,
I
diode
:
I
I
I
I
I
I
---c....__._I__,
-Rn
I
I
I
I
L ___________ _J
Figure 5-3-6
to circulator.
REFERENCES
[I] SOBOL, H., and F. STERZER, Solid-state microwave power sources. IEEE Spectrum, 4, 32,
April 1972.
[2] NAVON, D. H., Semiconductor Microdevices and Materials. Holt, Rinehart and Winston,
New York, 1986.
[3] SzE, S. M., Semiconductor Devices: Physics and Technology, Fig. 6 in Chapter 4. John
Wiley & Sons, New York, 1985.
[4] EARLY, J.M., Maximum rapidly switchable power density in junction triodes. IRE Trans.
on Electron Devices, ED-6, 322-335 (1959).
[5] JOHNSON, E. 0., Physical limitations on frequency and power parameters of transistors.
RCA Rev., 26, No. 6, 163-177, June 1965.
[6] DELOACH, B. C., JR., Recent advances in solid state microwave generators. Advances in
Microwaves, Vol. 2. Academic Press, New York, 1967.
[7] ANDERSON, R. L., Experiments on Ge-GaAs heterojunction. Solid State Electronics, 5, p.
341. Pergamon Press, London, 1962.
Problems
205
SUGGESTED READINGS
IEEE Proceedings. 70, No. 1, January 1982. Special issue on very fast solid-state technology.
LIAO, S. Y., Semiconductor Electronic Devices. Prentice-Hall, Inc., Englewood Cliffs, N.J.,
1990.
NAVON, D. H., Semiconductor Microdevices and Materials. Holt, Kmehart ano wmston, 1..,.ew
York, 1986.
SzE, S. M., Semiconductor Devices: Physics and Technology. John Wiley & Sons, New York,
1985.
PROBLEMS
Microwave Bipolar Transistors
the
the
the
the
the
the
W = 2 x 10- 5 cm
2 x 10- 4 cm
Le = 4 x 10- 4 cm
LE
PB=
PE=
Pc=
VE=
Ve=
20 fl-cm
1 fl-cm
10 fl-cm
0.4 v
0.5 V
A = 0.01 cm2
206
Chap. 5
Nd = 2 x 1019 cm- 3
Na = 3 X 10 17 cm- 3
Tp = 4 x 10- 6 sec
VE= 1.50
v
A = 4 x 10- 2 cm 2
Compute: a.
b.
c.
d.
e.
The
The
The
The
The
The
The
Nd = 5 x 10 18
Na = 6 X 10 16
6
Tp = 4 x 10VE= 0.80
A = 2 x 10- 2
cm- 3
em- 3
sec
v
cm2
207
Problems
lkfl
mA 25
20
iic
i,
0.01 F
u
10
8
(a)
----1----
8 10
20
(b)
Figure PS-11
nal source V, as shown in Fig. PS- I I (a). Its 1-V characteristic curve is shown in Fig.
P5-l l(b).
a. Find the negative resistance and forward resistances of the device.
b. Draw the load line on the 1-V curve.
c. Determine the quiescent operating point of the circuit by the values of voltage and
current.
d. Determine the new operating point by the values of voltage and current when a signal voltage of 14 V is applied to the circuit.
e. Draw the new load line on the 1-V curve.
f. Find the time constant of the circuit.
g. Compute v, i, in and i, as a function of time after the triggering signal is applied
and before the transition takes place.
h. Find the transition time T in microseconds.
i. Calculate v, i, in and i, immediately after transition.
5-12. A microwave tunnel diode has a negative resistance R. and the resonant circuit has a
circuit resistance Re. Derive an equation for the gain of a microwave tunnel-diode amplifier.
5-13. A certain microwave tunnel diode has a negative resistance of 69 + }9.7 !l. Determine the resonant-circuit impedance so that the microwave tunnel-diode amplifier will
produce a power gain of 15 dB.
Chapter 6
Microwave Field-Effect
Transistors
60 INTRODUCTION
After Shockley and his coworkers invented the transistor in 1948, he proposed in
1952 a new type of field-effect transistor (FET) in which the conductivity of a layer
of a semiconductor is modulated by a transverse electric field [ l]. In a conventional
transistor both the majority and the minority carriers are involved; hence this type of
transistor is customarily referred to as a bipolar transistor. In a field-effect transistor
the current flow is carried by majority carriers only; this type is referred to as a
unipolar transistor. In addition, the field-effect transistors are controlled by a
voltage at the third terminal rather than by a current as with bipolar transistors. Our
purpose here is to describe the physical structures, principles of operation, microwave characteristics, and power-frequency limitations of unipolar field-effect
transistors. Since the microwave field-effect transistor has the capability of amplifying small signals up to the frequency range of X band with low-noise figures, it has
lately replaced the parametric amplifier in airborne radar systems because the latter is
complicated in fabrication and expensive in production.
The unipolar field-effect transistor has several advantages over the bipolar
junction transistor:
1. It may have voltage gain in addition to current gain.
2.
3.
4.
5.
208
Its
Its
Its
Its
Sec. 6.1
209
Among the unipolar field-effect transistors are the junction field-effect transistors, metal-semiconductor field-effect transistors, high electron-mobility transistors,
and the metal-oxide-semiconductor field-effect transistors. All these devices are analyzed in this chapter.
Figure 6-1-1
210
Chap. 6
For a p -channel JFEf, the polarities of the two biasing voltages V8 and Vd are interchanged, the head of the arrow points away from the device, and the drain current Id
flows away from the device. Since electrons have higher mobility than holes, the
n -channel JFEf provides higher conductivity and higher speed and is preferred in
most applications to the p -channel JFEf.
_ qNd
where p
q
Nd
Es
Es
E0
= _ qNd
(6-1-1)
=
=
=
=
Integration of Eq. (6-1-1) once and application of the boundary condition of the
electric field E
= - ~~ = 0 at y = a yield
-dV =
dy
qNd
- - ( y - a)
(6-1-2)
Es
qNd 2
= -(y
2Es
2ay)
volts
= 0 at
(6-1-3)
Sec. 6.1
211
= a is expressed as
= qNda 2
volts
2E,
(6-1-4)
qNda 2
2E',
=-
IVg I -
I/Jo
(6-1-6)
The JFET has a conducting channel between the source and the drain electrodes when the gate bias voltage is zero. This is the ON state, and the transistor is
called a normally ON JFET. In order to reach the OFF state, a gate voltage must be
applied to deplete all carriers in the channel. As a result, this device is referred to as
the depletion-mode JFET or D-JFET.
Example 6-1-1:
Channel height:
Electron concentration:
Relative dielectric constant:
a= 0.1 m
= 8 x 10 17 cm- 3
Nd
Er =
11.80
qNda 2
2E,
V.=--=-----------'----~
12
P
= 6.66 volts
(6-1-7)
212
where .,,,
Z
L
W
Chap. 6
=
=
=
electron mobility
distance in z direction
length in x direction
= depletion-layer width
a = width between two p-n junctions
I"dR
Iddx
2q,,N"Z[a - W(x)]
(6-1-8)
W (x) =
112
qNd
(6-1-9)
Then
Wdw
= ~dV
(6-1-10)
qNd
2 2
2Z
q ,,NJ
Jw2 (a
- W)Wdw
LEs
"'J
,,~: z[a(W~ -
Boundary conditions.
Eq. (6-1-4) as
WT) -
~(W~ - W1)]
(6-1-12)
(6-1-13)
The depletion-layer width is
W
= a(V(x) +~~RI+
t/loy
12
(6-1-14)
Then
at x = 0 and V (x) = 0
(6-1-15)
and
at x = L and V (x) = Vd
(6-1-16)
Sec. 6.1
213
Substituting Eqs. (6-1-15) and (6-1-16) into Eq. (6-1-12) yields the drain current
d
where I,, =
q 2 N 2 Za 3
n
312
12
[VJ_ ~(VJ+
IVxl
+ l/J,,)' +~(IVxl+1/10)
]
I,, V.
3
V.
3
V.
LE,,
6 -l-l?)
Pinch off
I-
Saturation region
I
I
Linear
region
v, = o
1.0
J
0.8
-0.5
....,-,,
~
.c::
0.6
-;;
E
(3
-1.0
0.4
-1.5
0.2
-2.0
-2.5
2
10
Figure 6-1-2
Id=
VJ ~ (j VR I + I/Jo),
1,~d[1 - (IVRlv:
l/lor2J
(6-1-18)
g d -
aid I
- I,, [ 1 - (I Vg I + 1/10)'/
avd vg~constant v,,
v,,
(6-1-19)
aid /
avg
Vrconstant
= 1,, vd [ 1 _
2V~
v,,
IVg I +
)'1
I/lo
2
]
(6-1-20)
214
Chap. 6
Saturation region. At the pinch-off voltage, the drain current becomes saturated. By setting VP = Vd + I V8 I + I/Jo, the saturation drain current is given by
Id sat
. = Ip
[.!.3 -
(I Vg I+ "'") + ~(I
Vg I+ 1/10)
3
\!,
\!,
312
]
(6-1-21)
It should be noted that the gm in the saturation region is identical to the one in the
linear region.
For a small signal, the drain or output resistance is defined as
rd= avd
aid
(6-1-24)
vg~constant
f =~=
'
where C8
2TrC8
Ip/VP
_ 2nqNda
2TrZLt:s/(2a)
Trt:sL2
(6-1-26)
LZt:s is
. t he capacitance
.
between gate and source
= 2a
Pinch-off region. When an electric field appears along the x axis between
the drain and the source, the drain end of the gate is more reverse-biased than the
source end. Hence the boundaries of the depletion region are not parallel to the center of the channel, but converged as shown in Fig. 6-1-2. As the drain voltage Vd
and drain current Id are further increased, the channel is finally pinched off.
Breakdown voltage. As the drain voltage Vd increases for a constant gate
voltage V8 , the bias-voltage causes avalanche breakdown across the gate junction,
and the drain current Id increases sharply. The breakdown voltage is shown in Fig.
6-1-3 by the relationship
Sec. 6.1
215
1.2
1.0
V8 =
-0.5
<"!'.
.._,..,
c
t
0.6
'
c::
0.4
3
-1.0
- 1.5
0.2
-2.0
-2.5
JO
I2
Figure 6-1-3
Example 6-1-2:
Current of a JFET
Nd= 1 x
Na=
Er=
a=
L =
Z =
10 17 cm- 3
1 x 1019 cm-3
11.8
0.2 x 10- 4 cm
8 x 10- 4 cm
50 x 10- 4 cm
Vd = IOV
Vg = - 1.5 V
Compute:
a.
b.
c.
d.
e.
f.
The
The
The
The
The
The
Solution
a. The pinch-off voltage is
1.6
Vp
Chap. 6
216
Ip
9.80 mA
i/J = 26
0
10- ln [
I x 10 x I x 10
(1. X l0 10) 2
5
19
1=
0.936 V
Id
=
98 x
IO
-3[_2Q_
- ~3 (10 + 1.53.06+ 0.936)
3.06
312
+ ~ (1.5 + 0.936)
3
3 2
1 ]
3.06
6.80 mA
[I _ (1.5 3.06
+ o.936) + ~ ( 1.5 + o.936) 1 1
3
3.06
6 8 x IO _3 3
0.088 mA
IJ,at
3 2
4.8 GHz
62 METAL-SEMICONDUCTOR FIELD-EFFECT
TRANSISTORS (MESFETs}
In 1938, Schottky suggested that the potential barrier could arise from stable space
charges in the semiconductor alone without the presence of a chemical layer [2]. The
model derived from his theory is known as the Schottky barrier.
If the field-effect transistor is constructed with a metal-semiconductor
Schottky-barrier diode, the device is called a metal-semiconductor field-effect transistor (MESFET). The material may be either silicon or gallium arsenide (GaAs),
and the channel type may be either n channel or p channel.
Since GaAs MEFSETs have the capability of amplifying small signals up to the
frequence range of X band with low-noise figure, they have lately replaced the parametric amplifiers in airborne radar systems because the latters are complicated to
fabricate and expensive to produce.
The GaAs MESFET has higher electron mobility, higher electric field, and
higher electron saturation drift velocity than silicon devices, so its output power is
also greater. Another special feature is its lower noise figure, accounted for by its
Sec. 6.2
217
higher electron mobility. Therefore the GaAs MESFETs are very commonly used in
microwave integrated circuits for high-power, low-noise, and broadband amplifier
applications.
62 1 Physical Structures
The MESFET was developed by many scientists and engineers, such as Mead [3]
and Hooper [4], and it is sometimes also called the Schottky-barrier field-effect
transistor.
The unipolar transistor such as a GaAs MESFET can be developed by using either the epitaxial process or the ion implantation method and those methods are discussed in Chapter 12. Figure 6-2-1 shows schematically a simple MESFET in GaAs.
Source
Gate
Drain
Figure 6-2-1
Schematic diagram of a
GaAs MESFET.
In GaAs MESFETs the substrate is doped with chromium (Cr), which has an
energy level near the center of the GaAs bandgap. As Cr is the dominant impurity,
the Fermi level is pinned near the center of the bandgap. Thus, a very highresistivity substrate (near 108 ohm-cm) generally results, and it is commonly called
the semi-insulator GaAs substrate. On this nonconducting substrate a thin layer of
lightly doped n-type GaAs is grown epitaxially to form the channel region of the
field-effect transistor. In many cases a high resistivity GaAs epitaxial layer, called
the buffer layer, is grown between the n-type GaAs layer and the substrate. The
photolithographic process may be used to define the patterns in the metal layers such
as Au-Ge for source and drain ohmic contacts and in the Al layer for the Schottky
barrier-gate contact. The reason for using GaAs instead of Si is that GaAs has higher
electron mobility and can operate at higher temperature and higher power.
The GaAs MESFET can also be grown by using ion implantation. A thin ntype layer can be formed at the surface of the substrate by implanting Si or a donor
impurity Se from column VI of the element periodic table. However, the ion implantation process requires an anneal to remove the radiation damage. In either the
fully implanted device or the epitaxial device, the source and drain contacts may be
improved by further n+ implantation in these regions.
218
Chap. 6
After the fabrication processes are completed, the individual transistor is separated from the wafer, and this discrete transistor is called a chip device. The chip
device is then alloyed to a header to provide a contact to the collector region, and
Au or Al wires are bonded to the metallized regions to serve as leads to the emitter
and base. This bonded chip device is then named a package transistor. Figure 6-2-2
shows both the package and chip GaAs MESFET devices.
DXL 2501 A-P70
Standard Package
DXL 2501A
GaAs FET Chip Dimensions
l
..l_
tT
Terminal 4
Terminal 2
"iOllfCl'
~ource
E===t+~==:J
250 min
typ.
Terminal .1
gate
~
.L__
0.050 Typ
IT
l
~-
f-020Typ
-+-\
~ 0.070
004 Typ
All
dimL'll~iorr"
Figure 6-2-2
The MESFET device is of an interdigitated structure, fabricated by using an ntype GaAs epitaxial film about 0.15- to 0.35-m thick on a semi-insulating substrate about 100 . Then-channel layer is doped with either sulphur or tin in a doping concentration N between 8 x 10 16/cm 3 and 2 x l0 17/cm 3 The electron mobility
in the layer is in the range of 3000 to 4500 cm 2/Vs. The Schottky barrier-gate is
evaporated aluminum. The source and drain contacts are Au-Ge, Au-Te, or
Au-Te-Ge alloys. A contact metallization pattern of gold is used to bring the
source, drain, and gate contacts out to bonding pads over the semi-insulating substrate. A buffer layer of about 3 m with a doping concentration of 10 15 to 10 16 cm- 3
is often fabricated between the active n-type epitaxial layer and the semi-insulating
substrate.
In Fig. 6-2-3 and in Fig. 6-1-1 for JFET, a voltage is applied in the direction to
reverse-bias the p-n junction between the source and the gate, while the source and
the drain electrodes are forward-biased. Under this bias condition, the majority carriers (electrons) flow in then-type epitaxial layer from the source electrode, through
the channel beneath the gate, to the drain electrode. The current in the channel
Sec. 6.2
(;ate
219
Drain
:'rr,.=11--typ~l~:::.._,:---llL.----........,.,
.l'
vgs
I epitaxial
Contact t--G_a_A_s- - - - . . ; , - - - - - ' - - 4 "
metal
s
Figure 6-2-3
causes a voltage drop along its length so that the Schottky barrier-gate electrode becomes progressively more reverse-biased toward the drain electrode. As a result, a
charge-depletion region is set up in the channel and gradually pinches off the channel
against the semi-insulating substrate toward the drain end. As the reverse bias between the source and the gate increases, so does the height of the charge-depletion
region. The decrease of the channel height in the nonpinched-off region will increase
the channel resistance. Consequently, the drain current Id will be modulated by the
gate voltage VR. This phenomenon is analogous to the characteristics of the collector
current le versus the collector voltage Ve with the base current has a parameter in a
bipolar transistor. In other words, a family of curves of the drain current Id versus
the voltage Vd, between the source and drain with the gate voltage VR as a parameter
will be generated in an unipolar GaAs MESFET, as shown in Figure 6-2-4.
The transconductance of a field-effect transistor (FET) is expressed as
mhos
(6-2-1)
For a fixed drain-to-source voltage Vd,, the drain current Id is a function of the
reverse-biasing gate voltage VR. Because the drain current Id is controlled by the
field effect of the gate voltage VR, this device is referred to as the field-effect transistor. When the drain current Id is continuously increasing, the ohmic voltage drop between the source and the channel reverse-biases the p-n junction further. As a result, the channel is eventually pinched off. When the channel is pinched off, the
drain current Id will remain almost constant even though the drain-to-source voltage
vd is continuously increased.
Pinch-off voltage Vp. The pinch-off voltage is the gate reverse voltage that
removes all the free charge from the channel. Poisson's equation for the voltage in
the n channel, in terms of the volume charge density is given by
d2V
dy2
_ f!_
_ qNd
_ qNd
(6-2-2)
220
Chap. 6
35
~s = 0.0_
30
25
--0.5
<
20
-~
2
~
._,
--1.0
15
c:
Q
10
Figure 6-2-4
MESFET.
(6-2-3)
Sec. 6.2
Example 6-2-1:
221
a= 0.1 m
Nd = 8 x 10' 7 cm- 3
E, = 13.10
v.
qNda 2
l.6 x 10- 9 x 8 x 1023 x (0.1 x 10- 6 ) 2
= ~~~~~~~~~~~~~~2E.,
2 X 8.854 X 10- 12 X 13.10
=~~
6.00 volts
1. Intrinsic elements:
gm
Gd
= transconductance of the
= drain conductance
MESFET
R; = input resistance
C8 ..
= gate-source capacitance
Cxd
=
=
2. Extrinsic elements:
R8
R,
source-gate resistance
222
Chap. 6
Source
n-type
channel
Gate
,------------------,
Intrinsic
1
I
transistor
I
I
Drain
I
L
______ _
R,
Source
Figure 6-2-5
Source
=
Ze =
Cp
The values of these intrinsic and extrinsic elements depend on the channel
type, material, structure, and dimensions of the Schottky barrier-gate FET. The
large values of the extrinsic resistances will seriously decrease the power gain and
efficiency and increase the noise figure of the MESFET. It is advantageous to increase the channel doping N as high as possible in order to decrease the relative
influence of the feedback capacitance C8 d and to increase the transconductance gm
and the de open-circuit voltage gain. However, an increase in concentration N decreases the breakdown voltage of the gate. A doping of 10 18 per cubic centimeter
might be an upper limit.
Sec. 6.2
223
= Ip
Id
qNd,aZV".
=
1s the saturation current for the Shockley case
L
3
, = low-field mobility in square meters per volt-second
q = 1.6 x 10- 19 coulomb is the electron charge
Nd = doping concentration in electrons per cubic meter
a = channel height
Z = channel depth or width
L = gate length
V" = pinch-off voltage as defined in Eq. (6-2-3)
where Ip
u=
p
at V8
vd +V. Iv.8 1) 112 is the normalized sum of the drain and gate voltages with
"
(I~ 1)1
12
T'/
(6-2-4)
amperes
,~~I
Vs
v.,
v
=
=
Ex
+ ,
Figure 6-2-4 shows a plot of the drain current Id versus the drain voltage Vds
with the gate voltage V8 , as a parameter for a typical n-channel GaAs MESFET. The
drain current has a maximum at u = Um given by
u~ + 3um(; - p
where z
2
)
+ 2p 3 -
~=
(6-2-5)
Substitution of Eq. (6-2-5) into Eq. (6-2-4) yields the saturation drain current as
(6-2-6)
g
m
aIVx I
vd~ con"ant
gmax(Um - p)
+ z(u;;, - p 2 )
(6-2-7)
224
Chap. 6
In practice, the drain current and mutual conductance of a GaAs MESFET can be
expressed by
(6-2-8)
and
(6-2-9)
The velocity-field curves of a GaAs MESFET are very complicated. Figure
6-2-6 shows these curves in the saturation region [7]. The narrowest channel cross
section is located under the drain end of the gate. The peak electric field appears
near the drain. The drift velocity rises to a peak at x1, close to the center of the
channel, and falls to the low saturated value under the gate edge. To preserve current
continuity, heavy electron accumulation has to form in this region because the channel cross section is narrowing. In addition, the electrons are moving progressively
slower with increasing x. Exactly the opposite occurs between x2 and X3. The channel widens and the electrons move faster, causing a strong depletion layer. The
charges in the accumulation and depletion layers are nearly equal, and most of the
drain voltage drops in this stationary dipole-layer.
For a Si MESFET, the drain current is expressed as [7]
Id= Zqn(x)v(x)d(x)
(6-2-10)
Sec. 6.2
225
ID/Z (mA/cm)
+
ID
-IV
200
D
_J__
100
~--I
20
IO
I
I
-3mH
VG=-] V
VD (V)
I
I
I
I
I
I
I
Electric field
J _________ 0
I'
I
I
'
I
I
;~------ - - : - ---Vp
I
---+
I
I
I
I
Electron drift
velocity
---V
0 '---+-----'----r--+-_._
Cl
0.4
.Cl
0.2
Space charge
in channel
?.
0
~
-0.2
-0.4
Figure 6-2-6 Channel cross section, electric field, drift velocity, and space charge
distribution in the channel of a GaAs MESFIIT in the saturation region.
as shown in Fig. 6-2-7(c). When the drain voltage is increased beyond Vdsai. the depletion layer widens toward the drain as indicated in Fig. 6-2-7(d). When a negative
voltage is applied to the gate as shown in Fig. 6-2-7(e), the gate-to-channel junction
is reverse-biased, and the depletion layer becomes wider. As the gate voltage V8 , is
more negative, the channel is almost pinched off as shown in Fig. 6-2-7(f), and the
drain current h is nearly cut off.
226
Ohmic
contact
foe
Chap. 6
Nop~
n-Epi
Insulating
!'J_t
c
ll
V0
: } substrate
\.
x
(b)
(a)
Constant n
-.J..
--+
,,,,,.--Nogate
Xi
..J.--- Constant v
n=.Vv v=vs 1
1
(d)
(c)
Ve= 0
lil'----Vc <O
Vo
(e)
Figure 6-2-7
(f)
MESFEf.
Example 6-2-2:
Nd
8 x 10 17 cm- 3
a=O.l,m
Er=
13.1
L = 14 ,m
36 ,m
, = 0.08 m2/V s
=
800 cm 2 /V s
Sec. 6.2
vd = 5 volts
Vg = -2 volts
v., = 2 x 105 mis
Drain voltage:
Gate voltage:
Saturation drift velocity:
a.
b.
c.
d.
Solution
a. From Eq. (6-2-3), the pinch-off voltage is
Yr= 1.6 x 10- 19 x 8 x 1023 x 10- 14 /(2 x 8.854 x 10- 12 x 13.1)
= 5.52 volts
= 0.158
0 is
qNd1wZVr
2L
2)1/2 =l.126
)1/2
u= ( _
5 52
p = ( 5.52
= 0.60
u 2 = 1.268
u 3 = 1.428
p 2 = 0.362
p 3 = 0.217
x
_, x 3(1.268 - 0.362) - 2(1.428 - 0.217)
_
4 845
10
.
1 - 0.158(1.268 - 0.362)
4.845 x 10- 3 x 0.259
= 1.26 mA
227
228
Chap. 6
Hz
where gm = transconductance
.
dQ
C KS = gate-source capacitance
= dV
xs
L
Vs
=
=
I
VRd= constant
gate length
saturation drift velocity
It is interesting to note that the cutoff frequency of the lumped circuit analysis as
shown in Eq. (6-2-1 I) is different from the charge-carrier transit-time cutoff frequency as shown in Eq. (5-1-58) by a factor of one-half.
The maximum frequency of oscillation depends on the device transconductance
and the drain resistance in a distributed circuit. It is expressed [6] as
0 (1.. e. Vv
Vs
= 47TL
and Um
(6-2-12)
Hz
}max
(3)
= ";;j
1 3
~ 1.
We have
Vs(3)
/mas = Y L -;;
1 6
/
Hz
(6-2-13)
where y = 0.14 for Ep/Vs = 13 and y = 0.18 for Ep/Vs = 20 in the case of
GaAs.
It has been found experimentally [6] that the maximum frequency of oscillation for a gallium arsenide FET with the gate length less than IO mis
/max=
where L
33 x 10 3
L
Hz
(6-2-14)
Sec. 6.2
229
/co =
where
1
21TT
Vs
Hz
= 21Tl
(6-2-15)
Vs
It is evident that the gallium arsenide FET has a better figure of merit than the
silicon FET for an X-band amplifier because the saturation drift velocity V, is
2 x 107 cm/sec for GaAs at an electric field of 3 kV/cm and 8 x 106 cm/sec for silicon at 15 kV/cm. In comparing Eq. (6-2-15) with Eq. (6-2-11), the difference is a
factor of one-half.
The highest frequency of oscillation for maximum power gain with the input
and output networks matched is given [8] as
fmax =
J,.
Rs
Rd
RR
1 2
+R
) 1
Hz
R;
2.5 D
Rs= 2.5 fl
gm= 50 mU
Cv
0.60 pF
Solution
a. From Eq. (6-2-11), the cutoff frequency is
/cu
gm
27TC8 s
0.05
27T x 0.6 x 10- 12
= - - = ---------.,.
13.26 GHz
(6-2-16)
230
Chap. 6
j;., (
Rd
/max=2 R +Rff +R
.I
= 13.26 x 10
2
=
)112
9
(
450
)
2.5 + 3 + 2.5
1 2
49.73 GHz
63 1 Physical Structure
The basic structure of a HEMT is a selectively doped GaAs-AlGaAs heterojunction
structure as shown in Fig. 6-3- l.
An undoped GaAs layer and an Si-doped n-type AIGaAs layer are successively
grown on a semi-insulating GaAs substrate. A two-dimensional electron gas
(2-DEG) is created between the undoped and n-type layers. A buffer layer is sandwiched between the undoped GaAs layer and the semi-insulator substrate.
The HEMT can be fabricated by using the integrated-circuit techniques. Fig.
6-3-2 indicates the sequence for the self-aligned gate procedures in the fabrication of
large-scale integration HEMTs, including the E-(enhacement-mode) and D(depletion-mode) HEMTs.
The processing steps include the following:
1. Ohmic contact formation: The active region is isolated by a shallow mesa step
(180 nm), which is almost achieved in a single process, and can be made
nearly planar. The source and drain for E- and D-HEMTs are metallized with
Sec. 6.3
231
Source
SiN
Drain
Au Ge/Ni
,----------
2 DEG
Undopcd A!GaAs
Undoped GaAs
S.I. buffer
\
GaAs S.I.
\~---~
Ohmic contact
Insulator
~~~~~~~ii~~~Cap layer
y A!GaAs
If'.- - - - - - - - - - - - Semi-insulating
GaAs substrate
~~~~~~~~~--'
GaAs
A!GaAs
GaAs
Electron layer
Photoresist
F:'~"'J
S.I. GaAs
"-GaAs
(2)
(!)
Interconnect
metal
(3)
(4)
(5)
Figure 6-3-2 Processing steps for HEMT direct-coupled FET logic (DCFL) circuits. (After
M. Abe and others [9]; reprinted by permission of IEEE, Inc.)
232
Chap. 6
4. Gate metallization: Schottky contacts for the E- and D-HEMT gate are provided by depositing Al. The Schottky gate contacts and the GaAs top layer for
ohmic contacts are then self-aligned to achieve high-speed performance.
5. Interconnect metallization: Finally, electrical connections from the interconnecting metal, composed of Ti, Pt, and Au, to the device terminals are provided through contact holes etched in a crossover insulator film.
Figure 6-3-3 shows the cross-sectional view of a typical self-aligned structure
of E- and D-HEMTs forming an inverter for the direct-coupled FET logic (DCFL)
ciruit configuration [9].
E-HEMT
Ohmic
contact
IHI EMT
Gate
Interconnect
metal
Insulator (Si0 2 )
Insulator (Si0 2 )
Ga As
AIGaAs
Electron layer
Semi-insulating GaAs substrate
Ga As
Figure 6-3-3 Cross-sectional view of a DCFL HEMT. (After M. Abe et al. [9];
reprinted by permission of IEEE, Inc.)
The Fermi energy level of the gate metal is matched to the pinning point,
which is 1.2 eV below the conduction band. With the reduced AlGaAs layer thickness, the electrons supplied by donors in the AlGaAs layer are insufficient to pin the
surface Fermi level, and the space-charge region is extended into the undoped GaAs
layer. As a result, band bending is moving upward and the two-dimensional electron
gas (2-DEG) does not appear. When a positive voltage higher than the threshold
Sec. 6.3
233
voltage is applied to the gate, electrons accumulate at the interface and form a twodimensional electron gas.
The electron concentration can control D-(depletion-mode) and E-(enhancement-mode) HEMT operations. As temperature decreases, electron mobility, which
is about 8000 cm 2/V s at 300 K, increases dramatically to 2 x 105 cm 2/V s at
77 because of reduced phonon scattering. When the temperatures decrease further, the electron mobility of I. 5 x l 0 6 cm 2/V s at 50K and 2. 5 x l 0 6 cm 2/V s
at 4.5K have been demonstrated.
Ids
where q
n(z)
W
v (z)
qn(z)Wv(z)
(6-3-1)
electron charge
electron velocity
Current of a HEMT
W = 150 ,m
v(z) = 2 x 105 mis
n(z) = 5.21 x 10 15 m- 2
Solution
qn(z)Wv(z)
234
40
<.:
30
"
i::
~
u
..._.:;
A
10
A
A
A
2
-6
is:
3
vds
0.4
0.2
ov
20
0.6
Chap. 6
A
A
-0.2
-0.4
6.-0.6 V
-0.8 v
6
in volts
Note:
A
-
Figure 6-3-4
TABLE 631
/- V characteristics of a HEMT.
Device
Frequency (GHz)
Noise
Power
Speed
HEMT
GaAs MESFET
GaAs-AlGaAs HBT*
Si MOSFET
Si bipolar transistor
Up to 70
40
20
10
Very Good
Good
Good
Poor
Poor
Very Good
Good
Good
Very good
Poor
Excellent
Good
Excellent
Very poor
Good
Sec. 6.3
Lg
Rg
0.15nH
sn
Cgd
235
Ld
Rgd
O.Ol2pF 0.001
0.14 nH
Gate~ l----WV'v-----..---+-----.--'V>Nv--__,..~100''-----oDrain
c~, _l
0.18 pF
R,,
gm e-Jw!
54 ms
eds
0.08 pl'
1.8 ps
4.hQ
Soun:-l:
Figure 6-3-5
V,h )/ E]I/2
(6-3-2)
q = electron charge
Nd = donor concentration
l/Jms = metal-semicondcutor Schottky barrier potential between Al and GaAs
Ee = conduction bandedge difference between GaAs and AlGaAs
f = EaEr is the permittivity of AlGaAs
Er = dielectric constant of AlGaAs
Example 6-3-2:
Sensitivity of HEMT
= 0.13 v
Nd = 2 x 1024 m- 3
v,h
i/Jms
0.8 V
236
Chap. 6
Egg= 1.43 V
E8a = 1.80 V
Er = 4.43
GaAs bandgap:
AIGaAs bandgap:
AIGaAs dielectric constant
Compute:
E8a
E8 ,
d~
df
= - ( 2qNd(l/Jms - aEc =
v,h)/E JII 2
ddV,h I =
0.49 x 104]1/2
[
10-12
-70 mV/nm
70mV I nm
Sec. 6.4
237
238
Chap. 6
Metal
electrode
p-type
semiconductor
s
G
n-channel
Figure 6-4-1
Sec. 6.4
239
voltage must be applied to the gate to form an n channel for conduction. The
drain current is enhanced by the positive voltage. This type is called the
enhancement-mode (normally OFF) n-channel MOSFET.
2. n-Channel Depletion Mode (normally ON). If an n channel exists at equilibrium (that is, at zero bias), a negative gate voltage must be applied to deplete
the carriers in the channel. In effect, the channel conductance is reduced, and
the device is turned OFF. This type is called the depletion-mode (normally
ON) n-channel MOSFET.
3. p-Channel Enhancement Mode (normally OFF). A negative voltage must be
applied to the gate to induce a p channel for conduction. This type is called the
enhancement-mode (normally OFF) p-channel MOSFET.
4. p -Channel Depletion Mode (normally ON). A positive voltage must be applied
to the gate to deplete the carriers in the channel for nonconduction. This type is
called the depletion-mode (normally ON) p-channel MOSFET.
Figure 6-4-2 shows the four modes of the MOSFETs, and Fig. 6-4-3 illustrates
their electric symbols, output 1-V characteristics, and transfer characteristics.
(;
Enhancement
Enhancement
(normally off)
(normally off)
(;
Depletion
Depletion
(normally on)
(normally on)
(a)
(b)
240
Chap. 6
Type
Transfer
characteristic
Output
characteristic
G
n-channel
enhancement
(normally off)
lot~;;:,
so----lILD
tl
-
Vo __________.._
10
(normally on)
s~
s /)
(normally off)
Vo-
s~ILD
_..____VD
v~
-t
t
Iv
v,
/ I.
+
1
o
vG
(normally on)
0
VG
Ve
p-channel
depletion
ot~!:o
G
p-channel
enhancement
v,
VG
G
n-channel
depletion
S~D
-VD
v-~
t
t
G -
In
Figure 6-4-3 Electric symbols and output and transfer characteristics of the four
modes of MOSFETs.
Id
JLnC; { ( V8
21/fb -
112
[(Vd + 21/fb) 1l 2
(21/!b) 1l 2]}
(6-4-1)
where JLn
V8 = gate voltage
(E; - EF)/q is the potential difference between the Fermi level EF and
the intrinsic Fermi level ;
vd = drain voltage
Es = semiconductor permittivity
q = carrier charge
Na = acceptor concentration
I/lb =
In the linear region, the drain voltage is small and Eq. (6-4-1) becomes
Z
[(
Id= [_,JLnC; V8
)
(1
V1h Vd - 2 +
(6-4-2)
Sec. 6.4
241
40
Saturation
region
8
30
7
20
5
IO
4
3
Figure 6-4-4
IO
vd
in volts
I2
I4
I6
I8
or
Id =
where
Vih
LJJ-n C;(Vg
(6-4-3)
- v,h)Vd
= 21/Jb + ~; (EsqNal/Jb)'l 2
rnZ
(Vg - v,h )2
= T,nCi
Vdsat
(6-4-4)
Transconductance. The transconductance gm, which is also called the mutual conductance, in the linear region can be found from Eq. (6-4-3) as
242
Chap. 6
(6-4-5)
In the saturation region, the transconductance becomes
2mZ
=L
gm sat
(6-4-6)
gd
aid
av
Vg=constant
z ," C(V8 =L
v,h)
(6-4-7)
All equations derived here so far are based on the idealized n -channel MOSFET. For
an idealized p-channel MOSFET, all voltages V8, Vd, and Vih are negative, and the
drain current Id flows from the source to the drain. For a real n-channel MOSFEf
(say, Al-Oxide-Si structure), the saturation drain current is
ldsat
ZC;(Vg - v,h)Vs
(6-4-8)
gm= ZC;Vs
where Vs
(6-4-9)
th
</Jms _ Qf
C
q
i
2/,
'/'b
+2
,/, )1/2
C (Es qNa'/'b
(6-4-10)
where </>ms = <Pm - <Ps is the work function difference (in eV) between the metal
work function <Pm and the semiconductor work function <Ps
Q1 = fixed oxide charges
Example 6-4-1:
3 x 10 17 cm- 3
E,= 11.8
Na
4
d = 0.01 ,m
T = 300K
E;, =
Sec. 6.4
243
Solution
a. From Eq. (6-4-4), the surface potential for strong inversion is
!/Jsinv
3 x 1011 )
2 X 26 X 10- fn ( 1.
0.874 volt
lQIO
C; -
E; _
d-
8.854 X 10-iz _
2
O.OI x 10 _6
3.54 mF/m
v,h = 0.874 +
3.54
cfb
Gate
gm
vg
gin
Figure 6-4-5
cin
Drain
vg
gout
cout
vd
244
Chap. 6
The maximum operating frequency of a MOSFET in the linear region can be expressed as
(6-4-11)
= Lf-ln C; Vg = ZC;V
(6-4-12)
Characteristics of a MOSFET
l = 4 ,m
Z = 12 ,m
d = 0.05 ,m
V8 = 5 V
m=l
= 0.10 v
/.l-n = 1350 x 10- 4 m2/V. s
'Vs = 1.70 x 107 cm/s
V'ih
;, =
3.9
Sec. 6.5
245
Solution
a. The capacitance of the insulator Si0 2 is
C.
=~ =
d
6 91 x 10-4F/m2
0
= 6.91 mA
= ZC;Vs
=
= l.4lmU
Vs
27TL
1.7 X 105
21T x 4 x 10-6
6.76 GHz
As discussed in Section 6-2, the source, channel, and the drain of the MOS transistor are surrounded by a depletion region, so there is no need to isolate individual
components. This elimination of isolation regions in MOS transistors allows a much
greater packing density on a semiconductor chip than is possible with bipolar junction transistors. The MOSFET can be subdivided into two groups:
1. Then-channel MOSFET is commonly referred to as an NMOS.
2. The complementary MOSFET is usually called a CMOS. The CMOS provides
n -channel and p -channel MOSFETs on the same chip.
The fabrication technology for an NMOS is much simpler than for the bipolar transistor. The CMOS circuit has lower power dissipation than both the bipolar transistor
and the NMOS circuits. So both NMOS and CMOS devices are very useful in high
density integrated circuits.
651 NMOS Devices
246
Chap. 6
devices, however, have dominated the IC market since the 1970s because their electron mobility is higher than that of holes.
NMOS structure.
Polysilicon
Source
Buried
contact
Chan-stop
implant
EMDA
implant
DMD
P-su bstrate
From Fig. 6-5-1, it can be seen that two enhancement-mode (normally OFF)
devices (EMDA and EMDs) are in series with a depletion-mode (normally ON)
device (DMD). A field oxide (FOX) surrounds the transistors, and the gate and
source of the DMD are connected at the buried contact. An intermediate dielectric
layer separates the overlying metal layer from the underlying layers. In the JFET the
high input resistance is obtained from the reverse-biased p-n junction. In the MOSFET the extremely high input resistance ( ~ 10 14 ohms) is made possible by the insulator.
There are two basic structures for MOSFETs: the depletion type and the induced type as shown in Figs. 6-5-2 and 6-5-3, respectively. The major difference
between them is that with the terminals of the device open-circuited, the depletion
type has a conducting channel that links the drain to the source; the induced type has
a channel of opposite type to that of the drain and the source linking the two regions. When the device has p-type source and drain regions, it is called the pchannel MOSFET or PMOS.
Metal ohmic
Metal ohmic
contact
contact
I
I
Source
Gate
Vs
Ve
Source
11
Usually the
substrate is
connected
to the source L -
Drain
"
c""'"'"
- - ../
_ -
Drain
11
I'I
Source
Gate
Vs
VG
'- - - \ - .)
Transition
region edge
Usually the
substrate is
connected
to the source L -
__ _
cnanne1 ai
v -0
Drain
vD
p-Type substrate
Oxide
(insulator)
Drain I
1
+
G -
II
II
\.. __ \ _ )
Transition
region edge
- -
VsT
VST
Substrate terminal
Substrate terminal
(a) Cross-sectional view
la I Cross-sectional view
ID
1'U
11
\.. _ _ _ _ '_)
p-Type substrate
- -
Oxide
(insulator)
VD
,...-------"'\
I
\. - -
ID
Vc;s = VDc - Vp
VGS = VDC - Vp
I
I
11--------------} ~~j:~c~ment
);~;,~: ,
Vcs >O
Enhancement
mode
mode
----------------
mode
L'
__ vc;s =O
IU/I~~~~~~~~~~~~~~
VDc
V/J(;
( b) /- V curves
"'
~
"""
Figure 6-5-2
Figure 6-5-3
248
Chap. 6
The structures of enhancement and depletion modes refer to the relative increase or decrease of the majority carrier density in the channel connecting the
source to the drain. If a given gate bias tends to increase the majority carrier density
in the channel, the device is said to be operated in the enhancement mode (normally
OFF). Hence, if the gate in an NMOS is biased by a positive voltage with respect to
the substrate, which tends to drain more electrons into the n-type channel, it is said
to be operated in the enhancement mode. If a negative gate potential with respect to
the substrate is applied in order to diminish the electron density in the n channel,
then it is said to be operated in the depletion mode. For a p-channel MOSFEf
(PMOS), when the negative gate potential with respect to the substrate is biased, the
device is operated in the enhancement mode; and when the positive gate voltage is
applied, the device is operated in the depletion mode. When a MOS is operated in
the enhancement mode, the drain current is higher because the majority carrier density is higher. When, however, the MOS is operated in the depletion mode, the drain
current is lower because the majority carrier density is lower.
NMOS operation. The operation of an NMOS can be described as a logic
gate as shown in Fig. 6-5-4.
Two enhancement-mode devices (EMO) are in series with a depletion-mode
device (DMD), and the three transistors are connected between the positive power
supply Voo and ground Vss reference. The DMD is normally on at Vas = 0 and acts
as a current source for the two EMDs. Gates A and B of the two EMDs serve as inputs to the logic circuit, and the DMD's gate-source connection is the output electrode of the logic circuit. The output voltage of the two-input NANO circuit is low
only when both EMDs are turned on at their logic-high level.
NMOS devices can be also used as NOR-gate circuits, as shown in Fig. 6-5-5.
The two-input enhancement-mode NMOS devices are called drivers, and the
depletion-mode MOSFET, the load. When there is a logic 0-state input signal (low
voltage at v,,) at the driver device, the device has a very small channel current and
the output is at 1 state (the output voltage is close to Voo). But when the input signal
is at 1 state, the device conducts a large current with a small voltage drop across the
driver, and the output is at 0 state. The input and output are inverted and a NOR
logic is achieved.
CMOS structure. There are three structures of CMOS devices: n tub, p tub,
and twin tub. When a tub is formed in a p -type substrate, the device is called an n
tub, just as when in an n-type substrate it is called a p tub. If an n-tub and a p-tub
are combined on the same substrate, the device is referred to as a twin tub. A tub is
Sec. 6.5
EMDA
249
EMD 8
_fl_ .....-+-~DMD
I I
A
Inputs
AB
'-y-'
Output
(a)
Polysilicon
Inputs
Output
(b)
SiN
(C)
Figure 6-5-5
250
11-Channcl
Chap. 6
p-Channel
Oxide
Polysilicon
/>-Substrate
'--~~~~~~~------------.~~~---~~~~~(a) II tub
p-Gla'5
Nitride
Polysilicon
_____________
~f:i!'.:x!_
11+
AQ
_______ _
-Substrate
f'
or n
:-.ubstrate
also called a well, and it can be produced by extra diffusion steps. Figure 6-5-6
shows the structures of an n tub, a p tub, and a twin tub.
CMOS operation. The operation of a CMOS inverter is shown in Fig.
6-5-7. The p -channel transistor is formed in the n -type structure, and the n -channel
transistor is grown in the p region, which in turn is formed in the n-type substrate.
The p region acts as the n-channel transistor's substrate (back gate) and it is commonly referred to as a tub or well. The gates of the n - and p -channel transistors are
connected and serve as the input to the inverter. The common drains of each device
are the output of the inverter. The threshold voltage of the n- and p -channel transistors are V," and V,P, respectively, (V,P < O). Figure 6-5-7(c) shows the dependence
of the output voltage V,, on the input voltage V, of the CMOS inverter. For V, = 0,
the n -channel transistor is OFF (Vi ~ Vm), while the p -channel transistor is heavily
turned ON (that is, the gate-to-source voltage of the p channel is - Voo, which is
much more negative than V,p). Therefore, the output voltage is V0 = Voo. As the input voltage V, is increased above zero, the n-channel transistor eventually is turned
Sec. 6.5
251
11-Channe!
11-Channel
11-Substrate
I b) Device cross-section
JN
P\
"
"""
\ I
Cl
....Cl
3'
vrn
( d) Current curve
ON, while the p-channel transistor finally turns OFF. When VI > (Voo - IV,p I), the
output voltage is Vo = Vss.
