Microwave

Devices
and Circuits
Third Edition

SAMUEL Y. LlAO
Professor of Electrical Engineering
California State University, Fresno

PRENTICE HALL, fnglewood C l ~ sNew

, Jersey 0763.2
Contents

PREFACE

Chapter 0 INTRODUCTION

0- 1 Microwave Frequencies 1
0-2 Microwave Devices 2
0-3 Microwave Systems 3
0-4 Microwave Units of Measure 4

Chapter 1 INTERACTIONS BETWEEN ELECTRONS AND FIELDS 6

1-0 Introduction 6
1-1 Electron Motion in an Electric Field 6
1-2 Electron Motion in a Magnetic Field 10
1-3 Electron Motion in an Electromagnetic Field 12
Suggested Readings 15
Problems 15

Chapter 2 ELECTROMAGNETIC PLANE WAVES

2-0 Introduction 16
2- 1 Electric and Magnetic Wave Equations 17
Contents

2-2 Poynting Theorem 19

2-3 Uniform Plane Waves and Reflection 21
2-3-1 Uniform Plane Waves, 21
2-3-2 Boundary Conditions, 23
2-3-3 Uniform Plane- Wave Rejection, 24
2-4 Plane-Wave Propagation in Free Space and Lossless
Dielectric 29
2-4-1 Plane- Wave Propagation in Free Space, 29
2-4-2 Plane-Wave Propagation in Lossless Dielectric, 32
2-5 Plane-Wave Propagation in Lossy Media 32
2-5-1 Plane Wave in Good Conductor, 33
2-5-2 Plane Wave in Poor Conductor, 35
2-5-3 Plane Wave in Lossy Dielectric, 36
2-6 Plane-Wave Propagation in Metallic-Film Coating on Plastic
Substrate 41
2-6-1 Surface Resistance of Metallic Films, 42
2-6-2 Optical Constants of Plastic Substrates and Metallic Films, 42
2-6-3 Microwave Radiation Attenuation of Metallic-Film Coating on
Plastic Substrate, 44
2-6-4 Light Transmittance of Metallic-Film Coating on Plastic Sub-
strate, 46
2-6-5 Plane Wave in Gold-Film Coating on Plastic Glass, 47
2-6-6 Plane Wave in Silver-Film or Copper-Film Coating on Plastic
Substrate, 52
References 57
Suggested Readings 57
Problems 58

Chapter 3 MICROWAVE TRANSMISSION LINES

3-0 Introduction 61
3- 1 Transmission-Line Equations and Solutions 6 1
3-1-1 Transmission-Line Equations, 61
3-1-2 Solutions of Transmission-Line Equations, 64
3-2 Reflection Coefficient and Transmission Coefficient 67
3-2-1 Rejection Coeficient, 67
3-2-2 Transmission CoefJicient, 69
3-3 Standing Wave and Standing-Wave Ratio 71
3-3-1 Standing Wave, 71
3-3-2 Standing-Wave Ratio, 74
3-4 Line Impedance and Admittance 76
3-4-1 Line Impedance, 76
3-4-2 Line Admittance. 81
Contents

3-5 Smith Chart 82

3-6 Impedance Matching 89
3-6-1 Single-Stub Matching, 90
3-6-2 Double-Stub Matching, 92

3-7 Microwave Coaxial Connectors 96

References 98
Suggested Readings 98
Problems 98

Chapter 4 MICROWAVE WAVEGUIDES AND COMPONENTS

4-0 Introduction 102

4- 1 Rectangular Waveguides 103
4-1-1 Solutions of Wave Equations in Rectangular Coordinates, 104
4-1-2 TE Modes in Rectangular Waveguides, 106
4-1-3 TM Modes in Rectangular Waveguides, 111
4-1-4 Power Transmission in Rectangular Waveguides, 113
4-1-5 Power Losses in Rectangular Waveguides, 113
4-1-6 Excitations of Modes in Rectangular Waveguides, 116
4-1-7 Characteristics of Standard Rectangular Waveguides, 117
4-2 Circular Waveguides 119
4-2-1 Solutions of Wave Equations in Cylindrical Coordinates, 119
4-2-2 TE Modes in Circular Waveguides, 122
4-2-3 TM Modes in Circular Waveguides, 127
4-2-4 TEM Modes in Circular Waveguides, 129
4-2-5 Power Transmission in Circular Waveguides or Coaxial
Lines, 131
4-2-6 Power Losses in Circular Waveguides or Coaxial Lines, 133
4-2-7 Excitations of Modes in Circular Waveguides, 133
4-2-8 Characteristics of Standard Circular Waveguides, 135

4-3 Microwave Cavities 135

4-3-1 Rectangular-Cavity Resonator, 135
4-3-2 Circular-Cavity Resonator and Semicircular-Cavity Res-
onator, 136
4-3-3 Q Factor of a Cavity Resonator, 139

4-4 Microwave Hybrid Circuits 141

4-4-1 Waveguide Tees, 144
4-4-2 Magic Tees (Hybrid Trees), 146
4-4-3 Hybrid Rings (Rat-Race Circuits), 147
4-4-4 Waveguide Corners, Bends, and Twists, 148

4-5 Directional Couplers 149

4-5-1 Two-Hole Directional Couplers, 151
4-5-2 S Matrix of a Directional Coupler, 151
4-5-3 Hybrid Couplers, 154
Contents

4-6 Circulators and Isolators 156

4-6-1 Microwave Circulators, 1.58
4-6-2 Microwave isolators, 160
References 16 1
Suggested Readings 161
Problems 161

Chapter 5 MICROWAVE TRANSISTORS AND TUNNEL DIODES 166

5-0 Introduction 166

5- 1 Microwave Bipolar Transistors 169
5-1-1 Physical Structures, 170
5-1-2 Bipolar Transistor Configurations, 173
5-1-3 Principles of Operation, 178
5-1-4 AmpliJication Phenomena, 187
5-1-5 Power-Frequency Limitations, 190
5-2 Heterojunction Bipolar Transistors (HBTs) 193
5-2-1 Physical Structures, 193
5-2-2 Operational Mechanism, 194
5-2-3 Electronic Applications, 197
5-3 Microwave Tunnel Diodes, 198
5-3-1 Principles of Operation, 198
5-3-2 Microwave Characteristics, 201
References 204
Suggested Readings 205
Problems 205

Chapter 6 MICROWAVE FIELD-EFFECT TRANSISTORS

6-0 Introduction 208

6- 1 Junction Field-Effect Transistors ( JFETs) 209
6-1-1 Physical Structure, 209
6-1-2 Principles of Operation, 210
6-1-3 Current- Voltage (i-V ) Characteristics, 21 1
6-2 Metal-Semiconductor Field-Effect Transistors
(MESFETs) 2 16
6-2-1 Physical Structures, 21 7
6-2-2 Principles of Operarion, 218
6-2-3 Small-Signal Equivalent Circuit, 221
6-2-4 Drain Current id, 223
6-2-5 Cutoff Frequency f,, and Maximum Oscillation Frequency
f-7 228
Contents ix

6-3 High Electron Mobility Transistors (HEMTs) 230

6-3-1 Physical Structure, 230
6-3-2 Operational Mechanism, 232
6-3-3 Performance Characteristics, 233
6-3-4 Electronic Applications, 236
6-4 Metal-Oxide-Semiconductor Field-Effect Transistors
(MOSFETs) 237
6-4-1 Physical Structures, 237
6-4-2 Electronic Mechanism, 238
6-4-3 Modes of Operation, 238
6-4-4 Drain Current and Transconductance, 239
6-4-5 Maximum Operating Frequency, 243
6-4-6 Electronic Applications, 244
6-5 MOS Transistors and Memory Devices 245
6-5-1 NMOS Devices, 245
6-5-2 CMOS Devices, 248
6-5-3 Memory Devices, 252
6-6 Charge-Coupled Devices (CCDs) 254
6-6-1 Operational Mechanism, 255
6-6-2 Surface-Channel Charge-Coupled Devices (SCCDs), 258
6-6-3 Dynamic Characteristics, 259
References 26 1
Suggested Readings 262
Problems 263

Chapter 7 TRANSFERRED ELECTRON DEVICES (TEDs)

7-0 Introduction 269

7- 1 Gunn-Effect Diodes-GaAs Diode, 270
7-1-1 Background, 270
7-1-2 Gunn Effect, 271
7-2 Ridley-Watkins-Hilsum (RWH) Theory 273
7-2-1 Differential Negative Resistance, 273
7-2-2 Two-Valley Model Theory, 274
7-2-3 High-Field Domain, 280
7-3 Modes of Operation 284
7-3-1 Criterion for Classijjing the Modes of Operation, 285
7-3-2 Gunn Oscillation Modes (10'2/cm25 (noL) < 10'~/crn'),287
7-3-3 Limited-Space-Charge Accumulation (LSA) Mode
( f l > 2 X lo7 crnls), 289
7-3-4 Stable Amplijication Mode (noL < 10'2/crn2),290
7-4 LSA Diodes 291
7-5 InP Diodes 293
Contents

7-6 CdTe Diodes 296

7-7 Microwave Generation and Amplification 296
7-7-1 Microwave Generation, 296
7-7-2 Microwave AmpliJication, 298
References 300
Suggested Readings 30 1
Problems 302

Chapter 8 AVALANCHE TRANSIT-TIME DEVICES

8-0 Introduction 303

8- 1 Read Diode 304
8-1-I Physical Description, 304
8-1-2 Avalanche Multiplication, 305
8-1-3 Carrier Current I,(?)and External Current I,(t), 306
8-1-4 Output Power and Quality Factor Q , 308
8-2 IMPATT Diodes 309
8-2-1 Physical Structures, 309
8-2-2 Negative Resistance, 309
8-2-3 Power Output and Efficiency, 31 1
8-3 TRAPATT Diodes 3 14
8-3-1 Physical Structures, 314
8-3-2 Principles of Operation, 314
8-3-3 Power Output and Efficiency, 316
8-4 BARITT Diodes 3 17
8-4-1 Physical Description, 31 7
8-4-2 Principles of Operation, 317
8-4-3 Microwave Performance, 319
8-5 Parametric Devices 320
8-5-1 Physical Structures, 320
8-5-2 Nonlinear Reactance and Manley-Rowe Power Relations, 321
8-5-3 Parametric Amplifiers, 326
8-5-4 Applications, 330
References 33 1
Suggested Readings 332
Problems 333

Chapter 9 MICROWAVE LINEAR-BEAM TUBES ( 0 TYPE) 335

9-0 Introduction 335

9-1 Conventional Vacuum Triodes, Tetrodes, and Pentodes 338
9-1-1 Lead-Inductance and Interelectrode-Capacitance Effects, 338
Contents

9-1-2 Transit-Angle Effects, 339

9-1-3 Gain-Bandwidth Product Limitation, 340
9-2 Klystrons 341
9-2-1 Reentrant Cavities, 342
9-2-2 Velocity-Modulation Process, 345
9-2-3 Bunching Process
9-2-4 Output Power and Beam Loading, 354
9-2-5 State of the Art, 360
9-3 Multicavity Klystron Amplifiers 362
9-3-1 Beam-Current Density, 363
9-3-2 Output Current and Output Power of Two-Caviry Klystron,
369
9-3-3 Output Power of Four-Cavity Klystron, 371
9-4 Reflex Klystrons 373
9-4-1 Velocity Modulation 374
9-4-2 Power Output and EfJiciency, 376
9-4-3 Electronic Admittance, 379
9-5 Helix Traveling-Wave Tubes (TWTs) 382
9-5-1 Slow-Wave Structures, 384
9-5-2 Amplijication Process, 388
9-5-3 Convection Current, 391
9-5-4 Axial Electric Field, 392
9-5-5 Wave Modes, 394
9-5-6 Gain Consideration, 396
9-6 Coupled-Cavity Traveling-Wave Tubes 398
9-6-1 Physical Description, 398
9-6-2 Principles of Operation, 400
9-6-3 Microwave Characteristics, 402
9-7 High-Power and Gridded-Control Traveling-Wave Tubes 404
9-7-1 High EfJiciency and Collector Voltage Depression, 406
9-7-2 Normal Depression and Overdepression of Collector Voltage,
407
9-7-3 Two-Stage Collector Voltage Depression Technique, 410
9-7-4 Stabilization of Cathode and Collector Voltages, 412
References 4 17
Suggested Readings 4 18
Problems 419

Chapter 10 MICROWAVE CROSSED-FIELD TUBES (M TYPE)

10-0 Introduction 425

10-1 Magnetron Oscillators 427
10-1-I Cylindrical Magnetron, 42 7
10-1-2 Linear Magnetron, 436
Contents

10-1-3 Coaxial Magnetron, 442

10-1-4 Voltage-Tunable Magnetron, 443
10-1-5 Inverted Coaxial Magnetron, 444
10-1-6 Frequency-Agile Coaxial Magnetron, 447
10-2 Forward-Wave Crossed-Field Amplifier (FWCFA OR
CFA) 450
10-2-1 Principles of Operation, 450
10-2-2 Microwave Characteristics 455
10-3 Backward-Wave Crossed-Field Amplifier (Amplitron) 457
10-4 Backward-Wave Crossed-Field Oscillator
(Carcinotron) 46 1
References 466
Problems 467

Chapter 11 STRIP LINES

11-0 Introduction 472

1 1-1 Microstrip Lines 473
11-1-1 Characteristic Impedance of Microstrip Lines, 473
11-1-2 Losses in Microstrip Lines 477
11-1-3 Quality Factor Q of Microstrip Lines 484
1 1-2 Parallel Strip Lines 485
11 -2-1 Distributed Parameters 486
11-2-2 Characteristiclmpedance 486
11-2-3 Attenuation Losses 487
1 1-3 Coplanar Strip Lines 488
11-4 Shielded Strip Lines 489
References 49 1
Problems 492

Chapter 12 MONOLllHIC MICROWAVE INTEGRATED CIRCUITS 495

12-0 Introduction 495

12-1 Materials 497
12-1-1 Substrate Materials 498
12-1-2 Conductor Materials 498
12-1-3 Dielectric Materials 500
12-1-4 Resistive Materials 500
12-2 Monolithic Microwave Integrated-Circuit Growth 501
12-2-1 MMIC Fabrication Techniques, 502
12-2-2 Fabrication Example, 504
Contents xiii

12-3 MOSFET Fabrication 504

12-3-1 MOSFET Formation, 505
12-3-2 NMOS Growth, 506
12-3-3 CMOS Development, 508
12-3-4 Memory Construction, 510
12-4 Thin-Film Formation 5 14
12-4-1 Planar Resistor Film 514
12-4-2 Planar Inductor Film 516
12-4-3 Planar Capacitor Film 518
12-5 Hybrid Integrated-Circuit Fabrication 5 19
References 52 1
Suggested Readings 522
Problems 522

APPENDIX A

APPENDIX 8

INDEX
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 ex-
plain 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 expan-
sions as well as some corrections and deletions. The problems section has been en-
larged 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 het-
erojunction 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 IMPATT,
TRAPATT, and BARITT diodes and the parametric devices.
xvi Preface

Chapter 9 deals with microwave linear-beam tubes including klystrons, reflex

klystron, and traveling-wave tubes (TWTs).
Chapter 10 studies microwave crossed-field tubes such as magnetrons, for-
ward-wave crossed-field amplifiers, and the backward-wave crossed-field am-
plifiers-Amplitron and Carcinotron.
Chapter 11 explains strip lines including microstrip, parallel, coplanar, and
shielded strip lines.
Chapter 12 analyzes monolithic microwave integrated circuits (MMICs) in-
cluding MMIC growth, MOSFET fabrication, thin-film formation, and hybrid inte-
grated-circuit fabrication.
The arrangement of topics is flexible; an instructor may conveniently select or
order the several topics to suit either a one-semester or a one-quarter course. Numer-
ous problems for each chapter establish the reader's further understanding of the
subjects discussed. Instructors who have adopted the book for their courses may ob-
tain a solutions manual from the publisher.
The author is grateful to the several anonymous reviewers; their many valuable
comments and constructive suggestions helped to improve this edition. The author
would also like to acknowledge his appreciation to the many instructors and students
who used the first two editions and who have offered comments and suggestions. All
of this help was vital in improving this revision, and this continuing group effort is
sincerely invited. Finally, I wish to express my deep appreciation to my wife, Lucia
Hsiao Chuang Lee, and our children: Grace in bioengineering, Kathy in electrical
engineering, Gary in electronics engineering, and Jeannie in teachers education, for
their valuable collective contributions. Therefore, this revision is dedicated to them.
Samuel Y . Liao
Chapter 0

1 Introduction

The central theme of this book concerns the basic principles and applications of mi-
crowave 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.

0-1 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 re-
ally indicates the wavelengths in the micron ranges. This means microwave frequen-
cies are up to infrared and visible-light regions. In this revision, microwave frequen-
cies refer to those from 1 GHz up to lo6 GHz. The microwave band designation that
derived from World War I1 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 Chap. 0

TABLE 0.1 U.S. MILITARY MICROWAVE BANDS

Designation Frequency range in gigahertz

P band 0.225- 0.390

L band 0.390- 1.550
S band 1.550- 3.900
C band 3.900- 6.200
X band 6.200- 10.900
K band 10.900- 36.000
Q band 36.000- 46.000
V band 46.000- 56.000
W band 56.000-100.000

TABLE 0-2 U.S. NEW MILITARY MICROWAVE BANDS

Designation Frequency range in gigahertz Designation Frequency range in gigahertz

A band 0.100-0.250 H band 6.000- 8.000

B band 0.250-0.500 I band 8.000- 10.000
C band 0.500-1.000 J band 10.000- 20.000
D band 1.000-2.000 K band 20.000- 40.000
E band 2.000-3.000 L band 40.000- 60.000
F band 3.000-4.000 M band 60.000-100.000
G band 4.000-6.000

TABLE 0.3 IEEE MICROWAVE FREQUENCY BANDS

Designation Frequency range in gigahertz

HF
VHF
UHF
L band
S band
C band
X band
Ku band
K band
Ka band
Millimeter
Submillimeter

0-2 MICRO WAWE 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 capaci-
tance, and lead inductance appeared to limit the operation of vacuum tubes in mi-
crowave 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 3

cuits. In 1939 W. C. Hahn and G. F. Metcalf proposed a theory of velocity modula-

tion 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 . Rid-
ley 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 solid-
state 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.

0-3 MICROWAVE SYSTEMS

A microwave system normally consists of a transmitter subsystem, including a mi-

crowave oscillator, waveguides, and a transmitting antenna, and a receiver subsys-
tem 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 ade-
quate knowledge of the components involved is essential. Besides microwave
devices, the text therefore describes microwave components, such as resonators, cav-
ities, microstrip lines, hybrids, and microwave integrated circuits.

Microwave Transmitting Receiving Output to

source horn antenna horn antenna oscilloscope or
I I power meter

Figure 0-1 Microwave system.

Introduction Chap. 0

0-4 MICROWAVE UNITS OF MEASURE

Microwave measures can be expressed in different units, such as the CGS (centime-
ter-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 Interna-

TABLE 0-4 MKS UNITS

Quantity Unit Symbol

Capacitance farad = coulomb per volt F
Charge coulomb: A - s Q
Conductance mhos U
Current ampere = coulomb per second A
Energy joule J
Field volt per meter E
Flux linkage weber = volt. second 4J
Frequency cycle per second
Inductance henry = (V - s ) / A
Length meter
Power watt = joule per second
Resistance ohm
Time second
Velocity meter per second
Voltage volt

Note: 1 tesla = I weber/m2 = lo4 gausses = 3 X ESU

1 A (angstrom) = lo-'' m
1 p m (micron) = m

TABLE 0-5 PREFIXES

Prefix Factor Symbol

exa 10l8 E
eta loi5 P
tera 10l2 T
gigs 10' G
mega 1O6 M
kilo 10' k
hecto 102 h
deka 10 da
deci lo-' d
centi c
milli lo-' m
micro 10-0 P
nano n
pic0 P
femto 10-15 f
atto lo-'" a
Sec. 0.4 Microwave Units of Measure 5

tional Committee on Weights and Measures. They have been adopted by the Na-
tional 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 0-6 PHYSICAL CONSTANTS

Constant Symbol Value

Boltzmann constant k 1.381 X JI0K

Electronvolt eV 1.602 X 10-l9 J
Electron charge 9 1.602 X C
Electron mass m 9.109 X lo-)' kg
Ratio of charge to mass of an electron elm 1.759 x 10" Clkg
Permeability of free space Po 1.257 X Hlm or 4n X lo-' Hlm
Permittivity of free space Eo 8.854 X 10-l2 F/m
Planck's constant h 6.626 X J .s
Velocity of light in vacuum C 2.998 x lo8 mls

TABLE 0-7 TEMPERATURE SCALES

Rankine Fahrenheit Centigrade Kelvin

"R O F O C O K

Boiling polnt

Arnb~cntpolnt
Ice point

Absolutc rcro
Chapter 1

I Interactions Between
I Electrons and Fields

-eINTRODUCTION
1

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 dimen-
sions 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 interac-
tions.
Our primary purpose here is to provide the reader with a background for un-
derstanding the electron-field interactions in microwave devices that will be dis-
cussed in later chapters.

1 - j ELECTRON MOTION IN AN ELECTRIC FIELD

In describing fields and electron-field interactions, certain experimental laws of elec-

tricity 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 attrac-
tive or a repulsive force, depending on whether the charges are of opposite or like
sign. That is,

F = ---
Ql Q2
newtons
4 7 ~R
~20UR'2
Sec. 1.1 Electron Motion in an Electric Field

where Q = charge in coulombs

1
eO = 8.854 10-l2 = - X lop9F/m is the permittivity of free space
X
3 6 ~
R = range between the charges in meters
u = unit vector
It should be noted that since the MKS system is used throughout this text, a factor of
47r appears in the preceding equation.
The electric field intensity produced by the charges is defined as the force per
unit charge-that is,

If there are n charges, the electric field becomes

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 mo-
tion. So

where m = 9.109 X kg, mass of electron

a = acceleration in meters per second squared
v = velocity of electron in meters per second
e = 1.602 X 10-l9 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 ex-
periences a force - e E newtons. The differential equations of motion for an electron
in an electric field in rectangular coordinates are given by

where e / m = 1.759 x 10" C/kg is the ratio of charge to mass of electron and Ex,
E,, E, 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, 4, z ) are defined
as in Fig. 1 - 1 - 1 .
It can be seen that
Interactions Between Electrons and Fields Chap. 1

Figure 1-1-1 Cylindrical coordinates.

x = r cos 4
y = r sin 4
z =Z
and, conversely,
r = (x2 + y2)'I2

4 = tan-' (:) = sin-'

(x2
Y
+ y2)1'2 = cos-' (x2
X
+ Y2)1'2 (1 - 1 -7b)

A system of unit vectors, u,, ug, u,, in the directions of increasing r , 4, z, respec-
tively, is also shown in the same diagram. While u, is constant, u, and u4 are func-
tions of 4; that is,
u, = cos 4 u, + sin 4 u, (1 - 1 -8a)
u+ = -sin 4 u, + cos 4 u, (1 - 1 -8b)
Differentiation of Eqs. (1 - 1-8) with respect to 4 yields
du,
- = Ug
d4

The position vector p can be expressed in cylindrical coordinates in the form

p = ru, + zu, (1-1-9c)
Differentiation of Eq. (1 - 1 -9c) with respect to t once for velocity and twice for ac-
celeration yields
d p dr
v = - = -u, + r duL + --u
dz
=
dr d+ du, dz
--u + r - . - + -Uz
dt dt dt dt ' dt ' dt d+ dt
Sec. 1.1 Electron Motion in an Electric Field

Therefore the equations of motion for electrons in an electric field in cylindrical co-
ordinates are given by

where E,, E+, and E, are the components of E in cylindrical coordinates.

From Eq. (1-1-4) the work done by the field in carrying a unit positive charge
from point A to point B is

However, by definition, the potential V of point B with respect to point A is the

work done against the field in carrying a unit positive charge from A to B. That is,

V = -
I*" E.de (1-1-14)

Substitution of Eq. (1- 1- 14) in Eq. (1- 1- 13) and integration of the resultant yield

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,

If an electron starts from rest and is accelerated through a potential rise of V volts,
its final velocity is

Since d t is the increment of distance in the direction of an electric field E, the

change in potential dV over the distance dt? can be expressed as
IdvI = E d t (1-1-18)
10 Interactions Between Electrons and Fields Chap. 1

In vector notation it is
E = -VV
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.

1-2 ELECTRON MOTION IN A MAGNETIC FIELD

A charged particle in motion in a magnetic field of flux density B is experimentally

found to experience a force that is directly proportional to the charge Q , its velocity
v, the flux density B, and the sine of the angle between the vectors v and B. The
direction of the force is perpendicular to the plane of both v and B. Therefore the
force exerted on the charged particle by the magnetic field can be expressed in vec-
tor form as

Since the electron has negative charge, then

The motion equations of an electron in a magnetic field in rectangular coordinates

can be written

Since
v X B = (B,rv+ - B+v,)u, + (B,v, - B,u,)um + (B+v, - B,rv4)u, (1-2-4)
the equations of motion for electrons in magnetic field for cylindrical coordinates
can be given by
Sec. 1.2 Electron Motion in a Magnetic Field

0 - Y

"X
t Is-
"X
figure 1-2-1 ~ircvlarmotion of an
electron in a transverse magnetic field

Consider next an electron moving with a velocity of v, to enter a constant uni-

form magnetic field that is perpendicular to vx as shown in Fig. 1-2- 1. The velocity
of the electron is assumed to be

where u, is a unit vector in the x direction.

Since the force exerted on the electron by the magnetic field is normal to the
motion at every instant, no work is done on the electron and its velocity remains
constant. The magnetic field is assumed to be
B = B,u, (1 -2-7)
Then the magnetic force at the instant when the electron just enters the magnetic
field is given by

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 perpen-
dicular to the motion. At any point on the circle the outward centrifugal force is
equal to the pulling force. That is,

where R is the radius of the circle.

From Eq. (1-2-8) the radius of the path is given by

R = -mv meters
eB
The cyclotron angular frequency of the circular motion of the electron is
12 Interactions Between Electrons and Fields Chap. 1

The period for one complete revolution is expressed by

T = - 277
- = - 277m seconds
w eB
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 ve-
locity 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.

1-3 ELECTRON MOTION IN AN ELECTROMAGNETIC FIELD

If both electric and magnetic fields exist simultaneously, the motion of the electrons
depends on the orientation of the two fields. If the two fields are in the same or in
opposite directions, the magnetic field exerts no force on the electron, and the elec-
tron motion depends only on the electric field, which has been described in Section
1- 1. Linear-beam tubes (0-type devices) use a magnetic field whose axis coincides
with that of the electron beam to hold the beam together as it travels the length of
the tube. In these tubes the electrons receive the full potential energy of the electric
field but are not influenced by the magnetic field.
When the electric field E and the magnetic flux density B are at right angle to
each other, a magnetic force is exerted on the electron beam. This type of field is
called a crossedjield. In a crossed-field tube (M-type device), 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. The Lorentz force
acting on an electron because of the presence of both the electric field E and the
magnetic flux B is given by

The equations of motion for electrons in a crossed field are expressed in rectangular
coordinates and cylindrical coordinates, respectively, as
Sec. 1.3 Electron Motion in an Electromagnetic Field 13

where
d4
-= ",. = -eB is the cyclotron frequency
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. S o the equations of motion become simple and can easily be solved. An ex-
ample may show how to solve some of the preceding equations.

Example 1-3-1: Electron Motion in an Electromagnetic Field

The inner cylinder of radius a is the cathode and the outer shell with radius b is the
anode. A dc voltage Vo 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.

Anode
v = v,

Figure 1-3-1 Electron motion in an

electromagnetic field.

Solution

I. Write the equations of motion for electrons in cylindrical coordinates.

2. From (b)
Interactions Between Electrons and Fields Chap. 1

,.2- d4
dt
'
= - w,r2
2
+ constant
3. Application of the boundary conditions: At r = a ,

a2- d 4 = -1 w,a2 + constant

dt 2
1
3
dt
=0 constant = - -w,a2
2
Hence
d4 1
r2- = - w,(r2 - a 2 )
dt 2
4. The magnetic field does no work on the electrons:

5. For grazing the anode,

6 . The cutoff voltage is

This means that if Vo < Vk for a given Bo, the electrons will not reach the anode. Con-
versely, the cutoff magnetic field can be expressed in terms of Vo:

This implies that if Bo > Bk 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

GEWARTOWSKI,
Company, Princeton, N. J., 1965.
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. Prentice-
Hall, Inc., Englewood Cliffs, N.J., 1982.

PROBLEMS

1-1. 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 = t , 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.
1-2. A circular cavity is constructed of a center conductor and an outer conductor. The in-
ner center conductor has a radius of 1 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 emit-
ted 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 di-
rection 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 mWblm2, find the required supply
voltage so that the electrons just graze the anode and return to the cathode again.
Electromagnetic Plane
Waves

2-0 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 elec-
tric field, and a changing electric field will induce a magnetic field. Also, the in-
duced fields are not confined but ordinarily extend outward into space. The sinu-
soidal 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 prop-
agation 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
Sec. 2.1 Electric and Magnetic Wave Equations

2-1 ELECTRIC AND MAGNETIC WAVE EQUATIONS

The electric and magnetic wave equations can be basically derived from Maxwell's
equations, which in time domain are expressed as

where the vector operator V is expressed by

V = -u,
a a + -u,
+ -u, a (cartesian)
ax ay az

V =
a
-u,
a + -u,
+ -u4 a (cylindrical)
ar ra+ dz
a
V =-u, a
+-ue +- 1 a
-U+ (spherical) (2- 1-4c)
ar rae rsinea+
It should be noted that boldface roman letters indicate vector quantities or complex
quantities. The units of these field variables are

E = electricjeld intensity in volts per meter

H = magneticjeld intensity in amperes per meter
D = electric flux density in coulombs per square meter
B = magneticjux density in webers per square meter or in tesla
(1 tesla = 1 weber/m2 = lo4 gausses = 3 X ESU)
J = electric current density in amperes per square meter
p, = electric charge density in coulombs per cubic meter

The electric current density includes two components-that is,

J = Jc + JO (2- 1-5)
where J, = (TEis called the conduction current density
JO= the impressed current density, which is independent of the field
The current density source Jo may really be a current prescribed by an external
agency, but more often it is simply one which we know from measurements before
we start to find the fields. Alternatively, Jo is often the current on a surface or in a
region [I]. In most cases, however, the current density source JOmay not exist.
18 Electromagnetic Plane Waves Chap. 2

In addition to Maxwell's four equations, the characteristics of the medium in

which the fields exist are needed to specify the flux in terms of the fields in a specific
medium. These constitutive relationships are

where E = dielectric permittivity or capacitivity of the medium in farads per meter

er = relative dielectric constant (dimensionless)
E,, = 8.854 X lo-'' = 1 / ( 3 6 ~ )X F/m is the dielectric permittivity of
vacuum or free space
p = magnetic permeability or inductivity of the medium in henrys per meter
p, = the relative permeability or relative inductivity (dimensionless)
p0 = 41r X H/m is the permeability of vacuum or free space
u = conductivity of the medium in mhos per meter
If a sinusoidal time function in the form of ej"' is assumed, dldt can be re-
placed by jw. Then Maxwell's equations in frequency domain are given by

Taking the curl of Eq. (2- 1-1 1) on both sides yields

Substitution of Eq. (2- 1 - 12) for the right-hand side of Eq. (2- 1 - 15) gives

The vector identity for the curl of the curl of a vector quantity E is expressed as
V X V X E = -V2E + V(V.E) (2-1-17)
In free space the space-charge density is zero, and in a perfect conductor time-
varying or static fields do not exist. So
V . D = p, = 0 (2-1-18)
V-E=0 (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
Sec. 2.2 Poynting Theorem

where y = %p ( a +
j w ~ )= a + jp is called the intrinsic propagation
constant of a medium
a = attenuation constant in nepers per meter
p = phase constant in radians per meter
Similarly, the magnetic wave equation is given by
V2H = y2H
It should be noted that the "double del" or "del squared is a scalar operator-that
is,
V . V = V2 (2- 1-22)
which is a second-order operator in three different coordinate systems. In rectangu-
lar (cartesian) coordinates,

In cylindrical (circular) coordinates,

In spherical coordinates,

Also, the solutions of Eqs. (2- 1- 1) and (2- 1-2) solved simultaneously yield the elec-
tric and magnetic wave equations in the time domain as

2-2 POYNTING 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 Electromagnetic Plane Waves 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 = i (E X H*) (2-2-2)
Maxwell's equations in frequency domain for the electric and magnetic fields are
V X E = -jwpH (2-2-3)

Dot multiplication of Eq. (2-2-3) by H* and of the conjugate of Eq. (2-2-4) by E

yields

(V x H*) . E = (J* - jwcE*) - E (2-2-6)

Then subtraction of Eq. (2-2-5) from Eq. (2-2-6) results in
E . (V x H*) - H * . (V X E) = E . J* - j w ( ~ ( E 1 ~~-l . 1 ~ 1 , ) (2-2-7)
I
where E E* is replaced by E l 2 and H . H* by H 1,. I
The left-hand side of Eq. (2-2-7) is equal to -V . (E X H*) by the vector
identity. So
V - (E x H*) = -E . J* + j w ( ~E( l2 - pl H 1,) (2-2-8)
Substituting Eqs. (2-1-5) and (2-2-2) into Eq. (2-2-8), we have
-!E. J; = i u ~E*. + jw (;pH - H* - icE E*) +V P (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

where E l2 = u (I E I*) is the time-average dissipated power

i p ~. H* ,,
. = L p ( (H = W , is the time-average magnetic stored energy
E* = l c(1E 1,) = we is the time-average electric stored energy
1
- 2E - J,* = the complex power impressed by the source JOinto the field
Equation (2-2-10) is well known as the complex Poynting theorem or the Poynting
theorem in frequency domain.
Furthermore, let

p.~n = - i ( E - J8)do be the total complex power suplied by a source

within a region
(Pd) = 4wl E 12dv be the time-average power dissipated as heat inside
ti the region
Sec. 2.3 Uniform Plane Waves and Reflection 21

(Wm - We)=

P,,=
1f
r
(w, - w,)dz; be the difference between time-average magnetic

P ds
and electric energies stored within the region
be the complex power transmitted from the region

The complex Poynting theorem shown in Eq. (2-2- 10) can be simplified to
Pin = (Pd)+ j2w [(W,- K)] + Pt, (2-2- I I)
This theorem states that the total complex power fed into a volume is equal to the
algebraic sum of the active power dissipated as heat, plus the reactive power propor-
tional to the difference between time-average magnetic and electric energies stored
in the volume, plus the complex power transmitted across the surface enclosed by
the volume.

2-3UNIFORM PLANE WAVES AND REFLECTION

1 Uniform Plane Waves

2-3-

A plane wave is a wave whose phase is constant over a set of planes. A uniform
plane wave is a wave whose magnitude and phase are both constant. A spherical
wave in free space is a uniform plane wave as observed at a far distance. Its equi-
phase surfaces are concentric spheres, expanding as the wave travels outward from
the source, and its magnitude is constant.
Electromagnetic waves in free space are typical uniform plane waves. The
electric and magnetic fields are mutually perpendicular to each other and to the di-
rection of propagation of the waves. The phases of the two fields are always in time
phase and their magnitudes are always constant. The stored energies are equally di-
vided between the two fields, and the energy flow is transmitted by the two fields in
the direction of propagation. Thus a uniform plane wave is a transverse electromag-
netic wave or a TEM wave.
A nonuniform plane wave is a wave whose amplitude (not phase) may vary
within a plane normal to the direction of propagation. Consequently, the electric and
magnetic fields are no longer in time phase.
Since a uniform plane wave of electric or magnetic field has no variation of in-
tensity in a plane normal to the direction of propagation of the wave, then

if the direction of propagation is assumed in the positive z direction.

With the preceding assumptions and lossless dielectric-that is, a = 0-the
wave equations (2- 1-25) and (2- 1-26) in time domain for the electric and magnetic
intensities in free space for rectangular coordinates reduce to
22 Electromagnetic Plane Waves Chap. 2

in which the electric intensity is arbitrarily chosen in the x direction and the mag-
netic intensity in the y direction. With no loss in generality, it can be assumed that
the electric intensity is given by

The magnetic intensity can be obtained by inserting Eq. (2-3-3) into the curl equa-
tion
V X E = -jopoH (2-3-4)
For the assumed conditions, the curl equation reduces to

Differentiation of Eq. (2-3-3) with respect to z and substitution of the result in Eq.
(2-3-5) yield
7

1
- is accounted for in the derivation
1
up = -= 3 X 10' mls is phase velocity
G
= c, the velocity of light in vacuum
a=O
p = WG
is phase constant
The ratio of electric to magnetic intensities is given by
7

It is called the intrinsic impedance of free space.

Figure 2-3-1 shows uniform electric and magnetic plane waves in rectangular
coordinates. In general, for a uniform plane wave propagating in a lossless dielectric
medium (a = O), the characteristics of wave propagation would become
Sec. 2.3 Uniform Plane Waves and Reflection

Figure 2-3-1 Uniform plane waves

traveling in the z direction.

Boundary Conditions
2-3-2

Since Maxwell's equations are in the form of differential rather than algebraic equa-
tions, boundary conditions must be applied to a given problem if a specific solution
is required.
There are four basic rules for boundary conditions at the surface between two
different materials:

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,
Ampkre'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
3; M;t2
24

Medium 1

Dn 1
(a)
Medium 2

Dn 2
(b
Electromagnetic Plane Waves Chap. 2

Figure 2-3-2 Boundary conditions. (a)

Electric intensity. (b) Magnetic intensity.

So the boundary equations are

Erl = Er2
D,I = Dn2+ PS

I Hr2 + Js
H ~=
B ~ =I Bnz
If medium 2 is a perfect conductor (u = m, E, = 1 , p, = 1) and medium 1 is a per-
fect dielectric (vacuum or free space, u = 0, €0, po), then

2-3-3Uniform Plane-Wave Reflection

Normal-incidence reflection. The simplest reflection problem is that of a

uniform plane wave normally incident on a plane boundary between two dielectric
media with no surface-charge density and surface-current density. This situation is
shown in Fig. 2-3-3.
In medium 1 the fields are the sum of an incident wave plus a reflected wave,
respectively.
ELI) = Eo(e-JP~'+ rejPlz) (2-3- 11)

where pl = w -
v p ~ e ~
= J: - $ - intrinsic wave impedance of medium
I- = reflection coefficient
Sec. 2.3 Uniform Plane Waves and Reflection

Medium 1 Medium 2
t
Incident wave
p2 3 €2

k
p1 1 €1 Transmitted wave

Reflected wave

r - .. + Figure 2-3-3 Uniform plane-wave

reflection.

In medium 2 there are transmitted waves:

EL21 = Eo Te-jkz

where P2 = o v -
p 2 ~ 2

')'= d€,
fi L!@-
& = = intrinsic wave impedance of medium 2

T = transmission coefficient
For continuity of wave impedance at the boundary, the wave impedance is

Hence the reflection coefficient is given by

r=- 772 - 771 (2-3-16)

772 + 771
From the boundary condition the tangential components of electric field intensity
are continuous across the interface. Then

=~ ~ + (r) 1= E:~' (2-3-17)

Hence the transmission coefficient is expressed as

2772
T=l+r=- (2-3-18)
772 + 771

If medium 1 is lossless dielectric (that is, a = O ) , the standing-wave ratio is defined

as
26 Electromagnetic Plane Waves Chap. 2

The power density transmitted across the boundary is

Then
Ptr = pin=(] - r2)
where pi,, is the incident power density.
The incident power density minus the transmitted power density would yield
the reflected power density as
Pref = plncr2 (2-3-22)

Oblique-incidence reflection.
E Is in the Plane of Incidence. The plane of incidence is defined by the di-
rection 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 im-
pinging 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 po-
larization refer to the phenomenon of waves from horizontal and vertical antennas,
respectively, producing the corresponding orientations of wave polarization when

Normal
Incident wave Retlectcd wave
ox = P, sin 0,

rEx

Boundary
M2
"2 I

z Transmitted wave

Figure 2-3-4 Reflection and transmission of oblique incidence

Sec. 2.3 Uniform Plane Waves and Reflection 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 propaga-
tion. The polarization of a wave is an extremely useful concept for computing elec-
tromagnetic power flow. For example, a Poynting vector indicates that the power-
flow 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 H on the boundary. Thus
pi sin 6, = Pr sin 6, (2-3-23a)
pi sin 6, = p, sin 6, (2-3-23b)
From Eq. (2-3-23a), since Pi = P, = PI, the angle of reflection is equal to the an-
gle of incidence. This is

From Eq. (2-3-23b)

I

sin 6; p, vl
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, PI = p2 = p,,. If medium
2 is free space and medium 1 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

The components of magnetic intensity H are

Hx = 0

The wave impedance in the z direction is given by

Ex
Z, = - = q cos 6 (2-3-32)
HY
It should be noted that the subscripts of q and 0 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 Electromagnetic Plane Waves Chap. 2

the normal components at the boundary z = 0. In medium 1

ELI) = Eo cos o i [ e - j P ~ ' C O ~ e i + r e j P l z ~ O S @ 1r (2-3-33)

Eil'
z,= - = ql cos oi-
i+r
~ $ 1 ) l - r
The impedance must be equal to the z-directed wave impedance in region 2 at the
boundary. Substitution of Z, = 7 2 cos 0, in Eq. (2-3-35) yields
cos 0, - cos oi
r= 772

772 COS Or +
771

771 COS Oi

Then the transmission coefficient is given by

2772 COS or
T =
772 COS Or + 771 cos Oi

The preceding two equations are known as Fresnel's formulas for E in the plane of
incidence.
H Is in the Plane of Incidence. If H is in the plane of incidence, the com-
ponents of H are

The components of electric intensity E normal to the plane of incidence are

Ex = 0 (2-3-41)
EY= -77 I H ~ ~ - J P ~ (ei+zcos
X S ~ 0,)
~ (2-3-42)

E, = 0 (2-3-43)
The wave impedance in the z direction is given by

= - AE= - 77 = 77 sec
Hx cos 0
It should be noted that the subscripts of 7) and 0 have been dropped for the same rea-
son stated previously.
Fresnel's formulas for H in the plane of incidence are
sec Or - 71sec Oi
r= 772

772 sec Ot + q l sec Oi

2v2 sec 8,
T =
7 7 ~sec 8, + 771 sec Oi
Sec. 2.4 Plane-Wave Propagation in Free Space and Lossless Dielectric

2-4 PLANE-WAVE PROPAGATION IN FREE SPACE

AND LOSSLESS DIELECTRIC

2-4-1 Plane- Wave Propagation In Free Space

The electromagnetic wave being propagated in free space near the surface of the
earth is divided into two parts: the ground wave and the sky wave or ionosphere
wave. The ground wave is further divided into a direct wave, an earth-reflected
wave, and a surface wave. Figure 2-4- 1 shows the wave components of electromag-
netic wave from a nondirectional antenna to a receiving station.

sky wave + \
/ Direct wave
+
! Reflected wave
Transmitting Receiving
antenna antenna

Surface wave
,,-
Figure 2-4-1 Wave components near
Earth the surface of the earth.

The ionosphere is that region of the earth's atmosphere in which the con-
stituent 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 F layers 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 1012electrons per cubic meter at an altitude
between 90 and 1000 km. The E and F layers 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
F,, = 9 K Hz (2-4-1)
This means that a microwave of frequency F,, will be reflected back to the earth if
the electron density is equal to or higher than the required maximum electron den-
sity N,,, (electrons per cubic meter).
The sky wave reaches the receiving station after reflection from the iono-
sphere. 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 Electromagnetic Plane Waves Chap. 2

around the surface of the earth or guided by the ground-air interface. This compo-
nent 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 re-
ceiving station. It is the only wave considered in this book. The term free space will
be used to denote vacuum or any other medium having essentially the same charac-
teristics 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

where p, = transmitting power in watts

g, = transmitting antenna gain (numerical)
R = distance between antenna and field point in meters
The power received by the receiving antenna is given as

pr = p . ~ . = (')('g,) watts

A
where A, = -g, = effective antenna aperture in square meters
4lT
A
-- - A, = isotropic antenna aperture in square meters
4lT
gr = receiving antenna gain (numerical)
Figure 2-4-2 shows the relationships of electromagnetic energy transmission in free

--
space between two antennas.

Transmitting R Receiving

Figure 2-4-2 Electromagnetic energy

transmission between two antennas.

If the received power is expressed in terms of decibels, Eq. (2-4-3) becomes

P, = P. + G, + G, - 20 log (F) dBW (2-4-4)

where PI is in dBW, G, and G, are in decibels (dB). The term 20 log (4rR/A) is well
known as the free-space attenuation in decibels. It can easily be found from the stan-
Sec. 2.4 Plane-Wave Propagation in Free Space and Lossless Dielectric 31

dard nomograph shown in Fig. 2-4-3. For example, if the wavelength of a signal is
0 . 0 3 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 (4.rrR2)in Eq. (2-4-3) simply accounts for the fact that the power
density is inversely proportionally decreasing with the squared distance when the en-
ergy spreads out over free space. The factor (A2/47r)is the isotropic aperture of a re-
ceiving antenna. It does not imply that a higher-frequency wave decreases in magni-
tude 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.

Figure 2-4-3 Nomograph of free-space attentuation.

32 Electromagnetic Plane Waves Chap. 2

2-4-2 Plane- Wave Propagation in Lossless Dielectric

The lossless dielectric, which is often called the good or perfect dielectric, is charac-
terized by a = 0 . Hence the intrinsic impedance for a lossless dielectric can be ex-
pressed in terms of air. This is
7 7

The attenuation constant a is zero, and the phase constant P is given by

The phase velocity is expressed by

2-5 PLANE-WAVE PROPAGATION IN LOSSY MEDIA

The lossy media are characterized by u f 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 con-
ductivity. 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:
V2E= jwp(u + jm)E (2-5-1 )

For one dimension in the positive z direction, they become

The complex-frequency solutions would be given by

Ex = E d a z cos (wt - Pz) (2-5-4)

H, = 3e-az cos (WI - pz) (2-5-5)

rl
where y = djwp(cr + j w ~ =) a + jp
I
Sec. 2.5 Plane-Wave Propagation in Lossy Media

2-5-1 Plane Wave in Good Conductor

A good conductor is defined as one having a very high conductivity; consequently,

the conduction current is much larger than the displacement current. The energy
transmitted by the wave traveling through the medium will decrease continuously as
the wave propagates because ohmic losses are present. Expressed mathematically, a
good conductor requires the criterion

The propagation constant y is expressed as

=j U G J- j ~ for -
u
WE
*1

= (1 + j) u r f p u
Hence

The exponential factor e-"' of the traveling wave becomes e-' = 0.368 when

This distance is called the skin depth and is denoted by

Interestingly, at microwave frequencies the skin depth is extremely short and a piece
of glass with an evaporated silver coating 5.40-pm 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

= ,/u
JWP
+ WE
= $%
u
for u + WE

in which R , = is known as the skin effect and the magnitude of the con-
34 Electromagnetic Plane Waves Chap. 2

TABLE 2-5-1 TABLE OF CONDUCTIVITIES

Conductivity u
Substance TYP (mhoslm)
Quartz, fused insulator 10-l7 approx.
Ceresin, wax insulator 10-l7 approx.
Sulfur insulator 10-1" approx .
Mica insulator approx.
Paraffin insulator approx.
Rubber, hard insulator IO-l5 approx.
Glass insulator 10-lZ approx.
Bakelite insulator 10-9 approx.
Distilled water insulator approx.
Seawater conductor 4 approx.
Tellurium conductor 5 X lo2 approx.
Carbon conductor 3 X lo4 approx.
- -

Graphite conductor 1O5 approx .

Cast iron conductor lo6 approx.
Mercury conductor 1O6
Nichrome conductor 1O6
Constantan conductor 2 x lo6
Silicon steel conductor 2 X106
German silver conductor 3 x 106
Lead conductor 5 X106
Tin conductor 9 x lo"
Phosphor bronze conductor I 07
Brass conductor 1.1 x 107
Zinc conductor 1.7 x 107
Tungsten conductor 1.8 x 107
Duralumin conductor 3 xio7
Aluminum, hard-drawn conductor 3.5 x 107
Gold conductor 4.1 x 107
Copper conductor 5.8 x 107
Silver conductor 6.1 x 107

ductor surface resistance. The average power density for a good conductor is given
by
p =~(H(~R, (2-5- 12)
and the phase velocity within a good conductor is

The reflectivity and transmittance of a good conductor in vertical and horizontal po-
larizations are usually measured in terms of the grazing angle. The grazing angle 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 rC, = 90' - 8,;
then sin cC, = cos O,, sin 8, = cos I(,, sinZ8, + cos2 8, = I , and v, sin 8, = v2 sin 8,.
The vertical reflectivity of a good conductor for the tangential components of elec-
tric intensity as shown in Eq. (2-3-36) is simplified to
Sec. 2.5 Plane-Wave Propagation in Lossy Media 35

- (v2/v1cos - 7 1 sin $
r, = 772[1

772[1 - (v2/vI cos $)2]1/2+ q 1 sin $

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 direc-
tions. Therefore the reflectivity of a good conductor in vertical polarization is
sin $ - q2[1 - (v2/v1cos $)2]'/2
r, = 771
7 1sin $ + q2[1 - (vz/vl cos $)2]1/2
(2-5- I ~ C )

Similarly, the vertical transmittance of a good conductor for electric fields as shown
in Eq. (2-3-37) is given by
2772[1 - (v2lv1 cos l,b)2]1/2
T, =
772[1 - (v2/v1 cos $)2]1/2 + q1sin $
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 $
as
q2 sin $[I - (v2/vI cos J / ) ~ ] - ' / -
~ 771
I'h =
q2 sin $[1 - (v2/01 cos $)2]-1/2+ 77 1
Similarly, the transmittance of a good conductor for electric fields in horizontal po-
larization as shown in Eq. (2-3-46) can be expressed as
2772[1 - (v2/v, cos $)21-1/2
Th =
+ 77 1
q2 sin $ [ I - (v2lvl cos l,b)2]-1/2
In Fig. 2-3-4 it is assumed that medium 1 is free space or air and that medium 2 is
copper; then
I

The conductivity u of copper is 5.8 X lo7 mhoslm and its relative permeability is
unity. The magnitudes of reflectivity of copper for vertical and horizontal polariza-
tions are computed by Eqs. (2-5-13c) and (2-5-13e) against the grazing angle I,!Iof 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.

2-5-2 Plane Wave in Poor Conductor

Some conducting materials with low conductivity normally cannot be considered

either good conductors or good dielectrics. Seawater is a good example. It has a con-
ductivity of 4 mhoslm and a dielectric constant of 20. At some low frequencies the
36 Electromagnetic Plane Waves 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 con-
ductor are given by

2-5-3 Plane Wave in Lossy Dielectric

All dielectric materials have some conductivity, but the conductivity is very small
(u << W E ) . When the conductivity cannot be neglected, the electric and magnetic
fields in the dielectric are no longer in time phase. This fact can be seen from the
intrinsic impedance of the dielectric as -

The term a/(@€) is referred to as the loss tangent and is defined by

u
tan 8 = - (2-5-17)
WE

This relationship is the result of displacement current density leading conduction

current density by 90°, just as the current flowing through a capacitor leads the cur-
rent through a resistor in parallel with it by 90" in an ordinary electric circuit. This
phase relationship is shown in Fig. 2-5- 1 .

If the loss tangent is very small-that is, U / ( W E ) 4 1-the propagation con-

stant and intrinsic impedance can be calculated approximately by a binomial expan-
sion. Since

(2-5-18)
WE

then
Electromagnetic Plane Waves Chap. 2

E
where E, =-
€0

(CI = 90' - 8, is called the pseudo-Brewster angle

Equation (2-5-25) is applicable only for the tangential components of incident
and reflected fields that are in the same directions as shown in Fig. 2-3-4. For total
reflection (r, = 1 ) we set 8, = 90" in Eq. (2-3-25), and the incident angle is given

The angle specified by Eq. (2-5-24c) is called the critical incident angle for total
reflection. A wave incident on the boundary at an angle equal to or greater than the
critical angle will be totally reflected. There is a real critical angle only if
P I E ,> p 2 e 2 ,or, in the nonmagnetic material, if E I > E Z . Hence the total reflection
occurs only if the wave propagates from a dense medium into a less dense medium.
This is because the value of sin 6,. must be equal to or less than unity.
For vertical polarization, the normal components of electric fields are usually
used to determine the reflection coefficient. Therefore the reflectivity of a lossy
dielectric in vertical polarization is given by
- jx) Sin (1, - d ( E , - jx) - c0s2 (1,
rt = (Er
(2-5-26)
jx) sin (1, + d ( E r - jx) - cos2 I)
(E, -
Similarly, the vertical transmittance of a lossy dielectric for electric fields as shown
in Eq. ( 2 - 3 - 3 7 ) is
~v'(E, - jx) - cosZ (CI
T, = (2-5-27)
d ( E r - jx) - COS' (1, + (E, - jx) sin (CI

Horizontal polarization. The reflectivity of a lossy dielectric for electric

fields in horizontal polarization as shown in Eq. (2-3-45) becomes

r h= sin (1, - d ( ~ ,
- jx) - cos2 (CI
(2-5-28)
sin (CI + V(E,
- jx) - cos2 (1,
Similarly, the transmittance of a lossy dielectric for electric fields in horizontal po-
larization as shown in Eq. (2-3-46) is expressed as
2 sin (CI
Th =
sin (CI + V ( e r - jx) - cos2 (CI
The reflections of electromagnetic waves by such lossy dielectric materials as sea-
water, dry sand, and concrete cement are often of concern to many electronics engi-
Sec. 2.5 Plane-Wave Propagation in Lossy Media 39

neers. The conductivities (T 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 t,b of 0 to 90' at a frequency range of 0.1 to 40 GHz [2].

TABLE 2-52 CONDUCTIVITIES AND RELATIVE DIELECTRIC

CONSTANTS OF SEAWATER, DRY SAND, AND CEMENT

Seawater Dry sand Concrete cement

0 010 1 I 1 1 1 1 I I I I
10 20 30 40 50 60 70 80 90
Grazing angle J, In degrees

Figure 2-5-2 Magnitude of the reflectivity in vertical polarization versus

grazing angle for seawater.
Electromagnetic Plane Waves Chap. 2

GHz
1 I 1 1 I I 1 I 1
0.5
0 10 20 30 40 50 60 70 80 90
Grazing angle JI in degrees

Figure 2-5-3 Magnitude of the reflectivity in horizontal polarization versus

grazing angle for seawater.

Grazing angle J, in degrees

Figure 2-5-4 Magnitude of the reflectivity in vertical and horizontal

polarizations versus grazing angle for dry sand.
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 41

Grazing angle JI in degrees

Figure 2-5-5 Magnitude of the reflectivity in vertical and horizontal

polarizations versus grazing angle for concrete cement.

2-6 PLANE-WAVE PROPAGATION IN METALLIC-FILM

COATING ON PLASTIC SUBSTRATE*

In certain engineering applications, it is often desirable to use a metallic-film-coated

glass to attenuate optimum electromagnetic radiation at microwave frequencies and
also to transmit as much light intensity as possible at visible-light frequencies. Gen-
erally the coated metallic film should have a high melting point, a high electrical
conductivity, high adhesion to glass, high resistance to oxidation, and insensitivity
to light and water, as well as the capability of dissipating some power for deicing,
defogging, or maintaining certain temperature levels. The metallic-film coatings on
a plastic substrate are used in such applications as windshields on airplanes or auto-
mobiles, medical equipment in hospitals, and on dome windows of space vehicles,
missiles, and other military devices.

*Copyright O 1975 by the Institute of Electrical and Electronic Engineers, Inc. Light Transmit-
tance 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 Electromagnetic Plane Waves Chap. 2

2-6-1 Surface Resistances of Metallic Films

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 mean-
free-path, then the role of electron scattering becomes dominant. Fuchs [4] and
Sondheimer [5] considered the general form of the solution of the Boltzmann equa-
tion for the case of a conducting film and found the film conductivity ufin terms of
the bulk conductivity u ,the film thickness t, and the electron mean-free-path p:

The surface resistance of conducting films is generally quoted in units of ohms per
square because in the equation for resistance
specific resistivity X length -
--p4 (2-6-2)
R =
thickness X width tw
when units of length e and width w are chosen to have equal magnitude (that is, re-
sulting in a square), the resistance R in ohms per square is independent of the di-
mensions of the square and equals

According to the Fuchs-Sondheimer theory, the surface resistance of a metallic film

is decreased as the thickness of the film is increased.

2-6-2 Optical Constants of Plastic Substrates

and Metallic Films

The optical properties of materials are usually characterized by two constants, the
refractive index n and the extinction index k. The refractive index is defined as the

TABLE 2.6-1 SUBSTRATE MATERIALS

- -

Substrate material Refractive index n

Corning Vycor 1.458

Crystal quartz 1.540
Fused silica 1.458
Plexiglass 1.490
Polycyclohexyl methacrylate 1.504
Polyester glass 1.500
Polymethyl methacrylate 1.491
Zinc crown glass 1.508
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 43

ratio of the phase velocities of light in vacuum and in a medium. The extinction in-
dex 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 2.6-2 REFRACTIVE INDEX n AND EXTINCTION INDEX k OF THIN METALLIC

FILMS

Copper film Gold film Silver film

Wavelength
(A) n k n k n k
2000 1.427 1.215 1.13 1.23
2200 1.32 1.29
2300 1.38 1.31
2400 1.37 1.33
2500 1.39 1.34
2600 1.45 1.35
2700 1.51 1.33
2800 1.57 1.27
2900 1.60 1.17
3000 1.67 0.96
3100 1.54 0.54
3200 1.07 0.32
3300 0.30 0.55
3400 0.16 1.14
3500 0.12 1.35
3600 0.09 1.52
3700 0.06 1.70
3800
4000
4500 0.870 2.200 1.400 1.880
4920
5000 0.880 2.420 0.840 1.840
5460
5500 0.756 2.462 0.331 2.324
6000 0.186 2.980 0.200 2.897
6500 0.142 3.570 0.142 3.374
7000 0.150 4.049 0.131 3.842
7500 0.157 4.463 0.140 4.266
8000 0.170 4.840 0.149 4.654
8500 0.182 5.222 0.157 4.993
9000 0.190 5.569 0.166 5.335
9500 0.197 5.900 0.174 5.691
10,000 0.197 6.272 0.179 6.044

Source: Adapted from the American Institute of Physics Handbook by the American Institute of
Physics. Copyright 0 1972 by McGraw-Hill, Inc. (Used with permission of McGraw-Hill Book
Company .)
44 Electromagnetic Plane Waves Chap. 2

2-64 Microwave Radiation Attenuation of

Metallic-Film Coating on Plastic Substrate

A conductor of high conductivity and low permeability has low intrinsic impedance.
When a radio wave propagates from a medium of high intrinsic impedance into a
medium of low intrinsic impedance, the reflection coefficient is high. From electro-
magnetic plane wave theory in the far field, high attenuation occurs in a medium
made of material having high conductivity and low permeability. Good conductors,
such as gold, silver, and copper, have high conductivity and are often used as the
material for attenuating electromagnetic energy. Microwave radiation attenuation by
a metallic-film coating on substrate consists of three parts [7]:
Attenuation = A +R +C dB (2-6-4)
where A = absorption or attenuation loss in decibels inside the metallic-film
coating while the substrate is assumed to be nonabsorbing plastic glass
R = reflection loss in decibels from the multiple boundaries of a metallic-film
coating on substrate
C = correction term in decibels required to account for multiple internal
reflections when the absorption loss A is much less than 10 dB for
electrically thin film
Figure 2-6-1 shows the absorption and reflection of a metallic-film coating on a
plastic substrate.
Metallic
film
Air ---I t

a
RF energy
T;+[
Substrate

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 con-

stant y for a uniform plane wave in a good conducting material is given in Eq.
(2-5-7) as
y =a + jp = (1 + j)vnfpai for a- B o~ (2-6-5)
If the plastic substrate is assumed to be a nonabsorbing material, the absorption loss
A of the metallic-film coating on a substrate is related only to the thickness t of the
coated film and the attenuation a as shown:
A = 20 logloem'= 20(at) loglo e = 20(0.4343)(at)
= 8.686tf- dB (2-6-6)
where t = thickness of the film coating in meters
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 45

p = permeability of the film in henrys per meter

f = frequency in hertz
u-= conductivity of the coated film in mhos per meter

Since the thickness of the coated film is very thin-for example, 100 angstroms at
most (I% = lo-'' 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 21 - 20 log 21qg1 - 2010g 2 / 1 1

1% + PI IV + 7781 1% + 77.1
= 20 log IT^ + w I I P +
~a1Iqg+ %I dB
8177rlIqg llqa 1
where rlf = intrinsic impedance of the coated metallic film
qg = intrinsic impedance of the glass substrate
= intrinsic impedance of air or free space = 377 Cl

The intrinsic impedance of a metallic film is given by Eq. (2-5-1 1) as

I . 7

and the intrinsic impedance of a glass substrate is expressed in Eq. (2-5-16) as

qa -
r l - - - 194 n
g 2 . 377 for a, -G weq (2-6-9)
V5-z
where ug= about lo-'* mholm is the conductivity of the glass substrate
E, = 4.77 X lo-'' Flm is the permittivity of the glass substrate
E, = 3.78 is the relative permittivity of the glass substrate

Substituting the values of the intrinsic impedances w , q,, and qa in Eq. (2-6-7)
yields the reflection loss as [12]

- &1
- -

R 20 log 128.31 = 88 + 10 log (7) dB (2-6- 10)

Correction term C. For very electrically thin film, the value of the absorp-
tion loss A is much less than 10 dB and the correction term is given by [8]
C = 20 log 1 1 - p 10-*~'~(cos
0 - j sin 0) 1 (2-6- 11)

where p = -
:( + ):;1 1
for qa +

Over the frequency range of 100 MHz to 40 GHz, the angle 0 is much smaller
than l o so that cos 0 = 1 and sin 0 .= 0. Thus the correction term of Eq. (2-6-1 1)
46 Electromagnetic Plane Waves Chap. 2

can be simplified to
C = 2 0 log [3.54t*f] = -48 + 20 log [t%f] dB (2-6-12)
Finally, the total microwave radiation attenuation by a metallic-film coating on a
glass substrate, defined in Eq. (2-6-4) in the far field, becomes
Attenuation = 40 - 20 log (R,) 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 [ 1 2 ] .

2-6-4 Light Transmittance of Metallic-Film Coating

on Plastic Substrate

The complex refractive index of an optical material is given by [ 6 ] as

It is assumed that light in air is normally incident on a thin absorbing film Nl of

thickness t l and that it is transmitted through an absorbing substrate of complex re-
fractive index N2 and then emerges into air. The incidence and the emergence media
are dielectrics of refractive index no. The reflection loss between the substrate and
the air is small and, for convenience, is taken as zero. Figure 2-6-2 shows light
transmittance, reflection, and absorption through a thin absorbing metallic film and
a plastic substrate.
R LIGHT
o n

N2=n2-jk2 SUBSTRATE

Figure 2-6-2 Light transmittance,

reflection, and absorption through a thin
metallic film coated on plastic substrate.

Using the multireflection and transmission theory, the reflection loss is ex-
pressed by
a l e a + a2ePa+ a3 cos v + a4 sin v
R =
+
b l e a b2e-" + b3 cos v + b4 sin v
where a l = [(no - n J 2 + k:][(nl + n2)2 + ( k , + k J 2 ]
a2 = [(no + n d 2 + k?][(nl - n d 2 + (kl - k d 2 ]
a3 = 2{[ng - (n: + k:)][(n: + k:) - (n: + k:)] + 4nokl(nlk2 - n2kl)}
a4 = 4{[ni - (n: + k:)](nlk2 - n2kl) - nokl[(n: + k:) - (n: k:)]} +
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 47

C .
A - - is the wavelength in a vacuum
"f
c = 3 X 10' m/s is the velocity of light in a vacuum; f is the frequency in
hertz
y =-
4 r n lt l
Ao
6 1 = [(no + n d 2 + k:l[(nl + n J 2 + ( k ~+ k2)2]
b2 = [(no - n J 2 + k:l[(nl - n2)' + ( k ~- kJ2]
b 3 = 2{[n6 - (n: + k:)][(n: + k:) - (n: + k i ) ] - 4nokl(nl k2 - n2kl)}
b4 = 4{[na - (n: + k:)](nlk2- n2kl) + nokl[(ni + k:) - (n: + k:)]}
Transmittance T is given by [ 6 ] as

Absorption loss A is given by

and the total attenuation loss L is

When the concave surface of a plastic dome is uniformly coated with an elec-
tromagnetic interference shield of metallic film, however, the light is normally inci-
dent on the plastic substrate N 2 , transmits through the thin metallic film N I , 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 N2 or on the absorbing film N I . Thus
the total attenuation loss is the same in both cases.

2-6-5 Plane Wave in Gold-Film Coating on Plastic

Glass

Metallic-film coatings on plastic glasses have many engineering applications [9]. A

gold film, for example, is coated on the concave surface of a plastic-glass dome so
that an optimum amount of microwave radiation is attenuated by the gold film while,
at the same time, a sufficient light intensity is transmitted through the gold film.

Surface resistance. At room temperature the properties of bulk gold are

Conductivity: cr =4.10 X lo7 mhoslm

Resistivity: p = 2.44 x lo-* a - m
Electron mean-free-path: p = 570 A

It is assumed that the thickness t of the gold film varies from 10 to 100 A. Its surface
48 Electromagnetic Plane Waves Chap. 2

TABLE 2-64 SURFACE RESISTANCE OF GOLD FlLM

---

Thickaess t Conductivity uf Resistivity pf Surface resistance R,

(A) (mholm x lo7) (0-m x (Rlsquare)

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 Rlsquare for a
thickness of about 90 to 45 A. The surface resistance is decreased as the thickness of
the gold film is increased.

0 a
ca u
a
$5330-----
w a
O W

w
U I I I
I I I
o* I 1 1 1 I 1 1
V)
20 30 40 50 60 70 80 90 100 Figure 2-6-3 Surface resistance of gold
THICKNESS OF GOLD FILM IN A film versus thickness of gold film.

Microwave radiation attenuation. Substituting the values of the surface

resistances for gold films in Eq. (2-6-13) yields the microwave radiation attenuation
in decibels by the gold-film coating on a plastic glass. Figure 2-6-4 shows graphi-
cally the microwave radiation attenuation versus the surface resistance of the gold-
film coating. For a coated gold film having a surface resistance of 12 Rtsquare, the
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 49

\
0--- - Hawthorne's data
30d\ Liao's data
0----

01 1 1 I L I I 1
5 10 20 30 40
Surface resistance R, in ohms per square

Figure 2-6-4 Microwave radiation attenuation versus surface resistance of gold

film.

microwave radiation attenuation is about 19 dB. The data agree with Hawthorne's
conclusion [ 10,121.

Light transmittance. For the visible-light region, the values of the refrac-
tive index n and the extinction index k of a gold-film coating on a plastic glass de-
posited 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 transmit-
tance 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 relation-
ship for the visible-light region is shown in Fig. 2-6-6.
Figure 2-6-7 illustrates the relationship of light transmittance versus wave-
length for a given surface resistance of gold film. If a power dissipation of
5 Wlsquare 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 fllsquare.
50 Electromagnetic Plane Waves Chap. 2

FlLM
THICKNESS
IN A

FlLM
THICKNESS
IN A
Figure 2-6-5 Light transmittance T and
light attenuation loss L versus wave-
2000 3000 4000 5000 7000 10000 length A with film thickness t as
WAVELENGTH A IN A parameter for gold film.

5 90 -
0

E
Y -
O
Z
a
t
k80-----
5
Z
4
a ,
I-
I-
x
2 70 0 I
10 20 30 40
SURFACE RESISTANCE OF GOLD F l L M I N OHMS PER SQUARE
LIGHT TRANSMITTANCE DECREASES, AS SURFACE
RESISTANCE DECREASES
FOR 80 PER CENT OF LIGHT TRANSMITTANCE. Figure 2-6-6 Light transmittance
SURFACE RESISTANCE IS ABOUT 12 OHMS PER SQUARE versus surface resistance of gold film

The power dissipation can be expressed as

in which the voltage applied to the film-coating terminations is 28 V. The optimum

condition occurs at 19 dB of microwave radiation attenuation and 80% of light trans-
mittance.
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 51

WAVELENGTH a IN A
L I G H T T R A N S M I T T A N C E INCREASES AS SURFACE RESISTANCE
INCREASES
L I G H T T R A N S M I T T A N C E OF 80 PER CENT OCCURS A T 1 2
O H M S PER SQUARE

Figure 2-6-7 Light transmittance versus wavelength with surface resistance R, as

parameter for gold film.

Example 2-6-5: Calculation of a Gold-Film Coating

A gold film of 80 A is coated on a plastic substrate with a refractive index of 1.50. De-
termine:

a. The gold-film surface resistance in ohms per square

b. The microwave attenuation in decibels
c. The light transmittance T for red light of 0.69 p m
d. The light reflection loss R at A = 0.69 p m
Solution
a. From Eq. (2-6-1) the gold-film conductivity is

From Eq. (2-6-3) the gold-film surface resistance is

Electromagnetic Plane Waves Chap. 2

R = - 1= 1
" tuf 80 X x 1.03 x lo7
= 12.12 Cllsquares
b. From Eq. (2-6- 13) the microwave attenuation is
Attenuation = 40 - 20 log (R,)
= 40 - 20 log (12.12)
= 18 dB
c. From Fig. 2-6-5 the light transmittance T is estimated to be 75%.
d. From the same figure the light reflection loss R is about 25%.

2-6-6 Plane Wave in Silver-Filmor Copper-Film

Coating on Plastic Substrate*

Silver-film or copper-film coating on a plastic substrate has many uses in engineering

[ll]. The surface resistance, microwave radiation attenuation, light transmittance,
and optimum condition of both silver-film coating and copper-film coating can be
described in the same way as for fold-film coating.

Surface resistance. At room temperature the properties of bulk silver and

bulk copper are

Silver
Conductivity: a = 6.170 X lo7 mhoslm
Resistivity: p = 1.620 X lo-' R-m
Electron mean-free-path: p = 570 A

Copper
Conductivity: a = 5.800 x lo7 mhoslm
Resistivity: p = 1.724 X R-m
Electron mean-free-path: p = 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, respec-
tively, in ohms per square versus the thickness of the silver and copper films from 10
to I00 A.

*Copyright O 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
[ l I]; reprinted from IEEE Trans. on Electromagnetic Compatibility. EMC-18, No. 4 , 148- 153,
November 1976.
Sec. 2.6 Plane-Wave Propagation in Metallic-Film Coating on Plastic Substrate 53

TABLE 2-6-4 SURFACE RESISTANCE R, OF SILVER FlLM

Thickness r Conductivity uf Resistivity pf Surface resistance R,

(A) (mholm x 10') (R-m x lo7) (Rlsquare)

TABLE 2-6.5 SURFACE RESISTANCE R, OF COPPER FlLM

Thickness t Conductivity u, Resistivity pf Surface resistance R,

(A) (mhotm x lo7) (R-m x (Rlsquare)

0.52 5.20
0.54 6.00
0.58 7.25
0.62 8.86
0.68 11.33
0.75 15.00
0.86 21.50
1.05 35.00
1.37 68.50
2.31 231.00

"
10 20 30 40 50 60 70 80 90 100
Film thickness t in A

Figure 2-6-8 Surface resistance of silver and copper film versus thickness of film.
54 Electromagnetic Plane Waves Chap. 2

RF radiation attenuation. Substitution of the values of the surface resis-

tances of silver or copper films in Eq. (2-6-13) yields the microwave radiation atten-
uation in decibels by the silver-film or copper-film coating on a plastic substrate. Fig-
ure 2-6-9 shows the microwave radiation attenuation versus the surface resistance of
silver-film or copper-film coating, respectively.

I I I I
-
0--- Copper film

B 0- - - Silver film

- '
1 -

\I \
- - - - - - --- - -

-
I
9
I
I
d

I
I
I
I
I
I O
\

I I 1 I

Surface resistance R, of film coatings in ohms per square

Figure 2-6-9 Microwave radiation attenuation versus surface resistance of silver

and copper film.

Light transmittance. Light transmittance T and light reflection loss R of

silver-film and copper-film coatings are computed by using Eqs. (2-6-16) and
2-6-15), respectively. The values of the refractive index n and the extinction index k
of the silver-film and copper-film coatings deposited in a vacuum for the light-
frequency range are taken from Table 2-6-2. The refractive index no of air or vac-
uum is unity. The refractive index n2 of the nonabsorbing plastic glass is taken as
1.5. From the values of light transmittance T and light reflection loss R, absorption
loss A and total attenuation L are calculated. The results are illustrated in Figs. 2-6- 10
and 2-6- 1 1 for silver-film and copper-film coatings, respectively.
Optimum Condition. The light transmittance is increased as the surface re-
sistance 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 ra-
diation attenuation and 94% of light transmittance with a surface resistance of about
12 Rtsquare.
LIGHT TRANSMITTANCE T AND

0
I
-
0
LIGHT ATTENUATION L IN PERCENT
N
0
W
0 0
P
0
V I
0
~
0
O
0
D
0
d
W
0
O
56 Electromagnetic Plane Waves Chap. 2

- I 1 I 1
-
Silver film
-

--
'J - I
I
-
2 I
I
I
70 I I I I I

0 10 12 20 30 40
Surface resistance R, in ohms per square

Figure 2-6-12 Light transmittance versus surface resistance of silver and copper
films.

