08c

HIGH-FREQUENCY TRANSMISSION LINES
FERNANDO L. TEIXEIRA
*
The Ohio State University
Columbus, Ohio
KALADHAR RADHAKRISHNAN
Intel Corporation
Chandler, Arizona
WENG C. CHEW
University of Illinois at
UrbanaChampaign
Urbana, Illinois
1. INTRODUCTION
Transmission lines are used to guide electromagnetic
waves from one place to another. They are used for a
wide range of applications, and different types of lines are
suitable for different applications. The power transmission
lines that supply electricity do not transmit signals, but
just the power required to run the various electrical de-
vices. As a result, power lines are designed to minimize
the various losses incurred in traversing great distances.
There are also transmission lines that carry high-frequen-
cy signals varying in a nonrepetitive fashion for commu-
nication purposes. These are called high-frequency
transmission lines. The impetus in the design of such
lines is to maximize the operable frequency bandwidth
and minimize the attenuation of the signals. The former
preserves the quality of the signal being transmitted,
while the latter increases the spacing between repeater
stations, reducing the cost of transmission. Transmission
of signals is not limited to physically long lines. Trans-
mission lines are also used to send signals within a com-
puter or to feed an antenna from its source.
1.1. Historical Perspective
The earliest device to use electrical transmission lines was
the telegraph. The rst message was sent by Morse in the
year 1844. By the year 1866, a cable was laid across the
Atlantic Ocean, making it possible to send messages
across continents. A few years later, with the invention
of the telephone, transmission lines with a wider band-
width were employed. By 1890, as electricity became more
important, power lines were developed to transmit power
from the generators to the consumers. These lines which
use just a simple wire to transmit signals are referred to
as single-wire transmission lines.
The rst experiments involving two-conductor trans-
mission lines were carried out by Heinrich Hertz [1] to-
ward the end of the nineteenth century. His experiments
with the coaxial line demonstrated the validity of the elec-
tromagnetic theory of Faraday and Maxwell [2]. In 1897,
Lord Rayleigh [3] proved mathematically that it is possi-
ble to transmit waves through a hollow waveguide of rect-
angular or circular cross section. A different type of wave
propagation based on surface waves was analyzed by
Sommerfeld in 1899 [4]. This work was extended to
analyze wave propagation along dielectric cylinders by
Hondros and Debye in 1910 [5]. Marconi was the rst to
demonstrate wireless communication by sending signals
across the Atlantic Ocean in 1901. The primary function of
transmission lines in wireless communication is in trans-
ferring the signal from the transmitter to the antennas.
With the development of the radar in the 1930s, the
frequency of the signals were approaching the gigahertz
range. At these frequencies, the hollow waveguides pro-
vide a more practical, low-loss way of transferring energy.
Even though it was theoretically proved that it was pos-
sible to transmit waves through a hollow waveguide, no
experimental results were forthcoming until the year
1936. It was then that two men, George C. Southworth
and Wilmer L. Barrow, independently demonstrated wave
transmission through the waveguide [6]. The period dur-
ing World War II saw a steady growth in both theoretical
and experimental work on transmission lines and wave-
guides as practical communication devices.
After World War II, rectangular waveguides became
the dominant waveguiding structure. However, as engi-
neers sought components with larger bandwidths, other
structures were investigated. The next major milestone in
the history of transmission lines was the development of
the microwave printed circuit [7,8]. The microstrip line
was introduced to alleviate the bandwidth problem since
its fundamental mode operated at all frequencies. The
ease of fabrication was another attractive feature of the
microstrip line.
By the 1970s, most long-distance communication lines
were simple two-conductor transmission lines. As the de-
mand for high capacity and long-distance transmission
systems increased, optical bers were developed to re-
place the existing copper wire system [9]. Optical bers
had a much wider bandwidth than did the simple coaxial
line since they operated at optical frequencies. Their in-
herent low loss also made it possible to achieve a much
higher repeater spacing than conventional copper-based
transmission lines.
1.2. Comparison of Transmission Lines
Several factors need to be considered when choosing a
transmission line for a particular application. The most
important factors to be considered when comparing trans-
mission lines are
1. Frequency range and bandwidth
2. Attenuation
3. Power-handling capability
4. Ease of fabrication
Since there is no single transmission line that gives the
optimal performance in all these areas, the line that has
the best tradeoff for the given application must be chosen.
For low-frequency applications like power transmission, a
simple wire can be used to connect the source and the re-
ceiver. The frequency range of such lines is limited to low
frequencies because they start radiating energy as the
frequency is increased. At higher frequencies, the eld is
*
Corresponding author
HIGH-FREQUENCY TRANSMISSION LINES 1989
Previous Page
usually transmitted using coaxial lines that consist of a
wire inside a cylindrical outer conductor. The outer con-
ductor acts as a shield to minimize radiation losses. Such
coaxial lines can be used to send signals up to a few giga-
hertz. Even though the coaxial lines have large bandwidth,
they suffer from high attenuation and low power-handling
capability. A hollow metallic waveguide is more efcient at
high frequencies because of its high power-handling capa-
bility and low attenuation. However, unlike coaxial lines,
waveguides have a much smaller bandwidth.
It is also possible to transmit energy through dielectric
slabs and rods, without any conducting walls. Such wave-
guides are referred to as open or dielectric waveguides. For
example, an optical ber is a dielectric waveguide. As their
name suggests, they operate at optical frequencies. Since
they operate at such high frequencies, they have a very
large bandwidth. They also tend to radiate energy at dis-
continuities and bends.
The last transmission line to be discussed in this sec-
tion is the microstrip. A microstrip line consists of a thin
metallic strip placed on top of a dielectric substrate that is
supported from below by a conducting plate. Just like the
coaxial lines, microstrip has a large bandwidth but suffers
from high attenuation and low power-handling capability.
But their primary advantages are their ease of fabrication
and integration with complicated microwave circuits. The
properties of the various transmission lines discussed
above are summarized in Table 1. Some of them are illus-
trated in Fig. 1.
1.2.1. Two-Conductor Transmission Line. Different
types of two-conductor transmission lines are available
for different applications. The simplest of them is the
single-wire transmission line with Earth acting as the
implicit ground plane. Such lines are useful for low-fre-
quency applications. At higher frequencies, the radiation
losses make it impractical to use a single wire transmis-
sion line. At such frequencies, it is desirable to use a
coaxial line that consists of a wire surrounded by an ex-
ternal cylindrical conductor. The outer conductor acts as a
shield that prevents radiation losses. The principal mode
of operation in such lines is the TEM mode: a mode in
which both the electric and the magnetic elds are per-
pendicular to the direction of wave transmission. The
TEM mode can propagate at all frequencies, giving the
coaxial line a large bandwidth. Besides the coaxial line,
there also exists the two-wire balanced line that consists
of two identical lines placed next to each other. The twist-
ed-pair transmission line is another type of two-conductor
transmission line. In these lines, two insulated wires are
twisted together and encased, along with other similar
pairs, inside a metallic sheath. These lines are popular for
transmission of both analog and digital data. The twisting
in the wire reduces low-frequency interference, and the
Table 1. Comparison of Transmission Lines
Type Advantages Disadvantages
Single-wire transmission lines High power, inexpensive Low-frequency range, lossy
Coaxial lines Large bandwidth, small size Low power, high attenuation
Hollow waveguides Low attenuation, high power Limited bandwidth, large size
Optical waveguides Large bandwidth, low loss Radiate at bends, low power
Microstrip lines Large bandwidth, ease of fabrication Low power, high attenuation

+
Coaxial Line Rectangular Waveguide Circular Waveguide
Optical Waveguide Stripline Microstrip Line
Two-wire Balanced
Line (Twin Line)
+
+
Ridged Waveguide
r
+
1

2
Figure 1. Cross-sectional view of typical transmission line structures.
1990 HIGH-FREQUENCY TRANSMISSION LINES
use of different twist lengths in adjacent pairs reduces
crosstalk. The bandwidth of such twisted-pair lines used
in transmission of digital signals is presently around
6 MHz.
1.2.2. Hollow Waveguides. Hollow waveguiding struc-
tures are more efcient at transmitting high-frequency
signals. A simple hollow waveguide is a hollow metallic
cylinder lled with air or any dielectric medium. Such
waveguides, unlike coaxial lines and other simple trans-
mission lines, do not have a central conductor. As we shall
see later, these waveguides cannot support the TEM
mode. Instead, the wave propagates by bouncing between
the perfectly conducting walls. The modes inside a hollow
waveguide can be classied into TE and TM modes. In a
TE mode, the electric eld is transverse to the direction of
propagation. In TM modes, it is the magnetic eld that is
transverse to the direction of propagation. The TE and TM
modes are decoupled from each other. Any arbitrary eld
distribution within the waveguide can be expressed as a
linear combination of a TE component and a TM compo-
nent. Typically the eld inside the waveguide is a combi-
nation of several modes. For each waveguide mode, there
is a critical frequency below which the waveguide will not
propagate that mode. This critical frequency, also known
as the cutoff frequency, is inversely proportional to the size
of the waveguide cross section. Therefore, to propagate
low-frequency signals, the waveguide must be made big-
ger. This tends to restrict the bandwidth of the signal in a
waveguide.
1.2.3. Open Waveguides. The simplest form of an open
waveguide is the dielectric slab waveguide. It consists of a
dielectric coating on top of a ground plane. To simplify the
analysis of such a structure, the coating is usually as-
sumed to have an innite width with no eld variation
along that direction. For the wave to propagate within a
dielectric waveguide, its permittivity should be higher
than the permittivity of the surrounding medium. The
wave is then guided by total internal reection. As in the
case of hollow waveguides, the eld inside a dielectric slab
waveguide can also be decomposed into TE and TM modes.
However, the eld is not localized to the waveguiding
structure. Since these waveguides are not enclosed by
conducting walls, there exists a residual eld outside the
slab waveguide. The eld outside the guiding structure
decays exponentially, and hence the energy is localized to
the waveguide.
Apart from the slab waveguides, there also exist circu-
lar and rectangular dielectric waveguiding structures.
The optical ber is a simple and very popular example of
a circular dielectric waveguide. It has a dramatic impact
on modern telecommunications. Dielectric waveguides are
also very important in integrated optics and at the shorter
millimiter-wave frequencies. Unlike the closed wave-
guides, the TE and TM modes are coupled to each other
in circular and rectangular open waveguides; thus, none of
the modes are strictly TE or TM. Instead, the EH notation
is sometimes used to denote a mode whose TE component
dominates over its TM component, and vice versa for the
HE notation.
1.2.4. Microstrip Lines. Microwave integrated circuit
(MIC) transmission lines in general and microstrip lines
in particular have become popular for high-frequency ap-
plications because of their ease of fabrication and large
bandwidth. Moreover, they can be easily integrated with
microwave and millimiter-wave integrated circuits. A mi-
crostrip line consists of a thin metallic strip on top of a
dielectric substrate backed by a ground plane. The dielec-
tric substrate material needs to have a low-loss tangent
and a relatively large dielectric constant. The principal
mode of operation is not strictly TEM since the dielectric
does not uniformly surround the conducting strip. In-
stead, the fundamental mode is an approximation to the
TEM mode and is commonly referred to as the quasi-TEM
mode. Microstrip lines can be desiged for frequencies
ranging from a few gigahertz, or even lower, to many
tens of gigahertz.
2. PROPAGATION IN GUIDING STRUCTURES
When the transmission lines [1016] are small in terms of
the wavelengths of the transmitted signal, they can be
satisfactorily analyzed in terms of circuit theory parame-
ters, such as currents and voltages. In this case, the line is
treated using a distributed circuit model. However, as the
frequency increases and the operating wavelength be-
comes comparable with the dimensions of the transmis-
sion line, a more general analysis is necessary. The most
general approach to study the energy propagation along
transmission lines is based on Maxwells equations. By
solving the associated boundary-value problem dened by
the geometry and material of the transmission line, one
can, in principle, completely characterize the electrical
properties of the transmission line. In many cases, an an-
alytical solution is not possible, requiring the use of nu-
merical techniques.
In this section, we shall describe the basic properties of
the electromagnetic elds on a transmission-line structure
using Maxwells equations, and discuss how they can give
rise to the concepts of current and voltage along the trans-
mission line.
If the regions between the conductors in the transmis-
sion line are lled with a homogeneous material, it can
support purely transverse electric (TE), transverse mag-
netic (TM), or transverse electromagnetic (TEM) modes. A
TE wave is such that the electric eld E is everywhere
transverse to the direction of propagation, which will be
assumed to be the z axis. This is equivalent to saying that,
in the TE mode, the E eld is completely conned to the
transverse plane (cross section) of the waveguide. Analo-
gously, for the TM wave, the magnetic eld His transverse
to z, and for the TEM wave, both elds are transverse to z.
If the wave is TEM, it is customary to say that the wave-
guide is operating in the transmission-line mode. (Note:
The term transmission line is occasionally reserved for
structures that support the TEM mode. Here, we consider
transmission lines as any structure capable of guiding
waves; hence, the terms waveguides and transmission
lines are used interchangeably.) It can be shown that a
hollow waveguide cannot support a TEM wave. For the
transmission-line mode to exist, there needs to have at
least two conductors (one of them can be a ground plane).
In our analysis, we will assume harmonic elds and adopt
the e
iot
time convention.
2.1. TE and TM Modes
Because the elds of a general uniform transmission
line can be decomposed into TE and TM types, we can
characterize the TE wave with the ^ zz component of the
magnetic eld or H
z
, since H
z
a0 for this type of wave.
Similarly, we can characterize the TM wave with the ^ zz
component of the magnetic eld or E
z
since E
z
a0 for this
type of wave.
We shall rst derive the equations governing the E
z
and H
z
components of the elds in a uniform transmission
line with a general geometry lled with a homogeneous
material.
We start from Maxwells equations
VE=iomH (1)
VH= ioeE (2)
and let
E=E
s
^ zzE
z
; H=H
s
^ zzH
z
(3)
where the subscript s represents the transverse to z
components. Substituting Eq. (3) into Eqs. (1) and (2),
we have
V
s
^ zz
@
@z
_ _
(E
s
^ zzE
z
) =iom(H
s
^ zzH
z
) (4)
V
s
^ zz
@
@z
_ _
(H
s
^ zzH
z
) = ioe(E
s
^ zzE
z
) (5)
where V
s
stands for the transversal part of the nabla
operator:
V
s
^ xx
@
@x
^ yy
@
@y
Equating the s components of Eqs. (4) and (5), we have
V
s
^ zzE
z
@
@z
^ zz E
s
=iomH
s
(6)
V
s
^ zzH
z
@
@z
^ zz H
s
= ioeE
s
(7)
Substituting for E
s
from Eq. (7) into Eq. (6), we have
o
2
meH
s
=
ioeV
s
^ zzE
z
@
@z
^ zz V
s
^ zzH
z
@
@z
^ zz H
s
_ _
(8)
Using the vector identities ^ zz V
s
^ zz =V
s
, ^ zz ^ zz
H
s
= H
s
, and assuming that the eld has e
ik
z
z
dependence, so that (@
2
=@z
2
) k
2
z
; we can rewrite
Eq. (8) as
H
s
=
1
k
2
k
2
z
@
@z
V
s
H
z
ioe ^ zz V
s
E
z
_ _
(9)
where k
2
=o
2
me. By the same token, we have
E
s
=
1
k
2
k
2
z
@
@z
V
s
E
z
iom^ zz V
s
H
z
_ _
(10)
The factor k
z
is called the propagation constant. Equations
(9) and (10) allow us to derive all the other components of
the eld in a waveguide once we know the z components of
the eld.
If we equate the ^ zz components of Eqs. (4) and (5), we
have
V
s
E
s
=iomH
z
(11)
V
s
H
s
= ioeE
z
(12)
Substituting Eqs. (9) and (10) into Eqs. (11) and (12), we
have
(V
2
s
k
2
s
)H
z
=0 for TE waves (13)
(V
2
s
k
2
s
)E
z
=0 for TM waves (14)
where k
2
s
=k
2
k
2
z
: Therefore, H
z
and E
z
satisfy a two-di-
mensional scalar wave equation, also known as the re-
duced-wave equation. Once E
z
and H
z
are solved for from
Eqs. (13) and (14), we can derive all the other eld com-
ponents using Eqs. (9) and (10).
2.2. TEM Mode of a Transmission Line
For TEM waves, both E
z
and H
z
are zero. According to Eqs.
(9) and (10), H
s
and E
s
will be nonzero only if k
z
=k.
Therefore, all TEM waves, or TEM modes in a waveguide,
have e
ikz
dependence, where the propagation constant in
the guiding structure is equal to that in free space. Fur-
thermore, from Eqs. (11) and (12), we conclude that, for
TEM waves
V
s
E
s
=0; V
s
H
s
=0 (15)
These equations state that both the E
s
and H
s
are irro-
tational elds in the xy plane. Any irrotational eld can be
written as the gradient of a scalar function, as in the case
of the electrostatic eld E= Vf, where f is the electric
potential. In other words, E
s
is electrostatic in the xy
plane while H
s
is magnetostatic in the xy plane. Hence, we
can let
E
s
= V
s
f
s
(x; y)e
ikz
; H
s
= V
s
c
s
(x; y)e
ikz
(16)
Since V
.
E
s
=0 and V
.
H
s
=0 in the waveguide, f
s
and c
s
satisfy Laplaces equations
V
2
s
f
s
(x; y) =0; V
2
s
c
s
(x; y) =0 (17)
If we have perfect electric conductors, the boundary con-
ditions are ^ nnE
s
=0, and ^ nn
.
H
s
=0 on the metallic sur-
faces. These boundary conditions translate to
f
s
=constant (18)
^ nn
.
V
s
c
s
=
@
@n
c
s
=0 (19)
Equation (18) is known as the Dirichlet boundary condi-
tion, while Eq. (19) is the Neumann boundary condition.
The constants in Eq. (18) are the potentials on the con-
ductors, which may differ for different conductors.
At this point, it seems that f
s
and c
s
are decoupled, and
hence, the electric eld and the magnetic eld are inde-
pendent of each other. This could not be true, as the cou-
pling is expressed in Eqs. (6) and (7) (if we set E
z
=H
z
=0
for discussing TEM modes). Hence, the coupling of the
elds is only through the ^ zz variation of the elds:
@
@z
^ zz E
s
=iomH
s
(20)
@
@z
^ zz H
s
= ioeE
s
(21)
Furthermore, from these equations, we deduce that H
s
and E
s
are mutually orthogonal in the TEM mode. Since
the elds have e
ikz
dependence, we conclude that
^ zz E
s
=
m
e
_
H
s
=ZH
s
(22)
if we assume that the wave is propagating in only one di-
rection, then Z is also known as the intrinsic impedance of
the medium, and all TEM waves satisfy Eq. (22).
We can integrate Eq. (20) about a line contour around
one of the conductors that made up the transmission line
to obtain
@
@z
_
C
^ zz E
s
.
dl =iom
_
C
H
s
.
dl (23)
By Amperes law, we have
_
C
H
s
.
dl =I; the total current
on one of the conductors. For the left-hand side, we have
_
C
^ zz E
s
.
dl =
_
C
dl
.
^ zz V
s
f
s
=
_
S
dS^ zz
.
V
s
^ zz V
s
f
s
:
(24)
The second equality follows from Stokes theorem. Using
^ zz
.
V
s
^ zz V
s
f
s
=V
2
s
f
s
=0; we have
_
C
^ zz E
s
.
dl =
_
S
dSV
2
s
f
s
(25)
V
2
s
f
s
= r=e; the right-hand side of Eq. (25) evaluates to
Q/e, where Q is the total charge per unit length on one
conductor. Hence, Eq. (25) becomes
_
C
^ zz E
s
.
dl =
Q
e
(26)
Therefore, Eq. (23) becomes
d
dz
Q=iomeI (27)
Since the transverse eld is purely static, we can dene Q
=CV, where C is the capacitance per unit length and V=
V
1
V
2
. Hence, Eq. (27) becomes
d
dz
V =io
me
C
I (28)
Since me/C has the dimension of henries per meter, we can
dene L=me/C, an inductance per unit length, and Eq.
(28) becomes
dV
dz
=ioLI (29)
With manipulations similar to ^ zz (21), we obtain
dI
dz
=ioCV (30)
Equations (29) and (30) are the telegraphist equations for
a transmission line, which can also be derived from a cir-
cuits viewpoint.
Since the elds, and hence the voltage and current,
have e
ikz
dependence, where k=o

me
_
=o
LC
_
; we deduce
either from Eq. (29) or (30) that
V
I
=
L
C
_
=Z
0
(31)
if the wave is propagating only in the positive z direction.
Z
0
is also known as the characteristic impedance of a
transmission line. Since C and L are dependent on the
geometry of the transmission line, Z
0
is a geometry-de-
pendent impedance. This is unlike Z, the intrinsic imped-
ance. It can be easily shown from Eqs. (29) and (30) that
d
2
V
dz
2
o
2
LCV =0 (32)
d
2
I
dz
2
o
2
LCI =0 (33)
which are one-dimensional scalar wave equations, whose
solutions are given by
V(z) =V
e
ikz
V
e
ikz
(34)
I(z) =I
e
ikz
I
e
ikz
(35)
These constants are related through V
/I
=V
/I
=Z
0
.
The velocity of the wave on the line is given by
v =
1
me
_ =
1
LC
_ (36)
The time-average energy stored per unit length in a trans-
mission line for a single propagating wave is given by
W
e
) =
1
4
e
_
S
E
s
.
E
+
s
dS=
1
4
e
_
S
(V
s
f
s
)
2
dS (37)
W
m
) =
1
4
m
_
S
H
s
.
H
+
s
dS=
1
4
m
_
S
(V
s
c
s
)
2
dS (38)
where /W
e
S and /W
m
S are the time average energy
stored in the electric eld and the magnetic eld, respec-
tively. Using the fact that V
.
(fVf) =(Vf)
2
fV
2
f, we can
write
W
e
) =
1
4
e
_
C
1
C
2
f
@
@n
fdl =
1
4
(V
1
V
2
)Q=
1
4
CV
2
(39)
where V=V
1
V
2
, Q=CV. This equation could also be
derived from circuits theory. As we have shown before,
[E
s
[ =
m
e
_
[H
s
[; hence the time-average energy stored in the
magnetic eld is
W
m
) =W
e
) =
1
4
CV
2
(40)
Since V
2
=Z
2
0
I
2
=
L
C
I
2
; and me =LC; we can also write
W
m
) =
1
4
LI
2
(41)
Equation (41) can also be established by circuit theory. It
also establishes our definition of L as an inductance per
unit length.
The time-average power ow down a transmission line
is given by
P) =
1
2
+e
_
S
dS^ zz
.
(E
s
H
+
s
) (42)
Since H
s
=
(e=m)
_
^ zz E
s
from Eq. (22), we have
P) =
1
2
e
m
_ _
S
dS[E
s
[
2
=2vW
e
) =vW
e
W
m
) (43)
Hence, the time-average stored energy /W
e
W
m
S mov-
ing at velocity v contributes to power ow. Equation (43) is
also valid for non-TEM modes and more general (but loss-
less) situations, if v is replaced by v
g
, the group velocity.
The group velocity is dened as v
g
=(dk
z
/do)
1
. In the
TEM case, k
z
=k =o

me
_
; so that v
g
=v. The group velocity
is the velocity of energy propagation in a waveguide, and it
is also the signal velocity. Equations (29) and (30) can be
also derived from a simple circuits model assuming dis-
tributed inductances and capacitances along the line.
2.3. Lossy Transmission Lines
To characterize a lossy transmission line, we replace the
series impedance per unit length ioL by ioLR, and
the shunt admittance per unit length ioC by ioCG,
where R is the series resistance per unit length in the
conductor, while G is the shunt conductance per unit
length in the dielectric insulator. The series resistance re-
sults from the nite conductivity of the (imperfect) con-
ductors, while the shunt conductance results from the
imperfect insulation of the dielectrics. Usually, the series
resistance is a much more important attenuation factor
than the shunt conductance.
Figure 2 illustrates the distributed circuit model for the
transmission line. Together with L and C, R and G con-
stitute the primary transmission-line constants. In con-
trast, the velocity of propagation v and the characteristic
impedance of the line Z
0
are usually called secondary con-
stants. The telegraphist equations then become
dV
dz
=(ioL R)I = ZI (44)
dI
dz
=(ioC G)V = YV (45)
The characteristic impedance is now
Z
0
=
Z
Y
_
(46)
while the propagation constant becomes
k
2
= (ioL R) (ioC G);
k =o
LC 1
iR
oL
_ _
1
iG
oC
_ _
(47)
Hence k is complex and the wave e
ikz
is attenuating. More-
over, with loss, even if the primary constants are indepen-
dent of frequency, the secondary parameters will be
frequency-dependent.
Note that, strictly speaking, when we have a lossy
transmission line due to conductor losses, a pure TEM
wave cannot exist. This is because the axial current ow
meets a resistance, and hence, an axial component of the
electric eld is necessary now to drive a current in the con-
ductor. Therefore, the eld is only quasi-TEM. However, the
conductor loss can be regarded as a small perturbation of
the perfect conductor case, and the electromagnetic eld
in the lossy line will not differ much from that of a
lossless case.
The shunt conductance G in a lossy line can be found as
follows. If the capacitance per unit length between two
conductors is given by the formula
C=eK (48)
where K is a geometry-dependent factor, the shunt admit-
tance would be given by Y= ioC= ioeK. If now the
dielectric medium is lossy so that e =e
/
(is/o), then the
shunt admittance is given by
Y = ioe
/
K sK (49)
Hence, we identify G=sK. Note that the derivations in
Eqs. (20)(30) hold true even if e is complex. For this rea-
son, Y in Eq. (33) is exact.
The series resistance R can be found by calculating the
resistance of the conductor in a perturbative manner
when it is lossy. The skin effect will conne the current
to ow only on the surface of the conductor. Since the skin
depth in a conductor is d =
2=oms
_
, the current is con-
ned to ow in a thinner region at higher frequencies,
hence increasing this series resistance.
Another way of calculating transmission-line loss is via
a perturbation argument and the use of energy conserva-
tion. If a transmission line is lossy such that k=k
/
ik
//
,
and
V; I e
ik
/
zk
//
z
(50)
then, the power ow in a line, which is proportional to
|V|
2
or |I|
2
, is
P e
2k
//
z
(51)
By energy conservation, we obtain
dP
dz
= P
d
= 2k
//
P (52)
where P
d
is the power dissipated per unit length on the
line. Therefore, the attenuation constant k
//
can be derived
to be
k
//
=
P
d
2P
(53)
if we know P
d
. We can assume P to be close to that of a
lossless line in using Eq. (53) in a perturbative concept.
2.4. Distributed Parameters
From Eqs. (34) and (35), the impedance of a lossless trans-
mission line at any given point is given by
Z(z) =
V(z)
I(z)
=Z
0
e
ikz
G
0
e
ikz
e
ikz
G
0
e
ikz
(54)
where G
0
=V
/V
is the reection coefcient, which can

be found by using a known value of Z(z) at some point on
the line. For instance, if the line is terminated with a load
Z
L
at z =0, then Z(0) =Z
L
and, substituting in Eq. (54), we
have
G
0
=
Z
L
Z
0
Z
L
Z
0
(55)
Furthermore, the line impedance can be written at any
location along the line
Z(z) =Z
0
Z
L
iZ
0
tan bz
Z
0
iZ
L
tan bz
(56)
with zo0, while the reection coefcient everywhere
along the line (generalized reection coefcient) G(z) is
given by Eq. (55) with Z
L
substituted by Z(z) above. The
representation of the variation of the normalized imped-
ance along the line with the angle of the generalized re-
ection coefcient, that is, the relation
G=
Z Z
0
ZZ
0
for all Z, such that Re[Z]Z0, in the G plane, is called the
Smith chart [17,18]. It provides a useful graphical aid to
the analysis of transmission-line and matching problems
that are otherwise cumbersome if treated in analytical
form.
The addition of two waves, V
and V
, propagating in
opposite directions gives rise to a standing-wave pattern
along the line. The ratio of the minimum and maximum
values of the voltage of the standing-wave pattern denes
L
C
l <<
R
G
L
R
C G
L
R
C G
Figure 2. Distributed equivalent circuit representation of a transmission line supporting the TEM
mode. Each section Dl is modeled by a resistance R per unit length (O/m), an inductance L per unit
length (H/m), a capacitance C per unit length (F/m), and a conductance G per unit length (G/m).
the voltage standing-wave ratio (VSWR), given by
VSWR=
1[G
0
[
1 [G
0
[
(57)
Transmission lines are usually designed to maximize
transmission and minimize reection over the frequency
band of interest. Therefore, in practical design situations,
we have VSWRC1.
2.5. Transformation Matrices
In a circuit, a transmission line may interact with other
transmission lines, either by design, as in microwave in-
tegrated circuits (MICs), or as a result of undesired effects,
as in closely packed, high-frequency circuit interconnec-
tions [19,20] on computer boards. We use the term cou-
pling when such property is desired, or planned by design
[21,22]. Coupled integrated transmission lines form the
basic building blocks for a series of passive microwave el-
ements such as lters and directional couplers [15]. In the
case of an undesired interaction, we usually employ the
term crosstalk. In both cases, one can analyze the N-trans-
mission-line problem as a general microwave circuit con-
sists of many ports. A convenient way to characterize an
N-port network is to describe the network in terms of
impedance matrices or admittance matrices. For example,
if the N-port network can be characterized by a voltage
current pair at each port (owing into), then a column
vector of voltages can be dened and also a column vector
of currents. We can express a relationship between the
voltages and the currents as
V=
V
1
V
2
.
.
.
V
N
_
_
_
_
=
Z
11
Z
12
Z
1N
Z
21
Z
22
Z
2N
.
.
.
.
.
.
.
.
.
.
.
.
Z
N1
Z
N2
Z
NN
_
_
_
_
I
1
I
2
.
.
.
I
N
_
_
_
_
=
"
ZZ
.
I (58)
By the same token, we can express
I =
"
YY
.
V (59)
where
"
YY is the admittance matrix, which dene the Y pa-
rameters. For reciprocal circuits, it can be shown that
"
ZZ
and
"
YY are symmetric matrices, so that there are only three
independent matrix elements in a two-port network. For
lossless circuits, it can be shown that these matrices have
pure imaginary elements.
For high frequencies, however, it is more pertinent to
think about waves. Then, at each port, we can dene an
incident and a reected wave. For instance, we can dene
an incident voltage wave V
and a reected voltage wave

V
. A relationship can then be express between the re-

ected waves at all the ports to the incident waves at all
the ports. This constitutes the scattering matrix:
V=
V
1
V
2
.
.
.
V
N
_
_
_
_
=
S
11
S
12
S
1N
S
21
S
22
S
2N
.
.
.
.
.
.
.
.
.
.
.
.
S
N1
S
N2
S
NN
_
_
_
_
V
1
V
2
.
.
.
V
N
_
_
_
_
=
"
SS
.
V
(60)
It can be proved that the matrix
"
SS has to be symmetric
for reciprocal circuits, and that it has to be unitary if the
circuit is lossless. The S
ij
are called scattering or simply
S-parameters. S parameters are very important in the
characterization of a microwave circuits. CAD (computer-
aided design) packages for the analysis of complex net-
works are usually based on the availability of S parame-
ters (or Y parameters) for each of their building blocks
(such as the transmission lines).
When one needs to cascade a series of two-port net-
works, it is more convenient to work with chain matrices
or transmission matrices. A voltagecurrent transmission
matrix, usually called an ABCD matrix, relates the volt-
age and current at one port to the voltage and current at
the second port.
Written explicitly, we have
V
1
I
1
_ _
=
A
1
B
1
C
1
D
1
_ _
V
2
I
2
_ _
(61)
Notice that in the convention for the ABCD matrix, the
current at port 2 is owing out of the port rather than into
the port. In this manner, if we have a second transmission
matrix of a second network that relates V
2
, I
2
to V
3
, I
3
namely
V
2
I
2
_ _
=
A
2
B
2
C
2
D
2
_ _
V
3
I
3
_ _
(62)
then, when these two networks are cascaded together, the
resultant transmission matrix is the product of the two
matrices
V
2
I
2
_ _
=
A
1
B
1
C
1
D
1
_ _
A
2
B
2
C
2
D
2
_ _
V
3
I
3
_ _
(63)
For a two port network, it can be shown that
A=
Z
11
Z
12
; B=
Z
11
Z
22
Z
2
12
Z
12
(64a)
C=
1
Z
12
; D=
Z
22
Z
12
(64b)
for a reciprocal network. It is also readily veried that
ADBC=1 (65)
for this case. Hence, the determinant of a chain matrix is
always 1 for a reciprocal network.
3. HOLLOW WAVEGUIDES
In this section, we will focus on a typical class of high-fre-
quency transmission line, the hollow waveguide [10,23].
Hollow waveguides are specially useful in low-loss and
high-power systems, such as the feeding system of an an-
tenna. Hollow waveguides do not support a TEM mode,
and the analysis of the previous section deriving the cir-
cuit parameters (line and voltage) along the line in terms
of the TEM elds does not apply in this case. The absence
of a TEM mode in a hollow waveguide can be proved by
contradiction as follows. If we assume that it does, then
the magnetic eld H=H
s
has to end on itself due to the
absence of magnetic charges. It is clear that
_
C
H
s
.
dlO0
about any closed contour following the magnetic eld
lines. This is clearly in violation of Eq. (15) for a TEM
mode, which implies that if
_
H
s
.
dl =0 if C does not en-
close any conducting current. These two results are con-
tradictory, implying the absence of a TEM mode in a
hollow waveguide.
3.1. Rectangular Waveguides
The rectangular waveguide is the most commonly used
hollow waveguide. It is illustrated in Fig. 3. By adjusting
the aspect ratio, a to b, of the waveguide, one can obtain a
good bandwidth for single-mode propagation (the aspect
ratio of
1
2
is used in most practical waveguides). Further-
more, the analysis of this waveguide is the simplest.
3.1.1. TE Modes (H Modes). ATE mode in a rectangular
waveguide is characterized by H
z
satisfying Eq. (13) with
the requisite Neumann boundary condition. An H
z
that
will satisfy Eq. (13) with (@/@n)H
z
=0 on the waveguide
wall is
H
z
=H
0
cos
mp
a
x
_ _
cos
np
b
y
_ _
e
ik
z
z
(66)
where k
2
s
=(mp=a)
2
(np=b)
2
; k
z
=
k
2
k
2
s
_
: This last re-
lation is called the dispersion relation. A dispersion rela-
tion gives the propagation constant as function of
frequency: k
z
=k
z
(o). The transverse elds can be found
using Eqs. (9) and (10). The mode becomes evanescent or
nonpropagating when k
2
s
> k
2
, that is, when k
z
becomes
imaginary. Since k
2
=o
2
me, the cutoff frequency (i.e., the
frequency below which the mode is evanescent) is given by
o
mn;c
=
1
me
_
mp
a
_ _
2
np
b
_ _
2
_ _
1=2
(67)
The corresponding mode is usually labeled as TE
mn
(or
H
mn
) mode. Note that this is in marked contrast for the
case of TEM propagation. For the TEM mode, no cutoff
frequency exists. The wavelength of a wave at o
c
in the
medium denoted by m, e is the cutoff wavelength:
l
mn;c
=
2
[(m=a)
2
(n=b)
2
]
1=2
(68)
In waveguide conventions, a is assumed larger than b.
Then, the dominant mode (fundamental mode) with the
lowest cutoff frequency is the m=1, n=0 mode, also
known as the TE
10
mode (or H
10
mode). The TE
00
mode
is nonexistent because, otherwise, k
s
=0, and E
s
and H
s
would diverge from Eqs. (9) and (10) when k
s
-0. The
waveguide usually operates at those frequencies where
only the fundamental mode can propagate.
3.1.2. TM Modes (E Modes). A TM mode in a rectangu-
lar waveguide is characterized by E
z
satisfying Eq. (14)
with the requisite Dirichlet boundary condition. An E
z
that will satisfy Eq. (14) with E
z
=0 on the waveguide wall
is
E
z
=E
0
sin
mp
a
x
_ _
sin
np
b
y
_ _
e
ik
z
z
(69)
where k
2
s
=(mp=a)
2
(np=b)
2
; k
z
=
k
2
k
2
s
_
: The TM
mn
mode has the same cutoff frequency as the TE
mn
mode.
However, when either m=0, or n=0, the mode does not
exist since E
z
=0 then. Therefore, the lowest TM mode is
the TM
11
mode with a cutoff frequency above that of the
TE
10
mode. Given the z components of the elds, all other
eld components of a waveguide can be derived.
3.2. Circular Waveguides
Certain modes of a circular waveguide have less attenu-
ation from wall loss compared to a rectangular waveguide.
Hence, circular waveguides are sometimes preferred over
a rectangular waveguide. The geometry of a circular hol-
low waveguide is illustrated in Fig. 4.
3.2.1. TE Modes (H Modes). The E
z
component of a TE
mode satises Eq. (13) in cylindrical coordinates
1
r
@
@r
r
@
@r