The key feature of a CMOS gate is that in either logic state (Vo = Voo or Vss)
one of the two transistors is OFF and the current conducted between Voo and Vss is
negligible. Figure 6-5-7(d) shows the current loo as a function of V/. A significant
current is conducted through the CMOS circuit only when both transistors are ON at
the same switching time. The low power consumption of CMOS is one of its most
important contributions. The performance and simplification of circuit design are
other attractive features of the CMOS device. CMOS provide the circuit designer
with flexibility in designing circuits that are either static CMOS (a p-channel transistor for every n -channel transistor) or dynamic CMOS (unequal number of n - and p channel transistors) .
252
Chap. 6
Memories are devices that can store digital data or information in terms of bits
(binary digits). Many memory chips were designed and developed by using NMOS
devices. The major semiconductor memory categories are ROM, PROM, EPROM,
EEPROM, SRAM, BSRAM, and DRAM. The unit cells of the memory types are
shown in circuit schematics in Fig. 6-5-8 [14).
w _ _.....___ _---+_
la) ROM
IJ
IJ
(cl !::PROM
(bl PROM
idi H.PROM
w-----+--
le) SRAM
B
(fl BSRAM
lg) DRAM
Figure 6-5-8 Circuit diagrams of memory types. (After S. Asai [ 14 ]; reprinted by permission of JEEE, inc.)
ROM. The read-only memory (ROM) is also called mask ROM. During fabrication stage, the information is inscribed (and cannot be altered) in the form of
presence or absence of a link between the word (access) line and the bit (sense) line.
This causes the presence or absence of a readout signal on the bit line when the word
line is activated. The essential part of the ROM is the way the link is provided, since
the link determines the cell size (and thus the cost per bit) and the turnaround time
(TAT). Fast, high-density ROMs are in great demand for personal computers.
Sec. 6.5
253
254
Chap. 6
BSRAM. The bipolar SRAM (BSRAM) is the fastest of all the semiconductor memory types, but MOS SRAM is the fastest among MOS memories. Figure
6-5-9 shows the equivalent-circuit diagrams for bipolar SRAMs.
/J/l/)
RI/
Lateral
Vertical
pnp
fl/Jll
P-Substrate
Figure 6-5-9
Sec. 6.6
255
clock pulses, and so it is also called the charge-transfer device (CTD). CCDs have
many microwave applications, such as in infrared detection and imaging and digital
signal processing. There are three basic types of CCDs: surface-channel CCD
(SCCD), buried-channel CCD (BCCD), and junction CCD (JCCD). In the SCCD or
BCCD the charge is stored and transferred at the semiconductor surface or in the
semiconductor interior, respectively; whereas in the JCCD the store and transfer of
the charge packet occur at the p-n junction.
The motion of the charge packets in a charge-coupled device is transversely
controlled by the applied gate voltages. This phenomenon is similar to the carrier
motion in a microwave field-effect transistor like MESFET or MOSFET. In effect,
the CCD can be referred to as the field-effect CCD.
--------r-n-semiconductor
Depletion
edge
Vi
----------r--Depletwn
edge
I
I
v1
Distance-(a)
(b)
Figure 6-6-1 Energy-band diagrams of MIS structure. (From W. S. Boyle and G. E. Smith
[ 15 ]; reprinted with permission from The Bell System, AT&T.)
256
Chap. 6
The voltage applied to the metal electrode is negative with respect to the semiconductor and large enough to cause depletion. When the voltage is first applied at
t = 0, there are no holes at the insulator-semiconductor interface [see Fig.
6-6-1 (a)]. As holes are introduced into the depletion region, they will accumulate at
the interface and cause the surface potential to be more positive [see Fig. 6-6-l(b)].
Three-phase structure. The CCD can be constructed in the form of a typical three-phase structure as shown in Fig. 6-6-2 [16].
Vi=-SY
V2 =-IOV
V3 =-5V
17-type silicon
(a)
Silicon dioxide
17-type silicon
(b)
The CCD consists of a closely spaced array of MIS diodes on an n-type semiconductor substrate with a large negative gate voltage applied. Its basic function is to
store and transfer the charge packets from one potential well to an adjacent one. As
shown in Fig. 6-6-2(a), Vi = V3 and Vi is more negative. In effect, a potential well
with stored holes is created at gate electrode 2. The stored charge is temporary because a thermal effect will diffuse the holes out of the wells. Therefore the switching
time of the voltage clock must be fast enough to move all charges out of the occupied
well to the next empty one. When the voltage Vi is pulsed to be more negative than
the other two voltages Vi and Vi, the charge begins to transfer to the potential well at
gate electrode 3 as shown in Fig. 6-6-2(b).
Sec. 6.6
257
t=t
~~~~~~-1-~~~~-1-~~~~1---
V,
,
r-------
+++ r------,
~----, +++
L-.J
1..-.J
1..-.J
(a)
.,
+++r-.,
...
.__.,
I
_~
L-.,
.---.
1
~-~
L.--.
+++,----
...
I
_~
(b)
- - - . , T++ ~-----,
...
_~
r------,
1..-~
++1r---
1..-~
(c)
Figure 6-6-3
258
Chap. 6
p-S1
Jn-,ubtor
- - - - -~~ t
\\ell partial!)
filled" Ith
signal (!large
Figure 6-6-4 Physical structure and energy-band diagrams of a BCCD. (From D. F. Barie
[28]; reprinted with permission of the IEEE, Inc.)
There are two cases for the creation of a charge packet under deep depletion.
(Qsig = O). When the signal charge is zero, an empty well
is formed by the potential minimum at the semiconductor surface as shown in
Fig. 6-6-4(b). Gate voltage V8 and surface potential r/ls are related by
1. Zero Signal-Charge
Vg - Vlb
qNaW
(6-6-1)
Sec. 6.6
where
259
vfb =
V;
ftatband voltage
voltage across the insulator
2Esl/!s)l/2 mt
. he dep1et10n
.
'd h
Wl t
( qNa
2. Stored Signal-Charge (Q,;g > O). When a signal-charge packet is stored at the
semiconductor surface, the surface potential decreases and the potential well is
partially filled as shown in Fig. 6-6-4(c). The surface potential equation of Eq.
(6-6-2) becomes
V8
Vtb
(6-6-3)
The maximum charge density (electron or hole) that can be stored on an MOS
capacitor is approximately equal to
Nmax = CV11
q
(6-6-4)
71=}-E
If a single charge pulse with an initial amplitude Po transfers down a CCD register,
after n transfers, the amplitude Pn becomes
Pn
= PoT/n =
Po(! - nE)
for
E <{
(6-6-6)
260
Chap. 6
Frequency response. There are, in fact, upper and lower frequency limitations for CCDs. The potential well will not remain indefinitely, and thermally generated electrons (or holes) eventually fill the well completely. Also, the time stored
by the charge must be much shorter than the thermal relaxation time of the CCD's
capacitor. So the maximum frequency is limited by the channel length L.
Power dissipation.
p
Example 6-6-1:
nJVQmax
(6-6-7)
V = 10 volts
n = 3
Qmax
0.04 pC
10 MHz
12 W
Infrared detection and imaging. Because varying amounts of charge corresponding to information can be introduced into the potential wells at one end of
the CCD structure to emerge after some delay at the other end, the CCD is capable
of detecting and imaging the infrared light from a target [ 17]. Ten years ago, when
the CCD was first developed, only an array of 12 x 12 photodetectors could be used
to detect the infrared images of a target. Because of the availability of greatly improved CCDs, today an array of photodetectors, such as Indium Antimonide (InSb)
128 x 128, are used to form charge packets that are proportional to the light intensity of a target; and these packets are shifted to a detector point for detection, readout, multiplexing, and time delay and integration (TDI). In scanned IR systems TDI
is one of the most important functions performed by CCDs. In such a system the
scene is mechanically scanned across an array of detector elements. By using CCD
columns to shift the detector output signals (in the form of charge packets) along the
focal plane with the same speed as the mechanical scan moves the scene across the
array, the signal-to-noise ratio can be improved by the square root of the number of
detector elements in the TDI column.
Signal processing. The CCD can perform several analog and digital signal
processing functions, such as delay, multiplexing, demultiplexing, transversal
Chap. 6
References
261
filtering, recursive filtering, integration, analog memory, digital memory, and digital logic. Thus CCDs are being used widely in special applications for the very large
scale integration (VLSI) circuits.
Example 6-6-2:
a.
b.
c.
d.
Nmax = 2 x 10 12 cm- 2
;, = 3.9
d = 0.15 ,m
A = 0.5 x 10- 4 cm2
P = 0.67 mW
Solution
a. The insulator capacitance is
_
C; -
E; _
d-
F/
23 n
cm
10 12
J.6
10- 19
0.5
10- 4
16 pC
10 12 X I .6 X 10- 19
23 x 10- 9
14V
=-- =
nVQmax
0.67 x 10- 3
3 x 14 x 16 x 10- 12
I MHz
REFERENCES
[I] SHOCKLEY, W., A unipolar "field-effect" transistor. Proc. IRE, 40, No. I I, 1365- 1376,
November 1952.
[2) SCHOTTKY, W., Naturwiss, 26, 843, (1938).
[3] MEAD, C. A., Schottky barrier gate field-effect transistor. Proc. IEEE, 54, No. 2,
307-308, February 1966.
262
Chap. 6
[4] HOOPER, W.W., et al, An epitaxial GaAs field-effect transistor. Proc. IEEE, 55, No. 7,
1237-1238, July 1967.
[5] VANDERZIEL, A. et al., Gate noise in field-effect transistors at moderately high frequencies. Proc. IEEE, 51, No. 3, 461-467, March 1963.
[6] ZULEEG, R., and K. LEHOVEC, High frequency and temperature characteristics of GaAs
junction field-effect transistors in hot electron range. Proc. Symp. GaAs, Institute of
Physics Conf. Series No. 9, 240-245, 1970.
[7) LEHOVEC, K., and R. ZULEEG, Voltage-current characteristics of GaAs J-FEf's in the
hot electron range. Solid-State Electronics, Vol. 13, 1415-1426. Pergamon Press, London, 1970.
[8] DINGLE, R., et al., Electron mobilities in modulation-doped semiconductor heterojunction superlattices. Appl. Phys. Letters, Vol. 33, 665-667 (1978).
[9] ABE, M., et al., Recent advance in ultra-light speed HEMT technology. IEEE J. Quantum Electronics. QE-22, No. 9, 1870-1879, September 1986.
[10) TOGASHI, K., et al., Reliability of low-noise microwave HEMTs made by MOCVD. Microwave J., 123-132, April 1987.
[11) PAVLIDIS, D., and M. WEISS, The influence of device physical parameters on HEMT
large-signal characteristics. IEEE Trans. on Microwave Theory and Techniques, MTT36, No. 2, 239-249, February 1988.
[12) SzE, S. M., Physics of Semiconductor Devices, 2nd ed., p. 440, John Wiley & Sons,
New York, 1981.
[13] PARRILLO, L. C., VLSI Process Integration in VLSI Technology. Chapter II, ed. S. M.
Sze. McGraw-Hill Book Company, New York, 1983.
[14) ASAI, S., Semiconductor memory trend. Proc. IEEE, 74, No. 12, 1623-1635, December 1986.
[15) BOYLE, W. S., and G. E. SMITH, Charge couple semiconductor devices. Bell Syst. Tech.
J., 49, 587-593 (l 977).
[16] BOYLE, W. S., and G. E. SMITH, Charge-coupled devices-a new approach to MIS
device structures. IEEE Spectrum, 8, No. 7, 18027, July 1971.
[ 17) STECKL, A. J., et al., Application of charge-coupled devices to infrared detection and
imaging. Proc. IEEE, 63, No. I, 67-74, January 1975.
SUGGESTED READINGS
HEss, K., Advanced Theory of Semiconductor Devices. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1988.
IEEE Transactions on Electron Devices. Special issues on microwave solid-state devices.
ED-27, No. 2, February 1980.
ED-27, No. 6, June 1980.
ED-28, No. 2, February 1981.
ED-28, No. 8, August 1981.
IEEE Transactions on Microwave Theory and Techniques. Special issues on microwave solidstate devices.
MTT-21, No. II, November 1973.
MTT-24, No. 11, November 1976.
Chap. 6
263
Problems
PROBLEMS
JFETs
6-1. A silicon JFEf has a gate length of 5 , m. Calculate the maximum frequency of oscillation for the device.
6-2. Ann-channel silicon JFET has the following parameters:
Electron density:
Channel height:
Relative dielectric constant:
Nd = 2 x 10 17 cm- 3
a = 0.4 x 10- 4 cm
E, =
11.8
Nd = 5 x 10 17 cm- 3
Na = 8 X 10 18 cm- 3
11.8
a = 0.3 x 10- 4 cm
L = 6 x 10- 4 cm
Z = 45 x 10- 4 cm
J1.n = 520 cm 2 /Vs
vd = 8 v
V, = -1 V
E, =
Compute:
a. The pinch-off voltage in volts
b. The pinch-off current in mA
c. The built-in voltage in V
d. The drain current in mA
e. The saturation drain current at V,
f. The cut-off frequency in GHz
I V in mA
264
Chap. 6
Nd = 3 x 10 17 cm- 3
Na = 2 X !0 19 cm- 3
, = 11.8
a = 0.5 x 10- 4 cm
L = JO x I 0- 4 cm
Z = 20 x 10- 4 cm
vd = 1 v
V8 = -1.5 V
Determine:
Nd = 2 x !023 m- 3
a = 0.2 x 10- 6 m
11.8
L = 10 x 10- 6 m
Z = 60 x 10- 6 m
, =
vd
= 0 to 6 v
V8 = 0 to -3 V
5
Vs = 10 mis
1. The drain voltage Vd varies from 0 to 6 volts with an increment of 1 volt per step
2. The gate voltage V8 varies from 0 to -3 volts with a decrease of -0.5 volt per
step
3. The electron mobility, varies from 0.9 to 0.3 m 2/Vs with a decrease of 0.1 per
step
4. Use FI0.5 format for numerical outputs and Hollerith format for character outputs
5. Print the outputs in three columns for drain voltage Vd (volts), gate voltage V8
(volts), and drain current Id (mA)
MESFETs
6-7. A MESFET has a gate width of 5 ,m. Calculate the maximum frequency of oscillation
for the device.
Chap. 6
265
Problems
Nd = 2 x 10 17 cm- 3
E, = 13.10
a = 0.3 x 10- 4 cm
L = 10 x 10- 4 cm
Z = 50 x 10- 4 cm
vd = 1 v
Vg = -1.5 V
7
Vs = 10- cm/s
Compute:
L =
Z =
3 x 10 17 cm- 3
13.10
0.5 x 10- 4 cm
8 x 10- 4 cm
60 x 10- 4 cm
vd
6v
Nd
Er
a =
Vg = -2.0 V
7
Vs = I X I 0 cm/s
Calculate:
Electron concentration:
Channel height:
Relative dielectric constant:
Channel length:
Channel width:
Electron mobility:
Drain voltage:
Gate voltage:
Saturation drift velocity:
N = 2.38
x 1023 m- 3
a= 0.2 ,m
13.10
L = 10,m
Z = 60 ,m
, = 0.3 m2/V s
vd., = 6 volts
V8 = -3 volts
5
Vs = 10 mis
E, =
266
a.
b.
c.
d.
HEMTs
W = 140 m
v(z) = 3 X 105 mis
n(z)
6 x 10 15 m- 2
W = 100 m
v(z) = 4
x 105
mis
15
n(z) = 4 x 10 m- 2
Vrh
= 0.15
Nd
3 x 1024 m- 3
l/Jms
0.8 V
E,
= 4.43
Compute:
a. The conduction-band-edge difference between GaAs and AlGaAs
b. The sensitivity of the HEMT
6-17. A HEMT has the following parameters:
Threshold voltage:
Donor concentration:
AlGaAs dielectric constant:
Metal-semiconductor
Schottky barrier potential:
v,h
Nd
E,
l/Jms
0.12 v
5 x 1024 m- 3
= 4.43
=
= 0.8
Calculate:
a. The conduction-band-edge difference between GaAs and AlGaAs
b. The sensitivity of the HEMT
6-18. Describe the applications of a HEMT.
Chap. 6
Chap. 6
Problems
267
MOSFETs
6-19. A basic MOSFET is formed of Al metal, Si0 2 insulator, and Si semiconductor. The
insulator capacitance is 4 pf and the channel length Lis 12 ,m.
a. Determine the drain current Id sat in the saturation region with V,h = 2V and
Vgs = 4V for the enhancement mode.
b. Compute the transconductance gm for the same mode.
c. Calculate the drain current Id sat in the saturation region with V,h = -1. 5V and
Vgs = OV for the depletion mode.
d. Find the transconductance gm for the depletion mode.
6-20. An Al-Oxide-Si MOSFEf has an insulator capacitance of 3 pF and a channel length L
of 10 ,m. The gate voltage V,, is IO V and the threshold voltage is 1.5 V.
a. Determine the carrier drift velocity in real case.
b. Calculate the drain current.
c. Compute the carrier transit time.
d. Find the maximum operating frequency in GHz.
6-21. The insulator Si02 in a Si MOSFEf has a relative dielectric constant E;r of 11.8 and a
depth d of 0.08 ,m. The channel length L is 15 ,m and the channel depth Z is
150 , m. Calculate the insulator capacitance C;".
6-22. Compare the advantages and disadvantages of a GaAs MOSFEf with those of a Si
MOSFEf.
6-23. A certain p -channel MOSFET has the following parameters:
Doping concentration:
Relative dielectric constant:
Insulator relative dielectric
constant:
Insulator thickness:
Operating temperature:
Na = 2 x 10 17 cm- 3
Er = 11.8
4
d = 0.01 ,m
T = 320 K
E;, =
L = 5 ,m
Z = 10 ,m
d = 0.02 ,m
Vg = 8 V
v,h = 1.5 v
v, = 2 x 107 emfs
Er=
268
Chap. 6
6-25.
6-26.
6-27.
6-28.
6-29. A charge-coupled device has 484 (or 22 x 22 array) elements each with a transfer
inefficiency of 10- 4 and is clocked at a frequency of 100 KHz.
a. Determine the delay time between input and output.
b. Find the percentage of input charge appearing at the output terminal.
6-30. A surface-channel CCD is operated by a gate voltage VR of 10 volts. The insulator has a
relative dielectric constant e, of 6 and a depth d of 0.1 ,m. Determine the stored
charge density on this MOS capacitor.
6-31. A p-type surface-channel CCD (Al-Oxide-Si) has the following parameters:
Na = 10 14 cm -J
cPm,
= -0.85 eV
E;, =
W =I m
d=0.1,m
E;,
= 4
d = 0.3 ,m
A = 0.6 x 10- 4 cm 2
12
Nmax = 4 X 10
cm- 2
P = 0.8 mW
Chapter 7
Transferred Electron
Devices (TEDs)
10 INTRODUCTION
269
270
Chap. 7
operate with "hot" electrons whose energy is very much greater than the thermal energy. Because of these fundamental differences, the theory and technology of transistors cannot be applied to TEDs.
711 Background
After inventing the transistor, Shockley suggested in 1954 that two-terminal
negative-resistance devices using semiconductors may have advantages over transistors at high frequencies [ 1]. In 1961 Ridley and Watkins described a new method for
obtaining negative differential mobility in semiconductors [2]. The principle involved is to heat carriers in a light-mass, high-mobility subband with an electric field
so that the carriers can transfer to a heavy-mass, low-mobility, higher-energy subband when they have a high enough temperature. Ridley and Watkins also mentioned
that Ge-Si alloys and some III-V compounds may have suitable subband structures
in the conduction bands. Their theory for achieving negative differential mobility in
bulk semiconductors by transferring electrons from high-mobility energy bands to
low-mobility energy bands was taken a step further by Hilsum in 1962 [3]. Hilsum
carefully calculated the transferred electron effect in several III-V compounds and
was the first to use the terms transferred electron amplifiers (TEAs) and oscillators
(TEOs). He predicted accurately that a TEA bar of semi-insulating GaAs would be
operated at 373K at a field of 3200 V/cm. Hilsum's attempts to verify his theory experimentally failed because the GaAs diode available to him at that time was not of
sufficiently high quality.
It was not until 1963 that J. B. Gunn of IBM discovered the so-called Gunn effect from thin disks of n-type GaAs and n-type InP specimens while studying the
noise properties of semiconductors [4]. Gunn did not connect-and even immediately rejected-his discoveries with the theories of Ridley, Watkins, and Hilsum. In
1963 Ridley predicted [5] that the field domain is continually moving down through
the crystal, disappearing at the anode and then reappearing at a favored nucleating
Sec. 7.1
271
center, and starting the whole cycle once more. Finally, Kroemer stated [6] that the
origin of the negative differential mobility is Ridley-Watkins-Hilsum's mechanism
of electron transfer into the satellite valleys that occur in the conduction bands of
both the n-type GaAs and the n-type InP and that the properties of the Gunn effect
are the current oscillations caused by the periodic nucleation and disappearance of
traveling space-charge instability domains. Thus the correlation of theoretical predictions and experimental discoveries completed the theory of transferred electron
devices (TEDs).
712 Gunn Effect
A schematic diagram of a uniform n-type GaAs diode with ohmic contacts at the
end surfaces is shown in Fig. 7 -1-1. J. B. Gunn observed the Gunn effect in the
n-type GaAs bulk diode in 1963, an effect best explained by Gunn himself, who
published several papers about his observations [7 to 9). He stated in his first paper
[7] that
Above some critical voltage, corresponding to an electric field of 2000-4000 volts/cm,
the current in every specimen became a fluctuating function of time. In the GaAs specimens, this fluctuation took the form of a periodic oscillation superimposed upon the
pulse current. ... The frequency of oscillation was determined mainly by the specimen, and not by the external circuit. ... The period of oscillation was usually inversely proportional to the specimen length and closely equal to the transit time of electrons between the electrodes, calculated from their estimated velocity of slightly over
107 cm/s .... The peak pulse microwave power delivered by the GaAs specimens to a
matched load was measured. Value as high as 0.5 Wat l Gels, and 0.15 Wat 3 Gels,
were found, corresponding to 1-2% of the pulse input power.*
t' --
High-field domain
v = I Q5 m/s
::i.
0
Cathode -
Metal-coated
contact
/oo
4~
+
+
+
+
+
+ Anode
""l-75 ---1
m
272
Chap. 7
From Gunn's observation the carrier drift velocity is linearly increased from
zero to a maximum when the electric field is varied from zero to a threshold value.
When the electric field is beyond the threshold value of 3000 V/cm for the n-type
GaAs, the drift velocity is decreased and the diode exhibits negative resistance. This
situation is shown in Fig. 7-1-2.
2 x 10 7
/>
"'
>--------- ~------
>
--<~-Qo..~.2..
Y"
...,
107
.::::
I
I
I
I
I
I
I
c:
0
J
0
10
15
20
Field (kV/cm)
Figure 7-1-2 Drift velocity of electrons in n-type GaAs versus electric field.
(After 1. B. Gunn [8]; reprinted by permission of IBM, Inc.)
The current fluctuations are shown in Fig. 7-1-3. The current waveform was
produced by applying a voltage pulse of 16-V amplitude and 10-ns duration to a
specimen of n-type GaAs 2.5 x 10- 3 cm in length. The oscillation frequency was
4.5 GHz. The lower trace had 2 ns/cm in the horizontal axis and 0.23 A/cm in the
vertical axis. The upper trace was the expanded view of the lower trace. Gunn found
that the period of these oscillations was equal to the transit time of the electrons
through the specimen calculated from the threshold current.
Sec. 7.2
273
Gunn also discovered that the threshold electric field Eth varied with the length
and type of material. He developed an elaborate capacitive probe for plotting the
electric field distribution within a specimen of n-type GaAs of length L = 210 m
and cross-sectional area 3.5 x 10- 3 cm 2 with a low-field resistance of 16 fl. Current
instabilities occurred at specimen voltages above 59 V, which means that the
threshold field is
v
59
(7-1-1)
Eth = L =
x
_ x 102 = 2810 volts/cm
210
10 6
The fundamental concept of the Ridley-Watkins-Hilsum (RWH) theory is the differential negative resistance developed in a bulk solid-state III-V compound when either
a voltage (or electric field) or a current is applied to the terminals of the sample.
There are two modes of negative-resistance devices: voltage-controlled and currentcontrolled modes as shown in Fig. 7-2-l(a) and (b), respectively [5].
J
0
(a) Voltage-controlled mode
L
(b) Current-controlled mode
274
Chap. 7
High current
Low field
Low current
Figure 7-2-2 Diagrams of high field domain and high current filament. (From
B. K. Ridley [5 ]; reprinted by permission of the Institute of Physics.)
E1 Eo E2
Figure 7-2-3 Multiple values of current density for negative resistance. (From
B. K. Ridley [5 ]; reprinted by permission of the Institute of Physics.)
Sec. 7.2
275
}}lCll
::=
11
L.ower valley
Y}
Conduction
band
!J.F = 0.36 cV
Fx
= 1.43 eV
Forbidden
band
A
t:~s:-,
TABLE 7-21
} v~!~,~'"
Effective Mass
M,
Mobility
Separation
Valley
Lower
Upper
Mee= 0.068
Meu = 1.2
aE = 0.36 eV
aE = 0.36 eV
aE
the data for the two valleys in then-type GaAs and Table 7-2-2 shows the data for
two-valley semiconductors.
Electron densities in the lower and upper valleys remain the same under an
equilibrium condition. When the applied electric field is lower than the electric field
of the lower valley (E < Ee), no electrons will transfer to the upper valley as shown
in Fig. 7-2-5(a). When the applied electric field is higher than that of the lower valley and lower than that of the upper valley (Ee < E < Eu), electrons will begin to
transfer to the upper valley as shown in Fig. 7-2-S(b). And when the applied electric
TABLE 722
Gap energy
(at 300K)
E.(eV)
Separation energy
between two valleys
aE(eV)
Threshold
field
Eth(KV/cm)
Peak velocity
vp(I01 cmls)
Ge
GaAs
InP
0.80
1.43
1.33
2.3
3.2
10.5
1.4
2.2
2.5
CdTe
InAs
InSb
1.44
0.33
0.16
0.18
0.36
0.60
0.80
0.51
1.28
0.41
13.0
1.60
0.6
1.5
3.6
5.0
Semiconductor
Note: InP is a three-valley semiconductor: 0.60 eV is the separation energy between the middle and
lower valleys, 0.8 eV that between the upper and lower valleys.
276
v
w
00
(a)
v
.u/
0
Chap. 7
v
~u0
< t~
Figure 7-2-5
field is higher than that of the upper valley (Eu < E), all electrons will transfer to the
upper valley as shown in Fig. 7-2-S(c).
If electron densities in the lower and upper valleys are nc and nu , the conductivity of the n -type GaAs is
(7-2-2)
where e = the electron charge
=the electron mobility
n =nc + nu is the electron density
Example 7-2-1:
Electron density:
Electron density at lower valley:
Electron density at upper valley:
Temperature:
n = 10 18 cm- 3
ne = 10 10 cm- 3
nu = 108 cm- 3
T = 300K
Solution
u = e(enc + .n.)
= 1.6
1.28 mmhos
for ne
~nu
When a sufficiently high field Eis applied to the specimen, electrons are accelerated and their effective temperature rises above the lattice temperature. Furthermore, the lattice temperature also increases. Thus electron density n and mobility
are both functions of electric field E. Differentiation of Eq. (7-2-2) with respect to E
yields
(7-2-3)
Sec. 7.2
277
If the total electron density is given by n = n, + nu and it is assumed that c and /Lu
are proportional to P, where p is a constant, then
d
dE (nt
+ nu) =
dn
dE
dn,
dE
and
d
dP
rx dE
dE
dnu
dE
pEp-1
(7-2-4)
=0
(7-2-5)
P
,
p - oc pE
E
= -p
(7-2-6)
dnc
e(c - /Lu) dE
e(ncc
nuu)E
(7-2-7)
+-E
da
dE
(7-2-8)
da/dE
I+-a/E
(7-2-9)
-=a
Clearly, for negative resistance, the current density J must decrease with increasing
field E or the ratio of dJ!dE must be negative. Such would be the case only if the
right-hand term of Eq. (7-2-9) is less than zero. In other words, the condition for
negative resistance is
_ da/dE > l
(7-2-10)
a/E
Substitution of Eqs. (7-2-2) and (7-2-7) withf
p > 1
(7-2-11)
Note that the field exponent p is a function of the scattering mechanism and
should be negative and large. This factor makes impurity scattering quite undesirable
because when it is dominant, the mobility rises with an increasing field and thus p is
positive. When lattice scattering is dominant, however, p is negative and must depend on the lattice and carrier temperature. The first bracket in Eq. (7-2-11) must be
positive in order to satisfy unequality. This means that c > /Lu. Electrons must begin in a low-mass valley and transfer to a high-mass valley when they are heated by
the electric field. The maximum value of this term is unity-that is, c ~ /Lu. The
factor dnc/ dE in the second bracket must be negative. This quantity represents the
rate of the carrier density with field at which electrons transfer to the upper valley;
this rate depends on differences between electron densities, electron temperature,
and gap energies in the two valleys.
278
Chap. 7
...,
~
;;;
]th
v
I
I
""
"O
c
t
:l
l.i
Ju
Ultimate bulk
resistance
o~o--'--'-------"---L---------
E, E,h
Eu
Eu
Electric field E
(7-2-12)
Sec. 7.2
279
E""'u
- -!--:--------------~
I
--1
...,..-
: J...-I
--
..... - I
-'-----,l,--------.E
Qlo!::::,.:......J._.J.___ _ _ _ _ _
E,
Eu
Eth
Eu
Figure 7-2-7
dl
dE
dv
= qn dE
(7-2-13)
.," < 0
(7-2-14)
E""'
2x 10 1 ~----+--+-~A--____=::!~--11-----+--+---l--1----1----1
Experiment
Tl
0
~
~
I x I 0 7 L---'-.L.--'--J__------'--
Ci
IO
11
280
Example 7-2-2:
Chap. 7
Threshold field
Applied field
Device length
Doping concentration
Operating frequency
10 x 109 x 10 x 10- 6
105 m/sec
107 cm/sec
1020
109
= qnv = 1.6
= 3.2
10- 19
10
10- 5
x 106 A/m 2
= 320 A/cm 2
107
3200
Vd
In the last section we described how differential resistance can occur when an electric field of a certain range is applied to a multivalley semiconductor compound,
such as then-type GaAs. In this section we demonstrate how a decrease in drift velocity with increasing electric field can lead to the formation of a high-field domain
for microwave generation and amplification.
In the n-type GaAs diode the majority carriers are electrons. When a small
voltage is applied to the diode, the electric field and conduction current density are
uniform throughout the diode. At low voltage the GaAs is ohmic, since the drift velocity of the electrons is proportional to the electric field. This situation was shown
in Fig. 7-1-2. The conduction current density in the diode is given by
av
a = conductivity
(7-2-15)
Sec. 7.2
L
V
281
=
p =
applied voltage
charge density
v = drift velocity
U = unit vector
The current is carried by free electrons that are drifting through a background
of fixed positive charge. The positive charge, which is due to impurity atoms that
have donated an electron (donors), is sometimes reduced by impurity atoms that
have accepted an electron (acceptors). As long as the fixed charge is positive, the
semiconductor is n type, since the principal carriers are the negative charges. The
density of donors less the density of acceptors is termed doping. When the space
charge is zero, the carrier density is equal to the doping.
When the applied voltage is above the threshold value, which was measured at
about 3000 V/cm times the thickness of the GaAs diode, a high-field domain is
formed near the cathode that reduces the electric field in the rest of the material and
causes the current to drop to about two-thirds of its maximum value. This situation
occurs because the applied voltage is given by
=-
Ex dx
(7-2-16)
For a constant voltage V an increase in the electric field within the specimen must be
accompanied by a decrease in the electric field in the rest of the diode. The highfield domain then drifts with the carrier stream across the electrodes and disappears
at the anode contact. When the electric field increases, the electron drift velocity decreases and the GaAs exhibits negative resistance.
Specifically, it is assumed that at point A on the J-E plot as shown in Fig.
7-2-9(b) there exists an excess (or accumulation) of negative charge that could be
caused by a random noise fluctuation or possibly by a permanent nonuniformity in
doping in the n-type GaAs diode. An electric field is then created by the accumulated charges as shown in Fig. 7-2-9(d). The field to the left of point A is lower than
that to the right. If the diode is biased at point EA on the J-E curve, this situation implies that the carriers (or current) flowing into point A are greater than those flowing
out of point A, thereby increasing the excess negative space charge at A. Furthermore, when the electric field to the left of point A is lower than it was before, the
field to the right is then greater than the original one, resulting in an even greater
space-charge accumulation. This process continues until the low and high fields both
reach values outside the differential negative-resistance region and settle at points I
and 2 in Fig. 7-2-9(a) where the currents in the two field regions are equal. As a result of this process, a traveling space-charge accumulation is formed. This process,
of course, depends on the condition that the number of electrons inside the crystal is
large enough to allow the necessary amount of space charge to be built up during the
transit time of the space-charge layer.
The pure accumulation layer discussed above is the simplest form of spacecharge instability. When positive and negative charges are separated by a small distance, then a dipole domain is formed as shown in Fig. 7-2-10. The electric field in-
282
Chap. 7
0
0 E1
E2
EA
(a)
L ~---------\
(b)
"]
I
L
(c)
(E1)
Accumulation layer
formation completed
:~
I
I
I
I
f3
(d)
tI
Figure 7-2-9
Initial state
A
(e)
Formation of an electron
accumulation layer in GaAs. (After H.
Kroemer [ 16]; reprinted by permission of
IEEE, Inc.)
side the dipole domain would be greater than the fields on either side of the dipole in
Fig. 7-2-lO(c). Because of the negative differential resistance, the current in the
low-field side would be greater than that in the high-field side. The two field values
will tend toward equilibrium conditions outside the differential negative-resistance
region, where the low and high currents are the same as described in the previous
section. Then the dipole field reaches a stable condition and moves through the specimen toward the anode. When the high-field domain disappears at the anode, a new
dipole field starts forming at the cathode and the process is repeated.
Sec. 7.2
283
-1
f
l
(a)
Dipole
~
0
0
l
I-+(b)
E
E2
E1
0
0
( c)
ple increases above the threshold electric field and will drift with the carrier
stream through the device. When the electric field increases, the electron drift
velocity decreases and the GaAs diode exhibits negative resistance.
2. If additional voltage is applied to a device containing a domain, the domain
will increase in size and absorb more voltage than was added and the current
will decrease.
3. A domain will not disappear before reaching the anode unless the voltage is
dropped appreciably below threshold (for a diode with uniform doping and
area).
284
Chap. 7
1. Gunn oscillation mode: This mode is defined in the region where the product of
frequency multiplied by length is about 107 cm/s and the product of doping
multiplied by length is greater than l0 12/cm 2 In this region the device is unstable because of the cyclic formation of either the accumulation layer or the
high-field domain. In a circuit with relatively low impedance the device operates in the high-field domain mode and the frequency of oscillation is near the
intrinsic frequency. When the device is operated in a relatively high-Q cavity
and coupled properly to the load, the domain is quenched or delayed (or both)
before nucleating. In this case, the oscillation frequency is almost entirely determined by the resonant frequency of the cavity and has a value of several
times the intrinsic frequency.
Sec. 7.3
Modes of Operation
285
LSA
v;
e
~
.seo
///
"
"'CJ'"
::l
Quenched dornain
GUNN
J07
Transit-time
domain
_ _ L_ _ _ _ _ _
t::,'
Delayed domain
1012
I Ql3
JQ14
Figure 7-3-1 Modes of operation for Gunn diodes. (After J. A. Copeland [17];
reprinted by permission of IEEE, Inc.)
2. Stable amplification mode: This mode is defined in the region where the
product of frequency times length is about 107 cm/s and the product of doping
times length is between 10 11 and l0 12/cm2
3. LSA oscillation mode: This mode is defined in the region where the product of
frequency times length is above 107 cm/s and the quotient of doping divided by
frequency is between 2 x 104 and 2 x 105
4. Bias-circuit oscillation mode: This mode occurs only when there is either Gunn
or LSA oscillation, and it is usually at the region where the product of frequency times length is too small to appear in the figure. When a bulk diode is
biased to threshold, the average current suddenly drops as Gunn oscillation begins. The drop in current at the threshold can lead to oscillations in the bias
circuit that are typically 1 kHz to 100 MHz [18].
The first three modes are discussed in detail in this section. Before doing so, however, let us consider the criterion for classifying the modes of operation.
7-3-1 Criterion for Classifying the Modes
of Operation
The Gunn-effect diodes are basically made from an n-type GaAs, with the concentrations of free electrons ranging from 10 14 to 10 17 per cubic centimeter at room temperature. Its typical dimensions are 150 x 150 ,m in cross section and 30 ,m
Jong. During the early stages of space-charge accumulation, the time rate of growth
286
Chap. 7
where
(:J
(7-3-1)
= !._ =
<T
Anode
~
vt. 0) Q(x, t)
I
I
I
I
I
_,,,
H----+,c----~L-------~
11
11
11
/I
11
x=O
t!
t=O
i----x
>---
Vt
If Eq. (7-3-1) remains valid throughout the entire transit time of the spacecharge layer, the factor of maximum growth is given by
Growth factor
VTd
(7-3-2)
In Eq. (7-3-2) the layer is assumed to start at the cathode at t = 0, X = 0, and arrive at the anode at t = L/v and X = L. For a large space-charge growth, this factor must be larger than unity. This means that
n0 L
> - Ve1 n 1
(7-3-3)
This is the criterion for classifying the modes of operation for the Gunn-effect
2
diodes. For n-type GaAs, the value of v/(el "I) is about l0 12/cm , where I" I is
assumed to be 150 cm 2/V s.
Example 7-3-1:
Vd
l/Lnl
,
Sec. 7.3
Modes of Operation
Solution
287
EVd
elJLnl
= 1.19
=
105
x l0 16/m 2
1.19 x l0 12/cm 2
This means that the product of the doping concentration and the device length must be
n0 l > 1.19
l0 12/cm 2
Most Gunn-effect diodes have the product of doping and length (noL) greater than
10 12/cm 2 However, the mode that Gunn himself observed had a product noL that is
much less. When the product of noL is greater than l0 12/cm 2 in GaAs, the spacecharge perturbations in the specimen increase exponentially in space and time in accordance with Eq. (7-3-1). Thus a high-field domain is formed and moves from the
cathode to the anode as described earlier. The frequency of oscillation is given by
the relation [ 19]
f =
Vdom
Lerr
(7-3-4)
where Vc1om is the domain velocity and Leff is the effective length that the domain
travels from the time it is formed until the time that a new domain begins to form.
Gunn described the behavior of Gunn oscillators under several circuit
configurations [20]. When the circuit is mainly resistive or the voltage across the
diode is constant, the period of oscillation is the time required for the domain to
drift from the cathode to the anode. This mode is not actually typical of microwave
applications. Negative conductivity devices are usually operated in resonant circuits,
such as high-Q resonant microwave cavities. When the diode is in a resonant circuit,
the frequency can be tuned to a range of about an octave without loss of efficiency
[21].
As described previously, the normal Gunn domain mode (or Gunn oscillation
mode) is operated with the electric field greater than the threshold field (E > E1h).
The high-field domain drifts along the specimen until it reaches the anode or until
the low-field value drops below the sustaining field Es required to maintain Vs as
shown in Fig. 7-3-3. The sustaining drift velocity for GaAs is Vs = 107 cm/s. Since
the electron drift velocity v varies with the electric field, there are three possible domain modes for the Gunn oscillation mode.
Transit-time domain mode (JL = 107 cm/s). When the electron drift velocity
Vd is equal to the sustaining velocity Vs , the high-field domain is stable. In other
words, the electron drift velocity is given by
Vd
Vs
fL = 107 cm/s
(7-3-5)
Then the oscillation period is equal to the transit time-that is, To = r,. This situation is shown in Fig. 7-3-4(a). The efficiency is below 10% because the current is
collected only when the domain arrives at the anode.
288
Chap. 7
ll
v,=10 1 cm/s
E
Electric field E
11
-= T 1
( h) lklayL'd
lllllcll'
Tll >Tr
--------- -- F,11
------i--- - - - - - - - - - F,
Q...,,.__ _ _ _ _ ___.1_ _ _ _ _ _ _~
( c I ()uc11d1nl tnodc
Tn <Tr
- - - - - - - - - -- -
F111
- - - - - - - -- - - - - F,
(d I LSA
llHllk
Tl~
< Tl
Tif
-=-
3rd
Figure 7-3-4
Sec. 7.3
289
Modes of Operation
Delayed domain mode (106 emfs < fl < 107 emfs). When the transit time is
chosen so that the domain is collected while E < E1h as shown in Fig. 7-3-4(b), a
new domain cannot form until the field rises above threshold again. In this case, the
oscillation period is greater than the transit time-that is, To > T,. This delayed
mode is also called inhibited mode. The efficiency of this mode is about 20%.
Quenched domain mode (fl > 2 x 107 emfs). If the bias field drops below
the sustaining field Es during the negative half-cycle as shown in Fig. 7-3-4(c), the
domain collapses before it reaches the anode. When the bias field swings back above
threshold, a new domain is nucleated and the process repeats. Therefore the oscillations occur at the frequency of the resonant circuit rather than at the transit-time frequency. It has been found that the resonant frequency of the circuit is several times
the transit-time frequency, since one dipole does not have enough time to readjust
and absorb the voltage of the other dipoles [22, 23]. Theoretically, the efficiency of
quenched domain oscillators can reach 13% [22].
fl
5 x 106 emfs
(7-3-6)
(7-3-7)
fl
107 emfs
(7-3-8)
X 104
(7-3-9)
290
Chap. 7
Both the upper and lower boundaries of the LSA mode are indicated in Fig. 7-3-1.
The LSA mode is discussed further in Section 7-4.
7.3.4 Stable Amplification Mode (no L < 1012/cm 2 )
When the noL product of the device is less than about I0 12/cm 2 , the device exhibits
amplification at the transit-time frequency rather than spontaneous oscillation. This
situation occurs because the negative conductance is utilized without domain formation. There are too few carriers for domain formation within the transit time. Therefore amplification of signals near the transit-time frequency can be accomplished.
This mode was first observed by Thim and Barber [23]. Furthermore, Uenohara
showed that there are types of amplification depending on the fL product of the
device [24] as shown in Fig. 7-3-5.
JQ8
-=~
Quenched
domain
dd~'j~d
x
;;.,
(.)
;:i I 01
::>
O"
~1
First amplification
Relaxation oscillation
JQ6 ~~~~~~~~~~~--'-~-'--~-'--~...._....._..........
I 012
)QI!
)Q13
JQ14
The various modes of operation of Gunn diodes can be classified on the basis
of the times in which various processes occur. These times are defined as follows:
T,
Td
TK
To
=
=
Sec. 7.4
LSA Diodes
TABLE 731
Mode
Stable amplifier
Gun domain
291
Time relationship
To~ T,
TK $
T1
Doping level
n0 L < !0
11;,L > 1012
12
To= Tr
Quenched domain
TR$ T 1
To=
Delayed domain
n0 L > !0 12
Tr
TR$ T 1
noL >
To> T,
LSA
To< Tg
To> Td
2 x !04 <
1012
Nature of circuit
Nonresonant
Nonresonant;
constant voltage
Resonant;
finite impedance
Resonant;
finite impedance
Multiple resonances;
high impedance;
high de bias
74 LSA DIODES
The abbreviation LSA stands for the limited space-charge accumulation mode of the
Gunn diode. As described previously, if the product noL is larger than l0 12/cm 2 and
if the ratio of doping no to frequency f is within 2 x 105 to 2 x 104 s/cm3, the highfield domains and the space-charge layers do not have sufficient time to build up.
The magnitude of the RF voltage must be large enough to drive the diode below
threshold during each cycle in order to dissipate space charge. Also, the portion of
each cycle during which the RF voltage is above threshold must be short enough to
prevent the domain formation and the space-charge accumulation. Only the primary
accumulation layer forms near the cathode; the rest of the sample remains fairly homogeneous. Thus with limited space-charge formation the remainder of the sample
appears as a series negative resistance that increases the frequency of the oscillations
in the resonant circuit. Copeland discovered the LSA mode of the Gunn diode in
1966 [25]. In the LSA mode the diode is placed in a resonator tuned to an oscillation
frequency of
1
fo = (7-4-1)
To
The device is biased to several times the threshold voltage (see Fig. 7-4-1).
As the RF voltage swings beyond the threshold, the space charge starts building up at the cathode. Since the oscillation period To of the RF signal is less than the
domain-growth time constant T8 , the total voltage swings below the threshold before
the domain can form. Furthermore, since To is much greater than the dielectric relaxation time Td, the accumulated space charge is drained in a very small fraction of
the RF cycle. Therefore the device spends most of the RF cycle in the negative-resistance region, and the space charge is not allowed to build up. The frequency of oscillation in the LSA mode is independent of the transit time of the carriers and is determined solely by the circuit external to the device. Also, the power-impedance
product does not fall off as 1/ J6; thus the output power in the LSA mode can be
greater than that in the other modes.
