Microwave Materials PDF

Microwave Materials
Microwave Materials
EDITORS
V.R.K. Murthy
S. Sundaram
B. Viswanathan
Springer-Verlag Berlin Heidelberg GmbH

V.R.K. Murthy
Department of Physics
lIT Madras, Madras, INDIA
S. Sundaram
Microwave Millimeter Wave Wing
Defence Electronics Research Laboratory
Hyderabad, India
B. Viswanathan
Department of Chemistry
lIT Madras, Madras, India
Copyright @ 1994 Springer-Verlag Berlin Heidelberg

Originally published by Springer-Verlag Berlin Heidelberg New York in 1994
Softcover reprint of the hardcover 1st editioo 1994
All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording or otherwise, without the prior permission of the publisher
All export rights for this book vest exclusively with Narosa Publishing House.
Unauthorised export is violation of Copyright Law and is subject to legal action
ISBN 978-3-662-08742-8 ISBN 978-3-662-08740-4 (eBook)

DOI 10.1007/978-3-662-08740-4
Preface
Solid State Materials have been gaining importance in recent times

especially in the context of devices which can provide necessary
infrastructure and flexibility for various human endeavours. In this
context, microwave materials have a unique place especially in various
device applications as well as in communication networks. Various
technological developments are taking place in fine-tuning these materials
for specific applicatio"ns and in fixed band frequencies. Though the
science and technology of these materials has reached an advanced
stage, systematic attempts are still lacking in bringing all available
information in a single source. The present. volume is a modest attempt
in this direction, though it cannot be considered to be the one that
satisfies completely desired components and information required. The
editors have enlisted certain articles of interest in this area, especially
those dealing with measurement techniques, chapters dealing with
materials like Ferrites, YIGs, Radome and high Tc superconducting
materials which are of current interest. The editors are fully aware that
the coverages are not comprehensive either in scope or in depth. The
purpose of this volume is only to acquaint oneself of certain aspects of
a fast developing field. The editors will be grateful for any comments
or suggestions in this endeavour.
V.R.K. MURTHY
S. SUNDARAM
B. VISWANATHAN
Contents
Preface v
1. Materials and Processes in Microwave

Integrated Circuits Fabrication 1
T. Rs. Reddy
2. Materials and Technology for Microwave
Integrated Circuits 30
Bharathi Bhat and Shiban K. Koul
3. Metallization of Plastics by Electroless Plating 79
B. Viswanathan
4. Methods of Measurement of Dielectric Constant
and Loss in the Microwave Frequency Region 100
V.R.K. Murthy
5. Microwave Ferrites 112
a.p. Srivastava and Bijoy K. Kuanr
6. Microwave Lithium Ferrites 141
Pran Kishan
7. Single Crystal YIG and Allied Materials 164
T.R.N Kutty and S. Sundaram
8. Radome Materials 200
a.s. Mani
9. High Frequency Applications of High-T c
Superconductors 239
C.M. Srivastava
1
Materials and Processes in Microwave
Integrated Circuits Fabrication
T. Rs. Reddy
Hybrid Microelectronics Division
Defence Electronics Research Laboratory
Hyderabad. India
1. INTRODUCTION
Microwave Integrated Circuits (MICs), more precisely Hybrid Microwave
Integrated Circuits represent an important and significant step towards the
integration and miniaturization of the microwave subsystems and systems when
compared to those based on the waveguide/coaxial components. But the level
of integration is less than that encountered in the Monolithic Microwave Integrated
Circuits (MMICs). In spite of the advent of MMICs, however, MICs continued
to assert their place in the microwave systems due to some of their unique
features which cannot be fully met otherwise. Also MIC technology adopted
itself to the integration of MMICs functional chips in realising the microwave
systems. MICs span the range from printed circuit board where the integration
is limited to realising the conductor pattern, to the hybrid circuits (thin and
thick film) where some or all of the passive components are integrated along
with the conductor pattern. Since at microwaves frequencies the circuit dimensions
tend to be of the order of the wavelength, every element of the microwave
circuit influences its performance (Bhat and Koul1989 and Van Nie, Goedbloed
and Kersuzan 1981). This in tum reflects on the choice of the materials that
are used to fabricate the MICs. The quality requirements are more demanding
for the materials used for the fabrication of MICs than for the low frequency
circuits. In the present chapter efforts are made to highlight various aspects
related to some of the materials used for the fabrication of MICs.
2. TYPES OF MICs
Different types of circuit configuration have been used for the propagation of
microwaves and in realising various circuit elements such as inductors, capacitors
and resistors. Broadly these can be classified into two categories namely
distributed element and lumped element. While in the former case the realisation
of inductor and capacitor is related to specific configuration of the transmission
2 MICRowAVE MATERIALS
line, in the latter case these are realised as lumped elements analogous to those
in the low frequency circuits. 1n order that these behave as lumped elements
it should be ensured that their overall sizes do not exceed 1/10 of the wave-
length of the microwaves in the medium in which the elements are realised. In
real situation however the circuit tends to have mixed features while predominantly
tending to be either one of these types.
Most commonly employed circuit configurations include: (I) Suspended
line (strip line), (2) Microstrip line, (3) Coplanar line, (4) Slot line and (5)
Finline. Typical circuit configurations for some of these are shown in Fig. 1.
In almost all cases there is a conductor circuit pattern laid out on a dielectric
substrate. In some cases the passive components such as inductors, capacitors
and resistors are realised in situ and in others are attached to the circuit as
discrete components. Invariably the active components (devices) are attached
as separate elements and interconnections are made to the circuit lines
through one of the several means discussed below. Active devices can
Microstrip Line Strip Line
Coplanar
Slot Line
D Dielectric Substrate
_ Metallization
Fig. 1. Typical circuit configurations used for microwave integrated circuits.

Materials and Processes in MICs Fabrication 3
be discrete components such as diodes, transistors etc., as well as MMICs

(Reddy 1988).
During the fabrication of MICs a large variety of materials are used to
accomplish various tasks. These include the dielectric substrate, conductor and
resistor metallizations, materials for attaching the components to the circuit,
interconnecting materials, encapsulants etc. (Grovener 1989, Licari and Enlow
1988, Jowett 1982). Each of these will be discussed in detail in the following
sections.
3. SUBSTRATE
The substrate supporting the circuit is a dielectric with metallisation on either
side. The circuit pattern is printed on one or both sides by photolithographic
process. A brief description of the photolithographic fabrication process is in
order to appreciate the requirement of the materials for the substrate and the
metallization.
3.1 Photolithographic Fabrication

Figure 2 shows the flow chart of the photolithographic fabrication of circuit
pattern on the substrate. The substrate after proper cleaning is coated with a
thin layer of a photoresist. After pre-bake the substrate with the photoresist is
exposed to UV light through a photo mask which contains the replica of the
desired circuit pattern. After development, the circuit pattern is transferred on
to the photoresist exposing the metallization in the non-circuit area. This is
then removed by etching either chemically by dissolving the materials in the
spray of a suitable chemical reagent or by dry etching technique using plasma
with suitable reactive gases or gas mixtures. Chemical milling is a variant of
plasma etching in which the reactive ions/molecules/atoms are directed on to
the substrate in the form of a beam resulting in steep walls for the circuit lines
i.e. high etch factor. Photoresist is finally removed to reveal the delineated
circuit pattern on the substrate.
3.2 Dielectric Substrate

In contrast to the low frequency circuits the dielectric substrate not only provides
a support to the circuit pattern but also actively participates in the functioning
of the circuit. The dielectric constant and thickness of the substrate decide the
impedance of the transmission lines which are nothing but the conductor lines
printed on the dielectric substrates. For a given dielectric substrate the width
of the conductor pattern decide the impedance of the transmission line which
is typically 50 ohms. In addition, the material has to be of low loss. Tight
tolerances are demanded on the uniformity of the dielectric constant and thickness
over the substrate area in order to achieve the desired circuit performance. The
criticality increases with increase in the frequency.
The basic requirements for the substrate materials are:
1. The dimensional stability over the temperature of the operation.
2. Low dielectric loss.
3. Stable electrical properties such as dielectric constant.
4 MICRoWAVE MATERIALS
4. High DC resistance.
S. Easy machinability for sizing, making holes etc.
6. Provide smooth surface for metallization.
7. High thermal conductivity for high power applications.
8. Uniformity of these properties over the substrate area.
fmm:mi!m~m:m~ <- Metallization
Clean Substrate
<- Dielectric Base
W//##//////##//#M
<- Photoresist
~#//####///#/##//M
Photoresist Coating and

Pre-baking
W####////###///#/#
UV UGHT
1111111
~=:;:;::;:;:::;:=;;;:::;:;:;;=;;;;;;;;;! <- Photo Mask
W/##///##//#///////M
UV Exposure through
Photo Mask (Positive)
containing the Circuit
Pattern W#//###//////####n
Develop Photoresist
Pattern
and Post Bake
W////#///////#/#////#n:
Etching of the
Metallization
to delineate the
~//////##//////#//////#
Circuit Pattern
,.....""""_ _ _ _ _IiL......, <- Final printed

Photoresist Strip Circuit
and Cleaning
~////////#//////#/////h
Fig. 1. Photolithographic fabrication of microwave integrated circuits.

Substrate materials used in the fabrication of MICs can be broadly classified

into soft and hard substrates. Neither type meets all the requirements for the
ideal substrate. The selection of one type or the other depends on specific
application and the stringency of the requirement Table 1 lists properties of
these two types.
Table 1. Selected features of hard/soft substrates used for
microwave Integrated circuits
Feature Hard Soft
Dielectric Alumina, fused quartz Glass fiber reinforced (woven or

material berellia, aluminium nitride non-woven) or ceramic powder
f11led PTFE
Dielectric thickness 10 to 25 mil 1 to 250 mil
Dielectric constant 4-10 2-10
Minimum dissi-
pation factor see Table 2 0.0009 @ 10 GHz
Volume
resistivity see Table 2 > 1015 ohm em
Thermal conduc-
tivity see Table 2 2.8 BTU in/hr/f~/deg F
Maximum size
available 6" x 6" 36" x 48"
Sizing Cutting with diamond saw, Cutting with a shear, chiseling
scribing with diamond tool
or laser followed by dicing
Holes Ultrasonic or laser drilling High speed drilling with carbide
tools
Chiseling or punching
Metallization Large variety of multilayer Basically electroless/electroplated
metallization structures have or laminatedlrolled high purity
been evolved. Most common copper. Gold flash is provided to
being Cr/Au, Cr/Cu/Au, protect copper from environmental
NiCr/Au, TiW/Au, TaNl effects and to enable wire bonding
TiW/Au
Metallization Overall thickness 1/8 to 2 oz cu/sq ft
thickness 3 to 8 micron (4.5 to 74 micron)
Soft substrates are typically glass fibre reinforced and/or ceramic filled
(loaded) teflon based materials. The fillers or reinforcement materials control
the dielectric constant which ranges from 2.2 to 10. The teflon base provides
low dielectric loss. Mechanical working with these substrates is relatively easy
and lends itself for sizing and for making holes (Taconic Tech Topic). However,
when the thickness is small because of their flexible nature these substrates
need mechanical support which is normally provided by the aluminium base
(in some cases top as well) plate of the package. To provide perfect ground it
is necessary to hold the substrate to the base plate with screws at several
locations avoiding, of course, the circuit lines.
6 MICROWAVE MATERIALS
A variety of ceramic, glass or single crystal materials fall in the category of

hard substrates. Contrary to soft substrates hard substrates do not lend themselves
to easy mechanical working. Special methods like ultrasonic or laser machining
have to be used (Kugler and Culkin 1989). However due to their excellent
mechanical and electrical properties these are the preferred substrates in realising
high end microwave and millimetric wave circuits. 99.6% alumina is the most
common substrate material that is used with glazed or polish surfaces with
surface smoothness better than 3 micro inch CLA. Properties of different hard
substrates are summarised in Table 2 (Blum and Anzai 1989, Sunahara and
Takapatalci 1990).
Table 2. Typical properties of hard substrates used for microwave Integrated circuits
Property Alumina Fused Berellia Aluminium

quartz nitride
Composition 99.6% Si~ 99.9% A1N

All~ BeO
Dielectric constant 9.9 3.8 6.7 8.8
Dissipation factor @ 10 GHz 0.0001 0.0001 0.0002 0.0006
Volume resistivity (ohm em) >1014 >1015 1018 >1014
Dielectric strength (kV/mm) 22 SOO 14 14
Thermal expansion coefficient 6.3 0.5 6.4 4.5
(10-6 K-1)
Thermal conductivity (Wm-1 K-1) 38 1.4 280 170
Fluxural strength (MPa) 620 248 300
3.3 Thin Film Metallization

Metallization on the dielectric substrate is meant for realising the circuit pattern
or to provide a ground plane. A variety of metallization structures have been
employed depending on the requirement of the circuit as well as the nature of
the substrate. Important characteristics of the metallization are: (1) good
conductivity; (2) good adhesion to the substrate; (3) easy etchability; (4)
solderability; (5) wire bondable; (6) good surface finish; and (7) ability to
resist environmental degradation such as tarnishing etc.
For soft substrates the metallization is typically copper on both sides with
thickness ranging from 1/8 to 2 oz per sq ft i.e., 4.5 to 74 microns. The
metallization is applied on to the dielectric either by electroless deposition
followed by electroplating or by hot press lamination of a thin copper foil with
suitable pre-treatment of the surface of the dielectric followed by the application
of adhesive. Copper is of high purity which itself is formed by electrolessl
electro deposition on polished lead cylinders to provide smooth surface. Some
companies offer special laminations which include integral aluminium backing
plate to act as ground plane obviating the need for separate aluminium base
plate. This ensures perfect ground plane for the microwave circuits (Fig. 3). To
protect copper surface from the tarnishing effects due to the environment about
0.5 to 1 micron thick gold is eleCtroplated onto the copper metallization. This
is also necessary to enable wire bonding.
Copper (Electroplated/Rolled
1/8 to 20z)
Dielectric (4.5 to 500 mil)
Ground plane (Brass. Copper

or Aluminum)
Copper (ElectroplatedlRolled
1/8 to 2 oz)
Fig. 3. Commercial soft substrate (PTFE based dielectric) with integral
ground plane (MIs Rogers Corporation. USA).
Metallization structures employed on hard substrates are more complex for

the simple reason that no single metal meets all the requirements (Fig. 4).
0.5-1 Microns ~+-Conductorlayer2
200-300 A +-Barrier layer
3-7 Microns +-Conductor layer 1
200-300 A +-Barrier/Adhesive layer

200-3000 A +-Adhesive/resistive layer
200-625 Microns :.:-:-:.:-:-:-:.:.:-:.:-:-:-:.:.:.:.:-:-:-:-:-:.:-:-:.:.:-:-:.:-:-:.:-:-:.:-:.: +-Dielectric substrate
... .. . . . . . . . ... . .. . .. . . . ... . . .. . . . ... . . .

(Not to scale. Only top metallization shown)
Fig. 4. Generalised multilayer metallization scheme for microwave integrated circuits
on hard substrates (typical configurations with specifications are given in Table 3).
Adhesion of copper or gold to the substrates is poor. One has to provide an

intermediate base layer to improve the adhesion. Addition31 intermediate barrier
layer between the base layer and the top layer sometimes becomes necessary
to prevent inter-diffusion between these layers and to prevent the damage to
the base layer due to the attack by the moisture in the atmosphere which can
easily diffuse through the permeable top conductive layers. Resistance elements
in circuits are easily integrated on the hard substrates if a resistive layer is
incorporated in the metallization structures. These often play the role of the
base layer obviating the need for a separate one. Novel metallization structures
employing TiW have been envisaged for enhanced stability and environmental
hardness. Variety of metallization structures that are employed on the hard

substrates for realising MICs are given in Table 3. The characteristics of different
materials employed are summarised in Table 4.
Table 3. Typical specifications for commonly used metallization structures on

bard substrates for microwave Integrated circuits
Adhesive Resistive Barrierl Conductor Barrier Conductor

layer layer Adhesive layer layer 1 layer layer 2
(Angstroms) (Angstroms) (Angstroms) (Microns) (Angstroms) (Microns)
Cr/250 Au/3-8
Cr/250 Cu/3-7 Au/0.5-1
Cr/250 Cu/3-7 Ni!250 Au/0.5-1
NiCr/1500 Au/3-8
TiW/250 Au/3-8
TaN/3000 TiW/250 Au/3-8
Table 4. Properties of thin film materials used for conductor metallization
Mate- Functional Bulk Surface Skin Typical Interconnections

rial property resis- resistance depth thickness compatibility
tivity (ohms (Microns
fin GHz) fin GHz) Soldering Wire
(p.ohm em) bonding
Cu Conductor 1.7 O. 85xl0-2../f 2. 15"'f 3-7Jlm Excellent Possible@

Au Conductor 2.5 1.OOXI0-2../f 2.54"'f 3-7 Jlm Possible' Excellent
Al Conductor* 2.7 3-7Jlm No Very good
@ Gold plating is advised for better results.
# Leaching is a problem. Special solders are required.
* Does not require an adhesive layer.
Vacuum evaporation, Sputtering and Electroless/Electro deposition are the

common methods employed in the metallization of hard substrates. Choice of
the method to be employed is decided by the requirement, throughput and
finally the economics of the process (Holland 1965, Berry et al 1968, Maissel
and GIang 1970, Chopra and Kaur 1983, Stuart 1983).
Vacuum evaporation i& schematically illustrated in Fig. 5. In this, the material
to be deposited is heated to its evaporation temperature either by resistive heating
or by E-beam in a vacuum better than 10-5 Torr. The evaporated material is
deposited on to the substrate placed above the source material (line of sight). The
rate ofevaporation is controlled by the heating of the source materials (temperature).
E-beam evaporation is superior to resistive heating in several respects.
1. It is clean in the sense that the material itself acts as the container and
hence the contamination from the heater element or the container is
totally eliminated.
2. E-beam heating affords a better control of the rate of evaporation.
3. Refractory material such as tantalum also can be easily evaporated.

4. Large thicknesses can easily be achieved by the use of large size crucibles
holding large quantity of materials to be evaporated .
...... - - - ......... +- - - - - - SUBSTRATES

/ . . \ \ \ I I / I . . . . "-
\. \\\1//1 /
\.\.\ \ \ 1//1// <-- VACUUM CHAMBER
'~~--- ...-- - - ELECTRON BEAM
,...-.....- - - -..... <- - - BASE PLATE

,
,
'MATERIAL FOR EVAPORATION
!
TO HIGH VACUUM
Fig. 5. Schematic for E-beam evaporation.
Figure 6 shows a schematic representation of sputtering process. The chamber

is evacuated to better than 10-6 Torr and after thorough degassing. is back-
filled with argon to a pressure of 10-3 Torr. The argon ions produced in the
VACUUM ICHAMBER
WATER COOLm I SUBSlJA1t:S

CATHQDE I
\ I
-!-
\ ,
TARGET \ ,
(material to b..... .... \ ,
depOlfted)....~
.... , ,
''...l. /
RF
PLASMA- - - - -->~ht" SUPPLY
TO HIGH
<-- - ..
VACUUM L..=:==~;:::====== <-- GASES IN
ARGON
NEEDL£ VALVE NllROGEN
lHROTTlE VALVE OXYGEN
Fig. 6. Schematic fo, sputter deposition.

discharge set up between the two electrodes bombard the source material placed
on the cathode and dislodge (sputter) the material which in tum gets deposited
on the substrate placed on the anode which is closely placed. There are several
variants to this simple arrangement and systems. The discharge can be produced
by DC or AC (RF). The discharge can be enhanced by placing the plasma in
a magnetic field. In this case, it is referred to as magnetron sputtering. This
helps in improving rate of deposition for a given power. Gases other than
argon such as nitrogen, oxygen can be employed for reactive sputtering wherein
the gaseous ions react with the source (target) material to form a compound!
composite film. For example, tantalum target with oxygen results in tantalum
oxide film and the desired stoichemetry is achieved by adjusting the oxygen
partial pressure and the power.
Extremely well adhering films can be produced by sputtering due to the
simple fact that the energy of the sputtered atom is orders of magnitude more
than the evaporated ion. In either case, it is essential that the substrates are
heated to a temperature anywhere between 150°C and 300°C for improved
adhesion and to achieve films with good crystallinity and grain structure.
Throughput of the evaporation process is much higher than that in a sputtering
process. But this is not a real drawback with the advent of automatic cassette
to cassette loading through vacuum load locks. Another specific advantage of
sputtering is to be able to deposit composite, compound or alloy films true to
the composition existing in the source material, with the added ability to
manipulate the composition with reactive sputtering where possible.
Normally the sputtering or vacuum deposition is used to deposit the conductor
layer to a thickness of about 1 micron. Build up of this to 5-8 micron is done
by electroplating.
Electroless deposition followed by electroplating provide a cheaper alternative
for the metallization of the hard substrates. However, the quality of the film is
not as good as those prepared by sputtering or vacuum deposition. Good adhesion
is one of the main problems which in turn reflects on the difficulty in realising
fine line structures in the photolithographic fabrication~ Normally thin layer of
metallization is deposited by electroless process and the thickness is build up
by electroplating. Special formulations of electro less baths are available which
obviate the need for electroplating. Most of the commonly used metallization
structures can be deposited by the electroless technique. The surface of the subs-
trate is to be suitably prepared to initiate the electroless deposition. Typical process
steps for electroless deposition of copper on teflon substrate is given in
Table S (Prasad and Vijayaraghavan 1987). One of the main drawbacks of this
process is the difficulty in controlling the thickness of the film and its texture.
3.4 Thick Film Metallization

Thick film technology is a misnomer to the essentially print and frre technology,
and historically got its name due to the fact that the thickness of the film realised
by print and fire technology is of the order of 25 microns while those obtained
by evaporation and sputtering are of the order of 1 micron. Thick film technology
is a cheaper alternative to the thin film approach and is extensively used in the
MaJeriais and Processes in MICs Fabrication 11
Table 5. Process steps for electroless deposition
Operation Materials used Effect / Purpose
Cleaning Organic solventsIDetergents Removal of grease, dirt and

other foreign matter
Surface preparation Sodium naphthalene complex Surface becomes wettable with
water
Sensitization Stannous chloride + Concen- Adsorption of stannous
trated hydrochloric acid chloride on the surface
Activation Silver nitrate + Ammonium Deposition of silver on the
hydroxide sUrface through reduction by
adsorbed SnCl2
Electroless plating 1. Copper sulphate + Deposition of copper on
Trisodium citrate the surface
2. Sodium hydroxide
3. Formaldehyde
mix 1,2 and 3
fabrication of hybrid circuits in the lower frequency typically upto 1 GHz.

Briefly, the technology involves screen printing a conductive paste/ink on to
the substrate through a stainless steel or nylon screen/mesh on to which the
circuit pattern has been transferred using a masking material. After suitable
drying and pre-baking to drive away the solvents and other organic vehicles the
substrates are fired at temperatures of the order of 850-950°C. This leaves the
conductor pattern fused on to the substrate. Thick film paste/ink consists of the
conductor material in the form of fine particles or flakes suspended in a thixotropic
organic vehicle along with glass firt. Organic solvent is added to maintain the
desired consistency. Various processes that take place at different stages of proces-
sing are indicated in Fig. 7 (Holmes and Loasby 1976, Pitt 1981 and Haskard 1988).
The resulting film consists of the metal particles embedded in glass matrix
which binds itself to the substrate during the firing. The conductor particles
sinter together to form the conductor tracks but contains voids as well as glass
frit intrusions which is the very basis of adhesion. This puts a limit to the
suitability of thick film conductor tracks for microwave applications. Secondly,
as fired conductor tracks do not have the desired line definition due to the very
process of screen printing. Normal pastes can be used to realise circuits operating
upto 4 GHz by the use of photolithographic fabrication and etching for delineating
the circuit pattern instead of screen printing (Corkhill 1976 and Haskard 1988).
Special new formulations have been developed over the years which enabled
to push the frequency to the high end of microwaves region, i.e. upto 18 GHz.
These are the fritless pastes and organometallic formulations. The former is
conventional thick film paste which contains very little percentage of some
oxides instead of the usual glass frit (Corkhill 1976 and Haskard 1988). This
forms a very thin binding layer between the metal and substrate leaving behind
a better quality conductor track. Organo metallic formulations consist of the
metal ions complexed with an organic molecule which when fired is decomposed
to leave behind the metallization as a thin film. Contrary to the other thick film
formulations the thickness of the film obtained with organo metallic formulations
is of the order of 1 micron or less. Building up of the thickness can be achieved
by electroplating. As of now organo metallic formulations are available for
gold metallization (Haskard 1988).
In either case best results are obtained by screen printing the conductor over
the full substrate and realising the circuit pattern photolithographically.
Process step Effect Pattern
Screen printing Transfer of paste on

~
..................
:::::::::::::::
to the substrate
Settling at ambient Smoothenout the screen

Impression on the printed
fT .
..................
..................
...................
··"i1i"~
...................
'. .............
,, ..............
pattern
Pre-baking Removal of solvents

120°0
Firing schedule Organic vehicle burn-out

300-400°0
FIring schedule Fusing of the glass frit to

850-950°0 the substrate and sinterlng
of the metal particles
Fig. 7. Processes in thick film metal deposition.
3.5 Resistive and Dielectric Layers

As already mentioned resistive layers can easily be integrated on to the hard
substrates along with the conductor pattern (Siddall 1965 and Hughes 1987).
Historically NiCr (nichrome) is,the material used for the resistive layer though
to some extent chromium layer itself served the purpose. Tantalum nitride is
the newer material offering greater flexibility in controlling the sheet resistance
and better stability due to self passivation by the formation of thin oxide layer.
Tantalum oxinitride has been experimented with to arrive at low TCR values.
Added advantage being that these films as well as the dielectric layer tantalum
pentoxide (Ta20s) can be deposited by reactive sputtering with nitrogen and
oxygen using a single tantalum target. Table 6 summarises the characteristics
of different resistive films used in realising MICs.
Table 6. Properties of thin nlm resistor materials
Material Sheet TCR Remarks

resistance (ppmf'C)
(ohm per sq.)
Nichrome 10-500 < ±20 to ±150 Most popular and stable resis-
(Ni:Cr/80:20) tance material. But susceptible to
contamination
Tantalum upto 100 ±SO to ±150 Resistor trimming can be done
by partial anodising
Tantalum 30-200 - 60 to -200 Resistivity and TCR can be
nitride manipulated by incorporating
oxygen. Very stable resistors
Cr/SiO upto 600 < ±20 to ±200 Stable high resistivity cermets
Alz0 3 /NiCr/ 10-500 ... 0 Stable resistance and zero TCR

AlzO, sandwiches
Necessity for a dielectric layer is felt in realising certain circuit elements

such as parallel plate capacitors, cross overs particularly while realising a
spiral inductor. It is also used as a passivation layer for resistive elements.
Silicon dioxide and tantalum pentoxide are the most commonly used dielectric
materials while realising the MICs due to their low dielectric loss and ease of
deposition. The former can be deposited either by E-beam evaporation or
sputtering. The Ta20s layer is more conveniently deposited by reactive sputtering
using oxygen gas with tantalum target. Highly stable polymer films such as
polyimide and parylene are also used aslow dielectric constant alterna-
tives (Grovenor 1989). The characteristics of these materials are given in
Table 7.
4 ASSEMBLY
Having realised the circuit pattern on the substrate the next step involves
populating the printed circuit with discrete devices viz., the active components
like transistors, diodes and in some cases the passive components like capacitors,
resistors, inductors if these could not be conveniently integrated in the circuit.
Conventional leaded package types for these components that are used in the
Table 7. Properties of thin film dielectric materials
Material Dielectric Dissipation Dielectric Remarks

constant factor strength (V/cm)
SiO 2-6 0.01-0.04 1~ Precise control of composition

is difficult
Si~ 3.4 0.005 >1~ Most commonly used dielectric
for cross-overs, parallel plate
capacitor and passivation
TIlzO.s 21-25 0.001-0.01 >1~ Not suitable for the cross-overs,
but good for realising capacitors.
Process integration possible
while using TaN resistor elements
fabrication of the normal Printed Circuit Boards (PCBs) are not suitable for the
hybrids to get the full benefits of miniaturization. This is particularly so for
MICs which are meant to operate at higher frequencies. All the add on components
are to be surface mountable and at the high end of microwaves it is preferable
to integrate the passive components in the circuit and to use the active components
in the chip form. All the passive components are now available in the surface
mountable package or 'chip' form. Typical outline of some of these passive
components are shown in Figs. 8 and 9. In addition to the bare chip or 'die',
the active components also come in a variety of package types or with leads
attached (beam lead type) for use in MICs. Some of these types are shown in
Fig. 10.
The process of assembly of these components in the circuit involves: (a)
attaching the components on the circuit board to provide mechanical suppOrt,
and (b) to establish electrical contact between the component terminals and the
circuit conductor pattern.
Except in the case where the active components are used in the bare chip
form or in some types of passive chip components, the attachment of the
component terminals to the conductor circuit pattern for providing electrical
contact gives the necessary mechanical support and no additional mechanical
fixing is necessary. The bare chip mounted on the metallization provides the
mechanical support as well as electrical 'ground' or earthing or in some cases
electrical connection to active terminals of the component for example to the
collector terminal of transistor. The electrical connection between the connector
pads (bonding pads) on the die and the conductor lines on the circuit is done
by "wire bonding" a process by which fine dia wires or ribbons are attached
to the 'bonding' pads on the die and the conductor lines. Different methods
employed for attaching the components to the circuit and for interconnections
are summarised in Table 8 (Edwards 1991).
4.1 Soldering
This is one of the oldest and most common methods used for attaching the
discrete components to the circuit boards and a large varieties of materials and
GLASS
PASSIVATION
HIGH ADHESION
METALIZATION--+--
CERAMIC_---~-'C
SUBSTRATE
WRAP AROUND TERMINATIONS FOR SOLDER

OR CONDUCTIVE EPOXY
GLASS _ _ _ _ __ TION
PASSIVATION ~:::-:S-c:::::::~
CERA~.1IC-------')
SUBSTRATE
GOLD TOP TERMINATIONS FORWIREBONDING
Fig. 8. Outline of swface mount resistors (Mis Mini-Systems. Inc., USA).
techniques have been evolved over the years to suit the specific needs. Soldering
is a process by which two metal parts are joined together by the use of a low
melting point alloy which when melted forms an intermetallic compound with
either of the metal surfaces and bonds the two surfaces when cooled. Tin-lead
alloy has been the most commonly used solder for joining copper terminals and
lead commonly used in the electronic circuits. Important characteristics of
solder materials are:
1. Its melting point should be compatible with the components to be soldered
and the environment in which the circuit is to function.
2. It should be able to wet the surfaces to be soldered to enable intermetallic
diffusion/bonding to take place to establish good adhesion.
3. It should not leach or scavenge the materials being soldered. In this
regard the conventional tin-lead solders are not suitable for use with
gold metallization. The problem of leaching is overcome to a great extent
with the addition of 2% silver in the solder. Better alternatives are the
indium orland gold based solders but at a higher cost.
A large variety of solder materials have been evolved to meet specific requi-
rements. The characteristics of some typical solders used for the MIC applications
are listed in Table 9.
Different techniques used for carrying out soldering in MIC are: (1) Manually,
using soldering iron and (2) Reflow with hot gas, hot plate, IR, Vapour Phase,
Laser. Such methods as wave soldering and dip soldering, more common in the
case of PCB assembly, are not generally used in the fabrication of MICs.
_1~
,------------- -------------I-~
k .100 ~
-T
.035
~~-~~~
W .030
INDUCTOR
(DIMENSIONS IN INCHES)·
TERMINATIONS
ELECTRODES
(TYP DIMENSIONS mm)*

A - 0.25 T - 0.51
L - 2.0 W • 1.25
SURFACE MOUNTED CERAMIC MULTILAYER CAPACITOR
"NOT TO SCALE
Fig. 9. Outline of surface mount inductors and capacitors (MIs Philips Export
B.Y., Netherlands and MIs Piconics Inc., USA).
For good solder joints it is important that the surfaces to be soldered are
clean and free from oxide layers. To help the process of cleaning and to present
virgin surface to the solder material, flux materials are used along with the solder
material. Both resin based and water soluble (organic or inorganic) fluxes are
used. The latter are more active than the resin based fluxes but arc easily
removed by water cleaning which however is not always possible. Synthetic
W
---------7 \ ~
CI======~IL-~~~~========~r
0.05 3 - -t 0.Q15
0.12
r--' I
,I
l
..t'
0.25 4- ~ ~ 0.20 0.25
DIODE IN BEAM LEAD PACKAGE
(DIMENSIONS IN mm)'
GATE
SOURCE
A---...J
S
DRAIN
1.27,. [.51
f-6.35~ T
TYP
FET IN STRIPLINE PACKAGE

(DIMENSIONS IN mm)* 'NOT TO SCALE
Fig. 10. Outline of surface mount active components (MIs California Eastern
Laboratories, USA and MIs Avantek Inc., USA).
Table 8. Methods of assembly/Interconnection
Components with leads/tenninals Components in the fonn of chip/die with

bonding pads
(a) Die attachment

Solder Solder
Eutectic
Adhesive Adhesive
Welding (b) Wire/ribbon bonding
Table 9. Characteristics of some typical solders used for MICs
Solder material Melting point Remarks

("C)
Au 97%, Si 3% 370 Eutectic solder

Au 80%, Sn 20% 280 Suitable for use with gold metallization
In 80%, Ph 15%, Ag 5% 149 -do-
In 75%, Ph 25% 163 -do-
In 50%, Ph 50% 209 -do-
Sn 62%, Ph 36%, Ag 2% 179 -do-
Sn 63%, Ph 37% 183 Standard solder. Not suitable for gold
metallization
Sn 5%, Ph 95% 314 Not suitable for gold metallization
activated fluxes combine the high activity level typical of water soluble organic
fluxes with the benefit of solvent cleanability. The solvent cleaners normally
comprise of a blend of polar and nonpolar solvents like propyl alcohol and
chlorinated/fluorinated hydrocarbons.
The form in which the solder materials are used depends on the technique
used for the soldering. Table 10 summarises the same. The solder preform
mainly used for fixing the dies/chips to the substrate is nothing but a disc of
solder material with thickness in the range upto 500 microns and is either
circular or rectangular (square) in shape with sizes matching those of the dies.
It is essential that the die has metallization at the bottom.
Solder paste/cream is similar in characteristics to the thick film paste in the
sense that the solder material in the form of fine spherical particles is dispersed
in a thixotropic organic polymer along with a solvent to control the viscosity.
The paste of course does not contain any glass frit. The flux material is also
blended with the paste.
The solder paste can be applied by (1) sharp pin, (2) syringe and (3) screen
printing, to the circuit conductor terminations where the components have to
be attached.
Application using pin head is essentially a manual process and is only
acceptable for laboratory scale activity. There is: however absolutely no control
over the quantity of solder paste that is being applied and precise control of the
location is entirely dependent on the skill of the operator. Application through
syringe affords a better contrdl over the quantity of paste. Equipments are
available which precisely control the amount of the solder paste to be applied.
This can be predetermined by the choice of the nozzle and the other parameters
like the pressure in the cylinder. Simpler systems require the nozzle to be
moved to the desired location manually and the paste dispensed automatically.
More sophisticated systems are available where the nozzle is held on a movable
carriage of an X- Y table and its precise location over the substrate is computer
controlled. With such systems, the demand on the operator's skill is minimised.
Screen printing affords a precise control of both the quantity of the paste as
well as the location. This is a preferred method if a large number of identical
circuits are to be fabricated, but is essential if the circuit involves soldering
large number of compqnents or components with large number of pins with
narrow spacings/gaps. After application of solder cream/paste the components
to be soldered are placed in position and the paste is allowed to dry in air to
drive off the solvents. The adhesive nature of the paste helps in holding the
components in place before it goes for reflow.
Temperature of reflow, time and rate of heating are important in achieving
good solder joints. Rapid heating and/or cooling would result in the formation
of blisters and voids due to entrapment of volatile organic matter in the paste.
For large volume applications conveyerised reflow is convenient. IR reflow is
faster without heating too much the other circuit portions. But there is always
a danger of excessive heating of some components which have higher absorption
in the IR frequencies used for reflow. Vapour phase reflow enables uniform
heating of the whole circuit due to the intimate contact of the hot vapour of the
liquid. Flourinert fluorinated hydrocarbon with boiling point at 215°C is the
most common liquid used for vapour phase reflow. This has the added advantage
of cleaning the flux residues of the substrates.
Hot gas reflow involves directing hot gas (air) at the solder points through
a specially designed heated tool similar to soldering iron. Special tool tips are
provided so that all the leads of the component are heated simultaneously and
reflowed. Hot gas reflow has the advantage of heating only the solder points.
But basicallY,this is a slow process and useful for repair work wherein one
does not heat up the other solder joints in the circuits.
The advantage of laser reflow is also that the laser energy is directed only
to the solder joints. The energy and pulse width are so chosen that it is just
right for ensuring the reflow without causing the surrounding circuit/components
to heat up. Thus any heat sensitive components are protected. Different soldering
techniques are compared in Table 10.
4.2 Eutectic Die Bonding

This can be considered a special case of soldering in the sense that no external
material like solder is involved in joining the two materials. The materials to
be bonded together dissolve in each other forming a liquidus at a suitable
elevated temperature generally known as eutectic temperature, much below the
melting points of individual constituents. The most commonly used system is
a silicon die/chip eutectically bonded to gold metallization at about 370°C. In
this case the silicon die does not have any metallization on the bottom. The die
is brought into intimate contact with the metallization under gentle pressure
and rub to aid the eutectic bonding. Eutectic bonding provides very good
mechanical strength accompanied with good electrical and thermal conductivity.
Table 10. Comparision or dirrerent soldering techniques
Technique Form of solder Automation Thermal stress Remarks

material to the circuit
Melt Soldering
Soldering Thin wire-plain Strictly manual Localised heat- Laboratory scale
iron or flux cored ing of the term- assembly and
Preform inals to be sold- rework
ered
Reflow Soldering
Hot gas Preform Semi-automatic Only localised Laboratory scale
Cream/Paste heating of the assembly or batch
terminals to be production. Mainly
soldered useful for rework
Hot plate Preform Manual to total Whole circuit is From batch produ-
Cream/Paste automation by raised to the ction to full scale
use of convey- reflow tempera- production by inte-
erised system ture grating with auto-
mated dispensing
system for appli-
cation of solder
paste and compo-
nent pick and place
equipment
IR Cream/Paste -do- Only solder -do-

paste heats up for
quick reflow.
But there is a
danger of exces-
sive heating of
some compon-
ents which
absorb IR
Vapour Cream/Paste -do- Whole circuit is -do-

phase raised to retlow
temperature
Laser Cream/Paste Best utilised Only the solder Best suited for large
when automated terminals are production for the
with computer heated investment
control
The disadvantages however are: (1) the need to use high temperature for bonding
and (2) the strain on the die due to differential thermal expansion between the
die material and the substrate-particularly severe for large size dies.
4.3 Adhesive Bonding

An effective alternative to soldering is the use of conductive adhesives to
attach components to the substrate. This becomes essential if for some reason
we cannot subject the components to the temperatures used for soldering. Another
area where the adhesives come to rescue is when the two parts cannot be joined
together either by use of soldering or by eutectic method. Conductive epoxy
adhesives have been first introduced nearly two decades ago. Now it has grown
into a mature technology and several other adhesive formulations have come
up over the years. Epoxies dominated the field of adhesives used for
microelectronic applications. A large variety of formulations have been introduced
in the market to meet varied needs of the hybrid manufacturers. Other formulations
of adhesives are based on polyimides and the silicones (Jowett 1982 and Soane
and Martynenko 1989).
Epoxy resin adhesives are probably the most versatile bonding materials
available. These adhesives result in highly cross-linked molecular structures
with excellent tensile shear strength but poor peal or cleavage strength. Low
cure shrinkage and high resistance to creap under prolonged stress are typical
of epoxy adhesives. Their resistance to moisture and solvents is excellent. An
epoxy adhesive is primarily composed of an epoxy resin and a curing agent.
Secondary ingredients include reactive solvents to adjust viscosity. fillers such
as metal particles to make it conductive or ceramics such as alumina powder
to increase thermal conductivity. Epoxies are available in solid form or diluted
with solvents. The viscosity ranges from sprayable solvent system through
thixotropic paste suitable for screen printing. Single component epoxy adhesives
incorporate latent curing agents mixed with the epoxy resin. which become
active when the adhesive is heated. Variety of other fillers are used to adjust
the thermal expansion coefficient. to minimise the cure shrinkage and to adjust
the viscosity. The fillers are also used to reduce the exotherm from the cure
cycle thus reducing the thermal stress induced in the adhesive bonds.
Silicone adhesives are thermosetting materials. These have excellent heat
resistance. however their initial strength is considerably less than that of epoxies.
The adhesives are available in the form of viscous solutions either as two
component or single component formulations. Curing is normally at elevated
temperatures. But RTV (Room Temperature Vulcanizing) silicone adhesives
are also available. The desired electrical. mechanical and thermal properties
are achieved only after the materials are fully cured. Tensile shear strength is
low but their peal strength is relatively high compared to epoxies. Silicone
adhesives can withstand temperatures in excess of 300°C for several days.
Polyimide is also a thermosetting material but has the advantage of high
tensile shear strength similar to epoxy and high temperature capability like
silicone. One disadvantage however arises from the fact that it hardens into a
stiff glassy substance and there. could be problems with the differential thermal
22 MICROWAVB MATERIALS
expansion between the component and the substrate during thermal cycling.
This is particularly severe if the difference is appreciable and if the component
to be mounted is relatively large in size. This material is also available in
solution form in consistencies suitable for use with standard dispensing
equipments.
When the adhesive has to replace a solder it has to be conductive. This is
achieved by loading the adhesive with metal powder, usually silver or gold.
Silver-palladium filled adhesives can be used where silver migration is a severe
problem at the cost of lower conductivity. The metallic fillers in excess of
70% are used. The adhesives, in particular epoxies can be of two component or
single component type; solvent based or solid formulation. Curing time is
dependent on temperature. Higher temperature promotes fast curing. Single
component adhesives are convenient to use but need special storage conditions
to increase the shelf life. Adhesives are available in paste form to enable easy
and controlled dispensing through syringe or for screen printing. Important
characteristics that decide the choice of a good adhesive apart from providing
good adhesion are: (1) minimum shrinkage during curing, (2) minimum out-
gassing, (3) high stability over wider operating temperature range and (4) low
chloride ion content.
Minimum shrinkage during curing is important otherwise the component
will be under great stress. For high reliability, circuits are invariably hermetically
sealed in a metal or ceramic package. It is therefore essential that there is
minimum out-gassing of harmful substances like water vapour or ionic impurities
such as chloride ions which will corrode the aluminium metallization and fine
bonding wires. New epoxy formulations which are referred to as 'moisture
lock' are being introduced which act as getters to moisture (Estes et al 1985).
The requirement of low chloride ion content in the adhesives, as required by
new US Military specifications, has been questioned by one of the established
epoxy adhesive supplier and is still open to discussion on the acceptable limit
(Estes 1986 and Estes and Pernice 1989). Entrapment of the solvents within the
adhesive films during curing is one of the greatest problems in attaching the
component to the substrate, particularly so when a large size die is to be
attached. This not only causes poor performance but also results in reliability
problems (Estes 1984). Special epoxy adhesives have been devised which can
withstand continuous operating temperature of more than 175°C and intermittent
operating temperature of 300-400°C (Estes and Pernice 1989). The main
advantage of these over the polyimide formulation is· that these epoxies have
dual glass transition temperature. The lower Tg value around 80 to 900C imparts
non-brittle or resilient character to the adhesive thereby allowing for thermal
cycling from -55 to +155°C without micro cracks developing in the bond line.
The higher Tg (160 to 180°C) imparts high temperature strength to the adhesive
to enable thermosonic/thermocompression wire bonding to be successfully
completed without die floating occurring.
Successful use of adhesives for the Microwave Integrated Circuits operating
upto millimeter wave bands has been demonstrated (Hernandez 1974). One
point to note is the precise control of the amount of the adhesive to be used
and the curing schedule. It should be noted that longer curing schedule at
higher temperatures does not necessarily result in the best bond. One should
strictly follow the guidelines given by the manufacturers and supplement the
information with experimentation. To take care of large thermal expansion
mismatch one has to recourse to the elastomeric adhesives based on the silicones,
compromising on the tensile shear strength.
In addition to the electrically conductive adhesives, there are several occasions
to use electrically insulating and thermally conductive adhesive. In these cases
the adhesives are loaded with fine powders of alumina or boron nitride. The
latter provides- a higher conductivity by a factor of two. Another advantage of
the use of adhesives is that there are no flux residues to be cleaned which could
be detrimental to the reliability of the circuit.
Methods used for adhesive application are similar to those used for solder
paste or cream. Precise control in dispensing the adhesive is essential to avoid
bleedout of the excessive adhesive covering the bonding pads on and around
the chips. This hampers proper wire bonding. The curing is normally done at
elevated temperatures as prescribed by the suppliers in a thermostatically
controlled oven or on a hot plate. The latter may be preferable in preventing
crust formation on top of the epoxy while curing. This causes trapping of gases
from the uncured epoxy ultimately leading to splattering of the epoxy: Alternative
methods for curing are the use of IR and microwaves.
4.4 Wire Bonding

Wire bonding is a process wherein interconnection is provided between the
bonding pads on the semiconductor die/chip and the bonding pad terminations
of the conductor lines printed on the substrate. The techniques that are employed
for achieving this are: (1) Thermocompression bonding, (2) Ultrasonic bonding
and (3) Thermosonic bonding.
Gold and aluminium in the form of wire/ribbons are two metals which are
almost exclusively used for interconnection though to a lesser extent copper
and silver have also been experimented with.
4.4.1 Thermocompression Bonding

Thermocompression bonding involves bringing the wire to be bonded in intimate
contact with the bonding pad under a specified pressure at elevated tempe-
ratures for a specific time. Special tools are required to guide and hold in
position the fine bonding wires whose diameter is typically in the range of 0.7
to 6 mil. Temperatures in the range of 200 to 300°C and bond pressures of the
order of 100 kg/cm 2 are required to initiate plastic flow of the metal leading
to the bond formation due to solid state interdiffusion between the bonding
wire and the pad. Sright lateral mechanical scrubbing of the tool on the pad is
done to aid the bond formation. The duration of the bonding is of the order of
few milliseconds.
Capillary and wedge are the two commonly used types of tools for carrying
out the wire bonding. The capillaries are usually made of ceramic while the
wedges are made of tungsten carbide or titanium carbide.
4.4.1.1 BAIL BONDING

Wire bonding using a capillary is nonnally known as ball bonding. Figure II
shows the important stages in. the process of wire bonding. The first step in
the bonding process is the fonnation of a ball by the application of short heat
pulse to the tip of the wire. Earlier this was achieved with the help of a fine
tipped hydrogen flame. But in the recent times ·this is replaced by a short
electric discharge referred to as electronic flame-off. The size of the ball is
controlled by the discharge gap between the electrode and wire, the discharge
(a) BAll FORMATION

W LDIE:::J
RECUMP
(b) TARGETIING ON THE

Clt:!
(c) SQUASHING OF
BOND PAD OF THE DIE THE BALL FOR
BOND
FORMATION
o c
~ CONDUCTOR TRACK
(a) FORMATION OF THE LOOP
(d) CAPILLARY RAISES
TO THE lOOP HEIGHT
(f) FORMATION OF WEDGE (g) BREAKING OF WIRE. NOTE THAT

SHAPED BOND THE WIRE LAMP IS CLOSED
Fig. II. Graphic illustration of the bonding cycle of the ball bonding
(Wire) (MIs Kulicke & SoCfa Ltd. Inc., USA).
potential and the time. Optimum diameter of the ball is about 1.5 times the
wire diameter.
In thermocompression bonding both the bonding tool as well as the substrate
are kept at the desired elevated temperature. The bonding tool moves down and
the annular rim of the tool squashes the ball against the bonding pad for the
preset time leading to bond formation. As already mentioned, the temperature,
the bond force (pressure) and the duration of the bonding decide the quality of
bond. Excessive force squeezes the material of the wire from underneath the
tool leading to weakening of the wire itself. Optimum size of the squashed ball
is about 2 to 2.5 times the diameter of the wire. After the bond formation, the
tool is withdrawn to a preset height and the substrate is moved to bring the
second bond pad into position beneath the capillary tool. The tool is lowered
to make the second bond. It may be noted that the second bond cycle does not
involve the formation of ball. The wire is squashed by the rim of the capillary
to establish the bond. The process of tool lifting after the first bond, translation
of the substrate to the second bond position and lowering of the tool for the
second bond formation have to be precisely controlled to enable proper 'looping'
of the wire bond. This is essential not only to avoid shorting of the wire to the
sides of die or chip but also to maintain and control the impedance offered by
the bonding wire in the circuit which is impOrtant in the microwave integrated
circuits. Because of the circular symmetry of the first bond (ball bond) the tool
can be moved in any direction to carry out the second bond. The wire is held
by a clamp as the tool is moved upto a preset height after the second bond
thereby snapping the wire. The wire is cut at the place where it is squashed
most.
4.4.1.2 WEDGE BONDING

The bonding tool used for this is in the form of a wedge and the wire is bonded
to the interconnecting pad by stamping the wire by the foot of the wedge by
a preset force for the preset time. For thermocompression bonding here again
both the tool and the substrate are kept at the desired temperature. Because of
the directional nature of the wedge bond, the tool has to move only backwards
to make the second bond. Here again precise control of the tool movement is
essential for proper 'looping' of the bonding wire.
Wedge bonding is possible with both gold and aluminium wires whereas
ball bonding has not been possible with aluminium because of the difficulty in
the formation of the ball. Typical characteristics of the wedge and ball bonding
are given in Table 11. Ribbons instead of wires are used particularly for microwave
integrated circuits for a better control of the impedance to achieve closer matching
to the transmission lines printed on the substrate. Thermocompression bonding
is not suitable for aluminium wire bonding and on aluminium metallization due
to the contamination of the surface by a thin layer of oxide.
4.4.2 Ultrasonic Bonding

In contrast to thermocompression bonding, the energy required for ensuring
bonding between the metallization and the wire is supplied by ultrasonic energy
through the bonding tool. The high frequency scrubbing in addition to providing
localised heating causes plastic flow of the wire and the contact pad leading to
solid state interdiffusion. It also helps in breaking down the surface contaminants
if any, in particular the oxide layer on aluminium, helping in intimate contact
between the metal surfaces. This is a convenient method where the components
cannot be raised to the temperature needed for thermocompression bonding.
But it is not useful in bonding highly brittle dies such as GaAs chips.
4.4.3 Thermosonic Bonding

This technique uses both heat as well as ultrasonic energy for the bonding
process. This enables one to arrive at the best compromise in the values of
substrate/die temperature and ultrasonic energy both of which tend to be lower
than in either situations of thermocompression and ultrasonic bonding. In this
case the bonding tool is not heated. Thermosonic bonding is versatile, offering
wider latitude in the process parameters and is currently the most widely used
method for wire bonding. Typical range of parameters in the three cases of
wire bonding are given in Table 11.
Table 11. Comparision of wire/ribbon bonding techniques
Ball bonding Wedge bonding
Material Gold Gold Aluminium

Fonn Wire Wire/Ribbon Wire
Dimensions. 0.7 to 2 mil 0.5 to 3 mil 1 to 4 mil
(18 to 50 J.l1Tl) (12.5 to 75 Jlm)/ (25 to 100 J.l1Tl)
0.5 x 10 mil
(12.5 x 250 J.l1Tl)
Elongation ~% 4-6% 4-6%
Bonding Parameters
Temperature
Thennocompression 300-4OO°C 300-4OOoC Not possible
Ultrasonic Ambient Ambient Ambient
Thennosonic l00-2oo°C l00-2oo°C l00-20°C
Ultrasonic power 2 Watt 2 Watt 2 Watt
(maximum)
Force (typical) 30 g 40 g 30 g
for 1 mil wire
Time (sec) 10-100 10-100 10-100
Capabilities
Minimum bond pad size 50 J.l1Tl 20 J.l1Tl 20 J.l1Tl
Bond pull strength for 8-10 g 5g 5 g
1 mil wire
4.4.4 Materials for Bonding Wire/Ribbon

As already mentioned gold and aluminium in the form of wire/ribbon are the
two materials mostly used for interconnection between the bonding pads on the
die and the metallised terminations on the substrate. Silver and copper have
been considered as alternatives but have not replaced gold and aluminium. The
dimensions of the wire and the ribbon commonly employed are indicated in
Table 11. One of the important properties of the wire/ribbon that is significant
for the wire bonding is ductility. Higher. the ductility more plastic flow will
occur for any given applied pressure resulting in more intimate interface contact.
Oxidation is less important in case of noble metals in determining the bond
stability. Normally successful bonding is achieved between metals of similar
crystalline structures perhaps due to better mutual solubility in compression
bonding. For better bondability therefore most of the metalli7.8tion structures
are given a final layer of gold plating. However, gold wire bonding on aluminium
alloy metallization on silicon chip is a common feature. But a potential failure
mechanism in such configuration is the formation of brittle intermetallic phases
at the interface normally known as the purple plague. This is controlled to
some extent though not completely eliminated by alloying small percent of
palladium in the gold wire.
Elongation of the wire (the amount the wire stretches before it breaks expressed
in %) is very important since it decides the formation of good and stable bond
wire loop. Too much elongation results in the sagging of the loop leading to
bond failures. Too little elongation would make the formation of loop altogether
difficult. Typical values are given in Table 11.
5 CONCLUSION
The area of the materials used in the fabrication of Microwave Integrated
Circuits is very wide involving materials ranging from polymers, ceramics to
special high purity metals and alloys. Each of these serves a specific purpose
and possesses special characteristics. In this chapter it was only possible to
give an overall picture of the various processes involved in the fabrication of
MICs and the materials used therein. No attempt has been made to go into the
details of the preparation and characterisation of materials themselves. For
more detailed account the reader should consult the books and articles cited
under references. Secondly, as regards the characteristics of various materials
listed in the tables it should be emphasised that these are meant to be only
indicative of their typical values and cannot be used for any design calculations.
By the very nature of these materials there is bound to be a variation in these
characteristics depending on the detailed composition and the method of
preparation. It is therefore advised to use the values specified by the manufacturers
of the materials and if necessary and possible carry out inhouse characterisation
to confirm the listed specifications.
Another area which needs special attention is the defence and space electronics
wherein microwave circuits and systems find wide ranging applications. These
applications demand high reliability under adverse environmental conditions.
This implies that the system should have the ability to perform over a wide
temperature range and under conditions of high humidity, large vibration, high
acceleration, great impact etc. Detailed specifications of these and test methods
are given in US Military standards 883B and recently introduced standards
1772 or other similar standards. The need for the final subsystem to meet
stringent military and space standards implies that the materials used therein
have to perform well and meet the stipulated standards. One therefore has to
ensure that the materials used for the fabrication of microwave systems for
defence or space applications meet the stipulated standards by thorough screening
and characterisation as per the methods specified therein.
~EFERENCES
1. Berry RW, Hall PM and Harris MT (1968) 'Thin Film Technology' (Princeton,
D. Van Nostrand Co. Inc.).
2. Bhat Band Koul SK (1989) 'Stripline-like Transmission Lines for Microwave
Integrated Circuits' (New Delhi: Wiley Eastern Ltd).
3. Blum JB and Anzai K (1989) 'Aluminium Nitride Substrates for Hybrid
Microelectronics Application' Hybrid Circuit Technology, 6 No.8, p. 7-14.
4. Chopra KL and Kaur I (1983) 'Thin Film Device Applications' (New York:
Plenum Press).
5. Corkhill JR (1976) 'Thick Films at High Frequencies' in 'Hand Book of Thick
Film Technology' Eds. Holmes PJ and Loasby RG (Ayr, Scotland:
Electrochemical Publications Ltd.).
6. Edwards PR (1991) 'Manufacturing Technology in the Electronics Industry-
An introduction' (London: Chapman & Hall).
7. Estes RH and Pernice RF (1989) "Die Attach Adhesives-Evaluation of VceSAT
and 9jc Performance in Power Devices" Proceedings of ISHM International
Symposium on Microelectronics, Verginia, USA, p. 664-669.
8. Estes RH (1986) 'Adhesives for Military Hybrids' Hybrid Circuit Technology,
3 No.7, p. 21-24.
9. Estes RH, Kulesza FW and Banfield CE (1985) 'Recent Advances made in
Die-Attach Adhesives for Microelectronic Applications' Proceedings of ISHM
International Symposium on Microelectronics, Anaheim, USA. p. 391-401.
10. Estes RH (1984) 'The Effect of Porosity on Mechanical, Electrical and Thermal
Characteristics of Conductive Die-Attach Adhesives' Solid State Technology,
August p. 191-197.
11. Grovenor CRM (1989) 'Microelectronic Materials' (Bristol: Adam Hilger).
12. Haskard MR (1988) 'Thick Film Hybrids-Manufacture and Design' (Sydney:
Prentice-Hall of Australia Pty Ltd.).
13. Hernandez L (1974) 'Epoxy Techniques for Hybrid Microwave Integrated Circuits'
Proceedings of ISHM International Microelectronics Symposium, Boston, USA.
p.272-5.
14. Holland L (1965) 'Thin Film Microelectronics-The Preparation and Properties
of Components and Circuit Arrays' (London: Chapman and Hall Ltd.).
15. Holmes PJ and Loasl,>y RG (1976) 'Hand Book of Thick Film Technology'
(Ayr. Scotland: Electrochemical Publications Ltd.).
16. Hughes JE (1987, 'A Review of Thin Film Resistors and their Assembly Problems'
Hybrid Circuits, May.
17. Jowett CE (1982) 'Materials and Processes in Electronics' (London: Hutchinson
& Co (Publishers) Ltd.).
18. Kugler TR and Culkin TJ (1989) 'Ceramic Processing with Lasers' Hybrid
Circuit Technology, 6 No.8. p. 45-48.
19. Licari 11 and Enlow LR (1988) 'Hybrid Microcircuit Technology Hand Book:
Materials, Processes, Design, Testing and Production' (park Ridge, New Jersey:
Noyes Publications).
20. Maissel U and GIang R (1970) 'Hand Book of Thin Film Technology' (New
York: McGraw-Hill).
21. Pitt KEG (1981) 'An Introduction to Thick Film Component Technology' (Luton,
England: Mackintosh Publications Ltd.).
22. Prasad SD and Vijayaraghavan MS (1987) 'Fabricating Hybrid MICs on J71'FE
based Boards' Hybrid Circuit Technology 4 No.3, p. 19-21.
23. Reddy TRs (1988) 'MMICs in Hybrids' Proceedings of ISHM (India) '88 Third
Annual Conference of International Society for Hybrid Microelectronics (India
Chapter), Hyderabad, INDIA p. 31-45.
24. Siddall G (1965) 'The Properties of Passive Circuit Elements' in 'Thin Film
Microelectronics-The Preparation and Properties of Components and Circuit
Arrays' Ed. Holland L (London: Chapman and Hall Ltd.).
25. Soane DS and Martynenko Z (1989) 'Polymers in Microelectronics: F\Dldamentals
and Application' (Amsterdam: Elsevier Science Publishers B.V.).
26. Stuart RV (1983) 'Vacuum Technology, Thin Film and Sputtering: An
introduction' (Orlando, USA: Academic Press Inc.).
27. Sunahara K and Takabatake M (1990) 'Low-Temperature Fired Multilayer
Circuit Board for High Frequency Application' Hybrid Circuit Technology, 7
No.2, p. 35-41.
28. Taconic Tech Topic 'Processing of J71'FFJWoven Glass Laminates' ¥icrowave
Dielectrics Division (Petersberg, USA: MIs Taconic Plastics Ltd).
29. Van Nie AG, Goedbloed W and.Kersuzan G (1981) 'Relaibility and Degradation
of Microwave Integrated Circuits' in 'Reliability and Degradation' Eds. Howes
MJ & Morgan DV (New York: John Wiley & Sons Ltd) p. 363-439.
2
Materials and Technology for Microwave
Integrated Circuits
Bharathi Bhat and Shiban K. Koul

Centre for Applied Research in Electronics
Indian Institute of Technology. Delhi. New Delhi. India
1. INTRODUCTION
Advances over the past two decades in the planar techniques and technology
and also in the miniature microwave solid state devices compatible with this
technology have led to the realization of compact microwave integrated circuits
(MICs). These circuits are built on planar transmission lines such as the stripline.
microstrip line. suspended stripline. suspended microstrip. inverted microstrip.
slot line. coplanar waveguide and coplanar strips. The basic geometries of
these lines are shown in Fig. 1. For MICs. these planar transmission lines are
formed on substrates made of low loss dielectrics. ferrimagnetics and
semiconductors. A typical MIC consistS of one or more of the following parts:
(a) distributed elements (b) lumped elements in planar/discrete form.
(c) semiconductor devices and (d) special elements such as dielectric resonators
and ferrite discs. Two different technologies; viz. hybrid and monolithic are well
accepted for the fabrication of MICs.
In the hybrid technology. the passive circuit which consists of distributed
and/or planar lumped elements is realized as a conductor pattern on the substrate.
This is a single level process and involves metallization of the substrate through
vacuum evaporation/RF sputtering and defining the desired pattern through
acc~te photolithographic techniques. Resistors. if required. can be incorporated
by depositing resistive films at appropriate places. Semiconductor devices and
other disc:rete passive chip devices (capacitors and resistors) if any, are bonded
or soldered to the circuit pattern. Typical operating range of hybrid MICs is 1
to 30 GHz and with the adoption of more sophisticated photolithographic
techniques. their operation can be extended into the millimetre wave range up
to about 120 GHz. Nearly all coaxial and waveguide components. particularly
for operation at low power levels can now be realized in hybrid MIC form. The
advent of this t.echnology has resulted in considerable reduction in size and
weight. ease of mass production with improved reproducibility and potential
low cost of functional circuits. The technology offers the advantage of combining
Materials and Technology for Microwave Integrated Circuits 31
.. Ground
planes
(al Stripline (TEM model

Stri p conductor
Ground
plane
(bl Microstrip line (quas·,-TEM mode I
Strip conduc tor

Ground planes
J
dL (.,. . -. .---~..:.,..-~ ---:f- ·-;5\··.· :.... .,
10·. ·.·0:·;
f~~t-::~
~ -IWI- ' -
(cl Suspended stripline (quasi-TEM model