Example 2-6-6: Computation of a Copper-Film Coating

A copper film of 60 A is coated on a plastic substrate with a refractive index of 1.50.
Compute:

a. The copper-film surface resistance in ohms per square

b. The microwave attenuation in decibels
c. The light transmittance T for red light of 0.69 p m
d. The light reflection loss R at A = 0.69 p m
Solution
a. From Eq. (2-6- 1) the copper-film conductivity is

= 1.47 x lo7 mhoslm

From Eq. (2-6-3) the copper-film surface resistance is

R = - 1= 1
" tuf 60 X lo-'' x 1.47 x lo7
= 1 1.34 fllsquare

b. From Eq. (2-6-13) the microwave attenuation is

Attenuation = 40 - 20 log (R,)
Suggested Readings

= 40 - 20 log (1 1.34)
= 19 dB
c. The light transmittance T is estimated from Fig. 2-6- 11 to be 82%
d. From the same figure the light reflection loss R is about 18%.

REFERENCES

[I] ADLER,R. B., et al., Electromagnetic Energy Transmission and Radiation. P. 8. MIT
Press, Cambridge, Mass., 1969.
[2] LIAO,S. Y., Reflectivities of electromagnetic waves by seawater, dry sand, concrete ce-
ment, 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] F u c ~ s 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 Conf. 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.
[lo] HAWTHORNE, E. I., Electromagnetic shielding with transparent coated glass. Proc. IRE.,
42, 548-553, March 1954.
[ l I] 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 coat-
ing 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 Com-
pany, New York, 1966.
HAYT,W. H., Engineering Electromagnetics, 4th ed., Chapters 10 and 11. McGraw-Hill
Book Company, New York, 1981.
58 Electromagnetic Plane Waves 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 Appli-
cations. 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
(0, = V/;iZ 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 d/dz = p, and d/dt = jw for a sinusoidal wave propagating in the z di-
rection.
2-3. The electric field of a plane wave propagating in free space is given in complex nota-
tion by
E = 10-4e,(m1+20n~)
Ux + 1 0 - 4 ~ ] ( m r + n / 2 + 20nr)
U"
Where U, and U, 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 propaga-
tion.
2-4. 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 propagatinp through the
slab. Neglect the reflections from the back face of the slab.
2-5. The reflection and refraction of microwave propagating in the ionosphere are deter-
mined by the electron density in the ionosphere. If the electron density is assumed to
be lOI4 electrons per cubic meter, detemine the critical frequency for vertical incidence
so that the signal wave will be reflected back to the earth.
2-6. The conductivity u of copper is 5.8 x lo7 mhoslm and its relative permeability is
unity. Calculate the magnitudes of reflectivity of copper for vertical and horizontal po-
larizations against the grazing angle $ of 0 to 90" at a frequency range of l to 40 GHz.
The increment of the angle 4 is 10" each step and the increment of the frequency is
10 GHz each step.
2-7. 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.
2-8. Determine the pseudo-Brewster angle $ in terms of u, , u2, 17, , and q2 for a good con-
ductor. [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 $ 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 lo7 mhoslm, a resistivity of 2.44 x R-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 10' mhoslm, a resistivity of 1.620 x R-m,
and an electron mean-free-path of 570 A. Repeat Problem 2-1 1 for silver film.
2-13. Seawater has a conductivity of 4 mhoslm 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 (a = 2 x mholm and E, = 4) and copper
(a = 5.8 x lo7 mhoslm).
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 lo-' Vlm 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 mhoslm and a relative dielectric constant
of 10 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 lo7 mhoslm 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 deci-
bels by the copper at f = MHz.
2-19. A radar transmitter has an output power of 100 kW average. Calculate the power den-
sity in dBW/m2 at a range of 3000 m and the free-space attenuation in decibels at
f = 10 GHz.
2-20. Write a complete FORTRAN program to compute the magnitudes of reflectivity in ver-
tical 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 1 GHz, 1 GHz between 1 to 10
GHz, and 5 GHz between 10 to 40 GHz. Use F10.5 format for numerical outputs and
Hollerith format for character outputs. Print the outputs in three columns such as fre-
quency (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 nonabsorbing plastic glass for thicknesses of 10 to
100 A with an increment of 10 A each step. The wavelength varies from 2000 to
60 Electromagnetic Plane Waves 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 de-
posited in a vacuum are listed in Table 2-6-2. The refractive index n of the nonab-
sorbing plastic glass is 1.5. Use F10.5 format for numerical outputs, Hollerith format
for character outputs, and Data statements to read in the input values.
2-23. Write a complete FORTRAN program to compute the light transmittance and light
reflection for an aluminum-film coating on a nonabsorbing plastic substrate for thick-
nesses of 10 to 100 A and print out the results in percentages. Use F10.5 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 P2-23

Wave!ength Refractive Extinction Wavelength Refractive Extinction

(A) index n index k (A) index n index k
0000 0.00 0.00 0000 0.00 0.00

2-25. Repeat Problem 2-22 for a copper-film coating on a nonabsorb@ plastic glass for the
wavelengths from 4500 to 10,000 A with an increment of 500 A each step.
2-26. Repeat Problems 2-20 and 2-21 for dry sand.
2-27. Repeat Problems 2-20 and 2-2 1 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.
I Microwave Transmission
Lines

3-0 INTRODUCTION

Conventional two-conductor transmission lines are commonly used for transmitting

microwave energy. If a line is properly matched to its characteristic impedance at
each terminal, its efficiency can reach a maximum.
In ordinary circuit theory it is assumed that all impedance elements are lumped
constants. This is not true for a long transmission line over a wide range of frequen-
cies. Frequencies of operation are so high that inductances of short lengths of con-
ductors and capacitances between short conductors and their surroundings cannot be
neglected. These inductances and capacitances are distributed along the length of a
conductor, and their effects combine at each point of the conductor. Since the wave-
length is short in comparison to the physical length of the line, distributed parame-
ters cannot be represented accurately by means of a lumped-parameter equivalent
circuit. Thus microwave transmission lines can be analyzed in terms of voltage, cur-
rent, and impedance only by the distributed-circuit theory. If the spacing between
the lines is smaller than the wavelength of the transmitted signal, the transmission
line must be analyzed as a waveguide.

3- 1 TRANSMISSION-LINE EQUATIONS AND SOLUTIONS

-
3- 1 1 Transmission-Line Equations

A transmission line can be analyzed either by the solution of Maxwell's field equa-
tions 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 Microwave Transmission Lines Chap. 3

the time variable. In this section 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-circui t theory, the schematic circuit of a con-
ventional 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.

RAz LAz A RAz LAz B

Figure 3-1-1 Elementary section of a transmission line.

By Kirchhoff's voltage law, the summation of the voltage drops around the
central loop is given by
di(z, t )
G(Z, t ) = i ( z , t ) AZ~ + LAZ---
at
+ U ( ZI,) +-
a U- ')
(z,-
dz
(3-1-11

Rearranging this equation, dividing it by Az, and then omitting the argument ( z , t ) ,
which is understood, we obtain

Using Kirchhoff's current law, the summation of the currents at point B in Fig.
3- 1 - 1 can be expressed as
av(z + AZ, t )
i(z, t ) = v(z + Az, t ) G A z + C A Z at
+ i(z + Az, t )

By rearranging the preceding equation, dividing it by Az, omitting ( z , t ) , and as-

suming Az equal to zero, we have

Then by differentiating Eq. (3-1-2) with respect to z and Eq. (3-1-4) with respect to
I and combining the results, the final transmission-line equation in voltage form is
Sec. 3.1 Transmission-Line Equations and Solutions 63

Also, by differentiating Eq. (3- 1-2) with respect to t and Eq. (3-1-4) with respect to
z and combining the results, the final transmission-line equation in current form is
d2i
-=
az *
RGi + (RC + LC)-diat + LC, ata2i
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
V(Z,t) = Re V (z)eJ"' (3- 1-7)

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

y =a + jp (propagation constant) (3-1-11)

where V+ and I+ indicate complex amplitudes in the positive z direction, V- and I-
signify complex amplitudes in the negative z direction, a is the attenuation constant
in nepers per unit length, and P is the phase constant in radians per unit length.
If we substitute jw for a/dt in Eqs. (3-1-2), (3-1-4), (3-1-5), and (3-1-6) and
divide each equation by ej"', the transmission-line equations in phasor form of the
frequency domain become

in which the following substitutions have been made:

Z =R+ jwL (ohms per unit length)
Y = G + jwC (mhos per unit length)
y =~ Z =Ya + jp (propagation constant)
64 Microwave Transmission Lines Chap. 3

For a lossless line, R = G = 0, and the transmission-line equations are expressed as

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 dif-
ference is that the transmission-line equations are one-dimensional.

3-1-2 Solutions of Transmission-Line Equations

The one possible solution for Eq. (3-1- 14) is

v = V + ~ - Y Z + V - ~ Y Z = V+e-aze-iPz + V-eazeip: (3-1-23)
The factors V+ and V- represents complex quantities. The term involving e-jP'
shows a wave traveling in the positive z direction, and the term with the factor ejBzis
a wave going in the negative z direction. The quantity pz 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-7' - V-eYz) = Yo(V+e-"'e-JP' - V-ea'eJPz) (3-1-24)
In Eq. (3-1-24) the characteristic impedance of the line is defined as
7 I

The magnitude of both voltage and current waves on the line is shown in Fig. 3-1-2.

0 0
(a) Voltage wave ( b ) Current wave

Figure 3-1-2 Magnitude of voltage and current traveling waves.

Sec. 3.1 Transmission-Line Equations and Solutions

At microwave frequencies it can be seen that

R + oL and G 6 o C (3- 1-26)
By using the binomial expansion, the propagation constant can be expressed as

Therefore the attenuation and phase constants are, respectively, given by

Similarly, the characteristic impedance is found to be

From Eq. (3- 1-29) the phase velocity is

The product of LC is independent of the size and separation of the conductors and
depends only on the permeability p 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 Microwave Transmission Lines Chap. 3

Thus the numerical value of ~ / V Lfor Cair-insulated conductors is approximately

equal to the velocity of light in vacuum. That is,

When the dielectric of a lossy microwave transmission line is not air, the phase ve-
locity is smaller than the velocity of light in vacuum and is given by

In general, the relative phase velocity factory can be defined as

actual phase velocity
Velocity factor =
velocity of light in vacuum

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: Line Characteristic Impedance and Propagation Constant

A transmission line has the following parameters:

Calculate: (a) the characteristic impedance; (b) the propagation constant.

Solution
a. From Eq. (3-1-25) the line characteristic impedance is

b. From Eq. (3-1- 18) the propagation constant is

Sec. 3.2 Reflection Coefficient and Transmission Coefficient

3-2REFLECTION COEFFICIENT AND TRANSMISSION

COEFFICIENT

1 Reflection Coefficient
3-2-

In the analysis of the solutions of transmission-line equations in Section 3-1, the

traveling wave along the line contains two components: one traveling in the positive
z direction and the other traveling the negative z direction. If the load impedance is
equal to the line characteristic impedance, however, the reflected traveling wave
does not exist.
Figure 3-2-1 shows a transmission line terminated in an impedance Ze. It is
usually more convenient to start solving the transmission-line problem from the re-
ceiving rather than the sending end, since the voltage-to-current relationship at the
load point is fixed by the load impedance. The incident voltage and current waves
traveling along the transmission line are given by
V = V+e-" + V-e+yZ (3-2-1)

in which the current wave can be expressed in terms of the voltage by

If the line has a length of t?, the voltage and current at the receiving end become

The ratio of the voltage to the current at the receiving end is the load impedance.
That is,

"r z*
Sending
end end

Figure 3-2-1 Transmission line terminated in a load impedance.

68 Microwave Transmission Lines Chap. 3

The reflection coefficient, which is designated by r (gamma), is defined as

reflected voltage or current
Reflection coefficient =
incident voltage or current

If Eq. (3-2-6) is solved for the ratio of the reflected voltage at the receiving end,
which is V-eye, to the incident voltage at the receiving end, which is V+eYe,the re-
sult is the reflection coefficient at the receiving end:

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

I
where I re( is the magnitude and never greater than unity-that is, 1 rt 5 1. Note
that 8e 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

From Fig. 3-2-1 let z = t - d. Then the reflection coefficient at some point located
a distance d from the receiving end is

Next, the reflection coefficient at that point can be expressed in terms of the
reflection coefficient at the receiving end as

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 loss-
less 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
-2pd as shown in Fig. 3-2-3.
It is evident that rewill be zero and there will be no reflection from the re-
ceiving end when the terminating impedance is equal to the characteristic impedance
Sec. 3.2 Reflection Coefficient and Transmission Coefficient

Figure 3-2-2 Reflection coefficient for

lossy line.

Figure 3-2-3 Reflection coefficient for

lossless line.

of the line. Thus a terminating impedance that differs from the characteristic
impedance will create a reflected wave traveling toward the source from the termina-
tion. The reflection, upon reaching the sending end, will itself be reflected if the
source impedance is different from the line characteristic impedance at the send-
ing end.

Transmission Coefficient
3-2-2

A transmission line terminated in its characteristic impedance Zo is called a properly

terminated line. Otherwise it is called an improperly terminated line. As described
r
earlier, there is a reflection coefficient at any point along an improperly termi-
nated 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

Equation (3-2-13) will be verified later. The letter T represents the transmission
coefficient, which is defined as
70 Microwave Transmission Lines Chap. 3

T transmitted voltage or current - -Vtr - -1"

zz (3-2-14)
incident voltage or current Vinc IinC
Figure 3-2-4 shows the transmission of power along a transmission line where Pincis
the incident power, Prefthe reflected power, and P, the transmitted power.

F - : ~ - - P , ,

VR Load

I -a-
-2

Figure 3-2-4 Power transmission on a line

Let the traveling waves at the receiving end be

V+e-7" $-eye = V , , ~ - Y ~

Multiplication of Eq. (3-2-16) by Ze and substitution of the result in Eq. (3-2-15)

yield

which, in turn, on substitution back into Eq. (3-2-15), results in

The power carried by the two waves in the side of the incident and reflected waves is
(V+e-up)Z- (V-
Plnr = Pinc - Pref =
2Zo 2Zo
The power carried to the load by the transmitted waves is

By setting Pin,= Ptrand using Eqs. (3-2-17) and (3-2- 18), we have

This relation verifies the previous statement that the transmitted power is equal to
the difference of the incident power and reflected power.
Sec. 3.3 Standing Wave and Standing-Wave Ratio 71

Example 3-2-1: Reftection Coefficient and Transmission Coefficient

A certain transmission line has a characteristic impedance of 75 + jO.O1 R and is ter-
minated in a load impedance of 70 + j50 R. 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

a. From Eq. (3-2-17) the reflection coefficient is

b. From Eq. (3-2-18) the transmission coefficient is

Thus Eq. (3-2-21) is verified.

d. From Eq. (2-3-18) we obtain

3-3 STANDING WAYE AND STANDING-WAYE RATIO

3-3-1 Standing Wave

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
v = V+e-aze-jPz + V-emzejPz
= V+e - m z [ ~ (pz)
~s - j sin (pz)] + V-emz[cos (pz) + j sin (pz)] (3-3-1)
= (V+e-az + V-em') cos (pz) - j(V+ epaz - V-eaz) sin (pz)
72 Microwave Transmission Lines Chap. 3

With no loss in generality it can be assumed that V+e-"' and V-ea' are real. Then
the voltage-wave equation can be expressed as
V , = Voe-j@ (3-3-2)
This is called the equation of the voltage standing wave, where
V O= [ ( V +e-a' + cos2 (Pz) + (V+e-az - V-e-a')2 sin2 (/3z)]'I2 (3-3-3)
which is called the standing-wave pattern of the voltage wave or the amplitude of
the standing wave, and

4 = arctan
v+e-OL' - V - e
V +e-az + V - e a z
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 re-
spect to /3z and equating the result to zero. By doing so and substituting the proper
values of p z in the equation, we find that

1. The maximum amplitude is

V,,, = V+e-*' + V-ea' = V+e-"'(I + Irl) (3-3-5)
and this occurs at pz = n n , where n = 0, + 1 , k 2 , . . . .
2. The minimum amplitude is
V . = V +e-az - V-e"' = V+e-"'(1 - ( r ( )
mln (3-3-6)
and this occurs at pz = (2n - I ) n / 2 , where n = 0, & 1 , +2, . . . .
3. The distance between any two successive maxima or minima is one-half wave-
length, since

pz=nrr rm- -
z=-= nrr A
- n- ( n = 0, k l , 2 2 , . . .)
p 2r/A 2

Then

It is evident that there are no zeros in the minimum. Similarly,

I,, = I+e-"' + I-eaz = I+e-"'(1 + II'l) (3-3-8)

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 + # 0 and V - = 0, the standing-wave pattern becomes

V O= V + e - a z
Sec. 3.3 Standing Wave and Standing-Wave Ratio

Current

0
Figure 3-3-1 Standing-wave pattern in a lossy line.

1 rc Figure 3-3-2 Voltage standing-wave

0 pattern in a lossless line.

2. When V+ = 0 and V- # 0, the standing-wave pattern becomes

Vo = V-eaZ (3-3-1 1)
3. When the positive wave and the negative wave have equal amplitudes (that is,
I
I V+ePaz = I V- eazI) or the magnitude of the reflection coefficient is unity, the
standing-wave pattern with a zero phase is given by
V, = 2V+e-"' cos (pz) (3-3- 12)
which is called a pure standing wave

Similarly, the equation of a pure standing wave for the current is

I, = - j2Yo V+e-aZ sin (pz) (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:
us = (z, t) = Re[V,(z)eJ"'] = 2V+e-"' cos (pz) cos (wt) (3-3-14)
is = (z, t) = Re[I,(z)eJW']= 2YoV+e-"' sin (pz) sin (wt) (3-3-15)
The amplitudes of Eqs. (3-3-14) and (3-3-15) vary sinusoidally with time; the
74 Microwave Transmission Lines 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.

Figure 3-3-3 Pure standing waves of

voltage and current.

Standing. Wave Ratio

3-3-2

Standing waves result from the simultaneous presence of waves traveling in opposite
directions on a transmission line. The ratio of the maximum of the standing-wave
pattern to the minimum is defined as the standing-wave ratio, designated by p. That
is,
maximum voltage or current
Standing-wave ratio =
minimum voltage or current
( Vmax ( - I Imax 1
P=---
1 Vmin 1 1 Imin 1
The standing-wave ratio results from the fact that the two traveling-wave components
of Eq. (3-3- 1) 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 distinc-
tions are made between VSWR and ISWR.
When the standing-wave ratio is unity, there is no reflected wave and the line is
called a@t 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 low-
loss 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
Sec. 3.3 Standing Wave and Standing-Wave Ratio

and vice versa

This relation is very useful for determining the reflection coefficient from the
standing-wave ratio, which is usually found from the Smith chart. The curve in Fig.
3-3-4 shows the relationship between reflection coefficient 1 r 1 and standing-wave
ratio p .

1 2 3 4 5 6 7 8 9 10
VSWR p

Figure 3-3-4 SWR versus reflection coefficient.

As a result of Eq. (3-3-17), since I r I 5 1, the standing-wave ratio is a posi-

tive real number and never less than unity, p r 1. From Eq. (3-3-18) the magni-
tude of the reflection coefficient is never greater than unity.

Example 3-3-1: Standing-Wave Ratio

A transmission line has a characteristic impedance of 50 + jO.O1 R and is terminated
in a load impedance of 73 - j42.5 R. Calculate: (a) the reflection coefficient; (b) the
standing-wave ratio.
Solution

a. From Eq. (3-2-8) the reflection coefficient is

b. From Eq. (3-3-17) the standing-wave ratio is

76 Microwave Transmission Lines Chap. 3

3-4 LINE IMPEDANCE AND ADMITTANCE

3-4-1 Line Impedance

The line impedance of a transmission line is the complex ratio of the voltage phasor
at any point to the current phasor at that point. It is defined as

Figure 3-4- 1 shows a diagram for a transmission line.

Sending end Receiving end
I, Is I Ir 1,

ZQ

Figure 3-4-1 Diagram of a transmission line showing notations.

In general, the voltage or current along a line is the sum of the respective inci-
dent wave and reflected wave-that is,
V = Vinc+ Vref= V+e-Y2+ V-e" (3-4-2)
I = Iinc+ Iref= Yo(V+eOZ- V- eYz) (3-4-3)
At the sending end z = 0; then Eqs. (3-4-2) and (3-4-3) become
I,ZS =v+ + v- (3-4-4)
1,Zo = v+ - v- (3-4-5)
By solving these two equations for V+ and V-, we obtain
Is
v+ -- 2-(Z, + ZO)

v- = -Is2 (Z, - Z0)

Substitution of V+ and V- in Eqs. (3-4-2) and (3-4-3) yields

Is
I = -[(Zs
2Zo
+ Zo)e-Yz - (Z, - Zo)e"]
Sec. 3.4 Line Impedance and Admittance n
Then the line impedance at any point z from the sending end in terms of Z, and Zo is
expressed as

z = zo (Z, + Zo)e-?' + (Z, - Zo)eY2

(Z, + Zo)e-?' - (Z, - Zo)eY'
At z = [ the line impedance at the receiving end in terms of Z, and Zo is given by
(2, + Zo)e-?' + (2, - Zo)eYe
z, = zo (Z, + Zo)e-~'- (Z, - Zo)eye
Alternatively, the line impedance can be expressed in terms of Zp and ZO.At
z = 8, V, = IeZe; then

IpZe = V+e-ye + V-eye (3-4-12)

Solving these two equations for V+ and V- , we have

Then substituting these results in Eqs. (3-4-2) and (3-4-3) and letting z = f - d,
we obtain
I&
V = -[(Ze
2
+ Z0)e + (Zp - Zo)e-yd]
Yd
(3-4- 16)
Ie
I = -[(Ze + Zo)eYd- (Zp - Z0)ePd]
2Zo
Next, the line impedance at any point from the receiving end in terms of Ze and Zo
is expressed as

The line impedance at the sending end can also be found from Eq. (3-4-18) by let-
ting d = [:

It is a tedious task to solve Eqs. (3-4-10), (3-4-1 l), (3-4-18), or (3-4-19) for
the line impedance. These equations can be simplified by replacing the exponential
factors with either hyperbolic functions or circular functions. The hyperbolic func-
tions are obtained from
~ (yz) rC- sinh (yz)
e * =~ cosh (3-4-20)
Substitution of the hyperbolic functions in Eq. (3-4-10) yields the line impedance at
78 Microwave Transmission Lines Chap. 3

any point from the sending end in terms of the hyperbolic functions:
Z, cosh (yz) - Zo sinh (yz) Z, - Zo tanh (yz)
Z = Zo = Zo (3-4-21)
Zocosh(yz)-Z,sinh(yz) Zo-Z,tanh(yz)
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:
cosh (yd) + Zo sinh (yd) - Ze + Zo tanh (yd)
z = zo Zr
ZOcash (yd) + Zt sinh (yd) z O+ Ze ~ tanh
~ (yd)
(3-4-22)

For a lossless line, y = jp; and by using the following relationships between hyper-
bolic and circular functions
sinh (jpz) = j sin (pz) (3-4-23)
cosh (jpz) = cos (pz) (3-4-24)
the impedance of a lossless transmission line (ZO= Ro) can be expressed in terms of
the circular functions:
Z, cos (pz) - jRo sin (pz) Z, - jRo tan (pz)
Z = Ro = Ro (3-4-25)
Ro cos (pz) - j Z, sin (pz) RO- j Z, tan (pz)
and
Z, cos (pd) + jRo sin (Pd) -
- ze + IRo tan (pd) (3-4-26)
Z = Ro
R~ cos (pd) + jZe sin (pd) R O ~ +
o jZr tan (pd)

Impedance in terms of reflection coefficient or standing-wave ratio.

Rearrangement of Eq. (3-4-18) gives the line impedance-looking at it from the re-
ceiving end-as

in which the following substitution is made by

From Eq. (3-2-12) the reflection coefficient at a distance d from the receiving end is
given by
r = ree -2yd = lre ( e -2adej(Oe-2Dd)
(3-4-29)
Then the simple equation for the line impedance at a distance d from the load is ex-
pressed by

z = zo-11 -+rr
The reflected coefficient is usually a complex quantity and can be written
r = II'lej@ (3-4-31)
Sec. 3.4 Line Impedance and Admittance

where 1 r 1 = I rele-2ad
4 = Be - 2pd
The impedance variation along a lossless line can be found as follows:

Z(d) = Zo
I r leJ+ = Ro 1 + I r l(cos 4 + j sin 4 )
1+
1 r leJ+
1- 1 - 1 r ~(COS + + j sin 4 )
= R (d) + .jX (d) = I Z(d) 1ejed

1 + 2 ( r Jcos + + /rI2
where ( Z(d) 1 = RO
Ji- 2 , r l cos 4 + jrp

X(d) = Ro
r1
21 sin 4
1 - 21r1 cos 4 +lrlZ
& = arctan 6)
= arctan ( - I
21 I' 1 sin 4
)
Since 4 = 0e - 2pd, then 4 = Be - 21r if p d = 7r. However, C O S (-~ 21r)
~ =
cos 0e and sin(& - 27r) = sin 0e; then

It is concluded that the impedance along a lossless line will be repeated for every in-
terval at a half-wavelength distance.
Furthermore, the magnitude of a reflection coefficient 1 J? 1 is related to the
standing-wave ratio p by

The line impedance at any location from the receiving end can be written

This is a very useful equation for determining the line impedance in terms of stand-
ing-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 trans-
mission line is to take two measurements:

1. Measure the sending-end impedance with the receiving end short-circuited and
record the result:
Z,, = Zo tanh (y t ) (3-4-40)
80 Microwave Transmission Lines Chap. 3

2. Measure the sending-end impedance with the receiving end open-circuited and
record the result:
Z, = zo~ 0 t h( y e) (3-4-41)

Then the characteristic impedance of the measured transmission line is given by

Zo = vzzi (3-4-42)
and the propagation constant of the line can be computed from
7

Normalized impedance. The normalized impedance of a transmission line

is defined as

It should be noted that the lowercase letters are commonly designated for normal-
ized 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 nor-
malized impedance for a lossless line has the following significant features:

1. The maximum normalized impedance is

Here zmaxis 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
--
- IVminI - 1
-
- [rl-1
-
Zmin
RO 0 a 1 1+ 1r1 p
Here z,, 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, z,, or zmin is
repeated:

4. Since V,, and V,,,, are separated by a quarter-wavelength, z,, is equal to the
reciprocal of z,,, for every A/4 separation:
Sec. 3.4 Line Impedance and Admittance

3-4-2 Line Admittance

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

Then the normalized admittance can be written

Example 3-4-1: Line Impedance

A lossless line has a characteristic impedance of 50 Q and is terminated in a load resis-
tance of 75 Q. The line is energized by a generator which has an output impedance of
50 Q 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 quarter-
wave line. Then

From Eq. (3-4-26) the input impedance is

b. The reflection coefficient is

Then the instantaneous voltage at the load is

Vp = V+e-lPe(l + re) = 30(1 + 0.20) = 36 V
c. The instantaneous power delivered to the load is
82 Microwave Transmission Lines Chap. 3

3-5 SMITH CHART

Many of the computations required to solve transmission-line problems involve the

use of rather complicated equations. The solution of such problems is tedious and
difficult because the accurate manipulation of numerous equations is necessary. To
simplify their solution, we need a graphic method of arriving at a quick answer.
A number of impedance charts have been designed to facilitate the graphic so-
lution of transmission-line problems. Basically all the charts are derived from the
fundamental relationships expressed in the transmission equations. The most popular
chart is that developed by Phillip H. Smith [I]. The purpose of this section is to
present the graphic solutions of transmission-line problems by using the Smith chart.
The Smith chart consists of a plot of the normalized impedance or admittance
with the angle and magnitude of a generalized complex reflection coefficient in a
unity circle. The chart is applicable to the analysis of a lossless line as well as a lossy
line. By simple rotation of the chart, the effect of the position on the line can be de-
termined. To see how a Smith chart works, consider the equation of reflection
coefficient at the load for a transmission line as shown in Eq. (3-2-8):

Since I re 1 r 1, the value of re 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

which is also on or within the unity circle. Figure 3-5-1 shows circles for a constant
r
reflection coefficient and constant electrical-length radials pd.
From Eqs. (3-4-29) and (3-4-44) the normalized impedance along a line is
given by

With no loss in generality, it is assumed that d = 0; then

and

Substitution of Eq. (3-5-5) into Eq. (3-5-4) yields

1- I-? - I-?
Sec. 3.5 Smith Chart

Figure 3-5-1 Constant r circles and

electrical-length radials pd.

and

Equations (3-5-6) and (3-5-7) can be rearranged as

and

Equation (3-5-8) represents a family of circles in which each circle has a con-
stant resistance r. The radius of any circle is 1/(1 + r), and the center of any circle
is r/(l + 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 ( l l x ) , and the center of
any circle is at

r,=1 r . =X1- (where -03 I x I W)

All constant reactance circles are plotted in Fig. 3-5-3 according to Eq. (3-5-9).
Chap. 3

Figure 3-5-3 Constant reactance x circles.

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):

where

The Smith chart may also be used for normalized admittance. This is evident
since
1 1
YO= - = G o + jBo and Y =-=G + jB
Zo z
Then the normalized admittance is

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 (T,= 1,
ri = 0).
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 con-
stant 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 = lip, there is a Vminon the line.
8. At a point of z,,, = p , there is a V,, on the line.
9. The horizontal radius to the right of the chart center corresponds to V, , Imin,
z,, , and p (SWR) .
10. The horizontal radius to the left of the chart center corresponds to Vmin,Imax,
zmin , and 1/ 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.
Microwave Transmission Lines Chap. 3

Figure 3-5-4 Smith chart.

The magnitude of the reflection coefficient is related to the standing-wave ratio

by the following expression:

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: Location Determination of Voltage Maximum and Minimum from

Load
Given the normalized load impedance ze = 1 + j 1 and the operating wavelength
Sec. 3.5 Smith Chart

Figure 3-55 Diagram for Example 3-5-1.

X = 5 cm, determine the first V,,,, first V,,, from the load, and the VSWR p as shown
in Fig. 3-5-5.
Solution

1. Enter z p = 1 + j 1 on the chart as shown in Fig. 3-5-6.

2. Read 0.162A 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 O.162A toward the generator and first stop at
the voltage maximum on the right-hand real axis at 0.25A. Then

4. Similarly, move a distance from the point of O.162A toward the generator and
first stop at the voltage minimum on the left-hand real axis at 0.5A. Then
d2(Vmln)= (0.5 - 0.162)A = (0.338)(5) = 1.69 cm

Figure 3-5-6 Graphic solution for

Example 3-5-1.
Microwave Transmission Lines Chap. 3

5. Make a standing-wave circle with the center at ( I , 0) and pass the circle through
the point of 1 + 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: Impedance Determination with Short-Circuit Minima Shift

The location of a minimum instead of a maximum is usually specified because it can be
determined more accurately. Suppose that the characteristic impedance of the line Ro is
50 s2, and the SWR p = 2 when the line is loaded. When the load is shorted, the
minima shift 0.15A toward the load. Determine the load impedance. Figure 3-5-7
shows the diagram for the example.

Figure 3-5-7 Diagram for Example 3-5-2.

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.

Figure 3-5-8 Graphic solution for

Example 3-5-2.
Sec. 3.6 Impedance Matching 89

2. When the line is loaded, the first voltage minimum shifts 0.1% 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 . 1 9 from the minimum point along the distance scale
toward the load and stop at O.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

7. The load impedance is

3-6IMPEDANCE MATCHING

Impedance matching is very desirable with radio frequency (RF) transmission lines.
Standing waves lead to increased losses and frequently cause the transmitter to mal-
function. 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 nonreso-
nant; that is, its input impedance always remains at the same value Zo when the fre-
quency 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 charac-
teristic 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 imped-
ance 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.

L----J L----J

Figure 3-6-1 Matched transmission-line system.

90 Microwave Transmission Lines Chap. 3

For a low-loss or lossless transmission line at radio frequency, the characteris-

tic impedance ZOof the line is resistive. At every point the impedances looking in
opposite directions are conjugate. If ZOis 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 iron-
cored transformer is almost universally used as an impedance-matching device.
Occasionally an iron-cored transformer is also used at radio frequencies. In a practi-
cal 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 imped-
ances. 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.

1 Single-Stub Matching
3-60

Although single-lumped inductors or capacitors can match the transmission line, it is

more common to use the susceptive properties of short-circuited sections of trans-
mission lines. Short-circuited sections are preferable to open-circuited ones because
a good short circuit is easier to obtain than a good open circuit.
For a lossless line with Y, = Yo, maximum power transfer requires YII = Yo,
where Yll is the total admittance of the line and stub looking to the right at point 1- 1
(see Fig. 3-6-2). The stub must be located at that point on the line where the real
part of the admittance, looking toward the load, is YO.In a normalized unit yll must
be in the form

- I d +

Figure 3-6-2 Single-stub matching for Example 3-6-1

Sec. 3.6 Impedance Matching

if the stub has the same characteristic impedance as that of the line. Otherwise
YII = Yd f Ys= Yo (3-6-2)
The stub length is then adjusted so that its susceptance just cancels out the suscep-
tance of the line at the junction.

Example 3-6-1: Single-Stub Matching

A lossless line of characteristic impedance Ro = 50 R is to be matched to a load
Ze = 5 0 / [ 2 + j ( 2 + lh)]R by means of a lossless short-circuited stub. The charac-
teristic impedance of the stub is 100 R. Find the stub position (closest to the load) and
length so that a match is obtained.
Solution

1. Compute the normalized load admittance and enter it on the Smith chart (see
Fig. 3-6-3).

2. Draw a SWR circle through the point of ye so that the circle intersects the unity
circle at the point yd.

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.

Figure 3-6-3 Graphic solution for Example 3-6-1.

Microwave Transmission Lines 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
Y1I = Yd + Ys
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 junc-
tion are in parallel, as seen by the main line extending to the generator. The ad-
mittances must be converted to normalized values for matching on the Smith
chart. Then Eq. (3-6-2) becomes

4. The distance between the load and the stub position can be calculated from the
distance scale as

5. Since the stub contributes a susceptance of + j5.20, enter + j5.20 on the chart
and determine the rewired distance e from the short-circuited end (z = 0 ,
y = m), which corresdonds 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

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,

the susceptance of the stub will be

y: = - j5.2
7. The position of the stub from the load is

and the length of the short-circuited stub is

Double-Stub Matching
3-6-2

Since single-stub matching is sometimes impractical because the stub cannot be

placed physically in the ideal location, double-stub matching is needed. Double-stub
devices consist of two short-circuited stubs connected in parallel with a fixed length
between them. The length of the fixed section is usually one-eighth, three-eighths,
or five-eighths of a wavelength. The stub that is nearest the load is used to adjust the
susceptance and is located at a fixed wavelength from the constant conductance
Sec. 3.6 Impedance Matching 93

unity circle (g = 1) on an appropriate constant-standing-wave-ratio circle. Then the

admittance of the line at the second stub as shown in Fig. 3-6-4 is

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 prop-
erly, there will be no standing wave on the line to the left of the second stub mea-
sured from the load.

Figure 3-6-4 Double-stub matching for Example 3-6-2.

Example 3-6-2: Double-Stub Matching

The terminating impedance Ze is 100 + j 100 a,and the characteristic impedance Zo
of the line and stub is 50 a.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

1. Compute the normalized load impedance ze and enter it on the chart as shown in
Fig. 3-6-5:

2. Plot a SWR p circle and read the normalized load admittance 180" out of phase
with Ze on the SWR circle:

3. Draw the spacing circle of by rotating the constant-conductance unity circle

Microwave Transmission Lines Chap. 3

Figure 3-6-5 Graphic solution for Example 3-6-2

(g = 1) through a phase angle of 2pd = 2 p i = ~ i.rr toward the load. Now yll
must be on this spacing circle, since ya will be on the g = 1 circle ( y l l and y d ~
a
are A apart).
4. Move yp for a distance of 0.40A from 0.458 to 0.358 along the SWR p circle
toward the generator and read ydl on the chart:

5. There are two possible solutions for y ,, . They can be found by carrying ydl along
the constant-conductance (g = 0.55) circle that intersects the spacing circle at
two points:
b
y l l = 0.55 - jO.ll
y ; , = 0.55 - j1.88
6. At the junction 1- 1,
~ I =
I + yr~
Sec. 3.6 Impedance Matching

Then

Similarly,

7. The lengths of stub 1 are found as

8. The :A section of line transforms yll to y a and y l to y h along their constant

standing-wave circles, respectively. That is,

9. Then stub 2 must contribute

10. The lengths of stub 2 are found as

t , = (0.25 + 0.087)A = 0.337A
ei = (0.308 - 0 . 2 5 ) ~= 0 . 0 5 8 ~
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 l l 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 c m (0.25 in.) to about 2.54 cm (1 in.), depending on the power require-
ment. 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 Microwave Transmission Lines Chap. 3

3-7 MICROWA VE COAXIAL CONNECTORS

For high-frequency operation the average circumference of a coaxial cable must be

limited to about one wavelength, in order to reduce multimodal propagation and
eliminate erratic reflection coefficients, power losses, and signal distortion. The
standardization of coaxial connectors during World War I1 was mandatory for mi-
crowave operation to maintain a low reflection coefficient or a low voltage standing-
wave ratio (VSWR). Since that time many modifications and new designs for mi-
crowave connectors have been proposed and developed. Seven types of microwave
coaxial connectors are described below (see Fig. 3-7-1; TNC is not shown).

1. APC-3.5 The APC- 3.5 (Amphenol Precision Connector-3.5 mm) was

originally developed by Hewlett-Packard, but is now manufactured by Amphe-
nol. The connector provides repeatable connections and has a very low voltage
standing-wave ratio (VSWR). Either the male or female end of this 50-ohm
connector can mate with the opposite type of SMA connector. The APC-3.5
connector can work at frequencies up to 34 GHz.
2. APC-7 The APC-7 (Amphenol Precision Connector-7 mm) was also devel-
oped by Hewlett-Packard in the mid 1960s, but it was recently improved and is
now manufactured by Amphenol. The connector provides a coupling mecha-
nism without male or female distinction and is the most repeatable connecting
device used for very accurate 50-ohm measurement applications. Its VSWR is
extremely low, in the range of 1.02 to 18 GHz. Maury Microwave also has an
MPC series available.
3. BNC The BNC (Bayonet Navy Connector) was originally designed for mili-
tary system applications during World War 11. The connector operates very
well at frequencies up to about 4 GHz; beyond that it tends to radiate electro-
magnetic energy. The BNC can accept flexible cables with diameters of up to
6.35 mm (0.25 in.) and characteristic impedance of 50 to 75 ohms. It is now
the most commonly used connector for frequencies under 1 GHz.
4. SMA The SMA (Sub-Miniature A) connector was originally designed by
Bendix Scintilla Corporation, but it has been manufactured by Omni-Spectra
Inc. (as the OSM connector) and many other electronic companies. The main
application of SMA connectors is on components for microwave systems. The
connector is seldom used above 24 GHz because of higher-order modes.
5. SMC The SMC (Sub-Miniature C) is a 50-ohm connector that is smaller
than the SMA. The connector is manufactured by Sealectro Corporation and
can accept flexible cables with diameters of up to 3.17 mm (0.125 in.) for a
frequency range of up to 7 GHz.
6. TNC The TNC (Threaded Navy Connector) is merely a threaded BNC. The
function of the thread is to stop radiation at higher frequencies, so that the con-
nector can work at frequencies up to 12 GHz.
7. Type N The Type N (Navy) connector was originally designed for military
systems during World War I1 and is the most popular measurement connector
Sec. 3.7 Microwave Coaxial Connectors

New coupling nut

APC-3.5 male
Old coupling nut
BNC female
(b) APC-7
(a) APC-3.5

SMA female
BNC female

SMA male
BNC male
(d) SMA
(c) BNC

APC-3.5
male

Type N female
Type N male
SMC male
(plug)

(e) SMC (f) Type N

Figure 3-7-1 Microwave coaxial connectors.

98 Microwave Transmission Lines 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 1.07

REFERENCES

[I] SMITH,P. H.
Transmission line calculator. Electronics, 12, 29-3 1 (1939).
An improved transmission line calculator. Electronics, 17, 130- 133 and 3 18-325
(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 a]., 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.
L A O ,S. Y., Engineering Applications of Electromagnetic Theory.West Pubhsk~ngCompany,
St. Paul, Minn., 1988.
RIZZI,PETERA., Microwave Engineering: Passive Circuits. Prentice-Hall, Inc., Englewood
Cliffs, N.J., 1988.
SINNEMA, W., Electronic Transmission Technology: Lines, Waves, and Applications. Prentice-
Hall, Inc., Englewood Cliffs, N.J., 1979.

PROBLEMS

3-1. A transmission line has a characteristic impedance of 300 R and is terminated in a load
of 300 - j300 R. 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 R and is terminated in
a load of 100 R. 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 V,,, and minimum voltage V,,, on the line
c. The maximum current I,,, and minimum current I,,, on the line
d. The power transmitted by the line
3-3. A lossless line has a characteristic impedance of 75 R and is terminated in a load of
Problems 99

300 R . The line is energized by a generator which has an open-circuit output voltage of
20 V (rms) and output impedance of 75 R . The line is assumed to be 2 f 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 100 R and is terminated
in a load of 75 R . 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 R and its attenuation constant a is
0.0156 Nplm. 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 R , 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 Rlm, L = 5.2 X 10-8H/m, G =
6.2 x 10-3mho/m, and C = 2.13 X 10-'OFIm. The signal frequency is 4 GHz. Cal-
culate:
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 R and is terminated
in a load of 100 R . The line is energized by a generator of 20 V (rms) with an internal
resistance of 50 R . 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 ener-
gized by a generator which has an open-circuit output voltage of V, (rms) and an inter-
val impedance equal to the characteristic impedance of the line. Show that the sending-
end voltage is equal to the output voltage of the generator.
3-9. A lossless transmission line has a characteristic impedance of 300 R and is operated at
a frequency of 10 GHz. The observed standing-wave ratio on the line is 5.0. It is pro-
posed 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 R and is loaded by 60 - j60 R.
One stub is at the load, and the other is 3hj8 distance away from the first.
a. Determine the lengths in wavelength of the two short-circuited stubs when a match
is achieved.
100 Microwave Transmission Lines Chap. 3

b. Locate and crosshatch the forbidden region of the normalized admittance for possi-
ble match.
3-11. A lossless transmission line has a characteristic impedance of 300 fi and is terminated
by an impedance Zr. The observed standing-wave ratio on the line is 6, and the dis-
tance of the first voltage minimum from the load is O.166A.
a. Determine the load Zr.
b. Find the lengths in A of two shorted stubs, one at the load and one at A4! from the
load, which are required to match the load to the line.
3-12. A single-stub tuner is to match a lossless line of 400 R to a load of 800 - j 3 0 0a.
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.
3-13. A half-wave-dipole antenna has a driving-point impedance of 73 + j42.5 a. A loss-
less transmission line connected to a TV set has a characteristic impedance of 300 a.
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 an-
tenna. 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.]
3-14. A lossless transmission line has a characteristic impedance of 100 R and is loaded by
100 + j 100 R. A single shorted stub with the same characteristic impedance is in-
serted 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.
3-15. A double-stub matching line is shown in Fig. P3-15. The characteristic resistances of
the lossless line and stubs are 100 R, 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 so-
lutions .]

Figure P3-15
Problems 101

3-16. A lossless transmission line has a characteristic impedance Zo of 100 S1 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 characteris-
tic impedance Zo, < 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 Zol in terms of Zo.
3-17. A lossless transmission line has a characteristic impedance of 100 R 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 el 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.

Figure P3-18
I Microwave Waveguides
and Components

4-0 INTRODUCTION

In general, a waveguide consists of a hollow metallic tube of a rectangular or circu-

lar 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 WR-
90 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 [I].
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 particu-
lar 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 attenu-
ated to a negligible value in a relatively short distance. The dominant mode in a par-
ticular guide is the mode having the lowest cutoff frequency. It is advisable to choose
the dimensions of a guide in such a way that, for a given input signal, only the en-
ergy 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 Rectangular Waveguides 103

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.

4-1 RECTANGULAR WAVEGUIDES

A rectangular waveguide is a hollow metallic tube with a rectangular cross section.

The conducting walls of the guide confine the electromagnetic fields and thereby
guide the electromagnetic wave. A number of distinct field configurations or modes
can exist in waveguides. When the waves travel longitudinally down the guide, the
plane waves are reflected from wall to wall. This process results in a component of
either electric or magnetic field in the direction of propagation of the resultant wave;
therefore the wave is no longer a transverse electromagnetic (TEM) wave. Figure
4-1-1 shows that any uniform plane wave in a lossless guide may be resolved into
TE and TM waves.

Figure 4-1-1 Plane wave reflected in a

z waveguide. .

It is clear that when the wavelength A is in the direction of propagation of the

incident wave, there will be one component A, in the direction normal to the
reflecting plane and another A, parallel to the plane. These components are
A
A, = -
cos 8
A
Ap = -
sin 8
where 8 = angle of incidence
A = wavelength of the impressed signal in unbounded medium
A plane wave in a waveguide resolves into two components: one standing wave
in the direction normal to the reflecting walls of the guide and one traveling wave in
the direction parallel to the reflecting walls. In lossless waveguides the modes may
be classified as either transverse electric (TE) mode or transverse magnetic (TM)
mode. In rectangular guides the modes are designated TE,, or TM,,. The integer m
104 Microwave Waveguides and Components 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 they direction if the propagation of the wave is
assumed in the positive z direction.

4- 1= 1 Solutions of Wave Equations in Rectangular

Coordinates

As stated previously, there are time-domain and frequency-domain solutions for

each wave equation. However, for the simplicity of the solution to the wave equation
in three dimensions plus a time-varying variable, only the sinusoidal steady-state or
the frequency-domain solution will be given. A rectangular coordinate system is
shown in Fig. 4-1-2.

0 x Figure 4-1-2 Rectangular coordinates.

The electric and magnetic wave equations in frequency domain in Eqs.

(2-1-20) and (2-1-21) are given by
V2E = y2E (4-1-3)
V2H= y2H (4-1-4)
where y = d j w C L ( u+ j w ~ )= a + jp. These are called the vector wave equa-
tions.
Rectangular coordinates are the usual right-hand system. The rectangular com-
ponents of E or H satisfy the complex scalar wave equation or Helmholtz equation

The Helmholtz equation in rectangular coordinates is

This is a linear and inhomogeneous partial differential equation in three di-

mensions. By the method of separation of variables, the solution is assumed in the
form of
$ = 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 Rectangular Waveguides 105

(4- 1-7) yield

-- -.
Substitution of Eq. (4- 1-7) in Eq. (4- 1-6) and division of the resultant by Eq.

1 d2X + 1 d2Y + -
X d x 2 Ydy
1 d2Z
Zdz
7 = y2

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 kl, respectively; then the separation equation
is given by

The general solution of each differential equation in Eq. (4-1-8)

will be in the form of

X = A sin (kxx) + B cos (kxx) (4-1-13)
Y = C sin (k,y) + D cos (k,y) (4-1-14)
Z = E sin (k,z) + F cos (k,z) (4-1-15)
The total solution of the Helmholtz equation in rectangular coordinates is
(I, = [A sin (kxx) + B cos (kxx)][Csin (k,y) + D cos (k,y)]
X [E sin (k,z) + F cos (k,z)] (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, in the guide differs
from the intrinsic propagation constant y of the dielectric. Let

where kc = v k : + k; is usually called the cutoff wave number. For a lossless

dielectric, y 2 = -w2pe. Then
ys = ? d w 2 p ~- kc2 (4-1-18)
There are three cases for the propagation constant y, in the waveguide.

Case I. There will be no wave propagation (evanescence) in the guide if

o f p= ~ kf and y, = 0. This is the critical condition for cutoff propagation. The
cutoff frequency is expressed as
106 Microwave Waveguides and Components Chap. 4

Case 11. The wave will be propagating in the guide if w2pe > k? and

This means that the operating frequency must be above the cutoff frequency in order
for a wave to propagate in the guide.

Case 111. The wave will be attenuated if w 2 p < kz and

This means that if the operating frequency is below the cutoff frequency, the wave
will decay exponentially with respect to a factor of - a,z and there will be no wave
propagation because the propagation constant is a real quantity. Therefore the solu-
tion to the Helmholtz equation in rectangular coordinates is given by
$I = [A sin (k,x) + B cos (k,x)][C sin (k,y) + D cos ( k , y ) ] e - J P ~ z (4-1-22)

4-1-2 TE Modes in Rectangular Waveguides

It has been previously assumed that the waves are propagating in the positive z di-
rection in the waveguide. Figure 4-1-3 shows the coordinates of a rectangular
waveguide.

Figure 4-1-3 Coordinates of a rectan-

gular guide.

The TE, modes in a rectangular guide are characterized by E, = 0. In other

words, the z component of the magnetic field, Hz, must exist in order to have energy
transmission in the guide. Consequently, from a given Helmholtz equation,
V 2 H z = y2H, (4- 1-23)
a solution in the form of
Sec. 4.1 Rectangular Waveguides 107

H= = [A,,, sin (y)

+ [c. (y)
(y)] + D. (y)]e-~
B,,, cos x sin cos
(4- 1-24)
will be determined in accordance with the given boundary conditions, where
k, = m / a and k, = n ~ / are
b replaced. For a lossless dielectric, Maxwell's curl
equations in frequency domain are
V X E = -jopH (4- 1-25)
V X H = joEE (4-1-26)
In rectangular coordinates, their components are

With the substitution d/dz = - jp, and E, = 0, the foregoing equations are sim-
plified to
PgEy = - @ p H x (4-1-33)
P g Ex = W P H , (4- 1-34)

Solving these six equations for Ex, E,, H,, and H, in terms of Hz will give the TE-
108 Microwave Waveguides and Components Chap. 4

mode field equations in rectangular waveguides as

Hz = Eq. (4-1-24) (4- 1-44)

where k,2 = u 2 p e - Pi has been replaced.
Differentiating Eq. (4-1-24) with respect to x and y and then substituting the
results in Eqs. (4- 1-39) through (4- 1-43) yield a set of field equations. The boundary
conditions are applied to the newly found field equations in such a manner that either
the tangent E field or the normal H field vanishes at the surface of the conductor.
Since E, = 0, then dH,/dy = 0 at y = 0, b. Hence C, = 0. Since E, = 0, then
aH,/ax = 0 at x = O, a. Hence A, = 0.
It is generally concluded that the normal derivative of Hz must vanish at the
conducting surfaces-that is,

at the guide walls. Therefore the magnetic field in the positive z direction is given by

H2 = HO, cos (T) (3'

~
)T X
cos
e-jB$
(4- 1-46)

where Hoz is the amplitude constant.

Substitution of Eq . (4- 1-46) in Eqs. (4- 1- 39) through (4- 1-43) yields the TE,
field equations in rectangular waveguides as

E, = E ~ cos
, (--)
mnx
sin (?)e-jplz

m7~x nTY e-jpgz

Ey = Ee sin
cos (-j-)

H, = H ~ sin
, (T)
m x
cos (?)e-jpgz

sin (y)e-~@~'
m x
H~ = M cos (y)
Sec. 4.1 Rectangular Waveguides

wherem = 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 TE,, modes, is given
by

where a and b are in meters. The cutoff frequency, as defined in Eq. (4- 1- 19) for the
TE,, modes, is

The propagation constant (or the phase constant here) P,, as defined in Eq. (4-1- 18),
is expressed by

The phase velocity in the positive z direction for the TE,, modes is shown as

where v,, = 1 / 6 is the phase velocity in an unbounded dielectric.

The characteristic wave impedance of TE,, modes in the guide can be derived
from Eqs. (4- 1-33) and (4- 1-34):

where 77 = is the intrinsic impedance in an unbounded dielectric. The wave-

length A, in the guide for the TE,, modes is given by

where A = vp/ f is the wavelength in an unbounded dielectric.

Since the cutoff frequency shown in Eq. (4-1-54) is a function of the modes
and guide dimensions, the physical size of the waveguide will determine the propa-
gation of the modes. Table 4- 1- 1 tabulates the ratio of cutoff frequency of some
modes with respect to that of the dominant mode in terms of the physical dimension.
Whenever two or more modes have the same cutoff frequency, they are said to
be degenerate modes. In a rectangular guide the corresponding TE,, and TM,,
modes are always degenerate. In a square guide the TE,,, TE,,, TM,,, and TM,,
modes form a foursome of degeneracy. Rectangular guides ordinarily have dimen-
sions of a = 2b ratio. The mode with the lowest cutoff frequency in a particular
110 Microwave Waveguides and Components Chap. 4

TABLE 4-1-1 MODES OF (f,),/f, FOR a 2- b

Modes
f /fro TEi I TE21 TE12 TE22
a/b TElo TEol TMll TE2o TEo2 TM~I TMn TMz~ TE3o

1 1 1 1.414 2 2 2.236 2.236 2.828 3

1.5 1 1.5 1.803 2 3 2.500 3.162 3.606 3
2 1 2 2.236 2 4 2.828 4.123 4.472 3
3 1 3 3.162 2 6 3.606 6.083 6.325 3
P 1 a, w 2 m m m m 3

guide is called the dominant mode. The dominant mode in a rectangular guide wirh
a > b is the TElomode. Each mode has a specific mode pattern (or field pattern).
It is normal for all modes to exist simultaneously in a given waveguide. The sit-
uation 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: TEloin Rectangular Waveguide

An air-filled rectangular waveguide of inside dimensions 7 X 3.5 cm operates in the
dominant TElo mode as shown in Fig. 4-1-4.

Figure 4-14 Rectangular waveguide

for Example 4- 1 - 1.

a. Find the cutoff frequency.

b. Determine the phase velocity of the wave in the guide at a frequency of 3.5 GHz.
c. Determine the guided wavelength at the same frequency.
Solution

3 x lo8
.fC = - =
C

2a 2 X7 X lo-'
= 2.14 GHz
Sec. 4.1 Rectangular Waveguides

4-1-3 TM Modes in Rectangular Waveguides

The TM,, modes in a rectangular guide are characterized by Hz = 0. In other

words, the z component of an electric field E must exist in order to have energy
transmission in the guide. Consequently, the Helmholtz equation for E in the rectan-
gular coordinates is given by
V 2 E , = y2Ez (4- 1-59)
A solution of the Helmholtz equation is in the form of

= [A, sin (7)

+ B, cos (?)][C,, sin ( y )+ D,, cos ( y ) ] e - ~ @ g

which must be determined according to the given boundary conditions. The proce-
dures 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 E, = 0 at x = 0, a, then B, = 0, and for
E, = 0 at y = 0, b, then D, = 0. Thus the solution as shown in Eq. (4-1-60) re-
duces to

E~ = E~~sin (-h)
m x
sin ( y ) e - j ~ g (4-1-61)

where m = 1, 2, 3, . .
n = 1, 2, 3, . .
If either m = 0 or n = 0, the field intensities all vanish. So there is no TMol or
TMlo mode in a rectangular waveguide, which means that TElo is the dominant
mode in a rectangular waveguide for a > b. For Hz = 0, the field equations, after
expanding V x H = j w E,~ are given by

These equations can be solved simultaneously for E x , Ey , Hx , and Hy in terms of E, .

112 Microwave Waveguides and Components Chap. 4

The resultant field equations for TM modes are

E, = Eq. (4- 1-61)

where - o ',UE = -kf is replaced.

Differentiating Eq. (4-1-61) with respect to x or y and substituting the results
in Eqs. (4-1-68) through (4-1-73) yield a new set of field equations. The TM,, mode
field equations in rectangular waveguides are

E, = E ~ cos
, (--)
mrx
sin ( y ) e - j p g z

n r Y e-jpgz
E" = Eo, sin (~ Y
T X
) (-j-)
cos

E, = Eq. (4- 1- 61) (4- 1-76)

H, = H , sin (--)m r x cos ( y ) e - j o g z

H. = H@ cos (- I
mrx
sin ( y ) e - j p g z

Some of the TM-mode characteristic equations are identical to those of the TE

-
modes, but some are different. For convenience, all are shown here:
Sec. 4.1 Rectangular Waveguides

4- 1-4 Power Transmission in Rectangular

Waveguides

The power transmitted through a waveguide and the power loss in the guide walls
can be calculated by means of the complex Poynting theorem described in Chapter
2. It is assumed that the guide is terminated in such a way that there is no reflection
from the receiving end or that the guide is infinitely long compared with the wave-
length. From the Poynting theorem in Section 2-2, the power transmitted through a
guide is given by

For a lossless dielectric, the time-average power flow through a rectangular guide is
given by

Ex EY
where Z, = - = - -
Hy Hx
[ E l 2 = IExI2 + IEy12
l H I 2 = ( H , I 2 + IHyl2

For TE,, modes, the average power transmitted through a rectangular waveguide is
given by

For TM, modes, the average power transmitted through a rectangular waveguide is
given by

Pt, = (4-1-88)

where 7 = is the intrinsic impedance in an unbounded dielectric.

4- 1-5 Power Losses in Rectangular Waveguides

There are two types of power losses in a rectangular waveguide:

1. Losses in the dielectric

2. Losses in the guide walls

First we shall consider power losses caused by dielectric attenuation. In a low-

114 Microwave Waveguides and Components Chap. 4

loss dielectric (that is, u *

p ~ )the
, propagation constant for a plane wave traveling
in an unbounded lossy dielectric is given in Eq. (2-5-20) by
r

The attenuation caused by the low-loss dielectric in the rectangular waveguide for
the TE,, or TM,, modes is given by
U'I
cu, = for TE mode
2v1 - (fClfl2
a = -' I 1 -( f ) for TM mode (4- 1-90a)
2
As f % f, , 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 4 f,, the attenuation constant becomes very large and non-
propagation occurs.
Now we shall consider power losses caused by the guide walls. When the elec-
tric and magnetic intensities propagate through a lossy waveguide, their magnitudes
may be written
1 E 1 = 1 Eo2le-agz (4- 1-91)

where Eo, and Ho, are the field intensities at z = 0. It is interesting to note that, for a
low-loss guide, the time-average power flow decreases proportionally to e - 248'.
Hence

6 P, and 2agz < 1,

For So,,

Finally,

where Pi 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
I
Sec. 4.1 Rectangular Waveguides

where p = resistivity of the conducting wall in ohms-meter

a = conductivity in mnos per meter
6 = skin depth or depth of penetration in meters
The power loss per unit length of guide is obtained by integrating the power
density over the surface of the conductor corresponding to the unit length of the
guide. This is

P L = ; [ I H , ~ ~ ~ S Wlunitlength

where H, is the tangential component of magnetic intensity at the guide walls.

Substitution of Eqs. (4- 1-86) and (4- 1-97) in Eq. (4- 1-95) yields

where

Example 4-1-2: TEloMode in Rectangular Waveguide

An airfilled waveguide with a cross section 2 X 1 cm transports energy in the TElo
mode at the rate of 0.5 hp. The impressed frequency is 30 GHz. What is the peak value
of the electric field occurring in the guide? (Refer to Fig. 4-1-5.)
Solution The field components of the dominant mode TElocan be obtained by sub-
stituting rn = 1 and n = 0 in Eqs. (4-1-47) through (4- 1-52). Then

E, = E* sin (%)e-jflrZ H, = O

E, = 0 Hz = Hoz cos ( ~ ) p " K '

where Z, = opo/pK.

Figure 4-1-5 Rectangular waveguide

for Example 4- 1-2.
Microwave Waveguides and Components Chap. 4

The phase constant P, can be found from

9 x io16 4 x 10-~

The power delivered in the z direction by the guide is

P = ~e[i (I3 x I-I*)] .dx dy u,

= fl[ (E, sin (y)CJP*' U,J x (2 .dx dy u,

E@ sin (y))e*+~~g'u~)]

= /
b o

;E&"/'"Po o o (sin (a))'.. dy

=
1 P*
zE&-ab
'"Po

E, = 53.87 kV/m
The peak value of the electric intensity is 53.87 kV/m.

4- 1-6 Excitations of Modes in Rectangular

Waveguides

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 ex-
cite 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 be-
tween 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 feed-
ing 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 exci-
tation for various modes in rectangular waveguides are shown in Fig. 4-1-6.
In order to excite a TElo 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 TElomode in one direction only. The two antennas are placed a quarter-
wavelength 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 Rectangular Waveguides 117

I+;+
TElo mode

TM,, mode TMzl mode

Figure 4-1-6 Methods of exciting various modes in rectangular waveguides.

~==?gqlp!JT Waveguide

Figure 4-1-7 A method of launching a

TElo mode in one direction only.

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 re-
sulting 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.

-
4-1 7 Characteristics of Standard Rectangular
Waveguides

Rectangular waveguides are commonly used for power transmission at microwave

frequencies. Their physical dimensions are regulated by the frequency of the signal
being transmitted. For example, at X-band frequencies from 8 to 12 GHz, the out-
side dimensions of a rectangular waveguide, designated as EIA WR (90) by the
Electronic Industry Association, are 2.54 cm ( 1 .O in.) wide and 1.27 cm (0.5 in.)
118 Microwave Waveguides and Components Chap. 4

TABLE 4-1-7 CHARACTERISTICS OF STANDARD RECTANGULAR WAVEGUIDES

Physical dimensions Cutoff Recommended

. frequency for frequency
EIAa Inside, in cm (in.) Outside, in cm (in.) air-filled range for
designation waveguide TElo mode
WRb( 1 Width Height Width Height in GHz in GHZ
-
2300 58.420 29.210 59.055 29.845 0.257 0.32-0.49
(23.000) (11.500) (23.250) (1 1.750)
2100 53.340 26.670 53.973 27.305 0.281 0.35-0.53
(21.000) (10.500) (21.250) (10.750)
1800 45.720 22.860 46.350 23.495 0.328 0.41-0.62
(18.000) (9.000) (18.250) (9.250)
1500 38.100 19.050 38.735 19.685 0.394 0.49-0.75
(15.000) (7.500) (15.250) (7.750)
1150 29.210 14.605 29.845 15.240 0.514 0.64-0.98
(11.500) (5.750) (11.750) (6.000)
975 24.765 12.383 25.400 13.018 0.606 0.76-1.15
(9.750) (4.875) (10.000) (5.125)
770 19.550 9.779 20.244 10.414 0.767 0.96- 1.46
(7.700) (3.850) (7.970) (4.100)
650 16.510 8.255 16.916 8.661 0.909 1.14-1.73
(6.500) (3.250) (6.660) (3.410)
5 10 12.954 6.477 13.360 6.883 1.158 1.45-2.20
(5.100) (2.500) (5.260) (2.710)
430 10.922 5.461 11.328 5.867 1.373 1.72-2.61
(4.300) (2.150) (4.460) (2.310)
340 8.636 4.318 9.042 4.724 1.737 2.17-3.30
(3.400) (1.700) (3.560) (1.860)
284 7.214 3.404 7.620 3.810 2.079 2.60-3.95
(2.840) (1.340) (3.000) (1.500)
229 5.817 2.908 6.142 3.233 2.579 3.22-4.90
(2.290) (1.145) (2.418) (1.273)
187 4.755 2.215 5.080 2.540 3.155 3.94-5.99
(1.872) (0.872) (2.000) (1.000)
159 4.039 2.019 4.364 2.344 3.714 4.64-7.05
(1.590) (0.795) (1.718) (0.923)
137 3.485 1.580 3.810 1.905 4.304 5.38-8.17
(1.372) (0.622) (1.500) (0.750)
112 2.850 1.262 3.175 1.588 5.263 6.57-9.99
(1.122) (0.497) (1.250) (0.625)
90 2.286 1.016 2.540 1.270 6.562 8.20-12.50
(0.900) (0.400) (1.000) (0.500)
75 1.905 0.953 2.159 1.207 7.874 9.84-15.00
(0.750) (0.375) (0.850) (0.475)
62 1.580 0.790 1.783 0.993 9.494 11.90-18.00
(0.622) (0.31 1) (0.702) (0.391)
51 1.295 0.648 1.499 0.851 11.583 14.50-22.00
(0.510) (0.255) (0.590) (0.335)
42 1.067 0.432 1.270 0.635 14.058 17.60-26.70
(0.420) (0.170) (0.500) (0.250)

a Electronic Industry Association

Rectangular Waveguide
Sec. 4.2 Circular Waveguides 119

TABLE 4-1-7 CHARACTERISTICS OF STANDARD RECTANGULAR WAVEGUIDES (Cont.)

I Physical dimensions I Cutoff

frequency for
Recommended
frequency
EIAa Inside, in cm (in.) Outside, in cm (in.) air-filled range for
designation waveguide TElomode
WRb( 1 Width Height Width Height in GHz in GHZ
0.864 0.432 1.067 0.635 21.70-33.00
(0.340) (0.170) (0.420) (0.250)
0.71 1 0.356 0.914 0.559 26.40-40.00
(0.280) (0.140) (0.360) (0.220)
0.569 0.284 0.772 0.488 32.90-50.10
(0.224) (0.112) (0.304) (0.192)
0.478 0.239 0.681 0.442 39.20-59.60
(0.188) (0.094) (0.268) (0.174)
0.376 0.188 0.579 0.391 49.80-75.80
(0.148) (0.074) (0.228) (0.154)
0.310 0.155 0.513 0.358 60.50-91.90
(0.122) (0.061) (0.202) (0.141)
0.254 0.127 0.457 0.330 73.80-1 12.00
(0.100) (0.050) (0.180) (0.130)
0.203 0.102 0.406 0.305 92.20-140.00
(0.080) (0.040) (0.160) (0.120)
0.165 0.084 0.343 0.262 114.W173.00
(0.065) (0.033) (0.135) (0.103)
0.130 0.066 0.257 0.193 145.W220.00
(0.051) (0.026) (0.101) (0.076)
0.109 0.056 0.21 1 0.157 172.W261.00
(0.043) (0.022) (0.083) (0.062)
0.086 0.043 0.163 0.119 217.W333.00
(0.034) (0.017) (0.064) (0.047)

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.

4.2 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 mag-
netic (TM) mode. Several other types of waveguides, such as elliptical and reentrant
guides, also propagate electromagnetic waves.

4-2-1 Solutions of Wave Equations in Cylindrical

Coordinates

As described in Section 4-1 for rectangular waveguides, only a sinusoidal steady-

state or frequency-domain solution will be attempted for circular waveguides. A
cylindrical coordinate system is shown in Fig. 4-2- 1.
Microwave Waveguides and Components Chap. 4

Figure 4-2-1 Cylindrical coordinates.

The scalar Helmholtz equation in cylindrical coordinates is given by

Using the method of separation of variables, the solution is assumed in the form of
= R(r)@(4)Z(z) (4-2-2)
where R ( r ) = a function of the r coordinate only
@(+) = a function of the 4 coordinate only
Z(z) = a function of the z coordinate only

Substitution of Eq. (4-2-2) in (4-2- 1) and division of the resultant by (4-2-2) yield

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 i :

The solutions of this equation are given by

Z = Ae-Yn' + BeYn'
where y, = propagation constant of the wave in the guide.
Inserting yi for the third term in the left-hand side of Eq. (4-2-3) and multi-
plying the resultant by r2 yield

The second term is a function of 4 only; hence equating the second term to a con-
Sec. 4.2 Circular Waveguides

stant ( - n 2 ) yields

The solution of this equation is also a harmonic function:

@ = A, sin (n+) + B, cos (n+) (4-2-8)
Replacing the @ term by ( - n 2 ) in Eq. (4-2-6) and multiplying through by R, we
have

This is Bessel's equation of order n in which

This equation is called the characteristic equation of Bessel's equation. For a lossless
guide, the characteristic equation reduces to
p, = t-VW2/L€ - kt
The solutions of Bessel's equation are

where Jn(kcr)is the nth-order Bessel function of the first kind, representing a stand-
ing wave of cos (k,r) for r < a as shown in Fig. 4-2-2. Nn(kcr)is the n th-order Bes-
sel function of the second kind, representing a standing wave of sin (k,r) for r > a
as shown in Fig. 4-2-3.
Therefore the total solution of the Helmholtz equation in cylindrical coordi-
nates is given by
* = [CnJn(kcr) + D , Nn(kcr)][Ansin (n+) + B, cos (n+)]e2~BnZ(4-2-13)

Figure 4-2-2 Bessel functions of the first kind

Microwave Waveguides and Components Chap. 4

c nNt , , ( k , r )
A r g ~ ~ ~ n of

Figure 4-2-3 Bessel functions of the second kind.

At r = 0, however, k,r = 0; then the function N,, approaches infinity, so D, = 0.

This means that at r = 0 on the z axis, the field must be finite. Also, by use of
trigonometric manipulations, the two sinusoidal terms become

A,, sin ( n 4 ) + B,, cos (4)= VXG-Z cos [nm + tan-'(+)]

= F, cos ( n 4 ) (4-2- 14)

Finally, the solution of the Helmholtz equation is reduced to
q = ?O J,(k,r) cos (n4)e-jPg' (4-2- 15)

4-2-2 TE Modes in Circular Waveguides

It is commonly assumed that the waves in a circular waveguide are propagating in

the positive z direction. Figure 4-2-4 shows the coordinates of a circular guide.
The TE, modes in the circular guide are characterized by E, = 0.This means

Figure 4-2-4 Coordinates of a circular

waveguide.
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 H z = y2H2 (4-2-16)
Its solution is given in Eq. (4-2-15) by
Hz = HozJ,(k,r) cos (n+)e-jPg2 (4-2-17)

which is subject to the given boundary conditions.

For a lossless dielectric, Maxwell's curl equations in frequency domain are
given by
V xE = - jwpH (4-2-18)
V X H = jweE (4-2-19)

In cylindrical coordinates, their components are expressed as

When the differentiation a/dz is replaced by ( - jp,) and the z component of elec-
tric field E, by zero, the TE-mode equations in terms of Hz in a circular waveguide
are expressed as

E = j w p 1 dH,
k,2 r a+
Microwave Waveguides and Components Chap. 4

Hz = Ho,Jn(k,r)cos (n4)e-jPnz (4-2-31)

where k,2 = w 2 p c - Pi has been replaced.
+
The boundary conditions require that the component of the electric field E+ ,
which is tangential to the inner surface of the circular waveguide at r = a , must van-
ish or that the r component of the magnetic field H,, which is normal to the inner
surface of r = a , must vanish. Consequently

This requirement is equivalent to that expressed in Eq. (4-2-17):

= Ho,JA(k,a) cos (n+)e-"nz = 0 (4-2-32)

r=a

Hence
JA(k,a) = 0
where JL indicates the derivative of J, .
Since the Jn are oscillatory functions, the JA (k,a) are also oscillatory func-
tions. An infinite sequence of values of (k,a) satisfies Eq. (4-2-32).These points,
the roots of Eq. (4-2-32), correspond to the maxima and minima of the curves
J;(k,a), as shown in Fig. 4-2-2. Table 4-2-1 tabulates a few roots of JA(k,a) for
some lower-order n.

TABLE 4-24 pth ZEROS OF JA(K,a) FOR TE, MODES

The permissible values of k, can be written

kc = x:,
-
a
Substitution of Eq. (4-2-17) in Eqs. (4-2-26) through (4-2-31) yields the complete
field equations of the TEnpmodes in circular waveguides:
Sec. 4.2 Circular Waveguides 125

XApr
E+ = Eo+J:(--) cos (n+)e-Jp~~

Ear XLpr
H+ = - Jn(--) sin (n+)e-Jpgz
2,
XLpr
Hz = HozJn(T) cos (n+)e-jp~~

where Z, = E,/H+ = - E+/H, has been replaced for the wave impedance in the
guide and where n = 0, 1, 2, 3, . . . and p = 1, 2, 3, 4, . . . .
The first subscript n represents the number of full cycles of field variation in
one revolution through 27r rad of 4. The second subscript p indicates the number of
zeros of E+-that is, JL(XLpr/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):

The cutoff wave number of a mode is that for which the mode propagation constant
vanishes. Hence

kc = G!= Wc& (4-2-42)

a
The cutoff frequency for TE modes in a circular guide is then given by

and the phase velocity for TE modes is

where q = l / a = c / G is the phase velocity in an unbounded dielectric.

The wavelength and wave impedance for TE modes in a circular guide are
given, respectively, by

and
126 Microwave Waveguides and Components Chap. 4

where A = - -
- wavelength in an unbounded dielectric

'1 = &
f
= intrinsic impedance in an unbounded dielectric

Example 4-2-1: TE Mode in Circular Waveguide

A TE,, mode is propagating through a circular waveguide. The radius of the guide is
5 cm, and the guide contains an air dielectric (refer to Fig. 4-2-5).

Figure 4-2-5 Diagram for Example

4-2- 1.

a. Determine the cutoff frequency.

b. Determine the wavelength A, in the guide for an operating frequency of 3 GHz.
c . Determine the wave impedance Z, in the guide.
Solution

a. From Table 4-2- 1 for TEll mode, n = 1, p = 1 , and XiI = 1.841 = k,a. The
cutoff wave number is

The cutoff frequency is

b. The phase constant in the guide is

The wavelength in the guide is

Sec. 4.2 Circular Waveguides

c. The wave impedance in the guide is

z = -upo
=
(257 x 3 3 109)(4~
x lo-')
= 465 R
RP" 50.9

4-2-3 TM Modes in Circular Waveguides

The TM, modes in a circular guide are characterized by Hz = 0. However, the z

component of the electric field Ez must exist in order to have energy transmission in
the guide. Consequently, the Helmholtz equation for E, in a circular waveguide is
given by
V2E, = y2Ez (4-2-47)
Its solution is given in Eq. (4-2- 15) by
E, = EozJ,(k,r) cos (n+)e-JPgz
which is subject to the given boundary conditions.
The boundary condition requires that the tangential component of electric field
E, at r = a vanishes. Consequently,

Since J,(k,r) are oscillatory functions, as shown in Fig. 4-2-2, there are infinite
numbers of roots of J,(k,r). Table 4-2-2 tabulates a few of them for some lower-
order n.