1
r
2
@
2
@f
2
k
2
s
_ _
H
z
=0 (70)
with the boundary condition that (@/@r)H
z
=0, r =a,
where a is the radius of the waveguide. If we assume
that H
z
has e
7inf
, sinnf or cos nf dependence, where n is
an integer, we can replace @
2
/@f
2
by n
2
. The solution of
y
b
z
a
x
Figure 3. Geometry of a rectangular hollow waveguide.
Eq. (70) then becomes
H
z
=H
0
J
n
(k
s
r)e
infikzz
(71)
where J
n
is the Bessel function. We require that (@/@r)H
z
=0 at r =a, implying that
J
/
n
(k
s
a) =0 (72)
with k
z
=
k
2
k
2
s
_
: If the mth zero of J
/
n
(x) is dened to be
b
nm
such that J
/
n
(b
nm
) =0; then the values of possible k
s
are
k
s
=
b
nm
a
(73)
The subscript n denotes the orders of the Bessel function
J
n
(x) and the circular harmonic e
7inf
. The subscript m
denotes the mth zero of J
/
n
(x) discounting the zero at the
origin. The corresponding mode is usually denoted as the
TE
nm
mode. Cutoff occurs when k =o

me
_
ok
s
: From the
tabulated data for b
nm
, it follows that the TE
11
modes cor-
responding to the rst zero of J
1
(x) have the lowest cutoff
frequency. The cutoff frequency for the TE
nm
mode is given
by
o
nm;c
=
1
me
_
b
nm
a
(74)
and the corresponding cutoff wavelength is
l
nm;c
=
2p
b
nm
a (75)
3.2.2. TM Modes (E Modes). Similar to a TE mode, the
E
z
component of a TM mode has e
7inf
dependence, satis-
fying the equation
1
r
@
@r
r
@
@r

n
2
r
2
k
2
s
_ _
E
z
=0 (76)
with the boundary condition E
z
(r=a) =0. Hence
E
z
=E
0
J
n
(k
s
r)e
infik
z
z
(77)
with J
n
(k
s
a) =0. If we denote the mth zero of J
n
(x) by a
nm
,
then k
s
has possible values of
k
s
=
a
nm
a
(78)
where the subscript n denotes the order of the Bessel
function J
n
(x) and the subscript m denotes the mth zero of
J
n
(x), discounting the zero at the origin. The correspond-
ing mode is known as the TM
nm
mode. The cutoff frequen-
cy of the TM
nm
mode is given by
o
nm;c
=
1
me
_
a
nm
a
(79)
and the corresponding cutoff wavelength is
l
nm;c
=
2p
a
nm
a (80)
From the tabulated data for a
nm
, it follows that the lowest
TM mode is the TM
01
mode, but it has a higher cutoff fre-
quency compared to the TE
11
mode.
It can also be shown that the TE
01
mode has the lowest
loss at high frequencies. The TE
01
is axially symmetric
with E=
^
ffE
f
: An interesting property of this mode is that,
contrary to other modes, the loss decreases monotonically
as frequency increases.
The transverse eld components of a circular wave-
guide are easily obtained given the axial components. One
disadvantage of circular waveguides is that, because of its
circular symmetry, the wave polarization on the transver-
sal plane may change as the wave propagates, as a result
of small perturbations. One way to combat this problem is
to employ elliptical waveguides [24].
3.3. Other Closed Waveguides
If a metallic waveguide has an arbitrary cross section
whose shape does not fall on any of the curvilinear coor-
dinates, one has, in general, to resort to numerical tech-
niques to nd the modes and dispersion curves. These will
be discussed later. Among these other waveguides, ridge
waveguides are commonly used [25,26]. They are attrac-
tive because the frequency range of their dominant mode
exceeds that of a rectangular waveguide of same external
dimensions (the cutoff is lowered). The tradeoff is for an
increased loss from the walls.
When a uniform waveguide is lled with inhomoge-
neous materials [27], the guided modes of the structure
cannot be decomposed into TE and TM waves, except for
some very special cases. In other words, the E
z
and H
z
components of the elds are always coupled together. Such
modes are also called the hybrid modes. This coupling can
be shown from Maxwells equations [10]. The fact that the
E
z
and H
z
waves are in general coupled at a dielectric in-
terface is also known as the depolarization effect.
y
z
x
a
Figure 4. Geometry of a cylindrical hollow waveguide.
In general, inhomogeneous lled waveguides can be
used to make variable phase shifters and attenuators. If
the inhomogeneity lling the waveguide is nonreciprocal
(like ferrite), then isolators, gyrators, and attenuators can
be made.
An example of inhomogeneous lled waveguide is the
dielectric-slab-loaded waveguide. The dispersion relation
for this structure can be easily derived using the trans-
verse resonance condition [10].
4. OPEN WAVEGUIDES
In an open waveguide [2832] a wave is guided along a
structure, but the eld is not enclosed completely in the
structurethe eld extends to innity. However, the eld
decays exponentially away from the guiding structure;
hence, the energy of the wave is still localized around the
guiding structure. The guiding structure is often lled
with inhomogeneous medium. Therefore, many properties
of inhomogeneously lled waveguides are also true in
open, inhomogeneously lled waveguides.
4.1. Dielectric Waveguides
An example of an open waveguide is a dielectric slab
waveguide. The waveguide is made with dielectric coat-
ing on a ground plane, or in the case of optical thin-lm
waveguides, it is a coating of an optically more dense
medium on top of an optically less dense substrate, as
illustrated in Fig. 5.
Because of the symmetry of the geometry, we can de-
compose the eld inside such a waveguide into TM and TE
types [33]. The mode is guided by total internal reection.
This is possible only if e
1
4e
0
and e
1
4e
2
. At total internal
reection, the elds in regions 0 and 2 are evanescent, and
hence they decay exponentially away from the structure.
Therefore, most of the energy of the mode is still trapped
and localized in the vicinity of the structure.
If a TM wave is in a dielectric slab, we can write the
eld in region 1 as
H
1
= ^ yy[A
1
e
ik
1x
x
B
1
e
ik
1x
x
]e
ik
z
z
(81)
Equation (1) has the physical meaning that the wave in
region 1 is representable as bouncing waves. At x =0, the
upgoing wave is the reection of the downgoing wave;
hence, we have
A
1
=R
TM
12
B
1
(82)
where R
TM
12
is the TM reection coefcient at the 12 in-
terface. If there are subsurface layers, R
TM
12
could be the
generalized reection coefcient that includes subsurface
reections. Otherwise, it is just the single interface, Fres-
nel reection coefcient for a TM wave. At the upper in-
terface at x =d, we require that the downgoing wave be a
reection of the upgoing wave:
B
1
e
ik1xd
=R
TM
10
e
ik1xd
A
1
(83)
For nontrivial A
1
and B
1
, Eqs. (82) and (83) imply that
1 R
TM
10
R
TM
12
e
2ik
1x
d
=0 (84)
This is the guidance condition, sometimes known as the
transverse resonance condition for TM modes in a dielec-
tric slab, with
R
TM
ij
=
e
j
k
ix
e
i
k
jx
e
j
k
ix
e
i
k
jx
; k
ix
=
k
2
i
k
2
z
_
(85)
Because of Eq. (85), the guidance condition expressed by
Eq. (84) can be expressed entirely as a function of k
z
. We
can solve Eq. (84) either graphically or numerically on a
computer. Once the value of k
z
that satises Eq. (84) is
found, it can be used in Eq. (82) or (83) to nd a relation-
ship between A
1
and B
1
. The elds in region 0 and 2 can be
found easily:
H
0
= ^ yyT
TM
10
A
1
e
ik1xdik0x(xd) ikzz
(86a)
H
2
= ^ yyT
TM
12
B
1
e
ik
2x
x ikzz
(86b)
In other words, the eld in region 0 is a consequence of the
transmission of the upgoing wave in region 1, while the
eld in region 2 is a consequence of the transmission of the
downgoing wave in region 1. In the above, T
ij
is a trans-
mission coefcient with
T
TM
ij
=1R
TM
ij
(87)
In order for a mode to be trapped, the eld has to decay
exponentially in the x direction. Therefore, k
0x
and k
2x
have to be purely imaginary. In other words, we can nd
the values of k
z
for guidance only in the ranges k
z
4k
0
and
k
z
4k
2
.
For a symmetric waveguide where regions 0 and 2 are
the same, the rst few modes of the waveguides are as
sketched in Fig. 6. These modes are, in general, better
Region 0
Region 1
Region 2
x
d
y
z
1
>
2
1
>
3
Figure 5. Geometry of a dielectric slab waveguide.
The electromagnetic eld is guided by total internal
reection. This is possible only if e
1
4e
2
and e
1
4e
3
.
trapped when the contrast is high between region 1 and
regions 0 and 2, or when the frequency is high. We can
study the guidance of a TE mode that is the dual of a TM
mode in a similar fashion.
By a further manipulation of Eq. (84), and using the
definition of Fresnel reection coefcients, Eq. (84) can be
written as
a
0x
d
2
=
e
0
e
1
k
1x
d
2
tan
k
1x
d mp
2
_ _
(88)
where
a
0x
=
k
2
z
k
2
0
_
; k
1x
=
k
2
1
k
2
z
_
; andm=1; 2; 3; 4; . . . : The
left-hand side of Eq. (88) can be expressed in terms of
the k
1x
d variable
d
2
k
2
z
k
2
0
_
=
(k
2
1
k
2
0
)
d
2
_ _
2
k
1x
d
2
_ _
2
(89)
which is the equation of a circle. Equation (88) can hence
be solved graphically by plotting both sides of Eq. (88) as a
function of k
1z
d.
A mode in a dielectric has a part of its energy outside
the waveguide and inside the waveguide. Hence, the
group velocity of the mode is between that of the slab re-
gion and the outer region. When the frequency is very low,
the mode is weakly evanescent outside the dielectric slab
and the group velocity of a mode is closer to that of the
outer region because most of the energy of the mode is
outside the waveguide. When the frequency is high, the
mode is strongly evanescent outside and most of the en-
ergy of the mode is trapped inside the slab. Hence, the
group velocity of the mode is close to that of the dielec-
tric slab.
The TM
0
mode has no cutoff, since continuity of the
slope and eld amplitude can be satised for all frequen-
cies, and yet the eld is evanescent outside. This is not
true of the higher-order modes. For high frequencies, the
higher-order modes are well trapped inside the dielectric
waveguide. When the frequency is lowered, however, the
eld may cease to be evanescent outside the dielectric
slab. At the frequency when the eld becomes constant
outside, we have the cutoff. At frequencies lower than the
cutoff frequency, the mode leaks energy to outside the slab,
it becomes a leaky mode and is not guided at all.
Many other congurations of dielectric waveguides are
possible. Dielectric waveguides found many applications
in the millimiter-wave and optical frequency ranges. One
problem of these waveguides is the radiation loss incurred
at bends and discontinuities. This can be minimized by
utilizing a higher dielectric constant, but at the expense of
a much smaller size of the structures, which leads to man-
ufacturing problems. Analytical solutions in general are
not available for these structures, and a numerical anal-
ysis is necessary to nd the modes and dispersion rela-
tions. Moreover, in these structures, the TE and TM modes
are coupled by the boundary conditions at the dielectric
boundaries. For this reason, none of the modes are strictly
TE or TM, and the EH notation is used to denote a mode
whose TE component dominates over its TM component,
and vice versa for the HE notation.
Avery popular geometry of dielectric waveguides is the
circular dielectric waveguide, which is discussed next.
4.2. Circular Dielectric Waveguide
The geometry of a circular dielectric waveguide is illus-
trated in Fig. 7. An optical ber is a kind of circular di-
electric waveguide. Usually, an optical ber has a
protective cladding as well. In the ensuing analysis, we
will ignore the protective cladding, letting b-N.
In order for a mode to be guided, we require that
e
1
m
1
4e
2
m
2
. In other words, the light velocity in the core
region has to be slower than the light velocity in the clad-
ding. The eld outside the core region is evanescent for a
guided mode. Therefore, letting b-N affects the guided
mode slightly, especially if the mode is tightly bound to the
core; hence, it is a good approximation.
In the optical ber modes, except for the axisymmetric
modes, the TE and TM elds are coupled to each other by
the boundary conditions as in an inhomogeneously lled
waveguide. The z components of the eld are solutions to
H
y
H
y H
y
H
y
TM
0
Mode
TM
2
Mode TM
3
Mode
TM
1
Mode
d/2 d/2
d/2
d/2
d/2
x
d/2
x
d/2
x
x
d/2
Figure 6. Field distribution for different
modes in a dielectric slab waveguide. The
amplitude scale is arbitrary. The TM
0
has no cutoff frequency, but the other
modes have a nite cutoff frequency.
the wave equation in cylindrical coordinates for each of
the homogeneous regions.
We will analyze next the circular dielectric waveguide
using the weakly guided approximation [34]. The weakly
guided approximation is quite useful in analyzing optical
bers, since the dielectric constant contrast from core to
cladding is usually very low.
4.3. Weakly Guided Approximation
Starting from Maxwells equations, we can derive the vec-
tor wave equations governing electromagnetic elds in an
inhomogeneous waveguide
Vm
1
r
VEo
2
m
0
e
0
e
r
E=0 (90a)
Ve
1
r
VHo
2
m
0
e
0
m
r
H=0 (90b)
where m
r
=m(r)/m
0
=f
1
(r), and e
r
=e(r)/e
0
=f
2
(r). If we nd a
solution to Eq. (90a), the solution to the dual problem is
obtained by letting E-H, H-E, m
r
-e
r
, e
r
-m
r
. Howev-
er, if f
1
(r)af
2
(r), the dual problem corresponds to a differ-
ent waveguide with m
r
=f
2
(r) and e
r
=f
1
(r). In order for a
dual problem to be itself, we require that f
1
(r) =f
2
(r). If
this is the case, then a mode and its dual are degenerate.
Therefore, we can associate every mode in a waveguide
with a dual mode. However, f
1
(r) =1 usually, while
f
2
(r)a1; therefore, a mode is not exactly degenerate with
its dual.
However, in the case of a weakly guided dielectric
waveguide, e
r
-1; hence, a mode is nearly degenerate
with its dual. Furthermore, we can show that the vector
nature of the wave is unimportant. If m
r
=1, we can re-
write Eqs. (90a) and (90b) as
VVEo
2
m
0
e
0
e
r
E=0 (91a)
VVH(Vlne
r
) VHo
2
m
0
e
0
e
r
H=0 (91b)
where Vlne
r
=Ve
r
/e
r
is the effect of the polarization charg-
es at the dielectric interface. If the transverse and longi-
tudinal components of Eqs. (91a) and (91b) are extracted,
the following equations ensue:
V
2
E
s
V
s
[V
s
lne
r
.
E
s
] k
2
E
s
=0 (92a)
V
2
E
z
ik
z
(V
s
lne
r
)
.
E
s
k
2
E
z
=0 (92b)
V
2
H
s
(V
s
lne
r
) (V
s
H
s
) k
2
H
s
=0 (93a)
V
2
H
z
(V
s
lne
r
) (V
s
H
z
) (V
s
lne
r
)
( ^ zzik
z
H
s
) k
2
H
z
=0
(93b)
In the limit when e
r
-1, and o-N, the polarization
charge terms in Eqs. (92a) and (93a), which are propor-
tional to the rst derivative of the eld, can be ignored
with respect to the other terms, yielding
V
2
E
s
k
2
E
s
=0 (94a)
V
2
H
s
k
2
H
s
=0 (94b)
In other words, the wave guidance problem by a ber of
weak contrast in the high-frequency limit reduces to a
scalar problem.
In the limit when the contrast e
r
-1, the wave that
propagates in the ber is a TEM wave where E
z
=0. When
e
r
41, we see from Eq. (92b) that E
z
B|E
s
|/o. Therefore,
when o-N, then |E
z
|5|E
s
|. By the same token,
|H
z
|5|H
s
| when o-N. Therefore, when the contrast
is very low, and the frequency is very high, the mode is
quasi-TEM, and V
s
.
E
s
E0 and V
s
.
H
s
E0.
Equations (94a) and (94b) are equivalent to
(V
2
k
2
)f=0 (95)
where f is either E
x
, E
y
, H
x
or H
y
.
For example, if we assume cylindrical geometry, we can
let
f=
AJ
n
(k
1r
r)e
infik
z
z
; roa
BK
n
(a
2
r)e
infikzz
; r > a
_
(96)
The boundary conditions for f at the interface where k
2
=
o
2
m
0
e
0
e
r
displays a step discontinuity is
f
1
=f
2
(97a)
^ nn
.
Vf
1
= ^ nn
.
Vf
2
(97b)
These boundary conditions are derivable from Eq. (95)
alone.
Imposing these boundary conditions at r =a for the
weakly guided dielectric waveguide, whose eld is given
z
2a
Core
Cladding
b
1
,
1

2
,
2
0
,
0
y
Figure 7. Geometry of a circular dielectric waveguide. To obtain
a guided mode, the dielectric constant in the core should be great-
er than in the cladding, e
1
4e
2
. The eld is guided by total internal
reection. An optical ber is an example of circular dielectric
waveguide. In this case, the contrast between the dielectric con-
stants e
1
and e
2
is usually very low.
by Eq. (96), we have
AJ
n
(k
1r
a) =BK
n
(a
2
a) (98a)
AK
1r
J
/
n
(k
1r
a) =Ba
2
K
/
n
(a
2
a) (98b)
which yields
k
1r
J
/
n
(k
1r
a)
J
n
(k
1r
a)
=
a
2
K
/
n
(a
2
a)
K
n
(a
2
a)
(99)
Using the recurrence relationship that J
/
n
(x) =
J
n1
(x)
n
x
J
n
(x); and that K
/
n
(x) = K
n1
(x)
n
x
K
n
(x); we
can transform this to
k
1r
J
n1
(k
1r
a)
J
n
(k
1r
a)
=
a
2
K
n1
(a
2
a)
K
n
(a
2
a)
(100)
Similarly, using the recurrence relationship that
J
/
n
(x) =J
n1
(x)(n=x)J
n
(x); and K
/
n
(x) = K
n1
(x) (n=x)
K
n
(x); we have
k
1r
J
n1
(k
1r
a)
J
n
(k
1r
a)
=
a
2
K
n1
(a
2
a)
K
n
(a
2
a)
(101)
We can solve Eq. (11) graphically. The modes thus found
are collectively designated the LP
nm
mode. The lowest or-
der mode is the LP
01
mode (where LP stands for linearly
polarized). These LP modes can be shown to be a linear
combination of the EH and HE modes mentioned before.
The term weakly guided dielectric waveguide is a
misnomer. When Vis large, a mode can be tightly bound as
demonstrated by the phase velocity approaching that of a
core. Also, E
x
and E
y
are not independent of each other
since V
s
.
E
s
=0. The same statement applies to H
x
and H
y
.
5. MICROWAVE INTEGRATED CIRCUITS (MIC)
TRANSMISSION LINES
With the advent of integrated circuits at microwave and
millimiter-wave frequencies, microwave integrated circuit
(MIC) transmission lines have become increasingly im-
portant [15]. They are easily integrated with other micro-
wave integrated circuits, and have a low cost and better
reproducibility/reliability than do conventional lines.
Some examples are shown in Fig. 8. Not presented on
this gure but also representative of MICs lines are the
stripline and the inverted microstrip line [11]. Along with
some congurations of dielectric waveguides for millimiter
waves and integrated optics [29,32], such transmission
lines belong to a class usually called planar transmission
lines because the characteristics of these lines can be de-
termined from the dimensions on a single plane.
5.1. Quasi-TEM Approximation
The usual integrated circuits waveguiding structures can-
not support a TEM mode (an exception is the stripline,
whose design data can be determined exactly from an
electrostatic analysis for the TEM mode), for if they do, the
phase-matching condition will be violated at the interface
between the inhomogeneities (i.e, between the two dielec-
trics). This is because a TEM wave has the same phase
velocity of the medium in which the wave is traveling.
However, if the wavelength under consideration is
much larger than the transverse structure of the wave-
guide, we can show that the fundamental mode of such a
structure is almost TEM or quasi-TEM. A fundamental
mode is the mode that is propagating when o-0 or l-N.
The analysis under such assumptions is usually called
quasistatic analysis.
We can write Maxwells equations by separating out the
transverse and longitudinal components as
V
s
E
s
=ikZH
z
(102)
V
s
H
s
= ikZ
1
E
z
(103)
@
@z
^ zz E
s
^ zz V
s
E
s
=ikZH
s
(104)
@
@z
^ zz H
s
^ zz V
s
H
z
= ikZ
1
E
s
(105)
where Z =
m=e
_
: For a structure whose dominant trans-
verse dimension is much smaller than the wavelength, the
transverse variation of the eld would be more rapid than
the longitudinal variation. The x and y variations will vary
on the length scale of d. To emphasize this fact, we can
perform a coordinate stretching transformation by letting
x =dx
/
; y =dy
/
(106)
Under such a coordinate stretching transformation, we
obtain
V
s

1
d
V
/
s
(107)
Equations (104) and (105) become
ik
z
d^ zz E
s
^ zz V
/
s
E
z
=ikZdH
s
(108)
ik
z
d^ zz H
s
^ zz V
/
s
H
z
= ikZ
1
dE
s
(109)
r

r
r

r
Microstrip Line
Coplanar Waveguide Coplanar Strips
Slot Line
y
x
z
Figure 8. Cross-sectional view of different kinds of microwave
integrated circuit (MIC) waveguides.
where we have assumed e
ik
z
z
dependence of the eld.
When d/l-0, then kd-0. Since k
z
=
k
2
k
2
s
_
ok; we
also expect k
z
d-0, when d/l-0. Therefore, in the long-
wavelength limit, we obtain
(E
z
; ZH
z
) O(kd)(E
s
; ZH
s
) (110)
In other words, |E
z
|5|E
s
|, |H
z
|5|H
s
|, implying that
the eld is quasi-TEM. Therefore, we can write Eqs. (102)
and (103) as
V
s
E
s
=0 (111)
V
s
H
s
=0 (112)
Taking the divergence of Eqs. (104) and (105) and making
use of Eqs. (111) and (112), we have
V
s
.
eE
s
=0 (113)
V
s
.
mH
s
=0 (114)
Consequently, the transverse eld of a quasi-TEM mode is
essentially static. For this reason, the waveguide can be
analyzed as if a TEM mode were propagating on it. We can
solve the electrostatic problem to nd the line capacitance
C of the line. The line inductance L can be obtained by
solving the magnetostatic problem. Then, k
z
, the axial
wavenumber, can be found by
k
z
=o
LC
_
(115)
The preceding analysis indicates that when the wave-
length is long, the axial variation of the eld is slow com-
pared to the transverse variation. The transverse
variation of the eld has to be such that the eld can
match the boundary condition on the metallic conductors,
which is much smaller in dimension than the wavelength.
Hence, the transverse variation of the eld must balance
itself, resulting in Eqs. (111) to (113), which are the static
equations.
5.2. Capacitance and Effective Dielectric Constant of MIC
Lines
We shall discuss how the line capacitance of a MIC line,
such as a microstrip line, can be found. For other MIC
lines, the analysis follows in a similar manner [15]. There
is no closed-form solution for such a class of problem.
When e
r
=1 (i.e., the conductors are surrounded by homo-
geneous space and no dielectric interface is present), the
structure supports the TEM mode and one may solve such
problems by conformal mapping (as in the stripline case)
[35]. When h/w51, the problem can be solved by asymp-
totic matching [36]. However, to get an accurate value of C
for all w/h, a numerical analysis is preferable.
To nd the line capacitance of the MIC line, one can
rst solve for the charge distribution on the line. Then, the
capacitance can be easily found from the equation Q=CV,
where Q is the total charge per unit length on the line and
V is the voltage applied between the strip and the ground
plane. To nd the charge distribution, we can rst formu-
late an integral equation governing the charge distribu-
tion, and then solve it by numerical techniques [37].
Approximate closed-form formulas for C
eff
as a function
of line geometry and substrate material are also available
[38,39]. A potential f can be dened such that E
s
= V
s
f
and V
2
s
f=0; because E
s
is an electrostatic eld in the
long-wavelength limit.
To nd L, we make use of the fact that if e
r
=1, a pure
TEM mode propagates on the MIC line. In this case
k
z
=k
0
=o
L
0
C
0
_
(116)
Therefore, L
0
can be found once C
0
, the line capacitance
with e
r
=1, is known. C
0
can be also found by solving an
integral equation. Since L
0
remains unchanged when
e
r
a1, we have
k
z
=
C
C
0
k
0
(117)
An effective relative dielectric constant can be dened such
that
e
eff
=
C
C
0
(118)
which is a function of frequency. The physical meaning for
e
eff
is simple. It is the dielectric constant with which one
can ll the space homogeneously around a MIC line to
yield the same line capacitance as the inhomogeneously
lled line.
A characteristic impedance can also be dened as
Z
0
=
L
0
C
_
=
m
0
e
0
C
0
C
_
=
Z
a
0
e
r
_ (119)
where Z
a
0
is the characteristic impedance of the air-lled
line.
For the most commonly employed geometries and ma-
terials, closed-form empirical formulas for e
eff
and Z
0
are
available for design purposes and easy incorporation into
CAD algorithms [39].
5.3. Frequency-Dependent (Full-Wave) Theory of MIC Lines
The quasi-TEM model triumphs in predicting the value of
k
z
or the phase velocity in the long-wavelength limit. How-
ever, k
z
is in general frequency-dispersive. To obtain a k
z
that is valid at high frequencies as well (above a few GHz),
we need to solve the full-wave solution to the MIC line
problem. Approximate closed formulas (some essentially
empirical formulas, backed up by accurate characteriza-
tions) for the frequency-dispersive e
eff
(o) (and hence, k
z
and C) as a function of the static e
eff
, geometry, substrate,
and frequency are available [39]. These formulas are use-
ful for CAD applications, but have a number of drawbacks,
such as being applicable to only a specific class of MIC
lines (e.g., those based on alumina substrates) and some-
times requiring accurate static (TEM) data.
On the other hand, numerical methods [37] are the only
means of solving the full-wave problem for the general
microwave and millimiter-wave transmission-line prob-
lem. This is also true for integrated optics waveguides.
With the advances in computer technology, these methods
have become increasingly important. The most popular
numerical methods include the nite-difference and nite-
difference time-domain (FD and FDTD) methods, the
transmission-line method (TLM), the nite-element meth-
od (FEM), the boundary-element method (BEM), the
method of lines (MoL), the method of moments (MoM),
the beam propagation method (BPM), the effective index
method, and the spectral domain approach. There is no
such a thing as the best universal numerical method avail-
able. Each of these methods has its advantages and lim-
itations and is best suited for some given applications.
Each has its own variations and improvements. For a dis-
cussion of each and example applications, the reader is
referred to the texts by Sorrentino [37] and Huang and
Kong [40].
BIBLIOGRAPHY
1. H. Hertz, Electric Waves, Being Researches on the Propaga-
tion of Electric Action with Finite Velocity through Space
(transl. D. E. Jones), Macmillan, 1893 and Dover, 1962, New
York.
2. J. C. Maxwell, Electricity and Magnetism, 1873, p. 385; also in
Vol. 2, Section 690, p. 322, Academic Press, New York, 1935.
3. Lord Rayleigh, On the passage of electric waves through
tubes, Phil. Mag. 43:125132 (1897); reprinted in Collected
Papers, Paper 226, Vol. 4, Cambridge Univ. Press, Cambridge;
UK, 1903.
4. A. Sommerfeld, Fortpanzung elektrodynamischer Wellen an
einen zylindrishen Leiter, Ann. Phys. 67(Ser. 3):1223 (1899).
5. D. Hondros and P. Debye, Electromagnetische Wellen an die-
lectrischen Dra hten, Ann. Phys. 32(Ser. 4): 465 (1910).
6. K. S. Packard, The origin of waveguides: A case of multiple
rediscovery, IEEE Trans. Microwave Theory Tech. 32(9):961
969 (1984).
7. R. M. Barrett, Microwave printed circuitsthe early years,
IEEE Trans. Microwave Theory Tech. 32(9):983990 (1984).
8. H. Howe, Jr., Microwave integrated circuitsan historical
perspective, IEEE Trans. Microwave Theory Tech. 32(9):991
996 (1984).
9. T. Ikegami, Survey of telecommunications applications of
quantum electronicsprogress with optical ber communi-
cations, Proc. IEEE 80(3):411419 (1992).
10. R. E. Collin, Field Theory of Guided Waves, IEEE Press, Pis-
cataway, NJ, 1991.
11. I. J. Bahl, Transmission lines, in K. Chang, ed., Handbook of
Microwave and Optical Components, Wiley, NewYork, 1989,
pp. 159.
12. K. C. Gupta, Transmission-line discontinuities, in K. Chang.
ed., Handbook of Microwave and Optical Components, Wiley,
New York, 1989, pp. 60117.
13. F. A. Benson and T. M. Benson, Fields, Waves, and Transmis-
sion Lines, Chapman & Hall, New York, 1991.
14. B. C. Wadell, Tansmissmion Line Handbook, Artech House,
Norwood, MA, 1991.
15. T. Itoh, ed., Planar Transmission Line Structures, IEEE
Press, New York, 1987.
16. W. C. Chew, Theory of Microwaves and Optical Waveguides,
ECE431 Class Notes, Dept. Electrical and Computer Engi-
neering, Univ. Illinois at UrbanaChampaign, 1997.
17. G. Gonzalez, Microwave Transistor Ampliers, 2nd ed., Pren-
tice-Hall, Upper Saddle River, NJ, 1992.
18. P. Magnusson, G. C. Alexander, and V. K. Tripathi, Trans-
mission Lines and Wave Propagation, 3rd ed., CRC Press,
Boca Raton, FL, 1992.
19. A. Deutsch, G. V. Kopscay, P. J. Restle, H. H. Smith, G.
Katopis, W. D. Becker, P. W. Coteus, C. W. Surovic, B. J. Rub-
in, R. P. Dunner, Jr., T. Gallo, K. A. Jenkins, L. M. Terman, R.
H. Dennard, G. A. Sai-Halasz, B. L. Krauter, and D. R. Kne-
bel, When are transmission line effects important for on-chip
interconnections? IEEE Trans. Microwave Theory Tech.
45(10):18361846 (1997).
20. M. Celik, A. C. Cangellaris, and A. Yagmour, An all-purpose
transmission-line model for interconnect simulation in
SPICE, IEEE Trans. Microwave Theory Tech. 45(10):1857
1867 (1997).
21. N. Fache, F. Olyslager, and D. De Zutter, Electromagnetic and
Circuit Modelling of Multiconductor Transmission Lines, Ox-
ford Univ. Press, New York, 1993.
22. J. A. Brandao Faria, Multiconductor Transmission-line
Structures: Modal Analysis Techniques, Wiley, New York,
1993.
23. R. F. Harrington, Time-Harmonic Electromagnetic Fields,
McGraw-Hill, New York, 1961.
24. J. Kretzschmar, Wave propagation in hollow conducting
elliptical waveguides, IEEE Trans. Microwave Theory Tech.
18(9):547554 (1970).
25. S. B. Cohn, Properties of ridge wave guide, Proc. IRE 35:783
788 (1947).
26. J. Montgomery, On the complete eigenvalue solution of
ridged waveguide, IEEE Trans. Microwave Theory Tech.
19(6):457555 (1971).
27. P. B. Johns, The solution of inhomogeneous waveguide prob-
lems using a transmission-line matrix, IEEE Trans. Micro-
wave Theory Tech. 22(3):209215 (1974).
28. N. Dagli, Equivalent circuit representation of open guided-
wave structures, IEEE J. Quantum Electron. 26(1):98108
(1990).
29. N. Dagli and C. G. Fonstad, Universal design curves for rib
waveguides, J. Lightwave Technol. 6(6):11361145 (1988).
30. K. Ogusu, Numerical analysis of the rectangular dielectric
waveguide and its modications, IEEE Trans. Microwave
Theory Tech. 25(11):874885 (1977).
31. W. C. Chew and M. Nasir, A variational analysis of anisotro-
pic, inhomogeneous dielectric waveguides, IEEE Trans.
Microwave Theory Tech. 37(4):661668 (1989).
32. W. C. Chew, Analysis of optical and millimiter wave
dielectric waveguides, J. Electromag. Waves Appl. 3(4):
359377 (1989).
33. W. C. Chew, Waves and Fields in Inhomogeneous Media,
IEEE Press, Piscataway, NJ, 1995.
34. A. W. Snyder, Asymptotic expression for eigenfunctions and
eigenvalues of a dielectric or optical waveguide, IEEE Trans.
Microwave Theory Tech. 17(12):11301138 (1969).
35. P. Grivet, The Physics of Transmission Lines at High and Very
High Frequencies, Academic Press, London, 1970.
36. W. C. Chew and J. A. Kong, Asymptotic eigenequations and
analytic formulas for the dispersion characteristics of open
wide microstrip lines, IEEE Trans. Microwave Theory Tech.
29(9):933941 (1981).
37. R. Sorrentino, ed., Numerical Methods for Passive Microwave
and Millimiter Wave Structures, IEEE Press, NewYork, 1989.
38. S. Y. Poh, W. C. Chew, and J. A. Kong, Approximate formulas
for line capacitance and characteristic impedance of micro-
strip line, IEEE Trans. Microwave Theory Tech. 29(2):135142
(1981); correction in IEEE Trans. Microwave Theory Tech.
29(10):1119 (1981).
39. T. Edwards, Foundations for Microstrip Circuit Design, 2nd
ed., Wiley, New York, 1992.
40. W. P. Huang and J. A. Kong, eds., Progress in Electromagnet-
ics Research (PIER), Vol. 10, Methods for Modeling and Sim-
ulations of Guided-Wave Optoelectronic Devices, Elesvier,
New York, 1995.
HIGH-TEMPERATURE SUPERCONDUCTORS
RAAFAT R. MANSOUR
University of Waterloo
Waterloo, Ontario, Canada
1. INTRODUCTION
Superconductors, as opposed to normal or conventional
conductors, have the ability to conduct electrical current
with very small resistance, very small power loss, very
small generation of heat, and greatly reduced noise levels.
The resistance is very small but nite at microwave fre-
quencies and is truly zero at DC. Superconductivity was
rst understood as a very-low-temperature phenomenon
and was explained by John Bardeen, Leon Cooper, and
Robert Schrieffer in what has become known as the BCS
theory. According to this theory, at a particular low tem-
perature, called the transition temperature or critical tem-
perature (T
c
), the electrons in certain materials pair up
and form a single quantum state, acting like a frictionless
uid, and become superconducting.
While conventional low-temperature metallic super-
conductors have been known since the beginning of the
twentieth century, the extremely high cost of refrigeration
has limited their use in many applications. Until 1986 re-
searchers could not identify any materials that could be-
come superconducting with a critical temperature higher
than 23 K. The materials discovered up to that time, which
had critical temperatures lower than 23 K, were common-
ly known as low-temperature superconductor (LTS) mate-
rials. A new era in superconductivity began in 1986 with
the discovery of a new class of ceramic materials that are
superconductive at temperatures as high as 125K. Re-
ferred to as high-temperature superconductor (HTS) ma-
terials, these materials enerally have significant
advantages over LTS materials because they can be cooled
at economically and commercially feasible temperatures.
Superconductivity is certainly a multidisciplinary eld.
It spans almost the entire realm of electrical engineering
including microwave, power, electronic, and power and
computer engineering [1]. In microwave applications, the
HTS technology offers major breakthroughs in perfor-
mance of components and subsystems. Light weight, small
volume, and high performance, which are the properties of
the superconducting technology, are also the main drivers
in the design and construction of microwave systems. The
feasibility of using this technology to design microwave
components such as lters, multiplexers, receivers, delay
lines, couplers, antennas, and phase shifters with superior
performance has been already demonstrated. A review of
progress since 1990 is given in three IEEE-MTT Special
Issues on HTS microwave applications published in 1991
[2], 1996 [3], and 2000 [4]. Three books have also been
published on microwave superconductive devices [57].
Although the quest for room-temperature superconduc-
tor materials goes on, the quality of todays materials is
adequate for the development of advanced superconduc-
tive systems with superior RF performance for wireless
and space applications. In this article we provide rst a
brief historical overview summarizing the key develop-
ment milestones of the technology. We discuss the basic
material characteristics, highlighting the unique features
that are relevant to microwave applications. We then cov-
er the progress to date in various microwave applications
of high-temperature superconductivity.
2. HISTORY AND PHENOMENA OF
SUPERCONDUCTIVITY
The history of superconductivity can be divided into two
eras. The rst era started with H. Kamerlingh Onnes, a
Dutch physicist, who successfully liqueed helium in the
early twentieth century at around 4.2K. While investi-
gating the effect of such low temperatures on the proper-
ties of various metals in 1911, Onnes observed that the
ohmic resistance for mercury vanishes at about 4 K. He
made the same discovery with lead and tin shortly after.
The progress in understanding superconductivity and in
developing superconductors with transition temperatures
higher than 4 K was agonizingly slow. By switching from
single elements to alloys, researchers were eventually able
in 1941 to raise the transition temperature to 15 K, for
niobium nitride. It took, however, more than 30 years to
further raise the transition temperature to 23 K with the
discovery of the niobium germanium alloy in 1973. A
number of systems already use low-temperature super-
conducting devices. A comprehensive review of the appli-
cations of low temperature superconductivity is given in a
special issue published in January 1973 in the Proceed-
ings of the IEEE. The LTS materials are still being used
today in various applications. It was not until 1957 that a
theory nally emerged to give an acceptable microscopic
picture explaining why metals become superconductors at
low temperatures. Three physicists at the University of
Illinois, John Bardeen, Leon Cooper, and Robert Schrief-
fer, introduced what is now known as the BCS theory,
named after the initial letters of their surnames. The the-
ory was criticized from the start for failing to provide pre-
dictions that match experiments. Numerous revisions and
even quite different models were proposed.
HIGH-TEMPERATURE SUPERCONDUCTORS 2005
The second era commenced in 1986 with the
discovery of a new class of superconducting materials
with a then recordbreaking temperature of 35 K by Alex
Muller and George Bednorz of IBM Zurich. The com-
pound was a ceramic with a complex structure: an oxide
of lanthanum, barium, and copper. Muller and Bednorz
were awarded the 1987 Nobel Prize in Physics for their
discovery.
Early in 1987, Paul Chu and his colleagues at the Uni-
versity of Houston reported the real breakthrough in high-
temperature superconductivity (HTS), a yttriumbarium
copperoxide compound (YBa
2
Cu
3
O
7
) with a transition
temperature of 94 K. The material is commonly known
as YBCO. These new materials operate well at 77 K, al-
lowing the use of liquid nitrogen rather than liquid heli-
um. Liquid nitrogen is much cheaper and denser and has
a heat vaporization that is 60 times larger than that of
liquid helium. The ability to use liquid nitrogen rather
than liquid helium reduces the cost of cooling by a factor
of 1200.
In December 1987, Hiroshi Maedas group at Tsukuba
laboratories in Japan reported the bismuthstrontium
calciumcopperoxides (BiSrCaCuO), which have a T
c
of
110 K. The next breakthrough was announced in February
1988 at the World Congress on Superconductivity in Hous-
ton. Zhengzhi Sheng and Allen Hermann from the Uni-
versity of Arkansas reported success with a thallium
compound (TiBaCaCuO) having a transition temperature
of 120 K.
In 1993, Paul Chus group at the Texas Center for Su-
perconductivity at the University of Houston announced
the discovery of a mercury compound HgBaCaCuO having
a transition temperature of 161 K. For the rst time, su-
perconductors could conceivably be cooled by the house-
hold coolant Freon, which exists in liquid form at 147 K.
The stability of this compound is still however, in ques-
tion since it was developed in a procedure that involves a
static pressure of 150,000 atm. It is believed that chemical
substitution may be used to achieve the same result the
high pressure is now achieving. Figure 1 summarizes the
history of the superconductor materials.
Shortly after the discovery of the YBCO materials at
temperature near 90 K, scientists and engineers became
very interested in the prospects of employing the HTS
technology in the design of microwave components and
subsystems. The U.S. Naval Research Laboratory (NRL)
established a program known as the high-temperature
superconductivity space experiment (HTSSE) in 1989.
This program was a major catalyst to the development
of microwave superconductivity. It consisted of two
phases: the focus of the rst-phase HTSSE-I (19891992)
was to develop microwave HTS components such as res-
onators and lters [8], while the focus of the second
phase HTSSE-II (19921996) was to develop advanced
HTS subsystems [9]. More than 30 organizations from
the United States, Canada, and Europe provided devices
and subsystems to the HTSSE program [811]. The
results obtained from this comprehensive research pro-
gram did conclusively demonstrate that viable and robust
HTS subsystems could be developed, fabricated, and
cryogenically packaged for both ground and space appli-
cations [11].
In addition to the HTSSE program, several other
focused programs were led by government agencies in
the United States, Japan, Canada, and Europe on advanc-
ing HTS microwave technology. In particular, the U.S.
Defense Advanced Research Projects Agency (DARPA),
which sponsored several research programs in this
eld [12].
Another key factor that helped advance the develop-
ment of HTS microwave devices was the rapid progress
of the wireless industry in the 1990s. Several startup
companies emerged in the United States, Europe, and
Japan since the mid-1990s [1316] with a focus on the
commercialization of HTS microwave devices for wireless
applications.
[under pressure] Hg Ba Ca Cu O
Ti Ba Ca Cu O
Bi Sr Ca Cu O
Y Ba. Cu O
La Sr Cu O
Nb Ge
Nb N
Hg
LIQUID NITROGEN
MEISSNER
EFFECT
BCS
THEORY
JOSEPHSON
EFFECT
LIQUID HELIUM
1900
DISCOVERY
YEAR
160
140
120
100
80
60
40
20
T
R
A
N
S
I
T
I
O
N