Strip conductor
Air gap
~~~.,.......~.
.,.-
,. . . . ,. . .r./
.....,........,.....:....:...,.:..:.......:..~...:..:.....:..JI
~~----______J
Ground
plane
(dl Suspended microstrip (quasi-TEM model

Strip
conductor ///
/
Ground
plane
(eJInverted microstrip (quasi-TEM model
Fig. 1 (Contd.)
Conductors
(non-TEM mode)
Conductors
*~
T~
(g)Coplanar waveguide (quasi - TEM mode)
(h) Coplanar strips/quasi -TEM mode)
Fig. 1. Planar transmission lines for MICs and typical fields in their
cross-sectional planes, E field lines.- - - -H field lines.
multi-function circuits without the interconnecting wires thus permitting realization

of compact integrated modules with highly reliable performance. The monolithic
technology achieves much higher degree of circuit integration than the hybrid
technology and the circuits are popularly referred to as MMICs (monolithic
microwave integrated circuits). In this technology, active devices are grown in-
situ on or within a semiconductor substrate, and planar passive circuitry and
interconnections are then deposited on the substrate or grown in it. The fabrication
involves highly reproducible device technology using diffusion and/or ion
implantation, multilevel metallization and a composite process involving
photolithography/electron beam lithography. As compared with the hybrid
technology, the monolithic technology offers significant reduction in size and
weight; and improved reliability and reproducibility through elimination of
wire bonds. Embedding of active devices within the substrate and elimination
of wire bonds reduce the undesired parasitics. thereby improving the capability
for broadband performance. Furthermore. the ability to incorporate multi-
functional capability on a single chip pennits the realization of integrated receiver

front end and transmit-receive modules.
Monolithic technology is capital intensive and involves high cost manufacture.
MMICs can become cost-effective only when there is a requirement for large
volume production such as for DBS receiver systems and phased array radars.
Hybrid MICs, on the other hand, have proved to be cost effective and are
extensively used in a variety of practical radar, communication, navigation,
and sensing systems at microwave and even millimetre wave frequencies. The
applications 'are numerous; microwave test instrumentation, satellite
communication, electronic communication, electronic warfare, electronic scanning
through phased arrays, and direct broadcast satellite reception.
In the following, we provide a comprehensive review of the basic techniques
and technology of MICs. Planar transmission lines, substrate materials, basic
elements of planar circuits, and the hybrid and monolithic technologies are
covered.
2. PLANAR TRANSMISSION LINES

Planar transmission lines fonn the basic transmission media for microwave
integrated circuits whether the circuit is fabricated in hybrid or monolithic
form. Figure I illustrates the basic geometries of the various commonly used
planar transmission lines and also the typical field lines for the dominant mode
propagation. Besides these, several coupled conductor versions of structures
shown in Fig. 1 (a)-(e) and other variants are also used in a variety of circuit
applications. The analyses and electrical characteristics of such planar transmission
lines are available in several books and reprint volumes [1-7]. In this section,
we review the salient characteristics of the commonly used structures shown
in Fig. 1. Thro'ughout the section, we use the symbols Z and A. to denote the
characteristic impedance and guide wavelength, respectively, of the planar
transmission line under consideration, Za for the characteristic impedance of
the same planar transmission line with all dielectrics replaced by air, and .1.0
to denote the free space wavelength.
2.1 Stripline (Fig. la) [I, 2, 8, 9]

The basic structure of a stripline consists of a flat strip conductor situated
symmetrically between two large ground planes with the intervening space
homogeneously filled with a dielectric. The dominant mode of propagation in
a stripline is a pure transverse electromagnetic (TEM) mode; that is, the electric
and magnetic field components lie entirely transverse to the direction of
propagation. The fields concentrate around the strip conductor and decay rapidly
with distance away from the strip in the lateral direction. The characteristic
impedance Z and the guide wavelength A. of the homogeneous stripline are given
by
Z= za/..[i; (1)
(2)
where Er is the relative dielectric constant of the dielectric filling. The following
closed from expressions reported in the literature [8] are known to yield virtually
exact results for the characteristic impedance:
For w/b ~ O.S
.r;;-
ZvEr =29.979 In[ 2 (1(1 +_ f]
k) (3)
For w/b ~ O.S
Z..Je; 29.979tr 2 (4)

Er= 1[2 (1+..Ji?)]
n (1-..Ji?)
where
k = sech (trw/2b) (5a)
" = tanh (trw/2b) (5b)
In the above formulas, w is the strip width and b is the ground plane spacing.
The thickness of the strip conductor is assumed to be negligible.
For practical circuits, the dimensions of the stripline are chosen so as to
ensure operation in the TEM mode while keeping the losses to a minimum.
The attenuation in a stripline decreases with a decrease in Z for a constant ground
plane spacing b, and also with an increase in b for a constant Z [9]. The maximum
value of b is limited by the onset of the TE or TM mode, whichever propagates
frrst. The cut-off wavelengths of the low~st order TE and TM modes, denoted
as Ac,TE and Ac,TM, respectively, are given by [10]
A.c,TE =..Je; {2w + (nb/2)} (6)
(7)
The stripline is an excellent medium for realizing passive components. Because

of its operation in the TEM mode, excellent directivity can be achieved in
coupled line directional couplers and high isolation in other coupled line
components. The structure, however, is not convenient for incorporating chip
devices. Practical stripline circuits are provided with side metallic walls so as
to form a complete shielded box as shown in Fig. 2a. The side wall separation
c is chosen large enough (c/w ~ 10) so that the walls have negligible effect on
the propagation characteristics.
2.2 Microstrip Line (Fig. Ib) [2-7, 11-21]

Unlike the stripline, the microstrip is an inhomogeneous transmission line. It
consists of a dielectric substrate with a strip conductor on one surface and
metallization to form a ground plane on the reverse surface. The dominant
mode of propagation in this line is quasi-TEM; that is, the mode is dominantly
TEM with small field components along the direction of propagation. As shown
in Fig. Ib, the fields are confined to the vicinity of the strip conductor with a
larger concentration inside the dielectric substrate and less in the air region
T'.. '. . . ... . : .-. . . ... : -... .

L-~\-
:- . "-w~.-~-----··.
Metall ic
sh ield
\. C- - -----I
(a) Stripline
Metallic
Air ~shie ld
(bl Microstrip (c/w ~10 ,bfh=SI
Gasket
Metall ic
Air ' sh ield
:: :. ·.E ·:.·: : .... -:···. · ·. ··:· ·

AI r
f---- c - -----lld--
(e) Suspended stripline
Fig. 2. Examples of shielded planar transmission lines.
above. Larger the value of Er of the substrate. the greater is the relative concen-
tration of energy inside the substrate and lesser is the radiation. The characteristic
impedance and guide wavelength of a microstrip can be related to Z· and AO
in the form [11]
Z = Z'f..JEerr (8)
A = Aof..JEerr (9)
where Eeff is the effective dielectric constant (relative) of the microstrip medium.
It is defined as the relative dielectric constant of an equivalent homogeneous
microstrip which has the same phase velocity as the original (inhomogeneous)
microstrip. The value of Eeff lies in the range
(10)
depending on the value of w/h. For small strip widths (w/h « I), the fields are
distributed nearly equally in the substrate and air regions and hence Ecff approaches
the lower limit (1 + Er )/2. For large strip widths (w/h » I), the electric field
is confined mostly between the strip conductor and the ground plane. The
microstrip then resembles a parallel-plate capacitor with Ecrr approaching Er•
Quasi-static formulas for Z and Eerr

Closed form expressions for evaluating Z and Eerr of microstrip lines have been
reported by several investigators [12-14]. Under the quasi-static approximation
and assuming the strip thickness to be negligible, the expression for Z is given
by [15]
Z =(roN £eff) In[(h/w) F{w/ h) + {I + (2h/w)2jl/2] (lla)

where
F{w/h) = [6 + (2n- - 6) exp {- {30.666h/w)0.7528}] (llb)
The accuracy of these expressions is reported to be better than 0.01% for
w/h ~ 1 and 0.03% for w/h « 100 [14, 15]. The effective dielectric constant
is given by [14]
Eerr = {{Er + I)/2} + ({Er - 1)/2)} {I + {lOh/w)}-a·b (l2a)
a = [1 + (1/49) In {{W/h)4 ; {W/52h)2} + (1/18.7) In {l + {w/I8.lh3}]

(w/h) + 0.432
(I2b)
b = 0.564 (Er - 0.9 )0.053 (l2c)

Er + 3
The accuracy of this formula is reported to be better than 0.2% for ~ 128 Er
and 0.01 ~ w/h ~ 100 [14, 15]. The maximum frequency up to which the above
quasi-static formulas can be used is given by [15]
(13)
where h is in cm and fd is in GHz. From (13), it can be seen that the upper
frequency limit for quasi-static operation can be extended by choosing substrates
of lower thickness and lower dielectric constant.
Effect of Dispersion
At higher frequencies, the fields tend to concentrate more within the dielectric
substrate. Consequently, Eerr increases with an increase in frequency, with the
value approaching the substrate permittivity Er asymptotically in the limit
frequency tends to infinity [16]. The following closed form dispersion formula
can be used to compute the frequency dependent £CIT (denoted as £CIT (j) in the
range 2 ~ Er ~ 16, 0.06 ~ w/h ~ 16 andf~ 100• GHz [17]:
E
efT
(f) = [Fr -..re;; re-]
1 + 4F-LS
+ "Ie-efT
2
(14a)
where
F = 4k.J~: - 1) [0.5 + {1 + 2 log (1 + (w/h»}2] (14b)
and EefTis the quasi-static value given by (12). Similarly, the effect of dispersion
on Z can be obtained from the closed form expression given by [14]:
Z(f) =Z(Ecff(f) - 1) ...j{Ecff/EcfC(f}} (15)

EefT - 1
where Z is the quasi-static value given by (II).
Losses
The propagation loss (neglecting radiation) in a practical microstrip is primarily
due to two types of dissipative losses; conductor loss due to the finite resistivity
of the conductor and dielectric loss due to the fmite loss tangent of the dielectric
substrate. Microstrip losses have been studied by several investigators [18-20].
If we denote ae as the attenuation constant due to the conductor loss and ad
as the attenuation constant due to the dielectric loss, both expressed in dB/unit
length, then the total attenuation constant a is given by
a =(ae + t1cJ) dB/unit length (16)
The following closed form expressions due to Pucel et al [18] can be used to
compute ae in dB/unit length:
{ R; [32 -
(W'/h)']
a _ 1.38 hZ 32 + (w'/h}2 A w/h ~ 1
(17a)
c - 61 x 10-5 Z E~ [w' +R: 0ft67w'/h JA w/h ~ 1
f
. h h (w'/h) + l444
A= + (h/w') ~ + (125</_) + (l25/n) In (4.,./I)J w/h ~ 1/211:

1 + (h/w,) {I - (125t/trh) + (125/11:) In (2h/t}) w/h ~ 1/211:
(17b)
where
{ w/h + (I25/1r)(I/h) ~ + In (4nw/I)J w/h ~ 1/211: (18a)
w'/h = (18b)
w/h + (125/n}(t/h) {I + In (2h/t}) w/h ~ 1/211:
R: =RJl + (2/11:) tan-I {1.4(Li/6)2}] (18c)
Rs = ..j(CtJJl.o/2u) (18d)
6 =(1/Rs(1) (18e)
In these formulas, R. is the surface resistivity of the conductor, 6 is the skin
depth, (1 is the conductivity of the metal, Li is the rms surface roughness and
t is the thickness of the strip conductor.
The expression for l1ct in dB/unit length is given by [20]
ad =273(ErI.../Eeff(f)} (Eeff (f) - 1) (tan 6/).,0) (19)

Er - 1
Q-factor
The Q-factor of a microstrip corresponding to the conductor and dielectric
losses can be obtained from [15]
Q _ 11:..g;iiU} (20)
- ).,0 a
where a is the total attenuation constant specified in dB/unit length.

Maximum Frequency of Operation
The maximum frequency of operation of a microstrip is limited primarily by
the onset of higher order modes, higher losses and the requirement on stringent
fabricational tolerances. The frequency at which significant coupling takes
place between the quasi-TEM mode and the TM-surface wave mode is- given
by [21]
(21)
where Vo is the velocity of electromagnetic waves in free space. A major

advantage of the microstrip is that its surface is accessible for mounting passive
as well as active discrete devices. It is also a versatile medium for realizing a
variety of circuit forms and combining several circuit functions. Practical
microstrip circuits are housed in a shielded enclosure in order to suppress
radiation and provide electromagnetic shielding. Fig. 2b shows the cross-section
of a shielded microstrip. The dimensions of the enclosure are selected such that
the waveguide modes are below cut-off, and the top and side walls have practically
no effect on the propagation characteristics. Referring to the dimensional
parameters marked in Fig. 2b, these criteria can be met by choosing c/w ~ 10,
b/h ~5 and w, h « )J2.
2.3 Suspended Stripline (Fig. lc) [2, 5, 6, 22-27]

The suspended stripline is the most useful variant of the stripline. It is essentially
an inhomogeneous strip line in which the substrate carrying the strip conductor
is placed symmetrically between the two ground planes thereby creating an air
gap on either side of the substrate. The dominant mode of propagation is quasi-
TEM. The introduction of air gap results in the reduction of the effective
dielectric constant of the propagating medium. This configuration, therefore,
permits larger circuit dimensions leading to relaxed dimensional tolerances and
increased accuracy of circuit fabrication as compared with the microstrip. The
presence of the air gap also reduces the conductor loss in the ground plane,
because most of the electromagnetic energy gets concentrated in the dielectric
substrate. By choosing the substrate sufficiently thin, the effective dielectric
constant can be made close to that of air, thereby extending the frequency
range of operation in the dominant mode, which reduces to nearly TEM [22].
Because of the above advantages, the suspended stripline is ideally suited for
operation in the millimetre wave range. Practical suspended striplines require
some provision for supporting the dielectric substrate and also for electromagnetic
shielding. Fig. 2c shows the cross-section of a shielded structure with grooves
on the side walls to suspend the substrate.
Closed form expressions for Z and Eeff

The characteristic impedance and guide wavelength for a suspended stripline
(Fig. 2c) can be obtained from the following closed form expressions [27]:
For 0 < w < c/2
(22a)
where
..JEerr =[1 + (E - F In (w/b)) In (l/..je;W1 (22b)
and
Z= 601r [v + In {_6_} + {I + ~}1/21

R
(w/b) (w/b)
(22c)
where
E =0.2077 + 1.2177 (d/b) - 0.08364(c/b)
F = 0.03451 - ();1031 (d/b) + 0.01742(c/b)
v = - 1.7866 - 0.2035 (d/b) + 0,4750(c/b)
R = 1.0835 + 0.1007 (d/b) - 0.09457(c/b) (22d)
For c/2 < w < c
(23a)
where
..Jeerr = [1 + (E - Fin (w/b)) In (1I..je;)]-1 (23b)
and
Z = 1207t[V + R(w/b) + 1.393 + 0.667 In «w/b) + 1.444))-1] (23c)
where
E = 0.464 + 0.9467 (d/b) - 0.2063 (c/b)
F = 0.1424 + 0.3017 (d/b) - 0.02411 (e/b)
V =- 0.6301 - 0.07082 (d/b) + 0.247 (c/b)
R = 1.9492 + 0.1553 (d/b) - 0.5123 (e/b) (23d)
These expressions are valid for 1 S e/b S 2.5, I < Er < 4,0.1 < d/b < 0.5.
The accuracy of these expressions is within ± 2% for 0 < w < e/2 and ± 3%
for e/2 < w < e compared to rigorous methods.
In practice, the dimensions of the housing and grooves must be carefully
chosen so as to avoid propagation of undesired waveguide modes. The first
waveguide mode to appear can be either the TEIO-type or the distorted TEoI
mode depending on the dielectric constant and the dimensions of the enclosure.
For the TE IO type of mode, the cut-off frequency can be obtained from the
following relation [23] (refer Fig. 2c):
tan (05e Pc{£.)' tan (/3c..jEl) =(b/dl) ..j(e2/el) (24a)

where
el = [I - (d/b){(e, - I)/Er)]-l (24b)
e2 =[I - (d/dl){(e, - I)/Er)]-l (24c)
and Pc is the propagation constant at the cut-off frequency.
For the TEoI type of mode, the cut-off frequency can be obtained by solving
the following transcendental equation [24]':
(25)
where COe is the cut-off angular frequency and Vo is the free space velocity.
2.4 Suspended and Inverted Microstrip Lines (Fig. Id and Ie)

[II, 20, 28, 29]
The suspended microstrip (Fig. Id) is a variant of the microstrip with an air
gap between the substrate and the ground plane. The inverted microstrip
(Fig. Ie) differs from the suspended microstrip in that the strip conductor is
situated on a lower surface of the dielectric substrate facing the ground plane.
Both suspended and inverted microstrip lines can also be viewed as special
cases of the suspended stripline with one of the ground planes removed. These
structures retain the advantages of the suspended stripline in terms of achieving
larger strip dimensions and lower dissipative losses with respect to the microstrip.
In comparison with the suspended microstrip, the inverted microstrip has the
advantages of reduced radiation loss by virtue of having the strip conductor
below the substrate. However, because the air gap involved is too small (typically
on the order of 1 mm or less), incorporation of semiconductor devices becomes
very difficult in this configuration.
The analysis and propagation characteristics of both these structures are
reported in the literature [II, 20, 28, 29]. In the following, we provide closed
form expressions for evaluating the characteristic impedance Z and effective

dielectric constant EeIf[29]. The guide wavelength is given by Aol..jEeff'
Suspended Microstrip (Fig. Id)

Z =(60I..jEetr) In [f(u)/u + (1 + 4Iu2)1/2] (26a)
where
f(u) =6 + (2n - 6) exp [ - (30.666Iu)0.7528] (26b)
u =w/(a + b) (26c)
..jEetr =[1 + (dla).(al - ht In (wla» (II{£; _1)]-1 (27a)
where
al = [0.8621 - 0.1251 In (dla)t (27b)
bl = [0.4986 - 0.1397 In (dla)]4 (27c)
Inverted Microstrip (Fig. Ie)

The expression for the characteristic impedance is the same as (26a) withf(u)
=
given by (26b) and u wla. The effective dielectric constant is given by
..jEeff =[1 + (dla).(a2 - b,.ln (wla» (II{£; _1)]-2 (28a)

where
a2 = [0.5173 - 0.1515 In (dla)]2 (28b)
hz = [0.3092 - 0.1047 In (dla)]2 (28c)
The accuracy of (27) and (28) is reported to be within ± 1% for 1 < wla ~ 8,
0.2 ~ d/a ~ 1 and Er ~ 6.
As shown in the case of the suspended stripline at Fig. 2c, suspended and
inverted microstrip lines can also be housed in shielding enclosures. However,
unlike the suspended stripline, in these lines, the top wall is placed sufficiently
high so as not to take part in the propagation.
2.5 Siotline (Fig. It) [3, 30, 31]

The slgtline consists of a slot etched from the conducting layer on one surface
of a dielectric substrate with the opposite surface being bare. The structure is
thus complementary to that of a microstrip. The electric field lines are oriented
essentially across the slot whereas the magnetic field lines have both transverse
and longitudinal components. The dominant mode of propagation is non-TEM
and heDce, the characteristic impedance and guide wavelength are dependent
on frequency, although at a slow rate. As compared with the microstrip, the
slotline.is more dispersive. The slotline mode resembles the TEIO mode of a
rectangnlar waveguide but it differs from the waveguide in that it has no cut-
off frequency.
Based on the analysis of Cohn [30], Garg and Gupta [31] have obtained
closed form expressions for the characteristic impedance and guide wavelength
of slotline by curve fitting thtj numerically computed results. Their formulas

are given below:
For 0.02 S; w/h S; 0.2
Z = 7262 - 15.283 In (e r ) + SO lew/h) - o'~fk(w'h) - 0.1)
+ In {100(w/ h») {19.23 - 3.693 In (e r») - [0.13910 (e r) - 0.11

+ (w/h) (0.465 In (e r) + 144)]· (IL4 - 263610 (e r) - (h/AO) x 102]2
(29)
A/~ =0.923 -0.195 In (tr) + 0.2(w/h) - (0.126(w/h) + 0.02) In(hI~) x IOZ)
(30)
For 0.2 S; w/h S; 1.0
Z:: 113.19 - 23.257 In (tr) + 1.25(w/h) (114.59 - 22.531 In (tr»)
+ 20{(w/h) - 0.2) {I - (w/h») - [0.15 + 0.1 In (tr) + (w/h){- 0.79
+ 0.899 In (tr»)]- {l0.25 - 2.171 In (tr) + (w/h) {2.1 - 0.617 In (tr»)
- (hlJ..o) x IOZ)2 (31)
)JJ..o = 0.987 - 0.21 In (tr) + (w/h) (0.111 - 0.0022tr)

- (0.053 + 0.041(w/h) - 0.OO14tr) In (hlJ..o) x IOZ) (32)
The above expressions (29)-(32) are reported [31] to be accurate within 2%
over the range 9.7 S; tr S; 20 and 0.01 S; hlAo S; (hlJ..o)e, where (hlJ..o)e is the cut-
off value for the TEIO surface-wave mode on the slet line. The value of (h/'Ao)e
can be obtained from
(33)
The slotline is a convenient medium for shunt mounting of discrete devices.
It is particularly suitable for ferrite components that require regions of circularly
polarized magnetic field. One major disadvantage of the slotline is that its Q-
factor is low (=100) as compared with other transmission lines.
2.6 Coplanar Waveguide and Coplanar Strips (Fig. Ig and Ih) [3, 32-34]
The coplanar waveguide is basically a coupled slotline (Fig. Ig) in which the
signal is applied to the centre conductor with respect to the two ground conductors
on the two sides. The configuration of coplanar strips (Fig. Ih) is complementary
to the coplanar waveguide. Unlike the slotline, which is non-TEM in nature,
both the coplanar waveguide and coplanar strips propagate quasi-TEM mode.
At higher microwave frequencies, however, the contribution due to the longitudinal
magnetic field is sufficiently large resulting in non-TEM mode of propagation.
Both coplanar waveguide and coplanar strips have been analytically studied
by several authors [3,32-34]. The expressions for the characteristic impedance
Z and effective dielectric constant £elY of both these structures are given below
[34]:
Coplanar Waveguide (Fig. Ig)
z = 30 te K(k')
(34)
..Jed K(k)
e =1"+ «e _ 1)/2) K(k') K(k1) (35)

elf r K(k) K(ki)
where
Ie::; (sl(s + 2w») (36a)
kl = ({sinh (m-/4h»/(sinh (te (s + 2w)/4h»)) (36b)
k' =...J(1 - k 2) (36c)
k{ =~(1- k?) (36d)
In (35), K(k) represents a complete elliptic function of the first kind and K(k')
=
with k' ...J(1 - k 2 ) is its complementary function. An approximate expression
for K(k)/K(It) which is accurate to 8 ppm is given by
K(k) _ {[(lite) In (2(1 + «)/(1 - «»)]-1 , O:S k:S 0.7

(36e)
K(k') - [(lIte) In {2(1 + ..ff)/{l - ..ff))]-l , 0.7:S k:S 1
Coplanar Strips (Fig. Ih)

Z = 120te K(k) (37)
..JEelf K(k')
E =1 + {(e _ 1)/2) K(k') K(kt ) (38)

elf r K(k) K(k{)
where k, kit k' and k; are given by (36a) to (36d), respectively.

One major advantage of both the coplanar waveguide and coplanar strips is
that they permit simple integration of hybrid components both for series and
parallel connections. In the fabrication of MMICs, these coplanar structures
offer a convenient media in addition to the micros trip. Their Q-factors, however,
are low (.~100) as compared with that of a microstrip (.=250).
3. SUBSTRATES AND TRANSMISSION LINE CHARACTERISTICS

The selection of substrates for MICs depends on three major factors-namely,
the frequency of operation, the type of planar line used and the fabrication
technology; that is hybrid or monolithic. Other factors include, the desired
electrical, mechanical and thermal properties and COSL Hybrid MICs use dielectric
substrates which are generally isotropic, although anisotropic dielectric and
ferrimagnetic substrates are also used for certain special applications.
3.1 Substrates (or Hybrid MICs
General Requirements
The dielectric substrate for hybrid MICs must have the following general properties:
1. Low dielectric loss: The dielectric loss in a cirqJit is directly proportional
to the loss tangent (tan 6) of the substrate material at the operating frequency.
Typically values of the order of 5 x 10-4 or less would be desirable.
2. Fine surface finish: The conductor loss due to the metal mm at microwave
frequencies depends on the degree of surface finish of the substrate (in addition
to the resistivity of the metal), with the requirement on fineness of polish
becoming increasingly stringent with increasing frequency. Smooth surface is
also important for achieving fine line definition of the conducting pattern on
the substrate. The surface finish must be at least of the order of one skin depth
(typically 1 JIID or better at X-band).
3. Low temperature coefficient: In order to ensure stability of circuit

performance with temperature, the variation in dielectric constant and loss
tailgent with temperature must be as low as possible.
4. Uniformity of thickness and dimensional stability: Uniformity of thickness

in substrates from a batch and also their dimensional stability are important
from the point of view of reproducibility of circuit performance. These parameters
are of special importance in circuits realized in suspended and inverted microstrips
because they reflect on the uniformity of air gap between the substrate and the
bottom ground plane which is a critical.parameter in circuit design.
5. Easy machining: From the point of view of circuit manufacture, substrates

that can be cut and drilled easily are preferred.
6. High thermal conductivity: Higher thermal conductivity permits better heat

dissipation in the circuit. This is important in circuits operating at high power
levels.
7. Mechanical stability up to high temperature: This property is important

for circuits incorporating discrete components by soldering or thermo-compression
bonding.
Table 1 lists the properties of some of the substrate materials used in hybrid
MICs. These materials can be classified under three main categories: ceramic,
plastic and ferrimagnetic.
Ceramic Substrates
Ceramic substrates, in general, are hard and can be produced with good surface
polish. These substrates, however, cannot be easily cut or drilled.
Alumina possesses a good combination of dielectric, conductive, mechanical

Table 1. Properties of substrates usable In hybrid MICs
Material £, tan & x 10" Surface Thermal Thermal Dieleclric Special features
(approx.) at 10 GHz roughness conductivity expansion strength
ijun). (W/cm. K) (ppm/K) (KV/cm.)
(a) Ceramic Substrales

Alwnina (99.5%) 9.6-10.4 1 0.05 0.37 6.3 4 x 10' Au-coated substrates, widely used
(polished) in microwave range ~
Fused quartz (99.9%) 3.78 1 0.006-0.025 0.017 10 x 10' Reproducible substrates, minimwn ~
15'
thickness 0.05 mm, smooth surface, 0:;-
useful in mm-range, high cost 1::1
Sapphire £'.1 =9.6 1 0.005-0.025 0.42 6.0 4 x 10' Reproducible substrates, smooth [
(Single crystal) £'11 =11.6 surfaces, anisotropic, high cost ~
C')
Beryllia 6.3 60 0.05-1.25 2.5 6.1 High conductivity, rough surface,
BeD 98% compo\Dld substrate
g
Rutile (TiD:z) 85 40 0.25-2.5 0.05 7.5 Reduced size components, rough 1S'
surface, £, temperature sensitive
~
~
.,
(b) Plastic Substrales
RT-duroid 5880 2.2-2.3 3-7 0,075-1* 0.0026 16-100 1 x 10' C\l-coated, large sheet size, mini- ~
C')
-.
4.25-8.75** mwn thickness 0.1 mm, easily ~
machinable, low cost
RT-duroid 6010 10.2-10.7 10-15 0.0041 Cu-coated, high t;., easily
i
~
machinable, high loss ;;;
Kapton 3-3.5 FleJdble substrates, minimwn thick- ~
OQ
ness 0.025 mm, easily machinable, i:l
low cost
Cu-flon 2.1 4.5 1 x 10' i.C';\
~.
(c) Ferrimagnetic substrales I::
Ferrite/garnet 9-16 10 0.25 0.03 4 x 10' Porous, hard, brittle, anisotropic -.
~
under magnetic bias, useful for non-
~
reciprocal devices VI
*Average peak to valley difference in height (rolled eu).

**Average peak to valley difference in height (electro-deposited Cu).
and thennal properties. It has very low loss, negligible dispersion between
batches but is slightly anisotropic. Alumina substrates for MIC applications are
available in standard thicknesses 0.1, 0.254, 0.635, 1.27 and 2.54 mm with
dimensions in multiples of one inch (1" x 1" to 4" X 4"'). Because of its high
dielectric constant, alumina substrates are used extensively at lower microwave
frequencies.
Fused quartz has a lower dielectric constant than alumina and is available in
substrates with thickness as low as 0.05 mm and with a high degree of surface
finish (=0.006 J.llI1). With these features and its low loss property, fused quartz
offers an excellent medium for use at millimetre wave frequencies. The undesirable
features of fused quartz are, its high cost and highly brittle nature leading to
easy breakage during machining.
Sapphire is the mono-crystalline fonn of alumina. It is dielectrically anisotropic.

The substrate is nonnally cut with the c-axis nonnal to the ground plane. This
gives electrical properties that are independent of the direction of propagation
across the substrate. For electric fields along the c-axis, tTll = 11.6, and for fields
perpendicular to the c-axis, Erl. = 9.6. Its anisotropy is well defined, with negligible
variation from batch to batch, and hence is the most useful of all the anisotropic
materials. Sapphire substrates can be polished to achieve optical grade surface
finish, thereby offering very unifonn and precisely defined conductor patterns.
The material is optically transparent. Its thennal conductivity is about 30%
higher than that of alumina and hence may be preferred for mounting high
power active devices.
In microstrip active devices where high dissipation rates are encountered,
materials with a high thennal conductivity are required. Poly-crystalline BeO
which has a thermal conductivity of the order of 2.5 Wtcm k is ideal in such
applications. 'It can also be used as an insert in a combined substrate where
metal cannot be used for transfer of heat. The disadvantage of this material is
that it is highly toxic demanding special precautionary measures while machining.
Rutile (TiOz) has a very high dielectric constant but is temperature sensitive.
Plastic Substrates
Unlike the ceramic substrates, plastic substrates can be easily machined and
drilled. The most commonly used plastic substrate is the glass-fibre reinforced
polytetrafluoroethylene (PTFE) trade named as RT-Duroid 5880. It has a lower
=
dielectric constant (Er 2.2-2.3) than fused quartz and hence can be used at
millimetre wave frequencies. The loss tangent of this material is three to four
times higher than that of fused quartz, but this would have negligible effect on
circuit perfonnance since the overall loss is governed primarily by the conductor
loss (which is nearly ten times more than the dielectric loss). The thennal
expansion coefficient (: 10 ppmf'C) ofRT-duroid is compatible with the metal
compounds which are used for conductor patterns (: 10-30 ppmf'C for Cu,
Au) whereas that of fused quartz is much smaller (:0.5 ppmf'C). Duroid is
also mechanically stable up to temperatures as high as 350°C. It is a low cost
material and is available in larger sheets with standard thicknesses (1/64",

1/32", 1/16" and 1/4") and with copper metallization (~1O-30 Jlm) on one or
both sides. Although its surface finish is inferior to that of fused quartz, because
of the other advantages enumerated above, RT-duroid 5880 is widely used at
Ku-band and above.
For higher dielectric constant, ceramic-loaded plastic trade named RT-duroid
6010 is available. Its dielectric constant is close to that of alumina but the loss
tangent is nearly ten times higher. Since machining and drilling is easy, this
substrate can substitute alumina during the laboratory prototype development.
Ferrimagnetic Substrates
Ferrites and garnets have relative dielectric constants in the range 9 to 16 and
dielectric loss tangent of the order of 10-3 at 10 GHz. These materials are hard
and brittle. Substrates are normally cut from a bar and ground to the required
thickness and surface finish depending on the device application. In the presence
of a d.c biasing magnetic field, they exhibit gyromagnetic anisotropy thereby
permitting the realization of non-reciprocal microstrip devices.
3.2 Substrates for MMICs

Semiconductor substrates of Si and GaAs have dielectric constants close to that
of alumina but their dielectric loss tangents are nearly 10 to 100 times higher.
Hybrid MICs using these substtates would therefore be very lossy. Their suitability
lies in monolithic MICs where the active devices, passive components and
interconnections are all formed into the bulk or onto the surface of the same
semiconductor material. Table 2 lists the relevant properties of both Si and
GaAs. Of the two materials GaAs has emerged as a versatile and superior
material for MMICs. The resistivities of semi-insulating GaAs (10' to 109
a-cm) is nearly four orders of magnitude higher than that of semi-insulating
Si. A semi-insulating GaAs is therefore blore suitable as a low-loss dielectric
substrate for realizing passive circuitry and also for providing electrical isolation
between active devices in the circuit. The electron mobili~y in GaAs is more
than six times that of Si, thus making it suitable for realizing active semiconductor
devices (MESFETs) for operation at higher frequencies even up to 100 GHz.
The minority carrier life time for GaAs (~10-8 sec) is nearly five orders of
magnitude less than that for Si (~2.5 x 10-3 sec), which makes it well suited
for realizing fast switching p-i-n diodes. The electrical properties of the substrate
can be selected by doping so that it can act either as a semiconducting layer
or as an insulating mechanical support. The high-resistivity properties of semi-
insulating GaAs substrates are utilized for realizing passive components whereas
the high mobility properties of the semiconductor layer that is either epitaxially
grown onto the semi-insulating substrate or implanted into it are utilized for
realizing active devices.
3.3 Selection of Substrates

Considering the properties of various commercially available substrates, it is
clear that there is no single substrate which can meet all the requirements. For
00
"""
~
~
~
Table 2. Properties of substrates used In MMICs
Material t;.
(approx.)
tan I) x 104
at 10 GHz
Surface
roughness
Thennal
conductivity K
Resistivity
(a-em)
Electron
mobility
Thennal
expansion
Dielectric
strength
i
(,urn) (W/cmf'C) (em2/V-sec) (ppm/K) (KV/em)
Si 11.7-12 40-150 < 0.025 0.9 700·

Semi-insulating
Si 11.7-12 10-100 < 0.025 0.9 103-1OS 4.2 300
GaAs 12-13 10-60 < 0.025 0.3 4300·
Semi-insulating
GaAs 12-13 16 =0.025 0.3 107-1Q9 5.7 350
*At 1017/em2 doping.

a given application. one must choose a substrate which offers the best compromise
between the desirable properties.
For hybrid MICs at microwave frequencies. microstrip on alumina substrate
is the most extensively used planar transmission line. Alumina. with its high
dielectric constant keeps the circuit size small and its low loss tangent keeps
the losses low. The choice of substrate thickness depends on the desired impedance
level and the operating frequency. For operation up to about 20 GHz. substrates
of standard thickness 0.635 mm are employed. Typical characteristics of a
microstrip on alumina substrate are discussed in Sec. 3.4. It may be noted from
(21) that as the frequency of operation increases. microstrip circuits must adopt
progressively thinner substrates preferably with a lower dielectric constant.
Another factor which assumes importance as the frequency increases. is the
surface roughness of the substrate. As can be seen from (17) the conductor loss
increases considerably as the skin depth in the metal reduces to the order of
rms surface roughness and lower. Using alumina substrates of smaller thicknesses
0.2 to 0.3 mm a variety of integrated circuit components and subsystems upto
about 50 GHz have been reported [35-36].
=
Fused quartz. with its lower dielectric constant (t;. 3.8) and extremely smooth
surface finish (~0.006 mm which is nearly 50 times smoother than alumina).
and availability in sufficiently small thicknesses offers a superior alternative to
alumina at frequencies above 20 GHz. By using 0.12 mm thick Z-cut quartz
substrates. several microstrip components such as the up-converter and balanced
mixer. have been fabricated and integrated into hermetically sealed receivers
for operation in the W-band (75-100 GHz) [37]. Fused quartz also has the
advantage that it is compatible with most ferrite materials. This property allows
insertion of ferrite discs into a quartz substrate thereby enabling complete
integration of circulators with other components on a single substrate. Fused
quartz substrates. however are very expensive. They are difficult to machine
and drill.
An inexpensive alternative to fused quartz at millimetre wave frequencies is
= =
the copperclad RT-duroid 5880 (Er 2.22) or Cu-flon (t;. 2.1). Unlike fused
quartz. these substrates are flexible. and easy to machine and drill. Using these
substrates in shielded microstrip configuration. compact receivers have been
reported for operation in the V-band (50-75 GHz) [38. 39]. By using 0.127 mm
thick RT-duroid substrate in suspended stripline configuration. integrated circuit
mixers have been realized at W-band (75-110 GHz) [40] and even at D-band
(110-170 GHz) [41].
Among the dielectrically anisotropic materials. sappbfre is the most useful
because of its well defined and repeatable electrical characteristics. availability
in small thickness and optical grade finish. The anisotropic property can be
advantageously utilized for equalizing the even-and odd-mode phase velocities
in microstrip and suspended stripline coupled lines to achieve high directivity
in directional couplers [42]. Microstrip-like transmission lines and coplanar
lines with anisotropic substrates also find applications in electro-optic modulators
and surface-acoustic wave (SAW) applications [43].
Ferrites exhibit magnetic anisotropy under d.c magnetic bias. Planar
transmission lines-namely stripline, microstrip line, slot-line and coplanar lines

on magnetised ferrite substrates find applications in electronically variable phase
shifters [44]. Reciprocal as well ~ non-reciprocal phase shifting is possible by
suitably orienting the magnetic bias field either parallel or perpendicular to the
substrate plane depending on the type of circuit employed. Ferrite based striplines
and microstrip lines are also used in the realization of isolators and circulators [45].
The application of semiconductor materials is in MMICs and among the
various materials, GaAs offers the best properties. Chrome compensated semi-
insulating GaAs offers stable resistivities over the temperature range 850-
900°C which is required for the fabrication of microwave devices. The high
resistivities of the GaAs substrate allows the realization of active devices by
suitably doping either through epitaxial deposition or by ion implan~tion process.
The high electron mobility of GaAs permits short-gate length FET devices to
operate at frequencies as high as 100 GJIz. The GaAs MESFETs have found
wide application in amplifiers, mixers, switches and phase shifters. The versatility
of GaAs MESFETs in a variety of such functional circuits has added to the
popularity of GaAs substrates in MMICs.
3.4 Typical Transmission Line Characteristics

Table 3 provides a qualitative comparison of various planar transmission lines
considered in Sec. 2. The range of impedances and Q-factors listed are typical
values achievable with the commercially available substrates at microwave
frequencies. The upper and lower limits of impedances are governed by two
major factors. First, the accuracy of the photolithographic technique sets tolerances
on the minimum strip and slot widths that can be.achieved. The second factor
is the possible excitation of the higher order modes when the strip width
exceeds about one-quarter wavelength.
Of the various planar transmission lines listed, the most commonly used are,
the microstrip at microwave frequencies, and the suspended substrate transmission
line configurations--namely the suspended stripline and the suspended microstrip,
at millimetre wave frequencies. In the following, we discuss some ~ypical I
characteristics of the structures: .

Figures 3 to 5 illustrate typical variation in the effective dielectric constant,
characteristic impedance Z, and the attenuation constant a (equal to the sum of
the conductor and dielectric attenuation constants), respectively, of a microstrip.
= =
The substrates chosen are, alumina (t; 9.6), fused quartz (t; 3.8) and RT-
=
duroid (t; 2.22). From Fig. 3, we note that £ea of the microstrip increases
with an increase in wlh and also t;. This is an expected result because, in both
the cases, ,the eleCtric fietd concentration in the substrate region increases thereby
increasing £ea. Consequently both the impedance Z and the guide wavelength
A which are inversely proportional to ..JEeff decrease with an increase in wlh
and t; (see Figs. 3 and 4). For a given substrate thickness h and impedance level
Z, the strip width w increases with a decrease in t;. The normally encountered
impedance range in practical circuits is from 20 to 120 n and for these impedance
levels a microstrip on alumina-substrate of standard thickness 0.635 mm yields
convenient strip widths. We also note from (21) that with h = 0.635 mm and
Table 3. Comparison of various planar transmission lines
Configuration Typical Unloaded Dispersion Power Commonly used substrates/special features

impedance Q-factor handling
range (n) (Qu) capability
Stripline 20-120 high negligible high Substrates--RT duroid/for passive components, incon- ~
(- 500) venient for mounting discrete devices .,~
Microstrip line 20-120 medium small high Substrates-illurnina, fused quartz, RT-duroid for iS o
0:;--
(-:: 250) hybrid MICs; GaAs for MMICs, and ferrites for tl
non-reciprocal devices/most widely used transmission ::s
tl.
line at microwave frequencies
,..,~
Suspended stripline 40-150 high small high Substrates--fused quartz, RT-duroid/well suited for ;:so
millimetre wave components, inconvenient for shunt ::s
<:l
mounting of active devices C-
O<:>
Suspended microstrip 40-150 high small high Features similar to suspended stripline '<
Inverted micros trip high medium ~
.,
25-130 small Substrates--alumina, fused quartz, RT -duroid/suit-
able for higher microwave frequencies, inconvenient ~
rio
for mounting active devices
~
Slot line 50-200 low large low Normally used in conjunction with microstrip, suitable ~
tl
(- 100) for shunt mounting of chip deviceso Produces elliptically <:
~
polarized RF magnetic field, useful for non-reciprocal ;;-
devices on ferrite substrates ~
0<:>
Coplanar 40-150 low medium medium Easy connection of series and shunt elements, useful ~
waveguide for MMICs on GaAs and non-reciprocal components ~
tl.
on ferrite C)
~o
Coplanar Strips 40-250 low medium medium Easy connection of series and shunt elements, useful ....,
for MMICs on GaAs ;::
a"
VI
......
Er
L 0.8
2.22 - - _
---------
---- ---~ ---
~-----
-
3.18
6
9.6 -- ---... -. 0.6
-. --- -----
-----
IU 9·6
4
----- -- ..... -.-~ 0.4
3.78
2 0.2
212
-twl-
HiC..'.·.~ . ).:1
o 0.0
0.01 0.05 10
W/h
Fig. 3. Effective dielectric constant EeJf and nonnalized guide wavelength A. / Ao of micro-
strip versus w/h with t;. as parameter. t;. = 9.6 (alwnina), 3.78 (fused quartz),
2.22 (RT-duroid).
=
Er 9.6, the coupling of quasi-TEM mode to the spurious higher order mode
=
occurs at!c,TM 53.2 GHz, thereby ensuring operation only in the dominant
mode in the microwave band. The main factor which sets the limit on the
highest frequency of operation is the attenuation in the line which increases
rather rapidly with an increase in frequency.
From Fig. 5, which illustrates the attenuation versus the height of the substrate
in a 50 n microstrip, we note that the attenuation is much less with fused
quartz and RT-duroid substrates, than with alumina. However, even with these
substrates there is a limit on the frequency upto which microstrip geometry can
be used since the requirement on making the substrates thinner (for ensuring
operation only in the dominant mode) poses problems due to increased conductor
loss and critical dimensional tolerances.
The preblems of high conductor loss and critical dimensions encountered in
microstrips in the millimetre wave range are to some extent circumvented in the
suspended substrate configurations (Fig. 1(c)-(e». As discussed in Sec. 2.3,
introducing an air gap between the substrate and the ground plane reduces the
effective dielectric constant and consequently, the strip dimensions are nearly two
to three times larger than in a microstrip for the same characteristic impedance.
Fig. 6 which shows a typical comparison of the impedance characteristics of a
250~----------~----------~----------~
Er = 2.22 -jWr- h
,······Ir:.:.:.;.·.;.; .j.1
200~----------~----------4-----------~
3.78
1~~--------~~~~----~----------~
.,.
e
.c
9.6
-
o
N 100~-----------+~~----~~~----------~
50~----------+---------~~--~~----~
O~ __________ ~ ____________ - L_ _ _ _ _ _ _ _ _ _~
0·01 0·1 1 10
W/h
Fig. 4. Characteristic impedance Z versus wlh for microstrip with Er as parameter.
microstrip with those of a suspended microstrip and an inverted microstrip

illustrates this feature. The suspended substrate structures therefore offer relaxed
dimensional tolerances as compared with the microstrip. Another important
advantage of the air gap is the reduction in the conductor loss. As illustrated in
Fig. 7, the total dissipative loss incurred in a son microstrip in the 5-10 GHz range
is of the same order as that in a suspended or inverted microstrip around 60 GHz.
Of the three suspended substrate structures (Fig. 1(c)-{e» , the inverted microstrip
is the least utilized, because of the difficulty in incorporating active devices. The
suspended microstrip and the suspended stripline which offer easy accessibility to
the strip conductor pattern have emerged as popular configurations in the realization
of millimetre wave components.
Coplanar structures shown in Fig. 1(t)-{h) tend to be excessively lossy at
millimetre wave frequencies. Their utility is limited essentially to microwave
frequencies. Circuits using slot-microstrip combination can be easily realized
by etching slots on the ground plane side of a microstrip. While the microstrip
impedance decreases with an increase in the strip width, the slot impedance
increases with an increase in the slot width. The slot-microstrip combination,
therefore caters to a wide range of impedance levels, from approximately
20 to 200 n and offers considerable flexibility in the circuit design. The coplanar
waveguide and coplanar strips are suitable for MMICs built on semi-conductor
, - - 10GHz
- -- -- 20GHz
1· 0
--=...,e
'0
a
0· 1
0·01 L..-.-,-L,------~------_f_---~
0·01 0·1 5
h{m m)
Fig. S. Theoretical attenuation constant a ( = ex..: + ad) for a 50 n micros trip line
as a function of substrate height h with £,. as parameter:
I-alumina: E., = 9.6, tan ~ = 10-4, .d = 0.05 J.l1I1
2-fused quartz: E., =3.78, tan ~ = 10-4, .d =O.ot5 Jlm
3--RT-duroid: E., =2.22, tan ~ =5 X 10-4, .d =0.09 J.l1I1
substrates. Both these structures permit mounting of devices in series as well
as shunt configurations. The loss in these structures, however, is higher than
that in a microstrip. The microstrip is therefore preferred over these trans-
mission lines for MMICs and hence is more frequently used than the coplanar
structures.
4 BASIC CIRCUIT ELEMENTS FOR MICs

With the planar transmission lines as the basic transmission media, a typical
MIC incorporates one or more of the follOWing types of circuit elements-
discontinuities, transmission line sections in the form of stQbs and resonators,
planar lumped elements, dielectric disc resonators and transitions. In the following,
we briefly review some of the commonly encountered basic circuit elements in
these categories:
120
Suspended microstrip (a/d=1)
Inverted microstrip (a/d=1)
Microstrip (d =h)
VI 80
e
.c.
o
N
40
OL-______~--------~~-----LI----~--J--
0 6 8
W/d
Fig. 6. Comparison of impedance characteristics of microstrip (Fig. I b), suspended
microstrip (Fig. ld) and inverted micros trip (Fig. Ie), Er = 3.78.
4.1 Discontinuities
Discontinuities in MICs commonly involve a change in the strip conductor
width or slot width. Figure 8 shows some of the frequently encountered
discontinuities in circuits fabricated using the stripline, microstrip and their
variants and Fig. 9 shows the discontinuities encountered in slotlines. Two
types of effects may be identified in these discontinuities. One is the generation
of fringing electric fields at and in the vicinity of a discontinuity which can be
represented in terms of an equivalent capacitance. The second effect is the
change in the normal flow of current which can be represented in terms of an
equivalent inductance.
Microstrip Discontinuities [2, 4, 46, 47]
Open-end
Open-end and short circuit terminations are invariably required in any circuit.
In a microstrip an open circuit can be easily realized by abruptly terminating
a strip conductor whereas a short circuit requires drilling a hole through the
substrate and making a connection to the ground plane. Fig. lOa shows a open-
ended microstrip and the nature of fringing electric fields at the termination. At
lower microwave frequencies, these fringing fields and the resulting increase
in the electrostatic energy in the region of the open end may be modeled as a
capacitor C1 as shown in Fig. lOb. The capacitance, in tum, is equivalent to
extending the microstrip line by a length fll such that the input impedance of
this extended line section is equal to the capacitive reactance at the operating
frequency.
56 MICROWAve MATERIALS
0.25
0.20
Microstrip
-
- 0.15
eu
a:l
0·10
Inverted mlcrostrsp
0.05 -
'~ I aft! = 1) _---
---- .------.
----Z-.-
...
.",. --.",. .",..
,,-' ~.---
....;:::... ___ . Suspended microstrip
......... la/d=1)
'-
10 20 30 40
Frl!q IGHz)
FIg. 7. Comparison of attenuation constant a (= ex" + ad) versus frequency of 50 n
line in microslrip (Fig. Ib), suspended microslrip (Fig. Id) and inverted microslrip
(Fig. Ie)
lime, =Z cot (jJ&) (39)

where Z and fJ are the characteristic impedance and propagation constant,
respectively, of the microstrip line. Thus, as shown in Fig. lOb, the ideal open
circuit lies at a distance til from the physically terminated open end. Equivalently,
in a practical circuit, the physical length of the line must ~ shorter than the
theoretically calculated length of an open circuited line section by the correction
factor til.
Series-gap
A narrow series gap in a microstrip acts as an open circuit for de but presents
a finite impedance at microwave frequencies. The gap can therefore be used as
a dc block in active device circuits and as a RF coupling element between two
strip conductors. As illustrated in Fig. 8b, the series gap may be represented
as a x-network of capacitors. The shunt capacitors account for the fringing
"l
~~ z
~
TCl
~ 'p
la I Open end
(g
~III
~~ I
P P
I
z +Cp +Cp
I
,
z
P P
Ibl Series gap
Ls
111"'1-
~
z,
, T~s Z2
p:
Icl Step change in vidth P "
Lt
,
~
Z Z
I
I
p p
iP
Idl Transverse sUt
,
p
61 I I 61 I
~--P Z
,
I
I
I
'r
-reb
~.
I
: Z
I
~
t
I
P p'
leI Chamfered bend
If) T- junction :
"2
Fig. 8. Examples of discontinuities in the strip conductors of planar transmission
lines shown in Fig.l(a)-{e) and their equivalent circuits.
electric fields from the open end to the ground plane and the series capacitor
accounts for the fringing fields between the two open-ends (see Fig. 11). If the
line widths are different, then the two shunt capacitances would have different
values. For very large gaps, Cg tends to zero and Cp approaches the end-
capacitance of a open-circuited line.
58
a
MICROWAVE MATERIALS
p
=:JL. I
P
la I Shor ted slot
= ,
Zz
I
, 0 I
I
~ IPZ ~
Ibl Inductive strip
z~C: ! 1:n:
p p
P
lei StlP thang. in slot
width
~: :P2 CS
~~ II
,
•
Z Z Z Z
~ I
T I
IIII Transverse slit P, Pz
r'mI
Ls
Z
C
p
I
o
,
r I
<
P2
Cp
z
P,
Ibllnduttive notch
It I Capacitive strip
~ MetaUization
Fig. 9. Examples of slotline discontinuities.
Z,p Z,(1
I Ideal open
~circuit
I
i l
p
p p'
Ground plane
lb)
10)
Fig. 10. (a) Microstrip with an open end showing fringing electric field and
(b) Equivalent end capacitance C, and end correction <1/.
PI IP
I I
E-liIes
Ground plone
Grou nd pi ane
(oJ (bJ
Fig. 11. (a) Microstrip with a series gap and (b) Cross-section showing fringing
E-field lines.
Step Change in Width

Symmetric and asymmetrical step change in the strip conductor width is common
in most microstrip circuits, such as impedance transformers, directional couplers,
filters etc. Fig. 12 shows the nature of current flow lines on the strip conductor
around the step and the fringing electric field lines due to the excess charge
stored at the transition edge. Fig. 8c shows the equivalent circuit. The flexing
of current lines can be modeled as a series inductance Ls and the fringing electric
fields as a shunt capacitance Cs •
Transverse Slit in the Microstrip

A narrow slit in the strip conductor as shown in Fig. 8d can be used to realize
a pure series inductance.
Chamfered Bend
Bends in the strip conductor are generally required in order to accommodate
the circuit optimally in a given substrate area. A sharp bend in the strip gives
60 MICROWAWMATERIALS
Current flow
lines
Ground plane
Fig. 12. SymmetrlC:'microstrip step discontinuity showing current flow lines on the
strip (:Onductor and fringing E-field lines at the step.
rise to excess capacitance resulting in poor VSWR. In practice, the effect of the
discontinuity reactance is minimized by chamfering the bend as shown in
Fig. Se. The equivalent circuit for the bend section between the planes P and
P' can be modeled as a shunt capacitance with transmission line sections of
length .11 on both sides.
T-junction
T-junctions are required in most MICs, such as power dividers, filters, couplers
and amplifiers. At the T-junction, two different transmission lines meet at right
angles giving rise to fringing electric fields with respect to ground and also
distortion in current flow. The junction can therefore be modeled as a three-
port network with shunt capacitance and series inductances as shown in Fig. Sf.
Siotline Discontinuities [3, 5]

The slotline geometry is complementary to that of the microstrip. While a
terminated microstrip represents an approximate open circuit, an abruptly
terminated slot represents an approximate short circuit. Fig. 13a shows the
geometry of a shorted slot and the resulting current flow around the termination.
The additional magnetic energy stored due to the current flow can be modeled
by an inductor L.. The inductor, in tum can be modeled by a section of short
circuited transmission line of length .11 such that the inductive reactance is
equal to the input reactance of the shorted transmission line section. That is
roL, = Z tan (fi.1/) (40)
where Z and f3 are the characteristic impedance and propagation constant,

respectively, of the slotline. Therefore, a physical short at a certain plane P can
be considered to produce a perfect electrical short at a plane P' which is .11
away from the plane P (See Fig. 13b).
The equivalent circuits of several other slotline discontinuities are shown
in Fig. 9. In general, any rerouting of current flow due to the discontinuity
giving rise to excess magnetic stored energy is modeled as an inductor and any
Metalization
Current
flow lines
~lS
OR
'p
~ ,,
p p'
(a) (bl
Fig. 13. (a) Slotline with a shorted end showing current flow lines around the
short and (b) Equivalent end inductance L. and end correction til.
charge storage giving rise to excess electric stored energy is modeled as a

capacitor.
4.2 Lumped Constant Elements [48-52]

Lumped constant elements by definition must be very small in size; typically
less than )J20. Using the planar technique, lumped elements-namely, inductors,
capacitors and resistors have been realized up to about 18 GHz [48, 49]. At
lower microwave frequencies where the circuits built in microstrip or other
planar transmission lines become very large, the use of lumped inductors and
capacitors can reduce the circuit size to a considerable extent. The lumped
elements also facilitate achieving broadband performance in circuits by virtue
of the fact that their values remain fairly independent of frequency. Lumped
elements are more prevalent in MMICs than in hybrid MICs since they are
easily implementable in miniature from using the monolithic technology.
Planar Inductors
Figure 14 illustrates some typical layouts of planar inductors. The simple form
of planar inductor is a short section (I « 1../4) of very high impedance line
terminated in very low impedance lines as shown in Fig. 14a. A simple relationship
between the length I of the section and its inductance L can be obtained from
the expression for the input impedance of a section of short-circuited transmission
line. It is given by
PI = sin-1 (mLrz) (41)
where Z and P are the characteristic impedance and propagation constant,

respectively, of the hig.h impedance line. In microstrip, the characteristic
impedance of the inductive line is generally in the range 90 to 110 n and that
of the low impedance lines is in the range 10 to 15 n.
~22222222~
t-t«~
(a) (b)
(c) (d)
L R
Oo---'lW ''/1lIII0----0
O~-----------O
Equivalent circuit for (a) - (d)

Fig. 14. Configurations of planar inductors: (a) Short section of high impedance
line; (b) Single tum loop; (c) Spiral; and (d) Meander line.
A thin strip conductor on a dielectric substrate without the ground plane (or
with no influence of the ground plane) also acts as an inductor. Single turn
loop inductors of the form shown in Fig. 14b can be used for realizing lower
inductance values in the range 0.5 nH to about 3 nH. Spiral (Fig. 14c) and
meanderline (Fig. 14d) inductors depend on the mutual coupling between the
adjacent line segments to achieve higher inductance values upto about 50 nH
in a small area. Values higher than this are difficult to achieve in lumped form
because of the inter-element fringing capacitance. The Q-factor for a spiral is
typically 100 which is higher than that for a straight ribbon.
Planar Capacitors
A lumped capacitor can be realized as a smaUlength (I « i./4) oflow impedance
line (conductor patch) terminated in very high impedance lines as shown in
Fig. 15a. If Z and fJ denote the characteristic impedance and propagation constant,
Ibl
lei Idl
Dielee trie
~tilm
lei
~, If I film
~ Metallization
Fig. 15. Configurations of planar capacitors: (a) Short length of low impedance line
in microstrip; (b) Small gap in a micros trip conductor; (c), (d) Interdigital
line and (e), (f) Dielectric film overlay.
respectively, of the line, the length I of the·line can be obtained from the expression
for the input impedance of a section of open-circuited transmission line. That
is,
PI = sin-1 (roCZ) (42)
where the value of Z should be low (typically 10-15 n in microstrip). Such

capacitor patches cascaded alternately with high impedance inductor lines
(Fig. 14a) are commonly used as low pass filters in bias networks.
A narrow series gap in a microstrip conductor is useful for realizing very
low capacitances of the order of a fraction of a pF. Higher values up to a few
picofarads can be achieved in the interdigital capacitors shown in Fig. 15c and
d. In these capacitors, the capacitance is formed by the fringing field between
the interdigital gaps. The capacitance configurations shown in Fig. 15(a)-(d)
require a single metallization layer. Higher values of capacitances can be
obtained in overlay structures which use dielectric films such as silicon dioxide
(Fig. 15 (e) and (1). These overlay structures behave as parallel plate capacitors
if the film thickness is very small (0.5-1 Jlffi). Such capacitors offer capacitances
as high as 10 to 30 pF over a small area (= I mm 2) and can be easily realized
in monolithic technology.
Planar Resistors
Planar resistors are used in attenuators, as terminations in directional couplers,
power dividers/combiners and circulators and also as isolation resistors in power
dividers. Fig. 16 shows typical configurations of a series resistor in a microstrip,
terminating resistor for a microstrip and an implanted resistor in MMIC. These
resistors are made of high resistivity metal films or cermet films on a dielectric
R
-c::::::J---
Ground
Z Z
plane
la)
~Cr
Ib)
Hltal Hltal
RlSistor film
Ie)
Flg.I6. Configurations of planar resistors: (a) Series ~istor in amicrostrip;
(b) Microstrip tenninated in a resistor and (c) Implanted resistor
film for MMIC.
substrate. Resistors for MMICs are realized by forming isolated semiconductor

films (of appropriate thickness O.0S-{).S J.l.m) on a semi-insulating substrate.
The resistance value for very thin films « 1 J.l.m thick) can be obtained from
R =p.l/wd (43)
where Ps is the specific resistivity (O-cm) of the resistor film, and the dimensions
I, w, d, which denote the length, width, thickness, respectively of the film are
expressed in cm.
4.3 Resonators
Resonators for MICs can be classified into the following: lumped element
resonators, transmission line resonators and dielectric resonators.
Lumped Element Resonators

These are fabricated by combining the planar printed inductors and capacitors
described in Sec. 4.2. As examples, Fig. 17a shows a series resonant circuit
and Fig. 17b shows a parallel resonant circuit made up of a single turn inductor
and an interdigital capacitor. These resonators have rather low Q values typically
less than 100 over 4-12 GHz.
la) Ib)
Fig. 17. Examples of lumped resonant circuits: (a) Series resonant circuit and
(b) Parallel resonant circuiL
Transmission Line Resonators

Figure 18a shows a rectangular resonator and Fig. 18b shows a ring resonator
in microstrip. Their counterparts in slotline are shown in Fig. 18c and d. The
rectangular shaped resonators are
commonly used in filter networks. As discussed
in Sec. 4.2, t!te electrical length of these resonators is longer than the physical
length owing to the end-effect. For a half-wavelength resonator in microstrip
or slotline, resonances occur when
I + 2.11 =)./2 (44)
la) Ib)
Slat
Ie) Id)
Fig. 18. Examples of transmission line resonators: (a) Half-wavelength micros trip
resonator; (b) Microstrip ring resonator; (c) Half-wavelength slot resonator
and (d) Slot-ring resonator.
where I is the physical length of the resonator, AI is the end correction due to
the fringing fields at the termination and It is the guide wavelength in the
transmission line.
Unlike the rectangular resonators, the ring resonators do not need any end
correction. The resonance condition for the ring resonator is given by
n(2a + w) =nit (45)
where a is the radius of the inner ring, w is the width of the strip/slot and n
is an integer indicating the number of azimuthal full-wave variations of the
field. Ring resonators are useful in measurement techniques, particularly for
measuring the effective dielectric constant of planar transmission lines [53, 54].
Dielectric Resonators
Dielectric resonators for MIC application are made of low loss, high dielectric
constant (Er :-30-100) ceramic mixtures such as the titanates and zirconates
(e.g. Ba2Ti9020, (Zr-Sn) Ti04). Resonators made of these materials have two
important advantages over the lumped and transmission line resonators; viz-
very high Q-factors in the range 3000-10000 and excellent temperature stability.
They are widely used in low loss, narrow band filters and temperature stabilized
oscillators [55, 56].
For most applications, resonators are fabricated in the form of a cylinder,
although othor shapes such as the ring and rectangle are also used. Figure 19a
shows the geometry of an isolated cylindrical resonator. The commonly used
resonant mode is the lowest order circular symmetric mode and is denoted as
TEol6- The electric field lines form concentric circles around the z-axis and the
magnetic field lines lie in the meridian plane as shown in Fig. 19b. The resonant
frequency of a resonator is determined by its dimensions, relative dielectric
constant Er and the surrounding medium. For an isolated resonator, the resonant
frequency for the TEo16 mode is given approximately by [56]
Ir (GHz) = ~ 3~ [(aId) + 3.45]; a, d in mm (45)

"'VEr
where a and d are radius and height, respectively of the resonator. This relation
is reported to be accurate within about 2% in the range 0.5 < aid < 2; 30 < Er
< 50. As an example, typical dimensions of a cylindrical TEo16 mode resonator
having Er = 38 and resonating at 10 GHz are a = d = 2.45 mm.
Figure 19c shows the arrangement of a dielectric resonator coupled to a
microstrip. The TEo16 mode is easily excited because of the magnetic coupling
between the strip conductor and the dielectric resonator. The amount of coupling
can be adjusted by varying the lateral distance between the strip conductor and
the resonator. It may be noted that the presence of the microstrip alters the
resonant frequency of the resonator and also lowers its Q-factor [56].
4.4 Circuit Examples

The basic circuit elements considered in the preceding sections can be combined
in different ways to realize a variety of MIC components. Fig. 20 shows four
~-linIS
"
,,-,~
',' "
... -,
\
: /~ 1 ,-\ \H-lines
f ttl Ert i 1 \!
\ \ •• I' ' \ _/ '
\ ....... ,'1 \
_; "'-- /'
I
lal Ibl
ii=-lines Stri
con~uctor
,..!~-,
, .... -', '. Dielectric
r-T.."........,.,="'~.!..!3;.~'1"'!.,.:,'.,t.,_'_~ substrate
Icl
Fig. 19. Examples of a dielectric resonator in MIe: (a) Isolated disc

resonator; (b) Field lines for the dominant TEol8 mode and
(c) Resonator coupled to a microstrip line.
typical examples. The low pass filter shown in Fig. 20a is made up of series
inductors and shunt capacitors cascaded in an alternate fashion. The series
inductor is realized as a short section of high impedance line (narrow strip
width) and the shunt capacitor is realized as a short section of low impedance
line (wide strip). Figure 20b shows a four-port circuit incorporating a ring
having a circumference of 3)"/2. This circuit is popularly known as the rat-race
hybrid. Ports 1 and 3 are mutually isolated and so are ports 2 and 4. For a
signal fed to port I, equal power division takes place between ports 2 and 4
with signals differing in phase by 180°. For a signal fed to port 3 also, equal
power division takes place between ports 2 and 4 but the two signals remain
in phase. When two different signals (VI and V3) are fed to ports 1 and 3, the
sum signal appears at port 2 and the difference signal appears at port 4. These
properties are used in the design of balanced mixers, p-i-n diode phase shifters,
and comparator networks.
Figure 20c shows a 3 dB-power division network illustrating the use of a thin
film resistor for achieving good isolation between the two output ports (marked
2 and 3). Figure 20d is an example of a directional coupler circuit employing
coupling between a microstrip and a slotline to achieve broadband coupling
performance. The solid lines indicate the micros trip conductor pattern and the
10 I
~
-jv/If
Ie I Idl
Fig. 20. Examples of MIC components: (a) Low-pass filter. (b) Rat-race hybrid.
(c) 3 dB power divider. (d) Microstrip-slot direction81 coupler.
dotted lines represent the slot pattern etched from the ground plane side of the
microstrip. The slot line is parallel to the micros trip with its two ends terminated
in open circuits (indicated by dotted circles). Power fed to port 1 is equally divided
between ports 2 and 4 but the two signals differ in phase by 900 • Port 3 forms an
isolated port.
Figure 21 shows a typical microstrip layout of a two-stage FET amplifier
which includes in it step junctions. parallel-coupled line (which also serves as
a d.c block) and bias lines. The circuit incorporates d.c. block chip capacitors
at the microstrip gaps marked CI. Figure 22 shows the photograph of the
assembled microstrip amplifier.
5. FABRICATION TECHNOLOGY
Two different technologies; hybrid and monolithic have proved important for
the fabrication of MICs. Hybrid MICs permit the use of a wide variety of
planar transmission lines described in Sec. 2, whereas for MMICs, the
configurations are rather restricted to the microstrip and coplanar structures by
virtue of the process technology. The substrate materials appropriate for use in
the two technologies are already covered in Sec. 3. In this section, we provide
a brief review of the choice of other materials and some key features of the two
fabrication technologies [57-59].
Ground
Grou nd
o.c block FE! 1

cop oc i to r
~
Fig. 21. Layout of a two-stage FET amplifier in microstrip configuration.