TABLE 4-2-2 pth ZEROS OF Jn(K,a) FOR TM, MODES

For Hz = 0 and a/dz = - jp, , the field equations in the circular guide, after
expanding V X E = - j o p H and V X H = j w E, ~ are given by
Microwave Waveguides and Components Chap. 4

where kf = 0 2 p e - Pi has been replaced.

Differentiation of Eq. (4-2-48) with respect to z and substitution of the result
in Eqs. (4-2-50) through (4-2-55) yield the field equations of TM, modes in a circu-
lar waveguide:

Er = EaJ:($) cos (n$)e-jpaz

Xnpr
E, = EozJ,,(-;;) cos (n4)e-jpaZ (4-2-58)

Eo+
H. = -Jn
ZR
(a)XnPr
sin (n$)e-JPnz

where Z, = E,/H+ = - E4/Hr = P , / ( ~ E )and kc = Xnp/ahave been replaced and

wheren = 0, 1, 2, 3, . . . andp = 1, 2, 3, 4, . . . .
Some of the TM-mode characteristic equations in the circular guide are identi-
cal to those of the TE mode, but some are different. For convenience, all are shown
here:
I

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 TEll that has the smallest value of the
product, k,a = 1.841, as shown in Tables 4-2- 1 and 4-2-2.

Example 4-2-2: Wave Propagation in Circular Waveguide

An air-filled circular waveguide has a radius of 2 cm and is to carry energy at a fre-
quency of 10 GHz. Find all the TE, and TM, modes for which energy transmission is
possible.
Solution Since the physical dimension of the guide and the frequency of the wave re-
main constant, the product of (k,a) is also constant. Thus

Any mode having a product of (k,a) less than or equal to 4.18 will propagate the wave
with a frequency of 10 GHz. This is

The possible modes are

TEll(1.841) TMol(2.405)

4-2-4 TEM Modes in Circular Waveguides

The transverse electric and transverse magnetic (TEM) modes or transmission-line

modes are characterized by
E, = H, = 0
This means that the electric and magnetic fields are completely transverse to the di-
rection 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 be-
tween 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

become
Microwave Waveguides and Components Chap. 4

Figure 4-2-6 Coordinates of a coaxial

line.

where alar = - jp, and E, = H, = 0 are replaced.

Substitution of Eq. (4-2-71) in (4-2-73) yields the propagation constant of the
TEM mode in a coaxial line:
p, = OG (4-2-76)
which is the phase constant of the wave in a lossless transmission line with a dielec-
tric.
In comparing the preceding equation with the characteristic equation of the
Helmholtz equation in cylindrical coordinates as given in Eq . (4-2- 11) by
p, = V&Fx (4-2-77)
it is evident that
kc = 0
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

which is the velocity of light in an unbounded dielectric.

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
7

1)(TEM) = dE E
(4-2-80)

which is the wave impedance of a lossless transmission line in a dielectric.

Ampttre's law states that the line integral of H about any closed path is exactly
equal to the current enclosed by that path. This is

f
H .d t = I= loe-JD8z = 2.rrrH4 (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 two-
conductor 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 fol-
lows:

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.

4-2-5 Power Transmission in Circular Waveguides or

Coaxial Lines

In general, the power transmitted through circular waveguides and coaxial lines can
be calculated by means of the complex Poynting theorem described in Section 2-2.
For a lossless dielectric, the time-average power transmitted through a circular guide
can be given by

Ptr= - 1' + I I:], dr dm (4-2-82)

228 0

Er E4
where Z, = - = - - = wave impedance in the guide
Hm Hr
a = radius of the circular guide
Substitution of 2, for a particular mode in Eq. (4-2-82) yields the power transmitted
by that mode through the guide.
132 Microwave Waveguides and Components Chap. 4

For TE, modes, the average power transmitted through a circular guide is
given by

where 77 = a is the intrinsic impedance in an unbounded dielectric.

For TM, modes, the average power transmitted through a circular guide is
given by

For TEM modes in coaxial lines, the average power transmitted through a
coaxial line or two-open-wire line is given by

If the current carried by the center conductor of a coaxial line is assumed to be

the magnetic intensity induced by the current around the center conductor is given
by Ampkre's law as

The potential rise from the outer conductor to the center conductor is given by

The characteristic impedance of a coaxial line is

where 77 = a is the intrinsic impedance in an unbounded dielectric.

The power transmitted by TEM modes in a coaxial line can be expressed from
Eq. (4-2-86) as

Substitution of I V, 1 from Eq. (4-2-89) into Eq. (4-2-91) yields

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

4-2-6 Power Losses in Circular Waveguides

or Coaxial Lines

The theory and equations derived in Section 4-1-5 for TE and TM modes in rectan-
gular waveguides are applicable to TE and TM modes in circular guides. The power
losses for the TEM mode in coaxial lines can be computed from transmission-line
theory by means of

where PL = power loss per unit length

Pu = transmitted power
a = attenuation constant

- -
For a low-loss conductor, the attenuation constant is given by

4-2-7 Excitations of Modes in Circular Waveguides

As described earlier, TE modes have no z component of an electric field, and TM

modes have no z component of magnetic intensity. If a device is inserted in a circu-
lar guide in such a way that it excites only a z component of electric intensity, the
wave propagating through the guide will be the TM mode; on the other hand, if a
device is placed in a circular guide in such a manner that only the z component of
magnetic intensity exists, the traveling wave will be the TE mode. The methods of
excitation for various modes in circular waveguides are shown in Fig. 4-2-7.

Short-circuited Coaxial line

probe

Figure 4-2-7 Methods of exciting vari-

TMol mode TEll mode ous modes in circular wavquides.

A common way to excite TM modes in a circular guide is by a coaxial line as

shown in Fig. 4-2-8. At the end of the coaxial line a large magnetic intensity exists
in the 4 direction of wave propagation. The magnetic field from the coaxial line will
excite the TM modes in the guide. However, when the guide is connected to the
source by a coaxial line, a discontinuity problem at the junction will increase the
standing-wave ratio on the line and eventually decrease the power transmission. It is
often necessary to place a turning device around the junction in order to suppress the
reflection.
Microwave Waveguides and Components Chap. 4

Single stub Single stub

(a) (b)
Figure 4-2-8 Methods of exciting TM modes in a circular waveguide. (a) Coaxial
line with a slotted waveguide. (b) Coaxial line in series with a circular waveguide.

TABLE 4.2.8 CHARACTERISTICS OF STANDARD CIRCULAR WAVEGUIDES

EIAa Cutoff frequency Recommended

designation Inside diameter for air-filled frequency range
web( ) in cm (in.) waveguide in GHz for TE, mode in GHz

992 25.184 (9.915) 0.698 0.80-1.10

847 21.514 (8.470) 0.817 0.94-1.29
724 18.377 (7.235) 0.957 1.10-1.51
618 15.700 (6.181) 1.120 1.29- 1.76
528 13.411 (5.280) 1.311 1.51-2.07
45 1 11.458 (4.5 1 1) 1.534 1.76-2.42
385 9.787 (3.853) 1.796 2.07-2.83
329 8.362 (3.292) 2.102 2.42-3.31
28 1 7.142 (2.812) 2.461 2.83-3.88
240 6.104 (2.403) 2.880 3.31-4.54
205 5.199 (2.047) 3.381 3.89-5.33
175 4.445 (1.750) 3.955 4.54-6.23
150 3.810 (1.500) 4.614 5.30-7.27
128 3.254 (1.281) 5.402 6.21-8.51
109 2.779 (1.094) 6.326 7.27-9.97
94 2.383 (0.938) 7.377 8.49-11.60
80 2.024 (0.797) 8.685 9.97-13.70
69 1.748 (0.688) 10.057 11.60-15.90
59 1.509 (0.594) 11.649 13.40-18.40
50 1.270 (0.500) 13.842 15.90-21.80
44 1.113 (0.438) 15.794 18.20-24.90
38 0.953 (0.375) 18.446 21.20-29.10
33 0.833 (0.328) 21.103 24.30-33.20
28 0.714 (0.281) 24.620 28.30-38.80
25 0.635 (0.250) 27.683 31.80-43.60
22 0.556 (0.219) 31.617 36.40-49.80
19 0.478 (0.188) 36.776 42.40-58.10
17 0.437 (0.172) 40.227 46.30-63.50
14 0.358 (0.141) 49.103 56.60-77.50
13 0.318 (0.125) 55.280 63.50-87.20
II 0.277 (0.109) 63.462 72.70-99.70
9 0.239 (0.094) 73.552 84.80-1 16.00

"Electronic Industry Association

Circular Waveguide
Sec. 4.3 Microwave Cavities

4-2-8 Characteristics of Standard Circular

Waveguides

The inner diameter of a circular waveguide is regulated by the frequency of the sig-
nal being transmitted. For example: at X-band frequencies from 8 to 12 GHz, the in-
ner 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 characteris-
tics of the standard circular waveguides.

4-3 MICRO WAVE CAVITIES

In general, a cavity resonator is a metallic enclosure that confines the electromag-

netic energy. The stored electric and magnetic energies inside the cavity determine
its equivalent inductance and capacitance. The energy dissipated by the finite con-
ductivity of the cavity walls determines its equivalent resistance. In practice, the
rectangular-cavity resonator, circular-cavity resonator, and reentrant-cavity res-
onator 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.

4-3-1 Rectangular=CavityResonator

The electromagnetic field inside the cavity should satisfy Maxwell's equations, sub-
ject 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

Figure 4-3-1 Coordinates of a rectan-

gular cavity.
136 Microwave Waveguides and Components Chap. 4

walls. These functions can be found if

H;= H , cos (--)

mnx
cos (y)("7)
sin (TE,,) (4-3-1)

where m = 0, 1, 2, 3, . . . representsthe number of the half-wave periodicity in

the x direction
n = 0, 1, 2, 3, . . . represents the number of the half-wave periodicity in
the y direction
p = 1, 2, 3, 4, . . . represents the number of the half-wave periodicity in
the z direction
and
I?, = & sin
m x
sin (y)cm (F) (TMrnoP) (4-3-2)

wherem = 1, 2, 3, 4, . .
n = I , 2, 3, 4, . .
p = 0, 1 , 2 , 3 , . .
The separation equation for both TE and TM modes is given by

For a lossless dielectric, k 2 = 0 2 p e ;therefore, the resonant frequency is expressed

For a > b < d, the dominant mode is the T E ~ omode.I

In general, a straight-wire probe inserted at the position of maximum electric
intensity is used to excite a desired mode, and the loop coupling placed at the posi-
tion of maximum magnetic intensity is utilized to launch a specific mode. Figure
4-3-2 shows the methods of excitation for the rectangular resonator. The maximum
amplitude of the standing wave occurs when the frequency of the impressed signal is
equal to the resonant frequency.

Figure 4-3-2 Methods of exciting wave

modes in a resonator.

4-3-2 Circular-Cavity Resonator

and Semicircular-Cavity Resonator

Circular-cavity resonator. A circular-cavity resonator is a circular wave-

guide with two ends closed by a metal wall (see Fig. 4-3-3). The wave function in
the circular resonator should satisfy Maxwell's equations, subject to the same
Sec. 4.3 Microwave Cavities

Figure 4-3-3 Coordinates of a circular

x resonator

boundary conditions described for a rectangular-cavity resonator. It is merely neces-

sary to choose the harmonic functions in z to satisfy the boundary conditions at the
remaining two end walls. These can be achieved if

H, = H ~J,, (--)
XApr
cos (n4) sin (7) (TE,,~~) (4-3-5)

where n = 0 , 1 , 2, 3, . . . is the number of the periodicity in the 4 direction

p = 1, 2, 3, 4, . . . is the number of zeros of the field in the radial direction
q = 1, 2, 3, 4 , . . . is the number of half-waves in the axial direction
J, = Bessell's function of the first kind
Ho, = amplitude of the magnetic field
and

Ez = Eo,J, "P- cos (n+) cos (4-3-6)

("a
where n = 0 , 1 , 2, 3, . . .
p= l,2,3,4,. . .
q=0,1,2,3, . . .
Eo, = amplitude of the electric field
The separation equations for TE and TM modes are given by

2 = (7)
+
2
(TE mode)

2 = +( 7 (TM mode)

Substitution of k 2 = 02pe in Eqs. (4-3-7) and (4-3-8) yields the resonant frequen-
cies for TE and TM modes, respectively, as
Microwave Waveguides and Components Chap. 4

fr =
2 T G
der (yr+ (TM)

It is interesting to note that the TMllomode is dominant where 2a > d, and that the
TElllmode is dominant when d 1 2a.

Semicircular-cavity resonator. A semicircular-cavity resonator is shown

in Fig. 4-3-4. The wave function of the TE,, mode in the semicircular resonator can
be written
X&r
H , = H , J . ( ~ ) cos (n+) sin (y) (TE mode)

where n = 0, 1 , 2, 3, . . .
p = l , 2 , 3 , 4, . . .
q = 1 , 2 , 3 , 4, . . .
a = radius of the semicircular-cavity resonator
d = length of the resonator

4 Figure 4-3-4 Semicircular resonator.

The wave function of the TM,, mode in the semicircular-cavity resonator can be
written

E, = E ~ J . (5) sin (n+) cos (5,z) (TM mode)

where n = 1, 2, 3, 4, . . .
p = l , 2 , 3 , 4, . . .
q = o , 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:

fr = 277‘2
1
GJ(xa,2 + (7)' (TE,, mode) (4-3- 13)
Sec. 4.3 Microwave Cavities 139

fr =
1
2 m G
J(Xnp)' + (yr (TM,, mode) (4-3- 14)

However, the values of the subscripts n, p, and q differ from those for the circular-
cavity resonator. Also, it must be emphasized that the TElll mode is dominant if
d > a and that the TMllo mode is dominant if d < a.

4-34 Q Factor of a Cavity Resonator

The quality factor Q is a measure of the frequency selectivity of a resonant or an-

tiresonant circuit, and it is defined as
maximum energy stored - wW
-- (4-3- 15)
Q = 2~
energy dissipated per cycle P
where W is the maximum stored energy and P is 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:

where ( E 1 and I H I 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

where HI is the peak value of the tangential magnetic intensity and R, is the surface
resistance of the resonator.
Substitution of Eqs. (4-3-16) and (4-3-17) in Eq. (4-3-15) yields

Since the peak value of the magnetic intensity is related to its tangential and normal
components by
1 1 ~ +(
~ = IH,I2 ~ ~ 1 ~
where H,, is the peak value of the normal magnetic intensity, the value of I HIl2 at the
resonator walls is approximately twice the value of I H l2 averaged over the volume.
So the Q of a cavity resonator as shown in Eq. (4-3-18) can be expressed approxi-
mately by
wp (volume)
= 2R,(surface areas)
140 Microwave Waveguides and Components Chap. 4

An unloaded resonator can be represented by either a series or a parallel resonant

circuit. The resonant frequency and the unloaded Qo of a cavity resonator are

If the cavity is coupled by means of an ideal N: 1 transformer and a series inductance

L, to a generator having internal impedance Z,, then the coupling circuit and its
equivalent are as shown in Fig. 4-3-5.

Figure 4-3-5 Cavity coupled to a generator. (a) Coupling circuit. (b) Equivalent
circuit.

The loaded Qe of the system is given by

The coupling coefficient of the system is defined as

and the loaded Qe would become

Rearrangement of Eq. (4-3-24) yields

where QeXt= Qo/K = ooL/(KR) is the external Q.

There are three types of coupling coefficients:

1. Critical coupling: If the resonator is matched to the generator, then

K = l (4-3-26)
Sec. 4.4 Microwave Hybrid Circuits 141

The loaded Qe is given by

Q
e
--l 2 Q e x t = ~ Q o (4-3-27)
2. Overcoupling: If K > 1, the cavity terminals are at a voltage maximum in the
input line at resonance. The normalized impedance at the voltage maximum is
the standing-wave ratio p . That is
K = p (4-3-28)
The loaded Qe is given by

3. Undercoupling: If K < 1, the cavity terminals are at a voltage minimum and

the input terminal impedance is equal to the reciprocal of the standing-wave ra-
tio. That is,

The loaded Qt is given by

The relationship of the coupling coefficient K and the standing-wave ratio is

shown in Fig. 4-3-6.

Figure 4-3-6 Coupling coefficient ver-

Coupling coefficient K sus standing-wave ratio.

4-4 MICROWAYE HYBRID CIRCUITS

A microwave circuit ordinarily consists of several microwave devices connected in

some way to achieve the desired transmission of a microwave signal. The intercon-
nection of two or more microwave devices may be regarded as a microwave junc-
tion. Commonly used microwave junctions include such waveguide tees as the E-
142 Microwave Waveguides and Components Chap. 4

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: []: :[k: I:;[ VI = h111l+ h12V2

I2 = h2111+ h22V2
(4- 4- 1)
(4- 4- 2)

Y parameters: I:[ [i: I:;[ 11 = YIIVI+

1 2 = y21 Vl +
~ 1 2 ~ 2
y22 Vl
(4-4-3)
(4-4-4)

Collinear arms

Port 1

(a) E-plane tee (b) H-plane tee

(d) Hybrid ring

(c) Magic tee

Port 1 Port 3

Coupling holes
(e) Directional coupler (9 Circulator
Figure 4-4-1 Microwave hybrids. (a) E-plane tee. (b) H-plane tee. (c) Magic tee.
(d) Hybrid ring. (e) Directional coupler. (f) Circulator.
Sec. 4.4 Microwave Hybrid Circuits

- - - Figure 4-4-2 Two-port network.

z parameters: [] [
Vl

V2
=
=
Zll

~ 2 1 $1 v
1

V2
= Zll1, + 212 1 2
= ,721 11+ ,722 1 2
(4-4-5)
(4-4-6)

V l = A B VI = AV2 - BI2 (4-4-7)

A BCD parameters:
[ I l ] = [ C Dl[-:] I1=CV2-Dl2 (4-4-8)
All these network parameters relate total voltages and total currents at each of the
two ports. For instance,

(short circuit)

h12 =
v2
I l,=O
(open circuit)

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 frequen-
cies.
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- 1la)
(4-4- 1lb)
Figure 4-4-3 shows the S parameters of a two-port network.
II

Figure 4-4-3 Two-port network.

144 Microwave Waveguides and Components Chap. 4

4-4-f Waveguide Tees

As noted, waveguide tees may consist of the E-plane tee, H-plane tee, magic tee,
hybrid rings, corners, bends, and twists. All such waveguide components are dis-
cussed in this section.

Tee junctions. In microwave circuits a waveguide or coaxial-line junction

with three independent ports is commonly referred to as a tee junction. From the S -
parameter theory of a microwave junction it is evident that a tee junction should be
characterized by a matrix of third order containing nine elements, six of which
should be independent. The characteristics of a three-port junction can be explained
by three theorems of the tee junction. These theorems are derived from the equiva-
len t-circuit representation of the tee junction. Their statements follow
1. A short circuit may always be placed in one of the arms of a three-port junc-
tion 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 E-plane tee and H-plane tee are described below.

E-plane tee (series tee). An E-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 transmis-
sion 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, S l l , Sz2,
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 1 and port 2 of the collinear arm
will be in opposite phase and in the same magnitude. Therefore

It should be noted that Eq. (4-4-12) does not mean that S13is always positive and S23

Port 3

Port 2
Side arm

.,
Collinear
arms
Figure 4-4-4 E-plane tee
Sec. 4.4 Microwave Hybrid Circuits

Port 1
Out '7z2
F' Figure 4-4-5 Two-way transmission of
E-plane tee. (a) Input through main arm.
(b) (b) Input from side arm.

is always negative. The negative sign merely means that Sl3 and S23 have opposite
signs. For a matched junction, the S matrix is given by
0 s12 s
13
(4-4- 13)

From the symmetry property of S matrix, the symmetric terms in Eq. (4-4-13) are
equal and they are

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
S I I S +~ S 2 1 S f 2 + ss31S$= 0 (4-4- 15)
Hence

This means that either S 1 3or S f 3 , 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,

Substitution of Eq. (4-4- 14) in (4-4- 17) results in

IS12I2 = 1 - I (* = 1 - I l2
s
13 S23 (4-4-20)
Equations (4-4-19) and (4-4-20) are contradictory, for if s13= 0, then S23 is also
1
146 Microwave Waveguides and Components Chap. 4

zero and thus Eq. (4-4- 19) is false. In a similar fashion, if S23 = 0, then S13 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 diago-
nal elements of the S matrix of a tee junction are not all zeros.
In general, when an E-plane tee is constructed of an empty waveguide, it is
poorly matched at the tee junction. Hence SV # 0 if i = j. However, since the
collinear arm is usually symmetric about the side arm, ( S13 ( = ( S 2 3 I and S11 = SZZ.
Then the S matrix can be simplified to

(4-4-21)
s13 I S33

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 2

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

4-4-2 Magic Tees (Hybrid Tees)

A magic tee is a combination of the E-plane tee and H-plane tee (refer to Fig.
4-4-7). The magic tee has several characteristics:

Port 3

Figure 4-4-7 Magic tee.

Sec. 4.4 Microwave Hybrid Circuits 147

1. If two waves of equal magnitude and the same phase are fed into port 1 and
port 2, the output will be zero at port 3 and additive at port 4.
2. If a wave is fed into port 4 (the H arm), it will be divided equally between port
1 and port 2 of the collinear arms and will not appear at port 3 (the E arm).
3. If a wave is fed into port 3 (the E arm), it will produce an output of equal mag-
nitude and opposite phase at port 1 and port 2. The output at port 4 is zero.
That is, S 4 3 = S 3 4 = 0.
4. If a wave is fed into one of the collinear arms at port 1 or port 2, it will not
appear in the other collinear arm at port 2 or port 1 because the E arm causes a
phase delay while the H arm causes a phase advance. That is, Slz = Szl = 0.

Therefore the S matrix of a magic tee can be expressed as

The magic tee is commonly used for mixing, duplexing, and impedance measure-
ments. 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 transmit-
ters 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 oppo-
site 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 to-
gether; so double output power at port 2 is obtained for the antenna as shown in Fig.
4-4-8.
To antenna

Port Figure 4-4-8 Magic tee-coupled trans-

mitters to antenna.

4-4-3 Hybrid Rings (Rat-Race Circuits)

A hybrid ring consists of an annular line of proper electrical length to sustain stand-
ing 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.
Microwave Waveguides and Components Chap. 4

3xg
-
4

Port 2 Figure 4-4-9 Hybrid ring.

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

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.

4-4-4 Waveguide Corners, Bends, and Twists

The waveguide corner, bend, and twist are shown in Fig. 4-4-10. These waveguide
components are normally used to change the direction of the guide through an arbi-
trary angle.
In order to minimize reflections from the discontinuities, it is desirable to have
the mean length L between continuities equal to an odd number of quarter-wave-
lengths. That is,

where n = 0, 1, 2, 3, . . . , and A, is the wavelength in the waveguide. If the mean

length L is an odd number of quarter wavelengths, the reflected waves from both
ends of the waveguide section are completely canceled. For the waveguide bend, the
minimum radius of curvature for a small reflection is given by Southworth [2] as
R = 1.5b for an E bend (4-4-26)
Sec. 4.5 Directional Couplers 149

R = 1.5~ 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.

Figure 4-4-10 Waveguide corner, bend, and twist. (a) E-plane corner.
(b) H-plane corner. (c) Bend. (d) Continuous twist.

4-5 DIRECTIONAL COUPLERS

A directional coupler is a four-port waveguide junction as shown in Fig. 4-5-1. It

consists of a primary waveguide 1-2 and a secondary waveguide 3-4. When all
ports are terminated in their characteristic impedances, there is free transmission of
power, without reflection, between port 1 and port 2, and there is no transmission of
power between port 1 and port 3 or between port 2 and port 4 because no coupling
exists between these two pairs of ports. The degree of coupling between port 1 and
port 4 and between port 2 and port 3 depends on the structure of the coupler.
The characteristics of a directional coupler can be expressed in terms of its
coupling factor and its directivity. Assuming that the wave is propagating from port
1 to port 2 in the primary line, the coupling factor and the directivity are defined,
Primary '
Porl: l o
- waveguide Port 2
0
device
0
Port 3 O Port 4
Secondary waveguide Figure 4-5-1 Directional coupler.
150 Microwave Waveguides and Components Chap. 4

respectively, by
PI
Coupling factor (dB) = 10 loglo -
P4
P4
Directivity (dB) = 10 loglo-
P3
where PI = power input to port 1
P3 = power output from port 3
P4 = power output from port 4
It should be noted that port 2, port 3, and port 4 are terminated in their char-
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 distur-
bance, 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 pri-
mary 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.

Coupling holes
v Rotated
secondary

Primary waveguide Primary waveguide

(c) (d 1

Figure 4-5-2 Different directional couplers. (a) Two-hole directional coupler.

(b) Four-hole directional coupler. (c) Schwinger coupler. (d) Bethe-hole directional
coupler.
Sec. 4.5 Directional Couplers

4-5-1 Two-Hole Directional Couplers

A two-hole directional coupler with traveling waves propagating in it is illustrated in

Fig. 4-5-3. The spacing between the centers of two holes must be

where n is any positive integer.

Primary waveguide

Port l -----T----+-T------ Port 2

4- -+-----A_--+
Port 3 Canceled, - -A+ +Added Port 4
Figure 4-5-3 Two-hole directional
Secondary waveguide coupler

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 pro-
gressing from right to left) are out of phase by (2L/Ag)2n rad and are canceled at
port 3.

4-5-2 S Matrix of a Directional Coupler

In a directional coupler all four ports are completely matched. Thus the diagonal ele-
ments of the S matrix are zeros and

As noted, there is no coupling between port 1 and port 3 and between port 2 and
port 4. Thus

Consequently, the S matrix of a directional coupler becomes

Equation (4-5-6) can be further reduced by means of the zero property of the S ma-
trix, so we have
Microwave Waveguides and Components Chap. 4

S21 S& + S41 S?3 = 0

Also from the unity property of the S matrix, we can write
S12Sf2 + S14Sf4 = 1

Equations (4-5-7) and (4-5-8) can also be written

1 11 1 = 1 11 1
sl2 s14 s32 s34

( ~ 2 ( 1J S 2 3 ) = IS41 ( I S 4 3 1

Since S I Z= S2, , S I 4 = S4, , S 2 3 = S32, and S34 = S 4 3 , then

JSlzl = Is341

Is141 = Is231

Let
Sl2 = s34 = p
where p is positive and real. Then from Eq. (4-5-8)

p(S,*, + S41) =0
Let
S23 = s
41 = jq
where q is positive and real. Then from Eq. (4-5-9)

p2 + q2 = 1
The S matrix of a directional coupler is reduced to

Example 4-5-1: Directional Coupler

A symmetric directional coupler with infinite directivity and a forward attenuation of
20 dB is used to monitor the power delivered to a load Z p (see Fig. 4-5-4). Bolometer
1 introduces a VSWR of 2.0 on arm 4; bolometer 2 is matched to arm 3. If bolometer 1

Bolometer 2 Bolometer 1
Figure 4-5-4 Power measurements by directional coupler.
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 The wave propagation in the directional coupler is shown in Fig. 4-5-5

,
SWR = 2.0
Po* l Po* 2

Generator

2mW B +
-lmW r'

port 3 1 J port 4
SWR = 2.0
Figure 4-5-5 Wave propagation in the directional coupler.

a. Power dissipation at Zr.

1. The reflection coefficient at port 4 is

2. Since the incident power and reflected power are related by

where P+ = incident power and P - = reflected power, then

7 I

The incident power to port 4 is P a = 9 mW, and the reflected power from port
4 i s P ; = 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

and the power reflected from the load is

P2- = 100 x (1 mW) = 100 mW
5. The power dissipated in the load is
Microwave Waveguides and Components Chap. 4

b. The reflection coefficient is calculated as

7 7

Then the VSWR on arm 2 is

4-5-3 Hybrid Couplers

Hybrid couplers are interdigitated microstrip couplers consisting of four parallel strip
lines with alternate lines tied together. A single ground plane, a single dielectric,
and a single layer of metallization are used. This type of coupler, called a Lange hy-
brid coupler [3], has four ports, as shown in Fig. 4-5-6.

Figure 4-5-6 Lange hybrid coupler.

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 re-
turn 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 Directional Couplers 155

used in balanced amplifier circuitry for high-power and broad-bandwidth applica-

tions, as shown in Fig. 4-5-7.
Single-stage or cascaded double-stage GaAs MESFET 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:

and
Gain = I l2
S21 = a (S21n+ S 2 1 b l2 (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 ex-
pressed as

VSWR = + I for the input port

1 - ISIII
and

VSWR =
1-
+ S22
I S22 I
I for the output port

GaAs MESFET a

I GaAs MESFET b

Figure 4-5-7 Balanced amplifier with Lange couplers

Theoretically, if the two GaAs MESFET 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 MESFET chips are not actu-
ally measured and they may not be the same. When their characteristics are differ-
ent, 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 Microwave Waveguides and Components 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 re-
producibility.

Example 4-5-2: Operation of a Balanced Amplifier

A GaAs MESFET balanced amplifier with two Lange couplers has the following
parameters:

S parameters: SII,= S I I ~
S22a = S22b

Input signal power: En= 200 mW

Power gain of each GaAs chip: Gain = 10 dB

Determine: (a) the input and output VSWRs; (b) the output power in watts; (c) the lin-
ear 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
Po,,=200X 1 0 X 2 = 4 0 0 0 m W = 4 W
c. Because two GaAs chips are in parallel, the linear output power gain is
Gain = 10 log (2) = 3 dB

4-6 CIRCULATORS AND ISOLATORS

Both microwave circulators and microwave isolators are nonreciprocal transmission

devices that use the property of Faraday rotation in the ferrite material. In order to
understand the operating principles of circulators and isolators, let us describe the
behavior of ferrites in the nonreciprocal phase shifter.
A nonreciprocal phase shijier consists of a thin slab of ferrite placed in a
rectangular waveguide at a point where the dc magnetic field of the incident wave
mode is circularly polarized. Ferrite is a family of Me0 . Fez0 3 , where Me is a di-
valent iron metal. When a piece of ferrite is affected by a dc magnetic field, the fer-
rite exhibits Faraday rotation. It does so because the ferrite is nonlinear material and
its permeability is an asymmetric tensor [4], as expressed by

where
Sec. 4.6 Circulators and Isolators 157

which is the tensor magnetic susceptibility. Here x is the diagonal susceptibility and
K is the off-diagonal susceptibility.
When a dc magnetic field is applied to a ferrite, the unpaired electrons in the
ferrite material tend to line up with the dc field because of their magnetic dipole mo-
ment. However, the nonreciprocal precession of unpaired electrons in the ferrite
causes their relative permeabilities (p:, p;) to be unequal and the wave in the fer-
rite is then circularly polarized. The propagation constant for a linearly polarized
wave inside the ferrite can be expressed as [4]

where

The relative permeability p, changes with the applied dc magnetic field as given by

where ye = gyromagnetic ratio of an electron

Me = saturation magnetization
w = angular frequency of a microwave field
Hdc= dc magnetic field
p: = relative permeability in the clockwise direction (right or positive
circular polarization)
p; = relative permeability in the counterclockwise direction (left or negative
circular polarization)
It can be seen from Eq. (4-6-8) that if w = 1 ye l~ ~ , then p: is infinite. This
phenomenon is called the gyromagnetic resonance of the ferrite. A graph of p, is
plotted as a function of Hdcfor longitudinal propagation in Fig. 4-6-1.
If p: is much larger than p; (p: 9 p;), the wave in the ferrite is rotated in

Figure 4-6-1 Curves of p, versus Hdc

for axial propagation.
158 Microwave Waveguides and Components Chap. 4

the clockwise direction. Consequently, the propagation phase constant P+ for the
forward direction differs from the propagation phase constant P- for the backward
direction. By choosing the length of the ferrite slab and the dc magnetic field so that

a differential phase shift of 90" for the two directions of propagation can be ob-
tained.

4-6-1 Microwave Circulators

A microwave circulator is a multiport waveguide junction in which the wave can

flow only from the nth port to the (n +
1)th port in one direction (see Fig. 4-6-2).
Although there is no restriction on the number of ports, the four-port microwave
circulator is the most common. One type of four-port microwave circulator is a com-
bination of two 3-dB side-hole directional couplers and a rectangular waveguide
with two nonreciprocal phase shifters as shown in Fig. 4-6-3.

Port 4

Port 2 Figure 4-6-2 The symbol of a circulator.

Coupler 1 Coupler 2
,---1

Primary guide i--

I 7I I
I ' I I
I
I
Port 1 -----

Port 3 270'

Secondary guide i-- IL - - J I

Figure 4-6-3 Schematic diagram of four-port circulator.

Sec. 4.6 Circulators and Isolators 159

The operating principle of a typical microwave circulator can be analyzed with

the aid of Fig. 4-6-3. Each of the two 3-dB couplers in the circulator introduces a
phase shift of 90°, and each of the two phase shifters produces a certain amount of
phase change in a certain direction as indicated. When a wave is incident to port 1,
the wave is split into two components by coupler 1. The wave in the primary guide
arrives at port 2 with a relative phase change of 180". The second wave propagates
through the two couplers and the secondary guide and arrives at port 2 with a rela-
tive phase shift of 180". Since the two waves reaching port 2 are in phase, the power
transmission is obtained from port 1 to port 2. However, the wave propagates
through the primary guide, phase shifter, and coupler 2 and arrives at port 4 with a
phase change of 270". The wave travels through coupler 1 and the secondary guide,
and it arrives at port 4 with a phase shift of 90". Since the two waves reaching port
4 are out of phase by 180°, the power transmission from port 1 to port 4 is zero. In
general, the differential propagation constants in the two directions of propagation in
a waveguide containing ferrite phase shifters should be
w, - m3 = (2m +l ) ~ radls (4-6-10)
w2 - w4 = 2 n r radls (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 princi-
ples of operation remain the same. Figure 4-6-4 shows a four-port circulator con-
structed 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.

-----t

Phase shifter

Figure 4-6-4 A four-port circulator

A perfectly matched, lossless, and nonreciprocal four-port circulator has an S

matrix of the form

Using the properties of S parameters as described previously, the S matrix in Eq.

160 Microwave Waveguides and Components Chap. 4

(4-6-12) can be simplified to

4-6-2 Microwave isolators

An isolator is a nonreciprocal transmission device that is used to isolate one compo-

nent from reflections of other components in the transmission line. An ideal isolator
completely absorbs the power for propagation in one direction and provides lossless
transmission in the opposite direction. Thus the isolator is usually called uniline.
Isolators are generally used to improve the frequency stability of microwave genera-
tors, such as klystrons and magnetrons, in which the reflection from the load affects
the generating frequency. In such cases, the isolator placed between the generator
and load prevents the reflected power from the unmatched load from returning to the
generator. As a result, the isolator maintains the frequency stability of the generator.
Isolators can be constructed in many ways. They can be made by terminating
ports 3 and 4 of a four-port circulator with matched loads. On the other hand, isola-
tors can be made by inserting a ferrite rod along the axis of a rectangular waveguide
as shown in Fig. 4-6-5. The isolator here is a Faraday-rotation isolator. Its operating
principle can be explained as follows [ 5 ] . The input resistive card is in the y-z plane,
and the output resistive card is displaced 45" with respect to the input card. The dc
magnetic field, which is applied longitudinally to the ferrite rod, rotates the wave
plane of polarization by 45". The degrees of rotation depend on the length and di-
ameter of the rod and on the applied dc magnetic field. An input TEIOdominant
mode is incident to the left end of the isolator. Since the TElomode wave is perpen-
dicular to the input resistive card, the wave passes through the ferrite rod without at-
tenuation. The wave in the ferrite rod section is rotated clockwise by 45" and is nor-
mal to the output resistive card. As a result of rotation, the wave arrives at the output

Output
Direction waveguide
o f rotation
Resistive
vane

Magnetic
\ ield

Input Reflected Ferrite rod

waveguide

Figure 4-6-5 Faraday-rotation isolator.

Problems 161

end without attenuation at all. O n 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. T h e typical performance of these isolators is about 1-dB insertion loss in for-
ward transmission and about 20- to 30-dB isolation in reverse attenuation.

REFERENCES

[I] SAAD,T., and R . C. HANSEN, Microwave Engineer's Handbook, Vol. 1 . Artech House,
Dedham, Mass., 197 1 .
[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, 1 150- 1 15 1 , December 1969.
[4] S o o ~ o o 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, July-
August 1958.

SUGGESTED READINGS

LIAO,S. Y., Engineering Applications of Electromagnetic Theory. West Publishing Com-

pany, St. Paul, Minn., 1988.
, A . , Microwave Engineering: Passive Circuits. Prentice-Hall, Inc., Englewood
R I ~ Z IP.
Cliffs, N.J., 1988.

PROBLEMS

Rectangular waveguides
4-1. An air-filled rectangular waveguide has dimensions of a = 6 cm and b = 4 cm. The
signal frequency is 3 GHz. Compute the following for the TElo, T&I , TEII, and TMll
modes:
a. Cutoff frequency
b. Wavelength in the waveguide
c. Phase constant and phase velocity in the waveguide
d. Group velocity and wave impedance in the waveguide
4-2. Show that the TMol and TMlomodes in a rectangular waveguide do not exist.
4-3. The dominant mode TEIOis propagated in a rectangular waveguide of dimensions
a = 6 cm and b = 4 cm. The distance between a maximum and a minimum is
4.47 cm. Determine the signal frequency of the dominant mode.
4-4. A TEll mode of 10 GHz is propagated in an air-filled rectangular waveguide. The mag-
162 Microwave Waveguides and Components Chap. 4

netic field in the z direction is given by

(E)cos (2)
Hz = Ho cos
vz Aim
vz
The phase constant is P = 1.0475 radlcm, the quantities x and y are expressed in cen-
timeters, and a = b = V%are also in centimeters. Determine the cutoff frequency f, ,
phase velocity o, , guided wavelength A,, and the magnetic field intensity in the y di-
rection.
4-5. A rectangular waveguide is designed to propagate the dominant mode TElo at a fre-
quency 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.
4-6. An air-filled rectangular waveguide has dimensions of a = 6 cm and b = 4 cm. The
guide transports energy in the dominant mode TEloat a rate of 1 horsepower (746 J). If
the frequency is 20 GHz, what is the peak value of electric field occurring in the guide?
4-7. An impedance of (0.5 - j0.4)Zo is connected to a rectangular waveguide. A capac-
itive window with a susceptance jB = j0.4&is located at a distance of 0.2A from the
load.
a. Determine the VSWR on the line in the absence of the window.
b. Find the VSWR on the line in the presence of the window.
4-8. An air-filled rectangular waveguide with dimensions of 3 cm X 1 cm operates in the
TElo mode at 10 GHz. The waveguide is perfectly matched and the maximum E field
existing everywhere in the guide is lo3 Vlm. Determine the voltage, current, and wave
impedance in the waveguide.
4-9. The dominant mode TElo is propagated in a rectangular waveguide of dimensions
a = 2.25 cm and b = 1 cm. Assume an air dielectric with a breakdown gradient of
30 kV/cm and a frequency of 10 GHz. There are no standing waves in the guide. De-
termine the maximum average power that can be carried by the guide.
4-10. A rectangular waveguide is terminated in an unknown impedance at z = 25 cm. A
dominant mode TElois 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 di-
mensions of 7 x 3.5 cm. It operates in the dominant TElomode.
a. Determine the cutoff frequency.
b. Find the phase velocity in the guide at a frequency of 2 GHz.
c. Find the guided wavelength A, at the same frequency.
4-12. The electric field intensity of the dominant TElo mode in a lossless rectangular wave-
guide is
E? = E~ sin ' f
( ~ ) ~ - ~ f l ~for > f,

a. Find the magnetic field intensity H.

b. Compute the cutoff frequency and the time-average transmitted power.
Problems

Circular waveguides
4-13. An air-filled circular waveguide is to be operated at a frequency of 6 GHz and is to
have dimensions such that f, = 0.8 f for the dominant mode. Determine:
a. The diameter of the guide
b. The wavelength A, and the phase velocity v, in the guide
4-14. An air-filled circular waveguide of 2 cm inside radius is operated in the TEol mode.
a. Compute the cutoff frequency.
b. If the guide is to be filled with a dielectric material of E, = 2.25, to what value
must its radius be changed in order to maintain the cutoff frequency at its original
value?
4-15. An air-filled circular waveguide has a radius of 1.5 cm and is to carry energy at a fre-
quency of 10 GHz. Find all TE and TM modes for which transmission is possible.
4-16. A TEll wave is propagating through a circular waveguide. The diameter of the guide is
10 cm, and the guide is air-filled.
a. Find the cutoff frequency.
b. Find the wavelength A, in the guide for a frequency of 3 GHz.
c. Determine the wave impedance in the guide.
4-17. An air-filled circular waveguide has a diameter of 4 cm and is to carry energy at a fre-
quency of 10 GHz. Determine all TE, modes for which transmission is possible.
4-18. A circular waveguide has a cutoff frequency of 9 GHz in dominant mode.
a. Find the inside diameter of the guide if it is air-filled.
b. Determine the inside diameter of the guide if the guide is dielectric-filled. The rela-
tive dielectric constant is E, = 4.

Microwave cavities
4-19. A coaxial resonator is constructed of a section of coaxial line and is open-circuited at
both ends. The resonator is 5 cm long and filled with dielectric of E, = 9. The inner
conductor has a radius of 1 cm and the outer conductor has a radius of 2.5 cm.
a. Find the resonant frequency of the resonator.
b. Determine the resonant frequency of the same resonator with one end open and one
end shorted.
4-20. An air-filled circular waveguide has a radius of 3 cm and is used as a resonator for TEkl
mode at 10 GHz by placing two perfectly conducting plates at its two ends. Determine
the minimum distance between the two end plates.
4-21. A four-port circulator is constructed of two magic tees and one phase shifter as shown
in Fig. 4-6-4. The phase shifter produces a phase shift of 180". Explain how this circu-
lator works.
4-22. A coaxial resonator is constructed of a section of coaxial line 6 cm long and is short-
circuited at both ends. The circular cavity has an inner radius of 1.5 cm and an outer
radius of 3.5 cm. The line is dielectric-filled with E, = 2.25.
a. Determine the resonant frequency of the cavity for TEMw .
b. Calculate the quality Q of the cavity.
4-23. A rectangular-cavity resonator has dimensions of a = 5 cm, b = 2 cm, and d =
15 cm. Compute:
a. The resonant frequency of the dominant mode for an air-filled cavity
b. The resonant frequency of the dominant mode for a dielectric-filled cavity of
E , = 2.56
164 Microwave Waveguides and Components Chap. 4

4-24. An undercoupled resonant cavity is connected to a lossless transmission line as shown

in Fig. P4-24. The directional coupler is assumed to be ideal and matched on all arms.
The unloaded Q of the cavity is 1000 and the VSWR at resonance is 2.5.
a. Calculate the loaded Qp of the cavity.
b. Find the reading of bolometer 2 if bolometer 1 reads 4 mW.
c. Compute the power dissipated in the cavity.

VSWR Cavity

Hybrid circuits
4-25. A microwave transmission system consists of a generator, an overcoupled cavity, two
ideal but not identical dual directional couplers with matched bolometers, and a load
Zp . The lossless transmission line has a characteristic impedance & . The readings of
the four bolometers (1, 2, 3, and 4) are 2 mW, 4 mW, 0 and 1 mW, respectively. The
system is shown in Fig. P4-25.
a. Find the load impedance Ze in terms of ZO.
b. Calculate the power dissipated by Zt .
c. Compute the power dissipated in the cavity.
d. Determine the VSWR on the input transmission line.
e. Find the ratio of Qe/Qo for the cavity.

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 Zt as shown in
Fig. P4-26. Bolometer 1 introduces a VSWR of 2.0 on arm 1; 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 Zt .
b. Determine the VSWR on arm 3.
Problems

VSWK = 2.0
Bolometer 2 Bolon~ctcr1
n 2 1 n

I"'
-

4-27. A semicircular-cavity resonator has a length of 5 cm and a radius of 2.5 cm.

a. Calculate the resonant frequency for the dominant mode if the cavity is air-filled.
b. Repeat part (a) if the cavity is loaded by a dielectric with a relative constant of 9.
4-28. The impedance matrix of a certain lumped-element network is given by

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 MESFET balanced amplifier
circuit with the following parameters:
MESFET a: Reflection coefficients Sl1,= 0.74881-158.3'

Forward transmission S21a = 1.3500/-8.5O

coefficient
MESFET b: Reflection coefficients SlI b = 0.62101- 175.9'

Forward transmission = 1.2200/ - 19. lo

coefficient
Compute: a. The input and output VSWRs
b. The power gain in dB for the balanced amplifier
c. The power loss in dB if one MESFET fails
d. The linear output power gain in dB
Chapter 5

Microwave Transistors
and Tunnel Diodes

5-0 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 tran-
sistor (HBT), and the tunnel diodes. This group is discussed in this chapter. The
second group includes microwave field-effect transistors (FETs) such as the junc-
tion field-effect transistors (JFETs), metal-semiconductor field-effect transistors
(MESFETs), high electron mobility transistors (HEMTs), metal-oxide-semiconduc-
tor field-effect transistors (MOSFETs), the metal-oxide-semiconductor transistors
and memory devices, and the charge-coupled devices (CCDs). This group is de-
scribed 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), in-
dium 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 Introduction

'
TABLE 5.0.1 MICROWAVE SOLID-STATE DEVICES

Microwave BJT
Microwave transistor HBT
lhnnel diode

JFET
Field-effect MESFET, HEMT
transistors

Microwave
solid-state Gunn diode
devices Transferred electron
devices InP diodes
CdTe diodes

Read diode
Avalanche transit-
time devices TRAPATT diode
BARITT diode

ranges by changes in their temperature, optical excitation, and impurity content,

they are the natural choices for electronic devices. The properties of important
semiconductors are tabulated in Table 5-0-2.
The energy bands of a semiconductor play a major role in their electrical be-
havior. For any semiconductor, there is a forbidden energy region in which no al-
lowable states can exist. The energy band above the forbidden region is called the
conduction band, and the bottom of the conduction band is designated by E,. The
energy band below the forbidden region is called the valence band, and the top of

TABLE 5-0-2 PROPERTIES OF IMPORTANT SEMICONDUCTORS

Bandgap Mobility at 300°K

energy (eV) (cmz/V . s) Relative
dielectric
Semiconductor 0°K 300°K Holes Electrons constant

C
Ge
Si
AlSb
GaSb
GaAs
Gap
InSb
InAs
InP
CdS
CdSe
ZnO
ZnS
168 Microwave Transistors and Tunnel Diodes Chap. 5

the valence band is designated by E, . The separation between the energy of the low-
est conduction band and that of the highest valence band is called the bandgap en-
ergy E, , 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 di-
agram is shown in Fig. 5-0- 1.

-Distance Figure 5-0-1 Energy-band diagram.

In the 1970s, it seemed that microwave transistors would be useful for generat-
ing 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 in-
dicate 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 W in X band, and 50 W in 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 p m (microns). Thus only a limited voltage can be
applied to such a thin layer because of breakdown limitations. Consequently, the

1 2 4 6 810 20
Figure 5-0-2 Peak-power levels
Frequency (GHz) achieved by microwave diode.
Sec. 5.1 Microwave Bipolar Transistors 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 ei-
ther 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, communica-
tions, navigational and industrial electronics, and medical and biological equipment.
Representative applications for microwave solid-state devices are listed in Table
5-0-3.

TABLE 5-0-3 APPLICATIONS OF MICROWAVE SOLID-STATE DEVICES

Devices Applications Advantages

Transistor L-band transmitters for telemetry Low cost, low power supply, reliable,
systems and phased array radar high CW power output, light weight
systems
L- and S-band transmitters for
communications systems
TED C-, X-, and Ku-band ECM amplifiers Low power supply (12 V), low cost,
for wideband systems light weight, reliable, low noise, high
X- and Ku-band transmitters for radar gain
systems, such as traffic control
IMPATT Transmitters for millimeter-wave Low power supply, low cost, reliable,
communications systems high CW power output, light weight
TRAPATT S-band pulsed transmitters for phased High peak and average power, reliable,
array radar systems low power supply, low cost
BARIIT Local oscillators in communications Low cost, low power supply, reliable,
and radar receivers low noise

5-1 MICROWAVE BIPOLAR TRANSISTORS

The invention of the transistor (contraction for transfer resistor) by William Shock-
ley.and his coworkers at Bell Laboratory in 1948 had a revolutionary impact on elec-
tronic technology in general and on solid-state devices in particular. Since then tran-
sistors and related semiconductor devices have replaced vacuum tubes for lower-
power sources. Microwave power transistor technology has advanced significantly
during the past three decades. The microwave transistor is a nonlinear device, and its
principle of operation is similar to that of the low-frequency device, but requirements
for dimensions, process control, heat sinking, and packaging are much more severe.
For microwave applications, the silicon (Si) bipolar transistors dominate for
frequency range from UHF to about S band (about 3 GHz). As the technology im-
proves, the upper frequency limit for these devices is continuously being extended,
and at the present time the devices are capable of producing useful power up to
22 GHz. The majority of bipolar transistors of current interest are fabricated from
silicon, although GaAs devices offer prospects for improvements in operating fre-
170 Microwave Transistors and Tunnel Diodes Chap. 5

quency, in high temperatures, and in radiation hardness. The Si bipolar transistor is

inexpensive, durable, integrative, and offers gain much higher than available with
competing field-effect devices. It has moderate noise figure in RF amplifiers and llf
noise characteristics that are about 10-20 dB superior to GaAs MESFETs. For these
reasons, the Si bipolar transistors dominate in amplifier applications for the lower
microwave frequencies and are often the devices of choice for local oscillators.

-
5-1 1 Physical Structures

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 [I].

Emitter

(a) lnterdigitated (b) Overlay

Emitter metalization

p+ base diffusion

Base metalization
(c) Matrix

Figure 5-1-1 Surface geometries of microwave power transistor. (From H. Sobol

and F. Sterzer [ I ] : reprinted b y permission of IEEE, Inc.)

For high-frequency applications, the n-p-n structure is preferred because the

electron mobility (pn= 1500 cm2/V. s) is much higher than the hole mobility
( C L ~= 450 cm2/V . s). Figure 5-1-2 shows an example of the densities for an n-p-n
transistor. The density unit is in cm2/V . s.
Although there are many ways of fabricating a transistor, diffusion and ion im-
plantation are generally used. For example, the structure would typically start with a
lightly doped n-type epitaxial layer as the collector. The base region would be
formed by counter-doping the base region p-type by diffusion. The emitter would
Sec. 5.1 Microwave Bipolar Transistors 171

TABLE 5-1-1 FIGURE OF MERIT (M) OF VARIOUS SURFACE GEOMETRIES

Surface geometry and unit cell M = -EP

BA

lnterdigitated

2(e+W)
e ( +~ S )
I

w +s -1 P +p-I Matrix
I

Source: From H. Sob01 and F. Sterzer [I]; reprinted by permission of IEEE, Inc.

be formed by a shallow heavily doped n-type diffusion or by ion implantation. The

emitter and base contacts are generally located on the semiconductor surface in an
interdigital, planar arrangement. The interdigital geometry always provides for
n + 1 base fingers, where n is the number of emitter fingers. The number of fingers
varies with Jhe application, with more fingers required as the output power capabil-

Figure 5-1-2 Carrier densities of an n-p-n transistor.

172 Microwave Transistors and Tunnel Diodes Chap. 5

ity of the transistor increases. Additional fingers, however, increase the device para-
sitic~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 a p+-type region is devel-

Insulator layer
SiO, Emitter contact
Base contact
e m i t t e r junction

Base-collector junction

n-collector layer

( a ) Discrete n-p-n planar BJT

Emitter contact

Base contact

(b) Integrated c h ~ p t y p e11-1)-n BJ1

Figure 5-1-3 Schematic diagrams of bipolar junction transistors. (After D. Navon

121.)
Sec. 5.1 Microwave Bipolar Transistors 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.

5-1-2 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) configuration 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.

Figure 5-1-4 Common-base configurations for p-n-p and n-p-n transistors.

The CB configuration is also called the grounded-base configuration. For a p-

n-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 applica-
tions. Its input voltage VEBand output current Ic can be expressed in terms of the out-
put voltage Vcs and input current IEas
VEB= some function (Vcs , IE)
IC= some function (VcB, IE)

Common-emitter configuration. Most transistors have their emitter,

rather than their base, as the terminal to both input and output networks. Such a
174 Microwave Transistors and Tunnel Diodes Chap. 5

configuration is known as common-emitter (CE) or grounded-emitter configuration

as shown in Fig. 5-1-5.
In the CE configuration, the input current IBand the output voltage VcE are in-
dependent variables, whereas the input voltage VBEand output current Ic can be writ-
ten as
VBE= some function ( V C ~IB)
, (5-1-3)
IC = some function (VCE,IB) (5- 1-4)
The CE configuration is commonly used as a switch or pulse transistor amplifier.
This is because the transistor is open at the cutoff mode and is closed at the satura-
tion mode.

Figure 5-1-5 Common-emitter configurations for p-n-p and n-p-n transistors.

Common-collector configuration. Another transistor-circuit configuration

is called the common-collector or grounded-collector configuration as shown in Fig.
5- 1-6.
In a common-collector (CC) configuration, the output voltage of the load is
taken from the emitter terminal instead of the collector as in the common-base and
common-emitter configurations. When the transistor is cut off, no current will flow
in tl?e emitter terminal at the load. When the transistor is operating in a saturation
mode, the load current reaches toward its maximum. Therefore the CC configura-
tion transistor can also be used as a switch or pulse amplifier. The significant differ-
Sec. 5.1 Microwave Bipolar Transistors

Figure 5-1-6 Common-collector configuration for p-n-p and n-p-n transistors.

ence, however, between common-emitter and common-collector configurations is

that the common-collector amplifier has no voltage gain.

Hybrid-pi equivalent model. The hybrid-pi equivalent model is commonly

used in the normal active mode of the common-emitter configuration for small-
signal operation. Fig. 5-1-7 shows the hybrid-pi equivalent model of the common-
emitter configuration.
For a small-signal operation, the nonlinear or ac parameters of a hybrid-pi
equivalent model can be expressed as
Vce = hieib + hre~ce
i, = hfeib + hvevce

where hie = -
"ce=constant
176 Microwave Transistors and Tunnel Diodes Chap. 5

( a ) Common-emitter ti-p-n transistor

(b) Hybrid-pi model

;Itci
E
'b

g m v b r e + ~ [ c

( c ) Simplified hybrid-pi model at low frequency

ic

Figure 5-1-7 Hybrid-pi equivalent

model.

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 mea-
sured. 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 ATw, at the input terminal will
induce an incremental change of the collector current Ai, at the output terminal.
Then the mutual conductance (or transconductance) of a small-signal transistor is
defined by

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,
Sec. 5.1 Microwave Bipolar Transistors

and

Substitution of Eq. (5- 1- 12) into Eq. (5-1- 13) and differentiation of the resultant
yield

where VT = 26 mV at 300' K is the voltage equivalent of temperature

As the width of the base region is very narrow, Cb 9 Cc , the diffusion capaci-
tance in the base charge-storage is given by

and the total charges stored in the base is

Then the diffusion capacitance is expressed by

Cbfe=
qAWbnp(0) -
- gm-
wz
2VT 20,
The voltage across the diffusion capacitance can be written as

The small-signal input conductance of the emitter junction looking at the input of
the base is defined as

where hFE is the linear or dc common-emitter current gain factor and

gm is the mutual conductance.

Example 5-1-1: Equivalent Elements of a Hybrid-Pi Common-Emitter Circuit

A Si n-p-n bipolar transistor has the following parameters:

Collector current: I, = 6 rnA

Common-emitter current gain factor: ~ F =
E 120
Operational temperature: T = 300" K
Cross-sectional area: Wb = lo-' cm2
178 Microwave Transistors and Tunnel Diodes Chap. 5

Compute: (a) the mutual conductance g, ; (b) the input conductance gb and resistance
R,; (c) the electron diffusion coefficient D, ; and (d) the diffusion capacitance CLe.
Solution

a. The mutual conductance is

= 0.23 mho
26 x 10-1
b. The input conductance and resistance are
g m - 0.23
gb = - - - = 1.92 x mho
~ F E 120
Ri= 521 ohrns
c. The electron diffusion coefficient is

d. The diffusion capacitance is

5-1-3 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

and the collector is reverse-biased, the transistor is operated in the normal
mode as shown in Fig. 5-1-8(a). The term forward bias means that the posi-
tive polarity of the bias voltage is connected to the p side and the negative po-
larity to the n side for a p-n junction; the opposite obtains for reverse bias.
Most transistor amplifiers are operated in normal mode, and its common-base
current gain alpha is known as the normal alpha a~.
2. Saturation Mode. When both transistor junctions are forward-biased, the
transistor is in its saturation mode with very low resistance, and acts like a
short circuit, as shown in Fig. 5-1-8(b).
3. Cutoff Mode. If both transistor junctions are reverse-biased the transistor is
operated in its cutoff mode. As the current is cut off, the transistor acts like an
Sec. 5.1 Microwave Bipolar Transistors

e63 VEB 'CB

( d ) Inverse mode
VI-B v,,
( c ) Cutoff mode

VEB 'CB

( b ) Saturation motle
rn VCB v~
( a ) Normal mode

Figure 5-1-8 Operational modes of an n-p-n transistor

open circuit. Both the cutoff and saturation modes of a transistor are used as
switching devices for the OFF and ON states. Fig. 5- 1-8(c) shows the cutoff-
mode 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 a,.If the transistor is
symmetric, the normal alpha a~ is nearly equal to the inverse alpha a,.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. The resistivities of the semiconductor regions are low

2. The injected current densities are low
3. The space-charge layer widening effects can be ignored
4. The current and voltage of n-p junction diodes follow the basic equation of
I = Io(eVIV~
- 1) (5- 1-20)
Microwave Transistors and Tunnel Diodes Chap. 5

- Electron
---- flux
-
- J,,-
Hole flux

Figure 5-1-9 Current flow in an n-p-n transistor.

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 emit-
ter current IE consists of electron current InE crossing from the emitter into the base
and the hole current I p E 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 J E reach the collector
junction Jc because some of them combine with the holes in the p-type base. If Inc
is the electron current at the collector junction J c , there must be a recombination
current InE - Inc leaving the base. When the emitter is open-circuited, then IE = 0
and Inc = 0. As a result, the collector current Ic is equal to the reverse saturation
current Ic, because the junction between base and collector is reverse-biased. From
Fig. 5-1-9, we have

and

For an n-p-n transistor, Ice consists of holes moving across the collector junction Jc
from right to left (collector to base) and electrons crossing Jc in the opposite direc-
tion. Since the reference direction for Ico in Fig. 5-1-9 is assumed from left to right,
then, for an n-p-n transistor, Ico is positive for forward-biased Jc junction and Ico is
negative for reverse-biased Jc junction. The saturation current Ico at the Jc junction
of an n-p-n transistor is given by
AqDn,, AqDppn,
Ico = - -= Aqnf -
W LE
+ Dn +
(WNa LEN^
(5- 1-24)
Sec. 5.1 Microwave Bipolar Transistors 181

Also from Fig. 5-1-9, the sum of the three terminal currents should be zero and it is
IE + IC + IB =0 (5- 1-25)
Equation (5-1-25) can be verified by adding together Eqs. (5-1-21) through
(5- 1-23).

Current flow in common-base n-p-n transistor. In a common-base

configuration of an n-p-n transistor as shown in Fig. 5-1-9, the emitter n-p junction
is forward-biased and the collector p-n junction is reverse biased. Their total current
flow can be found from the basic diffusion equation. The steady-state diffusion equa-
tion for an n-p-n transistor at low-level injection is given by

and

where D, = electron diffusion constant

n,, = minority electron carrier density in the p-type base layer
npo = equilibrium minority electron carrier density in the p-type base layer
x = distance measured from the base region as shown in Fig. 5- 1- 10
r, = electron lifetime

t
0

1.1111ttc.r
0
-
Ihc
W
-
(-olltector
w:-----
r rgu~
r
ties unuer
-
.
1 ----.-.
5-1-10 Minority-carrier densi-
was .W-.:I.LL I I ---,:
rlurrrlal acclvt- L:--
gible recombination.
11eg11-

The general solution of Eq. (5- 1-26) can be written as

n, - nw = c I C - ~ I L n + c Z e x I L n (5- 1-27)
where C, and C2 are constants to be determined by the boundary conditions
L, = 2/7,0,is the electron diffusion length
n,, = n:lN, is the mass-action law
The boundary conditions at the edge of the emitter depletion layer in the base
side with a forward-biased emitter junction is
np(0) = np,e "E'"T (5- 1-28)
182 Microwave Transistors and Tunnel Diodes Chap. 5

where VE = forward-biased voltage across the emitter junction

VT = 26 X V at 300" k is the voltage equivalent of temperature
The boundary condition at the depletion-layer edge of the reverse-biased collector
junction is usually assumed to be
np(W) = 0 ( 5 - 1-29)
The two boundary conditions are shown in Fig. 5-1-10.
The general solution of Eq. (5-1-26) can be written as
sinh [(W - x)/L,]
np(x) = n,, ( e 'EIVT - 1)
[ sinh (WIL.)
+ npo(l - sinh (xlLn))]
Sinh (WIL,) (5- 1-30)

In almost all transistors, the base width is made very narrow (W 4 L,) 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 - 1 1 . (Note: sinh y A y, cosh y 5 1,
cothy lly, and sechy A 1 - y 2 / 2 fory 4 1 . )

1 - 5-1-11 Minority carrier distri-

Figure
0 w butions in base region as a function of
D~stance.\. WlI,. (After S. Sze 131.)

For W 4 L, , the minority carrier distribution of Eq. (5-1-30) can be sim-

plified to

Then the boundary conditions for holes in the emitter and collector depletion regions
can be expressed, respectively, as
Sec. 5.1 Microwave Bipolar Transistors 183

and
pc = pcoe-vclv~ at x = xc
where pE0 = equilibrium minority hole density in the emitter region
pco = equilibrium minority hole density in the collector region
Substituting Eqs. (5- 1-32) and (5-1-33) into Eq. (5-1-27) yields the minority distri-
butions in the emitter and collector regions as
pE(x) = PEO + PEO(e -~
V ~ / V 1) exp [ ( x + xE)/L,] for x 5 -xE (5-1-34)
and
pdx) = + pco exp [- ( x - xc)/L,]
~ C O for x L xc (5-1-35)
The total excess minority-carrier charge in the base region is given by

where A = cross section

The base recombination current for a carrier lifetime 7, is

According to the carrier injection processes, the minority electron density at

the reverse-biased collector junction Jc is usually assumed to be zero. That is,
npc = 0 at Jc for x = W (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 I,E which is injected from the emitter into the base at
x = 0 for L, S W is proportional to the gradient of the minority carrier density and
is expressed as
- AqD, npo
L, ') 'cosh (WIL.)
-
- -AqD,nf (,WT- 1 ) - AqD n2 A [email protected];
A vE/vT
Na W Nu W Na W
The electron current which reaches the collector at x = W is
-AqD,n,,
L,
1
sinh (WIL,)
[(evt-/VT
- I)
1
+ cosh (W/Ln)

- - AqD,nf (, VE/VT - 1 ) - AqD

A n2 = - A4Dnn:
- vE/vT
NuW Nu W Na W
The collector current Inc can also be expressed as
Microwave Transistors and Tunnel Diodes Chap. 5

where LE = diffusion length of the emitter and exp (VEIVT) S 1 is assumed

Similarly, the hole currents are

1) for x = -XE (5- 1-42)

and

for x = xc

where DE and Dc are the hole diffusion constants in the emitter and collector,
respectively
LC = 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 cur-
rent 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) + I,, (recombination-generation) (5-1-44)
The recombination-generation current can be computed from the following equation

where xd = depletion-layer width

ro = effective minority-carrier lifetime in the depletion layer
Figure 5-1- 12 shows current-voltage (I-V) characteristics of an ideal n-p-n bipolar
transistor for a common-base configuration.
There are three regions for the I-V characteristics of an n-p-n bipolar transis-
tor:

1. Active Region: In this region the emitter junction is forward-biased and the
collector junction is reverse-biased. The collector current Ic is essentially in-
dependent of collector voltage and depends only on the emitter current I E .
When the emitter current is zero, the collector current is equal to the reverse
saturation current Ic0 .
2. Saturation Region: In this region, as shown on the left side of Fig. 5-1-12,
both emitter and collector junctions are forward-biased. The electron current
flows from the n side across the collector junction to the p -type base. As a re-
sult, the collector current increases sharply.
Sec. 5.1 Microwave Bipolar Transistors

0.25 0 7 4 0 X
('ollcctor-to-bax voltagc VcB 111 volts

Figure 5-1-12 Current-Voltage (I-V) characteristics of an n-p-n transistor.

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: I-V Characteristics of an n-p-n Transistor

A silicon n-p-n transistor at 300' K has the following parameters:

Base width:
Diffusion length in emitter:
Diffusion length in collector:
Base resistivity:
Emitter resistivity:
Collector resistivity:
Emitter junction voltage:
Collector junction voltage:
Cross-section area:

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
Microwave Transistors and Tunnel Diodes Chap. 5

d. The equilibrium densities in the emitter, base, and collector regions

e. The terminal currents
Solution

a. The impurity densities are read from Fig. A-1 in Appendix A as

NdE = 1 X 1019 cm13 in the n-type emitter region
Nd = 1.5 x 1017 cm-3 in the p-type base region
Ndc = 3 x 1014 cm-3 in the n-type collector region
b. The mobilities are read from Fig. A-2 in Appendix A as
p p=~ 80 cm2/V- s in the emitter
pd = 105 cm2/V.s in the emitter
ppB=400 cm2/V.s in the base
pnc= 1600 cm2/V- s in the collector
c. The diffusion constants are computed to be
DpE = pPEVT= 80 x 26 X = 2.08 cm2/s
D& = pdVT = 105 X 26 X = 2.73 cm2/s
DpB = m V T = 400 X 26 X = 10.4 cm2/s
Dnc = pnCVT= 1600 X 26 X = 41.6 cm2/s
d. The equilibrium densities are
npB = n?/N& = (1.5 X 10'0)2/(1.5 X lOI7) = 1.5 X lo3 cm-3
PEO = nf/NdE = (1.5 x 1010)2/(1x lOI9) = 2.5 x lo4 ~ m - ~
pco = nf/Ndc = (1.5 X 10'0)2/(3x lOI4) = 7.5 X lo5 ~ m - ~
e. The terminal currents are computed as follows:
From Eq. (5-1-39), the electron current in the emitter is

x exp [0.5/(26 X

From Eq. (5-1-42), the hole current in the emitter is

IpE= A~DEPEO
--- (e VtlVr - 1) = AqD
2 n2,( v,/vTI - 1)
LE LEN^

From Eq. (5-1-24), the reverse saturation current in the collector is

Sec. 5.1 Microwave Bipolar Transistors

- AqD,n? --AqD,,p,,
l o =
N"w LE
= -13.104 X lo-" - 14.976 X lo-'"
= -1.312pA
From Eq. (5-1-40), the electron current which reaches the collector is

The emitter current is

IE = -IpE + InE= -33.67 X l o - @- 0.295 X lo-'
= -0.295 mA
The collector current is
Ic = -Ico - InC = 1.312 x lo-'* - (-0.295 X

= 0.295 mA
The current in the base terminal is
IB = 1p.E - (InE - 1°C) + ICO
= 33.67 X - (-29.46 X + 29.46 X + 1.312 X 10-l2
= 0.337 p A
Note: The recombination-generation currents in the space-charge regions are not
counted.

5-1-4 Amplification Phenomena

Bipolar transistors are usually used for signal amplification. The amplification phe-
nomena 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 in-
put current for a small signal in a bipolar junction transistor is known as the current
gain alpha a , or hfb. The current gain of a common-base p-n-p transistor is defined
by the current components crossing the emitter and collector junctions as

= - Ic + Ico = hfb
IE
where Ico = collector-junction reverse saturation current with zero emitter current
Since Ic and IE have opposite signs, the alpha a , as defined, as always positive. Typ-
ical numerical values of a are between 0.9 and 0.995.
188 Microwave Transistors and Tunnel Diodes Chap. 5

The emitter efficiency (or injection efficiency) is defined as

current of injected carriers at JE
Y = total emitter current

where InE = injected electron diffusion current at emitter junction J E

IpE = injected hole diffusion current at emitter junction JE
The transport factor P * is defined as
injected carrier current reaching JC
p* =
injected carrier current at JE

where Inc = injected electron diffusion current at Jc

InE = injected electron diffusion current at J E
Ln = a is the electron diffusion length
At the collector we have
Ic = -Ico - Inc
Then the current gain can be expressed as

In the normal active mode, the collector current is given by

Ic = -aIE - ICo (5- 1-5 1)
The current Ico is the current crossing the p-n junction, and it is expressed in Eq.
(5- 1-24). Then the complete expression of Ic for any Vc and IE is
= -a lE +~ ~ - eVcIVr)
~ ( l (5- 1-52)
where lo = -Ico is replaced.

Example 5-1-3: Silicon Bipolar Transistor

A silicon n-p-nbipolar transistor operates in common-base mode at 300" K and has the
following parameters:

Silicon intrinsic density: n, = 1.5 x 10" cm-3

Acceptor density in base region: N, = 5 x 1016 cm -
Donor density in emitter region: Nd = 5 X 1018 cm -
Hole lifetime: rP = 1 PS
Sec. 5.1 Microwave Bipolar Transistors

Electron lifetime:
Cross section:
Base width:
Emitter length:

Determine:

a. The mobilities p, and pp

b. The diffusion coefficients D, and D,
c. The emitter efficiency factor y
d. The transport factor /3
e. The current gain cx
Solution

a. The mobilities are read from Fig. A-2 in Appendix A as

p, = 200 cm2/V.s for N d E = 5 X 10'' cm-3
pp = 500 cm2/V.s for No = 5 X 1016cm-3
b. The diffusion coefficients are
D, = p,VT = 200 X 26 X = 5.20 cm2/s
D, = 500 x 26 x lo-' = 13.0 cm2/s
c. The emitter efficiency factor is

d. The transport factor is

e. The current gain is

a = /3*y = 0.904 x 0.997 = 0.90

Common-emitter n-p-n transistor. In the active region of a common-

emitter n-p-n transistor, the emitter junction is forward-biased and the collector
junction is reverse-biased. The base current is
Is = -(lc + IE) (5- 1-53)
Combining this equation with Eq. (5-1-5 l), we have

In the cutoff region, if l~ = 0,then IE = - IC and the collector current is given by

-Ico
I , = -1, = - = ICEO
1-a
190 Microwave Transistors and Tunnel Diodes Chap. 5

The actual collector current with the collector junction reverse-biased and the base
junction open-circuited is designated by the symbol I ~ o We . define the common-
emitter current gain /3 or hr, as

P = hf, =
Ic + Ico
Is

Since I E + JB + JC + JCO = 0 , we have

5-1-5 Power-Frequency Limitations

The question then arises as to whether microwave power transistors have any limita-
tions on their frequency and output power. The answer is yes. Several authors have
discussed this subject. Early [4] first introduced the power-frequency limitations in-
herent in (1) the limiting velocity of carriers in semiconductors and (2) the maxi-
mum fields attainable in semiconductors without the onset of avalanche multiplica-
tion. These basic ideas were later developed and discussed in detail by Johnson [ 5 ] ,
who made three assumpt\ons.

1. There is a maximum possible velocity of carriers in a semiconductor. This is

the "saturated drift velocity v," and is on the order of 6 X lo6 cmls for elec-
trons and holes in silicon and germanium.
2. There is a maximum electric field Em that can be sustained in a semiconductor
without having dielectric breakdown. This field is about lo5 Vlcm in germa-
nium and 2 X lo5 Vlcm in silicon.
3. The maximum current that a microwave power transistor can carry is limited
by the base width.

With these three postulates Johnson derived four basic equations for the power-
frequency limitations on microwave power transistors.

First equation: Voltage-Frequency Limitation:

Vm fT = - - 2 X 10" Vls for silicon

277 1 X 10" Vls for germanium
1
where fr = - is the charge-carrier transit-time cutoff frequency
2777
7 = - is the average time for a charge carrier moving at an average
0
velocity v to traverse the emitter-collector distance L
Vm = EmLmi,is the maximum allowable applied voltage
vs = maximum possible saturated drift velocity
Em = maximum electric field
With the carriers moving at a velocity v, of 6 X lo6 cm/s, the transit time can
Sec. 5.1 Microwave Bipolar Transistors 191

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 p m for overlay and matrix devices and to
nearly 250 p 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 satu-
rated velocity us and the electric field intensity will not be uniform.

Second equation: Current-Frequency Limitation:

where I , = maximum current of the device

x,=--- 1 - is the reactive impedance
WTCO 2~fTC0
Co = collector-base capacitance
It should be noted that the relationship 2 ~ f T ~=0 2 7 ~ f T=
~ 1 is used in deriving
Eq. (5- 1-59) from Eq. (5- 1-58). In practice, no maximum current exists because the
area of the device cannot be increased without bound. If the impedance level is zero,
the maximum current through a velocity-saturated sample might be infinite. How-
ever, the limited impedance will limit the maximum current for a maximum attain-
able power.

