# Chapter 9

Con…ned and Guided Waves
9.1 Introduction
Electromagnetic waves can be con…ned in a volume surrounded by a conductor wall as in microwave
cavities and also guided to propagate in conducting tubes and dielectric …bers. Parallel wire
transmission lines and coaxial cables are also used in transmitting low frequency waves. In this
chapter, electromagnetic waves in cavities and waveguides will be discussed. In waveguides, axial
electric and magnetic …elds 1
:
and H
:
satisfy scalar wave equation and can be readily solved by
incorporating the boundary conditions. Transverse …elds E
?
and H
?
can then be found in the
form of linear combination of the axial …elds. In conductor waveguides, electromagnetic waves can
be classi…ed into modes without axial magnetic …eld (TE mode) and modes without axial magnetic
…eld (TM mode). The axial phase velocity of electromagnetic waves in conductor waveguides with
a smooth wall is always larger than c. In slow waveguides with a periodically structured wall, the
phase velocity can be adjusted close to c as needed in linear accelerators. In dielectric …bers, there
can be no pure TE or TM modes because evanescent …elds outside the …ber do a¤ect mode structure
through the boundary condition at the …ber surface. Optical …bers used in communication all have
gradual radial (j) change in the index of refraction, :(j), and poses a problem of wave propagation
in a nonuniform medium.
1
9.2 Spherical Cavity
Electromagnetic waves in a spherical cavity surrounded by a conducting wall may be analyzed in
terms of the TE and TM spherical eigenvectors formulated in Chapter 6,
E = i.[r \c +\(r \c)], (9.1)
B = \(r \c) +/
2
r \c, (9.2)
where c and c both satisfy the Helmholtz equation
_
0
2
0r
2
+
2
r
0
0r
+
1
r
2
1
sin0
0
00
_
sin0
0
00
_
+
1
r
2
sin
2
0
0
2
0c
2
+/
2
_
_
_
c(r)
c(r)
_
_
= 0. (9.3)
Here / = .,c. Solutions for c and c are in the form of standing (non-propagating) wave,
c(r, 0, c, t), c(r, 0, c, t) = ,
|
(/r)1
n
|
(cos 0)
_
_
_
sin:c
cos :c
_
_
_
c
i.t
. (9.4)
Note that :
|
(/r) is discarded because the …elds should remain …nite at r = 0. As in mechanical
standing waves wherein the sum of potential energy and kinetic energy is constant, in electromag-
netic standing waves the sum of electric energy and magnetic energy is constant. This is guaranteed
because the magnetic …eld is out of phase with the electric …eld by ¬,2 as seen from Eqs. (9.1) and
(9.2). For TE modes without radial component, the boundary condition that 1
0
and 1
ç
vanish at
the inner wall r = a yields
,
|
(/a) = 0. (9.5)
Then the resonance frequency is given by
.
|c
=
cr
|c
a
, (9.6)
2
where r
|c
is the :-th root of ,
|
(r) = 0. Some roots are listed below.
| = 0 1 2 3
: = 1 ¬ 4.493 5.763 6.988
2 2¬ 7.725 9.095 10.417
3 3¬ 10.904 12.323 13.698
4 4¬ 14.066 15.515 16.924
For TM modes, the radial component of the magnetic …eld should vanish at r = a. This yields
d
dr
[r,
|
(/r)] = 0 at r = a, (9.7)
and the resonance frequency of TM mode is
.
|c
=
cr
0
|c
a
, (9.8)
where r
0
|c
is the :-th root of
o
oa
[r,
|
(r)] = 0.
| = 0 1 2 3
: = 1 ¬,2 2.744 3.870 4.973
2 3¬,2 6.177 7.443 8.722
3 5¬,2 9.317 10.713 12.064
4 7¬,2 12.486 13.921 15.413
The smallest eigenvalue is that of | = 1 TM mode,
.
TA,|=1
· 2.744
c
a
, /a = 2.744.
The Q value (cavity quality factor) is de…ned by
Q = .
time averaged energy stored in a cavity
time averaged rate of energy absorption by the wall
= .
_
\
j
0
2
H
2
d\
_
.j
0
2o
_
S
1
2
(n H)
2
do
=
2
c
_
\
H
2
d\
_
S
(n H)
2
do
, (9.9)
3
where the magnetic …eld is assumed to be in the form of standing wave H(r)c
i.t
, and
c =
_
2
.j
0
o
=
1
_
¬)j
0
o
, (9.10)
is the skin depth of the wall. Note that in the case of standing wave, the sum of electric and
magnetic energies is constant and equal to either
j
0
2
_
\
H
2
d\,
or
-
0
2
_
\
1
2
d\.