The LSA mode does have limitations. It is very sensitive to load conditions,
temperatures, and doping fluctuations [26). In addition, the RF circuit must allow
the field to build up quickly in order to prevent domain formation. The power output
292
Chap. 7
vd
Q =
'E"'
u =
.,,
"c
vd
,.,
;::;
0
-.;
>
.:::
.:
dvd
" =- dE
Cl
,h
Eb
Electric field
...
..---RF signal
"
E
f.:
J_
Figure 7-4-1
de bias
= riVolo = T/ (ME,hL)(noevoA)
(7-4-2)
Sec. 7.5
293
lnP Diodes
Example 7-4-1:
T/ = 0.06
M = 3.5
E,h = 320 kV/m
l = 121-tm
no =
1021 m3
(10 21
1.6
10- 19
(3.5
1.5
105
320
103
10- 8 )
12
10- 6 )
15 lnP DIODES
When Gunn first announced his Gunn effect in 1963, the diodes he investigated were
of gallium arsenide (GaAs) and indium phosphide (InP). The GaAs diode was described earlier in the chapter. In this section the n-type InP diode is discussed. Both
the GaAs diode and the InP diode operate basically the same way in a circuit with de
voltage applied at the electrodes. In the ordinary Gunn effect in the n-type GaAs,
the two-valley model theory is the foundation for explaining the electrical behavior
of the Gunn effect. However, Hilsum proposed that indium phosphide and some alloys of indium gallium antimonide should work as three-level devices [27]. Figure
7-5-1 shows the three-valley model for indium phosphide.
It can be seen that InP, besides having an upper-valley energy level and a
lower-valley energy level similar to the model shown in Fig. 7-2-4 for n-type GaAs,
also has a third middle-valley energy level. In GaAs the electron transfer process
from the lower valley to the upper valley is comparatively slow. At a particular
voltage above threshold current flow consists of a larger contribution of electrons
from the lower valley rather than from the upper valley. Because of this larger contribution from the lower-energy level, a relatively low peak-to-valley current ratio
results, which is shown in Fig. 7-5-2.
As shown in Fig. 7-5-3, the InP diode has a larger peak-to-valley current ratio
because an electron transfer proceeds rapidly as the field increases. This situation occurs because the coupling between the lower valley and upper valley in InP is
weaker than in GaAs. The middle-valley energy level provides the additional energy
loss mechanism required to avoid breakdown caused by the high energies acquired
by the lower-valley electrons from the weak coupling.It can be seen from Fig. 7-5-1
294
Chap. 7
Strong coupling
Conduction
band
Eg
= 1.33 eV
Figure 7-5-1
'Eu"'
Valence
band
'E"'u
Peak current
Tl
Valley current
0
Qj
T5
0
Qj
>
>
i::
0
gu
i::
!:::
u
<lJ
iii
Forbidtkn
gap
Peak current
iii
0
2 1h 4
IO
5
IO
Electric field (kV /cm)
Figure 7-5-3
tio for InP.
15
that the lower valley is weakly coupled to the middle valley but strongly coupled to
the upper valley to prevent breakdown. This situation ensures that under normal operating conditions electrons concentrate in the middle valley. Because InP has a
greater energy separation between the lower valley and the hearest energy levels, the
thermal excitation of electrons has less effect, and the degradation of its peak-tovalley current ratio is about four times less than in GaAs [28].
The mode of operation of InP is unlike the domain oscillating mode in which a
high-field domain is formed that propagates with a velocity of about 107 cm/s. As a
result, the output current waveform of an InP diode is transit-time dependent. This
mode reduces the peak-to-valley current ratio so that efficiency is reduced. For this
40
30
E
::t
.::
~
25
20
<I.I
.>:
.g
'.;::;
...,,,
15
<'.Jo
"-'."'?;
;>,
.;::
<I.I
~u
"'"~0
'
IO
':> /
<:
+
/0>
0'--~~~~'--~~....__~"--~~.>.-~"--'-----
IO
I5
20
30
Frequency of oscillator in GHz
40
reason, an operating mode is usually sought where charge domains are not formed.
The three-valley model of InP inhibits the formation of domains because the electron
diffusion coefficient is increased by the stronger coupling [28). From experiments
performed by Taylor and Colliver [29) it was determined that epitaxial InP oscillators operate through a transit-time phenomenon and do not oscillate in a bulk mode
of the LSA type. From their findings it was determined that it is not appropriate to
attempt to describe the space-charge oscillations in InP in terms of modes known to
exist in GaAs devices. Taylor and Colliver also determined that the frequencies obtained from a device were dependent on the active-layer thickness. The InP oscillator could be tuned over a large frequency range, bounded only by the thickness, by
adjusting the cavity size. Figure 7-5-4 shows the frequency ranges for the different
active-layer thicknesses and lines of constant electron velocity [28, 29). It can be
seen from the figure that only a few InP devices operate in the domain formation
area. In each case on the graph, the maximum efficiency occurs at about midband
[29). Table 7-5-1 summarizes the highest power and efficiencies for InP diodes.
TABLE 751
Frequency
(GHz)
5.5
8.5
10.75
13.8
15
18
18
25
26
29.4
33
37
28
11
II
5.4
5.4
5.4
Power
(W)
Operation
Efficiency
(%)
Reference
3.05
0.95
1.33
0.50
1.13
1.05
0.20
0.65
0.15
0.23
0.10
0.01
pulsed
pulsed
pulsed
pulsed
pulsed
pulsed
14.7
7.0
12.0
6.3
15
4.2
10.2
2.6
6
2
0.4
1
[28]
[29]
[28]
[29]
[28]
[29]
[28]
[29]
[28]
[29]
[29]
[28]
cw
pulsed
cw
pulsed
pulsed
cw
296
Chap. 7
76 CdTe DIODES
The Gunn effect, first observed by Gunn as a time variation in the current through
samples of n -type GaAs when the voltage across the sample exceeded a critical
value, has since been observed inn-type InP, n-type CdTe, alloys of n-GaAs and nGaP, and in InAs. In n-type cadmium telluride (CdTe), the Gunn effect was first
seen by Foyt and McWhorter [30], who observed a time variation of the current
through samples 250 to 300 ,m long with a carrier concentration of 5 x l0 14/cm 3
and a room temperature mobility of 1000 cm2/V s. Ludwig, Halsted, and Aven
[31] confirmed the existence of current oscillations in n-CdTe, and Ludwig has further reported studies of the Gunn effect in CdTe over a wider range of sample doping
levels and lengths [32]. It has been confirmed that the same mechanism-the fieldinduced transfer of electrons to a higher conduction band minimum (Gunn effect)applies in CdTe just as it does in GaAs. From the two-valley model theory in CdTe,
as in GaAs, the (000) minimum is the lowest in energy. The effective mass meff =
0.11 m (electron mass) and the intrinsic mobility, = 1100 cm 2/V sat room temperature. Hilsum has estimated that ( 111) minima are the next lowest in energy, being 0.51 eV higher than (000) minimum [33). In comparing the Gunn effect in CdTe
to that in GaAs, a major difference is the substantially higher threshold field, about
13 kV/cm for CdTe compared with about 3 kV/cm for GaAs [34]. Qualitatively, the
higher threshold can be thought of as associated with the relatively strong coupling
of the electrons to longitudinal optical phonons, which limits the mobility-and
hence the rate of energy acquisition from the applied field-and also provides an
efficient mechanism for transferring energy to the lattice, thereby minimizing the kinetic energy in the electron distribution.
The ratio of peak-to-valley current is another parameter of interest. In CdTe,
as in GaAs, the spike amplitude can be as large as 50% of the maximum total current. A similar maximum efficiency for CdTe and GaAs can be expected. Since the
domain velocities in CdTe and GaAs are approximately equal, samples of the same
length will operate at about the same frequency in the transit-time mode. The high
threshold field of CdTe combined with its poor thermal conductivity creates a heating problem. If sufficiently short pulses are used so that the heat can be dissipated,
however, the high operating field of the sample can be an advantage.
As described earlier in this section, if the applied field is less than threshold the
specimen is stable. If, however, the field is greater than threshold, the sample is unstable and divides up into two domains of different conductivity and different electric field but the same drift velocity. Figure 7-7-1 shows the stable and unstable regions.
At the initial formation of the accumulation layer, the field behind the layer decreases and the field in front of it increases. This process continues as the layer travels from the cathode toward the anode. As the layer approaches the anode, the field
Sec. 7.7
v
297
St~bk
Unstahk
. --+-t-.
rcg1011
I
region
I
I
Figure 7-7-1
velocity.
Electric field E
behind it begins to increase again; and after the layer is collected by the anode, the
field in the whole sample is higher than threshold. When the high-field domain disappears at the anode, a new dipole field starts forming again at the cathode and the
process repeats itself. Since current density is proportional to the drift velocity of
the electrons, a pulsed current output is obtained. The oscillation frequency of the
pulsed current is given by
f=~
(7-7-1)
Leff
where vd is the velocity of the domain or approximately the drift velocity of the electrons and Leff is the effective length that the domain travels. Experiments have shown
that the n-type GaAs diodes have yielded 200-W pulses at 3.05 GHz and 780-mW
CW power at 8. 7 GHz. Efficiencies of 29% have been obtained in pulsed operation
at 3.05 GHz and 5.2% in CW operation at 24.8 GHz. Predictions have been made
that 250-kW pulses from a single block of n -type GaAs are theoretically possible up
to 100 GHz.
The source generation of solid-state microwave devices has many advantages
over the vacuum tube devices they are beginning to replace. However, at present
they also have serious drawbacks that could prevent more widespread application.
The most important disadvantages are:
1.
2.
3.
4.
These problems are common to both avalanche diodes and transferred electron
devices [35].
Figure 7-7-2 shows the latest state-of-the-art performance for GaAs and InP
Gunn diodes [36]. The numbers adjacent to the data points indicate efficiency in percent. Gunn diode oscillators have better noise performance than IMPATTs. They are
298
Chap. 7
NUMBERS INDICATE
EFFICIENCY (%)
""'
0 30
~020
10
1/1
oY,
GaAs IMPATT
Si IMPATT
0.1
DATA FROM PUBLISHED RESULTS:
HUGHES
MA/COM
RAYTHEON
TRW
VARIAN
10
Figure 7-7-2 State-of-the-art performance for GaAs and InP Gunn diodes.
so
100
FREQUENCY (GHz)
500
Inc.)
used as local oscillators for receivers and as primary sources where CW powers of
up to 100 mW are required. InP Gunn diodes have higher power and efficiency than
GaAs Gunn diodes.
Sec. 7.7
299
In contrast to the stable amplifier, the Gunn-effect diode must oscillate at the
transit-time frequency while it is amplifying at some other frequency. The value of
n0 L must be larger than l0 12/cm 2 in order to establish traveling domain oscillations;
hence substantially larger output power can be obtained. Because of the presence of
high-field domains, this amplifier is called a traveling domain amplifier (TDA).
Although a large number of possible amplifier circuits exist, the essential feature of each is to provide both a broadband circuit at the signal frequency and a
short circuit at the Gunn oscillation frequency. In order to maintain stability with respect to the signal frequency, the Gunn diode must see a source admittance whose
real part is larger than the negative conductance of the diode. The simplest circuit
satisfying this condition is shown in Fig. 7-7-3. An average gain of 3 dB was exhibited between 5.5 and 6.5 GHz.
Gunn diodes have been used in conjunction with circulator-coupled networks
in the design of high-level wideband transferred electron amplifiers that have a
voltage gain-bandwidth product in excess of IO dB for frequencies from 4 to about
16 GHz. Linear gains of 6 to 12 dB per stage and saturated-output-power levels in
excess of 0. 5 W have been realized [39]. Table 7 -7-1 lists the performance of several amplifiers that have been designed since 1970.
Pulse generator
Bias
@-a
50-n coaxial
Circulator
TABLE 771
Frequency
band
3-dB bandwidth
(GHz)
Power gain
(dB)
Efficiency
4.5- 8.0
7.5-10.75
8.0-12.0
12.0-16.0
13.0-15.0
8
12
6
6
8
3
1.65
1.8
1.5
0.36
3
2.3
2.5
2.5
2
Ku
Source: After B.S. Perlman et al. [39]; reprinted by permission of IEEE, Inc.
(%)
300
Chap. 7
REFERENCES
[I] SHOCKLEY, W., Negative resistance arising from transit time in semiconductor diodes.
Bell System Tech. J., 33, 799-826, July 1954.
[2] RIDLEY, B. K., and T. B. WATKINS, The possibility of negative resistance effects in semiconductors. Proc. Phys. Soc., 78, 293-304, August 1961.
[3] HlLSUM, C., Transferred electron amplifiers and oscillators. Proc. IEEE, 50, 185-189,
February 1962.
[4] GUNN, J. B., Microwave oscillations of current in III-V semiconductors. Solid-state
Communications, 1, 89-91, September 1963.
[5] RIDLEY, B. K., Specific negative resistance in solids. Proc. Phys. Soc. (London), 82,
954-966, December 1963.
(6] KROEMER, H., Theory of the Gunn effect. Proc. IEEE, 52, 1736 (1964).
[7] GUNN, J. B., Microwave oscillations of current in III-V semiconductors. Solid-state
Communications, 1, 88-91 (1963).
(8] GUNN, J. B., Instabilities of current in III-V semiconductors. IBM J. Res. Develop., 8,
141-159, April 1964.
[9] GUNN, J. B., Instabilities of current and of potential distribution in GaAs and InP. 7th
Int. Conj. on Physics of Semiconductor "Plasma Effects in Solids," 199-207, Tokyo,
1964.
[10] KROEMER, H., Proposed negative-mass microwave amplifier. Physical Rev., 109, No. 5.
1856, March 1, 1958.
[11] KROEMER, H., The physical principles of a negative-mass amplifier. Proc. IRE, 47,
397-406, March 1959.
[12] COPELAND, J. A., Bulk negative-resistance semiconductor devices. IEEE Spectrum, No.
5, 40, May 1967.
[13] BUTCHER, P. N., and W. FAWCETT, Calculation of the velocity-field characteristics of gallium arsenide. Appl. Phys. Letters, 21, 498 (1966).
[14] CONWELL, E. M. and M. 0. VASSELL, High-field distribution function in GaAs. IEEE
Trans. on Electron Devices, ED-13, 22 (1966).
[15] RUCH, J. G., and G. S. KINO, Measurement of the velocity-field characteristics of gallium arsenide. Appl. Phys. Letters, 10, 50 (1967).
[16] KROEMER, H., Negative conductance in semiconductors. IEEE Spectrum, 5, No. 1, 47,
January 1968.
[17] COPELAND, J. A., Characterization of bulk negative-resistance diode behavior. IEEE
Trans. on Electron Devices, ED-14, No. 9, 436-441, September 1967.
[18] ELLIOTT, B. J., J.B. GUNN, and J.C. McGRODDY, Bulk negative differential conductivity
and traveling domains inn-type germanium. Appl. Phys. Letters, 11, 253 (1967).
[19] COPELAND, J. A., Stable space-charge layers in two-valley semiconductors. J. Appl.
Phys., 37, No. 9, 3602, August 1966.
[20] GUNN, J. B., Effect of domain and circuit properties on oscillations in GaAs. IBM J.
Res. Develop., 310-320, July 1966.
[21] HOBSON, G. S., Some properties of Gunn-effect oscillations in a biconical cavity. IEEE
Trans. on Electron Devices, ED-14, No. 9, 526-531, September 1967.
[22] THIM, H. W., Computer study of bulk GaAs devices with random one-dimensional doping fluctuations. J. Appl. Phys., 39, 3897 (1968).
Chap. 7
Suggested Readings
301
[23] THIM, H. W., and M. R. BARBER, Observation of multiple high-field domains in n-GaAs.
Proc. IEEE, 56, I IO {1968).
[24] UENOHARA, M., Bulk gallium arsenide devices. Chapter 16 in H. A. Watson (Ed.), Microwave Semiconductor Devices and Their Circuit Application. McGraw-Hill Book
Company, New York, 1969.
[25] COPELAND, J. A., CW operation of LSA oscillator diodes-44 to 88 GHz. Bell System
Tech. J., 46, 284-287, January 1967.
[26] WILSON, W. E., Pulsed LSA and TRAPATT sources for microwave systems. Microwave
J., 14, No. 8, August 1971, 87-90.
[27] G. B. L., Three-level oscillator in indium phosphide. Physics Today, 23, 19-20, December 1970.
[28] COLLIVER, D., and B. PREw, Indium phosphide: Is it practical for solid state microwave
sources? Electronics, lI0-113, April IO, 1972.
[29] TAYLOR, B. C., and D. J. COLLIVER, Indium phosphide microwave oscillators. IEEE
Trans. on Electron Devices, ED-18, No. IO, 835-840, October 1971.
[30] FoYT, A.G., and A. L. MCWHORTER, The Gunn effect in polar semiconductors. IEEE
Trans. on Electron Devices, ED-13, 79-87, January 1966.
[31] LUDWIG, G. W., R. E. HALSTED, and M. AVEN, Current saturation and instability in
CdTe and ZnSe. IEEE Trans. on Electron Devices, ED-13, 671, August-September
1966.
[32] LUDWIG, G. W., Gunn effect in CdTe. IEEE Trans. on Electron Devices, ED-14, No. 9,
547-551, September 1967.
[33] BUTCHER, P. N., and w. FAWCETT, Proc. Phys. Soc. (London), 86, 1205 (1965).
[34] OLIVER, M. R., and A.G. FoYT, The Gunn effect in n-CdTe. IEEE Trans. on Electron
Devices, ED-14, No. 9, 617-618, September 1967.
[35] HILSUM, C., New developments in transfered electron effects. Proc. 3rd Conf. on High
Frequency Generation and Amplification: Devices and Applications. August 17-19,
197 l, Cornell University.
[36] HIESLMAIR, H., ET AL., State of the art of solid-state and tube transmitters. Microwave
J., 26, No. IO, 46-48, October 1983.
[37] HAKKI, B. W., GaAs post-threshold microwave amplifier, mixer, and oscillator. Proc.
IEEE (Letters), 54, 299-300, February 1966.
[38] THIM, H. W., Linear microwave amplification with Gunn oscillators. IEEE Trans. on
Electron Devices, ED-14, No. 9, 520-526, September 1967.
SUGGESTED READINGS
EASTMAN, L. F., Gallium Arsenide Microwave Bulk and Transit-Time Devices. Artech House,
Dedham, Mass., 1973.
MILNES, A. G., Semiconductor Devices and Integrated Electronics. Van Nostrand Reinhold
Company, New York, 1980.
SooHOO, R. F., Microwave Electronics. Addison-Wesley Publishing Company, Reading,
Mass., 1971.
SzE, S. M., Physics of Semiconductor Devices, 2nd ed. John Wiley & Sons, New York, 1981.
302
Chap. 7
PROBLEMS
Transferred Electron Devices (TEDs)
7-1.
7-2.
7-3.
7-4.
7-5.
7-6.
7-7.
a. Spell out the following abbreviated terms: LSA, InP, and CdTe.
b. Describe in detail the principles of the following terms: Gunn effect, high-field domain theory, two-valley theory, and three-valley theory.
c. Discuss the differences between transferred electron devices and avalanche transittime devices.
Describe the operating principles of tunnel diodes, Gunn diodes, and LSA diodes.
Derive Eq. (7-2-11).
Describe the modes of operation for Gunn diodes.
Describe the Ridley-Watkins-Hilsum theory.
The LSA oscillation mode is defined between 2 x 104 and 2 x 105 ratios of doping
over frequency as shown in Fig. 7-3-1. Derive Eqs. (7-3-7) and (7-3-9).
For a transit-time domain mode, the domain velocity is equal to the carrier drift velocity and is about 107 cm/s. Determine the drift length of the diode at a frequency of
8 GHz.
Chapter 8
Avalanche Transit-Time
Devices
BO INTRODUCTION
Avalanche transit-time diode oscillators rely on the effect of voltage breakdown
across a reverse-biased p-n junction to produce a supply of holes and electrons. Ever
since the development of modern semiconductor device theory scientists have speculated on whether it is possible to make a two-terminal negative-resistance device.
The tunnel diode was the first such device to be realized in practice. Its operation depends on the properties of a forward-biased p-n junction in which both the p and n
regions are heavily doped. The other two devices are the transferred electron devices
and the avalanche transit-time devices. In this chapter the latter type is discussed.
The transferred electron devices or the Gunn oscillators operate simply by the
application of a de voltage to a bulk semiconductor. There are no p-n junctions in
this device. Its frequency is a function of the load and of the natural frequency of the
circuit. The avalanche diode oscillator uses carrier impact ionization and drift in the
high-field region of a semiconductor junction to produce a negative resistance at microwave frequencies. The device was originally proposed in a theoretical paper by
Read [1] in which he analyzed the negative-resistance properties of an idealized n+ p-i-p+ diode. Two distinct modes of avalanche oscillator have been observed. One is
the IMPATT mode, which stands for impact ionization avalanche transit-time operation. In this mode the typical de-to-RF conversion efficiency is 5 to 10%, and frequencies are as high as 100 GHz with silicon diodes. The other mode is the TRAPATT mode, which represents trapped plasma avalanche triggered transit operation.
Its typical conversion efficiency is from 20 to 60%.
Another type of active microwave device is the BARITT (barrier injected
transit-time) diode [2]. It has long drift regions similar to those of IMPATT diodes.
The carriers traversing the drift regions of BARITT diodes, however, are generated
303
304
Chap. 8
by minority carrier injection from forward-biased junctions rather than being extracted from the plasma of an avalanche region. Several different structures have
been operated as BARITT diodes, such as p-n-p, p-n-v-p, p-n-metal, and metal-nmetal. BARITT diodes have low noise figures of 15 dB, but their bandwidth is relatively narrow with low output power.
81 READ DIODE
81 1 Physical Description
The basic operating principle of IMPATT diodes can be most easily understood by
reference to the first proposed avalanche diode, the Read diode [l]. The theory of
this device was presented by Read in 1958, but the first experimental Read diode
was reported by Lee et al. in 1965 [3]. A mode of the original Read diode with a
doping profile and a de electric field distribution that exists when a large reverse bias
is applied across the diode is shown in Fig. 8-1-1.
vdc
Inactive
region
Avalanche
region
(a)
Silicon structure
n+
i (or v)
p+
E(x)
'O
""'
Ell
Ell
1.;:
(b)
Field distribution
u:l"
I
I
I
I
I
ee I
I
I
e
e
e
I
I
Ell
I
I
.\'
Distance
1020
M
(c)
Doping profile
.g
"'
l:::
c
5 x 1016
I
I
I
I
...:.>
1013
_ _ J__
0
Distance
Figure 8-1-1
Read diode.
Sec. 8.1
Read Diode
305
The Read diode is an n+ -p-i-p+ structure, where the superscript plus sign denotes very high doping and the i or v refers to intrinsic material. The device consists
essentially of two regions. One is the thin p region at which avalanche multiplication
occurs. This region is also called the high-field region or the avalanche region. The
other is the i or v region through which the generated holes must drift in moving to
the p+ contact. This region is also called the intrinsic region or the drift region. The
p region is very thin. The space between the n+ -p junction and the i-p+ junction is
called the space-charge region. Similar devices can be built in the p+ -n-i-n+ structure, in which electrons generated from avalanche multiplication drift through the i
region.
The Read diode oscillator consists of an n+ -p-i-p+ diode biased in reverse and
mounted in a microwave cavity. The impedance of the cavity is mainly inductive and
is matched to the mainly capacitive impedance of the diode to form a resonant circuit. The device can produce a negative ac resistance that, in turn, delivers power
from the de bias to the oscillation.
81 2 Avalanche Multiplication
When the reverse-biased voltage is well above the punchthrough or breakdown
voltage, the space-charge region always extends from the n+ -p junction through the
p and i regions to the i-p+ junction. The fixed charges in the various regions are
shown in Fig. 8-1-l(b). A positive charge gives a rising field in moving from left to
right. The maximum field, which occurs at the n+ -p junction, is about several hundred kilovolts per centimeter. Carriers (holes) moving in the high field near then+ -p
junction acquire energy to knock valence electrons into the conduction band, thus
producing hole-electron pairs. The rate of pair production, or avalanche multiplication, is a sensitive nonlinear function of the field. By proper doping, the field can be
given a relatively sharp peak so that avalanche multiplication is confined to a very
narrow region at the n+ -p junction. The electrons move into the n+ region and the
holes drift through the space-charge region to the p+ region with a constant velocity
7
VJ of about 10 emfs for silicon. The field throughout the space-charge region is
above about 5 kV/cm. The transit time of a hole across the drift i-region L is given
by
T
=-
(8-1-1)
VJ
= applied voltage
306
Chap. 8
=
=
=
=
(8-1-lb)
resistivity
electron mobility
semiconductor permittivity
maximum breakdown of the electric field
0,.
.5
..,.,,
100
"'
01)
"'
0,.
"
~
.,,,c
"""'"'
IO
ds
1014
1015
1016
1011
1018
Sec. 8.1
Read Diode
307
(a)
Space-charge region
Structure
n+
----L---j
(b)
Field distribution
(c)
Applied ac voltage
I0
WI
I
I
I
I
I
I
I
I
I
I
I
/ 0 (I)
(d)
I
1
'--~___.~~-.i.~---"--~~-L-~-'-~~
Figure 8-1-3 Field, voltage, and currents in Read diode. (After Read [ J];
reprinted by permission of the Bell System, AT&T Co.)
When the holes generated at the n+ -p junction drift through the space-charge
region, they cause a reduction of the field in accordance with Poisson's equation:
dE
dx
_!!_
(8-1-2)
I,(t)
= ~ = v~Q
(8-1-3)
308
where Q
vd
Chap. 8
It can be seen that the induced current I,(t) in the external circuit is equal to
the average current in the space-charge region. When the pulse of hole current Io(t)
is suddenly generated at the n+ -p junction, a constant current I,(t) starts flowing in
the external circuit and continues to flow during the time T in which the holes are
moving across the space-charge region. Thus, on the average, the external current
f,(t) because of the moving holes is delayed by T /2 or 90 relative to the pulsed carrier current Io(t) generated at then+ -p junction. Since the carrier fo(t) is delayed by
one-quarter of a cycle or 90 relative to the ac voltage, the external current f,(t) is
then delayed by 180 relative to the voltage as shown in Fig. 8-l-3(d). Therefore the
cavity should be tuned to give a resonant frequency as
27Tf
7T
=-
Then
f = J_ = .!}:!_
2T
2L
(8-1-4)
Since the applied ac voltage and the external current I,(t) are out of phase by 180,
negative conductance occurs and the Read diode can be used for microwave oscillation and amplification. For example, taking vd = 107 emfs for silicon, the optimum
operating frequency for a Read diode with an i-region length of 2.5 ,mis 20 GHz.
B 14 Output Power and Quality Factor Q
The external current I,(t) approaches a square wave, being very small during the
positive half cycle of the ac voltage and almost constant during the negative half cycle. Since the direct current Id supplied by the de bias is the average external current
or conductive current, it follows that the amplitude of variation of I,(t) is approximately equal to Id. If Va is the amplitude of the ac voltage, the ac power delivered is
found to be
W/unit area
(8-1-5)
(8-1-6)
Since the Read diode supplies ac energy, it has a negative Qin contrast to the positive Q of the cavity. At the stable operating point, the negative Q of the diode is
equal to the positive Q of the cavity circuit. If the amplitude of the ac voltage increases, the stored energy, or energy of oscillation, increases faster than the energy
Sec. 8.2
309
IMPATT Diodes
delivered per cycle. This is the condition required in order for stable oscillation to be
possible.
B2 IMPATT DIODES
B21 Physical Structures
A theoretical Read diode made of an n+ -p-i-p+ or p+ -n-i-n+ structure has been analyzed. Its basic physical mechanism is the interaction of the impact ionization
avalanche and the transit time of charge carriers. Hence the Read-type diodes are
called IMPATT diodes. These diodes exhibit a differential negative resistance by two
effects:
1. The impact ionization avalanche effect, which causes the carrier current fo(t)
and the ac voltage to be out of phase by 90
2. The transit-time effect, which further delays the external current I.(t) relative
to the ac voltage by 90
The first IMPATT operation as reported by Johnston et al. [4] in 1965, however, was obtained from a simple p-n junction. The first real Read-type IMPATT
diode was reported by Lee et al. [3], as described previously. From the small-signal
theory developed by Gilden [5] it has been confirmed that a negative resistance of
the IMPATT diode can be obtained from a junction diode with any doping profile.
Many IMPATT diodes consist of a high doping avalanching region followed by a
drift region where the field is low enough that the carriers can traverse through it
without avalanching. The Read diode is the basic type in the IMPATT diode family.
The others are the one-sided abrupt p-n junction, the linearly graded p-n junction
(or double-drift region), and the p-i-n diode, all of which are shown in Fig. 8-2-1.
The principle of operation of these devices, however, is essentially similar to the
mechanism described for the Read diode.
B22 Negative Resistance
Small-signal analysis of a Read diode results in the following expression for the real
part of the diode terminal impedance [5]:
2L 2
R =Rs + - VdEsA
where Rs
2/
1 - w w,
vd
L
A
cos()
()
(8-2-1)
310
_I
Chap. 8
"'I 1020
E
~
i:
Doping profile
.g
....
"'
1016
=
OJ
i:
u
0
p+
3 m
11
n+
"'I
E
~
i:
Doping profile
.g
"'....
=
"
u
i::
0
3 m
p+
1020
2i:
Doping profile
.g
;;:"'
;::;
i:
1013
0
3 m
w-
(8-2-2)
Vd
(2a 'vdlo)
EsA
112
(8-2-3)
Sec. 8.2
IMPATT Diodes
311
In Eq. (8-2-3) the quantity a' is the derivative of the ionization coefficient with respect to the electric field. This coefficient, the number of ionizations per centimeter
produced by a single carrier, is a sharply increasing function of the electric field.
The variation of the negative resistance with the transit angle when w > Wr is plotted in Fig. 8-2-2. The peak value of the negative resistance occurs near () = 'TT. For
transit angles larger than 'TT and approaching 37T /2, the negative resistance of the
diode decreases rapidly. For practical purposes, the Read-type IMPATT diodes work
well only in a frequency range around the 'TT transit angle. That is,
f = J_ = .!!.:!__
2T
rr
2rr
3rr
Transit angle ()
(8-2-4)
2L
Figure 8-2-2
transit angle.
(8-2-6)
312
Chap. 8
EsA
(8-2-8)
= 1 yield
(8-2-9)
It is interesting to note that this equation is identical to Eq. (5-1-60) of the powerfrequency limitation for the microwave power transistor. The maximum power that
can be given to the mobile carriers decreases as 1/ f. For silicon, this electronic
limit is dominant at frequencies as high as 100 GHz. The efficiency of the IMPATT
diodes is given by
(8-2-10)
NUMBERS INDICATE
EFFICIENCY (%)
""
0 30
~ 020
10
1/1
GaAs IMPATT
10
!a:
o~
Si IMPATT
20
10 0 18
10
1
1/12
0.1
DATA FROM PUBLISHED RESULTS:
HUGHES
MA/COM
RAYTHEON
TRW
VARIAN
10
50
FREQUENCY (GHz)
100
500
Sec. 8.2
IMPATT Diodes
313
For an ideal Read-type IMPATT diode, the ratio of the ac voltage to the applied
voltage is about 0.5 and the ratio of the ac current to the de current is about 2/1T, so
that the efficiency would be about l/1T or more than 30%. For practical IMPATT
diodes, however, the efficiency is usually less than 30% because of the space-charge
effect, the reverse-saturation-current effect, the high-frequency-skin effect, and the
ionization-saturation effect.
IMPATT diodes are at present the most powerful CW solid-state microwave
power sources. The diodes have been fabricated from germanium, silicon, and gallium arsenide and can probably be constructed from other semiconductors as well.
IMPATT diodes provide potentially reliable, compact, inexpensive, and moderately
efficient microwave power sources.
Figure 8-2-3 shows the latest state-of-the-art performance for GaAs and Si
IMPATTs [6]. The numbers adjacent to the data points indicate efficiency in percent.
Power output data for both the GaAs and Si IMPATTs closely follow the 1/ f and
1/ / 2 slopes. The transition from the 1/ fto the 1/ f2 slope for GaAs falls between 50
and 60 GHz, and that for Si IMPATTs between 100 and 120 GHz. GaAs IMPATTs
show higher power and efficiency in the 40- to 60-GHz region whereas Si IMPATTs
are produced with higher reliability and yield in the same frequency region. On the
contrary, the GaAs IMPATTs have higher powers and efficiencies below 40 GHz
than do Si IMPATTs. Above 60 GHz the Si IMPATTs seem to outperform the GaAs
devices.
Example 8-2-1:
vd = 2
x 107 cm/s
L = 6 ,m
V0max = 100 V
lomax = 200 mA
T/ = 15%
vbd
90
Compute: (a) the maximum CW output power in watts; (b) the resonant frequency in
gigahertz.
Solution
a. From Eq. (8-2-10) the CW output power is
71Pctc
2 x 105
f = 2L = 2 x 6 x 10 _6 = 16.67 GHz
314
Chap. 8
83 TRAPATT DIODES
831 Physical Structures
The abbreviation TRAPATT stands for trapped plasma avalanche triggered transit
mode, a mode first reported by Prager et al. [7]. It is a high-efficiency microwave
generator capable of operating from several hundred megahertz to several gigahertz.
The basic operation of the oscillator is a semiconductor p-n junction diode reversebiased to current densities well in excess of those encountered in normal avalanche
operation. High-peak-power diodes are typically silicon n+ -p-p+ (or p+ -n-n+) structures with then-type depletion region width varying from 2.5 to 12.5 ,m. The doping of the depletion region is generally such that the diodes are well "punched
through" at breakdown; that is, the de electric field in the depletion region just prior
to breakdown is well above the saturated drift-velocity level. The device's p+ region
is kept as thin as possible at 2.5 to 7.5 ,m. The TRAPATT diode's diameter ranges
from as small as 50 ,m for CW operation to 750 ,mat lower frequency for highpeak-power devices.
dE
=E
s dt
(8-3- l)
Sec. 8.3
315
....fl... -
p+
n
Charging
Plasma formation
<:
"'t::
;:!
.,,c:
u
"'
"'
Oil
Plasma extraction
Residual extraction
Charging
Voltage
A
>
Figure 8-3-1 Voltage and current
waveforms for TRAPATT diode. (After
A. S. Clorfeine et al. [8); reprinted by
permission of RCA Laboratory.)
Current
T/2
Time
in the external current. At point Ethe plasma is removed, but a residual charge of
electrons remains in one end of the depletion layer and a residual charge of holes in
the other end. As the residual charge is removed, the voltage increases from point E
to point F. At point Fall the charge that was generated internally has been removed.
This charge must be greater than or equal to that supplied by the external current;
otherwise the voltage will exceed that at point A. From point F to point G the diode
charges up again like a fixed capacitor. At point G the diode current goes to zero for
half a period and the voltage remains constant at VA until the current comes back on
and the cycle repeats. The electric field can be expressed as
E(x, t)
= Em - qNA x + !!._
,.
Es
(8-3-2)
where NA is the doping concentration of then region and xis the distance.
Thus the value of t at which the electric field reaches Em at a given distance x
into the depletion region is obtained by setting E(x, t) = Em, yielding
t
= qNAX
J
(8-3-3)
where
Vz
(8-3-4)
316
Example 8-3-1:
Chap. 8
Doping concentration:
Current density:
Solution
20 x 103
10-19 X 2
1015
6.25
101 cm/s
This means that the avalanche-zone velocity is much larger than the scattering-limited
velocity.
Thus the avalanche zone (or avalanche shock front) will quickly sweep across
most of the diode, leaving the diode filled by a highly conductng plasma of holes and
electrons whose space charge depresses the voltage to low values. Because of the dependence of the drift velocity on the field, the electrons and holes will drift at velocities determined by the low-field mobilities, and the transit time of the carriers can
become much longer than
L
Ts= Vs
(8-3-5)
The TRAPATT operation is a rather complicated means of oscillation, however, and requires good control of both device and circuit properties. In addition,
Sec. 8.4
BARITI Diodes
TABLE 831
317
Frequency
(GHz)
Peak power
(W)
Average power
(W)
0.5
1.0
1.0
2.0
2.0
4.0
8.0
600
200
400
100
200
100
50
3
1
2
1
2
Operating voltage
(V)
Efficiency
(%)
150
40
30
35
25
30
20
15
I JO
1 JO
80
80
80
60
the TRAPATT mode generally exhibits a considerably higher noise figure than the
IMPATT mode, and the upper operating frequency appears to be practically limited
to below the millimeter wave region.
84 BARITT DIODES
841 Physical Description
BARITT diodes, meaning barrier injected transit-time diodes, are the latest addition
to the family of active microwave diodes. They have long drift regions similar to
those of IMPATT diodes. The carriers traversing the drift regions of BARITT
diodes, however, are generated by minority carrier injection from forward-biased
junctions instead of being extracted from the plasma of an avalanche region.
Several different structures have been operated as BARITT diodes, including
p-n-p, p-n-v-p, p-n-metal, and metal-n-metal. For a p-n-v-p BARITT diode, the
forward-biased p-n junction emits holes into the v region. These holes drift with saturation velocity through the v region and are collected at the p contact. The diode
exhibits a negative resistance for transit angles between 1T and 27T. The optimum
transit angle is approximately 1.67r.
Such diodes are much less noisy than IMPATT diodes. Noise figures are as low
as 15 dB at C-band frequencies with silicon BARITT amplifiers. The major disadvantages of BARITT diodes are relatively narrow bandwidth and power outputs limited to a few milliwatts.
318
Chap. 8
vdc
~~~~~~---i1~1~~~~-~--.
M1"4., . . _ :~
-~~ml
--"
Figure 8-4-1 M-n-M diode. (After D. J. Coleman and S. M. Sze [13]; reprinted
by permission of the Bell System, AT&T Co.)
1. The rapid increase of the carrier injection process caused by the decreasing potential barrier of the forward-biased metal-semiconductor contact
2. An apparent 37T /2 transit angle of the injected carrier that traverses the semiconductor depletion region
The rapid increase in terminal current with applied voltage (above 30 V) as
shown in Fig. 8-4-2 is caused by thermionic hole injection into the semiconductor
as the depletion layer of the reverse-biased contact reaches through the entire device
thickness. The critical voltage is approximately given by
qNL2
Ve= 2Es
(8-4-1)
Sec. 8.4
319
BARITT Diodes
10-2
10-3
10-4
10-s
,-.,
.,e"'0.
10-6
.....
10-7
E
-.!
10-8
10-9
10-10
10-11
20
40
60
V (volts)
80
The current-voltage characteristics of the silicon MSM structure (PtSiSi-PtSi) were measured at 77 K and 300 K. The device parameters are L = 10
,m, N = 4 x 10 14 cm 3 , </>n1 = </>n2 = 0.85 eV, and area = 5 X 10- 4 cm2
The current increase is not due to avalanche multiplication, as is apparent from
the magnitude of the critical voltage and its negative temperature coefficient. At 77
K the rapid increase is stopped at a current of about 10 - 5 A. This saturated current
is expected in accordance with the thermionic emission theory of hole injection
from the forward-biased contact with a hole barrier height (<f>p2) of about 0.15 eV.
843 Microwave Performance
Continuous-wave (CW) microwave performance of the M-n-M-type BARITT diode
was obtained over the entire C band of 4 to 8 GHz. The maximum power observed
was 50 mW at 4.9 GHz. The maximum efficiency was about 1.8%. The FM singlesideband noise measure at 1 MHz was found to be 22.8 dB at a 7-mA bias current.
This noise measure is substantially lower than that of a silicon IMPATT diode and is
comparable to that of a GaAs transfer-electron oscillator. Figure 8-4-3 shows some
of the measured microwave power versus current with frequency of operation indicated on each curve for three typical devices tested.
The voltage enclosed in parentheses for each curve indicates the average bias
voltage at the diode while the diode is in oscillation. The gain-bandwidth product of
a 6-GHz BARITT diode was measured to be 19-dB gain at 5-mA bias current at
200 MHz. The small-signal noise measure was about 15 dB.
320
Current (mA)
Example 8-4-1:
Chap. 8
Figure 8-4-3 Power output versus current for three silicon M-n-M devices.
(After D. J. Coleman and S. M. Sze
[13 ]; reprinted by permission of the Bell
System, AT&T Co.)
Er = 11.8
N = 2.8 x 1021 m- 3
L = 6 ,m
Solution
a. From Eq. (8-4-1) the breakdown voltage is double its critical voltage as
1.6 x 10- 19 x 2.8 x 1021 x (6 x 10- 6 )2
qNL 2
vbd = - - =
= 154.36
Es
8.854 X 10- 12 X 11.8
vbd
Ebd = - =
L
154.36
= 2.573 x 107 V/m = 2.57 x 1()5 V/cm
6 x 10- 6
85 PARAMETRIC DEVICES
851 Physical Description
A parametric device is one that uses a nonlinear reactance (capacitance or inductance) or a time-varying reactance. The word parametric is derived from the term
parametric excitation, since the capacitance or inductance, which is a reactive
Sec. 8.5
Parametric Devices
321
parameter, can be used to produce capacitive or inductive excitation. Parametric excitation can be subdivided into parametric amplification and oscillation. Many of the
essential properties of nonlinear energy-storage systems were described by Faraday
[14] as early as 1831 and by Lord Rayleigh [15] in 1883. The first analysis of the
nonlinear capacitance was given by van der Ziel [16] in 1948. In his paper van der
Ziel first suggested that such a device might be useful as a low-noise amplifier, since
it was essentially a reactive device in which no thermal noise is generated. In 1949
Landon [ 17] analyzed and presented experimental results of such circuits used as
amplifiers, converters, and oscillators. In the age of solid-state electronics, microwave electronics engineers dreamed of a solid-state microwave device to replace
the noisy electron beam amplifier. In 1957 Suhl [18] proposed a microwave solidstate amplifier that used ferrite. The first realization of a microwave parametric amplifier following Suhl' s proposal was made by Weiss [ 19] in 1957. After the work
done by Suhl and Weiss, the parametric amplifier was at last discovered.
At present the solid-state varactor diode is the most widely used parametric
amplifier. Unlike microwave tubes, transistors, and lasers, the parametric diode is of
a reactive nature and thus generates a very small amount of Johnson noise (thermal
noise). One of the distinguishing features of a parametric amplifier is that it utilizes
an ac rather than a de power supply as microwave tubes do. In this respect, the parametric amplifier is analogous to the quantum amplifier laser or maser in which an ac
power supply is used.
BS:Z Nonlinear Reactance and Manley-Rowe Power
Relations
C=Q
(8-5-1)
If the ratio is not linear, the capacitive reactance is said to be nonlinear. In this case,
it is convenient to define a nonlinear capacitance as the partial derivative of charge
with respect to voltage. That is,
C(v)
= -iJQ
iJv
(8-5-2)
(8-5-3)
322
Chap. 8
In the operation of parametric devices, the mixing effects occur when voltages at
two or more different frequencies are impressed on a nonlinear reactance.
Small-signal method. It is assumed that the signal voltage Vs is much
smaller than the pumping voltage vp , and the total voltage across the nonlinear capacitance C (t) is given by
V = Vs+ Vp = Vs
COS
(wst) + Vp
COS
(wpt)
(8-5-4)
where Vs <SS VP . The charge on the capacitor can be expanded in a Taylor series about
the point Vs = 0, and the first two terms are
Q(v)
= Q(v1 +
vP)
dQ(vp) I
= Q(vp) + -----;J;;~:o
(8-5-5)
= dQ(vp) =
C(v)
dv
C(t)
(8-5-6)
C(vp)
Cn cos (nwpt)
(8-5-7)
n=O
Since vp is a function of time, the capacitance C(vp) is also a function of time. Then
C(t)
Cn cos (nwpt)
(8-5-8)
n=O
The coefficients Cn are the magnitude of each harmonic of the time-varying capacitance. In general, the coefficients C" are not linear functions of the ac pumping
voltage vP . Since the junction capacitance C (t) of a parametric diode is a nonlinear
capacitance, the principle of superposition does not hold for arbitrary ac signal amplitudes.
The current through the capacitance C(t) is the derivative of Eq. (8-5-5) with
respect to time and it is
i
= dQ = dQ(vp) + !{[C(t)vs]
dt
dt
dt
(8-5-9)
Sec. 8.5
323
Parametric Devices
Co[l
2y cos (wpt)]
(8-5-10)
The parameter y is proportional to the pumping voltage vp and indicates the coupling
effect between the voltages at the signal frequency fs and the output frequency fo .
Manley-Rowe power relations. Manley and Rowe [20] derived a set of
general energy relations regarding power flowing into and out of an ideal nonlinear
reactance. These relations are useful in predicting whether power gain is possible in
a parametric amplifier. Figure 8-5-1 shows an equivalent circuit for Manley-Rowe
derivation.
C(t)
Figure 851
In Fig. 8-5-1, one signal generator and one pump generator at their respective
frequenciesfs andJ;,, together with associated series resistances and bandpass filters,
are applied to a nonlinear capacitance C (t). These resonating circuits of filters are
designed to reject power at all frequencies other than their respective signal frequencies. In the presence of two applied frequenciesfs andJ;,, an infinite number of resonant frequencies of mfp nfp are generated, where m and n are any integers from
zero to infinity.