Fig. 22. Photograph of the two-stage amplifier.
5.1 Materials
Conductor materials
Conductor materials for the MICs must have the following desirable properties:
• high conductivity or low RF resistivity so that the ohmic loss is minimized.
• low temperature coefficient of resistance.

• strong adherence to the substrate material.
• good solderability and adaptability to bonding processes (such as thermo-
compression and ultrasonic bonding).
• resistant to oxidation.
Table 4 presents a list of conductor materials useful for MICs. The materials
listed under category A (Ag, Cu, Au, AI) are good conductors and are used
as conducting layers in MICs. For a given operating frequency, the thick-
ness of the deposited layer must be about three to six times the skin depth.
These conductors adhere well to the synthetic substrates but on ceramic
substrates such as alumina, the adhesion is very poor. The materials listed at
category B (Cr, Ta, Ti) have good adherence property but poor conduc-
tivity and those at category C(Mo, W) have moderate conductivity and fair
adhesion. The materials at category D (Pt, Pd) are barrier metals and are used
to separate a good conductor layer (category A) from the adhesive layer
(category B).
Table 4. Conductor materials for MICs [7, 57J
Material Conductivity Skin depth 0 Coefficient of Remarks

(1 X 10-7 at 10 GHz thermal expansion
(mhos/m) (Jltn) a.r (f'C X 106)
Category A
Ag 6.1 0.64 21 Good conductivity, poor
Cu 5.8 0.66 18 adhesion to ceramic
Au 4.1 0.79 15 substrates
AI 3.6 0.84 26
Category B
Cr 0.76 1.82 8.5 Poor conductivity, good
Ta 0.64 1.98 6.6 adhesion to ceramic
Ti 0.21 3.5 9.0 substrates
Category C
W 1.8 1.19 6 Moderate conductivity,
Mo 1.75 1.20 4.6 fair adhesion
CategoryD
Pt 0.94 1.64 9 Barrier metals as separa-
Pd 0.94 1.64 11 tors between categories
A and B
Substrates for MICs are metallized in two-or three-layers. The metal systems
commonly employed are, Cr-Au, Cr-Cu-Au, Ta-Au or Pd-Au for hybrid MICs
on ceramic substrates and Cr-Au, Ti-Pt-Au and Ti-Pd-Au for MMICs. These
materials are commonly deposited using evaporation or RF sputtering. For
MMICs, materials such as Ta, Mo and Pt are also deposited using the electron
beam evaporation.
Dielectric Films
Dielectric films are used in capacitors as overlays, in passive components such
as couplers and protective layers for active devices. The properties of commonly
used capacitances in the range 0.2 pF-lOO pF are required in MICs. Thin-film
dielectrics in MICs are listed in Table 5. The desirable properties for these
dielectrics are, reproducible dielectric constant, low loss tangent, sufficient break-
down voltage and low probability of developing pin holes. The quality factor
of silicon monoxide (SiO) is about 50 whereas values higher than 100 are
achievable in silicon-dioxide (SiOV, tantalum pentoxide (Ta20S) and silicon
nitride (Si3N4). These materials can be deposited using the techniques mentioned
in Table 5. For higher power MICs, breakdown voltages in excess of about
200 V are required and these are generally obtained with larger film thicknesses
in the range 0.5 to 1 J.LrD.
Table 5. Dielectric film materials for MICs [7]
Material Loss tangent Dielectric Deposition technique

(tan ~) strength in
106 V/cm
SiO 6-8 0.01~.03 0.4 Evaporation
Si~ 4 0.001~.01 10 Chemical vapour depo-
sition (CVD)
Ta20s 22-25 0,01 6 Anodization, evaporation
A12O, 7-10 0.2~.24 4 Anodization, evaporation
Si3N4 6.5-7.6 10 CVD, sputtering
Resistive films
Examples of planar resistors for hybrid MICs and MMICs are illustrated in
Fig. 16. The desirable properties for the resistive films are, good stability, low
temperature coefficient of resistance and good heat dissipation capability. The
layer thickness is generally less than one skin depth so that the resistance value
is independent of frequency. Table 6 lists the properties of important resistive
materials. Sheet resistivities in the range 5 to 1000 O/square are achievable with
these materials. Of the various materials listed, nickel-chrome (NiCr) and tantalum
(Ta) are preferred because of their high resilience against environment, heat
and aging.
Table 6. Resistor film materials for MICs [7, 58]
Material Resistivity Temp. coeff. of Stability Deposition

(ohms/square) resistance ('fore) technique
Cr 10-1000 - 0.1 to 0.1 Poor Evaporation

NiCr 40-400 0.002 to 0.1 Good Evaporation
Ta 5-100 - 0.01 to 0.01 Excellent Sputtering (in A.N)
Ti 5-2000 - 0.1 to 0.1 Fair Evaporation
Cr-SiO 5-600 - 0.005 to - 0.02 Fair Evaporation
(or cennet)
S.2 Hybrid Technology

The fabrication of a hybrid MIC basically involves the following three steps:
(a) metallization of the substrate, (b) photo-etching the circuit conductor pattern
and (c) soldering/bonding the discrete passive and active devices.
Figure 23 illustrates a typical thin film hybrid process involved in realizing
a microstrip pattern on alumina substrate.
Metallization
The substrate is first chemically cleaned in acetone and sulphuric acid. The
metal combination commonly adopted for alumina is Cr-Cu-Au. In order to
achieve good adhesion a seed layer of Cr of about 150A is first deposited
followed by a thin conducting layer (0.1 JlI11) on the substrate using either simple
evaporation or RF sputtering. The high conductivity Cu-Iayer is built up to
about 10 JlI11 either in the same vacuum run or by electroplating. Finally a flash
A. Metallization
k, :, 1. Uncoated alumina substrate, Clean
2. RF sputter both sides
Cr (-= 150A)-Cu (-= 10Mm)-Au (.:- 150A)
OR
RF sputter Cr (-= 150A-Cu (-= 0.1Mm)
i
Electroplate Cu (.::: 10Mm)-Au (-= 150A)
Photoruist 8. Etching the pattern
. ss s." '."'''''' . l - Cr -(U- Au 3. Apply (-ve) photoresist, spin, dry cure at
~
_ .' ... .. ., ' :...:. high temperature
E == = = : : : J l - cr-Cu- Au
4. Place photomask containing the pattern

Expose to UV light
., g P'''''''';'' 5. Develop by removing unexposed photoresist
....----- Ph ot ortsist
[",E;,~
Apply protective lacquer on lower surface,
Etch Au, Cu, Cr from top in sequence
P t ' 6.
~ ro tchvt
:E[
lac quer
St,i P
I '. '
1 0 . ,_ _ _. . . . .
..
."~
_~;...
.;
..,'.;.,
'
_ .' . .
. .
conductor 7.
(C r -eu-AU) Remove photoresist and lacquer
Ground plant
(Cr- Cu-AU)
Fig. 23. Typical hybrid MIe process for realizing a micros trip.
of Au (150 A) is given in order to prevent the Cu-surface from oxidation. The

Au-top surface is also excellent for ultrasonic bonding whereas a Cu-top surface
usuaIIy requires soldering for connecting active devices.
Etching the Pattern

The etching process for defining the conductor pattern involves different stages
owing to the different metaIIic layers. The substrate is coated with a photo-
resist (usually negative photo-resist) using a spinning system. The photo-resist
film must be thin for better resolution. The substrate is then dried and cured
at high temperature. The photo-mask containing the microwave circuit pattern
is placed on the substrate and exposed to ultra-violet light. The substrate is
then placed in a photo-resist developer. This wiII remove the photo-resist from
the exposed area and leave the resist on the desired pattern. In the case of a
microstrip pattern, the lower metaIIized surface which is to serve as a ground
plane is coated with a protective lacquer. The undesired metal film layers--Au,
Cu and Cr on the top surface are then etched away using suitable etchants in
sequence (Au etch: potassium iodine mixed with iodine; Cu etch: ferric chloride;
Cr etch: sodium hydroxide mixed with ferri-cyanide). Finally, the photo-resist
and lacquer are removed leaving the desired conducting pattern. Resistive films
if required as part of the circuit can be deposited on the substrate using evaporation
or sputtering.
When commercially available Cu-clad plastic laminates such as RT-duroid
are used, the first metaIIization step is eliminated. After cutting the laminate to
the desired size, it is cleaned first with a fine abrasive powder or metal polish
and then in acetone. The procedure for photo-etching the desired pattern is the
same as that described for ~lumina substrate except that a suitable etchant is
used for etching the copper. The pattern can then be electroplated with a thin
layer of Au in order to prevent Cu from oxidation.
Attachment of discrete components on plastic substrates is easy since the
substrate can be easily drilled to fit diodes and transistors which need one of
the terminals to be connected to ground. Several techniques are available for
attaching discrete devices-namely soft soldering, epoxy bonding using silver-
or gold-based epoxies and wire bonding using either themo-compression or
ultrasonic bonding. The first two techniques are commonly used for plastic
substrates and bonding is used for gold-coated alumina substrates.
5.3 Monolithic Technology

Unlike the hybrid technology which requires a single level metallization process,
the monolithic technology involves multi-level process for incorporating all
active devices, passive circuit elements and interconnections into the bulk or
onto the surface of a semi-insulating semiconductor substrate. Figure 24 iIIustrates
a typical MMIC process on GaAs substrate. The following are the process steps.
Active Layer Formation

The first step in the MMIC process is formation of an active layer (n-type) on
or into a semi-insulating GaAs substrate. Techniques used are either ion-implan-
74 MICRowAVB MATERIALS
,------'----------;;..z;;f-,_
- layer
I,Attive
'-r-_--_-_-_se_-~_i~_i~_;U_-I~_t_i~9_--_G_~_S--------' 1. Active layer formation
_nO -\L_ _ _ _ _ _-,

2. Mesa etch
s 0
(--- -----?::\ 3. Ohmic contacts: Source (S) and
drain (D)
Au-Ge/Ni evaporation or alloying
S G 0
(n~_
0""""
n-\L._ _ _ _ _--, 4. Schottky-gate (G) formation (Ti-Pt-
Au)
5. First level metallization (Ti-Pt-Au.

Mo-Au)
Inductor (L). lower plate of capacitor
(C). S-D overlay
[------:::6,
""i'!eno~
rzzza
J I
== 6. Resistor (R) formation
~ 7. Dielectric film/passivation
8. Second level metallization/air-bridge

interconnection
9. Via grounding backside metallization
Fig. 24. Typical GaAs MMIC process.

tation or epitaxial growth. In the ion-implantation technique, dopant atoms are

bombarded on the GaAs substrate and in the epitaxial technique, an additional
n-Iayer film is grown onto the substrate surface. As compared with the epitaxial
technique, ion implantation offers better control of doping profiles over a larger
substrate area and improved reproducibility.
Mesa Etch
Mesa etching is a process by which active device areas are isolated so that
current flow is restricted to the desired portions. This is achieved by etching
away the surface leaving mesas of active layer at the desired locations. The
thickness of the material removed is on the order of 0.8 p.m.
Ohmic Contact
Ohmic contacts are formed on the active n-Iayer by sequential evaporation of
Au, Ge and Ni followed by lift-off and alloying in H2 ambient at around 450°C.
The total layer thickness is approximately 20ooA.
Schottky-gate Formation
The gate material must possess properties of good adhesion to GaAs, good
electrical conductivity and thermal stability. The metal system that is commonly
used for gate formation in GaAs MMICs is Ti-Pt-Au. Gates are normally
defined by contact photolithography, and submicron gates by electron beam
lithography.
First-level Metallization
This is a overlay metallization which forms inductors, lower plates of MIM
(metal-insulator-metal) capacitors, transmission line sections and provides overlays
for ohmic contacts. The preferred metal systems are Ti-Pt-Au, Mo-Au, Cr-Pt-
Au. For example, sequential layers of Ti, Pt and Au are deposited and patterned
by lift-off. The pattern can also be formed by ion-milling through a photo-
resist mask to remove the unwanted metal. The thickness of this layer is typically
6OOOA.
Resistor Formation
Resistors are formed by using GaAs material or resistive films such as Ni, Cr,
Ti, NiCr and TaN. GaAs resistors are formed between two ohmic pads by
using the existing isolation and ohmic contact masks. Thin film resistors made
of materials such as Ni, Cr, TaN etc are realized by sputtering.
Dielectric Film Deposition and Passivation

Dielectric films are used as dielectric layers for MIM capacitors, cross-over
ins~lation and passivation of active device areas and resistors. The dielectric
materials commonly used for this purpose are silicon nitride (Si3N4) and silicon-
dioxide (SiOv. These are normally deposited either by plasma-enhanced chemical
vapour deposition or sputtering.
Second Level Metallization

The second level metallization is used to form top plates of capacitors,
interconnection of components and air-bridges. The metal combination used is
Ti-Pt-Au. These materials are evaporated in sequence and patterned by lift-off.
Gold electroplating to a thickness of about 10--15 J.lID is used for defining air-
bridge connections and for realizing transmission lines.
Via Grounding and Backside Formation

After completing the process of defining the circuit on the top surface, the
GaAs substrate is thinned to about 100-200 J.lID. The exact thickness is dictated
by the transmission line impedance. Next, via holes are etched through the
substrate using ion etching and the entire backside including the holes is metallized
to form the RF ground plane.
REFERENCES
1. H. Howe, Stripline Circuit Design, Artech House (MA), 1974.

2. B. Bhat and S.K. Koul, Stripline-like Transmission Lines for Microwave Integrated
Circuits, Wiley Eastern Ltd. (New Delhi), 1989.
3. K.C. Gupta, R. Garg and U. Bahl, Microstrip Lines and Slotlines, Artech House
(MA),1979.
4. T.C. Edwards, Foundationsfor Microstrip Circuit Design, John Wiley and Sons
(NY),1981.
5. I. Frey and K. Bhasin (Eds.), Microwave Integrated Circuits, (Reprint volume),
Artech House (MA), 1985.
6. T. ltoh (Ed.), Planar Transmission Line Structures, IEEE Press, NY, 1987.
7. Leo Young (Ed.), Advances in Microwaves, Vol. 8, Academic Press, NY, 1974.
8. M.A.R. Gunston, Transmission Line Imped4nce Data, Van Nostrand Reinhold
Co., (Lond.), 1972.
9. S.B. Cohn, Problems in strip transmission lines, IRE Trans. Microwave Theory
Tech., Vol. MTI-3, pp. 119-126, March 1955.
10. S. Ramo and I.R. Whinnery, Fields and Waves in Communication Electronicl,
John Wiley & Sons, NY, 1965.
11. M.V. Schneider, Microstrip lines for microwave integrated circuits, Bell Syst.
Tech, J., Vol. 48, pp. 1421-1444, May-June 1969.
12. H.A. Wheeler, Transmission line properties of parallel strips separated by a
dielectric sheet, IEEE Trans. Microwave Theory Tech., Vol. MTI-13, pp. 172-
185, March 1965.
13. H.A. Wheeler, Transmission line properties of a strip on a dielectric sheet on
a plane, IEEE Trans. Microwave Theory Tech., Vol. MTI-25, pp. 631-647,
Aug. 1977.
14. E.O. Hammerstad, Accurate models for microstrip computer-aided design, IEEE
MTI-S Int. Microwave symp. Digest, pp. 407-409,1980.
15. P. Bhartia and U. Bahl, Millimetre Wave Engineering and Applications, John
Wiley & Sons, NY, 1984.
16. EJ. Denlinger, A frequency dependent solution for microstrip transmission
lines, IEEE Trans. Microwave Theory Tech., Vol. MTI-19, pp. 30-39, Jan.
1971.
17. E. Yamashita. K. Atsuki and T. Ueda. An approximate dispersion formula of

micros trip lines for computer-aided design of microwave integrated circuits.
IEEE Trans. Microwave Theory Tech.• Vol. MTT-27. pp. 1036-1038. Dec.
1979.
18. R.A. Pucel. OJ. Masse and C.P. Hartwig. Losses in microstrip. IEEE Trans.
Microwave Theory Tech.. Vol. MTI'-16. pp. 342-350. June 1968. Also correction.
Vol. MTT-16. pp. 1064. Dec. 1968.
19. EJ. Denlinger. Losses in microstrip lines. IEEE Trans. Microwave Theory Tech.•
Vol. MTT-28. pp. 513-522. June 1980.
20. B.E. Spielman. Dissipation loss effects in isolated and coupled transmission
lines. IEEE Trans. Microwave Theory Tech.• Vol. MTT-25. pp. 648-656. Aug.
1977.
21. U. Bahl and O.K. Trivedi. A designer's guide to microstrip line, Microwaves.
Vol. 16. pp. 174-182. May 1977.
22. M.V. Schneider. Dielectric loss in integrated microwave circuits. Bell Syst.
Tech. 1.. Vol. 48. pp. 2325-2332. Sept. 1969.
23. S.B. Cohn and G.D. Osterhues. A more accurate model of the TEIO type waveguide
mode in suspended substrate. IEEE Trans. Microwave Theory Tech., Vol. MTT-
30. pp. 293-294. March 1982. Also correction in IEEE Trans. Microwave theory
tech., Vol. MTT-30. p. 1291. Aug. 1982.
24. F.E. Gardiol. Careful MIC design prevents waveguide modes. Microwaves.
Vol. 16. pp. 188. 190-191. May 1977.
25. 1.1. Smith, The even-and odd-mode capacitance parameters for coupled lines in
suspended substrate. IEEE Trans. Microwave Theory Tech., Vol. MTT-19.
pp. 424-431. May 1971.
26. U. Bahl and P. Bhartia. Characteristics of inhomogeneous coupled stripline.
IEEE Trans. Microwave Theory Tech., Vol. MTT-28, pp. 529-535, June 1980.
27. T.S. Shu, Xiao-xia Qi and Yun-yi Wang, Analysis equations for shielded suspended
substrate microslrip and broadside-arupled slriplines, IEEE MTT-S Int. Microwave
Symp., Digest. pp. 693-696, 1987.
28. S.K. Koul and B. Bhat, Characteristic impedance of microstrip-like transmission
lines for millimetre wave applications, Arch. Elek. Ubcrtragung., Vol. AEU-35,
pp. 253-258. June 1981.
29. P. Pramanick and P. Bhartia, CAD models for millimetre-wave rm-line and
suspended substrate microstrip lines. IEEE Trans. Microwave Theory Tech.•
Vol. MTT-33, pp. 1429-1435, Dec. 1985.
30. S.B. Cohn, Slotline on a dielectric substrate, IEEE Trans. Microwave Theory
Tech., Vol. MTT-17. pp. 768-778, Oct. 1969.
31. R. Garg and K.C. Gupta, Expressions for wavelength and impedance of slotline,
IEEE Trans. Microwave Theory Tech., Vol. MTT-24, p. 532, Aug. 1976.
32. C.P. Wen. Coplanar waveguide: A surface strip transmission line suitable for
non-reciprocal gyromagnctic device applications, IEEE Trans. Microwave Theory
Tech.• Vol. MTT-17, pp. 1087-1090. Dec. 1969.
33. J.B. Knorr and K.D. Kuchler, Analysis of coupled slots and coplanar strips on
dielectric substrate. IEEE Trans. Microwave Theory Tech., Vol. MTT-23 ,
pp. 541-548, July 1975.
34. G. Ghone and C. Naldi. Analytical formulas for coplanar lines in hybrid and
monolithic MICs. Electron. Lett., Vol. 20. No.4. pp. 179-181. Feb. 1984.
35. Y. Tokumitsu et al. 50 GHz IC components using alumina substrates. IEEE
Trans. Microwave Theory Tech., Vol. MTT-31, pp. 121-128, Feb. 1983.
36. T. Nakagami, H. Yatsuka and M. Ishizaki, Millimetre wave integrated circuits
using alumina substrates, Fujitsu Sci and Tech. I., Vol. 14, No.4, pp. 33-43,
Dec. 1978.
37. P.L. Lowbridge, C. Burnett and B. Prime, W-band high reliability hennetically
sealed microstrip receiver, Military Microwave Conf., London, Conf. Digest,
pp. 502-511, Oct. 1984.
38. M. Dydyk and B.D. Moore, Shielded microstrip aids V-band receiver designs,
Microwaves, Vol. 2, pp. 77-82, March 1982.
39. K. Chang et al, V-band low-noise integrated circuit receiver, IEEE Trans.
Microwave Theory Tech., Vol. MTT-31, pp. 146-154, Feb. 1983.
40. R.S. Tahim, G.M. Hayashibara and K. Chang, Design and performance of W-
band broadband integrated circuit mixers, IEEE Trans. Microwave Theory Tech.,
Vol. MIT-31, pp. 277-283, March 1983.
41. L.Q. Bui, N. Ton and D. Ball, A D-band millimetre-wave cross-bar mixer,
IEEE MIT-S Int. Microwave Symp.l>igest, pp. 55S-556, 1984.
42. N.G. Alexopoulos and S.A. Maas, Characteristics of microstrip directional couplers
in anisotropic substrates, IEEE Trans. Microwave Theory Tech., Vol. MIT-30,
pp. 1267-1270, Aug. 1982.
43. S.K. Koul and B. Bhat, Generalized analysis of microstrip-like transmission
lines and coplanar lines with anisotropic substrates for MIC, electro-optic
modulator and SAW applications, IEEE Trans. Microwave Theory Tech., Vol.
MIT-31 , p. 1051-1059, Dec. 1983.
44. S.K. Koul and B. Bhat, Microwave and Millimetre Wave Phase Shifters, Volume
I on D~lectric and Ferrite Phase Shifters, Artech House, MA, 1991.
. 45. A.I. Baden Fullce, Ferrites at Microwave FreqUD/C~s, lEE Electromagnetic Wave
Series 23, Peter Peregrinus Ltd, 1987.
46. R. Garg and I.I. BahI, Microstrip discontinuities, Int. 1. Electronics, Vol. 45,
pp. 81-87, 1978.
47. R.K. Hoffman, Handbook of Microwave Integrated Circuits, Artech House, MA,
1987.
48. M. Caulton, Lumped elements in microwave integrated circuits, in Advances in
Microwaves, Vol. 8, by L. Young and H. Sobol (Eds.), Academic Press, NY,
1974.
49. D.A. Daly et al, Lumped elements in microwave integrated circuits, IEEE Trans.
Microwave Theory Tech., Vol. MTT-15, pp. 713-721,1967.
SO. F.E. Terman, Radio Engineering Handbook, McGraw-Hill, NY, 1943.
51. E. Pettenpaul et al, CAD models of lumped elements on GaAs upto 18 GHz,
IEEE Trans. Microwave Theory Tech., Vol. MTT-36, pp. 294-304, 1988.
52. R.S. Pengelly and D. Richard, Design, measurement and application of lumped
elements up to I-band, Proc. 7th Eur. Microw. Conf., pp. 461-464, 1977.
53. I. Wolff and N. Knoppik, The micros trip ring resonator and dispersion
measurements on micros trip lines, Electron. Lett, Vol. 7, pp. 779-781, 1971.
54. K. Kawano and H. Tomimuro, Slot-ring resonator and dispersion measurement
on slot line, Electron Lett, Vol. 17, No. 24, pp. 916-917, 1981.
55. I.K. Plourde and C.L. Ren, Applications of dielectric resonators in microwave
components, IEEE Trans. Microwave Theory Tech., Vol. 29, No.8, pp. 754-
770,1981.
56. D. Kajfez and P. Guillon (Eds.), D~lectric Resonators, Artech House, MA, 1986.
57. H. Sobol, Application of integrated circuit technology to microwave frequencies,
Proc. IEEE, Vol. 59, pp. 1200-1211, 1971.
58. R.W.Bery,P.M.HallandM.T.H8rris,Thin-FiImTechnology,VanNostrand,1968.
59. R.A. PuccI (Ed.), MontJlithic Microwave Integrated Circuits, (A reprint volume),
IEEE Press, NI, 1985.
3
Metallization of Plastics by
Electroless Plating
B. Viswanathan
Department of Chemistry
Indian Institute of Technology, Madras, India
Of the available liquid state metallization processes for plastics, electroless

process has gained remarkable importance in the last 20 years, even though,
the administrator of the invention from NBS Dr. Blum did not anticipate any
tremendous commercial use for this process [1]. However, the following state-
ments from P&SF report [2] that "computer memory devices represent a growing
market for electroless nickel plating .... The US government has saved the
public several million dollars for the past few years by using electroless nickel.. ..
Electroless copper baths without formaldehyde may be the wave of the future
in circuit board plating" show that this anticipation was completely proved to
be wrong. In fact plating of plastics has become an important segment of the metal
finishing industry in US [3] and in many other developed and developing countries.
Shipley [4] predicts that commercial usage of electroless plating will grow
quite rapidly during the next 10 years (since 1984). Applications will increase
in number as design engineers and advanced engineering groups learn more
about the capabilities and advantages of electrolessly plated metals and alloys.
Starting froin nickel electroless plating by hypophosphite by Brenner and
Riddell [5] other metals and reducing agents have been used for the electro less
deposition on plastics. The available data given in Table I, show that only a
limited number of metals can be deposited by electroless method, however,
this method has gained commercial applications in many sectors like aerospace,
automotive, computers, electronics machinery and so on due to the unique
properties of this deposition technique. The main reasons for using electroless
method in preference to other methods especially electrolytic plating are:
• the feasibility of obtaining uniform deposits over irregular surfaces
• the possibility of direct deposition on nonconductors
• the flexibility of depositing on isolated areas
• the desirable characteristics of the deposits namely, less porous and
more corrosion resistant and unique deposit properties
• the possibility of bulk plating and semibulk racking.
Table 1. Typical metal coatings by electroless deposltlon-summary of available data
Deposit Year Mediwn Reducing agent First inventor Ref.
Ni-P 1946 Sodiwn Brenner & [13-17]

hypophosphite Riddel
CO-P 1947 alkali [17]
Ni Hydrazine [18]
hydrosulphides
Ni-B 1958 Dimethylamineborane [18, 19]
borohydrides
Co-Ni-P 1973 Sodiwn hypophosphite [17,20,2IJ
Ni-B-Th 1981 Matsuoka [21]
& Hayashi
Co-B acid/alkali Dimethylamineborane [21]
Cu 1947 alkali Formaldehyde Narcus [22,23,
alkali dimethylamineborane 24-32, 33,
alkali 34-36]
Pd-P 1968 alkali Sodiwn [22,37]
hypophosphite
Pd 1961 alkali Hydrazine Rhoda [6,38,39]
ter. amineboranes
Au 1970 Amineborane Okino [40, 41,42]
borohydride
Au-Cu 1982 Formaldehyde Molenaar [43]
Ag 1974 Dimethylamine Pearlstein [17,44]
borane
Sn 1980 Titaniwn Warwick & [45]
trichloride Shirley
Alloy 1970 Malloy [46]
Polyalloys 1965 Pearlstein & [47]
Weightman
Composites 1974 Parker [48]
The electtoless deposit is usually achieved by using a bath containing at

least five functional components, namely,
1. A metal salt as the source for the deposited metal.
2. A reducing agent.
3. A complexing agent.
4. A pH adjusting agent.
5. A stabilizing agent and other functional additives.
For each one of these components, one has a variety of substances to choose
from. The purpose of this chapter is to examine the science behind this process
especially that of electroless copper and to evaluate the relative merits of the
choices, especially in the context of metallization of plastics.
1. HISTORICAL BACKGROUND OF THIS DEVELOPMENT

As there are more than four important documents already available on this
Metallization ofPlastics by Electroless Plating 81
aspect [I, 2, 4, 6] we shall briefly consider the. available literature on the

developments of the electroless deposition technique. Riddell and Brenner invented
this process during their investigations on the electrodeposition of nickel-tungsten
and cobalt tungsten alloys on the interior of a liner using an ammoniacal
citrate-nickel-tungsten plating bath with an insoluble anode. In order to prevent
the oxidation of organic components which affected the quality of the deposit
they added various reducing agents to the bath. When they used hypophosphite
as the reducing agent, they observed that the current efficiency was higher and
at the same time the outside of the liner was also plated. This marked the begin-
ning of the autocatalytic chemical plating which is popularly known as 'electroless'
plating today. Though Brenner called it electrodeless process, Blum coined the
term 'electroless' to indicate that the process is similar to electro-plating in that
a metal coating is produced but the dissimilarity arises in that in the former
process no external electric current is involved. It should be remarked that
though the effect that sodium hypophosphite would reduce nickel and cobalt
salts to the metallic powder was known even 30 years earlier to this invention,
the achievement is that by a controlled autocatalytic process one could produce
a smooth, coherent metal coating. This was followed by a series of investigations
by Gutzeit and his group [7-10] on electroless deposition of Ni-P system,
wherein the role of many other additives like chelates, exaltants, buffers, stabilizers
were examined and operating conditions for obtaining a desirable rate of
deposition were formulated [7-12]. This series of studies has lead to the
development of other baths for electroless deposition of other metals and the
important inventions in the area of electroless deposition are summarized in
Table l. Electroless nickel using hypophosphite as reducing agent and as it is
generally known consists of Ni-P alloy with phosphorus content varying between
3 and 12% depending on both composition and operating parameters. The
phosphorus content is usually increased by lowering bath pH and by increasing
hypophosphite and orthophosphite concentrations. In many aspects the electroless
deposits exhibit different structure, physical and mechanical properties fatigue
strength and corrosion resistance as compared to that of electrolytic deposits.
A typical compilation of the comparison of the properties of electroless Ni-P
alloy and of electrodeposited nickel is given in Table 2 [4].
Electroless copper, has seen remarkable developments starting from Narcus
[23] who reported the practical reduction of copper on nonconductors in 1947
and Cahill [49] who described autocatalytic chemical reduction of copper
from alkaline copper baths using formaldehyde as the reducing agent. The
success story of copper baths is mainly due to the introduction of effective
stabilizing agents like mercaptobenzothiazole [50], thiourea [51] cyanide [52]
and sulphur compounds [53,54]. Because of the beuer corrosion resistance and
other favorable properties of electroless copper it is preferred over electroless
nickel on plastics.
The interest in electroless cobalt is mainly due to their magnetic properties
which can be exploited for recording storage and switching device applications.
Commercial usage of electroless cobalt is limited to magnetic applications
especially to their magnetic deposits on tapes, discs and wire substrates [55-62].
Table 2. Comparison of properties of deposits from electroless (Nt with

approximately 8% P) and electrolytic nickel [4]
Properties Electrolytic 99 + % Ni Electroless average 92% Ni/8% P

Cf)J11position 99 + Ni Average 92% Ni /8% P
Appearance Dull to bright Semibright
Structure Crystalline Amorphous
Density 8.9 Average 7.9
Thickness uniformity Varies ± 10%
Melting point l455°C 8900C
Hardness as plated 150-400 500-600
(kg/mm2)
Heat hardening No effect 1000
(kg/mm2)
Wear resistance Fair Excellent
Corrosion resistance Good (porous) Very good (few pores)
Relative magnetic 36% 4%
susceptibility
Electrical resistivity 7 60-100
(microohmlcm)
Thermal conductivity 0.16 0.01-0.02
(cal/cm/sec/C)
Coefficient of thermal 7.5 7.8
expansion
(in/in/F x 10)
Modules of elasticity 30 10
(Ksi)
Elongation 6-30% 2%
Internal Stress (ksi) ±1O ±1O
Friction coefficient vs
steel unlubricated galling 0.38
lubricated 0.2 U.2
Attempts to produce electroless deposits of polyalloys [63] as well as composites

[64] were successful in 1970 in addition to that of Au, Ag and tin.
Electroless gold plating continues to be an area of considerable activity. The
studies pursued now involves formulation of baths with long term stability,
increased plating rate and improved throwing power. Carbonates of sodium
and potassium in combination with amines like mono-, di-, and trialkanol,
amines and ethylene di-and tri-amines show higher throwing power in gold
plating baths [65]. Combined nickel and gold plating can also be achieved
using dimethylamine borane and hydrazine hydrate as reducing agents [66]. A
new addition called reduction promoters namely hydrochinone or pyrogallol
convert the oxidized thiourea species formed during gold reduction back to
thiourea and thus inhibit the precipitation of gold and accelerate gold plating
rate [67]. A German patent discloses a weakly acidic or neutral bath containing
in addition to KAu (CN)2, I-hydroxyethane, diphosphoric acid, diethanol-
ammonium salt of maleic acid hydrazide and triethanol amminium salt of 1-
hydroxylethane-l, I-diphosphonic acid [68, 69].
Metallization of Plastics by Electroless Plating 83
2. GENERAL SCHEME OF ELECTROLESS PLATING

Normally electroless plating involves a number of steps carried out in sequence
as shown in Fig. 1. Normally the sequence of operations involves first sensi-
tization followed by activation and then electroless deposition using chemical
reduction reaction with suitable controls either in the form of additives for
acceleration or retardation of the deposition rate. However one should remember
that there will be a number of rinsing and washing steps which are not explicitly
shown in the follow chart shown in Fig.1.
PI asticsl ceramics
moulded ,formed or
machined
Clean and
condition
Fig. 1. Sequence of general operations in electroless deposition.
3. SENSITIZATION AND ACTIVATION

Normal commercial procedures of sensitization for hydrophobic surfaces involves
two step immersion sequence in acidic stannous chloride solution followed by
immersion in an acidic palladium chloride solution. The first bath is called the
sensitizer and the second as activator. The chemistry involved in this process
of immersion is dispersion of finely divided palladium (0) which initiates the
autocatalytic plating process. The nucleation of palladium (0) is achieved by
the redox reaction taking place between Pd(ll) ions in the solution as well as
tin ions adsorbed on the substrate. This means one should achieve good adsorption
of sensitizers. Schlesinger and Kisel [70] have examined the adsorption properties
of acidified SnCl2 solution with additives like hydroquinene, triton x-loo, thiourea
and showed a non-linear relationship between the amount of sensitizer adsorbed
and the number of metal islands deposited from electroless nickel bath (citrate
bath) per unit area of the substrate. The effectiveness of commercially available
sensitizing solutions for electroless copper deposition on hydrophobic substrates
has been gauged by the measurement of the contact angles. Feldstein and
Weiner [71] have demonstrated that controlled additions of stannic ions with
or without sodium chloride especially with aging could yield superior wetting
and thus can yield uniform plating of hydrophobic substrates [72-74]. The
aging process in the stannic halide solutions probably favors the transformation
of an a-stannic acid to a p-stannic acid form. The colloidal nature of fj-stannic
acid is consistent with the view that the sensitizers together with the activators
form only colloids and not solution complexes [75]. Cohen and West [76-78]
postulate that a complex between Sn(II) and Pd(II) is first formed in which the
reduction of Pd(ll) to Pd(O) occurs subsequently forming a nucleus for the
growth of the colloid. The core of the colloid is a metallic alloy of Pd and
tin surrounded by stabilizing layers of SnCl2 and Sn (OHh. According to this
model, one should expect the deposition of metallic tin on Pd and the cyclic
voltammetric study [79] by Jean Horkans aimed at identifying the chemical
state of the catalyst on the activated surface showed that in addition to the
stabilizing stannous layer produced by the reduction of Sn(lV) metallic tin
also constitutes to the core of the active colloid. The catalytic activity of Pd(O)
that are formed seems to be prone to UV irradiation and irradiated Pd(O)
species appear to be incapable of becoming catalytically active in the metal
plating bath [80, 81].
The structural and composition of the colloidal catalyst particles have been
investigated by a variety of techniques like electron diffraction, Mossbauer
spectroscopy, Rutherford backscattering and photoelectron spectroscopy [82].
Though there is some convergence on the fcc structure of the active catalyst,
there still seems to be some uncertainty regarding the composition of
Pd-Sn alloy. It is certain that Pd/Sn ratio is greater than I, [83, 84]. The
various values of the ratio reported in literature could have arisen because the
analysis of the data have been performed in the different regions of the active
catalyst from the core to the outer layer depending upon the type of technique
employed.
In the sensitizing and activating step, certain accelerators like NaOH, EDTA
are added which are effective in removing the tin species from the surface and
increasing the initial rate of metal deposition. The uniform distribution of the
metal deposit is also dependent on the nature of the accelerator used, for example
NaOH and EDTA are favorable for uniform deposition of copper [85].
In the sensitization step as stated earlier, various types of tin compounds
including tin oxide (Sn02) and ~tannous and stannic chlorohydroxides,
oxyhydroxides and hydrated oxides have been identified. The formation of
these compounds and hence the extent of sensitization are dependent on the pH
of the solution [86]. The process sequence involves a rinsing step after the
activation and sensitization. There seems to be some difference of opinion on
the rinse water pH for the activation of plastics, Cohen and Meek [87] propose
that acidic solution rinse provides a cleaner, more catalytic surface while Ghorashi
=
[86] claim that a rinse water with pH 9 provides optimum conditions for
obtaining a uniformly activated surface. It is to be remarked that the final rinse
with appropriate pH provides the necessary type of species, hydroxide or chloride
species on the outer layer of the activator-sensitizer combination which favors
the catalytic reduction and deposition of the metal from the solution.
A new concept of the activation can be introduced based on the anchoring
the functional groups or ions on the polymer surfaces. For example Pd2+ can
be exchanged with polymer substrates with functional groups especially caIboxylic
groups which will avoid the formation of tin chloride shell over the activator
and thus promote in a facile manner electroless deposition [87a].
4. ELECTROCHEMISTRY OF ELECTROLESS COPPER

Electrochemical nature of the electroless deposition of metals has also received
considerable attention [88--91]. Paunovic [88] started it with his mixed potential
concept. According to this model, when at an electrode simultaneously two
different electrode reactions
OX1 + n1e -+ Red1 (1)
OX2 + n2e -+ Red2 (2)
take place, both reactions strive to attain the equilibrium state with its own
equilibrium potential Eeq but in this process attain a steady state mixed potential
(E MP). The characteristics of this mixed potential are that
(i) Both redox systems are removed from their equilibrium potentials
governed by the equations
111 =EMP - E1eq (3)
112 =EMP - E2eq (4)
(ii) Net electrochemical reaction occurs as they are removed from their
eqUilibrium by the establishment of the mixed potential.
(iii) The condition that the sum of cathodic current densities is equal to the
sum of the anodic current densities holds good, since net current cannot flow
in the isolated system.
-i +i
1 2
-
=i1 +i2 (5)
(iv) Since at the steady state mixed potential, the system is not in equilibrium
the change in free energy is not equal to zero.
Extending this mixed potential concept to electroless plating wherein one of
the redox couple is the metal electrode and the second one is the reducing agent
applied in electroless plating, the corresponding reactions can be written as
Mn+ + nMe -+ M (6)

<>t + nrectC -+ Red (7)
The establishment of mixed potential is shown pictorially in Fig. 2.
--- --
(+)
'"'(l-ttt})
'l9'- ....... _ -
EMF'
Eeq
Ale
c-)
- ....... ~
.... ,~!!.f)J
I.Q.....
........ .......
+
log i log i dep
Fig. 2. Current-Potential curves for the system with two different simultaneous
electrochemical reactions, Kinetic scheme.
Wherein it is seen that the potential of the redox couple is raised from its
reversible value E~ and the potential of the metal electrode is depressed
cathodically from its reversible value EAlso as to reach the mixed potential
EMP value. With the overall reaction at the electrode is represented as
Mn+ + Red -+ M + Ox (8)
In essence, the mixed potential can be considered to be the intersection
points of iM and i R• This implieSt at the steady state mixed potential, the rate
of reduction of the metal is equal to the rate of oxidation of the reducing agent
or
ieleclroless deposition = iM =ired (9)

This implies that one can a priori calculate the values of E Mp and deposition
=
rate idep if one knows the current potential functions i f (71) of the individual
electrode processes. The parameters which determine the current potential curves
for the partial electrode reactions (6) and (7) also determine E Mp and ideposition.
Parameters like the variation of the exchange current density (Fig. 3), concen-
= =
tration of reactants (since io 71FkCox 71FkCred, where ks are rate constants
and Cs are concentrations), temperature and the type of overvoltage (charge
transfer. mass transport or reaction overvoltage) that control the values of EMP
Metallization o/Plastics by Electroless Plating 87
+
(+)
Eeq
M
eq
ERed
(-)
+~--------------------~~~~~
logi 1dep logi2deplogi3dep
logi
Fig. 3. Variation of the rate of electroless deposition and the mixed
potential with the exchange current density.
as well as ideposilion. Another aspect that has been investigated is to account why
certain metals with negative free energy of adsorption of hydrogen can be
deposited using a specific reducing agent like formaldehyde [92, 93].
The solution chemistry of electroless copper bath has been probed by a
variety of electrochemical techniques. Using cyclic voltammetry technique below
the mixed potential value (- 0.7 V) Tam [94] has identified that methylene
glycolate (CH20 2H) and its copper (11) complex are the electroactive species.
in solution. The methleneglycolate anion is adsorbed on the copper sites with
the C-H bonds adjacent to the copper surface which then undergoes electro-
dehydrogenation to give rise to formate anion as well as molecular hydrogen
evolution [95]. As seen from the data in Table 1 only a limited variety of
reductants are used in electroless process namely hypophosphite, formaldehyde,
borohydride, dialkylamine borane and hydrazine. Normally electroless plating
is accompanied by hydrogen evolution though it is not directly related to metal
deposition. Van der Meeraker [96] proposed a universal mechanism for electroless
deposition which involves the dehydrogenation of the reductant as the first
step.
RH ~ R + H(10)
R + OW ~ ROH + e- (11)
H+H ~H2 (12)
H+OH ~ H20 + e- (13)
Mn+ + ne~ M (14)
where RH represents the reductant It is seen from this mechanism that hydrogen
atom combination step (12) indicates that metals with dehydrogenation-
hydrogenation catalytic activity can promote this reaction step favorably.
Extending this proposal Ohno et al [97] propose that the oxidation of the
reductants proceeds mainly along with hydrogen evolution step on copper,
silver and gold while hydrogen ionization mechanism predominates on Co, Ni,
Pd and Pt. The catalytic activity series of the metals for anodic oxidation of
reductants does not coincide with that for hydrogen electrode reaction.
Electroless copper obtained using hypophosphite as reductant is supposed to
proceed through the following steps
H 2 P02 ~ HP0 2 + H
Cu 2+ + 2H 2 P02 + 20H- catalyst surface ) CUO + 2H 2 PO :i + H2 (15)
when complexing agents like EDT A are used octahedral complexes are formed
while with ligands like oxalate, glycine square planar or square pyramidal type
complexes are proposed. Multinuclear (usually dimeric) complexes are reported
with citrate, tartrate, malate which did not show any ESR signal indicating
strong anti ferromagnetic coupling, and in the presence of these ligands significant
plating rates were observed [98, 99]. Eventhough complexation is one of the
essential steps in the electro less plating no definite relationship could be estab-
lished between the plating rate and the stability constant of the possible complexes
that are formed. However ligands which can form dimeric copper species exhibit
higher plating rates (Refer to the data assembled in Table 3 [100]).
The mixed potential theory proposed for electroless deposition does not
seem to predict correctly the experimentally obtained mixed potential and the
current potential (i IV) curve is not a simple sum of the half cell i IV responses
in the case of copper electroless deposition by formaldehyde in presence of
EDTA. When the two half cell reactions are
Cu (EDT A)2- + 2e- ~ Cu (metal) + EDTA4- (16)
2H2CO + 40W ~ 2HCOO- + H2 (gas) + 2H20 + 2e- (17)

strongly coupled then the assumption that the reduction of Cu (EDTA)2- is not
perturbed by the presence of HCHO or vice versa does not hold good. It is there-
fore necessary to invoke in the reaction mechanism of formaldehyde induced
reduction of copper, the formation of a Cu (EDT A)IHCHO complex on the elec-
trode followed by the reduction of the complex. This is one of the results reported
in literature which does not conform to the mixed potential model [101]. However,
recent studies by Hung and Ohno [102,103] have shown that reactions involved
in electroless copper deposition reduced by hypophosphite though complicated
still follows the mixed potential theory. The stability of HCHO bath for the
electroless copper deposit is a problem that has received attention. It is generally
believed that thick copper deposit can be obtained if the additives that have
propensity for adsorption on copper surface are excluded. Among the additives
NaCN was found to leave the chemical purity of the deposit in tact [104].
Table 3. Plating rates and stability constants with copper (II) and nickel (ll) Ions
= =
for various complexlng agents (M metal Ion and L ligand; K value
will change with pH and Ionic strength of the medium)
Complexing Stability Stability Plating rate Plating rate

agent constant constant =
pH 9.2, pH = 5.3
with Cul + with Nil + 65°C, mg/cml. h
log K log K mg/cml. h
Sodium citrate ML (5.9)

MlLz (13.2) ML (5.4) 7.6 0
Sodium potassium ML (3.39)
tartrate MlLz (1t24) ML (2.06) 6.6 O.S
Oxalic acid MLz (9.21)
ML (4.S4) ML (5.16) 0
Glycine ML (S.15) ML (5.7S) 0.19
MLz (15.03) MLz(1O·5S) 0.3
ML3 (14.0)
Malic acid ML (3.42) ML (3.17) 4.2 2.9
MlLz (S.O)
EDTA (sodium ML (IS.7) ML (1S.52) 0.7 0.8
salt) MLz (11.7)
Phthalic acid ML (3.15) ML (2.17) 0 0.9
Succinic acid ML (2.6) ML (1.6) 0.2 3.7
Lactic acid ML (2.45) ML (1.64) 0.2 1
MLz (4.08) MLz (2.76)
ML3 (4.3) ML3 (3.1)
Ethylenediamine ML (10.6) ML (7.35) 0.8 0.6
MLz (19.6) MLz (13.54)
ML3 (17.71)
Cyanide ML4 (25)
Diphosphate MLz (11)
Ammonia ML4 (14)
Pyridine ML4 (7)
Dipyridyl (22') ML3 (7)
2,9-Dimethylphenanth-
roline-(I, 10) MLz (11.7)
Data reproduced from [27] and [35].
5. CHEMISTRY OF ELECTROLESS COPPER BATHS

Compositions of a typical electroless copper bath is given. in Table 4. Though
free cupric ions can exist in acid medium, its presence in + 2 state requires a
complexing agent in alkaline medium used in the electroless plating baths.
Typical reducing agents like formaldehyde releases the necessary electrons for
the reduction of the metal ions in alkaline medium while in acid solutions it
is directly converted to carbondioxide according to the reaction
HCHO + HzO ~ COz + 2Hz (18)

Table 4. Typical bath compositions normally used In electroless plating
Deposit Bath specifications (concentration) End uses
Copper Copper sulphate (0.04 M) PCBs, MICs EMI

Ethylene diaminetetraacetic acid (0.10 M)
Rochelle salt (0.14 M)
Formaldehyde (0.20 M)
pH = 12.5
Temperature 65°C
Nickel Nickel sulphate or chloride (0.10 M) Aerospace, automotive,
Na3C6Hso, (0.10 M) computers, electronics
NaH 3POZ (0.20 M) and machinery
Triethanolamine (0.15 M)
pH = 5.7
Temperature 75°C
Gold KAu (CN}z (0.005 M) PCB, Connector taps,
Dimethylamine borane (DMB) (0.05 M) chips and selective
KOH (.80 M) plating
KCN (0.035 M)
lead acetate 15 ppm
K ZC0 3 (0.5-1.0 M)
pH = 5-7.5
Temperature 80°C
Palladium Palladium chloride (0.06 M) PCBs and connectors
Rochelle salt (0.09 M)
Ethylenediamine (0.04 M)
Sodium Hypophosphite (0.04 M)
pH = 8.5
Temperature 70°C
Silver Sodium silver cyanide (0.01 M) Contact switches, wave
Sodium cyanide (0.02 M) guides, reflectors
Sodium hydroxide (0.02 M)
DMAB (0.05 M)
However, the effective concentration of HCHO could be reduced in alkaline

medium due to a number of side reactions like the disproportionation reaction
(Cannizaro reaction)
2HCHO + OW ~ CH30H + HCOO (19)
which is responsible for the aging of the bath and its instability.
Addition of methanol may shift the reaction (19) to the left and thus make
available the necessary amount of formaldehyde for the reduction reaction. The
precipitating seeds in the bath could promote the reduction of cupric ions to
cuprous ions according to the reaction
2Cu2+ + HCHO + 50W ~ CuzO + HCOO- + 3H zO (20)
Since the bath does not contain complexing agents for cuprous ions and because
of low solubility of cuprous oxide the following reactions may also take place
in the bath
CU20 + H20 -+ Cuo + Cu2+ + 20H- (21)
2Cu+ -+ Cu2+ + Cuo (22)
These reactions result in fine particles of copper dispersed randomly in the
solution. Since cuprous ions are not easily reduced by formaldehyde the bath
instability also occurs. There are various ways one can avoid the formation of
cuprous oxide. They are: (1) agitation or aeration of the bath, (2) use of low
concentration of the constituents, (3) employing polyelectrolytes like cellulose,
ether, polyvinyl pyrolidine, polyvinyl alcohol, gelatin peptone polyamides etc
to cover the cuprous oxide to prevent the catalytic effect, (4) Decreasing the
free concentration of cupric ions by using suitable complexing agents and
(5) removing the nucleating impurities in the materials used especially in the
Rochelle salt.
6. NUCLEATION AND GROWTH OF ELECTROLESS DEPOSITS

The general description of conventional deposition processes cannot be applied
to electroless deposition obtained by the chemical reduction reaction taking
place at the interface. The chemical reaction is normally initiated at the catalytic
sites on the surface by a direct three dimensional nucleation step. The copper
nuclei thus formed are about 25 A in diameter, and form aggregates (nearly a
few hundreds angstroms thick) and the ultimate size of the aggregate is decided
by the energy considerations and at the limit of energetically stable condition
crystallization process sets in wherein unstable aggregates are transformed into
a relatively large grains. Repeated nucleation and recrystallization give rise to
twin faults between two adjacent nuclei with twin orientation having a common
(lll) boundary [105]. In the case of mixed electroless deposits like Ni-P,
Coop, Ni-Co-P though nucleation :akes place at the catalytically active sites,
the aggregates are in a liquid like solid solution of the species which undergoes
crystallization in the subsequent heat treatment steps [106].
The nucleation, growth and crystallization processes in electroless deposition
are dependent on the chemical reduction reaction and depending upon the
conditions employed, the structure of the crystallized phase of the metal deposit
can be different from what one obtains from simple metal deposits formed by
other methods.
7. KINETIC STUDY OF ELECTROLESS COPPER PLATING

The rate law [107] applicable to electroless copper plating from HCHO-Rochelle
salt baths has been empirically formulated as
Rate = 18.5 [CU2+]0.47 [011]°·18 [HCHO]O.07 exp [18.7 (T-313)fl1
which has been shown to be applicable to a limited extent. This type of rate
laws [l08] indicate a complex mechanism involving interaction among hydroxide
ions, HCHO, and cupric ion in the reduction process. Similar rate law has also
been found applicable when EDTA is used as complexing agent and HeHO as
the reducing agent [109]. Donahue et aI [110] have deduced the relationship
l/rx, =1/ro + KCx./ro

where 'Xs and '0 are the plating rates in excess of ligand (N, N, N', N'-tetrakis
(2-hydroxy-propyl) ethylenediamine (quadrol) and in the absence of excess
ligand respectively, K is the adsorption coefficient and Cxs is the concentration
of the ligand when it is present in excess. The applicability of the experimental
data to this equation was considered to imply that adsorption-inhibition effect
is operating in·the electroless plating of copper by formaldehyde [110]. From
a coulostatic method, Sato et al [111, 112] showed that the plating current
density of copper on glass substrates can be expressed as
iELP =K/Rp
where K is a constant and Rp is the polarization resistance. The deposition
rate decreases with aging of the bath while the proportionality constant K
(= 27.3 mV) is independent of the substrate used as well as aging of the bath.
It is therefore obvious that the kinetics of electroless plating though amenable
to various kinds of measurements is complicated and only empirical kinetic
laws can be generated on them which depend on a number of process parameters
like substrate used, additives added, aging of the bath and the nature of reducing
agent used. There is possibly a correlation between the rate of dissociation
of the complex (when the ligands like tartrate, EDTA, Quadrol, 2-diamine
N, N, N', N'-tetraacetic acid, triethanolamine [113] are used) and deposition
rate [114] and the hydro~yl concentration depenndence on the deposition
rate depends on the nature of the ligand used [115]. But for these few
generalizations the kinetics of electroless deposition still appears to be grossly
empirical.
8. THERMAL AND MECHANICAL PROPERTIES OF

ELECTROLESS COPPER
The electroless deposit for manufacturing printed circuit boards differs from
the conventional subtractive process in that (i) a permanent, non-strippable
resist is an integral part of the board, (ii) copper is not etched from the board,
(iii) conductor resolution is better (iv) uniform plating over the entire panel is
possible. However, the additive method has its own limitations. One of them
is that hair line cracks occurring at hole comers during thermal testing especially
for specific military specifications. The bath specifications were usually monitored
in bulk preparation by automatic controllers, for example concentration of
formaldehyde (bisulphate titration), copper (colorimetric analysis) and cyanide
(ion selective electrode) and pH, solution density and mixing potential of plating
solution. Typical mechanical data on two different electroless copper deposits
are assembled in Tables 5 and 6 [116, 117]. Deposits from electroless baths
containing EDTA and sufficient amount of glycine (upto 1 : 1 mole ratio so
that condensation product is formed and thus controls the release of formaldehyde
Table 5. Properties of electroiess copper deposits
Property Standard bathCa) High quality Crack free deposit<")

bathea)
Class 1 Class 2
Elongation % 2.5-3.0 6-12 4.3 7.1

Tensile strength (kg/mm2) 25 30-60 59 21.8
Grain size (dia.) parallel 0.3-0.7 0.3 0.7
to substrate (m)
Grain size (dia.) perpendi- 6-7 7 6
cular to substrate (m)
Cracks in through-hole Yes None No No
after thermal stress
ea)Data taken from [120].

(b>Data reproduced from [116].
or the free fonnaldehyde concentration) with HeHO have nearly equal elongation
values and greater tensile strength than electrolytic copper from acid sulphate
solutions containing no additives. The mechanical properties have a direct
correlation with the deposition rate. The internal stress in ~e electroless copper
is attributed to (a) entrapped hydrogen resulting from the plating process and
(b) fine particles of copper obtained from bath decomposition occluded in the
deposit [119].
Table 6. Typical properties of copper foils obtained by electroless

plating (data reported In [118])
Plating temp. °C 38 49
Plating rate m/hr 1.9 ·2.82 3.43 3.45 4.32 5.69
mil/hr 0.073 0.111 0.135 0.14 0.170 0.224
Tensile strength 52.8 37.5 35.8 45.4 49.2 31.0

(thousand PSI)
Tensile strength after 45.1 48.9
annealing (thousand PSI)
In electroless copper baths certain additives like cyanides and f, 10-phenan-

throHne are added as bath stabilizers. Though there are claims that addition of
cyanide improves the mechanical strength of the deposit Junginger et al [120,
121] have shown that bivalent copper complex is adsorbed on the extraneous
germs which are reduced to metallic copper by fonnaldehyde and the bath is
affected in the presence of oxygen and complexing agents. The texture of
deposit is dependent on the presence of surfactant (100 textured) while in the
absence of surfactant the growth is (111) textured. There are two forms of
hydrogen [122] (diffusible and residual) that exists in electroless copper deposits.
The diffusible molecular hydrogen in the microvoids impairs the ductility of
the metal deposit through its effect on lattice strain induced by high pressure and
through the effect of voids on the structural integrity of the deposit.
9. APPLICATIONS OF ELECTROLESS COATINGS
9.1 Printed Circuit Boards [123J

Printed circuit boards is one of the important elements of electronics industry.
Size, volume and circuit density are the factors wherein printed circuit boards
(PCB) technology scores over other processes. Printed circuitry is an accepted
technique by which interconnection of active components is made on solids or
flexible polymer substrates. These contacts are made by either substractive
(etch down) or additive process. Additive process is gaining importance due to
cost saving and other pollution problems associated with etch down operations.
In the additive printed circuitry either semi-additive or true additive procedures
are employed wherein the sequence of steps are slightly altered, in the semi-
additive process electroless plating precedes the masking while it is in the
reverse order in the additive process. In the additive printed circuitry technique
additives like monionic surfactants or cyanides are incorporated so as to obtain
copper'deposits of unusual physical properties like good surface conductivity,
good adhesion, and durability. One can also use the photochemical activity of
tin (II) compounds to create catalytically active sites including Pd (0) sites.
These developments in addition to the simplicity of the process are expected to
make this process more attractive and acceptable for printed circuitry manufacture.
9.2 Semiconductor Devices

The contact to silicon transistors are made using soft solders using intermediate
metallization. The intermediate metallization should have the following
characteristics:
(1) Good contact to both n- and p-type silicon.
(2) Good adhesion to the surface.
(3) The metal must be solderable.
(4) The metal-solder interface should be inert and should not form brittle
intermetallics.
In the silicon transistor chips the contacts are usually made using solder.
Direct soldering is not possible and intermediate metallization is necessary.
This metallization layer should have certain desirable characteristics as stated
above. Electroless nickel seems to meet the requirements. This technique has
also been extended to III-V compounds as well.
In addition production of a variety of miscellaneous electronic components
like capacitors, resistors, connectors, diodes, heat sinks, relays lead frames
employ electroless deposition to a great extent.
9.3 Microwave Integrated Circuits (MIC)

The electroless technique has now been extensively used to produce a variety
of passive and active microwave components on hard substrates like alumina,
titanate ceramics, ferrites as well as soft substrates like PTFE (poly tetra-
fluoroethylene). The replacement of Cr-Au metallization by electroless copper
can result in 15-25% decrease in microwave attenuation. Recently microwave
strip lines have also been designed by electroless copper plating [124]. The
usefulness of this technique as compared to copper clad PTFE is demonstrated
by the data given in Tables 7 and 8.
Table 7. Microwave Properties of commercially available copperclad PTFE

materials manufactured by different ftrms [124]
Material Designed Observed Dielectric Loss

resonance resonance constant tangent
frequency frequency (tan ~)
(GHz) (GHz)
Rogers corporation 8 7.67 2.41 0.0031

RT Duroid 5880
Taconic 8 7.96 2.24 0.0036
RT/Duroid 5880
Hindusthan Fluorocarbons 10 10.8 1.89 0.0022
3M company 10 9.7 2.35 0.0035
Taconic 10 9.75 2.325 0.0035
RT/DW'Oid 5880
Taconic 16 15.7 2.27 0.0036
Rogers corporation 18 18.94 2.01 0.Q1
RT/DW'Oid 5880
TableS. Microwave properties of electroless copper coated PTFE substrates [124]
Material Designed Observed Dielectric Loss

resonance resonance constant tangent
frequency frequency (tan 6)
(GHz) (GHz)
RT/DW'Oid 8 7.555 2,48 0.0081

RT/DW'Oid 10 9.878 2.26 0.0030
RT/Duroid 16 15.63 2.29 0.Q115
9.4 Electromagnetic Interrerence Shielding

Communication and control circuitry are highly vulnerable to interference by
electromagnetic or radiofrequency waves. Electromagnetic interference can
paralyse effectively HF communications, Ballisatic missile guidance and light
aircraft navigation. Electroless copper, or Ni-P or Ni-B plated ABS/polypropylene
oxide/polyester have been tested for shielding and copper coated with nickel
and phosphorus seems to provide maximum shielding.
9.S Thin Magnetic Film Devices

Cobalt-phosphorus films deposited on plastic substrates have been used in
magnetic recording tapes. In addition high density thin film magnetic discs for
storage applications in computers are produced using electro less cobalt deposit.
The Ni-P or cobalt deposits should have a high polish and uniformity, otherwise
the pits and non-uniformity will lead to loss of information and make the disc
scratchy and noisy.
9.6 Automobile Exterior Parts [125]

The production and use of plastics has grown dramatically during the last three
decades. Most of the electroplated exterior automotive plastic trim is fabricated
from acrylonitrile-butadiene-styrene (ADS) because of its low cost and good
physical properties. Some plated exterior automotive trim is molded from modified
polyphenylene oxide (MPPO) because of higher heat distortion temperature
and lower coefficient of thermal expansion than ADS type of plastics. The
direction of studies in this application is the development of improved stabilizers
and rate controllers like cyanides and other organic derivatives that will enable
rack processing.
10. CONCLUSION
Electroless plating has thus been receiving considerable attention in recent
times with respect to new bath formulations with improved features like long
term stability, increased plating rate, and increased throwing power from which
purer deposits or deposits with consistent composition and quality can be obtained.
Non-conventional methods like electroless gold plating on superconducting
YDa2Cu3O,.x ceramic oxides from non-aqueous baths are also attempted for
specific applications [126]. Plastics with thin uniformly metallized coatings is
used for various electronic and machinery applications and the need in this
sector is increasing more than what is expected. Structural studies of the deposits
still seem to interest scientists [127]. On the whole as expected, electroless
deposition appears to have a challenging future.
REFERENCES
1. A. Brenner. Plating and Surface Finishing, p. 24, July 1984.

2. P and SF report, Plating and Surface Finishing, p. 36 (1984).
3. D.A. Arcilesi, Plating and Surface Finishing, p. 90, June 1981.
4. C.R. Shipley, Jr., Plating and Surface Finishing, p. 92 (1984).
5. A. Brenner and G.E. RiddelJ, Proc. AES., 33, 23 (1946); 34, 156 (1947); US
Patent, 2, 532, 283; 2, 532, 284 (1950).
6. W. Goldie, 'Metallic Coatings on Plastics' Electrochemical Publications Ltd.,
Middlesex, England (1968).
7. G. Gutzeit and E.J. Ramirez, US Patent, 2, 658, 842 (1953).
8. G. Gutzeit and A. Krieg, US Patent, 2, 658, 841 (1953).
9. P. Talmey and G. Gutzeit, US Patent, 2, 762, 723 (1956).
10. G. Gutzeit, Plating, 46, 1158, 1275, 1317 (1959); 47, 63(1960).
11. E.B. Saubestre, Metal Finish., 60, 67 (June 1962); 6049 (July 1962); 60, 45
(August 1962); 60, 59 (September, 1962).
1~. F. Pearlstein, 'Electroless Plating' in F.A. Lowenheim Ed.;Modern Electroplating,
3rd Ed., John Wiley & Sons, NY 1974 Ch. 31.
13. K.S. Rajam and S.R. Rajagopalan, Technical Memorandwn, MT-TM-8601 (1986);
MT-TM-8801 (1988) NAL, gangalore, India.
14. K.S. Rajam, S.R. Rajagopalan, M.S. Hegdc and B. Viswanathan, Materials
Chemistry and Physics, 27, 141 {1991).
15. M. Papini and F. Papini, Thin Solid Films, 115, 1 (1984).
16. M. Schlesinger and J. Kisel, 1. Electrochem. Soc., 136, 1658 (1989).
17. S.L. Chow, N.E. Hedgecock, M. Schlesinger and 1. Rezek, J. Electrochem.
Soc., 119, 1614 (1972).
18. DJ. Levy, Electrochem. Tech., 1, 38 (1963).
19. J.W. Dini and P.R. Coronado, Plating, 54, 385 (1967). Deutsches Aulegeschrift
(DAS) 1, 137,918 (1958).
20. M. Gulla, US Patent, 3764, 352 (1973); 3, 832, 168 (1974).
21. M. Matsuoka and T. Hayashi, Plating and Surace Finishing, 68, 66 (July 1981).
22. A. Sergienko, US Patent, 3,418, 143 (1968).
23. H. Narcus, Met. Finish., 45, 64 (sep 1947).
24. M. Paunovic and R. Zeblisky, Plating and.Metal Finishing, 72, 52 (Feb. 1982).
25. K.F. Blurton, Plat. Met. Finishing, 52 (Jan, 1986).
26. S. Mizumoto, H. Nawafune, M. Kawasaki, A. Kinoshita and K. Araki, ibid., 48
(Dec, 1986).
27. R. Junginger, J. Electrochem. Soc., 135,2299 (1988).
28. C.I. Courduvelis and G. Sutcliffe, plating 67, 71 (Sep, 1980).
29. Y. Okinaka and H.K. Straschil, J. Electrochem. Soc., 133,2608 (1986).
30. S. Nakahara, Y. Okinaka and H.K. Sttaschil, J. Electrochem. Soc., 136, 1120 (1989).
"31. C.E. Baumgartner, Plating and Surface Finishing, 53 (Dec, 1989).
32. K. Kondo, J. Ishikawa, O. Takenaka, T. Matsubara and M. Irie, 1. Electrochem.
Soc., 137, 1859 (1990).
33. F. Pearlstein and R.F. Weightman, Plating, 474 (1972).
34. Aina Hung, Plating and Surface. Finishing, 62, (Jan, 1988); 74 (April, 1988).
35. Aina Hung, Ken-Ming Chen, J. Electrochem. Soc., 136, 72 (1989).
36. Aina Hung, and Izumi Ohno, J. Electrochem Soc., 137,918 (1990); Plating and
Surface Finishing, 54 (March, 1990).
37. F. Pearlstein and R.F. Weightman, Plating, 56, 158 (1969).
38. R.N. Rhoda, 1. Electrochem. Soc., 108, 707 (1961).
39. W.V. Hough, 1.L. Little and K.E. Warheit, US Patent, 4, 255, 194 (1981).
40. Y. Okinaka, Plating, 57, 914 (1970).
41. J.F. McCormack, US Patent, 3, 589, 916 (1971).
42. M.F. EI-Shazly and K.D. Baker, US Patent, 4, 337, 091 (1982).
43. A. Molenaar, 1. Electrochem, Soc., 129, 1917 (1982).
44. F. Pearlstein and R.F. Weightmen, Plating, 61, 154 (1974).
45. M.E. Warwick and B.J. Shirley, Trans, Inst. Met. Finish., 58, 9 (1980).
46. G.O. Mallory, US Patent, 4, 019, 910 (1977). G.O. Mallory and T.R. Horhn,
Plal Slu'. Finis., 66, 40 (April 1979). G.O. Mallory, Plating and Surface Finishing,
63, 34 (June 1976).
47. F. Pearlstein and R.F. Weightman, Electrochem Technol., 6, 427 (1965).
48. K. Baker, Plating, 61, 834 (1970).
49. A.E. Cahill, AES Proc., 44, 130 (1957).
50. F. Pearlstein US Patent 3, 222, 195 (1965).
51. M. Saito, Prod. Finish., 31, 7, 57 (1967).
52. R.1. Zeblisky, J.F. McCormack, J.D. Williamson and F.W. Schneble, US Patent,
3,095,309 (1963).
53. F.W. Schneble, R.J. Zeblisky, J.F. McCormack and J.D. Williamson, US Patent,
3,361,580 (1968).
54. E.B. Saubestre, Plating, 59, 563 (1972).

55. R.D. Fisher and W.H. Chilton, J. Electrochem. Soc., 109, 485 (1962).
56. L.D. Ransom, and V. Zentner, J. Electrochem Soc., Ill, 1423 (1964).
57. Y. Moradzadeh, J. Electrochem, Soc., 112,891 (1965).
58. M.G. Miksic, R. Travieso, A. Arcus and R.H. Wright, 1. Electrochem. Soc.,
113, 360 (1966).
59. 1.S. Sallo, Plating, 54, 257 (1967).
60. J.S. Judge, 1.R. Morrisoo, D.E. Speliotis and J.R. DePaw, Plating, 54, 533 (1967).
61. R.D. Fisher and W.H. Chilton, J. Electrochem. Soc., 121, 1023 (1974).
62. F. Pearlstein and R.F. Weightman, J. Electrochem. Soc., 121, 1023 (1974).
63. D.W. Baudrand, Plat. Surf. Finish., 66, 18 (Nov. 1979).
64. N. Feldstein, Mel. Finish., 81, 35 (Aug 1983).
65. US Patent, 4978559 (1990).
66. US Patent, 4985076 (1991).
67. US Patent, 4979988 (1990).
68. European Pataent, 418715 (1991).
69. German Patent, 284253 (1990).
70. M. Schiesinger and J. Kisel, J. Electrochem. Soc., 136, 1658 (1989).
71. N. Feldstein and 1.A. Weiner, Plating, 59, 140 (1972).
72. N. Feldstein and 1.A. Weiner, J. Electrochem. Soc., 119,668 (1972).
73. N. Feldstein, J.A. Weiner and G.L. Schnable, J. Electrochem. Soc., 119, 1486
(1972).
74. N. Feldstein, S.L. Chow, and M.Schlesinger, 1. Electrochem. Soc.,IW, 875 (1973).
75. E. Matijevic, A.M. Poskanzer and P. Zuman., Plating, 62, 958 (1975).
76. R.L. Cohen and K.W. West. Chern. Phys. Lett., 16, 128 (1982).
77. R.L. Cohen and K.W. West, J. Electrochem. Soc., 120,502 (1973).
78. R.L. Cohen and R.L. ~eek, 1. Colloid and Interface Sci., 55, 156 (1976).
79. Jean Horkans, 1. Electrochem. Soc., 130,311 (1983).
80. B.W. Baulis, C.C. Huang and M. Schlesinger, J. Electrochem. Soc., 126,394
(1979).
81. S.L. Chow, N.E. Hedgecock, M. Schlesinger and J. Rezek., J.Electrochem. Soc.,
119, 1013 (1972).
82. T. Osaka, H. Nagasaka and F. Goto., 1. Electrochem. Soc., 127,2343 (1980).
83. N. Feldstein and J.A. Weiner, 1. Electrochem. Soc., 120, 475 (1973).
84. J.1. Grunwald, S. Gottesfeld and D. Laser, Plating and Surf. Finish., 71 (Oct.