T
E
M
P
E
R
A
T
U
R
E

(
K
)
1920 1940 1960 1980 2000
Figure 1. History of superconductor materials.
2006 HIGH-TEMPERATURE SUPERCONDUCTORS
3. CHARACTERISTICS AND THEORIES OF
SUPERCONDUCTORS
Superconductors have unique characteristics that differ
from those of normal conductors. A much lower loss can be
obtained with superconductors corresponding to lower RF
losses. The surface resistance of superconductor materials
at microwave frequencies is typically two orders of mag-
nitude smaller than that of normal conductors. While the
surface resistance of a normal conductor varies with fre-
quency as (f)
1/2
, that of superconductors varies with fre-
quency as (f)
2
[7]. This frequency dependence is seen
experimentally and can be easily deduced from theory.
Superconductors have a surface resistance that is de-
pendent on the applied eld. The dependence is more pro-
nounced at high levels of elds, leading to nonlinearities
and generation of intermodulation products. The nonlin-
earity exists even at low power levels but is more pro-
nounced at high power levels. Such nonlinear
characteristics have a negative impact on the power-han-
dling capability of microwave superconductive devices
[1721]. The same nonlinear characteristics, however,
were exploited to develop novel HTS devices such as mix-
ers [7] and nonlinear transmission lines (NLTLs) [22]. An-
other characteristic of superconductor materials is that
they have frequency-independent penetration depth that
determines eld penetration into the materials rather
than the skin depth of normal conductors.
There are two other unique characteristics of super-
conductor material: the Meissner effect and ux quanti-
zation. The Meissner effect relates to the superconductors
ability to expel magnetic eld; specifically, superconduc-
tors are perfect diamagnets. This characteristic makes it
possible to suspend a magnet in midair over a piece of su-
perconductor. The Meissner effect also implies that the
transition to the superconducting state is sensitive to the
magnetic eld, and for increasing magnetic elds the ux
rst penetrates into and then ultimately destroys the
superconducting state. Correspondingly, a superconduct-
ing transmission line can be driven into the normal
state by a sufciently large current that exceeds the
critical current J
c
.
The ux quantization relates to the phenomenon that
magnetic eld generated from a current circulating in a
superconducting loop is quantized. The unit of quantiza-
tion is known as the uxon, or ux quantum and is given
by 2.07 10
15
Wb. The characteristic of ux quantiza-
tion makes it possible to build extremely sensitive mag-
netometer devices capable of measuring extremely low
levels of magnetic elds. Such magnetometers are known
as superconductor quantum interference devices (SQUIDs)
[23]. Table 1 summarizes the differences between super-
conductors and normal conductors.
For microwave applications, the two most important
parameters of HTS materials are the surface resistance R
s
and the critical current density J
c
, which represents the
maximum current that can be carried by the supercon-
ductor before switching to the normal state. There are
several techniques available to measure these two param-
eters [57]. Typical values for R
s
and J
c
for HTS thin lms
are 100200mO and 10
6
A/cm
2
, respectively.
The HTS thin lms are deposited on a low loss dielec-
tric substrate and are processed using standard photo-
lithographic techniques. The most widely used substrate
is lanthanum aluminate (LaAlO
3
) with a dielectric con-
stant of e
r
=24 and a loss tangent of tan d =3 10
5
. One
challenge with the LaAlO
3
substrate is that its crystal
structure exhibits twining, which makes the substrate ef-
fectively inhomogeneous. This can affect the ability to set
the frequency correctly in narrowband lter applications.
Another commonly used substrate for HTS lms is mag-
nesium oxide (MgO) with e
r
=9.5. MgO does not suffer
from twining problems, but on the other hand it does not
offer the same miniaturization, which the LaAlO
3
sub-
strate offers, because of its high dielectric constant. Sap-
phire substrates have also been used with various buffer
layers to provide the appropriate lattice match to HTS
lms. However, the surface resistance R
s
and current den-
sity J
c
are inferior to lms on LaAlO
3
or MgO substrates.
Dielectric materials typically exhibit a considerable im-
provement in loss tangent tan d on cooling to cryogenic
temperatures. For example, high-purity sapphire (e
r
=9.4)
at X-band frequencies exhibits tan d of 10
5
at 300K and
tan d of 10
7
at 77 K [24]. MgO and LaAlO
3
exhibit a
similar loss tangent improvement at cryogenic tempera-
tures [25] as compared to room-temperature operation.
This is considered a fringe benet of operating at cryo-
genic temperatures.
The HTS lms are deposited on 2- and 3-in. wafers,
which can vary in substrate thickness from 0.010 to 0.1 in.
(standard substrate thicknesses are 0.01 and 0.02 in.). The
HTS lms are typically deposited either by laser ablation
or by sputtering techniques [26]. Currently, 2- and 3-in.-
diameter wafers with HTS lms on one or both surfaces
are commercially available from various sources at prices
that are almost one-tenth the prices of the early 1990s.
The BCS theory deals with superconductors from a mi-
croscopic point of view. Two other well-known theories
Table 1. Differences between Superconductors and Normal Conventional Conductors
Characteristic Normal Conductor Superconductor
Surface resistance at 77K and f =5 GHz 5mO (Cu) 0.1mO (TBCCO)
Frequency dependence of surface resistance (f)
1/2
(f)
2
Field dependence of surface resistance
a
Constant R
s
pH
2
Field penetration Skin depth (d) Penetration depth (l)
Meissner effect Not applicable Applicable
Magnetic ux quantization Not applicable Applicable
a
The dependence on magnetic eld H is very weak at low levels of H.
were developed to deal with the macroscopic properties of
superconductors: the London theory [27] and Ginzburg
Landau (GL) theory [28]. These two theories have been
used in conjunction with Maxwells equation to model the
electromagnetic characteristics of superconductive micro-
wave devices [2931]. The London theory is based on the
two-uid model [27], which assumes that the current in
superconductor is carried by two parallel pathways, one of
the normal electrons and one of the superconducting elec-
trons whose densities are a function of temperature. The
London theory has been successful in explaining several of
the characteristics of superconductor materials. It does
not however, take into consideration the eld dependence
of the constituent parameters of the superconductor ma-
terials. The GL theory, on the other hand, is a more com-
prehensive macroscopic theory that accounts for eld
dependence allowing the characterization of the nonlin-
ear behavior of the superconductive materials.
4. HTS MICROWAVE FILTERS
Microwave lter networks represent a critical and sub-
stantive portion of any communication system. Such a
system, whether wireless or satellite, requires lters to
separate the signals received into channels for amplica-
tion and processing. The phenomenal growth in the tele-
communications industry has brought significant
advances in lter technology as new communication sys-
tems emerged, demanding equipment miniaturization
while requiring more stringent lter characteristics. The
emergence of the HTS technology has offered the possibil-
ity, for the rst time, to build lters that can compete with
traditional waveguide and dielectric resonator lters not
only in size but also in Q (quality factor).
Figure 2 illustrates a pictorial comparison between mi-
crowave lters realized using rectangular single-mode
waveguide technology, circular dual-mode waveguide
technology, dielectric resonator technology, and super-
conductor technology [32]. Two types of superconductive
lters are included in this gure: one using the hybrid
DR/HTS technology and the other based on HTS thin-lm
technology. All lters have the same order and are de-
signed to operate at the same center frequency of 4GHz. It
is clear that a significant reduction in size and mass can be
achieved with the use of HTS technology.
In addition to size and mass reduction, HTS lters also
offer improvement in the in-band insertion loss. The un-
loaded Q of a microwave resonator in general can be writ-
ten as Q
u
=(R
s
/GF tan d)
1
, where G is a factor
determined by the resonators geometry, which typically
increases as the resonator dimensions increase, while F is
a factor determined by the fraction of the electrical energy
of the cavity stored in the dielectric materials; R
s
is the
surface resistance and tan d is the loss tangent of the di-
electric material. It is clear that reducing either R
s
or tan d
or both of them can increase the unloaded Q
u
. Three main
types of HTS lter are reported in the literature:
1. HTS thin-lm planar lters
2. Hybrid dielectric/HTS lters
3. HTS thick-lm-coated lters
In both HTS thin-lm planar lters and thick-lm-
coated lters, the Q improvement is attributed mainly to
the reduction of R
s
by replacing normal metals with HTS
materials. On the other hand, the improvement in Q in
hybrid dielectric/HTS lters is attributed to the reduction
of R
s
as well as to the reduction of the loss tangent of the
substrate due to cooling.
4.1. HTS Thin-Film Planar Filters
Planar lters such as stripline, microstrip, and coplanar
line lters have been known since the mid-1970s [33].
These lters, however, have very limited applications be-
cause of their low Q values. Effectively any known planar
lter conguration can be realized in HTS technology by
replacing metal lms with HTS lms. This in turn would
increase the lter Q value by several orders of magnitude.
For example, a half-wavelength microstrip resonator
made of gold lms on lanthanum aluminate substrate
would typically have an unloaded Q value of 400. Replac-
ing the gold lms with HTS lms while using the same
substrate would provide an HTS resonator with an un-
loaded Q value of B30,000.
The emergence of the HTS technology has created the
opportunity for more innovation in planar lter congu-
rations. Several novel lter congurations have been pro-
posed allowing the realization of quite advanced lter
functions [3438]. Figure 3 illustrates a packaged HTS
10-pole elliptic function self-equalized planar lter. The
basic building block of this lter is a dual-mode lumped-
element resonator, which makes it easy to create elliptic
and self-equalization functions [39].
The emergence of several commercial computer soft-
ware packages for simulation and design of planar circuits
in the early 1990s played a key role in advancing the de-
velopment of HTS planar lters. Even though these com-
mercial packages do not factor in the physical
characteristics of the superconductor materials, they Figure 2. A comparison between various microwave lters.
have been successfully used in the design of planar HTS
lters, particularly for low-power applications.
4.2. Hybrid Dielectric/HTS Filters
For the conventional dielectric resonator shown in Fig. 4a,
the tangential electric eld of the HEE
11
mode vanishes at
the plane z =0 [40]. As a consequence, introducing a con-
ducting wall at this plane will only slightly perturb the
eld distribution of this mode. Thus the size of the dielec-
tric resonator lter operating in the HEE
11
mode may be
reduced by supporting the resonator directly on a con-
ducting surface (image plate) as shown in Fig. 4c. With the
use of dielectric support that lls the whole cavity, further
size reduction can be achieved as shown in Fig. 4d.
A normal-conducting image plate, however, signicant-
ly degrades the Q of the resonator, which in turn makes
the approach unfeasible for high-Q applications. On the
other hand, if the image plate is replaced by a surface
made of high-temperature superconductor materials, the
original unloaded resonator Q will be only slightly affect-
ed, allowing the use of this resonator structure to realize
compact size lters with a superior loss performance for
cryogenic applications [41,42]. The reduction in size is at-
tributed to the use of the image plate, while the improve-
ment in loss performance is attributed to the use of the
HTS image plate as well as to the increase in the unloaded
Q of the dielectric resonator as the temperature is lowered
from 300 to 77 K.
Several hybrid DR/HTS lters, which utilize the con-
cept illustrated in Fig. 4, were built and tested [21,38,
41,43]. Figure 5 shows a detailed conguration of an eight-
pole hybrid DR/HTS lter [29]. The lter consists of four
dielectric resonators operating in image-type dual modes.
The resonators are supported inside the lter housing
using two blocks made of low-loss low-dielectric constant
ceramic materials. The HTS material is in the form of
Figure 3. A 10-pole HTS thin-lm lter.
H
Dielectric support
Z=0
HEE
11
E
(a) (b)
(c) (d)
2H
H
r

r
Figure 4. Transition from conventional DR
cavity to an image DR cavity: (a) conventional
DR resonator; (b) eld distribution; (c) image
DRresonator; (d) image DR resonator with a DR
support.

92
26
44
40
38a
36
96
18
104
36
46
40
44
26
46
40
44
38c
102
30
98
38b
36
106
24
108
100
34
36
28
20
38d
40
42
44
32
94
Figure 5. A layout of the eight-pole hybrid DR/HTS lter. The
38s are dielectric resonators; the 40s are HTS wafers diced from a
standard 2-in. wafer; the 36s are dielectric supports; the 44s are
spring washers; the 26s, 102, and 104, are elements of the lter
housing; and the 28s are input/output probes.
small wafers (0.5 0.5 in. at C band), which are diced
from a standard 2-in. HTS thin lm deposited on 0.02-in.-
thick wafer of LaAlO
3
. The HTS lms are kept in contact
with the dielectric resonators using springs and metallic
plates bolted to the lter housing. A detailed description of
this type of lter is given in [44].
The drawback of hybrid DR/HTS lters is the mechan-
ical design complexity. These lters do, however, have
several advantages over their counterparts HTS thin-lm
planar lters:
1. No etching is required for the HTS material. Etching
may reduce the power-handling capability of the
HTS material and degrade its surface resistance.
The design also eliminates the need to deposit gold
contacts, which are required for thin-lm lters.
Also there is no need to generate circuit and contact
masks.
2. In patterned thin-lm lters, a good portion of the
HTS material is wasted as a result of the etching
process, while in hybrid DR/HTS lters the wafer is
diced into small pieces called shorting [short-cir-
cuiting] plates. The wafer is used efciently.
3. The lter has a good spurious performance. The im-
age plate helps not only in reducing the lter size
but also in eliminating spurious modes, making it
possible to design lters of this type with a spurious-
free window of more than 2 GHz width centered at a
frequency of 4GHz.
4.3. HTS Thick-Film-Coated Filters
Melt-processed YBCO thick lms were used to coat three-
dimensional structures for microwave applications
[45,46]. YBCO thick lms typically have a current densi-
ty of 2000 A/cm
2
and a relatively low value of surface re-
sistance of approximately 0.5 mO (5GHz at 77 K). The
feasibility of constructing a thick-lm YBCO 5.66-GHz
cavity resonator operating in TE
011
modes with unloaded
Q values of 715,000 at 77 K has been demonstrated [47].
Measured results of cavities operating at 10 and 7.5 GHz
have also been reported [48]. However, the TE
011
cavities
are too bulky to be used at frequencies below B12 GHz.
At lower frequencies, two resonator structures have been
proposed for thick-lm coating: coaxial resonators and
split-ring resonators [49]. Thick-lm coaxial resonators
have found limited use because of the need to coat all sur-
faces in the cavity to achieve reasonably high Qvalues. On
the other hand, the split-ring resonator has been adopted
by Illinois Superconductor Corporation to manufacture
thick-lm HTS lters for wireless applications [51].
Figure 6 illustrates a 3-pole HTS lter employing split-
ring resonators coated with HTS thick lms. Coupling be-
tween resonators is achieved through the use of irises to
provide the right coupling between resonators. Details on
this type of lter can be found in Ref. 50.
5. CRYOGENIC RECEIVERS
A layout of a typical receiver is shown in Fig. 7. In addition
to signal amplication, the receiver typically provides fre-
quency conversion to the transmit band. It consists of an
input lter, a low-noise amplier (LNA), an oscillator, and
a downconverter. The input lter limits the noise band-
width and provides a high rejection of out-of-band signals.
The insertion loss of this lter is a very critical parameter
since it adds directly to the overall noise gure of the re-
ceiver. In most applications, these lters are typically built
using high-Q lter structures to minimize the insertion
loss. For example, a C-band input lter in satellite receiv-
ers is typically congured as a cascade of a waveguide
lowpass lter and a waveguide bandpass lter. An HTS
lter with a similar performance would have
1
16
th the mass
and
1
100
th the volume of the conventional C-band wave-
guide input lter. It will also provide a 50% reduction in
insertion loss.
The LNA benets from cryogenic operation in two ways
[51,52]. The use of HTS-matching structures results in
decreased loss at the input of the LNA, which directly im-
proves the attainable noise gure. Additionally, a fringe
benet of operation at 77 Kis the reduction in the inherent
noise of the active component. The improvement is more
pronounced at high frequencies. Table 2 summarizes the
typical measured noise gure results of C-band and Ka-
band LNAs. The oscillator could also benet from using
HTS technology by employing high-Q HTS resonator as
the frequency-determining element. The benet in this
case would be phase noise reduction. However, cooling the
311
41
51
20
41
41
20
52
51
20
20
52
41
311
Figure 6. A three-pole lter employing split-ring resonators
coated with HTS thick lms.
RF
LO
HTS Resonator
Oscillator
Mixer
IF
LNA LPF
HTS
BPF
Figure 7. A block diagram for an RF receiver.
remaining oscillator components will have a negligible
impact on overall oscillator performance while adding a
large heat load on the system. Therefore the use of cryo-
genic technology has been limited in most applications to
the input lters and the rst LNA stage.
6. HTS DELAY LINES
Delay lines are useful devices in realizing transversal l-
ters for analog signal processing applications. Current
technologies for delay lines include surface acoustic
wave (SAW) devices and electromagnetic coaxial cables
in the form of coils. However, the small acoustic wave-
length and the minimum achievable linewidth limit the
SAW delay devices to low-frequency applications. Electro-
magnetic coaxial transmission lines with much larger
wavelengths can yield much higher frequency limits. On
the other hand, coaxial delay lines are bulky and have
high insertion losses. Often the use of ampliers and
equalizers is necessary to restore the original signal.
Superconducting delay lines [5355] offer the highest
bandwidth with the lowest loss in highly compact designs.
Several HTS planar transmission lines such as striplines
and microstrip and coplanar lines could be used to realize
delay lines. The high dielectric constant of lanthanum
aluminate substrates (e
r
=24) allows a significant amount
of delay to be placed on a single substrate, while the low
conductor loss of the HTS material keeps insertion loss
to a minimum. Figure 8 illustrates a delay line realized
using an HTS coplanar delay line [56]. A 100-ns HTS de-
lay line of this type measures 90 96 12 mm. A 100-ns
delay line realized using RG-141 cables would have a
length of 70 ft. Even with a cryocooler, the size and
weight reductions offered by the HTS technology are
very attractive.
7. ANTENNA AND BEAMFORMING NETWORKS
High-temperature superconductors offer new opportuni-
ties for the designers of antenna systems. Examples in-
clude antenna-matching elements, feed networks,
microstrip antennas, and superdirective arrays [5759].
In particular, beamforming networks can benet from
HTS technology in areas of both insertion loss reduction
and miniaturization. The current designs for beamform-
ing networks use either multilayer stripline technology or
coaxial technology. The stripline designs are compact in
size but exhibit very high insertion loss. The coaxial de-
signs, on the other hand, exhibit low loss but are known to
be very heavy and bulky.
Figure 9 illustrates a layout of an HTS 8 8 Butler
matrix designed to operate at L band [59]. The matrix
consists of folded thin-lm 901 hybrids and delay lines.
The size of the overall matrix is 75 50 12 mm. The
HTS Butler matrix is two orders of magnitude smaller
than its coaxial counterpart.
8. WIRELESS APPLICATIONS
Mobile communication systems require improved sensi-
tivity and selectivity to support the constant growth in
services, increased coverage, and larger numbers of
subscribers. Several companies have taken advantage of
Table 2. Measured Noise Figure Results ofKa-Band and
C-Band LNAs
Measured Noise Figure
Component At 300K (dB) At 77 K (dB)
Ka-band LNA 3.1 0.8
C-band LNA 1.0 0.25
Figure 8. A layout of an HTS delay line. Figure 9. An 8 8 HTS Butler matrix.
technical innovations in cryoelectronics to address these
improvements. Cryoelectronic solutions have matured to
the point where numerous eld trials under the auspices
of both RF equipment manufacturers and PCS carriers
have been proved to be successful [6063].
The performance of wireless base stations can be con-
siderably enhanced by incorporating cryogenic LNAs with
high-temperature superconductor lters. The HTS tech-
nology presents a viable solution of realizing small-size
high-order lters with low insertion loss. The overall size
of multiple HTS transceiver chains, including the associ-
ated cryocooler, is considerably less than that of an equiv-
alent conventional transceiver.
The capacity and coverage of a base station receiver are
determined primarily by the receiver selectivity and sen-
sitivity on the up-link and in part by the base station
transmitter power of the downlink. The selectivity can be
significantly increased with the use of high-order lters as
shown in Fig. 10. However, high-order lters, built using
conventional technology, would exhibit a very high pass-
band insertion loss, resulting in a reduced signal-to-noise
ratio and hence degrading the sensitivity performance of
the receiver.
A significant improvement in receiver sensitivity can
be achieved by designing the receiver to operate in a cryo-
genic environment. This virtually eliminates thermal
noise from the LNA and potentially improves the lter
loss performance. Although conventional dielectric reso-
nator lters can be cooled with some improvement in per-
formance, integrated receivers of this type are too large for
towertop mounting.
The low resistance of the HTS materials makes it pos-
sible to use the planar thin-lm technology to provide HTS
lters that are two orders of magnitude smaller in size
than conventional dielectric resonator lters. This signif-
icant reduction in physical size makes valuable space
available for other required electronic components, en-
abling service providers to enhance the utilization of ex-
isting base stations instead of developing additional base
stations. In addition, miniaturization can decrease deploy-
ment costs for new base stations, as less real estate is re-
quired to support base station. Figure 11 illustrates a
comparison between a conventional transceiver and an
HTS transceiver [63].
9. SPACE APPLICATIONS
The mass, volume, and power consumption of payload
electronic equipment are significant contributors to the
overall cost of space systems. HTS technology offers the
potential of large reductions in mass and volume of elec-
tronic equipment, leading to significant cost reduction of
satellite systems [39,64,65]. Todays space segment repre-
sents a market of many billions of dollars for commercial
and military applications and is growing. HTS technology
has the potential to accelerate the development and im-
plementation of new advanced satellite systems. It also
represents the potential of performance enhancements of
strategic defence communications systems.
Mass reductions have a dramatic impact on the
economics of a satellite program because launch costs
are related to satellite weight. Similarly, mass reductions
can be exploited to increase the capacity (by adding pay-
load electronics) or extending operational life (by increas-
ing station maintenance fuel). As a result, market factors
have constantly been pushing hardware suppliers to re-
duce mass and size of their products.
The current technology for satellite input multiplexers
is the dielectric resonator technology. A 60-channel HTS
Cryogenic package
HTS Filter LNA
Interference signal
Interference signal
Desired signal
Noise level
P
P
f f
Cryogenic receiver Conventional receiver
Figure 10. A cryogenic package consisting of an HTS lter
and an LNA.
Figure 11. A comparison between a conventional dielectric res-
onator transceiver (left) and an HTS transceiver (right).
multiplexer was built [39] to duplicate the requirements of
the Intelsat 8 C-band dielectric resonator multiplexer. The
channel lters are 10-pole self-equalized planar HTS thin-
lm lters of the type shown in Fig. 3.
Figure 12 illustrates the overall 60-channel HTS mul-
tiplexer, while Fig. 13 shows the experimental results of
four channel HTS lters having bandwidths of 34, 41, 72,
and 112 MHz, which meet the bandwidth requirements of
the Intelsat 8 program. More than 50% reduction in mass
and 50% reduction in size were achieved with the use of
HTS technology [39]. It should be mentioned that the re-
ductions in mass and size include the cryogenics (two
cryocoolers for redundancy).
There have also been efforts dedicated to the develop-
ment of high-power HTS output multiplexers [43,65].
Figure 14 illustrates the layouts of a four-channel super-
conductive C-band output multiplexer and a similar
conventional waveguide output multiplexer. For the con-
ventional C-band multiplexer, both the channel lters and
the manifold are built using the waveguide technology. For
the superconductive multiplexer, the channel lters are
hybrid DR/HTS lters, while the manifold is realized us-
ing coaxial technology. The superconductive multiplexer
occupies less than 5% of the volume of the waveguide
multiplexer (without the cryocooler).
The potential advantages of using HTS technology in
the design of output multiplexers are mass and volume
reduction as well as insertion loss improvement. The in-
sertion loss improvement can translate into improvement
in satellite EIRPor reduction in the DC power required for
power ampliers. However, the mass penalty of the cur-
rently available cryocooler and associated electronic con-
troller may overshadow any advantages gained with the
use of HTS technology for high-power-output multiplexer
applications [43].
10. CRYOCOOLER TECHNOLOGY AND
CRYOPACKAGING
Several cryocooler techniques exist that could meet the
temperature needs for microwave HTS devices and sub-
systems. These techniques can be divided into two main
categories: open-cycle and closed-cycle [66]. The open-cy-
cle techniques include cooling with a stored cryogen such
as liquid nitrogen or through JouleThomson gas expan-
sion. Open-cycle coolers are bulky and dissipate materials,
requiring frequent llup of the cryogen or compressed gas.
Therefore, they are ideal for laboratory environments or in
applications where regular equipment maintenance is
possible.
Closed-cycle coolers are self-contained refrigerators
that consume only electrical power, requiring no mainte-
nance over the designed lifetime. Closed-cycle cryocoolers
have been primarily developed for infrared devices and
military applications. The main design considerations for
a closed-cycle cryocooler are DC power consumption, size,
and reliability. In particular, the reliability of the cryo-
cooler remains the primary barrier to the widespread com-
mercial acceptance of superconducting devices.
In addition to the cryocooler, it is also essential to pro-
vide a cryopackage with a means to mechanically and
thermally attach the HTS microwave device to the cold
head of the cryocooler within an evacuated enclosure,
while providing electrical and RF cabling through the en-
closure. One major challenge is to ensure low loss in the
Figure 12. A 60-channel HTS multiplexer integrated with cryo-
coolers.
Frequency (GHz)
I
L