Each of the resonating circuits is assumed to be ideal. The power loss by the
nonlinear susceptances is negligible. That is, the power entering the nonlinear capacitor at the pump frequency is equal to the power leaving the capacitor at the other
frequencies through the nonlinear interaction. Manley and Rowe established the
power relations between the input power at the frequencies fs and fp and the output
power at the other frequencies mfp nfs .
From Eq. (8-5-4) the voltage across the nonlinear capacitor C(t) can be expressed in exponential form as
(8-5-11)
324
Chap. 8
Qm.n = Q"'.m.-n
The total voltage v can be expressed as a function of the charge Q. A similar Taylor
series expansion of v (Q) shows that
V
L L
Vm.nej(mwpt+nwst)
(8-5-14)
m=-oo n=-oo
dQ
dt
1=-=
2: 2:
J(mwp + nws)Qm.nej(mwp+nws)
m=-oo n=-oo
(8-5-16)
m=-oo n=-oo
where Im.n = j(mwp + nws)Qm.n and Im.n = I"'.m.-n. Since the capacitance C(t) is assumed to be a pure reactance, the average power at the frequencies mfP + nfs is
Pm,n
= (Vm,nf!,n + V!.nfm,n)
= (V"'.m.-nf-m.-n + V-m.-nf"'.m,-n) = P-m.-n
(8-5-17)
i i
m=-oon=-oo
mPm,n
+ Ws
mwp + nws
i i
m=-oon=-00
nPm.n
mwp + nws
=0
(8-5-19)
Since
Im.n/(mwp + nws)
= }Qm.n
Sec. 8.5
325
Parametric Devices
changed, since they are functions of the voltages Vm.n. Consequently, the frequencies JP and fs can be arbitrarily adjusted in order to require
(8-5-20)
i: i:
nPm.n
+ nws
=O
(8-5-21)
m=-oon=-oo 111Wp
i: i:
i: i:
mPm.n
+
-mPm.n
m=On=-oo l11Wp + nws
m=On=-oo -mwp - nws
=O
(8-5-22)
i: i:
=O
(8-5-23)
ii
=O
(8-5-24)
mPm,n
m=O n= -oo mjp + nfs
Similarly,
m=-oo
nPm,n
n=O mf,, + nfs
326
Chap.8
The power gain, which is defined as a ratio of the power delivered by the capacitor at a frequency of J;, + fs to that absorbed by the capacitor at a frequency off,
as shown in Eq. (8-5-26) is given by
Gain
=JP +
f,
f,
=fr!.
fs
(for modulator)
(8-5-27)
where J;, + fs = fo and (JP + fs) > JP > fs . The maximum power gain is simply
the ratio of the output frequency to the input frequency. This type of parametric
device is called the sum-frequency parametric amplifier or up-converter.
If the signal frequency is the sum of the pump frequency and the output frequency, Eq. (8-5-26) predicts that the parametric device will have a gain of
.
Ga m
f,
= Jp-;:----!
+ s
(for demodulator)
(8-5-28)
where f,
and P 0 1 are negative. In other words, the capacitor delivers power to the signal generator atf, instead of absorbing it. The power gain may be infinite, which is an unstable condition, and the circuit may be oscillating both atf, and/a. This is another
type of parametric device, often called a negative-resistance parametric amplifier.
327
Parametric Devices
Sec. 8.5
Idler circuit
Signal circuit
Figure 8-5-2
Parametric up-converter.
properties:
1. The ouptut frequency is equal to the sum of the signal frequency and the pump
frequency.
2. There is no power flow in the parametric device at frequencies other than the
signal, pump, and output frequencies.
Power Gain. When these two conditions are satisfied, the maximum power
gain of a parametric up-converter [21] is expressed as
Gain=~
JS
where fo = f,,
(1
+ \/l+x) 2
(8-5-30)
+ fs
=I!. (yQ)2
Q=
fo
2TTfsCRd
Moreover, Rd is the series resistance of a p-n junction diode and yQ is the figure of
merit for the nonlinear capacitor. The quantity of x/(l + \/l+x) 2 may be regarded as a gain-degradation factor. As Rd approaches zero, the figure of merit yQ
goes to infinity and the gain-degradation factor becomes equal to unity. As a result,
the power gain of a parametric up-converter for a lossless diode is equal to fol fs ,
which is predicted by the Manley-Rowe relations as shown in Eq. (8-5-27). In a
typical microwave diode yQ could be equal to 10. If fol fs = 15, the maximum gain
given by Eq. (8-5-30) is 7.3 dB.
328
Chap. 8
Noise Figure. One advantage of the parametric amplifier over the transistor
amplifier is its low-noise figure because a pure reactance does not contribute thermal
noise to the circuit. The noise figure F for a parametric up-converter [21] is given
by
2Td[ 1
1 ]
+ To yQ + (yQ)2
(8-5-31)
10 and y
Example 8-5-1:
2y.J4
(8-5-32)
fs
Calculate: (a) the power gain in decibels; (b) the noise figure in decibels; (c) the bandwidth.
Solution
a. From Eq. (8-5-30) the up-converter power gain is
J
ls
Power gam = -
100/25
+ Vi"'"+x)
= 9.55 = 9.80 dB
(1
= 25 x - - - r = = = = -
+ y' 1 + 100/25) 2
(1
+ 2Td[_l_ + _1_]
=l +2
2
T0
yQ
(yQ)
x 350[_!_
300
10
+ _1_] = 1. 26 = l
100
1 2
dB
329
Parametric Devices
Sec. 8.5
Parametric down-converter. If a mode of down conversion for a parametric amplifier is desirable, the signal frequency ls must be equal to the sum of the
pump frequency f,, and the output frequency fo . This means that the input power
must feed into the idler circuit and the output power must move out from the signal
circuit as shown in Fig. 8-5-2. The down-conversion gain (actually a loss) is given
by [21]
Gain=~
Jo
x
(I+ ~)2
(8-5-33)
Power Gain. The output power is taken from the resistance R; at a frequency
f; , and the conversion gain from ls to f; [21] is given by
Gain = 4 fi
ls
where ls
f,,
f;
R!I
R;
RTs
RTi
R8 R;
RTsRT;
(I - a) 2
(8-5-34)
=
=
=
=
=
=
=
signal frequency
pump frequency
fp - fs is the idler frequency
output resistance of the signal generator
output resistance of the idler generator
total series resistance at ls
total series resistance atf;
a = R/RTs
R = "/ /(wsw; C 2 RT;) is the equivalent negative resistance
330
Chap. 8
Bandwidth. The maximum gain bandwidth of a negative-resistance parametric amplifier [21] is given by
BW
If gain = 20 dB, f; = 4fs, and y
single-tuned circuits is about 0.03.
IT
=1
2\J~
(8-5-36)
Power Gain and Bandwidth. The power gain and bandwidth characteristics
of a degenerate parametric amplifier are exactly the same as for the parametric upconverter. With fs = f; and fp = 2fs, the power transferred from pump to signal
frequency is equal to the power transferred from pump to idler frequency. At high
gain the total power at the signal frequency is almost equal to the total power at the
idler frequency. Hence the total power in the passband will have 3 dB more gain.
Noise Figure. The noise figures for a single-sideband and a double-sideband
degenerate parametric amplifier [21] are given by, respectively,
F:
ssb
Fctsb
where Td =
To =
Rd =
R8 =
2TdRd
ToRg
TdRd
+ ToRg
(8-5-37)
(8-5-38)
It can be seen that the noise figure for double-sideband operation is 3 dB less than
that for single-sideband operation.
854 Applications
The choice of which type of parametric amplifier to use depends on the microwave
system requirements. The up-converter is a unilateral stable device with a wide
bandwidth and low gain. The negative-resistance amplifier is inherently a bilateral
and unstable device with narrow bandwidth and high gain. The degenerate parametric amplifier does not require a separate signal and idler circuit coupled by the diode
and is the least complex type of parametric amplifier.
Chap. 8
References
331
In general, the up-converter has the following advantages over the negativeresistance parametric amplifier:
1. A positive input impedance
2.
3.
4.
5.
At higher frequencies where the up-converter is no longer practical, the negativeresistance parametric amplifier operated with a circulator becomes the proper choice.
When a low noise figure is required by a system, the degenerate parametric amplifier
may be the logical choice, since its double-sideband noise figure is less than the optimum noise figure of the up-converter or the nondegenerate negative-resistance
parametric amplifier. Furthermore, the degenerate amplifier is a much simpler
device to build and uses a relatively low pump frequency. In radar systems the negative-resistance parametric amplifier may be the better choice, since the frequency required by the system may be higher than the X band. However, since the parametric
amplifier is complicated in fabrication and expensive in production, there is a tendency in microwave engineering to replace it with the GaAs metal-semiconductor
field-effect transistor (MESFET) amplifier in airborne radar systems.
REFERENCES
[l] READ, W. T., A proposed high-frequency negative-resistance diode. Bell System Tech.
J., 37, 401-446 (1958)
[2] COLEMAN, D. J., JR., and S. M. SzE, A low-noise metal-semiconductor-metal (MSM)
microwave oscillator. Bell SystemTech. J., 50, 1695-1699, May-June 1971.
[3] LEE, C. A., Er AL., The Read diode, an avalanche, transit-time, negative-resistance oscillator. Appl. Phys. Letters, 6, 89 (1965).
[4] JOHNSTON, R. L., B. C. DELOACH, and G. B. COHEN, A silicon diode microwave oscillator. Bell System Tech. J., 44, 369-372, February 1965.
[5] GILDEN, M., and M. E. HINES, Electronic tuning effects in the Read microwave
avalanche diode. IEEE Trans. on Electron Devices, ED-13, 5-12, January 1966.
[6] HIESLMAIR, H., Er AL., State of the art of solid-state and tube transmitters. Microwave
J., 26, No. 10, 46-48, October 1983.
[7] PRAGER, H. J., Er AL., High-power, high-efficiency silicon avalanche diodes at ultra high
frequencies. Proc. IEEE (Letters), 55, 586-587, April 1967.
[8] CLORFEINE, A. S., Er AL., A theory for the high-efficiency mode of oscillation in
avalanche diodes. RCA Rev., 30, 397-421, September 1969.
[9] DELOACH, B. C., JR., and D. L. ScHARFEITER, Device physics of TRAPATT oscillators.
IEEE Trans. on Electron Devices, ED-17, 9-21, January 1970.
332
Chap. 8
[ 10) LIU, S. G., and J. J. RISKA, Fabrication and performance of kilowatt L-band avalanche
diodes. RCA Rev., 31, 3, March 1970.
[I I] KosTICHACK, D. F., UHF avalanche diode oscillator providing 400 watts peak power and
75 percent efficiency. Proc. IEEE (letters), 58, 1282-1283, August 1970.
[12) WILSON, W. E., Pulsed LSA and TRAPATT sources for microwave systems. Microwave
J., 14, No. 8, 33-41, August 1971.
[13) COLEMAN, D. J., JR., and S. M. SzE, A low-noise metal-semiconductor-metal (MSM)
microwave oscillator. Bell System Tech. J., 1675-1695, May-June 1971.
[ 14] FARADAY, M., On a peculiar class of acoustical figures; and certain forms assumed by a
group of particles upon vibrating elastic surface. Phil. Trans. Roy. Soc. (London), 121,
299-318, May 1831.
[15] LoRD RAYLEIGH, and J. W. STRUTT, On the crispations of fluid resting upon a vibrating
support. Phil. Mag., 16, 50-53, July 1883.
[ 16) VAN DER ZIEL, A., On the mixing properties of nonlinear capacitances. J. Appl. Phys.,
19, 999-1006, November 1948.
[17) LANDON, V. D., The use of ferrite cored coils as converters, amplifiers, and oscillators.
RCA Rev., 10, 387-396, September 1949.
[18) SUHL, H., Proposal for a ferromagnetic amplifier in the microwave range. Phys. Rev.,
106, 384-385, April 15, 1957.
[19) WEISS, M. T., A solid-state microwave amplifier and oscillator using ferrites. Phys Rev.,
107, 317, July 1957.
[20) MANLEY, J. M., and H. E. RowE, Some general properties of nonlinear elements-Pt. I,
General energy relations. Proc. IRE, 44, 904-913, July 1956.
SUGGESTED READINGS
CHANG, K. K. N., Parametric and Tunnel Diodes. Prentice-Hall, Inc., Englewood Cliffs,
N.J., 1964.
DELoACH, B. C., and D. L. SCHARFETTER, Device physics of TRAPATT oscillators. IEEE
Trans. on Electron Devices, ED-17, No. I, 9-21, January 1970.
EASTMAN, L. F., Gallium Arsenide Microwave Bulk and Transit-Time Devices. Artech House,
Dedham, Mass., 1973.
HADDAD, G. I., ed., Avalanche Transit-Time Devices. Artech House, Dedham, Mass., 1973.
Chap. 8
Problems
333
HADDAD, G. I., ET AL., Basic principles and properties of avalanche transit-time devices. IEEE
Trans. on Microwave Theory and Techniques, MTT-18, No. 11, 752-772, November
1970.
MILNES, A.G., Semiconductor Devices and Integrated Electronics. Van Nostrand Reinhold
Company, New York, 1980.
PARKER, D., TRAPATT oscillations in a p-i-n avalanche diode. IEEE Trans. on Electron
Devices, ED-18, No. 5, 281-293, May 1971.
SzE, S. M., Microwave avalanche diodes. IEEE Proc., 59, No.8, 1140-1171, August 1971.
PROBLEMS
Avalanche Transit-Time Devices
8-1.
Spell out the following abbreviated terms: IMPATT, TRAPATT, and BARITT.
8-2.
8-3.
Describe the operating principles of the Read diode, IMPATT diode, TRAPATT diode,
and BARITT diode.
8-4.
An IMPATT diode has a drift length of 2 ,m. Determine: (a) the drift time of the carriers and (b) the operating frequency of the IMPATT diode.
8-5.
A Ku-band IMPATT diode has a pulsed-operating voltage of 100 V and a pulsedoperating current of 0.9 A. The efficiency is about 10%. Calculate:
a. The output power
b. The duty cycle if the pulse width is 0.01 ns and the frequency is 16 GHz
8-6.
Er = J J.8
N = 3 x 1021 m- 3
L = 6.2 ,m
Calculate: (a) the breakdown voltage and (b) the breakdown electric field.
Parametric Devices
8-7.
8-8.
The figure of merit for a diode nonlinear capacitor in an up-converter parametric amplifier is 8, and the ratio of the output (or idler) frequency fa over the signal frequency
f, is 8. The diode temperature is 300K.
a. Calculate the maximum power gain in decibels.
b. Compute the noise figure F in decibels.
c. Determine the bandwidth (BW) for 'Y = 0.2.
334
8-9.
Is=
fp
f,
f,
=
=
=
2 GHz
R;
1 kO
'Y
12 GHz
R8
1 KO
yQ
10 GHz
Rrs
lKO
5f,
Rr;
=
=
=
1 KO
=
=
0.35
Td
300K
= 0.01
10
pF
Chap. 8
Chapter 9
Microwave Linear-Beam
Tubes (0 Type)
90 INTRODUCTION
We turn now to a quantitative and qualitative analysis of several conventional vacuum tubes and microwave tubes in common use. The conventional vacuum tubes,
such as triodes, tetrodes, and pentodes, are still used as signal sources of low output
power at low microwave frequencies. The most important microwave tubes at
present are the linear-beam tubes (O type) tabulated in Table 9-0-1. The paramount
0-type tube is the two-cavity klystron, and it is followed by the reflex klystron. The
helix traveling-wave tube (TWT), the coupled-cavity TWT, the forward-wave amplifier (FWA), and the backward-wave amplifier and oscillator (BWA and BWO) are
also 0-type tubes, but they have nonresonant periodic structures for electron interactions. The Twystron is a hybrid amplifier that uses combinations of klystron and
TWT components. The switching tubes such as krytron, thyratron, and planar triode
are very useful in laser modulation. Although it is impossible to discuss all such
tubes in detail, the common operating principles of many will be described.
The advent of linear-beam tubes began with the Heil oscillators [I] in 1935
and the Varian brothers' klystron amplifier [ 1] in 1939. The work was advanced by
the space-charge-wave propagation theory of Hahn and Ramo [l] in 1939 and continued with the invention of the helix-type traveling-wave tube (TWT) by R.
Kompfner in 1944 [2]. From the early 1950s on, the low power output of linearbeam tubes made it possible to achieve high power levels, first rivaling and finally
surpassing magnetrons, the early sources of microwave high power. Subsequently,
several additional devices were developed, two of which have shown lasting importance. They are the extended interaction klystron [3] and the Twystron hybrid amplifier [4].
335
336
TABLE 901
Chap. 9
I
Cavity
Slow-wave structure
\
Resonant
Forward-wave
Backward-wave
~
Reflex
Klystron
Twystron
Coupled-cavity
TWT
In a linear-beam tube a magnetic field whose axis coincides with that of the
electron beam is used to hold the beam together as it travels the length of the tube.
0 -type tubes derive their name from the French TPO (tubes a propagation des
ondes) or from the word original (meaning the original type of tube). In these tubes
electrons receive potential energy from the de beam voltage before they arrive in the
microwave interaction region, and this energy is converted into their kinetic energy.
In the microwave interaction region the electrons are either accelerated or decelerated by the microwave field and then bunched as they drift down the tube. The
bunched electrons, in turn, induce current in the output structure. The electrons then
give up their kinetic energy to the microwave fields and are collected by the
collector.
0-type traveling-wave tubes are suitable for amplification. At present, klystron
and TWT amplifiers can deliver a peak power output up to 30 MW (megawatts) with
a beam voltage on the order of 100 kV at the frequency of 10 GHz. The average
power outputs are up to 700 kW. The gain of these tubes is on the order of 30 to
70 dB, and the efficiency is from 15 to 60%. The bandwidth is from 1 to 8% for
klystrons and 10 to 15% for TWTs.
Since the early 1960s, predictions have continued that microwave tubes will be
displaced by microwave solid-state devices. This displacement has occurred only at
the low-power and receiving circuit level of equipment. Microwave power tubes
continue, however, as the only choice for high-power transmitter outputs and are expected to maintain this dominant role throughout the next generation and beyond.
Figure 9-0-1 and 9-0-2 show the CW power and peak power state-of-the-art performances for various tube types [5].
The numbers by the data points represent efficiency in percent; the letter in
parentheses stands for the developer of the tube. Impressive gain has been made in
bandwidth of a single TWT device by Varian. More than 50 dB of gain is available
GYROTRON (U)
3
10
15 (R) (S)
/PPM COUPLED
CAVITYTWT
TWT(R)
,_______,
TWT(V)
1--~---~
ro
12 (H)
(W)
HDL~
VARIAN (V)
WATKINS JOHNSON (W) OROTRON
100
10
FREQUENCY (GHz)
o CFA
e TWT
6 KLYSTRON
GYROTRON
65 (U)
/GYROTRON
68 (U)
40 (H)
10 (U)
10
DATA FROM PUBLISHED RESULTS:
HUGHES (H)
RAYTHEON (R)
SIEMENS (S)
THOMPSON CSF (C)
USSR (U)
VARIAN (V)
10 1..-~-'---'--''-'-...UU-l...L-~-'---''--...........................__...___._~..__~
3
10
FREQUENCY IGHzl
100
337
338
Chap.9
in a 93- to 95-GHz TWT at the 50-watt average level by Hughes, and 2-kW peak
power at 30-dB gain is provided by Varian. The most impressive power achievements at very good efficiencies continue to occur in the gyrotron area. The Naval
Research Laboratory (NRL) reports results of work on a 30% bandwidth gyrotron
TWT. Tube reliability continues to improve as a result of low-temperature cathode
technology, better materials, and quality control in manufacture.
Conventional vacuum triodes, tetrodes, and pentodes are less useful signal sources at
frequencies above 1 GHz because of lead-inductance and interelectrode-capacitance
effects, transit-angle effects, and gain-bandwidth product limitations. These three effects are analyzed in detail in the following sections.
911 Lead-Inductance and
lnterelectrodeCapacitance Effects
At frequencies above 1 GHz conventional vacuum tubes are impaired by parasiticcircuit reactances because the circuit capacitances between tube electrodes and the
circuit inductance of the lead wire are too large for a microwave resonant circuit.
Furthermore, as the frequency is increased up to the microwave range, the real part
of the input admittance may be large enough to cause a serious overload of the input
circuit and thereby reduce the operating efficiency of the tube. In order to gain a better understanding of these effects, the triode ciruit shown in Fig. 9-1-1 should be
studied carefully.
Figure 9-1-1
Figure 9-1-1 (b) shows the equivalent circuit of a triode circuit under the assumption that the interelectrode capacitances and cathode inductance are the only
parasitic elements. Since C8p i C8k and wLk i 1/(wC8k), the input voltage Vin can
be written as
(9-1-1)
Sec. 9.1
339
l;n(l
'"
+ }wlkgm)
jwC8 k
(9-1-3)
Ym.
=h =
Vin
jwCgk
_ 2
l+"L
-w~~~+~~
JW k8m
(9-1-4)
in which wlkgm ~ 1 has been replaced. The inequality is almost always true, since
the cathode lead is usually short and is quite large in diameter, and the transconductance gm is generally much less than one millimho.
The input impedance at very high frequencies is given by
1
Zin =
LkCgkgm
.
1
J 3L 2 C
W
k gkgm
(9-1-5)
The real part of the impedance is inversely proportional to the square of the frequency, and the imaginary part is inversely proportional to the third order of the
frequency. When the frequencies are above 1 GHz, the real part of the impedance
becomes small enough to nearly short the signal source. Consequently, the output
power is decreased rapidly. Similarly, the input admittance of a pentode circuit is expressed by
(9-1-6)
where Cgs is the capacitance between the grid and screen, and its input impedance is
given by
1
. C8 k + Cgs
(9-1-7)
Zin= w 2 LkC8 k8m - 1 w 3Uqkg;,,
There are several ways to minimize the inductance and capacitance effects, such as a
reduction in lead length and electrode area. This minimization, however, also limits
the power-handling capacity.
912 Transit-Angle Effects
where r8
Vo
= de voltage
YVo
340
Chap. 9
When frequencies are below microwave range, the transit angle is negligible.
At microwave frequencies, however, the transit time (or angle) is large compared to
the period of the microwave signal, and the potential between the cathode and the
grid may alternate from 10 to 100 times during the electron transit. The grid potential during the negative half cycle thus removes energy that was given to the electron
during the positive half cycle. Consequently, the electrons may oscillate back and
forth in the cathode-grid space or return to the cathode. The overall result of transitangle effects is to reduce the operating efficiency of the vacuum tube. The degenerate
effect becomes more serious when frequencies are well above 1 GHz. Once electrons
pass the grid, they are quickly accelerated to the anode by the high plate voltage.
When the frequency is below 1 GHz, the output delay is negligible in comparison with the phase of the grid voltage. This means that the transadmittance is a large
real quantity, which is the usual transconductance gm . At microwave frequencies the
transit angle is not negligible, and the transadmittance becomes a complex number
with a relatively small magnitude. This situation indicates that the output is decreased.
From the preceding analysis it is clear that the transit-angle effect can be minimized by first accelerating the electron beam with a very high de voltage and then
velocity-modulating it. This is indeed the principal operation of such microwave
tubes as klystrons and magnetrons.
9-1 3 Gain-Bandwidth Product Limitation
In ordinary vacuum tubes the maximum gain is generally achieved by resonating the
output circuit as shown in Fig. 9-1-2. In the equivalent circuit (see Fig. 9-1-2) it is
assumed that rp ~ wLk . The load voltage is given by
V _
c- G
where G =
rp =
R =
L, C =
gmV8
j[wC - 1/(wL)]
(9-1-9)
1/rp + 1/R
plate resistance
load resistance
tuning elements
fr= 27TVLC
(9-1-10)
Sec. 9.2
Klystrons
341
Amax
at resonance by
(9-1-11)
Since the bandwidth is measured at the half-power point, the denominator of Eq.
(9-1-9) must be related by
G
1
wL
= wC - -
(9-1-12)
~~(--~-)-2_+_-f-_C
w2
~C + ~ ( ~ + L~
(9-1-13)
(9-1-14)
W2 -
w,
= -
for ( -G )
2C
1
~-
LC
(9-1-15)
9-2 KLYSTRONS
The two-cavity klystron is a widely used microwave amplifier operated by the principles of velocity and current modulation. All electrons injected from the cathode arrive at the first cavity with uniform velocity. Those electrons passing the first cavity
gap at zeros of the gap voltage (or signal voltage) pass through with unchanged velocity; those passing through the positive half cycles of the gap voltage undergo an
increase in velocity; those passing through the negative swings of the gap voltage undergo a decrease in velocity. As a result of these actions, the electrons gradually
bunch together as they travel down the drift space. The variation in electron velocity
in the drift space is known as velocity modulation. The density of the electrons in
the second cavity gap varies cyclically with time. The electron beam contains an ac
component and is said to be current-modulated. The maximum bunching should occur approximately midway between the second cavity grids during its retarding
342
Chap. 9
phase; thus the kinetic energy is transferred from the electrons to the field of the second cavity. The electrons then emerge from the second cavity with reduced velocity
and finally terminate at the collector. The charateristics of a two-cavity klystron amplifier are as follows:
1. The electron beam is assumed to have a uniform density in the cross section of
the beam.
2. Space-charge effects are negligible.
3. The magnitude of the microwave signal input is assumed to be much smaller
than the de accelerating voltage.
921 Reentrant Cavities
At a frequency well below the microwave range, the cavity resonator can be represented by a lumped-constant resonant circuit. When the operating frequency is increased to several tens of megahertz, both the inductance and the capacitance must
be reduced to a minimum in order to maintain resonance at the operating frequency.
Ultimately the inductance is reduced to a minimum by short wire. Therefore the
reentrant cavities are designed for use in klystrons and microwave triodes. A
reentrant cavity is one in which the metallic boundaries extend into the interior of
the cavity. Several types of reentrant cavities are shown in Fig. 9-2-3. One of the
commonly used reentrant cavities is the coaxial cavity shown in Fig. 9-2-4.
It is clear from Fig. 9-2-4 that not only has the inductance been considerably
decreased but the resistance losses are markedly reduced as well, and the shelfshielding enclosure prevents radiation losses. It is difficult to calculate the resonant
frequency of the coaxial cavity. An approximation can be made, however, using
transmission-line theory. The characteristic impedance of the coaxial line is given
by
ohms
(9-2-1)
..___,
..........
'--'
20
~
10
.......
- ..
~.
;::
::;:
........
,_:
:>
....
:>
......
........... .....
0..
er
,_
;::
0..
"'
~
1-J
....
.....
....
....
......
0.8
0.6
0.5
0.4
0.3
.......
0.2
0.05
0.4
0.2
~.1
0.6
10
20
FREQUENCY, GHz
Figure 9-21
Buncher cavity
Catcher cavity
Anode
Cathode
I Vo~I
G,,
Vo
j(tl)--+-
T
Bunched
to
Figure 9-2-2
E:3:
I
I
Distance scale
Time scale
I I
(b)
f 2 f3
f1
[[]
(a)
Collector
L+d L+2d
I, I
v(t1)
electron
beam
Vg
-11 +
.--J
Drift space
(c)
8
(d)
(e)
Figure 9-2-3 Reentrant cavities. (a) Coaxial cavity. (b) Radial cavity. (c) Tunable cavity.
(d) Toroidal cavity. (c) Butterfly cavity.
30
344
I.J:
I
I
........
r---r--r-
--?'\
1 \
I\
I
I
I
I
I
I
I I
II
I
I
1I
I I
: \
I :
,'
\I
I\
\I
..,_;.,.
\I
--~
1 \
/I\
"'----;-::...!
I
I
2b
Chap. 9
L~___.
(b) Cross-section of
coaxial cavity
The coaxial cavity is similar to a coaxial line shorted at two ends and joined at the
center by a capacitor. The input impedance to each shorted coaxial line is given by
(9-2-2)
Zin= j 211'
\j~en~
~ a tan(/3C)
(9-2-3)
Xin
W
= - 111'W
v~~ en~a
tan(/3 C)
(9-2-4)
Cg=d
(9-2-5)
At resonance the inductive reactance of the two shorted coaxial lines in series is
equal in magnitude to the capacitive reactance of the gap. That is, wL = 1/(wCg).
Thus
tan(/3C)
dv
wa2en(b/a)
(9-2-6)
Sec. 9.2
345
Klystrons
The radial reentrant cavity shown in Fig. 9-2-5 is another commonly used
reentrant resonator. The inductance and capacitance of a radial reentrant cavity is
expressed by
L
=fen~
21T'
C -_
[1T'a
E"o -
(9-2-7)
[)
0.765
]
4a 1-n--r========
v'f 2 + (b - a) 2
(9-2-8)
f,
r
where c
21T' ~
e v' f
o.765
+ (b - a) 2
Jen~}-112
a
(9-2-9)
Figure 9-2-5
(2;V,,
'J-;;:
= 0.593
10
;-;:-
Vo
mis
(9-2-10)
In Eq. (9-2-10) it is assumed that electrons leave the cathode with zero velocity.
When a microwave signal is applied to the input terminal, the gap voltage between
the buncher grids appears as
Vs = Vi sin (wt)
(9-2-11)
= -Vod = t1
to
(9-2-12)
346
V.
V.
= V1 sin wt
Chap. 9
t0 t 1
: I Buncher grids
-id*
I I
I I
I
Figure 9-2-6
buncher gap.
(Jg =
WT
wd
(t1 - to) = Vo
(9-2-13)
The average microwave voltage in the buncher gap can be found in the following
way:
1
(Vs)= T
Ill
to
:~ [ cos(wto) -
cos ( wo + ::) ]
(9-2-14)
Let
wd
wto + 2vo
(Jg
2
= wto + - = A
and
wd
2vo
= Og = B
2
(Vs > -_
V sin [wd/2vo)] . (
i
wd/( 2vo)
= 2 sin A
(9-2-15)
It is defined as
{3;
= sin [wd/(2v
wd/(2vo)
0 )]
= sin (Og/2)
Og/2
(9-2-16)
Note that {3; is known as the beam-coupling coefficient of the input cavity gap (see
Fig. 9-2-7).
It can be seen that increasing the gap transit angle (Jg decreases the coupling
between the electron beam and the buncher cavity; that is, the velocity modulation
Sec. 9.2
347
Klystrons
1.0
0.8
0.6
'
\ -/
-0.2
-0.4 0
""
I/
['..._ . /
71'
Figure 9-2-7
of the beam for a given microwave signal is decreased. Immediately after velocity
modulation, the exit velocity from the buncher gap is given by
(9-2-17)
where the factor {3; Vi/Vo is called the depth of velocity modulation.
Using binomial expansion under the assumption of
(9-2-18)
Eq. (9-2-17) becomes
. (
fJg)]
v (t 1) = Vo [ 1 + {3;Vi
Vo sm wto + 2
2
(9-2-19)
v(t1 )
v0 [ 1
. (
+ {3;Vi
Vo sm wt1
2
2fJg)]
(9-2-20)
348
Chap. 9
AL
Buncher
grid
L-----''1.---~--.~__,.__
__. - - - - ' - - - . - - - - . _
The distance from the buncher grid to the location of dense electron bunching
for the electron at th is
!:J..L =
Vo(td -
!:J..L =
!:J..L
Vmin(td -
= Vmax(fd
ta
(9-2-21)
lb)
and
fa)
fc)
= Vmax(fd
tc
Vmin(td -
are
fb
+ ;:)
- fb -
;:)
(9-2-22)
(9-2-23)
From Eq. (9-2-19) or (9-2-20) the minimum and maximum velocities are
Vmin
vo(l - ~~1 )
(9-2-24)
Sec. 9.2
349
Klystrons
Vmax
= Vo ( l + {3;2VoVi)
(9-2-25)
Substitution of Eqs. (9-2-24) and (9-2-25) in Eqs. (9-2-22) and (9-2-23), respectively, yields the distance
!lL
= v0(td
- tb)
{3; Vi
71'
{3; Vi
71'
(9-2-26)
and
fiL = v0(td - tb)
71'
{3; Vi
{3; Vi
71' ]
(9-2-27)
The necessary condition for those electrons at ta , tb, and tc to meet at the same distance !lL is
(9-2-28)
and
(9-2-29)
Consequently,
(9-2-30)
and
"L -_ vo-7TVo
wf3;Vi
(9-2-31)
It should be noted that the mutual repulsion of the space charge is neglected,
but the qualitative results are similar to the preceding representation when the effects
of repulsion are included. Furthermore, the distance given by Eq. (9-2-31) is not the
one for a maximum degree of bunching. Figure 9-2-9 shows the distance-time plot
or Applegate diagram.
What should the spacing be between the buncher and catcher cavities in order
to achieve a maximum degree of bunching? Since the drift region is field free, the
transit time for an electron to travel a distance of L as shown in Fig. 9-2-2 is given
by
1
= To[ I sin ( wt1 (9-2-32)
T = t2 - t1 = v
ti)
i)]
~~
where the binomial expansion of (I + x)~ 1 for Ix I ~ 1 has been replaced and
To = L/vo is the de transit time. In terms of radians the preceding expression can be
written
wT
i)
(9-2-33)
350
Acceleration
Chap.9
Output-gap
Catcher gap
Buncher gap
Figure 9-2-9
where
Oo
Applegate diagram.
wL
= - = 21T'N
(9-2-34)
Vo
is the de transit angle between cavities, N is the number of electron transit cycles in
the drift space, and
X =
~~1 Oo
(9-2-35)
. (
(Jg)]
+ T + To [ 1 - {3;2VoVi sm
wto + 2
(9-2-38)
Alternatively,
wt2 -
(Jg) (
(Jg)
(Oo + 2 = wto + 2 -
. (
X sm wto
(Jg)
+2
(9-2-39)
where (wto + Og/2) is the buncher cavity departure angle and wt2 - (Oo + Og/2) is
the catcher cavity arrival angle. Figure 9-2-10 shows the curves for the catcher cav-
Sec. 9.2
351
Klystrons
+ .!':
2
.,
Oh
i::
"'
-;
>
r----.
cz:...
fN
+
>~
.,"'....
..c:
u"'
7r
------'-----'-----'-------l
7r
0
2
o,
wto + 2
7r
- 7r
ity arrival angle as a function of the buncher cavity departure angle in terms of the
bunching parameter X.
Differentiation of Eq. (9-2-38) with respect to to results in
(9-2-40)
= - - - -lo- - - -
(9-2-41)
= - - - - -lo- - - - -
(9-2-42)
l2(to)
l2(t2)
1- X
COS
352
Chap. 9
X=O
-----::::::::'~
~~
/.. '/
"""--
-~~
~----
-'Ir
+'Ir
(e 0 + ~g)
The beam current at the catcher cavity is a periodic waveform of period 2TT / w
about de current. Therefore the current i2 can be expanded in a Fourier series and so
i2
= ao +
+ bn sin (nwt2)]
(9-2-43)
n=l
where n is an integer, excluding zero. The series coefficients ao, an, and bn in Eq.
(9-2-43) are given by the integrals
J"'
ao
an
= ;1
bn
= ;1
f"'
i2 d (wt2)
1TT
2 -.,,
_.,,
_.,,
(9-2-44)
ao =
1'TT
f"' lo d (wto) =
(9-2-45)
10
-.,,
an = -;
(9-2-46)
= -;
(9-2-47)
bn
cos (A B)
sin (A B)
= sin A cos B
and
cos A sin B
Sec. 9.2
353
Klystrons
the two integrals shown in Eqs. (9-2-40) and (9-2-41) involve cosines and sines of a
sine function. Each term of the integrand contains an infinite number of terms of
Bessel functions. These are
cos [ nX sin ( wt0 +
i)] =
+ 2[14(nX) cos
2(wto + i)]
4(wto + i)]
+ ...
(9-2-48)
and
sin [nx sin (wto +
i)] =
i)]
3( + i)]
+ 2[h(nX) sin
wto
+ ...
(9-2-49)
If these series are substituted into the integrands of Eqs. (9-2-46) and (9-2-47), re-
spectively, the integrals are readily evaluated term by term and the series
coefficients are
(9-2-50)
an
(9-2-51)
where Jn(nX) is the nth-order Bessel function of the first kind (see Fig. 9-2-12).
0.6
0.582
0.4
><:
5 0.2
.....~
....0
.,;::s
';;;
>
-0.2
-0.4
0
Argument of Jn(nX)
Figure 9-2-12
354
Chap. 9
i1 = Io
(9-2-52)
n=l
The fundamental component of the beam current at the catcher cavity has a magnitude
(9-2-53)
11 = 2/oli(X)
This fundamental component h has its maximum amplitude at
x=
1.841
(9-2-54)
L .
optimum
3.682vo Vo
wf3; Vi
(9-2-55)
It is interesting to note that the distance given by Eq. (9-2-31) is approximately 15%
less than the result of Eq. (9-2-55). The discrepancy is due in part to the approximations made in deriving Eq. (9-2-3 l) and to the fact that the maximum fundamental
component of current will not coincide with the maximun electron density along the
beam because the harmonic components exist in the beam.
(9-2-56)
where f3 0 is the beam coupling coefficient of the catcher gap. If the buncher and
catcher cavities are identical, then /3; = /30. The fundamental component of current
induced in the catcher cavity then has a magnitude
(9-2-57)
Figure 9-2-13 shows an output equivalent circuit in which Rsho represents the wall
resistance of catcher cavity, Rn the beam loading resistance, RL the external load resistance, and Rsh the effective shunt resistance.
Sec. 9.2
355
Klystrons
Figure 9-2-13
cuit.
The output power delivered to the catcher cavity and the load is given as
= (/3o/z) 2 R =
p
OU!
sh
f3ofz
Vi
(9-2-58)
where Rsh is the total equivalent shunt resistance of the catcher circuit, including the
load, and Vi is the fundamental component of the catcher gap voltage.
{30 /z Vz
=--
2/o Vo
(9-2-59)
in which the power losses to the beam loading and cavity walls are included.
If the coupling is perfect, {30 = I, the maximum beam current approaches
I2max = 2/o(0.582), and the voltage Vi is equal to Vo. Then the maximum electronic
efficiency is about 58%. In practice, the electronic efficiency of a klystron amplifier
is in the range of 15 to 30%. Since the efficiency is a function of the catcher gap
transit angle (JR, Fig. 9-2-14 shows the maximum efficiency of a klystron as a function of catcher transit angle.
60
~
50
40
c.,
u
.,....
;>,
.,:::
'.,
u
30
i;:
E
::I
;;:
:::;:"'
20
10
'Tr
'Tr
2
Catcher gap transit angle
(Jg
356
Chap. 9
Mutual conductance of a klystron amplifier. The equivalent mutual conductance of the klystron amplifier can be defined as the ratio of the induced output
current to input voltage. That is
IGm I ==
= 2{3ofoli(X)
izind
Vi
Vi
(9-2-60)
From Eq. (9-2-35) the input voltage Vi can be expressed in terms of the bunching
parameter X as
Vi
= 2Vo X
/30 Oo
(9-2-61)
/Gm/
{36 Oo li(X)
Go
(9-2-62)
g
~
.,u
10
...'"
""
"'
.......'"
i::
;::l
i::
0
i::
"".,
e....
'3080 V1
Bunching parameter X = - 2V0
Sec. 9.2
Klystrons
357
It can be seen from the curves that, for a small signal, the normalized
transconductance is maximum. That is,
IGml = {35Uo
Go
(9-2-63)
IGm I= 0.3l6{350o
(9-2-64)
Go
sh
(9-2-65)
(9-2-67)
where
The de power is
(9-2-68)
where Go = lo/Vo is the equivalent electron beam conductance. The power given by
the buncher cavity to produce beam bunching is
(9-2-69)
Chap. 9
358
Gs
Go
= Ro = F (8 )
Rs
(9-2- 70)
v2
v2
v2 +-2
v2
_2_=_2_+_2
2Rsh
2Rsho
2Rs
(9-2- 71)
2RL
Rsh
Rsho
Rs
RL
-=-+-+-
(9-2-72)
Finally, the loaded quality factor of the catcher cavity circuit at the resonant frequency can be written
l
I
l
I
(9-2-73)
-=-+-+QL
where QL
Qo
Qs
Qext
cf
--
r..:i""
"
=
g
"O
0=
0.21
0.20
f- -
--- ---
u
u
0.16
0.12
!::
~
0.08
"O
"
.~
'"ECi
0.04
0
--: /
0.4
0.8
/v
v ""
...............
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
1.2
1.6
2.0
2.4
Figure 9-2-16
angle.
- - - ---
2.8
3.2
: 3.6
3.5
e,.. radians
4.0
Sec. 9.2
359
Klystrons
Example 9-2-1:
Klystron Amplifier
Vo= 1000 V
Ro= 40 kfl
lo= 25 mA
= 3 GHz
d = 1 mm
l = 4cm
Rsh
= 30 kfl
Solution
a. For maximum Vi, J 1(X) must be maximum. This means l1(X) = 0.582 at
X = 1.841. The electron velocity just leaving the cathode is
v0
(0.593 x l06
)VVo = (0.593
106)Vi03
1.88
1 rad
107 mis
= w- =
21T(3 x 109 )
Vo
10- 3
1.88 X 101
=a =
/JI
fJO
Og/2
sin (1/2)
1/2
0. 952
Oo
= wT0 = w- =
21T(3 X 109 )
Vo
4 x 10- 2
01
1.8Xl
8
40 rad
_ 2VoX _ 2(103)(1.841) _
V
96 5
(0.952)(40) '
= {36 00 J 1(X) R =
A
v
Ro
sh
8 595
. n
Effi cte cy
= {30 12 Vi =
2/o V0
m
46 2 -io
360
Chap. 9
d. Calculate the beam loading conductance (refer to Fig. 9-2-13). The beam loading conductance GB is
GB =
8)
25 x 102
[(0.952) 2 - (0.952) cos (28.6)]
x 106 n
In comparison with RL and Rsho or the effective shunt resistance R,h , the beam
loading resistance is like an open circuit and thus can be neglected in the preceding calculations.
(a)
(b)
High efficiency and large power. In the 1960s much effort was devoted to
improving the efficiency of klystrons. For instance, a 50-kW experimental tube has
demonstrated 75% efficiency in the industrial heating band [8]. The VA-8840
klystron is a five-cavity amplifier whose operating characteristics are listed in Table
9-2-1.
One of the better-known high-peak-power klystrons is the tube developed
specifically for use in the 2-mile Stanford Linear Accelerator [9] at Palo Alto, California. A cutaway view of the tube is shown in Fig. 9-2-18. The operating characteristics of this tube are listed in Table 9-2-2.
The Varian CW superpower klystron amplifier VKC-8269A as shown in Fig.
9-2-19 has an output power of 500 kW (CW) at frequency of 2.114 GHz. Its power
Sec. 9.2
Klystrons
TABLE 921
361
VA-8840 OPERATING CHARACTERISTICS
Frequency
Power output
Gain
Efficiency
Electronic bandwidth (I dB)
Beam voltage
Beam current
5.9-6.45 GHz
14kW
52 dB
36%
75 MHz
16.5 kV
2.4 A
Frequency
RF pulse width
Pulse repetition rate
Peak power output
Beam voltage
Beam current
Gain
Efficiency
Weight of permanent (focusing) magnet
Electronic bandwidth (I dB)
2.856
2.5
60 to 360
24
250
250
50 to 55
about 36
363
20
GHz
s
pps
MW
kV
A
dB
%
kg
MHz
362
Chap. 9
gain is 56 dB and efficiency is 50 percent. The beam voltage is 62 V (de) and the
beam current is 16.50 A (de).
Long-life improvement. A new long-life klystron amplifier tube ~ith a design life in excess of ten years, three times the current design life, has been developed by the Electron Dynamics Division of the Hughes Aircraft Company. Key to
the long-life klystron was the development of a method of reducing the operating
temperature of the tube's cathode. The cathode is made of porous tungsten impregnated with barium, calcium, and aluminum oxides. The new cathode is coated with
a layer of osmium ruthenium alloy that lowers its work function, which is the temperature necessary for electrons to be emitted. This temperature reduction cuts evaporation of barium 10-fold and extends the life of the cathode.
Sec. 9.3
363
RF Output
RF Input
Intermediate cavities
Heater
Electron
beam
Magnet
coils
sides using the multistage techniques, the tube manufacturers have designed and produced multicavity klystron to serve the high-gain requirement as shown in Fig.
9-3-1.
In a multicavity klystron each of the intermediate cavities, placed at a distance
of the bunching parameter X of 1.841 away from the previous cavity, acts as a
buncher with the passing electron beam inducing a more enhanced RF voltage than
the previous cavity, which in turn sets up an increased velocity modulation. The
spacing between the consecutive cavities would therefore diminish because of the requirement of X being 1.841 and an increasing velocity modulating RF voltage as the
beam progresses through the various cavities. Keeping the intercavity distance constant, an increasing beam voltage Vo could be used in the subsequent cavities. Figure
9-3-2 shows the photograph of a four-cavity klystron amplifier.