1981).
85. J. Horkans, J. Kim, C. McGrath and L.T. Romankiw, 1. Electrochem. Soc.,134,
300 (1987).
86. H.M. Ghorashi, Plating and Surf. Finish., 694 (1975).
87. R.L. Cohen and R.L. Meek, Plating and Surf. Finish., 47 (1976).
87a. R.L. Jackson, 1. Electrochem. Soc., 135, 3172 (1988).
88. M. Paunovic, Plating, 51, 1161 (1968).
89. S.M. EI-Raghy and A.A. Abo-salama, 1. Electrochem. Soc., 126, 171 (1979).
90. F.M. Donahue, J. Electrochem. Soc., 119, 72 (1972); 121, 887 (1974).
91. P. Bindra and J. Tweedie, 1. Electrochem. Soc., 130, 1112 (1983).
92. P. Bindra and J. Rolden, J. Electrochem. Soc., 132,2581 (1985).
93. D. Vitkavage and M. Paunovic, Plating and Surf. Finish., 48 (April 1983).
94. T.M. Tam, J. Electrochem. Soc., 132, 806 (1983).
95. S. Gottesfeld, 1. Beery, M. Paffett, M. Hollander and C. Maggiore, 1. Electrochem.
Soc., 133, 1344 (1986).
96. J.E.A.M. Van Den Meeraker, J. Appl. Electrochem., 81, 395 (1981).
Metallization o/Plastics by Electroless Plating 99
97. I. Ohno, O. Wakabayashi and S. Haruyama, I. Electrochem. Soc., 132, 2323

(1985).
98. A. Hung. Plating and Surf. Finish., 62 (Ian, 1988).
99. A. Hung, Plating and Surf. Finish., 74 (April 1988).
100. A. Hung and K.M. Chen., I. Electrochem. Soc., 136,72 (1989).
101. B.1. Feldman and O.R. Melroy, I. Electrochem. Soc., 136, 640 (1989).
102. A. Hung and I. Ohno, I. Electrochem. Soc., 137, 918 (1990).
103. A. Hung and I. Ohno, Plating and Surf. Finish., 54 (March 1990).
104. C.E. Baumgartner, Plating and Surf. Finish., 53 (Dec. 1989).
105. R. Sard, I. Electrochem. Soc., 117, 864 (1970).
106. S.L. Chow, N.E. Hedgecock, M. Schlesinger and I. Razek, I. Electrochem.
Soc., 119, 1614 (1972).
107. F. L. Shippey and F.M. Donahue, Plating, 43 (Jan, 1973).
108. F.M. Donahue and F.L. Shippey, Plating, 135 (Feb., 1973).
109. A. Molenaar, M.F.E. Holdrinet and L.K.H. Van Beek., Plating, 238 (March, 1974).
110. F.M. Donahue, DJ. Sajkowski, A.C. Bosio, L.L. Schafer, 1. Electrochem. Soc.,
129,717 (1982).
111. N. Sato, M. Suzuke and Y. Sato, 1. Electrochem. Soc., 130 1485 (1983).
112. M. Suzuki, N. Sato, K. Kanno and Y. Sato., 1. Electrochem. Soc., 129,2183
(1982).
113. K. Kondo, I. Ishikawa, O. Takenaka, T. Matsubara and M. Irie, I. Electrochem.
Soc., 137, 1859 (1990).
114. M. Paunovic, 1. Electrochem. Soc., 124, 349 (1977).
115. S. Ohno, Plating, 58 (1971).
116. K.F. Blurton, Plating and surf. Finish., 52 (Jan., 1986).
117. M. Paunovic and R. Zeblisky, Plating and Surf. Finish., 52 (Feb., (1985).
118. S. Mizumoto, H. Nawafune, M. Kawasaki, A. Kinoshita and K. Araki, Plating
and Surf. Finish., 48 (Dec., 1986).
119. T. El-Mallah, S.M. El-Raghy, E.A. Abou-Saif, M.Y. Saada and A.E. El-Mehairy,
Plating and Surf. Finish., 61 (Oct., 1980).
120. K. Iunginger, I. Electrochem. Soc., 135, 2299 (1988).
121. R. Iunginger and O. Elsner, I. Electrochem. Soc., 135,2304 (1988).
122. Y. Okinaka and H.K. Straschil, I. Electrochem. Soc., 133,2608 (1986).
123. N. Feldstein, Plating, 146 (Feb., 1974).
124. V. Sundarajan, V. Sivasubramanian, K.C. lames Raju, B. Viswanathan and
V.R.K. Murthy, Indian I. Tech., 28, 609 (1990).
125. D.A. Arcilesi, Plating and Surf. Finish., 90 (Jan., 1981).
126. US Pataent 4971944 (1990).
127. K.S. Rajam, Indira Rajagopal, S.R. Rajagopalan and B. Viswanathan, Materials
Chemistry and Physics (in Press).
4
Methods of Measurement of Dielectric
Constant and Loss in the Microwave
Frequency Region
V.R.K. Murthy
Department of Physics
Indian Institute of Technology, Madras, India
In the recent years, there has been considerable interest in the microwave
properties of dielectric and magnetic materials usually termed as microwave
materials. Every cost effective microwave device is a success story in coping
with the limitations of the materials.
The measurements in microwave frequency region may be made purely for
scientific purposes for e.g., in connection with the investigation of relaxation
phenomena. However, due to the increasing importance of the microwave
materials in telecommunications, in the design of microwave circuit components
and quasi-optical components such as substrates, dielectric wave guides, radiation
absorbing materials (RAM), one needs high precision in measurements and in
some cases, the dielectric loss should be critically measured.
In this chapter, various methods are discussed to measure the dielectric
constant, loss, complex magnetic susceptibility, microwave conductivity on
powders and single crystals. These methods can also be used to study the
microwave characteristics of thin films. All these methods have been verified
on standard samples and found agreeable.
1. CAVITY RESONANT METHOD
1.1 Using a Rectangular Waveguide

Cavity perturbation method has been widely used to study the dielectric and mag-
netic parameters in the microwave frequency region. The dispersive and dissipative
terms of the materials are directly related to the change in the resonant frequency
and quality factor of the cavity from the respective empty cavity values.
In order to understand the overall properties of a material, the solutions of
Maxwell's equations in the particular configuration must be understood.
The standard perturbation analysis of this problem is found elsewhere [1,2].
As the resonan! cavity is analogous to an isolated lumped resonant circuit, the
Dielectric Constant and Loss in Microwave Frequency Region 101
standard expressions for calculating the values of capacitance and inductance

of a rectangular cavity excited in its fundamental mode have been arrived [3].
From the Maxwell's electromagnetic equations, for a rectangular waveguide
with broad dimension a, narrow dimension b and length I along X, Y, Z directions
respectively, the orientations of respective electric and magnetic fields can be
deduced as Ey, Hx and Hz (where Ez, Ex and Hy are zero). In this case, the
assumption is that the cavity is parallel resonant circuit with a capacitance Cy
corresponding to an average value of Ey and the resultant inductance Lxz from
the parallel combinations of two inductances Lx and Lz corresponding to the
average values of the respective magnetic fields Hx and Hz. From the low
frequency lumped circuit elements analogy
C _ (eoal)
y - b (1)
L _ JJ. oN 2 al (2)
xz - (a2 + 12)
Where 1= P A,J2 for TE lOp mode and N has usual meaning similar to the number
of turns in a solenoid. If N is replaced by l/rc, the above equations result to
10= 1 (3)
2tr.JLxz Cy
Where 10 is the frequency of the cavity.
1.1.1 Dielectric constant e' and loss e"

A sample of complex dielectric susceptibility Xc =X~ - jX; (equals to
e' - 1 - je") is kept at the maximum electric field location of the cavity. A
thin sample (for example a maximum of 3 mm diameter or side) is taken such
that the electric field is uniform through out the volume of the sample. In order
to simplify the problem, the sample is assumed to be cylindrical or rectangular
with uniform cross-sectional area S and length is greater than or equal to b so
that it will occupy the narrow section of the cavity.
After the introduction of the sample, the empty cavity's resonant frequency
and Q factor alter due to change in the overall capacitance and conductance of
the cavity without perturbing the inductance.
The resonant frequencies with and without the samples are/l and/o, correspon-
dingly the dielectric constant e' is given by
~2
e' = 1 + J 0 -2
121 V.
_c (4)
II 4V.
and the dielectric loss e" is given by
en= ~[/~
4V. if at
(..!. __Qo1)] (5)
where Vs is the volume of the sample, Vc is the volume of the cavity, Qh Qo

are the respective quality factors of cavity with and without the sample.
1.1.2 Magnetic permeabili" J.I.' and loss J.L"

In this case, the sample is introduced at maximum magnetic field location. The
sample is aligned along Hx for which Hz and Ey are least perturbed. The
permeability J.I.' i~ given by
J.I.' =1 +.1 Yc
K V.
(/011-11) (6)
J.I."= ~(~ -2~J (7)
where
K = 2a 2 (8)
a 2 + 12
The details of the experimental arrangement used in this type of measurement
is given in [3].
1.2 Measuremeut or Dielectric Constant and Loss on Small Single

Crystals Using a Circular Cavity
The principle is based on the perturbation theory developed by Buranov and
Shchegolev [4]. This method is proposed for measuring the dielectric constant
and loss of cylindrical samples whose length is smaller compared to wavelength,
while its transverse dimensions are comparable or less than the skinlayer thickness.
The cavity is a circular cavity as shown in Fig. 1 operating in TEclll mode
with its Q = 10,000. The appropriate equations for the perturbation caused by
insertion of a small sample at an electric field antinode are
Fig. 1. Device for mechanical movement of the sample. I-Sample, 2-Yokes,

3-Capron filament, 4-Pinions and 5-Top cover of the cavity.
Dielectric Conslanl and Loss in Microwave Frequency Region 103
.10.1 ae" (9)

T = 2/0 = [1 + n(e' - 1)]2 + (ne'')2
.1/0 = .1/ = a[(e' - 1)(1 + n(e' - 1) + (ne H

)]
(10)
10 [1 + n(e' - 1)]2 + (ne")2
where fo is the centre frequency of the cavity, .1 is the change in the half power
bandwidth of the cavity with sample and without, .1fis the change in the centre
frequency of the cavity with and without. For a cylindrical cavity operating in
=
TEall mode, die filling factor a 2.1 (volume of the sample/volume of the
cavity). The depolarization factor n is evaluated for (a) long cylindrical sample
(L » d) as
(11)
and (b) for an ellipsoid of revolution as
n = bc [ln~ -1] (12)

a2 b+ c
where a, b, c, are ellipsoid semiaxes. Consequently, the dielectric constant e'
and loss e" are evaluated by
(13)
(14)
These results assume that the sample is completely permeated by the

microwave field i.e the skin depth 0 > I, where I is the thickness of the sample
perpendicular to the field. There are several difficulties with this technique,
probably the most serious is that many poor conductors exhibit much higher
apparent conductivities (often 5 or 6 orders of magnitude) than at dc. Thus any
high conductivity microwave results should be checked and confmned by dc
or low frequency conductivity measurements. Nevertheless, this technique is
useful as an excellent screening method.
The other drawbacks have to do with geometrical factors, especially the
depolarization factor. However, this method can be comfortably used on
anisotropic crystals also.
1.3 Measurement or Dielectric Constant and Loss on Powder Samples

Using Rectangular Cavity
Many samples are available in powder form and some of them are prepared in
small quantities. The measuring technique and theory [5] in the foregoing part
are in general and may be applied to research on such powders.
The real and imaginary parts of the dielectric constant are given by
(15)
(16)
where the parameter ~refers to the change in frequency or Q factor, V e, Vs are

the volumes of the cavity and sample respectively. It can be seen that if we
take N = =
1/3, the equations are valid for a spherical sample and N 0, they
are valid for rod of infinite length.
However, it is explicitly agreed that one should analyze the depolarization
factor for the evaluation of dielectric parameters. However, by taking the powder
sample of known weight into a polyfoam cylindrical tube of known diameter
that can be used as a sample holder for cavity measurement, the depolarization
factor N can be calculated using Wurschchmid's method [6]. The corresponding
factors such as volume of the sample and cavity can be easily determined.
Further, the author has theoretically calculated e' and e" for various values
of N taking known weights of the sample into the polyfoam sample holder and
computed graph between various values of depolarization factor and corresponding
values of e' for each weight of the sample. Thus for a known value e' = 14
sample the experimentally measured values are fitted into the above equation
for the evaluation of N that has agreed with theoretial value of N. However, for
higher values of e', a slightly inevitable disagreement between theoretical and
experimental values of N could be seen which can be interpreted due to slight
changes in cylindrical shape of the polyfoam sample holder, inhomogeneity of
the sample packing and so on.
1.4 Microwave Conductivity Measurements on High Te Superconductors

Since the discovery of superconductivity with transition temperature (Te) above
90K in YBa2Cu307_x system, extensive studies have been carried out on
superconducting ceramic oxides. Even in superconducting state, the ac resistivity
does not vanish and it is increasing with frequency. The reason may be the
thermal dissociation of the superconducting fluid into normal electrons in between
these Tcs obtained by direct current (dc) and microwave frequency (ac) and
also due to the effect of thermally excited quasi particles.
The method of microwave conductivity measurement is described [7] using
a rectangular cavity.
In this study, the change in Q (quality factor) values were measured using
a rectangular cavity in frequency modulation mode. The value of Q for empty
cavity is of the order of 103 at room temperature.
Figure 2 shows the flow chart of the microwave conductivity measuring
system employed. A rectangular cavity of TEJ()6 mode with resonant frequency
at 9 GHz (X-band) was used. The cavity was coupled to the system and evacuated
I FM
MODULATION
MW SOURCE UL TRA- FAST
HY BRIO INTERFACER
INTEL'S
ANALOG ~
ISOLATOR 8255
TO DIGITAL \r-V'
1 CONVERTER
I ~~
FR EQUE NC Y ATTENUATOR HAS 1002
COUNTER
J.
CIRCULATOR
T
ULTRA HIGH
I SPEED HYBRID MIC R 0-
TRACK & HOLD PRO C ESSOR
r - --
I
----,
I
CRY STAL
DETECTOR
1
1- AMPLIFI E R
HTS 0025
SYSTEM
INTEL'S
8085
I I
I
I CAVITY
I
OSCILLOSCOPE T
I LOW NOISE
rt
I CURRENT - ........, LOW ORIFT
I I PRECISION
VOLTAGE
L
I ______ .JI CONVERTER IE- FET 52 J
AMPLIFIER
TEMPERATURE
& VACUUM l INTERFACER
CONTROLER DIGITAL TO
x-v INTEL'S
~
ANA LOG
SVSTEM ""- CONVERTER
RECORDER 8255
DAC 0080
Fig. 2. Block diagram of the experimental measurement unit employed.
to about 0.13Pa. The sample is kept at the maximum magnetic field in the
cavity, so that maximum current density can be achieved in the sample. The
frequency modulated power is fed to the cavity and the response of the cavity
is plotted using XY recorder. The microwave conductivity is calculated from
change in Q values
L1 ( -1) = -1 - - 1 = Po (moW) (17)

Q QI Q2
where Ql and Q2 are the quality factors of the cavity with and without the
sample, Po is Joule loss due to sample, ~ = 2tclo, 10 being the resonance
frequency of the empty cavity and W is the power stored in the cavity. The W
arid Po are calculated using the relations
- 0Os
W- e E2 abd (18)
(19)
where Eo is the maximum value of the electric field, a, b, d are the dimensions
=
of the cavity along breadth, width and length respectively, Rs (O'c5t 1 is the
surface resistivity of the sample and c5 is the skin depth at the frequency of
=
operation (2/.aJoJJoO')lfl asb. is the total surface area of the sample, Eo, JJo are
the pennittivity and penneability of vacuum. From the above relations, the
conductivity is given by
0'= 2p4 tr 4(as bs)2 2 (20)

(Q)~e~d6JL~) (ab)2 (L\ ~)
where P has usual meaning in TEmnp mode.
From the above relation one can see that conductivity is proportional to
square of the microwave loss (L\(1/Q»2.
2. DIELECTRIC MEASUREMENTS USING SWEPT

FREQUENCY TECHNIQUES
In this approach, a swept frequency dielectric measurement [8] is presented
which is shown to have advantages over conventional technique. This method
shows a basic accuracy of better than 5% depending on the material under test.
The most conventional methods [9, 10, 11) are simplified considerably if the
sample length is an odd or even number of quarter wavelengths in the material.
However this condition is usually difficult to achieve in practice since knowledge
of the wavelength inside the sample implies knowledge of the complex dielectric
constant one is trying to measure.
It becomes quiet obvious that for measuring dielectric constant, a very desirable
condition is to have a sample length of one quarter wavelength or multiple
thereof. It is rather awkward to change the physical dimensions of the sample
in a continuous manner, instead the electrical length of the sample can be
varied by changing the frequency. If now, the assumption is made that the
dielectric properties of the material under test do not change appreciably with
frequency over a relatively short range, then changing the electrical length by
sweeping over some band, has the same effect as changing the physical length
of the sample. In this way, it is possible to create conditions such as a quarter
wavelength which makes it relatively easy to calculate the complex dielectric
constant of the material. This approach forms the basis for the swept frequency
method. The experimental measurement is basicaUy a reflection measurement
which relies on interface reflections and the attenuation of the sample to give
information about the complex dielectric constant.
The basic experimental setup consists of a levelled microwave sweep generator
that sends the incident wave to the sample, which will reflect and transmit
some of the energy. If the reflected signal is detected and displayed versus
frequency, a superimposed standing wave pattern results.
The dielectric constant t' and loss tangent (tan 8 =e"/e') are given by the
following relations
E'=tE[I+~I- ~] (21)
where E =(cf2ll1f)2 and

tan 8= c , ~e'-Pln(l1) (22)
(2tc/l /e)
c =velocity of electromagnetic wave (light); I =
length of the sample;
P = (hP2 + IIPI)/(P I + PV; 12. 11 = frequencies of adjacent maxima in the
=
region of swept frequency. Ph P2 approximately the ratios of cut-off frequencies
to the actual frequencies at the adjacent maxima.
tl/=12 -II
11 = (I - R IT} I) (23)
(R - IT} I)
R =amplitude of reflection maximum.

(24)
Thus from three experimental quantities, viz, the sample length, the frequency
difference between two extremes and the amplitudes of maxima, one can evaluate
e' and tan 8. This method can also help one to decide whether or not there are
any inhomogenities present in the sample such as large voids and cracks.
3. MEASUREMENT OF DIELECTRIC CONSTANT AND LOSS ON

HIGH PERMITTIVITY MATERIALS
3.1 Ferroelectrics
Lanagan et al [12] have described this method for measuring on high permittivity
materials. This method can be essentially used on ferroelectrics and is based
upon S-parameters measurement using microwave network analyzer.
The specimen is a discontinuity for travelling waves and generally the incident
wave incidents at an angle on the air/dielectric boundary as shown in Fig. 3.
Since the incident wave has a reflection coefficient equal to unity and the
reflected wave can be described as complex quantity given by
(25)
where e" = e' - jE".

The remainder of the incident wave is transmitted through the dielectric
material until the next dielectric discontinuity is reached. The magnitude and
108 MICRoWAVB MATERIALS
INCIDENT WAVE
.
--------
(1 - f 2 ) e(-r I )
1'(1- f'2) e(-2rl)
S,1 5 21
p2(1_ p 2)e(-3rl)
p3(1_ p 2)e(-4rl)
Fig. 3. Plane wave incident to the interface of two different dielectric media.
phase of the transmitted wave as a function of the propagation constant r is

expressed as
(26)
21re'
p=;:;- (27)
where tan ~ = e" Ie'; ~ is free space wavelength and J,t is complex permeability
which is unity in case of nonmagnetic materials.
The sum of the reflected and transmitted waves can be expressed in terms
of the scattering parameters S11 and S21 which characterize two port device
such as a waveguide connected on both ends. The scattering parameters are
usually taken as complex quantities in terms of magnitude and phase given by
S21 = 1S211 e''9 (28)
The internal reflections shown in figure are accounted for by an infinite
geometric series and relations for S11 and S21 are given by
S11= P 1 - exp (-2rl)

(29)
1 - p2 exp (-2rl)
S21- (1 - p2) exp (-2rl) (30)

- 1 - p2 exp (-2r/)
I is the sample thickness.
In order to minimize the complication to solve for e from the above equation,
an approximation is made in the above equation which requires S21 measurement
at several frequencies. For high dielectric constant and high loss materials,
{i'- exp (- 2rl) can be assumed to be less than unity. The above equation is
simplified to the following equation S21 = (1 - (i'-) exp (- 2r/). A plot of
as
log S21 vs frequency is linear per the above equation. The dielectric constant
and loss interms of S21 phase and magnitude are given by
e'= [(!~) 2~r (31)
tan a= (.1 IS 1) [8.686 c]

21 (32)
.dv mJ'?
The parameter c is the speed of light in free space. (.11 s21 1/.dv) and (.d8/.dv)
are the slopes of magnitude and phase with respect to frequency v. The phase
difference is expressed interms of radians and magnitude difference in dB.
3.2 Dielectric Resonator Ceramics

The most important characteristics of a dielectric resonator are high dielectric
constant (e of the order 1(0) and high Q (> 10,(00) and temperature coefficient
of resonant frequency should be minimum. The operating frequency of a particular
mode of dielectric resonator can be controlled by the physical dimensions of
the sample and the dielectric constant. The dielectric constant can be measured
by the method suggested by Hakki and Colemon [13] and Courtney [14].
The measurement method involves placing the dielectric resonator (usually
taken in the form of a disc) between two perfectly conducting plates. When the
metal plates make contact with dielectric resonator, the field equations can be
expressed as a transcendental equation relating resonant frequency, dielectric
constant and size of the resonator. In general the resonator can support various
TE, TM and hybrid HEM modes. Any mode can be used to measure the
dielectric constant. However, TEont is most commonly used mode for the
evaluation of e' due to its insensitiveness to the presence of air gap between
the metallic plates and the dielectric resonator, the zero index in TEont refers
to the field variation along the azimuthal direction, n and I refer to the field
along the radial and axial direction respectively. However, TEoll mode can be
identified from all the existing modes both by mode chart and experimental
arrangement that can be further used for e evaluation [13].
The experimental arrangement is shown in Fig. 4. The short circuited resonator
is operated as a transmission resonator with small coupling antennas used to
couple in and bring out the power. The antennas are connected to the network
analyzer which can gauge the resonance frequencies of various modes. Among
these modes, TEoll can be easily identified.
The characteristic equation for TEont mode is
(33)
I I
I
I
I
I 1 I
I
I i
I
Fig. 4. The Courtney holder.
where Jo(a) and J 1(a) are two Bessel functions of the first kind of the orders
zero and one respectively; Ko(/3) and K1(/3) are the modified Bessel functions
of the second kind of orders zero and one respectively; f3 is the radial wave
number outside the dielectric resonator and a is the radial wave number inside
the dielectric resonator. The values of a and f3 are given by
2 ]1/2
f3 = 1~ [(~) -.1 . (34)
2]112
a= j~ [ E'- (~) (35)
Ao is the free space wavelength, D is the diameter of the resonator, L is the

=
length of the resonator, I 1, 2, 3... corresponds to the multiple half wavelengths
along the axial directions of the resonator.
The Q of the dielectric resonator is measured using a test holder given in
Fig. 5. The same can be used for temperature coefficient of resonant frequency
(tr). The measurement can be done with network analyzer. The Q can be calculated
by measuring the magnitude and phase of the reflection coefficient as a function
of frequency as described by Kaifez and Hwan [15]. The temperature coefficient
of resonant frequency is measured using the equation
",=--
1 i1f
fo i1T
(36)
where fo is the resonant frequency at the starting temperature.

Dielectric Constant and Loss in Microwave Frequency Region III
T
Fig. 5. Dielectric resonator in a cylindrical cavity.
Thus, in this chapter, emphasis was given to describe the various reliable
experimental methods to measure the dielectric constant, loss and conductivity
in the microwave frequency region. The materials include polymers, ceramic
oxides such as ferrites, titanates, dielectric resonators and high Tc superconductors.
REFERENCES
1. 1.S. Artman and T.E. Tamrnelwald,l. AppJ. Phys., 26, 1124, 1955.
2. B. Lax and K.I. Button, Microwave Ferrites and Ferrimagnetics, McGraw-Hill,
New York, 1962.
3. V.R.K. Murthy and R. Raman, Solid Stale Commn., 70, No.8, 847, 1989.
4. L.T. Buranov and I.F. Shchegoler, Translated from Pribory i Tekimilon
Eksperiments, No.2, 171, March-April, 1971.
5. Iroshi Kobayashi, lap. 11. Appl. Physics, 10, No.3, 345, 1971.
6. H. Zijlstra, Experimental methods in magnetism. 2 (North Holland), Ch. 2, p.70,
1967.
7. R. Pragasam, N.S. Raman, V.R.K. Murthy and B. Viswanathan, Physica Status
Solidi, Vol (a), 125, 583, 1991.
8. W.R. Tinga and E.M. Edwards, 11. Microwave Power, 3 (3), 112, 1968.
9. S. Roberts and A. Von Hipple, 11. Appl. Phys., 17,610, 1946.
10. W.H. Surber Ir. 11. Appl. Phys., 19,514, 1948.
11. C.O. Montgomery, Techniques o/microwave measurements, (McGraw-Hill Book
Co. Inc, New York), Ch. 10, 1947.
12. M.T. Lanagan, 1.H. Kim, D.C. Dube, S.1. lang and R.E. Newnham, Ferroelectrics,
82,91, 1988.
13. B.W. Hakki and P.D. Coleman, IRE Transactions on Microwave Theory and
Techniques, Vol MTT-8, pp. 401-410, 1960.
14. William E. Courtney. IEEE Transactions on Microwave Theory and Techniques,
Vol MTT-S, pp. 476-485, 1973.
15. D. Kaifez and E.I. Hwan, IEEE Trans. on Microwave Theory and Techniques,
Vol. MTT-32, pp. 666-670, July 1984.
5
Microwave Ferrites
G.P. Srivastava
Department of Electronic Science
South Campus, Benito luraz Road, New Delhi, India
Bijoy K. Kuanr
Zakir Husain College
lawaharlal Nehru Marg, New Delhi, India
1. INTRODUCTION
Emergence of ferrites arises due to the search for ferromagnetic materials with
low eddy current losses. The high resistivity of these materials is the primary
factor for controlling the eddy current losses, which is useful as cores for
transformers and inductors. The present status is that ferrites have been established
as materials of immense industrial use and ferrite devices find numerous applica-
tions in entire frequency range. At higb frequencies, as in the communication
field, the advantage of the ferrites become more pronounced specially at
microwave levels. Great bulk of the microwave applications will be quite
impossible without the assessment of ferrite and garnet materials.
The foundation of modem interest in ferrites was laid by Snoek [1] in 1946,
thereafter, the basic theory of magnetism developed by Neel [2] in 1948 was
the starting point for the rapid expansion of research and development activities
in this field of materials. The importance of ferrites is due to the fact that they
possess the combined properties of a magnetic material and an electric insulator
[3]. As a results. a high frequency electromagnetic wave can be propagated
through ferrites with very low attenuation. The work of Gorter [4], Goodenough
[5] and Blasse [6] showed that it is possible to make ferrites with different
substitution having a very wide range of saturation magnetisation and Neel
temperature values. These materials are also important from the point of view
of theoretical investigation because of their different type of spin arrangements
and exchange interactions.
Because of the large application of ferrites in microwave frequency range,
a designer of the ferrite devices has to know the detailed characteristics of the
material. Important parameters needed to characterise the ferrite material at
high frequencies are the resonance line-width (&I) and spin-wave line-width
(&I,~). &It is a measure of the power handling capability of the material.
Microwave Ferrites 113
There has been steady interest in developing new or improved latching type
high power ferrite phase shifters for use in phased array scanning antenna. The
device performance of phase shifter is Iimi~ by the ferrite material parameters
mentioned above. Inadequacies in currently available commercial ferrite and
garnet for use at high power levels has prompted the need for further material
development with improvement in power-handling capability as well as hysteresis-
loop properties. In addition, to the requirement for suitable hysteresis-loop
properties, materials for microwave latching phase shifters should have low
dielectric losses.
Lithium ferrite with chemical formula LiO.5F~s04 is an inverse spinel [7]
having a wide range of device applications. Several research programs have
been undertaken on this series of material to study its fundamental properties
and especially to develop high-power microwave materials from it. The utility
of Lithium ferrite and its substituted compositions as a class of materials for
application in microwave components and memory devices is due to the following
characteristics of the materials [8, 9].
1. Lithium ferrite can be prepared with a low value of tAH, with proper
substitutions of non-magnetic materials of the order of 50 oersted. This
is an extremely useful parameter for use in microwave devices.
2. Lithium ferrite can be used in high-power microwave devices. The
power-handling capability can be raised with appropriate non-magnetic
substitutions (like relaxing impurities) in its composition. Hence lithium
ferrite have come up as a low cost substitution for expensive garnets.
The unique properties of lithium ferrites which render them invaluable in
microwave applications is the employment of appropriate chemical substitution
in the basic formula unit. Although a wide range of substituted lithium ferrites
are available commercially, those in general used at microwave frequencies
include Ni, Zn, Mg, AI, Co, Ti etc.
The proper doping level and amount of dopants leads to material with high
Curie temperature, good density, low microwave dielectric losses, high hysteresis
loop squareness, cover a wide range of saturation magnetization values, low
stress sensitivity, high resistivity, high permeability and above all improved
microwave power handling capability.
2. FERROMAGNETIC RESONANCE (FMR) AT LOW POWER

When a ferrimagnetic material is subjected to the simultaneous action of a rf
field perpendicular to a dc magnetic field, under suitable conditions the sample
is found to absorb energy from the rf field. This phenomenon is referred to as
ferromagnetic resonance. The absorption of microwave power by the sample
occurs when the precessional frequency of the magnetization vector about the
direction of the externally applied dc field equals the frequency of rf field. This
resonant absorption of the electromagnetic radiation by the ferromagnetic material
is called FMR. In a usual FMR experiment the applied field has two components-
a large time independent field Happ and a small perpendicular "sinusoidally varying
field, 'sin Wt'. The first ferromagnetic resonance experiment was performed
by Griffithis [11] in 1946. Until today a lot of experiments have been made by
various workers for the complete understanding of relaxation processes in
ferromagnetic insulators. In a typical FMR experiment a sample is placed in a
uniform magnetic field large enough to magnetize it parallel to the field direction.
The absorption of rf power from the sinusoidally varying rf signal is a function
of Happ for fixed frequency. A resonance peak is obtained if the absorption is
plotted as a function of H app' The full width, in field units, at half maximum
absorption point is designated as the resonance line-width tJI. This simple
description of FMR contains the three basic ingredients common to resonance
phenomena are precession, resonant response and relaxation. The understanding
ofFMR was advanced considerably by Kittel [12] when he pointed out the role
of demagnetizing fields in determining the resonance conditions. The usual
resonance condition with r(gyromagnetic ratio) is satisfied for most materials
when demagnetizing [13, 14, 15] field are taken into accounL A phenomenological
description of the FMR absorption was given by Van Vleck [15]. He discussed
the observed line-width and various possible sources for line broadening.
Microwave phenomena have further expanded with discovery of YIG [16].
Sparkes [18] has made a detailed study of FMR in YIG and pointed out the
important role played by spin-waves. Kittel's theory was modified [19] for the
ferrimagnetic material which can be described by two sub-lattice systems of
oppositely directed magnetization. Further Brown and Park [20] had shown
that there were two types of responces, one in which the two sublattice magnetiza-
tions precesses about the applied field direction out of phase i.e. the exchange
resonance and the other when they precess in phase, which is the ferromagnetic
case. The size effect of FMR in ferrites was observed by Beljers and Polder
[21]; the resonance field was found to be a function of the diameter of the
ferrite sphere.
Yager et al [22] have obtained the linewidth of single crystal nickel ferrite.
It is 70 oersteds. Extensive studies were made on the variation of LlH with frequ-
ency [23], temperature [24], and ferrite composition [25]. The behaviour of
LlH with temperature could be attributed to the cation distribution [26] and the
valence state of the iron ion [27], in the ferrite. Lecraw and Spencer [28] had
shown that the linewidth in pure YIG was of the order of few millioersteds.
2.1 Spin Waves

In the usual FMR experiments the rf field is applied perpendicular to the dc
field and in resonance the uniform mode (i.e. the spins are parallel to one
another) alone is excited. This is a normal mode of the spin system. Spin wave
analysis has been used many times for the description of the resonance behaviour
of ferromagnetic spinel as well as garnet systems. The wavelength of the spinwave
is, under most conditions, much smaller than the wavelength of an electromagnetic
wave of the same frequency. Therefore, a spin wave can be excited by virtue
of its dipole moment and this is small because the positive and negative
contributions arising from different parts of the wave tend to cancel out to a
high degree. Other modes in which the spins do not precess in phase with each
other also exist. Such disturbances in the spin system are called spin waves and
Microwave Ferriles 115
=
is characterised by a wave vector k whose magnitude is I k I 21ft)., where
). is the spin wave wavelength.
=
For the uniform mode k O. The relaxation process assumed that energy is
scattered from the uniform precessional mode into higher order modes [29].
But Clogston et al [30] provided a clear picture for the understanding of the
exact mechanism of the energy transfer by taking into account the effect of the
dipole-dipole interaction. The dipole-dipole interactions broadens the spin wave
dispersion relation into a band, which was named as spin wave manifold. The
spinwave manifold may be defined as that region of the dispersion relation
=
where spin wave analysis is valid. In this manifold k 0 spin waves also exist,
which are degenerate with the uniform precession.
The relaxation process assumes that energy is scattered from the uniform
precessional mode into degenerate spin wave modes. The analysis of this scattering
process depends upon the knowledge of the relationship between wavelength
and resonant frequency of such modes. According to the theory of Clogston
[30] the relaxation process can be visualized as follows. In a resonance experiment
the rf field puts its energy into the uniform precession. Thus the system has an
= =
excess of uniform precession magnons (k 0). The k 0 magnon can relax to
the lattice through three channels as shown in Fig. 1. In the first process a
k = 0 magnon can be annihilated and a k = 0 degenerate magnon can be created
=
in a two magnon process, with the k 0 magnon eventually relaxing to the
lattice through processes involving magnons and phonons. Secondly a k 0 =
magnon can relax to the lattice via interactions with magnons other than the
degenerate magnons. In the third process, a k = 0 magnon can directly relax
to the lattice through processes not involving other magnons. The two magnon
process [31] has the dominant mechanism for relaxation in polycrystalline ferrites.
Fig. 1. General scheme for relaxation processes.
The derivation of the spinwave manifold hold as long as the wavelength of

the spinwave is much smaller than the size of the sample so that the surface
demagnetization fields of such magnons can be neglected. For large wave-
lengths these fields must be taken into account. Walker [32] solved the problem
by including these surface demagnetization fields and the resulting modes are
=
called the magnetostatic modes. The uniform mode has k 0 and therefore is
a magnetostatic mode. Walker's calculations showed that the magnetostatic
modes has a spread wider than the spinwave manifold i.e. the upper limit for
the Walker modes does not correspond to an extension of the upper limit of the
spinwave manifold. The bottom limits however, do coincide.
The spinwave manifold shown in Fig. 2 can be divided into two parts. A
very flat region where the exchange interaction plays no part is the first region
and the second is a curved region where the exchange effect is appreciable.
Irl [Ho-tNz- N >M]
Irl [140 -NZ M]
o K - -.......
Fig. 1. The spin wave spectrum.
Geschwind and Clogston [33] and Schlomann [34] have explained linewidths
in polycrystalline materials by calculating the coupling of the uniform precession
with degenerate spinwaves of medium k values (i.e. in the no exhange region).
They have neglected the effect of spinwave with high k values, where the exchange
effects are appreciable. In spinel ferrites the scattering arises from a distribution
of the magnetic ion on the octahedral sites. The coupling between the uniform
precession and the high k-spin waves is thought to be stronger in the spinel
than in the garnet. So the spinels have greater line width as compared to those
of garnets.
3. HIGH POWER FMR

Now it is a well known fact that ferrimagnetic materials display nonlinear
loss characteristic at high levels of peak microwave power. It has been
known since the work of Damon [35] and of Bloombergen and Wang [36] that
at high power the phenomena of ferromagnetic resonance are quite different
from those observed at low power levels. Two unexpected effects come to
light:
1. The main resonance line appeared to saturate and broaden steadily as

the signal power was increased beyond a threshold value.
2. Secondly, at a similar signal power, an additional rather broad absorption
peak appeared, a few hundred oersted below the dc magnetic field
required for the main resonance.
These two effects are shown here in Fig. 3.
Subsidiary Main
resonance resonance
c
I
o
...o
CII
.J:l
a
....
.Q
CII
a:
Biasing megnetic field, Ho

Fig. 3. Subsidiary absorption and premature saturation.
The presence of these nonlinear effects can seriously interfere with the
performance of microwave devices such as isolator, circulator and phase shifter.
The reduction of main resonance can impare the performance of resonance
isolators. The subsidiary resonance which appeared at low dc field cause serious
deterioration in the high power performance of phase shifter and circulators.
So it is important to understand the cause and cure of these nonlinearity.
Suhl [37] has explained these nonlinearities theoretically. Both the above
mentioned effects arise from power dependent coupling between the so-called
uniform mode of magnetic precession which is driven by the rf field and
certain spinwaves which become excited if the applied microwave field exceeds
a critical value. The spin waves which have the same frequency as that of the
applied field are responsible for the saturation of the main resonance and the
spin waves having the frequency one half of the signal frequency causes the
subsidiary absorption peak. The threshold value required for these non Ii neari ties
differ for the two types, but, both threshold (hcrj.) depend upon the linewidth,
the saturation magnetization, the geometry of the sample and the operating
frequency. In some cases the subsidiary absorption peak coincides with the
main resonance, this happens when threshold field are low.
Suhl [38, 40] presented a theoretical investigation on the behaviour of ferrites
at high microwave signal levels and showed that the nonlincariLies were connected
with two kinds of instability [39] of the uniform precessional motion of the
total magnetization against certain spinwave disturbances. And these disturbances
grow exponentially when the signal level exceed a certain threshold value.
Schlomann [41] had developed a theory of ferromagnetic resonance in which

dipolar interaction was taken into account by means of the spinwave formalism.
His theory predicts a very strong frequency and shape dependence of the line
width for the case in which the homogeneous mode of precession was approxi-
mately degenerate with long-wavelength spin waves propagating in direction
perpendicular to the dc field. The influence of inhomogeneities on the saturation
of the ferromagnetic resonance was investigated. Schlomann et al [42] observed
the nonlinear effect arising from spin wave instability in a microwave magnetic
field applied parallel to the dc field which termed as parallel pumping spin-
wave instability. They have investigated on cobalt and zinc substituted nickel
ferrite and single and polycrystals of rare earth substituted garnets. Their results
indicated the increase of spinwave linewidth with increasing wave number and
decreasing angle between propagation direction and dc magnetic field. LeCraw
et al [43] studied FMR at 9.3 and 3 GHz in single crystal YIG and observed
that hcnt was the lowest at room temperature, but as the temperature was increased
h crit rises quite abruptly. More works [44] on parallel pumping technique have
been made over a wide range of power levels and applied dc fields. A confluence
process was given to explain the susceptibility curves, obtained on single-
crystal YIG. Sounders and Green [45] measured subsidiary absorption on
polycrystalline aluminium substituted YIG and nickel ferrites, and shown that
spinwave linewidth was independent of aluminium content. Schlomann et al
[46] had analysed the power-handling capability of substituted garnets used in
circulators and phase shifters. They have defined the "high power figure of
merit" for the measure of the suitability of microwave ferrites, for high power
applications. They observed for ferrites, containing appreciable substitutions of
strong relaxers, a high figure of merit (Fhp) by reducing the saturation
magnetization. Single crystal nickle zinc and lithium zinc ferrites [47] were
studied at room temperature and 77°K and at 9.0 GHz. His experimental data
showed that spin wave Iinewidth reduced by a large amount when measured at
77°K.
For oblique pumping, the threshold microwave field amplitude (hcriU required
for the unstable growth of spinwave was investigated [48] with a linearly
polarized microwave field applied at an arbitrary angle with respect to the
static magnetic field. Green et a1 [48] performed their experiment on YIG spheres
at 9.2 GHz and used a pulsed magnetron as source with rectangular TE102
cavity. They have derived theoretical expressions for the oblique pumping hcrit>
and have obtained a fair agreement of their experimental data with the theory
by assuming a transverse demagnetizing factor Nt> greater than value of 1/3.
Extensive data on the oblique-pumping threshold as a function of pumping
angle have been obtained for external fields and the theory of parallel and
perpendicular pumping in saturated ferromagnetic insulator was extended to
include oblique pumping. Patton [49] has given a quite versatile theory which
was applicable to a large variety of microwave field configuration which were
not accessible on the basis of earlier theories [37, 41, 42]. Patton [49] has
developed equations for first order instability threshold and showed that sample
geometry and the pump configuration have a strong influence on the instability
threshold. The first order spinwave instability threshold has also been measured
[SO] as a function of sample shape (sphere, rod and thin disk), sample orientation,
pump configuration and magnetization state. The effect of sample shape and
pumping configuration on the threshold field for saturation was compared with
the theory. They have observed a consistent correlations between the data and
the theory which provided that domain shape plays the same role in determining
the threshold for partially magnetized samples that sample shape does for
saturated material.
Spinwave linewidth on polycrystalline YIG observed as a function of porosity
[51] pump configuration [52] and grain size [53, 60, 63] provides a clear
picture for the understanding of threshold mechanism. A spin wave transit time
model {55, 61] was given to explain the experimental data. Effect of nonmagnetic
inclusions on the spinwave linewidth in polycrystalline YIG was observed by
Schotter [57]. The threshold field for nonlinearity for first order processes on
the various configuration of static field angles were studied [54] on single
crystal YIG. Schouer [57] has studied the effect of rare-earth doped YIG
on spinwave linewidth and showed that the high power capability of ferrites
(,1HIc~ was increased with rare-earth ooping. Parallel pump [58] and subsidering
absorption [59] spinwave linewidth have been studied in porous polycrystalline
YIG (as a function of temperature) and single crystal YIG respectively. The
effect of crystalline anisbOphy for any arbitrary microwave pumping configuration
on spinwave instability theory [60] of single crystal YIG [62] provided a good
agreement with experimental data. Spin wave linewidth in polycrystalline YIG
[63] was compared to the theory of Sawado [62] for spinwave scattering by
nonmagnetic pores.
3.1 High Power Effects

Great technical interest lies in high power FMR. One can selectively excite
spinwave of a particular wave number through proper choice of dc field and
can couple energy into that spinwave. The basic thing is to pump microwave
energy into the spin system and observe the change in susceptibility with
increasing amplitude of microwave power. This process results in a greater
=
excitation of the k 0, i.e. uniform precession spinwaves, which in tum results
=
in an greater excitation of k 0 spinwaves to which the uniform precession is
=
coupled. If the rate of energy transfer to a particular k 0 spinwave exceed the
rate at which energy can be transferred from the spinwave, the amplitude of the
spinwave increases exponentially, hence there is a abrupt change in susceptibility.
The instability condition depends on the rate at which energy can be dissipated
=
from the particular k 0 spin wave. This rate is characterized by a linewidth
parameter called as spinwave linewidth, ,1Hk , which has to do with resonant
excitation at high power. ,1Hk is a linewidth only in the frequency swept linewidth
sense. The threshold microwave field amplitude herit at the break point of the
susceptibility curve can be related to the relaxation rate for spin-wave mode
with the lowest threshold for prescribed experimental conditions. ,1Hk is defined
as 'lDk/y, where ris the gyromagnetic ratio. It can also be defined in terms of
lifetime of a magnon (k) of certain k as IIrk. So, the measurement of nonlinearity
threshold provides informatiqn about the relation of a particular k = 0 spin waves.

The excitation of specific modes provide a detailed investigation of the spinwave
spectrum and the energy transfer mechanism with the spin system. At high
power, coupling between the microwave field and certain spinwave modes
leads to a non-linear behaviour on the response curve. The discovery of non-
linear phenomena provide a new insight into microwave relaxation process.
The instability threshold can be determined by two techniques those are
perpendicular or transverse pumping and the other is parallel pumping technique.
The configuration in which the microwave field is perpendicular to the dc
magnetic field, as in case of low power experiments, is the transverse pumping
and in case of parallel pumping, the rf field is applied parallel to the dc
magnetic field. Only the later technique is used in this investigation for
determination of instability threshold.
3.1.1 Transverse Pumping Instability

After the works of Damon [35] and Bloenbergen and Wang [36] two principle
non-linear phenomena were established in the usual FMR experiment (rf field
perpendicular to dc magnitude field) at high power. First one is the premature
saturation of the main resonance, i.e. the value of susceptibility, XN , at resonance
declines at a power level far fellow that for saturation. Second effect is often
labelled as subsidiary absorption i.e. the appearance of a subsidiary peak at
values of dc magnetic field below that required for main resonance. Both
effects are shown in Fig. 3. The origin of these peculiar high power effect was
first explained by Suhl [37]. In his theory, Suhl proposed that the parametric
excitation of spin-waves by the uniform mode was the cause of the observed
behaviour which is due to dipole-dipole interaction. At low power levels the
spinwave amplitudes stay essentially at their thermal values because of the
small amplitude of uniform precession. But as the power level increases, the
uniform precession amplitude grows and more and more energy is pumped into
the spin wave modes.
I. SUBSIDIARY ABSORPl10N
Suhl [37] proposed a mechanism for the high power effect which is bound on
the interaction of the uniform mode with spin waves of short wavelengths. In
"first-order Suhl process", where there is an appearance of a subsidiary peak,
=
energy from a single k 0 magnon is transferred to and creates two magnon,
having wave vector, k and - k respectively. This is shown is Fig. 4a.
Due to the conservation of energy principle these magnons must have one-
half the frequency of the k= 0 magnon. One particular pair, of mag non, among
many pairs with Wk = 0J/2 and with equal and opposite k, will be strongly couple
to the k = 0 mode and consequently lead to a nonlinearity in the absorption.
For a ellipsoid the critical field derived by Suh} [37] can be written as
rh . = 2Wk([W~ + (wn - Wk)2] [11~ + (w r - W)2]}1/2

(1)
c;nt WM sin Ok cos Ok (Wit + Wo wMNz + Wit l1kk2)
/
-
K
- ./ /
-
q"'
K=O
I.
~
1'=0
i!".O ./_
/-i!" -K
(a) ( b) (c)
Suhl first order Suhl second order Parallel pump
Fig. 4. Schematic representation of processes involved in the instability thresholds.
where 7]k =r ,1 H k 7]o =r.1H/2 ,and OJ r = OJo (Nz - Nt) OJ m•

It can be seen from the numerator of Eq. 1 that unless OJk =OJ /2 or OJ r =ro
or both, hcri' is as large as ro 2/«>M.. The condition for minimum of the above
equation can be obtained with rok =ro/2, because of momentum conservation,
to explain the nonlinear effects. Two cases can be considered, one in which the
subsidiary absorption occur at a low dc magnetic field than that required for
main resonance and the other when the main resonance and subsidiary absorption
coincide.
Case 1
This corresponds to the experimental observations when rok = fJJ/2 but ror not
close to ro, the subsidiary absorption occurs at an applied field less than that
required for main resonance. The threshold for this case can be written as [37]
h . - 2OJ7]k (roo - ror) (2)

r enl - roM sin 8 k cos 8k (ro/2 + OJo - roMNz )
The minimization of hent be done wiih ~ other than zero or 1C/2. This condition
simply tells that the eligible spinwave must propagate on a cone with an
intermediate value about the direction of propagation i.e. 8 k =45°. Thus the
subsidiary absorption is confined to a range of fields like
roo < (OJ/2 + roMNJ (3)
which is a necessary condition for experiments.
Case 2
Merging of the subsidiary absorption with the main resonance occur for
rok = ro/2 = ror. This condition occurs provided
NtroM > ro/2 (4)
Here the final result for threshold is
h '1 = 2ro 71k 7]0 (5)
en roM(ro/2 + ro - NtroM) sin 8k cos 8 k
This subsidiary absorption refers to a reduction in the susceptibility at the main

resonance.
II. PREMATURE SATURATION OF TIlE MAIN REsONANCE

The premature saturation of the main resonance is often called the "second
order Suhl process", which results from a 4 magnon process. Two uniform
=
precession magnons, k 0, are annihilated and a pair of magnon created having
the same energy and wave vector k and - k, as shown in Fig. 4b.
= =
Under practical conditions liJ liJt liJr is the only situation, for which hcril
is small enough. Though in principle this instability can occur when the driving
field differs from the resonance frequency. The uniform precession is most
=
strongly coupled with spinwave of ~ 0 i.e. the z-directed spinwave propagating
along the direction of the magnetization, have minimum threshold. The final
expression for hcril for a sphere, becomes
heril =dHa . (dHJ41rMs)1/2 (6)
The approximate value of the ratio of susceptibility is given by
n"/z;;= ..JPcritlPinI;; =hcnl/h (7)
The general instability threshold for subsidiary absorption, to a good approxi-

mation, is given by
(8)
The conventional interpretation of subsidiary absorption involves the excitation

=
of spinwaves with wave numbers k 0 propagating at an angle 8t with respect
to the static field, for fields below the butterfly curve minimum. In this field
range herit and k both are taken to increase with decreasing field. Whereas,
above the butterfly curve minimum k is about zero and ~ decreases [68].
Subsidiary absorption results on singal crystal YIG [67] showed a broad flat
minimum below the butterfly curve minimum, rather than the smooth increase
predicted theoretically.
III. PARAU.EL PuMPING

Schlomann et al [42] pointed out that the parametric excitation of spinwaves
and associated instability effects could also result when the microwave field is
parallel to the static field. In many respects the parallel pump threshold is
easier to understand physically than subsidiary absorption. The uniform mode
can also be driven by a longitudinal rf field at twice the uniform precession
frequency, provided the rf amplitude exceeds a critical level and the precession
path is elliptical. Some spin waves precess on ellipsoidal cones, rather than a
circular cone due to their dipolar interaction, and pair of these modes with
equal and opposite wave number can be excited by a sufficiently strong rf field
parallel to the dc field. In this case, the sample will absorb power only if the
rf field exceeds the instability threshold. Direct coupling between the parallel
pump microwave field and certain spin wave modes is possible, because of the
spinwave ellipticity. Figure 5 shows an elliptical cone.
Microwave Ferriles 123
Elliptical precession cone
Fig. 5. Schematic representation of elliptical precession cone for spin-

waves which leads to the parallel pwnp instability.
If the magnitude of M is constant the elliptical orbit results in a wobble in the

z-component denoted by ~mz. A spin wave mode at Wt can couple to the parallel
=
pump microwave field at (J) 2{J)k. The amplitude of ~mz is dependent on the
= =
polar angle 8 k : at 8 k 0, the amplitude of ~m z 0 i.e., the ellipticity is zero
and at 8 k = tr/2, lim" and hence ellipticity is maximum. The amplitude of
=
threshold field for instability, he, in terms of spin wave linewidth L1Hk 2r/Jr
is given by
(9)
where (J) is the operating frequency. From Eq. 9 it is evident that the threshold
=
diverges at ~ 0, because the ellipticity is then zero. The above expression
= =
has to be minimized for {J)k m/2. At small dc fields the line {J)k m/2 intersects
=
the spinwave spectrum for all values of {J)k, the lowest threshold is for 8 k Tr/2.
Hence, for the most susceptible magnons Eq. 9 becomes
(10)
From standard microwave theory the threshold field, hc' incident on the sample
is given by the relation [37, 38]
he - 25QI (1 - r.) (11)

- fogVc
where he is in Oe, Pc is the critical incident power in watts, fo is the magne-

tron operating fJ:equency in GHz, Vc is the volume of the cavity in cm3 , Ql is
the loaded Q of the cavity, rs
is the reflection coefficient related to return
loss by
r. =10- (rdB/20) (12)
g is the cavity geometrical parameter given by

g =(1/8)[1 + (dla)2] (13)

where d and a are respectively the length and the width of the cavity.
As .1H" is dependent on the wave number Ie, the increase in dc field decreases
the wave number and hence decreases he to a minimum. The wave number
becomes zero at a characteristic de internal field strength given by
HiIJI. =- 2trM. + [(2trm.)2 + (co/2r)2]l12 (14)
=
For higher fields there are no 8" tr /2 magnons satisfying the condition
=
co" co/2. Threshold field. he. increases rapidly as sin2 8" decreases. because
of the decrease of the magnitude of limz and hence the ellipticity of precession
cone.
3.2 High Power Microwave Measurements

The power handling capability of the ferrimagnetic materials can be determined
by finding the value of threshold microwave field. hent. at which the loss
characteristics of the materials become nonlinear. Two methods are available
for the determination of the critical microwave field for non-linearity threshold.
They are
1. Observation of deterioration of the trailing edge of the transmitted pulse.
2. Measurement of insertion loss versus incident power level at a given
biasing de field Happ.
The onset of nonlinearity can be observed by noting the distortion of the
trailing edge of the pulse. This method requires a perfect square pulse at low
power level. The CRO trace of the pulses transmitted at various power levels
through TEI02 reflection type microwave cavity containing the sample shows
that the nonlinearity becomes first noticeable at the trailing edge of the pulse.
With the increase .of microwave power a large region of the pulse was affected.
The nonlinearity sets quite abruptly. so the experimental determination of the
threshold field from the eRO traces was ambiguous. Here the threshold is
measured by the second method. The insertion loss of the test ~tion was
plotted against the rf power and the transition from linear to nonlinear state was
observed. The pumping configuration used for this study is the paraDel pumping.
3.2.1 X-Bani High Power Microwave Bench

The experiment can be performed around a reflection cavity high power
spectrometer. A pulsed magnetron (like BEL 4] S2A-l. India). can be used as
a high power source operating at a fixed frequency (like 9.4 GHz). The
instrumentation material research is shown in Fig. 6.
The pulse width and the pulse repetition frequency (PRF) should be chosen
after some important consideration. These are
(1) Pulse width should be larger than the time taken for the instability to
set in the material after the power is incident on it. This is dependent
on the relaxation time of the material which is of the order of few
nanoseconds.
Isolator He,
Magnetron
modulator
Circulator :. 1 /
40dB
fo = 9385 MHz
Peak power = 70 kW
r Mo;it~ri;;g - , - --l Ql=600
Pulse width = 1 }J sec arm
Prec
P.R.F = 120 Hz Attn
,I
,Frequ-
Isolator ency
~ Imeter
I
,I ~
Calib rated ~.
I
crystal
I ~
~
l,ood CaUb rated
crystal
To osc.illoscope , ~
L ________ J To oscilloscope
...i·
Fig. 6. X-Band high power bench: hc:ril measurement in reflection mode. ~
-
126 MICROWAVE MATFJUALS
(ii) Pulse repetition frequency should be chosen in such a way that the duty
cycle should be as low as possible i.e. of the' order of 10-5, to avoid sample
heating.
(iii) Pulse width should be larger than the response time of the cavity.
(iv) The leading and the trailing edges of the magnetron pulses should be
avoided.
A pulse width of 1 J.JS and PRF of 50 or 100 Hz has been found suitable for
all requirements.
A TEI02 rectangular microwave cavity of a low Q, should be used so that
a slight detuning caused by the insertion of the ferrite sample could not change
the reactance' of the cavity.
The RF power provided by the magnetron operating in the pulsed mode,
passes through a high power ferrite isolater into a 10 dB directional coupler.
The unused power of the main arm of the directional coupler is terminated in
a high power load in order to minimize frequency pulling effect due to mismatch.
The desired RF signal is incident on the cavity through a circulator and a
shorting switch, which preceed the cavity. The shorting switch helps to measure
the incident power as well as the reflected power from the cavity which is
monitered with calibrated diode detector connected to a CRO. The precession
variable attenuator which precede the detector is kept at a constant level such
that the crystal output is constant. This ensured that the deviation of crystal
behaviour from ideal square law does not affect the reading. The power absorbed
by the sample as a function of increasing input power level is measured at a
fixed biasing field HIpp. From the graph of the return loss versus input power
level the sharp increase in the cavity reflected power gives the onset of instability.
This procedure is repeated for various values of the applied dc field. The
threshold field for non-linearity (hJ was calculated using the standard relation
given in Eq. 11. These values of he was plotted as a function of Happ and the
curve obtained is usually called as "Butterfly curve". Spin wave line-width
&It is calculated from the ~ (minimum) value USing the relation in Eq. 10.
The crystal IN21 with mount and the isolator gives an output of 200 mV
on the CRO when the power incident is 200 mW. All measurements are
obtained with spherical samples of 2 to 3 mm diameter to permit the applica-
tion of perturbation technique. The sample is placed in a region of minimum
electric field and maximum magnetic field, by a teflon rod sample holder
inside the cavity. Such placement insured small dielectric losses and nearly
uniform microwave magnetic field throughout the sample. The cavity is
continuously tuned to the magnetron frequency at each biasing field by a
tuning screw.
4 MICROWA VE STUDIES
4.1 High Power Microwave Studies
4.1.1 Parallel Pumping Instability Threshold

The resu~ts obtained on parallel pumping technique at high power microwave
Microwave Ferrires 127
levels on Gd3+ substituted YIG and C02+ and AI3+ substituted UTi ferrites are
given. The microwave field is applied parallel to the static field. The method
of measurements is described in section 3.2.
Figure 7 shows the variation of critical fields, hcrit as a function of the static
applied field for some specimen of garnet and ferrites respectively. These
curves are called the "Butterfly Curves". Here the angle (yt) between the pump
field and the static field is zero degree. The well known butterfly shape of the
curves are apparent for all these samples. It is evident from the figure that the
heri! decrease to a minimum. In this field range, the instability corresponds to
Ok = n/2 spinwaves with k values which are gradually decreases to zero at the
minimum, this is the one with lowest relaxation rate. The applied static field
at which the wave number decreases to zero and the butterfly curve has a
minimum for a isotropic material is given in Eq. (14) [42],
2,
Co F.U
90 I- • •
x x
0.000
0.005 •
• fI 0.01
80 I- • •
•• 0.02
0.04
70 fo-
t 60 r-
•
8
QI
••
u
• • o
•
J:.
50 fo-
•
40fo-
•
301-
1 1 1 • 1
1 1 10
1 G
,
1
0 <:)
201- <:)
<:)
<:) <:)
101- x x
x x
<:)
.- •
• • • • • •x 1
x
I
o
-
1000 2000 3000
Happ
Fig. 7 (ConJd.)
'IO~------------------------------------------------------~
GJ+/F.U 411'I0Il5
X 0.0 1800
150
•
t:.
0.1
1.6
1120
860
• 2.2 530
f
~ 10
50
; ;; ;:)
500 1000 1500 Happ (0.) _ 2000 2500
Fig. 7. Variation of-critical fields as a function of static applied field

for some garnets and ferrites [68).
where
(15)
Haw is the applied static dc field, Nz is the longitutional demagenetising factor

and is equal to 1/3 for spheres. There is no change in the positions of the
experimentally observed [68] butterfly curve minima i.e. the minimum positions
of the butterfly curves for the samples of garnets and Co ferrites. But in case
of AI ferrites the butterfly curves have been shifted towards higher static field
than calculated from Eq. (14) by an amount approximately equal to the anisotropic
field (HJ. The shifting of the butterfly minima due to anisotropy of the material
with the addition of Al ions may be attributed in the foltowing way. Under the
influence of crystalline anisotropy the theoretical butterfly curve minimum
should occur at an internal field of [67]
H::- =- t (41fM. + H.) + [ :i- (41fM. + H.)2 + (;;. )

2]112
(16)
and
H.w =(Hullania + H.) - 41fM.Nz
Using Eq. (16) the theoretical butterfly minimum position is calculated and
observed are nearly the experimental value.
Above this minimum position threshold field increases rapidly with the
increase of static field for all the three series of samples [68]. The low level
substituted ferrites or garnets show smaller tendency for dipping at butterfly
minimum before the steep rise. This is in agreement with other published data
[31-33]. The experimental parallel pump data in these figures indicate that
IlHk is indeed Ie dependent. Moreover. increase of relaxing ions (like C02+ or
Gd~ below butterfly curve minima. makes IlHk a strong Ie dependent. Since
co the operating ffequency is fixed. the threshold (hmJ at each value of Happ
corresponds to a different Ie. These data clearly show that the threshold increases
substantially as the C02+ or Gd3+ content increases. The hen.. for maximum C02+
content i.e. 0.04 Co2+ ionS/F.U. is over ten times greater than the threshold for
the zero C02+ content sample. Gd3+ also increase lam, substantially. But the
increase of hen, by AI3+ is the least. Borghese [39] observed the variation of
IlHk for Mg-Mn ferrites with 2% cobalt with grain size. His results show that
IlHk is virtually independent of grain size. In the present investigation the
saturation magnetization is kept constant and the' average grain size is also
fixed for Co and AI ferrites. So the increase of lam, is only due to C02+ or AI3+
in addition to the LiTi ferrites. These results suggest that the relaxation mechanism
of low field loss contains no two magnon contribution. which is the major
effect of the ferromagnetic resonance r~iaxation.
In this calculation 8k is taken as 1f(1. which minimizes the critical field (hJ.
and for a isotropic material for parallel pumping he is given by
(17)
Under the influence of anisotropy the parallel pump threshold is given by

[68]
(18)
The square bracket term (SBT) represents the effect of anisotropy of the parallel
pump coupling. Eq. (17) is applicable to the isotropic materials like Co ferrites
and Gd YIG. but in case of AI ferrites where the crystalline anisotropy has a
large effect Eq. (17) does not hold good. So for these samples he is obtained by
Eq. (18). The experimental butterfly curves for AI ferrites do not show any
double minima. Sethares et al [10] observed double minima butterfly curves
for single YIG at 9.55 GHz. This double minima may be due to the spin-
=
wave switching phenomena. Setheres et al·ensure that only tPk 0 or 1r/2 are to
be selected.
In case of AI ferrites absence of any low field butterfly minimum ensures
=
the instability of spin wave with tPk 1r/2. The coupling of energy between fast
relaxers like Co or Gd through long range 4ipolar interaction is very strong.
Thus a large increase of instability is observed in case of Co ferrites and Qd
YIG. But in case of AI ferrites the coupling due to anisotropic field is not very
strong, thus a moderate increase of he occur.
For high static field, i.e. above butterfly curve minima the range of available
8k values becomes more restricted. At these external fields 8k starts decreasing
from 1r/2 to zero and k remains nearly zero.
The theoretical verification of the experimental data [68] are made by the
following models.
MODEL I
These parallel pump butterfly curves, for samples having large grain size (ao)
and small porosity (P), can be reproduced theoretically by considering the mean-
free-path or the transist-time model given by Vrehen et al [65] and Patton [63].
=
According to this model a generated spin wave having frequency COk coI2 travels