Time averaged dissipation rate at the wall is
1
2
_
.j
0
2o
(n H)
2
, (W/m
2
),
where the factor 1,2 is for rms (time average) value.
For the lowest order TM mode with | = 1, the azimuthal magnetic …eld is in the form
H
ç
= C,
1
(/r) sin0, (9.11)
where C is a constant. Then, for /a = 2.744,
Q
TA,|=1
=
a
c
_
1 ÷
,
0
(/a) ,
2
(/a)
,
2
1
(/a)
_
· 0.734
a
c
, (9.12)
where use is made of the integral,
_
r
2
,
2
|
(r) dr =
r
3
2
_
,
2
|
(r) ÷,
|1
(r) ,
|+1
(r)
¸
. (9.13)
Alternative expression is
Q
TA,|=1
= 0.734
a
c
= 0.32
`
c
. (9.14)
The atmospheric region between the ionosphere and ground earth can be regarded as a thin
4
cavity for electromagnetic waves. The simplest standing wave is the one which is azimuthally
symmetric (0,0c = 0) . The c component of the wave equation
_
\
2
+/
2
_
B = 0, / = .,c, (9.15)
yields
_
\
2
÷
1
r
2
sin
2
0
+/
2
_
1
ç
= 0, (9.16)
provided 1
v
= 1
0
= 0. Solution for 1
ç
(r, 0) is
1
ç
(r, 0) =
n
|
(r)
r
1
1
|
(cos 0) , (9.17)
where n
|
(r) satis…es
d
2
n
|
dr
2
+
_
/
2
÷
| (| + 1)
r
2
_
n
|
(r) = 0. (9.18)
Corresponding electric …eld is
1
v
= ÷
ic
2
.r
| (| + 1)
n
|
(r)
r
1
|
(cos 0) , (9.19)
1
0
= ÷
ic
2
.r
dn
|
(r)
dr
1
1
|
(cos 0) . (9.20)
In the narrow region between the ionosphere and ground, we may approximate
| (| + 1)
r
2
·
| (| + 1)
a
2
, (9.21)
where a is the earth radius. Then
n
|
(r) = cos
__
/
2
÷
| (| + 1)
a
2
(r ÷a)
_
, (9.22)
and the condition 1
0
= 0 at r = a yields a resonance frequency,
. =
_
| (| + 1)
c
a
. (9.23)
The fundamental resonance (known as Schumann resonance) frequency is approximately 10 Hz.
(Observed frequency is approximately 8 Hz.) The Schumann mode can be excited by lightning
5
discharges and is being continuously monitored as one of the most important electromagnetic
phenomenon of global scale. Note that the resonance frequency is essentially the inverse of the
transit time around the globe, t = 2¬a,c.
9.3 Perturbative Change in Cavity Resonance Frequency
For a given eigenmode, the eigenvalue /
2
of a cavity mode can be found from
/
2
0
=
_
\
(\E)
2
d\ ÷2
_
S
(n E) (\E)do
_
\
E
2
d\
, (9.24)
subject to the condition that \ E = 0 in the volume and n E = 0 on the wall surface. Note
that the normal vector n is directed away from the cavity volume. The surface integral evidently
vanishes for exact eigenmodes but is retained here for the purpose of variational calculation because
trial functions may not exactly satisfy the boundary conditions. If the cavity surface is only slightly
deformed, the unperturbed electric …eld may still be used in the majority of region except at the
deformation. Let us assume a small dent of volume \ and surface area o. The new eigenvalue
to order \ and o is
/
2
=
_
\ \
(\E)
2
d\ ÷2
_
S
0
+S
(n E) (\E)do
_
\ ·
E
2
d\
.
=
_
\
(\E)
2
d\ ÷
_
\
(\E)
2
d\ ÷2
_
S
0
+S
(n E) (\E)do
_
\
E
2
d\ ÷
_
\
E
2
d\
·
_
\
(\E)
2
d\
_
\
E
2
d\
_
1 +
_
\
E
2
d\
_
\
E
2
d\
_
÷
_
\
(\E)
2
d\
_
\
E
2
d\
÷2
_
S
(n E) (\E)do
_
\
E
2
d\
= /
2
0
_
1 +
_
\
E
2
d\
_
\
E
2
d\
_
÷
_
\
(\E)
2
d\
_
\
E
2
d\
÷2
_
S
(n E) (\E)do
_
\
E
2
d\
, (9.25)
6
where o
0
is the unperturbed surface on which n E = 0 and o is the surface of \ facing the
cavity interior. The surface integral can be converted into volume integral as
_
S
(n E) (\E)do = ÷
_
S
(n
0
E) (\E)do
= ÷
_
S
n
0
[E(\E)]do
= ÷
_
\
\ [E(\E)]d\
= ÷
_
\
(\E)
2
d\ +/
2
0
_
\
E
2
d\ (9.26)
where n
0
= ÷n is the normal vector directed away from \. Then, the change in the eigenvalue is
/
2
=
_
\
[(\E)
2
÷/
2
0
E
2
]d\
_
\
E
2
d\
. (9.27)
Since the electric and magnetic energy are identical in the cavity, we …nally …nd
/
2
· /
2
0
_
_
\
H
2
d\
_
\
H
2
d\
÷
_
\
E
2
d\
_
\
E
2
d\
_
, (9.28)
where the …rst term indicates the fraction of magnetic energy removed and second term the fraction
of electric energy removed from the cavity. Corresponding change in the resonance frequency is
.
.
0
·
1
2
_
_
\
H
2