364
Chap. 9
an infinitely wide beam, the electric fields (and, thus, the space-charge forces) are
constrained to act only in the axial direction. In a finite beam, the electric fields are
radial as well as axial with the result that the axial component is reduced in comparison to the infinite beam. With reduced axial space-charge force, the plasma frequency is reduced and the plasma wavelength is increased.
Mathematically, let the charge-density and velocity perturbations be simple sinusoidal variations in both time and position. They are
Charge density:
Velocity perturbation:
"V
(9-3-1)
(9-3-2)
(3, = ;
0
Wq
wp
= '-Vrep;;
mE;; is the plasma frequency and is a function of the electron-beam
(J =
density
phase angle of oscillation
Sec. 9.3
365
The electron plasma frequency is the frequency at which the electrons will oscillate in the electron beam. This plasma frequency applies only to a beam of infinite
diameter. Practical beams of finite diameter are characterized by plasma frequency
that is less than wp . This lower plasma frequency is called the reduced plasma frequency and is designated Wq The space-charge reduction factor R is a function of
the beam radius r and the ratio n of the beam-tunnel radius to the beam radius as
shown in Fig. 9-3-3 [IO].
1.0
0.6
0.4
0.2
"-
l_
3""
0.1
II
Q:;
0.06
0.04
0.02
0.01
0.1
0.2
0.5
1.0
2.0
5.0
10.0
{3,r = wr/u0
Figure 9-3-3
of Pergamon Press.)
= -po+ P
= Ifo + If
(9-3-3)
(9-3-4)
Chap. 9
366
= - lo+ l
(9-3-5)
(9-3-6)
(9-3-7)
VJ
where J = p"V0
= - op
-
(9-3-8)
ot
(9-3-9)
- op
ot
(9-3-10)
The beam-current density and the modulated velocity are expressed as [10]
J = "VoB cos (/3,z - wt) cos (wqt
+ 8)
(9-3-12)
and
"V
(9-3-13)
Sec. 9.3
367
In practical microwave tubes, the ratio of wq/w is much smaller than unity and the
second term in Eq. (9-3-12) may be neglected in comparison with the first. Then we
obtain
J = V 0 B cos (j3,z - wt) cos (wqt + O)
(9-3-14)
Example 9-3-1:
Four-Cavity Klystron
Beam voltage:
Beam current:
Operating frequency:
de electron charge density:
RF charge density:
Velocity perturbation:
lo
1.4 A
f = JO GHz
6
3
Po= 10- C/m
8
p= 10- C/m 3
'V
105 mis
Compute:
a.
b.
c.
d.
e.
f.
The
The
The
The
The
The
de electron velocity
de phase constant
plasma frequency
reduced plasma frequency for R = 0.4
de beam current density
instantaneous beam current density
Solution
'V0
(3, = 0.
x JOJO
714
wp = ( l.759 x 10 11 x .
x
_
10 12
8 854
)1/2
368
Chap. 9
The electrons leaving the input gap of a klystron amplifier have a velocity
given by Eq. (9-2-17) at the exit grid as
lf(t1)
=
=
= gap distance
where Vi
vo[1
(9-3-15)
Since the electrons under the influence of the space-charge forces exhibit simple harmonic motion, the velocity at a later time tis given by
lftoC
= Vo[ 1 +
- WpT)]
(9-3-16)
where f3q
= ;: is the
Example 9-3-2:
1 low
.
~ f3; Vi sm {f3qz) cos (13,z - wt)
2 voWq
=--
Vo= 18 kV
Io= 2.25 A
d = 1 cm
= 10 GHz
Vi = 10 V (rms)
/30 = {3; =
Po = 10-s C/m 3
Determine:
(9-3-17)
Sec. 9.3
369
Solution
a. The de electron velocity is
x 10- 10n
=
(9-3-18)
where V1 = magnitude of the input signal voltage. Then the magnitudes of the induced current and voltage in the output cavity are equal to
(9-3-19)
and
(9-3-20)
370
where f3o
Rsht
Chap. 9
= total
(9-3-21)
. _ Pout _
Power gam - Pin -
Pout
_ 1 ( low ) 4
IV, l 2/Rsh - 4 VoWq f3oRsh
Rsht
(9-3-22)
where Rsh = total shunt resistance of the input cavity. The electronic efficiency of a
two-cavity klystron amplifier is
T/
= Pout = Pout = !
Pin
Example 9-3-3:
lo Vo
(lo) (I Vi lw)
4 Vo
{3gRshl
VoWq
(9 _3_23 )
Vo= 20 kV
lo= 2 A
= 8 GHz
f3, = f3o = 1
Po = 10- 6 C/m 3
Vi = 10 V (rms)
R,h = 10 kO
R = 30 kO
Calculate:
a.
b.
c.
d.
e.
Solution
a. The plasma frequency is
wp =
Sec. 9.3
371
I/z I = 2 20
2
x 103 x 713 x I2 x 110 I = 0.3565 A
IVzl
II2IRshl
103 = 3.82 kW
T/ =
Pin
3.82 X 103
= 2 x 20 x 103 = 9 6 %
= _!_ (low
4
v.oWq
)zf331 V IR
Olsh
(9-3-24)
372
Chap. 9
(9-3-25)
This voltage produces a velocity modulation again and is converted into an RF convection current at the output cavity for four-cavity klystron as
low {3; IVi I
I14. I = -21 -v.
oWq
3
_ 1 (low )
-S
Vowq
and
Ih I = /3ol i4 I = ~ (
(9-3-26)
/30IVi IRsh
2
:::J
/381 Vi IR;h
(9-3-27)
:::JJ
/381 Vi IR;hRshl
(9-3-28)
The output power from the output cavity in a four-cavity klystron amplifier can be
expressed as
6
Pout
_ Il4 12 Rshl -_ 64
1 (low
4 2
VoWq ) /30121 Vi 12 RshRshl
(9-3-29)
where Rsht = total shunt resistance of the output cavity including the external load.
The multicavity klystrons are often operated with their cavities stagger-tuned
so as to obtain a greater bandwidth at some reduction in gain. In high-power
klystrons, the cavity grids are omitted, because they would burn up during beam interception. High-power klystron amplifiers with a power gain of 40 to 50 dB and a
bandwidth of several percent are commercially available.
Example 9-3-4:
Vo= lO kV
lo= 0.7 A
f = 4 GHz
{3; = /30 = l
Po = 5 X 10-s C/m3
Vi = 2 V (rms)
Rsh = 10 k!l
Rsh1 = 5 k!l
373
Reflex Klystrons
Sec. 9.4
Determine:
a.
b.
c.
d.
e.
The
The
The
The
The
plasma frequency
reduced plasma frequency for R = 0.6
induced current in the output cavity
induced voltage in the output cavity
output power delivered to the load
Solution
= [ l.759
10 11
5 x 10-5 ]1/2
_
X
_
= 0.997
8 854
10 12
109 rad/s
= 27T
4 x 109/(0.598 x 109 )
= 42.03
1(0~7x 42.03) x 1
8 10
3
1141
= -
x 121x(IOx103) 2
= 0.6365 A
103
= 2.03 kW
has a magnitude of unity with a phase shift of multiple 27T, the klystron will oscillate. However, a two-cavity klystron oscillator is usually not constructed because,
when the oscillation frequency is varied, the resonant frequency of each cavity and
the feedback path phase shift must be readjusted for a positive feedback. The reflex
klystron is a single-cavity klystron that overcomes the disadvantages of the twocavity klystron oscillator. It is a low-power generator of IO to 500-mW output at a
frequency range of I to 25 GHz. The efficiency is about 20 to 30%. This type is
widely used in the laboratory for microwave measurements and in microwave receivers as local oscillators in commercial, military, and airborne Doppler radars as well
as missiles. The theory of the two-cavity klystron can be applied to the analysis of
374
Chap. 9
the reflex klystron with slight modification. A schematic diagram of the reflex
klystron is shown in Fig. 9-4-1.
The electron beam injected from the cathode is first velocity-modulated by the
cavity-gap voltage. Some electrons accelerated by the accelerating field enter the repeller space with greater velocity than those with unchanged velocity. Some electrons decelerated by the retarding field enter the repeller region with less velocity.
All electrons turned around by the repeller voltage then pass through the cavity gap
in bunches that occur once per cycle. On their return journey the bunched electrons
pass through the gap during the retarding phase of the alternating field and give up
their kinetic energy to the electromagnetic energy of the field in the cavity. Oscillator output energy is then taken from the cavity. The electrons are finally collected by
the walls of the cavity or other grounded metal parts of the tube. Figure 9-4-2 shows
an Applegate diagram for the 1~ mode of a reflex klystron.
v0 [ 1
vVo
(9-4-1)
8
.
(
Vo sm
wti - 0 ) ]
/31V1
(9-4-2)
This expression is identical to Eq. (9-2-17), for the problems up to this point are
identical to those of a two-cavity klystron amplifier. The same electron is forced
back to the cavity z = d and time tz by the retarding electric field E, which is given
by
E=
Vr
Vo
+ Vi sin
(wt)
(9-4-3)
~~~~~~~~
This retarding field E is assumed to be constant in the z direction. The force equation for one electron in the repeller region is
2
m d z
dt 2
-eE
-e Vr
Vo
(9-4-4)
where E = -VV is used in the z direction only, Vr is the magnitude of the repeller
voltage, and IV1 sin wt I ~ (Vr + Vo) is assumed.
Integration of Eq. (9-4-4) twice yields
dz_ -e(Vr +Vo)
dt
mL
f' dt -_
11
-e(Vr +Vo)( _
mL
ti
(9-4-5)
Sec. 9.4
Reflex Klystrons
375
l---...-.
Repeller
Vs= V 1 sin wt
v,
+ tot1,t2
11-1---ld----I-
l +d
t0
t1
Electron beam is
accelerated during
this half cycle.
Figure 9-4-2
Electron beam is
decelerated during
this half cycle.
Applegate diagram with gap voltage for a reflex klystron.
376
at t
-e(V, + Vo)
mL
t1
at t
Chap. 9
ti, z
t1
= K1; then
(9-4-6)
On the assumption that the electron leaves the cavity gap at z = d and time t 1 with a
velocity of v(t,) and returns to the gap at z = d and time tz, then, at t = t2 , z = d,
t1
. (
(Jg)]
= tz - t, = e(V,2mL
+ Vo) v (t, ) = To'[ 1 + f3;V1
2Vo sm wt, - 2
(9-4-7)
where
T' _
0
2mLvo
e(V, + Vo)
(9-4-8)
w(t2
i)
(9-4-9)
where
Ob= wTb
(9-4-10)
2Vo
(9-4-11)
In order for the electron beam to generate a maximum amount of energy to the oscillation, the returning electron beam must cross the cavity gap when the gap field is
maximum retarding. In this way, a maximum amount of kinetic energy can be transferred from the returning electrons to the cavity walls. It can be seen from Fig.
Sec. 9.4
Reflex Klystrons
377
9-4-2 that for a maximum energy transfer, the round-trip transit angle, referring to
the center of the bunch, must be given by
(9-4-12)
n-
iz, = -Io -
(9-4-13)
n=l
The fundamental component of the current induced in the cavity by the modulated
electron beam is given by
iz
= - {3;/2 =
(9-4-14)
in which ()11 has been neglected as a small quantity compared with Oh. The magnitude of the fundamental component is
/z
= 2/of3J1(X ')
(9-4-15)
= Volo
(9-4-16)
2Vi /z
= Vilof3J,(X)
(9-4-17)
Vo
2X'
{3;(27Tn - 7T /2)
(9-4-18)
= 2VofoX' J,(X')
ac
27Tn - 7T /2
(9-4-19)
2X' J,(X')
27Tll - 7T 12
(9-4-20)
378
Chap. 9
The factor X 'J,(X ') reaches a maximum value of 1.25 at X' = 2.408 and J,(X ') =
0.52. In practice, the mode of n = 2 has the most power output. If n = 2 or 1 ~
mode, the maximum electronic efficiency becomes
.
- 2(2.408)1,(2.408) - 22 701.
Effic1encymax 27T (2) - 7T /2 . -10
(9-4-21)
0.8
~
~
0.6
0.4
0.2
0.5
1.0
1.5
2.5
2.0
3.0
3.5
4.0
2.408
X'
Figure 9-4-3
For a given beam voltage Vo , the relationship between the repeller voltage and
cycle number n required for oscillation is found by inserting Eqs. (9-4-12) and
(9-4-1) into Eq. (9-4-8):
Vo
(27Tn - 7T /2) 2 e
8w 2 L2
(9-4-22)
The power output can be expressed in terms of the repeller voltage Vr. That is,
p
le
ac -
(9 _4 _23 )
It can be seen from Eq. (9-4-22) that, for a given beam voltage Vo and cycle number
n or mode number N, the center repeller voltage Vr can be determined in terms of
the center frequency. Then the power output at the center frequency can be calculated from Eq. (9-4-23). When the frequency varies from the center frequency and
the repeller voltage about the center voltage, the power output will vary accordingly,
assuming a bell shape (see Fig. 9-4-4).
Sec. 9.4
379
Reflex Klystrons
10.04
I
:I:
".5G
.,c: lo.oo
::I
O'
10.02
9.98
.v
9.96
400
~
300
.5
;;
214
mode
mode
0.
200
....
~
100
314
I~'
I'
(\
\
I \
r\
mode/
'
0
11.
0
0
100
200
300
400
500
600
'
700
800
900
1000
= 2lof3J1 (X ')e-j80
t2
(9-4-24)
(9-4-26)
The amplitude of the phasor admittance indicates that the electronic admittance is a
function of the de beam admittance, the de transit angle, and the second transit of
the electron beam through the cavity gap. It is evident that the electronic admittance
is nonlinear, since it is proportional to the factor 211(X ')/X ', and X' is proportional
to the signal voltage. This factor of proportionality is shown in Fig. 9-4-5. When
the signal voltage goes to zero, the factor approaches unity.
380
1.0
~
~
......
N
....
--.........
0.8
0.6
(.)
c:
0.4
e:s
....
~ '-
""'
~i".
Chap. 9
0.2
0
0
"'
'-
~
4
Figure 9-4-5
factor.
The equivalent ciruit of a reflex klystron is shown in Fig. 9-4-6. In this circuit
L and C are the energy storage elements of the cavity; Ge represents the copper
losses of the cavity, Gb the beam loading conductance, and Ge the load conductance.
The necessary condition for oscillations is that the magnitude of the negative
real part of the electronic admittance as given by Eq. (9-4-26) not be less than the
total conductance of the cavity circuit. That is,
I-Gel~ G
where G
(9-4-27)
(9-4-28)
Since the electronic admittance shown in Eq. (9-4-26) is in exponential form, its
phase is 7T /2 when
is zero. The rectangular plot of the electron admittance Ye is a
spiral (see Fig. 9-4-7). Any value of for which the spiral lies in the area to the left
of line (-G - jB) will yield oscillation. That is,
eo
eo
~=G-0~=N~
~4m
where N is the mode number as indicated in the plot, the phenomenon verifies the
early analysis.
Oscillation region -
Nonoscillation region
/-
+jB,
I
I
;so
-c,
60
-80
80
+G,
Conductance
mhos
-;so
-G -jB /..- G
I
-;B,
Reflex Klystron
Vo=600V
R,h = l5kfl
fr= 9 GHz
The tube is oscillating at fr at the peak of then = 2 mode or I~ mode. Assume that the
transit time through the gap and beam loading can be neglected.
a. Find the value of the repeller voltage V, .
b. Find the direct current necessary to give a microwave gap voltage of 200 V.
c. What is the electronic efficiency under this condition?
Solution
a. From Eq. (9-4-22) we obtain
Vo
(V,
Vo)
(!...) (21Tn m
1T /2)
8w L2
II
(21T2 - 1T/2)2
X 10-3
(I. 759 x IO ) S( 21T x 9 x l09)2(I0_ 3) 2 - 0.832
381
382
(V,
Vo) 2
600
x _3
0 832
10
= _
v,
250
Chap. 9
= 0.721 x 106
1. Since
Vi= /zRsh
= 2/of1(X')Rsh
Vi
200
Since Kompfner invented the helix traveling-wave tube (TWT) in 1944 [11], its basic
circuit has changed little. For broadband applications, the helix TWTs are almost exclusively used, whereas for high-average-power purposes, such as radar transmitters,
the coupled-cavity TWTs are commonly used.
In previous sections klystrons and reflex klystrons were analyzed in some detail. Before starting to describe the TWT, it seems appropriate to compare the basic
operating principles of both the TWT and the klystron. In the case of the TWT, the
microwave circuit is nonresonant and the wave propagates with the same speed as
the electrons in the beam. The initial effect on the beam is a small amount of velocity modulation caused by the weak electric fields associated with the traveling wave.
Just as in the klystron, this velocity modulation later translates to current modulation, which then induces an RF current in the circuit, causing amplification. However, there are some major differences between the TWT and the klystron:
1. The interaction of electron beam and RF field in the TWT is continuous over
the entire length of the circuit, but the interaction in the klystron occurs only
at the gaps of a few resonant cavities.
2. The wave in the TWT is a propagating wave; the wave in the klystron is not.
3. In the coupled-cavity TWT there is a coupling effect between the cavities,
whereas each cavity in the klystron operates independently.
A helix traveling-wave tube consists of an electron beam and a slow-wave structure. The electron beam is focused by a constant magnetic field along the electron
beam and the slow-wave structure. This is termed an 0-type traveling-wave tube.
The slow-wave structure is either the helical type or folded-back line. The applied
signal propagates around the turns of the helix and produces an electric field at the
Sec. 9.5
383
center of the helix, directed along the helix axis. The axial electric field progresses
with a velocity that is very close to the velocity of light multiplied by the ratio of helix pitch to helix circumference. When the electrons enter the helix tube, an interaction takes place between the moving axial electric field and the moving electrons.
On the average, the electrons transfer energy to the wave on the helix. This interaction causes the signal wave on the helix to become larger. The electrons entering the
helix at zero field are not affected by the signal wave; those electrons entering the
helix at the accelerating field are accelerated, and those at the retarding field are decelerated. As the electrons travel further along the helix, they bunch at the collector
end. The bunching shifts the phase by 1T /2. Each electron in the bunch encounters a
stronger retarding field. Then the microwave energy of the electrons is delivered by
the electron bunch to the wave on the helix. The amplification of the signal wave is
accomplished. The characteristics of the traveling-wave tube are:
Frequency range:
3 GHz and higher
Bandwidth:
about 0.8 GHz
Efficiency:
20 to 40%
up to 10 kW average
Power output:
Power gain:
up to 60 dB
The present state of the art for U.S. high-power TWTs is shown in Fig. 9-5-1.
0.6
1-------C----t-4---+---+-+--------t------r--+-+-------<-r-<
01------+--r-+---+-->-+---+---l~j----+--+-+------+---t-~
021----t---l--+--+-+-+----l-'--'-!--+-+-+--f--+--l
010':-1--:"':0.2,....-;:'o.J=--'"-o"".~~~,~~~;~.~,.,.,o_-,20~.___.,o
FREQUENCY, GHz
384
Chap. 9
(a)
(b)
(d)
(c)
(c)
Figure 9-5-2 Slow-wave structures. (a) Helical line. (b) Folded-back line. (c)
Zigzag line. (d) Interdigital line. (e) Corrugated waveguide.
Slow-wave structures are special circuits that are used in microwave tubes to
reduce the wave velocity in a certain direction so that the electron beam and the signal wave can interact. The phase velocity of a wave in ordinary waveguides is
greater than the velocity of light in a vacuum. In the operation of traveling-wave and
magnetron-type devices, the electron beam must keep in step with the microwave
signal. Since the electron beam can be accelerated only to velocities that are about a
fraction of the velocity of light, a slow-wave structure must be incorporated in the
microwave devices so that the phase velocity of the microwave signal can keep pace
with that of the electron beam for effective interactions. Several types of slow-wave
structures are shown in Fig. 9-5-2.
The commonly used slow-wave structure is a helical coil with a concentric conducting cylinder (see Fig. 9-5-3).
It can be shown that the ratio of the phase velocity Vp along the pitch to the
phase velocity along the coil is given by
Vp
Yp2 + (1Td)2
= sin I/I
(9-5-1)
Sec. 9.5
Figure 9-5-3
where c
385
Helical slow-wave structure. (a) Helical coil. (b) One turn of helix.
p = helix pitch
d = diameter of the helix
Y,E[p
p
2
+ (rrd)2]
(9-5-2)
If the dielectric constant is too large, however, the slow-wave structure may intro-
duce considerable loss to the microwave devices, thereby reducing their efficiency.
For a very small pitch angle, the phase velocity along the coil in free space is approximately represented by
v =pc = ~
(9-5-3)
f3
1Td
Figure 9-5-4 shows the w-{3 (or Brillouin) diagram for a helical slow-wave
structure. The helix w-/3 diagram is very useful in designing a helix slow-wave structure. Once f3 is found, Vp can be computed from Eq. (9-5-3) for a given dimension
of the helix. Furthermore, the group velocity of the wave is merely the slope of the
curve as given by
aw
Vgr
(9-5-4)
= a{3
I ~=c
I (3
I
I
I
I
I
I
I
I
I-~-
f3
Figure 9-5-4
structure.
386
Chap. 9
E(x, y, z - L)
= E(x,
y, z)eH3oL
(9-5-5)
where E(x, y, z) is a periodic function of z with periodL. Since /30 is the phase constant in the axial direction, in a slow-wave structure /30 is the phase constant of average electron velocity.
It is postulated that the solution to Maxwell's equations in a periodic structure
can be written
E(x, y, z)
= f(x,
y, z)e-jf3oZ
(9-5-6)
where f (x, y, z) is a periodic function of z with period L that is the period of the
slow-wave structure.
For a periodic structure, Eq. (9-5-6) can be rewritten with z replaced by
z - L:
(9-5-7)
f(x, y, z - L)
= f(x,
y, z)
(9-5-8)
y, z)e-jf3ozejf3oL
(9-5-9)
E(x, y, z - L)
= f(x,
E(x, y, z - L)
= E(x, y,
z)ejf3oL
(9-5-10)
n= -oo
Sec. 9.5
387
where
1
En(X, y) = L
IL E(x, y, z)ej(Z7Tn/L)Z dz
(9-5-12)
{3n =Bo+
27Tn
(4-5-13)
-00
'.
'-2, -1, 0, l, 2, 3,
00
The quantities En(x, y)e-jf3.z are known as spatial harmonics by analogy with
time-domain Fourier series. The question is whether Eq. (9-5-11) can satisfy the
electric wave equation, Eq. (2-1-20). Substitution of Eq. (9-5-11) into the wave
equation results in
=0
(9-5-14)
It is evident from the preceding equation that if each spatial harmonic is itself a solution of the wave equation for each value of n, the summation of space harmonics also
satisfies the wave equation of Eq. (9-5-14). This means that only the complete solution of Eq. (9-5-14) can satisfy the boundary conditions of a periodic structure.
Furthermore, Eq. (9-5-1 l) shows that the field in a periodic structure can be
expanded as an infinite series of waves, all at the same frequency but with different
phase velocities Vpn That is
v
The group velocity
Vgr '
pn
defined by
f3n
Vgr
{30
+ (27Tn/ L)
= [d({3o +
Vgr
(9-5-16)
=-=-----
27rn/L)]-I
dw
=~
a{3o
(9-5-17)
which is independent of n.
It is important to note that the phase velocity Vpn in the axial direction decreases for higher values of positive n and f3o . So it appears possible for a microwave
of suitable n to have a phase velocity less than the velocity of light. It follows that
interactions between the electron beam and microwave signal are possible and thus
the amplification of active microwave devices can be achieved.
Figure 9-5-5 shows the w-{3 (or Brillouin) diagram for a helix with several
spatial harmonics. This w-{3 diagram demonstrates some important properties needing more explanation. First, the second quadrant of the w-{3 diagram indicates the
negative phase velocity that corresponds to the negative n. This means that the electron beam moves in the positive z direction while the beam velocity coincides with
388
w
Chap. 9
If= c
Forbidden region
4ir
6ir (3
the negative spatial harmonic's phase velocity. This type of tube is called a
backward-wave oscillator. Second, the shaded areas are the forbidden regions for
propagation. This situation occurs because if the axial phase velocity of any spatial
harmonic exceeds the velocity of light, the structure radiates energy. This property
has been verified by experiments [10].
(9-5-19)
To= -
Vo
/3n = w =
Vo
01
Vo To
(9-5-20)
In Eq. (9-5-20) the axial space-harmonic phase velocity is assumed to be synchronized with the beam velocity for possible interactions between the electron beam
and electric field. That is,
(9-5-21)
Vnp = Vo
= - eE
and
E =
- VV
(9-5-22)
Sec. 9.5
Control anode
Cathode
Heater
r-~H~e-at-er
__, ___..,...._,
389
Electron beam
focusing magnet
rCollector
supply
Helix
RF input
RF output
Gain or modulation
control voltage
Regulated
beam supply
+
Collector
supply
(a)
RF input
Anode
RF output
-----~
Cathode
Gun
assembly
Collector
Heater
(b)
Figure 9-5-6 Diagram of helix traveling-wave tube: (a) schematic diagram of helix
traveling-wave tube; (b) simplified circuit.
The electrons entering the retarding field are decelerated and those in the accelerating field are accelerated. They begin forming a bunch centered about those electrons
that enter the helix during the zero field. This process is shown in Fig. 9-5-7.
Since the de velocity of the electrons is slightly greater than the axial wave velocity, more electrons are in the retarding field than in the accelerating field, and a
great amount of energy is transferred from the beam to the electromagnetic field.
The microwave signal voltage is, in turn, amplified by the amplified field. The
bunch continues to become more compact, and a larger amplification of the signal
voltage occurs at the end of the helix. The magnet produces an axial magnetic field
to prevent spreading of the electron beam as it travels down the tube. An attenuator
placed near the center of the helix reduces all the waves traveling along the helix to
nearly zero so that the reflected waves from the mismatched loads can be prevented
from reaching the input and causing oscillation. The bunched electrons emerging
390
Chap. 9
.....u
c""'
u"'
<J t;:;
[jj
Accelerating force
Figure 9-5-7
from the attenuator induce a new electric field with the same frequency. This field,
in turn, induces a new amplified microwave signal on the helix.
The motion of electrons in the helix-type traveling-wave tube can be quantitatively analyzed in terms of the axial electric field. If the traveling wave is propagating in the z direction, the z component of the electric field can be expressed as
Ez
= E1 sin
(wt - {3pz)
(9-5-23)
~~ =
(9-5-24)
(wet + Oe)
(9-5-25)
(9-5-26)
Vo
+ Ve
COS
Then
dv
dt
where
Vo = de electron velocity
Ve = magnitude of velocity fluctuation in the velocity-modulated electron
beam
we = angular frequency of velocity fluctuation
Oe = phase angle of the fluctuation
(9-5-27)
For interactions between the electrons and the electric field, the velocity of the
velocity-modulated electron beam must be approximately equal to the de electron
velocity. This is
V =Vo
(9-5-28)
Sec. 9.5
391
(9-5-29)
Vo(t - to)
and
mVeWe sin (wet
(9-5-30)
Comparison of the left and right-hand sides of Eq. (9-5-30) shows that
eEi
Ve= - -
(9-5-31)
mw,
We
Oe
Vo)
It can be seen that the magnitude of the velocity fluctuation of the electron beam is
directly proportional to the magnitude of the axial electric field.
9.5.3 Convection Current
In order to determine the relationship between the circuit and electron beam quantities, the convection current induced in the electron beam by the axial electric field
and the microwave axial field produced by the beam must first be developed. When
the space-charge effect is considered, the electron velocity, the charge density, the
current density, and the axial electric field will perturbate about their averages or de
values. Mathematically, these quantities can be expressed as
V
p
1
= Vo + Vi ejwt-yz
= Po + Pi ejw1-yz
= - lo + li ejwt-yz
(9-5-32)
(9-5-33)
(9-5-34)
(9-5-35)
where y = a, + jf3, is the propagation constant of the axial waves. The minus sign
is attached to lo so that lo may be a positive in the negative z direction. For a small
signal, the electron beam-current density can be written
(9-5-36)
where - lo = povo, li = Pi Vo + po vi , and Pi Vi = 0 have been replaced. If an axial electric field exists in the structure, it will perturbate the electron velocity according to the force equation. Hence the force equation can be written
-dv =
dt
e
1.wt-yz
--E,e
m
dz a)V =
= ( -a + at
dt az
( JW
. - ')'Vo)Vie 1.wt-yz
(9-5-37)
-e/m
Ei
')'Vo
)W -
(9-5-38)
392
Chap. 9
In accordance with the law of conservation of electric charge, the continuity equation can be written
(9-5-39)
It follows that
(9-5-40)
PiVo
povi
(9-5-44)
gives
w e
lo
li=J-Ei
Vo m (jw - yv0 ) 2
(9-5-42)
where - lo = povo has been replaced. If the magnitude of the axial electric field is
uniform over the cross-sectional area of the electron beam, the spatial ac current i
will be proportional to the de current lo with the same proportionality constant for 11
and lo. Therefore the convection current in the electron beam is given by
.
I
(3,lo
E
- y)z t
= } 2Vo(}f3,
(9-5-43)
where (3, ""' w/vo is defined as the phase constant of the velocity-modulated electron
beam and vo = V(2e/m)Vo has been used. This equation is called the electronic
equation, for it determines the convection current induced by the axial electric field.
If the axial field and all parameters are known, the convection current can be found
by means of Eq. (9-5-43).
9.5.4 Axial Electric Field
The convection current in the electron beam induces an electric field in the slowwave circuit. This induced field adds to the field already present in the circuit and
causes the circuit power to increase with distance. The coupling relationship between the electron beam and the slow-wave helix is shown in Fig. 9-5-8.
ili
~I + iJ: <f:
{--____..
=:=
Lkctron heam
' - ()id-
il: -
Sec. 9.5
393
---
cav
ai
--at
az
(9-5-44)
Then
-yl
= - jwCV + yi
(9-5-45)
(9-5-46)
=-
(9-5-47)
Similarly,
-yV
jwL/
Elimination of the circuit current I from Eqs. (9-5-45) and (9-5-46) yields
(9-5-48)
If the convection-elecron beam current is not present, Eq. (9-5-48) reduces to a typical wave equation of a transmission line. When i = 0, the propagation constant is
defined from Eq. (9-5-48) as
Yo= jwvLC
(9-5-49)
and the characteristic impedance of the line can be determined from Eqs. (9-5-45)
and (9-5-47):
Zo=
(9-5-50)
394
Chap. 9
When the electron beam current is present, Eq. (9-5-48) can be written in terms of
Eqs. (9-5-49) and (9-5-50):
V
Since Ez =
- VV =
YYoZo i
y2 - y5
(9-5-51)
__ Y2 YoZo .
2
21
Y - Yo
EI
(9-5-52)
This equation is called the circuit equation because it determines how the axial electric field of the slow-wave helix is affected by the spatial ac electron beam current.
= jf3,
(9-5-54)
(9-5-55)
= (/oZo)I/3
(9-5-56)
4Vo
It can be seen from Eq. (9-5-55) that there are three forward traveling waves corresponding to e - jf3,z and one backward traveling wave corresponding to e+jf3,z. Let the
propagation constant of the three forward traveling waves be
jf3, - (3, cs
(9-5-57)
2c 313;(-13; -
2j13;cs + 13;c 2s 2)
(9-5-58)
Sec. 9.5
Since Cl>
395
= (- j)l/3
(9-5-59)
From the theory of complex variables the three roots of (- j) can be plotted in Fig.
9-5-9.
J
-I
---+--->1<,..-----1-.--- Real
Srr
- ! n =O
6'
n =I
-J
Figure 9-5-9
8 = (- j)l/3 =
(n
e-j((7T/2+2mr)/3)
= 0,
1, 2)
(9-5-60)
1 is
82
= e-j57T16 = - V3 - I!
2
(9-5-62)
c2
84 = - j -
(9-5-65)
Thus the values of the four propagation constants I' are given by
v; }{3.(1 ~)
v; }{3.(1 ~)
+
1'1 = -(3,C
1'2 = (3,C
1'3
= }(3,(1
- C)
(9-5-66)
(9-5-67)
(9-5-68)
1'4
=-
j(3,( 1 -
~)
(9-5-69)
396
Chap. 9
These four propagation constants represent four different modes of wave propagation
in the 0-type helical traveling-wave tube. It is concluded that the wave corresponding to '}'1 is a forward wave and that its amplitude grows exponentially with distance;
the wave corresponding to '}'2 is also a forward wave, but its amplitude decays exponentially with distance; the wave corresponding to '}'3 is also a forward wave, but its
amplitude remains constant; the fourth wave corresponding to '}'4 is a backward
wave, and there is no change in amplitude. The growing wave propagates at a phase
velocity slightly lower than the electron beam velocity, and the energy flows from
the electron beam to the wave. The decaying wave propagates at the same velocity as
that of the growing wave, but the energy flows from the wave to the electron beam.
The constant-amplitude wave travels at a velocity slightly higher than the electron
beam velocity, but no net energy exchange occurs between the wave and the electron
beam. The backward wave progresses in the negative z direction with a velocity
slightly higher than the velocity of the electron beam inasmuch as the typical value
of C is about 0.02.
V(z)
Vne-ynz
(9-5-70)
n=l
1(z)
lo Vn
-2: - -e-'>'n 1
2VoC 8~
2
n=I
(9-5-71)
Vi ()
Z =
"'.VoVn_
L,, 1 - - -e '>'n
n=l
2VoC 8n
(9-5-72)
Vi
+ Vi + Vi
lo
(Vi
z(O) = - 2VoC 2 8r
(9-5-73)
Vi
Vi)
+ 8~ + 8j
(9-5-74)
Sec. 9.5
397
. vo= -1
v,(O)
2VoC 81
(9-5-75)
82
The simultaneous solution of Eqs. (9-5- 73), (9-5-74), and (9-5-75) with i(O)
and v,(O) = 0 is
(9-5- 76)
Since the growing wave is increasing exponentially with distance, it will predominate over the total voltage along the circuit. When the length C of the slow-wave
structure is sufficiently large, the output voltage will be almost equal to the voltage
of the growing wave. Substitution of Eqs. (9-5-66) and (9-5-76) in Eq. (9-5- 70)
yields the output voltage as
- exp
V(C) = -V (O)
3
(v3
) [ jf3,. ( + 2C) C]
2 13,CC
exp -
(9-5-77)
The factor 13,C is conventionally written 2TTN, where N is the circuit length in electronic wavelength-that is,
a.
,._,,
and
21T
A,
(9-5-78)
V (C)
V(O)
3
r:
= - - exp ( v 3 TTNC)
(9-5- 79)
(C)
Ap == 10 log V
V(O)
2
1
-9.54
+ 47.3NC
dB
(9-5-80)
Vo= 3 kV
Beam voltage:
Beam current:
lo = 30 mA
Characteristic impedance of helix:
Circuit length:
N = 50
Frequency:
f = 10 GHz
Zo=lOO
398
Chap. 9
Determine: (a) the gain parameter C; (b) the output power gain Ar in decibels; and
(c) all four propagation constants.
Solution
a. From Eq. (9-5-56) the gain parameter is
3
13
13
_ (loZo)' _ (30 x 10- X 10)' _ 2 92
_2
C 4Vo
4 x 3 x 103
x 10
f3, =
W
Vo
/11 = -{3,C
=
27T X } 0!0
v:
/13
+ jf3,(l +
~)
2 92 x 10- 2)
-1.93 x 103 x 2.92 x 10- 2 x 0.87 + j 1.93 x 103( 1 + . 2
= -49.03
Y2 =
f3,C
v:
+ j 1952
+ jf3,(l +
~) =
49.03 + }1952
. (
y 4 = - jf3, I -
C )
4
= - J 1. 93
2
x 103 [ I - (2. 92 x 10-
)3]
= - j 1930
96 1 Physical Description
The term coupled cavity means that a coupling is provided by a long slot that
strongly couples the magnetic component of the field in adjacent cavities in such a
manner that the passband of the circuit is mainly a function of this one variable. Figure 9-6-1 shows two coupled-cavity circuits that are principally used in travelingwave tubes.
Sec. 9.6
Cavity
399
Electron
beam
Drift
tube
Coupling hole
(a)
Magnet
Iron
pole piece
(b)
Figure 9-6-1 Coupled-cavity circuits in the TWTs. (a) Basic coupled-cavity circuit. (b) Coupled-cavity circuit with integral periodic-permanent-magnet (PPM) focusing. (After J. T. Mendel [ 15 ]; reprinted by permission of JEEE.)
As far as the coupling effect is concerned, there are two types of coupledcavity circuits in traveling-wave tubes. The first type consists of the fundamentally
forward-wave circuits that are normally used for pulse applications requiring at least
half a megawatt of peak power. These coupled-cavity circuits exhibit negative mutual inductive coupling between the cavities and operate with the fundamental space
harmonic. The cloverleaf [12] and centipede circuits [13] (see Fig. 9-6-2) belong to
this type. The second type is the first space-harmonic circuit, which has positive mutual coupling between the cavities. These circuits operate with the first spatial harmonic and are commonly used for pulse or continuous wave (CW) applications from
one to several hundred kilowatts of power output [14]. In addition, the long-slot circuit of the positive mutual coupling-cavity circuit operates at the fundamental spatial
400
Chap. 9
...
""
' ,..
. .t
..
'
.-. ,,,;"
..-; ...
harmonic with a higher frequency mode. This circuit is suitable for megawatt power
output. Figure 9-6-3 shows several space-harmonic coupled-cavity circuits.
962 Principles of Operation
Any repetitive series of lumped LC elements constitute a propagating filter-type circuit. The coupled cavities in the traveling-wave tube are usually highly over-coupled,
resulting in a bandpass-filter-type characteristic. When the slot angle (O) as shown in
Fig. 9-6-l(a) is larger than 180, the passband is close to its practical limits. The
drift tube is formed by the reentrant part of the cavity, just as in the case of a
klystron. During the interaction of the RF field and the electron beam in the
traveling-wave tube a phase change occurs between the cavities as a function of frequency. A decreasing phase characteristic is reached if the mutual inductance of the
coupling slot is positive, whereas an increasing phase characteristic is obtained if the
mutual inductive coupling of the slot is negative [ 12].
The amplification of the traveling-wave tube interaction requires that the electron beam interact with a component of the circuit field that has an increasing phase
characteristic with frequency. The circuit periodicity can give rise to field components that have phase characteristics [16] as shown in Fig. 9-6-4. In Fig. 9-6-4 the
angular frequency (w) is plotted as a function of the phase shift (/3 C) per cavity. The
ratio of w to /3 is equal to the phase velocity. For a circuit having positive mutual in-
Sec. 9.6
401
TWT OPERATION
BEAM CHARACTERISTIC
-4r
-3w
-2w -w
0
.8L(o)
2..
3..
4,,
TWT
OPERATION
IS WITH THIS
BRANCH OF
BEAM CHARACTERISTIC
I
I
I
-4 ..
-3 ..
-2 ..
-..
0
.8L(b)
ductive coupling between the cavities, the electron beam velocity is adjusted to be
approximately equal to the phase velocity of the first forward-wave spatial harmonic.
For the circuits with negative mutual inductive coupling, the fundamental branch
component of the circuit wave is suitable for synchronism with the electron beam
and is normally used by the traveling-wave tube. The coupled-cavity equivalent circuit has been developed by Curnow [17] as shown in Fig. 9-6-5.
/c::::,
to
I
'
402
Chap. 9
In Fig. 9-6-5 inductances are used to represent current flow and capacitors to
represent the electric fields of the cavities. The circuit can be evolved into a fairly
simple configuration. Loss in the cavities can be approximately calculated by adding
resistance in series with the circuit inductance.
600
400
200
100
80
60
PEAK POWER
40
;;;
I=
<
g
!5
!50
..
0:
...~
20
10
8
6
4
0:
AVERAGE
POWER
.8
.6
.4
V 25 kV
.2
5 & 1
10
20
FREQUENCY (GHz)
30 40
60
eo
100
Sec. 9.6
403
the voltage, and the product of the beam current and the voltage determines the total beam power. That is,
/beam
KVcJ 12
(9-6-1)
Pbeam
KV5 12
(9-6-2)
60
CAVITY TUBES
40
20
10
~
~
-'
8
6
g
~
::>
CZ:
w
~
...~CZ:
jlj
V 20kV
.a
.8
...
2
.1
1.0
6 7 8 910
FREQUENCY (GHzl
20
40
eo
eo
100
404
Chap. 9
either a matched load or an input or output line in order to reduce gain variations
with frequency. Cavity sections are cascaded to achieve higher tube gain than can be
tolerated in one section of cavities. Stable gain greater than 60 dB can be obtained
over about 30% bandwidth by this method.
The overall efficiency of coupled-cavity traveling-wave tubes is determined by
the amount of energy converted to RF energy and the energy dissipated by the collector. Interaction efficiencies from lO to 40% have been achieved from coupledcavity traveling-wave tubes. Overall efficiencies of 20 to 55% have been obtained
[7].
10 L-L---l_j__ __J__---l--l-'"Ll-_J_--1---l---1-----l---l--+--l---+-1----+--
.::>.-
::>
"'
~
l?
"~
1
0.8
0.8
0.5
0.4
0.3
0.2
0.05
0.1
0.2
Figure 9-7-1
0.4
0.6
1
2
FREQUENCY, GHz
10
20
30
406
Chap. 9
After electrons are emitted from the cathode, the electron beam has a tendency
to spread out because of the electron-repelling force. On the other hand, the electron
beam must be small enough for effective interaction with the slow-wave circuit. Usually the diameter of the electron beam is smaller than one-tenth wavelength of the
signal. Coupled-cavity traveling-wave high-power tubes utilize a shadow-grid technique to control the electron beam; so the device is called a gridded traveling-wave
tube (GTWT). As shown in Fig. 9-7-3 [18], the electron emitter of a gridded
traveling-wave tube has two control electrodes: one shadow grid near the cathode
and one control grid slightly away from the cathode. The shadow grid, which is at
cathode potential and interposed between the cathode and the control grid, suppresses electron emission from those portions of the cathode that would give rise to
interception at the control grid. The control grid, which is at a positive potential,
controls the electron beam. These grids can control far greater beam power than
would otherwise be possible.
Figure 9-7-3
TWT.
Sec. 9.7
407
rate electrode may have two advantages. First, the collector can be made as large as
desired in order to collect the electron beam at a lower density, thus minimizing localized heating. If the collector were part of the slow-wave circuit, its size would be
limited by the maximum gap capacitance consistent with good high-frequency performance. Second, using a separate collector can reduce its potential considerably
below the beam voltage in the RF interaction region, thereby reducing the power
dissipated in the collector and increasing the overall efficiency of the device. Gridded traveling-wave high-power tubes have a separate collector that dissipates the
electrons in the form of heat. A cooling mechanism absorbs the heat by thermal conduction to a cooler surface.
The efficiency of a gridded traveling-wave high-power tube is the ratio of the
RF power output to the product of cathode voltage (beam voltage) and cathode current (beam current). It may be expressed in terms of the product of the electronic
efficiency and the circuit efficiency. The electronic efficiency expresses the percentage of the de or pulsed input power that is converted into RF power on the slowwave structure. The circuit efficiency, on the other hand, determines the percentage
of de input power that is delivered to the load exterior to the tube. The electron
beam does not extract energy from any de power supply unless the electrons are actually collected by an electrode connected to that power supply. If a separate power
supply is connected between cathode and collector and if the cavity grids intercept a
negligible part of the electron beam, the power supply between the cathode and collector will be the only one supplying power to the tube. For a gridded traveling-wave
tube, the collector voltage is normally operated at about 40% of the cathode voltage.
Thus the overall efficiency of conversion of de to RF power is almost twice the electronic efficiency. Under this condition the tube is operating with collector voltage
depression.
9 72 Normal Depression and Overdepression
of Collector Voltage
Most gridded traveling-wave tubes are very sensitive to variations of collector depression voltages below normal depression level, since the tubes operate close to the
knee of the electron spent beam curves. Figure 9-7-4 [18] shows the spent beam
curves for a typical gridded traveling-wave tube.
Under normal collector depression voltage Ve at -7 .5 kV with full saturated
power output, the spent beam electrons are collected by the collector and returned to
the cathode. Thus the collector current le is about 2.09 A. A small amount of electrons intercepted by the beam scraper or slow-wave circuit contribute the tube body
current for about 0.178 A. Very few electrons with lower kinetic energy reverse the
direction of their velocity inside the collector and fall back onto the output pole
piece. These returning electrons yield a current l, of 0.041 A, which is only a small
fraction of the body current lb. These values are shown in Fig. 9-7-4.
When the collector voltage is overdepressed from the normal level of -7.5 kV
to the worst case of about -11.5 kV, a greater number of the spent electrons inside
the collector reverse the direction of their velocity by a highly negative collector
voltage and fall back onto the grounded output pole piece because the potential of
the pole piece is 11.5 kV higher than the collector voltage. It can be seen from Fig.
.,"'
1.2
.,
c.
"'
=c.,
..
1.0
..