a distance 1 at a group velocity va before being destroyed. l is the mean-free-
path and is related to the grain size ao. It gives a life time for the spinwave,
l' = z/v, (19)
and the mean-free-path related spinwave Iinewidth can be expressed as
I
Nlk =-Y1'1 =-IylVal =-2Dk
l
(20)
where
=2yDk k + y4lrM.
A A
VI sin 8 k cos 8 k 8 k (21)
D is the exchange constant 5 x 10--9 Oe-cm2, ythe gyromagnetic ratio is equal

=
to 1.76 x 10' 0e--1-sec--1and 8 k 1r/2.
Polar component of VB i.e. the 2nd term in Eq. (21) is neglected, thus spin-
wave Iinewidth can be expressed as proportional to the k values.
Vrehen [65] also suggested that the mean-free-path was related to the grain
size of the sample, by observing the inverse grain size dependence of the
spinwave Iinewidth in fine grain samples. The application of this model with
the present experimental data is not suitable, as the samples have a quite larger
average grain size and porosity> 3%. The densification of ferrite is made
through the addition of Ba203, as a sintesing aid at the time of initial mixing
for sample preparation. This is due to the fact that the addition of bismuth
oxide lowers the sintesing temperature to l000-1050°C, where volatility ofLi 20
is minimal.
MODEL II
Scotter [57] has provided another model, which is a refinement of the transit-
time model. According to this model the spinwave linewidth is affected by the
size of the nonmagnetic pores, rather than the grain boundaries. Here the relaxation
has the form
1'- 4R (22)
- 3pvt>
where R is radius of the pore and p the porosity. The basic aim of Schotter's
model is that the pores in ferrites act as the scattering center for the travelling
spinwaves.
MODEL III
Sawado [62] has improved the pore scattering model by applying the screening
effects to the pore dipole potential. This method is used explicitly to explain
the experiment data here. The mean-free-path 1 as obtained by Sawado [62] for
a spinwave at k and 6" is given by
1 =f3!pa(6,,) (23)
where Pis the effective pore volume and is written as
P= 21l~+ 1 Po (24)
The static permeability III observed in this investigation at 10 KHz is 1000.

This value of III is used in calculating a(60. But for the calculation of mean-
free-path 1, the permeability used is the microwave permeability which is nearly
equal to one.
Po is the volume of the pore and is measured from the SEM photographs,
and is equal to (4f3)1CR3 , where R is the pore radius. This model has been used
for calculating the theoretical he values for all the ferrites and garnets. In case
of AI ferrites .effect of anisotropy is also added. The value of R obtained for
the sample is 0.17 to 0.20 microns which is of the same order obtained for YIG
by Sawado. The scattering cross-section of a pore given by Sawado is
a(6,,) =(4/9) (.!!!JJ!.)2

4tcyD
1(80 (25)
where
1(60 = f sin 8" d8" dtP" f 1P2 (cos 8)12 sin 8" (26)
1(6,,) is an integral which sums over all the angles into which the spinwave
may be scattered, 6" is the polar angle of the wave vector k. The spinwave
linewidth is then given by
(27)
Sawado [62] obtained the spinwave linewidth by integrating overall the Bt

values and given the expression for tJlk as
l1Hk =4:rl1 J V.(Ok)aT(O...) sin 0... dO ... d~ ... (28)
As the experimental data given here shows a strong k dependence, the scattering
cross section cannot be obtained by integrating overall allowed values of Bt.
Because an integration overall spinwave angles provide a wave vector-
independent linewidth. So the suggestions made by Silber and Patton [64]
seems to be true for the samples investigated under present study. In this work
tJlk is calculated by using Eq. (27) instead of Eq. (28).
Theoretical butterfly curves for parallel pumping conditions are obtained by
including the wave-vector k dependence of the Iinewidth, which can be expressed
as
(29)
where A and C are two constants. The value of these constants are calculated
in the following way. The values of A are calculated from the he value at the
=
butterfly minimum of the experimental data with 0... 90°, k 0 and at Haw =
around the butterfly curve minimum position, by using Eq. (10). The values of
constant C are obtained from low field side experimental data of the butterfly
curves, where 0 ... corresponds to tr/2 and for nonzero k values.
The most striking feature of the present investigation is the rapid increase
of A values with the addi"tion of Co2+ and Gd3+ contents except AI3+ content
samples. In addition, the parameter C reflects the degree of k dependence of
l1Hk • From the calculated C parameters it is clear that the increase of the relaxing
ions (Co+2 or Gd+3) l1Hk are more and more k-dependent.
The theoretical fitting of the low field experimental data are quite good for
the threshold data below butterfly minimum.
But for higher static field the constant mean-free-path model does not fit the
experimental data. So, the scattering of spinwave at grain boundaries cannot be
the only mechanism which limits the spinwave life time in polycrystalline
ferrites. In this investigation an attempt is made to fit the data with pore
scattering approach given by Silber and Patton [64]. For the external static
field beyond butterfly curve minima l1Hk may be expressed in the form of
(30)
where l1H...{p) is given in Eq. (28). The parameter B is calculated by setting
he equal to the experimental.
The agreement between the high field theoretical and experimental data,
though not quantitatively perfect, are still good. The trend of both the curves
are same. Only the difference is that the theoretical curves diverge more rapidly
than the experimental curves. This may be due to the fact that the onset of
absorption is not sharp for the static fields beyond 2000 Oe.
The results obtained for the samples under present investigations having
large average grain size and porosity> 3% suggest that pore scattering contributes
a large part to the spin wave threshold.
Thus the pore scattering theory originally given by Sawado [62J and latter
on modification by Silber and Patton [64J presents a good theoretical fit to our
experimental data. The present results show that the threshold minimum, he (min)
=
occurs only at 8 k 1C/2 and not at some value of 8 k less than 90°, as given
by scotter [57J. The integral 1(80 appearing in the expression for scattering
cross section 0'(80 shows the 8k dependence of linewidth at higher static field
i.e. above butterfly curve minimum. This integral is not only dependent on the
porosity alone it depends on the size of the pore present in polycrystalline
ferrite. Scanning Electron Microscope photrographs of these samples show that
the inclusion of C02+ ion is in the composition increases the collective presence
of smaller pores. The collective present of smaller pores increase the cross-
sectional area R/r times than a single larger pore, where Rand r are the radius
of the larger and small pores respectively. The increase of cross-section area
has a marked affect on the spin-wave line width.
4.1.2 Wave Vector (k) Dependence of tJH"

The wave vector k, dependence of tJHk are shown in Fig. 8. These curves are
obtained from the parallel pump butterfly curves. The value of k can be evaluated
by using the dispersion relation for an isotropic material (like Co and Gd) and
is expressed as
Ct>f = r[(H + Dk~ , (H + Dk2 + 4nMs sin 2 Ok) (31)

H is the internal field = Haw - Nz' 4(fM, and 8 k = rr/2 for the parallel pump
data.
.. 30
8
'"
:l:
<l 20
• •
10
. . -
•
o~______~~:=====c==~:==~~==~~~:~=:·=-j
1.0. 2.0 3.0 4.0
5
5.ox10
K-
Fig. 8. Wave vector (k) dependance on &fk [68],

For an anisotropic ferrite (AI-ferrite here) the <\ispersion relation is
(O~ = r[(H + H. + Dk + 4nM. sin ~k 8k HH + Dk2 + 4nM. cos ~~ sin 2 80

- (4nM.) sin 4 8 k sin 2 ~k cos2 ~k)] (32)
for ~k =n/2
(0: =r[(H + Dk2) . (H + H: + Dk2 + 41rM, sin 2 8 k )] (33)
The principal effect of anisotropic field on Al ferrite is that (Ok is dependent
on the azimuthal angle. 'k. whereas in isotropic Co ferrite or Gd-YIG (Ok is
independent of ~k. The gradient of tiH" vs k curves which are approximately
straight lines, are observed to increase rapidly with the increase relaxing ions
only. In this investigation the inverse k dependence was not observed. The results
are in agreement with the data observed for large grain samples earlier [60,
61]. The linear dependence of k with tiH" can be seen by visualizng the transit-
time or mean-free-path model. The modulous of group velocity of the spin
wave without the polar component is
Vg =2rDk (34)
1.e. Vg DC k
which implies that,
(35)
Furthermore, the sharp increase of the stope of &1" vs k curves for Co-ferrite
and Gd-YIG may be attributed due to the fact that, the addition of relaxing ions
makes 8" to increase from the unsubstituted value. From Eq. (8). the expression
of Vg the increase of 8" increases the" second term in this equation. Hence,
according to the relation
(36)
there is a sharp increase of tiH" value which can be expected, because the
values of r and I are constant.
The observed spinwave linewidth data can be qualitatively explained with
the idea of a constant mean-free-path model. When this model is applied to the
relaxation due to fast relaxing impurities, the strength of the relaxation interaction
for a particular relaxing ion should be taken into account The spin wave linewidth
may be expressed as
(37)
The values of the constant C are calculated from the hc(min) experimental values
from parallel pumping configuration. From energy and momentum considerations
the term on the right hand side of Eq. (37) is related to three mag non splitting
processes and can only contribute to the parallel pump linewidth. &I,,~ attributed
to k =0 modes at 8" =tr/2. The theoretical data have good agreement with the
experimental data.
The results obtained here for the k dependence of spinwave linewidth is due
to a transist-time phenomena. Such dependence leads to the high power capability
of ferrites, since relaxing ions are more effective at increasing the L1Hk -+o values.
4.1.3 Spinwave Linewidth L111" with the Variation of Co2+ and AI3+
Ferrite and Gd3+ YIG
=
Room temperature X-band spinwave linewidth, for k 0 spinwaves calculated
from parallel pump butterfly by curves data, are shown in Fig. 9. Using standard
20r------------------------,
t
.......
III
0 15
~:.c
X
<I
.&;
~ 10
~
III
-=
III
>
0
~
c:
'Q.
1Il
o 0.01 0.02 0.03 0.04

2+ J.
X(CO Ions/Formula Unit)-
OL-______ ~ ____ ~~ ______ _ L_ _ _ _ _ _~~-J
o 0.1 0.2
Al (Conc.)-
Fig. 9. Room temperature X-band spinwave linewidth, for k = 0 spin waves
calculated from parallel pump butterfly curves data [68].
theories of Suhl [40] and Schlomann [41] the heri' data obtained from parallel
pumping experiment is used to calculate the value of LlHt for k = 0 (8 t = Tr/2)
magnons. The standard relation between the critical microwave field amplitude
for spinwave threshold heri' and spinwave linewidth LlHt, for parallel pumping
is given by
he'nt -- (..!!!....)
(Om
(~)
sin 8f min
(38)
where (0 is the pump frequency, (Om is r4TrMs with rthe gyromagnetic ratio.
Fig'. 9 dipicts that, for C02+ addition from zero to 0.04 ionslF.U. the X-band
spin wave linewidth increases by more than an order of magnitude. LlHt increases
approximately linearly with increase of Gd3+ in YIG. But in case of AI3+ the
increase of LlHk though linear but increase is very slow and also at higher
concenttation of A13+, LlHt appears to attain a saturation value. The sharp increase
in the value of LlHk arises due to the fast relaxation effect of C02+ ions in
octahedral sites, which is based on their crystal field stabilization energy. The
higher charge Ti4+ ions on neighboring octahedral sites and low charge U 1+
ions on adjacent tetrahedral sites help, in the stabilization of C02+ in octahedral
sites. The clustering effect of Ti4+ ions around octahedral sites, if any, is expected
even in the absence of C02+ ions and is not going to affect the material behaviour
on addition of the fast relaxing C02+ ions. The observed data can be explained
by considering the strong relaxation the effect of C02+ on the anisotropy of the
material and the enhancement of spin-lattice relaxation rates. The anisotrophy
cancellation effects attributed to C02+ ions do not appear where Ti 4+ ions are
used as the dilutant in the basic composition of lithium ferrites. C02+ ions in
octahedral sites. provides a large positive contributions to the negative anisotrophy
of the iron sub lattices. Through magnetic superexchange interactions, the C02+
anisotrophy is communicated to the iron sublattices. This caused rotation of the
Fe 3+ spins in the adjacent tetrahedral and octahedral sites. If the C02+ spins are
decoupled from the Fe3+ by the presence of Li 1+ in tetrahedral and Ti4+ in
octahedral sites, then the isolated C02+ anisotropy effects is undetected at these
small concentrations. This give rise to the fact that C02+ now behaves as a
paramagnetic ion, and this coupled to the Fe3+ ions only through dipolar fields.
The exchange isolation cannot prevent the occurrence of increased spin-lattice
relaxation rates, due to the cross relaxation between paramagnetic ions, which
can transfer energy to fast relaxers like C02+ by means of longer range dipolar
interactions. This leads to a sharp increase of spin wave linewidth.
In case of Al ferrite the increase of M1'r. is very small but still appreciable
to use this material in medium power microwave devices. This slow increase
of LlH'r. may be attributed to the coupling of energy to Fe3+ ions via crystalline
anisotropy. The strong point of Al addition over Co in UTi ferrite is that AI3+
does not raise the magnetic losses unlike C02+. So a fine control of aluminium
can make this group of ferrite to best use for medium power microwave devices.
The large increase of LlHt in Gd-YIG over the concentration range studied
may be attributed [68] in the following way. Substitution of Gd3+ in YIG
occupies the sites reserved for nonmagnetic yttrium. For garnets the principal
relaxation mechanism is provided by the rare-earth ions. Gadolinium like other
rare-earth ions relax the magnetisation in a two stage process. In the first stage
energy is coupled from the ferric ion system into the spin system formed by
Gd3+ ions. This energy is then dissipated to the crystal lattice in a second stage.
The exchange interaction provides the energy required for coupling between
the ferric ions and Gd3+ ions. This exchange interaction is an order of magnitude
weaker than the interaction between ferric ions. Gd3+ will behave as a para-
magnetic ion. The cross relaxation between paramagnetic ions channel energy
to fast relaxer through the fluctuating field it produces on the ferric lattice.
Hence there is an increase of tJ.Ht •
4.2 Low Power Microwave Studies

The variation of X band resonance linewidth (tJ.H) with Gd-YIG and Co and
Al ferrites at room temperature are given in Fig. 10. It depicts that gadolinium
substitution has a much smaller effect on tJ.H. In case of Co-ferrite the increase
is slow, but in case of Al ferrite the increase in significant.
Upto 53% of gadolinium content the linewidth has increased to 190 Oe from
125 Oe for 33% Gd3+ content sample. For the highest concentration studied
here (2.2 Gd3+ /F.U.) tJ.H has increased to 420 Oe. The small increase of tJ.H
for low contents may be due to the increase of porosity alone, but for higher
concentration anisotropy broadening begins to contribute much. In higher
concentration range the linewidth can be calculated by the coupling of the
uniform precession with degenerate spinwaves of medium k-values. These
couplings are due to the variation in magnetocrystalline anisotropy due to the
granular structure of the material. The Gd3+ ion contains seven 4!electrons which
500
,....
II
9
:x: 300
<I
200
100.lt-----~
OL-__________- L____________L -_ _ _ _ _ _ _ _ _ _ ~
1.0 3+ 2.0 3.0

Gd /Formula unit-
Fig. 10 (Contd.)
400
t
CII
2 350
:I:
<I
300
0.00 0.01 0.02 0.03

Cobalt/Formula unit -
400
.x
f
.
.....
CD
:I:
<1
300
200~-----4------1-------~-----4--~
0.0 0.05 0.10 0.15 0.20
A\3+/Formula unit - -
Fig. 10. The variation of X-band resonance Iinewidth (LlH) with Gd-YIG
and Co and Al ferrites at room temperatures [68].
combine according to Hund's rule in such a way that the ion has no orbital
angular momentum. The spin orbit interaction is therefore very small and cannot
lead to a rapid relaxation as in the case of the other rare earth ions.
In case of C02+ ferrite the relaxation effect vary according to (4nMsrl. The
slope of the .1H vs C02+ content reflect the lower magnetisation of the samples
and the small increase of .1H with cobalt content is caused by the fast relaxation
mechanism of C02+ ions. The increase of .1/l with AI3+ may be due to poor
microstructure, increasing anisotropic field and also by increasing porosity.
The large value of .111 can be understood in terms of spin wave scattering. Due
to crystalline anisotropy the distribution of resonance field for the randomly
oriented crystallites induces the broadening of linewidth.
REFERENCES
1. J.L. Snoek, Phil. Tech. Rev., 8 (1946) p. 363.

2. L. Neel, Ann. Phys., 3 (1948) p. 137.
3. J.L. Snoek, "New Developments in Ferromagnetic Materials"; (Elsevier Press
Inc., New York 1949).
4. E.W. Gorter, Philips Research Reports, 9 (1954) p. 295.
5. J.B. Goodenough, Magnetism and the Chemical Bond. (Interscience-Wiley,
London, 1962).
6. G. Blasse, Philips Research Reports, Suppl. No.3 (1964).
7. P.B. Braun, Nature, 170 (1952) p. 1123.
8. P.B. Baba, G.M. Argentine, W.E. Coumey, G.F. Dionne and B.H. Temme,
Trans. IEEE 'Mag-6 (1972) p. 83.
9. G.O. White, C.A. Edmendson, R.B. Goldforb and C.E. Patton, J. Appl. Phys.
50 (1978) p. 2381.
10. J. C. Sethares, M.H. Seavey, T.C. Pumhagen and M.R. Stiglitz, J. Appl. Phys.,
35 (1964) p. 898.
11. J.H.E. Griffithis, Nature (London) 158 (1946) p. 670.
12. C. Kittel, Phys. Rev., 73 (1948) p. 155
13. C. Kittel, Phys. Rev., 76 (1949) p. 743.
14. D. Polder, Phys. Rev., 73 (1948) p. 1116.
15. J.H. Van Vleck, Physica, 15 (1949) p. 197.
16. S. Geller and M.A. Gilles, J. Phys. Chem. Solids 3 (1957) p. 30.
17. R.K. Wangshers, Phys. Rev., 93 (1954) p. 68.
18. M. Sparkes, Ferromagnetic Relaxation Theory (McGraw-Hill, New York) 1964.
19. J.H. Van Vleck, Phys. Rev., 78 (1950) p. 266.
20. F. Brown and D. Park, Phys. Rev., 93 (1954) p. 381.
21. H.G. Beljers and D. Polder, Nature 165 (1950) p. 800.
22. W.A. Yager, J.K. Galt, F.R. Merritt and E.A. Wood, Phys. Rev., 80 (1950)
p.744.
23. J.F. Dillon Jr., S. Geschwind and V. Jacoarino, Phys. Rev., 100 (1955) p. 75.
24. P.E. Tannenwald, Phys. Rev., 100 (1955) p. 1713.
25. R.W. Teals, 1. Appl. Phys., Suppl. to 33 (1962) p. 129S.
26. E.A. Turov, "Ferromagnetic Resonance Ch-Vi S.V. Vonsovskii (ed.) (Pergamon
Press Ltd, London, 1966) p.191.
27. W.A. Yager, 1.K. Galt, R.F. Merritt, Phys. Rev., 99 (1955) p. 1203.
28. R.C. Lecraw and E.G. Spencer, 1. Phys. Soc. Japan, 17 suppl B.I, (1962) p. 401.
29. F. Keffer, Phys. Rev., 88, (1952) p. 686.
30. A.M. Clogston, H. Suhl, L.R. Walker and P.W. Andersnon, Phys. Rev., 101
(1956) p. 903. J. Phys. Chem. Solids, 1 (1956) p. 129.
31. E. Schlomann, 1. Phys. Chem. Solids, 6 (1958) p. 242.
32. L.R. Walker, Phys. Rev., 105 (1957) p. 390.
33. Geschwind and Clogston, Phys. Rev., 108 (1957) p. 49.
34. Schlomann, 1. Phys, Chem. Solids, 6 (1958) p. 242.
35. R.W. Damon, Revs. Modem Phys., 25 (1953) p. 239.
36. N. Bloembergen and S. Wang, Phys. Rev., 93 (1954) 72.
37. H. Suhl, 1. Phys. Chem. Solids, 1 (1957) p. 207.
38. H. Suhl, Proc., IRE, 44 (1956) p. 1270.
39. C. Borghese and R. Roveda, J. Phys. Colloque Cl, Tome 32 (1971) p. 150.
40. H. Suhl, 1. Phys., 29 (3) (1958) p. 416.
41. E. Schlomann, J. Phys. Chem. Solids, 6 (1958) p. 242.
42. E. Schlomann, 1.1. Green and U. Milano, J. Appl. Phys. 31 Suppl. (1960)
No. 5 p. 386S.
43. R.C. Lecraw, E.O. Spencer and C.S. Porter, J. Appl. Phys., 29 (3) (1958)
p.326.
44. J.J. Green and E. SchIomann, J. Appl. Phys., 33 (S) No.3 (1962) p. 1358.
45. 1.H. Saunders and J.J. Green, J. Appl. Phys. Suppl., 33 (3), (1963) p. 1372.
46. E. Schlomann, 1.1. Green, J.H. Saunders, IEEE Trans. on Magnetics, MAO-l
(1965) p. 168.
47. 1.R. Samuel Dixon, J. Appl. Phys., 38 (1967) No.3 p. 1417.
48. J.J. Green and C.E. Patton, 1. Appl. Phys., 40 (i) (1969) p. 172.
49. C.E. Patton, 1. Appl. Phys., 40 (7) (1969) p. 2837.
50. C.E. Patton and J.J. Green, IEEE Trans. Mag., MAO-5 (3) (1969) p. 626.
51. C.E. Patton, Proceedings of the ICF, (1970) p. 524.
52. C.E. Patton, 1. Appl. Phys., 41 (1) (1970), p. 431.
53. C.E. Patton, J Appl Phys., 41 (4) (1970) p. 1637.
54. D.O. Scatter, 1. Appl. Phys., 32 (10) 1971) p. 4088.
55. Yu. M. Yakovlev, Yu N. Burdin, Yu R. Shilnikov and T.N. Bushueva, Soviet
Phys-Solid State, 12 (10) (1971) p. 247S).
56. C.E. Patton, IEEE Transactions on magnetics (1972) p. 433.
57. D.O. Scatter, J. Appl. Phys. 43 (11) (1972) p. 4813.
58. 0.0. White, C.E. Patton and C.A. Edmendson, J. Appl, Phys. 50 (3) (1979)
p.2118.
59. C.E. Patton and Wolfgang Jantz, J. Appl., Phys., 50 (11) (1979) p. 7082.
60. C.E. Patton, Phys, Stat. Sol. (b) 92 (1979) p. 171 and 93 (1979) p. 7082.
61. Om Prakash and C.M. Srivastava, Bull. Mater. Sci., 2 (3) (1980) p. 181.
62. Eikichi Sawado, Proceedings of ICF (1980) p. 818.
63. C.E. Patton, Proceeding of ICF (1980) p. 807.
64. L.M. Silber and C.E. Patton, IEEE Trans. Mag., MAO-18 (6) (1982) p. 1630.
65. Q.H.F. Vrehen, H.O. Beljers and 1.0.M. del au, IEEE Trans. Magn. MAO-5
(1969) p. 617.
66. J.J. Green and T. Kohane, SCP and Solid St. Tech. 7 (1964) p. 46.
67. Isidore Baby and Ernst Schlomann, J. Appl. Phy. Suppl. 33 (1962) p. 1377.
68. Bijoy Kumar Kuanr; Ph.D. Thesis, University of Delhi, 1992.
6
Microwave Lithium Ferrites
Pran Kishan
Solid State Physics Laboratory, Delhi, India
1. INTRODUCTION
Among the various magnetic oxide materials used in devices operating at micro-
wave frequency ranges lithium ferrite is presently occupying a prominent position.
Although its potential as rectangular loop material was being utilised for long
in memory devices the exploitation of its attractive feature as a microwave
material was much delayed due to excessive losses observed in early samples.
Many researchers working in the field put up a good deal of effort in the early
70s to obtain lithium ferrites with improved dielectric properties. Preparation
of low loss lithium ferrite was first reported in 1971 by Collins and Brown [1].
Considerable interest has persisted in lithium ferrites since then. With the
establishment of reproducible procedures for the preparation of lithium ferrites,
there has been no looking back and at this juncture lithium ferrites are among
the most wanted materials for the microwave frequency ranges. The family of
lithium ferrites has expanded to such an extent that it now covers a very wide
spectrum of properties for utilisation in microwave devices. The frequency
range of its use extends even to millimeter wave bands.
Braun [2] was first to show that lithium ferrite with chemical formula
Lio.sFe2.S04 is an inverse spinel. The divalent ion in this case being a composite
ion (Li ~~s Fe~:S). The compositional variations made possible by incorporation
of a variety of other metal ions in its formula unit have given rise to lithium
ferrite family as a class of material with many special features [3, 4]. These
features are:
1. Unsubstituted lithium ferrite (Lio.sF~.s04) has the highest Curie tempe-
rature (Tc • 640 0c) among the ferrimagnetic oxide materials. High Tc
signifies temperature stability of its magnetic properties. The substituted
compositions of lithium ferrites also inherit this property from the parent
material. A case of materials belonging to different ferrite families
and a comparison of Tc values for a typical value of magnetisation
(4nMs = 1000 Gauss) is illustrated in Fig. 1.
2. The lithium ferrites possess excellent rectangular loop characteristics.
From its earlier use in memory core industry the present application in
latching microwave devices is a natural extension.
1 UTi fERRITES
2 li TiZn fERR1TES
3 YN. GARNfTS
" Hg Hn Al fERRITES
I
3ID.
-
VI
VI
:::J
rg naa
l!J
VI
l:
5- 1aaa
,'1.,
I
I
lh
• '2ID laD It.. saa nD
CURIE TEMPERATURE (oC) ..
FIg. 1. Saturation magnetisation (4nMs) vs Curie temperature (Tc)
for different microwave ferrites.
3. The saturation magnetisation (4nMs) of unsubstituted lithium ferrite at

room temperature is - 3750 Gauss. It has become possible to vary this
parameter to cover values upto 5000 Gauss by. suitable compositional
adjustments. Before this NiZn ferrite system was the only one available
for such high values of 4nMs for microwave applications. Other properties
of technical importance such as remanence ratio, coercive force, resonance
linewidth etc. are possible to tailor in lithium ferrites by incorporating
appropriate substituents and controlling their concentration in the
composition. A notable feature is a series of low linewidth lithium
ferrites which have replaced high cost garnets in several microwave
applications.
4. Lithium ferrites are known to have a low ratio of magnetostriction to
magnetocrystalline anisotropy which means low stress sensitivity of
remanence. This property is further improved by the compositional
adjustment and is of great importance for microwave latching applications.
5. The dielectric constant (e') of the lithium ferrites with various su~titutions
fall in the range 14 to 20, the value being 15 for the parent member i.e
un substituted lithium ferrite. These values are higher than that of the
other micro-wave ferrite series (Garnets 15, MgMn ferrites 10 to 13 and
Microwave Lithium F'errites 143
NiZn ferrites 9 to 12). High e' values are significant from the view-
point of size and weight considerations of the device.
6. The dielectric loss tangents of lithium ferrites presently available are
comparable with other microwave ferrite series (tan c\ < 0.0005). It has
become possible with the advancements made in the processing methods
and from the compositional considerations.
The above features have contributed in the establishment of lithium ferrites
to their present status in microwave magnetics.
2. MATERIAL PROPERTIES
The choice of the ferrite for a microwave device is decided by the working
frequency, power level of operation and whether it works at resonance, far off
resonance or at remanence. The properties of the ferrite optimised in various
aspects determine the feasibility of the device and its level of performance. In
this connection the following parameters of the material are of technical
importance from the viewpoint of its utilisation in devices.
2.1 Dielectric Properties

It is the high resistivity of a ferrite which makes it a ceramic dielectric. The
microwave frequency electromagnetic waves can penetrate and propagate with
low dielectric loss in insulating materials. The permittivity of the ferrite is
expressed as
e = e' - je"
where e' is the dielectric constant of the ferrite material. It lies in the range 9
to 20 for all ferrites. The dielectric loss tangent given by
tan c\ =e"/e'
has values less than 0.001. For lithium ferrites e' has a high value in the range
14 to 20 and typical dielectric loss factors of 0.00025 are obtained.
The dielectric loss occurs primarily due to electrical conduction in the material.
The conduction by hopping mechanism takes place between Fe2+ and Fe3+ ions
present on equivalent crystallographic sites in the structure of the ferrite. The
desired values of resistivity of ferrite material for low dielectric loss at microwave
frequencies is more than 107 ohm cm. In lithium ferrite, represented as
Li 1+ 3+ 0 2 -
o.s Fe 2.S 4
there are no Fe2+ ions and a low loss factor is expected for it. However the
preparation of the ferrite requires high temperature ftring at 1150-1250 °C for
obtaining dense material. At these temperatures oxygen dissociation and lithia
volatility occurs resulting in formation of non-stoichiometric composition. The
consequence is reduction of Fe3+ ions into Fe2+ ions and high dielectric losses.
In addition the control of impurity ions of higher and lower valence in the
structure and reduction of macroscopic and microscopic heterogeneities (e.g.,
pores, non-magnetic inclusions etc.) are also very essential. Suitable techniques
have been developed to overcome it, namely sintering at lower temperatures
and employment of appropriate substitutions in the composition to inhibit Fe2+

formation. Annealing of sintered samples in oxygen atmosphere has also been
found to be beneficial in some cases.
2.2 Magnetisation
The magnetisation of ferrites arises from the antiparallel alignments of the
magnetisation of the two sublattices. The magnetic moments on each sublattice
are arranged parallel to each other and the difference between the magnetisation
of the two sublattices results in a net magnetic moment. The saturation
magnetisation (4nMs) is defined as the magnetic moment per unit volume when
all its individual megnetic domains are aligned by the application of an external
magnetic field. However the magnetic ordering is disturbed as the temperature
of the material is raised. Magnetisation decreases and ultimately disappears at a
characteristic temperature of the material called its Curie temperature (Tc).
It is the specific chemical formulation of the ferrite which primarily controls
its saturation magnetisation. The process parameters involved in the preparation
of the material also play an important role and for some cases the distribution
of ions on the two sublattices is affected by the firing schedule of the material.
The 4nMs of Lio.5F~504 which is - 3750 Gauss at room temperature can be
varied by substituting appropriate other metal ions in the crystal lattice. It has
become possible to prepare substituted lithium ferrite compositions with 4nMs
value covering a wide range from low value of a few hundreds to 5000 Gauss.
The Curie temperature, 640°C for the un substituted lithium ferrite, however
decreases with the increasing substitution levels.
Several series of lithium ferrites for pticrowave applications are available
with Tc varying from 80 to 6200C depending upon the 4nMs and other JrOperties.
The temperature variation of the magnetic properties of lithium ferrites are
superior to other ferrites due to their inherent property of high Tc. Only some
grades of garnets possess better temperature characteristics but their high cost
and limited range of magnetisation restrict their utility.
2.3 Magnetic Losses

The basic phenomena underlying the propagation of microwave frequency
electromagnetic waves in ferrites is the interaction between the magnetic field
of the incident wave and the magnetisation of the ferrite. Energy extracted from
the field in this process is called the magnetic loss and goes to the lattice as heat.
The microwave losses of magnetic nature are characterised by the imaginary
=
part, J/' of the complex permeability (p. p.' - jp.") of the ferrite which exhibits
a resonance behaviour at the angular frequency given by the relation
(() = rH
=
where r is the gyromagnetic ratio 2.8 MHz/Oe and II is the corresponding
=
total internal field. The magnetic loss factor (tan 8 m p."Ip.') for microwave
ferrites is desired to have low value of less than 0.001.
The total energy loss to the lattice due to magnetic interactions is also charac-
terised by the ferromagnetic resonance linewidth (.111), the half width of the
MicrowaveUthiumFerrites 145
resonance ("") peak, and it includes all magnetic Joss processes. The contributions
to tJl come from spin-lattice relaxation and magnetocrystalline anisotropy.
Presence of non-uniformities in the ferrite medium such as pores, non-magnetic
inclusions, cracks, surface roughness and variation in volume density of magnetic
ions, all tend to increase the tJl value. Suitable substitution of several ions like
ZnZ+ in the chemical formulation of lithium ferrite is also quite effective in
adjusting the tJl of the material. The presence of relaxing ions like eo2+ tend
to increase it Lithium ferrites of low linewidth values have been developed by
the chemical formulation techniques. This has led to the development of lithium
ferrites as low cost substitutes of garnets in several devices.
A number of microwave devices operate at fields away from the resonance
e.g. a latched phase shifter at remanent magnetisation. For such devices off
resonance linewidth is considered as the relevant loss parameter. The linewidth
measured as a function of external magnetic field is expressed as effective
linewidth (tlHea). The measurement of tlHeCf with external magnetic field for
lithium ferrite was first reported by Koelberger et al [6].
l.4 High Power Effect

The increase of incident rf power beyond a critical level causes the loss occurring
in the ferrite to rise dramatically. The corresponding threshold is connected to
the existence of a critical microwave field (hJ. This he value puts a limit on
the power handling capability of a particular microwave device. At high rf
magnetic fields the spin waves are also excited at frf"..quencies away from
resonance resulting in the non-linear loss behaviour of the ferrite. The critical
field is given as
=
where (J) is the incident frequency and co... r4nMs. The factor C has value
of the order of unity. The spinwave resonance linewidth tlHk is a material
parameter and is considered to be a measure of the critical field. For normal
microwave ferrites tlHk is 1 to 2 Oe. Higher values of tlHk for ferrites are obtained
by incorporation of fast relaxing ions in the ferrite composition. However some
increase in low-power magnetic loss accompanies it. Fine grained materials
also exhibit improved spinwave linewidths and better peak power handling
capability without any significant increase in low power loss.
l.S Hysteresis Loop Parameters

A ferrite sample, is made-up of randomly oriented microscopic magnetic domains.
Its magnetisation behaviour on the application of an external magnetic field is
described by hysteresis loop and is characterised by parameters like coercive
force (He) and remanence ratio (R =4nMr/4nMs, defined at drive field of
SHJ. A useful property of the hysteresis loop behaviour of the ferrite is that
its magnetisation can be latched to a desired value without maintaining any
continuous external field. The latched magnetization level can be easily adjusted
at any value between +4nMr and -4nMr by the external field. In a toroidal
sample this is done by driving a pulsed current through a threading wire wound
on the toroid which can produce a field H» He. Latched remanent positions
at any in-between values can be reached by partially switching. the toroid.
Several types of phase shifters and switches are based on the latched magnetisation
operation. The shape of the hysteresis loop plays an important role for such
devices.
Ferrite materials with rectangular loop characteristics (high remanence ratio
and low coercive force) are needed for this type of devices. The lithium ferrite
materials employed in the fabrication of latching devices have R > 0.85 and
He - 1 Oe. The switching coefficient and switching energy are two other
important parameters for such a device and are governed by the hysteresis loop
parameters of the material and the external circuit used in'the device.
Stresses produced in the ferrite during shaping, machining processes and
fabrication, affect the shape of the hysteresis loop and reduce the remanence
ratio through magnetostriction effect as shown in Fig. 2 [5]. Ferrites with low
stress sensitivity of remanence are essential for the microwave latching purposes.
41TMr
R= 417Mr
47TMs
-------------r~~~----------~H
He
-NORMAL
-WITH STRESS
Fig. 2. Hysteresis loop.
This is achieved in lithium ferrites through appropriate chemical fonnulations.

Sometimes thennal annealing of the ferrite samples after machining process is
carried out to release the stresses.
3. CHEMISTRY OF LITHIUM FERRITES

Lithium ferrite (Lio.sF~s04) is an inverse spinel with the cationic distribution
FeA [Lio.sFe1.5] .]104• where A and B repre'sent tetrahedral and octahedral sites
respectively. The lattice constant a .. 8.33 A. Its other properties of basic interest
Microwave Lithium Ferrites 147
are: saturation magnetization, Curie temperature, loss factors and the rectangularity
of the hysteresiS" loop.
The several series of lithium ferrites that have emerged out of the parent
composition have been prepared by substituting a number of other metal ions
in the ferrite composition. For microwave application the compositions that are
prepared are represented by quite complicaled chemical fonnulae. A typical
composition includes several cationic constituents which can be divalent, trivalent
and tetravalent ions, each included for one or more specific purposes. The
beneficial and deleterious effects of some of these ions has been discussed by
Baba et al [3]. Each ion has a different role to play in controlling the properties
of the material. The resulting composition is derived keeping in view the desired
properties of the product. This molecular engineering for lithium ferrites and
its bearing on the properties is discussed briefly below.
3.1 AluminiumlTitanium Substitutions

The 4nMs of a ferrite is generally varied by incorporating non-magnetic ions
for the magnetic ions in the fonnula unit. The substitution at octahedral site
leads to reduction in 4nMs values. Aluminium seems to be an obvious choice
for this purpose as it is often used in magnesium ferrites and garnets. The
chemical fonnula of the ferrite becomes Lio.sAlxFe2.s_~,o4. However several
problems were encountered by the earlier workers in preparing the ferrite and
the results were not satisfactory. Higher losses were observed and difficulties
in controlling porosity and microstructure were experienced due to higher sintering
temperatures required for synthesising the materials with the desired high densities.
Dionne [7] and Yakovlev et al [8] found that in addition to reduction in
magnetization, AI substitution results in increase of anisotropy. It has been
observed that in small contents it has some beneficial effect on the power
handling capability of the material. The effect of AI substitution in LiTi ferrites
on spin wave resonance Iinewidth is discussed in Section 6.
Titanium substitutions were found to be much more attractive as it eliminated
the problems experienced with AI. Improvements in sintering, control of porosity
and microstructure and lowering of anisotropy with Ti content up to an appreciable
level resulted in reduction of resonance Iinewidth and minimal effect on coercive
force was obtained. However being a tetravalent ion its inclusion in the fonnula
unit disturbs both the Li and Fe concentration
Lio.s+I/2Ti, F~.s -31/204

Ti ions occupy the octahedral sites and the cation distribution is described as
[9]
Li,/2Fel-I/2[Lio.sTi, Feu -I ]04
The detailed analyses have indicated that a small amount of Ti on tetrahedral
site cannot be ruled out. The titanium substituted lithium ferrites have a higher
dielectric constant (e') and values upto 20 have been obtained for Ti rich
compositions. It is apparent that the higher Ti concentration produces ferrites
of lower 4nMs and Tc values.
3.2 Zinc Substitution

Zinc ions are often included in the composition for enhancing the 4trMs values
in ferrites as Zn enters predominantly the tetrahedral site. The formula of
lithium ferrite with Zn substitution is represented by
Lio.H12 Zn,F~.S-z/2 0"
and the cation distribution is described by
Zn,Fel-l lLio.5-I/lFe1.S+IIiIO"
For higher z values (> 0.4) this substitution weakens the exchange interaction
and does not result in any further increase in 4trMs, on the other hand a fall
in magnetisation is observed. Similar behaviour of magnetisation is observed
in all spinel ferrites on zinc substitution.
Zn substitution is also very effective in controlling several other properties.
It promotes grain growth and densification during sintering and lowers the
anisotropy. All this results in decrease in resonance linewidth and coercive
force. Lower remanence ratio values are also obtained simultaneously [10, 11].
Resonance line widths of around 25 Oe have been achieved [12].
3.3 Simultaneous Zn and Ti Substitution

In lithium ferrites the practice being adopted is to include both Zn and Ti ions
simultaneously in the composition
Lio.5+I/2-c12Ti,zn,F~5-31~"
Studies on the effect of Zn and Ti substitutions on the static magnetic properties
in lithium ferrites have been carried out by several authors [5, 13-19-]. For a
typical LiZnTi ferrite composition the distribution of cations over the two
sublattices is represented as
forz>t
and
forz<t
The combined substitutional effect on material properties for Zn and Ti together

is consistent with the individual characteristics ofLiZn and LiTi ferrites. However
the simultaneous incorpOration of both non-magnetic Zn and Ti ions in the
composition brings down the Curie temperature appreciably, while it becomes
less ~ffective in lowering the 4trMs values compared to Ti substitution alone.
3.4 Manganese Substitution

A very essential requirement for microwave ferrites is their low conductivity
and dielectric loss tangent (tan Sc). Mn substitution is a well established practice
being adopted for inhibiting the formation of any Fe2+ ions. The role of Mn
ions is described by the well known buffering reaction
Mn 3+ + Fe2+ -+ Fe3+ + Mn 2+
+-
The reaction is favoured in the forward direction.
Another notable effect of Mn ions is in lowering the stress sensitivity of the
material. The magnetostriction constant (l) for Lio.sF~s04 is - 8 x 10~ [20].
It was observed long back by Baltzer [21] that the system (1 -x) Lio.sF~s04·
xLio.sMn2.s04 is said to have its saturation magnetostriction approach zero
value at x =0.12. Higher value of remanence ratio was also observed in the
neighbourhood of this composition. Like other non-magnetic ions Mn also
decreases the Curie temperature.
The Mn substitution is necessary in lithium ferrites particularly for latching
applications where low magnetrostriction is needed. Material becomes less
sensitive to stresses, which are introduced in the machining process of ferrite
parts by mechanical pressure in device fabrication and by thermal effects, by
Mn substitution in the ferrite composition.
3.S Nickel Substitution

A small amount of Ni is also often included in the composition as it has been
found useful in enhancing the remanence ratio. It is attributed to lowering of
magnetostriction [22].
3.6 Cobalt Substitution

The loss of microwave power in ferrite devices is found to increase non-
linearly beyond some critical incident power level. For high power devices the
ferrites capable of handling desired power levels are designed. For this purpose
the incorporation of divalent Co ions in very small content in the material has
proved very effective. Inclusion of Zn in the composition in the presence of Co
doped lithium ferrite is also found to increase tlHk • Zn alone does not have any
noticeable effect on this parameter [23]. An increase in tlHk value is, in general,
always accompanied by an increase in magnetic losses.
As a consequence of the above discussion the substitution of the several
cations in the composition leads to a chemical formula like
Lio.5+{I-I...JH:)12 ZIlzTi, Mn/lNihCocF~S-{a+31+2a+b+c)l204
The level of substitution of various ions is chosen with the aim to obtain the
material with desired properties. In practice the value of these cationic contents
is varied in limited ranges only e.g. z S 0.35, t S 1.0, a S 0.1 b S 0.1 and
C S 0.02.
Some work has been reported on mixed lithium-magnesium ferrites.
Individually both lithium ferrites and magnesium ferrites are considered to be
good microwave materials having rectangular hysteresis loop properties. Mixed
Li-Mg ferrite compositions containing Ti have been studied for their use in
latching applications. Typical compositions are described by Wang et al [24].
It is also reported that these materials have found applications in a number of
microwave devices.
The addition of Bi z0 3 which essentially acts as a sintering aid is discussed

separately in Section 4.
4. MATERIAL PREPARATION
Lithium ferrites were earlier finding application in memory cores due to its
rectangular hysteresis loop properties. Their wider use particularly for microwave
devices was restricted due to the difficulties experienced in sintering the material
at the high temperatures employed to achieve high densities in stoichiometric
form. The irreversible loss of lithia [25] and oxygen during sintering was the
main cause that made lithium ferrites technologically difficult to prepare. Low
resistivities of LiZn ferrites, sintered at 1 175°C, were observed even in the
case when firing was carried out in oxygen atmosphere [11]. This was attributed
to the loss of oxygen and lithia occurring at the high sintering temperatures
resulting in the reduction of Fe3+ to Fez+.
The above said hindrances in making lithium ferrites useful for microwave
frequencies were overcome by employing sintering aids like bismuth oxide,
vanadium pentoxide niobium oxide etc. [26]. The first two are low melting
materials and are effective by liquid phase sintering process. The addition of
NbzOs enhances the sintering by increasing bulk diffusion occurring due to
increased vacancy concentration caused by solubility of Nb5+ in the ferrites [27].
Bi z0 3 has been found to have an edge over others and has been adopted in
manufacturing of lithium ferrites for microwave applications. The addition of
Bi z0 3 in small amounts has been reported by several authors [28, 29, 2] for the
preparation of dense materials of both pure and substituted lithium ferrite
composition. This aspect of the lithium ferrite preparation, which has played
an important role in bringing up these materials to the present level, is discussed
below.
A lithium ferrite composition is prepared by the conventional ceramic
procedures. A small amount of Bi z0 3 is mixed either at the starting stage or
after the presintering. This makes the material sinterable at comparatively low
temperatutes (950-1050°C) where the loss of lithium and oxygen are minimal.
There have been speculations in the literature regarding exactly where and
in what form Biz~ locates in the polycrystalline ferrite bodies. Baba et al [2]
incorporated Bi3+ in the formula unit of the ferrite replacing Fe3+ ions. Similarly
Peshev and Pacheva [30, 31] included Bi3+ as a part of the formula unit although
they considered it merely illustrative of the composition. Simonet and Rermosin
[32] and Green et al [33] considered small quantities of Biz0 3 as a simple insoluble
additions segregating mainly at the ferrite grain boundaries. Microstructural
and thermal analyses carried out have confirmed this and have given a clear
picture of the role played by Biz0 3 [29, 30, 31].
In the chemical formula unit containing several substituents, like the one
given in Section 3, when a small amount of Biz0 3 is added, density upto 99%
of the X-ray density is achieved. Biz0 3 content of 0.25 to 1.0 % by weight has
been found to be most appropriate. Riger amount of Biz0 3 interferes with the
crystaUisation process and hampers grain growth. It leads to lowering of densities
and the material properties are adversely affected. Smaller amounts are quite
effective in lowering the sintering temperature and also improve several of

the properties of interest-.from the point of view of its utility in microwave
devices. Fig. 3 shows the properties of technical importance as a function of
Bi2~ content for a typical composition of microwave lithium ferrite. It is
seen that density, average grain size, remanence ratio, coercive force and
resonance linewidth are all affected by bismuth concentration. Addition of
t
-..
Ct
Ct
99
c:a
!
.....
97
c:a
-
95
f!
.3-
""
:1 ~~b:
a:
(b)
~
I:J
:>
~
(el
...t
UI
31D
-
0
:z:: lID
<I 180
5 ld)
'-'I
I
It
t
-
III
g
3
1
...5
....
...
::c 1
a:
0 to 2.0
BilO] CONTENT (Wt Y. J_

Fig. 3. Effect of bismuth oxide concentration on material properties.
IS2 MICROWAVE MATERIALS
Bi20 3 as a sintering aid also improves the mechanical properties of the sintered
ferrite.
The thermal analysis of Bi2~ rich mixtures with lithium ferrite shows that
it melts eutectically at around 770°C while the melting point of Bi20 3 itself is
820°C. The microstructural studies by electron microprobe technique have
depicted the grain boundary composition to be rich in bismuth while it does not
appear in the grains. It is inferred that eutectically melting composition of
Bi2~ rich ferrite is responsible for lowering the sintering temperature and
promoting grain growth by the liquid phase sintering process.
Presently addition of a small amount of Bi20 3 at the starting stage of the
lithium ferrite preparation is being adopted on a routine basis. For certain cases
such as low linewidth ferrites the concentration of Bi2~ is kept at very low
levels as it also acts as a unwanted non-magnetic impurity and proves deleterious
for this property. All lithium ferrite compositions containing Bi20 3 are calcined
at temperatures not exceeding 800°C followed by comminution and pressing.
The material gets sintered to high densities in the temperature range 1000 to
lOSO°C with well developed grain structure.
The other aspect in the material preparation of the ferrites is the microstructure
control. Several properties of the ferrites namely coercive force, remanent ratio
and resonance linewidth depend to some extent on the grain structure of the
material. Grain size effects on microwave ferrite magnetic properties have
been described by Inui and Ogasawara [3S]. Several new techniques are now
being used in the ferrite preparation to achieve the desired control on the
microstructure dependent parameters. For powder preparation besides conventional
methods other processes like co-precipitation, cryochemical and freeze drying,
solution spray draying, fluid bed etc. are used in practice. The primary aim in
all these methods is to prepare thoroughly dispersed, uniform and homogeneous
powders for carrying out most complete reaction to form the ferrite.
The second basic step in the preparation of polycrystalline ferrites is
densification. The general approach is to compact the powder in the required
shapes and f1l'e at high temperatures in a suitable atmosphere. The firing schedule
is accordingly adjusted for achieving optimal material properties. For lithium
ferrites containing Bi2~ as a sintering aid the firing is generally carried out
in air. For some special purposes oxygen atmosphere is also used.
The techniques of isostatic pressing, hot pressing, isostatic hot pressing etc.
are used for preparation of some special grades of materials like high power
ferrites. These methods give a better control over density and size of the crystal
grains.
5. TYPES OF LITHIUM FERRITES

As discussed in Section 3 a large variety of compositions of substituted lithium
ferrites has become possible and its properties are varied over very wide ranges.
For diverse microwave components operating at different frequency ranges
there are several series of lithium ferrites available to the design engineer to
chose from. The tailoring of material properties in lithium ferrites covers far
more sets of combination of properties than is possible in any other ferrimagnetic
oxide family. A number of manufacturers are now offering several different

types of lithium ferrites. These types are based 0JIl the substituents employed
and each one has a different set of properties with some special features.
The different grades of lithium ferrite material are broadly divided into the
following categories.
5.1 Rectangular Loop Type

There are two series of lithium ferrite material which fall under this type. The
major difference between them is the temperature stability of magnetic properties.
On the basis of chemical compositions these are classified as (a) Lithium
titanium ferrites and (b) Lithium titanium zinc ferrites
The LiTi series is characterised by high Curie temperatures and different
grades with 4nMs range upto 3500 Gauss are produced. All the members of the
series exhibit rectangular hysteresis loop with moderately high resonance
linewidths. These materials are employed in latching devices requiring high
temperature stability of performance.
The LiTiZn ferrite family has a wider range of 41rMs. The higher range i.e.
4000 to 5000 Gauss materials are essentially LiZn ferrites while the lower part
of series (41rMs< 3500 Gauss) are LiTi with some low level Zn content.'The
addition of Zn to ferrites lowers its anisotropy and the members of this series
have moderately low resonance linewidths. For obvious reasons the Tc of this
series of ferrites is lower than of the corresponding LiTi ferrite series. The Tc
value is highest (5000C) for the grade of material with 41rMs around 3500 Gauss
and decreases for both higher and lower 41rMs grades. These materials are also
used in latching devices and perform with a better loss characteristics.
Both LiTi and LiTiZn ferrites have low value of spin wave resonance linewidth
and are suitable for low power devices.
Figure 4 gives the temperature variation of magnetisation of some of the
members of the two series and bring out the difference in temperature stability
in the range of interest for the device engineer.
5.2 High Power Lithium Ferrites

This series is derived from the LiTi and LiTiZn ferrite series modified by
inclusion of some fast relaxing metal ions in the chemical composition.
The spin wave linewidth is increased in this way but the accompanying increase
in the magnetic loss factor and to some extent of coercive force cannot be
prevented. The high remanence ratio of the parent LiTi series is retained to a
large extent. High power lithium ferrites with .1Hk upto 7 Oe with compatible
loss characteristics are available.
These materials are specially suited for latching devices operating at high
microwave peak power levels. ,
In Section 6, the high power lithium ferrites are described in a little more
detail.
5.3 Low Linewidth Lithium Ferrites

The low resonance linewidth of lithium ferrite is obtained by l~wering the
- UTi FERRITE TYPE Ia

_____ U TiZn FERRITE TYPE I b
3000
t
2000
1000
o
-50 50 150 250
TEMPERATURE (Oe)
---
Fig. 4. Temperature variation Qf saturation magnetisation (4nMs)
for various grades of type Ia and lb.
anisotropy of LiTiZn fenites by appropriately enhancing the Zn concentration.

For 4trMs range upto 3000 Gauss. materials with t1H < 100 Oe have been
prepared and are commercially available. Very low t1H (25 Oe) has been achieved
in low 4trMs materials [12]. The materials of this series are no longer of
rectangular loop type and the curie temperatures are also quite low compared
to the other types discussed above.
This set of lithium ferrltes are suited for devices like circulators and isolators
and have been established as equivalent to garnet materials for use in the
fabrication of a number of microwave devices. Unlike garnets. lithium fenites
are easy to prepare and do not contain any costly ingredients.
A brief summary of the properties of the various types of lithium ferrites is
given in Table 1.
6. FOR HIGH POWER DEVICES

For enhancing the high power handling capability of ferrites generally the fast
relaxing divalent cobalt ions are substituted in the ferrite composition. The
C02+ ions occupy the octahedral sites and cause strong spin lattice interactions
arising from its unquenched orbital angular momentum. These ions also contribute
Table 1. Types of microwave lithium ferrites
Type 4nMs Tc (0C) E' M-l (Oe) M-lk (Oe) tan eSc X 10-4 tan c5u. X 10-4 R He (Oe)
(Gauss)
LiTi 800-3500 3()()....{i20 15-19 350-500 2 <5 <5 > 0.9 1-2
LiTiZn 250-3000 120-500 16-20 100-300 2 <5 <5 > 0.85 0.6-1.5
4000-5000 520-400 15 200-400 2 < 10 > 0.9 2-4 ~
High Power 600--2000 200-450 17-19 300-700 S7 <5 <10 > 0.85 1-2 ~.
Low Linewidth 800-3000 150-350 15-19 < 100 2 <5 S 10
~
~
t"'"'
§:
~.
~
~.
...,
VI
VI
-
strong positive anisotropy. The spin wave resonance linewidth (t1Hk) of the
ferrite is found to increase by almost an order with extent of the substitution
of C02+ as small as 0.05 ions/formula unit (FU) in the chemical composition
(Fig. 5).
25
-..
I I I I
Cii
0
lITi FERRITE
:
<I
20
•
~
4 lr'Hs .22506 (38)
4rr,.... 125116 (39) )(
i5 +
sa 15 - •
~ + •
""~
:.
a;
a..
111 ..
5
,. • •
.."
It
i
.11 I I I I
II 11.11 11.112 11.\13 11.114 US
Cos. IONS IF U _
Fig. 5. Variation of spinwave resonance linewidth (LiHt ) with cobalt content
Bannerjee et al [36], Brower and. Patton [37] and later Dionne [38] have
observed an apparent paradox in the properties of C02+ substituted lithium ferrites.
The presence of C02+ in the ferrite has simultaneously two types of effects
namely through anisotropy and relaxation. An anisotropy cancellation is expected
at a particular cobalt concentration, and a minimum in 1¥sonance linewidth
should be observed. The hysteresis loop properties are also supposed to reflect
this anisotropy effect. These authors have reported that such a situation is not
observed in titanium substituted lithium ferrites. A new model of cation clustering
by C02+ has been put forward for understanding the observed behaviour of
anisotropy effects [38].
Figure 5 shows that C02+ concentration of 0.01 ions per formula unit increases
t1Hk by more than two fold [38, 39]. A simultaneous increase of magnetic
losses also occurs. For devices working at moderate power levels, it would be
desirable to introduce less than 0.01 ions/FU (C02+) for synthesising the optimum
material. The control of cobalt content in these small concentrations on
reproducible basis in the preparation of ferrites is found to be rather difficult,
t1Hk values being extremely sensitive to C02+ concentration. In the recent work
[40,41] an alternate course of aluminium substitution in LiTi ferrites has been
suggested. AI3+ content upto 0.2 ions per formula unit have been used and
results obtained on several properties like spin wave resonance linewidth,
resonance linewidth, remanence ratio and coercive force are given. Fig. 6 depicts
4.. 0
o
I .,
2.0
1.0
D 0.1 0.2
All .. IONS / FU •
Fig. 6. Variation of spinwave resonance linewidth (tlHk ) with aluminium content.
the variation of ,1Hk with the AI concentration, the value increasing in a controlled
manner from 1.8 to 3.2 Oe for Al concentration varying from 0 to 0.2 ions/FU.
It is also noticed that there is a tendency of ,1Hk to attain a saturated value for
higher Al concentrations. In comparison the ferrite containing Co shows
monotonic rise in ,1H". The effect of aluminium concentrations on 41rMs and
Tc is rather large and has to be taken into consideration while arriving at the
appropriate composition by suitably adjusting the other cationic contents like
titanium.
The aluminium and cobalt substituted lithium ferrites have been evaluated
in reciprocal phase shifter configuration at moderately high peak power levels.
Fig. 7 shows the experimentally observed results [41]. For a composition free
of both A13+ and C02+ the onset of nonlinear effect occurs at around 100 Watt
peak power while it is higher than 300 W for aluminium containing ferrite
(0.1 ionslFU). The C02+ (0.004 ionslFU) is also effective in raising the threshold
power level but it is accompanied by some increase in insertion loss.
7 SINGLE CRYSTALS AND FILMS

Single crystal garnets have found extensive use in the microwave components
such as tunable oscillators, filters, limiters etc. YIG and its substituted
compositions grown mostly by solution growth methods are generally used in
the form of small spheres. Other single crystal materials like lithium ferrites,
having higher magnetisation values and technologically important for high
microwave frequencies, have not matched the quality of the single crystal
YIGs. Schloemann and Blight [42] have made use of lithium ferrite single
crystals for enhancing the usable bandwidths of stripline circulators.
The epitaxially grown films of ferrites are already finding applications in
1. Undopd
VI
00
2. Al : 0.1
-
3. (Q: o.OH
FREQ. 5·46Hz
2
r J
I
!iJ 0.2
&on
g
r
z
5 8.1
~
~
:>
------- -!.
~ 0 ----
:5
~ Rderente
ao::
100 200 300 400

PEAK POWER (WATTI~
Fig. 7. Peak power threshold results.

magnetostatic wave and bubble memory devices. But here also it is the YIG
material films, grown by liquid phase epitaxy (LPE) methods, which have
dominated the scene. It has been established that thin film approaches are
highly attractive for microwave and millimeter wave devices but the problems
cOnnected with producing films of required thickness and quality have restricted
the desired advancement. The motivation for thin film ferrite devices is also
provided by cost, size and weight considerations. There are several thin ferrite
film microwave devices reported in literature having potential for competing
with single crystal devices [42, 43]. The different aspects of ferrite thin films
are described by_ Glass [44].
The growth of lithium ferrite and lithium ferrite-aluminate films by LPE has
been investigated by Glass et al [45] and Van der Straten et at [46] respectively.
Interest in ferrite films has also extended to polycrystalline materials. Thick
films deposited from pastes containing lithium ferrite have been evaluated in
microstrip edge guided isolators and circulators [47]. Other techniques attempted
include ferrite plating [48] and arc plasma spray. Millimeter wave ferrite phase
shifters have been demonstrated by employing the latter method [49]. Experiments
have been carried out on deposition of ferrite films on GaAs substrates [50].
The rapidly growing field of monolithic microwave integrated circuits
(MMICs) lias created renewed interests in development of ferrite film devices.
From the viewpoint of compatibility it is desired that both ferrite and
semiconductro devices are fabricated on the same chip for realizing monolithic
integration. In this connection the three possible approaches for deposition are-
ferrite film on semiconductor substrate, semiconductor film on ferrite substrate
or both semiconductor and ferrite films on a mutually compatible substrate. As
mentioned above attempts have been initiated for deposition of ferrite films on
GaAs substrates, but the results are not very satisfactory. Other approaches are
also in a stage of infancy.
For deposition of single crystal films lattice match between the substrate
and film material is a basic requirement. The lattice constant of ferrite spinels
is - 8.4 A and of III-V compounds is - 5.9 A. The mismatch of lattice is quite
obvious. Schloemann [50] has pointed out that the lattice can match along a
[100] interface with the two lattices rotated by 45 0 with respect to each other.
In such a case the ratio of the two lattices should be "2 for a perfect match.
This requirement is coincidentally met by lithium ferrite (8.33 A) and GaAs
(5.64 A)/lnP (5.87 A), the ratio being quite close to "2. It has been- further
suggested that any residual mis~h of a few percent between the two lattices
can be absorbed by the deposition of suitable buffer layers. This approach is
expected to provide some interesting results for MMIC compatible ferrite
components.
8. FOR PHASE SHIFTER APPLICATIONS

Microwave ferrite phase shifters are extensively used for switching and control
of microwave signals in radar and communication systems. Their major field
of application is phased array antennas for electronic scanning at frequencies
above 3 GHz. These devices provide greater precision of operation than
semiconductor devices and simultaneously provide lower losses and higher

power ratings. However greater size, higher weight and lower switching time
have to be tolerated in ferrite based devices. A single axis scanning antenna
is made up of tens to hundreads of phase-shifters while a two axes antenna
comprises of a few thousand elements.
The three types of ferrite phase shifters which are considered suitable for
use in phased array antennas are:
1. Non-reciprocal toroidal latching type in rectangular wave-guide [52-54].
It finds use in wideband, fast switching applications.
2. ReciprQCai dual mode type in circular or rectangular wave-guide [55-56].
It is considered ideal for high frequency (above 5 GHz), short range
two axes scanning radar.
3. Rotary field type in circular waveguide [57]. This phase shifter is suited
for low side lobe, single axis scanning radar and is characterised by
high accuracy and high rf power handling capability.
The first two types i.e. toroidal and dual mode are currently being employed
in phased arrays. Both operate with remanent flux in the phasor section. The
different remanent states are obtained by the application of a special driver
circuit. An interface is used between the system and the device for converting
the digital command into current/voltage pulse which drives it to different
states. The phase changes are obtained by resetting to the reference limit point
B on the hysteresis loop (Fig. 2) through the saturation point A by applying a
large negative pulse. It is further driven up the major loop to a set point C for
the desired phase state.
The performance of a phase shifter is generally characterised by parameters
such as phase-shift range, insertion loss, frequency range, switching time,
switching energy, phase accuracy, peak power rating and temperature stability
besides cost, size and weight. The properties of the ferrite material used has a
bearing on most of these parameters. Different types of ferrite materials employed
in fabrication of these phase shifters are, in general, required to have rectangular
hysteresis loop with low coercive force and low loss factors. Magnesium-
manganese ferrites are often used but suffer from lack of temperature stability
and low rfpeak power handling capability. The polycrystalline garnets generally
do not have high remanence ratio but with suitably substituted compositions
good temperature stability and high power performance can be obtained. Lithium
ferrites with the versatility of tailoring the properties over wide ranges by
suitable compositional variations have emerged as the most suitable materials
for these phase-shifter applications. The important features of the lithium ferrites
responsible for this are (i) upper limit of 4n-Ms value upto 5000 Gauss which
makes them particularly useful for higher frequency ranges above the X-band
region (ii) the high dielectric constant (e') values, (iii) better temperature stability
due to high Curie temperatures, (iv) possible tailoring of properties for
enhancement of peak power handling capability and (v) low cost. Some of
these features have been described earlier in Section 1.
The LiTi ferrites of type Ia (Section 5) are more often used due to their
better temperature stability. Fa phase shifter with higher peak power specifications
lithium ferrites of type II are preferred. Based on the design--eonsideration in
a C-band phase-shifter lithium ferrite of 41t'Ms 1200 Gauss is used while an
X-band phase shifter employs material with 41t'Ms around 2000 Gauss. As a
thumb rule the magn~tisation of a material selected for the phasor is about
one-fifth of the centre frequency of the device consistant with the relation
4KMs/m« 1. In high frequency phase-shifters operating at 35 GHz and above
ferrites of 5000 Gauss are used. limited by the highest value of magnetisation
available among the ferrileS.
The shaping and machining of the ferrite in the various forms like toroids.
rods. yokes etc. for phase shifters is an important step in the use of ferrites for
this application. The production methods have been established for uniform
reproducibility alongwith good mechanical properties in the various configurations
of lithium ferrite materials.
The applications of both reciprocal and non-reciprocal phase shifters have
been extended as control elements in variable power dividers [57. 58]. Further
advancement of these devices fa antennas having polarization agility development
of a simultaneous dual polarization phase shifter has been reported [59].
The toroidal non-reciprocal and dual mode reciprocal~ largely based on
lithium ferrites. presently represent the major ferrite device market. A number
of phased array antenna programmes (e.g. PA1RIOT and AEGIS) have boosted