d\
_
\
H
2
d\
÷
_
\
E
2
d\
_
\
E
2
d\
_
. (9.29)
This result indicates that if a dent is made at a position on the inner wall where the magnetic energy
is concentrated, the eigenvalue (and thus the frequency) increases through an e¤ective decrease in
the inductance. If a dent is made where the electric energy is dominant, the resonance frequency
decreases through an e¤ective increase in the capacitance. This is analogous to a perturbation in
inductance and capacitance. If a conductor is inserted into an inductor, the inductance decreases
and if a conductor is inserted in a capacitor, the capacitance increases.
If a small object is placed in the volume of a cavity, it will perturb the electromagnetic …elds.
A resultant frequency shift may be analyzed in terms of dipole approximation for a body whose
dimension is su¢ciently small compared with the wavelength. For example, a conductor sphere of
7
Figure 9-1: A dent on a cavity wall with volume \ and area o. o
0
is the unperurbed area on
which n E
0
= 0 with E
0
the unperturbed …eld.
radius a induces an electric dipole moment
p = 4¬-
0
a
3
E(r
c
), (9.30)
and magnetic dipole moment
m = ÷2¬a
3
H(r
c
), (9.31)
where E(r
c
) and H(r
c
) are the …elds at the location of the sphere. The perturbed electric energy
is
c
c
= ÷p E(r
c
) = ÷4¬-
0
a
3
E
2
(r
c
), (9.32)
and perturbed magnetic energy is
c
n
= ÷m B(r
c
) = +2¬j
0
a
3
H
2
(r
c
). (9.33)
Then the magnetic energy removed is
÷2¬j
0
a
3
H
2
(r
c
).
and the frequency change is given by
.
.
0
= ÷
2-
0
E
2
(r
c
) +j
0
H
2
(r
c
)
-
0
_
\
E
2
d\
¬a
3
. (9.34)
8
9.4 TE and TM Modes in Conductor Waveguides
In an air-…lled waveguide, the electromagnetic …elds satisfy the vector wave equations,
_
\
2
÷
1
c
2
0
2
0t
2
_
E = 0, (9.35)
_
\
2
÷
1
c
2
0
2
0t
2
_
H = 0. (9.36)
Since we are interested in waves propagating along the waveguide extended in the .-direction, the
. and time dependence can be singled out in the form
E(r, t) = E(r
?
)c
i(Iz:.t)
, H(r, t) = H(r
?
)c
i(Iz:.t)
, (9.37)
where r
?
indicates the coordinates transverse to the .-axis. The axial components 1
:
and H
:
both
satisfy scalar wave equation,
_
\
2
÷
1
c
2
0
2
0t
2
_
1
:
= 0, (9.38)
_
\
2
÷
1
c
2
0
2
0t
2
_
H
:
= 0, (9.39)
Finding solutions to the axial …elds 1
:
and H
:
is su¢cient to determine electromagnetic …elds in
a waveguide, for the Maxwell’s equations
\E = i.j
0
H, \H = ÷i.-
0
1, (9.40)
yield transverse components entirely in terms of the axial components as follows,
E
?
=
i
(.,c)
2
÷/
2
:
(/
:
\
?
1
:
+.j
0
\
?
H
:
e
:
) , (9.41)
H
?
=
i
(.,c)
2
÷/
2
:
(/
:
\
?
H
:
÷.-
0
\
?
1
:
e
:
) . (9.42)
Modes having no axial electric …elds, 1
:
= 0, are called Transverse Electric (TE) modes and modes
having no axial magnetic …elds, H
:
= 0, are called Transverse Magnetic (TM) modes. As for
radiation …elds, these two modes form basic eigenfunctions.
9
9.5 Rectangular Waveguides
Figure 9-2: Rectangular waveguide.
We assume a conductor tube having a cross-section a /. The conductor walls are assumed to
be ideally conducting, that is, the skin-depth
c ==
_
2
.oj
0
=
1
_
¬)j
0
o
, (9.43)
is su¢ciently small and Ohmic dissipation is negligible. Later, we will relax this assumption in
evaluating the damping factor of electromagnetic waves con…ned in a waveguide. However, for
analyzing mode structure, the assumption of ideal conductor does not introduce signi…cant errors.
For TE modes, it is su¢cient to …nd the axial magnetic …eld H
:
(r, t) = H
:
(r, j)c
i(Iz:.t)
which
obeys the following Helmholtz equation,
_
0
2
0r
2
+
0
2
0j
2
+
_
.
c
_
2
÷/
2
:
_
H
:
(r, j) = 0. (9.44)
The boundary condition for H
:
(r, j) can be found from the vanishing tangential components of the
electric …eld at the wall of the waveguide,
\
?
H
:
(r, j) = 0 at r = 0, a and j = 0, /. (9.45)
10
This condition is satis…ed if we choose
H
:
(r, j) = H
0
cos
_