Third equation: Power-Frequency Limitation:

This equation was obtained by multiplying Eq. (5- 1-58) by Eq. (5-1-59) and replac-
ing VmI, by P,. 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 reac-
tance X, 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: Power Gain-Frequency Limitation:

where G , = maximum available power gain

Kh = K T / e is the thermal voltage
K = Boltzmann's constant, 1.38 X JIoK
192 Microwave Transistors and Tunnel Diodes Chap. 5

Figure 5-1-13 (P,,,X,.)'12versusf,.

(After B . C . De Loach [6]: reprinted
permission of Academic Press.)

T = absolute temperature in degrees Kelvin

e = electron charge (1.60 X C)
The maximum available power gain of a transistor was derived by Johnson [5]
as

where Z,,, and Zi, 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
Zout - Cin
Zin Cout

where Ci, is the input capacitance and Gout is the output (base-collector) capacitance.
When the maximum total carrier charges Qm move to the collector in a carrier-base
transit time rb and with a thermal voltage Vh,the input capacitance Cinand the emit-
ter diffusion capacitance Cd are related by

Gin = Cd -- - -
Q m
-- I m ~ b

Vth Vth

The output capacitance is given by

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 Heterojunction Bipolar Transistors (HBTs) 193

Example 5-1-4: Power-Frequency Limitation

A certain silicon microwave transistor has the following parameters:

Reactance: x,=ln
Transit-time cutoff frequency: fr = 4 GHz
Maximum electric field: Em = 1.6 x 10' V/cm
Saturation drift velocity: us = 4 x 10' cmls

Determine the maximum allowable power that the transistor can carry.
Solution. From Eq. (5- 1-60) the maximum allowable power is

5-2 HETEROJUNCTION BIPOLAR TRANSISTORS (HBTs)

Bipolar transistors can be constructed as homojunction or heterojunction types of

transistors. When the transistor junction is jointed by two similar materials such as
silicon to silicon or germanium to germanium, it is a homojunction transistor. The
transistor junction formed by two different materials, such as Ge to GaAs, is called a
heterojunction transistor. So far only the ordinary homojunction transistors have
been discussed. In this section we study the heterojunction transistors.

5-2-1 Physical Structures

When the lattice constants of two semiconductor materials are matched, they can be
formed together as a heterojunction transistor. This lattice condition is very impor-
tant because the lattice mismatch could introduce a large number of interface states
and degrade the heterojunction operation. Currently, Ge and GaAs are the two ma-
terials commonly used for heterojunction structures because their lattice constants
(a = 5.646 A for Ge and a = 5.653 for GaAs) are matched to within 1%. Since
each material may be either p type or n type, there are four possible heterojunction
combinations:

1. p -Ge to p -GaAs junction

2. p-Ge to n-GaAs junction
3. n-Ge to p -GaAs junction
4. n-Ge to n-GaAs junction

Fig. 5-2-1 shows the model diagram of a heterojunction transistor formed by n-Ge,
p -GaAs, and n-GaAs materials.
Microwave Transistors and Tunnel Diodes Chap. 5

Figure 5-2-1 Model diagram of a het-

erojunction transistor.

5-2-2 Operational Mechanism

When an n -Ge and a p -GaAs are isolated, their Fermi energy levels are not aligned,
as shown in Fig. 5-2-2 [ 7 ] .
In Fig. 5-2-2, the vacuum level is used as reference, the work function is de-
noted by 4, n-Ge is designated as 1, and p-GaAs is referred to as 2. The different
energies of the conduction-band edge and the valence-band edge are given by
AEc = XI - xz (5-2-1 )
and
AE, = Egz - E ~ -
I AEr
where y, = electron affinity in eV
E, = bandgap energy in eV

Vacuum level

E" I
AE,-----
E,, Figure 5-2-2 Energy-band diagram for
n-Ge O-GaAs isolated n-Ge and p -GaAs.

Example 5-2-1: Heterojunction Bipolar Transistor (HBT)

A Ge-GaAs heterojunction transistor has the following parameters:

Lattice constant: Ge a , = 5.646 A

GaAs az = 5.653 A
Electron affinity: Ge xi = 4.0 eV
GaAs ,y2=4.07 eV
Energy gap: Ge EgI = 0.80 eV
GaAs Eg2= 1.43 eV
Sec. 5.2 Heterojunction Bipolar Transistors (HBTs)

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

a. The lattice match is within 1%

b. The conduction-band differential is
AE, = XI - x2 = 4.0 - 4.07 = -0.07 eV
c. The valance-band differential is
E, = E,z - E,, - AE, = 1.43 - 0.8 - (-0.07)
= 0.70 eV

When the two semiconductor materials are jointed together, their Fermi energy lev-
els are aligned and their energy bands are depleted at the junction, as shown in Fig.
5-2-3 [7].

Figure 5-2-3 Energy-band diagram for an n-Ge-p-GaAs junction.

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 junc-
tion 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

where $or = barrier potential or portion of built-in voltage in n-Ge

(CI,2 = barrier potential or portion of built-in voltage in p-GaAs
196 Microwave Transistors and Tunnel Diodes Chap. 5

At the junction, the electric flux D is continuous. That is,

D = E,E~I%I= E , E ~ ~ % z (5-2-4)
= €l%l = €2%2

where E, permittivity of free space

=
E = permittivity
er = relative permittivity or dielectric constant
% = electric field
The space charges on both sides of the junction are equal and it is given by
xl Nd = x2Na (5-2-5)
where xl = depletion width in n-Ge
xz = depletion width in p -GaAs
Nd = donor density
Na = acceptor density
The electric fields in both sides can be written as

and

where Vl = portion of bias-voltage in n-Ge

V2 = portion of bias-voltage in p-GaAs

Substitution of Eq. (5-2-4) into Eqs. (5-2-6) and (5-2-7) results in

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 + $,2 + AE,/q) 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 $02
for hole injection. Therefore, the junction current can be approximated as shown in
Eq. (5-1-42) to be

where A = cross section

q = electron charge
Dp = hole diffusion constant
P,,, = minority or equilibrium hole density in n-Ge
Lp = hole diffusion length
V = bias voltage
VT = voltage equivalent of temperature
Sec. 5.2 Heterojunction Bipolar Transistors (HBTs)

5-2-3 Electronic Applications

Heterojunctions have been studied since 195 1 and have many applications in photon
devices as described in Chapter 10. The heterojunction bipolar transistor is a poten-
tial candidate for high-speed switching devices such as GaAs MESFETs. The analy-
sis described previously can be applied to the structures of Ge-GaAs and GaAs-
AlGaAs. In other heterojunction transistors, such as the Ge-Si structure, the lattice
A A
mismatch (a = 5.646 for Ge and a = 5.431 for Si) causes a high interface state
density and recombination- and tunneling-current components must be counted.

Example 5-2-2: n -Ge-p -GaAs-n -GaAs HBT

An n-Ge-p-GaAs-n-GaAs heterojunction transistor at 3000 K has the following
parameters:

Donor density in n-Ge region: N d = 5 x 10" cm-3

Acceptor density in p-GaAs region: N, = 6 x 1016 cm-3

Hole lifetime: rp = 6 X sec
Bias voltage at emitter junction: V,=1 V
Cross section: A =2 X cm2

Compute:

a. The built-in voltage in the p-GaAs side

b. The hole mobility
c. The hole diffusion constant
d. The minority hole density in the n-Ge region
e. The minority electron density in the p-GaAs region
f. The hole diffusion length
g. The emitter-junction current

Solution

a. The built-in voltage in the p-GaAs side is

b. The hole mobility is read from Fig. A-2 in Appendix A as

pp = 400 cm2/V s -
c. The hole diffusion constant is
Dp = 400 X 26 X = 10.40cm2/s
d. The minority hole density in n-Ge is
Microwave Transistors and Tunnel Diodes Chap. 5

e. The minority electron density in the p-GaAs region is

f. The hole diffusion length is

L, = d6 x x 10.40 = 7.90 x cm
g. The emitter-junction current is computed from Eq. (5-2-9) as

5-3 MICROWAVE TUNNEL DIODES

After the publication of Esaki's classic paper on tunnel diodes in 1958, the potential
of tunnel diodes for microwave applications was quickly established. Prior to 1958
the anomalous characteristics of some p-n junctions were observed by many scien-
tists, but the irregularities were rejected immediately because they did not follow the
"classic" diode equation. Esaki, however, described this anomalous phenomenon by
applying a quantum tunneling theory. The tunneling phenomenon is a majority car-
rier effect. The tunneling time of carriers through the potential energy barrier is not
governed by the classic transit-time concept-that the transit time is equal to the
barrier width divided by the carrier velocity-but rather by the quantum transition
probability per unit time. Tunnel diodes are useful in many circuit applications in
microwave amplification, microwave oscillation, and binary memory because of
their low cost, light weight, high speed, low-power operation, low noise, and high
peak-current to valley-current ratio.

5.3- 1 Principles of Operation

The tunnel diode is a negative-resistance semiconductor p-n junction diode. The neg-
ative resistance is created by the tunnel effect of electrons in the p-n junction. The
doping of both the p and n regions of the tunnel diode is very high-impurity con-
centrations of 10'' to loz0atoms/cm3 are used-and the depletion-layer barrier at the
junction is very thin, on the order of 100 A or cm. Classically, it is possible for
those particles to pass over the barrier if and only if they have an energy equal to or
greater than the height of the potential barrier. Quantum mechanically, however, if
the barrier is less than 3 A there is an appreciable probability that particles will tun-
nel through the potential barrier even though they do not have enough kinetic energy
to pass over the same barrier. In addition to the barrier thinness, there must also be
filled energy states on the side from which particles will tunnel and allowed empty
states on the other side into which particles penetrate through at the same energy
level. In order to understand the tunnel effects fully, let us analyze the energy-band
pictures of a heavily doped p-n diode. Figure 5-3-1 shows energy-band diagrams of
a tunnel diode.
Sec. 5.3 Microwave Tunnel Diodes

P F t i o n layer

rgG" lo;-

Distance
(a) Tunnel diode under zero-bias equilibrium

band

Eu

(b) Tunnel diode with applied forward bias

EF is the Fermi level representing the energy state with 50%

probability of being filled if no forbidden band exists
V , is the potential barrier of the junction
E, is the energy required t o break a covalent bond, which is
0.72 eV for germanium and I. I 0 eV for silicon
E, is the lowest energy in the conduction band
E, is the maximum energy in the valence band
V is the applied forward bias
F.B. stands for the forbidden band
B.H. represents the barrier height

Figure 5-3-1 Energy-band diagrams of tunnel diode.

200 Microwave Transistors and Tunnel Diodes Chap. 5

Under open-circuit conditions or at zero-bias equilibrium, the upper levels of

electron energy of both the p type and n type are lined up at the same Fermi level as
shown in Fig. 5-3-l(a). Since there are no filled states on one side of the junction
that are at the same energy level as empty allowed states on the other side, there is
no flow of charge in either direction across the junction and the current is zero, as
shown at point (a) of the volt-ampere characteristic curve of a tunnel diode in Fig.
5-3-2.

Tunnelling current (solid

curve) and injection
current (dashed curve)

I
Forward voltage

I-V characteristic
o f tunnel diode

V"
Forward voltage

Figure 5-3-2 Ampere-voltage characteristics of tunnel diode.

In ordinary diodes the Fermi level exists in the forbidden band. Since the tun-
nel diode is heavily doped, the Fermi level exists in the valence band in p-type and
in the conduction band in n-type semiconductors. When the tunnel diode is forward-
biased by a voltage between zero and the value that would produce peak tunneling
current /,(O < V < V,), the energy diagram is shown in part (1) of Fig. 5-3- 1(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 Microwave Tunnel Diodes 201

from the p type to the n type as shown in sector (1) of Fig. 5-3-2(a). As the forward
bias is increased to V, , 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 I,, 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 V, , 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 I, (or valley current) somewhere in the region where the tunnel-
diode characteristic meets the ordinary p-n diode characteristic. The ratio of peak
current to valley current (IplI,) can theoretically reach 50 to 100. In practice, how-
ever, this ratio is about 15.

5-3-2 Microwave Characteristics

The tunnel diode is useful in microwave oscillators and amplifiers because the diode
exhibits a negative-resistance characteristic in the region between peak current I,
and valley current I,. The I-V characteristic of a tunnel diode with the load line is
shown in Fig. 5-3-3.
Here the abc load line intersects the characteristic curve in three points. Points
a and c are stable points, and point b is unstable. If the voltage and current vary
about b, the final values of I and V would be given by point a or c, but not by b.
Since the tunnel diode has two stable states for this load line, the circuit is called
bistable, and it can be utilized as a binary device in switching circuits. However, mi-

't Load line

Figure 5-3-3 I-V characteristic of tun-

f I
Forward voltage ner diode with load line.
202 Microwave Transistors and Tunnel Diodes Chap. 5

crowave oscillation or amplification generated by the tunnel diode is our major con-
cern in this section. The second load line intersects the I-V curve at point b only.
This point is stable and shows a dynamic negative conductance that enables the tun-
nel diode to function as a microwave amplifier or oscillator. The circuit with a load
line crossing point b in 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. 5-3-3is given by

where R, is the magnitude of negative resistance.

For a small variation of the forward voltage about K, , the negative resistance is
constant and the diode circuit behavior is stable. A small-signal equivalent circuit for
the tunnel diode operated in the negative-resistance region is shown in Fig. 5-3-4.
Here Rs and L, denote the inductance and resistance of the packaging circuit of a
tunnel diode. The junction capacitance C of the diode is usually measured at the val-
ley point; R, is the negative resistance of the diode. Typical values of these parame-
ters for a tunnel diode having a peak current I, of 10 mA are

Yt_i-.
r---------
R, L, Tunnel diode 1

1
I

I I

I I
I
i-------- J
I Figure 5-3-4 Equivalent circuit of tun-
nel diode.

The input impedance Zi, of the equivalent circuit shown in Fig. 5-3-4is given
bv

For the resistive cutoff frequency, the real part of the input impedance Zinmust be
zero. Consequently, from Eq. (5-3-2) the resistive cutoff frequency is given by
I

For the self-resonance frequency, the imaginary part of the input impedance must be
zero. Thus,
Sec. 5.3 Microwave Tunnel Diodes 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.

r------1 r-------
I Tunnel I

I
v I c -R,;
I I
I I
- I
I
I
IL --------
diode
J
I
I I -------
L
diode
J
I

(a) Parallel loading (b) Series loading

Figure 5-3-5 Equivalent circuits of tunnel diodes.

Parallel loading. It can be seen from Fig. 5-3-5(a) that the output power in
the load resistance is given by

Pout =
v2
-
Re
One part of this output power is generated by the small input power through the tun-
nel diode amplifier with a gain of A, and this part can be written

Another part of the output power is generated by the negative resistance, and it is ex-
pressed as

Therefore

and the gain equation of a tunnel diode amplifier is given by

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. In the series circuit shown in Fig. 5-3-5(b) the power gain
A is given by
Microwave Transistors and Tunnel Diodes Chap. 5

The device remains stable in the negative-resistance region without switching if

Re < R, .
A tunnel diode can be connected to a microwave circulator to make a negative-
resistance amplifier as shown in Fig. 5-3-6. A microwave circulator is a multiport
junction in which the power may flow only from port 1 to port 2, port 2 to port 3,
and so on in the direction shown. Although the number of ports is not restricted,
microwave circulators with four ports are most commonly used. If the circulator is
perfect and has a positive real characteristic impedance Ro , an amplifier with infinite
gain can be built by selecting a negative-resistance tunnel diode whose input
impedance has a real part equal to -Ro and an imaginary part equal to zero. The
reflection coefficient from Fig. 5-3-6 is infinite. In general, the reflection coefficient
is given by

--- T"-nn-er -- --
1
I I
diode I
I
I
I
I
Circulator
I
I
I I
I I Figure 5-3-6 Tunnel diode connected
I I
Put L ----------- A 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. O., 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.
34 1. Pergamon Press, London, 1962.
Problems

SUGGESTED READINGS

IEEE Proceedings. 70, No. 1, January 1982. Special issue on very fast solid-state technol-
ogy.
LIAO,S. Y . , Semiconductor Electronic Devices. Prentice-Hall, Inc., Englewood Cliffs, N.J.,
1990.
NAVON,D. H . , Semiconductor Microdevices and Materials. Holt, Kinehart ana w msron, n e w
York, 1986.
SZE,S. M., Semiconductor Devices: Physics and Technology. John Wiley & Sons, New York,
1985.

PROBLEMS

Microwave Bipolar Transistors

5-1. A GaAs n-p-n transistor at 300" K has the following parameters:

Base width
Diffusion length in emitter
Diffusion length in collector
Base resistivity
Emitter resistivity
Collector resistivity
Emitter junction voltage
Collector junction voltage
Cross-section area

Find: a. the impurity densities in the emitter, base and collector

b. the mobilities in the emitter, base, and collector
c. the diffusion lengths in the emitter, base, and collector
d. the equilibrium densities in the emitter, base, and collector
e. the coupling coefficients
f. the emitter and collector currents

5-2. Derive Eq. (5-1-30) from Eq. (5-1-27).

5-3. Derive Eq. (5-1-31) from Eq. (5-1-30).
5-4. Verify Eq. (5-1-34).
5-5. Derive Eqs. (5-1-39), (5-1-40), and (5-1-41).
5-6. A certain silicon bipolar transistor has a maximum electric field intensity Em of
3 x lo5 voltslcm and its carrier has a saturated drift velocity us of 4 X lo6 cmlsec.
The emitter-collector length L is 4 microns.
a. Calculate the maximum allowable applied voltage.
b. Compute the transit time for a charge to transverse the emitter-collector length L.
c. Determine the maximum possible transit frequency.
206 Microwave Transistors and Tunnel Diodes Chap. 5

5-7. A bipolar transistor has voltage-frequency and current-frequency limitations as shown

in Eqs. (5-1-58) and (5-1-59). Derive the power-frequency limitation as shown in Eq.
(5- 1-60).
5-8. A bipolar transistor has a voltage-current-power limitation. Derive the power gain-
frequency relationship as shown in Eq. (5- 1-61).

Heterojunction Bipolar Transistors

5-9. An n-Ge-p-GaAs-n-GaAs heterojunction transistor at 300' K has the following
parameters:

Donor density in the n-Ge region: Nd = 2 x lOI9 ~ m - ~

Acceptor density in p-GaAs region: No = 3 x 10'' cm-3
Hole lifetime: 7p = 4 X sec
Bias voltage at emitter junction: VE = 1.50 V
Cross section: A =4 X cm2

Compute: a. The built-in voltage in the p-GaAs side

b. The hole mobility (read the value from Fig. 2-4-2)
c. The hole diffusion constant
d. The minority hole density in the n-Ge region
e. The minority electron density in the p-GaAs region
f. The hole diffusion length
g. The emitter-junction current

5-10. An n-Ge-p-GaAs-n-GaAs heterojunction transistor at 300" K has the following

parameters:

Donor density in the n-Ge region: Nd = 5 X 10l8 ~ r n - ~

Acceptor density in the p -GaAs region: No = 6 X 1016 em-3
Hole lifetime: rp = 4 X sec
Bias voltage across n-p junction: V, = 0.80 V
Cross section: A = 2 x 10-2cm2

Calculate: a. The built-in voltage in the p-GaAs side

b. The hole mobility (read from Fig. A-2 in Appendix A)
c. The hole diffusion constant
d. The minority hole density in the n-Ge region
e. The minority electron density in the p-GaAs region
f. The hole diffusion length
g. The emitter-junction current

Describe the electronic applications of an HBT.

Tunnel Diodes
5-11. A negative-resistance device is connected through a I-kR resistor in series and a
0.01-pF capacitor in parallel to a combination of a supply voltage V of 10 V and a sig-
Problems

Figure P5-11

nal source V, as shown in Fig. P5- 1l(a). Its I-V characteristic curve is shown in Fig.
P5-1 l(b).
a. Find the negative resistance and forward resistances of the device.
b. Draw the load line on the I-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 sig-
nal voltage of 14 V is applied to the circuit.
e. Draw the new load line on the I-V curve.
f. Find the time constant of the circuit.
g. Compute v , i, i, 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 u , i, i,, 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 am-
plifier.
5-13. A certain microwave tunnel diode has a negative resistance of 69 + j9.7 a. Deter-
mine the resonant-circuit impedance so that the microwave tunnel-diode amplifier will
produce a power gain of 15 dB.
1 Microwave Field-Effect
Transistors

6-0 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 [I]. 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, mi-
crowave characteristics, and power-frequency limitations of unipolar field-effect
transistors. Since the microwave field-effect transistor has the capability of amplify-
ing 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:

I. It may have voltage gain in addition to current gain.

2. Its efficiency is higher than that of a bipolar transistor.
3. Its noise figure is low.
4. Its operating frequency is up to X band.
5. Its input resistance is very high, up to several megohms.
Sec. 6.1 Junction Field-Effect Transistors (JFETs) 209

Among the unipolar field-effect transistors are the junction field-effect transis-
tors, metal-semiconductor field-effect transistors, high electron-mobility transistors,
and the metal-oxide-semiconductor field-effect transistors. All these devices are ana-
lyzed in this chapter.

6.1 JUNCTION FIELD-EFFECT TRANSISTORS (JFETs)

Unipolar field-effect transistors may be in the form of either a p-n junction gate or a
Schottky-barrier gate. The former is called a junctionjeld-effect transistor (JFET),
and the latter is referred to as a MESFET or metal-semiconductor Jield-effect transis-
tor. The JFET was originally proposed by Shockley [I]. Figure 6- 1- 1 shows the
schematic diagram and circuit symbol for an n-channel JFET.

-
6-1 1 Physical Structure

The n-type material is sandwiched between two highly doped layers of p-type mate-
rial that is designated P'. This type of device is called an n-channel JFET. If
the middle part is a p-type semiconductor, the device is called a p-channel JFET.
The two p-type regions in the n-channel JFET shown in Fig. 6-1-1 are referred
to as the gates. Each end of the n channel is joined by a metallic contact. In accor-
dance with the directions of the biasing voltages shown in Fig. 6- 1-1, the left-hand
contact which supplies the source of the flowing electrons is referred to as the
source, whereas the right-hand contact which drains the electrons out of the material
is called the drain. The circuit symbol for an n-channel JFET is also shown in Fig.
6- 1- 1 . The direction of the drain current Id is flowing from the drain to the device.

Figure 6-1-1 Schematic diagram and circuit symbol for an n-channel JFET.
210 Microwave Field-Effect Transistors Chap. 6

For a p -channel JFET, the polarities of the two biasing voltages VRand Vd are inter-
changed, 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 JFET provides higher conductivity and higher speed and is preferred in
most applications to the p -channel JFET.

6-1-2 Principles of Operation

Under normal operating conditions, when the gate voltage VR is zero, the drain cur-
rent Id is also zero. The channel between the gate junctions is entirely open. When a
small drain voltage Vd is applied between the drain and source, the n-type semicon-
ductor bar acts as a simple resistor, and the current Idincreases linearly with Vd . If a
reverse gate voltage VR is applied across the p-n gate junctions, the majority of free
electrons are depleted from the channel, and the space-charge regions are extended
into the channel. As the drain voltage Vd is further increased, the space-charge re-
gions expand and join together, so that all free electron carriers are completely de-
pleted in the joined region. This condition is called pinch off. When the channel is
pinched off, the drain current Idremains almost constant, but the drain 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. The Poisson's equation for the voltage
in the n channel, in terms of the volume charge density is given by

where p = volume charge density in coulombs per cubic meter

q = charge in coulombs
Nd = electron concentration in electrons per cubic meter
E, = permittivity of the material in farads per meter
E, = E , E , , er is the relative dielectric constant
E, = 8.854 X 10-l2 Flm is the permittivity of free space

Integration of Eq. (6-1-1) once and application of the boundary condition of the
dV
electric field E = - - = 0 at y = a yield
dy

a) volts per meter (6- 1-2)

Integration of Eq. (6- 1-2) once and application of the boundary condition V = 0 at
y = 0 result in

volts
Sec. 6.1 Junction Field-Effect Transistors (JFETs)

Then the pinch-off voltage V,, at y = a is expressed as

qNda
vp = - ~ E S
volts

where a is the height of the channel in meters.

Equation (6- 1-4) indicates that the pinch-off voltage is a function of the doping
concentration N d and the channel height a . Doping may be increased to the limit set
by the gate breakdown voltage, and the pinch-off voltage may be made large enough
so that drift-saturation effects just become dominant.
From Fig. 6- 1- 1, the pinch-off voltage under saturation condition can be ex-
pressed as
vp = v d + I VKI + $" (6- 1-5)
I
where V,] = absolute value of the gate voltage
$(, = built-in or barrier voltage at the junction
The saturation drain voltage is then given by
qNda
Vd sat = -- I V
RI - $0
~ E S

The JFET has a conducting channel between the source and the drain elec-
trodes 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: Pinch-Off Voltage of a Silicon JFET

A certain Si JFET has the following parameters:

Channel height: a = 0.1 p m

Electron concentration: Nd = 8 x 10'' cm-3
Relative dielectric constant: E, = 11.80

Calculate the pinch-off voltage

Solution From Eq. (6-1-6), the pinch-off voltage is

= 6.66 volts

6-1-3 Current- Voltage (I- V ) Characteristics

The drain current of an n-channel JFET is dependent on the drain and gate voltages
and the channel resistance. The n-channel resistance can be expressed as
Microwave Field-Effect Transistors Chap. 6

where p,, = electron mobility

Z= distance in z direction
L = length in x direction
W = depletion-layer width
a = width between two p-n junctions
The drain voltage across an elemental section dx of the channel is given by
Iddx
dV(x) = IddR =
2qpnNd Z [a - W ( x ) ]
From Eq. (6- 1 -6), the depletion-layer width can be written as

Then
W (x) = [2r.T[v(x) +-d
iV,
3
I + h] '"

Er
Wdw = - dV
4Nd
The drain current Idis expressed by

Substituting Eq. (6-1-10) into Eq. (6-1-11) and integrating from x = 0 to x = L

give the drain current as

Boundary conditions. The channel-to-gate voltage can be written from

Eq. (6- 1-4) as

The depletion-layer width is

Then
v*I + $0
a t x = Oand V(x) = 0 (6-1-15)

and
W : = a 2( +
~ d 1 1 + (kY) atx=LandV(x)=Vd (6-1-16)
Sec. 6.1 Junction Field-Effect Transistors (JFETs) 213

Substituting Eqs. (6- 1 - 15) and (6- 1 - 16) into Eq. (6- 1- 12) yields the drain current

p,,q* N:Za3 .
where I,, = 1s the pinch-off current.
LE,
Figure 6- 1-2 shows the normalized ideal current-voltage characteristics for a
pinch-off voltage of 3.2 volts [ I ] .

1 Pinch off

0 2 4 6 8 10
Drain voltage Vd in volts

Figure 6-1-2 Normalized ideal I-V characteristics of a typical JFET

Linear region. In the linear region at Vd + (1 VRI + $J, the drain current
can be expressed from Eq. (6- 1 - 12) as

The channel conductance (or the drain conductance) is given by

The mutual conductance (or transconductance) is expressed as

214 Microwave Field-Effect Transistors Chap. 6

Saturation region. At the pinch-off voltage, the drain current becomes sat-
urated. By setting Vp = Vd + I VRI + $", the saturation drain current is given by

and the corresponding saturation drain voltage as

Vd sat = Vp + I VY( + $, (6-1-22)

The mutual conductance in the saturation region is then expressed as

Ip - 2pnqNdaZ
where - -
VP L
It should be noted that the g, 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

Then the amplification factor for a JFET may be defined as

The cutoff frequency is given by

LZE, .
where C, = -is the 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 cen-
ter of the channel, but converged as shown in Fig. 6-1-2. As the drain voltage Vd
and drain current Idare further increased, the channel is finally pinched off.

Breakdown voltage. As the drain voltage Vd increases for a constant gate

voltage V, , the bias-voltage causes avalanche breakdown across the gate junction,
and the drain current Idincreases sharply. The breakdown voltage is shown in Fig.
6 - 1-3 by the relationship
Sec. 6.1 Junction Field-Effect Transistors (JFETs)

U r a ~ nboltdpc Vd In volt\

Figure 6-1-3 Breakdown voltage for a typical JFET.

Example 6-1-2: Current of a JFET

A silicon JFET at 300' K has the following parameters:

Electron density: Nd = 1 x IOl7

Hole density: N, = 1 x l o i 9cm-'
Relative dielectric constant: E, = 11.8
Channel height: a = 0.2 X cm
Channel length: L = 8 x 10-4cm
Channel width: Z = 50 x 10-4cm
Electron mobility: p, = 800 crn2/V.s
Drain voltage: Vd = lOV
Gate voltage: V,= - 1.5V
Compute:

a. The pinch-off voltage in volts

b. The pinch-off current in rnA
c. The built-in voltage in V
d. The drain current in rnA
e. The saturation drain current at V, = 0
f. The cutoff frequency
Solution

a. The pinch-off voltage is

Microwave Field-Effect Transistors Chap. 6

b. The pinch-off current is

800 x (1.6 x 10-'v)2(1x 10'7)2(50x 10-4)(0.2 x
I,, =
8 X X 8.854 X X 11.8

c. The built-in voltage is

d. The drain current is computed from Eq. (6- 1-17) as

e. The saturation drain current is computed from Eq. (6-1-21) to be

f. The cutoff frequency is computed from Eq. (6- 1-26) as

= 4.8 GHz
The operating frequency must be less than 4.8 GHz.

6-2 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-semiconductorjield-effect tran-
sistor (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 para-
metric 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 Metal-Semiconductor Field-Effect Transistors (MESFETs) 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.

6-2-1 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 ei-
ther the epitaxial process or the ion implantation method and those methods are dis-
cussed in Chapter 12. Figure 6-2- 1 shows schematically a simple MESFET in GaAs.

Source Gate Drain

P 7 P P
Contact
metal

Buffer
layer -
Depletion
Semi-insulating region in
GaAs substrate
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 high-
resistivity substrate (near lo8 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 A1 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 n-
type 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 im-
plantation 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 Microwave Field-Effect Transistors Chap. 6

After the fabrication processes are completed, the individual transistor is sepa-
rated 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 A1 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.

D X L 2501 A-P70 D X L l-SOlA

StClndardPackage GaAs FET Chip D ~ m e n s i o n s
Herinetlc nllcrostrlp Ilne packages

All i l ~ i n c n r ~ o narc
\ ~n 1nc11e\ t i l l t111ne11s101ib
arc ~n microns

Figure 6-2-2 Dimensions of package and chip GaAs MESFETs.

The MESFET device is of an interdigitated structure, fabricated by using an n-

type GaAs epitaxial film about 0.15- to 0.35-pm thick on a semi-insulating sub-
strate about 100 p . The n-channel layer is doped with either sulphur or tin in a dop-
ing concentration N between 8 X 10I6/cm3and 2 X 101'/cm3. The electron mobility
in the layer is in the range of 3000 to 4500 cm2/V.s. 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 sub-
strate. A buffer layer of about 3 p m with a doping concentration of to 1016cm-3
is often fabricated between the active n-type epitaxial layer and the semi-insulating
substrate.

6-2-2 Principles of Operation

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 car-
riers (electrons) flow in the n-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 Metal-Semiconductor Field-Effect Transistors (MESFETs) 219

Sourcc (;ate Drain

T T ?

Figure 6-2-3 Schematic diagram and circuit symbol of a GaAs MESFET.

causes a voltage drop along its length so that the Schottky barrier-gate electrode be-
comes 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 be-
tween 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 V,. This phenomenon is analogous to the characteristics of the collector
current I,. versus the collector voltage V, with the base current Ib as a parameter in a
bipolar transistor. In other words, a family of curves of the drain current Id versus
the voltage VdSbetween the source and drain with the gate voltage V, 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

For a fixed drain-to-source voltage Vds, the drain current Id is a function of the
reverse-biasing gate voltage V,. Because the drain current Id is controlled by the
field effect of the gate voltage V,, this device is referred to as the field-effect transis-
tor. When the drain current Idis continuously increasing, the ohmic voltage drop be-
tween the source and the channel reverse-biases the p-n junction further. As a re-
sult, 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 V,. 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
Microwave Field-Effect Transistors Chap. 6

Drain voltage Vds i l l volts

Figure 6-2-4 Current-voltage characteristics of a typical n-channel GaAs

MESFET.

where p = volume charge density in coulombs per cubic meter

q = charge in coulombs
Nd = electron concentration in electrons per cubic meter
E, = permittivity of the material in farads per meter
E, = E,E,, E, is the relative dielectric constant
E, = 8.854 X Flm is the permittivity of free space

Integration of Eq. (6-2-2) twice and application of the boundary condition yield the
pinch-off voltage at y = a as

qNda
v, = - volts
2 ~ 5

where a is the height of the channel in meters.

Equation (6-2-3) indicates that the pinch-off voltage is a function of the doping
concentration Nd and the channel height a. Doping may be increased to the limit set
by the gate breakdown voltage, and the pinch-off voltage may be made large enough
so that drift saturation effects just become dominant.
Sec. 6.2 Metal-Semiconductor Field-Effect Transistors (MESFETs) 221

Example 6-2-1: Pinch-Off Voltage of a MESFET

A certain GaAs MESFET has the following parameters:

Channel height: a = 0.1 p m

Electron concentration: Nd = 8 X 10'' cm-'
Relative dielectric constant: 6, = 13.10

Calculate the pinch-off voltage.

Solution From Eq. (6-2-3) the pinch-off voltage is

= 6.00 volts

Small-Signal Equivalent Circuit

6-2-3

For microwave frequencies, the metal-semiconductor field-effect transistor

(MESFET) has a very short channel length, and its velocity saturation occurs in the
channel before reaching the pinched path. The electronic characteristics of a
MESFET depend not only on the intrinsic parameters, such as g,, G d ,R,, CRs,and
, also on the extrinsic parameters R,, R,, Cds,RP and C,. Figure 6-2-5 shows
C X dbut
the cross section of a MESFET and its equivalent circuit.
The microwave properties of metal-semiconductor field-effect transistors were
investigated and analyzed by many scientists and engineers. The noise behavior of
these transistors at microwave frequencies has been investigated and measured by
van der Ziel and others [ 5 ] .

1. Intrinsic elements:
g, = transconductance of the MESFET
Gd = drain conductance
R, = input resistance
C,, = gate-source capacitance
CKd= gate-drain (or feedback) capacitance
2. Extrinsic elements:
R, = gate metallization resistance
R, = source-gate resistance
Cd,?= drain-source capacitance
R,, = gate bonding-pad parasitic resistance
Microwave Field-Effect Transistors Chap. 6

Source F L S 4 Gate ( + ~ d d Drain

r--p---------------1
I Intrinsic I
Gate Drain
'3 0

0 0
Source S So~~rcc

Figure 6-2-5 Cross section and equivalent circuit of a MESFET.

C, = gate bonding-pad parasitic capacitance

Zt = load impedance

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 in-
crease the channel doping N as high as possible in order to decrease the relative
influence of the feedback capacitance CRdand to increase the transconductance g,
and the dc open-circuit voltage gain. However, an increase in concentration N de-
creases the breakdown voltage of the gate. A doping of 1018 per cubic centimeter
might be an upper limit.
Sec. 6.2 Metal-Semiconductor Field-Effect Transistors (MESFETs)

6-2-4 Drain Current b

The drain current Idof a Schottky barrier-gate FET is expressed [6] as

3 ( p 2 - p2) - 2(p3 - p3)
Id = amperes (6-2-4)
1 + 7 ( p 2- p2)
qNdpaZVi .
where I, = is the saturation current for the Shockley case at V, = 0
3L
p = low-field mobility in square meters per volt-second
q = 1.6 X 10-l9 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)
v, + I VRI is the normalized sum of the drain and gate voltages with
.=( v, 1
respect to the pinch-off voltage
p = (y) 1/2
is the normalized gate voltage with respect to the pinch-off
voltage
Vp I = 2 is the normalized drift velocity with respect to the saturation
' = U,L as drift
velocity
v , = saturation drift velocity
a = is the drift velocity in the channel
1 + -PEX
Us
Ex = absolute value of the electric field in the channel
Figure 6-2-4 shows a plot of the drain current I d versus the drain voltage Vds
with the gate voltage V,, as a parameter for a typical n -channel GaAs MESFET. The
drain current has a maximum at u = u, given by

where z = pl V, I/(U,L) is the ratio between the small-field velocity extrapolated

linearly to the field V,/L and the saturation velocity
Substitution of Eq. (6-2-5)into Eq. (6-2-4)yields the saturation drain current as

The transconductance in the saturation region is given by

224 Microwave Field-Effect Transistors Chap. 6

In practice, the drain current and mutual conductance of a GaAs MESFET can be
expressed by

and

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 x, , 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 chan-
nel cross section is narrowing. In addition, the electrons are moving progressively
slower with increasing x. Exactly the opposite occurs between xz and x3. The chan-
nel 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)u (x)d (x) (6-2- 10)

where Z = channel depth or width

n(x) = Nd is the density of conduction electrons
v (x) = drift velocity
d(x) = conductive layer thickness
x = coordinate in the direction of the electron drift

The current-voltage characteristics of a Si MESFET are shown in Fig. 6-2-7 [7]. In

Fig. 6-2-7(a), there is no metal gate electrode. At the surface of the conducting
layer, the source and drain contacts are made. When a positive voltage Vd, is applied
to the drain, electrons will flow from source to drain. In Fig. 6-2-7(b), a metal gate
is added and shorted to the source. When a small drain voltage is applied, a deple-
tion layer is created. The current I,,, flowing from drain to source is indicated by Eq.
6-2-7. When the drain voltage Vd, is increased, the depletion layer becomes wider.
The resulting decrease in conductive cross section d must be compensated by an in-
crease of electron velocity ti to maintain a constant current through the channel. As
the drain voltage is increased further, the electrons reach their saturation velocity u,
Sec. 6.2 Metal-Semiconductor Field-Effect Transistors (MESFETs)

V,=3V I D / Z(mA/cm)

0 2 4
V, (V)

Electric field

Electron drift
velocity

I
I I '
! I I I

Figure 6-2-6 Channel cross section, electric field, drift velocity, and space charge
distribution in the channel of a GaAs MESFET in the saturation region.

as shown in Fig. 6-2-7(c). When the drain voltage is increased beyond Vd,,,, the de-
pletion 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 V,, is
more negative, the channel is almost pinched off as shown in Fig. 6-2-7(f), and the
drain current Id, is nearly cut off.
sw
Microwave Field-Effect Transistors Chap. 6

Ohmic
contact

+
C
!{--.-----I---
\
el":t'
layer
"D

,' N o gate ,/ N o gate

v, =0

VDSAT VD "G

x2
Ll
Constant n -4 4 -Constant IJ LI <L
n = .VD ' u = us,
(d)

"1 ,---
N o gate

Figure 6-2-7 Electric-field diagrams and current-voltage characteristics of a Si

MESFET.

Example 6-2-2: Current-Voltage (I-V) Characteristics of a GaAs MESFET

A typical n-channel GaAs MESFET has the following parameters:

Electron concentration: Nd 8 x 10"

=
Channel height: a = 0.1 p m
Relative dielectric constant: E, = 13.1
Channel length: L = 14 p m
Channel width: Z=36pm
Electron mobility: p = 0.08 m 2 / V s
= 800 cm2/V. s
Sec. 6.2 Metal-Semiconductor Field-Effect Transistors (MESFETs)

Drain voltage: Vd = 5 volts

Gate voltage: V , = -2 volts
Saturation drift velocity: v , = 2 X 10srn/s

a. Calculate the pinch-off voltage.

b. Compute the velocity ratio.
c. Determine the saturation current at V' = 0.
d. Find the drain current Id.

Solution

a. From Eq. (6-2-3), the pinch-off voltage is

V,, = 1.6 X lO-I9 x 8 X lo2' X 10-14/(2 X 8.854 X lo-'* X 13.1)
= 5.52 volts

b. From Eq. (6-2-4), the velocity ratio is

c. From Eq. (6-2-4), the saturation current at V, = 0 is

qNd~az&,
I,, =
2L

d. From Eq. (6-2-4), the u and p factors are

then the drain current is

228 Microwave Field-Effect Transistors Chap. 6

6-2-5 Cutoff Frequency f,, and Maximum Oscillation

Frequency f,,,

The cutoff frequency of a Schottky barrier-gate FET in a circuit depends on the way
in which the transistor is being made. In a wideband lumped circuit the cutoff fre-
quency is expressed [7] as

where g, = transconductance
C,, = gate-source capacitance = -
L = gate length
u, = saturation drift velocity
It is interesting to note that the cutoff frequency of the lumped circuit analysis as
shown in Eq. (6-2- 1 1) is different from the charge-carrier transit-time cutoff fre-
quency 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

where Rd = drain resistance

g, = device transconductance
E, = electric field at the pinch-off region in the channel
u, = saturation normalization of u
v,, p , and p are defined previously

For p = 0 (i.e. V,, = 0 ) and 71

We have
+ 1 , SO that$., = -0
47rL
and u,

where y = 0.14 for pE,,/v, = 13 and y = 0.18 for pE,/v, = 20 in the case of
GaAs.
It has been found experimentally [6] that the maximum frequency of oscilla-
tion for a gallium arsenide FET with the gate length less than 10 p m is

where L = gate length in meters. The best value of L is 0.5 p m .

Sec. 6.2 Metal-Semiconductor Field-Effect Transistors (MESFETs) 229

The maximum frequency of oscillation f,,,, is similar to the cutoff frequencyf,,,

determined by the transit time. From Eq. (5-1-58) the charge-carrier transit-time
cutoff frequency is

L .
where r = - 1s the transit time in seconds
0s
L = gate length in meters
v, = saturation drift velocity in meters per second
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 Ir, is
2 X lo7 cmlsec for GaAs at an electric field of 3 kVIcm and 8 X lo6 cmlsec for sili-
con at 15 kVIcm. In comparing Eq. (6-2-15) with Eq. (6-2-1 l), 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

J.0

= 1( R , + Rd
Imax R, + R, 1'" Hz

where Rd = drain resistance

R , = source resistance
R, = gate metallization resistance
R, = input resistance

Example 6-2-3: Cutoff Frequency of a MESFET

A certain GaAs MESFET has the following parameters:

a. Determine the cutoff frequency.

b. Find the maximum operating frequency.

Solution

a. From Eq. (6-2-1 I ) , the cutoff frequency is

230 Microwave Field-Effect Transistors Chap. 6

b. From Eq. ( 6 - 2 -16), the maximum frequency is

-
-
13'262 lo' ( 2 . 5 + 3 + 2.5
= 49.73 GHz

6.3 HIGH ELECTRON-MOBILITY TRANSISTORS (HEMTs)

The evolution of high-speed GaAs integrated circuits (ICs) is the result of continuous
technological progress utilizing the superior electronic properties of gallium arsenide
compared with those of silicon. Electron mobility in the MESFET channel with typ-
ical donor concentrations of about 10'' cm-3 ranges from 4000 to 5000 cm2/V. s at
room temperature. The mobility in the channel at 77" K is not too much higher than
at room temperature because of ionized impurity scattering. In undoped GaAs, how-
ever, electron mobility of 2 to 3 X lo5 cm2/V. s has been achieved at 77" K. The
mobility of GaAs with feasibly high electron concentrations for facilitating the fabri-
cation of devices was found to increase through modulation-doping technique:
demonstrated in GaAs-AlGaAs superlattices [8]. A high electron-mobility transistor
(HEMT), based on a modulation-doped GaAs-AlGaAs single heterojunction struc-
ture, was developed [9]. HEMTs have exhibited lower noise figure and higher gain at
microwave frequencies up to 70 GHz, and it is possible to construct HEMT am-
plifiers at even higher frequencies of operation. The major improvements over MES-
FETs include shorter gate lengths, reduced gate and source contact resistances, and
optimized doping profiles.

1 Physical Structure
6-3-

The basic structure of a HEMT is a selectively doped GaAs-AlGaAs heterojunction

structure as shown in Fig. 6-3- 1.
An undoped GaAs layer and an Si-doped n-type AlGaAs 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 sand-
wiched 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 High Electron-Mobility Transistors (HEMTs)

Source Gate SIN Drain

7 IItC; Undoped AlGaAs

Undoped GaAs

1 GaAs S.I.
S.I. buffcr
Figure 6-3-1 A cross section of a
HEMT (From K. Togashi et al. [lo];
reprinted by permission of Microwave
Journal.)

,Ohmic contact ,Insulator Photoresist

r GaAs \.

Semi-insulating
.
GaAs substrate

R l t (CCI2F2/He) Interconnect
Gate metal metal
+ + + t i /

I S.I. GaAs S.I. GaAs S.I. GaAs

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.)

an AuGe eutectic alloy and an Au overlay alloy to produce ohmic contacts

with the electron layer.
2. Opening gate windows: The fine gate patterns are formed for E-HEMTs as the
top GaAs layer and thin Alo.3Gao.7Asstopper are etched off by nonselective
chemical etching.
3. Selective dry etching: With the same photoresist process for formation of gate
patterns in D-HEMTs, selective dry etching is performed to remove the top
GaAs layer for D-HEMTs and also to remove the GaAs layer under the thin
A h 3Gao., As stopper for E-HEMTs.
232 Microwave Field-Effect Transistors Chap. 6

4. Gate metallization: Schottky contacts for the E- and D-HEMT gate are pro-
vided 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 intercon-
necting metal, composed of Ti, Pt, and Au, to the device terminals are pro-
vided 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 DHEMT
A A
I r \

Ohmic Interconnect Insulator ( S O 2 )

contact Gatc metal /
Insulator ( S i 0 2 )

GaAs
AlGaAs

Electron layel

Semi-~nsulatingGaAs substrate
I
Figure 6-3-3 Cross-sectional view of a DCFL HEMT. (After M. Abe et al. [9];
reprinted by permission of IEEE, Inc.)

6-3-2 Operational Mechanism

Since GaAs has higher electron affinity, free electrons in the AlGaAs layer are trans-
ferred to the undoped GaAs layer where they form a two-dimensional high-mobility
electron gas within 100 A of the interface. The n-type AlGaAs layer of the HEMT is
depleted completely through two depletion mechanisms (1 1):

1. Trapping of free electrons by surface states causes the surface depletion.

2. Transfer of electrons into the undoped GaAs layer brings about the interface
depletion.

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 thick-
ness, 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 High Electron-Mobility Transistors (HEMTs) 233

voltage is applied to the gate, electrons accumulate at the interface and form a two-
dimensional electron gas.
The electron concentration can control D-(depletion-mode) and E-(enhance-
ment-mode) HEMT operations. As temperature decreases, electron mobility, which
is about 8000 cm2/V. s at 300" K , increases dramatically to 2 X lo5 cm2/V.s at
77" because of reduced phonon scattering. When the temperatures decrease fur-
ther, the electron mobility of 1.5 x lo6 cm2/V.s at 50°K and 2.5 X lo6 cm2/V.s
at 4.5"K have been demonstrated.

Performance Characteristics
6-3-3

High electron mobility transistor (HEMT) amplifiers for 40 to 70 GHz have been
constructed. The 60-GHz amplifier exhibited a gain of 4.5 to 6.5 dB across the fre-
quency band 56 to 62 GHz, and had an associated noise figure of 6 dB measured at
57.5 GHz. The 72-GHz amplifier achieved a gain of 4 to 5 dB with a bandwidth of
2.5 GHz [ l l ] .

Current-voltage (I-V) characteristics. The drain current can be evaluated

from the following basic equation

where q = electron charge

n(z) = concentration of the two-dimensional electron gas
W = gate width
v (z) = electron velocity
Figure 6-3-4 shows the I-V characteristics of a HEMT amplifier with W =
150 p m and L, (gate length) = 0.6 prn at 30 GHz.
The major advantages of a HEMT are higher frequency, lower noise, and
higher speed. Table 6-3-1 compares HEMT data with other semiconductor elec-
tronic devices.

Example 6-3-1: Current of a HEMT

A HEMT has the following parameters:

Gate width: W = 150 krn

Electron velocity: v(z) = 2 X lo5 m/s
Two-dimensional electron-gas density: n(z) = 5.21 X lOI5 m-2

Determine the drain current of the HEMT.

Solution From Eq. (6-3-l),the drain current is
Ids = qn (z)Wu(2)
= 1.6 x 10-l9 x 5.21 x 1015 x 150 x x 2 x lo5
= 25 mA
Microwave Field-Effect Transistors Chap. 6

Vds In volts

Note:

-A for measured values

for predicted values

Figure 6-3-4 I-V characteristics of a HEMT.

TABLE 6-3-1 HEMT PERFORMANCE COMPARED WITH OTHER DEVICES

Device Frequency (GHz) Noise Power Speed

-

HEMT Up to 70 Very Good Very Good Excellent

GaAs MESFET 40 Good Good Good
GaAs-AlGaAs HBT* 20 Good Good Excellent
Si MOSFET 10 Poor Very good Very poor
Si bipolar transistor 1 Poor Poor Good

* HBT = heterojunction bipolar transistor

Equivalent circuit. In order to predict or calculate the values for a small- or

large-signal HEMT amplifier, the following high-frequency equivalent circuit,
shown in Fig. 6-3-5, may be useful.
It is difficult to compare the optimized performance of most major semicon-
ductor devices. The switching delay time of GaAs MESFETs is two or three times
longer than that of HEMTs. The GaAs-AlGaAs heterojunction bipolar transistor
(HBT) can achieve the same high-speed performance as the HEMT. The ultimate
speed capability, limited by cutoff frequency f~, is more than 100 GHz, and the
HBT also has the merit of flexible fan-out loading capability. The silicon MOSFET
and the bipolar transistor have excellent performance in threshold voltage uniformity
and controllability with no material problems, and they are easy to fabricate despite
complex processing steps. HEMTs are very promising devices for high-speed, very
large-scale integration (VLSI) with low-power dissipation, but they require new
Sec. 6.3 High Electron-Mobility Transistors (HEMTs)

LR R, C ~ d RYd R 'i L 'i

0.15 n H 8R 0.01 13F 0.001 52 2.2R 0.14nH
Gate - W T Drain

Cis --
0.18 p F - -

Rxr 1.8 11s

4. h n

Figure 6-3-5 Equivalent circuit of a HEMT.

technological breakthroughs to achieve the large-scale integration (LSI) quality of

GaAs-AlGaAs material. Such technologies include molecular-beam epitaxy (MBE),
organic metal-vapor phase epitaxy (OMPVE), and self-alignment device fabrica-
tions. The excellent controllability of MBE growth can regulate the ratio of standard
deviation of threshold voltage to the logic voltage swing.
The vertical threshold sensitivity is defined by the differential threshold
voltage to the thickness of the Al, Gal-, As layer and is expressed as

where Vh = threshold voltage in volts

e
= AlGaAs layer thickness in meters
q = electron charge
Nd = donor concentration
I),,,, = metal-semicondcutor Schottky barrier potential between A1 and GaAs
E, = conduction bandedge difference between GaAs and AlGaAs
E = E,G is the permittivity of AlGaAs
E, = dielectric constant of AlGaAs

Example 6-3-2: Sensitivity of HEMT

A HEMT has the following parameters:

Threshold voltage: Kh = 0.13 V

Donor concentration: Nd = 2 X m-3
Metal-semiconductor Schottky
barrier potential: hS
= 0.8 V
Microwave Field-Effect Transistors Chap. 6

GaAs bandgap: E,, = 1.43 V

AIGaAs bandgap: E,, = 1.80 V
AlGaAs dielectric constant E, = 4.43

Compute:

a. The conduction band-edge difference between GaAs and AlGaAs

b. The sensitivity of the HEMT

Solution

a. The conduction band-edge difference is

AE, = E,, - E,, = 1.80 - 1.43 = 0.37 V

b. The sensitivity of the HEMT is

6-3-4 Electronic Applications

The switching speed of a HEMT is about three times as fast as that of a GaAs MES-
FET. The largest-scale logic integrated circuit (IC) with HEMT technology has
achieved the highest speed ever reported among 8 X 8 bit multipliers. The switching
delay time of a HEMT is below 10 picosecond with a power dissipation reported at
about 100 pW. Therefore, HEMTs are promising devices for very large-scale inte-
gration, especially in very high-speed supercomputers, star wars, and space commu-
nications.
Information processing in the 1990s will require ultrahigh-speed computers,
having high-speed large-scale integration circuits with logic delays in the sub-
hundred-picosecond range. A 4K X 1 bit static random-access memory (SRAM)
device consists of a memory cell of 55 p m x 39 p m . Its normal read-write opera-
tion was confirmed both at 300" K and 77" K . The minimum address access time ob-
tained was 2.0 nanosecond with a chip dissipation power of 1.6 W and a supply
voltage of 1.54 V. The HEMT SRAMs have demonstrated better performance than
the SiMOS, BJT, and GaAs MESFET SRAMs.
Sec. 6.4 Metal-Oxide-SemiconductorField-Effect Transistors (MOSFETs) 237

6-4 METAL-OXIDE-SEMICONDUCTOR FIELD-EFFECT

TRANSISTORS (MOSFETs)

The metal-oxide-semiconductor field-effect transistor (MOSFET) is a four-terminal

device, and its current flow is controlled by an applied vertical electric field. The
four terminals are denoted as the source, gate, drain, and substrate. When the gate
bias voltage is zero, the two back-to-back p-n junctions between the source and
drain prevent current flow in either direction. When a positive voltage is applied to
the gate with respect to the source (source and substrate are common), negative
charges are induced in the channel to provide current flow. Since the current is con-
trolled by the electric field, this type of device is called the junctionjeld-effect tran-
sistor (JFET). The MOSFETs have superseded the bipolar junction transistors in
many electronic applications because their structures are simple and their costs are
low. In addition, the n-channel MOSFET (NMOS), p -channel MOSFET (PMOS),
complementary MOSFET (CMOS), logic-gate memories, and the charge-coupled
devices (CCDs) have emerged as important semiconductor electronic devices. All
these devices are discussed in this section.
The metal-insulator-semiconductor field-effect transistor (MISFET) may be
formed by a metal such as aluminum (Al) and a semiconductor, such as Ge, Si, or
GaAs, with an insulator such as S O 2 , Si3N4, or A1203 sandwiched between. If the
structure is in the form of Al-SiOz-Si, it is called a MOSFET. This device is very
useful in very large-scale integrated microwave circuits. In the near future, a
microwave-device chip of 1 - p m dimension containing 1 million or more devices
will be commercially available. The basic component of a MOSFET is the MIS
diode which was described previously.

6-401 Physical Structures

The metal-oxide-semiconductor field-effect transistor (MOSFET) is a four-terminal

device. There are both n -channel and p -channel MOSFETs. The n -channel MOS-
FET consists of a slightly doped p-type semiconductor substrate into which two
highly doped 'n sections are diffused, as shown in Fig. 6-4-1. These n+ sections,
which act as the source and the drain, are separated by about 0.5 p m . A thin layer
of insulating silicon dioxide (SiOz) is grown over the surface of the structure. The
metal contact on the insulator is called the gate.
Similarly, the p-channel MOSFET is made of a slightly doped n-type semi-
conductor with two highly doped p+-type regions for the source and drain. The heav-
ily doped polysilicon or a combination of silicide and polysilicon can also be used as
the gate electrode. In practice, a MOSFET is commonly surrounded by a thick oxide
to isolate it from the adjacent devices in a microwave integrated circuit. The basic
device parameters of a MOSFET are as follows: L is the channel length, which is the
distance between the two nf-p junctions just beneath the insulator (say, 0.5 p m ) , Z
is the channel depth (say, 5 p m ) , d is the insulator thickness (say, 0.1 p m ) , and r, is
the junction thickness of the n+ section (say, 0.2 p m ) .
Microwave Field-Effect Transistors Chap. 6

Metal

n-channel

Figure 6-4-1 Schematic diagram of an n-channel MOSFET.

6-4-2 Electronic Mechanism

When no voltage is applied to the gate of an n-channel MOSFET, the connection be-
tween the source electrode and the drain electrode corresponds to a link of two p-n
junctions connected back-to-back. The only current that can flow from the source to
the drain is the reverse leakage current. When a positive voltage is applied to the
gate relative to the source (the semiconductor substrate is grounded or connected to
the source), positive charges are deposited on the gate metal. As a result, negative
charges are induced in the p-type semiconductor at the insulator-semiconductor in-
terface. A depletion layer with a thin surface region containing mobile electrons is
formed. These induced electrons form the n-channel of the MOSFET and allow the
current to flow from the drain electrode to the source electrode. For a given value of
the gate voltage V,, the drain current Id will be saturated for some drain voltages Vd.
A minimum gate voltage is required to induce the channel, and it is called the
threshold voltage K h . For an n -channel MOSFET, the positive gate voltage V, must
be larger than the threshold voltage K h before a conducting n-channel (mobile elec-
trons) is induced. Similarly, for a p-channel MOSFET, the gate voltage V, must be
more negative than the threshold voltage Kh before the p-channel (mobile holes) is
formed.

6-4-3 Modes of Operation

There are basically four modes of operation for n-channel and p-channel MOSFETs.

1. n-Channel Enhancement Mode (normally OFF). When the gate voltage is

zero, the channel conductance is very low and it is not conducting. A positive
Sec. 6.4 Metal-Oxide-SemiconductorField-Effect Transistors (MOSFETs) 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 equi-
librium (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 I-V characteristics, and transfer characteristics.

Enhancement Enhancement
(normally off) (normally o f f )

p-channel

,I

Deplet~on Deplet~on Figure 6-4-2 Four modes of operation

(normally o n ) (normally o n )
for MOSFETs: (a) n channel and (b) p
(a) (b) channel.

6-4-4 Drain Current and Transconductance

Drain current. The drain current Id of the MOSFET is dependent of the

drain voltage Vd. It first increases linearly with the drain voltage in the linear region
and then gradually levels off to a saturated value in the saturation region. Figure
6-4-4 shows the current-voltage characteristic curves of an n-channel MOSFET [12].
240 Microwave Field-Effect Transistors Chap. 6

Electrical Output Transfer

Type charactenst~c
symbol charactenst~c

nchannel
enhancement
(normally o f f )
! 0 v, +
"G

nchannel
deplet~on
(normally on)
0 t
"G

pchannel
enhancement
i - "D 0
"G

(normally o f 0

pchannel
depletion i - "D
YG

Figure 6-4-3 Electric symbols and output and transfer characteristics of the four
modes of MOSFETs.

The drain current Id is given by [12] as

(6-4-1 )
where p, = electron carrier mobility
Ei
Ci = - is the insulator capacitance per unit area
d
ei = insulator permittivity
V, = gate voltage
(Clb = (E; - EF)/qis the potential difference between the Fermi level EF and
the intrinsic Fermi level E;
Vd = drain voltage
E, = semiconductor permittivity
q = carrier charge
N, = acceptor concentration
In the linear region, the drain voltage is small arid Eq. (6-4-1) becomes
Sec. 6.4 Metal-Oxide-SemiconductorField-Effect Transistors (MOSFETs)

I I I

(V, - Vth) = I0 V

9
40 - -
Saturation
region

-
-
.
4
4
-2
.
7

-
6

-
4

3
2
1
I I I 1
2 4 6 8 10 12 14 16 18
Vd in volts

Figure 6-4-4 Current-Voltage (I-V) curves of an n-channel MOSFET.

z
Id = -pnCi(Vg - V,h) Vd for Vd G (Vg - Vlh)
L

where v.. = 2$b + ~2 ( E ~ ~ N . ~ ) " '

In the saturation region, the drain current is given by
mZ
I d sat -pn Ci(Vg - Vlh)'
'
.
L
where m = 0.5 is the low doping factor, and Vdsat= Vg - Vthmay be assumed

Transconductance. The transconductance g,, which is also called the mu-

tual conductance, in the linear region can be found from Eq. (6-4-3) as
Microwave Field-Effect Transistors Chap. 6

In the saturation region, the transconductance becomes

2mZ
grn sat = -pn Ci(VR- Vth)
L
The channel conductance gd is given by

All equations derived here so far are based on the idealized n-channel MOSFET. For
an idealized p-channel MOSFET, all voltages V,, Vd, and Vrh are negative, and the
drain current Id flows from the source to the drain. For a real n-channel MOSFET
(say, Al-Oxide-Si structure), the saturation drain current is

and the transconductance becomes

where v, = L / r is the carrier drift velocity

The threshold voltage Vth is given by

where 4, =4, - 4, is the work function difference (in eV) between the metal
work function 4, and the semiconductor work function 4,
Qf = fixed oxide charges

Example 6-4-1: Threshold Voltage of an Ideal MOSFET

A certain p -channel MOSFET has the following parameters:

Doping concentration: N, =3 x 10" cm-'

Relative dielectric constant: 6, = 11.8
Relative dielectric constant
of SiOz : E,, = 4
Insulator depth: d = 0.01 p m
Operating temperature: T = 300°K

+,
a. Calculate the surface potential (inv) for strong inversion.
b. Compute the insulator capacitance.
c. Determine the threshold voltage.
Sec. 6.4 Metal-Oxide-Semiconductor Field-Effect Transistors (MOSFETs) 243

Solution

a. From Eq. (6-4-4),

the surface potential for strong inversion is

= 0.874 volt
b. From Eq. (6-4-I),
the insulator capacitance is

c. From Eq. (6-4-3),

the threshold voltage is

x 1.6 x 10-l9 x 3 x 1023 x 0 . 4 3 7 ) ~ ' ~

= 0.874 + 0.56 x lo3 X 14.80 x
= 1.70 volts

6-4-5 Maximum Operating Frequency

The maximum operating frequency of a MOSFET is determined by its circuit

parameters. A common-source equivalent circuit of a MOSFET is shown in Fig.
6-4-5.
where gin = input conductance due to the leakage current. Since the leakage
current is very small, say lo-'' A/cm2, gin is negligible
C i n= ZLCi is the input capacitance
Cfb = feedback capacitance
gout = gd is output conductance
Gout = C i C s / ( C i+ C,) is the sum of the two p-n junction capacitances in
series with the semiconductor capacitance per unit area

r 1:
C~
Gate Drain
P I)
P
Ck c'in <, Vd

X E,
Figure 6-4-5 Equivalent circuit of a common-source MOSFET.
244 Microwave Field-Effect Transistors Chap. 6

The maximum operating frequency of a MOSFET in the linear region can be ex-
pressed as

In the saturation region, V, 9 Vth, the transconductance is reduced to

z
grn sat = - pnC; VR = ZCius (6-4- 12)
L

6-4-6 Electronic Applications

The MOSFETs are often used as power amplifiers because they offer two advantages
over MESFETs and JFETs.

1. In the active region of an enhancement-mode MOSFET, the input capacitance

and the transconductance are almost independent of gate voltage, and the out-
put capacitance is independent of the drain voltage. This leads to very linear
(Class A) power amplification.
2. The active gate-voltage range can be larger because n-channel depletion-type
MOSFETs can be operated from the depletion-mode region (-V,) to the
enhancement-mode region (+ V,).

Example 6-4-2: Characteristics of a MOSFET

A certain n-channel MOSFET has the following parameters:

Channel length:
Channel depth:
Insulator thickness:
Gate voltage:
Doping factor:
Threshold voltage:
Electron mobility:
Electron velocity:
Relative dielectric
constant of SiOz: E,, = 3.9

a. Compute the insulator capacitance in FlmZ.

b. Calculate the saturation drain current in mA.
c. Determine the transconductance in the saturation region in millimhos.
d. Estimate the maximum operating frequency in the saturation region in GHz
Sec. 6.5 MOS Transistors and Memory Devices

Solution

a. The capacitance of the insulator S i 0 2 is

b. From Eq. (6-4-8), the saturation drain current is

= ZCi(Vx - Vt*)us
= 12 x 10-6 x 6.91 x x (5 - 0.1) x 1.7 x 105
= 6.91 mA
c. From Eq. (6-4-9), the transconductance in the saturation region is
g m sat = ZCtus
= 12 x 10-6 x 6.91 x 1 0 - ~ x 1.7 x 105
= 1.41 m U
d. From Eq. (6-4-13), the maximum operating frequency is

6-5 MOS TRANSISTORS AND MEMORY DEVICES

As discussed in Section 6-2, the source, channel, and the drain of the MOS transis-
tor 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 junc-
tion transistors. The MOSFET can be subdivided into two groups:
1. The n-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 tran-
sistor. 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.
6-5-1 NMOS Devices

The metal-oxide-semiconductor field-effect transistor (MOSFET) is the dominant

device used in very large-scale integration (VLSI) circuits. In the 1960s, the p-chan-
nel MOS (PMOS) was originally used in integrated circuits. n-channel MOS (NMOS)
246 Microwave Field-Effect Transistors Chap. 6

devices, however, have dominated the IC market since the 1970s because their elec-
tron mobility is higher than that of holes.

NMOS structure. Figure 6-5-1 shows a three-dimensional view of an

NMOS logic circuit.

- -- EMD, EMD,

NC -
DMD

- - -P-substrate
Figure 6-5-1 Three-dimensional view of an NMOS logic circuit. (After Parrillo [13];
reprinted by permission of the Bell Laboratory.)

From Fig. 6-5-1, it can be seen that two enhancement-mode (normally OFF)
devices (EMDA and EMDB) 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 MOS-
FET the extremely high input resistance (-1014 ohms) is made possible by the insu-
lator.
There are two basic structures for MOSFETs: the depletion type and the in-
duced 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 re-
gions. When the device has p-type source and drain regions, it is called the p-
channel MOSFET or PMOS.
248 Microwave Field-Effect Transistors Chap. 6

The structures of enhancement and depletion modes refer to the relative in-
crease 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 MOSFET
(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 den-
sity 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 (EMD) are in series with a depletion-mode
device (DMD), and the three transistors are connected between the positive power
supply VDDand ground Vss reference. The DMD is normally on at VGS= 0 and acts
as a current source for the two EMDs. Gates A and B of the two EMDs serve as in-
puts to the logic circuit, and the DMD's gate-source connection is the output elec-
trode of the logic circuit. The output voltage of the two-input NAND 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 I state (the output voltage is close to VDD).But when the input signal
is at I 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.

6-5-2 CMOS Devices

The complementary MOS (CMOS) is made of both NMOS and PMOS devices, and
its power consumption is quite low. In some CMOS designs, the NMOS circuit is in-
corporated in domino-CMOS to take advantage of the NMOS's high speed and the
CMOS's low power.

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 MOS Transistors and Memory Devices

Inputs Output
(a)

M V VDD
Inputs Output
(b)

SiN

eta1
OX

Figure 6-5-4 Two-input NAND logic

gate. (After Parrillo [13]; reprinted by
permission of the Bell Laboratory.)

Figure 6-5-5 NOR logic gate

250 Microwave Field-Effect Transistors Chap. 6

11-Channel 11-Channel
& &
Oxide Polys~licon

/)-Cla~s Nitride Polvsllicon A a

Polysilicon $01SiO,
11 or 11 wbjtratc
Figure 6-5-6 CMOS structures. (After
L . C . Parrillo [13]; reprinted by permis-
( c ) Iwin t u b sion of the Bell ~ a b o i a t o r ~ . )

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 com-
monly 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 transis-
tors are V,, and Vfp,respectively, (V, < 0). Figure 6-5-7(c) shows the dependence
of the output voltage V, on the input voltage K of the CMOS inverter. For 6 = 0,
the n-channel transistor is OFF (V, < V,,), while the p-channel transistor is heavily
turned ON (that is, the gate-to-source voltage of the p channel is -VDD, which is
much more negative than V,,,). Therefore, the output voltage is V, = VDD. As the in-
put voltage V, is increased above zero, the n-channel transistor eventually is turned
Sec. 6.5 MOS Transistors and Memory Devices

( b ) Device cross-section

(c) Vu vb Vl ( d ) Current curve

Figure 6-5-7 CMOS operation. (After Parrillo (131; reprinted by permission of

the Bell Laboratory.)

ON, while the p-channel transistor finally turns OFF. When K > (VDD - ( V, I), the
output voltage is V, = Vss.
The key feature of a CMOS gate is that in either logic state (V, = VDDor Vss)
one of the two transistors is OFF and the current conducted between VDD and Vss is
negligible. Figure 6-5-7(d) shows the current I D D as a function of h. 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 transis-
tor for every n-channel transistor) or dynamic CMOS (unequal number of n- and p -
channel transistors).
Microwave Field-Effect Transistors Chap. 6

6-5-3 Memory Devices

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].

B B R 11
(a) ROM (b) PROM (c) tPROM ((11 I I P K O V

B B B B B
(e) S R A M ( f ) BSRAM (g) DRAM

Figure 6-5-8 Circuit diagrams of memory types. (After S. Asai [14]; reprinted by permis-
sion of lEEE, Inc.)

ROM. The read-only memory (ROM) is also called mask ROM. During fab-
rication 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 MOS Transistors and Memory Devices 253

PROM. The programmable read-only memory (PROM) is one of the ROMs

that are field-programmable but lack "erase" capability. PROM uses cells with a fuse
that can be blown open electrically, or a p-n diode that can be short-circuited by an
avalanching pulse. A bipolar PROM has 64-Kbit p-n diode cells, and MOS PROMS
have higher density but slower speed.

EPROM. EPROM stands for the erasable programmable read-only memory.

In the EPROM cell, the presence or absence of charge in the floating gate of a
double-poly-gate MOSFET determines the logic state. Programming is done by in-
jecting energetic carriers generated by drain p-n junction avalanche breakdown,
thereby increasing the threshold voltage V,h of the memory transistor. The EPROM
is, therefore, also called the FAMOS (floating-gate avalanche injection MOS)
device. When ultraviolet light is shone on the device, the charge in the floating gate
is released, thereby erasing the memory. EPROMs are therefore packaged with a
glass window to permit the erasing operation to occur. The original p-channel
FAMOS cell, which has two transistors in series-one for storage and the other for
access, has been replaced by n-channel, single-transistor cells. The original
programming voltage of 30 V has been reduced to around 12 V. The most advanced
1-Mbit EPROMs today have a cell size of 19-29 p m 2 and an access time of
80-140 ns.

EEPROM. EEPROM designates the electrically erasable programmable

read-only memory, and it is the most sophisticated in its principle of operation. For a
ROM to be electrically erasable, it must have the means to inject and extract charge
carriers into and from a floating gate. The first proposal for an EEPROM used the
electron-trapping states at the nitride-oxide interface in a metal-nitride-oxide-silicon
(MNOS) structure with a very thin (about 2 nm) oxide. The recent dominant tech-
nology of EEPROM uses a floating gate separated from the silicon by an oxide
(about 150W thick). Programming and erasing in either type of EEPROM is achieved
by forcing the channel current to flow between the gate and substrate with the con-
trol biased and negative, respectively.
Both EPROM and EEPROM store charge on a conductive region in the middle
of a MOS-gate oxide, and are, therefore, critically dependent on MOS structure, es-
pecially high-field carrier transport in both silicon and oxide. Because carriers in the
floating island stay there even after the power supply is turned off, EPROM and
EEPROM are also nonvolatile memories.

SRAM. The random-access memory (RAM) is one of the largest memories.

In a RAM, memory cells are organized in a matrix form, and they can be accessed
in random order to read (retrieve) or write (store) data. A static random-access
memory (SRAM) can retain stored data indefinitely, and it can be implemented as a
flip-flop circuit to store one bit of information. Since SRAMs use flip-flop ciruits in
the memory cells, they are the most basic of all semiconductor memories. Active
devices for access and drive are either MOS transistors or bipolar transistors.
254 Microwave Field-Effect Transistors Chap. 6

BSRAM. The bipolar SRAM (BSRAM) is the fastest of all the semiconduc-
tor memory types, but MOS SRAM is the fastest among MOS memories. Figure
6-5-9 shows the equivalent-circuit diagrams for bipolar SRAMs.

Lateral Vertical
/I~/I t1/1?1

P-Substrate P-Su bstrate

( a ) Cros\-couplctl /I-,I-/I load cell ( b ) t . q u ~ w I c n tC I ~ C L I for

I ~ b~pol;~
SRAM
r

Figure 6-5-9 Circuit diagram for BSRAM.

DRAM. The first dynamic random-access memory (DRAM) used one-half of

the static memory cell, or three transistors-one for the driver, another for the load,
and the third for the access. This device has evolved to today's one-transistor, one-
capacitor DRAM cell, in which the single transistor is used as the access to the ca-
pacitive reservoir. Since the stored charge is gradually lost because of the space-
charge generation-recombination process, cells have to be read and refleshed at
predetermined intervals on the order of milliseconds (hence, the name dynamic).
The DRAM readout signal is small, on the order of 150-200 mV, and is subject to
various kinds of noise sources.

6-6 CHARGE-COUPLED DEVICES (CCDS)

The charge-coupled device (CCD) is a metal-oxide-semiconductor (MOS) diode

structure that was proposed in 1969 by Boyle and Smith [15, 161. The CCD can
move the charges in the MOS diode along a predetermined path under the control of
Sec. 6.6 Charge-Coupled Devices (CCDs) 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.

6-6-1 Operational Mechanism

A charge-coupled device (CCD) is an array of many MOS or MIS diodes. In opera-

tion, the information (or signal) is stored in the form of electrical charge packets in
the potential wells created in a MOS diode. Under the control of externally applied
voltage (i.e., gate voltage), the potential wells and the charge packets can be shifted
from one well to an adjacent one rapidly through the entire CCD structure.
Three separate mechanisms allow the charge packets to move from one well to
another: self-induced drift, thermal diffusion, and fringing field drift. Thermal diffu-
sion results in an exponential decay of the remaining charge under the transferring
electrode. The fringing field is the electric field in the direction of charge flow and it
can help speed the charge-transfer process considerably. Self-induced drift (or a
charge-repulsion effect) is only important at relatively large signal-charge densities.
It is the dominant mechanism in the transfer of the first 99% of the charge signal.