the production of ferrite phasors [60].
REFERENCES
1. T. Collins and A.E. Brown. 1. App. Phys.• 42. 3451. 1971.

2. P.O. Baba. G.M. Argentina. W.E. Courtney. G.F. Dionne and D.H. Temme.
Trans. IEEE. MAG-8. 83. 1972.
3. G.F. Dionne Proc. IEEE. 63. 777.1985.
4. P.B. Braun. Nature. 170. 1123. 1952.
5. G.O. White. C.A. Edmondson. R.B. Goldfarb and C.B. Patton. 1. Appl. Phys.,
SO. 238. 1978.
6. R.E. Koelberger. G.O. White and C.B. Patton, 1. PhYI. (Paris). 38 (Suppl. 4),
Cl-255. 1977.
7. G.F.Dionne, 1. App. Phys., 40. 4486, 1969.
8. Y.M. Yakovlev. E.V. Rubalskaya and B.L. Lapovok. SOy. Phys.-Sol. State,
10, 2301. 1969.
9. G. Blassc. Phil. Rei. Rep. (Suppl.), 8, 98. 1964.
10. R.G. West and A.C. Blankenship, 1. Am. Cer. Soc., 50.347,1967.
11. I.V. Young and 1. Smit,I. Appl. Phys., 42, 2344, 1971.
12. M. Sun and Y. Xiu. Trans. IEEE, MAG-lei, 672,1980.
13. G.F. Dionne, 1. Appl. Phys., 45,3621, 1974.
14. S.M. Zhilyakov, V.V. Ivolga, V.1. Mal 'tsev and E.P. Naiden, Sov. Phys.-Sol.
State, 19, 1817, 1977.
15. l.L. Dorman, T. Merceron and M. Nogues, Proc. Int. Conf. On Mossbauer
Effect (Iaipur, India) 193, 1981.
16. C.E. Patton, C.A. Edmondson and Y.H. Liu, 1. App. Phys., 53, 2431, 1982.
17. M. Rosenberg, P. Deppey, S. Dey, U. Jansen, C.E. Patton and C.A. Edmondson,
Trans. IEEE, MAG-18, 1616, 1982.
18. Pran Kishan and K.K. Laroia, Adv. in Cer., 15,299, 1985.
19. S. Phanjoubam, D. Kothari, J.S. Baijal and Pran Kishan, Sol. State. Commn.,
68, 549, 1988.
20. J~ Smit and H.P. Wijn, Ferrites, Wiley NY, 1959.
21. P.K. Baltzer, US Patent 3034987, 1962.
22. R.G. West, J. App. Phys., 34, 1113, 1963.
23. M.P. Bunina, E.V. Lebodera, Al. Pilshchikov and I.I. Silvestrobich, Sov. Phys.-
Sol. State 12, 341, 1970.
24. H. Wang et al Acta Eleetronica Simica, No. I, 64, 1.979.
25. G. Bandopadhyaya and R.M. Fulrath, 1. Am. Cer. Soc., 57, 483,1974.
26. H.B. 1m, Proc. ICF-3 (Kyoto, Japan), 281, 1980.
27. H.T. Kim and H.B. 1m, Trans, IEEE, MAG-18, 1541, 1983.
28. V.G. Glotov, Porosh Met. 10,27, 1970.
29. K.D. Dugar Zabon, Porosh Met., 11, 60, 1971.
30. P. Peshev and H. Pacheva, Mat. Res. Bull., IS, 1055, 1980.
31. P. Peshev and H. Pacheva, Mat. Res. Bull., IS, 1199 1980.
32. J.J.Green and H.J. Van Hook, Trans. IEEE, MTT-25, 155, 1977.
33. W. Simonet and A. Hermosin, Trans. IEEE, MAG-14, 903, 1978.
34. Pran Kishan, D.R. Sagar, S.N. Chatterjee, L.K. Nagpaul, Nitendar Kumar and
K.K. Laroia, Adv. In Cer. 15,207,1985.
35. T. Inui and N. Ogasawara, Trans, IEEE, MAG-l3, 1729, 1977.
36. S.K. Bannerjee, P.D. Baba, B.1. Evans and S.S. Hafner, J. Phys. (Paris) Colloq.
C. 145, 1971.
37. C,J. Brower and C.E. Patton, 1. Appl. Phys., 53, 2104, 1982.
38. G.F. Dionne, J. Appl. Phys., 57,3727, 1985.
39. B.K. Kuanr, P.K. Singh, Pran Kishan And G.P. Srivastava, Proc. ICF-5 (Bombay,
India), 1005, 1989.
40. Pran Kishan, N. Kumar, B.S. Matheru, Chandra Prakash, K.K. Jain and G.P.
Sharma, Proc. ICF-5 (Bcmbay, India), 949, 1989.
41. Pran Kishan, G.P. Sharma, Nitendar Kumar, K.K. Jain and B.K. Kuanr, Proc.
Tenth Int. Con!. on Microwave Ferrites (Poland), Sept. 1990.
42. E. SchIoemann and R.E. Blight, Trans. IEEE, MTT-34, 1394, 1986.
43. D.C. Webb, Trans. IEEE, MAG-34, 2799, 1988.
44. H.L. Glass, Proc. IEEE, 86, 157, 1988.
45. H.L. Glass, F.S. Steams and L.R. Adkins, Proc. ICF-3 (Kyoto, Japan), 39,1980.
46. P.J.M. van der Straten and R. Metselaar, J. Cryst. Growth, 48, 114, 1980.
47. C.K. Maiti, D. Bhattacharya and N.B. Chakravarti, Trans. IEEE, CHMT-8, 221,
1985.
48. M. Abe, Y. Tamaura, Y. Goto, N. Kiramura and M. Gomi, J. App. Phy., 61,3211,
1987.
49. E.W. Babbit and R.A. Stem, Trans. IEEE, MAG-IS, 1744, 1979.
50. M. Abe, T. Itoh, Y. Tamaura, Y. Gotoh and M. Gomi, Tran. IEEE, MAG-23,
3736, 1987.
51. E.F. Schloemann, Proc. IEEE, 76, 188, 1988.
52. M.A. Truehaft and L.M. Silber, Proc. IRE, 48, 1538, 1958.
53. W.J.lnce and E. Stem, Trans. IEEE, MIT-IS, 87, 1967.
54. G. Cattarin, C.D. Gregorio, 'U. Pattrucci, F. Pucci and R. Roveda, Rev. Tee.
Selenia, 8, 29, 1982.
55. C.R. Boyd, Trans. IEEE, MTT-18, 1119, 1970.
56. C.R. Boyd, Microwave J., 20,41, 1977.

57. D.S. Beese and D.E. Sharon, IEEE, MTT-S, InL Microwave Symp. Digest, 333,
1983.
58. C.R. Boyd, IEEE, MTT-S, Int. Microwave Sym. Digest, 735, 1986.
59. W.E. Hord and C.R. Boyd, IEEE, MTT-S, Int. Microwave Sym. Digest, 727,
1986.
60. G.P. Rodrigue, Proc. IEEE, 76, 121, 1988.
7
Single Crystal YIG and Allied Materials
T.R.N. Kutty
Materials Research Centre, Indian Institute of Science, Bangalore, India
S. Sundaram
Defence Electronics Research Laboratory, Hyderabad, India
1. INTRODUCTION
During the course of a wide search for the preparation of a magnetic material
of better performance than the conventional ferrites, Bertaut and Forrat [1]
found in 1956, that a mixed oxide of yttrium and ferric iron had peculiar
ferrimagnetic properties. It was found to have extremely low ferrimagnetic line
width and high saturation magnetization. In addition, it was found to be a good
insulator with comparatively very low electrical conductivity. This is an ideal
combination of properties indeed, which enables its application in magnetically
tuned devices useful throughout the microwave region.
It may be of interest to point out that such extraordinary and useful properties
were observed with the mixed oxides (Y203 and F~03) by Pauthenet (1956)
[2] while he was working for his doctorate dissertation. He found very low
magnetic line width (&I) of 5.7 Oe at 3.93 GHz for sintered mixed oxides of
iron and yttrium in the polycrystalline form. This unusual behaviour triggered
intense activity in this area during the last three decades [3-5].
On examinl;J.tion, the mixed oxide polycrystalline material was found to have
a structure identical with a garnet, in contrast to that of the spinel structure of
conventional ferrites. Garnets are a class of compounds with the general formula
X3Y2 (Z04h. In natural garnets, X will be a divalent element such as magnesium
Mg*), iron (Fe++), calcium (Ca++), manganese (Mn++), etc. Y will be invariably
aluminium (Al+++), iron (Fe+++) and chromium (Cr+++), Z is found to be silicon.
To represent the oxidation state of the component elements in such natural
garnets, they are called 2 : 3 : 4 garnets. In addition to such a combination as
is found in natural garnets, it is possible to have cations with different oxidation
states in synthetic garnets. One of the most important class of garnets containing
rare earth elements correspond to 3 : 3 : 3. All the three X, Y, Z components
will have the oxidation state of + 3 only.
Among the popular garnets, the following may be mentioned in connection
with the application to the fields of optics, radar and communication.
Single Crystal fIG and Allied Materials 165
X Y Z Formula Acronym
Y Fe+++ Fe+++ Y,Fez (Fe04h YIG
Y AI+++ AI+++ Y,All (AI04h YAG
Gd Fe+++ Fe+++ Gd,Fez (Fe04h GdIG
Gd AI+++ AI+++ Gd,AIl (AI04h GdAIG
Gd Ga+++ Ga+++ Gd,Gaz (Ga04), GOG
Varieties of combination of rare earth oxides and oxides of other metals of

appropriate valence have been tried to prepare mixed oxide crystals. Immense
amount of data have accumulated in literature concerning the preparation,
properties and technical applications of garnets.
A range of single crystal substituted hexagonal ferrites [6, 7, 8] having a
large number of phases has also been introduced in this decade for millimeter
wave tunable applications.
This chapter deals with magnetic properties of single crystal and YIG
allied materials, methods of preparation, their characterisation and typical
applications.
2. APPLICATIONS OF YIG AND OTHER ALLIED MATERIALS

The usefulness of rare earth iron garnets (ReIG) as magnetic material started
off on account of the low ferrimagnetic line width (.111) which enabled" realization
of microwave components [9, 10] such as bandpass and band-stop filters. YIG
tuned oscillators have found wide applications in modern electronic
instrumentation and systems due to their broad tuning range, excellent tuning
linearity and "clean" spectrum. ReIG have also found applications in harmonic
generators, limiters and also as programmable frequency sources/counters for
high resolution radar systems.
YIG spheres are commonly used as resonator elements in the above devices.
The tiny YIG sphere is the solid state equivalent of the microwave cavity
resonator. For realization of tunable filters above 30 GHz, conventional single
crystal YIG materials have got limitations and new resonator materials based
on hexagonal barium ferrites are used upto 75 GHz.
In an endeavor to harness its excellent acoustic properties (very low acoustic
loss surpassing that of quartz), YIG is also shaped into rods and plates for
delay devices.
Exceptional magnetostatic properties of YIG have been exploited for the
wave propagation at microwave frequencies. Consequently many signal processing
functions have been realised which include (i) Pulse compression, (ii) Variable
delay, (iii) Storage, (iv) Parametric amplification, (v) Pulse recall and
(vi) Delay lines with built-in isolation.The magnetostatic mode in YIG anteceded
indeed practical magneto acoustic behaviour. Optically, rare earth iron garnets
have interesting Faraday rotation properties. Apart from their application in the
form of spheres, rods and plates, rare earth iron garnets have found extensive
applications in the form of thin layers in several devices. A layer of iron garnet
of a few micron thickness is grown epitaxially on a non-magnetic single crystal
substrate such as gadolinium gallium garnet (GOG). This is used in the fabricaion
of magnetic memory domain devices. Such thin layer garnets have been extended
for the fabrication of a variety of surface devices.
In addition to the above practical applications, rare earth iron garnet crystals
provide basic parameters for theoretical predictions. It is possible to substitute
various metal ions (both magnetic and non-magnetic) at unambiguous sites in
the crystal and measure magnetic properties. These results could be verified
experimentally with the theoretically predicted values such as g-values, ferri-
magnetic resonance line widths and giant anisotropy. Thus, it is possible to
prepare tailor-made magnetic crystals of desired saturation magnetisation (41tMs),
compensation temperature and magnetic anisotropy.
3. CRYSTAL STRUCTURE OF GARNET

Garnet crystals belong to body centered cubic system [11-13], corresponding
to the space group la3d. There are 8 formula units per unit cell. It means that
there are 160 atoms in the unit cell. The crystal structure affords three types of
non-equivalent sites in the oxygen lattices which are occupied by metal ions. The
formula of a garnet may accordingly be represented by C3 A2 (D04h. Figs. 1 and
2 give the general arrangement of coordination polyhedra in the garnet structure.
The C ions are arranged on a bec sublattice; they are 8-coordinated ions,
occupying the center of a distorted dodecahedral polyhedra of oxygen atoms.
Fig. 1. Portion of the garnet structure with cation and oxygen atoms.
There are 24 C sites in each unit cell of garnet and have the point group 222
(D~. A metal ion in a-site is surrounded by 6 oxygen ions which are at the
comers of an octahedron. There are sixteen a-sites in a unit cell which have the
local symmetry of point group 3(S6)' The 4-coordinated D ions have a distorted
tetrahedral symmetry and these (d) sites correspond to the point group 4 (S4)'
There are 24 d-sites in one unit cell. These, together with 96 h-sites occupied
by oxygen ions with triangular coordination, constitute 160 atoms of the garnet
Fig. 2. (a) Corrdination polyhedra in garnet structure; (b) Alternating edge

sharing tetrahedra and dodecahedra chains in garnet.
cell. We can take yttrium iron garnet as a typical example and we find individual
(Fe04) tetrahedra share comers with (Fe06) octahedra. Both these polyhedra
share edges with (YOs) dodecahedra.
4. MAGNETIC PROPERTIES OF GARNETS
4.1 Saturation Magnetisation (41rM,)

The basic requirement for a garnet to be magnetic is that it should have magnetic
catidhs in sufficient proportions at d and a-sites. The d-site cations are grouped
into a sublattice and they are coupled ferromagnetically. The a-site and c-site
cation sublattices are similarly coupled ferromagnetically. On the other hand.
the d-site cation sublattice is antiferromagnetically coupled to a-site sublattice.
The net magnetic moment is given by the algebraic sum of all the contributions
made by the ions in three different sublattices.
(1)
where Md. M. and Me represent the respective sublattice magnetic moments.
In the case of yttrium iron garnet the yttrium ion is diamagnetic. so that the
magnetism is due to the trivalent iron ions only. 1:he iron occupies both 16 (a)
sites and 24 (d) sites while yttrium occupies the 24(c) sites. The net magnetic
moment of YIG per molecule will therefore be:
M = 3Fe3+ - 2Fe3+ = 5/lB (2)
168 MICROWAvE MATERIALS
This net magnetic moment of YIG shows typical Brillouin functional behaviour.
Substitution of paramagnetic or diamagnetic ions at different sites modifies the
magnetic properties of garnets. Any of the rare earth ions with an atomic number
above 61 (the rare-earth ions below samarium apparently have ionic radii too
large to fit into the garnet structure) can be substituted for yittrium. For instance,
if Gd3+ is substituted at c-sites, the term Me enters in Eq. (1). The magnetic
moment will be aligned antiparallel to the moment of the ions on the (d) sites.
It will diminish the net magnetic moment M. It is, therefore, reasonable to expect
that the saturation magnetisation 4teM. for gadolinium doped YIG will be lower
than that of pure YIG and it has been found to be so.
In addition, such doped YIG crystals are observed to show variation in the
saturation magnetisation (4teM.) with temperature. There is a temperature at
which M = 0 and this is called compensation temperature (Tcomp)' This Tcomp
shifts to higher values with increased substitution of Gd3+ for yttrium. Above
Tcomp' the contributions from c-sublattice diminishes.
Substitution of non-magnetic ions such as AI3+ at d-sites of YIG, provides a
means to regulate the shape of 4teMs vs T curves (Fig. 3).Thus, garnets with
.-..
HI
::J
to
~ 1700
VI
~
t; (a)
...
.8
to
...
N
C»
c:
f
-200 -40 120.
Temperature (e)
(b)
Tetnperatur1& (e)
•
Fig. 3. Saturation magnetization of substituted garnets: (a) Saturation magnetization
of yttrium gadolinium iron garnet (Y 3-3xGd3x) [F~] (Fe3) 0IZ; (b) Saturation
magnetization of yttrium gadolinium aluminium iron garnet (Yl.,Gdl.S) [Fez]
(F~5wAl,w) 0 12,
Single Crystal YIG and Allied Materials 169
minimum temperature coefficient for 4nMs values can be obtained by the combined
substitution at c and d-sites. It is also possible to substitute a-sites with metals
like gallium, indium and other trivalent elements. Such substitutions also will
modify the magnetic properties. With appropriate substitutions it is possible to
prepare a variety of garnet magnetic materials suitable for specific purposes. A
list of such materials reported in literature is given in Tables 1 and 2.
Table 1. Reported major substitutions In garnets
Ion Probable ionic Example

radii (A)
(a) Substitution at d-sites (Coordination No.4)
A13+ 0.39 AI-garnet
Fe3+ 0.49 Fe-garnet
Ge4 + 0.40 Ge-garnet
Si4+ 0.26 Si-garnet
Ga3+ 0.47 Ga-garnet
V S+ 0.355 Ca3Fe2 (Fel.s V I.S) 0 12
pS+ 0.17 (N aCa2) Mg 2P30 12
Ass+ 0.335 (N aCa2) Mn2 (AS 3)012
Li+ 0.59 Na3AI2Li3F12
(b) Substitution at a-sites (Coordination No.6)
Ce3+ 1.03 YFe-garnet
Dy3+ 0.91 Ge-garnet
Ho3+ 0.90
y 3+ 0.89
Er3+ 0.89
Tm 3+ 0.87
Yb3+ 0.86
Ln3+ 0.85
Mn2+ 0.82 (CaY2) (Mn2) (Ge3) 0 12
In3+ 0.80 YFe-garnet
Fe2+ 0.78 YFe-garnet containing Si
Zn2+ 0.75 Gd3Zn2 (GaGe2) 0 12
Sc3+ 0.73 Ge-garnet
Mg2+ 0.720 (MgGd2) (Mg 2) Ge3012
Zr4+ 0.71 Ca3Zr2 (VO.SGa2.S) 0 12
Sn4+ 0.69 Ca3Sn2 (Ga2Sn) 0 12
Mn3+ 0.65 (MnGd2) (CoMn) Ge3012
Fe3+ 0.64 Fe-garnet
Nos+ 0.64 Ca3 (NbHf) Ga3012
Ga3+ 0.62 Fe-garnet, Ga-garnet
Cr3+ 0.615 YFe-garnet
A13+ 0.53 AI-garnet
Co 3+ 0.53 Ca3 (SCl.sCOo.Z> Ge3012
(c) Substitution at c-sites (Coordination No.8)
S~+ 1.25 Sr3 SezGe30 12
Ca2+ 1.12 Si-, Ge-garnet
Bi3+ 1.11 YFe-garnet
(COn/d.)
Sm3+ 1.09 Fe-, Ga-garnets

Eu3+ 1.07 Ga, Fe-garnet
Cd2+ 1.07 Si, Ge, Fe-garnets
Gd3+ 1.06 AI-, Ga-, Fe-garnets
Th3+ 1.04 AI-, Fe-, Ga-garnets
Dy3+ 1.03 AI-, Fe-, Ga-garnets
H0 3+ 1.02
y3+ 1.02
&3+ 1.00
Tm 3+ 0.99
Yb3+ 0.98
l.n3 + 0.97
Mn2+ 0.93 Si-gamet, (Mn Y 2) (Mn2) Ge3012
Mg2+ 0.89 Mg3Al2Si3012
C02+ Unknown (CoGd 2) C~GC)°12
CU2+ (CuGd 2) Mn2Ge3012
Fe2+ FC)A12Si30 12
4.2 Ferrimagnetic Resonance Line Width (L1H)

It has already been mentioned that in a ferrimagnetic material (which is also
a good electrical insulator) like YIG, a net magnetic moment exists as a result
of the coupling between various sublattices. The resultant magnetic moment
can be aligned with an externally applied quasi static magnetic field (Ho). If an
alternating magnetic field (such as microwave - RF frequency) is simultaneously
applied perpendicular to the static field, the magnetic moment of the crystal
will precess around the static field. The rate at which the magnetic moment of
the material precess depends on the fundamental constants of the material and
magnitude of the static field. If the frequency of the RF field coincides with
the internal precession frequency, there is a strong interaction called
'Ferromagnetic Resonance' (see Figs. 4 and 5). The precession angle is controlled
by the strength of the RF field, coupled to the material, but is limited to a
maximum. The resonance (absor'ption)peak will have a line width characteristic
of the chemical composition as well as the physical condition of the specimen.
This parameter is the important guiding property for its applications in
devices. Ferromagnetic resonance absorption line width is expected to be the
lowest for the best performance of the material. Such materials have important
engineering advantages for certain microwave ferrite devices.
Narrowness of ferrimagnetic resonance line width L1H, is the major factor
which makes rare earth iron garnets more useful in microwave devices. This
is in general due to the perfect ordering of the constituent atoms in the lattice.
There are three important factors which contribute to the resonance line width
and they are:
(1) Scattering of nonuniform precession into degenerate modes of magnons
arising from the porosity and from the surface finish. For example, polycrystalline
YIG specimen of 79% density has a value of J1H ... 420 Oe whereas that of
99% density has a value of 15 Oe. For a single crystal specimen, polished with
5Jlm grit, the L1H value is only 10 Oe, while the best polished YIG specimen
has a line width of less than 1, viz. 0.1 Oe.
Table 2. Summary of YIG and hexagonal ferrltes parameten
Material 4Me II. Ou Frequency of Minimum use- Temperatures depen- Curie temp.
(Gauss) (Oe) Ou measure- ful frequency dance of resonant ("C)
ment (GHz) (GHz) frequency (MHz/"C)
YIG Y3FesOil 1,800 70 9,500 9 3.4 Can be made 0 to 277

fIrSt order by
GaYIG Y3Fe4GalOll 400 50 1,500 9 0.75 sphere alignment 167 ~
S·
OQ
Hexagonal Ferrites ~
M-phase BaFel:PlO 4,800 17,000 950 63 52 + 15 450 Q

M-phase 3,800 8,000 ~
BaFell.1ScsOl1 430 32 26 -0 340 is
M-phase SrFel10ll 4,700 18,700 950 62 57 + 16 460 -:::s
M-phase SrFeIO.~11.5010 3,300 25,500 670 31 75 + 25 410 ~
V-phase BIlIZn1Fe l10n 2,850 9.900 1,250 17 9 101
U-phase BalZnlF~lOll 3,890 9,300 500 26 30 400 ~
~
Z-phase Bal~F~041 3,900 4,500 310 35 17 380
Notes: 1. Sc and Al doplings can be varied continuously.

-
[
2. Y phase has planar anisotropy, tr (2.8 MHZ/Oe x He(He + Ha»la when the anisotropy plane is aligned with both dc and RF fields.
= ~
~
...
S·
I:;"'
- .J
--
a b
t
Z
t4
Me!
y Y
x x
C d
Z zHa
Mo
t YIG Sphere
y y
It f
Fig. 4. Ferrimagnetic re.sonance (a) Randomly oriented magnetic dipoles in the
unmagnetized ferrite, (b) Magnetic dipoles aligned under the influence of a
magnetic field, (c) An equivalent representation of b showing the combined
effect of the aligned dipoles, (d) Precession of the net magnetization vector
due to RF magnetic excitation, (e) Equivalent representation of precessing
magnetization vector, and (f) Tuned bandpass filter consisting of YIO sphere
at center of two mutually orthogonal loops (extracted from 'Microwave Theory
and Application' Hewlett Packard Company, p. 163).
Fig. 5. Electron precession rate versus flux density (extracted from' APPLICA nON
Notes on YIO RF Components', YIO-TEK Corporation, p. 17).
(2) Impurities, substituted in the lattice greatly influence the linewidth,

particularly those rare earth ions that occupy the yttrium positions. For instance,
a sample of YIO, containing 0.6 atom percent of terbium has as high a LlH
value as 200 Oe. Samarium" doping in YIO increases LlH to the extent of 22
Oe per atom percent of Sm.
(3) LlH can also be influenced by the slow relaxation in the lattice through
the redistribution between the ions. For example, when YIO is prepared in a
nonoxidising condition, a fraction of the Fe3+ ions can have a lower oxidation
state (Fe2~. It is also possible to generate Fe2+ ions deliberately in presence of
a bivalent impurities. For instance," if silicon is present in YIO as an impurity,
it occupies the position of Fe3+ in the d-sites, forming Si04 tetrahedra. This
creates charge imbalance in the crystal which is compensated by the formation
of an equivalent amount of Fe2+ ions in the a-site. This exchange of an electron
between Fe2+ and Fe3+ greatly increases the LlH values.
It is needless to emphasis that YIG devices used for microwave engineering
should have minimum LlH values and this could be achieved only with single
crystals of highest purity, except when impurities are introduced with a definite
purpose.
5. CRYSTAL GROWTH OF YIG

Single crystal YIG materials are generally grown by either flux or hydrothermal
or epitaxial techniques.' Among all the methods described in literature for the
crystal growth of YIG, the flux method stands unique as the surest and quickest
method for preparing single crystals. The different techniques are described
below.
5.1 Flux Method

Molten inorganic salts (or 'Flux' .as it is generally known) have high solvating
power for refractory crystals. The common molten solvents include KF, PbO,
PbF2, B2~ and mixtures of these solvents and owing to the corrosive nature
of the flux, preparation is carried out in platinum vessels only. In order to fix
the composition of the charge (flux material and the mixed oxides, yttrium
oxide and iron oxide) and also temperature limits for crystal growth, it is
necessary to have some idea on the Y2~-F~~ phase system. The phase diagram
was first investigated by Nielsen and Dearborn [5] and later, in detail, by Van
Hook [14]. The system can be described only by the use of additional components
F~~-Fe304-YFe03, because the partial reduction of F~3 to FC]04 takes place.
In fact, for the same reason, Y3Fes012(yIG) melts only incongruently around
1550°C at 1 atm. of~ pressure. The phase diagram of Van Hook [14] is given
in Fig. 6. Above 1385OC, garnet phase coexists with magnetite. The composition
of the melt is different from the coexisting solids and is oxygen deficient.
Therefore, Fe2+ concentration in the garnet phase will be higher and is a function
of oxygen pressure as well as the temperature. Thus, from the phase diagram,
it follows that pure Y3FeS012 cannot be produced by the melting technique.
Besides, flux method, when adopted, should be carried out at temperatures
corresponding to the stability field of hematite + garnet.
174 MICRowAVB MATFJUALS
Molten metal solwnts
1680r--------------------,
1640
Orthoferrite +
Liquid
1555 t 3·
iJ
~ 1520 Carnet
~ ~TPe+ U:uid
& 1480 Liquid 1469· ±2·
E 1t----~A~--------1 Ortho-
~ 1"0 I
Magnetite + Garnet
fer-ite
Garnet
1400 1387· +3-

Hematite + Gar net
20 30 40 50
V3FesO,2 VFe03
Mole .,. V203
Fig. 6. Details of YIG phase diagram.
5.1.1 Composition of Charge

The composition of charge depends upon type of flux used viz. PbO-PbFrB2~
or BaO-B 20 3•
5.1.1.1 PbO-PbFrB2~
The first flux material ever tried for growing single crystals of YIG happens
to be lead oxide, PbO. The phase diagram worked out by Nielsen and Dearborn
[5] showed that YIG is incongruently melting. The liquids region for crystal
growth tends towards the region rich in ferric oxide. It can be made out from
this that crystallization of YIG is possible only in the narrow region as shown
in" Fig. 7.
The solvent PbO has certain disadvantages because of its (a) high viscosity,
(b) high density and (c) high melting point Comparatively high temperatures
are required to dissolve the oxide, and at this temperature, lead oxide has a
tendency to decompose and get reduced to'metal and attacks platinum.
Attempts have been made to improve the perform8liCe of the above described
system by adding other compounds to the flux material, PbO. Addition of lead
fluoride (PbF~ brings down the temperature of fusion. The lowest eutectic is
found to be around 500°C. Lead fluoride will resist the reduction to give rise
to lead, and hence the flux becomes less corrosive to platinum. The density and
viscosity are also reduced considerably which improves the flow property of
£lZlJ areaIn which

PbO
YIG has been
grown
Fig. 7. A partial phase diagram suggested for the system PbO-Yz03-Fez03'
the molten mass. Crystals will sink to the bottom and it is possible to separate
flux more easily.
It may be pointed that lead fluoride is highly volatile under experimental
conditions. The loss due to evaporation has to be prevented to maintain the
composition and also protect the furnace from the fluoride attack. Addition of
boric oxide (or boric acid) will bring down the vapour pressure. The presence of
calcium oxide has been found to improve the size and quality of YIG crystals.
The solubility curve [15] for YIG in PbO-PbF2"B 20 3 is given in Fig. 8.
garnet+
fluxed melt
flux E ....... garnet

Fig. 8. Saturation concn. of Y3FeS01Z in PbO-PbFz-B z0 3 flux.
5.1.1.2 BaO-B 20 3 FLUX

In spite of the addition of boric oxide (or boric acid), the above flux material
is found to have appreciable amount of vapour pressure. In addition to the
corrosive character of lead compounds towards platinum and its alloys, lead
compounds are highly poisonous and injurious to health. It is, therefore, reasonable
to search for a noncorrosive, nonpoisonous and less volatile flux material for
growing the crystals of YIG. Such a system has been found to be barium oxide-
boric oxide mixture. Another advantage of such a system is its comparatively
lower density. The stability region for such a system has been worked out by
previous investigators [16-18] and is shown in Fig. 9. The figure shows that
YIG has a narrow region of stability corresponding to the ratio of barium oxide
to boric oxide 3.6 to 1 (BaO : B20 3 :: 3.6 : I) and (4.55 : I).
SaO
1923'
Fig. 9. Approximate phase diagram for the system BaO-B20 3-yttrium

iron garnet in weight percent.
From the study of solubilities, it has been found that the stability region for
YIG at high temperature is also narrow. YIG is stable between 1005 and
I 190°C. Above I 190°C the stable composition corresponds to YFe03 and below
l005°C, the composition corresponds to BaFe12C'19. From these considerations,
it becomes obvious, that the composition of the charge should be such that YIG
is stable over a convenient range of temperatures and the formation of other
phases is avoided.
5.1.2 Methods of Preparation of Pure Chemicals and Reagents

YIG and other crystals are to be grown with materials of highest purity. The
level of impurities present in such materials should be very much lower than
in the conventional reagents of analytically pure grade (below a few parts per
million). Yittrium oxide (Y 203), free from any paramagnetic impurities and
other detectable rare earth impurities, should be of 'five nines' purity (99.999).
Pure ferric oxide (F~03) can be prepared by either precipitation of ferric
hydroxide and subsequent dehydration or by the decomposition of ferric
ammonium sulphate at elevated temperature. In the first method, ferric hydroxide
when, freshly precipitated, has a great tendency to adsorb considerable amounts
of a variety of impurities owing to high surface activity and it is difficult to
remove them either by washing or dialysis. The second method yields good
quality ferric oxide but recrystallisation is necessary to purify further.
It is needless to emphasize that the flux materials for YIG crystal growth
should also be of highest purity (i) Lead monoxide (PbO) is prepared in
polypropylene vessels (to avoid contamination with silica from glassware),
starting from analytical grade hydrated lead acetate Pb (CH3COOh 3H20.
(ii) Starting material for the preparation of lead fluoride (PhF:z) is either lead
acetate or lead nitrate (Ph (N03h> of analytically pure grade, via lead carbonate.
(iii) Boric oxide (B2~)' one of the ingredients added to improve the performance
of the flux, can be replaced by boric acid which is added to the charge and
heated slowly to the required temperature. (iv) Barium borate (BaOB 20 3)
corresponding to the composition BaO : B20 3 : : 1 : 0.6, employed as an alternate
flux material, is prepared by heating barium carbonate (BaC03) and boric acid
over a selected temperature range for several hours and cooling the molten
mass. It is also necessary to prepare barium carbonate starting from very
pure barium chloride and ammonium carbonate. (v) Calcium oxide, a minor
constituent, is obtained by heating calcium carbonate in a platinum dish to
900°C. (vi) Vanadium pentoxide, used as an oxidative additive, is prepared by
slow decomposition of ammonium vanadate. (vii) Lead peroxide is prepared
by the oxidation of Pb (II) salt, such as lead nitrate, with sodium hypochlorite
solution.
5.1.3 Mixing Up of the Charge Material

Each component is finely powdered and ground in an agate mortar. The ground
mass is transferred to a polythene bottle. The mixture in the boule is agitated
for 24 hours and mixed intimately in a mechanically operated tumbling device.
The operation is repeated two to three times for every charge. The charge is
then transferred to the platinum crucible and kept at different temperatures
successively at 120, 350 and 500°C. Such a procedure will bring out uniform
composition through out the bulk and consolidate the charge.
5.1.4 Methods for Flux Growth
5.1.4.1 SOAKING AND COOUNG

Premelting of the charge, is carried out be slowly heating the Pt-crucible to the
melting point of the flux (500°C in the case of PbO-PbF:z) and then to 850°C,
in a period of six hours. This allows water and carbon dioxide, if any, to escape
freely and to compact the mass in the platinum crucible. Premelting reduces
the bulk volume of the charge to about 25% and this enables more of the
charge to be added, whenever necessary. After premelting, the platinum vessel
with its contents is weighed. This is to check the extent of flux evaporation at
the end of the processing.
Prior to crystallization by slow cooling, it is necessary that the constituents
of YIG, namely Y203 and F~03, completely dissolve in the molten flux. Any
individual solute particle acts as nucleating centre, when crystallization occurs,
thereby causing the formation of a large number of small crystals. Therefore,
the melt is maintained at a temperature well above the crystallization point for
long periods. The crucible is heated to 1275°C. The temperature is maintained
for 15-20 hr. This is called the soaking period.
After the soaking period, the melt is cooled to lower temperatures. The
cooling is not uniform in the early stages. Instead, a cooling rate of 5 deg per
hour is maintained, till the temperature is lowered by 50°C. This will initiate
nucleation and growth at a faster rate. However, not all nuclei will grow at the
same rate. Therefore, initial rapid cooling will result in a large number of
crystals of varying dimensions of which a few will be bigger ones. After
lowering the temperature by 50°C, the furnace (Fig. 10) is reheated to near
soaking temperature (1270°C). This cycling in temperature enables the dissolution
Insulation·
Heating _-l,?-~-7''-1''''
elements
Platinum
vessel
Fluxed
melt
Crystal
Pedestal
Stand
Oxygen in
Fig. 10. Furnace used for flux growth of YIG crystals by

acceleratory crucible rotation technique.
of smaller crystals leaving behind a few larger ones, which must have dissolved
only partially. The furnace is now cooled at a much slower rate (0.5-1 deg/hr).
The partially dissolved bigger crystals will now act as growth centres, thereby
restricting the development of newer nuclei. As the growth continues, the surface
area of the crystals increases. The larger amounts of constituents from the
solution can get deposited on the growing crystal. This brings about a continuous
change in the supersaturation, which can be balanced by an increased cooling
rate towards the end of the processing. However, it has to be mentioned that
the optimum growth rate does not increase to high values so as to maintain
larger cooling rates. This means that the cooling rate, after the initial cycling
may be 0.5-1 deg/hr which can be increased to 3 deg/hr when the temperature
of the furnace goes down below 1100°C. However, cooling rate of 5 deg/hr or
above can still be deterrent to the quality of the crystals.
Faster cooling rates are always found to result in crystals with lot of flux
inclusion. This is due to the fact that the number of the growing nuclei will be
more at the beginning of the processing. The number keeps on increasing as
the temperature is lowered at a faster rate. These fast growing nuclei interfere
by coalescing and most probably will contain in themselves a part of the
growing medium. When the growth is complete, the crystals so produced will
have small pockets of solidified flux materials in them. Also, the fast growth
of nuclei encourages dendritic growth. When the dendrites join arms, small
amounts of flux materials get entrapped and the crystals acquire a defective
core. Therefore, slow cooling, especially at the early stages is essential.
5.1.4.2 ANNEAuNo AND REcOVERY OF CRYSTALS

Studies on YIG-PbO-PbF2 system have shown that the YIG phase is not stable
below around 950°C. The garnet dissolves in the flux and yttrium orthoferrite,
YFe03, is formed. Below 900°C magnetoplumbite, PbFe12019 is formed as an
additional phase. The redissolution of garnet in the molten flux can be prevented
and crystals recovered by more than one method.
(i) The container can be taken out of the furnace at 950°C and the melt can
be poured off. The crystals stick to the bottom or sides of the platinum
vessel which is immediately placed back in the furnace. The crystals
are then annealed to room temp«Wdture.
(ii) In another method, crucible can be inverted in the furnace itself using
supporting rods. In the inverted position, the crystals remain at the top
of the platinum vessel stuck up to the sides and the melt flows down
to the bottom. Such inversion is possible with crucible assembly shown
in Fig. 11. The vessel has- to be cut with the help of a hydrogen flame
to collect the crystals.
(iii) A third method is to cool the melt below 950°C, comparatively rapidly
(50 deg/hr) so that little time is given for the dissolution of YIG and
development of YFe03 or PbFe12019 phases. Once the melt is below
the freezing point, the solidified mass is annealed to room temperature.
The crystals in this case will be embedded in the matrix of the flux and
they can be recovered by the preferential leaching.
The first method of recovering the crystals results in serious thermal shock,
however quickly the pouring off step can be completed. Besides, this procedure
exposes the operator directly to the vapours of the molten lead compounds and
to the radiating heat from the furnace. Therefore, only the other two methods
( b)
Fig. 11. Crucible assembly .
are generally adopted. It has been mentioned in literature that the third method
results in development of strain in the crystal while being embedded in the
solidified matrix of the flux. Investigations have shown that there is no difference
in the properties of crystals when method (ii) or (iii) is adopted, other parameters
being kept constant Since the third method is simpler for larger batch productions,
this procedure is generally followed.
The crystals have to be separated from the solidified flux. This is done by
leaching with dilute acids. Nitric acid is an ideal choice, since it dissolves the
solidified flux relatively faster and at the same time garnet crystals are not
attacked. The crucible is kept in boiling dilute nitric acid (3N). A charge of
about 500 g can be dissolved out by continuous boiling for two to thtee days.
In the case of procedure (i) and (ii), 2-3 hr of treatment with boiling nitric acid
(3N) removes the coating of the flux.
5.1.4.3 CONDmONS FOR FLUX OROwrn

It is necessary to maintain oxidizing conditions in the furnace during the flux
growth ofYIG. First, it will minimise the formation of Fe2+ ions in garnet which,
if present will increase conductance of YIG crystals through Fe2+ --+ Fe3+
hopping mechanism. Due to Fe2+ --+ Fe3+ exchange coupling, there will be
considerable increase in the ferrimagnetic resonance linewidth. Secondly, the
oxidizing conditions prevent the formation of lead through the reduction of Pb
ions. Lead will attack platinum crucibles as mentioned earlier.
Many of the earlier workers maintained oxidizing conditions by carrying
out crystal growth in oxygen atmospheres under pressure (5-100 atm). However,
this method is quite elaborate and tedious. On the other hand, proper additives
can be incorporated in the flux which will maintain oxidizing conditions.
(i) Addition ofV20 s (0.1-0.5 wt % of the total charge) sets up the V S+ ~ V4+
equilibrium which prevents the formation of Fe2+ ions. Vanadium even if
incorporated in YIG will result only in materials of low loss. Actually it has
been reported that the simultaneous substitution of Ca and V in YIG gives
material of extremely low loss.
(ii) Addition of Pb0 2 can be the other alternative. Since PbO is a major
component in the flux, a part of it can be replaced by Pb0 2 (1-2% by wt) which
will set in the Pb 4+ ~ Pb 2 + equilibrium. Sometimes, Pb02 is added in the charge
initially. Or otherwise the charge is kept in oxygen atmosphere at 400°C for
over 12 hrs. This will generate P~04 through the oxidation of PbO. At higher
temperatures, disproportionation of P~04 takes place as
(3)
Pb02 thus formed will be effective as mentioned earlier.

(iii) A stream of oxygen is directed to the bottom of the crucible during the
early stages of nucleation. This will act as the 'cold finger' for nucleation to
take place preferentially at the bottom. Besides, oxygen prevents the reducing
conditions to set in.
5.1.4.4 PARAMETERS THAT INFLUENCE SIZE AND QUAUfY OF CRYSTALS

Insufficient mixing and homogenization of the charges, prior to melting is
found to affect the quality of crystals. Crystals will be comparatively smaller
(1-3 mm size) even when given reasonable soaking periods (12-15 hr). Co-
precipitation of yttrium, ferric iron and lead from solutions as carbonate is
found to be an alternative method to attain easy homogeneity. Cooling rate
greatly influences crystal size as well as the number of inclusions they contain.
A continuously changing cooling rate, as shown in Fig. 12, is preferable to a
linear rate. Better crystals result by inducing nucleation at the bottom of the
crucible. This preferential nucleation can be effected by means of a stream of
oxygen projected to the bottom of the crucible. The crystals resulting from the
growth runs in which considerable PbF2 has evaporated are found to be smaller
in size with more inclusions. The size and yield of crystals have increased. The
inclusions also come down with the addition of PbF2 to PbO.
5.2 Hydrothermal Method

Solubility of many of the compounds which are insoluble in water under ambient
conditions can be increased considerably at higher temperatures under pressure.
This can be further enhanced by the addition of suitable reagents, generally
named as 'mineralisers'. By controlling the supersaturation at different parts of
the pressure vessel, which is achieved by adjusting the temperature gradient,
crystal growth can be carried out from aqueous solutions very much above the
boiling point of water. This technique called "hydrothermal method", can be
employed to grow single crystals at much lower temperatures than required for
the flux method. Hydrothermal method has been successfully employed in the
industrial preparation of quartz crystals. Quartz is grown from dilute sodium
carbonate or sodium hydroxide solutions around 400°C and about 1000 atm.
Crystal growth of YIa is an extension of this process. However, the solubility
of the constituent oxides, viz., Y203 and F~03 is very low in dilute alkali.
Therefore, the concentration of the mineraliser should be increased to much
higher value to obtain profitable growth rate. [19] Since the growth can be
carried out around 400°C, the problem of ferrous iron formation can be
considerably reduced.
The autoclave used in this method is schematically shown in Fig. 13. The
lower part is filled with polycrystalline YIa which is kept at a higher temperature.
The seed crystals are suspended from the closing piece of the liner, by means
of silver wire. In hydrothermal method, proper temperature conditions have to
be maintained so that the fluid phase fills the entire space available inside the
autoclave. It is necessary to keep a temperature gradient between the two ends
of the pressure vessel, such that the nutrient zone (lower part) is hotter than the
growth zone (upper part). A chromel-alumel thermocouple connected to a
potentiometric controller, regulates the furnace temperature. Thermocouples
positioned at the top and bottom of the autoclav,e are used for measurements.
The temperature gradient can be altered to any desired value by positioning the
base of the autoclave and adjusting the opening of the furnace. Temperature
gradient between 0 and 75°C can be obtained when the average working
temperature is around 400°C.
Crystal growth is canried out in 20% (by weight) sodium hydroxide solution.
The nutrient is mostly polycrystalline VIa. When the constituent oxides are
Growth stage
1200
r
T ·C ,..., 1.5"/hr
C;')
1100 ~.
1\"
(")
~
<oj
'""' 2.0·/hr
iii
-::s
C)
1000
~
~
End of the
growth rJ.? [
~
o 20 60 100 140 180 220 260 300 ~
TIME (Hours)- is'
t:;
Fig. 12. A typical cooling curve.
00
Vl
-
directly taken, the mole ratio of Y2~ and F~~ is kept close to 3:5. 75% of
the available volume inside the liner is filled 'with sodium hydroxide solution.
The seed crystals used are those obtained from the flux experiments. The baffle
opening is maintained around five percent. The space between the liner and the
autoclave is filled with distilled water (65-70% fill). Since sodium hydroxide
solution has lower pressure than that of pure water, lower percentage fill is
adequate to balance the pressure developed inside the liner.
The autoclave is assembled as shown in Fig. 13. It is closed with the cap,
followed by torquing the tightening screws. It is positioned in the furnace and
1. Body (QD 3'; ID i)

2 2.Cap (00 4·)
3.Closing plunger
4. Tightening screws)
5. Thrust washer
6. Con nect ion nut
7. Pressure connection
to bourdon gauge
8. Liner
9. Nutrient
10. Seed crystals
11. Thermocouple well
8 12:Saffle with controlled
opening.
12
11
Fig. 13. High pressure vessel for hydrothermal growth.
heated to 4000C. The temperature gradient is adjusted such that the bottom
temperature is 400°C, while the top thermocouple reads 380°C (temperature
gradient, liT - 20°C. The pressure within the liner is around 600 atm, while
outside the liner it is more then this value. The pressure vessel is maintained
for over 15 days, ltt the end of which the autoclave is cooled to room temperature
and the crystals mcovered. Growth of crystals is rather slow in the case of YIG
from 20 wt % sodium hydroxide. The average growth rate is less than 5 mils
per day. The growth rate increases with larger temperature gradient, but coherent
and uniform growth is obtained with tJ.T - 20°C.
YIG crystals grown by the hydrothermal method have minimum Fe2+ iron
content (invariably less than 0.02 atoms per formula unit). However the greatest
drawback is the presence of (OH) groups incorporated in the crystal. Infrared
spectra of hydrothermally grown YIG crystals show a broad band in the region
of 2800--3650 cm- t • This broad absorption is invariably absent for the flux-
grown specimens. Presence of (OR) group broadens the ferrimagnetic resonance
Iinewidth of the material. Therefore, its reduction is of. considerable importance.
It has been suggested that potassium hydroxide, though more corrosive, is a
better mineraliser [20]. The addition of calcium is also said to reduce the (OH)
content.
5.3 Epitaxial Technique

The adoption of 'Liquid phase Epitaxy (LPE)' techniques (generally used for
the fabrication of 'magnetic bubble memory' devices) for the growth of single
crystal YIG materials in the form of 'thinfilms' [21] has opened a new vista
in 'MICROWAVB MAGNETICS' which offers wide potential applications
[22] in the YHF and Microwave Frequency bands extending to millimetric
waves. This is the thermal growth of single crystal thin layers of magnetic
garnets on non-magnetic substrates by LPE.
YIG films can be grown on highly polished gadolinium gallium garnet
(ooG) utilizing vapour phase epitaxy (Y.P.E.) in a T-shaped reactor made of
fused silica. Source materials, yttrium and iron, are in the form of volatile
chlorides contained within platinum crucibles in the vertical leg of the T. Their
vapour pressures are controlled independently. These materials are carried into
the reaction zone, the downstream of the horizontal segment by dry HCI and
helium for the yttrium chloride and by helium for the ferrous chloride. Oxygen,
in a helium carrier, is introduced into the upstream side of the horizontal
segment GGG is chosen in the substrate since it is chemically' inert and provides
a good lattice match (within 0.007 A) and thermal expansion match from the
deposition temperature (approx. 1200°C) to room temperature.
Y.P.E. [21] gives rise to four oxides including YIG at different locations
within the reactor. It is possible to adjust the position and size of the YIG zone
to coincide with the location of the GOO subStrate. This shift of the reaction
zone is carried out by observing the colour of the oxide deposits on a fused
silica test plate introduced for a short time at the normal deposition location.
The length of the YIG zone is 5 to 7.5 cm and the film growth rate is 6 J.l.m per
hour. The Y.P.E. process suffers from the disadvantage of restricting the thickness
of YIG films to 20 J.l.M, due to depletion of source materials in the reactor. This
thickness is inadequate for useful microwave filters. However, ferrimagnetic
resonance Iinewidths (Mf) of less than 1 Oe have been measured at X-band.
Y.P.E. has been surpassed by liquid phase epitaxy (L.P.E.) which is an
extension of the bulk crystal flux growth technique. Garnet films grown by
L.P.E. have defect densities determined principally by the substrate defect
density which for good quality GOO is less than 5 per cm-I . Film thickness can
be held within 10% across 90% of the substrate area, with no detectable
composition variations or second phases. Films can be grown at growth rates
of 0.1 pm/min to more than 3 pm/min and to thickness of greater than 100 pm.
Epitaxial films of YIG have been grown by Adam et al [22] on GGG by
L.P.E. from a Y2~ (1.4 g) - F~~ (11.82 g) - B20 3 (4.04 g)-PbO (182.74 g)
flux. The following details have been reported by them. The growth technique
used horizontal dipping with axial rotation. The 200 g melt was contained in a
platinum crucible, and has a saturation temperature of 920°C. The 1.5 cm diameter
substrate was held by a platinum vacuum chuck, and was rotated at 100 rev/min
during growth and at 1000 rev/min, immediately after removal from the melt.
Using this technique, films of up to 100 pm thickness were grown at a rate of
approximately 2 pm/min at 900°C. The versatility of the L.P.E. system allows
the addition of rare earth ions or diamagnetic ions such as Ga3+.
6. CHARACTERISATION OF YIG CRYSTALS

Following physicochemical methods are employed to characterise the YIG
crystals: (I) Crystal morphology, (2) Density, (3) X-ray diffraction pattern,
(4) Infrared absorption spectrum, (5) Estimation oflead and ferrous iron contents,
(6) Determination of Curie temperature, (7) Saturation magnetisation and
(8) Ferrimagnetic resonance line width of YIG specimens of spherical geometry.
This involves the fabrication of highly polished spheres. Such a characterisation
will be helpful as guide to determine whether the YIG crystals could be used
in the devices. In tum, from the results of these characterizations, it is possible
to optimize the method of crystal growth and the fabrication of spheres to meet
the specific requirements. The quality of the crystals and the surface finish of
the spheres could thus be improved.
6.1 Crystal Morphology

Almost all well-formed crystals are rhombic dodecahedra, modified to various
degrees by icositeb"ahedral (211) faces [23}. The sharpest comers of the rhombic
dodecahedra result from the intersection of four (110) plane and point in six
equivalent (100) directions. The 8 remaining comers result from the intersection
of three (110) planes and point in (111) directions.
Chemical etching techniques provide a simple estimate of crystal perfection
with regard to dislocation. Suitable etchants for YIG include cold 20% (HCI),
85% (H3P04), UN03 (8 N)-acetic acid -H20) mixtures and dilute H2S04 (10 N).
Polished spheres ofYIG can be etched in HCl for 1 hr. Etch figures showed
(100), (110) and (111) holes [24}. Dislocation densities range froml()l to 1~/cm2.
6.2 Density
The gross density of YIG crystals is used as a rough measure of the extent of
inclusions present. The measured density can be compared with X-ray density.
Gross difference between these values indicates the presence of other phases
e.g; orthoferrite (YF~). Density of the crystals is measured by the displacement
method using specific gravity botfle. This involves the determination of the
volume of a liquid displaced by 'a known mass of the crystal. Bromofonn is

used as the displacing liquid (density =2.80 gm/cm 3 , at 2S C), so as to increase
the accuracy of measurement. The specific gravity bottle along with the liquid
is maintained at 25 C using a thennostat.
6.3 X-ray Diffraction Pattern
X-ray patterns of crushed single crystals are obtained by Debye-Scherrer method
using Cr-Karadiation (V-filtered) from a Rich-Seifert Unit. Philips Camera of
114.6mm diameter is used. In order to obtain the accurate cell edge value (ao),
those calculated from the high angle reflections are plotted against the Nelson-
Riley function (Fig. 14). It gives a straight line and is extrapolated to 8 =90
deg to read the actual ao value. For the specimen used for this plot (Fig. 14),
12.39
.......
......
~
12.34
12.33L--._ _ _ _-'--_ _ _ _..."....,~---__=_''=_---___:_:_!

o 0.10 2 0.20 2 0.30 0.40
l(CoS 9+ Cos9 9)
2 Sin 9
Fig. 14. Unit cell parameter of Y3FeS<>12'
=
ao 12.379 A. For crystals from various growth runs, ao ranges from 12.374
to 12.382 A [25, 26]. The accuracy of cell parameter for cubic crystals as
standardized by the above method is ± 0.002 A. This is a good indication of
the phase purity of YIG crystals. It may be mentioned that X-ray reflections
corresponding to YF~ or PbFe12019 were absent, so also those of flux materials
PbO and PbF2.
The quality of single crystals are tested from the X-ray rotation photographs.
Crystals shaped in the form of spheres (0.8 mm dia) are used for this purpose.
The sphere is oriented in the [100] axis. The reflections were found to be
un distorted. This indicates the strain-free nature of the crystal.
6.4 Infrared Absorption Spectrum
Infrared absorption spectra of rare earth iron garnets are fairly diagnostic of the
188 MICRoWAVB MATERIALS
purity of them. The important absorption bands for YIG is around 600 and
380 cm-I . The band around 600 cm-I can be tentatively assigned to asymmetric
stretching modes of (Fe04) tetrahedra and has the first overtones around
1150 cm-I [27]. However, it is more complex in the sense that lattice modes also
influences these absorption bands. The important point to be mentioned is that
infrared spectrum serves to detect the silicon impurities. As mentioned earlier,
silicon impurity cannot be completely eliminated. Since rare earth iron garnets
and natural silicate garnets are isomorphous, the silicon will be incorporated in
iron garnets as Si04 tetrahedra. In Fig. 15, the characteristic absorption spectrum
of a silicate garnet is compared with those of two YIG samples: (1) containing
c21
~
..
Oil
c
~
~ ;!
"
.
==
Oil
c
!
....
;!
Fig. 15. Infrared spectra of garnets.
considerable amount of Si impurity, (2) another crystal of minimum Si impurity.

The extra bands of sample (1) around 875 and 915 cm- 1 [28] coincides with those
of silicate garnets. The infrared spectrum of sample (2) also indicates that it has
minimum divalent impurities which could be mostly Fe2+ and to a less extent
Pb2+. Incorporation ofPb2+ in the garnet lattice is more difficult due to ionic size
difference. Fig. 15 also gives the spectrum in the region 700--1100 cm-1 for silicate
ions in YIG taken at higher concentration of the material in the matrix (25 mg of
YIG in 500 mg KCI).
6.S Estimation of Lead and Ferrous Iron Contents

The crushed single crystals are analysed for lead in the following way. A
known weight of the well powdered sample (- 320 mesh) is fused with potassium
hydrogen sulphate to fuming temperature. The solidified mass is digested with
2N nitric acid and then evaporated to near dryness. The residue is boiled with
water when all the salts excepting PbS04 dissolve. PbS04 formed is estimated
gravimetrically. The amount of lead present is directly proportional to flux
inclusion. The lead content has been found to be minimum « 0.15%) in well-
grown crystals. However, the percentage of lead is higher in those crystals
grown with a faster cooling rate, reaching upto 1.3%.
From the filtrate of the above experiment, yttrium and iron are determined.
=
High quality crystals gave the analysis Y2~ 43.92 and F~~ 55.89% =
= =
(calculated values for YIG are YA 44.08 and F~~ 55.92%). The analytical
data, again, confIrm the phase identity.
Ferrous iron in YIG is estimated by the vanadate method. For this purpose,
the single crystals are crushed and powdered to - 400 mesh in an agate mortar
under acetone. This method is normally adopted to prevent oxidation during
grinding. A known weight of the dry powder is treated with 10 ml standard
ammonium vanadate solution and 5 ml concentrated sulphuric acid. The solution
is kept at near boiling point on a sand-bath with minimum evaporation. After
about 2 hr, the solution is cooled, diluted with 4N H2S04 and titrated against
standard ferrous ammonium sulphate solution. The difference between direct
and the test titration values is proportional to the reducing capacity-which is
due to Fe2+ content in the sample. The duration of the experiment can be
reduced using a pressure vessel at 150°C. The well grown crystals with low Si
content have FeO content < 0.28%. This value increases for products from
crystal growth runs where oxidizing additive has not been added. FeO content
increases to 1.5% in some cases. The lowest FeO content (0.025 atom per
formula unit of YIG) observed corresponds to experiments in which V20S is
added to the charge. It has been found that Fe2+ content increases when Si is
intentionally added. It may be mentioned that Si-doped YIG has giant anisotropic
properties.
6.6 Determination of Curie Temperature

It is the temperature at which the ferrimagnetic crystal becomes a normal
paramagnetic compound. The Curie temperature of YIG crystals is determined
by the standard technique. An iron rod of about 12" long is held by a magnet,
such that the rod remained suspended in a tubular electric furnace. A few
crystals of YIG are attached to the bottom end of the rod and the position of
the rod is so adjusted that the crystals remain in the middle of the furnace. A
chromel-alumel thermocouple is positioned at the same level. The temperature
of the furnace is gradually increased at a rate of - 0.5°C/min. At a certain
temperature, the YIG crystals have fallen down due to the loss of magnetisation.
This temperature corresponds to the Curie point. The experiment can be repeated
with the same crystals. The Curie temperature is found to vary between 276
and 278°C. The reported value for YIG is 279-280°C [28].
6.7 Saturation Magnetisation

The most commonly used methods for evaluation of 4nMs [29] are the (i) ballistic
method (ii) vibration coil magnetometer (iii) the vibrating sample magnetometer
(iv) various force methods and (v) microwave methods. The Curie temperature
can be determined from the plot of 41tM. vs temperature. There are three
practicable ways of estimating 4trM. by microwave technique. Measurement of
the shift in the ferrimagnetic resonance field for a non-spherical specimen [30],
observation of magnetostatic modes [31] which is applicable to narrow line
width single crystals and study of magnetostatic cavity modes in the bulk
materials.
Saturation magnetisation (4trM.) can also be determined by measuring the
mass susceptibility of powdered YIG crystal at a field strength of 10,000 Oe
(around 290 x 10-6 cgs units/g). At this field, magnetisation saturation is achieved
and the M values correspond to M.aturalion • The 4trM. is related to the mass
susceptibility in the following way.
B =H + 4trM. (4)
B =H + 4tr",d . H (5)
Therefore. 4trM. =4tr",d . H (6)
where '" = mass susceptibility, d = density, H = field strength and B = magnetic
permeability
The 41tM. calculated in this manner is around 1750 Oe [32].
6.8 Fabrication or Spheres or YIG ror the Evaluation or

Ferrimagnetic Resonance Linewidth
Spheres are to be fabricated from YIG crystals for the evaluation of magnetic
properties as well as for use in various devices. The spheres are to have good
sphericity and maximum possible surface finish. The crystals are first cut using
diamond blades, the thickness of which should be as small as possible to
minimise loss of material. The width of the cut is 0.1~.2 mm, small cubes of
2.0-2.5 mm edge length are thus made by this method.
The cubes are ground into spheres by the air cyclone cavity method. Before
they are ground in the cavity, the cubes are tumbled in a metal cylinder containing
a slurry of 200 mesh carborundum powder in water for 24 hr. The edges and
comers of the cubes are rounded off by this procedure. The cubes are then
placed in the cavity, one at a time, the wall of which is lined with 400 grade
emery paper. The air is admitted into the cavity at a pressure of 50 psi. The
nozzle through which the air enters, is so adjusted that the air stream takes a
circular tum in the cavity (Fig. 16) which gives the crystal a swirling random
motion forcing it to graze the abrasive paper frequently. The air escapes through
the central holes. The continuous motion and random abrasion help the crystal
to attain spherical shape [33, 34].
After the crystal specimen has attained the spherical shape, the size regulation
and polishing are carried out by changing to finer emery papers in the order:
grade 1, 1/0,2/0, 3/0 and 4/0. As the finer emery is used, proper cleaning of
the cavity as well as the sphere is necessary, when the emery paper is changed.
Proper adjustment of air pressure and time of grinding result in very good
surface finish. The maximum reduction in size of the specimen takes place
Perspex
*---11- Air exit

Tightening
screw
G. .~
~;e~
~ir
Fig. 16. Cavity for grinding and polishing YIO spheres.
when emery paper of 1/0 and 2/0 grit size are used. Strict control of the
grinding time is necessary at this stage, to control the size. The size of the
sphere is checked at 1/2 hour intervals at 1/0 and 2/0 stages. With 3/0 and
4/0 emery, the surface acquires higher polish and air pressure is reduced to
5-10 psi. After the use of 4/0 emery, the spheres will be shining in reflected light.
The spheres are further polished with diamond paste of finer grit size in
steps of 2-4, 0-2 and 0-1/2 J.I.M grit. These abrasive compounds are first applied
to a soft carrier medium like wood, hard variety of paper or canvas cloth. The
diamond particles get embedded in the medium. The wall of the air cyclone
cavity is lined with this abrasive carrier.
Final polishing is carried out by a mechanochemical method using a suspension
of colloidal silica « 0.05 J.I.M grit) in 2 N KOH. This polishing mixture is
available in the trade name "Syton". The slurry is placed on a pelion polishing
pad and spheres are polished in it by simultaneous pressing and rolling for
6-10 hr. The finish of the surface is checked by viewing under reflecting
microscope (magnification, 50), whereby no scratches or pits should be seen.
6.9 Ferrimagnetic Resonance Linewidth (Ml) or YIG Materials

Linewidth of YIG materials can be determined by the following methods [9,
35,36]:
(i) Resonant cavity method.
(ii) Shorted line method.
(iii) Transmission method.
(i) Resonant cavity method is the most widely used technique for evaluation
of t1H of polycrystalline ferrites. The basic measurements to be made are of
resonant frequency and Q of the cavity containing the sample as the function
of the internal steady field. When a sample of the ferrite is placed in the cavity
at a position of maximum magnetic field and minimum electric field, the real
part of the permeability gives rise to a shift in the resonant frequency and
imaginary component to a fall in Q. The line width tJH is measured by measuring

the change in bias field corresponding to the 3 dB bandwidth. The exact solution
to the problem of ferrite loaded cavity is only possible in certain cases provided
the volume of the sample is very small compared to that of the cavity and
perturbation theory may be applied. The assumption is made that the fields
within the cavity are substantially unchanged by introduction of the specimen.
(a) However, very narrow linewidths as of single crystal YIG, cannot be
easily measured by the resonant cavity technique because, in the
neighbourhood of ferrimagnetic resonance the coupling, which is
proportional to (tJH)-l becomes so strong that perturbation theory no
longer holds good.
(b) Another disadvantage of the cavity technique is that the linewidth can
be measured only at a single frequency.
(ii) In shorted transmission [37] line method, the sphere is kept in a position
of maximum rf magnetic field, for instance a half wave length away from the
short-circuited end. The frequency is kept constant and the applied magnetic
field is varied to obtain ferrimagnetic resonance. Linewidth is determined by
measuring the change in bias field corresponding to the 3 db bandwidth.
The disadvantages of this method are:
(a) The linewidth can be measured only at a fixed frequency for a given
shorted line.
(b) Sophisticated instrumentation is required to observe precisely the 3 dB
bandwidth and a high order of signal frequency and amplitude stability
is essential.
(iii) In the transmission method [38], the 'band stop filter' principle is made use
of in measuring the linewidth. The sphere is placed in a RF transmission line
and transverse dc magnetic field is applied. At resonance, depending upon the
linewidth of the sphere, power will be absorbed by the sphere. This power loss
at resonance is made use of in measuring the linewidth of the sphere. By
varying the dc field, the resonant frequency can be varied so that linewidth can
be measured at any frequency of interest.
In evaluating the linewidth of a sphere, the required data are (i) the inser-
tion loss (Fig. 17) introduced by the sphere at resonance and (ii) the frequency
of resonance in MHz which can be measured accurately.
7. BASIC PARAMETERS OF YIG RESONATORS

The design of YIG tuned devices is centered around high quality resonators
fabricated out of single crystal ferrimagnetic materials. Highly polished spheres
of single crystal yttrium iron garnet (YIG) and gallium doped YIG (Ga YIG)
exhibit ferrimagnetic resonance.
The two significant characteristics of YIG described earlier, which dominate
the design of any magnetic, tunable device are:
(i) Saturation magnetisation (4nMs-gauss)
(ii) Resonance linewidth (tJH-oersted).
6
10
Ds in mills
SPHERE DIA IN MILLS
Fig. 17. Carter's chart of Qc vs sphere dia for a spherical YIG resonator.
The saturation magnetisation (4nMs) is a function of the number of electron

spins in the material per unit volume. Larger the 41tMs , easier it is to couple
the RF field to the resonator.
Linewidth is the width of resonance in oestered as the RF frequency is held
constant and applied dc field is varied. It can also be defined as the difference
between the two values of biasing field (at each side of resonance) for which
the imaginary part of the intrinsic susceptibility of the material equals the real
part, while the frequency is held constant. This factor depends upon the material
composition and the shape and surface finish of the resonator.
Q,. Factor: The unloaded Q.. of the spherical resonator is given by
Q = 10 X 10-6 (17)
u 2.8.:1 II
The unloaded Qu of the YIG resonator increases with frequency (f 0) for a given
value of II. It is desirable to have a low linewidth, to derive a high Qu.
7.1 Resonant Frequency

The resonant frequency for any ellipsoidal YIG resonator (neglecting anisotropy
effects) is given by
10 MHz =2.8 ..J[Ho - (Nz - NJ(411MJ][Ho - (N z - Ny) (411MJ] (8)
where NJ.' Ny and N z are demagnetizing factors in the x, y and z directions, Ho

is the biasing field in oersted and 411M. is in gauss.
= = =
For a sphere, NJ. Ny Nz 1/3. Hence,
(fo) = 2.8 Ho (9)
Thus, for a spherical resonator which is easier to prepare with precision, the
resonant frequency is independent of saturation magnetization and linearly
dependent only on biasing field Ho. In any other shape for the resonator, for
= = =
example, thin disc in the XY plane where NJ. Ny 0 and N z 1, the resonant
frequency
10 =2.8 (Ho - 4nM,) (10)
In this case, resonant frequency is influenced by saturation.
7.1.1 Resonance Limitations

An important factor in the phenomenon of Ferrimagnetic resonance, is critical
frequency!c which is determined by the value of 4nMs according to the equation
(11)
where Nt is transverse demagnetization factor. For a sphere, Nt = NJ. =Ny = 1/3
and hence
Ie ="32 4nM, (12)
Non-uniform coupling between the rf field and the precessing magnetic dipoles
occur below this frequency, giving rise to non-linear behaviour and increases
in losses. At frequencies lower than (1/2)!c, resonance cannot occur as it is not
possible to achieve magnetic saturation in the YIG resonator.
Also, resonator operation at temperatures close to or above the Curie tempe-
rature Te, of the material is not possible as the 4nM. drops to zero at Te.
7.2 RF Power Limitations

There are two major mechanisms which limit the amount of RF energy which
may be coupled to the YIG resonator. 'High level limiting' (RF level of the
order of 20 dBnJ is caused by the fact that the angle of electron precession has
a finite limit. The other RF power limiting mode is called 'coincident mode'
or 'low level limiting'. This type of resonator is magnetically saturated by the
biasing external magnetic field. An unlimited sub-harmonic mode absorbs all
the energy above the amount which the main mode can transfer in this condition.
The normal limiting levels for this mode are - 25 to -15 dbm • Because of these
power level limitations, YIG tuned devices are restricted to low power
applications.
7.3 Orientation and Temperature Effects on YIG Resonators

The internal field which adds to or subtracts from the external magnetic fields
is a critical consideration in the design of YIG resonators. The internal
magnetic field is very much influenced by two factors viz., orientation and
temperature.
7.3.1 Orientation Effects

YIG and GaYIG materials, which have cubic crystal structure, have three