a
r
_
cos
_

/
j
_
, (9.46)
where : and : are integers. Either : or : can be zero but not simultaneously. Substitution into
Eq. (9.44) yields a dispersion relation,
.
2
= c
2
_
_

a
_
2
+
_

/
_
2
+/
2
:
_
, (9.47)
which determines the axial wavenumber /
:
for a given wave frequency . and given dimensions of
the waveguide. The minimum frequency allowed for wave propagation (/
:
0) in a waveguide
occurs at /
:
= 0 and is given by
.
cna
= c
_
_

a
_
2
+
_

/
_
2
. (9.48)
This is called the cuto¤ frequency of the TE
na
mode. If a /, the smallest cuto¤ frequency is
that of the TE
10
mode (: = 1, : = 0),
.
c10
=
¬c
a
, or )
c10
=
c
2a
(Hz). (9.49)
The TE
10
mode is most commonly used in microwave communication.
The phase and group velocities along the wave guide are given by
.
/
:
=
c
_
1 ÷(.
cna
,.)
2
c, (9.50)
d.
d/
:
= c
_
1 ÷(.
cna
,.)
2
< c. (9.51)
In the waveguide, waves propagate along a zig-zag path being re‡ected at the walls. The velocity
along the zig-zag path is simply c, for the total wavenumber is
/ =
_
_