Energy band of MIS diode. A single MIS structure on an n-type semicon-

ductor (or p-type semiconductor) is the basic element of the CCD. Figure 6-6-1
shows the energy band diagrams for a MIS structure [15].

n-semiconductor

E'F

Distance +
(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 Microwave Field-Effect Transistors Chap. 6

The voltage applied to the metal electrode is negative with respect to the semi-
conductor 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-l(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 typi-

cal three-phase structure as shown in Fig. 6-6-2 [16].

+-Silicon dioxide

-17-type silicon

SiliconI dioxide

-n-type silicon

Figure 6-6-2 Cutaway of CCD. (From W . S . Boyle and G . E . Smith [16];

reprinted b y permission of the IEEE, Inc.)

The CCD consists of a closely spaced array of MIS diodes on an n-type semi-
conductor 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), V, = V3 and V2is more negative. In effect, a potential well
with stored holes is created at gate electrode 2. The stored charge is temporary be-
cause 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 V3 is pulsed to be more negative than
the other two voltages V, and V2, the charge begins to transfer to the potential well at
gate electrode 3 as shown in Fig. 6-6-2(b).
Sec. 6.6 Charge-Coupled Devices (CCDs) 257

Store and transfer of charge packets. A linear array of MIS diodes on

an n-type semiconductor is shown in Fig. 6-6-3 [15].

"3

v3
t=r3 V, I I I
I 1 I I 1
V~ . ... .....I.
... ..... . ..I I. !
.. .. ..... .... ...... .. ... . .... .... ...I
... ........
I
'
..... ... . ......................I.......!...
............ ..... .. ........I
.. ..... .. ......... .. .. ..... .. .. . .

Figure 6-6-3 Store and transfer of charge packets for a three-phase CCD. (From
W. S. Boyle and G . E . Smith [15]; reprinted by permission of The Bell System,
AT&T.)

For a three-phase CCD, every third gate electrode is connected to a common

line (see Fig. 6-6-3). At t = t l , a more negative voltage V, is applied to gate elec-
trodes 1, 4, 7, and so on, and less negative voltages V2 and V3 (V2 = V3) are applied
to the other gate electrodes. It is assumed that the semiconductor substrate is
grounded and that the magnitude of Vl is larger than the threshold voltage Vi, for the
production of inversion under steady-state conditions. As a result, positive charges
are stored in the potential wells under electrodes 1, 4, 7, . . . as shown in Fig.
6-6-3(a).
At t = tZ, when voltage V2 at gate electrodes 2, 5, 8, . . . is pulsed to be more
negative than Vl and V3, the charge packets will be transferred from the potential
wells at gate electrodes 1, 4, 7, . . . to the potential minimum under gate electrodes
2, 5, 8, . . . as shown in Fig. 6-6-3(b).
At t = t3, when voltages Vl = V3 and voltage V2 remains more negative, the
charge packets have been transferred one spatial position and the sequence is ready
to continue as shown in Fig. 6-6-3(c).
258 Microwave Field-Effect Transistors Chap. 6

6-6-2 Surface-Channel Charge-Coupled Devices

(SCCDs)

A surface-channel charge-coupled device (SCCD) is basically an MOS-diode struc-

ture as shown in Fig. 6-6-4(a) [28]. Its energy band diagrams are illustrated in parts
(b) and (c).

hell

it. l Q,,, = 0

h e l l parrldli)
t ~ l l e du ~ t l t
i ~ g n d l~ i l d r g c

Figure 6-6-4 Physical structure and energy-band diagrams of a BCCD. (From D. F. Barle
1281; reprinted with permission of the IEEE, lnc.)

There are two cases for the creation of a charge packet under deep depletion.

1. Zero Signal-Charge (QSi, = 0). 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 V, and surface potential 1,6~are related by
Sec. 6.6 Charge-Coupled Devices (CCDs)

where Vfi = flatband voltage

V, = voltage across the insulator

w= (z)
2Ess' 'I2 in the depletion width

Substitution of the depletion width into Eq. (6-6-1) yields

2. Stored Signal-Charge (Q,,, > 0). 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
1
V, - Vfi = -
Ci
Qsig
+-
Ci
+
( ~ E ~ ~ N ~ C C , rC,S
~)'"

The maximum charge density (electron or hole) that can be stored on an MOS
capacitor is approximately equal to

Nmax= -
ci v, for V, +1 (6-6-4)
4

6-6-3 Dynamic Characteristics

The dynamic characteristics can be described in terms of charge-transfer efficiency

q , frequency response, and power dissipation.

Charge-transfer efficiency q. The charge-transfer efficiency is defined as

the fraction of charge transferred from one well to the next in a CCD. The fraction
left behind is the transfer loss and it is denoted by E . Therefore the charge-transfer
efficiency is given by

If a single charge pulse with an initial amplitude Po transfers down a CCD register,
after n transfers, the amplitude P, becomes

where n equals the number of transfers or phases.

If many transfers are required, the transfer loss E must be very small. For ex-
ample, if a transfer efficiency of 99.99% is required for a three-phase, 330-stage
shift register, the transfer loss must be less than 0.01%. The maximum achievable
transfer efficiency depends on two factors: how fast the free charge can be trans-
ferred between adjacent gates and how much of the charge gets trapped at each gate
location by stationary states.
260 Microwave Field-Effect Transistors Chap. 6

Frequency response. There are, in fact, upper and lower frequency limi-
tations for CCDs. The potential well will not remain indefinitely, and thermally gen-
erated 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. The power dissipation per bit is given by

P = nf VQmax (6-6-7)

Example 6-6-1: Power Dissipation of a Three-Phase CCD

A three-phase CCD is operating under the following conditions:

Applied voltage: V = 10 volts

Number of phases: n = 3
Maximum stored charges: Q,,, = 0.04 pC
Clock frequency: f = 10MHz

Determine the power dissipation per bit

Solution From Eq. (6-6-7) the power dissipation per bit is

Charge-coupled devices (CCDs) have many microwave applications in electronic

components and systems, such as in infrared systems and signal processing.

Infrared detection and imaging. Because varying amounts of charge cor-

responding 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 im-
proved 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 inten-
sity of a target; and these packets are shifted to a detector point for detection, read-
out, 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 digi-
tal logic. Thus CCDs are being used widely in special applications for the very large
scale integration (VLSI) circuits.

Example 6-6-2: Design of an N-Type Three-Phase Surface-Channel CCD

The n-type three-phase surface-channel CCD has the following specifications:

Electron density: N,, = 2 x 1012cm-2

Insulator relative dielectric constant: ei,= 3.9
Insulator thickness: d = 0.15 p m
Insulator cross section: A = 0.5 x cm2
Power dissipation allowable per bit: P = 0.67 mW

a. Compute the insulator capacitance in farads per square centimeter.

b. Determine the maximum stored charges per well in coulombs.
c. Find the required applied gate voltage in volts.
d. Choose the clock frequency in megahertz.

Solution

a. The insulator capacitance is

b. The maximum stored charges per well is

Q,,, = NmB,qA = 2 X 1012 x 1.6 x lo-'' x 0.5 x
= 16pC
c. From Eq. (6-6-4) the required applied gate voltage is

d. From Eq. (6-6-7) the clock frequency is

P - 0.67 x lo-'
f =-- = 1 MHz
nVQmax 3 X 14 x 16 x 10-12

REFERENCES

[I] SHOCKLEY, W., A unipolar "field-effect" transistor. Proc. IRE, 40, No. 11, 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 Microwave Field-Effect Transistors 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 frequen-
cies. 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-FET's in the
hot electron range. Solid-State Electronics, Vol. 13, 1415- 1426. Pergamon Press, Lon-
don, 1970.
[8] DINGLE, R., et al., Electron mobilities in modulation-doped semiconductor heterojunc-
tion superlattices. Appl. Phys. Letters, Vol. 33, 665-667 (1978).
19) ABE,M., et al., Recent advance in ultra-light speed HEMT technology. IEEE J. Quan-
tum Electronics. QE-22, No. 9, 1870- 1879, September 1986.
[lo] TOGASHI, K., et al., Reliability of low-noise microwave HEMTs made by MOCVD. Mi-
crowave J., 123-132, April 1987.
[I 11 PAVLIDIS, D., and M. WEISS,The influence of device physical parameters on HEMT
large-signal characteristics. IEEE Trans. on Microwave Theory and Techniques, MTT-
36, No. 2, 239-249, February 1988.
[12] SZE,S. M., Physics of Semiconductor Devices, 2nd ed., p. 440, John Wiley & Sons,
New York, 1981.
[I 31 PARRILLO, L. C., VLSI Process Integration in VLSI Technology. Chapter 11, ed. S. M.
Sze. McGraw-Hill Book Company, New York, 1983.
[14] ASAI,S., Semiconductor memory trend. Proc. IEEE, 74, No. 12, 1623-1635, Decem-
ber 1986.
[15] BOYLE,W. S., and G. E. SMITH,Charge couple semiconductor devices. Bell Syst. Tech.
J . , 49, 587-593 (1977).
[I61 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.
[I71 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

H E S S , 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 solid-
state devices.
MTT-21, No. 1 1, November 1973.
MTT-24, No. I I , November 1976.
Chap. 6 Problems

MTT-27, No. 5, May 1979.

MTT-28, No. 12, December 1980.
MTT-30, No. 4, April 1982.
MTT-30, No. 10, October 1982.
LIAO,S. Y., Microwave Solid-State Devices. Prentice-Hall, Inc., Englewood Cliffs, N.J.,
1985.
LIAO,S. Y., Semiconductor Electronic Devices. Prentice-Hall, Inc., Englewood Cliffs, N.J.,
1990.
NAVON, D. H., Semiconductor Microdevices and Materials. Holt, Rinehart and Winston, New
York 1986.
PULFREY, DAVID L., and N. GARRY TARR.,Introduction to Microelectronic Devices. Prentice-
Hall, Inc., Englewood Cliffs, N.J., 1989.

PROBLEMS

JFETs
6-1. A silicon JFET has a gate length of 5 p m . Calculate the maximum frequency of oscil-
lation for the device.
6-2. An n-channel silicon JFET has the following parameters:

Electron density: Nd = 2 x 1017cm-3

Channel height: a = 0.4 x cm
Relative dielectric constant: 6, = 11.8

Compute the pinch-off voltage.

6-3. An n-channel silicon JFET at 300" K has the following parameters:

Electron density:
Hole density:
Dielectric constant:
Channel height:
Channel length:
Channel width:
Electron mobility:
Drain voltage:
Gate voltage:

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, = -1 V in mA
f. The cut-off frequency in GHz
264 Microwave Field-Effect Transistors Chap. 6

6-4. An n-channel silicon JFET at 300' K has the following parameters:

Electron density: Nd = 3 X 10'' cm13

Hole density: N, = 2 x 1019 cm-'
Relative dielectric constant: E , = 1 1.8
Channel height: a = 0.5 x cm
Channel length: L = 10 x 10-4cm
Channel width: Z = 20 x cm
Drain voltage: vd=7v
Gate voltage: V, = -1.5 V

Determine:
a. The electron mobility p,
b. The pinch-off voltage in V
c. The pinch-off current in mA
d. The drain current in mA
e. The saturation current in mA at V, = 0
f. The cutoff frequency in GHz
6-5. Verify Eq. (6-1-17) from Eq. (6-1-12).
6-6. Write a FORTRAN program to compute the drain current of an n-channel silicon
JFET. The device parameters are as follows:

Electron concentration: Nd = 2 x lo2' m-'

Channel height: a = 0.2 x lo-' m
Relative dielectric constant: E, = 11.8
Channel length: L = 10 X m
Channel width: Z = 60 X 10-'m
Drain voltage: Vd=Oto6V
Gate voltage: V* = O t o -3 V
Saturation drift velocity: v, = lo5 mls

The program specifications are

1. The drain voltage Vd varies from 0 to 6 volts with an increment of 1 volt per step
2. The gate voltage V, varies from 0 to -3 volts with a decrease of -0.5 volt per
step
3. The electron mobility p varies from 0.9 to 0.3 m2/V.s with a decrease of 0.1 per
step
4. Use F10.5 format for numerical outputs and Hollerith format for character out-
puts
5. Print the outputs in three columns for drain voltage Vd (volts), gate voltage V,
(volts), and drain current Id (mA)

MESFETs
6-7. A MESFET has a gate width of 5 p m . Calculate the maximum frequency of oscillation
for the device.
Chap. 6 Problems 265

6-8. A GaAs has a thickness of 0.40 p m and a doping concentration Nd of 5 X lOI7 ~ m - ~ .

The relative dielectric constant E, of GaAs is 13.10. Calculate the pinch-off voltage in
volts.
6-9. An n-channel GaAs MESFET at 300O K has the following parameters:

Donor density:
Relative dielectric constant:
Channel height:
Channel length:
Channel width:
Drain voltage:
Gate voltage:
Saturation drift velocity:

Compute:
a. The electron mobility (read from Fig. A-2 in Appendix A)
b. The pinch-off voltage in V
c. The saturation current at V, = 0
d. The drain current Id in mA
6-10. An n-channel GaAs MESFET at 300' K has the following parameters:

Donor density:
Relative dielectric constant:
Channel height:
Channel length:
Channel width:
Drain voltage:
Gate voltage:
Saturation drift velocity:

Calculate:
a. The electron mobility (read from Fig. A-2 in Appendix A)
b. The pinch-off voltage
c. The saturation current at V, = 0
d. The drain current Id in mA
A certain n-channel GaAs MESFET has the following parameters:

Electron concentration: N = 2.38 x lo2' m-3

Channel height: a = 0.2 p m
Relative dielectric constant: E, = 13.10
Channel length: L = 10pm
Channel width: Z=60pm
Electron mobility: p = 0.3 m2/V . s
Drain voltage: Vdr = 6 volts
Gate voltage: V, = -3 volts
Saturation drift velocity: v, = 10' m/s
266 Microwave Field-Effect Transistors Chap. 6

a. Calculate the pinch-off voltage.

b. Compute the velocity ratio.
c. Determine the saturation drain current at V, = 0.
d. Find the drain current I d .

HEMTs
6-12. Describe the structure of a HEMT and its fabrication processes.
6-13. Describe the operating principles of a HEMT.
6-14. A HEMT has the following parameters:

Gate width: W = 140 m

Electron velocity: v ( z ) = 3 x lo5 m/s
2-D electron-gas density: n ( z ) = 6 x lOI5 m-2

Compute the drain current of the HEMT.

6-15. A HEMT has the following parameters:

Gate width: W = 100m

Electron velocity: v ( z ) = 4 x lo5 m/s
2-D electron-gas density: n ( z ) = 4 x 10" m-2

Calculate the drain current of the HEMT.

6-16. A HEMT has the following parameters:

Threshold voltage: V,, = 0.15 V

Donor concentration: Nd = 3 x loz4m-'
Metal-semiconductor l,hrns = 0.8 V
Schottky barrier potential:
AlGaAs dielectric constant: 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: Fh= 0.12 V

Donor concentration: Nd = 5 x loz4m-3
AlGaAs dielectric constant: E, = 4.43
Metal-semiconductor
Schottky barrier potential: h, = 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 Problems

MOSFETs
6-19. A basic MOSFET is formed of Al metal, Si02 insulator, and Si semiconductor. The
insulator capacitance is 4 pF and the channel length L is 12 p m .
a. Determine the drain current I d ,, in the saturation region with = 2V and
V, = 4V for the enhancement mode.
b. Compute the transconductance g, for the same mode.
c. Calculate the drain current Id , in the saturation region with Vth = - 1.5V and
V, = OV for the depletion mode.
d. Find the transconductance g, for the depletion mode.
6-20. An Al-Oxide-Si MOSFET has an insulator capacitance of 3 pF and a channel length L
of 10 p m . The gate voltage V, is 10 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 SiOz in a Si MOSFET has a relative dielectric constant E,, of 11.8 and a
depth d of 0.08 p m . The channel length L is 15 p m and the channel depth Z is
150 p m . Calculate the insulator capacitance Ci, .
6-22. Compare the advantages and disadvantages of a GaAs MOSFET with those of a Si
MOSFET.
6-23. A certain p-channel MOSFET has the following parameters:

Doping concentration: N, = 2 X 10'' cm-3

Relative dielectric constant: E, = 11.8
Insulator relative dielectric
constant: E,, =4
Insulator thickness: d = 0.01 p m
Operating temperature: T = 320" K

a. Calculate the surface potential &(inv) for strong inversion.

b. Compute the insulator capacitance.
c. Determine the threshold voltage.
6-24. A certain n-channel MOSFET has the following parameters:

Channel length: L =5pm

Channel depth: Z = 10pm
Insulator thickness: d = 0.02 p m
Gate voltage: V,=8V
Threshold voltage: xh= 1.5V
Electron velocity: v, = 2 x 10' cm/s
Insulator relative dielectric
constant: E, = 4

a. Compute the insulator capacitance in rnF/m2

b. Calculate the saturation drain current in mA
268 Microwave Field-Effect Transistors Chap. 6

c. Determine the saturation transconductance in millimhos.

d. Estimate the maximum saturation operating frequency in GHz.

NMOS, CMOS, and Memory Devices

6-25. Describe the structures of NMOS, PMOS, and CMOS devices.
6-26. Explain the operational principles of the NMOS, PMOS, and CMOS devices.
6-27. Discuss the logic-gate operations of NMOS, PMOS, and CMOS.
6-28. Describe the major memory devices.

CCDs
6-29. A charge-coupled device has 484 (or 22 X 22 array) elements each with a transfer
inefficiency of 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 V' of 10 volts. The insulator has a
relative dielectric constant E, of 6 and a depth d of 0.1 p 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:

N, = loi4cm-' Q,, = C/cm2

dm,= -0.85 eV W = 1 Fm
E ; ~= 4 d = 0.1 p m

a. Determine the flat-band potential.

b. Calculate the voltage across the insulator.
6-32. A three-phase CCD has Q,,, = 0.06 pC operating at a clock frequency of 20 MHz
with 10 volts applied. Determine the power dissipation per bit.
6-33. A three-phase JCCD has a length L of 8 p m , an electric field Ex of 480 Vlcm, and an
electron mobility p, of 1000 cmZ/V- s. Calculate the signal carrier transit time.
6-34. An n-type three-phase SCCD has the following parameters:

Insulator relative dielectric E , ~= 4

constant:
Insulator thickness: d = 0.3 p m
Insulator cross section: A = 0.6 x cm2
Electron density: N,,, = 4 X 1012cm-*
Power dissipation allowable per bit: P = 0.8 mW

a. Compute the insulator capacitance in Flcm2.

b. Calculate the maximum stored charges per well in coulombs.
c. Find the required applied gate voltage in volts.
d. Choose the clock frequency in MHz.
Chapter 7

1 Transferred Electron
Devices (TEDs)

7.0 INTRODUCTION

The application of two-terminal semiconductor devices at microwave frequencies has

been increased usage during the past decades. The CW, average, and peak power
outputs of these devices at higher microwave frequencies are much larger than those
obtainable with the best power transistor. The common characteristic of all active
two-terminal solid-state devices is their negative resistance. The real part of their
impedance is negative over a range of frequencies. In a positive resistance the cur-
rent through the resistance and the voltage across it are in phase. The voltage drop
across a positive resistance is positive and a power of (I2R) is dissipated in the resis-
tance. In a negative resistance, however, the current and voltage are out of phase by
180". The voltage drop across a negative resistance is negative, and a power of
(-12R) is generated by the power supply associated with the negative resistance. In
other words, positive resistances absorb power (passive devices), whereas negative
resistances generate power (active devices). In this chapter the transferred electron
devices (TEDs) are analyzed.
The differences between microwave transistors and transferred electron devices
(TEDs) are fundamental. Transistors operate with either junctions or gates, but
TEDs are bulk devices having no junctions or gates. The majority of transistors are
fabricated from elemental semiconductors, such as silicon or germanium, whereas
TEDs are fabricated from compound semiconductors, such as gallium arsenide
(GaAs) , indium phosphide (InP), or cadmium telluride (CdTe) . Transistors operate
with "warm" electrons whose energy is not much greater than the thermal energy
(0.026 eV at room temperature) of electrons in the semiconductor, whereas TEDs
270 Transferred Electron Devices (TEDs) Chap. 7

operate with "hot" electrons whose energy is very much greater than the thermal en-
ergy. Because of these fundamental differences, the theory and technology of tran-
sistors cannot be applied to TEDs.

7-1 GUNN-EFFECT DIODES--GaAs DIODE

Gunn-effect diodes are named after J. B. Gunn, who in 1963 discovered a periodic
fluctuations of current passing through the n-type gallium arsenide (GaAs) specimen
when the applied voltage exceeded a certain critical value. Two years later, in 1965,
B. C. DeLoach, R. C. Johnston, and B. G. Cohen discovered the impact ionization
avalanche transit-time (IMPATT) mechanism in silicon, which employs the avalanch-
ing and transit-time properties of the diode to generate microwave frequencies. In
later years the limited space-charge-accumulation diode (LSA diode) and the indium
phosphide diode (InP diode) were also successfully developed. These are bulk
devices in the sense that microwave amplification and oscillation are derived from
the bulk negative-resistance property of uniform semiconductors rather than from
the junction negative-resistance property between two different semiconductors, as
in the tunnel diode.

-
7-1 1 Background

After inventing the transistor, Shockley suggested in 1954 that two-terminal

negative-resistance devices using semiconductors may have advantages over transis-
tors at high frequencies [I]. In 1961 Ridley and Watkins described a new method for
obtaining negative differential mobility in semiconductors [2]. The principle in-
volved 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 sub-
band when they have a high enough temperature. Ridley and Watkins also mentioned
that Ge-Si alloys and some 111-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 111-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 373°K at a field of 3200 V/cm. Hilsum's attempts to verify his theory ex-
perimentally 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 ef-
fect 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 immedi-
ately 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 Gunn-Effect Diodes-GaAs Diode 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 pre-
dictions and experimental discoveries completed the theory of transferred electron
devices (TEDs) .

7-1-2 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 91. 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 spec-
imens, this fluctuation took the form of a periodic oscillation superimposed upon the
pulse current. . . . The frequency of oscillation was determined mainly by the speci-
men, and not by the external circuit. . . . The period of oscillation was usually in-
versely proportional to the specimen length and closely equal to the transit time of elec-
trons between the electrodes, calculated from their estimated velocity of slightly over
lo7 cmls. . . . The peak pulse microwave power delivered by the GaAs specimens to a
matched load was measured. Value as high as 0.5 W at 1 Gc/s, and 0.15 W at 3 Gc/s,
were found, corresponding to 1-2% of the pulse input power.*

High-field domain

+ Anode

Figure 7-1-1 Schematic diagram for n -

type GaAs diode.

*After J. B. Gunn [ 7 ] ; reproduced by permission of IBM, Inc.

272 Transferred Electron Devices (TEDs) 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 Vlcm 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.

Field ( k V / c m )

Figure 7-1-2 Drift velocity o f electrons in n-type GaAs versus electric field.
(After J . 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 cm in length. The oscillation frequency was
4.5 GHz. The lower trace had 2 nslcm in the horizontal axis and 0.23 Alcm 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.

Figure 7-1-3 Current waveform o f n -

type G a A s reported b y Gunn. (After
J . B. Gunn [8]; reprinted by permission
of IBM. Inc.)
Sec. 7.2 Ridley-Watkins-Hilsum (RWH) Theory 273

Gunn also discovered that the threshold electric field Ethvaried 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 p m
and cross-sectional area 3.5 X cm2with a low-field resistance of 16 Currenta.
instabilities occurred at specimen voltages above 59 V, which means that the
threshold field is

7-2 RIDLEY-WATKINS-alLSUM (RWH) THEORY

Many explanations have been offered for the Gunn effect. In 1964 Kroemer [6] sug-
gested that Gunn's observations were in complete agreement with the Ridley-
Watkins-Hilsum (RWH) theory.
7-2-1 Differential Negative Resistance

The fundamental concept of the Ridley-Watkins-Hilsum (RWH) theory is the differ-

ential negative resistance developed in a bulk solid-state 111-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 current-
controlled modes as shown in Fig. 7-2-l(a) and (b), respectively [ 5 ] .

0 i:' 0 I-.
(a) Voltage-controlled mode (b) Current-controlled mode

Figure 7-2-1 Diagram of negative resistance. (From B. K. Ridley [5]; reprinted

b y permission of the Institute of Physics.)

In the voltage-controlled mode the current density can be multivalued,

whereas in the current-controlled mode the voltage can be multivalued. The major
effect of the appearance of a differential negative-resistance region in the current-
density-field curve is to render the sample electrically unstable. As a result, the ini-
tially homogeneous sample becomes electrically heterogeneous in an attempt to
reach stability. In the voltage-controlled negative-resistance mode high-field do-
mains are formed, separating two low-field regions. The interfaces separating low-
and high-field domains lie along equipotentials; thus they are in planes perpendicular
to the current direction as shown in Fig. 7-2-2(a). In the current-controlled
negative-resistance mode splitting the sample results in high-current filaments run-
ning along the field direction as shown in Fig. 7-2-2(b).
Transferred Electron Devices (TEDs) Chap. 7

High field High current

\ 1

Low field Low current

(a) High-field domain (b) High-current filament

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.)

Expressed mathematically, the negative resistance of the sample at a particular

region is
dl---d-J
- - negative resistance
dV dE
If an electric field Eo (or voltage Vo)is applied to the sample, for example, the
current density Jois generated. As the applied field (or voltage) is increased to E2 (or
V2), the current density is decreased to J2.When the field (or voltage) is decreased
to El (or V , ) ,the current density is increased to JI. These phenomena of the voltage-
controlled negative resistance are shown in Fig. 7-2-3(a). Similarly, for the current-
controlled mode, the negative-resistance profile is as shown in Fig. 7-2-3(b).

0 1 E0 E2 E
(a) Voltage-con trolled mode (b) Current-controlled modc

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.)

7-2-2 Two-Valley Model Theory

A few years before the Gunn effect was discovered, Kroemer proposed a negative-
mass microwave amplifier in 1958 [lo] and 1959 [l 11. According to the energy band
theory of the n-type GaAs, a high-mobility lower valley is separated by an energy of
0.36 eV from a low-mobility upper valley as shown in Fig. 7-2-4. Table 7-2-1 lists
Sec. 7.2 Ridley-Watkins-Hilsum (RWH) Theory

ill(,,, = I.?
p,, ' 1 X O c1llZ/l,..\
L owel- v a l l e y
iil,, = 0.0(18
p,

I:;
=

uI .
8000 c11i2/1

=
t
1.43 e V
' - \

'+ t
A/: = 0.36 eV
('ontiuct~on
band

Forhiddcn
band
I

h'
Valence Figure 7-2-4 Two-valley model of
ooo band electron energy versus wave number for
J n-type GaAs.

TABLE 7-24 DATA FOR TWO VALLEYS IN GaAs

Effective Mass Mobility Separation

Valley Me P AE
Lower MCp= 0.068 pt = 8000 cm2/V-sec AE = 0.36 eV
Upper Me, = 1.2 p, = 180 cm2/V-sec AE = 0.36 eV

the data for the two valleys in the n-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 val-
ley and lower than that of the upper valley (Ee < E < E,), electrons will begin to
transfer to the upper valley as shown in Fig. 7-2-5(b). And when the applied electric

TABLE 7-2-2 DATA FOR TWO-VALLEY SEMICONDUCTORS

Gap energy Separation energy Threshold

(at 300°K) between two valleys field Peak velocity
Semiconductor E,(eV) A E (eV) E~(KV/cm) o,(i07 cm/s)

Ge
GaAs
InP

CdTe
InAs
InSb
- --

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 Transferred Electron Devices (TEDs) Chap. 7

(a) E < L-@ (b) C; < I : < E,, (c) E, <t

Figure 7-2-5 Transfer of electron densities.

field is higher than that of the upper valley (E, < E), all electrons will transfer to the
upper valley as shown in Fig. 7-2-5(c).
If electron densities in the lower and upper valleys are ne and n u , the conduc-
tivity of the n-type GaAs is

where e = the electron charge

p =the electron mobility
n =ne +
nu is the electron density

Example 7-2-1: Conductivity of an n-Type GaAs Gunn Diode

Electron density: n= 10" cm-'

Electron density at lower valley: ne = 101° cm-3
Electron density at upper valley: nu = lo8 cm-'
Temperature: T= 300°K

Determine the conductivity of the diode.

Solution From Eq. (7-2-2) the conductivity is
a = e(pene + pun,)
= 1.6 X 10-'9(80@) X X 1016 + 180 X X lOI4)
= 1.6 x 10-l9 x 8000 x X 1016 for ne S nu
= 1.28 mmhos

When a sufficiently high field E is applied to the specimen, electrons are accel-
erated and their effective temperature rises above the lattice temperature. Further-
more, the lattice temperature also increases. Thus electron density n and mobility p
are both functions of electric field E. Differentiation of Eq. (7-2-2) with respect to E
yields
Sec. 7.2 Ridley-Watkins-Hilsum (RWH) Theory 277

If the total electron density is given by n = nt +

nu and it is assumed that pt and p ,
are proportional to EP,where p is a constant, then

dnr -
- - dn,
--
dE dE

and

Substitution of Eqs. (7-2-4) to (7-2-6) into Eq. (7-2-3) results in

Then differentiation of Ohm's law J = UE with respect to E yields

Equation (7-2-8) can be rewritten

Clearly, for negative resistance, the current density J must decrease with increasing
field E or the ratio of dJldE 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

Substitution of Eqs. (7-2-2) and (7-2-7) with f = n,/nt results in [2]

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 de-
pend on the lattice and carrier temperature. The first bracket in Eq. (7-2-1 1) must be
positive in order to satisfy unequality. This means that pt > p , . Electrons must be-
gin 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, pt & p , . The
factor dnt/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 Transferred Electron Devices (TEDs) Chap. 7

On the basis of the Ridley-Watkins-Hilsum theory as described earlier, the

band structure of a semiconductor must satisfy three criteria in order to exhibit nega-
tive resistance [ 121.

1. The separation energy between the bottom of the lower valley and the bottom
of the upper valley must be several times larger than the thermal energy (about
0.026 eV) at room temperature. This means that AE > kT or AE > 0.026 eV.
2. The separation energy between the valleys must be smaller than the gap energy
between the conduction and valence bands. This means that AE < E, . Other-
wise the semiconductor will break down and become highly conductive before
the electrons begin to transfer to the upper valleys because hole-electron pair
formation is created.
3. Electrons in the lower valley must have high mobility, small effective mass,
and a low density of state, whereas those in the upper valley must have low
mobility, large effective mass, and a high density of state. In other words,
electron velocities (dE/dk) must be much larger in the lower valleys than in the
upper valleys.

The two most useful semiconductors-silicon and germanium-do not meet

all these criteria. Some compound semiconductors, such as gallium arsenide
(GaAs), indium phosphide (InP), and cadmium telluride (CdTe) do satisfy these
criteria. Others such as indium arsenide (InAs), gallium phosphide (Gap), and in-
dium antimonide (InSb) do not. Figure 7-2-6 shows a possible current versus field
characteristic of a two-valley semiconductor.

Negative resistance region

resistance

E" E,
- E
Figure 7-2-6 Current versus field char-
Electric field E acteristic of a two-valley semiconductor.

A mathematical analysis of differential negative resistance requires a detailed

analysis of high-field carrier transports 113-141. From electric field theory the mag-
nitude of the current density in a semiconductor is given by
J = qnv (7-2-12)
where q =electric charge
n =electron density, and
v =average electron velocity.
Sec. 7.2 Ridley-Watkins-Hilsum (RWH) Theory

Figure 7-2-7 Electron drift velocity versus electric field

Differentiation of Eq. (7-2- 12) with respect to electric field E yields

The condition for negative differential conductance may then be written

where p, denotes the negative mobility, which is shown in Fig. 7-2-7.

The direct measurement of the dependence of drift velocity on the electric field
and direct evidence for the existence of the negative differential mobility were made
by Ruch and Kino [15]. Experimental results, along with the theoretical results of
analysis by Butcher and Fawcett [13], are shown in Fig. 7-2-8.

Electric field (kV/crn)

Figure 7-2-8 Theoretical and experimental velocity-field characteristics of a GaAs

diode.
280 Transferred Electron Devices (TEDs) Chap. 7

Example 7-2-2: Characteristics of a GaAs Gunn Diode

A typical n-type GaAs Gunn diode has the following parameters:

Threshold field E,, = 2800 Vlcm

Applied field E = 3200 Vlcm
Device length L = lOpm
Doping concentration no = 2 x lOI4 cm-'
Operating frequency f = 10GHz

a. Compute the electron drift velocity.

b. Calculate the current density.
c. Estimate the negative electron mobility.
Solution

a. The electron drift velocity is

vd = 10 X lo9 X 10 X = lo5 m/sec = lo7 cmlsec
b. From Eq. (7-2- 12) the current density is
J = qnv = 1.6 x 10-l9 x 2 X loz0 X 10 X lo9 X lo-'
= 3.2 X lo6 A/mZ
= 320 A/cm2
c. The negative electron mobility is

7-2-3 High-Field Domain

In the last section we described how differential resistance can occur when an elec-
tric field of a certain range is applied to a multivalley semiconductor compound,
such as the n-type GaAs. In this section we demonstrate how a decrease in drift ve-
locity 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 ve-
locity 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

where J = conduction current density

a = conductivity
Ex = electric field in the x direction
Sec. 7.2 Ridley-Watkins-Hilsum (RWH) Theory

L = length of the diode

V = applied voltage
p = 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

.=-I L

E, d~ (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 high-
field 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 de-
creases 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 accumu-
lated 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 im-
plies 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. Further-
more, 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 1
and 2 in Fig. 7-2-9(a) where the currents in the two field regions are equal. As a re-
sult 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 space-
charge instability. When positive and negative charges are separated by a small dis-
tance, then a dipole domain is formed as shown in Fig. 7-2-10. The electric field in-
Transferred Electron Devices (TEDs) Chap. 7

Accumulation layer

Figure 7-2-9 Formation of an electron

accumulation layer in GaAs. (After H.
Kroemer 1161; reprinted by permission of
(e) IEEE, Inc.)

side the dipole domain would be greater than the fields on either side of the dipole in
Fig. 7-2-10(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 spec-
imen 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 Ridley-Watkins-Hilsum (RWH) Theory

I -
Dipole

*
'\
I '
\
0 Figure 7-2-10 Formation of an elec-
tron dipole layer in GaAs. (After H.
o -ldl- L Kroemer [16]; reprinted by permission of
(d) IEEE, Inc.)

In general, the high-field domain has the following properties [12]:

1. A domain will start to form whenever the electric field in a region of the sam-
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
284 Transferred Electron Devices (TEDs) Chap. 7

4. The formation of a new domain can be prevented by decreasing the voltage

slightly below threshold (in a nonresonant circuit).
5. A domain will modulate the current through a device as the domain passes
through regions of different doping and cross-sectional area, or the domain
may disappear. The effective doping may be varied in regions along the drift
path by additional contacts.
6. The domain's length is generally inversely proportional to the doping; thus
devices with the same product of doping multiplied by length will behave simi-
larly in terms of frequency multiplied by length, voltagellength, and
efficiency.
7. As a domain passes a point in the device, the domain can be detected by a ca-
pacitive contact, since the voltage changes suddenly as the domain passes. The
presence of a domain anywhere in a device can be detected by a decreased
current or by a change in differential impedance.

It should be noted that properties 3 and 6 are valid only when the length of the
domain is much longer than the thermal diffusion length for carriers, which for
GaAs is about 1 p m for a doping of l o f 6per cubic centimeter and about 10 p m for
a doping of l o f 4per cubic centimeter.

7-3 MODES OF OPERATlON

Since Gunn first announced his observation of microwave oscillation in the n-type
GaAs and n-type InP diodes in 1963, various modes of operation have been devel-
oped, depending on the material parameters and operating conditions. As noted, the
formation of a strong space-charge instability depends on the conditions that enough
charge is available in the crystal and that the specimen is long enough so that the
necessary amount of space charge can be built up within the transit time of the elec-
trons. This requirement sets up a criterion for the various modes of operation of bulk
negative-differential-resistance devices. Copeland proposed four basic modes of op-
eration of uniformly doped bulk diodes with low-resistance contacts [17] as shown
in Fig. 7-3-1.

1. Gunn oscillation mode: This mode is defined in the region where the product of
frequency multiplied by length is about lo7 cmls and the product of doping
multiplied by length is greater than 10f2/cm2.In this region the device is un-
stable because of the cyclic formation of either the accumulation layer or the
high-field domain. In a circuit with relatively low impedance the device oper-
ates 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 de-
termined by the resonant frequency of the cavity and has a value of several
times the intrinsic frequency.
Sec. 7.3 Modes of Operation

no L doping x length ( c m - 2 ,

Figure 7-3-1 Modes of operation for Gunn diodes. (After J. A . Copeland 1/71;
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 lo7 cm/s and the product of doping
times length is between 10" and 10'2/cm2.
3. LSA oscillation mode: This mode is defined in the region where the product of
frequency times length is above lo7 cm/s and the quotient of doping divided by
frequency is between 2 X lo4 and 2 x lo5.
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 fre-
quency 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 be-
gins. 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, how-
ever, 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 concen-
trations of free electrons ranging from lOI4 to 10'' per cubic centimeter at room tem-
perature. Its typical dimensions are 150 X 150 p m in cross section and 30 p m
long. During the early stages of space-charge accumulation, the time rate of growth
286 Transferred Electron Devices (TEDs) Chap. 7

of the space-charge layers is given by

E E
where T~ = - = -is the magnitude of the negative dielectric relaxation time
a end p, I
E = semiconductor dielectric perrni ttivi ty
no = doping concentration
p, = negative mobility
e = electron charge
o = conductivity
Figure 7 - 3 - 2 clarifies Eq. ( 7 - 3 - 1 ) .

Figure 7-3-2 Space-charge accumula-

tion with a velocity of o.

If Eq. (7-3-1) remains valid throughout the entire transit time of the space-
charge layer, the factor of maximum growth is given by

Growth factor = Q(L Llv)

9
(7-3-2)
e c o , 0)
In Eq. (7-3-2) the layer is assumed to start at the cathode at t = 0, X = 0, and ar-
rive at the anode at t = L / u and X = L . For a large space-charge growth, this fac-
tor must be larger than unity. This means that

This is the criterion for classifying the modes of operation for the Gunn-effect
diodes. For n-type GaAs, the value of ~ u / ( ep, I)
l is about 10'2/cm2, where p, is I I
assumed to be 150 cm2/V . s.

Example 7-3-1: Criterion of Mode Operation

An n-type GaAs Gunn diode has the following parameters:

Electron drift velocity: vd = 2.5 X mls

Negative electron mobility: I p, ( = 0.015 m2/V . s
Relative dielectric constant: E, = 13.1

Determine the criterion for classifying the modes of operation.

Sec. 7.3 Modes of Operation

Solution From Eq. (7-3-3) the criterion is

= 1.19 X 10'6/m2
= 1.19 X 10'2/cm2
This means that the product of the doping concentration and the device length must be
noL > 1.19 x 10'2/cm2

7-3-2 Gunn Osci//afionModes (1Oi2/cm2 .s (noL ) < 1014/cmz)

Most Gunn-effect diodes have the product of doping and length (no L) greater than
10'2/cm2.However, the mode that Gunn himself observed had a product noL that is
much less. When the product of noL is greater than 10'Z/cm2in GaAs, the space-
charge perturbations in the specimen increase exponentially in space and time in ac-
cordance 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 [I91

where vdomis the domain velocity and Leffis 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

As described previously, the normal Gunn domain mode (or Gunn oscillation
mode) is operated with the electric field greater than the threshold field (E > Eth).
The high-field domain drifts along the specimen until it reaches the anode or until
the low-field value drops below the sustaining field E, required to maintain v, as
shown in Fig. 7-3-3. The sustaining drift velocity for GaAs is v, = lo7cmls. Since
the electron drift velocity v varies with the electric field, there are three possible do-
main modes for the Gunn oscillation mode.
Transit-time domain mode (fL = lo7 cmls). When the electron drift velocity
ud is equal to the sustaining velocity v,, the high-field domain is stable. In other
words, the electron drift velocity is given by

Then the oscillation period is equal to the transit time-that is, TO = 7,.This situa-
tion 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.
Transferred Electron Devices (TEDs) Chap. 7

0 Es E,h E Figure 7-3-3 Electron drift velocity

Elrctr~cfield E versus electric field.

J$~-T~
- _ - - -L- - - - - -. I:,,,
I-,
"'cl'

Figure 7-3-4 Gunn domain modes.

Sec. 7.3 Modes of Operation 289

Delayed domain mode (lo6 cmls < fL < lo7 cmls). When the transit time is
chosen so that the domain is collected while E < Ethas 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, r0 > 7,. This delayed
mode is also called inhibited mode. The efficiency of this mode is about 20%.
Quenched domain mode (fL > 2 X lo7 cmls). If the bias field drops below
the sustaining field E, 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 oscilla-
tions occur at the frequency of the resonant circuit rather than at the transit-time fre-
quency. 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, 231. Theoretically, the efficiency of
quenched domain oscillators can reach 13% [22].

Limited-Space-ChargeAccumulation (LSA)
7-3-3
Mode (fL > 2 x 10' cmls)

When the frequency is very high, the domains do not have sufficient time to form
while the field is above threshold. As a result, most of the domains are maintained
in the negative conductance state during a large fraction of the voltage cycle. Any
accumulation of electrons near the cathode has time to collapse while the signal is
below threshold. Thus the LSA mode is the simplest mode of operation, and it con-
sists of a uniformly doped semiconductor without any internal space charges. In this
instance, the internal electric field would be uniform and proportional to the applied
voltage. The current in the device is then proportional to the drift velocity at this
field level. The efficiency of the LSA mode can reach 20%.
The oscillation period r0should be no more than several times larger than the
magnitude of the dielectric relaxation time in the negative conductance region r d .
The oscillation indicated in Fig. 7-3-4(d) is TO = 3rd. It is appropriate here to define
the LSA boundaries. As described earlier, the sustaining drift velocity is lo7 cmls as
shown in Eq. (7-3-5) and Fig. 7-3-3. For the n-type GaAs, the product of doping
and length (noL) is about 10'2/cm2.At the low-frequency limit, the drift velocity is
taken to be
ue = fL = 5 X lo6 cmls (7-3-6)
The ratio of noL to fL yields

At the upper-frequency limit it is assumed that the drift velocity is

I u, = fL = 5 x lo7 cmls (7-3-8)
I
and the ratio of noL to fL is
I
I no
- = 2 x lo4 (7-3-9)
I f
r
290 Transferred Electron Devices (TEDs) 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 (noL < 1 0'2/cm2)

When the noL product of the device is less than about 10'2/cm2,the device exhibits
amplification at the transit-time frequency rather than spontaneous oscillation. This
situation occurs because the negative conductance is utilized without domain forma-
tion. There are too few carriers for domain formation within the transit time. There-
fore 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.

b------ High-field domain

-
.VI
10"

-6
5
M
-E
X
X

E 107
0.
2
---------
*
i
1
Relaxation oscillation

I 06
101'
noL-Carrier concentration x length ( ~ m - ~ )

Figure 7-3-5 Mode chart. (Afrer M. Uenohara [24]; reprinted by permission oj

McCruw -Hill Book Company.)
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:
r, = domain transit time
rd = dielectric relaxation time at low field
r, = domain growth time
r0 = natural period of oscillation of a high-Q external electric circuit
The modes described previously are summarized in Table 7-3- 1.
Sec. 7.4 LSA Diodes 291

TABLE 7-3.1 MODES OF OPERATION OF GUNN OSCILLATORS

Mode Time relationship Doping level Nature of circuit

Stable amplifier TO 2 r1 Nonresonant
Gun domain T~ 5 7, Nonresonant;
To = 7, constant voltage
Quenched domain T~ 5 7, Resonant;
7" = T, finite impedance
Delayed domain T~ 5 r1 Resonant;
finite impedance
LSA Multiple resonances;
high impedance;
high dc bias

7-4 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 10'2/cm2and
if the ratio of doping no to frequency f is within 2 X 10' to 2 X lo4 s/cm3, the high-
field 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 ho-
mogeneous. 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 = -70
(7-4- 1)

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 build-
ing up at the cathode. Since the oscillation period TO of the RF signal is less than the
domain-growth time constant T, , the total voltage swings below the threshold before
the domain can form. Furthermore, since 70is much greater than the dielectric re-
laxation time T ~ 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-resis-
tance region, and the space charge is not allowed to build up. The frequency of os-
cillation in the LSA mode is independent of the transit time of the carriers and is de-
termined solely by the circuit external to the device. Also, the power-impedance
product does not fall off as I / fg; 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 Transferred Electron Devices (TEDs) Chap. 7

0
or
O(
I II * Electric field
L
I I
I

t-
I I-
I

i? -I , R F signal - - .

i= -I
-
I
I
II
I
I
I
-I
I I
v d c bias Figure 7-4-1 LSA mode operation.

of an LSA oscillator can be simply written

where q =dc-to-RF conversion efficiency (primarily a function of material and

circuit considerations)
Vo =operating voltage
lo=operating current
M =multiple of the operating voltage above negative-resistance threshold
voltage
Eth=threshold field (about 3400 Vlcm)
L =device length (about 10 to 200 p m )
no =donor concentration (about lOI5 e /cm3)
e =electron charge (1.6 x 10-19 C)
uo =average carrier drift velocity (about lo7 cmls) .
A =device area (about 3 X to 20 X 10 - cm2)
For an LSA oscillator, no is primarily determined by the desired operating frequency
fo so that, for a properly designed circuit, peak power output is directly proportional
to the volume (LA) of the device length L multiplied by the area A of the active
layer. Active volume cannot be increased indefinitely. In the theoretical limit, this is
due to electrical wavelength and skin-depth considerations. In the practical limit,
however, available bias, thermal dissipation capability, or technological problems
associated with material uniformity limit device length. The power capabilities of
LSA oscillators vary from 6 kW (pulse) at 1.75 GHz to 400 W (pulse) at 51 GHz.
Sec. 7.5 InP Diodes 293

Example 7-4-1: Output Power of an LSA Oscillator

An LSA oscillator has the following parameters:

Conversion efficiency: r] = 0.06

Multiplication factor: M = 3.5
Threshold field: Eth = 320 kV/m
Device length: L = 12pm
Donor concentration: no = 10'' m3
Average carrier velocity: uo = 1.5 X lo5 m/s
Area: A = 3 x lo-' m2

Determine the output power in milliwatts.

Solution From Eq. (7-4-2) the output power is
P = q(ME,hL)(noevoA)= 0.06 X (3.5 X 320 X lo3 X 12 X

x (lo2' x 1.6 x 10-j9 x 1.5 x 105 x 3 x

= 0.06 X 13.44 X 7.2 X lo-' = 581 mW

7.5 InP 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 de-
scribed 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 dc
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 al-
loys 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 con-
tribution 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 oc-
curs 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
1 by the lower-valley electrons from the weak coupling.1t can be seen from Fig. 7-5- 1
Transferred Electron Devices (TEDs) Chap. 7

Strong coupling

Coi~ductlon
Lower band

I:orhicld~.n
gap

Valcnce
band

Figure 7-5-1 Three-valley-model energy level for InP diode.

4
Peak current
Peak current

Valley current

0 2Eth4 6 8 I0
Electric field (kV/cm) Electric field (kV/cm)

Figure 7-5-2 Peak-to-valley current ra- Figure 7-5-3 Peak-to-valley current ra-
tio for n-type GaAs. tio for InP.

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 op-
erating conditions electrons concentrate in the middle valley. Because InP has a
greater energy separation between the lower valley and the nearest energy levels, the
thermal excitation of electrons has less effect, and the degradation of its peak-to-
valley 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 lo7 cmls. 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
Figure 7-5-4 Active-layer thickness
versus frequency for InP diode. (From
B. C. Taylor and D . J . Colliver [29];
Frequency of oscillator in GHz reprinted by permission of IEEE, Inc.)

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 oscilla-
tors 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 ob-
tained from a device were dependent on the active-layer thickness. The InP oscilla-
tor 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, 291. 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 7-5.1 BEST PERFORMANCE OF InP DIODE

Frequency Thickness Power Operation Efficiency Reference

@Hz) (W) (%)

pulsed
pulsed
pulsed
pulsed
pulsed
pulsed
CW
pulsed
CW
pulsed
pulsed
CW
Transferred Electron Devices (TEDs) Chap. 7

7-6 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 in n-type InP, n-type CdTe, alloys of n-GaAs and n-
Gap, 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 p m long with a carrier concentration of 5 X 1014/cm3
-
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 fur-
ther 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 field-
induced 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 p = 1100 cm2/V s at room tem-
perature. Hilsum has estimated that (1 11) minima are the next lowest in energy, be-
ing 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 ki-
netic 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 cur-
rent. 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 heat-
ing 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.

7-7 MICROWAVE GENERATION AND AMPLIFICATION

7-7-1 Microwave Generation
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 un-
stable and divides up into two domains of different conductivity and different elec-
tric field but the same drift velocity. Figure 7-7-1 shows the stable and unstable re-
gions.
At the initial formation of the accumulation layer, the field behind the layer de-
creases and the field in front of it increases. This process continues as the layer trav-
els from the cathode toward the anode. As the layer approaches the anode, the field
Sec. 7.7 Microwave Generation and Amplification

Electric field E velocity.

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 dis-
appears 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

where v d is the velocity of the domain or approximately the drift velocity of the elec-
trons and L,ff 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. Low efficiency at frequencies above 10 GHz

2. Small tuning range
3. Large dependence of frequency on temperature
4. High noise

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 per-
cent. Gunn diode oscillators have better noise performance than IMPATTs. They are
298 Transferred Electron Devices (TEDs) Chap. 7

NUMBERS INDICATE
EFFICIENCY (%)
O' 30

GaAs IMPATT

0.1 7

HUGHES
- MAICOM
RAYTHEON
TRW Figure 7-7-2 State-of-the-art perfor-
VARIAN mance for GaAs and InP Gunn diodes.
1 1 (From H. Hieslrnair et a l . [36];reprinted
5 10 50 100 So(D by permission of Microwave Journal,
FREQUENCY (GHz) 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.

7-7-2 Microwave Amplification

When an RF signal is applied to a Gunn oscillator, amplification of the signal oc-

curs, provided that the signal frequency is low enough to allow the space charge in
the domain to readjust itself. There is a critical value of fL above which the device
will not amplify. Below this frequency limit the sample presents an impedance with
a negative real part that can be utilized for amplification. If rto L becomes less than
10'2/cm2, domain formation is inhibited and the device exhibits a nonuniform field
distribution that is stable with respect to time and space. Such a diode can amplify
signals in the vicinity of the transit-time frequency and its harmonics without oscilla-
tion. If this device is used in a circuit with enough positive feedback, it will oscil-
late. Hakki has shown that the oscillation diode can amplify at nearby frequencies or
can be used simultaneously as an amplifier and local oscillator [37]. However, the
output power of a stable amplifier is quite low because of the limitation imposed by
the value of noL .
Sec. 7.7 Microwave Generation and Amplification 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
noL must be larger than 10L2/cm2 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 amplzj7er (TDA).
Although a large number of possible amplifier circuits exist, the essential fea-
ture 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 re-
spect 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 exhib-
ited 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 10 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 sev-
eral amplifiers that have been designed since 1970.

Pulse ncncrator Bias

$ 5042 coaxial

Spectrum
analyzcr o H 1 1

Scope I
t
Circulator
GaAs
wafer Figure 7-7-3 Gunn-diode amplifier cir-
cuit. (After H. W. Thim 1381; reprinted
Detector diode by permission of IEEE, Inc.)

TABLE 7.7.1 CW GUNN-DIODE AMPLIFIER PERFORMANCE

Frequency 3-dB bandwidth Small signal gain Power gain Efficiency

band (dB) (dB)

C 4.5- 8.0 8 3 3
X 7.5-10.75 12 1.65 2.3
8.0-12.0 6 1.8 2.5
KU 12.0-16.0 6 1.5 2.5
13.0-15.0 8 0.36 2

Source: After B.S. Perlrnan et al. [39]; reprinted by permission of IEEE, Inc.
300 Transferred Electron Devices (TEDs) Chap. 7

REFERENCES
[ l ] 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 semi-
conductors. Proc. Phys. Soc., 78, 293-304, August 1961.
[3] HILSUM, C., Transferred electron amplifiers and oscillators. Proc. IEEE, 50, 185- 189,
February 1962.
[4] GUNN,J. B., Microwave oscillations of current in 111-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 111-V semiconductors. Solid-state
Communications, 1, 88-91 (1963).
[8] GUNN,J. B., Instabilities of current in 111-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. Conf. on Physics of Semiconductor "Plasma Effects in Solids," 199-207, Tokyo,
1964.
[lo] KROEMER, H., Proposed negative-mass microwave amplifier. Physical Rev., 109, No. 5.
1856, March 1, 1958.
[ l l ] 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. FAWCEIT,Calculation of the velocity-field characteristics of gal-
lium arsenide. Appl. Phys. Letters, 21, 498 (1966).
[I41 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 gal-
lium arsenide. Appl. Phys. Letters, 10, 50 (1967).
[I61 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 in n-type germanium. Appl. Phys. Letters, 11, 253 (1967).
[I91 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 dop-
ing 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, 110 (1968).
[24] UENOHARA, M., Bulk gallium arsenide devices. Chapter 16 in H. A. Watson (Ed.), Mi-
crowave Semiconductor Devices and Their Circuit Application. McGraw-Hill Book
Company, New York, 1969.
1251 COPELAND, J. A., CW operation of LSA oscillator diodes-44 to 88 GHz. Bell System
Tech. J . , 46, 284-287, January 1967.
1261 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, De-
cember 1970.
[28] COLLIVER, D., and B. PREW,Indium phosphide: Is it practical for solid state microwave
sources? Electronics, 110-1 13, April 10, 1972.
[29] TAYLOR, B. C., and D. J. COLLIVER, Indium phosphide microwave oscillators. IEEE
Trans. on Electron Devices, ED-18, No. 10, 835-840, October 1971.
[30] FOYT,A. G., and A. L. MCWHOWER, 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,
547-55 1, September 1967.
1331 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 Ampl$cation: Devices and Applications. August 17- 19,
1971, Cornell University.
1361 HIESLMAIR, H., ET AL.,State of the art of solid-state and tube transmitters. Microwave
J., 26, No. 10, 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.
S o o ~ o o 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.
Transferred Electron Devices (TEDs) Chap. 7

PROBLEMS

Transferred Electron Devices (TEDs)

1 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 do-
main theory, two-valley theory, and three-valley theory.
c. Discuss the differences between transferred electron devices and avalanche transit-
time devices.
7-2. Describe the operating principles of tunnel diodes, Gunn diodes, and LSA diodes.
7-3. Derive Eq. (7-2-1 1).
7-4. Describe the modes of operation for Gunn diodes.
7-5. Describe the Ridley-Watkins-Hilsum theory.
7-6. The LSA oscillation mode is defined between 2 X lo4 and 2 X lo5 ratios of doping
over frequency as shown in Fig. 7-3-1. Derive Eqs. (7-3-7) and (7-3-9).
7-7. For a transit-time domain mode, the domain velocity is equal to the carrier drift veloc-
ity and is about lo7 cmls. Determine the drift length of the diode at a frequency of
8 GHz.
Chapter 8

I Avalanche Transit-Time
Devices

8-0 INTRODUCTION
Avalanche transit-time diode oscillators rely on the effect of voltage breakdown
across a reverst-biased p-n junction to produce a supply of holes and electrons. Ever
since the development of modern semiconductor device theory scientists have specu-
lated 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 de-
pends 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 dc 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 mi-
crowave frequencies. The device was originally proposed in a theoretical paper by
Read [I] 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 opera-
tion. In this mode the typical dc-to-RF conversion efficiency is 5 to lo%, and fre-
quencies are as high as 100 GHz with silicon diodes. The other mode is the TRAP-
ATT 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 Avalanche Transit-Time Devices Chap. 8

by minority carrier injection from forward-biased junctions rather than being ex-
tracted 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-n-
metal. BARITT diodes have low noise figures of 15 dB, but their bandwidth is rela-
tively narrow with low output power.

8-1 READ DIODE

-
8-1 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 [I]. 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 dc electric field distribution that exists when a large reverse bias
is applied across the diode is shown in Fig. 8- 1- 1.

Space-charge region
Avalanche Inactive
Drift region
(a)

Silicon structure + i(or v ) -

(b)

Field distribution

I Distance

rrm

Distance

Figure 8-1-1 Read diode.

Sec. 8.1 Read Diode 305

The Read diode is an n f -p-i-p+ structure, where the superscript plus sign de-
notes 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 f - p junction and the i-p+ junction is
called the space-charge region. Similar devices can be built in the p+-n-i-n+ struc-
ture, 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 cir-
cuit. The device can produce a negative ac resistance that, in turn, delivers power
from the dc bias to the oscillation.

8-1-2 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- 1 (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 hun-
dred kilovolts per centimeter. Carriers (holes) moving in the high field near the n+-p
junction acquire energy to knock valence electrons into the conduction band, thus
producing hole-electron pairs. The rate of pair production, or avalanche multiplica-
tion, 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
vd of about lo7 cm/s 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

and the avalanche multiplication factor is

where V = applied voltage

Vb = avalanche breakdown voltage
n = 3-6 for silicon is a numerical factor depending on the doping of p+-n or
n+-p junction
306 Avalanche Transit-Time Devices Chap. 8

The breakdown voltage for a silicon p+-n junction can be expressed as

where p, = resistivity
p, = electron mobility
E, = semiconductor permittivity
Emax= maximum breakdown of the electric field
Figure 8-1-2 shows the avalanche breakdown voltage as a function of impurity at a
p+-n junction for several semiconductors.

1014 I 015 I o16 loi7 I o18

Figure 8-1-2 Breakdown voltage ver-
lmpurity density in ~ r n - ~ sus impurity doping.

8.1 -3 Carrier Current I,(t) and External Current I,(t)

As described previously, the Read diode is mounted in a microwave resonant circuit.

An ac voltage can be maintained at a given frequency in the circuit, and the total
field across the diode is the sum of the dc and ac fields. This total field causes break-
down at the n+-p junction during the positive half of the ac voltage cycle if the field
is above the breakdown voltage, and the carrier current (or the hole current in this
case) lo(t) generated at the n+-p junction by the avalanche multiplication grows ex-
ponentially with time while the field is above the critical value. During the negative
half cycle, when the field is below the breakdown voltage, the carrier current Io(t)
decays exponentially to a small steady-state value. The carrier current Io(t) is the
current at the junction only and is in the form of a pulse of very short duration as
shown in Fig. 8-1-3(d). Therefore the carrier current lo(t) reaches its maximum in
the middle of the ac voltage cycle, or one-quarter of a cycle later than the voltage.
Under the influence of the electric field the generated holes are injected into the
space-charge region toward the negative terminal. As the injected holes traverse the
drift space, they induce a current I,(t) in the external circuit as shown in Fig.
8- 1-3(d).
Sec. 8.1 Read Diode

(a)
Structure
+

(b)

Field distribution
I
I I I

voltage

(d)
[,,(I) and

Figure 8-1-3 Field, voltage, and currents in Read diode. (After Read [I];
reprinted by permission of the Bell System, AT&T Co.)

When the holes generated at the nf-p junction drift through the space-charge
region, they cause a reduction of the field in accordance with Poisson's equation:

where p is the volume charge density and E, is the semiconductor permittivity.

Since the drift velocity of the holes in the space-charge region is constant, the
induced current I,(t) in the external circuit is simply equal to
308 Avalanche Transit-Time Devices Chap. 8

where Q = total charge of the moving holes

vd = hole drift velocity
L = length of the drift i region
It can be seen that the induced current Ie(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 Ie(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
Ie(t) because of the moving holes is delayed by ~ / or 2 90' relative to the pulsed car-
rier current lo(t) generated at the n+-p junction. Since the carrier lo(t) is delayed by
one-quarter of a cycle or 90" relative to the ac voltage, the external current Ie(t) is
then delayed by 180" relative to the voltage as shown in Fig. 8-1-3(d). Therefore the
cavity should be tuned to give a resonant frequency as

Then

Since the applied ac voltage and the external current I&) are out of phase by 180°,
negative conductance occurs and the Read diode can be used for microwave oscilla-
tion and amplification. For example, taking vd = lo7 cmls for silicon, the optimum
operating frequency for a Read diode with an i-region length of 2.5 p m is 20 GHz.

8-1-4 Output Power and Quality Factor Q

The external current Ie(t) approaches a square wave, being very small during the
positive half cycle of the ac voltage and almost constant during the negative half cy-
cle. Since the direct current I d supplied by the dc bias is the average external current
or conductive current, it follows that the amplitude of variation of Ie(t) is approxi-
mately equal to I d . If V, is the amplitude of the ac voltage, the ac power delivered is
found to be
P = 0.707VaId W/unit area (8- 1-5)

The quality factor Q of a circuit is defined as

maximum stored energy

Q = w
average dissipated power

Since the Read diode supplies ac energy, it has a negative Q in contrast to the posi-
tive 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 in-
creases, the stored energy, or energy of oscillation, increases faster than the energy
Sec. 8.2 IMPAlT Diodes 309

delivered per cycle. This is the condition required in order for stable oscillation to be
possible.

8.2 IMPATT DIODES

8-2-1 Physical Structures

A theoretical Read diode made of an n+-p-i-p+or p+-n-i-nc structure has been ana-
lyzed. 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 lo(?)
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, how-
ever, 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.

8-2-2 Negative Resistance

Small-signal analysis of a Read diode results in the following expression for the real
part of the diode terminal impedance [5]:
2L2 1 1 - cos 8
R = R, + ---
V ~ E , A1 - w'/w? 8
where R, = passive resistance of the inactive region
vd = carrier drift velocity
L = length of the drift space-charge region
A = diode cross section
e, = semiconductor dielectric permittivity
310 Avalanche Transit-Time Devices Chap. 8

(a) Abrupt p-n junction

(b) Linearly graded

p-ti junction P + P II n+
I

Doping profile

(c) p-i-11 diode

nh 102"- -
-
s
E

Doping profile .s
-
L

3,
J
Figure 8-2-1 Three typical silicon
IMPATT diodes. (After R. L. Johnston et
2 1013--- I I al. 141; reprinted by permission of the
b
0 1 2 3 pm Bell System, AT&T C o . )

Moreover, 8 is the transit angle, given by

and w , is the avalanche resonant frequency, defined by

Sec. 8.2 IMPATT Diodes 311

In Eq. (8-2-3) the quantity a ' is the derivative of the ionization coefficient with re-
spect 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 > w , is plot-
ted in Fig. 8-2-2. The peak value of the negative resistance occurs near 8 = T . For
transit angles larger than .rr and approaching 3 ~ 1 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 .rr transit angle. That is,

5
+aII)

t
0 ?I 2rr 377
Figure 8-2-2 Negative resistance versus
Transit angle 0 transit angle.

Power Output and Efficiency

8-2-3

At a given frequency the maximum output power of a single diode is limited by

semiconductor materials and the attainable impedance levels in microwave circuitry.
For a uniform avalanche, the maximum voltage that can be applied across the diode
is given by
Vm = EmL (8-2-5)
where L is the depletion length and Em is the maximum electric field. This maximum
applied voltage is limited by the breakdown voltage. Furthermore, the maximum
current that can be carried by the diode is also limited by the avalanche breakdown
process, for the current in the space-charge region causes an increase in the electric
field. The maximum current is given by

Therefore the upper limit of the power input is given by

P, = I,V, = E ; E , U ~ A
312 Avalanche Transit-Time Devices Chap. 8

The capacitance across the space-charge region is defined as

Substitution of Eq. (8-2-8) in Eq. (8-2-7) and application of 2 n - f ~= 1 yield

It is interesting to note that this equation is identical to Eq. (5-1-60) of the power-
frequency limitation for the microwave power transistor. The maximum power that
can be given to the mobile carriers decreases as l / f.For silicon, this electronic
limit is dominant at frequencies as high as 100 GHz. The efficiency of the IMPATT
diodes is given by

Pa, Vu Iu

= P, = (E)(Z)

NUMBERS INDICATE
- EFFICIENCY (%)
\,30
GaAs IMPATT

- MAlCOM
RAYTHEON
- TRW
VARIAN Figure 8-2-3 State-of-the-art perfor-
1 1 L 1 1 1 1 1 I I 1
mance for GaAs and Si IMPATTs. (From
5 10 50 100 500 H. Hieslmair, et al. [6]; reprinted by
FREQUENCY (GHz) permission of Microwave Journal.)
Sec. 8.2 IMPAlT 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 dc current is about 2 / n , so
that the efficiency would be about I/T 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 gal-
lium 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 l / f and
l / f slopes. The transition from the l / f to the l / f 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: CW Output Power of an IMPATT Diode

An IMPATT diode has the following parameters:

Carrier drift velocity: ud = 2 X 10' cmls

Drift-region length: L=6pm
Maximum operating voltage: Voma, = 100 V
Maximum operating current: Io,,, = 200 mA
Efficiency: 77 = 15%
Breakdown voltage: Vbd = 90 V

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

P = 77Pdc=0.15 X 100 X 0.2 = 3W
b. From Eq. (8-2-4) the resonant frequency is
Avalanche Transit-Time Devices Chap. 8

8-3TRAPATT DIODES

1 Physical Structures
8-3-

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 reverse-
biased 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+) struc-
tures with the n-type depletion region width varying from 2.5 to 12.5 p m . The dop-
ing of the depletion region is generally such that the diodes are well "punched
through" at breakdown; that is, the dc 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 p m . The TRAPATT diode's diameter ranges
from as small as 50 p m for CW operation to 750 p m at lower frequency for high-
peak-power devices.

Principles of Operation
8-3-2

Approximate analytic solutions for the TRAPATT mode in pi-n-n+ diodes have
been developed by Clorfeine et al. [8] and DeLoach [9] among others. These analy-
ses have shown that a high-field avalanche zone propagates through the diode and
fills the depletion layer with a dense plasma of electrons and holes that become
trapped in the low-field region behind the zone. A typical voltage waveform for the
TRAPATT mode of an avalanche p'-n-n+ diode operating with an assumed square-
wave current drive is shown in Fig. 8-3-1. At point A the electric field is uniform
throughout the sample and its magnitude is large but less than the value required for
avalanche breakdown. The current density is expressed by

where E, is the semiconductor dielectric permittivity of the diode.

At the instant of time at point A , the diode current is turned on. Since the only
charge carriers present are those caused by the thermal generation, the diode ini-
tially charges up like a linear capacitor, driving the magnitude of the electric field
above the breakdown voltage. When a sufficient number of carriers is generated, the
particle current exceeds the external current and the electric field is depressed
throughout the depletion region, causing the voltage to decrease. This portion of the
cycle is shown by the curve from point B to point C. During this time interval the
electric field is sufficiently large for the avalanche to continue, and a dense plasma of
electrons and holes is created. As some of the electrons and holes drift out of the
ends of the depletion layer, the field is further depressed and "traps" the remaining
plasma. The voltage decreases to point D. A long time is required to remove the
plasma because the total plasma charge is large compared to the charge per unit time
Sec. 8.3 TRAPATT Diodes

Charging

Figure 8-3-1 Voltage and current

cI waveforms for TRAPATT diode. (After
0 712 T A. S . Clorfeine et al. [8]; reprinted by
Time permission of RCA Laboratory.)

in the external current. At point E the 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 F all 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 V A until the current comes back on
and the cycle repeats. The electric field can be expressed as

where NA is the doping concentration of the n region and x is 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 ) = E m , yielding

Differentiation of Eq. (8-3-3) with respect to time t results in

where v, is the avalanche-zone velocity.

316 Avalanche Transit-Time Devices Chap. 8

Example 8-3-1: Avalanche-Zone Velocity of a TRAPATT Diode

A TRAPATT diode has the following parameters:

Doping concentration: NA = 2 X 1015 cm-'

Current density: J = 20 kAlcm2

Calculate the avalanche-zone velocity.

Solution From Eq . (8-3-4) the avalanche-zone velocity is

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 de-
pendence of the drift velocity on the field, the electrons and holes will drift at veloc-
ities determined by the low-field mobilities, and the transit time of the carriers can
become much longer than

where v, is the saturated carrier drift velocity.

Thus the TRAPATT mode can operate at comparatively low frequencies, since
the discharge time of the plasma-that is, the rate Q / I of its charge to its current-
can be considerably greater than the nominal transit time 7,of the diode at high field.
Therefore the TRAPATT mode is still a transit-time mode in the real sense that the
time delay of carriers in transit (that is, the time between injection and collection) is
utilized to obtain a current phase shift favorable for oscillation.

Power Output and Efficiency

8-3-3

RF power is delivered by the diode to an external load when the diode is placed in a
proper circuit with a load. The main function of this circuit is to match the diode ef-
fective negative resistance to the load at the output frequency while reactively termi-
nating (trapping) frequencies above the oscillation frequency in order to ensure
TRAPATT operation. To date, the highest pulse power of 1.2 kW has been obtained
at 1.1 GHz (five diodes in series) [lo], and the highest efficiency of 75% has been
achieved at 0.6 GHz (1 1). Table 8-3- 1 shows the current state of TRAPATT diodes
1121.
The TRAPATT operation is a rather complicated means of oscillation, how-
ever, and requires good control of both device and circuit properties. In addition,
Sec. 8.4 BARllT Diodes 317

TABLE 8-3-1 TRAPATT OSCILLATOR CAPABILITIES

Frequency Peak power Average power Operating voltage Efficiency

(GHz) (W) (W) (V) (%)

0.5 600 3 150 40

1.O 200 1 110 30
1.o 400 2 110 35
2.0 100 1 80 25
2.0 200 2 80 30
4.0 100 1 80 20
8.0 50 1 60 15

Source: After W. E. Wilson [12]; reprinted by permission of Horizon House.

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.

8.4 BARITT DIODES

8-4-1 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 sat-
uration velocity through the v region and are collected at the p contact. The diode
exhibits a negative resistance for transit angles between 7r and 27r. The optimum
transit angle is approximately 1.h.
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 disad-
vantages of BARITT diodes are relatively narrow bandwidth and power outputs lim-
ited to a few milliwatts.

8-4-2 Principles of Operation

A crystal n-type silicon wafer with 11 (1-cm resistivity and 4 X loi4per cubic cen-
timeter doping is made of a 10-pm thin slice. Then the n-type silicon wafer is sand-
wiched between two PtSi Schottky barrier contacts of about 0.1 p m thickness. A
schematic diagram of a metal-n-metal structure is shown in Fig. 8-4-l(a).
The energy-band diagram at thermal equilibrium is shown in Fig. 8-4-l(b),
where +,,and +n2 are the barrier heights for the metal-semiconductor contacts, re-
Avalanche Transit-Time Devices Chap. 8

Metal 9 ( a ) M-11-M ctl~odc

(0t.nrr:y band u n d e r
Holes h ~ a conilltion
\

Figure 8-4-1 M-n-M diode. (After D . J . Coleman and S. M . Sze 1131; reprinted
b y permission of the Bell System, AT&T C o . )

spectively. For the PtSi-Si-PtSi structure mentioned previously, $, = $,2 =

0.85 eV. The hole barrier height $pz for the forward-biased contact is about 0.15 eV.
Figure 8-4-l(c) shows the energy-band diagram when a voltage is applied. The
mechanisms responsible for the microwave oscillations are derived from:

1. The rapid increase of the carrier injection process caused by the decreasing po-
tential barrier of the forward-biased metal-semiconductor contact
2. An apparent 3 ~ / transit
2 angle of the injected carrier that traverses the semi-
conductor 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

where N = doping concentration

L = semiconductor thickness
E, = semiconductor dielectric permittivity
Sec. 8.4 BARllT Diodes

Figure 8-4-2 Current versus voltage o f

a BARITT diode (PtSi-Si-PtSi). (After
D . J . Coleman and S . M . Sze [ I J ] ;
reprinted by permission of the Bell Sys-
V (volts) tem, AT&T Co.)

The current-voltage characteristics of the silicon MSM structure (PtSi-

Si-PtSi) were measured at 77" K and 300" K. The device parameters are L = 10
p m , N = 4 x 1014 cm3, +,, = +,2 = 0.85 eV, and area = 5 X 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 - 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 ( 4 ~of) about 0.15 eV.

8-4-3 Micro wave 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 single-
sideband 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 indi-
cated 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 Avalanche Transit-Time Devices Chap. 8

Figure 8-4-3 Power output versus cur-

rent for three silicon M-n-M devices.
(Ajier D . J. Coleman and S . M . Sze
[13]: reprinted by permission of the Bell
Current (rnA ) System, AT&T C o . )

Example 8-4-1: Breakdown Voltage of a BARITT Diode

An M-Si-M BARITT diode has the following parameters:

Relative dielectric constant of silicon: E, = 11.8

Donor concentration: N = 2.8 x 10'' m-3

Silicon length: L =6pm

Determine: a. the breakdown voltage; b. the breakdown electric field.