principal crystal axes; the (100), the (110) and (111) axes. For a given external
biasing field, Ho, YIG resonant frequency is influenced by given external biasing
field, H o, the YIG resonant frequency is influenced by the direction of these
axes relative to the direction of applied field. For a given YIG sphere, with
(111) 'Easy' axis parallel to Ho,
10= 2.8 (Ho - *!J (13)
with (100) 'hard' axis parallel to Ho,
10 = 2.8 ( Ho + 2 ~.) (14)
where Kl/M. is first order anisotropy in oersted.

If the sphere is rotated about a (110) crystal axis that is perpendicular to Ho,
the value of Ho, for a given resonance at 10 (MHz), is given by,
Ho =-2.8
10 - [2 - -5·
sm 8 - -15·
28M.
sm 2 8 ] -Kl (15)
where 8 is the angle between Ho and that 01(100) axis which becomes parallel
to Ho, as the sphare is rotated about the (110) axis. Thus, when the sphere is
oriented at the ± 270 points,
Ho = {~ MHz (16)
7.3.2 Temperature Effects

Temperature variations offset the resonance frequency due to variations of
K 1/M, with temperature, if the YIG resonator is in any arbitrary orientation
other than the above discussed. 41rMs and t1H of the material also very over
temperature and influence the performance of a YIG device.
8. MICROWAVE DEVICES USING SINGLE CRYSTAL YIG

The development of YIG tuned devices, employing single crystal YIG materials
came into vogue about two decades ago and has since gained considerable
importance due to the phenomenal improvements in the performance capa-

bilities of EW systems and modem electronic Instruments. The various func-
tions, such as electronic tuning, frequency selective limiting, frequency
discrination, signal distribution, signal isolation, swept signal generation
etc., are now being carried out, by these devices, to a high degree of precision
and linearity. YIG tuned solid state oscillators, are used as exciters for trans-
mitters, in EW systems, synthesizers and programmable frequency sources in
digitally tuned microwave receivers, and transmitters, and sweep generators
in microwave instrumentation. Automatic frequency counters, employing
YIG tuned harmonic generator-filter, offer stability and accuracy in freq-
uency measurements with digital read out. Panoramic receivers and wide
dispersion spectrum analyzers, using these YIG devices, show better display
capabilities with elimination of false signal identification and image rejec-
tion. Four of the typical applications of YIG devices in systems are described
below:
8.1 RF Tuners
A typical RF tunci comprises of a YIG tuned filter-oscillator combination driven
by suitable electronic drivers with associated isolators and a mixer preamplifier
IF filter combination. The operation is based on the 'superhet' principle wherein
the input RF signal selected by the YIG filters is down converted to an IF
signal for further processing. A signal command enables both the filter and
oscillator to be driven to appropriate frequencies, with a fixed IF frequency
off-set, throughout the tuning range of the tuner. An instantaneous bandwidth
of about 20 MHz over a large dynamic range and linear tunability over octaves
or multioctave frequencies has made these YIG tuners vital and critical
components for ECM receivers.
8.2 YIG Filter Banks

Multi-channel YIG filter bands, either fixed tunable or tunable are in use in
multichannel preselectors for applications in microwave communication receivers
or channelised receivers. Here, the YIG multicouplers select the frequency pass
band which is channelised to the receiver through PIN diode switches. Each
frequency channel may be switched on or off in nanoseconds in any sequence
or combination for either frequency or time frequency hopping modulated
signals. The YIG filter bank offers, as compared to others types, lesser insertion
loss at minimum volume and weight.
8.3 YIG Tuned Oscillator, Multiplier/Filter

A high spectral purity YIG tuned multiplier/filter for applications as simulators
and general local oscillators offers a multi-octave frequency output, by multiplying
signals from a 2-7 GHz YIG tuned oscillator and passing these signals through
a YIG filter tuned to the desired multiple. The resultant output will be having
a high spectral purity of better than - 50 dB in harmonic and sub-harmonic and
tunable linearity over a wide frequency band.
8.4 Staloc
To overcome mistracking problems due to frequency drift limitations. c1osed-
loop' operation of the octave and multioctave filters have been developed using
a unique tracking technique called "STALOC" (Self-tracking Automatic Lock-
on Circuit). These filters. automatically lock on and track a CW reference
frequency at a predetermined offset frequency. These filters find applications
in active ECM and jammer set-on receivers.
8.S MSW Devices

An exciting technology has emerged in this decade. which offers wide potential
applications. especially signal processing in the UHF and Microwave frequency
bands extending to millimetric waves. This promising technology is based on
magnetostatic wave (MSW) propagation in magnetically biased epitaxial films
of YIG. Some of the important devices are MSW delay lines (Fig. 18) dispersive
lines. wideband tunable filters. narrow and filters. tunable oscillators and signal
to noise enhancers which are fast finding applications. in military radars and
microwave instrumentation.
Magretic Field Ho
Dielectric
substrate
Ground Plane
l:' X
Fig. 18. Basic configuration of a MSW delay line.
9. TRENDS IN MATERIAL TECHNOLOGY

TOday's microwave tunable devices owe much to the phenomenal development
of "MICROWAVE MAGNETICS" and associated technologies. over the years.
Their applications in microwave systems radar. in communication and Defence
applications have enhanced the system capabilities beyond imagination. Direct
signal processing of microwaves extending to millimetric waves and multioctave
frequency capabilities are the results of innovations in single crystal YIG and
allied materials. New materials viz.• hexagonal ferrites [39] and single crystal
lithium ferrite films [40] are now in the forefront to extend the usefulness of
single crystal materials to millimeter wave applications. [41] especially from
18 GHz to 75 GHz.
REFERENCES
1. F. Bertaut and F. Forrat, Compt. Rend. Acad. Sci., 242, 382 (1956).
2. G. Pauthenet, Doctorate Dissertation, University of Paris (1956).
3. S. Geller and M.A. Gilleo, Acta. Cryst., 10,239 (1951).
4. W.H. Grodkiewicz, E.F. Dearborn and L.G. Van Uitert, Cryst. Growth, Ed.
H.S. Peiser, Pergaman Press, N.Y., 1961, p. 441.
5. 1.W. Nielsen and E.F. Dearborn, 1. Phys. Chern. Solids,S, 202 (1958).
6. G.P. Rodrigue "Magnetic materials for millimeter wave applications" IEEE
Transactions on magnetics., Sept. 1963, pp. 351-356.
1. G. Winkler and H. Dotsch, 'Hexagonal Ferrites at Millimeter wave length'
Proceedings of the 9th European Microwave Conference, 1919, pp. 13-22.
8. P. Roschmann, M-Lernke, W. Tolksdorf and F. Welz 'Aristotropy field and
FMC linewidth in single crystal AI, Ga, and Sc substituted Hexagonal ferrites
with M structure. Materials Research Bulletin. 19, 1984, pp. 385-392.
9. W.H. Van Aulock, "Handbook of Microwave Ferrite Materials", Academic
Press, N.Y. (1965).
10. 1. Helszajn, "Principles of Microwave Ferrite Engineering", Wiley-Interscience,
N.Y. (1969).
11. E.A. Giess, B.E. Argyle, D.C. Cronemeyer; E. Klokholm, T.R. McGuire, D.F.
O'kane, T.S. Plasketl and V. Sadagopan, AlP Conference Proceedings Series
NO.5 (17th conference on Magnetism and Magnetic Materials, Chicago (1911).
12. S. Standley, "Oxide Magnetic Materials", Oxford University Press (1969).
13. F.F.Y. Wang in Treatise on Material Science and Technology, Vol. 2., Academic
Press, N.Y. (1973), pp. 213-387.
14. H.1. Van Hook, 1. Amer. Ceram. Soc., 44., 208 (1961).
15. W. Tolkdorf, Acta Electronica, 17,57 (1914).
16. R.A. Laudise, in 'Art and Science of Growing Crystals' Ed. J.J. Gilman, Wiley
Intcrscicnce, N.Y. (1963).
17. R.C. Linares, J. Applied Physics, j~, 433 (1964).
18. M.1. Kestigian, 1. Am. Cram. Soc., 50, 165 (1961).
19. R.A. Laudise and E.D. Kolb, J. Amer, Ceram. Soc., 45, 51 (1962).
20. E.D. Kolb, D.L. Wood, E.G. Spencer and R.A. Laudise, J. Appl. Phys., 38,
1021 (1961).
21. J.E. Pullian, G.R. Archer 1.L. and Besser P.1. 'Magnetic Oxide films' IEEE
Trans on Magnetics MAG-5, pp. 111-21, 1969.
22. 1.0. Adam, Prof. 1.H. Collins and J.M. Owens 'Microwave device application
of epitaxial magnetic garnets'. The Radio and Electronics Engineer Vol. 45,
No. 12, pp. 138-148, Dec. 1915.
23. A.B. Chase and J.A. Osmer, Jour. Cryst. Growth,S, 239 (1969).
24. V.N. Vertoprakhon, V.D. Zamozhskii and P.V. Klertosv, Sov. Phys. Crystalogr.,
13, 113 (1968).
25. S. Geller, G.P. Espinosa and P.B. Grandall, 1. Appl. Cryst., 2, 86 (1969).
26. W.1. Cruft, Amer. MinLTal, 50, 1634 (1965).
27. B. Cockayne, 1. Amer, Ceram. Soc., 49, 204 (1966).
28. K.A. Wickersheim, R.A. Lefever and B.M. Hanking, 1. Chcmp. Phys., 32, 211
(1960).
29. Measurement of the properties of ferrite materials Miss Sp. Maxwell. The Marconi
Review First Quarter 1910, pp. 2-11.
30. C. Kittel; Phys. Review 13, p. ISS, 1948.
31. M. Magid. IEEE Trans., IM-I3-p. 329, 1964.
32. G.R. Harrison and L.R. Hodges (Jr.), J. Appl. Phys. 33, 1375 (1962).
33. W.L. Bond, Rev. Sci. Instr., 22, 344 (1951).
34. J.W. Jeffery, "Methods in X-ray crystallography", Academic Press (1971).
35. M. Sparks, "Ferromagnetic relaxation theory", McGraw-Hill, N.Y. (1964).
36. A.H. Harrish, "The physical principles in magnetism", Wiley, N.Y. (1965).
37. Lax and Hutton, Microwave Ferrite and Ferrimagnetics, N.Y., McGraw-Hill
1962.
38. Mathaii, Magnetically tunable band stop filters' IEEE transactions on Microwave
theory and Techniques. Vol. MTT-13, pp, 203-212, March 1965.
39. Hexagonal Ferrites for Millimeter Wave Application, D.B. Nicholson, Hewlett-
'Packard Journal, 41, Oct 1990, pp. 59-67.
40. Microwave Magnetics and SAW Devices RADC Electromagnetics Sciences
Division, Microwave Journal, Apri11979, pp. 20-21.
41. D.B. Nicholson, R.J. Matreci and M.J. Levernier, 26.5 to 75 GHz, Preselected
mixers based on magnetically tunable barium ferrite filters, Hewlett-Packard
Jour., 41, OcL 1990,49-58.
8
Radome Materials
a.s. Mani
Institute of Annament Technology
Defence Research and Development Organisation
Ministry of Defence. Pune. India
1. INTRODUCTION
1.1 Definition
According to IEEE radome is defined as "an enclosure for protecting an antenna
from normal effects of its physical environment, generally intended to leave
the electrical performance of the antenna unaffected" [1]. In simple terms, a
radome is a housing for an antenna. Because of the reason that radome is
essentially a protective device, it is required to provide necessary structural
strength, but yet, not deteriorate the electromagnetic performance of the antenna
under operational conditions.
1.2 General Information

Since early 1940s, when airborne radomes started finding applications, tremendous
technological advances have been made in the field of radome materials and design
aspects. A wide variety of design ideas have been attempted depending on applica-
tion. A study on the subject of radomes encompasses aspects related to electrical
design, mechanical design, fabrication, analysis, testing and other related areas.
Various authors have contributed to the literature on the subject [2-9]. Microwave
Engineers responsible for developing radome for a specific application should
be aware of the above aspects and their implications. In this chapter, materials
used for radome are discussed with brief reference to these aspects.
1.3 Radome Wall Configurations

An important aspect of radome design is the selection of proper wall configuration.
Fig. 1 shows the configurations that can be used for a radome with uniform
wall thickness.
1.3.1 Monolithic Wall

This consists of a single layer of homogeneous radome material. The thin wall
radome structure implies the wall thickness to be less than about 1/20th of a
Radome Materials 201
RAOOME WALL CONFIGURATIONS
I I
-II-
A/20
--\ t--
12
I
--+l f--
nA/2
m~ m
Fig. 1. Commonly used radome wall configurations.
wavelength in the material corresponding to highest frequency of operation.

Electrically, this can be modelled as a low pass filter and hence radome design
based on thin wall configuration can be useful in some of the broadband
applications [10]. However, this is subject to structural adequacy of the material
in the environment faced by the radome. The configuration can be adapted
only if environmental and strength requirements are lenient. It is observed that
excellent electrical performance at high angles of incidence can be achieved if
the electrical thickness (dIe) of the radome is in the region of 0.02. When thin
wall configuration is feasible, and considered adequate to meet structural and
aerodynamic loads, the radome can have the following features: (a) Light weight;
(b) High electrical transmission coefficient; (c) Broadband frequency coverage;
(d) Insensitive to changes in polarization; (e) Insensitive to variations in angle
of incidence; (0 Small refraction of incident waves; (g) Material parameters
not critical; and (h) Manufacturing methods not critical.
If the wall thickness corresponds to one half wavelength in the material, the
configuration is called half wavelength configuration. Such configuration exhibits
nearly 100% electrical transmission for normal incidence, if the loss tangent of
the material is low. Maximum transmission at any desired angle of incidence
can be achieved by properly choosing the thickness of the wall. For frequencies
between 5 and 25 GHz, the thickness of half wave radome varies in the region
of 15 mm to 3 mm for a material with dielectric constant of 4.0. Since for
many applications, such thickness may provide adequate strength and rigidity,
the half wave radome wall may be considered suitable.
When half wavelength wall configuration is not suitable for any reason, the
wall thickness can be made to be a multiple of )./2. Such configurations can
be a low loss structure at the design frequency. However, the performance of
the structure is quite sensitive to loss tangent of the material.
1.3.2 Multilayer Wall

In spite of the variety of available materials, it is often impossible to design a
satisfactory radome using single homogeneous sheet of dielectric. The problem
may be due to inadequate mechanical strength, stiffness or weather resistance

property of the material, though otherwise electrically suitable. It may happen
that the thin wall design may have insufficient strength, while half wave design
may prove to be excessively heavy or unnecessarily strong or stiff making the
design undesirable for an airborne radar use. Sometimes, design of monolithic
wall configuration may also lead to impracticable machining tolerances not
normally achievable by using available equipment. Under these conditions,
multiple layer configuration is adopted for radome wall.
Some of the simple multilayer structures that can be used in radome wall
design are A-Sandwich, B-Sandwich and C-Sandwich structures. Out of these,
A-Sandwich is most popular, being one of the simplest of the multilayer structures.
This has three alternate layers of high and low dielectric constant materials.
The core will be of low density, low strength and low dielectric constant
material, whereas the skins on either side use high density materials. The main
advantage of the configuration is its high strength to weight ratio. Broadband
radomes can be designed using A-Sandwich design [11]. However, such designs
are more sensitive to variations in incident angle and polarization.
B-Sandwich structure has outer layers of low dielectric constant and a central
core of high dielectric constant material. In general, it is found that such B-Sand-
wich structures have good transmission characteristics over wide range of incident
angles. But unfortunately, such structures are not found suitable from structural
strength considerations, and hence are not very popular as radome designs.
C-Sandwich is a five layer structure, wherein alternate layers of low and
high dielectric constant materials are used. This configuration is observed to
have good structural rigidity and also have good electrical characteristics over
wide frequency range. However, because of complex design and fabricational
aspects involved in such radomes, this has been used only where essential.
1.3.3 Other Radome Wall Configurations

Apart from the above common types, a few other types of radome walls have
also been reported in literature. Conducting wires in the form of reactive loads
have been added to the radome wall structure [12]. These allow more degrees
of freedom in the radome design and can be used to advantage for specific
purposes. A grooved wall structure as shown in Fig. 2 has also been reported
to give better frequency bandwidth with low depolarization [13]. Gradient
dielectric radome wall can also provide broad band properties [14]. Dielectric
Fig. 2. Doubly grooved radome wall.

rings or patches have been used to reduce abberations due to antenna scanning
within the radome [15].
Apart from the above wall configurations, which were essentially made of
dielectric materials, other wall configurations have also been reported. These
are the metal space frame radomes capable of quick assembly at site for ground
applications [16] and perforated metal radome for missile applications. The
latter is reported to have better rain erosion resistance, high strength to weight
ratio and low static charge build-up [17].
1.4 Radome .Design Aspects

Radome design must consider aspects of both mechanical and electrical
requirements of the system. Mechanical considerations include structural stability,
aerodynamics and the environmental conditions the antenna is likely to be
subjected. Electrically, the radome must ensure least distortion in the intrinsic
antenna radiation pattern. Both mechanical and electrical aspects of radome
design allow much flexibility when considered individually. In a realistic design,
however, the high interdependence of the two areas usually cuts the flexibility
drastically. Many trade-offs between competing criteria must be made often
requiring an interactive, multidisciplinary design approach, as shown in Fig. 3.
1.4.1 Aspects of Mechanical Design

In general, the solution to the mechanical design problems lie in proper selection
of radome geometry, materials and fabrication techniques. Radomes installed
on aircrafts are often primary structures whose failure can endanger the aircraft.
In such cases, the structural design of the radome is of paramount importance
and should consider aerodynamic and inertial loads created by aircraft speed
and flight profile, radome shape and sudden changes in motion as may be
encountered during manoeuvres, landing, breaking or in gusty winds. The radome
will be subjected to unsymmetrical pressure patterns due to vehicle manoeuvres,
such as yaw, banking and also due to vibrations. These will result in compressive,
tensile and shear stresses both in the circumferential and longitudinal directions
of an airborne nosecone radome. From typical pressure loading curves, shear
and bending moment at different stations on radome can be computed. These,
together with other forces, enable estimation of the stresses likely to be developed
on the radome surface. The shape, material and fabrication process will be
chosen depending on the values of these stresses.
1.4.2 Aspects of Electrical Design

Major aspects requiring attention in the electrical design of radome are
transmission efficiency, boresight error, bandwidth requirements and reflections
due to radome. In addition, depolarization, emergence of flash lobes, wave
trapping, errors due to radome asymmetry are some of the other electrical effects
which have to be considered for exhaustive electrical analysis of the radome.
In general, proper weightages are given to these parameters and appropriate
parameter index as a functions of these variables is computed. Optimisation of
the variables is then carried out to yield best parameter index. Optimised values
Electrical Aero~~amic Environmental
Requirements Structurol Requirements

Requirements
I
~ ~
SELECT
Primary materials. Relax
Wall struc. ture. specs
Fabrication techniques
STUDY
i) Better
IEvalUate structure I Relax
materials
fo- materials
ii)Wall
Not ok
I I
Over design
specs
structure
Ok
I
Evaluate
VARY Bectrical performance I
I
1 Materials
I-- 11 Wall shape
mAntenna Not ok
interaction Ok
I Manufacture prototype
I
!
I ~
Electrical
te~ts
lOther
tests
Not ok
I a< I Ok[ Not ok
!
Output
Fig. 3. Iterative procedure involved in radome design.
of different variables are then used to fabricate prototype which is then refined
for further optimisation.
Electrical performance of the radome largely depends on range of incident
angles, which the electromagnetic waves make with the radome wall. In addition,
polarization of the antenna enclosed in the radome and the electrical characteristics
namely dielectric constant and loss tangent of the material decide the electrical
performance of the radome.
Radome shape and location of the antenna relative to the radome dictate the
range of incident angles. In the case of ground based or ship-borne radomes,
the shape can be chosen on the basis of optimum electrical performance. In
case of airborne radomes, it is generally optimised on the basis of aerodynamic
considerations. Since shape is an important electrical factor in radome design,
it has been found to be convenient to classify radomes on this basis. A normal-
incidence radome is identified as the one in which most of the area illuminated
by the antenna is approximately normal to the radiated beam of antenna. The
transmission properties in such case can be approximated by those exhibited by
plane dielectric sheets traversed by plane waves normal to the surface. A large
spherical radome, in which the antenna is located at the centre of the sphere
will yield performance of an ideal normal-incidence radome. Streamlined radomes,
as the name implies, have large fineness ratio (ratio of length to base). Some
of the practical shapes used for airborne radome designs are shown in Fig. 4 [3].
t----------A----------t
600 LOG SPRIAL
Fig. 4. Typical streamlined radome shapes.
Travel of electromagnetic wave through a layer of radome wall can be

analysed by using normal laws of reflection and refraction at each interface.
Each layer of homogenous dielectric can be considered to be made of two
interfaces, one where wave enters the dielectric medium and the other where
it leaves the medium. Relevant equations useful for analysing the reflection
and transmission properties of the wall configuration are given in Table 1.
Geometry involved in the analysis is shown in Fig. 5.
Generally, it is observed that reflection for perpendicular polarization (R 1) is
larger than that for parallel polarization (R ll ). For normal incidence Rll and Rl
are equal in magnitude. Rll tends to zero at Brewster angle for low loss materials.
Insertion phase shift is defined as the shift in phase introduced by the material
for a plane wave incident on its surface. This is dependent on the polarization
of the incident wave and the characteristics of the material. Difference in
insertion phaseshift for two orthogonal polarizations is an important factor in
analysing depolarization characteristics due to radome.
Table 1. Equations governing radome electrical design
Incidence on Flat Dielectric Sheet

At Each Interface
rl = [cos 8 - (E r - sin 2 8)1I2] /[cos 8 + (E r - sin 2 8)1/2J
rl1 = [Er cos 8 - (Er - sin 2 8)1/2]/[E r cos 8 + (E r - sin 2 8)1I2]

8s =tan-I [(t;.)II2] for parallel polarisation
Homogenous Flat Sheet

R _ rill - exp [-2j Pod (Er - sin 28)ltlJ)
1 - l-r?exp[-2jP od(E r -sin 28)1I2]
T, _ (1- r?)exp[-j Pod [cos 8- (Er - sin 28)]1I2)

1- 1- r?exp[-2jP od (Er _ sin 28)1I2]
R _ rl1 (1 - exp [-2j Pod (Er - sin 28)1/2J)

11- l-r?lexp[-2jPod(Er-sin28)1I2]
T, _ (1 - r?l) exp [-j Pod [cos 8 - (E r - sin 2 0)]1I2)
11 - 1 _ r?l exp [-2jP od (Er - sin 28)1I2]
Monolithic Radome Wall

Thin Wall Radome
Power tr. coeff., Tl = 1 - [nd (t;. - 1)/Ao cos 0]1
Diff. in insertion phase-shift.1 11 -.11 = nd (t;. - If sinlO/Aot;. cos 0
Multiple Half Wave Radome
Radome Thickness d = n).012 (t;. - sinl 0]112
For normal incidence do = n).rJ2 (t;.)112
Tl = 1 - [nn (t;. + 1) tan c5 /2(t;.) 112]
For haff wave radome n= 1
8a: Brewster angle;

(t;. : ReI. dielectric constant of dielectric sheet; 0: Angle of incidence;
r, R: Ref. coeff; T: Tr. coeff; 1: Perpendicular polarisation; 11: Parallel polarisation)
Equations given in Table 1 can be used to plot the electrical performance

of radome wall configuration of different materials. Typical curves for some of
the wall configurations are shown in Figs. 6 and 7. Such curves help in selection
of material for optimum radome performance.
2. RADOME MATERIALS
2.1 Material Selection

Materials used for fabrication of radomes are primarily selected for adequate
strength and good electrical characteristics. The final application or role of the
radome in system also greatly influences the selection of material. Whereas the
Fig. 5. Plane wave reflection and transmission at a dielectric interface.
temperature typically encountered on the.surface of ground based radomes can

be in the range of - 55 to + 80°C, airborne radomes may be subjected to much
higher temperatures, depending on type of aircraft and flight conditions. Fig. 8
shows the estimated equilibrium temperatures on the surface of a 30° conical
radome for different aircraft speeds and at different altitudes. For missile radomes,
selection of materials should also cater for both thermal shocks and thermal
gradients.
Ground radomes will be required to stand pressures resulting from high
wind speeds. The pressure of air at velocity V impinging at normal incidence
=
on a flat surface is P pV2/2, where p is mass density of air. In general, pressure
distribution will depend on the shape of objects in nearby vicinity. In the
region 0 to 30 m from ground, the variation in pressure can be about 30%,
increasing with height above ground level. In the case of airborne radomes,
one has to consider the flight profile and likely manoeuvres of the aircraft. In
addition, any pressure differentials must also be considered. Underwater radomes
have to be specially designed for these pressure differentials.
Impact strength is another important consideration in the selection of materials.
In particular, the airborne radome design should consider impact loads from
birds, stones, hails etc. If the radome is mounted on the belly of aircraft, the
radome must withstand impact due to sand and dust particles. An important
aspect in the design of radomes of this type is to cater for adequate ground
clearance during landing and take off. Apart from the above, loads due to ice,
8=4
12.5
FREQ 9375MHz I
I
: 120'10.1.
1
I
I
I I I
elO 20",.1.
I I
I
E ~----,/~--------~ 75 I
fJ)
fJ)
10",_ ~eI/ I
I
'"z 10',• .I.
I
I
I
'"
!.! I
...
:J:
I
I
I I ,
-' I I 1201•
-' I I lIel
; 7.5 I 10'101 \
I 5.0 lIell \
I 20'10.1.
I
I PARALLEL POLARISATION------
I PER P£NDICULAR
/10'" lI el
5.0·~--'-;;----TO---'-;;---T:"""--'
0' 2c1 40' O' 10' 20' 30' 4r! 5r! 6r! 7r! 80" 9
ANGLES OF INCIIENCE - ANGLES a= INCIDENCE-
(a) Monolithic AJ2 wall (b) A-Sandwich wall
Fig, 6, Power reflection curves.
snow and hails must also be considered. Shock, vibration and acceleration
levels may form some of the other considerations in selection of materials for
the radome.
Radomes are expected to have large electromagnetic transmission coefficient.
However, this performance may get drastically affected by any moisture absorbed
by the materials forming the radome wall. Dielectric properties of the water
vary in the region 80 to 40 over the frequency region of 1 to 20 GHz. The loss
tangent in the same frequency region vary from about 0.06 to 0.9. Hence, any
possibility of absorption of moisture in the radome wall structure must be
totally avoided. Apart from selecting the materials impervious to absorption of
moisture, it will be advisable to treat the surface of the radome so that rain
water does not stick to surface. If the thickness of water layer over the radome
is comparable to wavelength of electromagnetic radiation, the loss will no
more be negligible. From practical data collected on ground based radomes,
'?adome Materials 209
CORE THICKNESS
lmm / 2mm \
-4~4
//
CORE
/ / /
SKINS OF f=4
CORE OF E = 1.15
100
t 10",1
f 75
oil
~
-"
:5"
l"
o
u
50
Parallel polartsatoon
Pl!rpenchcular
I ! I I I I I
0' 10' 20' 30' 40' 50' &0' 70' 80' 90'
Angles of Incidence _
Fig. 7. Power reflection of double sandwich.
the thickness of water layer formed on its surface can be expressed by

t = 0.48 (QR)O.32, where Q is the rain rate in incheslhour and R the radius of
radome in feet [18]. Losses due to such water films can go upto about 2 dB
depending on frequency. This emphasises the importance of not only avoiding
moisture absorption by the radome materials but also water film formation on
surface of radome.
Radomes being an external hardware on any platform, are likely to be subjected
to lot of natural or man-created abuse. A major source of such abuse of serious
nature arises from corrosion due to chemical fuels or other contaminants with
which the airborne or surface radomes are likely to come in contact. In addition
radomes for naval use will be required to counter corrosion due to sea
environment. Proper material selection to take care of these aspects is essential
in radome design. For airborne radomes, static electricity build-up needs to be
minimised and this can be done by selecting proper treatment of the outer
surface of radome. All coatings used for protection of radome will effect the
--
/ V V -/
d0Jr
400'c
~' I. ~ ~
.-
/ i11r- -
'c Te= TjJ ( 1 + O,174M l

/ il V 1/
T}I = Ambient air temperature / I
r-- r--
1 -r--
on hot day after proper
correction III i/ V
8 / / / I
'§
.~
V / / I
c
o 20<ie / I I /
Q. V V I V
i / / / /
c
i /
'-I--
V VVV
M0- l/ / / I
/ / / /
J
V V ;/
/ / VV
V / '/ /
.
oe
/
l,c L /-- 1--
.I
o 2 3
Free stream mach number_
Fig. 8. Estimated skin temperature vs machnwnber.
electrical performance of the radome, and it is necessary to take their effects

while designing the radome.
Electrical parameters of interest in the selection of radome materials are the
dielectric constant and loss tangent. Best radome performance is obtained by
choosing low loss and low dielectric constant materials. Variation of these
values over the range of frequencies of interest dictate the performance of the
radome's bandwidth. These electrical characteristics together with the mechanical
Strength characteristics discussed earlier form the basis for selection of materials
for radome design.
2.2 Composites
Most of the materials used for radome fall under the category of composite
materials. A working definition of composite materials can be "a material
system composed of mixture of combination of two or more macro constituents
RadomeMaterials 211
differing in form and/or material composition and that are essentially insoluble
in each other" [19].
In principle, composites can be constructed of any combination of two or
more materials. But typical composites are those in which a structura1 constituent
called reinforcement is embedded in a matrix. Because the different constituents
are intermixed, there is always a contiguous region. It may simply be an interface
i.e., a common boundary of the constituents, or a distinct added phase called
an interphase. When such interphase is present, there are two interfaces, one
between each surface of the interphase and its adjoining constituent as shown
in Fig. 9. By suitable combination of the constituents, the composite material
can be designed to have the right combination of desired properties. These
properties may be related to structural strength, temperature characteristics or
the electrical parameters.
INTERPHASE
(BONDING AGENT
FIBER
~J~==+-INTERFACE
Ftg. 9. Composites showing constituents and interphases.
The high strength to weight ratio and the capability of tailoring the mechanical
and electrical properties of the composites m~e them very attractive for
fabrication of radomes. The constituents of the composites are selected based
on the required radome characteristics. Since the radome is required to be
electromagnetically transparent, the electrical conductivity of both the
reinforcement and the matrix selected for the composite have to be extremely
low. In addition they must provide the necessary stiffness to the final composite
structure.
2.2.1 Fiber Reinforcement

One of the most common types of reinforcement used in composites is the
fiber reinforcement. Glass fibers, aramide fibers and ceramic fibers are the type
of fibers generally used in radome fabrication. Out of these glass fibres are
most popular.
Whereas mass glass has a strength of 1000 kg/sq cm, the fibrous glass has
strength of upto" 35,000 kg/sq cm. The stiffness or Young's modulus remain
without much change at around 7 x lOS. Basically, the change in strength comes
due to less defect in the atomic structure of the fibrous glass.
Different types of glass fibers that can be used as reinforcement are shown
in Table 2 [20]. One major problem in characterizing the fibers is that some
of the properties noticeably tensile strength and resistivity may depend on
many conditions other than the glass itself. For instance, glass fibers drawn
singly under clean conditions and tested within minutes of drawing will have
Table 2. Typical properties of glasses (from [20))
Property E-glass S-glass D-glass
Specific gravity 2.54 2.49 2.16

Virgin tensile strength (at 72°P) 3.45 Gpa 4.6 Gpa 2.41 Gpa
Yield strength (at l000 0 P) 0.83 Gpa 1.9Gpa
Modulus of elasticity (at 72°P) 724 Gpa 855 Gpa
Coeff. of linear expansionrp 2.8 x 10-6 1.6 x 10-6 1.7 X 10-6
Elastic elongation 4.8% 5.4% 4.7%
Specific heat 0.192 0.176 0.175
Dielectric constant 6.13 5.21 4.0
Loss tangent .004 .007 .006
considerably higher strength than same type of fiber that has been drawn as a
strand, sized, and woven on a reel and stored for some period. Also characteristics
such as resistivity may depend very much on extraneous factors such as water
content, surface treabnent etc. E-glass is a low alkali composition glass primarily
for electrical applications. It has 52-56% of Si02, 12-16% of Al20 3 and
16-25% ofCaO. MgO and B20 3 vary from 0-6% to 8-13%. S-Glass is a high
tensile strength glass which has significant strength to weight ratio, superior
strength retention at elevated temperature and high fatigue limit. Typical
composition of S-glass is 65% Si02 , 25~ Al20 3 and 10% MgO.
D-glass is an improved dielectric glass developed for high performance
radome application. The dielectric constant of D-glass is around 4.0, compared
to 6.11 for E-glass and 5.2 for S-glass. The loss tangent is also 0.003, compared
to 0.006 for E-glass and 0.007 for S-glass. However, it has lower strength and
substantially lower Young's modulus.
Basic raw material for glass fiber is pure molten glass which is extruded
through a bushing containing large number of holes. The emergent ends of
glass are drawn and sized. Sizing is a process wherein a material is applied on
its surface to fill pores and to modify the surface properties to improve adhesion.
The range of fiber diameter commonly produced vary from 0.003 to 0.02 mm.
Sometimes, conductive additives are added to the fiber to avoid electrostatic
charge build-up. Such additives will deteriorate the performance of radomes.
As reinforcements, glass fibers can be used in the form of rovings, chopped
strands, mats, fabrics and woven rovings.
Aramide fibers under the trade mark of Kevlar is marketed by E.I. Du Pont.
It is an organic compound of carbon, hydrogen, oxygen and nitrogen. The
relative mechanical properties of Kevlar and other reinforcements generally
used in composites is shown in Fig. 10. Kevlar 29 is a low density, high
strength aramide fiber useful for ballistic protection. Kevlar 49 is characterised
by low density, and high tensile strength and modulus. Kevlar 49 is useful in
fabrication of high performance composite applications where light weight,
high strength and stiffness are important. It can also resist shattering upon
impact and inhibit propagation of cracks. Its stress-strain behaviour is linear
upto ultimate failure in tension at 2344 MPa and 1.8% elongation. Its specific
5 15
KEVLAR 29 KEVLAR49
• •
.6
"'0 10 :5
RESIN IMI'REGNATED STRANDS
20
•
S.GLASS
•
HTGRAPHITE
E
15 ",u
•
BORON ~
OTHER ORGANICS
HM GfW'Hrrl' 10
• E GLASS
• STEEL
5
• ALUMINUM
2 3 4 5 6 7
SPECf'lC TENSILE MODUWS.106 in
Fig. 10. Properties of reinforcing fibers.
modulus is about 4 to 5 times higher than of glass fiber. In spite of these

properties, there seems to be some disagreement on its performance under
prolonged conditions. Recent reports have indicated that Kevlar does not perform
well when used on top of surfaces that are exposed substantially to extremes
of weather such as snow and hot sun [21]. Water retention in Kevlar fiber
ranges from about 1 to 4% and because of this reason, extreme care should be
taken during fabrication of electromagnetic windows using Kevlar. In electrical
characteristics, Kevlar compares very well with E-glass fiber. The dielectric
constant and loss tangent are about 3.5 and 0.005, respectively.
2.2.2 Resin Matrix

Selection of resin for use in a particular application depends on a number of
facto~ompatibility with reinforcement, capability to withstand required
temperature and humidity, exposure to ultraviolet rays and usability in the
selected fabrication process including necessary tack and drape. Other parameters
which determine the selection of resin are its viscosity and wettability to filaments,
shrinkage, cure, liberation of volatiles during processing, toxicity and shelf
life. The types of resins usable for radome applications are the polyesters,
epoxies and polyimides.
Polyester resin is the earliest type of resin which has been popular for use
in composites. It is cheap and can be cured under room temperature conditions
with proper promoter. It can be used with glass fiber in wet lay-up process or
by vacuum impregnation to make laminates. Benzoyl peroxide and amine
accelerator can be used to affect room temperature curing. Polyester resin is
useful for applications requiring moderate structural strength.
Epoxy resins are characterised by better mechanical properties including
good adhesive strength and chemical resistance. They are used for low pressure
214 MICRowAvE MATERIALS
laminating upto 100 lb/sq in. requiring cure at 350 OF. The maximum usable
temperature can be upto 420 of under intermittent conditions. However, it is
felt that there is some degradation of properties after long time exposure to
humidity which limits the service temperature to about 275 of. Epoxy resins
are preferred in many applications due to versatility in terms of pot life, curing
conditions, viscosity etc. They are used in filament wound radomes and also
in sandwich constructions.
Polyimides and polybenzimidazoles are useful for high temperature
applications. They have excellent mechanical and electrical properties over
wide temperatures and can be used in applications requiring long term exposure
over elevated temperatures. For high temperature applications condensation-
type polyimides are useful. These resins are however difficult to process and
are also costlier than other types by about 5 to 20 times. Polyimide laminates
can become water absorbent unless properly sealed.
2.2.3 Other Aspects of Composites

Performance behaviour of composites not only depend on the materials
constituting the composite, but also many other factors-mainly the form and
structural arrangement of the constituents and also their interaction. Vast literature
is available giving detailed information on these aspects. A few of the major
factors which affect performance of radomes are discussed briefly.
2.2.3.1 FIBER-REsIN BOND

The permanence of the mechanical properties of composites, especially their
resistance to degradation upon environmental exposure depends on the fiber-
resin bond. Sometimes a coupling agent may be used for forming a good bond
between the reactive group of the resin and the hydroxyl group which may be
formed on the surface of the fiber. A probable explanation is that the coupling
agent forms a coating which is more than a molecular monolayer thick and
prepares an intermediate zone around the fiber with distinct properties that are
beneficial. The ideal coupling agent should provide a low modulus flexible
layer at the interface that will improve adhesive strength of the fiber-resin bond
and reduce the number of voids in the material.
2.2.3.2 FIBER-MATRIX RATIO

Both quantity and arrangement of fibers affect the composite properties. In
general, strength increases directly in relation to amount of reinforcement. The
properties of unidirectional composites can be expressed in terms of volume
fraction.
The suggested expressions for the electrical conductivity and permittivity in
the longitudinal directions are given by mixture rule [22]
O'c = Vr O'r + VrnO'rn Conductivity
Ec = VrEr + Vrn Em Permittivity
where Vr and Vrn are respective fiber and matrix volume fractions (Vr + Vrn =1).
Similarly Young's modulus is given by
Ec =VrEr + Vm Em Young's Modulus

However, practical values of permittivity of a composite can better be
approximated by using empirical fonnula [23]:
log Ec = Vm log Em + Vr log Er
Generally, the fiber strength and stiffness greatly exceeds matrix strength
and stiffness. Hence, longitudinal mechanical properties are almost entirely
governed by the fibers. Compressive strength in the fiber direction is less than
the tensile strength, particularly if the fibers are thin or not very straight. For
glass reinforcements, the compressive strength may vary from 50 to 75% of the
tensile strength. For plastic reinforced with aromatic fibers, however, compressive
strength is generally 20 to 25% of tensile strength, due to low compressive
strength of the fiber itself.
As regards longitudinal thermal expansion coefficient, the phase with higher
modulus tends to impose its expansion coefficient on the final composite property.
It can be approximated as [22]
Linear Thennal Expansion
Strength changes due to causes such as water pick-up will also be governed by
similar equation.
The transverse properties are generally governed by the properties of the
matrix phase. The transverse strength is mostly lower than that of matrix, being
about 50%, but the nature of the fiber matrix interface is particularly important
here. Transverse Young's modulus, transverse conductivity and pennittivity
are also dependent on fiber-volume fraction. The transverse thennal expansion
can be approximated by
Transverse Thennal Expansion
2.2.3.3 SHORT FIBER REINFORCEMENT

In case the fibers are not continuous, the aspect ratio of the fibers affect the
longitudinal properties of the composite. This is because the matrix being less
stiff, cannot transfer stress from fiber to fiber. The stiffer the fibers in relation
to the matrix, the .larger their length to diameter ratio must be if the Young's
modulus is to approach that of continuous composite with the same fiber-
volume fraction. Same is true for electrical properties such as permittivity and
conductivity too; the length of fiber required for satisfactory transfer of flux,
current etc. increases with the ratio of the property between the two phases.
Aspect ratios of the order of few hundreds for most fibers can give strength
close to that obtainable with corresponding continuous fiber composites.
Satisfactory Young's modulus can however be obtained by shorter length fibers.
In case bundles of chopped fibers are used in place of individual short
fibers, aspect ratio of fiber bundle is more important. Strength and stiffness of
such materials are generally not impressive.
2.2.3.4 FIBER ORIENTATIoN

The relationship between strength of the composite and orientation of the
reinforcing fiber is shown in Fig. 11. Continuous parallel strands give the
highest strength range, bidirectional arrangement gives a middle-strength range
and random arrangement gives lowest strength range. Hence by arranging the
orientation of fibers it is possible to obtain unidirectional or isotropic properties.
GLASS fiBERS
I~ ~UNIDIRECTIONAl
IJIIII~
I~I~IL--- EHJ3
I I I1 I±lli
BI DIRECTIONAL
,~i~ ~ PSEUlXJ"'lllOPK:
VOLUME Of" GLASS ..
Fig. 11. Relationship between strength and arrangement of reinforcement.
For a balanced crossply wherein the two sets of fibers are orthogonal to
each other, the Young's mudulus in X and Y directions are equal and has
approximately the average value of the unidirectional composite in the two
directions. However, it has low Young's modulus in the + 45° and - 45°
directions. A laminate with plies in 0°, + 60° and - 60° has nearly the same
Young's modulus in every direction. Same is true for electrical characteristics
such as permittivity or conductivity in different directions of laminates.
2.2.3.5 MULTIDIRECTIONAL WEAVING

Multidirectional woven products approach ideal isotropy in construction, resulting
in isotropic physical and mechanical properties. Applications requiring
multidirectional woven structures usually are those where extremes in temperatures
and highly stressed states are encountered. Very tough fibers such as ceramic
or quartz are only used in such applications. Desired composite properties can
be properly tailored into the structure during the weaving process. Variation in
structural requirements in any of three orthogonal directions can be altered by
properly altering volume fraction of yarn in the direction. The variations can
be from 10 to 60% in the desired direction. In most of the cases, yarn bundles
are used in each direction instead of individual filament or yarn. Blocks are
woven with yarn reinforcements as per requirements which can be used for the
final application. It is also possible to have multiyarn system where different
types of yarns can be used in different directions. Preforms of such hybrid
multidirectional types find applications in space. Schematic of 3D, 4D and 5D

arrays are shown in Fig. 12. The 7D woven blocks can have yarns in 3 orthogonal
directions and 4 diagonal directions.
3-D ORTHOGONAL WEAVE
Fig. 12. Multidimensional woven structures.
2.2.3.6 FABRICATION METHODS

Since the performance of a radome is very much dependent on the type of
process used in its fabrication , it is necessary for the microwave engineer responsible
for radome design to be aware of different fabricational methods and their
relative merits. This is shown in Table 3.
Filament winding technique is a popular method since it lends itself to
automatic equipment. Selective orientation of reinforcing fibers is possible in
this method and hence it offers good design flexibility. Also good control over
resin to glass ratio can be obtained in this method. Because of this, uniform
dielectric constant of the laminated material and good repeatability can be
obtained in this method. By proper adjustment of circumferential and loop
windings, mechanical properties in selective directions can be achieved if required.
Table 3. Manufacturing methods of radomes (from [3])
Radome wall Vacuum Autoclave Pressu(C Filament Matched die Foam-in-

configuration bag bag winding moulding place
A. Solid wall
1. Halfwave 1 2 3 3 4
2. Thin-wall 1 1 3 1 4
B. Multilayer
1. A-Sandwich 1 3 3 3 4 2
2. C-Sandwich 1 3 3 3 4 3
3. Multilayer 1 3 3 4 4 4
1: Most widely used; 2: Frequently used (dictated by type of materials); 3: Little used
(dictated by specialised requirements); 4: Not applicable.
Matched die moulding uses a pair of dies. Fabrication of these dies and their
alignment is an important step, since final dimension and finish depend on
these dies. Matched die moulding can give accurate and repeatable products of
high strength and electrical homogeneity. The technique involves high initial
cost outlay in terms of tools and dies. Modification of moulding tools to change
radome shape or thickness is difficult and expensive. Matched die moulding
technique is suitable for fabrication of radomes based on foam-in-place core or
sharply curved thick honeycomb.
In vacuum bag moulding process, dry glass fabric is wetted with resin and
laid on male mould. A plastic film bag is placed over the lay-up and after proper
sealing, the space between lay-up and plastic bag is evacuated by connecting
to a vacuum source. Atmospheric pressure is utilised in moulding the required
part. Generally the required thickness is built up by laying a number of layers
of reinforcing fabric. It is important to use resin with proper viscosity, so that
the resin and air can be easily squeezed from the lay up. This is a low cost
technique but getting uniform resin content over complete surface requires
skill and practice. This technique is useful only for radomes which are not
subjected to high stresses.
Principle used in autoclave technique is similar to vacuum bag moulding but
it uses much greater pressures. Also the moulding can be carried out under
elevated temperature condition depending on resin requirement. Special autoclave
equipment needs heavy capital investment. Preimpregnated cloth called prepregs
can be used in this technique and can give accurate, repeatable products with
consistent performance.
2.3 Honeycomb Core Materials [24]

Typical core materials used in sandwich radomes have low dielectric constant
of less than 1.5 and loss tangent less than about 0.02. Most common of them
is the non-metallic honeycomb structural material. The basic A-sandwich
construction using honeycomb core is shown in Fig. 13. The honeycomb core
material is made of hexagonal shaped cells. The cell sizes usually used range
Sandwich C onstrucflon
Fig. 13. A-sandwich panel construction with honeycomb core.
from 3/16" to 1/2" with densities ranging from 2 to 10 lb/cft. The core to skin
adhesive rigidly joins the sandwich components and allows the complete unit
to act as single entity with high torsional rigidity. The rigidity obtained by the
sandwich construction can be understood by comparing it with an I-beam. The
facings of sandwich panel acts as flanges -and the core as the web of the 1-
beam. While loading on one of the facings is in th6 compression, the other is
in tension. The core resists the shear load and increases the stiffness of the
structure. In comparison to the web of I-beam, the honeycomb core gives
<;ontinuous support of the facings. Fig. 14 shows the comparison of sandwich
structures with normal monolithic structures.
.E
Monolithic Wall Sand wich Wa ll
1.
I 1+ 1 1 1 1 11 1111 1 1 1~t
1
Relative StiH ness(D) 100 700 3700
Relative Strength 100 350 '25
Relative Weight 100 103 10&
Fig. 14. Sandwich vs monolithic wall-strength comparison.
Glass fabrics, Nomex paper and Kevlar are some of the materials used for
honeycomb core in radome applications. Dielectric constant of different types
of honeycomb as a function of density is shown in Fig. 15. Apart from normal
standard hexagonal cells, over-expanded and flexible cell honeycombs shown
in Fig. 16 are useful in fabricating intricately shaped radome structures. The
Dielectric. constant at 0 incidence angle

1.20,r------r----,------,r----r---~_r_--_r_--__.__,,_-__,
- - Prirotlel polarization
- - - - Perpendicular polarizaf
c .........
.EVI .......... ... .....
u
6 1.12
... .....
-
.!rl
~
L..
Gi 1.08 .._..
0
t04
tOOL...--L_ _- L_ _--1_ _- - l L -_ _.L-_ _-L-_ _...L-_ _.-J

2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
Honeycomb Density - PCF
Fig. 15. Dielectric constant vs honeycomb density.
Hexagonal cor.. ""II Over Expandt!d cor.. c...11 FI ..xible cor ... ""II
Fig. 16. Types of honeycomb core cells.
over-expanded configuration is a hexagonal honeycomb which has been over

expanded in W direction providing a rectangular cell configuration. This facilitates
curving or forming in the L-direction. This process tends to increase W shear
properties when compared to normal hexagonal honeycomb. The flex core cell
configuration provides good formability into compound curvature with reduced
antielastic curvature, and without buckling the cell walls. Higher shear strengths
are possible with these flexible core honeycomb compared to hexagonal
honeycomb structures where curvatures of very tight radii are to be formed.
Typically the Nomex aramide fiber, the main constituent of the non-metallic
honeycomb softens at about 450 OF, and hence the properties drop off rapidly
at this temperature. However, when returned to ambient conditions. the honeycomb
is expected to regain its original strength. Typical properties of honeycomb
strength at higher temperatures is shown in Fig. 17.
An important aspect to be considered during fabrication of sandwich radomes
.....~ 100
----.
" ~"
0
I-
Cl.I
a.~ 80
~ -
....... c
0
E .~ 60
o
....e
I::.
.r: ."," 1/2t0100

...... \ hO~rs exposure
0 0"1 40
...... c 500 hour 5
c ~
~ ..... exposure
III
"- 20
&
o
100 200 300 400 500
Ex~sure and Test tern perature ~F
Fig. 17. Temperature effects on honeycomb.
using honeycomb cores is the need to properly bond the core to the skins. In
order to achieve a good attachment, the adhesive must have the upique qualities
of surface wetting and controlled flow during early stages of cure. This controlled
flow prevents the adhesive from flowing down the cell wall and leaving a low
strength top skin attachment. Adequate attention must also be paid for closing
the sandwich structure at the edges, to avoid moisture pickup through these edges.
Transfer of heat through a sandwich panel is dependent upon the basic
principles of convection, conduction and radiation. Non-metallic honeycomb
cores used for radomes minimise heat flow from the outer skin. When the
structure is to act also as a heat shield, flow of heat can further be reduced by
filling the cells, so that convection within the cells can be eliminated. Special
foams are sometimes used, where possible water accumulation in the cells due
to skin damage must be prevented.
2.4 Foam Materials
2.4.1 Rigid Foam

Rigid foams with closed cells are useful as core material in sandwich radomes.
Depending on structural strength required for the radome, the density of the
foam can be chosen. Low molecular weight poliyols are used with isocyanates,
resulting in high degree of cross linkage for obtaining the high strength foam.
Control of density is done by controlling water added during crosslin king
reaction. High temperature polyurethane foam prepared from TDI and alkyol
triaUycyaniviate copolymer has higher strength [25]. Typical densities used in
radomes are of the order 50 to 200 kg/cu m. Compressive strengths of these
foams are nearly twice that of normal foams. Typical strength and dielectric
constant properties of the foam are given in Fig. 18.
30~------------------------~
I /compressive
I' 1.25
Dielettrif
constan
-
1.20
c:
1.15 gVI
£ c:
2' 0
u
-
CII
u
!;
III
10 1.10 'i:
~
'ii
1.05 C
o 0.5 0.10 0.15 0.20

Density.gms/cu em
Fig. 18. Properties of rigid polyurethane foam.
Intricately shaped structural sandwich radomes can be moulded with foam-

in-place core. In such cases, the foam is built up between the skins in the
required fonn and density. However, extreme care has to be taken to ensure
that foaming action is proper and fills the complete volume in which the
sandwich structure is to be realised. Costly tooling and heavy metallic moulds
are required to withstand high pressures which are likely to be built up during
the foaming process.
Alternatively, the required shape can be moulded from plain sheets by heating
it upto the foam softening temperature. This temperature is a function of the
foam materials and density. Unifonn heating in thennostatically controlled air
circulated oven to attain equilibrium temperature is preferred. The sheet is then
removed from the oven and placed on preheated male metal mould. A female
mould conforming to the outside profile of the core shape is then pressed over
the foam sheet. This technique is called thermo forming and intricate doubly
curved rigid foam cores for sandwich radome can be fabricated by this technique.
Skins are then bonded to the shaped core to make the sandwich structure.
Radome of tangent ogive and other shapes fabricated by this method have
shown good electrical and mechanical performance [26].
2.4.2 Syntactic Foam

Syntactic foams are produced from low density glass microbaloons. The hollow
glass microspheres range in density from 5 to 50 Ib/cft and in size from 0.02
to 0.2 mm diameter. These microbaloons are mixed with proper resin to yield
low density foams used as core. The syntactic foam serves to make the sandwich
structure lightweight, increase its stiffness and improve crack resistance. Typical
dielectric constant of this type of foam range from 1.5 to 2.5, densities from
0.5 to 1 g/cu cm and loss tangent from 0.005 to 0.01. Syntactic foam can have
more ordered cells so that sandwiches with thinner skins can be built improving
the electrical efficiency [27]. Both additive and condensation polymides can be
used in conjunction with syntactic foam to build high performance radomes.
Unpublished reports have shown good phase characteristics for radomes built
using syntactic foam.
2.S Materials for Ground Based Radomes

Ground based radomes enclosing large antennas would normally be spherical
or part-spherical structures. The structures can either be air supported or can
be rigid. The air support inflatable radomes can be lightweight and designed
for mobile stations. They will require continuous air pressure to maintain support
of the radomes. The risk factor for these radomes is rather high since the
structure can collapse due to accidental fabric failure or due to abnormal wind
or snow conditions. The rigid radomes are generally made of panels joined
together by means of metallic or dielectric members. For easy fabrication, all
the panels are designed to be identical and can be either flat or curved. The
panels can be of thin wall, halfwave or sandwich configuration.
For air supported radomes, thin rubberised fabrics or thin plastics such as
dacron can be used. Thin low loss glass fabrics coated on both sides with teflon
can also be utilised. These materials have dielectric constants of the order 3 to
3.5 and loss tangents of about 0.01. Typical thickness range from 0.5 to 1.5
mm. Outer coating of hypalon or tedlar is used to give better weather resistance
characteristics. RA YDEL is a composite fabric of high modulus reinforcing
yams, coupled with a matrix based on perfluoropolymer resin [28]. It is a
registered trademark product of Chemical Fabrics Corporation and has been
used for air supported and pretensional radome designs. With a dielectric constant
of 2.7 and loss tangent of 0.005 at·24 GHz, it can be used to design low loss
broadband ground based radomes. The special resin used in its formulation
gives it the capacity to perform well even in adverse rain conditions.
Stretched polypropylene film material has dielectric constant of 2.3 and loss
tangent of 0.003 and can be used for making homogeneous tuned wall radomes
[29]. It has excellent fragment-defeat properties, is self bonding under heat and
pressure, and can provide high level of ballistic protection against fragmentation
munitions. In order to maintain high ballistic properties, the film panels must
be constructed free from entrapped air and moisture with uniform interply and
proper polymer orientation. Polyurethane foam glass fiber laminate panels with
suitable flange attachment can also be used as panels. Where sandwich structures
are essential from strength consideration, A-Sandwich with proper cores are
used. Wherever metallic supports for membranes are used, the design has to
take care of the effects due to aperture blockage.
2.6 Materials for Missile Radomes
2.6.1 General Considerations

Seeker missiles are the precision guided weapons which have enormous potential
in modern warfare. The fully autonomous active seeker has many operational
advantages but presents great design challenges, compared to semi-active or
passive types. Fig. 19 shows a simplified block diagram of seeker electronics.

Highly integrated transmitter-receiver modules having small size and weight
are sometimes gimbally mounted along with the antenna inside the seeker
RADOME
SIGNAL GUIDANCE
RECEIVER t - -...... PROCESSO
ANSMI
POWER PRIME FOWER
SUPPLY
Fig. 19. Typical missile seeker head electronics.
head. The output is the guidance data which is fed to the autopilot for guiding
the missile path.
Sensor antenna and radome form an important subsystem of the missile
seeker. Compared to the microwave antenna, millimetric wave sensors offer
better resolution and adverse weather capacity. Millimetric wave sensors are
also being considered for re-entry interceptors for exo-atmospheric operations.
Seekers operating on radiometric principle require wider rf bandwidth sensors.
For better target discrimination, multi-mode seekers are used where information
from more than one sensor is used for guidance. In all cases, radome design
must be compatible with sensors.
2.6.2 Radome Requirements

Radome for missiles must satisfy electrical, mechanical and aerodynamic
requirements. Halfwave radome with hemispherical dome offers good electrical
performance, but the aerodynamic drag may not be acceptable. Halfwave radome
may cater for 5 to 20% frequency bandwidth, depending on material dielectric
constant and fineness ratio. Broadband antiradiation homer antenna will require
multilayer radome wall.
Amplitude distortion caused by radome material results in transmission loss
and subsequent degradation of system sensitivity. Phase distortion introduced
by radome causes error in boresight and boresight slope. Proportional navigational
guidance used in some missiles is sensitive to target position error slope and
any error introduced in boresight slope by the radome will affect the overall
performance. General guideline for acceptable values for missile seeker radomes
are transmission loss of 1 dB, boresight error of 1.0 milliradian and boresight
error slope of 0.05 deg/deg. Frequency modulation is sometimes preferred to
conventional pulse modulation in missile seekers since it offers high average
power and good short range performance. In such cases, transmitter-receiver
isolation and front-end VSWR become important design demands. Reflections
from radome must be minimised in such cases for all antenna scan angles.
Seekers employing common aperture for dual mode sensors present many
challenges in radome design. In such cases, the radome design will be required
to be optimised at several wavelengths which may be widely separated.
2.6.3 Materials
Conventional resin glass composites can be used only for low speed missiles,
which do not demand operation in high temperatures. For speeds of the order
of Mach 3 or above, it is necessary to use inorganic materials based on silicon,
glass, or ceramics. Apart from capability to withstand high temperature without
deterioration in performance, these materials can be designed to provide adequate
12
11
10
..... 9
c
E
VI
c
0
u 8
U
.....
L..
u
.!! 7
Q.I
0
Pyroceram 9606
6
Raycerom III
5
3
0 SOO 1000 1500 2000 2500
Temperature
0.010
..... 0.008
c
f
VI
VI
0.006
0.004
0
...J 0.002
0
0 500 1000 1S00 2000 2500
Temperature
Fig. 20. Electrical characteristics of some inorganic materials.
protection against rain erosion. Electrical properties of some of the inorganic

materials useful in radome fabrication are shown in Fig. 20.
2.6.3.1 SOUD WAll. RADoME MATERIALS

Two important types of materials which can be used for solid wall high
temperature radomes are 99+% pure alumina and glass ceramics. Pyroceram
9606 is a glass ceramic available from Coming Glass Co. useful as a high
temperature missile radome material. Some of the main features of alumina
and glass ceramic materials are shown in Table 4. Whereas dielectric constant
Table 4. Typical properties of alumina and glllss ceramic materials
Alumina Glass ceramic

(Pyroceram 9606)
Specific gravity 3.7 to 3.9 g/cu cm 2.62 g/cu cm
Variation over 25 to 500"C
Dielectric constant 9.6 to 10.3 remains fairly constant at 5.7
Loss tangent 0.0001 to 0.0005 0.0002 to 0.001
Flexural strength 270-250 x 106 N/m2 235-200 x 106 N/m2
Young's modulus 380-350 x 109 N/m2 120 N/m2
of alumina is about 9.6, that of glass ceramic is 5.65. Hence for halfwave
design, physical thickness of glass ceramic radomes is larger than of alumina.
Due to increased physical thickness, its mechanical properties are adequate in
most cases, even though some of the mechanical strength characteristics of
glass ceramic materials are poorer. Also, though density of glass ceramic is
less, it has only marginal weight advantage over similar alumina design. Both
materials being hard, can withstand rain erosior) and thermal shocks well. For
alumina, thermal differentials in excess of 400°C can result in failure. Above
500°C, mechanical strength of pyroceram deteriorates more drastically than of
alumina. In its pure form, permittivity of alumina increases from 9.6 at room
temperature to 11.4 at lOOO°C. By adding titanate, it can remain nearly constant
at about 11.6 over complete temperature range.
2.6.3.2 SIUCA BASED RADOME MATERIALS

Fused silica formed by sintering slip casting process is the less dense form of
silica. At a specific density of 2.2 g/cu cm, its dielectric constant value of 3.5
remains fairly constant over large temperature range. However, being porus in
nature, methods of preventing water absorption have to be incorporated in making
radomes out of fused silica. The material has also poor rain erosion resistance.
Silicon nitride has excellent strength and erosion resistance characteristics.
The dielectric constant is dependent on the density of the material varying
from about 8 for a density of 3.2 g/cu cm to 5.7 for a density of 2.4 g/cu cm.
Reaction sintered silicon nitride can be prepared to give much lower densities.
Gas evolution taking place during this process allows formation of porus
structures. The amount of porosity, and thus dielectric constant is dependant on
composition and process parameters. However, a hard impervious coating is
required for these low density material to provide resistance to water absorption
and to sand/rain erosion. The loss tangent can generally be kept less than 0.005
in all cases. It is observed that variations in dielectric constant with temperature
in silicon nitride are intrinsic in nature and little improvement can be achieved
by process variation. On the other hand, the loss is predominantly controlled
by intergranular phases and impurities. Monolithic radomes using reaction bonded
silicon nitride can be used for narrow band applications.
By having a two layer structure of low or medium density silicon nitride
core as a thickbase, with a thin skin of higher density silicon nitride, it is
possible to achieve acceptable radome performance over a wide range of
frequencies. The high density outer skin can provide erosion resistance and can
also form a suitable base for moisture sealent.
Chemical vapour deposited silicon nitride (CVDSN) has the ~ntial as a
multiple frequency window material and can be used in conjunction with sensors
operating at different wavelengths.
2.6.4 Frangible Glass as Radome Material

Frangible materials, as referred here, are those which shatter into a multitude
of small fragments in precisely predictable fashion. Chemically strengthened
glasses and glass ceramics can be designed to break with very small but
concentrated amount of energy. One of the methods by which frangibility of
materials can be exploited in the field of missiles is by employing a frangible
dome over a dielectric hemispherical radome. The outer frangible dome can be
aerodynamically shaped so that it can protect the inner parts of the missile
from the rigours of atmospheric flight in lower atmosphere. Then, as the missile
reaches less dense regions, the frangible dome would be triggered to be blown
off, exposing the simple low dielectric' hemispherical dome. Being electrically
superior, the hemisphere can perform much better and guide the missile to the