a
_
2
+
_

/
_
2
+/
2
:
. (9.52)
The phase velocity along the .-axis larger than c is due to a /
:
which is smaller than /. The group
velocity along the .-axis corresponds to the axial component of c.
11
For TM modes, the axial electric …eld 1
:
(r, j) satisfying the Helmholtz equation
_
0
2
0r
2
+
0
2
0j
2
+
_
.
c
_
2
÷/
2
:
_
1
:
(r, j) = 0, (9.53)
and the boundary conditions
1
:
= 0 at r = 0, a and j = 0, / (9.54)
is given by
1
:
(r, j) = 1
0
sin
_

a
r
_
sin
_

/
j
_
, (9.55)
where : and : are nonzero integers. The cuto¤ frequency of the TM mode is
.
cna
= c
_
_

a
_
2
+
_

/
_
2
, : _ 1, : _ 1. (9.56)
9.6 TE
10
Mode
In this section, some properties of the TE
10
mode will be studied. We assume an electric …eld in
the form
1
j
(r, ., t) = 1
0
sin
_
¬
a
r
_
c
i(Iz:.t)
. (9.57)
Corresponding magnetic …eld can be found from
H =
1
i.j
0
\E
= ÷
/
:
.j
0
1
0
sin
_
¬
a
r
_
c
i(Iz:.t)
e
a
÷i
¬,a
.j
0
cos
_
¬
a
r
_
c
i(Iz:.t)
e
:
. (9.58)
The magnetic …eld pro…le seen from top is shown below for the case /
:
= ¬,a. Also shown is the
pro…le of the surface current on the lower wall,
J
c
= n H. (9.59)
12
x
1 0.8 0.6 0.4 0.2 0
z
2
1.5
1
0.5
0
Figure 9-3: Magnetic …eld pro…le of the TE
10
mode in a rectangular waveguide.
The power transmitted in the waveguide can be found by integrating the Poynting vector in the
axial direction given by
o
:
= ÷1
j
H

a
=
/
:
.j
0
1
2
0
sin
2
_
¬
a
r
_
, (9.60)
1 =
_
o
0
dr
_
b
0
dj o
:
(r) (9.61)
=
/
:
.j
0
1
2
0
_
o
0
sin
2
_
¬
a
r
_
_
b
0
dj (9.62)
=
/
:
.j
0
1
2
0
a/
2
, (peak value). (9.63)
The rms power is
1
vnc
=
/
:
.j
0
1
2
0
a/
4
(9.64)
=
a/
4
_
-
0
j
0
_
1 ÷(.
c10
,.)
2
1
2
0
(9.65)
=
a/
4
1
2
0
7
TE
10
, (W) (9.66)
13
x
1 0.8 0.6 0.4 0.2 0
z
2
1.5
1
0.5
0
Figure 9-4: Surface current density pattern on the upper inner surface.
where
7
TE
10
=
_
j
0
,-
0
_
1 ÷(.
c10
,.)
2
, (9.67)
is the impedance of the TE
10
mode.
In practice, the …nite (not in…nite) conductivity of the wall causes dissipation of electromagnetic
energy and the power decays as the wave propagates along the waveguide. To evaluate the damping
factor, we recall that the Poynting ‡ux normal to a conductor wall is
o
?
= 7 [H
t
[
2
=
_
÷i.j
0
o
[H
t
[
2
, (9.68)
where
7
c
=
_
÷i.j
0
o
, (9.69)
is the surface impedance of the conductor and H
c
is the tangential component of the magnetic …eld
at the wall surface. The real part of o
?
indicates net dissipation,
Re
_
÷i.j
0
o
[H
t
[
2
=
_
.j
0
2o
[H
t
[
2
. (9.70)
14
For the TE
10
mode, the tangential component of the magnetic …eld at the side walls at r = 0 and
a is
H
:
(r = 0, a) = (i
¬,a
.j
0
1
0
c
i(Iz:.t)
. (9.71)
Therefore, the time averaged loss rate per unit length of the side walls is
÷
d1
1
d.
= 2
_
.j
0
2o
_
¬,a
.j
0
_
2
1
2
0
2
/. (9.72)
Similarly, on the top and bottom walls,
÷
d1
2
d.
= 2
_
.j
0
2o
_
_
¬,a
.j
0
_
2
+
_
/
:
.j
0
_
2
_
1
2
0
4
a
=
1
2
-
0
j
0
_
.j
0
2o
1
2
0
a, (9.73)
and the total loss rate is
÷
d1
d.
=
-
0
j
0
_
.j
0
2o
_
/
_
.
c
.
_
2
+
a
2
_
1
2
0
. (9.74)
The power damping factor is given by
÷
1
1
d1
d.
= 2
_
.-
0
2o
a + 2/(.
c
,.)
2
a/
_
1 ÷(.
c
,.)
2
= 2c, (9.75)
and the …eld intensity decays as 1
0
c
c:
along the waveguide. For example, the damping factor in a
copper waveguide (o · 6 10
7
S/m) having a cross-section 2 1 cm
2
excited at a frequency ) = 9
GHz is approximately 2c = 0.04/m. The microwave power c-folds in a distance of 1,2c = 25 m.
Example 1 Rectangular Cavities
Electromagnetic …elds con…ned in a rectangular wave guide closed at . = 0 and 1 form complete
standing waves in all directions, r.j, .. We assume 1 a / without loss of generality. It is
su¢cient to consider TM modes because after suitable coordinates changes, TM modes can be
recovered from TE modes. The axial electric …eld 1
:
(r, t) must vanish at r = 0, a and at j = 0, /
and thus may be assumed to be
1
|na
:
(r, j) = 1
0
sin
_