(
d
B
)
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
4.25 4.2 4.15 4.1 4.05 4 3.95 3.9 3.85 3.8 3.75
Figure 13. Experimental results of four channels of the HTS
60-channel C-band input multiplexer.
Conventional C-Band
Output Multiplexer
HTSC-Band
Output Multiplexer
Figure 14. A comparison between a C-band four-channel HTS
output multiplexer and its waveguide counterpart.
electrical connections without allowing too much heat to
travel from the outside room-temperature environment to
the HTS component. It is difcult to nd commercially
available RF cables that can reasonably meet the two con-
ditions described above. For example, having a cable with
low conduction heat load requires the use of a small-di-
ameter cable. On the other hand, small-diameter cables
typically have a high RF insertion loss. The choice of the
RF and electrical connections to the package has a major
impact on the size and cooling power of the cryocooler.
Another major design issue is to guarantee that the en-
closure maintains its vacuum over the operational lifetime
of the system. Cryopackging is typically one of the main
issues in the design of microwave HTS subsystems.
11. CONCLUSIONS
The HTS technology offers the potential of large reduction
in mass and volume of microwave equipment. It could also
provide performance discrimination not attainable with
other technologies. The technology has grown very fast
and merged with commercial and defense applications.
This article has presented a summary of the main HTS
microwave applications. I strongly urge newcomers to the
eld to continue reading about the subject to understand
more about the potential challenges and limitations of this
revolutionary technology.
BIBLIOGRAPHY
1. Special Tenth Anniversary Plenary Sessions, IEEE Trans.
Appl. Superconduct. 7(2):69145 (June 1997).
2. P. H. Carr and B. R. McAvoy, eds., IEEE-MTTspecial issue on
microwave applications of superconductivity, IEEE Trans.
Microwave Theory Tech. MTT-39 (Sept. 1991).
3. M. Nisenoff and W. Meyers, eds., IEEE-MTT special issue on
the microwave and millimeter wave applications of high tem-
perature superconductivity, IEEE Trans. Microwave Theory
Tech. MTT-44 (July 1996).
4. J. J. Whelehan and G. C. Liang, eds., IEEE-MTT special issue
on microwave and communication applications at low tem-
peratures, IEEE Trans. Microwave Theory Tech. MTT-48
(July 2000).
5. H. Weinstock and M. Nisenoff, eds., Microwave Superconduc-
tivity, Kluwer Academic Publishers, The Netherlands, 2001.
6. M. J. Lancaster, Passive Microwave Device Applications of
High Temperature Superconductors, Cambridge Univ. Press,
Cambridge, MA, 1997.
7. Z. Y. Shen, High Temperature Superconducting Microwave
Circuits, Artech House, Norwood, MA, 1994.
8. J. C. Ritter, M. Nisenoff, G. Price, and S. A. Wolf, High tem-
perature superconductivity space experiment (HTSSE), IEEE
Trans. Magn. 27(2):25332537 (March 1991).
9. T. G. Kawecki et al., The high temperature superconductivity
space experiment (HTSSE-II) design, IEEE Trans. Microwave
Theory Tech. MTT-44:11981212 (July 1996).
10. G. W. Mitschang, Space applications and implications of high
temperature superconductivity, IEEE Trans. Appl. Supercon-
duct. 8(2):6973 (June 1995).
11. M. Nisenoff and W. J. Meyers, On-orbit status of the high
temperature superconductivity experiment, IEEE Appl. Su-
perconduct. Trans. 11:799805 (March 2001).
12. F. Patten, US programs in HTS electronics and radar appli-
cations, Workshop on Superconducting Microwave Applica-
tions to Mobile Communications, Asia-Pacific Microwave
Conf., Yokohama, Japan, Dec. 1998.
13. R. B. Hammond et al., HTS wireless lters; past, present and
future performance Microwave J. 41:94107, 1998.
14. A. I. Braginski, Superconducting electronics coming to
market, IEEE Trans. Appl. Supercond., 9:28252836,
1998.
15. M. J. Rowell, Recommended directions of research and de-
velopment in superconducting electronics, IEEE Trans.
Appl. Supercond. 9:28372848, 1999.
16. D. G. Smith and V. K. Jain, Superconducting lters for wire-
less communications; A reappraisal, IEEE Trans. Appl. Su-
percond., 9, pp. 40104013, 1999.
17. D. E. Oates, A. C. Alfredo, S. M. David, and A. M. Sami,
Stripline resonator measurements of Zs vs H
f
in thin lms,
IEEE Trans. Microwave Theory Tech. MTT-39:15221529
(Sept. 1991).
18. P. P. Nguyen et al., Nonlinear surface impedance for YBCO
thin lm: Measurements and a coupled-grain model, Phys.
Rev. B, 48:64006412 (Sept. 1993).
19. C. Wilker et al., Nonlinear effects in high temperature super-
conductors: 3
rd
order intercept from harmonic generation,
IEEE Trans. Appl. Superconduct. 5(2):16651670 (June
1995).
20. O. G. Vendik, I. B. Vendik, and T. B. Samoilova, Nonlinearity
of superconducting transmission line and microstrip resona-
tor, IEEE Trans. Microwave Theory Tech. 45(2):173178 (Feb.
1997).
21. R. R. Mansour, B. Jolley, S. Ye, F. S. Thomson, and V. Dokas,
On the power handling capability of high temperature super-
conductive lters, IEEE Trans. Microwave Theory Tech. MTT-
44:13221338 (July 1996).
22. G. Coutts, R. R. Mansour, and S. K. Chaudhuri, High tem-
perature superconducting nonlinear transmission lines,
(Dec. 2000).
23. J. Clark, Principles and applications of SQUIDs, Proc. IEEE
77(2):12081223 (1989).
24. N. Klein et al., High-Q dielectric resonator devices at cryo-
genic temperatures, IEEE Trans. Appl. Superconduct.
9(2):35733576 (1999).
25. N. Klein et al., Properties and applications of HTS-shielded
dielectric resonators: A state-of-the-art report, IEEE Trans.
Microwave Theory Tech. 44:13691373 (July 1996).
26. J. D. Doss, Engineers Guide to High Temperature Supercon-
ductivity, Wiley, New York, 1989.
27. T. P. Orlando and K. A. Delin, Foundations of Applied Super-
conductivity, Addison-Wesley, Reading, MA, 1991.
28. T. Van Duzer and C. W. Turner, Principles of Superconductive
Devices and Circuits, Prentice-Hall, Englewood Cliffs, NJ,
1999.
29. C. W. Lam, D. M. Sheen, S. M. Ali, and D. E. Oates, Modeling
the nonlinearity of superconducting strip transmission lines,
IEEE Trans. Appl. Superconduct. 2(2):5865 (June 1992).
30. R. M. A. Megahed and S. M. El-Ghazaly, Nonlinear analysis of
microwave superconductor devices using full-wave electro-
magnetic model, IEEE Trans. Microwave Theory Tech.
43(11):25902598 (Nov. 1995).
31. J. Kuster, R. Dill, and P. Russer, Filed theory investigation of
high-Tc superconducting coplanar waveguide transmission
lines and resonators, IEEE Trans. Microwave Theory Tech.
39:15661574 (1991).
32. R. R. Mansour et al., Design considerations of superconduc-
tive input multiplexers for satellite applications, IEEE Trans.
Microwave Theory Tech. MTT-44:12131228 (July 1996).
33. I. Bahl and P. Bhartia, Microwave solid state circuit design, in
I. Bahl and P. Bhartia, eds., Filters and Multiplexers, Wiley,
New York, 1988, Chap. 6.
34. M. Lancaster et al., Miniature superconducting lters, IEEE
Trans. Microwave Theory Tech. MTT-44:13391346 (1996).
35. K. F. Raihan, R. Alvarez, J. Costa, and G. Hey-Shipton, High-
ly selective HTS band pass lter with multiple resonator
cross-couplings, IEEE MTT-S Int. Microwave Symp. Digest,
June 2000, pp. 661664.
36. S. Ye and R. R. Mansour, A novel split-resonator high power
HTS planar lter, IEEE MTT-S Int. Microwave Symp. Digest,
June 1997, pp. 299302.
37. G. Tsuzuki, M. Suzuki, and N. Sakakibara, Superconducting
lter for IMT-2000 band, IEEE Trans. Microwave Theory
Tech. MTT-48:25192525 (Dec. 2000).
38. R. R. Mansour, S. Ye, S. Peik, V. Dokas, and B. Fitzpatrick,
Quasi dual-mode resonators, IEEE Trans. Microwave Theory
Tech. MTT-48(12):24762482 (Dec. 2000).
39. R. R. Mansour et al., A 60 channel superconductive input
multiplexer integrated with pulse-tube cryocoolers, IEEE
Trans. Microwave Theory Tech. MTT-48(7):11711180 (July
2000).
40. D. Kajfez and P. Guilon, Dielectric Resonators, Artech House,
Norwood, MA, 1986.
41. R. R. Mansour, V. Dokas, G. Thomson, W. C. Tang, and C.
Kudsia, A C-band superconductive input multiplexer for com-
munications satellites, IEEE Trans. Microwave Theory Tech.
MTT-42:24722479 (Dec. 1994).
42. J. A. Curtis and S. J. Fiedziuszko, Hybrid dielectric/HTS res-
onator and their applications, IEEE MTT-S Int. Microwave
Symp. Digest, 1991, pp. 447450.
43. R. R. Mansour, S. Ye, V. Dokas, B. Jolley, W. C. Tang, and C.
Kudsia, System integration issues of high power HTS output
multiplexers, IEEE Trans. Microwave Theory Tech. MTT-
48:11991208 (July 2000).
44. R. R. Mansour and V. Dokas, Miniaturized Dielectric Reso-
nator Filters and Method of Operation Thereof at Cryogenic
Temperatures, U.S. Patent 5,498,771 (1996).
45. N. M. Alford et al., Low surface resistance in YBCO melt-
processed thick lms, Nature 349:680 (1991).
46. T. W. Button et al., The processing and properties of high Tc
thick lms, IEEE Trans. Magn. 27:14341437 (1991).
47. T. W. Button and N. M. Alford, High Q YBCO cavities, Appl.
Phys. Lett. 60:13781380 (1992).
48. T. W. Button et al., Properties and applications of thick lm
high temperature superconductors, IEEE Trans. Microwave
Theory Tech. MTT-44:13561360 (July 1996).
49. W. N. Hardy et al., Split-ring resonator for use in magnetic
resonance from 2002000MHz, Rev. Sci. Instrum. 52(2):213
(1981).
50. R. D. Lithgow and J. M. Peters, Electromagnetic Resonant
Filter Comprising Cylindrically Curved Split Ring Resona-
tors, U.S. Patent 6,616,540A (1997).
51. H. H. S. Javadi et al., Jet Propulsion Laboratory/NASA Lewis
Research Center space qualied hybrid high temperature super-
conducting/semiconducting 7.4GHz, low noise downconverter,
(July 1996).
52. R. Romano, R. R. Mansour, and F. Thomson, A hybrid super-
conductive/semiconductive microwave receiver, IEEE Trans.
Appl. Superconduct. 7:30673070 (June 1997).
53. W. G. Lyons et al., High temperature superconductive wide-
band compressive receivers, IEEE Trans. Microwave Theory
Tech. MTT-44:12581278 (July 1996).
54. S. H. Talisa et al., High temperature superconducting space-
qualied multiplexers and delay lines, IEEE Trans. Micro-
wave Theory Tech. MTT-44:12291239 (July 1996).
55. G. C. Liang et al., Space qualied superconductive digital in-
stantaneous frequency-measurement subsystem, IEEE
Trans. Microwave Theory Tech. MTT-44:12891299 (July
1996).
56. D. Oates, Analog devices in military systems, Microwave Su-
perconductivity, Chapter 12, Kluwer Academic Publishers,
The Netherlands, 2001.
57. H. Chaloupka et al., Miniaturized high temperature super-
conductor microstrip patch antenna, IEEE Trans. Microwave
Theory Tech. MTT-39:15311529 (Sept. 1991).
58. C. Kuester, H. Chaloupka, and J. Knauth, A high tempera-
ture superconducting anti-jam GPS antenna array, Proc. Mil-
itary Communications Conf., 1999 (MILCOM 1999), 1999,
Vol. 1, pp. 671674.
59. S. Peik and R. R. Mansour, High temperature superconductor
beam forming network, IEEE MTT-S Int. Microwave Symp.
Digest, June 1999, pp. 10151019.
60. G. Koepl, A receiver front end for wireless base stations,
Microwave J. 39(4) (April 1996).
61. G. Koepl, Superconductor improve coverage in wireless net-
work, Microwave RF 37(4):94107 (April 1998).
62. R. W. Simon, Superconductor lters in wireless base stations,
Commun. Syst. Design (July 1997).
63. J. P. Simmons and J. Madden, Practical HTS/cryogenic sys-
tems for wireless applications, Microwave J. (Oct. 1996).
64. J. S. Fiedziuszko, J. A. Curtis, and S. C. Holme, Low-loss
multiplexers with planar dual-mode HTS resonators, IEEE
Trans. Microwave Theory Tech. MTT-44:12481257 (July
1996).
65. M. Klauda et al., Superconductors and cryogenics for future
communication systems, IEEE Trans. Microwave Theory
Tech. MTT-48(7):12271239 (July 2000).
66. W. A. Little, Recent advances in low cost cryogenic coolers for
electronics, Proc. Semiconductor Thermal Measurement and
Management Symp., 2000, pp. 110111.
HISTORY OF WIRELESS COMMUNICATION
PATT LEGGATT
In the earlier decades of the nineteenth century, magnetic
and electric forces were believed to act instantaneously at
a distance, a view prevalent in continental Europe and
especially in Germany. But the obvious philosophical
difculties inherent in instantaneous action at a distance
made this concept less acceptable to a few in the United
Kingdom in general and to Michael Faraday in particular.
HISTORY OF WIRELESS COMMUNICATION 2015
1. FARADAYS WAVES
In 1831 Faraday conducted a series of experiments on the
magnetic effects of electric currents, during which he
made his important discovery of electromagnetic induc-
tion, whereby an electric potential was induced in a
conductor subjected to a changing magnetic eld. In the
course of these experiments, Faraday came to the conclu-
sion that magnetic and electrostatic forces were not in-
stantaneously effective at a distance but required a nite
time for their transmission. We know now that electro-
magnetic energy is transmitted at the speed of light, but
while continuing his experiments into 1832, Faraday had
no means of detecting or measuring the very small time
intervals necessary to conrm his theories. Furthermore
he suggested, with remarkable intuition, that transmis-
sion of such forces took the form of some kind of wave
motion. To establish his prior claim to the notion of wave
motion, Faraday deposited a written statement in a sealed
envelope with the Royal Society in 1832. The envelope was
nally opened more than 100 years later in 1937 by the
then-president of the Royal Society, and found to contain,
inter alia, the following words: I am inclined to compare
the diffusion of magnetic forces from a magnetic pole to the
vibrations upon the surface of disturbed water, or those of
air in the phenomenon of sound.
Faraday, the son of a blacksmith, was a brilliantly
imaginative experimenter and theoretician but had only
self-taught scientic education and no knowledge of
mathematics. Because of this, his theories were regarded
with some disdain by many contemporary scientists.
However, there was one young mathematician, James
Clerk Maxwell, who was most impressed by Faradays
concepts of magnetic elds and lines of force as set out in
his paper Experimental researches in electricity, read to
the Royal Society in 1851.
2. MAXWELLS ANALYSIS
Maxwell already had an interest in this area, triggered
while he was at Cambridge by the ideas of William
Thomson, afterward to become Lord Kelvin; and Maxwell
determined to submit Faradays concepts to detailed
mathematical analysis. As his studies continued, he con-
centrated his attention on the possible nature of the
medium through which electromagnetic forces could be
propagated, wishing to devise a mechanical model exhi-
biting appropriate characteristics. This was a complex
task, but by about 1862 he came up with a system of
very minute rapidly spinning eddies or vortices, each
surrounded by a layer of even more minute particles
revolving in a direction opposite to that of the vortices
themselves.
The vortices, with their outer layers of particles, could
interact with one another in a manner analogous to a train
of gear wheels, so that energy imparted to one vortex
would be transferred to others and so progress through
the medium.
Maxwell devised this mechanical model as one that
could exhibit the sort of behavior necessary to embrace
Faradays concepts of lines of force and wavelike transmis-
sion of energy; but it was a convenient model only, and he
was not so fanciful as to suggest that it represented
anything much related to reality. Nevertheless it served
his purpose of enabling rigorous mathematical analysis,
culminating in his 1864 paper to the Royal Society, A
dynamical theory of the electro-magnetic eld. One of
Maxwells main hopes had been to derive the electromag-
netic nature of light and heat, and this was triumphantly
achieved in particular by one of the consequences of his
theories, that electromagnetic wave propagation would
travel at a velocity very close to the value for the speed
of light that had been experimentally determined by
others at the time.
3. THE MAXWELLIANS
Maxwells analysis was deeply mathematical, making use
of Hamiltonian quaternionic calculus and high-order dif-
ferential equations, and at the time of his nal publication
in 1873, there were very few physicists with the intellec-
tual ability to understand it. There were only three of
note: Oliver Lodge and Oliver Heaviside in England, and
the Irish professor George Francis FitzGerald of Trinity
College, Dublin.
In 1873 Oliver Lodge, then a student at University
College, London, attended a lecture by Maxwell at a
meeting of the British Association and obtained a copy of
his Treatise published in that year. Lodge did not study
this closely until 1876, but when he did he quickly came to
realize that Maxwells equations implied not only the
electromagnetic nature of light and heat but also that
there could be a whole spectrum of radiation with wave-
lengths both above and below those of visible light. Lodge
was probably the rst to appreciate that such electromag-
netic waves could perhaps be generated electrically; and
in 1879, after Maxwells death, he began to give serious
attention to this possibility.
FitzGerald too was studying Maxwells theories at
about this time and, in a paper to the Royal Dublin Society
in 1882, suggested that electromagnetic radiation of about
10 wave-length could be generated by discharging a con-
denser (Leyden jar) through a circuit of very low resis-
tance and low inductance. Furthermore, he claried
Maxwells analysis to the extent of showing that the
equations also led to the laws of reection and refraction
that had already been developed in the wave theories of
light.
Another physicist who contributed signicantly in this
way was Oliver Heaviside, now best remembered for his
1902 theory of a reecting ionized layer in the upper
atmosphere, which, when physically veried some 20
years later, was named after him. Heaviside was a rst-
class mathematician and was already interested in Max-
wells papers. He reformulated much of the analyses in
appreciably simpler terms, changing Maxwells rather
convoluted systems into notation of his (Heavisides) own
devising. He developed his own operational calculus and
vector algebra and although contemporary mathemati-
2016 HISTORY OF WIRELESS COMMUNICATION
cians found it difcult to grasp at rst, it is in this form
that Maxwells equations are familiar to students of today.
These three adherents and interpreters of the equa-
tions called themselves Maxwellians. They were scien-
tists, and what was crucial to them was that the
electromagnetic radiation implicit in the equations indi-
cated that the unsatisfactory theories of instantaneous
action at a distance could be discarded. While the theore-
tical analyses were their prime interest, they naturally
hoped for experimental verication of the real existence of
such radiation, and it was Lodge who pursued this most
effectively.
4. OLIVER LODGE
Oliver Lodge was born in Staffordshire, England in 1851,
the son of a pottery merchant. He received his scientic
education at University College, London and while there
attended a lecture by James Clerk Maxwell showing that
oscillatory electrical discharges resulted in electromag-
netic wave radiation propagating with the velocity of light.
This red Lodge with an abiding interest in what was
eventually to become wireless.
Lodge quickly rose to be an eminent physicist, gaining
his D.Sc. degree in 1877 and being appointed Professor of
Experimental Physics at the University College of Liver-
pool in 1881. In 1887 he was elected Fellow of the Royal
Society, the highest honor in British science, and he was
knighted as Sir Oliver Lodge in 1902. He died at age 89 in
1940.
In 1887 Lodge undertook a series of investigations into
lightning discharges and protection against them. He
found by experiment that lightning ashes were oscilla-
tory and followed this up with further experiments invol-
ving spark discharges from Leyden jars. In early 1888 he
developed a famous experiment that he called the recoil
kick. This involved generating oscillations by discharge of
a Leyden jar capacitor into a long pair of wires and
observing that a much greater spark would occur at the
end of the wires when they were of a suitable length. He
correctly surmised that this was due to a standing-wave
pattern along the wires with a voltage anti-node at the
end. Furthermore, he understood that this condition
would be satised when the wires were a half-wavelength
long (or a multiple of this) and was thus able to determine
experimentally the wavelength of the oscillations.
It can be seen therefore that he was conversant with
the principles of resonance and tuningor syntony as he
called itsaying in 1888, The natural period of oscilla-
tions in the wires will then agree with the oscillation
period of the discharging circuit, and the two will vibrate
in unison, like a string or a column of air resounding to a
reed.
Although his experiments were largely conned to
oscillations along wires, Lodge knew well that the electro-
magnetic waves were propagated in the space surround-
ing the wires rather than in the wires themselves. He
knew also, being familiar with Clerk Maxwells mathema-
tical analysis, that the waves would be radiated into space
and travel at the speed of light.
Thus we can see that Lodge had demonstrated experi-
mentally the existence of electromagnetic waves as pre-
dicted by Maxwells equations. But he had dealt only with
waves guided along wires, since he had not then devised
any means of detecting such radiation in free space.
In fact Lodge was not the rst to observe electromag-
netic waves along wires. In 1870 Wilhelm von Bezold
observed such phenomena, detecting the waves by pat-
terns formed by dust particles under the inuence of
electrostatic elds. However, he did not relate these ob-
servations to Maxwells theories and his work attracted
little notice.
5. HEINRICH HERTZ
At the same time that Lodge was undertaking his experi-
ments, there were even more effective investigations being
carried out by Heinrich Hertz in Germany. Hertz, the son
of a lawyer, was born in February 1857. He turned
eventually to science with a years course at Munich,
transferring in 1878 to the University of Berlin. Here he
studied under Professor von Helmholtz, who gave much
encouragement to one he recognized as an outstanding
pupil. Finally, having been awarded his doctorate, Hertz
was appointed assistant professor at the Physics Institute
of Berlin in 1880.
Hertz had been brought up in the instantaneous
action at a distance school of thought, but fairly early in
his career he was introduced to Maxwells theories by
professor von Helmholtz, who encouraged him to attempt
experimental proof of Maxwells postulated displacement
current in air or empty space. Hertz did not immediately
take this up, but he was intrigued by the possibility that
the concepts of displacement currents and electromag-
netic waves could fundamentally change action at a dis-
tance theories.
A few years later he had been appointed professor at
the Technical High School at Karlsruhe and found there in
a collection of old physical apparatus a pair of Knock-
enhauer spirals, at coils wound in wooden frames. Ex-
perimenting casually with these, he noticed that
discharging a Leyden jar through one of the coils gave
rise to a small spark across the open terminals of the other
some distance away. This revived his earlier interests, and
he began to devote effort to theoretical clarication of
Maxwells equations. He became increasingly convinced
that the equations could indeed give the true explanation
of electric and magnetic eld phenomena, and by 1884 he
wrote I think we may infer without error that if the
choice rests only between the usual system of electromag-
netics and Maxwells, the latter is certainly to be pre-
ferred. But the physical existence of electromagnetic
waves, and especially their nite velocity of propagation,
needed to be established by practical demonstration, so
Hertz undertook a series of experiments culminating in
the famous ones of 1887/88 that proved the point beyond
all doubt. It is true that others before him had, rather
accidentally, observed electromagnetic radiation, notably
Mahlon Loomis in the United States in 1872 and David
Hughes in England in 1879, but neither of them under-
stood what was happening or were familiar with Max-
wells work.
Hertzs investigations into the subject had involved
generation, detection, and measurement of waves in free
space, rather than along wires. Lodge generously ac-
knowledged that Hertz experiments were superior to his
own and a more convincing proof of the validity of Clerk
Maxwells theories.
After his experiments Hertz undertook further theore-
tical interpretation and development of the Maxwell con-
cepts, much helped by his correspondence with
FitzGerald, Lodge, and Heaviside, which revealed signi-
cant earlier work by these Maxwellians that he had not
previously heard of. Hertzs papers in 1890 were particu-
larly important in the eld of theoretical physics and were
inuential in setting the scene for the later achievements
of Lorentz and Einstein. Hertz died in 1894 at the early
age of 36.
Hertz was not primarily an experimentalist seeking to
demonstrate the existence of electromagnetic waves. He
was a theoretical physicist who conducted his famous
experiments as a means of justifying his rm conclusion
on a matter of fundamental scientic importance. The
waves were not important to Hertz for their own sake;
he saw them simply as affording proof that Maxwells
equations gave the true picture and that hitherto accepted
theories of action at a distance must therefore be regarded
as obsolete.
The three Maxwellians, Lodge, FitzGerald, and Heavi-
side, were of a like mind. They were satised with the fact
that Hertz had experimentally demonstrated the real
existence of electromagnetic radiation, and neither they
nor Hertz himself concerned themselves with any possible
practical applications such as communication.
Someone soon did, however; Richard Threlfall, as pre-
sident of the Australasian Association for the Advance-
ment of Science, proposed in 1890 using Hertzian waves
for communication purposes. But no one else saw this as a
practical proposition at the time, the range of a few yards
achieved by Hertz in his Karlsruhe laboratory not seem-
ing to offer very much.
6. WIRELESS TELEGRAPHY BEGINS WITH MARCONI
Five or six years went by with nothing very signicant
happening until Hertz death in 1894. But ironically, this
was one further event involving Hertz that had a most
profound impact, for in that year the 20-year-old Gu-
glielmo Marconi, on holiday in the Italian Alps, read an
obituary describing the work of Hertz. He was immedi-
ately inspired to consider whether Hertzian waves might
not form the basis of a wireless telegraph communication
system, and dedicated himself to this idea for the rest of
his life.
At the end of his holiday Marconi returned to the
family home in the Villa Griffone near Bologna, and at
once commenced experiments in his attic workshop, where
he had long since occupied himself with the electrical
devices that had fascinated him from boyhood.
It is a well-known story how Marconi improved his
apparatus and techniques to achieve greater and greater
ranges during 1894/95; how he came to England in 1896 to
make further progress; and how he spanned the Atlantic
with the letter S in Morse in 1901. This story need not be
told again here, but Marconis relationship with Oliver
Lodge is perhaps less well known and may usefully be
described.
7. LODGE AND COHERERS
Although it was Hertz who rst contrived to demonstrate
the existence of electromagnetic waves in free space and
that they exhibited reection and refraction in the same
way as light, it is true to say that Lodge knew and
understood as much or more about the nature and beha-
vior of the waves than did anyone else in the latter years of
the nineteenth century.
As already mentioned, Lodge was thoroughly conver-
sant with resonance and tuning, and he also well under-
stood the principles underlying radiation from antennas of
various congurations. Although he knew that waves
must be radiated into free space, he lacked any means of
detecting them; and it was use of the crude, but just
adequate, resonant sparkgap detector that enabled Hertz
to effect his splendidly successful experiments.
But in 1889, the year after Hertz demonstrations,
Lodge made a discovery that was to prove a crucial step
forward on the path to a practical wireless communication
system. During his investigations into lightning and the
analogous effects of spark discharges from Leyden jars,
Lodge observed that two iron spheres or other metal
surfaces very close together would at times fuse together
to form a conducting path when subjected to a Hertzian
wave pulse. He called this arrangement a coherer, saying
that it formed an astonishingly sensitive detector of
Hertzian waves. Later, in 1893, he was made aware of
the work of Edouard Branly in France, who had observed
similar cohering effects with a glass tube lled with metal
lings. Lodge immediately tried this for himself and found
it very much more sensitive than his own iron spheres. In
fact the phenomenon of coherence resulting from nearby
spark discharges had been independently noticed by
others some years earlier, including Guitard in 1850,
Varley in 1866, and Onesti in 1874; but these predecessors
had not in fact ascribed the effects to Hertzian waves, and
neither had Branly in his 1890 experiments. In 1902
Lodge, in conjunction with Alexander Muirhead, invented
a new form of coherer in which a knife-edged steel wheel
grazed the surface of a small pool of mercury covered with
a lm of oil: an incoming radio pulse ruptured the thin oil
lm and allowed low-resistance contact between the steel
and the mercury. This type of coherer was at least as
sensitive as any other and a good deal more stable in
operation.
Lodge did not at the time attempt to put his coherers to
practical use, but Marconi used a lings version in his
early apparatus. It was this, plus the use of elevated
antennas and connection to Earth, which enabled Marconi
to develop and steadily improve his equipment to the point
where it could be seen as a workable wireless communica-
tion system.
8. LODGE VERSUS MARCONI
Marconi came to England in 1896 as a young man of
nearly 22, with little theoretical knowledge of wireless
principles but full of enthusiasm and dedication. He had a
air for publicity and in December gave a demonstration
to members of the public, carrying a blackbox around the
audience that rang a bell whenever a key at the end of the
hall was depressed. Representatives of the press were
present, and the next day Marconi was headline news,
hailed as the inventor of wireless. In fact Marconi was
always quite modest and did not claim to be an inventor
but rather that he took up others peoples ideas and
inventions and improved them. But Lodge, and other
scientists who had made signicant contributions in the
eld, was naturally indignant about the adulation of a
young Italian upstart: Lodge may thus be forgiven his
testy comment in later years that It was stale news to me
and to a few others, but he then added more generously,
But whereas we had been satised that it could be done,
Marconi went on enthusiastically and persistently till he
made it a practical success.
Another area of interplay between Marconi and Lodge
was tuning. As already mentioned, Lodge had a full
understanding of the principles of resonance and tuning
and in May 1897 he applied for a patent on his syntony
ideas.
Tuning, and the resultant ability to separate one
transmission from another, was of course a vital necessity
for the development of Marconis wireless system and he
tried many different circuit arrangements with gradually
increasing success. Finally in 1900 his famous four
sevens patent was granted. It may appear strange that
he was granted this patent when Lodge had registered his
some years earlier; but it seems that it was not considered
that one infringed the other, and it was ruled in court that
the two were complementary rather than duplications.
Nevertheless the existence of Lodges patent was seen as
an embarrassment by the Marconi Company, especially
when its validity was extended in 1911 by a further 7
years. Accordingly, the Marconi Company negotiated with
Lodge and bought his patent for a considerable sum.
9. THE LODGEMUIRHEAD SYNDICATE
Another source of rivalry was the appearance of the
LodgeMuirhead Syndicate with a competing wireless
telegraphy system. Although Lodge was basically unin-
terested in commercial exploitation of his Hertzian wave
experiments and discoveries, such a venture was sug-
gested to him by Dr. Alexander Muirhead after he had
attended one of Lodges lectures in 1894. This eventually
resulted in formation of the LodgeMuirhead Syndicate in
which Lodge provided the scientic ideas and Muirhead
a very able telegraph engineerthe design of practical
equipment.
By 1903 the Syndicate was ready with a well-designed
and effective system incorporating the Lodge steelmer-
cury coherer, but they found themselves up against the
Marconi Companys monopoly of coastal stations in the
United Kingdom that it had negotiated with Lloyds of
London from 1901, and the companys contracts with
many shipping lines for exclusive use of Marconi equip-
ment and operators.
Faced with this situation, the Syndicate could nd only
limited markets in the military eld and in a few overseas
countries. So despite its technical excellence, the Lodge
Muirhead system was not a commercial success and was
wholly bought out by the Marconi Company in 1911,
together with the Lodge tuning patent referred to pre-
viously.
10. CONTINUOUS-WAVE TELEGRAPHY
Early radiotelegraphy was effected by spark trains pro-
duced as Morse signals by a telegraph key. The spark-
induced oscillation bursts making up a dot or dash were
randomly phased.
It was realized that continuous wave trains could be
advantageous in offering narrower bandwidth transmis-
sion and hence the possibility of more precise tuning of
transmitter and receiver, and various developments were
undertaken to achieve this. The rst was Marconis syn-
chronous spark discharger wherein multiple sparkgaps on
a rotating disk ensured that the oscillations from one gap
would be so phased as to continue the oscillations from the
preceding gap, giving a reasonable approximation of a
continuous wave.
An alternative approach was the Poulsen system, based
on earlier work by Duddell, using the negative resistance
characteristic of an arc to generate continuous oscillation
in a parallel resonant circuit. This was quite widely used
by the German Telefunken company for a radiotelephone
system, but was not always successful due to the difculty
of maintaining a steady arc discharge.
Another alternative was the high-frequency alternator.
Difculties here arose from the very high rotational
speeds necessary to produce even quite modest radio-
frequencies; but Ernst Alexanderson of American General
Electric designed very successful machines capable of
100 kHz at powers of hundreds of kilowatts. A later
development in Germany was the Goldschmidt HF, alter-
nator, which reduced the need for excessive rotational
speed by an ingenious system of frequency multiplication
within the machine. With suitable coils and pole pieces,
and moderate speed, the rotor could produce a frequency
of perhaps 15 kHz; this would be induced in the stator,
setting up a rotating eld therein. The stator eld was
arranged to rotate in the direction opposite that of the
rotor, so that an oscillation of double the frequency, 30kHz
was induced in the stator and 30 15 =45 kHz in the
rotor. This 45 kHz was selected by a tuned lter resulting
in a frequency of 60 kHz in the stator; the latter frequency
was applied to the aerial for transmission. The Gold-
schmidt patents were bought up by the Marconi Company
in 1913.
A difculty with any of these machines was to maintain
very precise alternator rotational speed, since even a
small uctuation would seriously detract from the poten-
tial for exact frequency stability and receiver selectivity.
Particularly difcult was avoidance of speed changes with
electrical load when the Morse key was operated; and this
was compensated to some extent by a subsidiary key
contact adjusting the alternator driving motor eld
when the key was depressed.
A nal problem was that a continuous wave would by
itself produce no sound in receiver headphones. To render
Morse signals audible required the wave to be modulated
at audiofrequency. Crude modulation could be provided by
a ticker, a chattering contact that broke up the contin-
uous wave into audiofrequency groups. A better solution
was to modulate the signals with a frequency about 1kHz
different from the main carrier to produce an audible
beat note. This was achieved by Fessendens heterodyne
arrangement or by Goldschmidts tonewheel.
11. RADIOTELEPHONY
Most early workers were content to communicate in Morse
code telegraphy, but Reginald Fessenden, a Canadian
working in the United States, considered radiotelephony
much preferable. In 1900 he achieved speech transmission
over a distance of 1mi using a spark transmitter with a
spark repetition frequency of 10,000/s. But modulating
speech on a spark signal has been memorably described as
like printing a newspaper on a roll of stair carpet, and
articulation and background noise were very unsatisfac-
tory.
Fessenden realized that a continuous wave carrier was
necessary for satisfactory speech modulation, and he
initially experimented with arc-based oscillators, but
found a HF alternator more satisfactory. On Christmas
Eve 1906, and again on New Years Eve, he successfully
transmitted speech and music from his Brant Rock station
using a HF alternator. Remarkably, his experimental
speech transmissions shortly before these two events
were heard by chance at Fessendens receiving station at
Machrihanish in Scotland.
In the very early years of the twentieth century the
only radiowave detector was the coherer; but this, which
was an ON/OFF device, was suitable only for reception of
Morse transmissions. Fessenden sought a continuously
operating detector that would be suitable for demodulat-
ing speech signals on a continuous-wave carrier, and in
1903 invented his liquid barretter, an electrolytic gas
generator that did indeed respond continuously to the
amplitude of a received carrier. This was used in conjunc-
tion with his 1906 speech and music transmissions,
although by this time the Marconi magnetic detector
was also in use and capable of amplitude demodulation,
as were Flemings diode of 1904 and the crystal detectors
introduced by Braun, Austen, Pickard, Pierce, and
Dunwoody from 1906.
Lee de Forest developed an arc-based radio telephone
in 1907/08, which was supplied in some quantity to the
U.S. Navy.
12. VACUUM TUBES
On the basis of Thomas Edisons discovery in 1883 of
unilateral conduction or electric current between an in-
candescent lament and an adjacent plate electrode,
John Ambrose Fleming in England devised his oscillation
valve in 1904. In 1906 de Forest introduced a third grid
electrode between lament and plate, thus inventing the
triode tube: but he had little understanding of the operat-
ing principles of his invention, and it was the detailed
studies of Howard Armstrong and Irving Langmuir that
developed it into a reliable and practical device by 1915.
Also in 1915, Armstrong developed the regeneration
principle and hence the triode oscillator, and this nally
fullled the need for convenient and stable generation of
continuous radiofrequency oscillations that could readily
be amplitude modulated for radiotelephony applications.
Furthermore, the triode could be congured as a sensitive
detector of such signals.
Thus was opened the way to modern electronics and, a
few years later, to entertainment broadcasting of speech
and music.
FURTHER READING
M. Adams, The Race for the Radiotelephone, AWA Review, Vol. 10,
The Antique Wireless Association, Bloomeld, NY, 1996.
W. J. Baker, A History of the Marconi Company, Methuen,
London, 1970.
A. Briggs, The Golden Age of Wireless, Oxford Univ. Press, New
York, 1965.
L. Coe, Wireless Radio: A Brief History, McFarland, Jefferson, NC,
1996.
H. M. Dowsett, Wireless Telegraphy and Broadcasting, Gresham
Publishing, London, 1923.
H. M. Dowsett, Handbook of Technical Instruction for Wireless
Telegraphists, The Wireless Press, London, 1925.
O. E. Dunlap, Marconi, The Man and His Wireless, Arno Press,
New York, and The New York Times, 1971.
W. H. Eccles, Continuous Wave Wireless Telegraphy, Wireless
Press, London, 1921.
J. J. Fahie, A History of Wireless Telegraphy, Arno Press, New
York, and The New York Times, 1971.
H. M. Fessenden, Fessenden Builder of Tomorrows, Arno Press,
New York, l974.
G. M. Garratt, The Early History of Radio, IEE, London, 1994.
J. V. Hinshaw, Marconi and His South Welleet Wireless, Cha-
tham Press, Chatham, MA, 1969.
W. P. Jolly, Marconi, Constable, London, 1972.
T. S. W. Lewis, Empire of the Air, Harper Collins, New York, 1991.
T. L. Mayes, Wireless Communication in the United States: The
Early Development of American Radio Operating Companies,
New England Wireless and Steam Museum, East Greenwich,
RI, 1989.
J. G. OHara and W. Pricha, Hertz and the Maxwellians, IEE,
London, 1987.
V. J. Phillips, Early Radio Wave Detectors, IEE, London, 1980.
P. Rowlands and J. P. Wilson, Oliver Lodge and the Invention of
Radio, PD Publications, Liverpool, UK, 1994.
R. Stanley, Book of Wireless Telegraphy, Longmans, London, 1919.
G. E. Sterling, The Radio Manual, Van Nostrand, New York,
1928.
G. F. J. Tyne, Saga of the Vacuum Tube, Howard W. Sams,
Indianapolis, 1977.
HORN ANTENNAS
EDWARD V. JULL
University of British Columbia
Vancouver, British Columbia
Canada
A horn antenna is a length of conducting tube, ared at
one end, and used for the transmission and reception of
electromagnetic waves. For an efcient transition between
guided and radiated waves, the horn dimensions must be
comparable to the wavelength. Consequently horns are
used mostly at centimeter and millimeter wavelengths. At
lower or higher frequencies they are inconveniently large
or small, respectively. They are most popular at micro-
wave frequencies (330 GHz), as antennas of moderate di-
rectivity or as feeds for reectors or elements of arrays.
Since acoustic horns have been in use since prehistoric
times, the design of horns as musical instruments was a
highly developed art well before the appearance of the rst
electromagnetic horns. This occurred shortly after Hertz
in 1888 rst demonstrated the existence of electromag-
netic waves. Experimenters placed their sparkgap sources
in hollow copper tubes (Figs. 1a, 5a). These tubes acted as
highpass lters for microwave and millimeter-wave radi-
ation from the open end. In London in 1897 Chunder Bose
used rectangular conducting tubes with collecting fun-
nels, or pyramidal horns (Fig. 1d) in his demonstrations
at 5 and 25 mm wavelengths [1]. Thus the electromagnetic
horn antenna was introduced but this early beginning of
microwave invention closed with Marconis demonstration
that longer wavelengths could be received at greater dis-
tances. Horns were too large to be practical at those wave-
lengths, and it was almost 40 years before microwave
horns reappeared with the need for directive antennas for
communications and radar. Horns alone were often not
sufciently directive but combined in an array or with a
lens (Fig. 4a), or more often a parabolic reector (Figs. 4b,
4c) highly directive antenna beams are obtained.
1. RADIATING WAVEGUIDES AND HORNS
Horns are normally fed by waveguides supporting only the
dominant waveguide mode. For a rectangular waveguide
(Fig. 1a) with TE
01
mode propagation only, these dimen-
sions in wavelengths l are l/2oaol and bEa/2. Open-
ended waveguides have broad radiation patterns, so when
used as a feed for a reector, there is substantial spillover,
or radiation missing the reector and radiation directly
backward from the feed. To increase the directivity of a
radiating waveguide and its efciency as a reector feed,
for example, its aperture dimensions must be enlarged, for
the beamwidth of an aperture of width abl is proportion-
al to l/a radians.
This waveguide enlargement by a are characterizes
horns. The aperture elds of a horn are spherical waves
originating at the horn apex (Fig. 2). The path from the
horn apex to the aperture plane at a distance x from the
aperture center of a horn of slant length is
r=(( cos a)
2
x
2
)
1=2
- cos a
x
2
2 cos a
(1)
a
b
(a)
a
b
(b)
b
a
(c)
b
a
(d)
l
E
l
H
l
E
l
H
Figure 1. (a) Open-ended rectangular waveguide; (b) E-plane
sectoral horn; (c) H-plane sectoral horn; (d) pyramidal horn.
HORN ANTENNAS 2021
when x5 cos a. Thus the phase variation in radians
across the aperture for small are angles a is appro-
ximately kx
2
=(2), where k=2p/l is the propagation
constant. This quadratic phase variation increases with
increasing are angle, thus reducing directivity
increase due to the enlarged aperture dimension. It is
convenient to quantify aperture phase variation by the
parameter
s =
(1 cos a)
l
-
d
2
8l
; d5 (2)
which is the approximate difference in wavelengths be-
tween the distance from the apex to the edge (x =d/2) and
the center (x =0) of the aperture. The radiation patterns of
Figs. 3a, 3b [2] show the effect of increasing s on the E- and
H-plane radiation patterns of sectoral and pyramidal
horns. The main beam is broadened, the pattern nulls
are lled, and the sidelobe levels are raised over those for
an in-phase aperture eld (s =0). With large are angles
radiation from the extremities of the aperture can be so
out of phase with that from the center that the horn di-
rectivity decreases with increasing aperture width.
The adverse effects of the are can be compensated by a
lens in the aperture (Fig. 4a), but because that adds to the
weight and cost and because bandwidth limitations are
introduced by matching the lens surfaces to reduce reec-
tions, it is seldom done. Instead, a combination of aperture
width and are length in wavelengths is chosen that pro-
vides maximum axial directivity or minimum beamwidth.
This is an optimum horn design. To achieve higher di-
rectivity or narrower beamwidth for a given aperture
width, a longer horn is required.
d
x
x
0
l
Figure 2. Effect of horn are on the aperture eld phase of a
horn.
0.9
0.8
0.7
A
b
s
o
l
u
t
e

v
a
l
u
e

o
f

r
e
l
a
t
i
v
e

f
i
e
l
d

s
t
r
e
n
g
t
h
0.6
0.5
0.4
0.3
0.2
0.1
0 0
1/8
1/4
1/2
3/4
3/4
b
b
2
8
1
s =1
s =
Maximum phase deviation
in wavelengths
=
0 0.5 1.0 1.5 2.0 3.0 2.5
0
1/8
0 0.5 1.0 1.5 2.0 3.0 2.5
1.0
0.9
0.8
0.7
A
b
s
o
l
u
t
e

v
a
l
u
e

o
f

r
e
l
a
t
i
v
e

f
i
e
l
d

s
t
r
e
n
g
t
h
(b) (a)
0.6
0.5
0.4
0.3
0.2
0.1
0
1.0

a
2
8
s =
1/4
1/2
3/4
s = 1
b
sin
l
H
l
E
lH
lE
a
sin
Figure 3. Universal radiation patterns of sectoral and pyramidal horns ared in the (a) E and (b)
H planes. The parameter s =b
2
=8l
E
in (a) and a
2
=8l
H
in (b); 2ps/l is the maximum phase dif-
ference between the elds at the center and the edge of the aperture. (Copyright 1984, McGraw-
Hill, Inc. from Love [2].)
2022 HORN ANTENNAS
Sectoral horns (Figs. 1b, 1c) are rectangular wave-
guides ared in one dimension only. The incident wave-
guide mode becomes a radial cylindrical mode in the ared
region of the horn. Since radiation pattern beamwidths
are inversely proportional to aperture dimensions in
wavelengths, sectoral horns have beams that are narrow
in the plane containing the broad dimension. Such fan-
shaped beams may be useful for illuminating elongated
parabolic reectors or parabolic cylinder reectors.
A pyramidal horn (Fig. 1d) is ared in both waveguide
dimensions and so is more adaptable both as a reector
feed and on its own. The forward radiation pattern may be
calculated quite accurately from Kirchhoff diffraction the-
ory for all except small horns. The TE
01
rectangular wave-
guide mode yields an aperture eld uniform in one
dimension (in the E plane) and cosinusoidal in the other
(the H plane). A comparison of parts (a) and (b) of Fig. 3
shows that this results in a higher sidelobes in the E plane
and, for a square aperture, a narrower beam. Pyramidal
horns are relatively easily constructed, and for all except
small horns their axial gain can be predicted accurately.
Consequently, they are used as gain standards at micro-
wave frequencies; that is, they are used to experimentally
establish the gain of other microwave antennas by com-
paring their response to the same illuminating eld.
Most of the preceding remarks on open-ended rectan-
gular waveguides and pyramidal horns also apply to open-
ended circular waveguides and conical horns (Figs. 5a,
5b). For propagation of the lowest-order mode (TE
11
) only
in a circular waveguide, the interior diameter must be
0.59loao0.77l. This mode has a uniform aperture eld
in the E plane and a cosinusoidal distribution in the or-
thogonal H plane. This appears, modied by a quadratic
phase variation introduced by the are, in the aperture
eld of the horn. Consequently the E-plane radiation pat-
tern of the horn is narrower, but with higher sidelobes
than the H-plane pattern and the radiated beam is ellip-
tical in cross section. In addition, cross-polarized elds
appear in pattern lobes outside the principal planes.
2. HORN FEEDS FOR REFLECTORS
Many renements to horns arise from their use as efcient
feeds for parabolic reectors, particularly in satellite and
space communications and radio astronomy. The phase
center, where a horns far radiation eld appears to orig-
inate, must be placed at the focus of the reector (Fig. 4b).
This phase center is within the horn on the horn axis and
depends on the are angle and aperture distribution. For
both rectangular and conical horns the position of the
phase center is not the same in the E and H planes, or
planes containing the electric and magnetic eld vectors,
respectively. A phase center can be calculated from the
average of the positions of the E- and H-plane phase cen-
ters or determined from the position of the feed that max-
imizes the gain of the reector antenna.
For efcient aperture illumination the feed horn radi-
ation pattern should approximately match the shape of
the aperture, and illuminate it essentially uniformly and
with minimal spillover, or radiation missing the reector.
Pyramidal horns may seem suitable for rectangular aper-
tures because their beams are rectangular in cross section,
and conical horns may seem a natural choice for a circular
aperture, but efcient aperture illumination is not ob-
tained in either case, because their principal plane pat-
terns differ. Both horns have high E-plane pattern
sidelobes and low H-plane sidelobes. A dual (TE
11
/TM
11
)-
mode conical horn provides equal E- and H-plane beam-
widths and equally low sidelobes, and is an efcient feed
for a circular aperture over a narrow frequency band (see
Love [3], p. 195; Ref. 3 also contains reprints of most ear-
lier significant papers on horn antennas). A broadband
f
f
f
(a)
(b)
(c)
Figure 4. (a) Horn aperture eld phase correction by a lens;
(b) parabolic reector fed by a horn; (c) horn reector antenna
(f =focal length of the lens or reector).
HORN ANTENNAS 2023
solution achieves an axisymmetric beam with annular
corrugations on the interior surfaces of a conical horn
(Fig. 5c). These produce a horn aperture eld distribution
that is approximately cosinusoidal across the conical horn
aperture in all directions and hence an axisymmetric ra-
diation pattern with low sidelobes. Such corrugations in
the E-plane interior walls only of a pyramidal horn will
produce a cosinusoidal E-plane aperture distribution, and
consequently similar E-plane and H-plane radiation pat-
terns for a square horn aperture.
A feed for a small circular reector that is more easily
constructed than a corrugated conical horn but with a less
axisymmetric radiation pattern, is an open-ended circular
waveguide ringed by a recessed disk of approximately
quarter-wavelength-deep corrugations (Fig. 5d). These
corrugations suppress backradiation from the feed and
so improve the aperture illumination over that of a simple
open circular waveguide [3, pp. 181, 226]. Combined with
dual-mode excitation, this arrangement provides a simple
and efcient feed for a front-fed paraboloidal reector.
3. RADIATION FROM APERTURES
The far-eld radiation pattern of an aperture can be cal-
culated exactly from the Fourier transform of the tangen-
tial elds in the entire aperture plane. Either electric or
magnetic aperture elds may be used but for apertures in
space, a combination of the two gives the best results from
the usual assumption that aperture plane elds are con-
ned to the aperture and negligible outside it. This aper-
ture eld is assumed to be the undisturbed incident eld
from the waveguide. For apertures with dimensions larger
than several wavelengths, a further simplifying assump-
tion usually made is that the aperture electric and mag-
netic elds are related as in free space.
3.1. Rectangular Apertures
With the abovementioned assumptions, at a distance
much greater than the aperture dimensions, the radiated
electric eld intensity of a linearly polarized aperture eld
E
x
(x, y, 0) in the coordinates of Fig. 6a is
E(r; y; f) =A(r; y; f)
_
b=2
(b=2)
_
a=2
(a=2)
E
x
(x; y; 0)e
j(k
1
x k
2
y)
dx dy
(3)
Here
k
1
=k sin y cos f
k
2
=k sin y sinf
_
(4)
and
A(r; y; f) =j
e
jkr
2lr
(1 cos y)(
yy cos f
ff sin f) (5)
is a vector dening the angular behavior of the radiation
polarization for an aperture in space. For an aperture in a
conducting plane, it is more accurate to use
A(r; y; f) =j
e
jkr
lr
(
yy cos f
ff sin f cos y) (6)

which, since it is based on the aperture plane electric
elds only, fully satises the assumption of a vanishing
tangential eld in the aperture plane outside the aperture.
Consequently radiation elds of open-ended waveguides
and small horns can be calculated accurately from (3) with
(6) if they are mounted in a conducting plane. Clearly (5)
and (6) differ significantly only for large angles y off the
beam axis.
(b)
(c)
(d)
(a)
a
l
a
Figure 5. (a) Open-ended circular waveguide; (b) conical horn;
(c) corrugated horn; (d) circular waveguide with corrugated
ange.
2024 HORN ANTENNAS
If the aperture eld is separable in the aperture coor-
dinatesthat is, if in (3), E
x
(x, y, 0) =E
0
E
1
(x)E
2
(y), where
E
1
(x) and E
2
(y) are eld distributions normalized to E
0
,
the double integral is the product of two single integrals
E(r; y; f) =A(r; y; f)E
0
F
1
(k
1
)F
2
(k
2
) (7)
where
F
1
(k
1
) =
_
a=2
(a=2)
E
1
(x)e
jk
1
y
dx (8)
F
2
(k
2
) =
_
b=2
(b=2)
E
2
(y)e
jk
2
y
dy (9)
dene the radiation eld.
4. OPEN-ENDED WAVEGUIDES
4.1. Rectangular Waveguides
With the TE
10
waveguide mode the aperture eld
E
x
(x; y; 0) =E
0
cos
py
a
(10)
in (7) yields the following equations for (8) and (9):
F
1(k
1
)
=b
sin
k
1
b
2
_ _
k
1
b
2
(11)
F
2
(k
2
) =a
cos
k
2
a
2
_ _
p
2
(k
2
a)
2
_
_
_
_
_
_
_
_
(12)
This denes the radiation pattern in the forward hemi-
sphere p/2oyop/2, 0ofo2p. If the aperture is in
space, then (5) is used for A(r; y; f), but this is not an ac-
curate solution since the aperture dimensions are not
large. Rectangular waveguides mounted in conducting
planes use (6) for A(r; y; f) in (7), which then accurately
provides the far eld. The pattern has a single broad lobe
with no sidelobes. For large apertures plots of the normal-
ized E-plane (f=0) and H-plane (f=p/2) patterns of (7)
appear in Figs. 3a and 3b for those of a horn with no are
(s =0), but without the factor (1cos y)/2 from (5) or cos y
from (6).
4.2. Circular Waveguides
The dominant TE
11
mode eld in circular waveguide pro-
duces an aperture eld distribution, which in the aperture
coordinates r
/
; f
/
of Fig. 6b is
E(r
/
; f
/
) =E
0
rr
/
J
1
(k
c
r
/
)
k
c
r
/
cos f
/
ff
/
J
/
1
(k
c
r
/
) sin f
/
_ _
(13)
where J
1
is the Bessel function of the rst kind and order,
J
/
1
is its derivative with respect to its argument k
c
r
/
, and
k
c
a/2 =1.841 is its rst root; E
0
is the electric eld at the
aperture center (r
/
=0). Since (13) is not linearly polar-
ized, its use in (3) provides only part of the total radiated
far eld. The total eld
E(r; y; f) =jkaE
0
J
1
(1:841)
e
r
jkr
yy cos f
J
1
k
/
a
2
_ _
k
/
a
2
_
ffsin f cos y
J
/
1
k
/
a
2
_ _
1
k
/
a
3:682
_ _
2
_
_
(14)
in which k
/
=ksiny.
In the E and H planes (f=0 and p/2) the cross-polar-
ized elds cancel and the patterns shown in Fig. 14a are
similar to those of (11) and (12), respectively, but with
slightly broader beams and lower sidelobes for the same
aperture dimensions. As with rectangular waveguides,
open-ended circular waveguide apertures are insufcient-
ly large for (14) to represent all the radiated elds accu-
rately. In the principal planes (f=0, p/2), it can give a
reasonable approximation for the copolarized elds but
fails to accurately represent the cross-polarized eld
patterns in f=p/4. This is evident from a comparison of
x
x
y
y
z
z
r
r
a
2

b
a
Figure 6. Coordinates for radiation from
(a) rectangular and (b) circular apertures.
HORN ANTENNAS 2025
numerical results from approximate and exact solutions
[4, p. 233].
5. PYRAMIDAL AND SECTORAL HORNS
5.1. Radiation Patterns
A pyramidal horn fed by a rectangular waveguide sup-
porting the TE
10
mode has an incident electric eld in the
aperture of Fig. 6a that is approximately the mode distri-
bution modied by a quadratic phase variation in the two
aperture dimensions:
E
x
(x; y; 0) =E
0
cos
py
a
_ _
exp jk
x
2
2
E
y
2
2
H
_ _ _ _
(15)
With (15), Eq. (3) becomes
E(r; y; f) =A(r; y; f)E
0
I
1
(k
1
)I
2
(k
2
) (16)
where (5) is used for A(r; y; f) and
I
1
(k
1
) =
_
b=2
(b=2)
exp j
px
2
l
E
k
1
x
_ _ _ _
dx (17)
I
2
(k
2
) =
_
a=2
(a=2)
cos
py
a
_ _
exp j
py
2
l
H
k
2
y
_ _ _ _
dy (18)
The E-plane (f=0) and H-plane (f=p/2) radiation pat-
terns are, respectively
E
y
(r; y)
E
y
(r; 0)
=
1 cos y
2
I
1
(k sin y)
I
1
(0)
(19)
E
y
(r;y)
E
y
(r; 0)
=
1 cos y
2
I
2
(k sin y)
I
2
(0)
(20)
These integrals can be reduced to the Fresnel integrals
C(u) jS(u) =
_
u
0
e
j(p=2)t
2
dt (21)
which are tabulated and for which computer subroutines
are available. For example, we obtain
I
1
(k sin y)
I
1
(0)
=
e
2
j(p
E
=l) sin
2
y
C(u
2
) C(u
1
) j S(u
2
) S(u
1
) [ ]
C(u) jS(u)
(22)
with
u=
b
2l
E
_ (23)
u
2
1
= u
2
E
l
_
sin y (24)
Figure 3a shows plots of the magnitude of (22) for various
values of the E-plane are parameter s =b
2
=8l
E
, while
Fig. 3b shows corresponding plots of |I
2
(k siny)/I
2
(0)| for
the H-plane are parameter s =a
2
/8l
H
. For no are
(s =0) the patterns are those of a large open-ended rect-
angular waveguide supporting only the TE
10
mode. The
effect of the are is to broaden the mainbeam, raise the
sidelobes, and ll the pattern nulls. For larger values of s,
there is enhanced pattern beam broadening and eventu-
ally a splitting of the mainbeam on its axis.
These curves also represent the radiation patterns
of the E/H-plane sectoral horns of Figs. 1b and 1c. For
an E-plane sectoral horn (
H
o), the E-plane pattern is
given by (19) and the H-plane pattern approximately by
(12). For an H-plane sectoral horn (
E
o), the E-plane
pattern is given approximately by (11) and the H-plane
pattern by (20).
In comparing parts (a) and (b) of Fig. 3 it is evident that
E-plane beamwidths of a square aperture are narrower
than H-plane beamwidths. For horns of moderate are
angle and optimum horns the E-plane half-power beam-
width is 0.89l/b radians and the H-plane half-power
beamwidth is 1.22l/a radians. E-plane patterns have min-
imum sidelobes of 13.3 dB below peak power, while
H-plane pattern minimum sidelobes levels are 23.1 dB.
The universal patterns of Fig. 3 can also be used to
predict the approximate near-eld radiation patterns of
horns by including the quadratic phase error; which is a
rst-order effect of nite range r. This is done by including
exp j
p
rl
(x
2
y
2
)
_ _
(25)
in (15). Then the near-eld principal plane patterns of a
pyramidal horn are given by (17) and (18) with
E
;
H
re-
placed by
/
H
=
r
H
r
H
(26)
and
/
E
=
r
E
r
E
(27)
These near-eld effects are analogous to decreasing the
are length of a horn with a xed aperture width. The
mainbeam broadens, nulls are lled in, and sidelobes rise.
5.2. Limitations and Extensions
Results from (16) do not apply to small horns and are lim-
ited to the forward direction (yo901). They are most ac-
curate on and around the beam axis (y =0), becoming
progressively less accurate as y increases. The simplest
method for extending the analysis is by the uniform geo-
metric theory of diffraction [e.g., 3, p. 66], which provides
the edge-diffracted elds in the lateral and rear directions,
which receive no direct illumination from the aperture.
Only the edges normal to the plane of the pattern contrib-
ute significantly to the E-plane pattern, but the rear
H-plane pattern requires contributions from all four ap-
erture edges and so is difcult to calculate this way.
2026 HORN ANTENNAS
While the geometry of the pyramidal horn dees rigor-
ous analysis, numerical methods have been used with
some success for open waveguides and small horns. For
larger horns this approach becomes computationally in-
tensive, but some results from Liu et al. [5] are shown in
Fig. 7 and compared with measurements and approximate
computations. Their numerical computations and mea-
surements by Nye and Liang [6] of the aperture elds
show that higher-order modes need to be added to the
dominant mode eld of (15) and that the parabolic phase
approximation of (1) improves as the aperture size in-
creases.
5.3. Gain
Pyramidal horns are used as gain standards at microwave
frequencies because they can be accurately constructed
and their axial directive gain reliably predicted from a
relatively simple formula. The ratio of axial far-eld
power density to the average radiated power density
from (16) yields
G=G
0
R
E
(u)R
H
(v; w) (28)
where G
0
=32ab/(pl
2
) is the gain of an in-phase uniform
and cosinusoidal aperture distribution. The reduction of
this gain due to the phase variation introduced by the E-
plane are of the horn is
R
E
(u) =
C
2
(u) S
2
(u)
u
2
(29)
where the Fresnel integrals and their argument are de-
ned by (21) and (23). Similarly the gain reduction factor
due to the H-plane are of the horn is
R
H
(v; w) =
p
2
4
[C(v) C(w)]
2
[S(v) S(w)]
2
(v w)
2
(30)
25
(a)
(b)
15
5
5
15
G
a
i
n