target. Thus the frangible dome could function as a protective cover for the
electrically superior radome of low mechanical strength. By properly choosing
the glass ceramic material for the outer dome it can be designed to not only
give protection to the inner radome, but also enable a strong signal to guide the
missile on its initial flight path.
2.7 Materials for Simple Electromagnetic Windows

Sometimes, simple rf transparent window will be required to protect the antenna,
with no major structural requirements or aerodynamic loads. Thermal plastics
can be used as EM windows for such applications. Typical thermal plastic
materials that are available commercially are given in Table 5, along with their
main electrical characteristics. These materials are generally used in thin wall
configuration. Thermal plastics can be moulded into the required shape by
vacuum forming or injection moulding and this offers economically viable
solution when large quantities are involved.
Noryl is a polyphenylene oxide and is useful for small ECM antennas such
as cavity backed Archemedian spirals. Since thin wall configuration is used,
the axial ratio of the original antenna can be preserved. Open ended waveguides
in large phased arrays can be protected by having Noryl protection caps. Teflon
Table 5. Some thermal plastics useful as EM windows

Manufacturer Dielectric Loss tangent
constant
NORYL General electric 2;58 .005

(Polyphenylene oxide)
KYDOX (PVC Acrylic) Rohn and Hass 3.44 .008
TEFLON E.I. Dupont 2.1 .0005
(Tetrafluoro ethylene)
and Lexan are other thermoplastics which can be used in the form of thin flat
sheets or in the contour conforming to the platform. The dielectric constant of
Teflon is so low, that larger thickness of protective cover can be used compared
to glass laminates. It can be used even at subsonic speeds, if certain amount of
erosion can be tolerated.
2.8 PROTECTIVE FILMS AND COATINGS FOR RADOMES

In order to protect radomes in certain operational environments, it is advisable
to use protective films or coatings. Some of the important coatings of relevance
for radomes are for protection against erosion due to rain, sand and dust,
moisture pick-up, static electricity build-up, ultraviolet radiations etc.
2.8.1 Protection Against Rain and Sand Erosion

To combat erosion due to rain, and sand, neoprene coating of about 0.015"
(0.4 mm) thickness on radomes have been used in the past. The dielectric constant
of Neoprene is about 2.8 and its effect must be accounted for in predicting the
radome performance. Under prolonged heavy rain conditions at high speeds,
radomes with these coatings start eroding. Hence such radomes require periodic
recoats.
Polyurethane is another useful coating material which can be used for reducing
erosion due to rain. These are created by reacting isocyanatic groups containing
an active hydrogen, such as primary or secondary amines or other methanes.
Such a coat used as top coat on a three layer coating system could have 5 to
8 years of service. It is generally used over an epoxy, polyamide or polysulfide
primer. Special coating and curing procedures are generally recommended for
these coatings. Successive coats should be applied when the previous coat is
set-to-touch but not tack-free. Polyurethane coatings are about 4 times tougher
than neoprene coating systems [30].
Long exposure to ultraviolet radiations can cause deterioration in case of
ground radomes, especially if they are made of foam. A white protective coating
of epoxy or acrylic loaded with titanium dioxide can give protection against
these radiations.
3. EFFECTS OF MATERIAL VARIABLES AND TOLERANCES

For proper engineering design, it is essential to have a knowledge of the effects
of variables and tolerances on the final performance of the radome. A few of
them having direct bearing on the electrical aspects of radome are discussed
briefly here.
3.1 Errors in Manufacturing Thickness

Theoretical power reflection percentages for different errors in thickness for a
halfwave radome sheet is shown in. Fig. 21. These curves are for a flat monolithic
dielectric sheet of·dielectric constant 4.0 designed to operate 9375 MHz. An
error in thickness of 0.0005" results in 1% power reflection at 6()0 angle of
20".
POLARISATION: PERPENDICULAR
DIELECTRIC CONST .4.0
FREQUENCY :9375 MHz
z
o
;::
til
...
...J
.u
a:
a:
IIJ
3: 10 .,.
~
o~~~-----===t==---------L---------~~~
o 0.25 0.50 0.75
ERROR IN THICKNESS (mm)_
Fig. 21. Effect of solid laminate thickness errors on power reflection.
incidence; this value rises to 8% at 80°. For a corresponding sheet at Q band

(37.5 GHz) these values are 5% and 28% at 60° and 80°. Need for control of
wall thickness for streamlined radomes having large angles of incidence is
obvious. Even for sandwich radomes, small errors in thickness of skins or core
are not often serious at low angles of incidence. But if the shape of the radome
is such that high angles of incidence are likely to be encountered, strict tolerances
on sheet thickness is essential.
3.2 Thickness Change Due to Erosion

Sometimes, eventhough tight quality control measures have been imposed to
ensure proper wall thickness, deterioration in performance over a period of
time may be noticed due to erosion of the radome wall. Airborne radomes
flying at supersonic speeds in heavy rain are subjected to erosion, and subsequent
poor electrical performance. A change of the order of 0.005" is not unreasonable
for a neoprene coated radome. In such cases, power reflection coefficient of
5% at X-band and upto 20% at Q band for 80° radome can be expected. Periodic
maintenance with recoating is the possible solution for overcoming this.
3.3 Thermal Expansion or Radome

Radome thickness can vary due to the thermal expansion. It is estimated that
a change of about 100°C normally encountered during a supersonic flight can
result in about 4% rise in reflection of power at X-band. More serious problem,
than the increase in wall thickness due to thermal expansion, is the differential
expansion between materials used in different parts of a radome. Stiffener rings,
ablation boots and similar parts of a radome are generally made of ~etal alloys
whose thermal expansion coefficient are different from materials used in radome
fabricaton. Careful engineering design is required for countering such problems.
3.4 Effects Due to Variation in Dielectric Constant

Dielectric constant is the single most important parameter to affect the radome
electrical performance. Hence this -is required to be well controlled during the
manufacture and the life-time of radome. Fig. 22 gives some idea of the effect
of variation in dielectric constant on a resonant halfwave radome. The change
in the value of dielectric constant of radome wall can be either due to processes
80
t 70
....
~
:... 60
z
Q
~ 50
~
11)
z 40
c(
a: a:
to- L!J
ffi 30 ~ 60
~
~
2 20~~ ______~______~_____
8.400 8.775 9.150 8.400 8.775 8.1S(
FREQUENCY(GHz) --....,..
~ FREQUENCY(GHz)-
Fig. 12. Effect of variation in dielectric constant on radome performance.
or materials involved during fabrication, or can be due to external environmental

effects. Strict control over the characteristics of resins, reinforcement and volume
ratios during the fabrication of composites is essential to maintain the final
electrical performance. Parameters such as fiber orientation which have been
discussed earlier are also important.
The curing procedure involved in the fabrication of radome requires application

of simultaneous heat and pressure in proper sequence. Whereas heat is used to
facilitate and control chemical reaction, pressure is applied to squeeze any
excess resin and to minimize void content. Selection of proper curing process
depends on materials involved. Various related aspects such as resin flow,
pressure distribution during curing, compaction and formation of voids have
Cure Pre ssure (psig) Cure Pressure (p sig)

8055 20 0 10 80 55 20 0
III
~200
:z: T 300/978
·iii 8
t5
~
~
ffi 150 :z:
t; 6
a::
>- z
<f'J w
W 100 a::
2: >-
<f'J
<f'J
lfl<f'J
IF.
w
g: 50
T300/976
2r ~~;;:'io DATA
~
o O~_-,--I_-l-_~
u
Ol:----:!--~----,J.
o 5 10 15
o 5 10 15 VOid Content (0'0)
Void Content (0/0}
~ 8,---------,
..,
.l<
!2 6
><
34-
::;)
c
~ 21-
mO'-----.JI'--___,JI_~
VI
o 5 10 15
Void Content ('/,)
15.----------.
\ T 300/976
b 2.4psio "t:J
C
.8
~ ~
::!
It..
10 ~4.30
.... VI
z
Z
....
III 8 4.20
Z !::1
o II::
u .... 4.10
c
(5
5 II!
-'
> ~ '.00
I DATA
-MODEL
O~---~--~ 2 3 5
o 50
Void Content ('/, by volume)
CURE PRE SSURE (psig)
Fig. 23. Relationship between mechanical properties and curing parameters.

been investigated in detail through suitable m~el studies [31, 32]. Experimental
measurements to relate the void content with cure pressure are done through
photomicrographic analysis [33]. The relationship between cure pressure, void
content and compression strength for a thermoset resin composite system is
given in Fig. 23. Factors during operation, such as temperature variations and
any moisture absorption would result in change of dielectric constant If moisture
content is high, pressures produced by heated vapours of air trapped in the
voids can cause variations in its electrical characteristics, as well as result in
physical change. This aspect must be viewed seriously if sandwich radomes
using honeycomb core is subjected to large temperature variations.
3.5 Minimising Radome Errors

Though some of the errors pointed out earlier cannot be totally eliminated, it
is essential that adequate precautions are taken at all stages including selection
of materials, design, fabrication and operational maintenance.
At the material selection stage, it is advisable to select proper compatible
resins and fibers taking into account the strength requirements. It is essential
to have a perspective idea of the environmental and other stresses to which the
radome is likely to be subjected during its life-time. Errors due to irregular
properties of basic radome materials can be avoided by checking -these values
just before use. Apart from mere selection of materials, attention must be paid
to proper control of fabrication and curing methods to ensure homogeneity,
proper bonding and to minimise voids. Matched die moulding with heavy
pressures can keep dimensional thickness errors during manufacture within
tolerable limits. Use of woven socks or preforms of required shape can eliminate
increase in thickness at joints or qverlaps.
To avoid ingress of moisture, proper sealing at the edges is essential. The
humidity problem might be minimised by keeping the radome covered while
not in use. For airborne radomes, this will leave the radome surface dry and
free of surface moisture before take off. It would be advisable to inspect visually
for any damage or erosion periodically, so that corrective measures like repair
or total replacement can be taken before catastrophic failure takes place.
4. RADOME MATERIAL MEASUREMENTS

Standard measurement techniques which are applicable for characterising
mechanical and electrical properties of materials are also directly applicable
for radome materials. Excellent review of electrical characterisation of materials
with lots of reference is given in literature [34]. There are however certain
specific types of measurements which are relevant to radomes during fabrication
and development and these are dealt briefly in the following paragraphs.
4.1 Flat Panel Tests

It is advisable that properties of all constituent materials are measured before
fabrication of radome. A proper record of these results will help in analysing
any malfunction of radome subsequently. For aiding this, it is common practice
to make flat panels using the same materials under the same working conditions
as are used for actual radome fabrication. All other parameters are also kept
identical so that the flat panels truly characterise the radome wall.
Electrically, the flat panel is tested for transmission loss and insertion phase
by following free space method. A typical set-up wherein measurements can
be carried out using a network analyzer is shown in Fig. 24. Corrections to
accommodate effects due to multiple reflections can be incorporated to improve
accuracy [35]. By rotating the sample panel, transmission coefficient can be
Test Panel
(>30 lC 3D)
-
Adjustable
e,
~
1
0
*
ill 30 •
Reference
PhOse Network
Locked Analyzer
Source
Fig. 24. Free space panel measurement set-up.
obtained as a function of incident angle. Curve fitting can be used to estimate

value of dielectric constant within an error of 0.004 to 0.009 and loss tangent
within 0.00003 [36].
4.2 Pattern Comparison Method

The ultimate purpose of radome is to ensure minimum distortion to antenna
radiation pattern. Hence one of the practical methods of checking the performance
of radome is to plot radiation characteristics of the antenna with and without the
radome. By noting the differences in major features of the two radiation patterns,
effects such as loss in gain, beam broadening, boresight shift, depolarisation
due to radome can be studied and the radome accepted for qualification. Though
lot of automation can be built in, the basic set-up for the measurement is shown
in Fig. 25. Reflections caused by the radome, can also be checked by following
the principle of substitution. These set-ups can also be used to optimise the
location of the antenna within the radome for certain critical applications.
4.3 Perrormance Tests as Userul Production Tools

During the production of highly sophisticated radomes of complex design, it is
Fig. 25. Pattern comparison method for evaluating radome.
necessary to verify the correctness of the steps followed in the fabrication. Any
manufacturing error due to oversight or fall in quality of workmanship may
prove to be costly to be rectified when detected at the final stage. Even after
production, it is worth checking the performance of the radome using some
gauges prior to shipment and installation with antenna assembly. These gadgets
are normally evolved from simple original concepts to suit the specific
requirements.
4.3.1 Concept Based on Loading of Resonant Circuit

Two instruments called Osculating Cavity Dielectrometer (OCD) and Osculating
Electrode Dielectrometer (OED) were developed to permit monitoring of
fabrication for E-3A AWACS radomes [37]. Since the radome design involved
C-sandwich with step-tapered glassply lay-ups, and machine tapered honeycomb
cores, precision monitoring of wall thickness was required. Both OCD and
OED use rf oscillators with rf resonant circuits. The basic principles of OCD
and OED are shown in Fig. 26. OED uses interdigital ~lectrodes, and the OCD
a 1" diameter antenna aperture through which the rf interacts with the radome
INTER DIGITAL
ELECTRODES
OSCULATING ELECTRODE DIELECTROMETER
10dB PADS
r-:OC==D~O==S::""I"""LL'-'/Ii.""';rOR~
~---I &POWER SUPPLY 1H DIA APERTURE
(AVANTEIQ
OSCULATING CAVITY DIELECTROMETER
Fig. 26. OeD/OED equipment set-up.

skin. The oscillator frequency shift is directly related to electrical thickness of

the material. The OED is placed closed to radome surface and conforms with
radome curvature, whereas the OCD is spaced approximately 1/8" above the
radome surface. OED is more sensitive to radome surface roughness, but less
sensitive to radome curvature. Tests are carried out at different locations on the
radome surface to check uniformity.
4.3.2 Concepts Based on Reflections from Radome

Microwave s~thoscope based on reflection principle is used to monitor reflections
of power as shown in Fig. 27. The reflected power over the test frequency band
can be plotted as dB below 100% reflection and can serve to monitor the elec-
trical thickness of radome. Comparison made with theoretical reflections or
NETWORK
~--+ ANALYZER
PHASE
MAGNITUDE
DISPlAY
COAX
LOW-REfLECTION
HORN
HIGH-DIRECTIVITY
WIG DIRECTIONAL COUPLERS
IREC~;D",II----------IN~~~:~~ZEER
x- y
RECORDER
AMPliTUDE PHASE
Fig. 27. Microwave hom stethoscope.
calibrated panels can qualify the radome. The same principle can also be used
for assessing the quality of anti-static coatings and also any defects in radome
wall such as moisture pick-up or wall thickness variations due to erosion [38].
A simple gauge based on same principle uses short circuit stubs and shaped
grooves on its flange to make measurements more reliable and accurate [39].
A flexible conducting foil is held in intimate contact with the internal surface
of radome and acts as a back plate of the cavity. The gauge has been found to
be useful for pre-shipment checks.
S. USEFUL SPECIFICATIONS, STANDARDS, TEST METHODS

AND DATABASES
Some of the specifications, standards and test methods connected with radome
materials and related aspects are as follows:
5.1 Specifications
5.1.1 Materials
L-P-383 Plastic Material. Polyester Resin, Glass Fiber Base, Low
Pressure Laminated
MIL-C-7439 Coating System, Elastomeric, Rain Erosion Resistant,
and Rain Erosion Resistant with Anti-Static Treatment,
for Exterior Aircraft and Missile Plastic Parts
MIL-R-7575 Resin, Polyester, Low Pressure Laminating
MIL-C--8073 Core Material, Plastic Honeycomb, Laminated Glass
Fiber Base, for Aircraft Structural Applications
MIL-C--8087 Core Material, Foamed in-Place Polyester Disocyanate
Type, Interchangeability and Replaceability of Compo-
nent Parts for Aircraft and Missile
MIL-C-9084 Cloth, Glass, Finished, for Polyester Resin Laminates
MIL-R-9299 Resin, Phenolic, Laminating
MIL-R-9300 Resin, Epoxy, Low Pressure Laminating
MIL-R-25042 Resin, Polyester, High Temperature Resistant, Low
Pressure Laminating
MIL-P-25395 Plastic Materials, Heat Resistant, Low Pressure, Lami-
nated Glass Fiber Base, Polyester Resin
MIL-P-25421 Plastic Materials, Glass Fiber Base-Epoxy Resin, Low
Pressure Laminated
MIL-R-25506 Resin, Silicone, Low Pressure Laminating
MIL-P-25515 Plastic Materials, Phenolic Resin, Glass Fiber Base,
Laminated
MIL-P-25518 Plastic Materials, Silicone; Resin, Glass Fiber Base, Low
Pressure Laminated
MIL-C-27315 Coating Systems, Elastomeric, Thermally Reflective and
Rain Erosion Resistant
MIL-C--81773 Coating, Polyurethane, Aliphatic Weather Resistant
MIL-C--83231 Coatings, Polyurethene, Rain Erosion, Resistant for
Exterior Aircraft and Missile Plastic Parts
MIL-C--83286 Coating, Urethane, Aliphatic Isocyanate, for Aerospace
Applications
MIL-Y--83370 Yam, Roving and Cloth, High Modulus, Organic Fiber
5.J.2 Construction and Other Aspects

MIL-A--8860 Airplane Strength and Rigidity, General Specification for
MIL-A--8869 Airplane Strength and Rigidity, Special Weapons Effects
MIL-B-5087 Bonding, Electrical and Lightning Protection, for
Aerospace Systems
MIL-M--8856 Missiles, Guided: Strength and Rigidity: General
Specification for
MIL-P-9400 Plastic Laminate Materials and Sandwich Construction,
Glass Fiber Base, Low Pressure Aircraft Structural,
Process Specification Requirements
MIL-R-7705B Radomes, General, Specifications
MIL-S-9041 Sandwich Construction, Plastic Resin, Glass Fabric Base,
Laminated Facings and Honeycomb Core for Aircraft
Structural Applications
MIL-S-25392 Sandwich Consb'Uction, Plastic Resin, Glass Fabric Base,

Laminated Facings and Polyurethane Foamed-in-Place
Core for Aircraft Structural Applications
5.2 Standards and Test Methods

FED-STD-406 Plastics, methods of testing
MIL-STD-401 Sandwich consb'Uctions and core materials; general test
methods
MIL-STD--810 Environmental test methods
ARTC-4 Electrical test procedures for Radomes and Radome
materials
ASTMD 3039-76 Test method for tensile properties of fiber-resin
composites
ASTM D 790-84 Test method for flexural properties of Wlreinforced and
reinforced plastics and electrical insulating material.
ASTM D 3355-74 Test method for fiber content of unidirectional fiber-
resin composites by electrical resistivity
ANSI/ASTM-D-234<h1!4 Standard test method for apparent interlaminar shear
strength of parallel fiber composites by short beam
method.
5.3 Other Information [40)

PLASPE An engineering and marketing database from 'plastics
technology' magazine: can be accessed by subscribers
using a terminal and modem via telephone
HIMP High temperature material properties data bank operated
by Centre for Information and numerical data analysis
and synthesis (CINDAS) of Purdue University, in
conjunction with US Dept. of Defence.
EPOS Engineering plastics on screen-A data base on mecha-
nical, thermal, electrical and other properties of injection
mouldable materials from ICI, UK.
POLYFACT An on-line menu driven interactively formatted database
provided by E.I. Dupont.
REFERENCES
1. IEEE standard 145-1983, Institute of Electrical and Electronics Engineers Inc,

345, East 47th Street, New York.
2. T.E. Tic~ (ed) "Techniques of Airborne Radome Design", AFAL--TR-66-391,
Vols. 1 and 2, Dec. 1966.
3. J.D. Walton, Jr (ed) "Radome Engineering Handbook Design and Principles",
New York: Marcel Dekker, 1976.
4. W. Cady, M. Karelitz, and L. Turner, "Radar Scanners and Radomes.", New
York: McGraw-Hill, 1948.
5. A.W. Rudge, K. Milne, A.D. OJiver and P. Knight, "Handbook of antenna
design", (Ch. 14, Vol 2). London: Peter Peregrinus, 1983.
6. R.C. Johnson and H. Jasik, in "Antenna Engineering Handbook", (Ch. 44). New
York: McGraw-Hill, 1984.
7. M.I. Skolnik. "Radar Handbook". (Ch 14).• New York: McGraw-Hili. 1970.
8. Proc. Symposia on Electtomagnetic Windows. Georgia Inst. of Technology.
Atlanta-Held on number of years starting from 1955.
9. International Conferences on Electromagnetic Windows. Paris-Held on number
of Years starting from 1967.
10. G.S. Mani et al. "Electrical Design Report of 14 GHz Radome". DLRL report
no. DLRL. : R : 86 : 020 (Restricted). March 1986.
11. G .S. Mani and Ramsingh. "Design and fabrication of radomes for a broadband
spiral antenna". DLRL technical notes (Restricted). Vol 2. March 1981.
12. R.H. Cary. "Some novel techniques for avoiding obscurations". lEE Conf.
Publication NO. ISS. 1971.
13. E.L. Rope and G.P. Tricoles. "Anisotropic Dieleclrics. Tilted grooves on nat
sheets and an axially symmetric radomes". IEEE International Symp. Antenna
Propag. Dig .• pp 610-611. JlUle 1979.
14. D.L. Loyet. "Broadband radome design techniques". Proc. 13th EM Window
Symposium. Georgia Institute of Technology. pp 169-173. 1976.
IS. D.C. Seller, "Abberation of dieleclric patches and rings". Proc. Int. Conf. on
EM Windows. Paris. 1967.
16. A. Cohen. P. Davis. S.C. Nilo and J.F. Orabona. "A 150 feet metal space frame
radome". Proc. OSU-WALDC Radome Symp. WADC Tech. Report 57-314.
17. Pelton and Munk. "A Streamlined Metallic Radome". IEEE Trans. Ant. and
Prop.• Vol. A. pp. 22. no. S. pp. 799-803.1974.
18. A. Cohen and A.P. Sondski. "Effect of rain on satellite commlUlication earth
terminals and rigid radomes". Microwave Journal. Vol 9. no. 9. pp 111-121.
Sep.1966.
19. M.M. Schwartz. "Composite Materials Handbook.... New York: McGraw-Hili.
1983.
20. G. Lubin. "Handbook of Advanced Composites". New York: Van Nostrand
Reinhold. 1982.
21. "Kevlar Use". Aviation Week and Space Technology. pp. 37. April 7. 1986.
22. 1. Cook. "Reinforcements for plastics intended for electrical and electronic
applications". Journal of Naval Sciences. Vol. 1. No.3. pp. 221-233. July 1975.
23. R.H. Cary. "Avionic Radome Materials". AGARD report no. AGARD AR-7S.
24. ''The basics on bonded sandwich construction". Hexcel Corporation Book Part
no. TSB 120 m:td TSB 124.
25. G.M. Brydon and R.H. Cary. "Some Radome materials with superior dielectric
and temperature characteristics". Proc. Int. Conf. on EM Windows. Paris. 1967.
26. G.S. Mani. et al. "Development of Broadband Radomes for Electronic Equipment
Pod". DLRL report No. DLRL R : 86 : 024 (Reslricted). April 1986.
27. M.S. Cray and M.G. Taylor. "Syntactic foam-A new method for radome
construction". Presentation given at Fourth Int. Conf. on EM Windows. Barcol.
France.
28. M.B. Punnett. "Raydel-Microwave transmissive fabric radome composites".
Proc Int. Conf. on EM Windows. Georgia Inst. of tech. pp. 117-123. 1978.
29. Joseph 1. Prifti. "Hardened TlUled-wall plastic radome for military radar".
30. "Astrocoat-Sterling erosion resistant coating system". Sterling Lacquer Mfg.
Co.• 3150. Brannon Avenue. St. Louis. M063139. USA.
31. G.S. Springer. "A model for curing process of epoxy malrix composites". Progress
in Science and Engineering of Composites. Japan Society of Composite Materials
pp 25-35. 1982.
32. T.G. Gutowski, "A resin flow/fiber defonnation model for composites", Advances
in Technology in Materials and Processes, SAMPE, pp. 925-934, 1985.
33. 1.M. Tanged et al, "Effects of cw-e pressure on resin flow, voids and mechanical
properties", Jour. of Composite Materials, Vol. 21, pp. 421-440, May 1987.
34. M.N. Afsar, 1.R. Birch and R.N. Clarke, "The measurements of properties of
materials", Proc: IEEE, Vol 74, No. I, pp. 183-199, January 1986.
35. W. W. Ho, "High temperatw-e millimetric dielectric characterization of radome
materials", Proc. SPIB-Int. Soc. opt. Engrs, Vol. 362, pp. 190-195, 1982.
36. F.I. Shimabukuro et ai, "A quasi-optical method for measuring complex
permittivity of materials", IEEEE Trans. on Microwave Theory and Tech, Vol
MTT--32,no. 7,pp. 659-665, 1984.
37. T. Larry Norin and Luis L. Oh, "Monitoring AWACS radar radome fabrication",
Proc. 14th EM Window Symposium, Georgia Inst. of Tech., pp 137-145, 1978.
38. N.R. Ray, "A portable radome tester", Proc. 12th Symp. on EM Windows,
pp 117-124, June 1974.
39. Leon 1. Ricardei and Seymong Sutlcin, "A microwave radome perfonnance
verification gauge", IEEE Trans. on Ant. and Prop., Vol. 39, no. 2, pp. 131-
142, Feb 1991.
40. Gladius Lewis, "Selection of Engineering Materials", New Jersey, Prentice-
Hall,199O.
9
High Frequency Applications of High-Tc
Superconductors
C.M. Srivastava
Advanced Centre fot Research in Electronics and Department of Physics
Indian Institute of Technology. Bombay. India
1. INTRODUCTION
The current investigations on the technological impact of the high temperature
superconductors (HTSC) indicate that the most promising applications of these
materials are in the high-frequency devices. The replacement of waveguides
and coaxials circuits by microwave integrated circuits more than two decades
ago heralded a revolution in the field of microwave circuits. The main reasons
for the widespread use of microwave integrated circuits are increased reliability,
improved reproducibility and significant reduction in size, weight and cost.
These gains are of vital importance in space and satellite electronic circuits.
Further in military applications, where equipments are subjected to severe shock
and vibrations, MIe devices provide .better mechanical integrity than waveguides
bolted together.
Even before the advent of HTSe superconductors were used in some special
high frequency circuits and cavities [1,2,3]. The reason is that lower conductor
losses result in lower noise, higher speed and wider bandwidth. Amongst the
major systems applications is the use of superconducting microwave cavities
for accelerating charged particles. The possibility of using HTSe in super-
collider experiment is currently being examined.
Several successful attempts [4, 5, 6] have been made to develop passive
microwave devices like resonators, filters and delay lines using HTSe. These
devices are in the stripline configurations. A strip line is a thin narrow strip of
metal which is placed a short distance above a metal ground plane. The strip
is placed parallel to the ground plane and is separated from it by a thin dielectric
slab. The transmission occurs via the transverse electromagnetic (TEM) mode
much in the same manner as in a two wire line. A stripline is lossier than a
waveguide but it is smaller and cheaper and hence is extensively used.
The losses in stripline primarily originate from the dielectric and the conductor.
With improvement in the microwave dielectric materials the conductor losses
in stripline devices today dominate even when the best quality copper or gold
High Frequency Applications ofHigh-T. Superconductors 241
is used. With the advent of HTSC and use of liquid nitrogen as coolant for the
device it is possible to reduce the conductor losses and enhance the quality
factor of the device in a cost effective manner.
The possibility of complete elimination of refrigeration below ambient
temperature exists in a space environment where temperature close to 100 K
can be obtained from radiative cooling. If Tc can be further pushed by about
30 K from the existing maximum of 125 K for thallium based compounds it
should be possible to use many of these devices in aerospace industry without
additional refrigeration.
Although today several high Te superconducting compounds are known, most
of the high frequency applications have been attempted on YBa2Cu3O,(YBCO)
films. Only recently some results on Tl-Ca-Ba-Cu-O thin films have also been
reported.
There are however some problems which need to be solved before the rapid
deployment of the HTSC microwave device is possible. The fabrication of
these devices require development of good quality epitaxial thin films with
high Je , low RI , long term stability, methods of patterning without affecting the
superconducting properties, ohmic contacts and large size spbstrates of good
qUality. The device performance is critically dependent on these parameters of
the film. Further, the type of correlation which exists between the structural
and microwave properties of HTSC films is not clear at present. It is therefore
not possible to predict the device performance from the measurements on the
unpatterened films.
In this article, we discuss the electromagnetic properties of good conductors
and superconductors. Specially, we examine the dependence of the propagation
characteristics on surface· resistivity for a microstripline. We then describe
some of the microwave components which have been successfully fabricated
using HTSC materials. Finally, we give the current trends and future potential
applications.
2. ELECTROMAGNETIC PROPERTIES
2.1 Normal Metals-Anomolous skin effect

The dc resistivity of a good conductor like copper changes by 3 orders of
magnitude on cooling from 300 to 0 K. However, in the same temperature
range the high frequency surface resistance decreases only by a factor of 7.
This is due to .the anomolous skin effect At very low temperatures the mean-
free-path I, of pure annealed copper is about 1 mm. The magnetic field H
inside the conductor is parallel to the surface and though it is much larger
than the electric field its intensity gets rapidly attenuated inside the conductor
(/ oc e-l/6 ), where 6 is the skin depth and is given by
6 =[2/Jl.c tv 0'] 112 (1)

Here Jl.c is the permeability of the conductor, tv is the frequency and 0' is the
conductivity. The magnetic field at low temperatures thus penetrates a distance
much less than I - 1 mm (6 « I). The surface electrons which are the active
carriers of charge transport cannot then interact with the external rf field and
exchange energy. This leads to anomolous skin effect and results in high value
for the high-frequency surface resistance compared to the d.c. resistance.
In the region where anomolous skin effect is absent (8) I), the normal state
surface resistance is given by
R. =_1
(18
=(tr /J,tcP)ll2 (2)
For copper at 300 K, p = 1.7 J,tO-cm andJl.c- Jlo = 4trx 10-7 HIm. For 10 GHz
we obtain R. = 40 mn. At 77 K, (1cu (77)/(1cu (300) - 16. We then have at
77 K,R.-10 mn.
The dependence of surface resistance on temperature and frequency for
copper has been measured by Hammond et al [7]. It is found that R, varies as
ro1/2 for the metals in the normal state. Further, since (1 increases as temperature
is lowered, R. decreases with temperature at constant frequency.
1.1 Superconductors-Sudace impedance

In superconductors the electrons at the surface cannot travel deep into the
materials and are confmed within a sheath of thickness A, where A. is the Landon
penetration depth. The microwave properties of superconductors can only be
understood through quantum mechanics. The surface impedance Z. =R. + j X.
depends on the mechanism of charge transport in the superconductor. In the
extreme local limit the surface impedance is given by Ginsburg [8]
Z.(ro) = (1 + J) [:~rl2 (3)
where
(1sc =(11 - j(12 (3a)
is the conductivity of the superconductor. Using the BeS weak coupling theory
[9] Mattis and Bardeen [10] have obtained the expressions for (11 and (12. Since
these expressions involve complicated integrations over the quasi-particle energy
states we adopt a simplified semi-classical model which to some extent reproduces
the results of quantum mechanical calculations.
For superconductors, zero resistivity leads to the acceleration equation
mv=eE
=
Since J nev, where n is the number of charge carriers per unit volume, we
can write this equation as
. ne2 (4)
J=-E
m
We consider the two fluid model [11] in which the current is assumed to be
a superposition of a normal (resistive) I n and a superconductive component J.
J = I n + J. (5)
with
JD = O'E (Sa)
where (J is the normal state conductivity of the metal. Then, neglecting
displacement current
v x H = O'E + J. (6)
and so
v x V x H = O'V X E + V X J.
or
(7)
From Eq. (4)
. ne 2 ne 2J.1. • (8)
V x J. = - V x E = - - - H
m m
Using Eq. (7) and (8), we obtain
V2 H = 0'J.LiI + 12 H (9)
where .t = (m/ne 2J.1.)112 is the Landon penetration depth. If we assume

H - Ho exp (jOJt) exp (- yz) (10)
- .t2 + :li
Y2 -..L ~2
(11)
where ~ is the ~kin depth (Eq. 1) and y is the propagation constant in the
superconductor.
The surface impedance is defined by the ratio of the electric field Ell to the
magnetic field HII tangential to the conductor surface
Ell . jOJJ.I.
Z. = -H =R. +JX. = - - (12)
II Y
In the normal state (T > TJ .t = 00 and Eq. (11) gives
y= 1+ i = UOJJ.l.O')1I2 (13)
Using Eqs. (12) and (13) we obtain the expression for R, which agrees with
Eq. (2).
For the superconductor ( T < Tc), .t is finite and can be very small at low
temperatures compared to ~, so Eq. (11) determines the characteristics of
microwave propagation. Substituting from Eq. (11) into Eq. (12), we obtain
_ OJJ.I..t [(1 +~) -1]112 (14)

R. - ..fi 4.t4
1 + ~4
(A.« 0) (15)
(16)
-COJU (A.« 0) (17)

From Eqs. (15) and (16), we obtain
(18)
and
R2 +X2
(19)
A. = CO J.l. (~;
_ ;;)1/2
The conclusion thatR, is proportional to co 2 and XI is proportional to ro has
been derived by Hartwig [12].
Many studies of the temperature dependence of surface resistivity of
superconducting rums have been reported [7, 13, 14]. This data can be compared
with the expression for R, in Eq. (15). According to two fluid model a oc 14
and A.2 oc (1 -1 4rl, where 1 =Tae• So R; can be expressed as [1]
J.l.2an(1~)A.~C02 t4
R, - 2 (1 _ (4) (20)
The theory by Mattis and Bardeen [10] gives for T < Te ,
,co, n a co exp (-T Ll/kB n

R( 2
(21)
Measurements of R, on YBCO films by Fathy et al [15] indicate that Rs

varies as co 2 as expected. However, Kobayashi et al [13] found that both Rs and
XI in YBCO films vary as coo. s. Recent measurements of microwave losses in
YBCO films by Pakulis et al [14] show that the temperature dependence of Rs
given by Eq. (20) is not applicable and the loss is proportional to I/~
throughout the temperature range, 0 < I < 1. A complete understanding of the
temperature and frequency dependence of microwave losses in high Te ceramic
superconductors is not yet available probably due to the presence of intergranular
losses which dominate at low temperature.
Figure 1 shows the surface resistance of a thin film of Tl-Ca-Ba-Cu-O
measured by Hammond et al [7] as a function of temperature at 9.55 GHz. This
is compared with the Rs for a similar copper film. At 77 K the Tl-film has at
least 20 times lower loss than copper for fields upto 10 Gauss.
High Frequency Applications ofHigh-Tc Superconductors 245
I00c-------+---------,
9.55 GHz
-
E
aE
- -
10
CF HC Cu ..... --
.....
~ .....
B
1/1
'iii
III
II:
t:. #: V 262
III
•'*
x :ij: V 266
.9
5
III
V 302
0.1 ~_----.JL__ ____1_ ___l._ _~
o 100 200
Tllmperature (K)
Fig. 1. The measured surface resistance vs temperature at low power for three
Tl-Ca-Ba-Cu-O thin films. The complete temperature scan for film # V302 is
shown. The other two fJ.J.ms were measured at 4.2. 77 and 150 K. At 77 K the
R. for all three films is at least 20 times smaller than copper at the same
temperature [7].
2.3 Residual Losses

The surface resistance of most superconductors can be expressed as
Rexp = R(ro, 1) + Rres (22)
where R(ro, 1) is given by Eq. (20) or (21). Rres is a sample dependent constant
whose origin is yet not known. Amongst several possible causes of residual
resistance are the presence of normal conducting materials, tunneling across
cracks, generation by phonons by the rf field and grain boundary scattering of
charge carriers. It is observed that for T,;/T > 5, Rres invariably dominates.
3. ATTENUATION AND DISPERSION IN MICROSTRIP LINES

The analysis of the propagation of the electromagnetic fields in a stripline
using superconductors has been carried out by Kautz [16]. Improved analysis
including the effect of dispersion on the transmission line performance has
been given by Whitaker et aI [17], Kwon et aI [18] and Ekholm and McKnight
[19].
In this section, we discuss the theory of microstrip transmission line and
calculate the propagation constant r which has its real and imaginary parts
r= a + jp (23)
where a is the attenuation constant and Pis the phase factor. The geometry of
the microstrip line is shown in Fig. 2. a is affected by both. conductor and the
dielectric losses, but Pis primarily determined by the line geometry and depends
only to a small extent on the conductor. It is convenient to take
Ground
plane
Fig. 2. Geometry of a micros trip line. The ground plane and the strip mayor may not
be of the same metal. The dielectric substrate is a low microwave loss material.
Ow' is the width and ',' is the thickness of the strip and 'h' is the thickness of
the dielectric substrate.
(24)
where the subscripts c, d and m stand for conductor, dielectric and modal
contributions to a and p.
Using the transmission line equation
(25)
for the propagation of a sinusoidal voltage V (X, m)eimt on the microstrip line,
it can be shown that r is approximately given by
r= [zy]l/2 = .£. (26)

Zo
where Z is the series impedance, Y is the shunt admittance and Zo is the charac-
teristic impedance. For the microstrip line [19]
Z=jmfJo gl + 2Zsg2 (27)
y = meo (jEreff + Er tan O)/gl (28)

where Eo and fJo are the permittivity and permeability of free space, 4 is the
surface impedance, Ereff is the effective dielectric constant, Er and tan 8 are the
relative permittivity and loss tangent of the substrate and gl and g2 are functions
depending on the geometry of the line.
The attenuation constant 11c depends on Zo and the surface resistivity, Zs of
the ground plane and the strip. If both are made of the same material
(29)
Detailed analysis of the losses in microstrips using normal metals has been
given by Pucel et al [20]. Caulton [21] has given a simplified formula for such
striplines
High Frequency Applications 0/High-Tc Superconductors 247
a R. (1.. + 1 ).
= Zo m-I (30)
c w 2h+w
Here Rs is the surface resistivity of the conductor. w is the width of the strip
and h is the thickness of the dielectric substrate.
The characteristic impedance ~ is given in terms of Er. w and h by an empirical
relation obtained by Wheeler [22]
(31)
This gives similar results as the simplified expressions obtained for Zo by
Schneider [23] for the two limits of wide (w/h > 1) and narrow (w/h < 1) microstrip
lines.
(32)
1
Zo=--60ln -+-
.JEcff
(8hw 4hw) (33)
where
Er + 1 Er - 1 ( 10h)-1/2
Ecff=-2-+-2- 1+ W (34)
The effective dielectric constant is given by Eq. (31) only upto the cut-off
frequency
f c -- c
4h (Er _ 1)1/2
(35)
Eq. (35) gives the region (0 </ </J in which the lowest order longitudinal TEl
mode for the microstrip dominates. For /> !C. £eff becomes frequency dependent
and is not given by Eq. (34).
Dispersion at high frequencies has been analyzed by Mittra and Itoh [24]
and Atwater [25]. At very high frequencies the transmission line behaves as a
waveguide with the wave completely confined to the dielectric region.
Consequently the effective dielectric constant £eff approaches Cr as the frequency
is increased to a very high value.
It is convenient to relate Eeff to a dielectric filling factor q by
Ecff - 1 (36)
q = Er - 1
For / > !c as Eeff increases with frequency and approaches Er. q -+ 1.

=
For a 25 mil alumina substrate with Er 10, Ie from Eq. (35) is 40 GHz.
= = =
With hlw 1 Eq. (34) gives eerr 6.86. This gives q 0.65. Yamashita et al
[26] have discussed numerical methods for the estimate of £err beyond Ie. This
is needed to estimate the modal phase factor Pm in Eq. (24)
Pm = 211:1 ~eeff (f) (37)

c
The expression for phase factor due to the conductor is
PI =1m {~~} 82 m-1 (38)
For a superconductor from Eqs. (12) and (17),
= =
For I 10 GHz, A. 5000 Aand Zo = 50 n since 82 - 1 we obtain Pc - 1 rad/m.
For £err if) - 16, from Eq. (37), Pm - 102 rad/m. Thus Pm » Pc.
It can be shown that in general Pc « Pm. It is therefore not necessary to
include Pc in the calculation of the phase velocity. We then have
211:1 c
vpb =-- =
Pm ~eeff(f)
(39)
Ekholm and McKnight [19] have calculated the phase velocity as a function
of frequency for a microstrip line with wlh = =
I, h 0.5 mm and Er 27.0. =
They have used numerical approximation of Yamashita et al [26] to estimate
eerr (f). From Eq. (35)1c is nearly 20 GHz. Their result for Vpb vslis shown
in Fig. 3. The decrease in Vpb occurs due to increase in £err (f) in the frequency
band between 1 and 100 GHz.
7-10 7
7
5-10
10 8 1010
Frequency (Hz)
Fig. 3 Phase velocity as a function of frequency for the microstrip with

w =h » I and h =0.5 mm, E, = 27 [19].
High Frequency Applications of High-Tc Superconductors 249
From our approximate analysis the low frequency phase velocity limit
should be C/.JEeff' where Eeff is given by Eq. (34) while the high frequency
phase velocity limit should be c/Fr. These two values are 7.55 x 107 m/s and
5.77 x 107 m/s respectively which are close to the values given in Fig. 3
obtained from a more accurate analysis.
It may be noted that the dispersion in phase velocity arising from Eeff (f )
spreads over a large frequency band, 1 to 100 GHz. If a pulse propagates on
the microstrip line which has dominant frequency components in this region
severe distortion in the shape of the pulse will occur due to this dispersion.
The transition from the higher to lower phase velocity occurs near 10 GHz.
At this frequency the wavelength within the substrate is A/ Fr - 0.57 cm and
is comparable to the transverse dimension of the microtrip. The dispersion
therefore is primarily arising from geometry related modal effects which
determines the redistribution of energy from the TEl mode to higher order modes
for f > !c. On the other hand the attenuation in the system is determined both
by the contributions by the conductor and the dielectric.
The loss in the dielectric substrate which fills the region between the conductors
in the transmission line is small. An approximate expression for ad in Eq. (24)
is given by [19]
ad~ n Er tan
~
0 ~
1
m-I (40)
"V Eeff 11.0
where Ao is the free space wavelength and tan 0 is the loss tangent of the dielectric.
Lanthanum aluminate (LaAl03) and lanthanum gallate (LaGa03) have been
extensively used as substrates in HTSC microstrip line. At 77 K these materials
have [27] Er - 23 and tan 0 - 2 X 10-4. Using Eqs. (34) and (40) for stripline
= =
with w/h 1, we obtain l1.! 1.08 dB/m.
For a copper microstrip line on LaAI03 substrate at 80 K and 9.55 GHz, Rs
has a value [7] nearly of 10 mn. With w/h = 0.25 and w ::;: 0.5 mm, we get a
value of about 56 n for Zo from Eq. (33) and obtain lXc - 3.5 dB/m from
Eq. (30).
Compared to copper at 80 K TI-Ca-Ba-Cu-O film deposited on LaAl03 has
Rs - 0.5 mn [7]. In this case then <Xc would be 20 times smaller than the copper
microstripline. It would then also be smaller than the attenuation ~ produced
by the dielectric.
3.1 Superconducting Microstrip Line and Waveguide

Structure-Potential and problems
There are several transmission line structures which have been used in the
microwave circuit integration. Some of these [28] are microstrip, suspended
stripline, inverted stripline, slot line, coplanar waveguide, trapped inverted
stripline, Fin line and image line. Amongst these microstrip is the most commonly
used structure. Specially the superconducting devices are generally made in
this structure due to its simplicity.
An analysis of the attenuation and dispersion of microstrip line of YBCO
on yttria-stabilised zirconia substrates on the basis of the theory developed in
Sec. 3 shows that at 60 K the HSTC line should be significantly less attenuating
than the coper line at the same temperature. The application of HTSC microstrip
lines at frequencies above 10 GHz appears to be most useful on account of
very high attenuation of cooled copper at these frequencies.
Winter and Rose [29] have studied the transmission properties of high Tc
superconductor waveguide. Their results show that these have the potential of
100 GHz of band-width for transmission over long distances with low attenuation.
They also show that the millimeter waveguide would have the advantage over
optical systems since they can have much lower carrier frequency which would
result in wider dynamic range.
Despite attractive advantages widespread use of high Tc superconductors in
high-frequency devices have not been possible due primarily to the difficulties
of fabrication of reproducible high quality films for microstrip lines. Likewise
HTSC microwave cavities for accelerating charged particles have been
investigated but the progress is slow due to difficulties arising from the techniques
used for their fabrication.
One of the major problems arising from imperfection in structure due to the
methods used for synthesis is the discrepancy between theoretical and experimental
results which in the case of attenuation constant may be of a few orders of
magnitude. The disagreement may also be due to the inadequacy of the theoretical
models like those used in Sec. 3 [30] for the analysis of the properties of HTSC
materials.
3.2 Substrate for Microstrip Lines

The substrates which are often used in HTSC microstrip lines are MgO. SrTi03•
AI2~' LaGa03 and LaAl~. Their properties are given in Table 1.
Table 1. Mlcrostrlp line substrate materials
Material t;. tan 8 Ie (GHz) Ref.
MgO 9.65 4 x 10-4 51 [31]

Al:z~ 8.30 1 x 10-4 55 [32]
LaGaO] 23.1 2 x 10-4 32 [33]
LaAl~ 23.2 5 x 10-4 32 [34]
zrO:z: y* 27.0 6 x 10-4 29.0 [35]
SrTi~ 1875.0 IOZ 3.6 [33]
*Ytttia-stabilized zirconia.
The most commonly used HTSC material is TBCO. The films are generally
prepared by laser ablation deposition followed by post-deposition thermal
processing.
In Table 1. we have also given reference to the work in which these substrates
have been used to make a microstrip line. We have also indicated in the table
=
the cut-off frequency !C. obtained from Eq. (35) with h 0.5 mm. At frequencies
greater than!c the microstrip line supports a variety of waveguide modes in
addition to the quasi-TEM mode of low frequencies. This has important effects
High Frequency Applications ofHigh-T, Superconductors 251
on the .distribution of the pulse 'shape during transmission if it comprises of

frequency components above !c.
3.3 Thin Film Microstrip Resonators

A 35 GHz ring resonator has been fabricated by Bhasin et al [36] from laser
ablated YBCO films on LaAI03 substrates. The structure of ~e resonator is
given in Fig. 4. Below 60 K the HTSC strip performed better than the gold.
At 25 K the measured Q of HTSC resonator was 600 which was 1.5 times the
Q of identical resonator made of gold.
W = Super conducting strip line
W= 143 l-1 m
t1 =0.5 )1m
t s =254pm
tt =0.1 )1m
·t 2 =1.0)lm
Fig. 4. 35 GHz ring resonator structure fabricated from laser ablated YBCO thin film
on lanthanum aluminate substrate. The film is c-axis aligned and its Tc is
89.8 K. The metal ground plane is deposited by first evaporating 100 A. of Ti
for adhesion followed by 1 micron of gold [36].
A high Q TI-Ca-Ba-Cu-O thin film microstrip resonator has been fabricated

by Hammond et al [7]. They measure a loaded Q of 7300 at 2.6 GHz and 77 K
which is 20 times higher than of an identical silver resonator. On increasing
the frequency to 7.3 GHz the Qat 77 K of HTSC resonator was 6000 which
was better by a factor of 10 than of the silver resonator. The HTSC resonator Qs
decreased monotonically as the input power was increased. At effective power
levels in the resonator upto 100 Watts in frequency range 2 to 7 GHz the Q was
at least three times higher than the silver resonator at all frequencies.
The analysis of Q is performed [37] in the coplanar geometry in terms of
the quality factor Qc due to conductor losses, Qr to radiation and Qd to losses
due to dielectric substrate
-1 = -
1 1
+- 1
+- (41)
Q ac Qr C?!
Radiation losses are reduced by choosing smaller lateral waveguide dimensions.
But this lowers Qc and hence an optimization procedure is required. Qr generally

is in the range of 10,000. Qd is approximately given by (tan 0)-1. From
Table I, Qd is thus in the range of 3000 to 10,000. The conductor quality factor
can be described as
(42)
where g is a geometry dependent factor which can be calculated and Rs is the

surface resistance. Qc depends on the structural imperfections of the HTSC
films such as grain boundaries, twinning, point defects, surface roughness and
some intrinsic parameters which are so far unknown and contribute to Rres of
Eq. (22).
3.4 Superconducting Filters

Currently, narrow band MIC'filters cannot be used in communication systems
on account of their low unloaded Q (50 to 5(0) compared to the coaxial and
waveguide structures for which Q is two orders of magnitude larger (5000 to
10,000). A low Q filter introduces high midband loss and degradation in selectivity
and gain at the band edges. It is expected that the replacement of copper by
HTSC will permit MIC structures whose performance will be comparable to
those of waveguides at least at low power levels. We have seen in the previous
section that microstrip resonators with HTSC film can achieve Q > 7000.
Bonetti and Williams [27] have developed techniques which help in the
design and development of microstrip filters on high Er substrates. They have
used this technique to design and fabricate a 4-pole, C-band, 150 MHz band-
pass filter with a 0.05 ripple Chebychev response built on a lanthanum galate
substrate. Experimental results at liquid nitrogen show excellent agreement
with the predicted performance.
4. HIGH-Te THIN FILM MICROWAVE DETECTORS AND MIXERS

It is noted in Sec. 3 that HTSC microstrip transmission line can provide distortion
free propagation of electrical transients with a bandwidth of more than
100 GHz. The HTSC films used in these applications have to be of very high
quality deposited on a substrate generally with c-axis orientation. In contrast
for active superconductive devices which can operate at liquid nitrogen
temperatures use has invariably been made of granular films. Such films have
been successfully tested as sensitive microwave and far-infrared detectors and
mixers [38, 39]
Konopka et al [40] have fabricated granular YBCO and BI-Ca-Sr-Cu-O
(BCSCO) detectors which operate over the frequency range of 24 to 110 GHz
at 77 K. The sensitivity achieved at 110 GHz is comparable to that of the
crystalline detectors.
Mixing experiments of two microwave signals in YBCO-on-MgO and BCSCO-
on-MgO detectors in the 25 GHz range were carried out successfully [40].
YBCO mixer output was linear at the signal levels between -10 and -50 dBm.
The mixing action disappeared at 80 K when the onset of superconductivity in
the film was 85 K. By tuning both signal and local oscillator microwave sources,
IF frequencies could be varied from 50 MHz to 5 GHz. This could be achieved

without any decrease in the mixer output upto 3 GHz. The detector response
time was estimated to be less than 40 ps.
High Tc fIlms with substantial granularity are produced either by dc magnetron
sputtering or by chemical deposition from nitrate precursors.
The physi~al processes responsible for detection and mixing effects are only
partially knOWD. The granular films comprise of a collection of superconducting
grains interconnected by grain boundary Josephson junctions and regions of
the normal non-superconducting phase. The general behaviour of the detector
has to be analyzed on the basis of this complex structure. Absence of a full
understanding of the physical mechanism makes the device optimization extremely
difficult.
The detection sensitivity increases with the decrease in temperature but
there is a limit to the detectors lowest-usable temperature (typically 50 K) on
account of the complicated transport mechanism at low temperature in which
quantum effects dominate.
s. MICROWAVE SUPERCONDUCTING CAVITIES FOR

PARTICLE ACCELERATORS
We have seen that in the microwave frequency range superconductors remain
substantially free of resistance at frequencies upto 100 GHz. This is particularly
true of type I superconductors. Even for type II superconductors, the Meissner
effect may persist metastably above Hel to a superheating critical field Hsh
because it takes a finite time to nucleate a flux line. Experimentally, for type
I superconductors Hsb is found close to H I-{K, where K is the Abrikosov constant
=
and is given by K AI~. Here, A. is the London penetration depth and ; is the
coherence length. For the extreme type II superconductor, theory predicts
Hsh - 0.75 He » H el • The ability of superconductors to support lossless high
frequency surface currents below Hsh is exploited in many devices, the most
important amongst these is the electron linear accelarator called Linac.
The principle of Linac is simple. A cluster of electrons are accelerated by
passing them through a series of resonant cavities. The time taken for the
electrot:' to travel from one cavity to the next is the half period of the rf source.
The electric fields in the successive cavities act in the same sense upon the
particle because the velocity is such that it takes just the right amount of time
to reach the cavity when the electric field is oriented for forward acceleration
[I].
The electrons are introduced in linac at high speeds close to velocity of
light. Let the initial energy of the electrons be 1 MeV. It is proposed to accelerate
them to 1 GeV. If the resonant cavity has a volume Vof 1/30 m3 and the field
gradient E in the cavity is 8 MV1m, nearly 400 resonant cavities are needed tc
reach this energy. It is, therefore, necessary to have cavities with high Q to reduC(
the operational cost.
The peak power Ep' stored in the cavity is
(43
where the pennittivity of free space, Eo =(_1_)

361r
X 10-9 Fm-I • With E =8 MV/m
and V = 1/30 m3, Ep - 9.5 J per cavity. The energy loss per cavity is given
by
w= Epl (44)
Q
For copper cavity, Q - 6 x 1<t. Iff - 1 GHz, W - 170 kW. The total power
consumption for 400 cavities is then 68 MW. At present such high powers can
be produced at high frequencies only in the pulse mode. If the surface resistance
can be reduced by three to four orders of magnitude the Q can be substantially
improved and the power dissipation reduced to a level that continuous operation
of a Linac would be possible. This possibly can be achieved with the help of
a HTSC superconductors at 77 K.
We note that the performance of the microwave cavity is limited by two
parameters-the critical field (He) and the surface resistance (R.). It is in fact
the superheating critical field Hsh, which determines the peak power handling
capability of the cavity without loosing the superconducting properties through
the relation
Emu _ Hsh (~: )112 (45)
For Nb, H.h - 1600 Oe, so Emu - 60 MV/m. In HTSC material, H.h is much
higher. For example, in YBCO the thermodynamic critical field is 27,000 Oe
and the superheating critical field is 20,000 Oe. One can, therefore, take for
HTSC that H.h is 10 times higher than Nb so the accelerating voltage capability
of a HTSC cavity increases to 600 mV/m. This leads to substantial reduction
in the number of cavities required to ahieve the same energy.
The improvement in Q due to reduction in surface resistance when HTSC
cavity is used leads to lower power dissipation and thereby to lower cost of
operation. The Q of the cavity is given by
Q = Jlo VI (46)
R.A
where f is the frequency and V /A is the volume to surface ratio of the cavity.
Assuming VIA - 0.1 m, f =1 GHz, R. - 10-7 n approximate for a superconductor
for Tc/T > 5, we obtain Q - 108 • This is four orders of magnitude larger than
for a copper cavity.
Attempts to use HTSC cavities in Linac are in progress [41] and it is likely
that soon attractive economic benefits will be realised in this application.
6. CONCLUSION
The high frequency properties of high temperature superconductors have been
dis~ussed and in particular their surface impedance has been examined as a
High Frequency Applications ofHigh-Tc Superconductors 255
function of frequency and temperature. The possible advantages of microstrip

transmission lines made of high temperature superconductors over copper or
gold are described through a detailed study of the attenuation and dispersion
of microwave transients at 77 K. It is shown that the performance of HTSC
passive microwave components show great promise for improved performance
over their metallic counterparts. However, there may be some problems associated
with the fabrication of good quality, reproducible thin films of HTSC on dielectric
substrates which may delay early utilization of these devices in microwave
systems. A brief outline of application of granular superconductivity observed
in HTSC films to active devices like detectors and mixers is given. Finally the
possible utilization of HTSC cavities in linear particle accelerator has been
discussed.
REFERENCES
1. J.E.C. Williams, 'Superconductivity and its applications', Pion Limited, London,

NW2,1970.
2. V.L. Newhouse, 'Superconducting devices' in 'Superconductivity' Vol. n, Ed.
R.D. Parks. Marcel Dekker Inc, N.Y. 1969.
3. T. van Duzer and C.W. Turner, 'Principles of superconducting Devices and
Circuits', Elsevier, N.Y. Chapter 2. 1981.
4. P.H. Carr, 'Potential microwave applications of high temperature superconductors',
Microwave Journal, p. 91-93, Dec. 1987.
5. D.E. Oates, A.C. Anderson and B.S. Shih, 'Superconducting stripline resonators
and high Tc materials, IEEE MTT-S Digest p. 627-630, 1989.
6. J.M. Pond and C.M. Krowne, 'Slow wave properties of superconducting microstrip
transmission lines', IEEE MTT-S Digest p. 449-452, 1988.
7. R.B. Hammond, G.V. Negrete, M.S. Schmidt, M.l Moskowitz, M.M. Eddy,
D.D. Strother and D.L. Skoglund, 'Superconducting Tl-Ca-Ba-Cu-O Thin film
micros trip resonator and its power handling performance at 77 K', IEEE MTT-
S Digest, p. 867-870, 1990.
8. D.M. Ginsburg, 'Calculation of the electromagnetic absorption edge in
superconducting alloys', Phys. Rev. 151, pp. 241-245, 1966.
9. J. Bardeen, L.N. Cooper and lR. Schrieffer, 'Theory of Superconductivity',
Phys. Rev. 108 1175 (1957).
10. C.C. Mattis and 1 Bardeen, 'Theory of anomolous skin effect in normal and
superconducting metals', Phys. Rev. 111,412-417, 1958.
11. F. London, 'Superfluids, Macroscopic Theory of superconductivity', Vol. II,
New York, Dover publications, 1960.
12. W.H. Hartwig, 'Superconducting resonators and devices', Proc. IEEE, 61, 1
pp. 58-70, Jan. 1973.
13. Y. Kobayashi, T. Imai and H. Kayano, 'Microwave measurement of surface
impedance of High-Tc superconductors', IEEE MTT-S Digest, p. 281-2841990.
14. E.J. Pakulis, R.L. Sandstrom, P. Chaudhari and R.B. Laibouitz, 'Temperature
dependance of microwave losses in Y-Ba-Cu-O films', Appl. Phys. Lett, 57 (9)
940-9421990.
15. A. Fathy, D. Kalokitis, V. Pendrick, E. Belohoubek, T. Venkatesan, L. Nazar,
J.B. Barner, A. Findikoglu, A. Inam, X.D. Wu, X.x.xi and W.L. McLean,
'Microwave Properties and Modelling of High-Tc Superconducting Thin Film

Meander Line', IEEE MTI-S Digest p. 859-862, 1990.
16. R.L. Kautz, 'Picosecond pulses on superconducting striplines', J. Appl. Phys.
49 pp. 308-314, 1977.
17. IF. Whitaker, T.B. Norris, G.A. Mourou and T.Y. Hsiang, 'Pulse dispersion
and shaping in microstrip lines', IEEE Trans. Microwave Theory Tech. Vol.
MTT-35, pp. 41-47 1987.
18. O.K. Kwon, B.W. Langley, R.F.W. Pease and M.R. Beasley 'Superconductors
as very high-speed system level interconnects', IEEE Electron Device Leu.
Vol. EDL-8, pp. 582-585, 1987.
19. E.B. Ekholm and S.W. McKnight, 'Attenuation and Dispersion for High-Tc
Superconducting Microstrip Lines', IEEE Trans. Microwave Theory Tech. MTI-
38,387-395,1990.
20. R.A. Pucel, D.J. Masse and C.P. Hartnig, 'Losses in Microstrip', IEEE, Trans.
Microwave Theory Tech. MTI-16, 342-346, 1968.
21. M. Coulton, 'Film Technology in Microwave integrated circuits', Proc. IEEE.
59, pp. 1481-1489, 1971.
22. H.A. Wheeler, 'Transmission line propagation of a strip on a dielectric sheet on
a plane'-Trans. IEEE, Microwave Theory Tech., MTI-25, 631-647, 1977.
23. M.V. Sneider, 'Microstrips for microwave integrated circuits', B.S.T. Jour. 48,
1421-1244,1969.
24. R. Miura and T. Itoh, 'A new technique for the analysis of the dispersion
characteristic of microstrip lines' , IEEE Trans. Microwave Theory Tech. MTI-
19, pp. 47-56, 1971.
25. H.A. Atwater, 'Tests of micros trip dispersion formulas', IEEE Trans. Microwave
Theory Tech. Vol. 36, pp. 619~21, 1988.
26. E. Yamashita and K. Atsuki, 'Analysis of microstrip-like transmission line by
nonuniform discretization of integral equations', IEEE Trans. Microwave Theory
and Tech. MTI-24, pp. 195-200, 1976.
27. R.R. Bonetti and A.E. Williams, 'Preliminary design steps for thin-film
superconducing fields', IEEE MTT-S Digest, 273-276, 1990.
28. D. Venkatsubbiah, 'Practical aspects of passive components while operating
from dc to microwave frequencies' in Microwave Technology and Applications'
Eds. RVS Sitaram, G.P. Srivastava and P.G. Nair, Tata McGraw-Hill Publ. Co.,
New Delhi 1988, pp. 186-194.
29. IH. Winters and C. Rose, 'High-Tc superconductor waveguides: Theory and
applications' IEEE, Trans. Microwave Theory Tech. MTT-39, pp. 617-623,
1991.
30. T.E. van Deventer, P.B. Katechi, J.Y. Josefowicz and D.B. Rensch, 'High
frequency characterisation of high-temperature superconducting thin film lines',
IEEE MTT-S Digest, 285-288, 1990.
31. M. Wintergill, I Fontanella, C. Andeen and D. Schuele, 'The temperature variation
of dielectric constant of pure CaP, SrF, BaF and MgO', I Appl. Phys. 129
1561, 1963.
32. K. Char, N. Newman, S.M. Garrison, R.W. Barton, R.C. Taber, S.S. Landerman
and R.D. Jacowitz, 'Microwave surface resistance of epitaxial YBa Cu 0 thin
film on supphire', Appl. Phys. Lett. 57,409, 1990.
33. D.W. Cooke, E.R. Gray, R.I. Houlton, H.H.S. Javadi, Maez, B.L. Bennett, B.
Rusnak, E.A. Meyer, P.N. Arendt, IG. Berry, D.R. Brown, F.H. Garzon, 1.0.
Raistrick, A.D. RoUett, B. Bolmaro, N.F. Elliott, N. Klein, G. Muller, S. Orbach,
.H. Piel, J.Y. Josefowicz, D.B. Rensch, L. Drabeck and G. Gruner, 'Surface
High Frequency Applications ofHigh- T. Superconductors 257
resistance of YBa Cu 0 films deposited on LaGaO substrate', Physica C 162-

164, p. 1537, 1989.
34. K.H. YOlDlg, G.V. Negrete, R.B. Hammond, A. Inam, R. Ramesh, D.L. Hart
and Y. Yonezawa, 'Clear correlations observed between YBa:zCu30-B thin f11m
properties and GHz microwave resonator performance', Appl. Phys. Lett. 58
p. 1789, 1991.
35. A.C. Anderson, B.Y. Tsam, I.W. Steimbeck and M.S. Dilorio, 'RF surface
resistance of YBCO thin films', MIT Lincoln Labotatory Quaterly Tech. Rep.
Mar. 11, 1988.
36. K.B. Bhasin, C.M. Chorey, I.D. Warner, R.R. Romanofsky and V.O.Heinen,
'Performance and Modeling of superconducting ring resonators at millimeter-
wave frequencies', IEEE MTT-S Digest, pp. 269-272, 1990.
37. R. Dill, 1. Otto, G. Riha, P. Russer, L. Schultz, G. Solkner, A.A. Valenzuela
and E. Wolfgang, 'Test chip for high temperature superconductor passive devices',
IEEE MTT-S Digest, pp. 863-866, 1990.
38. 1. Konopka, R. Sobolewski, A. Konopka and S.I. Lewandowski, 'Microwave
detection and mixing in Y-Ba-Cu-O thin films at liquid nitrogen temperatures',
Appl. Phys. Lett., 53, pp. 796-798, 1988.
39. M. LelDlg, P.R. Broussard, I.H. Claassen, M. Osofsky, S.A. Wolf and U.Strom,
'Optical detection in thin granular films of Y -Ba-Cu-O at temperature between
4.2 and 100 K' Appl. Phys. Lett. 51, pp. 2046-2048, 1987.
40. I. Konopka, R. Sobolewski, G. lung, W. Kula, P. Gierlowski, A. Konopka and
S.J. Lewandowski, 'Microwave detectors based on granular high-Tc thin films' ,
IEEE Trans. Microwave Theory Tech. 38, pp. 160-165, 1990.
41. H. Padamsee et al, 'Microwave superconductivity for particle accelerators-How
the high Tc superconductivity measures up' in 'Superconductivity and its
applications', Eds. H.S. Kwok and D.T. Shaw Elsevier, New York (1988)
p.249.

Microwave Materials PDF

Uploaded by

Copyright:

Available Formats

Microwave Materials PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Microwave Materials PDF

Uploaded by

Copyright:

Available Formats

Microwave Materials

Springer-Verlag Berlin Heidelberg GmbH

Copyright @ 1994 Springer-Verlag Berlin Heidelberg

ISBN 978-3-662-08742-8 ISBN 978-3-662-08740-4 (eBook)

Solid State Materials have been gaining importance in recent times

1. Materials and Processes in Microwave

Microstrip Line Strip Line

Fig. 1. Typical circuit configurations used for microwave integrated circuits.

be discrete components such as diodes, transistors etc., as well as MMICs

3.1 Photolithographic Fabrication

3.2 Dielectric Substrate

Photoresist Coating and

,.....""""_ _ _ _ _IiL......, <- Final printed

Fig. 1. Photolithographic fabrication of microwave integrated circuits.

Substrate materials used in the fabrication of MICs can be broadly classified

Feature Hard Soft

Dielectric Alumina, fused quartz Glass fiber reinforced (woven or

A variety of ceramic, glass or single crystal materials fall in the category of

Property Alumina Fused Berellia Aluminium

Composition 99.6% Si~ 99.9% A1N

3.3 Thin Film Metallization

Dielectric (4.5 to 500 mil)

Ground plane (Brass. Copper

Metallization structures employed on hard substrates are more complex for

3-7 Microns +-Conductor layer 1

200-300 A +-Barrier/Adhesive layer

200-625 Microns :.:-:-:.:-:-:-:.:.:-:.:-:-:-:.:.:.:.:-:-:-:-:-:.:-:-:.:.:-:-:.:-:-:.:-:-:.:-:.: +-Dielectric substrate

... .. . . . . . . . ... . .. . .. . . . ... . . .. . . . ... . . .

Adhesion of copper or gold to the substrates is poor. One has to provide an

hardness. Variety of metallization structures that are employed on the hard

Table 3. Typical specifications for commonly used metallization structures on

Adhesive Resistive Barrierl Conductor Barrier Conductor

Table 4. Properties of thin film materials used for conductor metallization

Mate- Functional Bulk Surface Skin Typical Interconnections

Cu Conductor 1.7 O. 85xl0-2../f 2. 15"'f 3-7Jlm Excellent Possible@

Vacuum evaporation, Sputtering and Electroless/Electro deposition are the

3. Refractory material such as tantalum also can be easily evaporated.

...... - - - ......... +- - - - - - SUBSTRATES

'~~--- ...-- - - ELECTRON BEAM

,...-.....- - - -..... <- - - BASE PLATE

Figure 6 shows a schematic representation of sputtering process. The chamber

WATER COOLm I SUBSlJA1t:S

Fig. 6. Schematic fo, sputter deposition.

3.4 Thick Film Metallization

Table 5. Process steps for electroless deposition

Operation Materials used Effect / Purpose

Cleaning Organic solventsIDetergents Removal of grease, dirt and

fabrication of hybrid circuits in the lower frequency typically upto 1 GHz.

Process step Effect Pattern

Screen printing Transfer of paste on

Settling at ambient Smoothenout the screen

Pre-baking Removal of solvents

Firing schedule Organic vehicle burn-out

FIring schedule Fusing of the glass frit to

3.5 Resistive and Dielectric Layers

Table 6. Properties of thin nlm resistor materials

Material Sheet TCR Remarks

Alz0 3 /NiCr/ 10-500 ... 0 Stable resistance and zero TCR

Necessity for a dielectric layer is felt in realising certain circuit elements