a
r
_
sin
_

/
j
_
cos
_

1
.
_
, (9.76)
15
where cos
_

1
.
_
is chosen so that r and j components of the electric …eld vanish at . = 0 and 1.
The resonance frequency of (|, :, :) is thus given by
.
|na
= c
_
_

a
_
2
+
_

/
_
2
+
_

1
_
2
, (9.77)
where :, : are nonzero integers and | = 0, 1, 2, The lowest resonance frequency is
.
011
= c
_
_
¬
a
_
2
+
_
¬
/
_
2
. (9.78)
9.7 Excitation of Rectangular TE Modes
Electromagnetic waves in waveguides can be excited by a current source placed in the guide. To
illustrate the general methodology, we consider a thin vertical current placed at r = a,2 in a
rectangular waveguide. The j component of the wave equation,
_
\
2
÷
1
c
2
0
2
0t
2
_
E =
1
-
0
\j +j
0
0J
0t
. (9.79)
is
_
\
2
÷
1
c
2
0
2
0t
2
_
1
j
= j
0
0J
j
0t
, (9.80)
where the charge density j has been ignored. Since the transverse electric …eld of TE modes can
be generated by the axial magnetic …eld H
:
as
E
?
=
j
0
/
2
?
0
0t
\
?
H
:
e
:
, (9.81)
we obtain
1
j
= ÷
j
0
/
2
?
0
0t
0H
:
0r
. (9.82)
Substitution into Eq. (9.82) yields
_
\
2
÷
1
c
2
0
2
0t
2
_
1
/
2
?
0H
:
0r
= ÷J
j
= ÷1(t)c
_
r ÷
a
2
_
c(.). (9.83)
16
Let us assume TE
n0
modes are excited by the current. Then the axial magnetic …eld may be
expanded as
H
:
(r, ., t) =
1

n=1
¹
a
(., t) cos
_
¬:
a
r
_
, (9.84)
where ¹
n
(., t) is an expansion function and the eigenfunction for TE
na
mode is recalled,
H
na
:
(r, j) = cos
_

a
r
_
cos
_

/
j
_
. (9.85)
Multiplying Eq. (9.83) by
sin
_
¬:
0
a
r
_
,
and integrating over the cross section area a /, we …nd
_
\
2
:
÷
1
c
2
0
2
0t
2
÷/
2
n0
_
¹
n0
(., t) =
2¬:
a
2
1(t)c(.) sin
_

2
_
, (9.86)
with /
2
n0
= (:¬,a)
2
.
If the current is an impulse, 1(t) = ¡
0
c(t) where ¡
0
is the amount of charge transferred, the
Laplace transform of ¹
n0
(., t) obeys
_
d
2
d.
2
÷
:
2
c
2
÷/
2
n0
_
¹
n0
(., :) =
2¬:
a
2
¡
0
c(.) sin
_