(
d
B
)
25
35
25
15
5
5
15
G
a
i
n

(
d
B
)
25
35
45
55
180 150 120 90 60 30 0
(degrees)
E- plane
Moment method
Approximate
Measured
Moment method
Approximate
Measured
30 60 90 120 150 180
180 150 120 90 60 30 0
(degrees)
H- plane
30 60 90 120 150 180
Figure 7. Calculated and measured (a) E-plane and (b)
H-plane radiation patterns of a pyramidal horn of di-
mensions a=4.12l, b =3.06l,
E
=10:52l
E
=9:70l
(Copyright 1993, IEEE, from Liu et al. [5].)
HORN ANTENNAS 2027
where
v
w
=
a
2l
H
_
1
a
l
H
2
_
(31)
A plot or R
E
and R
H
in decibels as a function of the pa-
rameter 2d
2
=l, where d is the appropriate aperture di-
mension b or a and the slant length
E
or
H
, respectively,
is shown in Fig. 8. Calculation of the gain from (28) is ac-
curate to about 70.1dB for 22 dB standard gain pyrami-
dal horns: optimum horns with dimensions of at least 5l.
For 18-dB-gain horns, the accuracy is about 70.2 dB, and
for 15-dB horns, 70.5dB. Since optimum gain pyramidal
horns have an aperture efciency of approximately 50%,
the gain is approximately
G=0:5
4p
l
2
ab (32)
For an E-plane sectoral horn
H
oand R
H
(v, w)1 the
axial gain is then G
E
=G
0
R
E
(u), an inaccurate formula
because aperture dimension a is less than a wavelength. A
result that includes the fact that aperture electric and
magnetic elds are not related by free-space conditions
and that interaction occurs across the narrow aperture of
the horn is
G
E
=
16ab
l
2
(1 (lg=l)
R
E
(u
/
) exp
pa
l
1
l
l
g
_ _ _ _
(33)
where
u
/
=
b
2l
g
E
_ (34)
and
l
g
=
l
1 (l=2a)
2
_ (35)
is the guide wavelength. The accuracy of (33) is compara-
ble to that of (28) for the horns of similar b dimension.
The gain of an H-plane sectoral horn, obtained by let-
ting
E
o so that R
E
(u) 1, is G
H
=G
0
R
H
(v, w). It
probably is reasonably accurate, but there appears to be
no experimental evidence available to verify it.
The near-eld gain of pyramidal and sectoral horns can
be calculated from the preceding expressions by replacing
E
and
H
by (26) and (27), respectively.
6. CONICAL HORNS
The aperture eld of a conical horn fed by a circular wave-
guide supporting the TE
11
mode is approximately
E(r
/
; f
/
) exp
jkr
/2
2
_ _
(36)
where E(r
/
; f
/
) is given by (13) and is the slant length of
the horn. Numerical calculation of the radiation patterns
is necessary. In the example of Fig. 9 [7] with a are angle
a =51 and aperture width a=4l, the E-plane (f=0)
1.0
2.5
2.0
G
a
i
n

r
e
d
u
c
t
i
o
n

(
D
B
)
1.5
1.0
R
H
R
E
0.5
0.0
2.0 3.0 4.0
2 d
2
/l
5.0 6.0 7.0 8.0
Figure 8. E- and H-plane are and near-eld gain reduction fac-
tors R
E
and R
H
of pyramidal and sectoral horns in decibels.
(Copyright 1981, IEE, from Jull [11].)
50 25 0
deg
40
30
20
dB
10
50 25
0
Figure 9. Copolar and cross-polar radiation patterns for a con-
ical horn with dimensions a=4l; =23l - - - - E-plane,
H-plane cross-polarization. (Copyright 1994, IEE,
from Olver et al. [7].)
2028 HORN ANTENNAS
pattern is narrower than the H-plane (f=p/2) pattern as
in square rectangular horns. The cross-polar (f=p/4) ra-
diation pattern peak level is 18.7 dB relative to the copo-
lar pattern peak levels, a level typical of conical horn
apertures large than about 2l. Smaller conical horns can
have more axisymmetric patterns. E- and H-plane pat-
terns have equal beamwidths for an aperture diameter a
=0.96l, and cross-polarized elds cancel for a=1.15l.
This makes small conical horns efcient as reector feeds
and as array elements with high polarization purity.
Larger conical horns are similar to rectangular horns
in their lack of axial pattern symmetry. Optimum gain
conical horns have an aperture efciency of about 54% and
half-power beamwidths in the E and H planes of 1.05l/a
and 1.22l/a radians, respectively, for aperture diameters
of more than a few wavelengths.
7. MULTIMODE AND CORRUGATED HORNS
Lack of axisymmetric radiation patterns make rectangu-
lar and conical horns inefcient reector feeds. Conical
horns also have unacceptably high cross-polarization lev-
els if used as reector feeds in a system with dual polar-
ization. Multimode and corrugated horns were developed
largely to overcome these deciencies. In a dual-mode
horn in [3, p. 195], this is done by exciting the TM
11
mode, which propagates for circular waveguide diameters
a41.22l, in addition to the TE
11
mode, which propagates
for a40.59l. The electric eld conguration of these
modes in a waveguide cross section is shown in Figs. 10a
and 10b. Added in phase and in the right proportion,
cross-polarized and aperture perimeter elds cancel, while
the copolar elds around the aperture center add, yielding
the aperture eld conguration of Fig. 10c. These mixed
mode elds are linearly polarized and taper approximate-
ly cosinusoidally radially across the aperture. This yields
the essentially linearly polarized and axisymmetric radi-
ation patterns desired.
Partial conversion of TE
11
to TM
11
elds can be effected
by a step discontinuity in the circular waveguide feed, as
in Fig. 10d, or by a circular iris or dielectric ring in the
horn. The TM
11
/TE
11
amplitude ratio depends on the ratio
of waveguide diameters, and the relative phase of the
modes depends on the length of larger-diameter circular
waveguide and the horn. This dependence limits the fre-
quency bandwidth of the horn to about 5%. A multimode
square pyramidal horn has similarly low sidelobe levels in
its E- and H-plane radiation patterns because of an es-
sentially cosinusoidal aperture distribution in both E and
H planes [2]. This is achieved by excitation of a hybrid
TE
21
/TM
21
mode either by an E-plane step discontinuity
or by changes in the E-plane are. With their bandwidth
very limited, dual-mode horns have largely been replaced
by corrugated horns in dual-polarization systems, except
where a lack of space may give an advantage to a thin-
walled horn.
Corrugated horns have aperture elds similar to those
of Fig. 10c and consequently similar radiation patterns,
but without the frequency bandwidth limitations of the
multimode horn. This is achieved by introducing annular
corrugations to the interior walls of a conical horn. There
must be sufcient corrugations per wavelength (at least 3)
that the annular electric eld E
f
is essentially zero on the
interior walls. The corrugations make the annular mag-
netic eld H
f
also vanish. This requires corrugation
depths such that short circuits at the bottom of the
grooves appear as open circuits at the top, suppressing
axial current ow on the interior walls of the horn. This
groove depth is l/4 on a plane corrugated surface or a
curved surface of large radius. For a curved surface of
smaller radius, such as near the throat of the horn, the
slot depths need to be increased; for example, for a surface
radius of 2l, the depth required is 0.3l. Usually slots are
normal to the conical surface in wide-are horns but are
often perpendicular to the horn axis with small ares. To
provide a gradual transition from the TE
11
mode in the
waveguide to a hybrid HE
11
mode in the aperture, the
depth of the rst corrugation in the throat should be about
0.5l so that the surface there resembles that of a conduct-
ing cone interior. Propagation in corrugated conical horns
can be accurately calculated numerically by mode match-
ing techniques. The aperture eld is approximately
E
x
(r
/
) =AJ
0
(k
c
r
/
) exp
jkr
/2
2
_ _
(37)
where k
c
a/2 is 2.405, the rst zero order Bessel function
J
0
; is the slant length of the horn; and A is a constant.
This aperture eld is similar to that of Fig. 10c, and the
resulting E and H patterns are similarly equal down to
about 25 dB. Some universal patterns are shown in
Fig. 11. Cross-polarization elds are also about 30dB from
the axial values, but now over a bandwidth of 21 or more.
Broadband axisymmetric patterns with low cross-po-
larization make corrugated horns particularly attractive
as feeds for reectors. Low cross-polarization allows the
use of dual polarization to double the capacity of the sys-
tem. Another notable feature for this application is that
the position of the E- and H-plane pattern phase centers
coincide. Figure 12 shows the distance of the phase center
(d)
(a)
TE
11
(b)
TM
11
(c)
0.85 TE
11
+ 0.15 TM
11
Figure 10. Excitation of axisymmetric linearly polarized aper-
ture elds in a stepped conical horn. (Copyright 1984, McGraw-
Hill, Inc. from Love [2].)
HORN ANTENNAS 2029
from the horn apex, divided by the slant length, of small
are angle conical [8] and corrugated [9] horns for values
of the phase parameter s given by (2). For a conical horn
the E-plane phase center is significantly farther from the
aperture than the H-plane phase center. Thus, if a conical
horn is used to feed a parabolic reector, the best
location for the feed is approximately midway between
the E- and H-plane phase centers. With a corrugated
horn such a compromise is not required, so it is inhe-
rently more efcient.
Corrugated horns may have wide are angles, and their
aperture size for optimum gain decreases correspondingly.
For example, with a semiare angle of 201, the optimum
aperture diameter is about 8l, whereas for a semiare
angle of 701 it is 2l. Wide-are corrugated horns are some-
times called scalar horns because of their low cross-po-
larization levels.
For radio astronomy telescope feeds and other space-
science applications, efcient corrugated horns have been
made by electroforming techniques for frequencies up to
640 GHz. Their axisymmetric radiation patterns with very
low sidelobe levels resemble Gaussian beams, which is of-
ten essential at submillimeter wavelengths.
8. PROFILE HORNS
Most corrugated horns are conical with a constant are
angle. Figure 13 shows a prole conical horn in which the
are angle varies as on a sine-squared or similar curve
along its length. This arrangement provides a horn short-
er than a conical corrugated horn of similar beamwidth,
with a better impedance match due to the curved prole at
the throat and an essentially in-phase aperture eld dis-
tribution due to the prole at the aperture. Consequently
the aperture efciency is higher than that of conical cor-
rugated horns. The phase center of the horn is near the
aperture center and remains nearly xed over a wide fre-
quency band. Radiation patterns of a short prole horn
similar to that of Fig. 13, but with hyperbolic prole
curves, are shown in Fig. 14 [10]. A Gaussian prole curve
has also been used. All produce patterns similar to those of
a Gaussian beam, such as is radiated from the end of an
optical ber supporting the HE
11
mode. The performance
of this small horn as a feed seems close to ideal, but larger-
prole horns may exhibit higher sidelobe levels due to ex-
citation of the HE
12
mode at the aperture.
9. HORN IMPEDANCE
Antennas must be well matched to their transmission
lines to ensure a low level of reected power. In micro-
wave communications systems levels below 30 dB are
commonly required.
0
10
20
30
R
e
l
a
t
i
v
e

p
o
w
e
r
,

d
B
40
0 1 2 3 4 5 6
a sin
7 8 9 10 11 12
S=0
0.1
0.2
0.3
0.4
0.5
Figure 11. Universal patterns of small-are-angle corrugated

horns as a function of the parameter s =a
2
/8l. (Copyright 1984,
McGraw-Hill, Inc. from Love [2].)
1.0
0.8
0.6
0.4
0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
S=
a
2
8l
lph
l

l
ph
a
l
Figure 12. Normalized distance of the phase center from the
apex of conical (- - - - E plane, H plane and corrugated ()
horns). (Data from Milligan [8] and Thomas [9].).
L
a
4
2
Figure 13. A prole corrugated horn. (Copyright 1994, IEE, from