. I
::>
/~X
l)
lo'
II
//
0.8
0.6
0.4
JII
~/I
I
--------------Current I, caused by returned electrons /
""7 8
o
2
3
4
5
6
0.2
A\
Body current lb
-r
O._.'""'~:a-ii-i;1""'511""5!!!!!Sl~-~-=:!!:-o=:!!:=i:::::...--L~....L~....L~.J-~.J..._~L-~L-__J
10
11
12
13
14
9-7 -4 that when the collector voltage is overdepressed from - 7. 5 to - 11. 5 kV, the
collector current is decreased sharply from 2.01 to 1.14 A and the body current is
increased rapidly from 0.237 to 1.110 A. The body current consists of two parts:
One part is the current caused by the electrons intercepted by the circuit or the beam
scrapers; another part is the current caused by the electrons returned by the overdepressed collector voltage. Figure 9-7-5 [18] shows the impact probability of returned electrons by certain overdepressed collector voltage.
Sec. 9.7
409
5.29
mm
Collector snout
Collector snout
Figure 9-7-5
voltage.
Example 9-7-1:
Determine:
410
Chap. 9
d. The heat in calories associated with the returning electrons (a factor for converting joules to calories is 0.238)
e. The temperature T in degrees Celsius for the output iron pole piece [Hint:
T = 0.238Vlt/(mass x specific heat)]
f. Whether the output iron pole piece is melted
Solution
0.85
8
_ = 5.31 x 10 1 electrons per second
x
10 19
b. The energy is
VI,
= 44.51 calories
e. The temperature is
0.238Vl,t
mass x specific heat
T=-------
44.51
250 x 10- 3 x 0. 108
l648.52C
When the spent electron beam arrives in the collector, the kinetic energies of each
electron are different. Under the normal operation at a collector voltage of about
40% of the cathode voltage, very few electrons will be returned by the negative collector voltage. Consequently, the tube body current is very small and negligible because the returned electrons are the only ones intercepted by the cavity grids and the
slow-wave circuit. When the collector is more negative, however, more electrons
with lower energy will reverse their direction of velocity and fall onto the output
Sec. 9.7
411
pole piece. Thus the tube body current will increase sharply. Since electrons of various energy classes exist inside the collector, two-stage collector voltage depression
may be utilized. Each stage is biased at a different voltage. Specifically, the main
collector may be biased at 40% depression of the cathode voltage for normal operation, but the collector snout may be grounded to the output pole piece for overdepression operation. As a result, the returned electrons will be collected by the collector snout and returned to the cathode even though the collector voltage is
overdepressed to be more negative. Since the collector is cooled by a cooling mechanism, the overheating problem for overdepression is eased. Figure 9-7-6 shows a
structure of two-stage collector voltage depression, and Fig. 9-7-7 depicts a basic
interconnection of a gridded traveling-wave tube with its power supplies [18].
Collector snout
Collector
~ Returned
E_l_ec_tr_o_n_b_e_am_h_ol_e_ __,-,.,-,..-~ ~ ~~ =-=:_~> electrons
Figure 9-7-6 Diagram for two-stage
collector depression.
Solenoid
supply
Cathode
power
supply
Modulator
Collector
Vk
Cathode -Vk
--
GTWT
Body current lb
Collector
power
supply
+
Collector current
412
Chap. 9
-:-
lb
-vk
Cathode
GTWT
Regulated
_collector+
power
supply VP
Figure 9-7-8
Collector
le
-~
~-----Rectifier
Transformer -
Filter
Voltage regulator
r-~-1
I
I
I
I
I
I
I
I
I
I
I
! vk
>
-IO
I
I
I
l __ , __ _JI
I
'--1'
~~-10-3vk
r--.L--,
I
+ I
I
I
I
I
I
I
I
I
! vk
I
I
L__,__ I
.J
-vk
Negative to cathode
....
~
Co)
Figure 9-7-9
-::-
Vk
+
Reference
414
Chap. 9
collector depression voltage is the difference of the two regulated supply voltages.
The cathode supply provides the tube body current and the collector supply yields
the collector current. The ratio of the collector current over the body current is
about 10 for an operation of 40% voltage depression. Thus the power delivered to
the tube by the collector supply is about four times larger than the power furnished
by the cathode supply. An electrical transient may occur in the circuit when power
supplies are just being switched on or off. The reasons for an electrical transient may
be caused by two factors:
1. Regulator in series with the collector power supply: In this method a voltage
regulator is incorporated in series with the collector supply as shown in Fig.
9-7 -10 so that the output voltage of the collector supply may be regulated at a
certain level with respect to ground. Since the output voltage of the cathode
supply is highly regulated at a certain level, the difference between the two
regulated voltages will produce a well-regulated voltage with respect to ground
at the collector electrode.
2. Regulator in parallel with the collector supply: In this method a voltage regulator is inserted in parallel with the collector supply as shown in Fig. 9-7-11 so
that a regulated voltage with respect to ground at normal depression may be
achieved at the collector terminal.
-I
Transformer
Rectifier
Voltage regulator-------1~
Filter
I
I
l VP
I
I
I
I
I
I
I
I
I
ill
Sl=
ill
I
I
I
.L---,
Differential
a"mplifier
~---- -10-3 VP
:
I
I
I
_J
L_T_
"'~
..i::
Q,
,:,
I
I
..i::
_ _L_1
"'"'
I
I
~VP I
I
I
-10-3 VP
Reference
I
I
I
I
I
I
I
L_....,.... _
_J
-VP
Negative to cathode
....
Cl'I
Figure 9-7-10
I-Transformer
...""'
Rectifier
Filter
Voltage regulator
0)
-Vi,
5
1--::--1
I
I
I
I
I
I
I
I
I
I
Q)
(.)
"'~
"'
'?e
..c
Collector
I
I
2VP
I
I
I
I
I
I
I
I
L __ , __ J
..C~llll
Q)
.....
r-- _i_ __ 1
I
I
I
I
I
I
I
.
I
1
I 2 VP
I
I
I
I
-10-J
Vo
Ref
-::-
L--~--J
I
-vk
Cathode
Figure 9-7-11
Chap. 9
417
References
3. Regulator between the cathode voltage and the collector voltage: In this
method the collector depression voltage is regulated with respect to the
cathode voltage as shown in Fig. 9-7-12. If the collector voltage is overdepressed above the normal depression value (absolute value), differential amplifier 2 tends to adjust the cathode voltage below its fixed level (absolute
value). When the cathode voltage is dropped, the collector voltage is readjusted to its normal depression level with respect to ground.
+
Reference
-10- 3 vk
~-
+ -,
I
I
I
Differential
amplifier 2
I
I
lI
Differential
amplifier I
Vk
_21
I
I
I
I
I
I
I
I
I
I
I
I
1- + -,
I
I
I
I
2VP
I
I
I
I
L-..,--...J
!0 3 R 1
I
I
I
L-,---l
r-...L-,
,---L---,
I
I
I
I
I
I
P'k
I
I
I
I
I
I
l _
I
I
I
I
I
I
: t VP
:
I
I
I
I
I
L-1-_J
L-,--...J
REFERENCES
[l] WARNECKE, R.R., et al., Velocity modulated tubes. In Advances in Electronics, Vol. 3.
Academic Press, New York, 1951.
[2] CHODOROW, M., and C. SussKIND, Fundamentals of Microwave Electronics. McGrawHill Book Company, New York, 1964.
[3] CHODOROW, M., and T. WESSEL-BERG, A high-efficiency klystron with distributed interaction. IRE Trans. Electron Devices, ED-8, 44-55, January 1961.
[4] LARUE, A. D., and R.R. RuBERf, Multi-Megawatt Hybrid TWT's at S-band and Chand. Presented to the IEEE Electron Devices Meeting, Washington, D.C., October
1964.
418
Chap. 9
[5] HIESLMAIR, H., et al., State of the art of solid-state and tube transmitters. Microwave J.,
26, No. 10, 46-48, October 1983.
[6) FEENBERG, E., Notes on velocity modulation. Sperry Gyroscope Laboratories Report.
5521-1043, Chapter I, 4I-44.
[7] STAPRANS, A., et al., High-power linear beam tubes. Proc. IEEE, 61, No. 3, 299-330.
March I973.
[8] LIEN, E. L., High efficiency klystron amplifier. In Conv. Rec. MOGA 70 (8th Int. Conf.,
Amsterdam, September I970).
[9] MERDINIAN, G., and J. V. LEBACQZ, High power, permanent magnet focused, S-band
klystron for linear accelerator use. Proc. 5th int. Conj. of Hyper-frequency Tubes (Paris,
September I 964).
[10] BECK, A. H. W., Space charge wave and slow electromagnetic waves. p. 106, Pergamon
Press, New York, 1985.
[I I] KoMPFNER, R., The traveling-wave tube as amplifier at microwaves. Proc. IRE, 35,
124- I 27, February I 947.
[I2] CHODOROW, M., and R. A. CRAIG, Some new circuits for high-power traveling-wave
tubes. Proc. IRE, 45, Il06-II18, August I957.
[13] RouMBANIS, T., et al., A megawatt X-band TWT amplifier with I8% bandwidth. Proc.
High-Power Microwave Tubes Symp., Vol. 1 (The Hexagon, Fort Monmouth, N.J.,
September 25-26, I962).
[I4] RuETZ, A. J., and W. H. YocoM, High-power traveling-wave tubes for radar systems.
IRE Trans. Mil. Electron., MIL-5, 39-45, April I96l.
[15] MENDEL, J. T., Helix and coupled-cavity traveling-wave tubes. Proc. IEEE, 61, No. 3,
280-298, March 1973.
[16] BRILLOUIN, L., Wave Propagation in Periodic Structures, 2nd ed. Dover, New York,
1953.
[17] CURNOW, H.J., A general equivalent circuit for coupled-cavity slow-wave structures.
IEEE Trans. on Microwave Theory and Tech., MTT-13, 671-675, September I965.
[ 18] LIAO, S. Y., The effect of collector voltage overdepression on tube performance of the
gridded traveling-wave tubes. Report for Hughes Aircraft Company, El Segundo, Calif.,
August 1977.
SUGGESTED READINGS
COLLIN, R. E., Foundations for Microwave Engineering, Chapter 9. McGraw-Hill Book Company, New York, 1966.
GANDHI, 0. P., Microwave Engineering and Applications, Chapters 9, 10, and I I. Pergamon
Press, New York, 1981.
GEWARTOWSKI, J. W., and H. A. WATSON, Principles of Electron Tubes, Chapters 5, 6, and 10
to I3. D. Van Nostrand Company, Princeton, N.J., I965.
GILMOUR, A. S., JR., Microwave Tubes. Artech House, Dedham, Mass., I986.
IEEE Proceedings, 61, No. 3, March I973. Special issue on high-power microwave tubes.
LIAO, S. Y., Microwave Electron Tubes. Prentice-Hall, Inc., Englewood Cliffs, N.J., I988.
Chap. 9
Problems
419
PROBLEMS
Vacuum Tubes
9-1. A vacuum pentode tube has five grids: a cathode, a control grid, a screen grid, a suppressor grid, and an anode plate as shown in Fig. P9- J.
a. Sketch the equivalent circuit.
b. Derive an expression for the input impedance Zin in terms of the angular frequency
w and the circuit parameters.
c. Determine the transit-angle effect.
Figure P9-l
Klystrons
Vo= 1200 V
lo= 28 mA
J = 8 GHz
d =I mm
l = 4cm
R,h = 40 kfl (excluding beam loading)
a. Find the input microwave voltage Vi in order to generate a maximum output voltage
V2 (including the finite transit-time effect through the cavities).
b. Determine the voltage gain (neglecting the beam loading in the output cavity).
420
Determine:
a. The input voltage (rms)
b. The output voltage (rms)
c. The power delivered to the load in watts
9-4. A two-cavity amplifier klystron has the following parameters:
Beam voltage:
Beam current:
Frequency:
Gap spacing in either cavity:
Spacing between centers of cavities:
Effective shunt impedance:
Vo=900V
lo= 30 mA
f = 8 GHz
d = 1 mm
l = 4cm
R,h = 40 kfl
Determine:
Vo= 20 kV
lo= 2 A
f = 9 GHz
Po = 10- 6 C/m 3
p = 10- 8 C/m 3
"V
= 105 m/s
Determine:
Vo= 30 kV
lo= 3 A
d = 1 cm
f
Vi
= 8 GHz
= 15 V(rms)
{3; = {30 = l
Po = 10- 7 C/m 3
Chap. 9
Chap. 9
Problems
421
Compute:
a. The de electron velocity
b. The de electron phase constant
c. The plasma frequency
d. The reduced plasma frequency for R = 0.4
e. The reduced plasma phase constant
f. The transit time across the input gap
g. The modulated electron velocity leaving the input gap
9-8. A two-cavity klystron amplifier has the following parameters:
Beam voltage:
Beam current:
Operating frequency:
Beam coupling coefficient:
de electron charge density:
Signal voltage:
Cavity shunt resistance:
Total shunt resistance including load:
Vo= 30 kV
lo= 3 A
f3; =
10 GHz
= 1
/30
1
Po= 10- C!m 3
Vi
15 V(rms)
lkfi
= 10 kfi
=
Rsh =
Rshr
Calculate:
a. The plasma frequency
b. The reduced plasma frequency for R = 0.4
c. The induced current in the output cavity
d. The induced voltage in the output cavity
e. The output power delivered to the load
f. The power gain
g. The electronic efficiency
9-9. A four-cavity klystron amplifier has the following parameters:
Beam voltage:
Beam current:
Operating frequency:
Beam coupling coefficient:
de electron charge density:
Signal voltage:
Cavity shunt resistance:
Total shunt resistance including load:
Determine:
a. The plasma frequency
b. The reduced plasma frequency for R = 0.4
c. The induced current in the output cavity
d. The induced voltage in the output cavity
e. The output power delivered to the load
f. The electronic efficiency
Vo=
lo=
f =
f3; =
Po =
Vi
1.5 A
2 GHz
/30
= 1
1o- 6 Clm 3
= 2 V(rms)
Rsh =
Rshr
20 kV
2 kfi
= 1 kfi
422
Chap. 9
Reflex Klystrons
Vo = 300 Y
lo= 20 mA
Vi = 40 v
Determine:
c. The effficiency
9-11. A reflex klystron operates under the following conditions:
Vo= 500 V
R,h
= 20 kfl
f,
= 8 GHz
The tube is oscillating atf, at the peak of then = 2 mode or I~ mode. Assume that the
transit time through the gap and the beam loading effect can be neglected.
a. Find the value of repeller voltage V, .
b. Find the direct current necessary to give microwave gap voltage of 200 V.
c. Calculate the electronic efficiency.
9-12. A reflex klystron operates at the peak of the n = 2 mode. The de power input is
40 mW and Vi/Vo = 0.278. If 20% of the power delivered by the beam is dissipated in
the cavity walls, find the power delivered to the load.
9-13. A reflex klystron operates at the peak of the n = I or~ mode. The de power input is
40 mW and the ratio of Vi over Vo is 0.278.
a. Determine the efficiency of the reflex klystron.
b. Find the total output power in milliwatts.
c. If 20% of the power delivered by the electron beam is dissipated in the cavity walls,
find the power delivered to the load.
Traveling-Wave Tubes (TWTs)
Vo= 2 kV
lo= 4 mA
f = 8 GHz
N = 50
Zo = 200
Chap. 9
423
Problems
Beam current:
Beam voltage:
Characteristic impedance of helix:
Circuit length:
Frequency:
9-16.
9-17.
9-18.
9-19.
9-20.
lo= 50 mA
Vo= 2.5 kV
Zo = 6.75 fl
N = 45
= 8 GHz
Determine:
a. The gain parameter C
b. The output power gain AP in decibels
c. All four propagation constants
d. The wave equations for all four modes in exponential form
An 0 -type traveling-wave tube operates at 2 GHz. The slow-wave structure has a pitch
angle of 5. 7. Determine the propagation constant of the traveling wave in the tube. It
is assumed that the tube is lossless.
An 0-type helix traveling-wave tube operates at 8 GHz. The slow-wave structure has a
pitch angle of 4.4 and an attenuation constant of 2 Np/m. Determine the propagation
constant 'Y of the traveling wave in the tube.
In an 0-type traveling-wave tube, the acceleration voltage (beam voltage) is 3000 V.
The characteristic impedance is 10 fl. The operating frequency is 10 GHz and the
beam current is 20 mA. Determine the propagation constants of the four modes of the
traveling waves.
Describe the structure of an 0-type traveling-wave tube and its characteristics; then explain how it works.
In an 0-type traveling-wave tube, the acceleration voltage is 4000 V and the magnitude
of the axial electric field is 4 V/m. The phase velocity on the slow-wave structure is
1.10 times the average electron beam velocity. The operating frequency is 2 GHz. Determine the magnitude of velocity fluctuation.
Gridded Traveling-Wave Tubes (GTWTs)
424
Chap. 9
RFPac
RFPac
Pdc
Volo
11=--=--
If the cathode voltage is -18 kV and the collector voltage is depressed to - 7 .5 kV determine the efficiency of the Gf WT.
Chapter 10
Microwave Crossed-Field
Tubes (M Type)
100 INTRODUCTION
In the previous chapter, several commonly used linear-beam tubes were described in
detail. In these tubes, the de magnetic field that is in parallel with the de electric
field is used merely to focus the electron beam. In crossed-field devices, however,
the de magnetic field and the de electric field are perpendicular to each other. In all
crossed-field tubes, the de magnetic field plays a direct role in the RF interaction
process.
Crossed-field tubes derive their name from the fact that the de electric field and
the de magnetic field are perpendicular to each other. They are also called M -type
tubes after the French TPOM (tubes apropagation des ondes a champs magnetique:
tubes for propagation of waves in a magnetic field). In a crossed-field tube, the electrons emitted by the cathode are accelerated by the electric field and gain velocity,
but the greater their velocity, the more their path is bent by the magnetic field. If an
RF field is applied to the anode circuit, those electrons entering the circuit during
the retarding field are decelerated and give up some of their energy to the RF field.
Consequently, their velocity is decreased, and these slower electrons will then travel
the de electric field far enough to regain essentially the same velocity as before. Because of the crossed-field interactions, only those electrons that have given up
sufficient energy to the RF field can travel all the way to the anode. This phenomenon would make the M-type devices relatively efficient. Those electrons entering the circuit during the accelerating field are accelerated by means of receiving
enough energy from the RF field and are returned back toward the cathode. This
back-bombardment of the cathode produces heat in the cathode and decreases the
operational efficiency.
426
Chap. 10
I
I
I
Nonresonant
Standing wave
Reentrant
Backward wave
r1-i
"'tA
Forward wave
Reentrant
Magnetron
Maser effect
Resonant
Nonreentrant
Reentrant
Nonreentrant
i"'mL.. I
Amplitron
Carcinotron
Gyrotrons
Sec. 10.1
Magnetron Oscillators
427
Hull invented the magnetron in 1921 [l], but it was only an interesting laboratory
device until about 1940. During World War II, an urgent need for high-power microwave generators for radar transmitters led to the rapid development of the magnetron to its present state.
All magnetrons consist of some form of anode and cathode operated in a de
magnetic field normal to a de electric field between the cathode and anode. Because
of the crossed field between the cathode and anode, the electrons emitted from the
cathode are influenced by the crossed field to move in curved paths. If the de magnetic field is strong enough, the electrons will not arrive in the anode but return instead to the cathode. Consequently, the anode current is cut off. Magnetrons can be
classified into three types:
1. Split-anode magnetron: This type of magnetron uses a static negative resistance between two anode segments.
2. Cyclotron-frequency magnetrons: This type operates under the influence of
synchronism between an alternating component of electric field and a periodic
oscillation of electrons in a direction parallel to the field.
3. Traveling-wave magnetrons: This type depends on the interaction of electrons
with a traveling electromagnetic field of linear velocity. They are customarily
referred to simply as magnetrons.
Negative-resistance magnetrons ordinarily operate at frequencies below the microwave region. Although cyclotron-frequency magnetrons operate at frequencies in
microwave range, their power output is very small (about 1 W at 3 GHz), and their
efficiency is very low (about 10% in the split-anode type and 1% in the single-anode
type). Thus, the first two types of magnetrons are not considered in this text. In this
section, only the traveling-wave magnetrons such as the cylindrical magnetron, linear (or planar) magnetron, coaxial magnetron, voltage-tunable magnetron, inverted
coaxial magnetron, and the frequency-agile magnetron will be discussed.
1011 Cylindrical Magnetron
428
Chap. 10
,-ANODE
END HAT/
Figure 10-1-1
Electron
path
dzr - r (d<P)z
dt 2
dt
= .!__Er m
dt
dt
dt
(10-1-1)
(10-1-2)
where !... = 1.759 x 10 11 C/kg is the charge-to-mass ratio of the electron and
m
Bo = B, is assumed in the positive z direction.
Rearrangement of Eq. (10-1-2) results in the following form
d ( 2 d<P) _ e
dr _ 1 d 2
dt r dt - mB,r dt - 2Wc dt (r)
where
We=
(10-1-3)
yields
r2
d</>
1
dt
= 2wcr + constant
2
(10-1-4)
Sec. 10.1
429
Magnetron Oscillators
= a, where a is the radius of the cathode cylinder, and ~ = 0, constant = - ~Wea 2 The angular velocity is expressed by
at r
dcfJ
dt
= !We
2
(1 - a2)
(10-1-5)
r2
Since the magnetic field does no work on the electrons, the kinetic energy of the
electron is given by
1
-mV 2 = eV
2
(10-1-6)
= 2e v
m
"V; + "V ~
= (dr)
dt
+ (r dc/J)
dt
(10-1- 7)
at r = b, where bis the radius from the center of the cathode to the edge of the anode, V = Vo, and dr /dt = 0, when the electrons just graze the anode, Eqs. (10-1-5)
and ( l 0-1-7) become
~~ = ~we(1 - ::)
b2 (d<P)2
dt
= 2e Vo
(10-1-8)
(10-1-9)
(10-1-10)
The electron will acquire a tangential as well as a radial velocity. Whether the electron will just graze the anode and return toward the cathode depends on the relative
magnitudes of Vo and Bo. The Hull cutoff magnetic equation is obtained from Eq.
(10-1-10) as
(10-1-11)
This means that if Bo > B0c for a given Vo , the electrons will not reach the anode.
Conversely, the cutoff voltage is given by
V0c
8~Bijb 2 (1 -::r
(10-1-12)
This means that if Vo < Voe for a given Bo, the electrons will not reach the anode.
Equation (10-1-12) is often called the Hull cutoff voltage equation.
430
Example 10-1-1:
Chap. 10
Conventional Magnetron
Vo = 26 kV
Io = 27 A
Bo = 0.336 Wb/m 2
a = 5 cm
b = 10 cm
Anode voltage:
Beam current:
Magnetic flux density:
Radius of cathode cylinder:
Radius of vane edge to center:
Compute:
a. The cyclotron angular frequency
b. The cutoff voltage for a fixed Bo
c. The cutoff magnetic flux density for a fixed Vo
Solution
a. The cyclotron angular frequency is
x 10 11 x 0.336
= 5.91
x 10 10 rad
1 ~S
= 139.50 kV
15~2) r1
= 14.495 mWb/m 2
= =-;;;
Sec. 10.1
431
Magnetron Oscillators
= 27T = 27Tm
w
(10-1-15)
eB
Since the slow-wave structure is closed on itself, or "reentrant," oscillations are possible only if the total phase shift around the structure is an integral multiple of 27T
radians. Thus, if there are N reentrant cavities in the anode structure, the phase shift
between two adjacent cavities can be expressed as
"' =
'f'n
27Tm
(10-1-16)
7T
(7T mode)
(10-1-17)
Figure 10-1-3 shows the lines of force in the 7T mode of an eight-cavity magnetron.
It is evident that in the 7T mode the excitation is largely in the cavities, having opposite phase in successive cavities. The successive rise and fall of adjacent anodecavity fields may be regarded as a traveling wave along the surface of the slow-wave
structure. For the energy to be transferred from the moving electrons to the traveling
field, the electrons must be decelerated by a retarding field when they pass through
each anode cavity. If L is the mean separation between cavities, the phase constant
of the fundamental-mode field is given by
a _ 27Tn
fJO -
NL
(10-1-18)
Tr
432
Chap. 10
Maxwell's equations subject to the boundary conditions. The solution for the fundamental </>component of the electric field has the form [l]
(10-1-19)
where E1 is a constant and /30 is given in Eq. (10-1-18). Thus, the traveling field of
the fundamental mode travels around the structure with angular velocity
d</>
dt
/30
(10-1-20)
When the cyclotron frequency of the electrons is equal to the angular frequency of the field, the interactions between the field and electron occurs and the
energy is transferred. That is,
We = /30
d</>
dt
(10-1-21)
Power output and efficiency. The efficiency and power output of a magnetron depend on the resonant structure and the de power supply. Figure 10-1-4
shows an equivalent circuit for a resonator of a magnetron.
Y,
G,
B'
A'
'---r----~~~~~~---.,~~~~~~"---y-J
Beam
Resonator
Load
Each resonator of the slow-wave structure is taken to comprise a separate resonant circuit as shown in Fig. 10-1-4. The unloaded quality factor of the resonator is
given by
Qun
= woC
Gr
(10-1-22)
where w 0 = 2TTfo is the angular resonant frequency. The external quality factor of
the load ciruit is
Sec. 10.1
Magnetron Oscillators
433
Qex
woC
Ge
(10-1-23)
woC
+Ge
(10-1-24)
= G,
+ G,
T/c =Ge
(10-1-25)
Ge
1
Gex
QexfQun
The maximum circuit efficiency is obtained when the magnetron is heavily loaded,
that is, for Ge ~ G,. Heavy loading, however, makes the tube quite sensitive to the
load, which is undesirable in some cases. Therefore, the ratio of Qc/ Qe, is often chosen as a compromise between the conflicting requirements for high circuit efficiency
and frequency stability.
The electronic efficiency is defined as
Pgen
T/ = - =
e
where
Pgen
Pde
Vo
lo
P1os1
Pde
Volo - P1ost
Volo
(10-1-26)
= anode voltage
= anode current
= power lost in the anode circuit
Volo -
Volo - lo 2e /3 2
P1ost
w5
E~.,
+ B'!
(10-1-27)
= !N IVl2woC
2
where N
V
Emax
M1
Qc
sin(/3n~) / (/3n~)
operation
/3 = phase constant
Bz = magnetic flux density
L = center-to-center spacing of the vane tips
434
Chap. 10
_ NL 2woC
2MrQe
Pgen
(10-1-28)
Emax
Pgen
T/e
Example 10-1-lA:
= Volo =
(10-1-29)
IomMrQe
+ BzeNL 2 woC
Pulsed Magnetron
Vo= 5.5 kV
lo= 4.5 A
f = 9 x 109 Hz
Anode voltage:
Beam current:
Operating frequency:
Resonator conductance:
Loaded conductance:
Vane capacitance:
Duty cycle:
Power loss:
2 x 10- 4 mho
2.5 x 10- 5 mho
2.5 pF
0.002
= 18.50 kW
G, =
Ge =
C =
DC=
P1oss
Compute:
a.
b.
c.
d.
e.
f.
Solution
w,
Qun
12
= 628
Sec. 10.1
435
Magnetron Oscillators
Q =
12
5655
1
5655/707
1 l.1 J%
_ %
25 25
State of the art. For many years, magnetrons have been the high-power
sources in operating frequencies as high as 70 GHz. Military radar relies on conventional traveling-wave magnetrons to generate high-peak-power RF pulses. No other
microwave devices can perform the same function with the same size, weight,
voltage, and efficiency-range advantage as can the conventional magnetrons. At the
present state of the art, a magnetron can deliver a peak power output of up to
40 MW with the de voltage in the order of 50 kV at the frequency of 10 GHz. The
average power outputs are up to 800 kW. Its efficiency is very high, ranging from 40
to 70%. Figure I0-1-5 shows the state of the art for U.S. high-power magnetrons.
20
10
.......
,.:
~::>
....
....
....
........
...
"'
......
.. ..~ w
""
...
<(
0.8
...
0.6
0.5
...
0.4
....
0.3
0.2
0.05
0.1
0.2
o.4
o.e
10
FREQUENCY. GHz
Figure 10-1-5
20
30
436
Chap. 10
Beacon magnetrons-miniature conventional magnetrons-deliver peak outputs as high as 3.5 kW, yet weigh less than two pounds. These devices are ideal for
use where a very compact, low-voltage source of pulsed power is required, such as
in airborne, missile, satellite, or Doppler systems. Most of the beacon magnetrons
exhibit negligible frequency shift and provide long-life performance under the most
severe environmental and temperature conditions.
The Litton L-5080 pulse magnetron, shown in Fig. 10-1-6, is a typical vanestrap magnetron oscillator. It has a maximum peak output power of 250 kW at a frequency range from 5.45 to 5.825 Ghz. Its duty cycle is 0.0012.
Figure 10-1-6 Photograph of Litton L-5080 magnetron. (Courtesy of Litton Electron Tube Division.)
dt
dt
(10-1-30)
B;c;nxn;J
+
Et -
Cathode
Sec. 10.1
437
Magnetron Oscillators
(10-1-31)
(10-1-32)
where.!!....
m
B,
Ex
(10-1-33)
Equation (10-1-33) shows that, regardless of space charges, the electron velocity
parallel to the electrode surface is proportional to the distance of the electron from
the cathode and to the magnetic flux density B,. How far the electron moves from
the cathode depends on B, and on the manner in which the potential V varies with x,
which in turn depends on the space-charge distribution, anode potential, and electrode spacing.
If the space charge is assumed to be negligible, the cathode potential zero, and
the anode potential Vo , the differential electric field becomes
dV =Vo
dx
where Vo
d
(10-1-34)
d x =!!_(Vo_ B dy)
dt 2
m d
'dt
(10-1-35)
d x
dt 2
(e )
2
m B,
e Vo
x - m
d =O
(10-1-36)
Solution of Eq. (10-1-36) and substitution of the solution into Eq. (10-1-33)
yield the following equations for the path of an electron with zero velocity at
cathode (origin point) as
X
Vo
= -d[l 8 zWc
COS
(wet)]
(10-1-37)
438
Chap. 10
(10-1-38)
z =0
(10-1-39)
2Vom = d
B;ed
(10-1-40)
d 2 B;
2m
K = - - = - = 1.14 x 10- 11
Vo
e
(10-1-41)
When the value of K is less than 1.14 x 10- 11 , electrons strike the anode; when the
value is larger than 1.14 x 10- 11 , they return to the cathode. Figure 10-1-8 shows
the electron path.
~,~'L
z
Figure 10-1-8
magnetron.
Cathode
From Eq. (10-1-40), the Hull cutoff voltage for a linear magnetron is given by
1e
V0c =--Bod
2m
(10-1-42)
where Bo = Bz is the magnetic flux density in the positive z direction. This means
that if Vo < V0c for a given Bo, the electrons will not reach the anode.
Similarly, the Hull cutoff magnetic flux density for a linear magnetron is expressed as
B0c =.!_
d -V2-;Vo
(10-1-43)
This means that if Bo > Boe for a given Vo, the electrons will not reach the anode.
Sec. 10.1
439
Magnetron Oscillators
Example 10-1-2:
Linear Magnetron
Vo = 10 kV
lo = 1 A
Bo = 0.01 Wb/m 2
d = 5 cm
Anode voltage:
Cathode current:
Magnetic flux density:
Distance between cathode and anode:
Compute:
a. The Hull cutoff voltage for a fixed Bo
b. The Hull cutoff mangetic flux density for a fixed Vo
Solution
a. The Hull cutoff voltage is
V0c = ~ x 1.759 x 10 11 x (0.01) 2 x (5 x 10- 2 ) 2
=
22.00 kV
B0c
1
(2 x 10 x 10 )
5 x 10- 2 x 1.759 x 10 11
1 2
= 6.74 mWb/m 2
Hartree condition. The Hull cutoff condition determines the anode voltage
or magnetic field necessary to obtain nonzero anode current as a function of the
magnetic field or anode voltage in the absence of an electromagnetic field. The
Hartree condition can be derived as follows and as shown in Fig. 10-1-9.
The electron beam lies within a region extending a distance h from the
cathode, where h is known as the hub thickness. The spacing between the cathode
and anode is d. The electron motion is assumed to be in the positive y direction with
a velocity
1 dV
lfv =
Va-=-
-TL
(10-1-44)
Bo dx
----------~~"---------
B0
01
Ex
Beam boundary
Cathode
-T
l_
~ Th
I
d
_i_
440
where Bo
V
= Bz is the magnetic
= potential
Chap. 10
tm "V;
(10-1-45)
eV
(dV)
dx
2eV B fi
m
(10-1-46)
(10-1-47)
Vv = dx
Integration of Eq. (10-1-47) yields the potential within the electron beam as
eB5 x2
2m
( 10-1-48)
V(h)
= 0 at x = 0.
_!__B6h 2
2m
The po-
(10-1-49)
and
dV
dx
Ex= - -
2
= --Boh
(10-1-50)
f
f
Vo= -
Ex dx
= -
Ex dx -
= V(h)
Ex dx
+ !...B"fih(d - h)
m
= !...s5h(d - h/2)
m
(10-1-51)
The electron velocity at the hub surface is obtained from Eqs. (10-1-44) and
(10-1-50) as
"Vv(h)
.
= -m Boh
( 10-1-52)
Sec. 10.1
Magnetron Oscillators
441
For synchronism, this electron velocity is equal to the phase velocity of the slowwave structure. That is,
w
/3
- = -Boh
(10-1-53)
m w2
wBod
Voh
= ~ - 2e 132
(10-1-54)
This is the Hartree anode voltage equation that is a function of the magnetic flux
density and the spacing beween the cathode and anode.
Example 10-l-2a:
Linear Magnetron
Vo = 15 kV
lo = 1. 2 A
f = 8 GHz
Bo = 0.015 Wblm 2
h = 2. 77 cm
d = 5 cm
Anode voltage:
Cathode current:
Operating frequency:
Magnetric flux density:
Hub thickness:
Distance between anode and cathode:
Calculate:
x 103 mis
/iw =
0. 73
108 mis
0.73
= 5.475
108
x 104
= 39.60 kV
0.015
1.515 x 104
10- 2
2 x 1.759 x 10 11
(O 73
108) 2
442
Chap. 10
External
cavity
Coupling slots
(a) Cross section
Magnetic flux
density
Cavity mode
attenuator
Coupling
slot
TE 011 mode
electric field
lines
Vane
resonator
RF output
Tuning
piston
Inner
circuit-mode
attenuator
(b) Cutaway view
Sec. 10.1
Magnetron Oscillators
443
1T-mode operation, the electric fields in every other cavity are in phase, and so they
couple in the same direction into the surrounding cavity. As a result, the surrounding coaxial cavity stabilizes the magnetron in the desired 1T-mode oepration.
In the desired TEo11 mode, the electric fields follow a circular path within the
cavity and reduce to zero at the walls of the cavity. Current flow in the TEo 11 mode
is in the walls of the cavity in circular paths about the axis of the tube. The undesired modes are damped out by the attenuator within the inner slotted cylinder near
the ends of the coupling slots. The tuning mechanism is simple and reliable. As the
straps are not required, the anode resonator for the coaxial magnetron can be larger
and less complex than for the conventional strapped magnetron. Thus cathode loading is lower, and voltage gradients are reduced.
The Varian SFD-333TM magnetron, as shown in Fig. 10-1-11 is a typical Xband coaxial magnetron. It has a minimum peak power of 400 kW at a frequency
range from 8.9 to 9.6 GHz. Its duty cycle is 0.0013. The nominal anode voltage is
32 kV, and the peak anode current is 32 A.
444
Chap. 10
Anode circuit
Conttol electrode
sole. Power output can be adjusted to some extent through the use of the control
electrode in the electron gun. At high-power levels and high frequencies, the bandwidth percentage is limited. However, at low-power levels and low frequencies, the
bandwidth may approach 70%.
10-1 5 Inverted Coaxial Magnetron
A magnetron can be built with the anode and cathode inverted-that is, with the
cathode surrounding the anode. For some time, the basic problem of mode suppression prevented its use. In an inverted coaxial magnetron, the cavity is located inside
a slotted cylinder, and a resonator vane array is arranged on the outside. The cathode
is built as a ring around the anode. Figure 10-1-13 shows the schematic diagram of
an inverted coaxial magnetron.
Anode vane
Vo
Cylindrical-cavity
resonator TE 011
445
Magnetron Oscillators
Sec. 10.1
(r
!!!_
rdt
where!!..._
m
Bo
d<fJ)
dt
(10-1-55)
= !_Bz d<fJ
m
(10-1-56)
dt
!!_
dt
where We
(r
d</J)
dt
dt
dt
(r
2)
(10-1-57)
= !!._Bz
is the cyclotron angular frequency. Integration of Eq. (10-1-57)
m
yields
(10-1-58)
at r
= b,
stant
= 0'
con-
(10-1-59)
Since the magnetic field does no work on the electrons, the kinetic energy of the
electron is given by
!m"V2 = eV
(10-1-60)
2e V -_
"lr 2
fr+
"lr 2 _
f <I> -
(dr)
-
dt
d</J)
r-
dt
(10-1-61)
at r = a, where a is the radius from the center of the cylinder to the edge of the anode, V = Vo, and dr/dt = 0.
When the electrons just graze the anode, Eqs. (10-1-60) and (10-1-61) become
(10-1-62)
(d</J)
a2 dt
-_ 2e V.o
m
(10-1-63)
446
Chap. 10
= 2e
m Vo
( 10-1-64)
The electron will acquire a tangential as well as a radial velocity. Whether the electron will just graze the anode and return back toward the cathode depends on the relative magnitudes of the anode voltage Vo and the magnetic flux density Bo. The cutoff condition can be obtained from Eq. ( 10-1-64) as
e
2 (
Voe= -B6a
1 - -b2)2
8m
a2
( 10-1-65)
This means that if Vo < V0c for a given Bo, the electrons will not reach the anode.
Equation (10-1-65) is often called the Hull cutoff voltage equation. Similarly, the
magnetic cutoff condition is expressed by
(10-1-66)
This means that if Bo > B0c for a given Vo , the electrons will not reach the anode.
Equation (10-1-66) is called the Hull cutoff magnetic equation.
The advantage of an inverted coaxial magnetron design is that the cathode current density can be reduced to one-tenth of that used in cathode-centered magnetrons. Thus the millimeter magnetron is a practical and long-life device. The output waveguide can be in the circular electric mode that has extremely low
transmission loss. Figure 10-1-14 compares the inverted coaxial magnetron with a
conventional magnetron designed for the same frequency. It should be noted that the
cathode sizes are quite different.
Anode
Cathode
Conventional rising
sun magnetron
Inverted coaxial
magnetron
Sec. 10.1
Magnetron Oscillators
Example 10-1-5:
447
Vo= IO kV
lo= 2 A
a= 3 cm
b = 4 cm
Bo = 0.01 Wb/m 2
Determine:
a. The cutoff voltage for a fixed Bo
b. The cutoff magnetic flux density for a fixed Vo
Solution
a. The cutoff voltage is
v'
l
x I.759 x 10 11 x (0.01) 2 x (3 x 10- 2)2
8
= -
= l.20 kV
'
-(8 x
10 x J03)1;2[ 3
l.759 x 10 11
= 0.0289 Wb/m 2
448
Chap. 10
.
Ag11e excurs10n
= -N
T
(10-1-67)
Sec. 10.1
Magnetron Oscillators
I _
One cycle of
r--
agile rate
449
_I
~
Figure 10-1-16
excursion.
where N
= number of pulses placed on the target during one radar scan, say, 20,
whichever is smaller
'T
=DC
(10-1-68)
where DC = duty cycle is the ratio of the pulse duration over the repetition period
for a pulse. The duty cycle is defined as
Out c cle =
Pulse duration
= ~ =
y y
Pulse repetition period
T
(I0-1-69)
= ~
(10-1-70)
where the 2 in the denominator is counted for the fact that two excursions through
the agile frequency range occur during each cycle of agile rate.
Example 10-1-6:
Frequency-Agile Magnetron
Determine:
a.
b.
c.
d.
e.
JLS
450
Chap. 10
Solution
a. The agile excursion is
Agile excursion =
14
_ x
_ = 70 MHz
0 2
10 6
fr
1
X 10-6
DC
0.001
10-6
= -:;: = 0.20
= 5 MHz
f = --:; = 0.20 x
= 5 kHz
14
= -f = -=
5000
2 8 ms
.
x _
0 0028
= 178.57 Hz
In the emitting-sole tube, the current emanated from the cathode is in response to
the electric field forces in the space between the cathode and anode. The amount of
Sec. 10.2
451
INPUT-OUTPUT SEPARATOR
AND DRIFT SPACE
RF OUTPUT
CIRCUIT
ENERGY
DIRECTION
(a)
RF
OUTPUT
ACCELERATING
ANODE
CATHODE
Rf ATTENUATOR
(b)
Figure 10-2-1
452
Chap. 10
+Strap
- Strap
Figure 10-2-2
current is a function of the dimension, the applied voltage, and the emission properties of the cathode. The perveance of the interaction geometry tends to be quite
high, about 5 to 10 x 10- 10 , which results in a high-current and high-power capability at relative low voltage. In the injected-beam tube the electron beam is produced
in a separate gun assembly and is injected into the interaction region.
The beam-circuit interaction features are similar in both the emitting-sole and
the injected-beam tubes. Favorably phased electrons continue toward the positively
polarized anode and are ultimately collected, whereas unfavorably phased electrons
are directed toward the negative polarized electrode.
In linear-beam interaction, as discussed for traveling-wave tubes in Sec. 9-5,
the electron stream is first accelerated by an electric gun to the full de velocity; the
de velocity is approximately equal to the axial phase velocity of the RF field on the
slow-wave structure. After interaction occurs, the spent electron beam leaves the interaction region with a low-average velocity. The difference in velocity is accounted
for by the RF energy created on the microwave circuit. In the CFA, the electron is
exposed to the de electric field force, magnetic field force, and the electric field
force of the RF field, and even to the space-charge force from other electrons. The
last force is normally not considered in analytic approaches because of its complexity. Under the influence of the three forces, the electrons travel in spiral trajectories
in a direction tending along equipotentials. The exact motion has been subject to
much analysis by means of a computer. Figure 10-2-3 shows the pattern of the electron flow in the CFA by computerized techniques [2]. It can be seen that when the
spoke is positively polarized or the RF field is in the positive half cycle, the electron
speeds up toward the anode; while the spoke is negatively polarized or the RF field is
Sec. 10.2
453
\
\
/">(
c 'synchronous
Figure 10-2-3
in the negative half cycle, the electrons are returned toward the cathode. Consequently, the electron beam moves in a spiral path in the interaction region.
The total power generated in a given CFA is independent of the RF input
power, as long as the input power exceeds the threshold value for spoke stability at
the input. The power generated can be increased only by increasing the anode
voltage and current. Neglecting circuit attenuation, the output power of the CFA is
equal to the sum of the input power and the power generated in the interaction region. That is, the power gain of a CFA is given by
(10-2-1)
where
Pout
P;n
Pgen
= P;n + Pgen
= RF input power
= RF power induced into the anode circuit by electrons
Therefore, the CFA is not a linear amplifier but rather is termed a saturated
amplifier.
The efficiency of a CFA is defined as the product of the electronic efficiency T/e
and the circuit efficiency 1/c . The electronic efficiency T/e is defined as in Eq.
10-1-29). The overall efficiency is then expressed as
1/ = T/cT/e =
(10-2-2)
454
Chap. 10
= !1._ = Pout
T/
- Pin
(10-2-3)
Pgen
T/e
P.
OU!
[>,.
me
-2af
P,
fe
f
-2a(f-</>)
d.4..
'f'
= A e-2ae + Pgen (I _
'"
(10-2-4)
e-2ae)
2af
2a f
Pgen
(10-2-5)
"V=~
B
(10-2-6)
E;;,.x
where
Emax
= 2{3 2Zc
(10-2-7)
Sec. 10.2
455
The power loss per spoke due to the electron motion toward the anode at any position is derived from Eq. (10-2-6) as
(10-2-8)
where Vs0 =de voltage per spoke
/,{) =de current per spoke
Since the power varies linearly with position, the average power loss over the
entire circuit length is
(10-2-9)
fa0
(w)
m
2e fJ
me /j2
f3 Zc
fa0
(g + 1)l
g _
Pgen
(10-2-10)
l+ B
Ia0
2
mf3
Zc
2e
(gg -+ 1)
(10-2-11)
Crossed-Field Amplifier
Va0 =
2 kV
fa0 =
1.5 A
456
Electronic efficiency:
RF input power:
Chap. 10
T/e = 20%
Pin= 80W
Calculate:
a. The induced RF power
b. The total RF output power
c. The power gain in decibels
Solution
a. The induced RF power is
"
?gen = 0.20
103
1.5 = 600 W
+ 600 = 680 W
Pin
80
20
10
8
......
,.:
50
..
"
...
IC
~
"
<
....
J,,!l
0.8
0.8
0.5
0.4
0.3
0.2
0.05
0.1
0.2
0.4
0.8
F REOUENCY. GHz
Figure 10-2-4
10
20
30
Sec. 10.3
457
The trade name of the backward-wave crossed-field amplifier (BWCFA) is Amplitron, and its schematic diagram is shown in Fig. 10-3-1.
Cathode
Magnetic flux
density 8 0
Anode
cavity
Contact
points
Input
(strap line)
Figure 10-3-1
Amplitron.