Solution

a. From Eq. (8-4- 1) the breakdown voltage is double its critical voltage as

b. The breakdown electric field is

8-5 PARAMETRIC DEVICES

8-5-1 Physical Description

A parametric device is one that uses a nonlinear reactance (capacitance or induc-

tance) 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 ex-

citation 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 (161 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, mi-
crowave electronics engineers dreamed of a solid-state microwave device to replace
the noisy electron beam amplifier. In 1957 Suhl [18] proposed a microwave solid-
state amplifier that used ferrite. The first realization of a microwave parametric am-
plifier following Suhl's proposal was made by Weiss [19] in 1957. After the work
done by Suh1 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 dc power supply as microwave tubes do. In this respect, the para-
metric amplifier is analogous to the quantum amplifier laser or maser in which an ac
power supply is used.

8-5-2 Nonlinear Reactance and Manley-Ro we Power

Relations

A reactance is defined as a circuit element that stores and releases electromagnetic

energy as opposed to a resistance, which dissipates energy. If the stored energy is
predominantly in the electric field, the reactance is said to be capacitive; if the
stored energy is predominantly in the magnetic field, the reactance is said to be in-
ductive. In microwave engineering it is more convenient to speak in terms of
voltages and currents rather than electric and magnetic fields. A capacitive reactance
may then be a circuit element for which capacitance is the ratio of charge on the ca-
pacitor over voltage across the capacitor. Then

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,

The analogous definition of a nonlinear inductance is

a@
L (i) = -
di
322 Avalanche Transit-Time Devices 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 v, is much

smaller than the pumping voltage v p , and the total voltage across the nonlinear ca-
pacitance C ( t ) is given by
u = v, + vp = V, cos (w,t) + Vp cos ( w p t ) (8-5-4)
where V, -=3V p. The charge on the capacitor can be expanded in a Taylor series about
the point v, = 0, and the first two terms are

For convenience, it is assumed that

where C ( v p ) is periodic with a fundamental frequency of w p . If the capacitance

C ( u p ) is expanded in a Fourier series, the result is

C (v,) = C C , cos (nwpt)

n=O

Since up is a function of time, the capacitance C ( v p )is also a function of time. Then

C(t) = C C , cos (nwpt)

n=O

The coefficients C , are the magnitude of each harmonic of the time-varying capaci-
tance. In general, the coefficients C , are not linear functions of the ac pumping
voltage u p . 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 am-
plitudes.
The current through the capacitance C ( t ) is the derivative of Eq. (8-5-5) with
respect to time and it is

It is evident that the nonlinear capacitance behaves like a time-varying linear

capacitance for signals with amplitudes that are much smaller than the amplitude of
the pumping voltage. The first term of Eq. (8-5-9) yields a current at the pump fre-
quencyf, and is not related to the signal frequency&.

Large-signal method. If the signal voltage is not small compared with the
pumping voltage, the Taylor series can be expanded about a dc bias voltage Vo in a
junction diode. In a junction diode the capacitance C is proportional to
(40- V)-'I2 = Vil/*,where 40is the junction barrier potential and V is a negative
Sec. 8.5 Parametric Devices

voltage supply. Since

[Vo + V, cos ( o p t ) ] - ' / 2= V G " ~ for Vp -=S Vo

the capacitance C ( t ) can be expressed as

C ( t ) = Co[l + 2 y cos ( o p t ) ]
The parameter y is proportional to the pumping voltage up and indicates the coupling
effect between the voltages at the signal frequency & 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.

Figure 8-5-1 Equivalent circuit for Manley-Rowe derivation.

In Fig. 8-5- 1, one signal generator and one pump generator at their respective
frequenciesf, andf, , 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 frequen-
cies. In the presence of two applied frequencies& andf, , an infinite number of reso-
nant frequencies of mf, & nf, 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 capac-
itor 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 f, and f, and the output
power at the other frequencies mf, -+ n& .
From Eq. (8-5-4) the voltage across the nonlinear capacitor C ( t ) can be ex-
pressed in exponential form as
324 Avalanche Transit-Time Devices Chap. 8

The general expression of the charge Q deposited on the capacitor is given by

In order for the charge Q to be real, it is necessary that

The total voltage v can be expressed as a function of the charge Q . A similar Taylor
series expansion of v ( Q ) shows that

In order for the voltage v to be real, it is required that

The current flowing through C ( t ) is the total derivative of Eq. (8-5-12) with
respect to time. This is

where I,,, = j (mop + ~ J , ) Q ,and

, ~ I,,, = I*,.-, . Since the capacitance C ( t ) is as-
sumed to be a pure reactance, the average power at the frequencies mf, + nf, is

Then conservation of power can be written

Multiplication of Eq. (8-5-18) by a factor of (mop+ nos)/(mwp+ no,) and rear-

rangement of the resultant into two parts yield

Since
+ nos) =
~ m . n / ( ~ p jQm.n
then Pm,,/(moP+ no,) becomes - jVm,nQ,*,n-jV-m.-nQTm,-n and is independent of
opor o, . For any choice of the frequencies fp and5 , the resonating circuit external
to that of the nonlinear capacitance C ( t ) can be so adjusted that the currents may
keep all the voltage amplitudes Vm,,unchanged. The charges Q , , are then also un-
Sec. 8.5 Parametric Devices 325

changed, since they are functions of the voltages V,,,. Consequently, the frequen-
cies & and& can be arbitrarily adjusted in order to require

Equation (8-5-20) can be expressed as two terms:

Since P,,, = P -,,-,, then

Similarly,
c2
m=-m n=o
npm.,
mfp + nfs
=0

where w, and o,have been replaced by f, andf, , respectively.

Equations (8-5-23) and (8-5-24) are the standard forms for the Manley-Rowe
power relations. The term P,,, indicates the real power flowing into or leaving the
+
nonlinear capacitor at a frequency of mf, nf, . The frequency f, represents the fun-
damental frequency of the pumping voltage oscillator and the frequency f, signifies
the fundamental frequency of the signal voltage generator. The sign convention for
the power term P,,, will follow that power flowing into the nonlinear capacitance
or the power coming from the two voltage generators is positive, whereas the power
leaving from the nonlinear capacitance or the power flowing into the load resistance
is negative.
Consider, for instance, the case where the power output flow is allowed at a
frequency of f, + f, as shown in Fig. 8-5-1. All other harmonics are open-
circuited. Then currents at the three frequencies&, f, , andf, + f, are the only ones
existing. Under these restrictions m and n vary from - 1 through zero to + 1, re-
spectively. Then Eqs. (8-5-23) and (8-5-24) reduce to

Po.I +--=o
- PII
fs h+fS
where Pl.oand Po,l are the power supplied by the two voltage generators at the fre-
quencies f, and f, , respectively, and they are considered positive. The power P I q l
flowing from the reactance into the resistive load at a frequency off,+ f, is consid-
ered negative.
326 Avalanche Transit-Time Devices Chap. 8

The power gain, which is defined as a ratio of the power delivered by the ca-
pacitor at a frequency off, + f, to that absorbed by the capacitor at a frequency off,
as shown in Eq. (8-5-26) is given by

Gain = -- f0-+

f,
' 5
(for modulator)

where f, + f, = f, and ( f , + f,) > f, > f, . 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 fre-
quency, Eq. (8-5-26) predicts that the parametric device will have a gain of
5
Gain = - (for demodulator) (8-5-28)
f, + fs

+
where f, = f, f, and f, = f, - f,. This type of parametric device is called the
parametric down-converter and its power gain is actually a loss.
If the signal frequency is at f , , the pump frequency at f, , and the output fre-
quency at f , , where f, = f, +f , , the power P I , Isupplied at f, is positive. Both P,,o
and Po.l are negative. In other words, the capacitor delivers power to the signal gen-
erator atf, instead of absorbing it. The power gain may be infinite, which is an un-
stable condition, and the circuit may be oscillating both atf, and f, . This is another
type of parametric device, often called a negative-resistance parametric amplifier.

8-5-3 Parametric Amplifiers

In a superheterodyne receiver a radio frequency signal may be mixed with a signal

from the local oscillator in a nonlinear circuit (the mixer) to generate the sum and
difference frequencies. In a parametric amplifier the local oscillator is replaced by a
pumping generator such as a reflex klystron and the nonlinear element by a time-
varying capacitor such as a varactor diode (or inductor) as shown in Fig. 8-5-2.
In Fig. 8-5-2, the signal frequency5 and the pump frequency f, are mixed in
the nonlinear capacitor C. Accordingly, a voltage of the fundamental frequenciesf,
andf, as well as the sum and the difference frequencies mf, k nf, appears across C .
If a resistive load is connected across the terminals of the idler circuit, an output
voltage can be generated across the load at the output frequencyf, . The output cir-
cuit, which does not require external excitation, is called the idler circuit. The out-
put (or idler) frequency f, in the idler circuit is expressed as the sum and the differ-
ence frequencies of the signal frequency 5 and the pump frequency f, . That is,
fo = mfp 'nf. (8-5-29)
where m and n are positive integers from zero to infinity.
If f, > f, , the device is called a parametric up-converter. Conversely, if
fo < f,, the device is known as a parametric down-converter.
Sec. 8.5 Parametric Devices

Signal circuit Idler circuit

Figure 8-5-2 Equivalent circuit for a parametric amplifier.

Parametric up-converter. A parametric up-converter has the following

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 = 6
-
X
fs (1 +G X ) '

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 + G)' may be re-
garded 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 fo/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 fo/fs = 15, the maximum gain
given by Eq. (8-5-30) is 7.3 dB.
328 Avalanche Transit-Time Devices 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

where Td = diode temperature in degrees Kelvin

To = 300°K is the ambient temperature in degrees Kelvin
yQ = figure of merit for the nonlinear capacitor
In a typical microwave diode yQ could be equal to 10. If f,/f, = 10 and Td =
300"K, the minimum noise figure is 0.90 dB, as calculated by using Eq. (8-5-31).
Bandwidth. The bandwidth of a parametric up-converter is related to the
gain-degradation factor of the merit figure and the ratio of the signal frequency to
the output frequency. The bandwidth equation [21] is given by
r

Iff,/& = 10 and y = 0.2, the bandwidth (BW) is equal to 1.264.

Example 8-5-1: Up-Converter Parametric Amplifier

An up-converter parametric amplifier has the following parameters:

Ratio of output frequency over signal frequency: f , / f , = 25

Figure of merit: yQ = 10
Factor of merit figure: y = 0.4
Diode temperature: Td = 3500K

Calculate: (a) the power gain in decibels; (b) the noise figure in decibels; (c) the band-
width.
Solution
a. From Eq. (8-5-30) the up-converter power gain is
X 100/25
Power gain = f-o = 25 x
f s (1 +6 ) ~(1 + 41 + 100/25)2

b. From Eq. (8-5-31) the noise figure is

c. From Eq. (8-5-32) the bandwidth is

7
Sec. 8.5 Parametric Devices 329

Parametric down-converter. If a mode of down conversion for a paramet-

ric amplifier is desirable, the signal frequency 5 must be equal to the sum of the
pump frequency f, and the output frequency f,. 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 [211

Gain = f,
- X
fo (1 +G ) 2

Negative-resistance parametric amplifier. If a significant portion of

power flows only at the signal frequency5 , the pump frequencyf,, and the idler fre-
quency$, a regenerative condition with the possibility of oscillation at both the sig-
nal frequency and the idler frequency will occur. The idler frequency is defined as
the difference between the pump frequency and the signal frequency, $ = f, - fs .
When the mode operates below the oscillation threshold, the device behaves as a bi-
lateral negative-resistance parametric amplifier.

Power Gain. The output power is taken from the resistance Ri at a frequency
$ , and the conversion gain from f, to [21] is given by

where f, = signal frequency

f, = pump frequency
= f, - f, is the idler frequency
R, = output resistance of the signal generator
R, = output resistance of the idler generator
Rrs = total series resistance at fs
Rn = total series resistance at$
a = R/Rr,
R = y2/(wSwiC 2RTi)is the equivalent negative resistance

Noise Figure. The optimum noise figure of a negative-resistance parametric

amplifier [21] is expressed as

where y Q = figure of merit for the nonlinear capacitor

To = 300°K is the ambient temperature in degrees Kelvin
Td = diode temperature in degrees Kelvin
It is interesting to note that the noise figure given by Eq. (8-5-35) is identical to that
for the parametric up-converter in Eq. (8-5-31).
330 Avalanche Transit-Time Devices Chap. 8

Bandwidth. The maximum gain bandwidth of a negative-resistance paramet-

ric amplifier [21] is given by
I

Bw = -YJ&
2 f, gain
If gain = 20 dB, j = 4 5 , and y = 0.30, the maximum possible bandwidth for
single-tuned circuits is about 0.03.

Degenerate parametric amplifier. The degenerate parametric amplifer or

oscillator is defined as a negative-resistance amplifier with the signal frequency equal
to the idler frequency. Since the idler frequency J is the difference between the
pump frequency f, and the signal frequency f, , the signal frequency is just one-half
the pump frequency.

Power Gain and Bandwidth. The power gain and bandwidth characteristics
of a degenerate parametric amplifier are exactly the same as for the parametric up-
converter. With f, = $ and &, = 2 5 , 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,

where Td = average diode temperature in degrees Kelvin

To = 300°K is the ambient noise temperature in degrees Kelvin
Rd = diode series resistance in ohms
R, = external output resistance of the signal generator in ohms
It can be seen that the noise figure for double-sideband operation is 3 dB less than
that for single-sideband operation.

8-5-4 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 paramet-
ric 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 negative-
resistance parametric amplifier:

1. A positive input impedance

2. Unconditionally stable and unilateral
3. Power gain independent of changes in its source impedance
4. No circulator required
5. A typical bandwidth on the order of 5%

At higher frequencies where the up-converter is no longer practical, the negative-

resistance 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 op-
timum 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 nega-
tive-resistance parametric amplifier may be the better choice, since the frequency re-
quired 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 ten-
dency in microwave engineering to replace it with the GaAs metal-semiconductor
field-effect transistor (MESFET) amplifier in airborne radar systems.

REFERENCES

[I] 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 System Tech. J., 50, 1695- 1699, May-June 1971.
[3] LEE,C. A , , ET AL.,The Read diode, an avalanche, transit-time, negative-resistance 0s-
cillator. Appl. Phys. Letters, 6, 89 (1965).
[4] JOHNSTON, R. L., B. C. DELOACH, and G . B. COHEN, A silicon diode microwave oscilla-
tor. 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., ET AL.,State of the art of solid-state and tube transmitters. Microwave
J., 26, No. 10, 46-48, October 1983.
[7] PRAGER, H. J., EI.AL.,High-power, high-efficiency silicon avalanche diodes at ultra high
frequencies. Proc. IEEE (Letters), 55, 586-587, April 1967.
[8] CLORFEINE, A . S., ET 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 Avalanche Transit-Time Devices Chap. 8

[lo] LIU,S. G., and J. J. RISKA,Fabrication and performance of kilowatt L-band avalanche
diodes. RCA Rev., 31, 3, March 1970.

[I 11 KOSTICHACK, D. F., UHF avalanche diode oscillator providing 400 watts peak power and
75 percent efficiency. Proc. IEEE (Letters), 58, 1282-1283, August 1970.

[I21 WILSON, W. E., Pulsed LSA and TRAPATT sources for microwave systems. Microwave
J . , 14, No. 8, 33-41, August 1971.

[I31 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.

[I41 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.

[I51 LORDRAYLEIGH, and J. W. STRUTT, On the crispations of fluid resting upon a vibrating
support. Phil. Mag., 16, 50-53, July 1883.

[I61 VANDERZIEL,A., On the mixing properties of nonlinear capacitances. J. Appl. Phys.,

19, 999- 1006, November 1948.

[I71 LANDON,V. D., The use of ferrite cored coils as converters, amplifiers, and oscillators.
RCA Rev., 10, 387-396, September 1949.

[I81 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, 904-9 13, July 1956.

[2 I] BLACKWELL, L. A., and K . L. KOTZEBUE, Semiconductor-Diode Parametric Amplifiers,

41, 42, 45, 53, 57, 62, 70. Prentice-Hall, Inc., Englewood Cliffs, N.J., 1961.

SUGGESTED READINGS

CHANG, K . K. N., Parametric and Tunnel Diodes. Prentice-Hall, Inc., Englewood Cliffs,
N.J., 1964.
DELOACH, B. C., and D. L. SCHARFEITER, Device physics of TRAPATT oscillators. IEEE
Trans. on Electron Devices, ED-17, No. I, 9-21, January 1970.
L. F., Gallium Arsenide Microwave Bulk and Transit-Time Devices. Artech House,
EASTMAN,
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, 1140-1 171, August 1971.

PROBLEMS

Avalanche Transit-Time Devices

8-1. Spell out the following abbreviated terms: IMPATT, TRAPATT, and BARITT.
8-2. Derive Eq. (8-2-9).
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 p m Determine: (a) the drift time of the car-
riers 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 pulsed-
operating 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. An M-Si-M BARITT diode has the following parameters:

Relative dielectric constant of Si: 6, = 11.8

Donor concentration: N = 3 x lo2' m-3
Si length: L = 6.2 p m

Calculate: (a) the breakdown voltage and (b) the breakdown electric field.

Parametric Devices
8-7. a. Describe the advantages and disadvantages of the parametric devices.
b. Describe the applications of the parametric amplifiers.
8-8. The figure of merit for a diode nonlinear capacitor in an up-converter parametric am-
plifier is 8, and the ratio of the output (or idler) frequency f, over the signal frequency
f,is 8. The diode temperature is 300°K.
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 Avalanche Transit-Time Devices Chap. 8

8-9. A negative-resistance parametric amplifier has the following parameters:

a. Calculate the power gain in decibels.

b. Compute the noise figure F in decibels.
c. Determine the bandwidth (BW).
Chapter 9

Microwave Linear-Beam
I Tubes (0 Type)

9-0 INTRODUCTION

We turn now to a quantitative and qualitative analysis of several conventional vac-

uum 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 ( 0 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 am-
plifier (FWA), and the backward-wave amplifier and oscillator (BWA and BWO) are
also 0 -type tubes, but they have nonresonant periodic structures for electron inter-
actions. 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 [ l ] in 1935
and the Varian brothers' klystron amplifier [I] in 1939. The work was advanced by
the space-charge-wave propagation theory of Hahn and Ramo [I] in 1939 and con-
tinued 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 linear-
beam 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 impor-
tance. They are the extended interaction klystron [3] and the Twystron hybrid am-
plifier [4].
335
336 Microwave Linear-Beam Tubes (0Type) Chap. 9

TABLE 9-0.1 LINEAR BEAM TUBES (0TYPE)

Linear-beam tubes (0-type)

Cavity Slow-wave structure

I
Resonant +
I-
Forward-wave Backward-wave

Klystron H e l ~ xTWT
EL BWA, BWO

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 dc 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 deceler-
ated 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
70dB, 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 ex-
pected 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 perfor-
mances 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)

15 (R) (S)
PPM
/,-- COUPLED
CAVITY TWT

AIR COOLED PPM-, 12 (H)

HELIX TWT

\ 16 EIA (V)

BWO

lo
GATA
FROM PUBLISHED RESULTS: EIO
HUGHES (H)
RAYTHEON (R)
SIEMENS (S)
THOMPSON CSF (C)
USSR (U)
VARIAN (V) HDL \ Figure 9-0-1 CW power state-of-the-
WATKINS JOHNSON (W) O R O T R ~ ~ \ art performance for various microwave
1 1 1 1 1 1 111 I 1 , 1 1 1 1 1 \ a A L tubes. (After C. Hieslmair 151, reprinted
1 10 100
FREOUENCY (GHz) by permission of Microwave Journal.)

o CFA
TWT
a KLYSTRON
m GYROTRON

GYROTWT

- SIEMENS (S)

- THOMPSON CSF (C)

USSR (U)
VARIAN ( V ) Figure 9-0-2 Peak power s t a t e - o f - t h e -
art performance for various microwave
1 10 tubes.(After C. Hieslmair 151, reprinted
FREOUENCY IGHz) by permission of Microwave Journal.)
337
338 Microwave Linear-Beam Tubes (0Type) 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 achieve-
ments 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.

9-1 CONVENTIONAL VACUUM TRIODES, TETRODES, AND

PENTODES

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 ef-
fects are analyzed in detail in the following sections.

-
9-1 1 Lead-Inductance and
Interelectrode-Capacitance Effects

At frequencies above 1 GHz conventional vacuum tubes are impaired by parasitic-

circuit 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 bet-
ter understanding of these effects, the triode ciruit shown in Fig. 9-1-1 should be
studied carefully.

Figure 9-1-1 Triode circuit (a) and its equivalent (b).

Figure 9-1-l(b) shows the equivalent circuit of a triode circuit under the as-
sumption that the interelectrode capacitances and cathode inductance are the only
parasitic elements. Since C,,, 4 CRkand wLk 4 l/(wCpk), the input voltage Vi, can
be written as
VIn= V, + V k = V, + jwLkgmVg (9-1-1)
Sec. 9.1 Conventional Vacuum, Triodes, Tetrodes, and Pentodes 339

and the input current as

Iln = jwCRk VR
Substitution of Eq. (9- 1-2) in Eq. (9- 1- 1 ) yields

The input admittance of the tube is approximately

Iin=
y.~n = -V,, 1 +jwC$
jwLkgm = + jwCRk
w 2 ~ k ~ R k g m

in which wLkg, 4 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 transconduc-
tance g, is generally much less than one millimho.
The input impedance at very high frequencies is given by

The real part of the impedance is inversely proportional to the square of the fre-
quency, 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 ex-
pressed by
Yin = w L k CRkgm+ jw (CRk + Cgs) (9- 1-6)
where CRsis the capacitance between the grid and screen, and its input impedance is
given bv

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.

9-1-2 Transit-Angle Effects

Another limitation in the application of conventional tubes at microwave frequencies

is the electron transit angle between electrodes. The electron transit angle is defined
as
wd
6 -wTg = - (9-1-8)
00

where r R = d / v Ois the transit time across the gap

d = separation between cathode and grid
vo = 0.593 X 1 0 6 Gis the velocity of the electron
Vo = dc voltage
340 Microwave Linear-Beam Tubes (0Type) 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 poten-
tial 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 transit-
angle 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 compari-
son with the phase of the grid voltage. This means that the transadmittance is a large
real quantity, which is the usual transconductance g, . 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 de-
creased.
From the preceding analysis it is clear that the transit-angle effect can be mini-
mized by first accelerating the electron beam with a very high dc voltage and then
velocity-modulating it. This is indeed the principal operation of such microwave
tubes as klystrons and magnetrons.

9-1-3 Gain-BandwidthProduct 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 r,, wLk . The load voltage is given by

where G = l/rp + 1/R

rp = plate resistance
R = load resistance
L, C = tuning elements
The resonant frequency is expressed by

Figure 9-1-2 Output-tuned circuit of a

- pentode.
Sec. 9.2 Klystrons

and the maximum voltage gain Amaxat resonance by

Since the bandwidth is measured at the half-power point, the denominator of Eq.
(9- 1-9) must be related by

The roots of this quadratic equation are given by

Then the bandwidth can be expressed by

Hence the gain-bandwidth product of the circuit of Fig. 9-1-2 is

It is important to note that the gain-bandwidth product is independent of frequency.

For a given tube, a higher gain can be achieved only at the expense of a narrower
bandwidth. This restriction is applicable to a resonant circuit only. In microwave
devices either reentrant cavities or slow-wave structures are used to obtain a possible
overall high gain over a broad bandwidth.

9-2 KLYSTRONS

The two-cavity klystron is a widely used microwave amplifier operated by the princi-
ples of velocity and current modulation. All electrons injected from the cathode ar-
rive 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 ve-
locity; 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 un-
dergo 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 oc-
cur approximately midway between the second cavity grids during its retarding
342 Microwave Linear-Beam Tubes (0Type) Chap. 9

phase; thus the kinetic energy is transferred from the electrons to the field of the sec-
ond 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 am-
plifier are as follows:

1. Efficiency: about 40%.

2. Power output: average power (CW power) is up to 500 kW and pulsed power is
up to 30 MW at 10 GHz.
3. Power gain: about 30 dB.

Figure 9-2-1 shows the present state of the art for U.S. high-power klystrons.
Figure 9-2-2 shows the schematic diagram of a two-cavity klystron amplifier. The
cavity close to the cathode is known as the buncher cavity or input cavity, which ve-
locity-modulates the electron beam. The other cavity is called the catcher cavity or
output cavity; it catches energy from the bunched electron beam. The beam then
passes through the catcher cavity and is terminated at the collector. The quantitative
analysis of a two-cavity klystron can be described in four parts under the following
assumptions:

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 dc accelerating voltage.

9-2-1 Reentrant Cavities

At a frequency well below the microwave range, the cavity resonator can be repre-
sented by a lumped-constant resonant circuit. When the operating frequency is in-
creased 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 shelf-
shielding 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

&=-
1
2-
J' E
b
en-
a
ohms
M

10
8

z
< 3
b
I-
0 2
a
W
P
2
2 1
08

0.6

05
04L

03

U
0.2
0.05 1.1 0.2 0.4 0.6 1 2 4 8 10 20 30
FREOUENCY. GH1

Figure 9-2-1 State of the art for U.S. high-power klystrons.

-L-
Buncher cavity Catcher cavity
RF output

Cathode Drift space Collector

R
- + o d L+d L+2d
Distance scale
& -- Time scale t -
to t l '2 3

Figure 9-2-2 Two-cavity klystron amplifier

Figure 9-2-3 Reentrant cavities. (a) Coaxial cavity. (b) Radial cavity. (c) finable cavity.
(d) Toroidal cavity. (c) Butterfly cavity.
Microwave Linear-Beam Tubes (0Type) Chap. 9

(a) Coaxial cavity (b) Cross-section of

coaxial cavity
Figure 9-2-4 Coaxial cavity and its equivalent.

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
Z,, = jZo tan (P &) (9-2-2)
where & is the length of the coaxial line.
Substitution of Eq. (9-2-1) in (9-2-2)results in
r

The inductance of the cavity is given by

r

and the capacitance of the gap by

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 = l/(wC,).
Thus

where v = 1 1 6 is the phase velocity in any medium.

The solution to this equation gives the resonant frequency of a coaxial cavity.
Since Eq. (9-2-6) contains the tangent function, it has an infinite number of solu-
tions with larger values of frequency. Therefore this type of reentrant cavity can
support an infinite number of resonant frequencies or modes of oscillation. It can be
shown that a shorted coaxial-line cavity stores more magnetic energy than electric
energy. The balance of the electric stored energy appears in the gap, for at reso-
nance the magnetic and electric stored energies are equal.
Sec. 9.2 Klystrons 345

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

The resonant frequency [ l ] is given by

where c = 3 X lo8 mls is the velocity of light in a vacuum.

9-2-2 Velocity-Modulation Process

When electrons are first accelerated by the high dc voltage Vo before entering the
buncher grids, their velocity is uniform:

vo = @ = 0.593 X i o 6 f i m/s (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
V, = V, sin (ot) (9-2- 11)
where V1 is the amplitude of the signal and Vl 4 VO is assumed.
In order to find the modulated velocity in the buncher cavity in terms of either
the entering time to or the exiting time tl and the gap transit angle 8, as shown in
Fig. 9-2-2 it is necessary to determine the average microwave voltage in the buncher
gap as indicated in Fig. 9-2-6.
Since VI 4 VO, the average transit time through the buncher gap distance d is
Microwave Linear-Beam Tubes (0Type) Chap. 9

I
v, V, = V, sin wt t o tl

*t
Figure 9-2-6 Signal voltage in the
buncher gap.

The average gap transit angle can be expressed as

The average microwave voltage in the buncher gap can be found in the following
way:

11
V1
(V,) = V, sin(wt)dt = - -[cos (or,) - cos(wto)]
to
wi-

="[
wr
cos(ot0) -
(
cos wo +-
31
Let

and

Then using the trigonometric identity that cos (A - B) - cos (A + B) = 2 sin A

sin B, Eq. (9-2-14) becomes

(V,) = VI
sin (8,/2)
sin (wto + 2) (9-2-15)

It is defined as
sin [od/(200)] -
- sin (8,/2)
Pi' wd/(2v,,) 8,/2

Note that pi is known as the beam-coupling coefjcient of the input cavity gap (see
Fig. 9-2-7).
It can be seen that increasing the gap transit angle 0, decreases the coupling
between the electron beam and the buncher cavity; that is, the velocity modulation
Sec. 9.2 Klystrons

'a

Figure 9-2-7 Beam-coupling coefficient versus gap transit angle

of the beam for a given microwave signal is decreased. Immediately after velocity
modulation, the exit velocity from the buncher gap is given by

where the factor piV,/Vo is called the depth of velocity modulation.

Using binomial expansion under the assumption of

Eq. (9-2- 17) becomes

Equation (9-2-19) is the equation of velocity modulation. Alternatively, the equation

of velocity modulation can be given by
348 Microwave Linear-Beam Tubes (0Type) Chap. 9

9-2-3 Bunching Process

Once the electrons leave the buncher cavity, they drift with a velocity given by Eq.
(9-2-19) or (9-2-20)along in the field-free space between the two cavities. The ef-
fect of velocity modulation produces bunching of the electron beam-or current
modulation. The electrons that pass the buncher at Vs = 0 travel through with un-
changed velocity vo and become the bunching center. Those electrons that pass the
buncher cavity during the positive half cycles of the microwave input voltage V,
travel faster than the electrons that passed the gap when Vs = 0 . Those electrons that
pass the buncher cavity during the negative half cycles of the voltage V, travel slower
than the electrons that passed the gap when V, = 0 . At a distance of AL along the
beam from the buncher cavity, the beam electrons have drifted into dense clusters.
Figure 9-2-8 shows the trajectories of minimum, zero, and maximum electron ac-
celeration.

The distance from the buncher grid to the location of dense electron bunching
for the electron at tb is
AL = vo(td - tb) (9-2-21)
Similarly, the distances for the electrons at ta and t, are

AL = vmin(td- ta) = Vmin

From Eq. (9-2-19)or (9-2-20)the minimum and maximum velocities are

Sec. 9.2 Klystrons

Substitution of Eqs. (9-2-24) and (9-2-25) in Eqs. (9-2-22) and (9-2-23), respec-
tively, yields the distance

and

The necessary condition for those electrons at t, , tb , and t, to meet at the same dis-
tance AL is

and

Consequently,

and

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
bv
T = t 2 - t I = - -L - T,[l - @ sin ( w t l -
v (tl) 2Vo
+)I (9-2-32)

where the binomial expansion of (1 + x)-' for I x 1 G 1 has been replaced and
To = L / v o is the dc transit time. In terms of radians the preceding expression can be
written
wT = wtz - wtl = 60 - X sin (9-2-33)
Microwave Linear-Beam Tubes (0Type) Chap. 9

Acceleration Output-gap
voltage

Figure 9-2-9 Applegate diagram.

where oL
90 = - = 2rN (9-2-34)
Do

is the dc transit angle between cavities, N is the number of electron transit cycles in
the drift space, and
x = - pie oVI
2vo
is defined as the bunching parameter of a klystron.
At the buncher gap a charge dQo passing through at a time interval dfois given
by
dQo = Iodto (9-2-36)
where 10 is the dc current. From the principle of conservation of charges this same
amount of charge dQo also passes the catcher at a later time interval d l 2 . Hence

where the absolute value signs are necessary because a negative value of the time ra-
tio would indicate a negative current. Current i2 is the current at the catcher gap.
Rewriting Eq. (9-2-32) in terms of Eq. (9-2-19)yields

[ ;{
t 2 = t 0 + 7 + T o 1--
sin (oro+ +)]
Alternatively,

ot2 - (90 + 2) = (oh + 2) - X sin (oro+ 2) (9-2-39)

+
where (oto 9,/2) is the buncher cavity departure angle and wt2 - (90 + Og/2) is
the catcher cavity arrival angle. Figure 9-2-10 shows the curves for the catcher cav-
Sec. 9.2 Klystrons

-* -R
- 0 II
- n
2 2
9
wt,, +J
2 Figure 9-2-10 Catcher arrival angle
Buncher cavity departure angle versus buncher departure angle.

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

The current arriving at the catcher cavity is then given as

In terms of t2 the current is

In Eq. (9-2-42) the relationsip of t 2 = to + T + TOis used-namely, ot2 = oto +

WT + WTO= W ~ O+ Og + 00. Figure 9-2-1 1 shows curves of the beam current i2(t2)
as a function of the catcher arrival angle in terms of the bunching parameter X.
Microwave Linear-Beam Tubes (0Type) Chap. 9

Catcher cavity arrival angler, wtr - 'o + $)

Figure 9-2-11 Beam current iz versus catcher cavity arrival angle.

The beam current at the catcher cavity is a periodic waveform of period 27r/w
about dc current. Therefore the current i2 can be expanded in a Fourier series and so
rn

i2 = a. + 2 [a, cos (not4 + b, sin ( m t z ) ]

n= l
(9-2-43)

where n is an integer, excluding zero. The series coefficients a o , a , , and b, in Eq.

(9-2-43) are given by the integrals

Substitution of Eqs. (9-2-37)and (9-2-39)in Eq. (9-2-44)yields

a, = - [ lo cos + nf& + n80) + nX sin ( w o + $)] d(uto) (9-2-46)

b, = I in t o + dg+ n o ) +n sin ( t o + )I t o (9-2-47)

By using the trigonometric functions

cos ( A k B) = cos A cos B T sin A sin B
and
sin ( A * B) = sin A cos B + cos A sin B
Sec. 9.2 Klystrons 353

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

+ 2[J4(nx)cos ,(,to +): ]

and

sin [nX sin ( w , + :)] = 2[Jl(nX)sin (wto + $)]

+ 2[J$nX) sin 3(wt0 + 2)]
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
a, = 210Jn(nX)cos (ne, + nee) (9-2-50)
b, = 210Jn(nX)sin (4+ d o ) (9-2-51)

where Jn(nX)is the nth-order Bessel function of the first kind (see Fig. 9-2-12).

Argument of J,(nX)

Figure 9-2-12 Bessel functions J,(nX).

354 Microwave Linear-Beam Tubes (0Type) Chap. 9

Substitution of Eqs. (9-2-45), (9-2-50), and (9-2-51) in Eq. (9-2-43) yields

the beam current i2 as
m

i2 = 10 + n= 1
210 J,(nX) cos [nw (t2 - 7 - TO)] (9-2-52)

The fundamental component of the beam current at the catcher cavity has a magni-
tude
If = 210 J,(X) (9-2-53)
This fundamental component If has its maximum amplitude at

The optimum distance L at which the maximum fundamental component of current

occurs is computed from Eqs. (9-2-34), (9-2-35), and (9-2-54) as

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 approxima-
tions made in deriving Eq. (9-2-31) 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.

9-2-4 Output Power and Beam Loading

The maximum bunching should occur approximately midway between the catcher
grids. The phase of the catcher gap voltage must be maintained in such a way that
the bunched electrons, as they pass through the grids, encounter a retarding phase.
When the bunched electron beam passes through the retarding phase, its kinetic
energy is transferred to the field of the catcher cavity. When the electrons emerge
from the catcher grids, they have reduced velocity and are finally collected by the
collector.

The induced current in the catcher cavity. Since the current induced by
the electron beam in the walls of the catcher cavity is directly proportional to the
amplitude of the microwave input voltage V, , the fundamental component of the in-
duced microwave current in the catcher is given by

where Po is the beam coupling coefficient of the catcher gap. If the buncher and
catcher cavities are identical, then P, = P o . The fundamental component of current
induced in the catcher cavity then has a magnitude

Figure 9-2-13 shows an output equivalent circuit in which Rshorepresents the wall
resistance of catcher cavity, Re the beam loading resistance, RL the external load re-
sistance, and RSh the effective shunt resistance.
-
Sec. 9.2 Klystrons

POI,

Figure 9-2-13 Output equivalent cir-

cuit.

The output power delivered to the catcher cavity and the load is given as

PO", =-
(PO1 2 > 2
Rsh = -
Po 1 2 V2
2 2
where Rshis the total equivalent shunt resistance of the catcher circuit, including the
load, and Vz is the fundamental component of the catcher gap voltage.

Efficiency of klystron. The electronic efficiency of the klystron amplifier is

defined as the ratio of the output power to the input power:
PO", Po12 V2
Efficiency = - = -
Pin 210VO
in which the power losses to the beam loading and cavity walls are included.
If the coupling is perfect, Po = 1, the maximum beam current approaches
IZmax = 2I0(0.582), and the voltage V2 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 8, , Fig. 9-2-14 shows the maximum efficiency of a klystron as a func-
tion of catcher transit angle.

Figure 9-2-14 Maximum efficiency of

Catcher gap transit angle Or klystron versus transit angle.
356 Microwave Linear-Beam Tubes (0Type) Chap. 9

Mutual conductance of a klystron amplifier. The equivalent mutual con-

ductance of the klystron amplifier can be defined as the ratio of the induced output
current to input voltage. That is

From Eq. (9-2-35) the input voltage Vl can be expressed in terms of the bunching
parameter X as

In Eq. (9-2-61) it is assumed that PO = pi. Substitution of Eq. (9-2-61) in Eq.

(9-2-60) yields the normalized mutual conductance as

where Go = Io/Vo is the dc beam conductance. The mutual conductance is not a

constant but decreases as the bunching parameter X increases. Figure 9-2- 15 shows
the curves of normalized transconductance as a function of X.

P O ~ Vl
Bunching parameter X = ---
O Figure 9-2-15 Normalized trans-
2v~ conductance versus bunching parameter.
Sec. 9.2 Klystrons 357

It can be seen from the curves that, for a small signal, the normalized
transconductance is maximum. That is,

For a maximum output at X = 1.841, the normalized mutual conductance is

The voltage gain of a klystron amplifier is defined as

where Ro = Vo/lo is the dc beam resistance. Substitution of Eqs. (9-2-57) and

(9-2-61) in Eq. (9-2-65) results in
A v = Gm Rsh (9-2-66)

Power required to bunch the electron beam. As described earlier, the

bunching action takes place in the buncher cavity. When the buncher gap transit an-
gle is small, the average energy of electrons leaving the buncher cavity over a cycle
is nearly equal to the energy with which they enter. When the buncher gap transit
angle is large, however, the electrons that leave the buncher gap have greater aver-
age energy than when they enter. The difference between the average exit energy
and the entrance energy must be supplied by the buncher cavity to bunch the elec-
tron beam. It is difficult to calculate the power required to produce the bunching ac-
tion. Feenberg did some extensive work on beam loading [6]. The ratio of the power
required to produce bunching action over the dc power required to perform the elec-
tron beam is given by Feenberg as

where

The dc power is
Po = V;Go
where Go = lo/Vo is the equivalent electron beam conductance. The power given by
the buncher cavity to produce beam bunching is
358 Microwave Linear-Beam Tubes (0 Type) Chap. 9

where GB is the equivalent bunching conductance. Substitution of Eqs. (9-2-68) and

(9-2-69) in Eq. (9-2-67) yields the normalized electronic conductance as

Figure 9-2-16 shows the normalized electronic conductance as a function of the

buncher gap transit angle. It can be seen that there is a critical buncher gap transit
angle for a minimum equivalent bunching resistance. When the transit angle 8, is
3.5 rad, the equivalent bunching resistance is about five times the electron beam re-
sistance. The power delivered by the electron beam to the catcher cavity can be ex-
pressed as

As a result, the effective impedance of the catcher cavity is

Finally, the loaded quality factor of the catcher cavity circuit at the resonant fre-
quency can be written

where QL = loaded quality factor of the whole catcher circuit

Qo = quality factor of the catcher walls
QB = quality factor of the beam loading
Qe,, = quality factor of the external load

Bunoher gap transit angle 0,. radians

Figure 9-2-16 Normalized electronic conductance versus buncher gap transit

angle.
Sec. 9.2 Klystrons 359

Example 9-2-1: Klystron Amplifier

A two-cavity klystron amplifier has the following parameters:
Vo = 1000 V Ro = 40 k f l
lo= 25 m.4 f = 3GHz

Gap spacing in either cavity: d=lmm

Spacing between the two cavities: L =4cm
Effective shunt impedance, excluding beam loading: Rsh = 30 kQ

a. Find the input gap voltage to give maximum voltage V2.

b. Find the voltage gain, neglecting the beam loading in the output cavity.
c. Find the efficiency of the amplifier, neglecting beam loading.
d. Calculate the beam loading conductance and show that neglecting it was justified
in the preceding calculations.
Solution
a. For maximum V2, JI(X) must be maximum. This means JI(X) = 0.582 at
X = 1.841. The electron velocity just leaving the cathode is
uo = (0.593 x 1 0 6 ) f i = (0.593 x lo6)- = 1.88 x lo7 mls
The gap transit angle is
d lo-3
o, = w- = 2 ~ ( 3x lo9) = I rad
vo 1.88 x lo7
The beam-coupling coefficient is

The dc transit angle between the cavities is

4 x 10-2
o0 = wTo = w-L = 2~ (3 x lo9) = 40 rad
uo 1.88 x lo7
The maximum input voltage Vl is then given by

b. The voltage gain is found as

c. The efficiency can be found as follows:

1, = 210JI(X) = 2 x 25 X x 0.582 = 29.1 X A
V2 = POf2Rsh= (0.952)(29.1 X 10-3)(30 X lo3) = 831 V
/30I2 V2 - (0.952N29.1 X 10-3)(831) = 46.29b
Efficiency = --
210 vo 2(25 x 10-3)(103)
Microwave Linear-Beam Tubes (0Type) Chap. 9

d. Calculate the beam loading conductance (refer to Fig. 9-2-13). The beam load-
ing conductance Ge is

- po cos - = [(0.952)2 - (0.952) cos (28.6")]

2
= 8.8 x lo-' mho
Then the beam loading resistance RB is

In comparison with R L and Rshoor the effective shunt resistance R s h , the beam
loading resistance is like an open circuit and thus can be neglected in the preced-
ing calculations.

9-2-5 State of the Art

Extended Interaction. The most common form of extended interaction has

been attained recently by coupling two or more adjacent klystron cavities. Figure
9-2-17 shows schematically a five-section extended-interaction cavity as compared
to a single-gap klystron cavity.

Figure 9-2-17 Comparison of a five-

@ @ A
(b)
gap extended interaction'cavity with a
single-gap klystron cavity. (a) Five-gap
coupled cavity resonator. (b)Single-gap
klystron cavity. (After A. Staprans et al.
[7]; reprinted by permission of IEEE.)

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-884D
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, Cali-
fornia. A cutaway view of the tube is shown in Fig. 9-2-18. The operating character-
istics 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 9-2-1 VA-884D OPERATING CHARACTERISTICS

Frequency 5.9-6.45 GHz

Power output 14 kW
Gain 52 dB
Efficiency 36%
Electronic bandwidth (1 dB) 75 MHz
Beam voltage 16.5 kV
Beam current 2.4 A

Figure 9-2-18 Cuts\ -MW

S-band permanent-ma sed
klystro!n. (Courtesy oj Linear
Accelerator Center .)

TABLE 9.2-2 OPERATING CHARACTERISTICS OF THE

STANFORD LINEAR ACCELERATOR CENTER HIGH-POWER
KLYSTRON [9]

Frequency 2.856 GHz

RF pulse width 2.5 P S
Pulse repetition rate 60 to 360 PPS
Peak power output 24 MW
Beam voltage 250 kV
Beam current 250 A
Gain 50 to 55 dB
Efficiency about 36 %
Weight of permanent (focusing) magnet 363 kg
Electronic bandwidth (1 dB) 20 MHz
Microwave Linear-Beam Tubes (0Type) Chap. 9

Figure 9-2-19 Photograph of Varian

VKC-8269A CW super-power klystron
amplifier. (Courtesyof Varian Associ-
ates, Inc.)

gain is 56 dB and efficiency is 50 percent. The beam voltage is 62 V (dc) and the
beam current is 16.50 A (dc).

Long-life improvement. A new long-life klystron amplifier tube with a de-

sign life in excess of ten years, three times the current design life, has been devel-
oped 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 impreg-
nated 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 tem-
perature necessary for electrons to be emitted. This temperature reduction cuts evap-
oration of barium 10-fold and extends the life of the cathode.

9-3 MULTlCAVlTY KLYSTRON AMPLIFIERS

The typical power gain of a two-cavity klystron amplifier is about 30 dB. In order to
achieve higher overall gain, one way is to connect several two-cavity tubes in cas-
cade, feeding the output of each of the tubes to the input of the following one. Be-
Sec. 9.3 Multicavity Klystron Amplifiers

RF Output

~agnet
coils

Figure 9-3-1 Schematic diagram of a four-cavity klystron amplifier. (Courtesy of

Varian Associates, Inc.)

sides using the multistage techniques, the tube manufacturers have designed and pro-
duced 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 re-
quirement of X being 1.841 and an increasing velocity modulating RF voltage as the
beam progresses through the various cavities. Keeping the intercavity distance con-
stant, an increasing beam voltage Vocould be used in the subsequent cavities. Figure
9-3-2 shows the photograph of a four-cavity klystron amplifier.

9-3-1 Beam-Current Density

When the two-cavity klystron amplifier was discussed in Section 9-2, it was as-
sumed that the space-charge effect was negligible, and it was not considered because
of the assumed small density of electrons in the beam for low-power amplifier. How-
ever, when high-power klystron tubes are analyzed the electron density of the beam
is large and the forces of mutual repulsion of the electrons must be considered.
When the elecrons perturbate in the electron beam, the electron density consists of a
dc part plus an RF perturbation caused by the electron bunches. The space-charge
forces within electron bunches vary with the size and shape of an electron beam. In
Microwave Linear-Beam Tubes (0Type) Chap. 9

Figure 9-3-2 Photograph of a four-

cavity klystron amplifier VA-890H.
(Courtesy of Varian Associates, Inc.)

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 com-
parison to the infinite beam. With reduced axial space-charge force, the plasma fre-
quency is reduced and the plasma wavelength is increased.
Mathematically, let the charge-density and velocity perturbations be simple si-
nusoidal variations in both time and position. They are
Charge density: = B cos @,z) cos (w,t + 8) (9-3-1)
Velocity perturbation: '3'" = -C sin @,z) sin (w,t + 8) (9-3-2)
where B = constant of charge-density perturbation
C = constant of velocity perturbation
w
p = - is the dc phase constant of the electron beam
Yo
w, = Rw, is the pertubation frequency or reduced plasma frequency
R = w , / o , is the space-charge reduction factor and varies from 0 to 1

the plasma frequency and is a function of the electron-beam

density
8 = phase angle of oscillation
Sec. 9.3 Multicavity Klystron Amplifiers 365

The electron plasma frequency is the frequency at which the electrons will os-
cillate 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 w, . This lower plasma frequency is called the reduced plasmu fre-
quency and is designated w, . 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 [lo].

Figure 9-3-3 Plasma-frequency reduction factor R [I01 (Reprinted by permission

of Pergamon Press.)

As an example of the effect of the reduced space-charge forces, if per =

w r / V 0 = 0.85 and n = 2, the reduced factor is R = 0.5. Therefore, o, = 2w,,
which means that in a klystron the cavities would be placed twice as far apart as
would be indicated by infinite-beam calculation. The total charge density and elec-
tron velocity are given by
ptot = -po +P (9-3-3)
I r , = qr, + qr (9-3-4)
where po = dc electron charge density
p = instantaneous RF charge density
TO= dc electron velocity
7r = instantaneous electron velocity perturbation
366 Microwave Linear-Beam Tubes (0Type) Chap. 9

The total electron beam-current density can be written as

Jtot = - JO+ J (9-3-5)
where JO= dc beam-current density
J = instantaneous RF beam-current perturbation
The instantaneous convection beam-current density at any point in the beam is
expressed as

where J = pVo - poV

Jo = poVo is replaced
p T = very small is ignored
In accordance with the law of conservation of electric charge, the continuity
equation can be written as

where J = p V o - poV is in positive z direction only. From

a'
- -- - w ~sin $.z - w t ) cos (wqt + 6 )
az (9-3-9)
+ pepoCcos (P,z - w t ) sin (w,t + 6)
and
a'
- - = - w ~sin l6.z - w t ) cos (w,t + 6)
at

+ wqB cos (P,z - w t ) sin (wqt + 0)

then we have
w , B = PepoC (9-3- 11)

The beam-current density and the modulated velocity are expressed as [lo]
J = VOB cos (P,z - w t ) cos (w,t + 0)
+ ~ww V sin (Pez - w t ) sin (w,t
, B + 6)
and
pi VI
Q = -To- cos (P,z) sin (P.z - w t )
2Vo
Sec. 9.3 Multicavity Klystron Amplifiers 367

In practical microwave tubes, the ratio of w,/w is much smaller than unity and the
second term in Eq. ( 9 - 3 - 1 2 ) may be neglected in comparison with the first. Then we
obtain
J = 7roB cos @,z - w t ) cos ( o , t 0) + (9-3- 14)

Example 9-3-1: Four-Cavity Klystron

A four-cavity klystron VA-828 has the following parameters:

Beam voltage: Vo = 14.5 kV

Beam current: l o = 1.4A
Operating frequency: f = 10GHz
dc electron charge density: po = C/m3
RF charge density: p = C/m3
Velocity perturbation: Ir = lo5 m/s

Compute:

a. The dc electron velocity

b. The dc phase constant
c. The plasma frequency
d. The reduced plasma frequency for R = 0.4
e. The dc beam current density
f. The instantaneous beam current density
Solution

a. The dc electron velocity is

T o = 0.593 X 1062/14.5 x lo3 = 0.714 x 10' m/s
b. The dc phase constant is

c. The plasma frequency is

d. The reduced plasma frequency for R = 0.4 is

w, = 0.4 X 1.41 X 10' = 0.564 x 10' radls
e. The dc beam current density is
Jo = X 0.714 X 10' = 71.4 A/m2
f. The instantaneous beam current density is
J = x 0.714 X 10' - X lo5 = 0.614 A/m2
368 Microwave Linear-Beam Tubes (0Type) 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

~ ( t ,=) v O [ l + 2Vo
sin (on)]

where Vl = magnitude of the input signal voltage

d
7 = -= tl - to is the transit time
vo
d = gap distance
Since the electrons under the influence of the space-charge forces exhibit sim-
ple harmonic motion, the velocity at a later time t is given by

vt0t= ~ , [ 1+
p, Vl sin (01)cos (WPt - w,,T)
I
where w,, = plasma frequency. The current-density equation may be obtained from
Eq. (9-3-14) as

J = - -1 -pi
JOW V, sin @,z) cos w e z - ~ t )
2 Vow,
W
p,
where =
vo is the plasma phase constant.
Example 9-3-2: Operation of a Four-Cavity Klystron
A four-cavity CW klystron amplifier VA-864 has the following parameters:

Beam voltage: VO= 18 kV

Beam current: lo= 2.25 A
Gap distance: d=lcm
Operating frequency: f = 10GHz
Signal voltage: Vl = 10 V (rms)
Beam coupling coefficient: P o = PI = 1
dc electron beam current density: po = C/m3

Determine:

a. The dc electron velocity

b. The dc electron phase constant
c. The plasma frequency
d. The reduced plasma frequency for R = 0.5
e. The reduced plasma phase constant
f. The transit time across the input gap
g. The electron velocity leaving the input gap
Sec. 9.3 Multicavity Klystron Amplifiers

Solution

a. The dc electron velocity is

"Vo = 0.593 x lo6- = 0.796 x 10' m/s
b. The dc electron phase constant is
p, = 2n x 101°/(0.796 X 10') = 7.89 X 102rad/m
c. The plasma frequency is
w, = [1.759 x 10" x 10-'/(8.854 X 10-'2)]1'2= 1.41 X lo7 md/s
d. The reduced plasma frequency is
w, = 0.5 x 1.41 x lo7 = 0.705 x lo7 radls
e. The reduced plasma phase constant is
p, = 0.705 x 107/(0.796 x 10') = 0.088 md/m
f. The transit time across the gap is
T = 10-2/(0.796 X 10') = 0.1256 ns
g. The electron velocity leaving the input gap is
V(tl) = 0.796 X 108[1 + 1 X 10/(2 X 18 x lo3) sin (27~X 101° X 1.256
x lo-'0)]
= 0.796 X 10' + 2.21 x lo4 m/s

9-3-2 Output Current and Output Power

of Two-Cavity Klystron

If the two cavities of a two-cavity klystron amplifier are assumed to be identical and
are placed at the point where the RF current modulation is a maximum, the magni-
tude of the RF convection current at the output cavity for a two-cavity klystron can
be written from Eq. (9-3-17) as
1 loo
Ii2I =--PilVlI
2 Vow,
where Vl = magnitude of the input signal voltage. Then the magnitudes of the in-
duced current and voltage in the output cavity are equal to
1 loo
1121 = P0li2I = --Pi]V,]
2 Vow,
and
1 loo
I vzI I , -pal
= IZ I R ~=~ -
2 Vow,
vl I R ~ ~ ,
370 Microwave Linear-Beam Tubes (0Type) Chap. 9

where po = pi is the beam coupling coefficient

= total shunt resistance of the output cavity in a two-cavity klystron
amplifier including the external load
The output power delivered to the load in a two-cavity klystron amplifier is
given by

The power gain of a two-cavity klystron amplifier is then expressed by

Pout Pout 1 10,
Power gain = - = =-(-)PdR,.Rsh, (9-3-22)
PI" ( VI I2/Rsh 4 Vowq
where RSh = total shunt resistance of the input cavity. The electronic efficiency of a
two-cavity klystron amplifier is

,,=pin==-
0 " '

10 v
o 4
PO"'
' (")- (I-
vo
Vl Iw)2pQRshl
vow,

Example 9-3-3: Characteristics of Two-Cavity Klystron

A two-cavity klystron has the following parameters:

Beam voltage: VO= 20 kV

Beam current: Io=2A
Operating frequency: f = 8 GHz
Beam coupling coefficient: p, = p o = 1
dc electron beam current density: po = C/m3
Signal voltage: V, = 10 V (rms)
Shunt resistance of the cavity: R,h = 10 kK?
Total shunt resistance including load: R = 30 kK?

Calculate:

a. The plasma frequency

b. The reduced plasma frequency for R = 0.5
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

Solution

a. The plasma frequency is

o, = [1.759 x 10" X 10-6/(8.854 X 10-L2)]'12= 1.41 X 10srad/s
Sec. 9.3 Multicavity Klystron Amplifiers

b. The reduced plasma frequency is

w, = 0.5 x 1.41 X 10' = 0.705 X lo8 radls
w/w, = 27r x 8 X 109/(0.705 x lo8) = 713
c. The induced current in the output cavity is

d. The induced voltage in the output cavity is

1 ~ I 2= 1 1 2 IRshl = 0.357 X 30 X lo3 = 10.71 kV
e. The output power delivered to the load is
Pout= 1 1 2 12R,hl = 0.3572 X 30 X lo3 = 3.82 kW
f. The power gain is
Gain = 1/4[2/(20 x lo3) x 71312 x l 4 x 10 x lo3 x 30 x 103
= 3.83 x lo5 = 55.8 dB
g. The electronic efficiency is

9-3-3Output Power of Four-Cavity Klystron

High power may be obtained by adding additional intermediate cavities in a two-

cavity klystron. Multicavity klystrons with as many as seven cavities are commer-
cially available, although the most frequently used number of cavities is four. Each
of the intermediate cavities functions in the same manner as in the two-cavity
klystron amplifier.
Let us carry out a simplified analysis of the four-cavity klystron amplifier. The
four cavities are assumed to be identical and they have same unloaded Q and beam
coupling coefficient Pi = Po. The two intermediate cavities are not externally
loaded, but the input and output cavities are matched to their transmission lines. If V,
is the magnitude of the input cavity-gap voltage, the magnitude of the RF convection
current density injected into the first intermediate cavity gap is given by Eq. (9-3-
17). The induced current and voltage in the first intermediate cavity are given by
Eqs. (9-3-19) and (9-3-20). This gap voltage of the first intermediate cavity pro-
duces a velocity modulation on the beam in the second intermediate cavity. The RF
convection current in the second intermediate cavity can be written with 1 Vl I re-
placed by ( V2 ( . That is,
1 low
Ii3I = --PoIV21
2 vow,
372 Microwave Linear-Beam Tubes (0Type) Chap. 9

The output voltage of the second intermediate cavity is then given by

=- (-)
1 low
4 Vow,
Pdl VI IR:h
This voltage produces a velocity modulation again and is converted into an RF con-
vection current at the output cavity for four-cavity klystron as
1 low
Ii4I = --P,IV3)
2 vow,

=- (-)
1 low3
8 Vow,
Vl IR2

and

1 = pol 1 = -8 (-)Vow,
1 low
114 i4 v1I R ; ~
The output voltage is then

I V4 I = I I4IRShl = -81 (-)Vow,

low
1881 VI IRLR,

The output power from the output cavity in a four-cavity klystron amplifier can be
expressed as

where Rshl = 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 in-
terception. 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: Output Power of Four-Cavity Klystron

A four-cavity klystron has the following parameters:

Beam voltage: vo = 10 kV
Beam current: I0 = 0.7 A
Operating frequency: f = 4GHz
Beam coupling coefficient: P I = P o = 1

dc electron beam current density: pa = 5 X Clm3

Signal voltage: Vl = 2 V (rms)
Shunt resistance of cavity: RSh = 10 kcl
Total shunt resistance including load: Rsh, = 5 kcl
Sec. 9.4 Reflex Klystrons

Determine:

a. The plasma frequency

b. The reduced plasma frequency for R = 0.6
c. The induced current in the output cavity
d. The induced voltage in the output cavity
e. The output power delivered to the load
Solution

a. The plasma frequency is

5 x
8.854 X lo-''
I"' = 0.997 X lo9 radls

b. The reduced plasma frequency is

w, = 0.6 X 0.997 X lo9 = 0.598 x lo9 radls
w/w, = 27~X 4 x 109/(0.598 x lo9) = 42.03
c. The induced current in the output cavity is

= 0.6365 A
d. The induced voltage in the output cavity is
IV4I = I141Rsh/ = 10.63651 X 5 X 10'
= 3.18 kV
e. The output power is
Po,, = I14(2R,h/= 10.6365)ZX 5 X lo3 = 2.03 kW

9-4 REFLEX KLYSTRONS

If a fraction of the output power is fed back to the input cavity and if the loop gain
has a magnitude of unity with a phase shift of multiple 27r, the klystron will oscil-
late. 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 two-
cavity klystron oscillator. It is a low-power generator of 10 to 500-mWoutput at a
frequency range of 1 to 25 GHz. The efficiency is about 20 to 30%. This type is
widely used in the laboratory for microwave measurements and in microwave receiv-
ers 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 Microwave Linear-Beam Tubes (0Type) 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 re-
peller space with greater velocity than those with unchanged velocity. Some elec-
trons 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. Oscilla-
tor 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.

9.4- 1 Velocity Modulation

The analysis of a reflex klystron is similar to that of a two-cavity klystron. For sim-
plicity, the effect of space-charge forces on the electron motion will again be ne-
glected. The electron entering the cavity gap from the cathode at z = 0 and time t~
is assumed to have uniform velocity

The same electron leaves the cavity gap at z = d at time tl with velocity

[ tz
~ uo 1 + -sin (wtl -
v ( t= $)]
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=
+
V, Vo + V I sin ( o t )
(9-4-3)
L
This retarding field E is assumed to be constant in the z direction. The force equa-
tion for one electron in the repeller region is

where E = -VV is used in the z direction only, Vr is the magnitude of the repeller
voltage, and ( VI sin wt 1 + (Vr + Vo)is assumed.
Integration of Eq. (9-4-4) twice yields
Sec. 9.4 Reflex Klystrons

RF output

Anode
Cathode Repeller

_____--- ---

I)

vo
V, = V , sin w l
v,
I

+I1-
10 f l , f 2

to = time for electron entering cavity gap dt z = 0

rl =time for same electron leaving cavity gap at z = d
t 2 = time for same electron returned by retarding field Figure 9-4-1 Schematic diagram of a
z = d and collected on walls of cavity reflex klystron.

)ring this

I .
accelerated during decelerated during
I this half cycle.- this half cycle. -

Figure 9-4-2 Applegate diagram with gap voltage for a reflex klystron.
376 Microwave Linear-Beam Tubes (0Type) Chap. 9

at t = t l , dzldt = v ( t l ) = K I; then

at t = t l , z = d = K2; then

On the assumption that the electron leaves the cavity gap at z = d and time tl with a
velocity of u(tl) and returns to the gap at z = d and time t 2 , then, at t = t 2 , z = d ,
-e(V, + Vo)
0 = (t2 - tJ2 + v(tJ(t2 - tl)
2mL
The round-trip transit time in the repeller region is given by

where
T1=t2-tl=
2mL
v ( t l ) = T; 1
e (v,+ VO) [ +- ( sin wtl - -
)I (9-4-7)

is the round-trip dc transit time of the center-of-the-bunch electron.

Multiplication of Eq. (9-4-7) through by a radian frequency results in

where
w (t2 - t ~ =) 66 + XI i 3
sin a t l - -

06 = wT6 (9-4-10)
is the round-trip dc transit angle of the center-of-the-bunch electron and

is the bunching parameter of the reflex klystron oscillator.

9-4-2 Power Output and Efficiency

In order for the electron beam to generate a maximum amount of energy to the oscil-
lation, 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 trans-
ferred 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

where VI + Vo is assumed, n = any positive integer for cycle number, and N =

n - $ is the number of modes.
The current modulation of the electron beam as it reenters the cavity from the
repeller region can be determined in the same manner as in Section 9-2 for a two-
cavity klystron amplifier. It can be seen from Eqs. (9-2-30) and (9-4-9) that the
bunching parameter X ' of a reflex klystron oscillator has a negative sign with respect
to the bunching parameter X of a two-cavity klystron amplifier. Furthermore, the
beam current injected into the cavity gap from the repeller region flows in the nega-
tive z direction. Consequently, the beam current of a reflex klystron oscillator can be
written
rn

i2, = -I0 - 210Jn(nX') cos [n (wt2 - 66 - OR)] (9-4-13)

n= 1

The fundamental component of the current induced in the cavity by the modulated
electron beam is given by
i2 = - p,12 = 2 1 0 P i J ~ ( Xcos
1 ) (wt2 - 6;) (9-4-14)
in which OR has been neglected as a small quantity compared with 6 6 . The magni-
tude of the fundamental component is

The dc power supplied by the beam voltage VOis

and the ac power delivered to the load is given by

From Eqs.(9-4-10), (9-4-1I ) , and (9-4-12) the ratio of VI over Vo is expressed by

Substitution of Eq. (9-4-18) in Eq. (9-4-17) yields the power output as

Therefore the electronic efficiency of a reflex klystron oscillator is defined as

Pa, 2XrJl(X')
'Efficiency = - =
Pdc 27~n- n / 2
378 Microwave Linear-Beam Tubes (0Type) Chap. 9

The factor X 'JI(X') reaches a maximum value of 1.25 at X ' = 2.408 and JI(X') =
0.52. In practice, the mode of n = 2 has the most power output. If n = 2 or 1 a
mode, the maximum electronic efficiency becomes
2(2.408)Jj(2.408)
Efficiency,,, = = 22.7%
2lT (2) - ~ / 2
The maximum theoretical efficiency of a reflex klystron oscillator ranges from 20 to
30%. Figure 9-4-3 shows a curve of X ' JI(X') versus X'.

Figure 9-4-3 X 'J , ( X ') versus X '

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):

The power output can be expressed in terms of the repeller voltage V , . That is,
-

It can be seen from Eq. (9-4-22) that, for a given beam voltage VOand cycle number
n or mode number N, the center repeller voltage V, can be determined in terms of
the center frequency. Then the power output at the center frequency can be calcu-
lated 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 Reflex Klystrons 379

Repeller voltage in volts

Figure 9-4-4 Power output and frequency characteristics of a reflex klystron.

9-4-3 Electronic Admittance

From Eq. (9-4- 14) the induced current can be written in phasor form as

The voltage across the gap at time t2 can also be written in phasor form:

The ratio of iz to V2 is defined as the electronic admittance of the reflex klystron.

That is,

The amplitude of the phasor admittance indicates that the electronic admittance is a
function of the dc beam admittance, the dc 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 2J1(Xt)/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 Microwave Linear-Beam Tubes (0 Type) Chap. 9

Figure 9-4-5 Reflex klystron saturation

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; G, 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,

where G = G, + Gb + Ge = 1/RShand Rshis the effective shunt resistance.

Figure 9-4-6 Equivalent circuit of a

reflex klystron.

Equation (9-4-26) can be rewritten in rectangular form:

Since the electronic admittance shown in Eq. (9-4-26) is in exponential form, its
phase is r / 2 when 86 is zero. The rectangular plot of the electron admittance Ye is a
spiral (see Fig. 9-4-7). Any value of 86 for which the spiral lies in the area to the left
of line (-G - jB) will yield oscillation. That is,

where N is the mode number as indicated in the plot, the phenomenon verifies the
early analysis.
Figure 9-4-7 Electronic admittance spiral of a reflex klystron.

Example 9-4-1: Reflex Klystron

A reflex klystron operates under the following conditions:
Vo=600V L = Imm

R, = 15kR = 1.759 X 10" (MKS system)

m

The tube is oscillating atf, at the peak of the n = 2 mode or 1f 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

Microwave Linear-Beam Tubes (0 Type) Chap. 9

V, = 250 V
b. Assume that Po = 1 . Since
VZ = I2Rsh = 2IoJ,(X ')Rsh
the direct current I. is

"2 - 200
I - = 11.45 mA
- 2Jj(X1)R,h 2(0.582)(15 X 10')
c. From Eqs. (9-4-1 I ) , ( 9 - 4 - 1 2 ) ,and (9-4-20) the electronic efficiency is
2 X ' JI(X1) -
-
2(1.841)(0.582)= 19.49%
Efficiency =
27rn - 7r/2 27r(2) - 7r/2

9-5 HELIX TRAVELING-WAVE TUBES (TWTS)

Since Kompfner invented the helix traveling-wave tube (TWT) in 1944 [ l l ] , its basic
circuit has changed little. For broadband applications, the helix TWTs are almost ex-
clusively 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 de-
tail. 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 veloc-
ity modulation caused by the weak electric fields associated with the traveling wave.
Just as in the klystron, this velocity modulation later translates to current modula-
tion, which then induces an RF current in the circuit, causing amplification. How-
ever, 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 struc-

ture. 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 Helix Traveling-Wave Tubes (TWTs) 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 he-
lix pitch to helix circumference. When the electrons enter the helix tube, an interac-
tion 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 interac-
tion 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 de-
celerated. As the electrons travel further along the helix, they bunch at the collector
end. The bunching shifts the phase by 7r/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%
Power output: up to 10 kW average
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.

Figure 9 - 5 1 State of the art for U.S.

high-power TWTs.
384 Microwave Linear-Beam Tubes (0 Type) Chap. 9

9-5-1 Slow-Wave Structures

As the operating frequency is increased, both the inductance and capacitance of the
resonant circuit must be decreased in order to maintain resonance at the operating
frequency. Because the gain-bandwidth product is limited by the resonant circuit,
the ordinary resonator cannot generate a large output. Several nonresonant periodic
circuits or slow-wave structures (see Fig. 9-5-2) are designed for producing large
gain over a wide bandwidth.

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 sig-
nal 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 con-
ducting cylinder (see Fig. 9-5-3).
It can be shown that the ratio of the phase velocity up along the pitch to the
phase velocity along the coil is given by
*P

:- y
- P
m -
= sin +
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs)

Figure 9-5-3 Helical slow-wave structure. (a) Helical coil. (b) One turn of helix.

where c = 3 x 10' m/s is the velocity of light in free space

p = helix pitch
d = diameter of the helix
$I = pitch angle

In general, the helical coil may be within a dielectric-filled cylinder. The phase
velocity in the axial direction is expressed as

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 ap-
proximately represented by

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 struc-
ture. Once p is found, up 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

I
I
I
I
I
/
I = E =P
" p lrd

Figure 9-5-4 w - p diagram for a helical

P structure.
386 Microwave Linear-Beam Tubes (0 Type) Chap. 9

In order for a circuit to be a slow-wave structure, it must have the property of

periodicity in the axial direction. The phase velocity of some of the spatial harmon-
ics in the axial direction obtained by Fourier analysis of the waveguide field may be
smaller than the velocity of light. In the helical slow-wave structure a translation
back or forth through a distance of one pitch length results in identically the same
structure again. Thus the period of a helical slow-wave structure is its pitch.
In general, the field of the slow-wave structure must be distributed according to
Floquet's theorem for periodic boundaries. Floquet's periodicity theorem states that:

The steady-state solutions for the electromagnetic fields of a single propagating mode in
a periodic structure have the property that fields in adjacent cells are related by a com-
plex constant.

Mathematically, the theorem can be stated

where E ( x , y, z) is a periodic function of z with periodl. Since PO is the phase con-

stant in the axial direction, in a slow-wave structure PO is the phase constant of aver-
age 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-jPoZ (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:

E(x, y, z - L) = f (x, y, z - L)~-@O(~-~) (9-5-7)

Since f ( x , y , z - L ) is a periodic function with period L , then
f ( x , y , z - L) = f ( x , y, z)
Substitution of Eq. (9-5-8) in (9-5-7) results in
E(x, y, z - L ) = f ( x , y , z)e-j@'eJPoL (9-5-9)
and substitution of Eq. (9-5-6) in (9-5-9) gives
E(x, y, z - L ) = E ( x , y , z)ejPoL (9-5-10)
This expression is the mathematical statement of Floquet's theorem, Eq. ( 9 - 5 - 5 ) .
Therefore Eq. (9-5-6) does indeed satisfy Floquet's theorem.
From the theory of Fourier series, any function that is periodic, single-valued,
finite, and continuous may be represented by a Fourier series. Hence the field distri-
bution function E ( x , y , z ) may be expanded into a Fourier series of fundamental pe-
riod L as
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs) 387

where
E,(x, y) = -
:6
are the amplitudes of n harmonics and
E (x, y , z)ej(2"n'L)Z
dz (9-5- 12)

is the phase constant of the nth modes, where n = - a , . . . , -2, - 1, 0, 1, 2, 3,

...,a.
The quantities E,(x, y)e-'PnZ 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

Since the wave equation is linear, Eq. (9-5-14) can be rewritten as

It is evident from the preceding equation that if each spatial harmonic is itself a solu-
tion 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 solu-
tion of Eq. (9-5- 14) can satisfy the boundary conditions of a periodic structure.
Furthermore, Eq. (9-5- 11) 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 up, . That is

The group velocity v,, , defined by v,, = dw/ap , is then given as

which is independent of n.
It is important to note that the phase velocity v,, in the axial direction de-
creases for higher values of positive n and P o . 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 o - P diagram demonstrates some important properties need-
ing more explanation. First, the second quadrant of the w - p diagram indicates the
negative phase velocity that corresponds to the negative n. This means that the elec-
tron beam moves in the positive z direction while the beam velocity coincides with
388 Microwave Linear-Beam Tubes (0Type) Chap. 9

W-
C
Forbidden region

-6n
- -4n 3 0 -
2n 4n & 0 Figure 9-5-5 o-fl diagram of spatial
L L L L L L harmonics for helical structure.

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 [lo].

9-5-2 Amplification Process

A schematic diagram of a helix-type traveling-wave tube is shown in Fig. 9-5-6.

The slow-wave structure of the helix is characterized by the Brillouin diagram
shown in Fig. 9-5-5. The phase shift per period of the fundamental wave on the
structure is given by

where Po = w / v o is the phase constant of the average beam velocity and L is the pe-
riod or pitch.
Since the dc transit time of an electron is given by

the phase constant of the nth space harmonic is

In Eq. (9-5-20) the axial space-harmonic phase velocity is assumed to be synchro-

nized with the beam velocity for possible interactions between the electron beam
and electric field. That is,

Equation (9-5-20) is identical to Eq. (9-5-13). In practice, the dc velocity of the

electrons is adjusted to be slightly greater than the axial velocity of the electromag-
netic wave for energy transfer. When a signal voltage is coupled into the helix, the
axial electric field exerts a force on the electrons as a result of the following rela-
tionships:
F = -eE and E = -VV (9-5-22)
Sec. 9.5 Helix Traveling-Wave Tubes ( W s )

Electron beam
Control anode

or

RF input Slow-wave structure RF output

Anode (helix)
7 1 I f

Gun
assembly

Collector
Heater ----/

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 accelerat-
ing 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 dc velocity of the electrons is slightly greater than the axial wave ve-
locity, 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
Microwave Linear-Beam Tubes (0Type) Chap. 9

Retarding field
Retarding force

I Accelerating force

Figure 9-5-7 Interactions between electron beam and electric field.