Olver et al. [7].)
2030 HORN ANTENNAS
The impedance behavior of a horn depends on the mis-
match at the waveguide/horn junction and at its aperture.
For an E-plane sectoral horn, reections from these dis-
continuities are comparable in magnitude, and since they
interfere, the total reection coefcient oscillates with fre-
quency and the input voltage standing-wave ratio (VSWR)
may vary from 1.05 at high frequencies to 1.5 at the lowest
frequency. With E-plane sectoral horns aperture reection
is much stronger than junction reection, so their VSWRs
increase almost monotonically with decreasing frequency.
An inductive iris in the waveguide near the E-plane horn
junction can match its discontinuity. A capacitive iris may
be similarly used for an H-plane sectoral horn. Aperture
reections in these horns may be matched with dielectric
covers.
Pyramidal horns of sufcient size and optimum design
tend to be inherently well matched to their waveguide
feeds because the E/H-plane aperture and are disconti-
nuities partially cancel. For example, a 22-dB-gain horn
has a VSWR of about 1.04 and an 18 dB horn a VSWR of
less than 1.1.
Conical horns fed by circular waveguides supporting
the dominant TE
11
mode have an impedance behavior
similar to that of pyramidal horns of comparable size fed
by rectangular waveguides. The waveguide/horn disconti-
nuities of both horns may be matched by an iris placed in
the waveguide near the junction. A broader bandwidth
match is provided by a curved transition between the in-
terior walls of the waveguide and the horn. Broadband
reduction of aperture reection may be similarly reduced
by a curved surface of a few wavelengths radius. Such
aperture-matched horns also have lower sidelobe levels
and less backradiation in their E-plane patterns than do
conventional pyramidal and conical horns. Their H-plane
are patterns are affected little by such aperture matching
because the electric eld vanishes at the relevant edges.
For dual-mode and corrugated horns there are also
negligible elds at the aperture edges and hence little dif-
fraction there. Corrugated horns with initial groove
depths near the throat of about a half-wavelength and
which gradually decrease to a quarter-wavelength near
the aperture, as in Fig. 13, are well matched at both throat
and aperture. For most well-designed corrugated horns a
VSWR of less than 1.25 is possible over a frequency range
of about 1.51. Dual-mode horns using a step discontinu-
ity as in Fig. 10d may have a VSWR of 1.21.4. If an iris is
required for a match, the frequency bandwidth will, of
course, be limited. Conical and pyramidal horns using
are angle changes to generate the higher-order modes
can have VSWRs less than 1.03 and require no matching
devices.
BIBLIOGRAPHY
1. J. F. Ramsay, Microwave antenna and waveguide techniques
before 1900, Proc. IRE 46:405415 (1958).
2. A. W. Love, Horn antennas, in R. C. Johnson and H. Jasik,
eds., Antenna Engineering Handbook, 2nd ed., McGraw-Hill,
New York, 1984, Chap. 15.
3. A. W. Love, ed., Electromagnetic Horn Antennas, IEEE Press,
Piscataway, NJ, 1976.
4. R. E. Collin, Antennas and Radiowave Propagation, McGraw-
Hill, New York, 1985.
5. K. Liu, C. A. Balanis, C. R. Birtcher, and G. C. Barber, Anal-
ysis of pyramidal horn antennas using moment methods,
IEEE Trans. Anten. Propag. 41:13791389 (1993).
6. J. F. Nye and W. Liang, Theory and measurement of the eld
of a pyramidal horn, IEEE Trans. Anten. Propag. 44:1488
1498 (1996).
7. A. D. Olver, P. J. B. Clarricoats, A. A. Kishk, and L. Shafai,
Microwave Horns and Feeds, IEE Electromagnetic Waves Se-
ries, Vol. 39, IEE, London, 1994.
8. T. Milligan, Modern Antenna Design, McGraw-Hill, New
York, 1985, Chap. 7.
9. B. MacA. Thomas, Design of corrugated horns, IEEE Trans.
Anten. Propag. 26:367372 (1978).
10. R. Gonzalo, J. Teniente, and C. del Rio, Very short and
efcient feeder design from monomode waveguide, IEEE
(b) (a)
0
20
40
60
90 45 0
45 90
0
20
40
dB
60
90 45 0
45 90
Figure 14. (a) Far eld radiation patterns of TE
11
mode and (b) radiation patterns of a prole
corrugated horn of aperture a=15.8mm and length L=26.7mm at 30GHz (- - - - - E-plane, ____
H-plane yycross-polarization). (Copyright 1997, IEEE, from Gonzalo et al. [10].)
HORN ANTENNAS 2031
Antennas and Propagation Soc. Int. Symp. Digest, Montreal,
1997, pp. 468470.
11. E. V. Jull, Aperture Antennas and Diffraction Theory, IEE
Electromagnetic Waves Series, Vol. 10, IEE, London, 1981.
HTS FILM GROWTH
KOOKRIN CHAR
VLADIMIR MATIJASEVIC
Conductus, Inc.
Thin lms of superconductors are of interest primarily for
electronics applications. High-performance superconduct-
ing thin lms are essential, for example, for high-frequen-
cy passive devices and Josephson junction circuits. Thin
lms are also important for fundamental studies of super-
conducting materials, where specially designed experi-
ments can take advantage of the thin-layer geometry
and the capability of layering various combinations of ma-
terials, as well as lithographically dening ne features in
them. Additionally, there is currently growing interest in
using thin-lm deposition technology for deposition of
thick lms on polycrystalline substrates for power appli-
cations.
In the decade since the discovery of high-temperature
superconductivity (HTS) in cuprate compounds, thin-lm
materials have played an important role in the evolution
of this eld. The discovery of HTS has led to a rapid de-
velopment of many different thin lm deposition tech-
niques. HTS thin lms are now routinely made in
hundreds of laboratories around the world, and thin-lm
research is actively pursued in areas such as new super-
conducting materials and epitaxial oxide heterostruc-
tures. More recently, advances in deposition processes
have brought about the synthesis of larger-area HTS
lms, making fabrication less expensive and allowing for
high-throughput manufacturing of single-superconduct-
ing-layer lms.
In this article, we review the physical vapor deposition
(PVD) techniques most frequently used for HTS thin-lm
synthesis. Chemical vapor deposition techniques are re-
viewed in another article of this encyclopedia. Here, we
review only the methods and issues that are relevant for
lm deposition. This does not cover the details of the lm
growth processes, nor the physical properties of HTS lms
and their applications. The reader is referred to other ref-
erences for a more complete discussion of HTS thin lms
and their applications [1,2].
1. MATERIALS
The HTS material of choice for applications is still YBa
2-
Cu
3
O
7
(YBCO), the rst cuprate discovered to supercon-
duct above 77 K. There are several compelling reasons
why this material is still the most suitable for many of the
applications. Some of the fundamental ones are its smaller
conduction anisotropy, higher superconducting critical
currents in a magnetic eld, and greater chemical stabil-
ity than in other HTS compounds. However, the most im-
portant reason is the ease of fabricating high-quality,
single-phase YBCO thin lms.
In this review, YBCO will be considered the canonical
high-T
c
superconductor, and the discussion will be limited
to this example, partly for reasons of conciseness and
partly because of the prominence of this compound in
the HTS community. Additionally, unless otherwise men-
tioned, one is generally considering growth of c-axis-ori-
ented lms, i.e. where the CuO
2
layers are parallel to the
substrate.
Table 1 reviews the HTS compounds and methods used
to date to prepare these thin lms. For a review of mate-
rials, we refer the reader to Refs. 2 and 3. There has been a
substantial effort related to BiSrCaCuO thin lms
(2212 and 2223 phases) by a variety of techniques, partic-
ularly in Japan. Some industrial work in the United
States has also focused on TlBaCaCuO (mostly
2212), primarily for passive electronics applications. One
of the obvious reasons for looking at these other com-
pounds is their higher critical temperatures. Neverthe-
less, the advantages of YBCO have been hard to surpass,
particularly at temperatures below 65 K. Substantial thin-
lm work has also been devoted to LaSrCuO, NdCe
CuO, HgBaCaCuO, and innite-layer compounds,
mostly driven by academic interest in basic properties of
these HTS materials.
2. THERMODYNAMIC ISSUES
Film growth is inherently a nonequilibrium process. Nev-
ertheless, thermodynamic stability is important as a driv-
ing force for the reactions taking place during growth.
HTS compounds possess large unit cells that can have ad-
ditional complications, such as a wealth of metal atom de-
fects and oxygen nonstoichiometries. These complex
structures require temperatures for formation close to
their melting points, typically (0.80.9) T
m
, much higher
Table 1. HTS Materials Systems, Methods Used to
Fabricate Thin Films, and Applications of the Thin Films
Material T
c
(K)
Thin-Film
Deposition
Technique Applications
Y(R)BaCuO 123 8595 In situ: sputtering,
PLD, evapora-
tion, MOCVD;
ex situ
RF devices,
Josephson
Junctions
BiSrCaCuO
2201, 2212, 2223
10115 In situ: sputtering,
PLD, evaporation
Josephson
junctions
TlBaCaCuO
1201, 1212, 1223,
2201, 2212, 2223
20125 1212: in situ
sputtering;
ex situ
RF devices
HgBaCaCuO 95133 Ex situ
1201, 1212, 1223
LaSr(Ba)CuO 2040 In situ
NdCeCuO 2030 In situ
2032 HTS FILM GROWTH
than is usual for growth of epitaxial metals and semicon-
ductors. The intricacies of lm growth and the thermody-
namics of phases produced are certainly not sufciently
well understood at present. Below, we summarize some of
the key thermodynamic issues as they are currently ac-
cepted.
2.1. Oxygen Stability
HTS compounds are oxides and, as such, require the pres-
ence of oxygen during synthesis. Although it is technically
possible to deliver oxygen through the substrate, in prac-
tice, oxygen is delivered as a gas impinging on the growth
surface together with the cation species. Figure 1 shows
the pressuretemperature thermodynamic stability dia-
gram for the bulk YBCO compound. One can see from the
diagram that a certain minimum pressure must be main-
tained for the stability of YBCO. Below that oxygen pres-
sure, depicted by the line labeled d
1
, the YBCO compound
is not stable and will decompose. For the typical temper-
atures during lm formation, this translates into a min-
imum oxygen pressure of about 100mPa (1mtorr). At
higher oxygen pressures, there is another decomposition
line for YBCO, labeled d
2
, seldom not reached during in
situ growth of thin lms. During cooldown of lms to room
temperature, this second decomposition is seldom ob-
served because it is strongly limited by diffusion kinetics.
However, the stability diagram also implies that the
YBCO compound is metastable at room temperature and
pressure. In fact, most HTS compounds are believed not to
be thermodynamically stable at lower temperatures. Ref-
erences 4 and 5 have a more detailed discussion of the
YBCO stability lines. Hammond and Bormann [6] argued
that growth of thin lms is optimal close to the high-tem-
perature thermodynamic decomposition line of the com-
pound. In fact, the empirical data conrm this, as can be
seen also in Fig. 1, where successful lm growth regions
are denoted. For processes that possess a more reactive
form of oxygen, such as atomic oxygen or ozone, the equiv-
alent decomposition line is shifted compared with that in
Fig. 1. Such activated oxygen species have been proven
helpful for growth of YBCO when the total pressure is
lower than about 1 Pa (10mtorr).
2.2. Compositional Phase Diagram
The HTS compounds contain typically three to ve differ-
ent metal species in addition to oxygen. Some of these
materials are line compounds (and this is presumed to be
the case for YBCO) and, as such, do not accept a solid so-
lution of atoms in their chemical formula. This means that
one will always be synthesizing a number of phases in
addition to the desired HTS material. For YBCO, the
Gibbs phase rule implies two other compounds as impu-
rity phases. Figure 2 shows the present understanding of
the ternary phase diagram for YBaCu oxides at the low
oxygen pressures relevant for in situ thin-lm growth
[7,8]. The corners of the triangles dene the phases which
will be present for any composition in that triangle.
There are some caveats to the description given above.
First, as already stated, lm growth is not an equilibrium
process, and therefore metastable phases could be formed
during synthesis. Second, the HTS material that is formed
during lm growth may not be, and probably is not, ex-
actly like the bulk material. This has now been well doc-
umented in the literature for YBCO lms. In fact, it is also
fairly well established that YBCO lm growth does not
match exactly the thermodynamic phase diagram shown
in Fig. 2. In particular, the Y
2
BaCuO
5
phase has not been
observed in in situgrown lms. On the other hand, CuO
and Y
2
O
3
second phases are commonly observed in YBCO
lms prepared close to 1:2:3 stoichiometry and are not
predicted by the phase diagram.
0
1
2
3
4
5
6
7
8
9
0.7
1000
d
2
d
1
m
1
900 800 700 600 500
0.8 0.9 1.0
T (C)
1000/T (K
1
)
1.1 1.2 1.3 1.4
6
5
4
3
2
1
0
1
2
l
o
g
[
p
O
2
(
M
P
a
)
]
l
o
g
[
p
O
2
(
m
t
o
r
r
)
]
Sputtering
Evaporation
with activated O
Sputtering
Evaporation
Laser
ablation
Figure 1. Thermodynamic stability diagram for YBCO in oxygen
pressure and temperature. The shaded elliptical areas indicate
regimes where in situ lm growth for YBCO has been reported
successful. The rectangles show ex situ lm growth regimes.
Y
2
O
3
YBa
2
Cu
3
O
y
BaCu
2
O
2
Cu
2
Y
2
O
5
Cu
2
O
BaO
YBa
3
Cu
2
O
y
Y
2
BaCuO
5
Figure 2. Thermodynamic ternary phase diagram for YBaCu
oxides at temperatures below 8501C and oxygen pressures below
one atmosphere.
HTS FILM GROWTH 2033
3. EPITAXY
Due to the significant anisotropy of the HTS materials, the
easy direction for electrical current ow is in the ab plane,
along the CuO
2
planes. For devices which require current
to ow parallel to the substrate, the growth direction of
the HTS lms has to be in the c-axis direction. In order to
achieve c-axis-oriented growth, good epitaxial growth of
all grains is necessary. Furthermore, a high-angle, in-
plane grain boundary of two c-axis-oriented grains has
been found to behave as a superconducting weak link.
This implies that one needs to align all grains in the plane
of the substrate as well as out of the plane. The require-
ments of in-plane epitaxy are very well illustrated in the
case of yttria-stabilized zirconia (YSZ) substrates where
the YBCO lms grow in the c direction, but the c axis
grains can have two major different orientation in the ab
plane, 451 apart [15]. In the case of applications that re-
quire the current ow to be perpendicular to the sub-
strate, a-axis growth has been implemented. Many
multilayer Josephson devices have been tried in this
way; a-axis growth is typically initiated with a lower sub-
strate temperature, which in turn reduces the mobility of
ions necessary for the c-axis growth. Because of the lower
growth temperature, lms are formed that are structur-
ally less perfect. To overcome this problem, many workers
have used a technique of starting growth at a lower tem-
perature for a-axis nucleation, making a template, and
then raising the temperature for the remainder of lm
deposition [16].
3.1. Substrates
The most frequently used substrate for fundamental re-
search is SrTiO
3
because of its good thermal and lattice
constant match with most HTS materials. However, high
dielectric loss, small wafer size, and high cost of SrTiO
3
render it unfavorable for technological applications. La-
AlO
3
has emerged as an alternative to SrTiO
3
as a low-loss
perovskite substrate, although its critical dielectric prop-
erties are not under complete control. MgO substrates, on
the other hand, have an even lower loss for microwave
applications. However, the reproducibility of the MgO sur-
face quality has been a problem in the earlier days, but
more recently, the surface quality has improved. For ap-
plications where very low loss is required, MgO is the best
HTS substrate at present. MgO and LaAlO
3
are commer-
cially available in wafer sizes larger than 5 cm (2in.).
More conventional substrates such as Si and sapphire
have been used and found to be incompatible with thick
HTS thin lms, mainly due to microcracking in the HTS
lms resulting from the thermal expansion mismatch be-
tween the substrates and the HTS materials. The maxi-
mum thickness of YBCO one can grow without significant
microcracking is about 50 nm on Si and about 600nm on
sapphire. Table 2 lists the substrates most commonly used
today for growth of HTS lms. References 17 and 18 more
extensively discuss HTS-compatible substrates and buffer
layers.
In most cases, excellent epitaxial alignment can be
achieved by depositing oxide buffer layers, such as CeO
2
,
before HTS growth. The CeO
2
buffer layer has been found
to suppress the unwanted a-axis nucleation for the c-axis
growth on substrates such as SrTiO
3
and LaAlO
3
. On sub-
strates such as YSZ and r-plane sapphire, the CeO
2
buffer
layer promotes in-plane alignment of each c-axis grain,
thereby eliminating potential high-angle grain bound-
aries. Infact, by using an appropriate buffer layer and a
seed layer, one can create a process to produce 451-angle
grain boundaries in a controlled fashion. On Si substrates,
the YSZ buffer layer was found to grow epitaxially and to
be very effective in preventing chemical reaction between
HTS materials and Si.
3.2. Ion-Beam-Assisted Deposition
More recently, a new technique, ion-beam-assisted depo-
sition (IBAD), has been implemented to grow HTS layers
on polycrystalline substrates. By bombarding the growing
lm surface with an ion beam incident at a specific angle
to the substrate, one is able to achieve partial in-plane
crystalline alignment of the deposited material where the
substrate has none. Iijima et al. [19] rst utilized this
technique to grow in plane textured zirconia buffer layers
on substrates made of polycrystalline Ni-based alloys.
YBCO was then deposited by PLD. Due to the absence of
high-angle grain boundaries in YBCO, the critical cur-
rents attained are much higher than without the IBAD
process. Other groups, including the one at Los Alamos
National Laboratory, has followed on this work and
achieved critical currents over 10
6
A/cm
2
at 77 K [20].
More recently a group at Stanford University has demon-
strated very good IBAD results with 10nm thick MgO
lms [21]. While more development is needed to establish
whether the IBAD process is a viable one for large-scale
applications, it certainly opens up a whole new range of
substrate materials for HTS lm growth.
4. FILM GROWTH METHODS
Most of the methods currently employed for fabrication of
HTS thin lms involve formation of the cuprate crystal
structure during lm deposition. These are usually called
in situ processes, and they will represent most cases cov-
ered in this review. Chemical vapor deposition processes,
which are also in situ growth, are covered in another
Table 2. Substrates and Buffer Layer for HTS Thin Films
Substrate Orientation Buffer Layer
SrTiO
3
(10 0) None
Ceo
2
LaAlO
3
(10 0) CeO
2
Pseudocubic None
Mgo (10 0) None
SrTiO
3
YSZ (10 0) CeO
2
Y
2
O
3
Al
2
O
3
(1 1 02) CeO
2
MgO
Si (10 0) YSZ/Y
2
O
3
YSZ/CeO
2
chapter of this volume. Historically, the processes that
were developed rst, required a postdeposition anneal, or
simply postanneal, in order to crystallize the material.
Such methods are now less frequently utilized, since they
are not well suited for fabrication of multilayer structures
and are thus technologically more limiting. However, they
are still used for compounds such as TlBaCaCuO or
HgBaCaCuO, where the in situ processes are often im-
practical because of the required high vapor pressures of
Hg and Tl. Finally, there are other processes such as sol-
gel, plasma spray deposition, and liquid-phase epitaxy
that are generally used for fabrication of thick lms (thick-
er than 1 mm); these also will not be covered in this brief
review.
4.1. Postanneal Growth Methods
In the postanneal, or ex situ, growth process, the metallic
elements are deposited on a substrate in the correct com-
position as an amorphous or multilayer lm, usually in
compound form with oxygen and possibly uorine. Subse-
quent annealing in air or oxygen at a high temperature,
typically Z8501C for YBCO, forms an epitaxial lm of the
desired HTS phase [9]. A very commonly used postanneal
method for growth of YBCO lms is achieved by using
BaF
2
in the precursor lm. Water is then required in the
annealing step in order to eliminate the uorine and start
the HTS growth process. Since the critical growth step
takes place separately from deposition, the actual tech-
nique used to deposit the layers is of secondary impor-
tance. Sputtering and evaporation are typically used.
Ex situ methods have initially focused on synthesis un-
der atmospheric oxygen pressure and correspondingly
high temperatures. This region of phase space is depicted
in Fig. 1 with the rectangle in the upper left corner. Later
work has shown that lower-temperature ex situ growth
can also be achieved if the oxygen pressure is reduced at
the same time (see also the lower rectangle in Fig. 1). Such
lms grown under lower oxygen pressures were observed
to have properties closer to the in situ grown lms [10,11].
4.2. In situ Growth Methods
Dominant methods is use today for physical vapor depo-
sition of HTS are sputtering and laser ablation [also re-
ferred to as pulsed-laser deposition (PLD)]. Both of these
methods are most commonly done from a single target and
as such, became rather popular mostly because they are
relatively simple to implement and fairly reproducible in
the lms they produce. In addition, targets for PLD are
relatively inexpensive and easy to fabricate, and hence,
the technique is well suited for investigation of many dif-
ferent materials. Less widespread today, but currently
growing in popularity, are coevaporation and molecular
beam epitaxy. After the initial slow start in contending
with a sufciently oxidizing environment in high vacuum,
evaporation methods have now emerged as an established
way to grow high-quality HTS lms.
The following is a list of in situ physical vapor deposi-
tion techniques used for deposition of HTS compounds,
which will be covered in this review:
1. Sputtering
On-axis magnetron
Off-axis magnetron
Inverted cylindrical magnetron
Ion-beam sputtering
2. Laser ablation
On-axis
Off-axis
Laser-MBE
3. Evaporation
Flash evaporation
Reactive coevaporation
ReactiveMBE
For more discussion on various PVD techniques, we also
refer the reader to Refs. 1214.
5. SPUTTERING
Sputtering is a very commonly used technique for metal
deposition in semiconductor and magnetic storage indus-
tries. As such, it was applied early on to HTS materials.
The rst event of significance was the use of a single com-
posite target, which became commonplace for HTS mate-
rials because of the inherent difculty of reproducing the
metal atom stoichiometry. The major complication com-
pared to more conventional sputtering of metals is the en-
ergetic negative-ion (oxygen) bombardment of the
substrate, due to the ionic nature of the oxide target.
This causes resputtering and degradation of the sample.
This has been recognized even before the advent of HTS
and has been fully investigated by Rossnagel and Cuomo
[22]. Two types of solution emerged: (1) Using a higher gas
pressure in order to thermalize the energetic species [23]
and (2) off-axis sputtering, where one eliminates the high-
energy particles coming directly from the target [24,25].
Figure 3 shows a schematic of the two processes. Other
variations include on-axis unbalanced-magnetron sputter-
ing [26] and inverted cylindrical magnetron (ICM) sput-
tering [27].
(a) (b)
Sputtering
target
Substrate
Figure 3. Schematic for the (a) on-axis and (b) off-axis sputter
deposition processes.
The energies of ejected atoms from the sputtered tar-
gets are in the range of tens of electron volts, and they get
scattered by the background Ar and O
2
gas. The angular
distribution of each cation species is different, and the
sweet spot for the proper cation composition is relatively
small. In order to cover a 5-cm (2-in.) Wafer uniformly
from a single 5-cm target in an off-axis geometry, one has
to resort to some sort of scanning method, such as rotation
of the wafer. The deposition rate is very low, not only be-
cause the deposition rate for oxides is much lower than for
metals but also because of the off-axis geometry. It takes
several hours to deposit a few hundred nanometers of lm.
Larger targets and a number of sputter guns depositing
simultaneously have been used at several laboratories to
increase the deposition rate. When all the deposition con-
ditions are optimum, sputtering has demonstrated a ca-
pability to produce YBCO lms of excellent crystallinity
and surface condition. However, the deposition conditions
have to be changed as the targets erode, because the
changing target surface geometry results in a different
plasma distribution. In addition, the substrate heating
method for off-axis sputtering is not as simple as in the
case of the off-axis laser ablation or evaporation, which
will be described later.
6. LASER ABLATION
Laser ablation, or pulsed-laser deposition (PLD), is a rel-
atively new technique that gained much popularity be-
cause it is ideally suited for deposition at a high oxygen
pressure. The relative ease of this technique in depositing
multicomponent oxides (and nitrides) has made it espe-
cially effective in exploring new materials for HTS elec-
tronics, such as epitaxial dielectrics or barrier layers. A
short-wavelength (170260-nm) excimer laser is focused
onto a rotating target of the material to be deposited. Un-
der the energy of the laser beam (0.12J per pulse), the
matter emitted from the target forms a plume that carries
it to the substrate at supersonic velocities. In general, a
higher gas pressure is required during laser ablation, due
to the very high energy of the vaporized material from the
surface of the target. The laser plume glows brightly from
the target, and deposition is usually done near the end of
the plume, about 5cm away (see Fig. 4). Again, as in the
case of sputtering, the various species scatter differently,
and therefore the sweet spot of the deposition process is
relatively small, usually one to two centimeters, depend-
ing on the geometry. The deposition rate per laser pulse
ranges from a fraction of an angstrom to a few angstroms.
The technique was found to be fairly reproducible from the
early days and has been used extensively for research and
development purposes since 1989. Even though other
techniques are potentially more manufacture-friendly, la-
ser ablation is still very actively used for prototyping de-
vices made up of several complex materials, such as
superconductors, ferroelectric oxides, and magnetic ox-
ides.
A particular problem associated with PLD is the depo-
sition of micrometer-sized droplets, so-called boulders, on
the grown lm. These particles originate at the target and
are emitted from the action of the laser pulse. Avariety of
procedures have been utilized to reduce this problem, so
that the boulder density can be very low in the best lms.
Such procedures include target preparation (such as fre-
quent polishing), defocusing the laser spot, mechanically
chopping the plume, and spatially ltering the beam.
Although PLD can produce high deposition rates (up to
tens of nanometers per second), the area on which one
deposits is small. A straightforward way to increase the
deposition area is to scan wafers over the plume either by
moving the substrate vertically and horizontally or by ro-
tating the substrate. This method has been tried in several
laboratories and is still pursued. Another scheme utilizes
a rotating cylindrical target with a linear laser prole to
obtain larger deposition areas.
Another major difculty of these techniques is uniform
heating of large wafers. Most of the heating methods for
small-area laser ablation have been to mount a substrate
on a heated metal surface with silver paste, which is dif-
cult to extend to larger sizes. In order to overcome this
difculty of heating a large wafer, an off-axis laser ablation
technique has been developed (see Fig. 4b). A large wafer
(58 cm) can be mounted parallel to the direction of the
plume inside a relatively simple blackbodylike heater. The
deposition takes place when the atoms collide with the
background pressure and are scattered to the surface of
the wafer. By rotating the wafer and selecting an appro-
priate pressure for the geometry, a fairly uniform deposi-
tion was achieved over 5-cm wafers. This technique allows
for simultaneous deposition on both surfaces of the wafer,
which is an important benet for microwave applications
requiring a ground plane. A drawback of this technique is
its low deposition rate because of the off-axis geometry.
Typical conditions for deposition of a 5-cm wafer resulted
in a deposition rate about a factor of 10 lower than for the
on-axis deposition.
By using laser uence just enough to evaporate a few
atomic layers of the target in a low-oxygen environment
(less than 10
2
PaE10
4
torr) and at the same time using
a sequence of metal or metal oxide targets, one can obtain
a process similar to molecular-beam epitaxy (MBE) by
evaporation, here called laser MBE. Some in situ diag-
nostic tools can then be used to characterize the growth of
(a) (b)
Ablation
target
Laser
beam
Substrate
Figure 4. Schematic for the (a) on-axis and (b) off-axis laser ab-
lation processes.
the materials. This technique has mainly been used to
grow articially layered superconducting materials, such
as the innite-layered superconductor [28,29].
7. EVAPORATION AND MOLECULAR-BEAM EPITAXY
Deposition of HTS thin lms by evaporation follows a tra-
dition of such deposition of metal lms. The added com-
plexity here is the required partial pressure of oxygen
during growth. One can distinguish several variations in
the evaporation approaches: ash evaporation, reactive
coevaporation of metals, and sequential deposition by
MBE. Except for the rst process, these evaporation tech-
niques utilize individual metallic sources. However, in at-
tempting to control individual sources, two difculties
arise. One is the need for very ne control of individual
sources, including development of species-specific sensors.
The second is the need to work at a low enough pressure to
minimize beam scattering and at the same time achieve
the highly oxidizing thermodynamic conditions required
for the growth of these compounds. For an in situ process,
one is conned to work at a pressure of molecular oxygen
above 100 mPa (1mtorr) or to use a more reactive form of
oxygen supply than molecular oxygen.
7.1. Flash Evaporation
The simplest approach to evaporation of HTS materials is
to evaporate the compound in small batches, namely, in
ashes of evaporant material. Since the material does not
melt congruently, it is not possible to establish a contin-
uously constant rate of metal uxes, but for short enough
intervals, one can average out the compositional varia-
tions. Usually, the evaporant material is a powder of
YBCO located in a feed mechanism that drops small quan-
tities onto an evaporation source [30]. In most cases, the
lms require a postannealing treatment to oxidize the lm
sufciently, since the fast deposition does not allow for
sufcient incorporation of oxygen. This method is at
present less significant.
7.2. Reactive Coevaporation
Historically, the rst attempts to make lms by evapora-
tion utilized a high molecular oxygen pressure. The prob-
lems associated with a high background oxygen pressure
are rate control of the individuals species and the degra-
dation of the sources. In order to circumvent this difculty,
several approaches were taken. One was to accommodate
a high differential pressure between the sources and the
sample by introduction of nozzles in close proximity to the
sample. Another method is to utilize a more reactive spe-
cies for oxygen incorporation, such as atomic oxygen or
ozone.
The evaporation technique allows one to tune the com-
position of the lm by adjusting the relative rates of the
sources. Several groups have worked over the past years
on developing process control for coevaporation and study-
ing YBCO lm properties as a function of metal atom
composition. Figure 5 shows SEM micrographs of lms
with various metal compositions. The general nding is
that lms that are grown slightly Y- and Cu-rich have
better performance than lms with exactly 1 : 2: 3 stoic-
hiometry. This is true for a majority of in situ techniques
and not just evaporation. The reasons for this are still not
completely clearin particular, whether it is a materials
issue or a process control issue. It is known that super-
conducting properties of lms degrade significantly as one
goes into the Ba-rich composition. It is possible that pro-
cess uctuations around the desired metal atom stoic-
hiometry are responsible for degradation of lms close to
1: 2: 3 composition.
Probably the most technologically significant advance
in HTS thin-lm evaporation has been the large-area
heater developed for coevaporation by the group of
H. Kinder at the Technical University of Mu nich (Germany)
and depicted in Fig. 6. They used a blackbody-type rotat-
ing-disk heater, similar to the one used by others in PLD
deposition, but they added an oxidation pocket [31]. This
heater has a narrow slit that allows for a differential pres-
sure between the oxidation pocket and the rest of the
chamber of about 1000:1. The deposition and the oxidation
processes are therefore separated, and a low pressure in
the chamber improves the stability of thermal evaporation
sources. Uniform YBCO lms were grown on wafers up to
Figure 5. Morphology of YBCO lms as a function of their com-
position during reactive coevaporation.
Oxidation pocket
Deposition zone
Sealing slit
O
2
Figure 6. Schematic of the heater for large-area HTS deposition
by evaporation.
20 cm in diameter using this technique. More importantly,
this method is significantly more cost-effective and has
higher throughput than any of the other PVD processes.
7.3. Molecular-Beam Epitaxy
In general, one can distinguish between reactive molecu-
lar-beam epitaxy (MBE) approaches to HTS lm deposi-
tion and mere reactive evaporation by the lower
background pressure of the former technique. Another
difference is that MBE deposition is usually done sequen-
tially rather than simultaneously. Typically, MBE systems
have multiple sources with individual shutters, as well as
some in situ monitoring tools (see Fig. 7). Reference 32 has
a thorough discussion of relevant issues is reactive MBE of
HTS lms. The work by a number of groups in this eld
has focused on careful atomic layering to produce very
smooth lms, as well as customized growth of new mate-
rials and heterostructures containing these phases. The
reactive MBE technique also lends itself to careful control
of chemical doping of these materials, an issue that is
very important for the superconducting properties of the
cuprates.
7.4. In situ Monitoring Diagnostics
Evaporation and, especially, MBE, with their inherent low
background pressure, lend themselves well to vacuum
techniques for in situ deposition monitoring and lm di-
agnostic tools such as reection high-energy electron dif-
fraction (RHEED). RHEED has been shown to be
particularly helpful in controlling deposition of BiSr
CaCuO lms, where the growth occurs in blocks of sub-
unit cells, but where many similar phases are possible.
RHEED has been less helpful to date in the growth of
YBCO, which occurs in blocks of unit cells and where in-
tergrowths are more difcult to tailor, due to the higher
stability of the primary phase. Terashima and coworkers
[33] have shown that growth of YBCO can exhibit oscilla-
tions in the RHEED pattern commensurate with unit cell
deposition, suggesting that growth proceeds in a unit-cell-
by-unit-cell fashion, rather than in smaller building
blocks. More recently, RHEED has also been extended to
high-pressure processes such as PLD [34].
Another area of technological development has been in
the use of optical absorption techniques for measurement
and control of atomic uxes. Both hollow-cathode lamp
systems [35] and tunable diode lasers [36] have been used
to monitor uxes in situ in close proximity to the sub-
strate. Such monitoring of individual uxes is a prerequi-
site for careful control of HTS growth.
8. CONCLUDING REMARKS
In spite of a wealth of research, growth of cuprate lms
has remained a complicated matter. This is due to the
materials rather complex multicomponent crystal struc-
tures. They are prone to a variety of defects and growth
morphologies. Much work still remains to be done until
lms are better understood and more reproducible.
As already described, HTS lms are now routinely
made with out-of-plane as well as in-plane alignment on
single-crystalline substrates. Aligned lms are still not
routine on polycrystalline substrates, but there has been
progress in this area as well. This has been manifested in
high critical currents in lms, particularly for YBCO. Typ-
ical critical currents in good quality YBCO lms are over
2 10
6
A/cm
2
at 77 K and above 10
7
A/cm
2
at low temper-
atures. There has been much work in characterizing de-
fect structures, but a good understanding of their
inuence on physical properties is still lacking. One
growth structure that seems to be nearly ubiquitous in
YBCO lms is the spiral, which forms during growth and
at the core of which is a screw dislocation [36]. Figure 8
Q
u
a
rtz
m
ic
ro
b
a
la
n
c
e
Atomic
absorption
light in
Atomic
absorption
light out
RHEED
screen
Electron
gun
Wafer
holder
Source furnaces
Mass
spectrometer
Ozone
beam
Leak
valve
Turbo
pump
Figure 7. Schematic for the reactive MBE process used in HTS
lm growth.
Figure 8. An STM image of a laser ablated YBCO lm, showing
spiral growth structures. (Image courtesy of Prof. Darrell
Schlom.).
shows a scanning tunneling microscopy (STM) image of a
surface of a YBCO lm with a spiral. Steps in this image
have a height of one unit cell of YBCO. Such spirals have
been identied as contributing of pinning of vortices, im-
portant for high critical currents.
Maximally oxygen-doped YBCO bulk materials, made
at higher temperatures than thin lms, have a supercon-
ducting transition temperature of 92 K and a resistivity of
35 mO cm at 100K. Although having nominally the same
crystal structure as the bulk material, YBCO lms often
have different electronic properties. For example, lower T
c
(8590k) and expanded c-axis lattice constants (1.170
1.172 nm) are frequently found in thin lms. Additionally,
ux-pinning centers appear to be highly dispersed in in
situgrown lms of YBCO, although the structural origin
of these centers has not yet been identied. It is important
to have a close feedback loop between lm deposition and
structural and electrical characterization in order to fur-
ther improve HTS lms.
Technology for deposition and monitoring of HTS thin
lms has developed significantly since the late 1980s. At
the same time, the materials understanding of lm nucle-
ation, growth, and oxide interfaces has started to unfold.
Together, these developments in the future will bring new
capabilities for HTS lm growth.
BIBLIOGRAPHY
1. J. M. Rowell, A decade of progress towards a high tempera-
ture superconducting electronics technology, Solid State Co-
mmun. 102:269282 (1997).
2. T. R. Lemberger, Films of high-temperature oxide supercon-
ductors, in D. M. Ginsberg, ed., Physical Properties of High
Temperature Superconductors III, World Scientific, Singa-
pore, 1992.
3. C. P. Poole, H. A. Farach, and R. J. Creswick, Superconduc-
tivity, Academic Press, San Diego, 1995.
4. T. B. Lindemer et al., Decomposition of YBa
2
CU
3
O
7 x
and
YBa
2
Cu
4
O
8
for po
2
r0.1MPa, Physica C 178:93 (1991).
5. R. K. Williams et al., Oxidation induced decomposition of
YBa
2
Cu
3
O
7 x
, J. Appl. Phys. 70:906913 (1991).
6. R. H. Hammond and R. Bormann, Correlation between the in-
situ growth conditions of YBCO thin lms and the thermo-
dynamic stability criteria, Physica C 162164:703704 (1989).
7. R. Beyers and B. T. Ahn, Thermodynamic considerations in
superconducting oxides, Rev. Mater. Sci. 21:335 (1991).
8. P. Karen, O. Braaten, and A. Kjekshus, Chemical phase dia-
grams for the YBa
2
Cu
3
O
7
family, Acta Chem. Scand. 46:805
840 (1992).
9. M. Naito et al., Thin lm synthesis of the high T
c
oxide su-
perconductor YBa
2
Cu
3
O
7
by electron beam codeposition, J.
Mater. Res. 2:713725 (1987).
10. R. Feenstra et al., Effect of oxygen pressure on the synthesis
of YbaCuO
7x
thin lms by post deposition annealing, J.
Appl. Phys. 69:65696585 (1991).
11. J. M. Phillips et al., Comparison of Ba
2
YCu
3
O
7d
thin lms
grown on various perovskite substrates by coevaporation, J.
Mater. Res. 7:26502657 (1992).
12. R. G. Humphreys et al., Physical vapor deposition techniques
for the growth of YBa
2
Cu
3
O
7
thin lms, Superconduct. Sci.
Technol. 3:3852 (1990).
13. J. A. Alarco et al., High quality YBCO thin lmslaser de-
position, co-evaporation, and device fabrication, Phys. Scripta
44:95101 (1991).
14. C. H. Stoessel et al., Thin-lm processing of high-T
c
super-
conductors, J. Superconduct. 6(1):117 (1993).
15. S. M. Garrison et al., Observation of two in-plane epitaxial
states in YBa
2
Cu
3
O
7d
lms on yttria-stabilized ZrO
2
, Appl.
Phys. Lett. 58:21682170 (1991).
16. H.-U. Habermeier et al., Preparation and properties of YBCO
thin lms with the c-axis aligned in the lm plane, in L. Co-
rrera, ed., High Temperature Superconductor Thin Films,
North-Holland, Amsterdam, 1992, pp. 343352.
17. A. Perrin and M. Sergent, in A. Narlikar, ed., Studies of High
Temperature Superconductors, Vol. 7, Nova Science Publish-
ers, Commack, NY, 1991.
18. J. Qiao and C. Y. Yang, High-T
c
superconductors on buffered
silicon: Materials properties and device applications, Mater.
Sci. Eng. R14:157202 (1995).
19. Y. Iijima et al., In-plane aligned YBa
2
Cu
3
O
7 x
thin lms de-
posited on polycrystalline metallic substrates, Appl. Phys.
Lett. 60:769771 (1992).
20. X. D. Wu et al., High current YBa
2
Cu
3
O
7 d
thick lms on
exible nickel substrates with textured buffer layers, Appl.
Phys. Lett. 65:19611963 (1994).
21. C. P. Wang et al., Deposition of in-plane textured MgO
on amorphous Si
3
N
4
substrates by ion-beam-assisted deposi-
tion and comparisons with ion-beam-assisted deposited
yttria-stabilized-zirconia, Appl. Phys. Lett. 71:29552957
(1997).
22. S. M. Rossnagel and J. J. Cuomo, Negative ion effects during
magnetron and ion beam sputtering of YBa
2
Cu
3
O
x
, in Thin
Film Processing and Characterization of High-Temperature
Superconductors, American Institute of Physics Conf. Proc.
165, AIP, New York, 1988, pp. 106113.
23. U. Poppe et al., Direct production of crystalline superconduct-
ing thin lms of YBa
2
Cu
3
O
7
by high-pressure oxygen sput-
tering, Solid State Commun. 66:661665 (1988).
24. C. B. Eom et al., Synthesis and properties of YBa
2
Cu
3
O
7
thin
lms grown in situ by 90 degrees off-axis single magnetron
sputtering, Physica C 171:354382 (1990).
25. J. R. Gavaler et al., Critical parameters in the single-target
sputtering of YBa
2
Cu
3
O
7
, J. Appl. Phys. 70:43834391 (1991).
26. N. Savvides and A. Katsaros, In situ growth of epitaxial
YBCO thin lms by on-axis unbalanced direct current mag-
netron sputtering, Appl. Phys. Lett. 62:528530 (1993).
27. X. X. Xi et al., Preparation of YBa
2
Cu
3
O
7
thin lms by in-
verted cylindrical magnetron sputtering, J. Less-Common
Metals 151:349355 (1989).
28. T. Kawai et al., Superconducting articial lattices grown by
laser MBE, in L. Correra, ed., High Temperature Supercon-
ductor Thin Films, North-Holland, Amsterdam, 1992.
29. H. Koinuma and M. Yoshimoto, Controlled formation of oxide
materials by laser molecular beam epitaxy, Appl. Surface Sci.
75:308319 (1994).
30. Y. Yasuda et al., YBaCuO superconducting thin lms pre-
pared by plasma-assisted ash evaporation, Appl. Phys. Lett.
55:307309 (1989).
31. P. Berberich et al., Homogeneous high quality YBa
2
Cu
3
O
7
on
3
/ /
and 4
/ /
substrates, Physica C 219:497 (1994).
32. D. G. Schlom and J. S. Harris, Jr., MBE growth of high T
c
superconductors, in R. F. C. Farrow, ed., Molecular Beam Epi-
taxy: Applications to Key Materials, Noyes, Park Ridge, IL,
1995.
33. T. Terashima et al., Reection high-energy electron diffrac-
tion oscillations during epitaxial growth of high-temperature
superconducting oxides, Phys. Rev. Lett. 65:26842687 (1990).
34. G. J. H. M. Rijnders et al., In situ monitoring during pulsed
laser deposition of complex oxides using reection high ener-
gy electron diffraction under high oxygen pressure, Appl.
Phys. Lett. 70:18881890 (1997).
35. J. N. Eckstein, I. Bozovic, and G. F. Virshup, Atomic layer-by-
layer engineering of high-T
c
materials and heterostructure
devices, Mater. Res. Soc. Bull. 19(9): 4450 (1994).
36. W. Wang et al., Direct atomic ux measurement of electron-
beam evaporated yttrium with a diode-laser-based atomic ab-
sorption monitor at 668nm, Appl. Phys. Lett. 71:3133 (1997).
37. C. Gerber et al., Screw dislocations in high-T
c
lms, Nature
350:279280 (1991).
HYBRID CAD TECHNIQUES
F. ARNDT
University of Bremen
Bremen, Germany
1. INTRODUCTION
To meet the increasing demand for enhanced waveguide
components and antennas in wireless terrestrial and
satellite-based communications, and for radar applica-
tions, accurate and efcient CAD tools are required. Due
to efforts during the 1990sthe decade of global 3D
electromagnetic (EM) eld solversthe task of accurate
analysis of components can be considered as being largely
solved. Current challenges for CAD software relate to a
new quality of efciency and speed that allow direct, fast
component optimizations in order to meet desired speci-
cations within reasonable timeframes.
Many approaches to improve the efciency of EM-based
eld solvers have been reported so far, such as reduced-
order models (ROMs) applied for nite-element (FE) [1] or
nite-difference (FD) [2] methods, fast multipole and
adaptive integral methods (FMM, AIM) for method-of-
moment (MoM) solutions of scattering and microstrip
circuit problems [310], or combined subgrid conformal
techniques for FD methods [11,12].
The issue discussed in this article is the utilization of
advanced hybrid MM/FE/MoM/FD techniques, which ty-
pically can go beyond the efciency possibilities of the
involved single methods.
In computational electromagnetics, the idea of combin-
ing different variations to hybrid techniques or hybrid
methods to expand their capabilities is not new.
Hybrid techniques have already been utilized long ago
for handling electromagnetic scattering and radiating
problems for large structures [7,13,14]. Examples are
the combination of MoM with an asymptotic method
[e.g., geometric theory of diffraction (GTD)], [7,1316], or
the combination of FE with the boundary integral equa-
tion method [7,17,18].
We apply in our article the hybrid notation for
combining the fast mode-matching (MM) method with
EM-based space or surface discretization methods, such
as the nite-element (FE) and nite-difference (FD) meth-
ods and the method-of-moments (MoM), to combine the
efciency of the MM with the exibility of the FE, MoM,
and FD techniques.
Due to its efciency, the mode-matching (MM) method
[1945] has been widely employed for designing wave-
guide components, where modal expansions can be de-
rived analytically. For structures that are not separable in
Cartesian, cylindrical, or elliptical coordinate systems,
FE, FD, or MoM techniques provide the desired exibility.
However, because of rather high requirements concerning
central processing unit (CPU) time for the CAD of compo-
nents using pure FE, MoM, or FD methods, for reasonably
fast optimizations of typical industrial components, which
often require a high number of iterations to meet given
specications, more efcient techniques are desirable. A
very effective approach for solving this problem utilizes
hybrid methods based on MM/FE/MoM/FD techniques,
hence retaining the specic advantages of all proven EM
methods while largely avoiding their disadvantages.
Hybrid mode-matching/method-of-moment (MM/MoM)
techniques have been applied for horn antennas for a long
time [46,47], where the inner horn structure is calculated
by the MM method and the radiation problem is solved by
MoM [4650]. Hybrid MM/FE techniques have been intro-
duced more recently for the analysis of dual-mode lters
[5153]. In these applications, the cross section eigenvalue
problem of arbitrarily shaped homogeneous waveguide
structures (e.g., cross-iris) is solved by a 2D FE method
while the MM is applied for the calculation of the scatter-
ing parameters at the discontinuities with arbitrary con-
tour. The full-wave combination of all discontinuities is
achieved by the generalized scattering matrix (GSM). A
more detailed overview on the MM/FE method is given in
Ref. 54. More recent applications describe the MM/FE
CAD of wafe-iron, evanescent-mode, and combline lters
[5557], of coaxial feeds [58,59], and the analysis of some
discontinuities [60,61]. Modal expansion FDTD and trans-
mission-line matrix (TLM) techniques have been proposed
[62,63], and a hybrid MM/FD-FD method has been pre-
sented [64]. The extension of hybrid methods to MM/FE/
MoM and MM/FE/FD has been presented [65,66], and
hybrid techniques exploiting the full efciency and ex-
ibility spectrum of MM/FE/TransFE/MoM/FD methods
have been introduced [67,68].
In this article, we describe the recent state of the art of
hybrid MM/FE/MoM/FD techniques employed for the
rigorous, efcient CAD and optimization of waveguide
components, coupled horns, and waveguide slot array
antennas, and discuss some typical applications and ad-
vanced aspects. Representative application examples will
demonstrate the versatility and computational speed of
these hybrid methods. Their accuracy is veried by avail-
able measurements or reference calculations.
2. THEORY
2.1. MM/FE Method
A very large class of waveguide components in microwave
techniques is typically composed of stepwise homogeneous
2040 HYBRID CAD TECHNIQUES
waveguide sections (Fig. 1). Because of its high numerical
efciency, the mode-matching (MM) CAD technique has
long been advantageously applied for the design of corre-
sponding components such as for iris or metal-insert-
coupled lters, junctions, couplers, diplexers, multiplex-
ers, polarizers, transformers, and orthomode transducers
[1945]. An overview of many of the earlier research work
in MM techniques is given in Ref. 36.
The basic principle of the MM technique can be eluci-
dated at a step discontinuity (Fig. 1). Matching the
tangential E- and H-eld components along the trans-
verse surface of the general step discontinuity, which is
assumed to be located at z =0, yields the relation between
the incident and scattered modal wave amplitude coef-
cients a and b
Z
I
i
_
(a
I
i
b
I
i
) =
o
j =1
C
ij
Z
II
j
_
(a
II
j
b
II
j
) i =1 o
o
i =1
C
ij
Y
I
i
_
(a
I
i
b
I
i
) =
Y
II
j
_
(a
II
j
b
II
j
) j =1 o
(1)
where Z=1/Yare the modal wave impedances and C
ij
are
the elements of the frequency independent coupling ma-
trix
C
ij
=
_
S
II
_
e
I
T
i
e
II
j
dS=
_
S
II
_
h
I
T
i
h
II
j
dS; (2)
e and h are the transversal eigenvectors
e=
u
z
.
V
t
C TE modes
V
t
C TM and TEM modes
_
_
_
h=u
z
.
e
(3)
with u
z
the unit vector in the z direction, and the
potentials C are solutions of the 2D Helmholtz equation
V
2
t
Ck
2
c
=0 (4)
in case of TE and/or TM modes, with the separation
condition
k
2
c
b
2
=k
2
(5)
where jb =g is the propagation factor and k the free-space
wavenumber; V
2
t
=V
2
(@
2
=@z
2
) denotes the transversal
Laplace operator. For TEM modes, C are solutions of the
Laplacian equation
V
2
t
C=0 (6)
The potentials C are suitably normalized
__
S
V
T
t
C
i
V
i
C
i
dS=1 (7)
The Dirichlet and Neumann boundary conditions
are
C
TM
=0 G
E
@C
TM
@n
=0 G
M
(TM modes)
@C
TE
@n
=0 G
E
C
TE
=0 G
M
(TE modes)
C
TEM
i
=
constO0 G
E;i
0 else G
E
_
_
_
@C
TEM
@n
=0 G
M
(ith TEM mode)
(8)
where G denotes an electric (E) or magnetic (M)
wall. Equation (1) is represented in form of a matrix
equation
a
I
b
I
=V(a
II
b
II
)
V
T
(a
I
b
I
) =a
II
b
II
V=[
Y
I
_
][C][
Z
I
_
]
Y
I
_
=diag(
Y
I
_
)
Z
I
_
=diag(
Z
I
_
)
(9)
which can be solved with regard to the amplitudes of
b in order to yield the generalized scattering matrix
(GSM)
b=Sa (10)
In case of rectangular, circular, circular coaxial, and
elliptical waveguide discontinuities, the cross section ei-
genvectors required for the MM technique are given
y
z
a
I
a
II
b
I
Metallic wall
S
I
S
II
b
II
Figure 1. Step discontinuity between two homogeneous wave-
guides.
HYBRID CAD TECHNIQUES 2041
analytically [1945]. For more general cross sections,
the corresponding eigenvalue problem has to be solved
numerically, where formulations by a 2D FE method
[5257] are preferred, leading to a fast convergent direct
solution without requiring search algorithms that are
usually necessary for cross section eld matching [69] or
for boundary integral equation techniques [70].
2.1.1. 2D FE Scalar Approach (Nodal Elements) for
Waveguides with Ideally Conducting Boundaries. Nodal
function expansion of scalar potentials achievesaccord-
ing to the authors experiencethe most efcient approach
(and free of spurious modes) for metallic boundaries and
homogeneous cross sections, such as for typical waveguide
discontinuities shown in Fig. 1.
A standard generalized matrix eigenvalue problem for
the considered waveguide cross sections of more general
shape is obtained [51,52] as follows
[Kk
2
c
M] C
=0 (11)
where
K
JK
=
_
O
_
V
t
N
J
V
t
N
K
dO
M
JK
=
_
O
_
N
J
N
K
dO; C
=(C
1
; C
2
; ; C
N
)
T
and the potentials C
are approximated by their nodal

values C
K
and rst-order Lagrangian interpolation poly-
nomials N
K
(x, y) by
C(x; y) -
K
C
K
N
K
(x; y) (12)
The generalized eigenvalue problem (11) is transformed
into standard form by a Cholesky separation technique,
and is solved iteratively by the Lanczos algorithm after
suitable preconditioning, which includes the boundary
problem for the TEM wave of coaxial sections as a special
case.
Matching the transverse elds at the common interface
of the general waveguide step discontinuity leads to the
corresponding generalized scattering matrix (GSM)
[52,54], which achieves the stable, reliable, and efcient
full-wave combination of all involved parts.
2.1.2. Line Integral Formulation of Coupling Inte-
grals. The frequency independent coupling integrals of
the MM technique can be formulated in terms of line
integrals by using the common denitions of the trans-
versal eigenvectors [54]
k
2
cII
k
2
cII
k
2
cI
_
G
II
C
II
@C
I
@n
dc TE TE
0 TE TM; TE TEM
_
G
II
C
II
@C
I
@t
dc TMTE; TEMTE
k
2
cI
k
2
cI
k
2
cII
_
G
II
C
I
@C
II
@n
dc TMTM
_
G
II
C
I
@C
II
@n
dc TMTEM; TEMTEM
0 TEMTM
(13)
In case of degenerate modes (k
cI
=k
cII
), an adequate
expression for the limiting case k
cI
-k
cII
is formulated.
Line integral formulations (13) of coupling integrals
are applied for discontinuities, which consistat least at
one port sideof waveguides providing analytical expres-
sions for the eigenvectors. The accuracy is nearly identical
to the corresponding area integral formulations, and there
is a slight reduction in calculation time concerning the
individual coupling integrals.
However, for discontinuities of waveguides with arbi-
trary cross section on both sides, where the eigenvectors
are numerical solutions of the corresponding 2D FE
eigenvalue problem, the accuracy of line integral calcula-
tions can typically be one order of magnitude lower than
for results using area integral formulations. This is due
mainly to errors caused by nite discretization in the
numerical approximations of the line integral contour.
Moreover, the factor (1 k
2
cI
=k
2
cII
), which occurs in the
line integral formulations (13), can increase the error still
further, since the determination of eigenvalues k
c
is also of
only limited accuracy. Hence, in such cases, in contrast to
the opinion presented in Ref. 61, area integral formula-
tions [54] are preferable.
2.1.3. 2D FE Vector Approach (Edge Elements) for Inho-
mogeneous Cross Sections. For inhomogeneous cross sec-
tions, such as for waveguide structures containing
dielectrics (Fig. 2), triangular edge elements for transver-
sal eld expansion are preferred because of their useful
features such as eliminating spurious modes (which can
occur at inhomogeneous cross sections containing dielec-
trics), closely approximating material boundaries, and
reducing effects of singularities.
Following the approach in Refs. 71 and 72, we expand
the vector eld within a triangular edge element (Fig. 2)
as
E
e
=
3
i =1
N
e
i
E
e
i
(14)
where E
e
i
denotes the tangential eld along the ith edge
and N the normalized vector function
N
e
i
=l
e
i
(L
e
i
VL
e
j
L
e
j
VL
e
i
) (15)
where l
i
is the length of ith edge and L
j
(x, y) the area
coordinate of the jth node [71]. The Ritz procedure [71] is
applied to formulate the generalized eigenvalue problem,
which is reduced to tridiagonal form by the Lanczos
procedure [72,73]. The system of equations arising in
each Lanczos iteration step is solved by sparse matrix
Cholesky decomposition using the minimum-degree algo-
rithm [73].
It should be noted that in contrast to the opinion
expressed in Ref. 60, for homogenous cross sections (con-
taining only metallic boundaries), the scalar nodal ele-
ment approach [52,54] achieves the most efcient
formulation for this kind of problem.
2.2. MM/Transnite FE Method for Planar Waveguides with
Arbitrary Contour
Many waveguide components (such as a T-junction with
septum, Fig. 3) are planar in the H or E plane with
arbitrary contour. Instead of solving the complete struc-
ture by the transnite element (TransFE) method [74], for
such components, a hybrid MM/TransniteFE method is
applied [67], which combines again the efciency and
exibility advantages of the involved methods. For the
planar element under consideration, the 2D Helmholtz
equation can be expressed by an adequate functional [74].
Its solution area is subdivided into an inner region to be
discretized and the port region. The elds are approxi-
mated by a suitable set of basis functions, which are
solutions of the Helmholtz equation in either the H or in
the E plane. Assuming that the elds at the ports are
expressed in terms of normalized forward- and backward-
traveling modes, an expression containing the modal S
parameters can be derived. The resulting equation system
is solved by the iterative biconjugate gradient (BCG)
procedure.
2.3. MM/MoM Method for Apertures and Arbitrarily Shaped
3D Waveguide Discontinuities
For horn antenna structures with arbitrarily shaped outer
contour, and for arbitrary 3D metallic N-port waveguide
structures (Fig. 4), a hybrid MM/MoM is advantageously
applied [4650,76]. Enforcing the continuity of the tan-
gential magnetic eld at the aperture(s) in the usual way
[7578] yields the following equation for the magnetic
surface current density M
H
i
t
=H
I
t
{M] H
II
t
{M] (16)
where H
i
t
denotes the tangential magnetic eld incident
from the waveguide side(s), and H
I
t
, H
II
t
are the tangential
elds in regions I and II, respectively (Fig. 4) caused by
the magnetic surface current density.
For the solution of (16), the MoM is employed. The
magnetic surface current density on the aperture surface
is approximated by Q linear independent basis functions
M
q
on S
a
M(r) =Z
0
Q
q =1
V
q
M
q
(r) (17)
where V
q
are the expansion coefcients and Z
0
denotes the
free-space impedance. Introducing a set of P=Q linear
Figure 2. Step discontinuity to waveguide with inhomogeneous
cross section (circular waveguide containing dielectric cylinder);
mesh; vector basis function for a triangular element.
Figure 3. Planar waveguide structure with
arbitrarily shaped contour solved by the MM/
transnite FE method. Example: T junction
with septum.
independent test functions W
p
on S
a
results in a system of
linear equations, which leads to the elements of the modal
admittance matrix for region I [46,50,76].
Calculation of the admittance matrix for region II is
based on the KirchhoffHuygens principle. The electro-
magnetic eld is calculated using the magnetic surface
current densities on S
a
and the electric surface current
densities on S
a
and S
b
:
M(r) =
nE(r) r on S
a
0 r on S
b
_
(18a)
J(r) =nH(r) r on S
a
and S
b
(18b)
For numerical calculations, the electric eld integral
equation (EFIE) is preferred:
E(r) = rot
_
S
M(r
/
)G
0
(r; r
/
)dS
/
1
joe
0
rot rot
_
S
J(r
/
)G
0
(r; r
/
)dS
/
(19)
Advantages include the possibility of direct applicability
of numerically available eigenvectors e
i
in region I as
basis functions u
z
e
i
for the magnetic surface current
density. In (19), G
0
is the free-space Greens function, and
the electric surface current densities are expanded in
J(r) =
S
s =1
I
s
J
s
(r) r on S
a
and S
b
(20)
For the electric surface current densities, the Rao
WiltonGlisson (RWG) basis functions [79] for triangular
patches are chosen, which yields the required high ex-
ibility. The Galerkin method is applied for the numerical
solution [4650,76], which results in the modal admit-
tance matrix of region II.
Using the orthonormality of the eigenvectors, the fol-
lowing expression for the scattering coefcients is derived
[50,76]
s
ij
= Z
0
Q
q =1
A
iq
V
qj
d
ij
(21)
where V
qj
is the expansion coefcient for the basis function
M
q
for excitation with mode j, d
ij
is the Kronecker delta,
and
A
iq
=
_
S
M
q
.
( ^ zz e
i
)dS (22)
2.4. MM/FD Method
For arbitrary 3D structures containing dielectric material,
such as the example of Fig. 5, the MM/nite-difference
(FD) method is applied. When calculating the structure as
a whole, the FD time-domain (FDTD) method is used;
when combining parts with the MM, the FD frequency
domain (FDFD) method is preferred[109].
Microwave structures, such as lters and feeds, often
include dielectric 3D objects of arbitrary shape, as well as
areas of rather different eld intensities. Hence, the
numerical effort for the FD computation can be high for
accurate results if no additional renement approaches
are utilized. Several subgrid techniques have been re-
ported [11,12,8084]. Whereas usual subgrids mostly ty-
pically require additional interpolations schemes at the
grid interfaces that can reduce exibility, the literature
[11,12,64,80,81,108,109] presents an efcient direct sub-
grid method without interpolation.
For the corresponding CAD applications described in
this article, the direct subgrid FD technique is generalized
to include both arbitrarily curved metallic and dielectric
surfaces by a formulation based on combining the subgrid
with a locally conformal FD algorithm [85] and with a
modied intersection approach [86] for arbitrary dielectric
(a) (b)
n
z
^
J
n
n
M
M
I
II
II
(3)
(1)
(2)
S
b
S
b
S
a
S
3
a
S
1
a
S
2
a
I
I
I
Figure 4. Denitions of regions I and II for the investigated horn
and arbitrarily shaped 3D waveguide structures.
Figure 5. Arbitrary 3D structures containing dielectric material.
Example: Lateral coax feed with dielectric coating and compensa-
tion posts.
boundaries. The high exibility of this mesh generator is
characterized by the fact that all three techniques can be
combined [12] arbitrarily, also at one cell if necessary. For
the FDTD, S-parameters are extracted via the matrix
pencil method [87,88,105] which reduces the number of
involved timesteps as compared with more standard
methods.
2.5. Multilevel Fast Multipole Algorithm (MLFMA) and
Adaptive Integral Method (AIM)
Matrix equations for N unknowns resulting from MoM
solutions [e.g., Eqs. (16)(20)] require O(N
2
) operations
when applying standard MoM solutions by using an
iterative solver [6,7]. Hence, for a high number of un-
knowns arising in large-scale problems, such as slot
arrays, it is imperative to apply fast integral equation
solvers, such as the multilevel fast multipole algorithm
(MLFMA) or the adaptive integral method (AIM). The
MLFMA allows matrix vector products to be effected in
O(N log N) operations [37,10,89]. Applications of MLFMA
have been restricted mainly to scattering problems so far.
MLFMA applications to apertures with an electric eld
integral equation (EFIE) approach have been introduced
in Refs. 90 and 91; a combined eld integral equation
(CFIE) MLFMA solution for coupled apertures was pre-
sented more recently [92,93].
Fast integral solvers based on the adaptive integral
method (AIM) have proved appropriate for large-scale
problems [9498]. The method retains the advantages of
the conjugate-gradient fast Fourier transform (CGFFT)
technique allowing matrix vector products to be effected in
O(Nlog N) operations as well as exible modeling techni-
ques involving adequate basis functions [97]. Applications
of AIM have been restricted mainly to scattering problems
and microstrip antennas [9497]. Initial investigations of
the applicability of the AIM method to aperture radiating
problems, such as slot arrays, have been presented more
recently [98,113].
For application to, for example, the slot array problem
(Fig. 6), the formulation is subdivided into three calcula-
tion steps. In the rst step, the MoM is applied to calculate
the integral equation for the radiating aperture(s)
within the corresponding external (horn or slot array)
structure geometries. The eigenvectors of rectangular and
circular apertures are analytically given or can be calcu-
lated for arbitrarily shaped apertures by the FE approach
[51,54].
In the second step, the AIM is utilized for the efcient
solution of the matrixvector product. The third step
comprises the calculation of the radiation eld by
taking into account the generalized scattering matrix
composed of the complete radiating and feed parts. The
computation scheme for the example of slot arrays is
shown in Fig. 6.
The solution is based on the magnetic eld integral
equation (MFIE) setup in the aperture areas of the slot
array [113]:
^ zz H
inc
^ zz H
1
{M
z
] = ^ zz H
2
{M
z
] (23)
For the solution of (23), the magnetic surface current
density is discretized using the aperture eigenvectors e
j
as
basis functions
M
z
= ^ zz
j
M
j
e
j
(24)
Scalar multiplication by weighting function and inte-
gration over the apertures yields the matrix equation in
the usual form [99]
Z
.
M=I (25)
with
(Z)
ij
=
__
w
i
.
H
2
{M
z
]dA
__
w
i
.
H
1
{M
z
] dA;
(I)
i
=
__
w
i
.
H
inc
dA;
(M)
j
=M
j
(26)
and the excitation vector
(I)
i
= 2
__
^ zz
.
(e
i
h
inc
)dA: (27)
The matrix elements can be formulated in their mixed-
potential form, which alleviates the problem with singu-
larities [100]:
(Z)
ij
=
__
^ zz
.
(e
i
h
j
)dA
2jk
4pZ
F
__
e
i
.
__
e
jkR
R
e
j
/
dA
/
dA
2
jk4pZ
F
__
( ^ zz
.
rot
t
e
i
)
__
e
jkR
R
( ^ zz
.
rot
t
/
e
j
/
)dA
/
dA:
(28)
The principle of the adaptive integral method [9498]
consists in the representation of distributed sources by an
[a
i
]
[b
i
]
Feed
network
[S
F
]
Radiation
network
[S
H
]
[b
2
i
] [b
2
o
]
[a
H
]
[a
P
i
]
[b
P
i
]
[a
P
o
]
[b
P
o
]
[b
H
]
[a
2
i
] [a
2
o
]
[a
1
o
]
[b
1
i
]
[a
1
i
]
[b
1
o
]
:
:
:
:
:
:
:
.
.
.
.
.
Slot 1
[S
1
]
Slot 2
[S
2
]
Slot P
[S
P
]
Figure 6. Computation scheme of external and internal cou-
plings of waveguide fed arrays.
Figure 7. Application of the MM/FE method. Rectangular wa-
veguide with a ridge discontinuity [101]. Waveguide a=
19.05mm, b =9.524mm; ridge w=5.08mm, thickness s =
1.016mm, d=1.905mm. Measurements [101]. Hybrid MM/FE
method CPU speed 500 frequency points o2s (2-GHz P4 PC, 41
modes): 4 ms per frequency point.
Figure 8. Accuracy and dynamic range of the MM/FE method.
Optimized C-band wafe-iron lter; built without tuning screws.
CPU speed o0.1 s per frequency point, 2-GHz P4 PC, 152 modes.
[Measured results (boldface return-loss curve in small window)
courtesy of Bosch Telecom (now TESAT Spacecom GmbH), F.-J.
Goertz, D. Wolk, Backnang, Germany, and Dr. D. Schmitt (now
European Space Agency (ESA), Noordwijk, The Netherlands).]
Figure 9. Application of the MM/FE method for direct combina-
tion of FE eigenvector sections. Septum polarizer, dimensions
according to Ref. 102 (waveguide a=b =48.26mm). Power split of
square waveguide orthogonal modes to rectangular waveguide
output ports, H01 return loss, isolation, H10 return loss. Small
window: axial ratio. CPU time 34s, 500 frequency points, with
o0.1s per frequency point (14s for eigenvalue problemof the four
ridge sections, 20s MM dynamic calculation for the 500 frequency
points), 2-GHz P4 PC, 120 modes.
Figure 10. Extending the MM/FE and MM/MoM methods to
coaxial structures. Typical feed elements for combline lters:
direct post, loop, disk, slope, sheetmetal coupling, transformer
coupling.
equidistant mesh of point sources. The electric eld
strength of the jth wave mode is, for instance, given
by
e
xj
=
n
L
x
j
n d( r
n
);
e
yj
=
n
L
y
j
n
d( r
n
);
^ zz
.
rot
t
e
j
=
n
L
f
j
n
d( r
n
)
(29)
The weights for mesh points are determined by the
identity of the moments. For an aperture with a partial
grid of N
x
N
y
points, we obtain by the requirement
Nx1
nx =0
N
y1
ny =0
L
x
nxny
(x
n
x
x
0
)
m
x
(y
n
y
y
0
)
m
y
=
__
Ap
j
e
x
(x x
0
)
mx
(y y
0
)
my
dA
(30)
\0 _ m
x
oN
x
. 0 _ m
y
oN
y
a linear equation system for the weights of the x compo-
nents of the electric eld strength in the equivalent source
mesh. For ~rr
0
the centerpoint of the aperture leads to
minimum numerical effort, although other choices are
also possible. This equivalent representation is inserted
in (28), and one obtains
(Z
AIM
)
ij
=
2jk
4pZ
F
n
/
L
x
in
g(~rr
n
~rr
n
/ )L
x
jn
/
2jk
4pZ
F
n
/
L
y
in
g(~rr
n
~rr
n
/ )L
y
jn
/
2
jk4pZ
F
n
/
L
f
in
g(~rr
n
~rr
n
/ )L
f
jn
/
(31)
For this expression, application of the fast Fourier
transformation (FFT) is possible. For the near couplings,
the accuracy of the expression has to be improved by
taking into account corrections due to the method of
moments.
The matrix is not stored explicitly; hence, an iterative
solver has to be applied, which is based on a fast matrix
Figure 11. Milled rectangular combline lter for 2.32GHz mid-
band frequency without tuning screws. Dimensions (mm): wave-
guide housing a=22.816, b =21.729; feed SMA, inner conductor
height 4.5, distance wall to center rst post 10.4, posts 8.928
8.928, heights 20.35 (1st), 19.96 (2nd), distances (post plane to
post plane) 20.47, 23.76 (Mesh renement in post section factor 5).
CPU time 20s, 200 frequency points (0.1s per frequency
point),125 modes, 2-GHz P4 PC. (This gure is available in full
color at http://www.mrw.interscience.wiley.com/erfme.)
0
10
20
30
40
50
60
70
80
26 26.2 26.4 26.6 26.8 27 27.2 27.4 27.6
f in GHz
s11(h 10)(h10)
s21(h 10)(h10)
s22(h 10)(h10)
Scattering parameters of "WR_38_oversized_3_cavity_filter"
Figure 12. Advanced optimized WR-38 dual-mode lter utilizing
TE
20
-mode excitation in oversized cavities [37]. Application of the
MM/transnite element method for arbitrarily shaped planar
structures. CPU time 60s for 500 frequency points, 34 modes, 2-
GHz P4 PC. (This gure is available in full color at http://
www.mrw.interscience.wiley.com/erfme.)
vector product
(Z : M)
i
=(Z
MoM
: M)
i