Schematic diagram of
The anode cavity and pins comprise the resonator circuits. A pair of pins and
the cavity are excited in opposite phase by the strap line. The electron beam and the
electromagnetic waves interact in the resonant circuits. The BWCFA can deliver
3-MW pulse with 10-,s duration at S-band, and the tube gain reaches 8 dB.
The highly successful QK434 Amplitron produced by Raytheon exhibited stable gain as high as 16 dB, power output levels ranging from a few hundred kilowatts
to 3 MW, and efficiencies ranging from 60% for normal power levels to 76% for
high-power, low-gain operation. The tube is commonly used in air surveillance radar
and military pulsed radar. Figure 10-3-2 pictures the QK434 Amplitron.
The two-stage superpower Amplitron, also manufactured by Raytheon, generated 425 kW of CW power at an efficiency of 76%. The gain was 9 dB and the band-
458
Chap. 10
width was 5% at a mean frequency of 3 GHz. The tube was used for all high-datarate transmission in the Apollo program. Figure 10-3-3 shows the two-stage
superpower Amplitron.
The circuit and electronic equations for the M-type amplifier and oscillator
were developed by several authors [4, 5]. The basic secular equation including the
space-charge effect is given by
(y2 - y[i)(j{3. - y) [(j{3. - y)2
+ {3~]
{3.
y)
+ jl
'Vo =
~a Bm]H
2
(10-3-1)
Sec. 10.3
w,
= .!!_Bo is
m
=
H2 =
Bo
vom
459
=A
a =
exp ( - jyy)
= }{3
(10-3-2)
'Y
= }{3 ( 1 + p)
(10-3-3)
(10-3-4)
The right-hand term is small, as it is not much different from C, so that if either one
of the left-hand terms is also small, Eq. (10-3-4) will be satisfied.
If {3, = {3 is assumed, the first factor is small and the solution for p becomes
P
If {3, /3m
by
a
1
+ ll'
)
2
1;2(
{3 )
/3m
1;2
(10-3-5)
= {3 is assumed, the second factor is small and the solution for p is given
( -{3 + 1)H2
p 2 -- +-1 (1 a)2
2
- 4 1
+ ll'
/3m -
(10-3-6)
From the definition of p, gain is only obtained when p is imaginary. From Eq.
(10-3-5), this only happens when a is positive. In Eq. (10-3-6), pis imaginary when
{3, + f3m = {3 and a is less than unity. This last condition comes from rewriting Eq.
(10-3-6) as
(13)
1 1- a
p=l(l+a)1f2/3m
1;2
(10-3- 7)
460
Chap. 10
= - i/3 (1 +
I'
= /3m
+ ..!_
-J 2 (1
is
+a
(10-3-8)
p)
a2)1;2
(/3'/3m)
i;2
(10-3-9)
This gives rise to increasing backward waves for all values of a except a = -1. Because H rxc 312 , Eqs. (10-3-5), (10-3-7), and (10-3-9) show that the gain per unit
length in M -type devices is lower than in 0 -type devices using similar circuits because C 312 < c.
Example 10-3-1:
Amplitron Characteristics
Anode voltage:
Anode current:
Magnetic flux density:
Operating frequency:
Characteristic impedance:
Determine:
a. The de electron-beam velocity
Solution
a. The de electron-beam velocity is
'Vo= 0.593
106
(15
108 mis
e
-B0
m
1.759
10 11
0.2
= 35.18
109 rad/s
Sec. 10.4
461
'Vo
f3m
109
35 18
X
= 484.57
0.726 X 103
rad/m
= (/oZo)'/3 = (
4V0
3 X 50
x 15 x 103
)'/3 = 0.136
RF signal
output
Accelerator
--. I L_ __ _s~o~-:::v_:~!_r~:_t~~---__j_i_
B8
Electron beam
E
- - - - Sole
-..1'. ......---I
Collector
+
vco
462
Circuit
V~B~-
Chap. 10
Sole
Figure 10-4-2
the circuit, electrons at position C are closer to the circuit, and electrons at position
D are closer to the sole. However, electrons at position C have departed a greater
distance from the unperturbed path than have electrons at position D. Thus, the elec-
trons have lost a net amount of potential energy, this energy having been transferred
to the RF field. The reason for the greater displacement of the electrons moving toward the circuit is that these electrons are in stronger RF fields, since they are closer
to the circuit. Electrons at position G have moved so far from the unperturbed position that some of them are being intercepted on the circuit. The length from position
A through position G is a half cycle of the electron motion.
Circular M-Carcinotron. The M-Carcinotrons are generally constructed in
the circular reentrant form as shown in Fig. 10-4-3. The slow-wave structure and
sole are circular and nearly reentrant to conserve magnet weight. The sole has the
appearance of the cathode in a magnetron.
The Litton L-3721 M-BWO, shown in Fig. 10-4-4, is a typical M-type
backward-wave oscillator (M -BWO or Carcinotron) with a minimum power of 200
Wat frequency range from 1.0 to 1.4 GHz.
Collector
RF attenuator
Grid Accelerating
\....-'.'.~:ZZZmzZZZZ:;i;:::-..._,
anode
RF output
cb~l-JiH--
Slow-wave
structure
Sole
- - RFwave
463
Sec. 10.4
"t'"'JI'
7~
L-3721
Figure 10-4-4 Photograph of Litton L3721 BWO. (Courtesy of Litton Company, Electron Tube Division.)
Wide
In the circular configurations, the delay line is terminated at the collector end
by spraying attenuating material on the surfaces of the conductors. The output is
taken from the gun end of the delay line which is an interdigital line. Clearly, in this
case, the electron drift velocity has to be in synchronism with a backward-space harmonic.
As in the case of 0-type devices, the only modification in the secular equation
is a change of sign in the circuit equation. If this change is made in Eq. (I0-3-1), we
write
(10-4-1)
'Yo = j{3
'}' = jk +
(10-4-2)
(/3 2
k2
= jf3{3,k 2
j2({3, - k)e]
((3, - k + l ~ a 2/3m) H2
(10-4-3)
A solution of Eq. (10-4-3) for synchronism can be obtained by setting {3 = (3, and
(3, - k = (3,b ', where b' is a small number so that terms like b ' 2 and b 'E may be
neglected. This yields
2E
..
j{3,b -
(b'
2a
{3{3,k {3; + l + 0'.2
= (3,k
1 2a
Ji. H2
+ a 2 /3m
= 2(3,kD 2
1)
/3m
(10-4-4)
464
Chap. 10
where
D
E
b'
.,
5 (5 - jb)
f3
= 1 + a2 f3m H
(10-4-5)
= /3eD5
= bD
= -1 or 5 2
(10-4-6)
(10-4-7)
-
JM
+ 1=0
(10-4-8)
= j ({3, + b) + f3,D51
= j(/3, + b) + f3,D52
(10-4-9)
(10-4-10)
where the l>'s are the roots of Eq. (10-4-8) and they are
,;,
= j
Ul
,;,
u2
.b-Vb 2 +4
2
.b+Vb 2 +4
2
To determine the amplification of the growing waves, the input reference point is set
at y = 0, and the output reference point is taken at y = e. It follows that at y = 0,
the voltage at the input point can be computed as follows:
V1(0) + V2(0)
Vi(O)
<>1
= V (O)
+ Vz(O) = O
<>2
(10-4-11)
(10-4-12)
V1(0)
V(O)
1 - 52/<>1
V2(0)
1 - l>1/l>2
-V (O)
51 V(O)
<>2
-()zV (O)
51
(10-4-13)
(10-4-14)
+ V2(0) exp(-y2C)
(10-4-15)
Sec. 10.4
465
or
(10-4-17)
0,/02 = b - w+4
b + v'b 2 + 4
(10-4-18)
and
(10-4-19)
Then
4)
(10-4-20)
o,
2{3,Dt'
= (2n +
1)1T
(10-4-21)
= 2n + 1
4
Vo= 20 kV
Io= 3.5 A
Bo = 0.3 Wb/m 2
f = 4 GHz
Zo = 50 0
D = 0.8
b = 0.5
Compute:
a.
b.
c.
d.
e.
(10-4-22)
466
Chap. 10
Solution
a. The de electron velocity is
"Vo = 0.593
.o.5 - Y(o.5) 2 + 4
= }
" _ .o.5 +
u2-j
= - J0.78
v'(o.s) + 4 -_ J. 1. 28
2
'}'2
j 301.00
N = 1.5625
and
e -_ /3:
2rrN _
-
for n = 1
21T x 1.5625 _
300
- 3.27 cm
REFERENCES
[l] HUTTER, R. G. E., Beam and Wave Electrons in Microwave Tubes. D. Van Nostrand
Company, Princeton, N.J., 1960.
[2] SKOWRON, J. F., The Continuous-Cathode (Emitting-Sole) Crossed-Field Amplifier.
Proc. IEEE, 61, No. 3, March 1973, 330-56.
[3] BROWN, W. C., The microwave magnetron and its derivatives. IEEE Trans. on Electron
Devices, ED-31, No. 11, November 1984, 1595-1605.
[4] PIERCE, J. R., Traveling-Wave Tubes, P. 210. D. Van Nostrand Company, Princeton,
N.J., 1950.
[5] BECK, A.H. W., Space-Charge Waves and Slow Electromagnetic Waves, P. 250. Pergamon Press, New York, 1958.
[6] GEWARTOWSKI, J. V., and H. A. WATSON, Principles of Electron Tubes, P. 391. D. Van
Nostrand Company, Princeton, N. J. , 1965.
Chap. 10
467
Problems
PROBLEMS
Magnetrons
10-1. Describe the principle of operation for a normal cylindrical magnetron and its characteristics.
10-2. A normal cylindrical magnetron has the following parameters:
Inner radius:
Outer radius:
Magnetic flux density:
10-3.
10-4.
10-5.
10-6.
10-7.
Ra = 0.15 meter
Rb = 0.45 meter
Bo = 1.2 milliwebers/m 2
B =
Bylly
Determine the trajectory of an electron with an initial velocity 'Vo in the z direction.
Vo
= E2Uz =-du,
f
Figure PI0-7
468
Vo=
lo=
Bo =
a=
b =
Anode voltage:
Anode current:
Magnetic flux density:
Radius of cathode cylinder:
Radius of vane edge to center:
32 kV
84 A
0.01 Wb/m 2
6 cm
12 cm
Compl,lte:
Vo= 22 kV
lo= 28 A
10 GHz
G, = 3 x 10- 4 If
Ge = 3 X 10- 5 If
C = 3 pF
DC= 0.001
Ploss
= 200 kW
Compute:
The angular resonant frequency
The unloaded quality factor Qun
The loaded quality factor Qe
The external quality factor Q.,
The circuit efficiency
The electronic efficiency
10-10. A linear magnetron has the following parameters:
a.
b.
c.
d.
e.
f.
Anode voltage:
Anode current:
Magnetic flux density:
Distance between cathode and anode:
Vo = 20 kV
lo = 17 A
8 0 = 0.01 Wb/m 2
d = 6 cm
Calculate:
Vo= 32 kV
lo= 60 A
10 GHz
Chap. 10
Chap. 10
469
Problems
Bo = 0.01 Wb/m 2
h = 3 cm
d = 6 cm
Compute:
Anode voltage:
Cathode current:
Anode radius:
Cathode radius:
Magnetic flux density:
Vo= 30 kV
lo= 25 A
a= 2.5 cm
b = 5 cm
Bo = 0.01 Wb/m 2
Determine:
a. The cutoff voltage for a fixed Bo
b. The cutoff magnetic flux density for a fixed Vo
10-13. A frequency-agile coaxial magnetron has the following parameters:
Pulse duration:
Duty cycle:
Pulse rate on target:
Determine:
a. The agile excursion
b. The pulse-to-pulse frequency separation
c. The signal frequency
d. The time for N pulse
e. The agile rate
Crossed-Field Amplifiers (CFAs)
10-14.
10-15.
10-16.
10-17.
Va0 = 3 kV
la0 = 3 A
1/e = 25%
Pin= 100 W
470
Vao = 1.80 kV
la0 = 1.30 A
T/e = 22%
Pin= 70 W
Calculate:
f = 9 GHz
Vo= 20 kV
lo= 3.5 A
Bo = 0.3 Wb/m 2
Zo = 50 !1
Compute:
Vo= 25 kV
lo= 4 A
Bo = 0.35 Wb/m 2
Zo = 50 !1
Calculate:
a. The de electron-beam velocity
b. The electron-beam phase constant
c. The cyclotron angular frequency
d. The cyclotron phase constant
e. The gain parameter
Carcinotrons
= 10 GHz
Chap. 10
Chap. 10
471
Problems
Operating frequency:
Anode voltage:
Anode current:
Magnetic flux density:
Characteristic impedance:
D factor:
b factor:
= 8 GHz
Vo= 25 kV
lo= 4 A
Bo = 0.35 Wb/m 2
Zo = 50 ohms
D = 0.75
b = 0.50
Calculate:
a. The de electron velocity
b. The electron-beam phase constant
c. The delta differentials
d. The propagation constants
e. The oscillation condition
Chapter 11
Strip Lines
11 0 INTRODUCTION
Prior to 1965 nearly all microwave equipment utilized coaxial, waveguide, or parallel strip-line circuits. In recent years-with the introduction of monolithic microwave integrated circuits (MMICs)-microstrip lines and coplanar strip lines have
been used extensively, because they provide one free and accessible surface on which
solid-state devices can be placed. In this chapter parallel, coplanar, and shielded
strip lines and microstrip lines, which are shown in Fig. 11-0-1 [1], are described.
lL,
0
Ground plane
(a) Schematic diagram
Figure 11-0-1
472
Sec. 11.1
Microstrip Lines
473
T
c,
(a)
Figure 11-1-1
line.
"
(b)
474
Strip Lines
Chap. 11
line is a function of the strip-line width, the strip-line thickness, the distance between the line and the ground plane, and the homogeneous dielectric constant of the
board material. Several different methods for determining the characteristic
impedance of a microstrip line have been developed. The field-equation method was
employed by several authors for calculating an accurate value of the characteristic
impedance [3 to 5). However, it requires the use of a large digital computer and is
extremely complicated. Another method is to derive the characteristic-impedance
equation of a microstrip line from a well-known equation and make some changes
[2]. This method is called a comparative, or an indirect, method. The well-known
equation of the characteristic impedance of a wire-over-ground transmission line, as
shown in Fig. 11-1-1 (b), is given by
Zo
where
60
4h
= ~ / Ind
V Er
for h
(11-1-1)
Er
h
d
= the height from the center of the wire to the ground plane
= diameter of the wire
If the effective or equivalent values of the relative dielectric constant Er of the ambient medium and the diameter d of the wire can be determined for the microstrip
line, the characteristic impedance of the microstrip line can be calculated.
Effective dielectric constant Ere. For a homogeneous dielectric medium,
the propagation-delay time per unit length is
Td
= v"
= ~ = 3.333
ns/m
(11-1-2)
E
1.016
ns/ft
(11-1-3)
where
0
= 41T
Eo
In transmission lines used for interconnections, the relative permeability is 1. Consequently, the propagation-delay time for a line in a nonmagnetic medium is
Td
= 1.106~
ns/ft
(11-1-4)
The effective relative dielectric constant for a microstrip line can be related to the
relative dielectric constant of the board material. DiGiacomo and his coworkers discovered an empirical equation for the effective relative dielectric constant of a microstrip line by measuring the propagation-delay time and the relative dielectric constant of several board materials, such as fiberglass-epoxy and nylon phenolic [6].
Sec. 11.1
475
Microstrip Lines
= 0.475Er + 0.67
(11-1-5)
Ere
is the effec-
ere= 0.475e,
_,,,.
+ 0.67
,-v
3
~v
= 0.67w (o.8 + ;)
(11-1-6)
Zo
87
ln [ 5.98h ]
Yer+ 1.41
0.8w + t
(11-1-7)
476
Strip Lines
1.3
I. 2
I. I
~-
1.0
.r:
:9
0.9
0.8
-,- -
--,---- __j_ -1
---+--
_l--f- -+
: -+
0. 7
-'=l
"
-:l
2
:::l
-:l
0.5
'-'
0
.2
"'
0.4
,,.
t'
--
+ --
+--
I- -- _J _
0. I
,- '
0.2
f--,----+-
L _,
0.3
0:::
'
-1=1-=1~~
~+-- --+-
'_;
c:
d=o.67(08+
0.6
t--~---~-----+-
"'
E
t-
'---'---'~-'-~-'-----'-~-'-~'---'-~-'----'
0. I 0.2
0.3
0.4 0.5
0.6 0.7
0 8 0.9
1.0
where
Chap. 11
Equation ( 11-1-7) is the equation of characteristic impedance for a narrow microstrip line. The velocity of propagation is
c
v=
3 x 108
yz.= VZ,
(11-1-8)
m/s
The characteristic impedance for a wide microstrip line was derived by Assadourian
and others [8] and is expressed by
h
Zo-- w
f;-
377
h
---E
~W
for (w
h)
(11-1-9)
Sec. 11.1
Microstrip Lines
477
Since the dielectric constant of the materials used does not vary excessively
with frequency, the dielectric constant of a microstrip line can be considered independent of frequency. The validity of Eq. ( 11-1-7) is doubtful for values of dielectric thickness h that are greater than 80% of the line width w. Typical values for the
characteristic impedance of a microstrip line vary from 50 n to 150 n, if the values
of the parameters vary from Er = 5.23, t = 2.8 mils, w = 10 mils, and h = 8 mils
to Er = 2.9, t = 2.8 mils, w = 10 mils, and h = 67 mils [2].
Example 11-1-1:
Er =
Zo =
87
Ye,+ 1.41
In [ 5.98h ]
0.8w + t
_
87
In [ 5.98 x 7 ]
o.8 x 10 + 2.8
- V5.23 + 1.41
= 45.78 n
,.
L.
0
t_
478
Strip Lines
Chap. 11
=-
dP/dz
2 P (z)
= ad +
ac
(11-1-11)
where ad is the dielectric attenuation constant and ac is the ohmic attenuation constant.
The gradient of power in the z direction in Eq. (11-1-11) can be further expressed in terms of the power loss per unit length dissipated by the resistance and
the power loss per unit length in the dielectric. That is,
_ dP(z)
dz
=-~(!VI*)
dz
=!(RI)!*
~II/ 2 R
~o-V*V
+ !/V/ 2 u
=Pc+ pd
(11-1-12)
Np/cm
(11-1-13)
Pc
ac=2P(z)
Np/cm
(11-1-14)
and
Sec. 11.1
Microstrip Lines
479
ad=~~
Np/cm
(I 1-1-15)
where (J' is the conductivity of the dielectric substrate board in U/cm. This dielectric
constant can be expressed in terms of dielectric loss tangent as shown in Eq.
(2-5-17):
(J'
(I 1-1-16)
tan()= -
WE
(11-1-17)
Since the microstrip line is a nonmagnetic mixed dielectric system, the upper dielectric above the microstrip ribbon is air, in which no loss occurs. Welch and Pratt [9]
derived an expression for the attenuation constant of a dielectric substrate. Later on,
Pucel and his coworkers [IO] modified Welch's equation [9]. The result is
ad=
q(J'
4.34--
\j~
\/Ze
Eo
v'z.:
dB/cm
(11-1-18)
In Eq. ( 11-1-18) the conversion factor of 1 Np = 8. 686 dB is used, Ere is the effective dielectric constant of the substrate, as expressed in Eq. (I 1-1-5), and q denotes
the dielectric filling factor, defined by Wheeler [3] as
q
Ere -
Er -
(11-1-19)
= 27. 3
(qEr) tanAg ()
dB/ Ag
Ere
where A8 =
A8 =
(11-1-20)
Ere
If the loss tangent, tan (), is independent of frequency, the dielectric attenuation per wavelength is also independent of frequency. Moreover, if the substrate
conductivity is independent of frequency, as for a semiconductor, the dielectric attenuation per unit is also independent of frequency. Since q is a function of Er and
w/h, the filling factors for the loss tangent qEn/Ere and for the conductivity q/\/Ze
are also functions of these quantities. Figure 11-1-5 shows the loss-tangent filling
factor against w/h for a range of dielectric constants suitable for microwave inte-
480
Strip Lines
e,
Chap. 11
I 05 (rutile)
0.98
0.96
qe,
,;;
0.94
0.92
0.90
0.88
0
0.2 0.4
0.6 0.8
1.0 1.2
w/h
grated circuits. For most practical purposes, this factor is considered to be 1. Figure
11-1-6 illustrates the product adp against w / h for two semiconducting substrates, silicon and gallium arsenide, that are used for integrated microwave circuits. For design purposes, the conductivity filling factor, which exhibits only a mild dependence
on w/h, can be ignored.
Ohmic losses. In a microstrip line over a low-loss dielectric substrate, the
predominant sources of losses at microwave frequencies are the nonperfect conductors. The current density in the conductors of a microstrip line is concentrated in a
sheet that is approximately a skin depth thick inside the conductor surface and exposed to the electric field. Both the strip conductor thickness and the ground plane
thickness are assumed to be at least three or four skin depths thick. The current density in the strip conductor and the ground conductor is not uniform in the transverse
plane. The microstrip conductor contributes the major part of the ohmic loss. A diagram of the current density J for a microstrip line is shown in Fig. 11-1-7.
Because of mathematical complexity, exact expressions for the current density
of a microstrip line with nonzero thickness have never been derived [10]. Several researchers [8] have assumed, for simplicity, that the current distribution is uniform
and equal to I /w in both conductors and confined to the region Ix I < w/2. With
this assumption, the conducting attenuation constant of a wide microstrip line is
given by
ac =
8.686Rs
Ziw
dB/cm
for - > 1
h
(11-1-21)
Sec. 11.1
Microstrip Lines
481
390
380
~
370
C:
x
360
E
u
i:Q
::;::,
350
Cl.
""
340
<:j
3 201,0------4-- - - - + - - - - - - + - - - - - - + - - 0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
w/h
Bottom of strip
Top of strip
where Rs
Rs
o = Y-;;;f,_;;
' ; is the skin depth in cm
For a narrow microstrip line with w/ h < 1, however, Eq. (11-1-21) is not applicable. The reason is that the current distribution in the conductor is not uniform,
as assumed. Pucel and his coworkers [10, 11] derived the following three formulas
from the results of Wheeler's work [3]:
482
acZoh
Rs
Strip Lines
8.68
2'1T
[l _(W ')2] [l
4h
t)]
47TW
+ -h + -h- ( In--+
-
W1
'TTW 1
for - :::; h
27T
2
acZoh = 8.68 [l _
Rs
2'1T
Chap. 11
(~) ] [l + ~ + ~
4h
(in 2h _
(
(11-1-22)
!..)
J
h
1
for 2'1T
< - :::; 2
h
(I 1-1-23)
and
_ae_Zo_h
Rs
8. 68
{w'
h + ;2 In [ 2'1Te (w'
lh + 0.94)]}
x [ I + -h + -h- ( In -2h - -t)]
W
'TTW
h
1
[-w' + _w--''/-'-('TT_h'--)]
w'lh + 0.94
w
for 2 :::; h
(I 1-1-24)
w'
=
=
aw
=;
2.718
w
+ ilw
(in
(I 1-1-25)
~w + 1)
2
ilw = ; (in th
1)
2t
w
'TT
for- < - :::; h
h
2
w
for h
'TT
-
(I 1-1-26)
(11-1-27)
The values of ae obtained from solving Eqs. (11-1-22) through (I 1-1-24) are plotted
in Fig. 11-1-8. For purposes of comparison, values of ae based on Assadourian and
Rimai's Eq. (11-1-21) are also shown.
Radiation losses. In addition to the conductor and dielectric losses, microstrip line also has radiation losses. The radiation loss depends on the substrate's
thickness and dielectric constant, as well as its geometry. Lewin [12] has calculated
the radiation loss for several discontinuities using the following approximations:
1. TEM transmission
2. Uniform dielectric in the neighborhood of the strip, equal in magnitude to an
effective value
3. Neglect of radiation from the transverse electric (TE) field component parallel
to the strip
4. Substrate thickness much less than the free-space wavelength
Sec. 11.1
Microstrip Lines
483
w/11
Lewin' s results show that the ratio of radiated power to total dissipated power for an
open-circuited microstrip line is
Pract
P,
(11-1-28)
Zo
Ere
+ 1_
Ere
Ere 2Ere
1 In
y'Z, +
VZ. VZ. -
1
1
(11-1-29)
in which Ere is the effective dielectric constant and Ao = c/ f is the free-space wavelength.
The radiation factor decreases with increasing substrate dielectric constant. So,
alternatively, Eq. ( 11-1-28) can be expressed as
Prad
P,
Rr
Zo
(11-1-30)
Strip Lines
484
Chap. 11
Rr
240ir 2
(:Jz
(11-1-31)
F (Ere)
The ratio of the radiation resistance Rr to the real part of the characteristic
impedance Zo of the microstrip line is equal to a small fraction of the power radiated
from a single open-circuit discontinuity. In view of Eq. (11-1-28), the radiation loss
decreases when the characteristic impedance increases. For lower dielectric-constant
substrates, radiation is significant at higher impedance levels. For higher dielectricconstant substrates, radiation becomes significant until very low impedance levels
are reached.
dB/cm
and that the characteristic impedance of a wide microstrip line, as shown in Eq.
(11-1-9), is
Zo
= !!_
W
~ = ?1;- !!_
v~
fl
VEr W
.\ = _lQ_
!yr;,
(11-1-32)
cm
39.5
(;J
(11-1-34)
fGttz
Rs=#=
27T
fl/square.
(11-1-35)
485
Sec. 11.2
(11-1-36)
= 4780h~
(11-1-37)
For 25-mil alumina at 10 GHz, the maximum Q, achievable from wide microstrip
lines is 954 [ 13].
Similarly, a quality factor Qd is related to the dielectric attenuation constant:
(11-1-38)
where ad is in dB/ AR.
Substituting Eq. (11-1-20) into Eq. (11-1-38) yields
Qd _
Ao
= __
- ~ tan (}
tan (}
(11-1-39)
where Ao is the free-space wavelength in cm. Note that the Qd for the dielectric attenuation constant of a microstrip line is approximately the reciprocal of the dielectric loss tangent (} and is relatively constant with frequency.
A parallel strip line consists of two perfectly parallel strips separated by a perfect
dielectric slab of uniform thickness, as shown in Fig. 11-2-1. The plate width is w,
the separation distance is d, and the relative dielectric constant of the slab is Erd.
A.
Figure 11-2-1 Schematic diagram of a
parallel strip line.
Strip Lines
486
Chap. 11
= -cd
H/m
(11-2-1)
where /.Le is the permeability of the conductor. The capacitance between the two
conducting strips can be expressed as
C
EJW
(11-2-2)
F/m
= 2Rs =
w
~w -v~
~
fi/m
(11-2-3)
adw
U/m
(11-2-4)
11 22 Characteristic Impedance
The characteristic impedance of a lossless parallel strip line is
Z0
~
= 377
-Ve w -V~ w
f"i:.. =
<!:__
for w
<!:__
~d
(11-2-5)
vp=~=Vlc=~=
m/s
487
Sec. 11.2
Zo =
~C = w
377 !!_
for w
(11-2-7)
11 23 Attenuation Losses
The propagation constant of a parallel strip line at microwave frequencies can be expressed by
for R
<{
wL
and
<{
wC
(11-2-8)
~ (R .ff+ G ~)
Np/m
(11-2-9)
and
(3
= wvLc
(11-2-10)
rad/m
Substitution of the distributed parameters of a parallel strip line into Eq. ( 11-2-9)
yields the attenuation constants for the conductor and dielectric losses:
ac
= !R
=! ~
2 \ti d \j~
0.
Np/m
(l l-2-11)
and
Np/m
Example 11-2-1:
(11-2-12)
A lossless parallel strip line has a conducting strip width w. The substrate dielectric
separating the two conducting strips has a relative dielectric constant E,d of 6 (beryllia
or beryllium oxide BeO) and a thickness d of 4 mm.
Calculate:
a. The required width w of the conducting strip in order to have a characteristic
impedance of 50 n
b. The strip-line capacitance
c. The strip-line inductance
d. The phase velocity of the wave in the parallel strip line
Solution
488
Strip Lines
377 d
w =
v'Z,; Zo
377 4
=
\/6
Chap. 11
x 10- 3
50
12.31 x 10- 3 m
b. The strip-line capacitance is
edw
c=d =
= 163.50 pF /m
c
3 x 108
v =--=
v'Z,;
\/6
A coplanar trip line consists of two conducting strips on one substrate surface with
one strip grounded, as shown in Fig. 11-3-1. The coplanar strip line has advantages
over the conventional parallel strip line (see Section 11-2) because its two strips are
on the same substrate surface for convenient connections. In microwave integrated
circuits (MICs) the wire bonds have always presented reliability and reproducibility
problems. The coplanar strip lines eliminate the difficulties involved in connecting
the shunt elements between the hot and ground strips. As a result, reliability is increased, reproducibility is enhanced, and production cost is decreased.
The characteristic impedance of a coplanar strip line is
Z _ 2 Pavg
o----n;-
(11-3-1)
where Io is the total peak current in one strip and Pavg is the average power flowing in
the positive z direction. The average flowing power can be expressed as
Pavg
= 1Re
JJ (E
H*) llz dx dy
(11-3-2)
Sec. 11.4
489
Example 11-3-1:
A coplanar strip line carries an average power of 250 mW and a peak current of
100 mA. Determine the characteristic impedance of the coplanar strip line.
From Eq. (11-3-1), the characteristic impedance of the coplanar strip line is
Solution
2 x 250 x 10- 3
Zo = (100 x 10-3)2 = 50 fl
A partially shielded strip line has its strip conductor embedded in a dielectric
medium, and its top and bottom ground planes have no connection, as shown in Fig.
11-4-1.
Figure 11-4-1
line.
Z _ 94.15
0
where K
= 1_
(w K + 8.854,
Ct )-i
v;, d
(I 1-4-1)
1
t/ d
ci tance in pF / m
490
Strip Lines
Chap. 11
200
70
180
60
160
50
:
No
140
40
120
30
100
20
80 L--~~L-~J..._~~~_.:>o...L..3o.......i::L..:...__.u.....J......J-'-~~--'-~-'-~-l---'-~-' IO
0.1
0.2
0.3
0.4 0.5 0.6
0.8 1.0
2.0
3.0
4.0
w
d
Example 11-4-1:
Calculate:
a. The K factor
b. The fringe capacitance
c. The characteristic impedance of the line.
2.56
w = 25 mils
t = 14 mils
d = 70 mils
E, =
Chap. 11
References
491
Solution
a. Using Eq. (11-4-1), the K factor is obtained:
14)-I -- 1.25
K -- ( l - -t)-I -- (1-- -
10
C1 =
=
8.854 x 2.56
.
[2 x 1.25 In (1.25
3 1416
+ 1) - (1.25 - 1) In (l.25 2
1)]
15.61 pF/m.
Zo
94.15 [25 (
)
15.61
]-I
25
70 t.
+ 8.854 x 2.56
\12.56
50.29
REFERENCES
[l] LIAO, S. Y., Engineering Applications of Electromagnetic Theory, Chapter 3. West Publishing Co., St. Paul, Minn, 1988.
[2] KAUPP, H. R., Characteristics of microstrip transmission lines. IEEE Trans. on Electronic Computers, EC-16, No. 2, 185-193, April 1967.
[3] WHEELER, H. A., Transmission-line properties of parallel strips separated by a dielectric
sheet, IEEE Trans. on Microwave Theory and Techniques, MTT-3, No. 3, 172-185,
March 1965.
[4] BRYANT, T. G., and J. A. WEISS, Parameters of microstrip transmission lines and of coupled pairs of microstrip lines, IEEE Trans. on Microwave Theory and Techniques,
MTT-6, No. 12, 1021-1027, December 1968.
[5] STINEHELFER, H. E., An accurate calculation of uniform microstrip transmission lines.
IEEE Trans. on Microwave Theory and Techniques, MTT-16, No. 7, 439-443, July
1968.
[6] DIGIACOMO, J. J., et al., "Design and Fabrication of Nanosecond Digital Equipment,"
RCA, March 1965.
[7] SPRINGFIELD, A., Simplified Theory of Microwave Transmission Systems, F. Assodourian
and E. Rimol. pp. 1651-1657, IRE Proceeding, December 1952.
[8] AssooouRIAN, F., and E. RIMOL, Simplified theory of mirowave transmission systems.
Proc. IRE, 40, 1651-1657, December 1952.
[9] WELCH, J. D., and H.J. PRATT, Losses in microstrip transmission systems for integrated
microwave circuits, NEREM Rec., 8, 100-101, (1966).
[10] PuCEL, R. A., D. J. MASSE, and C. P. HARTWIG, Losses in microstrip. IEEE Trans. on
Microwave Theory and Techniques, MTT-16, No. 6, 342-350, June 1968.
492
Strip Lines
Chap. 11
PROBLEMS
Microstrip Lines
Er
Zo
v
Zo~
L=-=--
and
l~
C=-=-
Zov
Zoe
Chap. 11
Problems
493
Erd=2.l
w = 26 mm
d = 5 mm
ac = 4.l x 101 U/m
= lO GHz
Determine the
a. Surface resistance of the gold strip
b. Characteristic impedance of the strip line
c. Phase velocity
11-7. A gold parallel strip line has the following parameters:
Relative dielectric constant of polyethylene:
Strip width:
Separation distance:
Erd
= 2.25
w=25mm
d = 5mm
Calculate the
a. Characteristic impedance of the strip line
b. Strip-line capacitance
c. Strip-line inductance
d. Phase velocity
Coplanar Strip Lines
Erd
= 10
w = 4mm
t = l mm
Strip Lines
494
TEM-mode field intensities:
Ey
Chap. 11
H, =
Find the
a. Average power flow
b. Peak current in one strip
11-9. A shielded stripline has the following parameters:
Relative dielectric constant of the
insulator polyethylene:
Strip width:
Strip thickness:
Shield depth:
= 2.25
w = 2mm
t = 0.5 mm
d = 4mm
Erd
Calculate the
a. K factor
b. Fringe capacitance
c. Characteristic impedance
11-10. A shielded strip line is made of a gold strip in a polystyrene dielectric insulator and
has the following parameters:
Relative dielectric constant of
polystyrene:
Strip width:
Strip thickness:
Shield depth:
Determine the
a. K factor
b. Fringe capacitance
c. Characteristic impedance
Erd
= 2.56
0.7 mm
t=I.4mm
d = 3.5 mm
w =
Chapter 12
Monolithic Microwave
Integrated Circuits
120 INTRODUCTION
Integrated circuits are a combination of active and passive elements that are manufactured by successive diffusion or ion implantation processes on a semiconductor
substrate. The active elements are generally silicon planar chips. The passive elements are either thin or thick film components. In thin films, a thin film of conducting (resistor) or nonconducting (capacitor) material is deposited on a passive insulated substrate, such as ceramic, glass, or silicon dioxide, by vacuum deposition.
Thick film refers to films more than several thousand angstroms (A) thick. Such
films are used almost exclusively to form resistors, and the pattern is usually defined
by silk-screening.
The integrated-circuit (IC) complexity has advanced from small-scale integration (SSI) for up to 100 components per chip, to medium-scale integration (MSI) for
up to 1000 components per chip, to large-scale integration (LSI) for up to 105 components per chip, and finally the very large-scale integration (VLSI) for more than
1 million components per chip. Recently, the integrated circuit has advanced to the
ultralarge-scale integration (ULSI) stage. For example, a 32-bit microprocessor chip
contains more than 0.15 x 106 components, and a 1-megabit dynamic randomaccess memory (DRAM) chip contains more than 2.2 x 106 components. In this
chapter, we discuss the integrated circuit devices.
Electronic circuits can be classified into three categories according to circuit
technology as shown in Fig. 12-0-1.
1. Discrete circuit (DC): The conventional electrical or electronic circuit, in
which the elements are separately manufactured and then connected together
495
496
Chap. 12
. - - - - + - - - - - - + - - - a Vs (SV)
130rl
4krl
(I) Discrete circuit where elements
Inputs
o Output
Bo
Co
Collector Base
Multiple emitters
Silicon
dioxide
P substrate
TOP
METAL
(a)
Figure 12-0-1 Discrete circuit, integrated circuit and microwave integrated circuit. (From
M. Caulton et al. [ 1]; reprinted by permission of IEEE, inc.)
Sec. 12.1
Materials
497
MMICs are suitable for space and military applications because they meet the requirements for shock, temperature conditions, and severe vibration. A major factor
in the success of MMICs has been the advances in the development of microwave
solid-state devices as described previously. In this chapter the basic materials and
processes necessary for fabrication of MMICs are described. Three general types of
circuits can be utilized for hybrid MMICs: distributed microstrip lines, lumpedelement (inductors and capacitors) circuits, and thin-film circuits. These three types
are discussed in Section 4.
121 MATERIALS
The basic materials for monolithic microwave integrated circuits, in general, are
subdivided into four categories:
498
Chap. 12
sapphire
2. Conductor materials-aluminum, copper, gold, and silver
3. Dielectric films-AbOJ, SiO, Si02, SiJN4, and Ta20s
4. Resistive films-Cr, Cr-SiO, NiCr, Ta, and Ti
1211 Substrate Materials
Material
PROPERTIES OF SUBSTRATES
tan (} x IO''
at IO GHz
Relative
dielectric
constant
(E,)
Thermal
conductivity
K(W/cm 0 C)
Alumina
IO
Bery Ilia
Ferrite/garnet
l
2
6
13-16
2.5
0.03
16
13
O.Q3
4
4
l
5
100
9.3-ll.7
0.01
0.02
0.4
GaAs
Glass
Rutile
Sapphire
0.3
Applications
Microstrip, suspended
substrate
Compound substrated
Microstrip, coplanar,
compound substrate
High frequency, microstrip,
monolithic MIC
Lumped element
Micros trip
Microstrip, lumped element
The ideal conductor materials for monolithic microwave integrated circuits should
have the following properties [2]:
Sec. 12.1
1.
2.
3.
4.
5.
499
Materials
High conductivity
Low temperature coefficient of resistance
Good adhesion to the substrate
Good etchability and solderability
Easily deposited or electroplated
Table 12-1-2 shows the properties of some widely used conductor materials for
microcircuits [4]. These materials not only have excellent conductivity, but they can
also be deposited by a number of methods and are capable of being photoetched.
They are used to form both the conductor pattern and the bottom ground plane. The
conductor thickness should be equal to at least four skin depths, to include 98% of
the current density. It can be seen from Table 12-1-2 that good electrical conductors
have poor substrate adhesion, whereas poor electrical conductors have good substrate
adhesion. Aluminum has relatively good conductivity and good adhesion. It is possible to obtain good adhesion with high-conductivity materials by using a very thin
film of one of the poorer conductors between the substrate and the good conductor.
Some typical combinations are Cr-Au, Cr-Cu, and Ta-Au. A typical adhesion layer
may have a surface resistivity ranging from 500 to 1000 fl/square without loss. The
choice of conductors is llSually determined by compatibility with other materials required in the circuit and the processes required. For small losses, the conductors
should be of the order of three to five skin depths in thickness. That is, thick films of
TABLE 1212
PROPERTIES OF CONDUCTORS
Coefficient
of thermal
expansion
(a,/C x 106 )
Adherence
to
dielectric
film or
substrate
Material
Skin depth
8 at GHZ
(m)
Surface
resistivity
(fl/sq x 10- 7 v'])
Ag
1.4
2.5
21
Poor
Cu
1.5
2.6
18
Very poor
Au
1.7
3.0
15
Very poor
Very good
Good
Al
1.9
2.6
3.3
4.7
26
4.6
Mo
2.7
4.7
6.0
Good
Cr
Ta
2.7
4.0
4.7
7.2
9.0
6.6
Good
Very good
Method of
deposition
Evaporation,
screening
Evaporation,
plating
Evaporation,
plating
Evaporation
sputtering,
vapor
phase,
electronbeam
evaporation
Electron-beam
evaporation,
sputtering
Evaporation
Electron-beam
sputtering
500
Chap. 12
the good conductor (about 10 m thick) are required. Films of this thickness can be
achieved by evaporation or plating or by any of the standard thick-film processes.
1213 Dielectric Materials
Dielectric materials are used in monolithic microwave integrated circuits for blockers, capacitors, and some couple-line structures. The properties of dielectric materials should be
1. Reproducibility
2. Capability of withstanding high voltages
3. Ability to undergo processes without developing pin holes
4. Low RF dielectric loss
Some of the dielectrics used in microcircuits are shown in Table 12-1-3; SiO, Si02,
and Ta20s are the most commonly used. Thin-film Si02 with high-dielectric Q can
be obtained by growing the pyrolitic deposition of Si02 from silane and then densifying it by heat treatment. Si02 can also be deposited by sputtering. With proper
processing, Si02 capacitors with Qs in excess of 100 have been fabricated with good
success. Capacitors fabricated with Si02 films have capacitances in the range of 0.02
to 0.05 pf/square mil. Thin-film SiO is not very stable and can be used only in noncritical applications, such as bypass capacitors. In power microwave integrated circuits, capacitors may require breakdown voltages in excess of 200 volts. Such capacitors can be achieved with films on the order of 0.5 to 1.0 m thick with low
probability of pin holes or shorts.
TABLE 1213
Relative
dielectric
constant
Material
Method of deposition
(,)
Dielectric
strength
(V/cm)
SiO
Si02
Si 3 N
Evaporation
Deposition
Vapor phase,
sputtering
Anodization,
evaporation
Anodization,
sputtering
6-8
4
7.6
6.5
7-10
4 x 105
101
101
10 7
4 x 106
30
100-1000
6 x 106
100
AliO)
Ta20s
22-25
Microwave
Q
Sec. 12.2
501
1. Good stability
2. Low temperature coefficient of resistance (TCR)
3. Adequate dissipation capability
4. Sheet resistivities in the range of IO to 1000 per square
Table 12-1-4 lists some of the thin-film resistive materials used in monolithic
integrated circuits. Evaporated nichrome and tantalum nitride are the most widely
used materials. The exact temperature coefficient of resistance achieved depends on
film formation conditions. Thick-film resistors may be utilized in circuits incorporating chip components. The thickness of the thick film is in the range of 1 to 500 m.
The term thick film refers to the process used, not to the film thickness. Thick-film
techniques involve silk-screening through a mask, such as the printing and screening
of silver or gold in a glass frit, which is applied on the ceramic and fired at 850 C.
Microwave thick-film metals are sometimes several micrometers thick, thicker than
those of low-frequency integrated circuits.
TABLE 1214 PROPERTIES OF RESISTIVE MATERIALS
Material
Cr
NiCr
Ta
Cr-SiO
Ti
Method of
deposition
Resistivity
(ll/square)
(%/OC)
TCR
Stability
Evaporation
Evaporation
Sputtering
Evaporation
or cermet
Evaporation
10-1000
40-400
5-100
-600
-0.100-+0.10
+0.002-+0.10
-0.010-+0.01
-0.005--0.02
Poor
Good
Excellent
Fair
5-2000
-0.100-+0.10
Fair
502
Chap. 12
Monolithic technology is not well suited to microwave integrated circuits because the processing difficulties, low yields, and poor performance have seriously
limited their applications. To date, the hybrid form of technology is used almost exclusively for microwave integrated circuits in the frequency range of 1 to 15 GHz.
Hybrid MMICs are fabricated on a high-quality ceramic, glass, or ferrite substrate.
The passive circuit elements are deposited on the substrate, and active devices are
mounted on the substrate and connected to the passive circuit. The active devices
may be utilized in chip form, in chip carriers, or in small plastic packages. The resistivity of microwave integrated circuits should be much greater than 1000 0-cm
for good circuit performance.
1. Thermal oxides
2. Dielectric layers
3. Polycrystalline silicon
4. Metal films
Epitaxial growth. In epitaxy technology, single-crystal semiconductor layers grow on a single-crystal semiconductor substrate. The word epitaxy comes from
the Greek epi (on) and taxis (arrangement). The epitaxial process offers an important means of controlling the doping profiles so that device and circuit performances
can be optimized. There are three types of epitaxy.
Sec. 12.2
503
1. Vapor-phase epitaxy (VPE) is the most important technique for silicon and
GaAs devices.
2. Molecular-beam epitaxy (MBE) is a process involving the reaction of one or
more thermal beams of atoms or molecules with a crystalline surface under
ultrahigh vacuum conditions ( ~ 10- 1 torr). MBE can achieve precise control in
both chemical composition and doping profiles. Single-crystal multilayer structures with dimensions of the order of atomic layers can be made by the MBE
method.
3. Liquid-phase epitaxy (LPE) is the growth of epitaxial layers on crystalline substrates by direct precipitation from the liquid phase. This process is particularly
useful for growing GaAs and related IIl-V compounds. LPE is suited to grow
thin epitaxial layers (2:0.2 m) because it has a slow growth rate. It is also
useful to grow multilayered structures in which precise doping and composition
controls are required.
Lithography. Lithography is the process of transferring patterns of geometric shapes on a mask to a thin layer of radiation-sensitive material, which is known
as resist, for covering the surface of a semiconductor wafer. The resist patterns
defined by the lithographic process are not permanent elements of the final device
but only replicas of circuit features. There are four types of lithography technology:
1.