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 quantita-
tively analyzed in terms of the axial electric field. If the traveling wave is propagat-
ing in the z direction, the z component of the electric field can be expressed as
E, = El sin (wt - Ppz) (9-5-23)
where El is the magnitude of the electric field in the z direction. If t = to at z = 0,
the electric field is assumed maximum. Note that Pp = w / v p is the axial phase con-
stant of the microwave, and up is the axial phase velocity of the wave.
The equation of motion of the electron is given by

m -dv
= - eEl sin (wt - P,z)
dt
Assume that the velocity of the electron is
v = vo + v, cos (wet + 0,)
Then
dv
- = -vewe sin (wet + 0,) (9-5-26)
dt
where vo = dc electron velocity
v, = magnitude of velocity fluctuation in the velocity-modulated electron
beam
w, = angular frequency of velocity fluctuation
0, = phase angle of the fluctuation
Substitution of Eq. (9-5-26) in Eq. (9-5-24) yields
muewesin (wet + w,) = eE1 sin ( o t - P,z) (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 dc electron
velocity. This is
v z=5 vo (9-5-28)
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs) 391

Hence the distance z traveled by the electrons is

z = vo(t - to) (9-5-29)
and
mv,we sin (wet + 8,) = eE1 sin [wt - PPvo(t - to)] (9-5-30)
Comparison of the left and right-hand sides of Eq. (9-5-30) shows that

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 quanti-
ties, 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 dc
values. Mathematically, these quantities can be expressed as

+
where y = a, jpe is the propagation constant of the axial waves. The minus sign
is attached to Jo so that Jo may be a positive in the negative z direction. For a small
signal, the electron beam-current density can be written
J = PO = - Jo + J l eJw'-Y' (9-5-36)
+
where -Jo = povo, J I = p l v o p o o l , and p l v l .= 0 have been replaced. If an ax-
ial electric field exists in the structure, it will perturbate the electron velocity accord-
ing to the force equation. Hence the force equation can be written

where dzldt has been replaced by vo . Thus

392 Microwave Linear-Beam Tubes (0Type) Chap. 9

In accordance with the law of conservation of electric charge, the continuity equa-
tion can be written

V .J +-
ap = ( - yJI + jwp l)eJw'-yz
at
=0 (9-5-39)
It follows that
j~Jl (9-5-40)
P I = --w
Substitution Eqs. (9-5-38)and (9-5-40)in
JI = PIVO + POVI
gives
J~ = J -w -
e Jo (9-5-42)
vo m ( j w - yvo)2 - 5
where - Jo = 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 dc current l o with the same proportionality constant for JI
and Jo . Therefore the convection current in the electron beam is given by

where pe -= w/uo is defined as the phase constant of the velocity-modulated electron

beam and 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 slow-
wave 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 be-
tween the electron beam and the slow-wave helix is shown in Fig. 9-5-8.

L + I , + L 1-
C ' C' =: v C' --
--
), : ;Transnlission line
C. ;

-
-
A *
Figure 9-5-8 Electron beam coupled to
equivalent circuit of a slow-wave helix.
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs) 393

For simplicity, the slow-wave helix is represented by a distributed lossless

transmission line. The parameters are defined as follows:

L = inductance per unit length

C = capacitance per unit length
I = alternating current in transmission line
V = alternating voltage in transmission line
i = convection current

Since the transmission line is coupled to a convection-electron beam current, a

current is then induced in the line. The current flowing into the left-end portion of
the line of length dz is i, and the current flowing out of the right end of dz is
[i + (ai/az) dz]. Since the net change of current in the length dz must be zero, how-
ever, the current flowing out of the electron beam into the line must be
[-(dildz) dz]. Application of transmission-line theory and Kirchhoff's current law
to the electron beam results, after simplification, in

Then

in which a/dz = -y and a/at = jw are replaced.

From Kirchhoff 's voltage law the voltage equation, after simplification, is

Similarly,

Elimination of the circuit current I from Eqs. (9-5-45) and (9-5-46) yields

If the convection-elecron beam current is not present, Eq. (9-5-48) reduces to a typ-
ical wave equation of a transmission line. When i = 0, the propagation constant is
defined from Eq. (9-5-48) as

and the characteristic impedance of the line can be determined from Eqs. (9-5-45)
and (9-5-47):
7
394 Microwave Linear-Beam Tubes (0Type) 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):

Since E, = - VV = - (aV/dz) = yV, the axial electric field is given by

El= --y2yozo
y2 - yz
This equation is called the circuit equation because it determines how the axial elec-
tric field of the slow-wave helix is affected by the spatial ac electron beam current.

9-5-5 Wave Modes

The wave modes of a helix-type traveling-wave tube can be determined by solving

the electronic and circuit equations simultaneously for the propagation constants.
Each solution for the propagation constants represents a mode of traveling wave in
the tube. It can be seen from Eqs. (9-5-43) and (9-5-52)that there are four distinct
solutions for the propagation constants. This means that there are four modes of trav-
eling wave in the 0-type traveling-wave tube. Substitution of Eq. (9-5-43) in Eq.
(9-5-52) yields
y2yozo p,zo
( y 2 - Y;)( jpe - Y ) =~ - j
2vo
Equation (9-5-53)is of fourth order in y and thus has four roots. Its exact solutions
can be obtained with numerical methods and a digital computer. However, the ap-
proximate solutions may be found by equating the dc electron beam velocity to the
axial phase velocity of the traveling wave, which is equivalent to setting
yo = jpc (9-5-54)
Then Eq. (9-5-53)reduces to

where C is the traveling-wave tube gain parameter and is defined as

It can be seen from Eq. (9-5-55)that there are three forward traveling waves corre-
sponding to e - JPcz and one backward traveling wave corresponding to e+'Pe'. Let the
propagation constant of the three forward traveling waves be

where it is assumed that CS + 1.

Substitution of Eq. (9-5-57)in Eq. (9-5-55) results in
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs) 395

Since C6 < 1, Eq. (9-5-58) is reduced to

6 = (- j ) ' / "

From the theory of complex variables the three roots of (- j ) can be plotted in Fig.
9-5-9.

I -J Figure 9-5-9 Three roots of (- j ) .

Equation (9-5-59) can be written in exponential form as

6 = (- j ) l / 3 = e-j[(*/2+2n*)/31 ,( = 0, 1, 2 ) (9-5-60)
The first root 6 , at n = 0 is
7

The second root 62 at n = 1 is

7

The third root S3 at n = 2 is

8, = e-i3w/2 =j

The fourth root 64 corresponding to the backward traveling wave can be obtained by
setting
Y = - jPe - Pee64 (9-5-64)
Similarly,
.C2
84 = -1, (9-5-65)

Thus the values of the four propagation constants y are given by

396 Microwave Linear-Beam Tubes (0Type) 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 correspond-
ing to yl is a forward wave and that its amplitude grows exponentially with distance;
the wave corresponding to y2 is also a forward wave, but its amplitude decays expo-
nentially with distance; the wave corresponding to y3 is also a forward wave, but its
amplitude remains constant; the fourth wave corresponding to y4 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.

9-5-6 Gain Consideration

For simplicity, it is assumed that the structure is perfectly matched so that there is no
backward traveling wave. Such is usually the case. Even though there is a reflected
wave from the output end of the tube traveling backward toward the input end, the
attenuator placed around the center of the tube subdues the reflected wave to a mini-
mum or zero level. Thus the total circuit voltage is the sum of three forward
voltages corresponding to the three forward traveling waves. This is equivalent to
3

The input current can be found from Eq. (9-5-43) as

in which C6 4 1, El = yV, and y = j&(l - C6) have been used.

The input fluctuating component of velocity of the total wave may be found
from Eq. (9-5-38) as

where El = yV, C6 4 1 , P,vo = w , and vo = V ( 2 e / m ) v ohave been used.

To determine the amplification of the growing wave, the input reference point
is set at z = 0 and the output reference point is taken at z = e. It follows that at
z = 0 the voltage, current, and velocity at the input point are given by
Sec. 9.5 Helix Traveling-Wave Tubes (TWTs) 397

The simultaneous solution of Eqs. (9-5-73), (9-5-74), and (9-5-75) with i(0) = 0
and u1(0) = 0 is

Since the growing wave is increasing exponentially with distance, it will predomi-
nate over the total voltage along the circuit. When the length e 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

v(e) - "O' 3 exp

- ($bee) [- exp jPe(l + g)t] (9-5-77)

The factor pee is conventionally written 27rN, where N is the circuit length in elec-
tronic wavelength-that is,

The amplitude of the output voltage is then given by

v (0) exp ( ~ ' 3?~Nc)
v (e) = -3
The output power gain in decibels is defined as

where NC is a numerical number.

The output power gain shown in Eq. (9-5-80) indicates an initial loss at the cir-
cuit input of 9.54 dB. This loss results from the fact that the input voltage splits into
three waves of equal magnitude and the growing wave voltage is only one-third the
total input voltage. It can also be seen that the power gain is proportional to the
length N in electronic wavelength of the slow-wave structure and the gain parameter
C of the circuit.

Example 9-5-1: Operation of Traveling-Wave Tube (TWT)

A traveling-wave tube (TWT) operates under the following parameters:

Beam voltage: Vo = 3 kV
Beam current: l o = 30 mA
Characteristic impedance of helix: Zo = 10
Circuit length: N = 50
Frequency: f = 10 GHz
398 Microwave Linear-Beam Tubes (0Type) Chap. 9

Determine: (a) the gain parameter C; (b) the output power gain A, in decibels; and
(c) all four propagation constants.
Solution

a. From Eq. (9-5-56) the gain parameter is

b. From Eq. (9-5-80) the output power gain is

A, = -9.54 + 4.73 NC = -9.54 + 47.3 X 50 x 2.92 X lo-' = 59.52 dB
c. The four propagation constants are

9.6 COUPLED-CA VITY TRAVELING-WAVE TUBES

Helix traveling-wave tubes (TWTs) produce at most up to several kilowatts of aver-

age power output. Here we describe coupled-cavity traveling-wave tubes, which are
used for high-power applications.

9-6-1 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. Fig-
ure 9-6-1 shows two coupled-cavity circuits that are principally used in traveling-
wave tubes.
Sec. 9.6 Coupled-Cavity Traveling-Wave Tubes

Electron
Drift beam
Cavity tube A

Magnet Iron
pole piece

Copper

Figure 9-6-1 Coupled-cavity circuits in the TWTs. (a) Basic coupled-cavity cir-
cuit. (b) Coupled-cavity circuit with integral periodic-permanent-magnet(PPM)fo-
cusing. (After J. T. Mendel (151; reprinted by permission oflEEE.)

As far as the coupling effect is concerned, there are two types of coupled-
cavity 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 mu-
tual inductive coupling between the cavities and operate with the fundamental space
harmonic. The cloverleaf [I21 and centipede circuits [I31 (see Fig. 9-6-2) belong to
this type. The second type is the first space-harmonic circuit, which has positive mu-
tual coupling between the cavities. These circuits operate with the first spatial har-
monic 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 cir-
cuit of the positive mutual coupling-cavity circuit operates at the fundamental spatial
Microwave Linear-Beam Tubes (0Type) Chap. 9

Figure 9-6-2 Centipede and cloverleaf coupled-cavity circuits. (After A. Stuprans

et al. [7]; reprinted by permission of IEEE.)

Figure 9-6-3 Space-harmonic coupled-

cavity circuits. (After A . Staprans et al.
(71; reprinted by permission of IEEE,
Inc.)

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.

9-6-2 Principles of Operation

Any repetitive series of lumped LC elements constitute a propagating filter-type cir-

cuit. The coupled cavities in the traveling-wave tube are usually highly over-coupled,
resulting in a bandpass-filter-type characteristic. When the slot angle (8) 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 fre-
quency. 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 elec-
tron 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 compo-
nents that have phase characteristics [I61 as shown in Fig. 9-6-4. In Fig. 9-6-4 the
angular frequency (o)is plotted as a function of the phase shift (P t )per cavity. The
ratio of o to p is equal to the phase velocity. For a circuit having positive mutual in-
Sec. 9.6 Coupled-Cavity Traveling-Wave Tubes

T W T OPERATION
IS WlTH THlS BRANCH
BEAM CHARACTERISTIC \ OF u - 0 7

6L-
(0) TWT
OPERATION
IS WlTH THlS
BRANCH OF
Figure 9-6-4 w -p diagrams for
coupled-cavity circuits. (a) Fundamental
forward-wave circuit (negative mutual
inductance coupling between cavities).
(b) Fundamental backward-wave circuit
(positive mutual inductive coupling be-
tween cavities). (After A. Staprans et al.
[7]; reprinted by permission of IEEE,
Inc .)

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 cir-
cuit has been developed by Curnow [I71 as shown in Fig. 9-6-5.

Figure 9-6-5 Equivalent circuits for a

slot-coupled cavity. (After A. Staprans et
al. [7]; reprinted by permission of IEEE,
Inc.)
402 Microwave Linear-Beam Tubes (0Type) 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.

9-6-3 Microwave Characteristics

When discussing the power capability of traveling-wave tubes, it is important to

make a clear distinction between the average and the peak power because these two
figures are limited by totally different factors. The average power at a given fre-
quency is almost always limited by thermal considerations relative to the RF propa-
gating circuit. However, the peak RF power capability depends on the voltage for
which the tube can be designed. The beam current varies as the three-half power of

PPM FOCUSED COUPLED CAVITY TUBES

--
-
-
-
-

.2 - -
Figure 9-6-6 Peak and average power
capability of typical TWTs in field use.
.I
1 : : b A : 'i A
FREQUENCY(GHZ)
& I 'ebl100 (After J . T. Mendel [15]; reprinted by
permission of IEEE., Inc.)
Sec. 9.6 Coupled-Cavity Traveling-Wave Tubes 403

the voltage, and the product of the beam current and the voltage determines the to-
tal beam power. That is,

where Vo is the beam voltage and K is the electron-gun perveance.

For a solid-beam electron gun with good optics, the perveance is generally
considered to be between 1 to 2 X lo-('. Once the perveance is fixed, the required
voltage for a given peak beam power is then uniquely determined. Figures 9-6-6 and
9-6-7 demonstrate the difference between peak and average power capability and the
difference between periodic-permanent-magnet (PPM) and solenoid-focused designs
[151.
Coupled-cavity traveling-wave tubes are constructed with a limited amount of
gain per section of cavities to ensure stability. Each cavity section is terminated by
100 I I 1 1 1 1 1 1 1 I 1 1 l l l l -

80-

60- SOLENOID FOCUSED COUPLED

C A V l l Y TUBES
40-

20 -

lo -
$ O:

$r O -

-3 '-l -

8 -
8
a

Y
2-

5
1 -- -
.a - V-ZOkV

-
-
.B
-
1 -
-
.l -
Figure 9-6-7 CW power capability of
.I I I a a 1 1 1 1 1 I , , I , I L TWTs operating at nearly 20 kV. (After
1.o 2 3 4 5 8 7 8 8 1 0 20 m en so roo J . T. Mendel [ I S ] ; reprinted by permis-
FREQUENCY IGHZI sion of IEEE., Inc.)
404 Microwave Linear-Beam Tubes (0Type) 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 col-
lector. Interaction efficiencies from 10 to 40% have been achieved from coupled-
cavity traveling-wave tubes. Overall efficiencies of 20 to 55% have been obtained
[71.

9-7 HIGH-POWER AND GRIDDED-CONTROL

TRAVELING-WAWE TUBES

Coupled-cavity traveling-wave tubes are the most versatile devices used for am-
plification at microwave frequencies with high gain, high power, high efficiency, and
wide bandwidth. The present state of the art for U.S. high-power gridded tubes is
shown in Fig. 9-7-1. High-power TWTs have four main sections: electron gun for
electron emission, slow-wave structure for effective beam interaction, magnetic cir-
cuit for beam focusing, and collector structure for collecting electron beams and dis-
sipating heat energy. Specifically, the physical components of a coupled-cavity
traveling-wave tube consist of an electron emitter, a shadow grid, a control grid, a
modulating anode, a coupled-cavity circuit, a solenoid magnetic circuit, and a col-
lector depression structure (see Fig. 9-7-2).

Figure 9-7-1 State of the art for U.S. high-power gridded tubes.
406 Microwave Linear-Beam Tubes (0Type) 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. Usu-
ally 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 tech-
nique 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, sup-
presses 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.
Shadow grid Control Focus Modulating Anode

Figure 9-7-3 Electron emitter with control electrodes for gridded high-power
TWT.

In general, an anode modulation technique is frequently used in traveling-wave

tubes to eliminate voltage pulsing through the lower unstable beam voltage and to
reduce modulator power requirements for high-power pulse output. In gridded
traveling-wave tubes the modulator applies a highly regulated positive grid drive
voltage with respect to the cathode to turn the electron beam on for RF am-
plification. An unregulated negative grid bias voltage with respect to the cathode is
used to cut the electron beam off. Thus the anode modulator acts as a pulse switch
for the electron beam of the gridded traveling-wave tube.
The anode of the electron gun is operated at a voltage higher than that of the
slow-wave structure to prevent positive ions, formed by the electron beam in the re-
gion of the slow-wave structure, from draining toward the cathode and bombard-
ing it.

9-7-1 High Efficiency and Collector Voltage

Depression

After passing through the output cavity, the electron beam strikes a collector elec-
trode. The function of the collector electrode could be performed by replacing the
second grid of the output cavity with a solid piece of metal. However, using a sepa-
Sec. 9.7 High-Power and Gridded-Control Traveling-Wave Tubes 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 lo-
calized 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 per-
formance. 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. Grid-
ded 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 con-
duction 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 cur-
rent (beam current). It may be expressed in terms of the product of the electronic
efficiency and the circuit efficiency. The electronic efficiency expresses the percent-
age of the dc or pulsed input power that is converted into RF power on the slow-
wave structure. The circuit efficiency, on the other hand, determines the percentage
of dc input power that is delivered to the load exterior to the tube. The electron
beam does not extract energy from any dc power supply unless the electrons are ac-
tually 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 col-
lector 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 dc to RF power is almost twice the elec-
tronic efficiency. Under this condition the tube is operating with collector voltage
depression.

9-7-2 Normal Depression and Overdepression

of Collector Voltage

Most gridded traveling-wave tubes are very sensitive to variations of collector de-
pression 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 V, 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 I, is about 2.09 A. A small amount of elec-
trons 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 I, of 0.041 A, which is only a small
fraction of the body current I b . 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.
Collector voltage -V, in kilovolts

Figure 9-7-4 Spent beam curves for a typical gridded TWT.

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 overde-
pressed collector voltage. Figure 9-7-5 [18] shows the impact probability of re-
turned electrons by certain overdepressed collector voltage.
Sec. 9.7 High-Power and Gridded-Control Traveling-Wave Tubes

Output pole piece

Returned
electrons

Collector snout
I
- I
I 12.77 rnm A

a indicates that most electrons are ret~irncd

by low overdepression voltage
@ indicates that rnost electrons are returned
by high overdcprcssion voltagc
@ indicates that most electrons are returned
by higher overdepression voltage

Figure 9-7-5 Impact probability of returned electrons by overdepressed collector

voltage.

Example 9-7-1: Gridded Traveling-Wave Tube (GTWT)

A gridded traveling-wave tube is operated with overdepression of collector voltage as
follows:

Overdepression collector voltage: V, = - 11 kV

Returned current: I, = 0.85 A
Mass of heated iron pole piece: mass = 250 mg
Specific heat H of iron at 20°C: H = 0.108 calorieslg-"C

Determine:

a. The number of electrons returned per second

b. The energy (in eV) associated with these returning electrons in 20 ms
Microwave Linear-Beam Tubes (0 Type) Chap. 9

c. The power in watts for the returning electrons

d. The heat in calories associated with the returning electrons (a factor for convert-
ing joules to calories is 0.238)
e. The temperature T in degrees Celsius for the output iron pole piece [Hint:
T = 0.238VIt/(mass x specific heat)]
f. Whether the output iron pole piece is melted

Solution

a. The number of electrons returned per second is

0.85
I, = = 5.31 x 1018electrons per second
1.6 X lO-I9

b. The energy is
w = ~t = VI,~ = 11 x 10" x 5.31 x io18 x 20 x 10-3
= 1.168 X lo2' eV
c. The power is
P = VI, = 11 X lo3 X 0.85 = 9.35 kW
d. The created heat is
H(heat) = 0.238Pt = 0.238V1,t
= 0.238 X 11 X lo3 X 0.85 X 20 X

= 44.51 calories
e. The temperature is
0.238VIrt
T =
mass x specific heat

f. The output iron pole piece is melted.

9-7-3 Two-Stage Collector Voltage Depression

Technique

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 col-
lector voltage. Consequently, the tube body current is very small and negligible be-
cause 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 High-Power and Gridded-Control Traveling-Wave Tubes 411

pole piece. Thus the tube body current will increase sharply. Since electrons of vari-
ous 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 opera-
tion, but the collector snout may be grounded to the output pole piece for overde-
pression operation. As a result, the returned electrons will be collected by the col-
lector 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 mech-
anism, 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 -
,
,
Output pole piece Collector

------
- - - - -- - - -2 Returned
Electron beam hole - - - - - - - - - - ) electrons

-VC Figure 9-7-6 Diagram for two-stage

collector depression.

Solenoid
supply

I .
Cathode t -
power Modulator
supply I

71
Collec

i
h n

Vk t Cathode -Vk
v y v
CTWT
I Body current Ib
-vc I
Collector
power
supply

Figure 9-7-7 Basic interconnection of a gridded traveling-wave tube with its

power supplies.
412 Microwave Linear-Beam Tubes (0Type) Chap. 9

9-7-4 Stabilization of Cathode and Collector

Voltages

The cathode voltage of a gridded traveling-wave tube is negative with respect to

ground so that the electrons can be emitted from the cathode. In order to maintain a
constant beam power for a uniform gain, the cathode voltage must be constant. In
addition, the phase shift through the tube is directly related to the beam velocity;
thus high resolution and low ripple are required in the cathode voltage power supply
to avoid undesirable phase-shift variations. Consequently, the cathode power supply
of the gridded traveling-wave high-power tube is usually regulated for better than 1%
over line and load changes and is also well filtered because of the critical require-
ments on the cathode voltage with respect to ground. The cathode power supply pro-
vides the tube body current. Under normal operation the body current is very small
in comparison with the collector current. Figure 9-7-8 shows a basic interconnec-
tion for a gridded traveling-wave tube with two regulated power supplies.

Regulated
cathode +

power

1
-Vk
Cathode -!I! -
Modulator

h!~- !I
!

Regulated
GTWT > Collector

- collector + -%
power
supply v,,

Figure 9-7-8 Interconnections for a gridded traveling-wave high-power tube.

Figure 9-7-9 illustrates a voltage-regulator circuit for a cathode power supply.

The voltage regulator indicated in the circuit consists of two devices: one differential
amplifier and one tetrode tube. The solid-state differential amplifier amplifies the
difference between the preset reference voltage and the voltage that is one-
thousandth of the output voltage. The reference voltage is adjustable to a preset
level. The output voltage of the differential amplifier drives the control grid of the
regulator tube between cutoff and saturation in order to nullify the difference
voltage.
As shown in Fig. 9-7-8 the negative terminal of the collector power supply is
connected to the cathode and the positive terminal to the collector electrode. The
414 Microwave Linear-Beam Tubes (0Type) 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. Load changes: When the load of a generator is suddenly increased, a larger

current is needed. Since the generator cannot meet the demand instanta-
neously, the output voltage of the generator drops momentarily. Conversely,
when the load of a generator is suddenly decreased, the output voltage drops
accordingly.
2. Switching on or off: When the switch of a generator is just turned on or off, the
armature current in the armature conductors produces armature reaction. The
nature of armature reaction reduces the terminal voltage for lagging loads.

When an electrical transient is created in the circuit, the collector voltage is

overdepressed. As a result, the spent electrons inside the collector reverse the direc-
tion of their velocity by the highly negative collector voltage and fall back onto the
grounded output pole piece. The tube body current is sharply increased and the col-
lector current is greatly decreased. When the returned electrons impact the output
pole piece, the pole piece will be damaged by high heat. The damage of the output
pole piece creates a mismatch in the interaction circuit and degrades the performance
of the tube. In particular, the tube gain, efficiency, bandwidth, and power output are
affected accordingly by the circuit mismatch when the collector voltage is overde-
pressed below the normal depression level. Furthermore, the large body current may
burnout the solid-state differential amplifier of the cathode voltage regulator and
vary the electron beam of the gridded tube. If the damage is beyond the tolerance of
the gridded traveling-wave high-power tube, the tube may cease to function.
In order to maintain a constant collector depression voltage, the collector
voltage must be regulated. There are three possible ways to do so [18]:

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 regula-
tor is inserted in parallel with the collector supply as shown in Fig. 9-7-1 1 so
that a regulated voltage with respect to ground at normal depression may be
achieved at the collector terminal.
,-dJ,
asinos aseqd-aai q~
Chap. 9 References 417

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 overde-
pressed above the normal depression value (absolute value), differential am-
plifier 2 tends to adjust the cathode voltage below its fixed level (absolute
value). When the cathode voltage is dropped, the collector voltage is read-
justed to its normal depression level with respect to ground.

+-
I
-
-

- vk .
l

-
388R2
-1 -"c
Differential -10-3 vk
Differential r - T -7
I I amplifier 2 II II
I I I I
I I
p'k 1 II +VP
I
l
I
I I I I
I I I I
1 - 1 I - I
L--,--J L-T--l
I IO~R, I
r-L- r-A-7
I + ; 1
I
+ 1
I I I
I I
I
I I II
I
1 ;Vk I
I
II iVP I
I I I
I I
I I I I
L - , _- - - I
I I 1 - 1
L--,--J
I
I -
A
-
-vk
-
I
I

Negative of cathode supply Negative of collector supply

Figure 9-7-12 Regulator between cathode voltage and collector voltage.

REFERENCES

[ I ] WARNECKE, R. R., et a]., Velocity modulated tubes. In Advances in Electronics, Vol. 3 .

Academic Press, New York, 1951.
[2] CHODOROW, M., and C. SUSSKIND, Fundamentals of Microwave Electronics. McGraw-
Hill Book Company, New York, 1964.
[3] CHODOROW, M., and T. WESSEL-BERG, A high-efficiency klystron with distributed inter-
action. IRE Trans. Electron Devices, ED-8, 44-55, January 1961.
[4] LARUE, A. D., and R. R. RUBERT, Multi-Megawatt Hybrid TWT's at S-band and C-
band. Presented to the IEEE Electron Devices Meeting, Washington, D.C., October
1964.
418 Microwave Linear-Beam Tubes (0Type) 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 , 41-44.
[7] STAPRANS, A., et al., High-power linear beam tubes. Proc. IEEE, 61, No. 3, 299-330.
March 1973.
[8] LIEN,E. L., High efficiency klystron amplifier. In Conv. Rec. MOGA 70 (8th Int. Conf.,
Amsterdam, September 1970).
[9] MERDINIAN, G., and J. V. LEBACQZ, High power, permanent magnet focused, S-band
klystron for linear accelerator use. Proc. 5th Int. Conf. of Hyper-frequency Tubes (Paris,
September 1964).
[lo] 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- 127, February 1947.
[I21 CHODOROW, M., and R. A. CRAIG,Some new circuits for high-power traveling-wave
tubes. Proc. IRE, 45, 1106- 1 1 18, August 1957.
[13] ROUMBANIS, T., et al., A megawatt X-band TWT amplifier with 18% bandwidth. Proc.
High-Power Microwave Tubes Symp., Vol. 1 (The Hexagon, Fort Monmouth, N.J.,
September 25-26, 1962).
[I41 R u m , A. J., and W. H. Y o c o ~ ,High-power traveling-wave tubes for radar systems.
IRE Trans. Mil. Electron., MIL-5, 39-45, April 1961.
[I51 MENDEL, J. T., Helix and coupled-cavity traveling-wave tubes. Proc. IEEE, 61, No. 3,
280-298, March 1973.
[I61 BRILLOUIN, L., Wave Propagation in Periodic Structures, 2nd ed. Dover, New York,
1953.
[I71 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 1965.
[I81 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., Foundationsfor Microwave Engineering, Chapter 9. McGraw-Hill Book Com-

pany, New York, 1966.
GANDHI, 0. P., Microwave Engineering and Applications, Chapters 9, 10, and 11. Pergamon
Press, New York, 1981.
GEWARTOWSKI, J. W., and H. A. WATSON, Principles of Electron Tubes, Chapters 5, 6 , and 10
to 13. D. Van Nostrand Company, Princeton, N.J., 1965.
GILMOUR, A . S., JR., Microwave Tubes. Artech House, Dedham, Mass., 1986.
IEEE Proceedings, 61, No. 3, March 1973. Special issue on high-power microwave tubes.
LIAO,S. Y., Microwave Electron Tubes. Prentice-Hall, Inc., Englewood Cliffs, N.J., 1988.
Chap. 9 Problems

PROBLEMS

Vacuum Tubes
9-1. A vacuum pentode tube has five grids: a cathode, a control grid, a screen grid, a sup-
pressor grid, and an anode plate as shown in Fig. P9- 1.
a. Sketch the equivalent circuit.
b. Derive an expression for the input impedance Z,, in terms of the angular frequency
w and the circuit parameters.
c. Determine the transit-angle effect.

Figure P9-1

Klystrons
9-2. The parameters of a two-cavity amplifier klystron are as follows:

Beam voltage: V, = 1200 V

Beam current: I,, = 28 mA
Frequency : f = 8 GHz
Gap spacing in either cavity: d=lmm
Spacing between the two cavities: L = 4 cm
Effective shunt resistance: RSh= 40 k f l (excluding beam loading)
a. Find the input microwave voltage V , in order to generate a maximum output voltage
VZ (including the finite transit-time effect through the cavities).
b. Determine the voltage gain (neglecting the beam loading in the output cavity).
c. Calculate the efficiency of the amplifier (neglecting the beam loading).
d. Compute the beam loading conductance and show that one may neglect it in the
preceding calculations.
9-3. A two-cavity amplifier klystron has the following characteristics:

Voltage gain: 15 dB
Input power: 5 mW
Total shunt impedance of input cavity Rsh: 30 k n
Total shunt impedance of output cavity Rsh: 40 k f l
Load impedance at output cavity Rp: 40 k f l
420 Microwave Linear-Beam Tubes (0 Type) Chap. 9

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: Vo = 900V

Beam current: lo= 30 mA
Frequency: f = 8 GHz
Gap spacing in either cavity: d=lmm
Spacing between centers of cavities: L = 4 cm
Effective shunt impedance: Rsh = 40 kK?

Determine:
a. The electron velocity
b. The dc electron transit time
c. The input voltage for maximum output voltage
d. The voltage gain in decibels
9-5. Derive Eq. (9-2-49).

Multicavity Klystrons
9-6. A four-cavity klystron amplifier has the following parameters:

Beam voltage: V0 = 20 kV
Beam current: Io=2A
Operating frequency: f = 9 GHz
dc charge density: po = C/m3
RF charge density: p = lo-' C/m3
Velocity perturbation: ?'" = lo5 m/s

Determine:
a. The dc electron velocity
b. The dc phase constant
c. The plasma frequency
d. The reduced plasma frequency for R = 0.5
e. The beam current density
f. The instantaneous beam current density
9-7. A four-cavity CW klystron amplifier has the following parameters:

Beam voltage: Vo = 30 kV
Beam current: Io=3A
Gap distance: d = lcm
Operating frequency: f = 8GHz
Signal voltage: VI = 15 V(rms)
Beam coupling coefficient: P I. = P0 -
-1
dc electron charge density: po = lo-' C/m3
Chap. 9 Problems

Compute:
a. The dc electron velocity
b. The dc 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:
dc electron charge density:
Signal voltage:
Cavity shunt resistance:
Total shunt resistance including load:

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:
dc 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
Microwave Linear-Beam Tubes (0 Type) Chap. 9

Reflex Klystrons
9-10. A reflex klystron operates at the peak mode of n = 2 with

Beam voltage: Vo = 300 V

Beam current: lo= 20 mA
Signal voltage: VI = 40 V

Determine:
a. The input power in watts
b. The output power in watts
c. The effficiency
9-11. A reflex klystron operates under the following conditions:

Vo 500 V
=
R,,, 20 kfl
=
f, = 8 GHz
L = 1 mm is the spacing between repeller and cavity

The tube is oscillating at f, at the peak of the n = 2 mode or 1f 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 dc power input is
40 mW and Vl/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 = 1 or f mode. The dc power input is
40 mW and the ratio of Vl 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)

9-14. A traveling-wave tube (TWT) has the following characteristics:

Beam voltage: Vo = 2 kV
Beam current: I. = 4 m A
Frequency: f =8GHz
Circuit length: N = 50
Characteristic impedance: Zo = 20 L
!

Determine:
a. The gain parameter C
b. The power gain in decibels
Chap. 9 Problems

9-15. A TWT operates under the following parameters:

Beam current: lo = 50 mA
Beam voltage: Vo = 2.5 kV
Characteristic impedance of helix: ZO = 6.75 l2
Circuit length: N = 45
Frequency : f = 8 GHz

Determine:
a. The gain parameter C
b. The output power gain A, in decibels
c. All four propagation constants
d. The wave equations for all four modes in exponential form
9-16. 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.
9-17. 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 Nplm. Determine the propagation
constant y of the traveling wave in the tube.
9-18. In an 0 -type traveling-wave tube, the acceleration voltage (beam voltage) is 3000 V.
The characteristic impedance is 10 0. 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.
9-19. Describe the structure of an 0-type traveling-wave tube and its characteristics; then ex-
plain how it works.
9-20. 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. De-
termine the magnitude of velocity fluctuation.

Gridded Traveling-Wave Tubes (GTWTs)

9-21. The current I, caused by the returning electrons at an overdepression voltage of
- 11.5 kV in a GTWT is about 0.973 A as shown by the spent beam curve in Fig.
9-7-4.
a. Calculate the number of electrons returned per second.
b. Determine the energy in electron volts associated with these returning electrons in
1 ms for part (a).
c. Find the power in watts for the returning electrons.
9-22. The output iron pole piece of a GTWT has the following characteristics:

1. The specific heat H at 20°C is 0.108 cal/g°C.

2. A factor for converting joules to calories is 0.238.
3. The mass of the heated iron pole piece is assumed to be 203.05 mg.
4. The duration time t of the collector depression transient voltage of - 11.5 kV is
15 ms.
5. The melting point of iron is 1535°C.
424 Microwave Linear-Beam Tubes (0Type) Chap. 9

Using these characteristics,

a. Calculate the heat in calories associated with the returning electrons at an overde-
pression voltage of - 1 1.5 kV.
b. Compute the temperature T in degrees Celsius for the output iron pole piece. [Hint:
T = 0.238VIt/(mass x specific heat).]
c. Determine whether the output iron pole piece is melted.
9-23. The efficiency of a GTWT is expressed as

If the cathode voltage is - 18 kV and the collector voltage is depressed to -7.5 kV de-
termine the efficiency of the GTWT.
1 Microwave Crossed-Field
Tubes (M Type)

10-0 INTRODUCTION

In the previous chapter, several commonly used linear-beam tubes were described in
detail. In these tubes, the dc magnetic field that is in parallel with the dc electric
field is used merely to focus the electron beam. In crossed-field devices, however,
the dc magnetic field and the dc electric field are perpendicular to each other. In all
crossed-field tubes, the dc magnetic field plays a direct role in the RF interaction
process.
Crossed-field tubes derive their name from the fact that the dc electric field and
the dc magnetic field are perpendicular to each other. They are also called M-type
tubes after the French TPOM (tubes a propagation des ondes a champs magnktique:
tubes for propagation of waves in a magnetic field). In a crossed-field tube, the elec-
trons 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 dc electric field far enough to regain essentially the same velocity as before. Be-
cause 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 phe-
nomenon would make the M-type devices relatively efficient. Those electrons enter-
ing 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 Microwave Crossed-Field Tubes (M Type) Chap. 10

In this chapter, several commonly used crossed-field tubes such as magnetrons,

forward-wave crossed-field amplifiers (FWCFAs), backward-wave crossed-field am-
plifiers (BWCFAs or amplitrons), and backward-wave crossed-field oscillators
(BWCFOs or carcinotrons) are studied.

Cylindrical magnetron: The cylindrical magnetron was developed by Boot and

Randall in early 1940.
Coaxial magnetron: The coaxial magnetron introduced the principle of inte-
grating a stabilizing cavity into the magnetron geometry.
Voltage-tunable magnetron: The voltage-tunable magnetron has the cathode-
anode geometry of the conventional magnetron, but its anode can be tuned
easily.
Inverted magnetron: The inverted magnetron has the inverted geometry of the
conventional magnetron with the cathode placed on the outside surrounding
the anode and microwave circuit.
Forward-wave crossed-field amplijier (FWCFA): The forward-wave crossed-
field amplifier is also called the M-type forward-wave amplifier.
Backward-wave crossed$eld amplijier (BWCFA):The backward-wave crossed-
field amplifier was developed by Raytheon Company in 1960 and its trade
name is Amplitron. It is a broadband, high-power, high-gain, and high-
efficiency microwave tube, and it has many applications such as in airborne
radar systems and spaceborne communications systems.
Backward-wave crossed-jield oscillator (BWCFO): In 1960, the French Com-
pany developed a new type of crossed-field device in which an injection gun
replaced the conventional cylindrical cathode of the magnetron. Its trade name
is Carcinotron, and it is also called the M-type backward-wave oscillator.

All of these crossed-field electron tubes are tabulated in Table 10-0-1.

Crossed-field tubes ( M type)

I
I
Nonresonant
I
Maser effect
~esonant

I
Standing wave

I
Reentrant
Reentrant
I
Nonreentrant
I
Reentrant
Sec. 10.1 Magnetron Oscillators

10-1 MAGNETRON OSCILLATORS

Hull invented the magnetron in 1921 [I], but it was only an interesting laboratory
device until about 1940. During World War 11, an urgent need for high-power mi-
crowave generators for radar transmitters led to the rapid development of the mag-
netron to its present state.
All magnetrons consist of some form of anode and cathode operated in a dc
magnetic field normal to a dc 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 dc mag-
netic field is strong enough, the electrons will not arrive in the anode but return in-
stead 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 resis-

tance 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 mi-

crowave 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, lin-
ear (or planar) magnetron, coaxial magnetron, voltage-tunable magnetron, inverted
coaxial magnetron, and the frequency-agile magnetron will be discussed.

10-1= 1 Cylindrical Magnetron

A schematic diagram of a cylindrical magnetron oscillator is shown in Fig. 10-1-1.

This type of magnetron is also called a conventional magnetron.
In a cylindrical magnetron, several reentrant cavities are connected to the
gaps. The dc voltage Vois applied between the cathode and the anode. The magnetic
flux density Bo is in the positive z direction. When the dc voltage and the magnetic
flux are adjusted properly, the electrons will follow cycloidal paths in the cathode-
anode space under the combined force of both electric and magnetic fields as shown
in Fig. 10-1-2.

Equations of electron motion. The equations of motion for electrons in a

cylindrical magnetron can be written with the aid of Eqs.(l-2-5a) and (1-2-5b) as
428 Microwave Crossed-Field Tubes (M Type) Chap. 10

Figure 10-1-1 Schematic diagram of a cylindrical magnetron

.Electron
path

Figure 10-1-2 Electron path in a cylin-

drical magnetron.

where = 1.759 x loL1Clkg 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

$ $)
(r2 =
e d r l d
;B,r dl = 20' ;(r2)

e
where w, = -B, is the cyclotron angular frequency. Integration of Eq. (10-1-3)
m
yields

,.2 d4 = - w,r2
-
dt 2
+ constant (10-1-4)
Sec. 10.1 Magnetron Oscillators 429

at r = a , where a is the radius of the cathode cylinder, and = 0, con-

dt
I
stant = --o,a2. The angular velocity is expressed by
2

Since the magnetic field does no work on the electrons, the kinetic energy of the
electron is given by

However, the electron velocity has r and C#I components such as

at r = b, where b is the radius from the center of the cathode to the edge of the an-
ode, V = Vo, and drldt = 0, when the electrons just graze the anode, Eqs. (10- 1-5)
and (10-1-7) become

Substitution of Eq. (10- 1-8) into Eq. (10- 1-9) results in

The electron will acquire a tangential as well as a radial velocity. Whether the elec-
tron 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

This means that if Bo > Bo, for a given Vo, the electrons will not reach the anode.
Conversely, the cutoff voltage is given by

This means that if Vo < VO, for a given Bo, the electrons will not reach the anode.
Equation (10- 1- 12) is often called the Hull cutoff voltage equation.
430 Microwave Crossed-Field Tubes (M Type) Chap. 10

Example 10-1-1: Conventional Magnetron

An X-band pulsed cylindrical magnetron has the following operating parameters:

Anode voltage: Vo = 26 kV
Beam current: 10= 27 A
Magnetic flux density: Bo = 0.336 Wblm2
Radius of cathode cylinder: a = 5cm
Radius of vane edge to center: b = 10 cm

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

e
w, = -Bo = 1.759 x 10" x 0.336 = 5.91 X 101° rad
m
b. The cutoff voltage for a fixed BOis

= 139.50 kV
c. The cutoff magnetic flux density for a fixed Vo is

Cyclotron angular frequency. Since the magnetic field is normal to the

motion of electrons that travel in a cycloidal path, the outward centrifugal force is
equal to the pulling force. Hence
m y-2
- - eyB (10-1-13)
R
where R = radius of the cycloidal path
Ir = tangential velocity of the electron
The cyclotron angular frequency of the circular motion of the electron is then
given by
Sec. 10.1 Magnetron Oscillators

The period of one complete revolution can be expressed as

Since the slow-wave structure is closed on itself, or "reentrant," oscillations are pos-
sible only if the total phase shift around the structure is an integral multiple of 271
radians. Thus, if there are N reentrant cavities in the anode structure, the phase shift
between two adjacent cavities can be expressed as

where n is an integer indicating the nth mode of oscillation.

In order for oscillations to be produced in the structure, the anode dc voltage
must be adjusted so that the average rotational velocity of the electrons corresponds
to the phase velocity of the field in the slow-wave structure. Magnetron oscillators
are ordinarily operated in the 2. mode. That is,
4, = 2. (2.mode) (10-1-17)
Figure 10-1-3 shows the lines of force in the 2. mode of an eight-cavity magnetron.
It is evident that in the 2. mode the excitation is largely in the cavities, having oppo-
site phase in successive cavities. The successive rise and fall of adjacent anode-
cavity 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

The traveling-wave field of the slow-wave structure may be obtained by solving

Figure 10-1-3 Lines of force in n

mode of eight-cavity magnetron.
432 Microwave Crossed-Field Tubes (M Type) Chap. 10

Maxwell's equations subject to the boundary conditions. The solution for the funda-
mental 4 component of the electric field has the form [I]

where El is a constant and Po is given in Eq. (10-1-18). Thus, the traveling field of
the fundamental mode travels around the structure with angular velocity

where *
dt
can be found from Eq . ( 1 0- 1- 19).
When the cyclotron frequency of the electrons is equal to the angular fre-
quency of the field, the interactions between the field and electron occurs and the
energy is transferred. That is,

Power output and efficiency. The efficiency and power output of a mag-
netron depend on the resonant structure and the dc power supply. Figure 10-1-4
shows an equivalent circuit for a resonator of a magnetron.
where Ye = electronic admittance
V = RF voltage across the vane tips
C = capacitance at the vane tips
L = inductance of the resonator
G, = conductance of the resonator
G = load conductance per resonator

< v
c J Figure 10-1-4 Equivalent circuit for
Beam Resonator Load one resonator of a magnetron.

Each resonator of the slow-wave structure is taken to comprise a separate reso-

nant circuit as shown in Fig. 10- 1-4. The unloaded quality factor of the resonator is
given by

where oo= 27rfo is the angular resonant frequency. The external quality factor of
the load ciruit is
Sec. 10.1 Magnetron Oscillators

Then the loaded Qt of the resonant circuit is expressed by

The circuit efficiency is defined as

- Ge - 1
Gex 1 + Qex/Qun
The maximum circuit efficiency is obtained when the magnetron is heavily loaded,
that is, for Ge S- G,. Heavy loading, however, makes the tube quite sensitive to the
load, which is undesirable in some cases. Therefore, the ratio of Qe/Qex is often cho-
sen as a compromise between the conflicting requirements for high circuit efficiency
and frequency stability.
The electronic efficiency is defined as

where P,,. = RF power induced into the anode circuit

Pdc = VOIOpower from the dc power supply
Vo = anode voltage
lo= anode current
PloSt= power lost in the anode circuit
The RF power generated by the electrons can be written as
Pgen = VOIO- P~ost
m
= VOlO- lo-
2ep
+ ELax
-
BI

where N = total number of resonators

V = RF voltage across the resonator gap
Emax= M I I V I/L is the maximum electric field
MI = sin(&:)/(&:) = 1 for small 6 is the gap factor for the rr-mode
operation
p = phase constant
B, = magnetic flux density
L = center-to-center spacing of the vane tips
434 Microwave Crossed-Field Tubes (M Type) Chap. 10

The power generated may be simplified to

The electronic efficiency may be rewritten as

Example 10-1-lA: Pulsed Magnetron

An X-band pulsed conventional magnetron has the following operating parameters:

Anode voltage: vo = 5.5 kV

Beam current: l o = 4.5 A
Operating frequency: f =9x 1 0 9 ~ ~
Resonator conductance: G, = 2 x mho
Loaded conductance: Gp = 2.5 X mho
Vane capacitance: C = 2.5 pF
Duty cycle: DC = 0.002
Power loss: P~o,s= 18.50 kW

Compute:

a. The angular resonant frequency

b. The unloaded quality factor
c. The loaded quality factor
d. The external quality factor
e. The circuit efficiency
f. The electronic efficiency

Solution

a. The angular resonant frequency is

w, = 2 X 9 x lo9 = 56.55 x lo9 rad
b. The unloaded quality factor is

c. The loaded quality factor is

Sec. 10.1 Magnetron Oscillators

d. The external quality factor is

e. The circuit efficiency is

f. The electronic efficiency is

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 conven-
tional 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 dc 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 10-1-5 shows the state of the art for U.S. high-power magnetrons.

Figure 10-1-5 State of the art for U.S. high-power magnetrons.

436 Microwave Crossed-Field Tubes (M Type) Chap. 10

Beacon magnetrons-miniature conventional magnetrons-deliver peak out-

puts 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 vane-
strap magnetron oscillator. It has a maximum peak output power of 250 kW at a fre-
quency 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 Elec-

tron Tube Division.

10-1.2 Linear Magnetron

The schematic diagram of a linear magnetron is shown in Fig. 10-1-7. In the linear
magnetron as shown in Fig. 10-1-7, the electric field Ex is assumed in the positive x
direction and the magnetic flux density B, in the positive z direction. The differential
equations of motion of electrons in the crossed-electric and magnetic fields can be
written from Eqs. (1-3-2a, b, c) as

Anode
I
I

I Cathode
Figure 10-1-7 Schematic diagram of a
linear magnetron.
Sec. 10.1 Magnetron Oscillators

where = 1.759 X 10" Clkg is the charge-to-mass of an electron

m
B, = magnetic flux density in positive z direction
Ex = electric field in positive x direction
In general, the presence of space charges causes the field to be a nonlinear
function of the distance x, and the complete solution of Eqs. (10-1-30) through
(10- 1-32) is not simple. Equation (10- 1-3l), however, can be integrated directly.
Under the assumptions that the electrons emit from the cathode surface with zero
initial velocity and that the origin is the cathode surface, Eq. (10-1-31) becomes

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 elec-
trode spacing.
If the space charge is assumed to be negligible, the cathode potential zero, and
the anode potential VO,the differential electric field becomes

where Vo = anode potential in volts

d = distance between cathode and anode in meters
Substitution of Eq. (10- 1-34) into Eq. (10-1-30) yields

Combination of Eqs. (10-1-33) and (10-1-35) results in

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
Vo
x = -[I - cos (w,t)]
Bzwcd
Microwave Crossed-Field Tubes (M Type) Chap. 10
vo
[w,t - sin (act)] (10-1-38)
Y=B,w,d

e
where o, = -B, is the cyclotron angular frequency
m
f, = 2.8 X lo6 B, is the cyclotron frequency in Hz
Equations (10- 1-37) through (10- 1-39) are those of a cycloid generated by a
point on a circle of radius Vo/(B,o,d) rolling on the plane of the cathode with angu-
lar frequency o, . The maximum distance to which the electron moves in a direction
normal to the cathode is 2Vom/(Bled).When this distance is just equal to the anode-
cathode distnce d , the electrons graze the anode surface, and the anode current is cut
off. Then the cutoff condition is
2Vom
-- -d
Bled
Let a constant K equal to
d 2 B l 2m
K = - - - = 1.14 X lo-" (10-1-41)
Vo e
When the value of K is less than 1.14 X lo-", electrons strike the anode; when the
value is larger than 1.14 x lo-", they return to the cathode. Figure 10-1-8 shows
the electron path.

@t Anode

Bz d

Cathode
A Xi Figure 10-1-8 Electron path in a linear
magnetron.

From Eq. (10-1-40),the Hull cutoff voltage for a linear magnetron is given by

where Bo = B, is the magnetic flux density in the positive z direction. This means
that if Vo < Vo, for a given Bo , the electrons will not reach the anode.
Similarly, the Hull cutoff magnetic flux density for a linear magnetron is ex-
pressed as
I

This means that if Bo > Boc for a given Vo, the electrons will not reach the anode.
Sec. 10.1 Magnetron Oscillators 439

Example 10-1-2: Linear Magnetron

A linear magnetron has the following operating parameters:

Anode voltage: V, = 10 kV
Cathode current: lo= 1A
Magnetic flux density: Bo = 0.01 Wb/m2
Distance between cathode and anode: d = 5 cm

Compute:

a. The Hull cutoff voltage for a fixed Bo

b. The Hull cutoff mangetic flux density for a fixed V,

Solution

a. The Hull cutoff voltage is

Vo, = 4 X 1.759 X 10" X (0.01)~X (5 X

= 22.00 kV
b. The Hull magnetic flux density is

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

-' - - - - - - - - - -Anode
---- - - - - - - - -
T
I

-T
I
_e

I
---+
-
Beam bounddry

Cathode
------..-
T

t z J

Figure 10-1-9 Linear model of a magnetron

440 Microwave Crossed-Field Tubes (M Type) Chap. 10

where Bo = B, is the magnetic flux density in the positive z direction

V = potential
From the principle of energy conservation, we have

Combining Eqs. (10- 1-44) and (10- 1-45) yields

This differential equation may be rearranged as

Integration of Eq. (10-1-47) yields the potential within the electron beam as

where the constant of integration has been eliminated for V = 0 at x = 0. The po-
tential and electric field at the hub surface are given by

and

The potential at the anode is thus obtained from Eq. (10-1-50) as

VO = - I," Ex d x

= - 6" I Ex d x -
d

E,rdx

The electron velocity at the hub surface is obtained from Eqs. (10-1-44) and
(10-1-50) as
Sec. 10.1 Magnetron Oscillators 441

For synchronism, this electron velocity is equal to the phase velocity of the slow-
wave structure. That is,

For the T-mode operation, the anode potential is finally given by

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 10-1-2a: Linear Magnetron

A linear magnetron has the following operation parameters:

Anode voltage: VO= 15 kV

Cathode current: lo = 1.2A
Operating frequency: f = 8 GHz
Magnetric flux density: Bo = 0.015 Wb/m2
Hub thickness: h = 2.77 cm
Distance between anode and cathode: d = 5 cm

Calculate:

a. The electron velocity at the hub surface

b. The phase velocity for synchronism
c. The Hartree anode voltage

Solution

a. The electron velocity is

Q = 1.759 X 10" X 0.015 X 2.77 X lo-'
= 0.73 x lo3 m/s
b. The phase velocity is

c. The Hartree anode voltage is

442 Microwave Crossed-Field Tubes (M Type) Chap. 10

10-1-3 Coaxial Magnetron

The coaxial magnetron is composed of an anode resonator structure surrounded by

an inner-single, high-Q cavity operating in the T h l l mode as shown in Fig.
10-1-10.
The slots in the back walls of alternate cavities of the anode resonator structure
tightly couple the electric fields in these resonators to the surrounding cavity. In the

Electric field
in cavity
Unstrapped
resonators

.ouDlin/g slots' External

cavity

( a ) Cross section

Magnetic flux
Cavity mode

Stabilizing cavity
Vane
resonator

Cathode

Tuning
piston

attenuator

( b ) Cutaway view

Figure 10-1-10 Schematic diagram of a coaxial magnetron. (Courtesy of Varian

Associates, Inc.)
Sec. 10.1 Magnetron Oscillators 443

T-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 surround-
ing coaxial cavity stabilizes the magnetron in the desired r-mode oepration.
In the desired TEo,I 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 T b I Imode
is in the walls of the cavity in circular paths about the axis of the tube. The unde-
sired 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 load-
ing is lower, and voltage gradients are reduced.
The Varian SFD-333TM magnetron, as shown in Fig. 10- 1- 11 is a typical X-
band 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.

Figure 10-1-11 Photograph of Varian

SFD-333TM coaxial magnetron
(Courtesy of Varian Associates, Inc.)

10-1-4 Voltage-Tunable Magnetron

The voltage-tunable magnetron is a broadband oscillator with frequency changed by

varying the applied voltage between the anode and sole. As shown in Fig. 10-1-12,
the electric beam is emitted from a short cylindrical cathode at one end of the
device. Electrons are formed into a hollow beam by the electric and magnetic forces
near the cathode and then are accelerated radically outward from the cathode. The
electron beam is then injected into the region between the sole and the anode. The
beam rotates about the sole at the rate controlled by the axial magnetic field and the
dc voltage applied between the anode and the sole.
The voltage-tunable magnetron uses a low-Q resonator, and its bandwidth may
exceed 50% at low-power levels. In the T-mode operation, the bunch process of the
hollow beam occurs in the resonator, and the frequency of oscillation is determined
by the rotational velocity of the electron beam. In other words, the oscillation fre-
quency can be controlled by varying the applied dc voltage between the anode and
Microwave Crossed-Field Tubes (M Type) Chap. 10

Sole

Anc

Emitter-
View of anode from sole Figure 10-1-12 Cross-section view of
a voltage-tunable magnetron. (Courtesy
electrode'
of Varian Associates, Inc.)

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 band-
width 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 suppres-
sion 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
"0

Figure 10-1-13 Schematic diagram of

an inverted coaxial magnetron.
(Courtesy of Varian Associates, Inc.)
Sec. 10.1 Magnetron Oscillators 445

Mathematically, the motion equations of the electrons in an inverted coaxial

magnetron can be written from Eqs. (1-2-5a, b, c) as

where = 1.759 X 10" Clkg is the charge-to-mass ratio of electron

m
Bo = B, is assumed in the positive z direction
Rearrangement of Eq. (10- 1-56) results in the following form

where w, = + B , is the cyclotron angular frequency. Integration of Eq. (10-1-57)

m
yields

r 2d4
- = - 1 wcr2 + constant
dt 2

at r = b , where b is the radius of the cathode cylinder, and *dt

= 0, con-
1
stant = - - w,b2. The angular velocity is expressed by
2

Since the magnetic field does no work on the electrons, the kinetic energy of the
electron is given by
4mQ2 = eV (10-1-60)
However, the electron velocity has r and 4 components such as

at r = a , where a is the radius from the center of the cylinder to the edge of the an-
ode, V = Vo , and drldt = 0.
When the electrons just graze the anode, Eqs. (10-1-60) and (10-1-61) become
446 Microwave Crossed-Field Tubes (M Type) Chap. 10

Substitution of Eq. (10- 1-62) into Eq. (10- 1-63) results in

The electron will acquire a tangential as well as a radial velocity. Whether the elec-
tron will just graze the anode and return back toward the cathode depends on the rel-
ative magnitudes of the anode voltage Vo and the magnetic flux density Bo. The cut-
off condition can be obtained from Eq. (10- 1-64) as

This means that if Vo < Vo, 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

This means that if Bo > Bo, 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 cur-
rent density can be reduced to one-tenth of that used in cathode-centered mag-
netrons. Thus the millimeter magnetron is a practical and long-life device. The out-
put 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
I \
\ Cathode 1 Anode

Figure 10-1-14 Comparison of a con-

ventional magnetron with an inverted
sun magnetron
Inverted coaxial coaxial magnetron. (Courtesy of Varian
magnetron Associates, Inc.)
Sec. 10.1 Magnetron Oscillators 447

Example 10-1-5: Inverted Coaxial Magnetron

An inverted coaxial magnetron has the following parameters:

Anode voltage: VO = 10 kV
Cathode current: Io=2A
Anode radius: a = 3cm
Cathode radius: b = 4cm
Magnetic flux density: Bo = 0.01 Wb/m2

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

= 1.20 kV
b. The cutoff magnetic flux density is

10-1-6 Frequency-Agile Coaxial Magnetron

A frequency-agile coaxial magnetron differs from a standard tunable magnetron.

The frequency agility (FA) of a coaxial magnetron is defined as the capability to
tune the output frequency of the radar with sufficiently high speed to produce a
pulse-to-pulse frequency change greater than the amount required effectively to ob-
tain decorrelation of adjacent radar echoes. The frequency-agile magnetron, to-
gether with appropriate receiver integration circuits, can reduce target scintillation,
increase the detectability of target in a clutter environment, and improve resistance
to electronic countermeasures (ECM). The increase of the pulse-to-pulse frequency
separation will improve the radar system performance. Furthermore, the greater the
pulse-to-pulse frequency separation, the more difficult it will be to center a jamming
transmitter on the radar frequency for effective interference with system operation.
448 Microwave Crossed-Field Tubes (M Type) Chap. 10

The frequency-agile coaxial magnetrons are classified into three types:

1. Dither magnetrons: The output RF frequency varies periodically with a con-

stant excursion, constant rate, and a fixed center frequency.
2. Tunableldither magnetrons: The output RF frequency varies periodically with a
constant excursion and constant rate, but the center frequency can be manually
tuned by hand or mechanically tuned by a servomotor.
3. Accutune magnetrons: The output RF frequency variations are determined by
the waveforms of an externally generated, low-level voltage signal. With
proper selection of a tuning waveform, the accutune magnetron combines with
features of dither and tunableldither magnetrons, together with a capability for
varying the excursion, rate, and tuning waveform. Figure 10- 1- 15 shows an
accutune magnetron.

Figure 10-1-15 X-band accutune mag-

'd netron VMX-1430. (Courtesy of Varian

Associates, Inc.)

The X-band frequency-agile coaxial magnetron VMX-1430 is a typical agile

magnetron. Its pulse voltage is 15 kV, and pulse current is 15 A. Its maximum duty
cycle is 0.001 1, and accutune range is 1 GHz. Its center frequency is 9.10 GHz, and
peak output power is 90 kW. The agile rate and agile excursion are shown in Fig.
10-1-16.
From Fig. 10-1-16, it can be seen that the agile rate is the number of times per
second that the transmitter frequency traverses the agile excursion and returns to its
starting frequency. Similarly, the agile excursion is defined as the total frequency
variation of the transmitter during agile operation.
The number of pulses that can be effectively integrated cannot be greater than
the number of pulses placed on the target during one scan of the antenna. Therefore,
the antenna beamwidth and scan rate become factors that must also be considered in
determining the integration period of the radar. Consequently, a design value for
agile excursion can now be expressed in terms of radar operating parameters as
N
Agile excursion = - (10- 1-67)
7
Sec. 10.1 Magnetron Oscillators

Figure 10-1-16 Agile rate and agile

I excursion

where N = number of pulses placed on the target during one radar scan, say, 20,
whichever is smaller
T = shortest pulse duration used in the system

The frequency, or pulse repetition rate (PRR), is given by

where DC = duty cycle is the ratio of the pulse duration over the repetition period
for a pulse. The duty cycle is defined as
Pulse duration T
Duty cycle = --= ~ f (10-1-69)
Pulse repetition period T
Hence, the agile rate can be written as
1
Agile rate = -
2T
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

A frequency-agile coaxial magnetron has the following operating parameters:

Pulse duration: T =0.20, 0.40, 0.80 p s

Duty cycle: DC = 0.001
Pulse rate on target: N = 14 per scan

Determine:

a. The agile excursion

b. The pulse-to-pulse frequency separation
c. The signal frequency
d. The time for N pulses
e. The agile rate
Microwave Crossed-Field Tubes (M Type) Chap. 10

Solution

a. The agile excursion is

Agile excursion = l4 = 70 MHz

0.2 x 10-6
b. The pulse-to-pulse frequency separation is
1 1
= 5 MHz
fp = ;= 0.20 X 10-6
c. The signal frequency is
DC 0.001
= 5 kHz
= 7= 0.20 x 10-6
d. The time for 14 pulses per second is

e. The agile rate is

1
Agile rate = = 178.57 Hz
2 X 0.0028

10-2 FOR WARD-WAYE CROSSED-FIELD AMPLIFIER

(FWCFA OR CFA)

The crossed-field amplifier (CFA) is an outgrowth of the magnetron. CFAs can be

grouped by their mode of operation as forward-wave or backward-wave types and by
their electron stream source as emitting-sole or injected-beam types. The first group
concerns the direction of the phase and group velocity of the energy on the mi-
crowave circuit. This can be seen from the w-/3 diagram in Fig. 9-5-5 as discussed
in Sec. 9-5. Since the electron stream reacts to the RF electric field forces, the be-
havior of the phase velocity with frequency is of prime concern. The second group
emphasizes the method by which electrons reach the interaction region and how they
are controlled. This can be seen in the schematic diagrams of Fig. 10-2-1.
In the forward-wave mode, the helix-type slow-wave structure is often selected
as the microwave circuit for the crossed-field amplifier; in the backward-wave mode,
the strapped bar line represents a satisfactory choice. A structure of strapped
crossed-field amplifier is shown in Fig. 10-2-2.

10-2-1 Principles of Operation

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 Forward-Wave Crossed-Field Amplifier (FWCFA or CFA)

INPUT-OUTPUT SEPARATOR
AND DRIFT SPACE

Figure 10-2-1 Schematic diagrams of CFAs. (After J. F. Skowron [ 2 ] , reprinted

by permission of IEEE, Inc.)
452 Microwave Crossed-Field Tubes (M Type) Chap. 10

Figure 10-2-2 Diagram of a strapped CFA. (After J . F. Skowron 121, reprinted by

permission of IEEE, Inc.)

current is a function of the dimension, the applied voltage, and the emission proper-
ties of the cathode. The perveance of the interaction geometry tends to be quite
high, about 5 to 10 x lo-'', which results in a high-current and high-power capabil-
ity 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 dc velocity; the
dc 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 in-
teraction 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 dc 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 complex-
ity. 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 elec-
tron 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 Forward-Wave Crossed-Field Amplifier (FWCFA or CFA)

Figure 10-2-3 Motion of electrons in CFA. (After J. F. Skowron [Z], reprinted by permis-
sion of IEEE, Inc.)

in the negative half cycle, the electrons are returned toward the cathode. Conse-
quently, 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 re-
gion. That is, the power gain of a CFA is given by
Pout pln + pgen -
- 1 + -Pgen
g = P,, = P," Pin
where Po,, = P,, + Pgen
PI, = RF input power
P,,, = RF power induced into the anode circuit by electrons
Therefore, the CFA is not a linear amplifier but rather is termed a saturated
amplijier .
The efficiency of a CFA is defined as the product of the electronic efficiency q,
and the circuit efficiency v c . The electronic efficiency T, is defined as in Eq.
10-1-29). The overall efficiency is then expressed as
454 Microwave Crossed-Field Tubes (M Type) Chap. 10

where Pout = RF output power

Pin = RF input power
q e = Pgen/Pdc
Pdc = VdId is dc power
Vd = anode dc voltage
Id = anode dc current
The circuit efficiency is defined as

where P,,, = q, Vd Id
Since the power generated per unit length is constant, the output power is given by

where a =circuit attenuation constant

4! =circuit length in 4 direction
Substitution of Eq. (10-2-4) in Eq. (10-2-3) results in the circuit-efficiency
equation as

The term Pin/Pgen becomes negligible for high-gain CFA.

Assume that the input signal is sufficiently strong for spoke stability, that the
RF power grows linearly with distance along the circuit, that the dc current for
the spoke is constant, and that the back-bombardment loss is not considered, then
the electronic-efficiency equation can be derived as follows. The average drift elec-
tron velocity at any position is given by

where E =totall electric field at the position under consideration

B =magnetic flux density at the position
The power flow at any position is related to the RF field and the beam-coupling
impedance of the circuit by

where Em,, =peak electric field

p =o/V is phase constant
Z,. =beam-coupling impedance
Sec. 10.2 Forward-Wave Crossed-Field Amplifier (FWCFA or CFA) 455

The power loss per spoke due to the electron motion toward the anode at any posi-
tion is derived from Eq. (10-2-6) as

where V, =dc voltage per spoke

I, =dc current per spoke
Since the power varies linearly with position, the average power loss over the
entire circuit length is

By using Eqs. (10-2-I), (10-2-9), and 7rp = o,the total power loss for all the
spokes is given by

where g = power gain.

Finally, the electronic efficiency is given by

10-2-2 Microwave Characteristics

The crossed-field amplifier (CFA) is characterized by its low or moderate power

gain, moderate bandwidth, high efficiency, saturated amplification, small size, low
weight, and high perveance. These features have allowed the CFA to be used in a va-
riety of electronic systems ranging from low-power and high-reliability space com-
munications to multimegawatt, high average power, coherent pulsed radar. Figure
10-2-4 shows the present state of the art for the U.S. high-power forward-wave
crossed-field amplifiers.
The Raytheon QKS-1541 amplifier, shown in Fig. 10-2-5, is a typical
forward-wave CFA with an average power of 14 kW at frequency range from 2.9 to
3.1 GHz. Its pulse width is 28 s. Its typical peak anode voltage is 44 to 56 kV and
peak anode current is 58 A.

Example 10-2-1: Crossed-Field Amplifier

A CFA operates under the following parameters:

Anode dc voltage: Vd = 2 kV
Anode dc current: Id = 1.5 A
456 Microwave Crossed-Field Tubes (M Type) Chap. 10

Electronic efficiency: 7,= 20%

RF input power: Pin = 80 W

Calculate:

a. The induced RF power

b. The total RF output power
c. The power gain in decibels

Solution

a. The induced
J
RF power is
P,,, = 0.20 x 2 x lo3 x 1.5 = mw
b. The RF output power is
Po,, = P,, - P,,, = 80 + 600 = 680 W
c. The power gain is
Pout 680
g = - = - = 8.50 = 9.3 dB
P,, 80

FREQUENCY. CHI

Figure 10-2-4 State of the art for U.S. high-power CFAs.

Sec. 10.3 Backward-Wave Crossed-Field Amplifier (Amplitron)

Figure 10-2-5 Photograph of Raytheon

QKS- 1541 CFA (Courtesy of Raytheon
Company, Microwave Tube Operation.)

10-3 BACKWARD-WAVE CROSSED-FIELD AMPLIFIER

(AMPLITRON)

The trade name of the backward-wave crossed-field amplifier (BWCFA) is Am-

plitron, and its schematic diagram is shown in Fig. 10-3-1.

Magnetic flux Cathode

Figure 10-3-1 Schematic diagram of

Amplitron.

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-ps duration at S-band, and the tube gain reaches 8 dB.
The highly successful QK434 Amplitron produced by Raytheon exhibited sta-
ble 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, gener-
ated 425 kW of CW power at an efficiency of 76%. The gain was 9 dB and the band-
Microwave Crossed-Field Tubes (M Type) Chap. 10

Figure 10-3-2 Photograph of QK-434 Amplitron. (From W. C. Brown [3];

reprinted by permission of IEEE, Inc.)

width was 5% at a mean frequency of 3 GHz. The tube was used for all high-data-
rate transmission in the Apollo program. Figure 10-3-3 shows the two-stage
superpower Amplitron.

Figure 10-3-3 Photograph of a two-

stage superpower Amplitron. (From
W. C. Brown [3]; reprinted by permis-
sion of IEEE, Inc.)

The circuit and electronic equations for the M-type amplifier and oscillator
were developed by several authors [4, 51. The basic secular equation including the
space-charge effect is given by
( y 2 - y W P e - Y ) [(jPe - Y ) +~ PkI

where yo = circuit propagation constant

y = harmonic wave propagation constant
0 .
p = - is the electron-beam phase constant
olr,
I

g/, = J2
m
Vo is the dc electron-beam velocity
Sec. 10.3 Backward-Wave Crossed-Field Amplifier (Amplitron)
or e
&=%=% BOis the cyclotron phase constant

e
o,.= -BO is the cyclotron angular frequency
m
Bo = crossed-magnetic flux density
+ a2)42C3
(s)
HZ = 2(1

= "'is the gain parameter

4 =A exp ( - j y y ) + B exp ( j y y ) is the wave equation
A exp ( - j w ) - B exp ( j y y ) - . 1 d 4 .
a = - J - - is a factor
A exp ( - j w ) + B exp ( j y y ) y 4 dy
In general, there are five solutions of y from Eq. (10-3-1 ) . Let
YO = jP (10-3-2)
Y = jP(1 + P ) (10-3-3)
where p is a very small constant ( p 4 1 ) .
Substituting Eqs. (10-3-2) and (10-3-3) in Eq. (10-3-1) we have

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 Be A p is assumed, the first factor is small and the solution for p becomes

If Be 2 pm P is assumed, the second factor is small and the solution for p is given
by

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),p is imaginary when
pe + pm = /3 and a is less than unity. This last condition comes from rewriting Eq.
(10-3-6) as
460 Microwave Crossed-Field Tubes (M Type) Chap. 10

The solution for an M-type backward-wave amplifier can be obtained by set-

ting
Y = -.@(I + P) (10-3-8)

The only possible solution for p + p, = p, is

This gives rise to increasing backward waves for all values of a except a = - 1. Be-
cause H a C 3 / ' , 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 be-
cause C3l2< C.

Example 10-3-1: Amplitron Characteristics

An Amplitron has the following operating parameters:

Anode voltage: Vo = 15 kV
Anode current: Io=3A
Magnetic flux density: Bo = 0.2 Wb/m2
Operating frequency: f = 8GHz
Characteristic impedance: & = 50 R

Determine:

a. The dc electron-beam velocity

b. The electron-beam phase constant
c. The cyclotron angular frequency
d. The cyclotron phase constant
e. The gain parameter

Solution

a. The dc electron-beam velocity is

yo= 0.593 X lo6 X (15 X lo3)'/' = 0.762 X lo8 mls

b. The electron-beam phase constant is

c. The cyclotron angular frequency is

Sec. 10.4 Backward-Wave Crossed-Field Oscillator (Carcinotron)

d. The cyclotron phase constant is

e. The gain parameter is

10-4 BACKWARD- WAYE CROSSED-FIELD OSCILLATOR

(CARCINOTRON)

The backward-wave crossed-field oscillator of M-Carcinotron has two configurations:

linear M-carcinotron and circular M-carcinotron.

Linear M-Carcinotron. The M-Carcinotron oscillator is an M-type

backward-wave oscillator. The interaction between the electrons and the slow-wave
structure takes place in a space of crossed field. A linear model of the M-
Carcinotron oscillator is shown in Fig. 10-4-1 .

RF signal RF
output termination
Accelerator
I Slow-wave structure

....
f
Collector Figure 10-4-1 Linear model of an M -
Carcinotron oscillator. (From J. V .
Gewartowski and H . A . Watson [6];

v, ", - reprinted by permission of Van Nostrand

Company .)

The slow-wave structure is in parallel with an electrode known as the sole. A

dc electric field is maintained between the grounded slow-wave structure and the
, negative sole. A dc magnetic field is directed into the page. The electrons emitted
from the cathode are bent through a 90" angle by the magnetic field. The electrons
interact with a backward-wave space harmonic of the circuit, and the energy in the
circuit flows opposite to the direction of the electron motion. The slow-wave struc-
ture is terminated at the collector end, and the RF signal output is removed at the
electron-gun end. Since the M-Carcinotron is a crossed-field device, its efficiency is
very high, ranging from 30 to 60%.
The perturbed electrons moving in synchronism with the wave in a linear M-
Carcinotron are shown in Fig. 10-4-2.
Electrons at position A near the beginning of the circuit are moving toward the
circuit, whereas electrons at position B are moving toward the sole. Farther down
Microwave Crossed-Field Tubes (M Type) Chap. 10

Circuit

Sole

Figure 10-4-2 Beam electrons and electric field lines in an M-Carcinotron.

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 to-
ward 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 posi-
tion 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
W at frequency range from 1.0 to 1.4 GHz.