2jk
4pZ
F
n
L
x
in
:
1
:{g(~rr
m
)]:
j
L
x
jm
M
j
_ _ _ _
n
2jk
4pZ
F
n
L
y
in
:
1
:{g( r
m
)]:
j
L
y
jm
M
j
_ _ _ _
n
2
jk4pZ
F
n
L
f
in
:
1
:{g(~rr
m
)]:
j
L
f
jm
M
j
_ _ _ _
n
(32)
3. TYPICAL APPLICATION EXAMPLES
The extremely high CPU speed that can be achieved
by applying hybrid methods is demonstrated by a
simple example, a rectangular waveguide with a ridge
discontinuity (Fig. 7), a structure originally reported in
Ref. 101. The hybrid MM/FE method takes less than 2s for
1
4
2
3
w
d
1
t
p
k
0
0
6.5 7
IS
11
I (FE / MM)
IS
33
I (FE / MM)
IS
44
I (FE / MM)
IS
11
I (FDTD)
7.5
Frequency [GHz]
Magic T4
I
S
i
j
I

/
d
B
S
i
j
[
d
B
]
8 8.5 9 9.5 10 10.5
f/GHz
11
5
10
15
20
25
30
35
40
45
50
7 8 9 10 11
10
20
50
30
40
S11
S33
S44
Figure 13. Application of the hybrid MM/MoM method. Com-
pensated magic T, dimensions (mm): waveguide 12.725.4, l =
8.225, d=3.175, k =3.556, w=8.382, t =0.7937, p=0, height of
post element 12.7. Structure: triangular mesh. MM/MoM results,
comparison with MM/FE [59] and FDTD [109] results.
0
5
10
15
20
25
30
35
40
6.6 6.8 7 7.2 7.4 7.6 7.8 8.2
f in GHz
8.4 8.6 8.8 9.2 9.4 9.6 9.8 10 9 8
Scattering parameters of "WR112_asymn_Tee_stepped_omt_opt_opt3"
|
S
|

i
n

d
B
s11 (h 10)(h 10)
s21 (h 10)(h 10)
s11 (h 01)(h 01)
s31 (h 01)(h 01)
Figure 14. Optimized broadband WR-112 taper/branching ortho-
mode transducer (OMT) with stepped T junction in the branching
region for improved bandwidth behavior. Return loss of both
polarizations. (This gure is available in full color at http://
www.mrw.interscience.wiley.com/erfme.)
500 frequency points (i.e., B4ms per frequency point, 41
modes, 2-GHz P4 PC), with the time required for the 2D
FE eigenvalue solution included.
The excellent accuracy, high dynamic range, and high
CPU speed achieved by the MM/FE method are demon-
strated by an optimized and measured C-band wafe-iron
lter (WR-229) Fig. 8 for space applications. The lter has
been fabricated without any tuning screws. The advanced
lter utilizing a large gap width for high-power applica-
tions has been optimized by applying the evolution strat-
egy [30,54], and has been multipacting tested with
41.5 kW by the European Space Agency (ESA).
The next example, a septum polarizer (Fig. 9), relates
to a typical MM/FE application, where the ridge sections
(i.e., FE eigenvector sections) are directly connected to
each other. Henceas outlined in Section 2for the
corresponding coupling coefcients, calculations via area
integrals are preferred to line integrals in this case. The
dimensions are chosen according to Ref. 102, where cross
section eld-matching has been applied for solution of the
eigenvalue problem [103], which is considerably more
time-consuming than applying the MM/FE technique
because of the search algorithm involved in the MM
eigenvalue solution problem.
The MM/FE and MM/MoM methods have been ex-
tended for inclusion of typical coaxial feed structures in
Fig. 10, where TEM mode expressions (Sections 2.1,2.3)
have to be taken into account. This achieves the possibility
of fast CAD and optimization of combline and interdigital
lters/diplexers with rectangular post elements. The rec-
tangular post structure in combline/interdigital lters can
be utilized for low-cost fabrication, such as by computer-
controlled milling techniques [110,112] directly, or the
rectangular structure can be converted into corresponding
circular posts by applying well-known formulas, such as
those in Ref. 111, of identical characteristic impedances
(square/circular rod between two plates). This yields good
starting points for further optimizations by applying, for
example, the MM/FD or MM/MoM methods for circular
posts.
For accuracy and computing speed verication, a sim-
ple direct post-fed four-resonator rectangular combline
lter for 2.32 GHz midband frequency has been optimized
and fabricated by a computer-controlled milling technique
without any tuning screws (Fig. 11). Advanced CAD tools
based on these hybrid techniques [68], allow the corre-
sponding data le to be directly and conveniently loaded
from the graphical output after optimization. Very good
(a)
(b)
0
10
30
40
50
60
70
80
1.9 1.91 1.92 1.93 1.94 1.95
f in GHz
1.96 1.97
1.98 1.99 2
20
|
S
|

i
n

d
B
Scattering parameters of "combcomtoppost_corrugated_23nov03"
Figure 15. Advanced combline lter with rounded cavities and
corrugated reentrant resonators directly optimized toward the
goal of 23dB return loss: (a) structure; (b) S-parameter results.
(This gure is available in full color at http://www.mrw.
interscience.wiley.com/erfme.)
0
10
20
30
40
|
S
|

i
n

d
B
50
60
70
9.5 9.6 9.7 9.8 9.9 10 10.1 10.2
f in GHz
10.3 10.4 10.5 10.6 10.7 10.8 10.9 11
Scattering parameters of "ex_filter_cx_opt"
s11(t 00)(t 00)
s21(t 00)(t 00)
s22(t 00)(t 00)
Figure 16. WR-90 H-plane iris seven-resonator lter with
rounded corners, fed by topwall SMA connectors. Optimization
of the whole lter including the effect of the coax feeds. (This
gure is available in full color at http://www.mrw.interscience.
wiley.com/erfme.)
agreement between theory and measurements can be
achieved. For the MM/FE calculation, 125 modes have
been taken into account; the CPU speed is 0.1s per
frequency point on a 2-GHz P4 PC, which underlines,
again, the efciency of the hybrid MM/FE method also for
this class of lters, making their optimization a straight-
forward task.
An application example for the MM/transnite FE
method for planar waveguides with arbitrary contour is
the oversized cavity lter shown in Fig. 12. The TE
20
mode
excited by offset irises can be controlled to achieve a dual-
mode behavior with stopband poles [37].
The application of the MM/MoM method for 3D wave-
guide structures is demonstrated in Fig. 13 with the
example of a magic T compensated by a post and an iris.
The CPUspeed is about 8s per frequency point on a 2-GHz
P4 PC. Good agreement with reference calculations using
the FDTD method [59] and the MM/FE generalized scat-
tering matrix separation (GSMS) technique [105] can be
achieved. In this way, rather complicated 3D structures,
such as wafe-iron lters with round teeth [76] can be
modeled.
Figure 14 demonstrates the application of the MM/
MoM technique for an advanced broadband taper/branch-
ing orthomode transducer (OMT) with a tapered branch-
ing section. The B20% bandwidth achieved compares well
with the usual 10% of more standard designs [104].
The MM/MoM technique can also advantageously be
applied for more complicated components such as a comb-
line lter with rounded cavities and reentrant resonators
for convenient tuning (Fig. 15).
An application example for the hybrid MM/transnite
FE/MoM technique with high practical importance is
shown in Fig. 16: a WR-90 H-plane iris lter with rounded
corners, fed by a coax standard 50 O subminiature version
A (SMA) connector transition from the topwall. Because of
the high CPUspeed (1s per frequency point, 2-GHz P4 PC,
102 modes), the lter can be directly optimized taking the
effect of the coax feed rigorously into account.
An antenna application example of the MM/MoM tech-
nique is the design of an axially corrugated (choked)
Figure 17. Axially corrugated (choked) horn optimized for broad-
band low cross-polarization. Lower plot shows the peak cross-
polarization level as a function of frequency. (This gure is available
in full color at http://www.mrw.interscience.wiley.com/erfme.)
Matched
Slot
0
0 20 40 60 80 100 120 140 160 180
10
20
30
40
50
10
0
0 20 40 60
theta/[deg]
d
B
80 100 120 140 160 180
20
30
40
50
60
TE
10
|E|
|E|
max
dB
/
MLFMM
AIM
[107]
Measured
Figure 18. Traveling-wave slot array antenna with 21 slots
according to Elliott [107]. Dimensions given in Ref. 102. The
authors own calculation with fast MoM techniques: adaptive
integral method (AIM) and multilevel fast multipole method
(MLFMM). Comparison with reference calculations [107] and
measurements [106]. (This gure is available in full color at
http://www.mrw.interscience.wiley.com/erfme.)
horn (Fig. 17). Because of the consequent utilization of the
hybrid technique, the CPU speed achieved (B5s per
frequency point, 3-GHz P4 PC) allows the convenient
direct optimization toward desired broadband low cross-
polarization levels.
Figure 18 shows a traveling-wave slot array antenna
with 21 slots in the broad waveguide wall [106]. The
utilization of the MM/MoM AIM and MLFMM techniques
including the internal coupling provides results that agree
well with measurements reported in Ref. 106 and with
reference calculations using the more standard MoM
spectral-domain technique [107].
4. CONCLUSION
Fast EM-based CAD tools yielding accurate, optimized
designs within adequately short response times are indis-
pensable for solving challenging optimization tasks of
advanced modern microwave and millimeter-wave compo-
nent applications. This goal is achieved by hybrid CAD
techniques, which typically go beyond the possibilities of
single methods, by expanding the capabilities and enhan-
cing the efciency. The hybrid CAD techniques described
here combine advantageously the efciency of the mode-
matching (MM) method with the exibility of nite-ele-
ment (FE), nite-difference (FD), and method-of-moments
(MoM) techniques. Typical application examples demon-
strate their versatility and high calculation speed. The
accuracy is veried by available measurements and re-
ference calculations.
BIBLIOGRAPHY
1. J. E. Bracken, D.-K. Sun, and Z. J. Cendes, S-domain methods
for simultaneous time and frequency characterization of
electromagnetic devices, IEEE Trans. Microwave Theory
Tech. MTT-46:12771290 (Sept. 1998).
2. A. S. Rong, H. Yang, X. H. Chen, and A. Cangellaris, Efcient
FD-TD modeling of irises/slots in microwave structures and
its application to the design of combline lters, IEEE Trans.
Microwave Theory Tech. MTT-49:22662275 (Dec. 2001).
3. W. C. Chew, J.-M. Jin, C.-C. Lu, E. Michielssen, and J. M.
Song, Fast solution methods in electromagnetics, IEEE Trans.
Anten. Propag. AP-45:533543 (March 1997).
4. J.-S. Zhao, W. C. Chew, C.-C. Lu, E. Michielssen, and J. Song,
Thin-stratied medium fast-multipole algorithm for micro-
strip structures, IEEE Trans. Microwave Theory Tech. MTT-
46:395403 (April 1998).
5. F. Ling, C.-F. Wang, and J.-M. Jin, An efcient algorithm for
analyzing large-scale microstrip structures using adaptive
integral method combined with discrete complex-image
method, IEEE Trans. Microwave Theory Tech. MTT-48:832
839 (May 2000).
6. A. F. Peterson, S. L. Ray, and R. Mittra, Computational
Methods for Electromagnetics, IEEE Press, New York, 1998.
7. W. C. Chew, J.-M. Jin, E. Michielssen, and J. Song, Fast and
Efcient Algorithms in Computational Electromagnetics, Ar-
tech House, Boston, 2001.
8. R. Coifman, V. Rokhlin, and S. Wandzura, The fast multipole
method for the wave equation: A pedestrian prescription,
IEEE Anten. Propag. Mag. 35:712 (June 1996).
9. E. Bleszynski, M. Bleszynski, and J. Jaroszevicz, AIM: Adap-
tive integral method for solving large-scale electromagnetic
scattering problems, Radio Sci. 31:12251251 (Sept.Oct.
1996).
10. M. F. Gyure and M. A. Stalzer, A prescription for the multi-
level Helmholtz FMM, IEEE Comput. Sci. Eng. 5:3547
(JulySept. 1998).
11. R. Lotz and F. Arndt, Locally conformed subgrid FD-FD
technique for the analysis of 3D waveguide structures with
curved metallic objects, 1999 Int. Microwave Symp. Digest,
June 1999, Vol. 3, pp. 12771280.
12. F. Arndt, R. Lotz, and J. Ritter, Advanced FD-TD techniques
for the CAD of microwave components, Electromagnetics
23:153168 (Feb.March 2003).
13. G. A. Thiele and T. H. Newhouse, A hybrid technique for
combining moments methods with the geometrical theory of
diffraction, IEEE Trans. Anten. Propag. AP-23:6269 (Jan.
1975).
14. W. D. Burnside and P. H. Pathak, A summary of hybrid
solutions involving moment methods and GTD, in B. J. Strait,
ed., Applications of the Method of Moments to Electromag-
netic Fields, SCEEE Press, St. Cloud, FL, 1980.
15. J. J. H. Wang, Generalized Moment Methods in Electromag-
netics, Wiley, New York, 1991.
16. G. Tiberi, A. Monorchio, G. Manara, and R. Mittra, Hybridiz-
ing asymptotic and numerically rigorous techniques for
solving electromagnetic scattering problems using the char-
acteristics basis functions, Int. Antenna and Propagation
Symp. Digest, June 2003, Vol. 3, Session 114.
17. J. Liu and J. M. Jin, A novel hybridization of higher order
nite element and boundary integral methods for electro-
magnetic scattering and radiation problems, IEEE Trans.
Antennas Propag. AP-49:17941806 (Dec. 2001).
18. H. Rogier, F. Olyslager, and D. de Zutter, A hybrid nite
element integral equation approach for the eigenmode ana-
lysis of complex anisotropic dielectric waveguides, Radio Sci.
31:9991010 (JulyAug. 1996).
19. W. C. Hahn, A new method for the calculation of cavity
resonators, J. Appl. Phys. 12:6268 (Jan. 1941).
20. J. R. Whinnery and H. W. Jamieson, Equivalent circuits for
discontinuities in transmission lines, Proc. IRE. 32:98116
(Feb. 1944).
21. P. J. B. Clarricoats and K. R. Slinn, Numerical solution of
waveguide-discontinuity problems, Proc. IEE. 114:878886
(July 1967).
22. A. Wexler, Solution of waveguide discontinuities by modal
analysis, IEEE Trans. Microwave Theory Tech. MTT-15:508
517 (Sept. 1967).
23. G. L. James, Analysis and design of TE11-to-HE11 corrugated
cylindrical waveguide mode converters, IEEE Trans. Micro-
wave Theory Tech. MTT-29:10591066 (Oct. 1981).
24. H. Patzelt and F. Arndt, Double-plane steps in rectangular
waveguide and their application for transformers, irises and
lters, IEEE Trans. Microwave Theory Tech. MTT-30:771
777 (May 1982).
25. F. Arndt, D. Ellermann, H. W. Haeusler, and J. Strube, Field
theory analysis and numerical synthesis of symmetrical
multiple-branch waveguide couplers, Frequenz 36:262266
(Oct. 1982).
26. F. Arndt, J. Bornemann, D. Grauerholz, and R. Vahldieck,
Theory and design of low-insertion loss n-line lters,
IEEE Trans. Microwave Theory Tech. MTT-30:155163 (Feb.
1982).
27. R. Vahldieck, J. Bornemann, F. Arndt, and D. Grauerholz,
Optimized waveguide E-plane metal insert lters for milli-
meter-wave applications, IEEE Trans. Microwave Theory
Tech. MTT-31:6569 (Jan. 1983).
28. F. Arndt, B. Koch, H.-J. Orlok, and N. Schroeder, Field theory
design of rectangular waveguide broad-wall metal-insert slot
couplers for millimeter-wave applications, IEEE Trans. Mi-
crowave Theory Tech. MTT-33:95104 (Feb. 1985).
29. U. Tucholke, F. Arndt, and T. Wriedt, Field theory design of
square waveguide iris polarizers, IEEE Trans. Microwave
Theory Tech. MTT-34:156160 (Jan. 1986).
30. H. Schmiedel and F. Arndt, Field theory design of rectangular
waveguide multiple-slot narrow wall couplers, IEEE Trans.
Microwave Theory Tech. MTT-34:791798 (July 1986).
31. J. Dittloff, F. Arndt, and D. Grauerholz, Optimum design of
waveguide E-plane stub-loaded phase shifters, IEEE Trans.
Microwave Theory Tech. MTT-36:583587 (March 1988).
32. J. Dittloff and F. Arndt, Rigorous design of septate E-plane
multiplexers with printed circuit elements, 1988 IEEEMTT-S
Int. Microwave Symp. Digest, New York, May 1988, pp. 431
434.
33. J. Dittloff and F. Arndt, Computer-aided design of slit-coupled
H-plane T-junction diplexers with E-plane metal-insert l-
ters, IEEE Trans. Microwave Theory Tech. MTT-36:1833
1840 (Dec. 1988).
34. F. Alessandri, G. Bartolucci, and R. Sorrentino, Admittance
matrix formulation of waveguide discontinuity problems:
Computer-aided design of branch-guide directional couplers,
IEEE Trans. Microwave Theory Tech. MTT-36:394403 (Feb.
1988).
35. F. Arndt, J. Dittloff, U. Papziner, D. Fasold, N. Nathrath, and
H. Wolf, Rigorous eld theory design of compact and light-
weight broadband diplexers for satellite communication sys-
tems, Proc. 19th European Microwave Conf., London, Sept.
1989, pp. 12141219.
36. R. Sorrentino, ed., Numerical Methods for Passive Microwave
and Millimeter Wave Structures, IEEE Press, New York,
1989.
37. F. Arndt, T. Duschak, U. Papziner, and P. Rolappe, Asym-
metric iris coupled cavity lters with stopband poles, 1990
MTT-S Int. Microwave Symp. Digest, Dallas, May 1990, pp.
215218.
38. U. Papziner and F. Arndt, Field theoretical computer-aided
design of rectangular and circular iris coupled rectangular or
circular waveguide cavity lters, IEEE Trans. Microwave
Theory Tech. MTT-41:462471 (Feb. 1993).
39. M. Guglielmi, G. Gheri, M. Calamia, and G. Pelosi, Rigorous
multimodule network representation of inductive steps, IEEE
Trans. Microwave Theory Tech. MTT-42:317326 (Feb. 1994).
40. A. P. Orfanidis, G. A. Kyriacou, and J. N. Sahalos, A mode-
matching technique for the study of circular and coaxial
waveguide discontinuities based on closed-form coupling in-
tegrals, IEEE Trans. Microwave Theory Tech. MTT-48:880
883 (May 2000).
41. F. Alessandri, M. Giordano, M. Guglielmi, G. Martirano, and
F. Vitulli, A new multiple-tuned six-port Riblet-type direc-
tional coupler in rectangular waveguide, IEEE Trans. Micro-
wave Theory Tech. MTT-51:14411448 (May 2003).
42. J. Kocbach and K. Folgero, Design procedure for waveguide
lters with cross-couplings, 2002 Int. Microwave Symp. Di-
gest, June 2002, Vol. 3, pp. 14491452.
43. L. Accatino and M. Mongiardo, Hybrid circuit-full-wave
computer-aided design of a manifold multiplexers without
tuning elements, IEEE Trans. Microwave Theory Tech. MTT-
50:20442047 (Sept. 2002).
44. P. Matras, R. Bunger, and F. Arndt, Modal scattering matrix
of the general step discontinuity in elliptical waveguides,
(March 1997).
45. M. Mongiardo, P. Russer, C. Tomassoni, and L. Felsen,
Analysis of N-furcation in elliptical waveguides via the gen-
eralized network formulation, IEEETrans. Microwave Theory
Tech. MTT-47:24732478 (Dec. 1999).
46. E. Ku hn and V. Hombach, Computer-aided analysis of corru-
gated horns with axial ring or ring-loaded radial slots, Proc.
Int. Conf. Antennas Propagation (ICAP 83)Part I, UK,
March 1983, pp. 293296.
47. K. Wolff, T. Wriedt, F. Arndt, and U. Tucholke, Hybrid eld
design of square potter horns, Proc. Int. Conf. Antennas
Propagation (ICAP 87)Part I, UK, March 1987, pp. 210
213.
48. T. Wriedt, K.-H. Wolff, F. Arndt, and U. Tucholke, Rigorous
hybrid eld theoretic design of stepped rectangular wave-
guide mode converters including the horn transitions into
half-space, IEEE Trans. Anten. Propag. AP-37:780790 (June
1989).
49. K. Liu, C. A. Balanis, and C. R. Birtcher, Analysis of pyrami-
dal horn antennas using moment methods, IEEE Trans.
Anten. Propag. AP-41:13791389 (Oct. 1993).
50. R. Bunger, R. Beyer, and F. Arndt, Rigorous combined mode-
matching integral equation analysis of horn antennas with
arbitrary cross-section, IEEE Trans. Anten. Propag. AP-
47:16411648 (Nov. 1999).
51. R. Beyer and F. Arndt, Field theory design of circular
waveguide dual-mode lters by a combined mode matching
nite element method, EuMC Int. Microwave Symp. Digest,
Cannes, France, Sept. 1994, pp. 437442.
52. R. Beyer and F. Arndt, Efcient modal analysis of waveguide
lters including the orthogonal mode coupling elements by an
MM/FE method, IEEE Microwave Guided Wave Lett. 5:13
(Jan. 1995).
53. J. R. Montejo-Garai and J. Zapata, Full-wave design and
realization of multicoupled dual-mode circular waveguide
lters, IEEE Trans. Microwave Theory Tech. MTT-43:1290
1297 (June 1995).
54. F. Arndt, R. Beyer, J. M. Reiter, T. Sieverding, and T. Wolf,
Automated design of waveguide components using hybrid
mode-matching/numerical EM building blocks in optimiza-
tion-oriented CAD frame-worksstate-of-the-art and recent
advances, IEEE Trans. Microwave Theory Tech. MTT-45:747
760 (May 1997).
55. F. Arndt, R. Beyer, W. Hauth, D. Schmitt, and H. Zeh,
Cascaded wide stop band wafe-iron lter designed with a
MM/FE CAD method, EuMC Int. Microwave Symp. Digest,
Munich, Germany, Oct. 1999, pp. 186189.
56. F. Arndt and J. Brandt, Fast hybrid MM/FE CAD tool for the
design and optimization of advanced evanescent mode lters,
MIOP Int. Microwaves and Optronics Symp. Digest, Stutt-
gart, Germany, May 2001, pp. 369372.
57. F. Arndt and J. Brandt, MM/FE CAD and optimization of
rectangular combline lters, EuMC Int. Microwave Symp.
Digest, Milan, Italy, Sept. 2002, pp. 335338.
58. J. Rubio, J. Arroyo, and J. Zapata, Analysis of passive micro-
wave circuits by using a hybrid 2-D and 3-D nite-element
mode-matching method, IEEE Trans. Microwave Theory
Tech. MTT-47:17461749 (Sept. 1999).
59. R. Beyer and F. Arndt, Efcient MM/FEG SMS technique for
the CAD of broadband lateral coax feeds in rectangular
waveguide, IEEE MTT-S Int. Microwave Symp. Digest, Bos-
ton, June 2000, pp. 109112.
60. D. Arena, M. Ludovico, G. Manara, and A. Monorchio,
Analysis of waveguide discontinuities using edge elements
in a hybrid mode matching/nite elements approach,
IEEE Microwave Wireless Compon. Lett. 11:379381 (Sept.
2001).
61. V. Crino, C. Tomassoni, and M. Mongiardo, Line-integral
formulation of the hybrid MM/FEM technique, in IEEE
MTT-S Int. Microwave Symp. Digest, Seattle, WA, June
2002, pp. 20332036.
62. F. Alimenti, P. Mezzanotte, L. Roselli, and R. Sorrentino,
Efcient analysis of waveguide components by FDTD com-
bined with time domain modal expansion, IEEE Microwave
Guided Wave Lett. 5:351353 (Oct. 1995).
63. M. Righi, M. Mongiardo, R. Sorrentino, and W. Hoefer,
Efcient TLM diakoptics for separable structures, Proc. Int.
MTT-S, Atlanta, GA, June 1993, Vol. 1, pp. 425428.
64. R. Lotz and F. Arndt, FD-FD GSM technique for the CAD and
optimization of combline lters, IEEE MTT-S Int. Microwave
Symp. Digest, Phoenix, May 2001, pp. 12531256.
65. F. Arndt and A. Enneking, Hybrid MM/FE/MoMtechnique for
waveguide horn antennas and slot arrays, Proc. Int. Conf.
Electromagnetics in Advanced Applications (ICEAA01), Tor-
ino, Italy, Sept. 2001, pp. 569572.
66. F. Arndt, J. Brandt, and J. Ritter, MM/FE/FD CAD method for
the optimization of waveguide lters including structures of
arbitrary shape and coax feeds, EuMC Int. Microwave Symp.
Digest, London, Sept. 2001, pp. 319322.
67. F. Arndt, J. Brandt, V. Catina, A. Enneking, and J. Ritter, Fast
hybrid MM/FE/TransFE/MoM/FD CAD of waveguide compo-
nents and aperture antennas, Proc. 18th Annual Review of
Progress in Applied Computational Electromagnetics (ACES),
Monterey, CA, March 2002, pp. 1-151-23.
68. MiG, A CAD tool utilizing fast hybrid MM/FE/MoM/FD
techniques, Microwave J. (cover feature article) 45:178182
(Sept. 2002).
69. J. Strube and F. Arndt, Rigorous hybrid-mode analysis of the
transition from rectangular waveguide to shielded dielectric
image guide, IEEE Trans. Microwave Theory Tech. MTT-
33:391401 (May 1985).
70. W. Schroeder and M. Guglielmi, Boundary integral equation
approach to multi-mode Y-matrix characterization of multi-
ridged sections in circular waveguide, IEEE MTT-S Int.
Microwave Symp. Digest, San Fracisco, June 1996, pp.
18491852.
71. J. Jin, The Finite Element Method in Electromagnetics. Wiley,
New York, 1993.
72. O. C. Zienkievicz, The Finite Element Method, McGraw-Hill,
Berkshire, UK, 1977.
73. Y. Saad, Numerical methods for large eigenvalue problems,
in Algorithms and Architectures for Advanced Scientic
Computing. Manchester Univ. Press, Manchester, UK,
1992.
74. Z. J. Cendes and J. F. Lee, The transnite element method for
modeling MMIC devices, IEEE Trans. Microwave Theory
Tech. MTT-36:16391649 (Dec. 1988).
75. A. Joestingmeier and A. S. Omar, Analysis of inhomogen-
eously lled cavities coupled to waveguides using the VIE
formulation, IEEE Trans. Microwave Theory Tech. MTT-
41:12071214 (June/July 1993).
76. R. Bunger and F. Arndt, Moment-method analysis of arbi-
trary 3D metallic N-port waveguide structures, IEEE Trans.
Microwave Theory Tech. MTT-48:531537 (April 2000).
77. J. R. Mautz and R. F. Harrington, A generalized network
formulation for aperture problems, IEEE Trans. Anten. Pro-
pag. AP-24:870873 (Nov. 1976).
78. H. Auda and R. F. Harrington, A moment solution for
waveguide junction problems, IEEE Trans. Microwave Theory
Tech. MTT-31:515520 (July 1983).
79. S. M. Rao, D. R. Wilton, and A. W. Glisson, Electromagnetic
scattering by surfaces of arbitrary shape, IEEE Trans. Anten.
Propag. AP-30:409418 (May 1982).
80. J. Ritter and F. Arndt, A generalized 3D subgrid technique for
the nite-difference time domain method, 1997 Int. Micro-
wave Symp. Digest, June 1997, Vol. 3, pp. 15631566.
81. R. Lotz, J. Ritter, and F. Arndt, 3D subgrid technique for the
nite difference method in the frequency domain, 1998 Int.
Microwave Symp. Digest, June 1998, Vol. 3, pp. 17391742.
82. S. S. Zivanovic, K. S. Yee, and K. K. Mei, A subgridding
method for the time-domain nite-difference method to solve
Maxwells equations, IEEE Trans. Microwave Theory Tech.
MTT-39:471479 (March 1991).
83. I. S. Kim and W. J. R Hoefer, A local mesh renement
algorithm for the time domain nite difference method using
Maxwells curl equations, IEEE Trans. Microwave Theory
Tech. MTT-38:812815 (June 1990).
84. S. Kapoor, Sub-cellular technique for nite-difference time-
domain method, IEEE Trans. Microwave Theory Tech. MTT-
45:673677 (May 1997).
85. S. Dey and R. Mittra, A locally conformal nite-difference
time-domain (FD-TD) algorithm for modeling three-dimen-
sional perfectly conducting objects, IEEE Microwave Guided
Wave Lett. 7:273275 (Sept. 1997).
86. N. Kaneda, B. Houshmand, and T. Itoh, FDTD analysis of
dielectric resonators with curved surfaces, IEEE Trans. Mi-
crowave Theory Tech. MTT-45:16451649 (Sept. 1997).
87. Y. Hua and T. K. Sarkar, Matrix pencil method for estimating
parameters of exponentially damped/undamped sinusoids in
noise, IEEE Trans. Acoust. Speech Signal Process. 38:814824
(May 1990).
88. J. Ritter and F. Arndt, Efcient FD-TD/matrix pencil method
for the full-wave scattering parameter analysis of waveguid-
ing structures, IEEE Trans. Microwave Theory Tech. MTT-
44:24502456 (Dec. 1996).
89. J. M. Song and W. C. Chew, Multilevel fast-multipole algo-
rithm for solving combined eld integral equations of electro-
magnetic scattering, Microwave Opt. Tech. Lett. 10 (1):1419
(Sept. 1995).
90. R. Bunger and F. Arndt, MLFMM analysis of ridged wave-
guide horns, Proc. Millennium Conf. Antennas Propagation
(AP2000), Davos, Switzerland, April 2000, p. 1418.
91. I. Rullhusen and F. Arndt, MLFMA analysis of coupled
rectangular apertures, Proc. JINA Conf., Nice, France, Nov.
2002, pp. 459462.
92. I. Rullhusen and F. Arndt, Efcient MLFMA analysis of
coupled rectangular and circular apertures in a nite screen,
IEEE MTT-S Int. Microwave Symp. Digest, Philadelphia,
June 2003, pp. 19711974.
93. I. Rullhusen and F. Arndt, MLFMA analysis of nite aperture
arrays including reector, IEEE AP-S Int. Symp. Digest,
Columbus, OH, Paper 125.8, June 2003.
94. S. S. Bindiginawale, J. L. Volakis, and H. Anastassin, Scatter-
ing from planar structures containing small features using
the adaptive integral method (AIM), IEEE Trans. Anten.
Propag. AP-46:18671878 (Dec. 1998).
95. E. Bleszynski, M. Bleszynski, and T. Jaroszevicz, AIM: Adap-
tive integral method for solving large scale electromagnetic
scattering and radiation problems, Radio Sci. 31:12251251
(Sept./Oct. 1996).
96. F. Ling, C. F. Wang, and J. M. Jin, Application of the adaptive
integral method to scattering and radiation analysis of arbi-
trarily shaped planar structures, J. Electromagn. Waves Appl.
12:10211037 (1998).
97. F. Ling, C. F. Wang, and J. M. Jin, An efcient algorithm for
analyzing large-scale microstrip structures using adaptive
integral method combined with discrete complex image
method, IEEE Trans. Microwave Theory Tech. MTT-48:832
839 (May 2000).
98. I. Rullhusen and F. Arndt, AIM analysis of slot arrays,
Proc. Int. Conf. Electromagnetics in Advanced Applica-
tions (ICEAA03), Torino (Turin, Italy), Sept. 2003, pp.
321324.
99. R. Bunger, R. Beyer, and F. Arndt, Rigorous combined mode-
matching integral equation analysis of horn antennas with
arbitrary cross section, IEEE Trans. Anten. Propag. AP-
47:16411648 (Nov. 1999).
100. D. R. Wilton, S. M. Rao, A. W. Glisson, D. H. Schaubert,
O. M. Al-Bundak, and C. M. Butler, Potential integrals for
uniform and linear source distributions on polygonal and
polyhedral domains, IEEE Trans. Anten. Propag. AP-
32:276281 (March 1984).
101. R. Mansour, R. S. K. Tong, and R. H. McPhie, Simplied
description of the eld distribution in nlines and ridge
waveguides and its application to the analysis of E-plane
discontinuities, IEEE Trans. Microwave Theory Tech. MTT-
36:18251832 (Dec. 1988).
102. J. Bornemann and V. A. Labay, Ridge waveguide pola-
rizer with nite and stepped thickness septum, IEEE
Trans. Microwave Theory Tech. MTT-43:17821787 (Aug.
1995).
103. J. Bornemann and F. Arndt, Transverse resonance, standing
wave, and resonator formulation of the ridge waveguide
eigenvalue problem and its application to the design of E-
plane nned waveguide lters, IEEE Trans. Microwave The-
ory Tech. MTT-38:11041113 (Aug. 1990).
104. J. Uher, J. Bornemann, and U. Rosenberg, Waveguide Com-
ponents for Antenna Feed Systems: Theory and CAD, Artech
House, Boston, 1993.
105. J. Ritter and F. Arndt, Efcient FD-TD/matrix pencil method
for the full-wave scattering parameter analysis of waveguid-
ing structures, IEEE Trans. Microwave Theory Tech. MTT-
44:24502456 (Dec. 1996).
106. R. S. Elliott, On the design of traveling-wave-fed longitudi-
nal shunt slot arrays, IEEE Trans. Anten. Propag. AP-
27:717720 (Sept. 1979).
107. A. Enneking, R. Beyer, and F. Arndt, Rigorous analysis of
large nite waveguide slot arrays including the mutual
internal and external higher-order mode coupling, AP-S Int.
Symp. Digest, July 2000, pp. 7477.
108. R. Lotz, J. Ritter, and F. Arndt, 3D subgrid technique for
the nite difference method in the frequency domain, 1998
Int. Microwave Symp. Digest, June 1998, Vol. 3, pp. 1739
1742.
109. R. Lotz, J. Ritter, and F. Arndt, Locally conformed subgrid
FD-FD technique for the analysis of 3D waveguide structures
with curved metallic objects, 1999 Int. Microwave Symp.
Digest, June 1999, Vol. 3, pp. 12771280.
110. A. M. K. Saad, Antipodal ridge evanescent wave-guide lter,
Proc. 17th European Microwave Conf. (EuMC) Proceedings,
Rome, Italy, Sept. 1987, pp. 157162.
111. T. S. Saad, Microwave Engineers Handbook, Vol. 1, Dedham,
MA, Artech House, 1971.
112. G. L. Matthaei, L. Young, and E. M. T. Jones, Microwave
Filters, Impedance-Matching Networks, and Coupling Struc-
tures, McGraw-Hill, New York, 1964.
113. F. Arndt, I. Rullhusen, V. Catina, J. Brandt, and W. Wessel,
Fast hybrid mode-matching/integral equation methods for
the analysis of slot arrays, Proc. Int. Symp. on Electromag-
netic Theory, URSI EMTS, Pisa, May 2004, pp. 4547.
HYBRIDS AND COUPLERS
PRAKASH BHARTIA
Natel Engineering Co., Inc.
Chatsworth, California
PROTAP PRAMANICK
K & L Microwave
Salisbury, Maryland
1. INTRODUCTION
Hybrids and couplers form an indispensable component
group in modern MIC technology. With the inventions of
planar transmission lines such as stripline, microstrip
line, nline, dielectric image line, and their derivatives,
hybrid and coupler technology has undergone a substan-
tial change since the mid-1960s as a result of the rapidly
growing applications of MICs in the electronic warfare,
communications, and radar industries.
Despite the fact that the basic philosophy behind the
operation of such couplers remains the same as in couplers
designed using conventional transmission lines, their an-
alyses and syntheses are quite involved. This is because
most of these lines support hybrid modes due to inhomo-
geneity in conguration. However, the present-day anal-
ysis and synthesis techniques for such hybrids and
couplers are believed to have gained maturity.
This article describes the design aspects of planar hy-
brids and couplers in as self-contained a presentation as
possible within a limited space. In what follows in this
section, we present the basics of hybrids and couplers and
discuss different types of hybrids and couplers and their
applications. The next section describes the design of
matched hybrid tees (Ts), hybrid rings, and 901 hybrids.
This is followed by a section on coupled-line couplers, both
the TEM and the distributed types, as well as other mis-
cellaneous types of couplers. The nal section includes
various aspects of coupler design, such as losses and im-
provement of directivity.
1.1. Basics of Hybrids and Couplers
A hybrid or a directional coupler can in principle be rep-
resented as a multiport network. In such a network, the
2054 HYBRIDS AND COUPLERS
port into which the electrical power is fed is called the in-
cident port. The ports through which the desired amounts
of coupled power are extracted are called coupled ports,
while the rest of the ports are called isolated ports. Al-
though hybrids and couplers having up to six ports nd
applications in many systems, we will restrict ourselves
mostly to discussion of four-port networks without loss of
generality.
Consider the four-port network shown in Fig. 1. If P
1
is
the power fed into port 1 (which is matched to the gener-
ator impedance) and P
2
, P
3
, and P
4
are the powers avail-
able at the ports 2, 3, and 4, respectively (while each of the
ports is terminated by its image impedance), the coupling
coefcient is dened as
C= 10 log
P
n
P
1
n=3; 4 (1)
If port 3 happens to be the desired coupled port, the cou-
pling coefcient is given by
C= 10 log
P
3
P
1
dB (2)
If port 4 is the desired uncoupled port, the desired isola-
tion is given by
I = 10 log
P
4
P
1
dB (3)
The transmission to the primary port 2 is given by
T =10 log
P
2
P
1
dB (4)
The measure of directivity between the coupled and the
uncoupled ports is given by
D=I C (5)
As a general practice, the performance of a hybrid or a
directional coupler is specied in terms of its coupling, di-
rectivity, and the characteristic impedance at the center
frequency of its band of operation. These data enable the
circuit designer to calculate the structural parameters of
the coupler.
1.2. Types of Hybrids and Couplers
Hybrids use directly connected circuit elements. Couplers
can be parallel-coupled or aperture-coupled.
Figure 2a shows the simplest form of a hybrid. This is
a branchline coupler, consisting of two mainlines coupled
by two
1
4
l line sections spaced
1
4
l apart, where l is the
wavelength. Such a branchline hybrid can also be in cir-
cular form as shown in Fig. 2b. In either case, the total
length of all the lines is one wavelength.
Figure 3 shows a parallel-coupled coupler. It consists of
two coupled transmission lines. The lines can have any
form depending on the application.
A parallel-coupled line coupler (Fig. 4) can be of two
types: the TEM type or the distributed type. In the former,
the coupled transmission lines support a pure TEM or a
quasi-TEM mode. In the latter, the mode supported by the
coupled lines are non-TEM in nature. While all of these
types of couplers are realized using planar IC technology,
there can be another type of coupler using aperture cou-
pling, through some common ground plane (as used in
conventional aperture-coupled waveguide couplers as
shown in Fig. 5) with two similar or dissimilar types of
planar transmission lines.
1
4
2
3
Figure 1. Four-port network.
Z
r
Series arm
Shut arm
1
2
3
4
Z
r
Z
o
Z
p
Z
o
2
3
/4
1
Z
o
/4
Z
o
4
/4
Z
p
Z
o
(a)
(b)
/4
Figure 2. (a) Single-section branchline hybrid; (b) circular form
of branchline hybrid.
HYBRIDS AND COUPLERS 2055
To achieve tighter coupling over a wider bandwidth,
multisections of the previously mentioned couplers can be
cascaded in tandem (to be discussed in a later section).
1.3. Applications
Virtually all kinds of microwave circuits use hybrids or
couplers in one form or another. In general, the areas of
application can be divided into two parts:
1. Passivetuners, delay lines, lters, and matching
networks (sometimes an array of couplers may be
needed for a desired performance of the network).
2. Active circuitsused mainly as directional couplers
in balanced ampliers, mixers, attenuators, modu-
lators, discriminators, and phase shifters
2. DESIGN OF HYBRIDS
2.1. 901 Hybrid
The simplest 901 hybrid is the branchline hybrid shown in
Fig. 2a. For a certain input power at port 1, under match
condition, there will be 901 phase difference between the
waves appearing at ports 2 and 3, at the center frequency
at which each arm is exactly a quarter-wavelength long.
This 901 phase difference varies over 751 for a 10%
change in frequency around the center frequency. The
coupling bandwidth is 20%, but its usable bandwidth is
limited to 10%, due to an unacceptable change in the iso-
lation over a bandwidth exceeding 10%. Ideally such cou-
plers can be designed for 39 dB coupling.
There are three main arm losses in branchline hybrids:
losses due to the portion of power coupled to the secondary
arm (1) and the power dissipated as a result of dielectric
(2) and conductor (3) losses.
Branchline hybrids can be realized using virtually all
kinds of planar transmission lines, for example, stripline,
microstrip line, slotline, nline, and image line. However,
the basic design principle is extremely simple and the
same in all cases.
The coupling factor is determined by the ratio of the
impedance of the shunt and series arms and is optimized
to meet the proper match over the required bandwidth.
l
g
Port 2 Port 4
Port 1 Port 3
Figure 3. Direct-coupled coupler.
Line 1
Port 2 Port 1
Port 4 Port 3
Line 2
Z
l 0
Figure 4. Parallel-coupled line coupler.
(2n1)
(a)
(b)
4
/g
(2n1)
4
/g
Figure 5. Aperture-coupled couplers: (a) narrow-wall coupling;
(b) broad-wall coupling.
For 901 hybrids, the following conditions hold good (see
Fig. 2a):
P
2
P
3
=
Z
0
Z
p
_ _
2
(6)
Z
0
Z
r
_ _
2
=
Z
0
Z
p
_ _
2
1 (7)
Although the two-branch hybrid is the most fundamen-
tal structure, it has a very narrow bandwidth. This dis-
advantage can be overcome by using multisection hybrids,
but in most cases, planar transmission lines require too
wide a range of impedances for this purpose. These may
sometimes be difcult to realize physically. On the other
hand, very wide linewidths may require unreasonable as-
pect ratios at higher frequencies due to shorter wave-
lengths. The physically unrealizable high-impedance line
can be avoided by using a modied hybrid ring [1].
Complete analytical design techniques for such hybrids
using Chebyshev and Zolotarev [2,3] functions are avail-
able. However, such techniques may sometimes be unsuit-
able for the design of planar circuits because of the wide
impedance range problem, as previously mentioned. Mu-
raguchi et al. [4] presented a computer-aided design tech-
nique that is most suitable for an optimum design.
2.2. Ring Form of Branchline Hybrid
This is a circular ring version of the square 901 hybrid
described in the previous section. Therefore, the discus-
sion of the previous section is equally valid for hybrid
rings. The conguration is shown in Fig. 2b and is partic-
ularly advantageous in the realization of microstrip phase
detectors and balanced mixers, with all ports matched.
2.3. Matched Hybrid T (Ratrace Hybrid)
A matched hybrid T is a special kind of ring form of the
branch-line coupler in which the circumference is an odd
multiple of
3
2
l. As a result, the phase response is 01/1801.
The simplest version of a matched hybrid T is shown in
Fig. 6. Ports AB, BC, and CD are separated by 901, and
ports A and D are three quarter-wavelengths away from
each other.
Because of the impedance and the phase relationships
shown in the structure, any power fed into port C splits
equally into two parts that add up in phase at ports B and
D and out of phase of port A. As a result port A is isolated
from the input. Similarly, power fed at port A divides
equally between ports B and D with 1801 phase difference,
and port C remains isolated.
The frequency response of a typical
3
2
wavelength hy-
brid T is shown in Figs. 7a and 7b. It is observed that it
offers around 20% bandwidth in terms of matching, split,
and isolation. Moreover, the 1801 phase relationship is
much more frequency-sensitive than is the 01 phase rela-
tionship. Still, such a response is quite adequate for ap-
plications in mixers, SSB (single-side band) generators,
and other components.
For high-frequency applications, a circumferential
length of
3
2
wavelengths may pose fabricational difculties
due to unrealizable aspect ratios of the transmission lines.
This problem is overcome by having a ring of circumfer-
ential length of two wavelengths, where ports A and D are
located
5
4
l away on the ring.
The design of a hybrid amounts to realizing the re-
quired transmission-line sections with proper phase ve-
locities and characteristic impedances. Direct synthesis
equations can be used in cases of stripline [5], microstrip
[6], suspended microstrip, and inverted microstrip [7],
which support pure TEM or quasi-TEM modes of propa-
gation.
Realizations of the hybrids using non-TEM transmis-
sion lines such as slotline, nline, or image line use iter-
ative techniques with the help of accurate analysis
equations [79] depending on the mode of propagation.
However, closed-form design equations are also available.
Microstrip and stripline hybrids have been realized and
successfully used in commercially available microwave-
balanced mixers and other circuit components. Develop-
ment of nline and image-line hybrids are still at the ex-
perimental stage [10].
3. COUPLED-LINE DIRECTIONAL COUPLERS
The hybrids described in the previous sections are inher-
ently of narrow bandwidth. Broadbanding of microwave
couplers is achieved in a number of different ways, de-
pending on the application.
Broadband couplers are either aperture-coupled or par-
allel-coupled. Aperture coupling is used very successfully
in conventional waveguide techniques for realizing direc-
tional couplers with large bandwidths. Although aper-
ture coupling is a convenient way of realizing directional
couplers using two different planar transmission lines,
directional couplers using the same kind of planar trans-
mission lines are realized more efciently using parallel
coupling.
3.1. Directional Couplers Using Aperture-Coupled Lines
Figure 8 shows a simple two-hole directional coupler. Elec-
tromagnetic energy is coupled from the primary guide to
/4
/4
/4
3/4
D
A
B
2
Z
0
Z
0
z
0
Z
0
Z
0
C
Figure 6. Hybrid ring.
the secondary guide due to the eld radiated by the excit-
ed electric and magnetic dipoles generated at the holes by
the propagating electromagnetic wave in the primary
guide. The holes are spaced such that the round-trip phase
shift of a wave through them should be 1801. Therefore the
backward-traveling waves in the secondary guide will be
completely out of phase to cancel each other at port 3. If
the coupled lines have the same propagation constant,
then the forward-traveling waves in the secondary guide
will be of the same phase regardless of hole spacing and
are added at port 4.
Such aperture-coupled directional couplers can be re-
alized using various kinds of planar transmission-line
combinations, for example (1) microstripmicrostrip, (2)
microstripimage line, (3) image lineimage line, (4) im-
age linetrough guide, or (5) trough guidetrough guide.
Some possible combinations are shown in Fig. 9.
3.2. TEM Line Directional Couplers
When the center conductors of two coaxial lines support-
ing a pure TEM mode of propagation are brought into
close proximity of each other, electromagnetic energy is
coupled from one line to the other. This property has given
rise to a class of broadband planar directional couplers.
Most such couplers use striplines or microstrip lines that
support pure TEM or quasi-TEM modes, respectively.
Planar TEM-line directional couplers can be either
edge-coupled or broadside-coupled as shown in Figs. 10a
and 10b.
In general, such coupled TEM lines support two modes
that interact to give rise to the coupling. These are the
even and odd modes. Properties of the coupled lines can
be evaluated by suitable linear combinations of even
and odd modes.
V
S
W
R
S
p
l
i
t
I
s
o
l
a
t
i
o
n
1.0
0.7 0.8 0.9 1.0 1.1 1.2 1.3
1.1
1.2
1.3
1.4
1.5
20
30
40
10
20
30
40
50
A
D-B
C-D
C-B
Port C
B
C
D
Normalized Frequency
(a)
P
h
a
s
e