2.
3.
4.
Electron-beam lithography
Ion-beam lithography
Optical lithography
X-ray lithography
Etching and photoresist. In the processes of making MICs, a selective removal of Si02 is required in order to form openings through which impurities can be
diffused. The photoetching method used for this removal is shown in Fig. 12-2-1.
Ultraviolet radiation
Substrate
(silicon chip)
(a) Masking and exposure to
ultraviolet radiation
Polymerized
Substrate
(silicon chip)
(b) Photoresist
Figure 12-2-1
Photoetching process.
504
Chap. 12
process usually involves the printing and silk-screening of silver or gold through a
metal mask in a glass frit, which is applied on the ceramic and fired at 850 C. After
firing, the initial layer may be covered with gold.
Deposition. Three methods-vacuum evaporation, electron-beam evaporation, and de sputtering-are commonly used for making MMICs.
For example, the photoresist technique can be used to remove the oxide layer in
related areas. As shown in Fig. 12-2-2, the fabrication procedures include the following:
Sec. 12.3
MOSFET Fabrication
505
(a) Deposition
(b) Mask
Oxide
Window
(d) Etching
most important device for very large-scale integrated circuits (VLSICs) such as microprocessors and semiconductor memories. Its basic fabrication processes can be
described in three areas: MOSFEf formation, NMOS growth, and CMOS development.
1231 NIOSFET Formation
MOSFEfs can be fabricated by using the following steps as shown in Fig. 12-3-1.
1. Oxidation: Select the p-type substrate and form a Si02 layer on the surface.
2. Diffusion: Open two windows by using the photoresist technique and diffuse
an n+ -layer through the windows.
3. Etching: Remove the center oxide region by the photoetching technique.
4. Oxidation: Again expose the entire surface to dry oxygen so that the Si02
covers the top surface.
5. Deposition: Deposit phosphorous glass over the surface to cover the oxide
layer.
506
Chap. 12
Etch
Si0 2
p-Type Si substrate
/J
(a) Oxidation
Etch
/!
/!
(d)
(e) Formation of
phosphorous glass
Oxidation
Al
Al
,,
(g) Metalhlation
Al
Etch
/!
6. Etching: Open two windows above the two n+ -type diffused regions by using
the photoetching method.
7. Metallization: Now see that aluminum metallization is carried out over the entire surface of the device.
8. Etching: Finally, etch away the unwanted metal and attach the metal contacts
to the diffused gate, drain, and source regions.
It can be seen that there is only one diffusion process in the fabrication of a
MOSFEf compared to the three required for the bipolar junction transistor. Therefore, MOSFEf fabrication is more efficient and less expensive than the BJT. These
attributes make MOSFEf integrated circuits attractive.
1232 NMOS Growth
Then-channel MOS (NMOS) logic gate was discussed in Section 6-5-1, and its fabrication processes are described as shown in Fig. 12-3-2.
507
MOSFET Fabrication
Sec. 12.3
Isolation mask
r----,
l_ ______r----L_:--1
f
A
Resist
'lo
,-~--,
Boron implant
,---- -----,
Boron implant
FOX
Nitride
oxide
.t
ri._____.I ~1
r - - - - .__\__,r,
p-Substrate
p-Chan stop
p-Substrate
Gate oxide
(b)
(a)
r---rr::
fl.____. ~1
A
A
Resist
/
Arsenic
implant
-~-------
-,
1A
t t
p-Substrate
p-Substrate
(c)
(d)
1A
FOX
p-Substrate
(e)
508
Chap. 12
Sec. 12.3
MOSFET Fabrication
509
t/
x x
x
n-Tub
xxxxxxxxxxxxxx
(a) Epitaxy and diposition
v-Epitaxy
Si0 2
(b) Implantation
(c)
Oxidation
11-Tub
p-Tub
(d) Deposition
(e) Implantation
v
P-Glass
(f)
Deposition and
metallization
Figure 12-3-3 CMOS fabrication. (After L. C. Parrillo [7); reprinted by permission of the Bell Laboratories.)
510
Chap. 12
1.
2.
3.
4.
SRAM cell with transistor load. Figure 12-3-4 shows the structure for a
single static RAM cell. Figure 12-3-4(b) shows a layout for the SRAM cell. The
width-to-channel-length ratio of the depletion-mode load transistor (1/5) is adjusted
to provide enough current drive to meet the speed requirements of the cell without
causing excessive steady-state (quiescent) power consumption. To minimize the cell
area, buried contacts (diffusion to polysilicon contact) are required. Figure 12-34(a) shows a six-transistor (n-channel) cell, which uses a cross-coupled inverter pair
(flip-flop, T1 to T4) to store 1 bit of information. The numbers next to the transistors
indicate the relative width-to-channel-length ratios. A pair of access transistors (T4
and Ts) transmits data into and out of the cell when the word and bit lines are simultaneously activated. The loads for the flip-flop are depletion-mode transistors (T1 and
Tz) with their sources and gates tied together as a NANO circuit. The data (logic 1
or 0) are retained in the cell by the positive feedback existing in the flip-flop circuit.
When the gate of T4 is at a high potential, its drain voltage is low (~VT). This
voltage, in return, is fed to the gate of T3, and keeps T3 off. The drain of T3 is then
tied to the high potential of T1 (which is always on), so is the gate of T4. This state of
the cell defines a logic 1, which is retained unless new data are entered.
SRAM cell with resistor load. When the depletion-mode load transistors
are replaced by high-valued resistors, the structure becomes a resistor MOS (RMOS)
as shown in Fig. 12-3-5.
Sec. 12.3
MOSFET Fabrication
511
Bit line
(Al)
(Al)
r--"----,
;;:;;;;;:;;;;;;;;;t;~~~T:!~~}
--+
Word line
(polysilicon)
Isolation
~ Polysilicon
Direct diffusion
to polysilicon
contact
Diffusion-polysilicon
~contact
i-1
Al-Si contact
L _ _. window
(b)
(a)
Figure 12-3-4 Structure of a single SRAM cell. (After R. W. Hunt [8]; reprinted by permission of John Wiley & Sons.)
Word line
Vee
R1
I~
Vcc polysilicon
(first level)
Ion-implanted
poly silicon
resistor
(second level)
Rz
"'
~
T3
T4
1T2
} Si0
iS
Polysilicon
gate
-:-
(a)
p-Substrate
(b)
Figure 12-3-5 SRAM cell with resistor load. (After L. C. Parrillo [7] and T. Ohzone et al.
[9]; reprinted by permission of the Bell Lab.)
512
Chap. 12
High-valued resistors reduce the power consumption, and they can be made in
a relatively small space by using polysilicon which has been ion-implanted to
provide the proper resistance. The polysilicon resistors (R1 and Rz) can be made in
the same single layer of polysilicon (gate and interconnect) by masking the polysilicon resistor regions from the high-impurity doping used in the gate and interconnect
portions of the polysilicon level. Using this type of fabrication process, SRAM-cell
areas can be reduced to half the cell area required in conventional transistor load
cells. Figure 12-3-5(a) shows the circuit diagram of a SRAM cell with polysilicon
resistor load (R1 and Rz), and (b) the device' cross section. The first-level polysilicon
is used for gate and routing power supply Vee. The second-level polysilicon is used
for resistor load directly over an active transistor. The connection to the drain and
Vee is made directly from an implanted polysilicon resistor.
DRAM cell with storage capacitor. Since a SRAM device consists of a
large number of static RAM cells, it can dissipate a great deal of electrical power.
For these reasons large memory chips use dynamic RAM cells that require only one
transistor and one storage capacitor per bit of information. Figure 12-3-6 shows the
diagrams of a basic SRAM cell with storage capacitor.
Polysilicon
cell plate
Diffused
column line
Row line
- [ Access trlistor
Storage
capacitor
''*'I-+-- Capacitor
~~~~ru--i
A
Column line
Access transistor
(polysilicon)
(b)
(a)
Storage
gate
Transfer
gate
Poly silicon
Source
(c)
(d)
Figure 12-3-6 Diagrams of DRAM cell with storage capacitor. (After L. C. Parrillo [7] and
R. W. Hunt [8]; reprinted by permission of the Bell Laboratories.)
Figure 12-3-6(a) shows the DRAM cell layout. A diffusion process forms the
bit line (source/drain) and also the source of the access transistor. The capacitance of
the diffused bit line (or the junction capacitance) and its resistance can be limiting
factors for the DRAM performance. In order to minimize these parasitic effects, it is
necessary to use MoSiz for word lines and A 1 for bit lines in fabricating advanced
Sec. 12.3
513
MOSFET Fabrication
memory chips such as the 256-kilobit DRAM. To increase the charge-storage capacity, the use of thinner gate insulators with higher dielectric contacts (e.g., Si 3 N4 and
Ta20s with dielectric constants of 8 and 22, respectively) are being explored. When
word and bit lines are simultaneously addressed (or activated to a high voltage), the
access transistor is turned on and the charge is transferred into the storage capacitor
if it had no initial charge (stored "zero"); little charge is transferred to the storage
capacitor if it had been fully charged initially (stored "one"). The amount of charge
that the bit line must supply to the storage capacitor is measured by the sensing circuitry, and this information is used to interpret whether a "zero" or "one" has been
stored in the cell. The sense circuitry then stores full charge in the capacitor if the
charge was there originally, or it fully depletes the capacitor if little charge existed
originally. The information in the cell is thus "refleshed" after it is read.
Silicon
nitride
Silicon
dioxide
11-Type silicon
(a)
(b)
(C)
(cl)
Second polysilicon
layer
First polysilicon
layer
---------- -----..~.r----.
(e)
If)
lg)
lhl
514
Chap. 12
2. Etching: The silicon nitride-pad oxide layers are then removed in a selective
etchant that does not attack silicon, and the first gate oxide is grown [Fig.
12-3-7(d)].
3. Deposition: The first-level polysilicon layer is deposited and patterned as
shown in Fig. 12-3-7(e).
4. Oxidation and deposition: The second gate oxide is grown and the secondlevel polysilicon is deposited and patterned as shown in Fig. 12-3-7(f).
5. Implantation: The exposed gate oxide region is implanted with an n-type dopant. A thick layer of silicon dioxide (P-glass) is deposited next, and the contact windows are opened in the oxide to reach the second-level polysilicon as
shown in Fig 12-3-7(g).
6. Deposition: Finally, a layer of aluminum is deposited and patterned as shown
in Fig. 12-3-7(h). A protective coating of silicon nitride can be deposited on
the wafer to seal it from contaminations.
The choice of lumped or distributed elements for amplifier matching networks depends on the operating frequency. When the frequency is up to X band, its wavelength is very short, and a smaller lumped element exhibits a negligible phase shift.
Because of the advanced thin-film technology, the size of lumped elements can be
greatly reduced and their operating frequencies can reach up to 20 GHz. Beyond that
distributed elements are preferred. In monolithic microwave integrated circuits
(MMICs), lumped resistors are very useful in thin-film resistive terminations for
couplers, lumped capacitors are absolutely essential for bias bypass applications, and
planar inductors are extremely useful for matching purposes, especially at lower microwave frequencies where stub inductors are physically too large [ 11].
1241 Planar Resistor Film
Sec. 12.4
Thin-Film. Formation
515
Planar resistors can be grouped into semiconductor films, deposited metal films, and
cermets.
Planar resistors based on semiconductors can be fabricated by forming an isolated band of conducting epitaxial film on the substrate by mesa etching or by isolation implant of the surrounding conducting film. Another way is by implanting a
high-resistivity region within the semi-insulating substrate. Metal-film resistors are
formed by evaporating a metal layer over the substrate and forming the desired pattern by photolithography. Cermet resistors are formed from films consisting of a
mixture of metal and a dielectric. Figure 12-4-1 shows several examples of planar
resistor design.
Resistor
"""'~
3 ::;,:,:::.,"'
substrate
Figure 12-4-1
resistors.
( c) Deposited resistor
Configurations of planar
= fp,
wt
where
e=
w
Ps
t
=
=
=
ohms
(12-4-1)
516
Chap. 12
Contacts
Resistive
material
Low-resistivity
material p,
Figure 12-4-2
When units of length f and width w are chosen to have equal magnitude, the result
is in a square. Therefore, the resistance R in ohms per square is independent of the
dimension of the square.
Example 12-4-1:
Solution
10 x 2.44 x 10- 8
x
x
_ = 0.244 fl/square
10 7
1
10
(b) S line
( d) Circular spiral
Sec. 12.4
Thin-Film Formation
Ribbon inductor.
517
[12]:
L
= 5.08 x 10- 3
where
e[
en(w
+ 1.19 + 0.022(w;
t)]
nH/mil
(12-4-2)
Round-wire inductor.
= 5.08 x
10- 3 f[fn(f/d)
+ 0.386] nH/mil
(12-4-3)
where d
L
where do
n
= 0.03125 n 2do
(12-4-4)
nH/mil
= separation in mils
J-
1.76
nH/mil
(12-4-5)
518
where A
n
Chap. 12
= 8.5 AI n 513
nH
(12-4-6)
Example 12-4-2:
n= 5
Number of turns:
Separation:
Film width:
s = 100 mils
w = 50 mils
The inductance is
L = 0.03125 (5) 2 x 2.5 (5)(50
= 1464.84 nH/mil
+ 100)
Metals
Dielectric
f - - - - - - - - -.....!:/..._,__ substrate
Substrate
lz:z:z:z:z:z:z:z====:doi---Ground plane
(b) lnterdigitated capacitor
where
ew
EoEr
farads
Eo
=
=
=
(12-4-7)
Sec. 12.5
519
lnterdigitated capacitor. The interdigitated capacitor consists of a singlelayer structure, and it can be fabricated easily on substrates as microstrip lines with
values between 0.1 and 15 pF as shown in Fig. 12-4-S(b). The capacitance can be
approximated as [ 13]
Er + l
=- [ (N
- 3)A1
A1J
pF/cm
(12-4-8)
N = 8
E, = 13.10
h = 0.254 cm
= 0.00254 cm
w = 0.051 cm
Solution
The capacitance is
C =
=
13.10 + l
0.
x 0.00254x[(8 - 3) x 0.089 + 0.10]
051
0.380 pF/cm
In the hybrid integrated circuit (HIC), semiconductor devices and passive circuit elements are formed on a dielectric substrate. The passive circuits are either distributed
elements or lumped elements or a combination of both. The distributed and lumped
elements are formed by using a thin- or thick-film process. The distributed circuit elements are generally single-layer metallization. The lumped elements are either fabricated by using multilevel deposition and plating techniques or are attached to the
substrate in chip form. Hybrid integrated circuits have been used almost exclusively
in the frequency range of 1 to 20 GHz for satellite communications, phased-array
radar systems, electronic warfare measures, and other commercial or military electronic systems because they can offer higher reliability, greater reproducibility, better performance, smaller size, and lower cost than conventional electronic circuits or
systems.
520
Chap. 12
Hybrid integrated circuits can be classified into two types: hybrid IC and
miniature hybrid IC.
1. Hybrid IC: This type uses the distributed circuit elements that are fabricated
on a substrate using a single-layer metallization technique. Other circuit elements such as resistors, inductors, capacitors, and semiconductor devices are
added to the substrate.
2. Miniature hybrid IC: The miniature hybrid IC uses multilevel elements. All
passive elements such as resistors, inductors, and capacitors are deposited on
the substrate; the semiconductor devices are attached to the substrate.
Hybrid ICs use a single-layer metallization technique to form the circuit components on a substrate. There are two techniques-plate-through and etchback-for
fabricating the hybrid integrated circuits.
Plate-through technique. The plate-through technique begins with a substrate coated with a thin layer of evaporated metal. The second step is to form a
thick photoresist as shown in Fig. 12-5-l(a).
After etching
Figure 12-5-1 Fabrication of hybrid ICs. (After M. Kumar and/. J. Bahl [13};
reprinted by permission of John Wiley & Sons.)
Chap. 12
References
521
The thickness of the photoresist is similar to the thickness of the final metal
film required. After defining a pattern in the photoresist, the second layer is plated
up to the desired thickness with precise definition, and only in the areas where metal
is required. Finally, the photoresist layer is removed, and the thin seed metal is
etched with very little undercut from the undesired areas.
Etchback technique. The etchback technique utilizes a thick metal layer
obtained completely by evaporation or by a combination of a thin evaporated layer
and a thick plated layer as shown in Fig. 12-5-l(b). A thin photoresist layer is used
as a mask to define the circuit pattern. Finally, the unwanted areas of the metal are
etched away. This technique results in thicker conductors than the plate-through
technique.
REFERENCES
[l] CAULTON, M., et al., Status of lumped elements in microwave integrated circuits:
present and future. IEEE Trans. on Microwave Theory and Techniques, MTT-19, No. 7,
July 1971.
[2] KEISTER, F. Z., An evaluation of materials and processes for integrated microwave circuits. IEEE Trans. on Microwave Theory and Techniques, MTT-16, No. 7, 469-475,
July 1968.
[3] CAULTON, M., and H. SOBOL, Microwave integrated circuit technology: A survey. IEEE
J. Solid-State Circuits, SC-5, No. 6, 292-303, December 1970.
[4] SOBOL, H., Applications of integrated circuit technology to microwave frequencies.
Proc. IEEE, 59, No. 8, 1200-1211, August 1971.
[5] SOBOL, H., Technology and design of hybrid microwave integrated circuits. Solid State
Technology, 13, No. 2, 49-59, February 1970.
[6] NANAVATI, R. P., Semiconductor Devices, p. 415, fig. 11-10. Intext Education Publishers, Scranton, PA, 1975.
[7] PARRILLO, L. C., VLSI process integration. Chapter 11, p. 464, fig. 14; p. 484, fig. 29
in S. M. Sze (Ed.) VLSI Technology. John Wiley & Sons, New York, 1983.
[8] HUNT, R. W., Memory design and technology. Chapter 11, p. 474, fig. 20 in S. M. Sze
(Ed.) VLSI Technology. John Wiley & Sons, New York, 1983.
[9] 0HZONE, T., et al., A 2k X 8 static RAM. P. 475, fig. 21 in S. M. Sze (Ed.) VLSI
Technology. John Wiley & Sons, New York, 1983.
[IO] OLDHAM, W. G., The fabrication of microelectronic circuits. P. 488, fig. 19 in S. M.
Sze (Ed.) VLSI Technology. John Wiley & Sons, New York, 1983.
[I I] PucEL, ROBEIIT A., Design considerations for monolithic microwave circuits. IEEE
Trans. on Microwave Theory and Techniques, MTT-29, No. 6, 513-534, July 1981.
[12] KUMAR, M., and I. J. BAHL, Microwave integrated circuits. Chapter 15, p. 794, fig.
15-4 in I. Bahl (Ed.) Microwave Solid-State Circuit Design. John Wiley & Sons, New
York, 1988.
[13] YOUNG, L., Advances in Microwaves, pp. 148-158. Academic Press, New York, 1974.
522
Chap. 12
SUGGESTED READINGS
BAHL, I., ed. Microwave Solid-State Circuit Design. John Wiley & Sons, New York, 1988.
HESS, K., Advanced Theory of Semiconductor Devices. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1988.
LIAO, S. Y., Semiconductor Electronic Devices. Prentice-Hall, Inc., Englewood Cliifs, N.J.,
1990.
STREETMAN, B. G., Solid-State Electronic Devices, 3rd ed. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1989.
SzE, S. M., ed., VLSI Technology. John Wiley & Sons, New York, 1983.
PROBLEMS
Materials
12-6.
12-7.
12-8.
12-9.
12-13. Discuss the discrete, integrated, and monolithic microwave integrated circuits.
12-14. Analyze the hybrid MMICs.
12-15. Describe the hybrid IC techniques.
Appendix A
TABLE A1
CONDUCTIVITY
<T
Conductor
<T
Insulator
Silver
Copper
Gold
Aluminum
Tungsten
Zinc
Brass
Nickel
Iron
Bronze
Solder
Steel (stainless)
Nichrome
Graphite
Silicon
Water (sea)
6.17 x 101
5.80 x 101
4.10 x 101
3.82 x 101
1.82 x 101
1.67 x 101
1.50 x 101
1.45 x 101
1.03 x 101
1.00 x 101
0.70 x 101
0.11 x 101
0.10 x 101
7.00 x 10
1.20 x 103
3-5
Quartz
Polystyrene
Rubber (hard)
Mica
Porcelain
Diamond
Glass
Bakelite
Marble
Soil (sandy)
Sands (dry)
Clay
Ground (dry)
Ground (wet)
Water (distilled)
Water (fresh)
Ferrite (typical)
<T
2 x
10- 10-2 -
2 x
10-17
10-16
10-15
10-14
10-13
10-13
10-12
10-9
10-
10-5
10-
10-
10- 5
10-3
10-
10-3
10-2
523
524
Appendix A
TABLE A-2
Material
f,
Material
Air
Alcohol (ethyl)
Bakelite
Glass
Ice
Mica (ruby)
Nylon
Paper
Plexiglass
Polyethylene
Polystyrene
Porcelain (dry process)
Quartz (fused)
Rubber
1
25
4.8
4-7
4.2
5.4
4
2-4
2.6-3.5
2.25
2.55
6
3.80
2.5-4
Sands (dry)
Silica (fused)
Snow
Sodium chloride
Soil (dry)
Styrofoam
Teflon
Water (distilled)
Water (sea)
Water (dehydrated)
Wood (dry)
Ground (wet)
Ground (dry)
Water (fresh)
TABLE A3
f,
4
3.8
3.3
5.9
2.8
1.03
2.1
80
20
1
1.5-4
5-30
2-5
80
RELATIVE PERMEABILITY,
Diamagnetic material
Bismuth
Paraffin
Wood
Silver
Paramagnetic material
0.99999860
0.99999942
0.99999950
0.99999981
Aluminum
Beryllinum
Nickel chloride
Manganese sulphate
1.00000065
1.00000079
1.00004
1.0001
TABLE A4
Ferromagnetic material
TABLE AS
Nickel
Cast iron
Cobalt
Machine steel
Ferrite (typical)
Transformer iron
Silicon iron
Iron (pure)
Mumetal
Supermalloy
PHYSICAL CONSTANTS
Boltzmann constant
Electronvolt
Electron charge
Electron mass
Ratio of charge to mass
of an electron
Planck's constant
x
x
x
x
x
10- 23 J/K
10- 19 J
10- 19 c
10- 31 kg
10- 11 C/kg
k
eV
q
m
e/m
1.381
1.602
1.602
9.109
1.759
6.626 x 10- 34 J - s
50
60
60
300
1,000
3,000
4,000
4,000
20,000
100,000
525
Appendix A
TABLE A&
Semiconductor
Geo9Sio1
Ge
GaAs
Ge
Zn Se
Ge
Zn Se
GaAs
AlAs
GaAs
GaP
Si
Al Sb
GaSb
Ga Sb
InAs
ZnTe
GaSb
ZnTe
lnAs
ZnTe
Al Sb
CdTe
PbTe
CdTe
InSb
ZnTe
CeSe(hex)
Energy
gap
(eV)
0.77
0.66
1.43
0.66
2.67
0.66
2.67
1.43
2.15
1.43
2.25
1.11
1.6
0.68
0.68
0.36
2.26
0.68
2.26
0.36
2.26
1.6
1.44
0.29
1.44
0.17
2.26
1.7
Lattice
constants
0
(A)
(5.63)
5.658
5.654
5.658
5.667
5.658
5.667
5.654
5.661
5.654
5.451
5.431
6.136
6.095
6.095
6.058
6.103
6.095
6.103
6.058
6.103
6.136
6.477
6.52
6.477
6.479
6.103
4.3(Yz)
(6.05)
Energy
gap
structu re
Indirect
Indirect
Direct
Indirect
Direct
Indirect
Direct
Direct
Indirect
Direct
Indirect
Indirect
Indirect
Direct
Direct
Direct
Direct
Direct
Direct
Direct
Direct
Indirect
Direct
Indirect
Direct
Direct
Direct
Direct
Expansion
coefficient
at 300K
(x 10- 6 0 C- 1)
Heterojunction
preferred
doping
5.7
5.8
5.7
7.0
5.7
7.0
5.8
5.2
5.8
5.3
2.33
3.7
6.9
6.9
4.5(5.3)
8.2
6.9
8.2
4.5(5.3)
8.2
3.7
5
19.8
5
4.9
8.2
2.45-4.4
n
p
n
p
n
p
n
p
p
n
n
p
n/p
p/n
n
p
p
n
p
n
p
n
p/n
n/p
p
n
p
n
Typical
dopants
P, As, Sb
Al, Ga, In
Se, Te
Al, Ga,
Al, Ga,
Al, Ga,
Al, Ga,
Zn, Cd
Zn
Se, Te
Se, Te
Al, Ga,
In
In
In
In
In
Se, Te/Zn, Cd
Zn, Cd/Se, Te
Se, Te
Zn, Cd
Cu
Se, Te
Cu
Se, Te
Cu
Se, Te
Li, Sb, P/I
Cl, Br/Na, K
Li, Sb
Se, Te
Cu
Cl, Br, I
Electron
affinity
(eV)
(4.1)
4.13
4.07
4.13
4.09
4.13
4.09
4.07
3.5
4.07
4.3
4.01
3.65
4.06
4.06
4.9
3.5
4.06
3.5
4.9
3.5
3.65
4.28
4.6
4.28
4.59
3.5
4.95
Source: After Milnes & Freucht, 116. Reprinted with permission from Heterojunctions and MetalSemiconductor Junctions, Academic Press, N.Y., 1972.
U1
N
OI
TABLE A7
E,
(eV)
Si
Ge
SiC(a)
AIP
AJAs
AJSb
GaP
GaAs
GaSb
lnP
lnAs
InSb
ZnS
Zn Se
ZnTe
CdS
CdSe
CdTe
PbS
Pb Se
PbTe
1.1 J
0.67
2.86
2.45
2. J6
1.6
2.26
1.43
0.7
1.28
0.36
O.J8
3.6
2.7
2.25
2.42
1.73
1.58
0.37
0.27
0.29
/.Ln
(cm2/V-sec)
J350
3900
500
80
J80
200
300
8500
5000
4000
22,600
JQ5
J 10
600
250
650
1050
575
JOOO
J600
J.Lp
(cm 2/V-sec)
p
(!1-cm)
480
J900
300
J50
400
JOOO
JOO
200
J700
JOO
J5
JOO
200
JOOO
700
O. J
5
J
4 x 10
0.04
8 x 10- 3
O.Q3
0.06
100
109
JOO
JQ5
1010
5 x 10- 3
10-3
10-2
q
Transition
Doping
i
i
i
i
i
i
i
d
d
d
d
d
d
d
d
d
d
d
i
i
i
n,p
n,p
n,p
n,p
n,p
n,p
n,p
np
n,p
n,p
n,p
n,p
n
n
p
n
n
n,p
n,p
n,p
n,p
Lattice
D
D
z
z
z
z
z
z
z
z
z
z.w
z
z
W,Z
H
H
H
(A)
5.43
5.66
3.08
5.46
5.66
6.J4
5.45
5.65
6.09
5.87
6.06
6.48
5.409
5.67J
6.lOJ
4.J37
4.30
6.482
5.936
6.J47
6.452
E,
J 1.8
J6.
10.2
10.9
11
J 1.1
J3.2
J5.7
J2.4
J4.6
J7.7
8.9
9.2
10.4
8.9
10.2
10.2
J6J
280
360
Density
(g/cm 3 )
Melting
point
(oC)
2.33
5.32
3.2J
2.40
3.60
4.26
4.J3
5.3J
5.6J
4.79
5.67
5.78
4.09
5.65
5.5J
4.82
5.8J
6.20
7.6
8.73
8.J6
J4J5
936
2830
2000
J740
1080
J467
J238
7J2
1070
943
525
J650t
JlOOt
J238t
J475
J258
J098
J119
108J
925
527
Appendix A
TABLE A8
J,(x)
Ji(x)
J,(x)
J,(x)
J,(x)
0.00
.02
.04
.06
.08
.JO
.12
.14
.16
.18
.20
.22
.24
.26
.28
.30
.32
.34
.36
.38
.40
.42
.44
.46
.48
.50
.52
.54
.56
.58
.60
.62
.64
.66
.68
.70
.72
.74
.76
.78
.80
.82
.84
.86
.88
.90
0.000
+.OJO
.020
.030
.040
.050
.060
.Q70
.080
.090
.0995
.J09
.119
.129
.139
.148
.158
.1675
.177
.187
.196
.205
.215
.224
.233
.242
.251
.260
.269
.278
.287
.295
.304
.312
.321
.329
.337
.345
.353
.361
.369
.3765
.384
.3915
.399
.406
0.92
.94
.96
.98
1.00
1.02
1.04
1.06
1.08
I. JO
1.12
1.14
1.16
1.18
1.20
1.22
1.24
1.26
1.28
1.30
1.32
1.34
1.36
1.38
1.40
1.42
1.44
1.46
1.48
1.50
1.52
1.54
1.56
1.58
1.60
1.62
1.64
1.66
1.68
1.70
1.72
1.74
1.76
1.78
1.80
1.82
1.84
0.413
.420
.427
.4335
.440
.4465
.453
.459
.465
.471
.477
.482
.488
.493
.498
.503
.508
.513
.5175
.522
.526
.5305
.534
.538
.542
.5455
.549
.552
.555
.558
.561
.563
.566
.568
.570
.572
.5735
.575
.5765
.578
.579
.580
.5805
.581
.5815
.582
.582
1.86
1.88
1.90
1.92
1.94
1.96
1.98
2.00
2.02
2.04
2.06
2.08
2.10
2.12
2.14
2.16
2.18
2.20
2.22
2.24
2.26
2.28
2.30
2.32
2.34
2.36
2.38
2.40
2.42
2.44
2.46
2.48
2.50
2.52
2.54
2.56
2.58
2.60
2.62
2.64
2.66
2.68
2.70
2.72
2.74
2.76
2.78
2.80
2.82
2.84
0.582
.5815
.581
.5805
.580
.579
.578
.577
.575
.574
.572
.570
.568
.566
.564
.561
.559
.556
.553
.550
.547
.543
.540
.536
.532
.5285
.524
.520
.516
.511
.507
.502
.497
.492
.487
.482
.476
.471
.465
.4595
.454
.448
.442
.435
.429
.423
.416
.4JO
.403
.396
2.86
2.88
2.90
2.92
2.94
2.96
2.98
3.00
3.02
3.04
3.06
3.08
3.JO
3.12
3.14
3.16
3.18
3.20
3.22
3.24
3.26
3.28
3.30
3.32
3.34
3.36
3.38
3.40
3.42
3.44
3.46
3.48
3.50
3.52
3.54
3.56
3.58
3.60
3.62
3.64
3.66
3.68
3.70
3.72
3.74
3.76
3.78
3.80
3.82
3.83
0.389
.3825
.375
.368
.361
.354
.3465
.339
.3315
.324
.316
.309
.301
.293
.285
.277
.269
.261
.253
.245
.237
.229
.221
.212
.204
.196
.1865
.179
.171
.1625
.154
.146
.137
.129
.121
.112
.104
.0955
.087
.079
.070
.062
.054
.0455
.037
.029
.021
.013
.005
.000
3.84
3.86
3.88
3.90
3.92
3.94
3.96
3.98
4.00
4.JO
4.20
4.30
4.40
4.50
4.60
4.70
4.80
4.90
5.00
5.05
5.JO
5.16
5.20
5.26
5.30
5.32
5.34
5.36
5.38
5.40
5.47
5.50
5.56
5.60
5.66
5.70
5.80
5.90
6.00
6.JO
6.20
6.30
6.40
6.60
6.70
6.80
6.90
7.00
7.01
-0.003
.Oil
.019
.027
.035
.043
.051
.058
.066
.J03
.139
.172
.203
.231
.2565
.279
.2985
.315
.3275
.334
.337
.341
.343
.345
.346
.346
.346
.346
.346
.345
.343
.341
.3375
.334
.3285
.324
.311
.295
.277
.256
.233
.208
.182
.125
.095
.065
.035
.005
.000
'Eu
I
1014
10 13
1015
1016
1011
1018
1019
1020
I 021
Figure A-1
10 2
104
-;r
103
:0
0
"
10 2
10 4
10 2 L.L.l...LJ.J..UJ.L__J_LLllJllL-1......L.LliillJ.._..J.._Ll.J.illlJ~t...:I:,.....JJ!J
1016
1018
iot9
1014
1015
Impurity concentration (cm- 3 )
(Reprinted from Sze's Physics of Semiconductor Devices, 1981, by permission of John Wiley & Sons, Inc.)
Figure A-2
Appendix B
UNITS OF llllEASUREllllENT
In field intensity measurements the units of measure and the conversion from one
unit to another are the essential parts of the process. A few widely used units are described here:
1. dB-The decibel (dB) is a dimensionless number that expresses the ratio of two
power levels. It is defined as
dB = 10
log10(~:)
(B-1)
The two power levels are relative to each other. If power level P2 is higher than
power level P1, dB is positive and vice versa. Since P = V 2 /R, when their
voltages are measured across the same or equal resistors, the number of dB is
given by
dB
20
log10(~:)
(B-2)
The voltage definition of dB has no meaning at all unless the two voltages under consideration appear across equal impedances. Above 10 GHz the
impedance of waveguides varies with frequency, and the dB calibration is limited to power levels only. Table B-1 shows the conversion of voltage and
power ratios to dB.
2. dBW-The decibel above I watt (dBW) is another useful measure for expressing power level P2 with respect to a reference power level P1 of I W. Similarly,
if the power level P2 is lower than I W, the dBW is negative.
530
TABLE B1
Appendix B
CONVERSION OF VOLTAGE AND POWER RATIOS TO DECIBELS
Voltage
ratio
Power
ratio
1.000
0.989
.977
.966
.955
.944
.933
.923
.912
.902
.891
.881
.871
.861
.851
.841
.832
.822
.813
.804
.794
.785
.776
.767
.759
.750
.741
.733
.724
.716
.708
.700
.692
.684
.676
.668
.661
.653
.646
.638
.631
.624
.617
.610
.603
1.000
0.977
.955
.933
.912
.891
.871
.851
.832
.813
.947
.776
.759
.741
.724
.708
.692
.676
.661
.646
.631
.617
.603
.589
.575
.562
.550
.537
.525
.513
.501
.490
.479
.468
.457
.447
.437
.427
.417
.407
.398
.389
.380
.372
.363
-dB+
0.0
.1
.2
.3
.4
.5
.6
.7
.8
.9
1.0
I. I
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3.0
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
4.0
4.1
4.2
4.3
4.4
Voltage
ratio
Power
ratio
Voltage
ratio
Power
ratio
1.000
1.012
1.023
1.035
1.047
1.059
1.072
1.084
1.095
1.109
1.122
1.135
1.148
1.161
1.175
1.189
1.202
1.216
1.230
1.245
1.259
1.274
1.288
1.303
1.318
1.334
1.349
1.365
1.380
1.390
1.413
1.429
1.445
1.462
1.479
1.496
1.514
1.531
1.549
1.567
1.585
1.603
1.622
1.641
1.660
1.000
1.023
1.047
1.072
1.096
1.122
1.148
1.175
1.202
1.230
1.259
1.288
1.318
1.349
1.380
1.413
1.445
1.479
1.514
1.549
1.585
1.622
1.660
1.698
1.738
1.778
1.820
1.862
1.905
1.950
1.995
2.042
2.089
2.138
2.188
2.239
2.291
2.344
2.399
2.455
2.512
2.570
2.630
2.692
2.754
0.596
.589
.582
.575
.569
.562
.556
.550
.543
.537
.531
.525
.519
.513
.507
.501
.496
.490
.484
.479
.473
.468
.462
.457.
.452
.447
.442
.437
.432
.427
.422
.417
.412
.407
.403
.398
.394
.389
.385
.380
.376
.372
.367
.363
.359
0.355
.347
.339
.331
.324
.316
.309
.302
.295
.288
.282
.275
.269
.263
.257
.251
.246
.240
.234
.229
.224
.219
.214
209
.204
.200
.195
.191
.186
.182
.178
.174
.170
.166
.162
.159
.155
.151
.148
.145
.141
.138
.135
.132
.129
-dB+
4.5
4.6
4.7
4.8
4.9
5.0
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
6.0
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6.8
6.9
7.0
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
8.0
8.1
8.2
8.3
8.4
8.5
8.6
8.7
8.8
8.9
Voltage
ratio
Power
ratio
1.679
1.698
1.718
1.738
1.758
1.778
1.799
1.820
1.841
1.862
1.884
1.905
1.928
1.950
1.972
1.995
2.018
2.042
2.065
2.089
2. ll3
2.138
2.163
2.188
2.215
2.239
2.265
2.291
2.317
2.344
2.371
2.399
2.427
2.455
2.483
2.512
2.541
2.570
2.600
2.630
2.661
2.692
2.723
2.754
2.786
2.818
2.884
2.951
3.020
3.090
3.162
3.236
3.3ll
3.388
3.467
3.548
3.631
3.715
3.802
3.890
3.981
4.074
4.159
4.265
4.365
4.467
4.571
4.677
4.786
4.898
5.012
5.129
5.248
5.370
5.495
5.623
5.754
5.888
6.026
6.166
6.310
6.457
6.607
6.761
6.918
7.079
7.244
7.413
7.586
7.762
531
Appendix B
TABLE B1
Voltage
ratio
(CONTINUED)
Power
ratio
-dB+
Voltage
ratio
Power
ratio
Voltage
ratio
Power
ratio
-dB+
Voltage
ratio
Power
ratio
0.355
.351
.347
.343
.339
0.126
.123
.120
.118
.115
9.0
9.1
9.2
9.3
9.4
2.818
2.851
2.884
2.917
2.951
7.943
8.128
8.318
8.511
8.710
0.211
.209
.207
.204
.202
0.0447
.0437
.0427
.0417
.0407
13.5
13.6
13.7
13.8
13.9
4.732
4.786
4.842
4.898
4.955
22.39
22.91
23.44
23.99
24.55
.335
.331
.327
.324
.320
.112
.110
.107
.105
.102
9.5
9.6
9.7
9.8
9.9
2.985
3.020
3.055
3.090
3.126
.200
.197
.195
.193
.191
.0398
.0389
.0380
.0372
.0363
14.0
14.1
14.2
14.3
14.4
5.012
5.070
5.129
5.188
5.248
25.12
25.70
26.30
26.92
27.54
.316
.313
.309
.306
.302
.JOO
.0977
.0955
.0933
.0912
10.0
IQ.I
10.2
10.3
10.4
3.162
3.199
3.236
3.273
3.311
8.913
9.120
9.333
9.550
9.772
10.000
10.23
10.47
10.72
10.96
.188
.186
.184
.182
.180
.0355
.0347
.0339
.0331
.0324
14.5
14.6
14.7
14.8
14.9
5.309
5.370
5.433
5.495
5.559
28.18
28.84
29.51
30.20
30.90
.299
.295
.292
.288
.283
.282
.279
.275
.272
.269
.0891
.0871
.0851
.0832
.0813
.0794
.0776
.0759
.0741
.0724
10.5
10.6
10.7
10.8
10.9
11.0
I I. I
11.2
11.3
11.4
3.350
3.388
3.428
3.467
3.508
3.548
3.589
3.631
3.673
3.715
11.22
11.48
I 1.75
12.02
12.30
12.59
12.88
13.18
13.49
13.80
.178
.176
.174
.172
.170
.168
.166
.164
.162
.160
.0316
.0309
.0302
.0295
.0288
.0282
.0275
.0269
.0263
.0257
15.0
15.1
15.2
15.3
15.4
15.5
15.6
15.7
15.8
15.9
.266
.263
.260
.257
.254
.0708
.0692
.0676
.0661
.0646
11.5
11.6
11.7
11.8
11.9
3.758
3.802
3.846
3.890
3.936
14.13
14.45
14.79
15.14
15.49
.159
.157
.155
.153
.151
.251
.248
.246
.243
.240
.0631
.0617
.0603
.0589
.0575
12.0
12.1
12.2
12.3
12.4
3.981
4.027
4.074
4.121
4.169
15.85
16.22
16.60
16.98
17.38
.150
.148
.146
.145
.143
.0251
.0246
.0240
.0234
.0229
.0224
.0219
.0214
.0209
.0204
16.0
16.1
16.2
16.3
16.4
16.5
16.6
16.7
16.8
16.9
5.623
5.689
5.754
5.821
5.888
5.957
6.026
6.095
6.166
6.237
6.310
6.383
6.457
6.531
6.607
6.683
6.761
6.839
6.918
6.998
31.62
32.36
33.11
33.88
34.67
35.48
36.31
37.15
38.02
38.90
39.81
40.74
41.69
42.66
43.65
44.67
45.71
46.77
47.86
48.98
.237
.234
.232
.229
.227
.224
.221
.219
.216
.214
.0562
.0550
.0537
.0525
.0513
.0501
.0490
.0479
.0468
.0457
12.5
12.6
12.7
12.8
12.9
13.0
13.1
13.2
13.3
13.4
4.217
4.266
4.315
4.365
4.416
4.467
4.519
4.571
4.624
4.677
17.78
18.20
18.62
19.05
19.50
19.95
20.42
20.89
21.38
21.88
.141
.140
.138
.137
.135
.0200
.0195
.0191
.0186
.0182
17.0
17.1
17.2
17.3
17.4
7.079
7.161
7.244
7.328
7.413
50.12
51.29
52.48
53.70
54.95
.133
.132
.130
.129
.127
.0178
.0174
.0170
.0166
.0162
17.5
17.6
17.7
17.8
17.9
7.499
7.586
7.674
7.762
7.852
56.23
57.54
58.88
60.26
61.66
532
TABLE B1
Appendix B
(CONTINUED)
Voltage
ratio
Power
ratio
-dB+
Voltage
ratio
Power
ratio
0.126
.125
.123
.122
.120
.119
.118
.116
.115
.114
.112
.111
.IIO
.I08
.I07
0.0159
.0155
.0151
.0148
.0145
.0141
.0138
.0135
.0132
.0129
.0126
.0123
.0120
.0118
.0115
18.0
18.1
18.2
18.3
18.4
18.5
18.6
18.7
18.8
18.9
19.0
19.1
19.2
19.3
19.4
7.943
8.035
8.128
8.222
8.318
8.414
8.511
8.6IO
8.7IO
8.811
8.913
9.016
9.120
9.226
9.333
63.IO
64.57
66.07
67.61
69.18
70.79
72.44
74.13
75.86
77.62
79.43
81.28
83.18
85.11
87.10
Voltage
ratio
Power
ratio
O.I06
.I03
.104
.102
.IOI
.100
0.0112
.0110
.OI07
.DI05
.0102
.0100
I0-3
I0-4
IO-'
I0-6
I0-7
10-s
I0-9
IO-IO
IO-II
I0-12
I0-2
I0-3
I0-4
10-s
10-6
-dB+
19.5
19.6
19.7
19.8
19.9
20.0
30
40
50
60
70
80
Voltage
ratio
Power
ratio
9.441
9.550
9.661
9.772
9.886
I0.000
89.13
91.20
93.33
95.50
97.72
100.00
I03
IO
IO'
I06
I07
I08
I09
IOIO
IOll
1012
102
I03
IO
90
100
110
120
IO'
106
3. dBm-The decibel above 1 milliwatt (dBm) is also a useful measure of expressing power level P 2 with respect to a reference power level P1 of 1 milliwatt (mW). Since the power level in the microwave region is quite low, the
dBm unit is very useful in that frequency range. It is customary to designate
milli by a lowercase letter m and mega by an uppercase letter M.
4. dBV-The decibel above 1 volt (dBV) is a dimensionless voltage ratio in dB
referred to a reference voltage of 1 V.
5. dBV-The decibel above 1 microvolt (dBV) is another dimensionless
voltage ratio in dB referred to a reference voltage of 1 V. The field intensity
meters used for the measurements in the microwave region often have a scale
in dB V, since the power levels to be measured are usually extremely low.
6. V /m-Microvolts per meter (V /m) are units of 10- 6 V /m expressing the
electric field intensity.
7. dBV /m-The decibel above 1 microvolt per meter (dBV /m) is a dimensionless electric field intensity ratio in dB relative to 1V / m. This unit is also
often used for field intensity measurements in the microwave region.
8. V /m/MHz-Microvolts per meter per megahertz (V /m/MHz) are units of
10- 6 V / m per broadband electric field intensity distribution. This is a twodimensional distribution, in space and in frequency.
9. dBV /m/MHz-The decibel above 1 microvolt per meter per megahertz
(dBV /m/MHz) is a dimensionless broadband electric field intensity distribution ratio with respect to 1 V /m/MHz.
10. V /MHz-Microvolts per megahertz per second of bandwidth ( V /MHz)
are units of 1o- 6 V s of broadband voltage distribution in the frequency do-
533
Appendix B
main. The use of this unit is based on the assumption that the voltage is evenlly distributed over the bandwidth of interest.
Figure B-1 shows the conversion of dB scales in power and voltage:
dBW = -30 + dBm
=
dBW
-60
+ dBW
dBW
dBm
dBV
= -60 + dBmV
= -120 + dBV
dBmV
dBV
30
60
60
120
-30
30
-60
60
-30
-120
-60
-60 -
Figure Bl
',
dBV
(B-1)