Collector
RF attenuator I Gfid Accelerating

RF output

- Electron beam
RF wave

Figure 10-4-3 Schematic diagram of a circular M-Carcinotron. (Courtesy of

Raytheon Company, Microwave Tube Operation.)
Sec. 10.4 Backward-Wave Crossed-Field Oscillator (Carcinotron)

Figure 10-4-4 Photograph of Litton L-

3721 BWO. (Courtesy of Litton Com-
L.3721 7%' Wide pany, Electron Tube Division.)

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 har-
monic.
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. (10-3- l), we
write
YO = jP (10-4- I)

By eliminating negligible terms, we obtain for the Carcinotron

A solution of Eq. (10-4-3) for synchronism can be obtained by setting P = P, and

pe - k = P,b ', where b ' is a small number so that terms like b '* and b ' E may be
neglected. This yields
464 Microwave Crossed-Field Tubes (M Type) Chap. 10

where

b ' = bD (10-4-7)
6(6 - jb) = -1 or a2 - jb6 + 1=0 (10-4-8)
5
As a result, we have reduced the number of waves to two, with propagation
constants given by

where the 6's are the roots of Eq. (10-4-8) and they are

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 = t. It follows that at y = 0,
the voltage at the input point can be computed as follows:

Solving Eqs. (10-4-1 1) and (10-4-12) simultaneously we have

Then the voltage at the output point y = t! is given by

V (0) = VI(O)exp(- y ~ t )+ V2(0) exp(- y2e) (10-4- 15)
= V (0)[6, exp(- y l t ) - a2 exp(-y2t)]l(6, - 6 2)
The term in the square bracket is the inverse of the voltage gain of the device. Oscil-
lation takes place when this is zero. That is,
Sec. 10.4 Backward-Wave Crossed-Field Oscillator (Carcinotron)

= exp[ - PeDe(6z - 62)]

From Eq. (10-4-8) we have

and
62 - = jub2 +4
Then

Equations (10-4-18)and (10-4-20)can only be satisfied simultaneously if b = 0 and

61 = -62. Then

where n = any integer numbers. If we introduce N as usually defined through

pet = ~ T Nthe
, oscillation condition becomes

Example 10-4-1: Carcinotron Characteristics

A circular carcinotron has the operating parameters:

Anode voltage: VO = 20 kV
Anode current: 10 = 3.5 A
Magnetic flux density: Bo = 0.3 Wb/m2
Operating frequency: f = 4GHz
Characteristic impedance: Zo = 50 R
D factor: D = 0.8
b factor: b = 0.5

Compute:

a. The dc electron velocity

b. The electron-beam phase constant
c. The delta differentials
d. The propagation constants
e. The oscillation condition
Microwave Crossed-Field Tubes (M Type) Chap. 10

Solution

a. The dc electron velocity is

To= 0.593 x lo6 x (20 x 103)1'2= 0.8386 X 10' m/s
b. The electron-beam phase constant is

c. The delta differentials are

d. The propagation constants are

e. The oscillation occurs at

DN=1.25 forn=l
then N = 1.5625

and

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. Perga-
mon Press, New York, 1958.
[6] GEWAFTOWSKI, J. V., and H. A. WATSON,Principles of Electron Tubes, P. 391. D . Van
Nostrand Company, Princeton, N. J., 1965.
Chap. 10 Problems

PROBLEMS

Magnetrons
10-1. Describe the principle of operation for a normal cylindrical magnetron and its charac-
teristics.
10-2. A normal cylindrical magnetron has the following parameters:

Inner radius: R, = 0.15 meter

Outer radius: Rh = 0.45 meter
Magnetic flux density: Bo = 1.2 milliwebers/m2

a. Determine the Hull cutoff voltage.

b. Determine the cutoff magnetic flux density if the beam voltage Vo is 6000 V .
10-3. It is assumed that in a normal cylindrical magnetron the inner cylinder of radius a
carries a current of loin the z direction (i.e., I = leu,) and the anode voltage is Vo.
The outer radius is b. Determine the differential equation in terms of the anode
voltage Vo and the current l o .
10-4. Compare the cutoff conditions for an inverted cylindrical magnetron (i.e., the inner
cathode voltage is Voand the outer anode is grounded) with a normal cylindrical mag-
netron. It is assumed that the magnetic field does no work on the electrons.
10-5. It is assumed that electrons in an inverted cylindrical magnetron leave the interior of
the coaxial cathode with initial velocity caused by thermal voltage V, in volts. Find
the initial velocity required for the electrons to just hit the anode at the center con-
ductor.
10-6. It is assumed that electrons in an inverted cylindrical magnetron leave the interior of
the coaxial cathode with zero initial velocity. Find the minimum velocity for an elec-
tron to just graze the anode at the center conductor.
10-7. In a linear magnetron the electric and magnetic field intensities as shown in Fig. P10-
7 are given by

Determine the trajectory of an electron with an initial velocity Qo in the z direction.

468 Microwave Crossed-Field Tubes (M Type) Chap. 10

10-8. An X-band pulsed cylindrical magnetron has the following parameters:

Anode voltage: Vo = 32 kV
Anode current: 1, = 84 A
Magnetic flux density: Bo = 0.01 Wb/m2
Radius of cathode cylinder: a=6cm
Radius of vane edge to center: b = 12 cm

Compute:
a. The cyclotron angular frequency
b. The cutoff voltage for a fixed Bo
c. The cutoff magnetic flux density for a fixed Vo
10-9. An X-band pulsed conventional magnetron has the following parameters:

Anode voltage:
Anode current:
Operating frequency:
Resonator conductance:
Loaded conductance:
Vane capacitance:
Duty cycle:
Power loss:

Compute:
a. The angular resonant frequency
b. The unloaded quality factor Q,,
c. The loaded quality factor Qt
d. The external quality factor Q,,
e. The circuit efficiency
f. The electronic efficiency
10-10. A linear magnetron has the following parameters:

Anode voltage: Vo = 20 kV
Anode current: l o = 17 A
Magnetic flux density: Bo = 0.01 Wb/m2
Distance between cathode and anode: d = 6 cm

Calculate:
a. The Hull cutoff voltage for a fixed BO
b. The Hull cutoff magnetic flux density for a fixed Vo
10-11. A linear magnetron has the following parameters:

Anode voltage:
Anode current:
Operating frequency:
Chap. 10 Problems

Magnetic flux density: Bo = 0.01 Wb/m2

Hub thickness: h=3cm
Distance between anode and cathode: d = 6 cm

Compute:
a. The electron velocity at the hub surface
b. The phase velocity for synchronism
c. The Hartree anode voltage
10-12. An inverted coaxial magnetron has the following parameters:

Anode voltage: Vo = 30 kV
Cathode current: lo = 25 A
Anode radius: a = 2.5 cm
Cathode radius: b=5cm
Magnetic flux density: Bo = 0.01 Wb/m2

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: T = 0.30, 0.60, 0.90 ps

Duty cycle: DC = 0.001 1
Pulse rate on target: N = 15 per scan

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. Derive Eq. (10-2-4).
10-15. Derive the circuit-efficiency Eq. (10-2-5).
10-16. Derive Eqs. (10-2-8), (10-2-10), and (10-2-1 1).
10-17. A CFA has the following operating parameters:

Anode dc voltage: Vd = 3 kV
Anode dc current: Id=3A
Electronic efficiency: 7,= 25%
RF input power: Pin= 100 W

Compute:
a. The induced RF power
470 Microwave Crossed-Field Tubes (M Type) Chap. 10

b. The total RF output power

c. The power gain in dB
10-18. A CFA operates under the following parameters:

Anode dc voltage: Va0 = 1.80 kV

Anode dc current: I d = 1.30A
Electronic efficiency: 7,= 22%
RF input power: P,. = 70 W

Calculate:
a. The induced RF power
b. The total RF output power
c. The power gain in dB

Arnplitrons
10-19. An Amplitron operates under the following parameters:

Operating frequency: f = 9GHz

Anode voltage: Vo = 20 kV
Anode current: lo= 3.5 A
Magnetic flux density: Bo = 0.3 Wb/m2
Characteristic impedance: ZO= 50 Cl

Compute:
a. The dc electron-beam velocity
b. The electron-beam phase constant
c. The cyclotron angular frequency
d. The cyclotron phase constant
e. The gain parameter
10-20. An Amplitron has the following operating parameters:

Operating frequency: f = 10GHz

Anode voltage: VO= 25 kV
Anode current: Io=4A
Magnetic flux density: Bo = 0.35 Wb/m2
Characteristic impedance: Zo = 50 Cl

Calculate:
a. The dc electron-beam velocity
b. The electron-beam phase constant
c. The cyclotron angular frequency
d. The cyclotron phase constant
e. The gain parameter

Carcinotrons
Chap. 10 Problems

10-21. A circular Carcinotron operates under the following parameters:

Operating frequency: f =8GHz

Anode voltage: Vo = 25 kV
Anode current: Io=4A
Magnetic flux density: Bo = 0.35 Wb/m2
Characteristic impedance: Zo = 50 ohms
D factor: D = 0.75
b factor: b = 0.50

Calculate:
a. The dc electron velocity
b. The electron-beam phase constant
c. The delta differentials
d. The propagation constants
e. The oscillation condition
Chapter 11

Strip Lines

1 1-0 INTRODUCTION

Prior to 1965 nearly all microwave equipment utilized coaxial, waveguide, or paral-
lel strip-line circuits. In recent years-with the introduction of monolithic micro-
wave integrated circuits (MM1Cs)-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 [I], are described.

Strip
conductor
/

Ground plane

( a ) Schematic diagram (h) Field lines

Figure 11-0-1 Schematic diagrams of strip lines.

Sec. 11.I Microstrip Lines

11-1 MICROSTRIP LINES

Chapter 3 described and discussed conventional transmission lines in detail. All

electrical and electronic devices with high-power output commonly use conventional
lines, such as coaxial lines or waveguides, for power transmission. However, the mi-
crowave solid-state device is usually fabricated as a semiconducting chip with a vol-
ume on the order of 0.008-0.08 mm3. The method of applying signals to the chips
and extracting output power from them is entirely different from that used for
vacuum-tube devices. Microwave integrated circuits with microstrip lines are com-
monly used with the chips.The microstrip line is also called an open-strip line. In
engineering applications, MKS units have not been universally adopted for use in
designing the microstrip line. In this section we use either English units or MKS
units, depending on the application, for practical purposes.
Modes on microstrip line are only quasi-transverse electric and magnetic
(TEM). Thus the theory of TEM-coupled lines applies only approximately. Radia-
tion loss in microstrip lines is a problem, particularly at such discontinuities as short-
circuit posts, corners, and so on. However, the use of thin, high-dielectric materials
considerably reduces the radiation loss of the open strip. A microstrip line has an ad-
vantage over the balanced-strip line because the open strip has better interconnec-
tion features and easier fabrication. Several researchers have analyzed the circuit of a
microstrip line mounted on an infinite dielectric substrate over an infinite ground
plane [2 to 51. Numerical analysis of microstrip lines, however, requires large digital
computers, whereas microstrip-line problems can generally be solved by conformal
transformations without requiring complete numerical calculations.

--
1 1 1 1 Characferisfic Impedance of Microstrip Lines

Microstrip lines are used extensively to interconnect high-speed logic circuits in dig-
ital computers because they can be fabricated by automated techniques and they
provide the required uniform signal paths. Figure 11-1- 1 shows cross sections of a
microstrip line and a wire-over-ground line for purposes of comparison.
In Fig. 11- 1- 1(a) you can see that the characteristic impedance of a microstrip

Figure 11-1-1 Cross sections of (a) a microstrip line and (b) a wire-over-ground
line.
474 Strip Lines Chap. 11

line is a function of the strip-line width, the strip-line thickness, the distance be-
tween 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 51. 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. 1 1-1- 1(b), is given by

60 4h
Z 1
-- forh S - d

where E, =dielectric constant of the ambient medium

h = the height from the center of the wire to the ground plane
d = diameter of the wire
If the effective or equivalent values of the relative dielectric constant E, of the ambi-
ent 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

where p is the permeability of the medium and E is the permittivity of the medium.
In free space, the propagation-delay time is
Tdf = 6= 3.333 ns/m or 1.016 ns/ft (11-1-3)

where

In transmission lines used for interconnections, the relative permeability is 1. Con-

sequently, the propagation-delay time for a line in a nonmagnetic medium is
Td = 1 . 1 0 6 G 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 dis-
covered an empirical equation for the effective relative dielectric constant of a mi-
crostrip line by measuring the propagation-delay time and the relative dielectric con-
stant of several board materials, such as fiberglass-epoxy and nylon phenolic [6].
Sec. 11.I Microstrip Lines 475

The empirical equation, as shown in Fig. 11-1-2, is expressed as

E, = 0.4756, + 0.67 (11-1-5)
where E, is the relative dielectric constant of the board material and ereis the effec-
tive relative dielectric constant for a microstrip line.

Figure 11-1-2 Effective dielectric con-

stant as a function of relative dielectric
constant for a microstrip line. (After
H. R., Kaupp (21; reprinted by permis-
Relative dielectric constant. e, sion of IEEE, Inc.)

Transformation of a rectangular conductor into an equivalent circular

conductor. The cross-section of a microstrip line is rectangular, so the rectangu-
lar conductor must be transformed into an equivalent circular conductor. Springfield
discovered an empirical equation for the transformation [ 7 ] .His equation is

where d = diameter of the wire over ground

w = width of the microstrip line
t = thickness of the microstrip line
The limitation of the ratio of thickness to width is between 0.1 and 0.8, as indicated
in Fig. 11-1-3.

Characteristic impedance equation. Substituting Eq. (11-1-5) for the

dielectric constant and Eq. ( 1 1- 1-6) for the equivalent diameter in Eq. ( 1 1 - 1 - 1 )
yields

Z 1n [ 5'98h ] for (h < 0 . 8 ~ ) (11-1-7)

O
-
- a
87
0 . 8+
~ t
Strip Lines Chap. 11

0.1 Figure 11-1-13 Relationship between a

round conductor and a rectangular con-
0
0 0.1 0.3 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 ductor far from its ground plane. (After
H . R . Kaupp 121; reprinted by permis-
Ratio o l strlp thickness to w ~ d t h .r / 1 1 , sion of IEEE, Inc.)

where er = relative dielectric constant of the board material

h = height from the microstrip line to the ground
w = width of the microstrip line
t = thickness of the microstrip line
Equation (1 1-1-7) is the equation of characteristic impedance for a narrow mi-
crostrip line. The velocity of propagation is

The characteristic impedance for a wide microstrip line was derived by Assadourian
and others [8] and is expressed by
r

zo="----
w E
- 377
GW
for (w * h)
Limitations of Equation (11-1-7). Most microstrip lines are made from
boards of copper with a thickness of 1.4 or 2.8 mils (1 or 2 ounces of copper per
square foot). The narrowest widths of lines in production are about 0.005-0.010 in.
Line widths are usually less than 0.020 in.; consequently, ratios of thickness to
width of less than 0.1 are uncommon. The straight-line approximation from Eq.
(1 1-1-6) is an accurate value of characteristic impedance, or the ratio of thickness to
width between 0.1 and 0.8.
Sec. 11.I 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 inde-
pendent of frequency. The validity of Eq. (1 1- 1-7) is doubtful for values of dielec-
tric 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 fl to 150 fl, if the values
of the parameters vary from E, = 5.23, t = 2.8 mils, w = 10 mils, and h = 8 mils
to E, = 2.9, t = 2.8 mils, w = 10 mils, and h = 67 mils [2].

Example 11-1-1: Characteristic Impedance of Microstrip Line

A certain microstrip line has the following parameters:

E, = 5.23
h = 7 mils
t = 2.8 mils
w = 10 mils

Calculate the characteristic impedance Zo of the line.

Solution

-
1 1 1-2 Losses in Microstrip Lines

Microstrip transmission lines consisting of a conductive ribbon attached to a dielec-

tric sheet with conductive backing (see Fig. 11-1-4) are widely used in both mi-
crowave and computer technology. Because such lines are easily fabricated by
printed-circuit manufacturing techniques, they have economic and technical merit.

'tor

Figure 11-1-4 Schematic diagram of a

microstrip line.
478 Strip Lines Chap. 11

The characteristic impedance and wave-propagation velocity of a microstrip

line was analyzed in Section 11- 1- 1. The other characteristic of the microstrip line is
its attenuation. The attenuation constant of the dominant microstrip mode depends
on geometric factors, electrical properties of the substrate and conductors, and on
the frequency. For a nonmagnetic dielectric substrate, two types of losses occur in
the dominant microstrip mode: (1) dielectric loss in the substrate and (2) ohmic skin
loss in the strip conductor and the ground plane. The sum of these two losses may be
expressed as losses per unit length in terms of an attenuation factor a . From ordi-
nary transmission-line theory, the power carried by a wave traveling in the positive z
direction is given by

where Po = ( V+ 12/(2&) is the power at z = 0.

The attenuation constant a can be expressed as

where adis the dielectric attenuation constant and a, is the ohmic attenuation con-
stant.
The gradient of power in the z direction in Eq. (11-1-11) can be further ex-
pressed 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,

where a. is the conductivity of the dielectric substrate board.

Substitution of Eq. (1 1- 1-12) into Eq. (1 1-1- 11) results in

and

Dielectric losses. As stated in Section 2-5-3, when the conductivity of a

dielectric cannot be neglected, the electric and magnetic fields in the dielectric are
no longer in time phase. In that case the dielectric attenuation constant, as expressed
in Eq. (2-5-20), is given by
Sec. 11.1 Microstrip Lines 479

where u is the conductivity of the dielectric substrate board in ~ / c m This

. dielectric
constant can be expressed in terms of dielectric loss tangent as shown in Eq.
(2-5-17):
u
tan 8 = -
WE

Then the dielectric attenuation constant is expressed by

cud =
2
6 tan 0 Np/cm (11-1-17)

Since the microstrip line is a nonmagnetic mixed dielectric system, the upper dielec-
tric 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,
Puce1 and his coworkers [lo] modified Welch's equation [9]. The result is
r

In Eq. (1 1- 1-18) the conversion factor of 1 Np = 8.686 dB is used, E, is the effec-

tive dielectric constant of the substrate, as expressed in Eq. (1 1-1-5), and q denotes
the dielectric filling factor, defined by Wheeler [3] as

We usually express the attenuation constant per wavelength as

c u p 27.3 (E) tan 8

dB/&

A0
where A, = -and A. is the wavelength in free space, or
VL
C
A, = -and c is the velocity of light in vacuum.
fG
If the loss tangent, tan 8,is independent of frequency, the dielectric attenua-
tion per wavelength is also independent of frequency. Moreover, if the substrate
conductivity is independent of frequency, as for a semiconductor, the dielectric at-
tenuation per unit is also independent of frequency. Since q is a function of E, and
w l h , the filling factors for the loss tangent qa/cr, and for the conductivity q / G
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-
Strip Lines Chap. 11

E, = 105 (rutile)

-
----- Er=
1

50
, 1 1 1 , 1 1 1
-----_
1 [ , 1 1 , ,

-
Figure 11-1-5 Filling factor for loss
0.88 1 1 1 1 l l l l 1 1 1 l l 1 1 1 1 1 1 tangent of microstrip substrate as a func-
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 tion of w l h . (After R . A . Puce1 et al.
/ l o / ; reprinted by permission of IEEE,
Inc.)

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, sil-
icon and gallium arsenide, that are used for integrated microwave circuits. For de-
sign purposes, the conductivity filling factor, which exhibits only a mild dependence
on wlh, 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 conduc-
tors. 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 ex-
posed 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 den-
sity 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 dia-
gram 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 [lo]. Several re-
searchers [8] have assumed, for simplicity, that the current distribution is uniform
and equal to I/w in both conductors and confined to the region ( x ( < w/2. With
this assumption, the conducting attenuation constant of a wide microstrip line is
given by
8.686RS w
a,. = - dB/cm for - > 1
Zow h
Sec. 11.1 Microstrip Lines

Figure 11-1-6 Dielectric attenuation factor of microstrip as a function of w l h for

silicon and gallium arsenide substrates. (After R. A . Pucel et al. [lo]; reprinted by
permission of IEEE, Inc.)

Bottom of strip

Top of strip

Figure 11-1-7 Current distribution

on microstrip conductors. (After R. A .
Pucel et a l . [lo]; reprinted by permis-
sion of IEEE, Inc.)

where R, = *
L- is the surface skin resistance in fl/square,

d$ is the skin depth in cm

For a narrow microstrip line with w l h < 1 , however, Eq. (1 1- 1-21) is not ap-
plicable. The reason is that the current distribution in the conductor is not uniform,
as assumed. Pucel and his coworkers [ l o , 111 derived the following three formulas
from the results of Wheeler's work [3]:
482 Strip Lines Chap. 11

w 1
for - I- (11-1-22)
h 2%-

and
~-Z o h- 8.68 w' wr/(rh)
- -+
Rs r.i+:ln12~eig+o.94)rr;[ h %
W '+ 0 . 9 4

x
[
h
1 + 1 + 7
w ITW
(I n - - - L:) w
for 2 r -
h
where a, is expressed in dB/cm and

2t w
Aw =
IT
(1, + 1) for- < -
h h
IT
5 -
2
(11-1-26)

AW =
%-
(In 7+ 1) for -
h
W IT
2 -
2

The values of a, obtained from solving Eqs. (1 1- 1-22) through (1 1- 1-24) are plotted
in Fig. 11- 1-8. For purposes of comparison, values of a<based on Assadourian and
Rimai's Eq. (1 1- 1-21) are also shown.

Radiation losses. In addition to the conductor and dielectric losses, mi-

crostrip 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 dieIectric 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

Figure 11-1-8 Theoretical conductor attenuation factor of microstrip as a function

of w l h . (After R. A . Puce1 et al. [lo];reprinted by permission of lEEE, Inc.)

Lewin's results show that the ratio of radiated power to total dissipated power for an
open-circuited microstrip line is

where F (E,) is a radiation factor given by

-
+1
F (Ere)= --
E,, € - 1 VEre+ 1
ere 2Ere& ln G- 1
in which E,, is the effective dielectric constant and AO = c / f is the free-space wave-
length.
The radiation factor decreases with increasing substrate dielectric constant. So,
alternatively, Eq. (1 1- 1-28) can be expressed as
484 Strip Lines Chap. 11

where R, is the radiation resistance of an open-circuited microstrip and is given by

The ratio of the radiation resistance R, to the real part of the characteristic
impedance Zoof the microstrip line is equal to a small fraction of the power radiated
from a single open-circuit discontinuity. In view of Eq. (1 1- 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 dielectric-
constant substrates, radiation becomes significant until very low impedance levels
are reached.

-
1 1 1-3 Quality Factor Q of Micros trip Lines

Many microwave integrated circuits require very high quality resonant circuits. The
quality factor Q of a microstrip line is very high, but it is limited by the radiation
losses of the substrates and with low dielectric constant. Recall that for uniform cur-
rent distribution in the microstrip line, the ohmic attenuation constant of a wide mi-
crostrip line is given by Eq. (1 1-1-21) as

and that the characteristic impedance of a wide microstrip line, as shown in Eq.
(11-1-9), is
7

The wavelength in the microstrip line is

where f is the frequency in GHz.

Since Q, is related to the conductor attenuation constant by

where a, is in dBlA,, Qc of a wide microstrip line is expressed as

where h is measured in cm and R, is expressed as

7 7
Sec. 11.2 Parallel Strip Lines 485

Finally, the quality factor Q, of a wide microstrip line is

Qc = 0.63h (11-1-36)

where a is the conductivity of the dielectric substrate board in U/m.

For a copper strip, a = 5.8 x lo7U /m and Q, becomes

For 25-mil alumina at 10 GHz, the maximum Q, achievable from wide microstrip
lines is 954 [I 31.
Similarly, a quality factor Qd is related to the dielectric attenuation constant:

where a d is in dB/A, .
Substituting Eq. (1 1- 1-20) into Eq. (1 1-1-38) yields

Ao 1
Qd = < tan o
z-
tan 0

where A. is the free-space wavelength in cm. Note that the Qd for the dielectric at-
tenuation constant of a microstrip line is approximately the reciprocal of the dielec-
tric loss tangent 0 and is relatively constant with frequency.

11.2 PARALLEL STRIP LINES

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 ~ , d .

Figure 11-2-1 Schematic diagram

parallel strip line.
Strip Lines Chap. 11

11-2-1 Distributed Parameters

In a microwave integrated circuit a strip line can be easily fabricated on a dielectric

substrate by using printed-circuit techniques. A parallel stripline is similar to a
two-conductor transmission line, so it can support a quasi-TEM mode. Consider a
TEM-mode wave propagating in the positive z direction in a lossless strip line
( R = G = 0). The electric field is in the y direction, and the magnetic field is in the
x direction. If the width w is much larger than the separation distance d, the fringing
capacitance is negligible. Thus the equation for the inductance along the two con-
ducting strips can be written as

where pc is the permeability of the conductor. The capacitance between the two
conducting strips can be expressed as

where ed is the permittivity of the dielectric slab.

If the two parallel strips have some surface resistance and the dielectric sub-
strate has some shunt conductance, however, the parallel stripline would have some
losses. The series resistance for both strips is given by
I

where R, = ~ ( ~ f p , . ) / ais, the conductor surface resistance in fl/square and a, is

the conductor conductivity in U/m. The shunt conductance of the strip line is

G = -a d w U/m ( 1 1-2-4)
d
where adis the conductivity of the dielectric substrate.

1 1-2-2 Characteristic Impedance

-
The characteristic impedance of a lossless parallel strip line is
7

The phase velocity along a parallel strip line is

W 1 - 1 C
for p, = p~ (1 1-2-6)
" " = P = E - Gm/s
=iz
The characteristic impedance of a lossy parallel strip line at microwave frequencies
*
( R wL and G 4 wC) can be approximated as
Sec. 11.2 Parallel Strip Lines 487

377 d
for w S d

1 1-2-3 Attenuation Losses

The propagation constant of a parallel strip line at microwave frequencies can be ex-
pressed by
y = V(R+-jwL)(G +-jwC) for R < wL and G + wC

Thus the attenuation and phase constants are

and

Substitution of the distributed parameters of a parallel strip line into Eq.(ll-2-9)

yields the attenuation constants for the conductor and dielectric losses:
7 -

and

Example 11-2-1: Characteristics of a Parallel Strip Line

A lossless parallel strip line has a conducting strip width w. The substrate dielectric
separating the two conducting strips has a relative dielectric constant erd 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 R
b. The strip-line capacitance
c. The strip-line inductance
d. The phase velocity of the wave in the parallel strip line
Solution

a. From Eq. (1 1-2-5) the width of the conducting strip is

Strip Lines Chap. 11

b. The strip-line capacitance is

c. The strip-line inductance is

d. The phase velocity is

i 1-3 COPLANAR STRIP LINES

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 in-
creased, reproducibility is enhanced, and production cost is decreased.
The characteristic impedance of a coplanar strip line is

where lois the total peak current in one strip and Pa,, is the average power flowing in
the positive z direction. The average flowing power can be expressed as

Pavg = $Re (E x H*) .u, dx dy (1 1-3-2)

where Ex =electric field intensity in the positive x direction

H, =magnetic field intensity in the positive y direction
* =conjugate
Sec. 11.4 Shielded Strip Lines

Figure 11-3-1 Schematic diagram of a

coplanar strip line.

Example 11-3-1: Characteristic Impedance of a Coplanar Strip Line

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.
Solution From Eq. (1 1 -3-I), the characteristic impedance of the coplanar strip line is

11-4 SHIELDED STRIP LINES

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.

I
1 Figure 11-4-1 Partially shielded strip
line.

The characteristic impedance for a wide strip (wid P 0.35) [14] is

1
where K = -
1 - t/d
t = the strip thickness
d = the distance between the two ground planes
8.8546,
Cf = 7 [2K In ( K + 1) - ( K - 1) In ( K 2 - l)] and is the fringe capa-
citance in pF/m
490 Strip Lines Chap. 11

Figure 11-4-2 shows the characteristic impedance Zo for a partially shielded

strip line, with the t l d radio as a parameter.

Figure 11-4-2 Characteristic impedance Zo of a partially shielded strip line with

the t l d ratio as a parameter. (After S. Cohn 1141; reprinted by permission of IEEE,
Inc.)

Example 11-4-1: Characteristic Impedance of a Shielded Strip Line

A shielded strip line has the following parameters:

Dielectric constant of the insulator

(polystyrene): E, = 2.56
Strip width: w = 25 mils
Strip thickness: t = 14 mils
Shield depth: d = 70 mils

Calculate:

a. The K factor
b. The fringe capacitance
c. The characteristic impedance of the line.
Chap. 11 References

Solution

a. Using Eq. (11-4-I),the K factor is obtained:

b. From Eq. (11-4-l), the fringe capacitance is

= 15.61 pF/m.
c. The characteristic impedance from Eq. (11-4-1) is

REFERENCES

[l] LIAO,S. Y., Engineering Applications of Electromagnetic Theory, Chapter 3. West Pub-
lishing Co., St. Paul, Minn, 1988.
[2] KAUPP,H.R., Characteristics of microstrip transmission lines. IEEE Trans. on Elec-
tronic 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 cou-
pled 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., SimpliJied Theory of Microwave Transmission Systems, F . Assodourian
and E. Rimol. pp. 1651-1657, IRE Proceeding, December 1952.
[8] ASSODOURIAN, F., and E. RIMOL,Simplified theory of rnirowave 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).
[lo] 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

[ l 11 WCEL,R. A., D. J. MASSE,and C. P. HAK~WIG, Correction to "Losses in microstrip."

IEEE Trans. on Microwave Theory and Techniques, MTT-16, No. 12, 1064, December
1968.
[12] LEWIN,L., Radiation from Discontinuities in Strip-Line, IEEE Monograph No. 358E,
February 1960.
[13] VENDELINE, G. D., Limitations on stripline Q . Microwave J., 63-69, May 1970.
[14] COHN,S., Characteristic impedance of the shielded-strip transmission line. IRE Trans.
on Microwave Theory and Techniques, MTT-2, No. 7, 52, July 1954.

PROBLEMS

Microstrip Lines
11-1. A rnicrostrip line has the following parameters:

E, = 5.23 and is the relative dielectric constant of the fiberglass board material
h = 0.8 mils
t = 2.8 mils
w = 10 mils

Write a FORTRAN program to complete the characteristic impedance & of the line.
Use a READ statement to read in the input values, the F10.5 format for numerical
outputs, and the Hollerith format for character outputs.
11-2. Since modes on microstrip lines are only quasi-transverse electric and magnetic
(TEM), the theory of TEM-coupled lines applies only approximately. From the basic
theory of a lossless line, show that the inductance L and capacitance C of a microstrip
line are

L = -zo= - zo<
u C

and

c = -1= - <
zov zoc
where Zo = characteristic impedance of the microstrip line
u = wave velocity in-the microstrip line
c = 3 x 10' m/s, the velocity of light in vacuum
relative dielectric constant of the board material
E, =
11-3. A microstrip line is constructed of a perfect conductor and a lossless dielectric board.
The relative dielectric constant of the fiberglass-epoxy board is 5.23, and the line
characteristic impedance is 50 a.Calculate the line inductance and the line capaci-
tance.
11-4. A microstrip line is constructed of a copper conductor and nylon phenolic board. The
relative dielectric constant of the board material is 4.19, measured at 25 GHz, and its
thickness is 0.4836 mm (19 mils). The line width is 0.635 mm (25 mils), and the line
thickness is 0.071 mm (2.8 mils). Calculate the
Chap. 11 Problems

a. Characteristic impedance Zo of the microstrip line

b. Dielectric filling factor q
c. Dielectric attenuation constant ad
d. Surface skin resistivity R, of the copper conductor at 25 GHz
e. Conductor attenuation constant a ,
11-5. A microstrip line is made of a copper conductor 0.254 mm (10 mils) wide on a G-10
fiberglass-epoxy board 0.20 mm (8 mils) in height. The relative dielectric constant E,
of the board material is 4.8, measured at 25 GHz. The microstrip line 0.035-mm
(1.4 mils) thick is to be used for 10 GHz. Determine the
a. Characteristic impedance Zo of the microstrip line
b. Surface resistivity R, of the copper conductor
c. Conductor attenuation constant a,
d. Dielectric attenuation constant ad
e. Quality factors Q, and Qd

Parallel Striplines
11-6. A gold parallel stripline has the following parameters:

Relative dielectric constant of teflon: crd = 2.1

Strip width: w = 26 mm
Separation distance: d=5mm
Conductivity of gold: u, = 4.1 X 10' u / m
Frequency: f = 10GHz

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: = 2.25

Strip width: w = 25 mm
Separation distance: 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

11-8. A 50-R coplanar strip line has the following parameters:

Relative dielectric constant of alumina: erd = 10

Strip width: w = 4mm
Strip thickness: t=lmm
Strip Lines Chap. 11

TEM-mode field intensities: E, = 3.16 x lo3 sin (7) e-jpz

):(
HI = 63.20 sin e-Jpz

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: crd = 2.25
Strip width: w=2mm
Strip thickness: t = 0.5 mm
Shield depth: d=4mm

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: erd = 2.56
Strip width: w = 0.7 mm
Strip thickness: t = 1.4 mm
Shield depth: d = 3.5 mm

Determine the
a. K factor
b. Fringe capacitance
c. Characteristic impedance
Chapter 12

I Monolithic Microwave
Integrated Circuits

12-0 INTRODUCTION

Integrated circuits are a combination of active and passive elements that are manu-
factured by successive diffusion or ion implantation processes on a semiconductor
substrate. The active elements are generally silicon planar chips. The passive ele-
ments are either thin or thick film components. In thin films, a thin film of conduct-
ing (resistor) or nonconducting (capacitor) material is deposited on a passive insu-
lated 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 integra-
tion (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 lo5 com-
ponents 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 lo6 components, and a 1-megabit dynamic random-
access memory (DRAM) chip contains more than 2.2 X lo6 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
Monolithic Microwave Integrated Circuits Chap. 12

Discrete circuit where elements

are separately made

Inputs {

Silicon
Collector Base Multiple emitters dioxide

Integrated circuit for J , of (1)

I I N collector I \

I
I 1

P substrate

CAPACITOR
INDUCTOR 8 CONNECTOR M ~ A ~ ~ ~ ~ ~ , ! ~
CAPACITORS METALLIZATION

TOP
METAL

C r OR TI\
Microwave integrated circuit

Figure 12-0-1 Discrete circuit, integrated circuit and microwave integrated circuit. (From
M . Caulton et al. [ I ] ; reprinted by permission of IEEE, Inc.)
Sec. 12.1 Materials 497

by conducting wires, is now referred to as a discrete circuit. The word discrete

means separately distinct.
2. Integrated circuit (IC): The integrated circuit consists of a single-crystal chip of
semiconductor, typically 50 X 50 mils in cross section, containing both active
and passive elements and their interconnections.
3. Monolithic microwave integrated circuit (MMIC): The word monolithic is
derived from the Greek monos (single) and lithos (stone). Thus a monolithic
integrated circuit is built on a single crystal. Such circuits are produced by the
processes of epitaxial growth, masked impurity diffusion, oxidation growth,
and oxide etching. Monolithic integrated circuits, like conventional integrated
circuits, can be made in monolithic or hybrid form. However, MMICs are
quite different from the conventional ICs. The conventional ICs contain very
high packing densities, whereas the packing density of a typical MMIC is quite
low. An MMIC whose elements are formed on an insulating substrate, such as
glass or ceramic, is called a j l m integrated circuit. An MMIC, which consists
of a combination of two or more integrated circuit types, such as monolithic or
film, or one IC type together with discrete elements, is referred to as a hybrid
integrated circuit.

Monolithic microwave integrated circuits offer the following advantages over

discrete circuits:

1. Low cost (because of the large quantities processed)

2. Small size
3. Light weight
4. High reliability (all components are fabricated simultaneously, and there are
no soldered joints)
5. Improved reproducibility
6. Improved performance

MMICs are suitable for space and military applications because they meet the re-
quirements 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, lumped-
element (inductors and capacitors) circuits, and thin-film circuits. These three types
are discussed in Section 4.

12-1 MATERIALS

The basic materials for monolithic microwave integrated circuits, in general, are
subdivided into four categories:
498 Monolithic Microwave Integrated Circuits Chap. 12

1. Substrate materials-alumina, beryllia, ferrite/garnet, GaAs, glass, rutile, and

sapphire
2. Conductor materials-aluminum, copper, gold, and silver
3. Dielectric films-A1203, SiO, SiO2, Si3N4,and Ta205
4. Resistive films-Cr, Cr-SiO, NiCr, Ta, and Ti

-
12-1 1 Substrate Materials

A substrate of monolithic microwave integrated circuits is a piece of substance on

which electronic devices are built. The ideal substrate materials should have the fol-
lowing characteristics [2]:
1. High dielectric constant (9 or higher)
2. Low dissipation factor or loss tangent
3. Dielectric constant should remain constant over the frequency range of interest
and over the temperature range of interest
4. High purity and constant thickness
5. High surface smoothness
6. High resistivity and dielectric strength
7. High thermal conductivity
Table 12-1- 1 lists the properties of some popular substrates that have been used
for MMICs [3,4]. The selection of a substrate material also depends on the expected
circuit dissipation, the circuit function, and the type of circuit to be used.
TABLE 12-1-1 PROPERTIES OF SUBSTRATES

Relative
dielectric Thermal
tan 0 x 104 constant conductivity
Material at 10 GHz (er) K (W/cm°C) Applications

Alumina 2 10 0.3 Microstrip, suspended

substrate
Beryllia 1 6 2.5 Compound substrated
Ferritelgarnet 2 13-16 0.03 Microstrip, coplanar,
compound substrate
GaAs 16 13 0.03 High frequency, microstrip,
monolithic MIC
Glass 4 5 0.01 Lumped element
Rutile 4 100 0.02 Microstrip
Sapphire 1 9.3-11.7 0.4 Microstrip, lumped element

Source: From H. Sobol [4]; reprinted by permission of IEEE, Inc.

12-1-2 Conductor Materials

The ideal conductor materials for monolithic microwave integrated circuits should
have the following properties [2]:
Sec. 12.1 Materials

1. High conductivity
2. Low temperature coefficient of resistance
3. Good adhesion to the substrate
4. Good etchability and solderability
5. 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 possi-
ble 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 fllsquare without loss. The
choice of conductors is usually determined by compatibility with other materials re-
quired 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 12-1-2 PROPERTIES OF CONDUCTORS

Adherence
Coefficient to
Skin depth Surface of thermal dielectric
6 at GHZ resistivity expansion film or Method of
Material (~m) x 10-'gf) ((Y,/"c x lo6) substrate deposition

21 Poor Evaporation,
screening
18 Very poor Evaporation,
plating
15 Very poor Evaporation,
plating
26 Very good Evaporation
4.6 Good sputtering,
vapor
phase,
electron-
beam
evaporation
Mo 2.7 4.7 6.0 Good Electron-beam
evaporation,
sputtering
Cr 2.7 4.7 9.0 Good Evaporation
Ta 4.0 7.2 6.6 Very good Electron-beam
sputtering

Source: After H. Sobol [4]; reprinted by permission of IEEE, Inc.

500 Monolithic Microwave Integrated Circuits Chap. 12

the good conductor (about 10 p m thick) are required. Films of this thickness can be
achieved by evaporation or plating or by any of the standard thick-film processes.

12-1-3 Dielectric Materials

Dielectric materials are used in monolithic microwave integrated circuits for block-
ers, capacitors, and some couple-line structures. The properties of dielectric materi-
als 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, S O z ,
and TazOs 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 densi-
fying 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 non-
critical applications, such as bypass capacitors. In power microwave integrated cir-
cuits, capacitors may require breakdown voltages in excess of 200 volts. Such capac-
itors can be achieved with films on the order of 0.5 to 1.0 p m thick with low
probability of pin holes or shorts.

TABLE 12-1-3 PROPERTIES OF DIELECTRIC MATERIALS

Relative
dielectric Dielectric
constant strength Microwave
Material Method of deposition (Vfcm) Q
SiO Evaporation 6-8 4 x lo5 30
SiOz Deposition 4 1o7 100- 1000
Si3N4 Vapor phase, 7.6 10'
sputtering 6.5 1o7
AlzOz Anodization, 7-10 4 X lo6
evaporation
T~zOS Anodization, 22-25 6 X lo6 100
sputtering

Source: After H. Sobol 141; reprinted by permission of IEEE Inc.

12-1-4 Resistive Materials

Resistive materials are used in monolithic microwave integrated circuits for bias net-
works, terminations, and attenuators. The properties required for a good microwave
resistor are similar to those required for low-frequency resistors and should
be [51
Sec. 12.2 Monolithic Microwave Integrated-CircuitGrowth

1. Good stability
2. Low temperature coefficient of resistance (TCR)
3. Adequate dissipation capability
4. Sheet resistivities in the range of 10 to 1000 fl 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 incorporat-
ing chip components. The thickness of the thick film is in the range of 1 to 500 pm.
The term thickjlm 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 12-1-4 PROPERTIES OF RESISTIVE MATERIALS

Method of Resistivity TCR

Material deposition (Rlsquare) (%l°C) Stability

Cr Evaporation 10- 1000 -0.100-+O. 10 Poor

NiCr Evaporation 40-400 +0.002-+O. 10 Good
Ta Sputtering 5- 100 -0.010-+0.01 Excellent
Cr-SiO Evaporation -600 -0.005--0.02 Fair
or cermet
Ti Evaporation 5-2000 -0.100-+O. 10 Fair

Source: From H. Sobol [4]; reprinted by permission of IEEE, Inc

12.2 MONOLITHIC MICROWAVE INTEGRATED-CIRCUIT

GROWTH

Like lower-frequency integrated circuits, monolithic microwave integrated circuits

(MMICs) can be made in monolithic or hybrid form. In a monolithic circuit, active
devices are grown on or in a semiconducting substrate, and passive elements are
either deposited on the substrate or grown in it. In the hybrid circuit active devices
are attached to a glass, ceramic, or substrate, which contains the passive circuitry.
Monolithic integrated circuits have been successful in digital and linear applications
in which all required circuit components can be simultaneously fabricated. In most
cases, the same device, such as bipolar or metal-oxide-semiconductor (MOS) tran-
sistors, can be used for amplifiers, diodes, resistors, and capacitors with no loss in
performance. Many digital circuits used in computers require large arrays of identi-
cal devices. Thus the conventional ICs contain very high packing densities. On the
other hand, very few applications of microwave integrated circuits require densely
packed arrays of identical devices, and there is little opportunity to utilize active
devices for passive components.
502 Monolithic Microwave Integrated Circuits Chap. 12

Monolithic technology is not well suited to microwave integrated circuits be-

cause the processing difficulties, low yields, and poor performance have seriously
limited their applications. To date, the hybrid form of technology is used almost ex-
clusively 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 re-
sistivity of microwave integrated circuits should be much greater than 1000 a-cm
for good circuit performance.

12.2-1 MMlC Fabrication Techniques

Monolithic microwave integrated circuits (MMICs) can be fabricated by using differ-

ent techniques such as diffusion and ion implantation, oxidation and film deposition,
epitaxial growth, lithography, etching and photoresist, and deposition.

Diffusion and ion implantation. Diffusion and ion implantation are the
two processes used in controlling amounts of dopants in semiconductor device fabri-
cations. The process of diffusion consists of diffusing impurities into a pure material
in order to alter the basic electronic characteristics of the pure material. Ion implan-
tation is used to dope the substrate crystal with high-energy ion impurities. Both
processes are used to dope selectively the semiconductor substrate to produce either
an n- or p-type layer. Until 1970, selective doping was performed mainly by the dif-
fusion method at elevated temperatures. Since 1970, many doping operations have
been conducted by ion implantation. In this process the dopant ions are implanted
into the semiconductor by using a high-energy ion beam. The advantages of the ion-
implantation method are precise control of the total amount of dopants, the im-
provement of reproducibility, and reduced processing temperature. Both diffusion
and ion implantation can be used for fabricating discrete and integrated devices be-
cause these processes are generally complementary to one another.

Oxidation and film deposition. To fabricate discrete and integrated

devices or circuits many different types of thin films are used. There are four groups
of thin films:

1. Thermal oxides
2. Dielectric layers
3. Polycrystalline silicon
4. Metal films

Epitaxial growth. In epitaxy technology, single-crystal semiconductor lay-

ers grow on a single-crystal semiconductor substrate. The word epitaxy comes from
the Greek epi (on) and taxis (arrangement). The epitaxial process offers an impor-
tant means of controlling the doping profiles so that device and circuit performances
can be optimized. There are three types of epitaxy.
Sec. 12.2 Monolithic Microwave Integrated-CircuitGrowth 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 (-lo-' torr). MBE can achieve precise control in
both chemical composition and doping profiles. Single-crystal multilayer struc-
tures 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 sub-
strates by direct precipitation from the liquid phase. This process is particularly
useful for growing GaAs and related 111-V compounds. LPE is suited to grow
thin epitaxial layers ( 2 0 . 2 pm) 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 geomet-

ric 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. Electron-beam lithography
2. Ion-beam lithography
3. Optical lithography
4. X-ray lithography

Etching and photoresist. In the processes of making MICs, a selective re-

moval 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.
Poly tncri~cd
Ultraviolet radiation

Mask
Photoresist
SiO,.

(a) Masking and exposure to (b) Photoresist

ultraviolet radiation Figure 12-2-1 Photoetching process.

During the photolithographic process the substrate is coated with a uniform

film of Kodak photoresist (KPR), which is a photosensitive emulsion. A mask for
the desired openings is placed over the photoresist, and ultraviolet light exposes the
photoresist through the mask as shown in Fig. 12-2-](a). A polymerized photoresist
is developed, and the unpolymerized portions are dissolved by using trichloro-
ethylene after the mask is removed; see Fig. 12-2-l(b). The SiO2, which is not
covered by the photoresist, can be removed by hydrofluoric acid. The thick-film
504 Monolithic Microwave Integrated Circuits 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 evapora-

tion, and dc sputtering-are commonly used for making MMICs.
Vacuum Evaporation. Here the impurity material to be evaporated is placed
in a metallic boat through which a high current is passed. The substrate with a mask
on it and the heated boat are located in a glass tube in which a high vacuum at a pres-
sure of lop6to lop8 torr is maintained. The substrate is heated slightly while the heat
is evaporating the impurities, and the impurity vapor deposits itself on the substrate,
forming a polycrystalline layer on it.
Electron-Beam Evaporation. In another method of evaporating the impurity
a narrow beam of electrons is generated to scan the substrate in the boat in order to
vaporize the impurity.
dc Sputtering. The third method of vacuum deposition is known as dc sput-
tering or cathode sputtering. In a vacuum, the crucible containing the impurity is
used as the cathode and the substrate as the anode of a diode. A slight trace of argon
gas is introduced into the vacuum. When the applied voltage between cathode and
anode is high enough, a glow discharge of argon gas is formed. The positive argon
ions are accelerated toward the cathode, where they dislodge atoms of the impurity.
The impurity atoms have enough energy to reach the substrate and adhere to it.

12-2-2Fabrication Example

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 fol-
lowing:

1. Deposition. An oxide layer is deposited on the semiconductor material, and

then a photoresist layer is deposited to cover the oxide on top of the semicon-
ductor chip.
2. Mask. Ultraviolet light is used to shine through a precision photographic mask
to the photoresist.
3. Chemical etching. Chemical etching with hydrofluoric acid is used to remove
the selected oxide region.
4. Etching. The photoresist is finally dissolved with an organic solvent in the
oxide leaving the desired opening.

12-3 MOSFET FABRICA TlON

In recent years, the metal-oxide-semiconductor field-effect transistor (MOSFET) has

superseded the bipolar junction transistor in many electronic applications. This is be-
cause the structure of the MOSFET is simple and its cost is low. The MOSFET is the
Sec. 12.3 MOSFET Fabrication
Ultraviolet light
1 Transparent

Oxide

mask (a negative)
( a ) Deposition ( b ) Mask

( c ) Chemical etching ( d ) Etching

Figure 12-2-2 Illustration of photoresist technique. (After P. R. Nanavari [ 6 ] ; courtesy of

Intext Educational Publishers.)

most important device for very large-scale integrated circuits (VLSICs) such as mi-
croprocessors and semiconductor memories. Its basic fabrication processes can be
described in three areas: MOSFET formation, NMOS growth, and CMOS develop-
men t .

12-3-1 MOSFET Formation

MOSFETs 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 SiOz layer on the surface.
2. Diffusion: Open two windows by using the photoresist technique and difise
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 SiOz
covers the top surface.
5. Deposition: Deposit phosphorous glass over the surface to cover the oxide
layer.
506 Monolithic Microwave Integrated Circuits Chap. 12

I p-Type Si substrate I I I' I I 11 I

( a ) Oxitiation ( b ) n-Type diffusion ( c ) Masking, etching

( d ) Oxidation ( e ) F o r m a t ~ o nof ( f ) Masking and etching

phosphorous glass

(g) Metall~~at~on ( h ) Maaking and etching

Figure 12-3-1 Fabrication steps for making MOSFETs. (After P. R. Nanavan

[ 6 ] ; courtesy of Intext Educational Publishers.)

6. Etching: Open two windows above the two nc-type diffused regions by using
the photoetching method.
7. Metallization: Now see that aluminum metallization is carried out over the en-
tire 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
MOSFET compared to the three required for the bipolar junction transistor. There-
fore, MOSFET fabrication is more efficient and less expensive than the BJT. These
attributes make MOSFET integrated circuits attractive.

12-3-2 NMOS G ~ wth

o

The n-channel MOS (NMOS) logic gate was discussed in Section 6-5-1, and its fab-
rication processes are described as shown in Fig. 12-3-2.
Sec. 12.3 MOSFET Fabrication 507

Isolation mask Buried contact mask

0 c2=a7
A A

i Resist i Boron implant i Boron ~ m p l a n t

' Nitride

Gate oxide
~ C h a stop
n p-Substrate

Depletion implant mask

Resist
f--- --------.
J * -r - 7

Window mask
m

P-Glass
EMD, EMD, DMD f

FOX

p-Substrate
\ Buried
contact
I
Figure 12-3-2 NMOS fabrication processes. (After L. C. Parrillo 171; reprinted
by permission of AT&T Bell Laboratories.)
508 Monolithic Microwave Integrated Circuits Chap. 12

1. Deposition and implantation: The starting p-type substrate is lightly doped

and then an oxide (SiO*) layer is grown on the top of the substrate. A silicon-
nitride (Si3N4)layer is deposited on the oxide surface. An isolation mask is
used to define the active areas covered by SiOz - Si3N4,and the isolation or
field areas are etched by plasma or reactive ion etching.
2. Implantation: Boron ions are implanted as the channel stop to prevent inver-
sion under the field oxide (FOX). The wafer is then put in an oxidation furnace
to grow a thick layer of FOX.
3. Implantation: After the nitride-oxide layers are cleaned, a thin gate oxide
(about 20 nm thick) is formed. Using the photoresist to mask the enhancement-
mode device (EMD), an n-channel implant is made to form the depletion-
mode device (DMD).
4. Deposition: The polysilicon is deposited and patterned as the gates. The gates
are also used as the self-aligned mask for source and drain arsenic implanta-
tion.
5. Metallization: Metal films are evaporated and etched to produce the electrode
contacts.

12-3-3 CMOS Development

The CMOS device was described in Section 6-5-2, and its fabrication processes are
explained as follows (see Fig. 12-3-3).

1. Epitaxy: The starting material is a lightly doped n epitaxy over a heavily

doped n' substrate.
Deposition: A composite layer of SiOz (pad) and SiN4 (nitride) is defined,
and silicon is exposed over the intended n-tub region. Phosphorus is implanted
as the n-tub dopant at low energy, and enters the exposed silicon, but is
masked from the adjacent region by the Si3N4 layer.
2. Implantation: The wafers are then selectively oxidized over the n-tub regions.
The nitride is stripped and boron is implanted for the p-tub.
3. Oxidation: The boron enters the silicon through the thin pad oxide but is
masked from the n tub by the thick SiOz layer there. All oxides are then
stripped and the two tubs are driven in.
4. Deposition: n+ polysilicon is deposited and defined, and the source and drain
regions are implanted.
5. Implantation: Phosphorus is selectively implanted into the n-channel source
and drain regions at a higher dose so that it overcompensates the existing
boron.
6. Deposition and metallization: A phosphorus glass layer is then deposited. Af-
ter windows are dry-etched in the P-glass, aluminum metallization is defined
using dry etching.
Sec. 12.3 MOSFET Fabrication

S13i.44
Jx
P 1, ,,-Tub 1
rxxx X' x x XI
X X X
f
X X x x X X X x X x x

( a ) Epitaxy and diposition

C _ _ - - _ _ _ _ _ - v-Epitaxy
_-__-_-- ----

X X X X X X X X X X X X X X

( b ) Implantation
I v I

(c) Oxidation

Field oxide

............
(d) Deposition
V

it .+I
X
Resist

n +
ttr;:x
I?

...... .............
xxxxx 11-Tub
/)-Tub
( e ) Implantation

( f ) Deposltlon and ----------

Figure 12-3-3 CMOS fabrication. (After L . C . Parrillo [7]; reprinted by permis-

sion of the Bell Laboratories.)
510 Monolithic Microwave Integrated Circuits Chap. 12

12-3-4 Memory Construction

As discussed in Section 6-5-3, there are seven types of memory devices depending
on their different structures. Their basic functions are to store in, read out, and write
out data. The fabrication processes of a memory are complicated. In this section,
some basic fabrication techniques are described.
The random access memory (RAM) device is the most basic memory, and it
can be subdivided into static RAM (SRAM) and dynamic RAM (DRAM). Among
memory chips, the RAM device has the highest component density per chip. In a
RAM any bit of information in a matrix of bits can be accessed independently. Indi-
vidual rows of memory bits are accessed by a conductive word line which may be a
diffusion, polysilicon, or metal line. Similarly, individual columns of bits in the
matrix are accessed by a bit line. The acronym RAM is generally used to refer to
randomly addressable memories into which data can be written and retrieved
indefinitely. The read-only-memory (ROM), however, stores data permanently but
cannot accept new information. Static RAMs retain their data indefinitely unless the
power to the circuit is interrupted. Dynamic RAMs require that the charge (data)
stored in each memory cell be "refleshed" periodically to retain the stored data. The
fabrication processes of memory devices are described as follows:

1. SRAM cell with transistor load

2. SRAM cell with resistor load
3. DRAM cell with storage capacitor
4. DRAM cell with double-level polysilicon layer

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 (115) 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 12-3-
4(a) shows a six-transistor (n-channel) cell, which uses a cross-coupled inverter pair
(flip-flop, TI 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 T5) transmits data into and out of the cell when the word and bit lines are simul-
taneously activated. The loads for the flip-flop are depletion-mode transistors (TI and
T2) with their sources and gates tied together as a NAND 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 ( @ V T ) .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 TI (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
-
Bit line Bit line
(All Vss (Al) (A0

I
-
I I 1 I
+I vcc
(diffusion)

Word line Word line

(polysilicon)

-
Bit line vss Bit line

Direct diffusion Polysilicon

to polysilicon
contact Diffusion-polysilicon
contact
r-7 AI-Si contact
L-J window

Figure 12-3-4 Structure of a single SRAM cell. (After R. W . Hunt [&I;reprinted by permis-
sion of John Wiley & Sons.)

Word line
Ion-implanted
polysilicon
Vcc polysilicon resistor
(first level) (second level)

T4

Polysilicon
gate
m-Substrate

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 Monolithic Microwave Integrated Circuits 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 (RI and R2) can be made in
the same single layer of polysilicon (gate and interconnect) by masking the polysili-
con 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 (R, and Rz), and (b) the device' cross section. The first-level polysilicon
is used for gate and routing power supply VCC.The second-level polysilicon is used
for resistor load directly over an active transistor. The connection to the drain and
Vcc 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.
Diffused Polysllicon
column line cell plate
I ROW line
Aluminum
Access transistor

1
I
77 1storage
row line

Capacitor

I
Column line

Access transistor
(polysilicon)
(b)

Transfer gate Storage gate

P +
p+ d= n + ! -
/ \
Source Inversion regions

Figure 12-3-6 Diagrams of DRAM cell with storage capacitor. (After L . C. Parrillo [7] and
R. W . Hunt [a]; reprinted by permission of the Bell Laboratories.)

Figure 12-3-6(a) shows the DRAM cell layout. A diffusion process forms the
bit line (sourceldrain) 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 A1 for bit lines in fabricating advanced
Sec. 12.3 MOSFET Fabrication 513

memory chips such as the 256-kilobit DRAM. To increase the charge-storage capac-
ity, the use of thinner gate insulators with higher dielectric contacts (e.g., Si3N4and
Ta205 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 cir-
cuitry, 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 Silicon
nitride dioxide
\ /

Second polys~licon
layer

Contact

Figure 12-3-7 Diagrams of double-level polysilicon DRAM cell. (After L . C.

Purrillo [ 7 ] and W . G . Oldham [ l o ] ; reprinted by permission of the Bell Lab.)
514 Monolithic Microwave Integrated Circuits Chap. 12

DRAM cell with double-level polysilicon layers. Figure 12-3-7 shows

three-dimensional diagrams of the fabrication processes for the double-level polysili-
con DRAM cell. Fabrication processes for the double-level polysilicon DRAM cell
follow:

1. Oxidation: The first step is to perform selective oxidation of silicon using sili-
con nitride-pad oxide layers as the oxidation mask in Fig. 12-3-7(a), (b), and
(c).
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 second-
level 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 do-
pant. A thick layer of silicon dioxide (P-glass) is deposited next, and the con-
tact 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.

12-4 THIN-FILM FORMATION

The choice of lumped or distributed elements for amplifier matching networks de-
pends on the operating frequency. When the frequency is up to X band, its wave-
length 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 mi-
crowave frequencies where stub inductors are physically too large [l l].

12-4-1 Planar Resistor Film

A planar resistor consists of a thin resistive film deposited on an insulating substrate.

Thin-film resistor materials are aluminum, copper, gold, nichrome, titanium, tanta-
lium, and so forth, and their resistivity ranges from 30 to 1000 ohms per square.
The planar resistors are essential for terminations for hybrid couplers, power com-
biners or dividers, and bias-voltage circuits. Some design considerations should be
aimed to include the following:
Sec. 12.4 Thin-Film.Formation 515

1. the sheet resistivity available

2. the thermal stability or temperature coefficient of the resistive material
3. the thermal resistance of the load
4. the frequency bandwidth

Planar resistors can be grouped into semiconductor films, deposited metal films, and
cermets.
Planar resistors based on semiconductors can be fabricated by forming an iso-
lated band of conducting epitaxial film on the substrate by mesa etching or by isola-
tion 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 pat-
tern 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.
Res~stor
Metdl

substrdte

(a) Implanted res~stor

( b ) Mesa resistor

Metal --+

-
Ef2?2s?
3 <p,< 100

Semi-insulating
substrate
Figure 12-4-1 Configurat~onsof planar
(c) Depos~tedres~stor resistors.

The resistance of a planar resistor, as shown in Fig. 12-4-2, can be expressed

as

R = -e p . ohms
wt

where I!? = length of the resistive film

w = width of the film
p, = sheet resistivity of the film in ohm-meter
t = film thickness in meters
Monolithic Microwave Integrated Circuits Chap. 12

Contacts

/ \

Figure 12-4-2 Thin film resistor.

When units of length t 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: Resistance of a Planar Resistor

A planar resistor has the following parameters:
Resistive film thickness: t = 0.1 pm
Resistive film length: e = lOmm
Resistive film width: w = 10 mm
Sheet resistivity of gold film: ps = 2.44 X lo-' 0-m

Calculate the planar resistance.

Solution The planar resistance is

12-4-2 Planar Inductor Film

Planar inductors for monolithic circuits can be realized in a number of configurations

some of which are shown in Fig. 12-4-3.
Typical inductance values for monolithic circuits range from 0.5 to 10 nH. The
equations of inductance are different for different configurations.

(a) Meander line ( b ) S line

73(c) Square spiral (d) Circular spiral

Figure 12-4-3 Configurations of planar
inductors.
Sec. 12.4 Thin-Film Formation 517

Ribbon inductor. The inductance of a ribbon inductor can be expressed as

[12]:

where t = ribbon length in mils

t = ribbon thickness in mils
w = ribbon width in mils

Round-wire inductor. The inductance of a round-wire inductor is given by

L = 5.08 X 10-3t[tn(t/d) + 0.3861 nH/mil (12-4-3)
where d = wire diameter in mils
t = wire length in mils
Circular spiral inductor. The inductance of a circular spiral inductor is ex-
pressed as
L = 0.03125 n2do nH/mil (12-4-4)
where do = 5di = 2.5n(w + s) in mils
n = number of turns
s = separation in mils
w = film width in mils
Figure 12-4-4 shows the schematic diagram of a circular spiral inductor.

F dFigure 12-44
O Circular spiral inductor.

Circular loop inductor. The inductance of a single-turn flat circular loop

inductor is given by
518 Monolithic Microwave Integrated Circuits Chap. 12

Square spiral inductor. The inductance of a square spiral inductor can be

written as
L = 8.5 A; n 5 / 3 nH (12-4-6)
where A = surface area in cm2
n = number of turns

Example 12-4-2: Calculation of a Planar Circular Spiral Inductor

A circular spiral inductor has the following parameters:
Number of turns: n = 5
Separation: s = 100 mils
Film width: w = 50 mils

Compute the inductance.

Solution The inductance is

12-4-3 Planar Capacitor Film

Two types of planar capacitors commonly used for MMICs are the metal-o,:le-
metal capacitor and the interdigitated capacitor, as shown in Fig. 12-4-5.

'Metals

( ( substrate
-
I Ground plane

(a) Metal-oxide-metal capacitor (b) Interdigitated capacitol

Figure 12-4-5 Schematic diagrams of planar capacitors.

Metal-oxide-metal capacitor. The metal-oxide-metal capacitor has three

layers; the middle dielectric layer is sandwiched by the top and bottom electrode lay-
ers as shown in Fig. 12-4-5(a). The capacitance can be expressed as

farads

where E, =8.854 X 10-l2 Flm

E, =relative dielectric constant of the dielectric material
4 = metal length
w = metal width
h = height of the dielectric material
Sec. 12.5 Hybrid Integrated-Circuit Fdbrication 519

lnterdigitated capacitor. The interdigitated capacitor consists of a single-

layer structure, and it can be fabricated easily on substrates as microstrip lines with
values between 0.1 and 15 pF as shown in Fig. 12-4-5(b). The capacitance can be
approximated as [ 131

where N = number of fingers

A, = 0.089 pFIcm is the contribution of the interior finger for h > w
A2 = 0.10 pFIcm is the contribution of the two external fingers for h > w
4! = finger length in cm
w = finger-base width in cm

Example 12-4-3: Computations of a Planar Capacitor

An interdigitated capacitor fabricated on a GaAs substrate has the following parame-
ters:
Number of fingers: N = 8
Relative dielectric constant of GaAs: E , = 13.10
Substrate height: h = 0.254 cm
Finger length: e = 0.00254 cm
Finger-base width: w = 0.051 cm
Compute the capacitance.
Solution The capacitance is

12-5 HYBRID INTEGRA TED-CIRCUIT FABRICA TlON

In the hybrid integrated circuit (HIC), semiconductor devices and passive circuit ele-
ments 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 el-
ements are generally single-layer metallization. The lumped elements are either fab-
ricated 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 elec-
tronic systems because they can offer higher reliability, greater reproducibility, bet-
ter performance, smaller size, and lower cost than conventional electronic circuits or
systems.
520 Monolithic Microwave Integrated Circuits 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 ele-
ments 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 com-
ponents 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 sub-

strate 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).

Substrate with evaporated seed metal After plating and etching

(thick photoresist form for plating)
( a ) Plate-through technique

V/
Cu, Au, Ag

Substrate with thick evaporated metal After etching

(thin photoresist for etch mask)
(b) Etchback technique

Figure 12-5-1 Fabrication of hybrid ICs. (After M. Kumar and I . J. Bahl [ 1 3 ] ;

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

[I] 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 cir-
cuits. 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 Publish-
ers, 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.
181 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] OHZONE, 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.
[lo] 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,ROBERTA., Design considerations for monolithic microwave circuits. IEEE
Trans. on Microwave Theory and Techniques, MTT-29, No. 6, 5 13-534, July 1981.
[I21 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 Monolithic Microwave Integrated Circuits 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
12-1. List the basic materials for MMICs.
12-2. List the basic characteristics required for an ideal substrate material.
12-3. List the basic properties provided by ideal conductor, dielectric, and resistive materials
used in MMICs.

MMIC Growth
12-4. Describe the MMIC techniques.
12-5. Explain the photoresist process.

MOSFET Fabrication
12-6. Describe the basic fabrication processes for MOSFETs.
12-7. Explain the NMOS growth.
12-8. Analyze the CMOS formation.
12-9. Describe the memory construction.

Thin-Film Formation
12-10. Describe the resistor-film growth.
12-11. Explain the inductor-film formation.
12-12. Discuss the capacitor-film development.

Hybrid MMlCs
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 Am1 CONDUCTIVITY a IN MHOS PER METER

Conductor u Insulator u
Silver Quartz
Copper Polystyrene
Gold Rubber (hard)
Aluminum Mica
Tungsten Porcelain
Zinc Diamond
Brass Glass
Nickel Bakelite
Iron Marble
Bronze Soil (sandy)
Solder Sands (dry)
Steel (stainless) Clay
Nichrome Ground (dry)
Graphite Ground (wet)
Silicon Water (distilled)
Water (sea) Water (fresh)
Ferrite (typical)
524 Appendix A

TABLE A-2 DIELECTRIC CONSTANT-RELATIVE PERMITTIVITY E,

Material E, Material E,

Air Sands (dry)

Alcohol (ethyl) Silica (fused)
Bakelite Snow
Glass Sodium chloride
Ice Soil (dry)
Mica (ruby) Styrofoam
Nylon Teflon
Paper Water (distilled)
Plexiglass Water (sea)
Polyethylene Water (dehydrated)
Polystyrene Wood (dry)
Porcelain (dry process) Ground (wet)
Quartz (fused) Ground (dry)
Rubber Water (fresh)

TABLE A-3 RELATIVE FERMEABILITY p,

Diamagnetic material P, Ferromagnetic material

Bismuth 0.99999860 Nickel

Paraffin 0.99999942 Cast iron
Wood 0.99999950 Cobalt
Silver 0.9999998 1 Machine steel
Paramagnetic material P, Ferrite (typical)
Transformer iron
Aluminum 1 ,00000065 Silicon iron
Beryllinum 1.00000079 Iron (pure)
Nickel chloride 1.00004 Mumetal
Manganese sulphate 1.0001 Supermalloy

TABLE A-4 PROPERTIES OF FREE SPACE

Velocity of light in vacuum c 2.997925 X lo8 meter per second

Permittivity eo 8.854 x 10-l2 farad per meter
Permeability po 47T x lo-' henry per meter
Intrinsic impedance 70 377 or 12077 ohms

TABLE Am5 PHYSICAL CONSTANTS

-- -- - -

Boltzmann constant k 1.381 X JI°K

Electronvolt eV 1.602 x 1 0 - ' 9 ~
Electron charge 9 1.602 X 10-I9C
Electron mass m 9.109 x 10-3'kg
Ratio of charge to mass elm 1.759 X 10-"C/kg
of an electron
Planck's constant h 6.626 X 10-34J - s
Appendix A 525

TABLE A-6 ENERGY GAP STRUCTURES OF SEMICONDUCTORS

Energy Expansion Hetero-

Energy Lattice gap coefficient junction Electron
Semi- gap cons$ants struc- at 300°K preferred Typical affinity
conductor (ev) (A) ture (x "C-I) doping dopants (ev)

Gee 9 S ~ I O Indirect P, As, Sb

Ge Indirect Al, Ga, In
GaAs Direct Se, Te
Ge Indirect Al, Ga, In
ZnSe Direct Al, Ga, In
Ge Indirect Al, Ga, In
ZnSe Direct Al, Ga, In
GaAs Direct Zn, Cd
AlAs Indirect Zn
GaAs Direct Se, Te
Gap Indirect Se, Te
Si Indirect Al, Cia, In
AlSb Indirect Se, Te/Zn, Cd
GaSb Direct Zn, Cd/Se, Te
GaSb Direct Se, Te
InAs Direct Zn, Cd
ZnTe Direct Cu
GaSb Direct Se, Te
ZnTe Direct Cu
InAs Direct Se, Te
ZnTe Direct Cu
AlSb Indirect Se, Te
CdTe Direct Li, Sb, P/I
PbTe Indirect Cl, Br/Na, K
CdTe Direct Li, Sb
InSb Direct Se, Te
ZnTe Direct Cu
CeSe(hex) Direct Cl, Br, I

Source: After Milnes & Freucht, 116. Reprinted with permission from Heterojunctions and Metul-
Semiconductor Junctions, Academic Press, N.Y., 1972.
TABLE A-7 PROPERTIES OF SEMICONDUCTOR MATERIALS
--

Melting
Density point
Eg
(eV)
PCL,
(cm2/V-see)
PP
(cd/V-sec)
P
(a-cm) Transition Doping Lattice (9) cr (g/cm3) 3("(

Si 1.11
Ge 0.67
SiC(a) 2.86
A1P 2.45
AlAs 2.16
AlSb 1.6
Gap 2.26
GaAs 1.43
GaSb 0.7
InP 1.28
InAs 0.36
InSb 0.18
ZnS 3.6
ZnSe 2.7
ZnTe 2.25
CdS 2.42
CdSe 1.73
CdTe 1.58
PbS 0.37
PbSe 0.27
PbTe 0.29
Appendix A

TABLE A-8 FIRST-ORDER BESSEL FUNCTION VALUES

x JI(~) x JI(x) x J I ~ ) x J I ~ ) x Jl(x)

Impurity concentration

Figure A-1 Resistivity versus impurity concentration for Si and GaAs.

Impurity concentration icm-')

(Reprinted from Sze's Physics of Semiconductor Devices, 1981, by permission of John Wiley & Sons, Inc.)

Figure A-2 Mobility for Ge, Si, and GaAs

Appendix B

UNITS OF MEASUREMENT

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 de-
scribed here:

1. dB-The decibel (dB) is a dimensionless number that expresses the ratio of two
power levels. It is defined as

The two power levels are relative to each other. If power level P2 is higher than
power level P I , dB is positive and vice versa. Since P = V2/R, when their
voltages are measured across the same or equal resistors, the number of dB is
given by

The voltage definition of dB has no meaning at all unless the two voltages un-
der consideration appear across equal impedances. Above 10 GHz the
impedance of waveguides varies with frequency, and the dB calibration is lim-
ited to power levels only. Table B-1 shows the conversion of voltage and
power ratios to dB.
2. dBW-The decibel above 1 watt (dBW) is another useful measure for express-
ing power level P2 with respect to a reference power level P I of 1 W. Similarly,
if the power level P2 is lower than 1 W, the dBW is negative.
530 Appendix B

TABLE 6-1 CONVERSION OF VOLTAGE AND POWER RATIOS TO DECIBELS

Voltage Power Voltage Power Voltage Power Voltage Power

ratio ratio -dB+ ratio ratio ratio ratio -dB+ ratio ratio
Appendix B 531

TABLE B-1 (CONTINUED)

Voltage Power Voltage Power Voltage Power Voltage Power

ratio ratio ratio ratio ratio ratio ratio ratio

0.126 2.818
,123 2.851
,120 2.884
,118 2.917
,115 2.951
,112 2.985
,110 3.020
,107 3.055
,105 3.090
,102 3.126
,100 3.162
,0977 3.199
,0955 3.236
,0933 3.273
,0912 3.311
,0891 3.350
,087 1 3.388
,0851 3.428
,0832 3.467
,0813 3.508
.0794 3.548
,0776 3.589
.0759 3.631
,074 1 3.673
,0724 3.715
,0708 3.758
,0692 3.802
,0676 3.846
,0661 3.890
,0646 3.936
,0631 3.981
,0617 4.027
,0603 4.074
,0589 4.121
,0575 4.169
,0562 4.217
,0550 4.266
,0537 4.315
.0525 4.365
,0513 4.416
.050 1 4.467
.0490 4.519
,0479 4.571
,0468 4.624
.0457 4.677
532 Appendix 6

TABLE 6.1 (CONTINUED)

Voltage Power Voltage Power Voltage Power Voltage Power

ratio ratio -dB + ratio ratio ratio ratio -dB+ ratio ratio

0.126 0.0159 18.0 7.943 63.10 0.106 0.0112 19.5 9.441 89.13
,125 ,0155 18.1 8.035 64.57 ,103 .0110 19.6 9.550 91.20
,123 ,0151 18.2 8.128 66.07 .I04 .0107 19.7 9.661 93.33
,122 ,0148 18.3 8.222 67.61 ,102 ,0105 19.8 9.772 95.50
.I20 ,0145 18.4 8.318 69.18 ,101 ,0102 19.9 9.886 97.72
.I19 .0141 18.5 8.414 70.79 ,100 .0100 20.0 10.000 100.00
,118 ,0138 18.6 8.511 72.44 lo-' 30 10'
,116 ,0135 18.7 8.610 74.13 lo-4 40 102 104
,115 ,0132 18.8 8.710 75.86 50 1O5
,114 ,0129 18.9 8.811 77.62 lo-3 lo-6 60 lo3 1O6
,112 ,0126 19.0 8.913 79.43 lo-7 70 1o7
,111 ,0123 19.1 9.016 81.28 lo-8 80 lo4 1O8
,110 ,0120 19.2 9.120 83.18 lo-9 90 lo9
,108 ,0118 19.3 9.226 85.11 lo-5 10-10 100 lo5 1010
,107 ,0115 19.4 9.333 87.10 lo-" 110 10"
10-l2 120 1O6 10l2

3. dBm-The decibel above 1 milliwatt (dBm) is also a useful measure of ex-

pressing power level P2 with respect to a reference power level PI of 1 milli-
watt (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. dBpV-The decibel above 1 microvolt (dBpV) is another dimensionless
voltage ratio in dB referred to a reference voltage of 1 pV. The field intensity
meters used for the measurements in the microwave region often have a scale
in dBpV, since the power levels to be measured are usually extremely low.
6. pV/m-Microvolts per meter (pV/m) are units of 10-6V/m expressing the
electric field intensity.
7. dBpV/m-The decibel above 1 microvolt per meter (dBpV/m) is a dimen-
sionless electric field intensity ratio in dB relative to 1pV/m. This unit is also
often used for field intensity measurements in the microwave region.
8. pV/m/MHz-Microvolts per meter per megahertz (pV/m/MHz) are units of
10-6V/m per broadband electric field intensity distribution. This is a two-
dimensional distribution, in space and in frequency.
9. dBpV/m/MHz-The decibel above 1 microvolt per meter per megahertz
(dBpV/m/MHz) is a dimensionless broadband electric field intensity distribu-
tion ratio with respect to 1 pV/m/MHz.
10. pV/MHz-Microvolts per megahertz per second of bandwidth (pV/MHz)
are units of 10-6V- s of broadband voltage distribution in the frequency do-
Appendix 6 533

main. The use of this unit is based on the assumption that the voltage is even-
lly distributed over the bandwidth of interest.
Figure B-1 shows the conversion of dB scales in power and voltage:
dBW = -30 + dBm dBV = -60 + dBmV
= -60 + dBpW = -120 + dBpV (B- 1)

dBW dBm dBuW dBV dBmV dBuV

dBV = -60 + dBmV

= -1 20 + dBwV
(a) Power scale (b) Voltage scale

Figure B-1 Conversion of dB scale in power and in voltage.