(
d
e
g
)
0.8 0.9 1.0 1.1 1.2
10
0
10
160
170
180
190
200
(b)
A
B
C
D
B-D Input at A
B-D Input at C
Figure 7. (a) Response curves for
3
2
ratrace magic T (power split and isolation are expressed in
decibels); (b) phase response curves for
3
2
ratrace magic T.
Port 1
Port 3
Port 2
Port 4
Primary
waveguide
(2n1)
4
/g
Secondary waveguide
Cancel Add
Figure 8. Schematic of a two-aperture directional coupler.
TEM-line couplers can be reduced to a two-port net-
work as shown in Fig. 11a. There is a plane of symmetry
that becomes a perfect magnetic wall for the incident sig-
nals of equal amplitude and same phase at ports 1 and 4.
The plane of symmetry becomes a perfect electric wall
for the incident signals of equal amplitude but of exactly
opposite phases at the same ports. Therefore each mode
corresponds to a two-port network as shown in Figs. 11b
and 11c for even and odd modes, respectively. Analysis of
the directional coupler is accomplished by analyzing these
two networks and superimposing the responses as shown
in Fig. 11d.
The S-matrix equation of a Four-port symmetric and
reciprocal network can be written as
b
1
b
2
b
3
b
4
_
_
_
_
=
S
11
S
12
S
13
S
14
S
12
S
11
S
14
S
13
S
13
S
14
S
11
S
12
S
14
S
13
S
12
S
11
_
_
_
_
a
1
a
2
a
3
a
4
_
_
_
_
(8)
where
S
11
=
1
2
(G
e
G
o
) (9a)
S
12
=
1
2
(T
e
T
o
) (9b)
S
13
=
1
2
(T
e
T
o
) (9c)
S
14
=
1
2
(G
e
G
o
) (9d)
Evaluation of the even- and odd-mode reection and
transmission coefcients is done from the corresponding
effective dielectric constants and the characteristic imped-
ances:
G
i
=
A
i
B
i
=Z
0
C
i
Z
0
D
i
A
i
B
i
=Z
0
C
i
Z
0
D
i
(10a)
T
i
=
2
A
i
B
i
=Z
0
C
i
Z
0
D
i
(10b)
where the transmission matrix
A
i
B
i
C
i
D
i
_ _
=
cos y
i
jZ
0i
sin y
i
j sin y
i
Z
0i
cos y
i
_
_
_
_
i =e; 0 (11)
Knowing Z
0,e
, y
e
, Z
0,o
, y
o
, and Z
0
and the system charac-
teristic equation, we can calculate the performance of the
directional coupler using (9) and (10).
Microstrip
Microstrip
Microstrip
lines
Image
lines
Image
line
Trough lines
Aperture
coupling
Aperture
coupling
Aperture
coupling
Dielectric
slabs
Waveguide
(a) (b)
(c) (d)
(e) (f)
Figure 9. Possible combinations of various aperture-coupled
lines: (a) microstripwaveguide coupling; (b) microstripmicro-
strip line; (c) microstripimage line; (d) image lineimage line; (e)
trough linetrough line; (f) dielectric guidedielectric guide.
W
h
Substrate
Strip conductor
Ground plane
Ground plane
Air
S
r
2
r
2
r
1
2W
t
2b 2S
(a)
(b)
Figure 10. (a) Edge-coupled microstrip lines; (b) broadside-cou-
pled microstrip line.
3.2.1. Coupled TEM Line. For coupling between the
purely TEM lines shown in Fig. 12, we have the follow-
ing special case
y
e
=y
o
=y (12)
which means equal, even-, and odd-mode phase velocities,
and
Z
0
=
Z
0;e
Z
0;o
_
(13)
G
e
= G
o
=
j[(Z
0;e
=Z
0;o
)
1=2
(Z
0;o
=Z
0;e
)
1=2
] sin y
S
(14)
where
S=2 cos y j
Z
0;e
Z
0;o
_ _
1=2
Z
0;o
Z
0;e
_ _
1=2
_ _
sin y (15)
Substitution of (14) and (15) into (9) gives
S
11
=0 (16a)
S
12
=T
e
(16b)
S
13
=0 (16c)
S
14
=G
e
(16d)
Such couplers, known as backward-wave couplers, are in
general a quarter-wavelength long at the center frequency
or y =
1
2
p.
From (13) and (14) the coupling coefcient is given by
C= 20 log [S
14
[ = 20 log
Z
0;e
Z
0;o
Z
0;e
Z
0;o
dB (17)
Moreover, for a system impedance Z
0
of 50 O, the matching
condition (condition for a perfect input match) gives
Z
2
0
=Z
0;e
Z
0;o
(18)
Therefore, for a specied coupling C the design equations
become
Z
0;e
=Z
0
110
c=20
1 10
c=20
_ _
1=2
(19a)
Z
0;o
=Z
0
1 10
c=20
110
c=20
_ _
1=2
(19b)
Once Z
0,e
and Z
0,o
are known, the physical dimensions of
the coupler can be obtained using the equations for cou-
pled TEM transmission lines [6,20].
3.2.2. Coupled Quasi-TEM Line. In coupled quasi-TEM
lines, for example, in microstrip, the odd-mode phase ve-
locity is different from the even-mode phase velocity.
Therefore, the condition given by (12) does not hold
good. However, for weak couplings, (12) can be assumed
to be approximately true. Hence the designer may go
ahead and determine the initial design using (17)(19).
However, as the coupling gets tighter, the previous equa-
tions tend to be less valid. In such a case, the condition for
+

e
/2
e
/2
1
4
1
a
1
e
/2
e
/2
+
4
+

o
/2
1
o
/2
4
1
(
e
+
o
)
1
(
e
o
)
4
b
2
b
1
a
4
3
2
3
o
/2
o
/2
2
3
3
2
2
a
2
b
2
a
3
b
3
Coupling
structure
(CS)
(CS)
(a)
(b)
(c)
(d)
(CS)
E wall
H wall
(CS)
(
e
+
o)
(
e
o)
Figure 11. Schematic of directional coupler: (a) wave parame-
ters; (b) even-mode excitation; (c) odd-mode excitation; (d) input
excitation.
In
Out
Out
2
1
3
4
Figure 12. Coupling between two TEM lines.

input matching becomes
Z
0
=
Z
0;e
sin y
e
Z
0;o
sin y
o
Z
0;e
sin y
o
Z
0;o
sin y
e
_ _
1=2
Z
0;o
; Z
0;e
_
(20)
and the electrical length at the center frequency is
y =
1
2
(y
e
y
o
) =
2p
l
0
e
ee
_

e
eo
_
2
l =90
(21)
where l is the physical length of the coupler.
Single-section quarter-wave parallel-coupled line cou-
plers are used extensively in many applications. They are
usually of narrow bandwidth of approximately one octave.
To obtain the desired coupling at the band edges, the cou-
pler has to be designed for overcoupling at the center fre-
quency.
3.2.3. Frequency Response of a Single-Section Cou-
pler. Using the analysis equations (9)(17), it can be
shown that the frequency response of the coupling coef-
cient is given by
C=(y) =
jC sin y
1 C
2
_
cos y j sin y
(22a)
where C is the midband coupling for a matched, loosely
coupled coupler. The approximate frequency response of a
quasi-TEM coupler is shown in Fig. 13. The response fol-
lows Eq. (22a) when the coupled lines support TEM mode.
The general expression for the directivity or the unde-
sired coupling is given by [11]
D=
pD(1 [x[
2
)
4[x[
_ _
2
(22b)
where
D=
b
e
b
o
b
o
x =
r
e
1 r
2
e
_ _
r
o
1 r
2
o
_ _
r
e
=
Z
0;e
Z
0
Z
0;e
Z
0
r
o
=
Z
0;o
Z
0
Z
0;o
Z
0
and b
e
and b
o
are the even- and odd-mode propagation
constants, respectively, D=0 for b
e
=b
o
, that is, for a TEM
coupler.
3.2.4. Multisection Couplers for Wider Bandwidth. For
many applications, the single-section coupler proves to be
of inadequate bandwidth. Therefore, the designer should
have recourse to a multisection design. A multisection
coupler is a cascaded combination of more than one single-
section coupler, each a quarter-wavelength long at the
center frequency of the band. The number of sections to be
used depends on the tolerable insertion loss, bandwidth,
and the available physical space.
Multisection couplers can be either symmetric or asym-
metric around the center section, as shown in Fig. 14.
3.2.5. Symmetric Coupler. The symmetric coupler gives
901 phase difference between the direct and the coupled
output ports under matched conditions. In what follows
we present a direct synthesis technique.
Figure 14a shows a multisection symmetric coupler.
The coupling factor for an n-section symmetric coupler
with weak coupling can be written as [12]
C(y) =
V
2
V
1
=C
1
sin(ny) (C
2
C
1
) sin[(n 2)y]
(C
i
C
i1
) sin[(n 2i 2)y]
(C
[(n1)=2]
C
[(n1)=2]
) sin y
(23)
If the desired coupling response is maximally at,
then C
i
must satisfy a set of
1
2
(n 1) linear equations
Out
In
In
Out
Out
Coupling
factors
1
1
2
2
4
3
4
3
C
1
C
2
C
1
C
2
C
2
C
1

(a)
(b)

Figure 14. Multisection (a) symmetric and (b) asymmetric cou-
plers.
Midband
C = C(/)
(max)
Z
0e
Z
0o
Z
0e
+ Z
0o
Figure 13. Approximate frequency response of a quasi-TEM
line.
obtained from
d
r
C(y)
dy
r
_ _
y =p=2
=0 r =2; 4; 6; . . . ; n 1 (24)
Note that n is always an odd integer.
The design concept is based on the fact that the
backward-coupled wave of a TEM coupler corresponds to
the reected wave of a quarter-wave lter. Therefore the
designer of a TEM-wave coupler has to synthesize only a
two port in place of a four port with the reection coef-
cient response the same as the desired coupling coefcient
response of the four-port directional coupler.
In terms of the midband VSWR R of the quarter-wave
lter, the coupling coefcient is given by [12]
C
0
=
R 1
R1
(25)
3.3. Multiconductor Couplers
The interdigital coupler, or multiconductor coupler, in-
vented by Lange [13], has always been a popular compo-
nent in planar circuits. Figure 15 shows a four-element
interdigital coupler, although in certain applications the
number of elements may be greater than four. The coupler
is usually designed for 3dB coupling and the output phas-
es are in quadrature. Obviously, the best realization is in
the microstrip form.
An interdigital coupler has advantages because of its
small size and relatively large line separation when
compared with the two-coupled line device and has a
much larger bandwidth when compared with branchline
couplers.
Interdigital couplers are used for balanced MIC ampli-
ers, balanced mixers, and binary power divider trees.
3.3.1. Design of Interdigital Couplers. Kajfez et al. [14]
have described a simplied design technique for the in-
terdigital coupler. The proposed technique serves many
practical purposes but seems to be inadequate for an ac-
curate design. So far the method due to Presser [15]
has been found to be the most accurate and simplest.
Moreover, it has the provision for nite stripthickness
correction.
Consider the layout shown if Fig. 15. The designer is
supplied with the desired coupling coefcient C and the
system characteristic impedance Z
0
. The length of the
coupled region l has to be a quarter wavelength at the
center of the band.
The main design equations for an N element (N even)
coupler are written as
R=
Z
0;o
Z
0;e
(26a)
C=
(N 1)(1 R
2
)
(N 1)(1 R
2
) 2R
(26b)
Z=
Z
0;o
Z
0
R[(N 1) R][(N 1)R1]

_
1 R
(26c)
3.4. Distributed-Type Couplers
Distributed-type coupling takes place between two adja-
cent transmission lines supporting purely non-TEM
modes. For example, distributed-type couplers can be re-
alized using two open dielectric waveguides (or image
guides) or nlines.
In general, two distributed-type coupled lines can be
represented as shown in Fig. 16. Under the assumption
that all four ports are matched and the coupling structure
has the length l, the ratio of the elds in the two lines can
be shown to be [16]
E
b
(l)
E
a
(l)
= tan (b
e
b
o
)
l
2
_ _
(27)
where b
e
and b
o
are the even- and the odd-mode phase
constants, respectively.
Therefore for complete transfer of power from line a to
line b requires
(b
e
b
o
)
L
2
=
p
2
(28)
or
L=
p
b
e
b
o
(29)
Z
0
=50 Z
0
=50
S
W
l /2
l /2
Z
0
=50 Z
0
=50
Figure 15. Lange coupler.
The scattering coefcients of the structure can be written
as a function of l normalized by L as
[S
12
[ = cos
p
2
l
L
_ _
(30a)
[S
13
[ = sin
p
2
l
L
_ _
(30b)
Equation (30b) shows that the required length for 3dB
coupling is one half of L.
The preceding equations are based on the assumptions
that the bent portions of the guides have no effect on cou-
pling. Branch junction effects near the bends are negligible,
but the assumed uncoupled lines zr0 and zZz
0
do get cou-
pled. This extra coupling can be taken into consideration by
dening the effective length of the coupler as
l
eff
=l
2L
p
_
z
/
z
0
[b
e
(z) b
o
(z)] dz (31)
The integration limit z
/
is chosen to be the point at which
the coupling is practically negligible.
From the preceding discussions it appears that the de-
sign of any distributed coupler requires a precise knowl-
edge of the even- and the odd-mode phase constants of the
coupled lines. Figures 16b and 16c show the cross sections
of a coupled nlines and image lines, respectively, as ex-
amples of two commonly used distributed couplers.
3.5. Wilkinson Couplers, Power Dividers, and Combiners
A Wilkinson coupler [17,18] is a special case of couplers
that do not have isolated ports and offer broad bandwidth
and equal phase characteristics at each of its output ports.
Figure 17 shows the schematic diagram of a Wilkinson
coupler. The output ports isolation is obtained by series
terminating the output ports. Each quarter-wave line has
the characteristic impedance of
2
_
Z
0
and the output is
terminated by a resistor of 2Z
0
O, where Z
0
is the system
impedance.
A Wilkinson power divider offers a bandwidth of about
one octave. The typical frequency response is shown in
Fig. 18. An adequately at response is obtained over more
than one octave band, but at the band edges the isolation
is affected by the load impedance.
The performance of a Wilkinson coupler can be further
improved, depending on the availability of space, by the
addition of a
1
4
l transformer in front of the power division
step. The output VSWR of the uncompensated coupler is
better than the output VSWR of the compensated circuit.
3.5.1. Multisection Wilkinson Coupler. The octave
bandwidth of a single-section coupler proves to be inade-
quate in many applications; therefore Cohn [18] proposed
the use of multisections for bandwidth expansion. The
use of multisections makes it possible to obtain a decade
bandwidth.
Z=0
a
Plane of
symmetry
1
4
2
3
y
y
W
W
y
x
d
Waveguide
Fin structure
Electric wall
Magnetic wall
Even mode
Odd mode
Two-orthogonal
modes
E
y
, H
x
E
y
, H
x
E
y
, H
x
x
x
y y
y
E
y
,H
x
2S
2a 2a
b b
x
Z
0
Z
l
b
(a)
(b)
(c)
2S
2a 2a
y
2S
Odd mode Even mode
2a 2a
Figure 16. (a) Schematic of distributed coupler; (b) coupled n-
lines and eld distributions; (c) coupled image lines and eld dis-
tributions.
Z
0
Z
0
Z
0
2Z
0
/4
2Z
0
2Z
0
Figure 17. Single-section Wilkinson coupler.
A multisection Wilkinson power divider coupler con-
sists of a number of quarter-wave sections with resistive
terminations at the end of every section, as shown sche-
matically in Fig. 19. Larger bandwidth and greater isola-
tion are obtained when a large number of sections are
used.
The characteristic impedances of the sections are ob-
tained from the normalized impedances for
1
4
l transformer
sections for a 21 transformer. This can be done with the
help of charts in Ref. 5, which presents design curves up to
four sections. Similar gures are also available for higher
numbers of sections. Having obtained the impedance of
each section, we can obtain values of the terminating re-
sistors for each section.
For a two-section divider the values of the terminating
resistors are given by
R
2
=
2Z
1
Z
2
[(Z
1
Z
2
)(Z
2
Z
1
cot
2
f)]
1=2
(32a)
R
1
=
2R
2
(Z
1
Z
2
)
R
2
(Z
1
Z
2
) 2Z
2
(32b)
where
f=
p
2
1 0:707
f
2
f
1
f
2
f
1
_ _ _ _
(32c)
and f
1
, f
2
are the upper and lower band edge frequencies of
operation.
3.6. Other Couplers
3.6.1. Tandem Coupler. The design of a multisection
coupler with tight coupling over a broad bandwidth re-
quires some of its sections to have tighter coupling than
the overall coupling. This invariably leads to the physi-
cally unrealizable spacings between the two conductors or
severely reduced directivity due to significant mechanical
discontinuities in the sections. To solve this problem, in a
restricted physical space, various combinations of sym-
metric and asymmetric couplers are tandemed [5,19].
Since in the majority of applications, the tightest coupling
may be 3 dB, two couplers may be connected in tandem to
achieve the goal. Figure 20a shows the symmetric tandem
of two 8.34-dB couplers, while Fig. 20b shows the sym-
metric tandem of two asymmetric couplers. When tande-
med, each 8.34-dB coupler gives an overall coupling of
3 dB. This conguration offers high power-handling capa-
bility and often represents a good choice, provided the
particular application does not require maximum band-
width with very low loss.
Figure 21 shows the conguration of an asymmetric
tandem of symmetric couplers. As in the symmetric tan-
dem of symmetric couplers, the quadrature phase shift is
maintained between the outputs.
As is apparent from the interconnections, the 901 phase
relationship is maintained in the symmetric case, but in
the asymmetric case, the phase relation depends on the
number of sections.
4. DESIGN CONSIDERATIONS
4.1. Losses in Hybrids
The total loss in hybrids can be estimated from the com-
bined dielectric and conductor losses in the individual
lines. These are obtained using closed-form equations for
microstrip- and stripline-type transmissions lines [20].
V
S
W
R
1.5
1.2
1.0
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
Input
Output
Uncompensated
50
50
50
70.7
70.7
Figure 18. Frequency response of Wilkinson coupler. (After
Howe [5]. Reprinted with permission of Artech House.)
/4
(typical)
Z
o
Z
o
Z
o
Z
N
Z
1
Z
N1
R
N1
R
1
R
N
Figure 19. Multisection Wilkinson coupler.
Figure 20. Tandem coupler: (a) Symmetric tandem of symmetric
couplers; (b) symmetric tandem of asymmetric couplers.
For other types of transmission lines, numerical tech-
niques are used [21].
The attenuation due to the even mode a
e
c
is always less
than that due to the odd mode a
o
c
. In coupled lines the loss
is given by the average of the losses due to even and odd
modes. In almost all planar coupled lines the conductor
loss greatly exceeds the dielectric loss [20].
As in striplines and microstrip lines, the primary con-
tributors to losses in dielectric-based planar waveguide
couplers are the dielectric loss and the metallic loss. Such
losses are always computed numerically from the associ-
ated eld equations [22].
4.2. Directivity Improvement
Because of the inhomogeneity in dielectric structure, the
directivity of microstrip couplers offers a poor bandwidth
resulting from different odd- and even-mode phase veloc-
ities. There are several ways to equalize the phase veloc-
ities, which are described as follows.
4.2.1. Use of a Shield. These structures nearly equalize
even- and odd-mode phase velocities. The cross section of
the structure is shown in Fig. 22. This conguration es-
sentially redistributes the eld with a substantial amount
of the eld in the airdielectric medium above the coupled
strips. For d=h the phase velocities are exactly equal, and
each is equal to
v
e
=v
o
=
c
(e
r
1)=2
_ (33)
Strictly speaking, improvement in directivity is obtained
at the cost of manufacturing difculties in this case.
4.2.2. Use of Lumped Capacitors. Using lumped capac-
itors at the ends of the coupled section is the simplest way
of achieving equal phase velocities. The structure is shown
in Fig. 23. This effectively increases the odd-mode phase
angle by [16]
Dy
o
=2pf
0
(C
1
C
2
)Z
0;o
(34)
where f
0
is the center frequency of the coupled line.
4.2.3. Use of a Dielectric Overlay. The presence of an-
other dielectric layer of the same permittivity as that of
the substrate reduces the odd-mode phase velocity to a
large extent without considerably affecting the even-mode
phase velocity [23]. Thus, by controlling the thickness and
width of the overlay, the even- and the odd-mode phase
velocities can be equalized within 1% over quite a broad
band. As shown in Fig. 24, the overlay covers the two
strips where the main coupling takes place. The overlay is
bonded with the help of some kind of epoxy. However, this
method is practically cumbersome because of its poor re-
peatability. To overcome this problem a multilayer struc-
ture using different dielectric constants materials has
been proposed [23].
In Isolation
Z
(1,3) Z
(1,3) Z
(1,2)
Z
(1,2) Z
(1,1)
Z
(3,1)
-3dB -3dB
Z
(2,2)
Z
(2,3) Z
(2,1)
Figure 21. Asymmetric tandem of symmetric coupler.
d
h
S W W
r
Figure 22. Parallel-coupled microstrip with grounded shield.
50
50
50
50
W
0
2W
0
W
0.43W
C
1
S
C
2
l
Figure 23. Lumped-capacitor compensation of microstrip cou-
pler.
h

r
r
S W W
Figure 24. Parallel-coupled microstrip with overlay com-
pensation.
BIBLIOGRAPHY
1. A. K. Agrawal, and G. F. Mikucki, A printed circuit hybrid ring
directional coupler for arbitrary power divisions, IEEE Trans.
Microwave Theory Tech. MTT-34: 14011407 (Dec. 1986).
2. R. Levy, Zolotarev branch-guide couplers, IEEE Trans. Mi-
crowave Theory Tech. MTT-21: 9599 (Feb. 1973).
3. R. Levy, and J. Helszajn, Specific equation for one or two section
quarter-wave matching networks for stub resistor loads, IEEE
Trans. Microwave Theory Tech. MTT-30: 5562 (Jan. 1982).
4. O. Muraguchi, M. Y. Takeshi, and Y. Naito, Optimum design
of 3-dB branch line couplers using microstrip lines, IEEE
Trans. Microwave Theory Tech. MTT-31: 674678 (Aug. 1983).
5. H. Howe, Stripline Circuit Design, Artech House, Dedham,
MA, 1974.
6. K. C. Gupta, R. Garg, and I. J. Bahl, Microstrip Lines and
Slotlines, Artech House, Dedham, MA, 1979.
7. P. Pramanick, and P. Bhartia, Analysis and sythesis equa-
tions for suspended and inverted microstriplines, Arch. Elek.
U
bertragung. 39: 323326 (Sept. 1985).

8. P. Pramanick, and P. Bhartia, Computer aided design models
for millimeter wave nlines and suspended substrate micro-
strip lines, IEEE Trans. Microwave Theory Tech. MTT-33:
14291435 (Dec. 1985).
9. P. Bhartia, and I. J. Bahl, Millimeter Wave Engineering and
Applications, Wiley, New York, 1984, pp. 300346.
10. S. K. Koul, private communication, Center for Advanced
Research in Electronics, Indian Institute of Technology, New
Delhi, India, 1985.
11. T. C. Edwards, Foundations for Microstrip Circuit Design,
Wiley, New York, 1981, pp. 147148.
12. G. L. Matthaei, L. Young, and E. M. T. Jomes, Microwave
Filters Impedance Matching Networks and Coupling
Structures (reprinted), Artech House, Dedham, MA, 1980.
13. J. Lange, Interdigited stripline quadrature hybrid, IEEE
Trans. Microwave Theory Tech. MTT-17: 11501151 (Dec.
1969).
14. D. Kajfez, Z. Paunovic, and S. Paulin, Simplied design of
Lange coupler, IEEE Trans. Microwave Theory Tech. MTT-26:
806808 (Oct. 1978).
15. A. Presser, Interdigited microstrip coupler design, IEEE
Trans. Microwave Theory Tech. MTT-26: 801805 (Oct. 1978).
16. P. Bhartia, and I. J. Bahl, Millimeter Wave Engineering and
Applications, Wiley, New York, 1984, pp. 358382.
17. E. Wilkinson, An N-way hybrid power divider, IEEE Trans.
Microwave Theory Tech. MTT-8: 116118 (Jan. 1960).
18. S. B. Cohn, A class of broadband 3-port TEM hybrids, IEEE
Trans. Microwave Theory Tech. MIT-16: 110118 (Feb. 1968).
19. W. A. Davis, Microwave Semiconductor Circuit Design, Van
Nostrand, New York, 1983.
20. R. Garg, and I. J. Bahl, Characteristics of coupled microstrip,
IEEE Trans. Microwave Theory Tech. MTT-27: 700705 (July
1979).
21. T. N. Trinh, and R. Mittra, Suspended H-waveguide and its
mm-wave applications, IEEE Int. Microwave Symp. Digest,
1983, pp. 305308.
22. B. Shelad, and B. E. Spielman, Broadband (7-18 GHz) 10-dB
overlay coupler for MIC applications, Electron. Lett. 11:
175176 (April 1975).
23. M. Horno, and F. Medina, Multilayer planar structures for
high-directivity directional coupler design, IEEE Trans.
Microwave Theory Tech. MTT-34: 14421449 (Dec. 1986).

08c

Uploaded by

Document Information

Original Description:

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

08c

Uploaded by

Copyright:

Available Formats

HIGH-FREQUENCY TRANSMISSION LINES

is the reection coefcient, which can

and a reected voltage wave

. A relationship can then be express between the re-

ff sin f cos y) (6)

Figure 11. Universal patterns of small-are-angle corrugated

Figure 13. A prole corrugated horn. (Copyright 1994, IEE, from

are approximated by their nodal

Figure 12. Coupling between two TEM lines.

R[(N 1) R][(N 1)R1]

bertragung. 39: 323326 (Sept. 1985).

You might also like