2
_
, (9.87)
which can be solved as
¹
n0
(., :) =
¬:¡
0
a
2
1
_
(:,c)
2
+/
2
n0
exp
_
÷
_
(:,c)
2
+/
2
n0
[.[
_
sin
_

2
_
. (9.88)
Inverse Laplace transformation yields
¹
n0
(., t) =
_
_
_

0
a
J
0
_

a
_
(ct)
2
÷.
2
_
sin
_

2
_
, ct [.[
0, ct < [.[
(9.89)
Note that the pulse front propagates at the speed c. Of course, the pulse shape is severely distorted
as it propagates over a large distance because of the dispersion of the electromagnetic modes
con…ned in a waveguide.
For a steady oscillating current 1(t) = 1
0
c
i.t
, we may assume ¹
n0
(., t) = ¹(.)c
i.t
, where
17
¹
n0
(.) satis…es
_
d
2
d.
2
+
.
2
c
2
÷/
2
n0
_
¹
n0
(.) =
2¬:
a
2
1
0
c(.) sin
_

2
_
. (9.90)
Noting
d
2
d.
2
c
iIj:j
= 2i/c(.) ÷/
2
c
iIj:j
,
we …nd the following steady state solution,
¹
n0
(.) =
¬:1
0
ia
2
_
(.,c)
2
÷/
2
n0
sin
_

2
_
c
i
_
(.¸c)
2
I
2
m0
j:j
. (9.91)
9.8 Circular Waveguide
Figure 9-5: Circular waveguide. The lower …gure shows qualitatively the electric and magnetic …eld
pro…les.
Let us now consider electromagnetic waves con…ned in a conductor tube of radius a. The
Helmholtz equation for the axial electric and magnetic …elds is
_
0
2
0j
2
+
1
j
0
0j
+
1
j
2
0
2
0c
2
+
_
.
c
_
2
÷/
2
:
_
_
_
1
:
H
:
_
_
= 0. (9.92)
18
General solutions bounded at the axis j = 0 are
1
:
(j, c), H
:
(j, c) = J
n
(/j)c
inç
, (9.93)
where : is an integer and
/
2
=
_
.
c
_
2
÷/
2
:
0. (9.94)
(The case /
2
< 0 will be considered later for slow wave circular waveguides.) For TE modes, the
boundary condition H
j
(j = a) = 0 requires
d
dj
J
n
(/j) = 0 at j = a, (9.95)
and for TM modes, H
:
(j = a) = 0 requires J
n
(/a) = 0. In the table, roots of J
0
n
(r) = 0 relevant
to TE modes and roots of J
n
(r) = 0 relevant to TM modes are shown. [Note the special case
J
0
0
(r) = ÷J
1
(r).]
Roots of J
0
n
(r) = 0
: = 0 1 2
1st 3.832 1.841 3.053
2nd 7.016 5.331 6.706
3rd 10.173 8.536 9.969
Roots of J
n
(r) = 0
: = 0 1 2
1st 2.405 3.832 5.136
2nd 5.520 7.016 8.417
3rd 8.654 10.173 11.620
The dispersion relation of the TE modes is
/a = a
_
_
.
c
_
2
÷/
2
:
= r
0
na
, (9.96)
or
.
2
= c
2
_
_
r
0
na
a
_
2
+/
2
:
_
, (9.97)
and the cuto¤ frequency is
.
c
=
c
a
r
0
na
. (9.98)
The smallest root is the …rst root of J
0
1
(r) = 0, r
11
· 1.841. This is the fundamental mode in a
circular waveguide and corresponds to the TE
10
mode in a rectangular waveguide. Through gradual
tapering, the rectangular TE
10
mode can be converted to circular TE
11
mode.
19
Example 2 Cylindrical Cavity
Consider TE modes in a circular waveguide closed at . = 0, 1.The axial magnetic …eld may be
assumed to be
H
:
(r) = H
0
J
n
(/j)c
inç
sin
_

1
.
_
,
where
/ =
_
_
.
c
_
2
÷/
2
:
.
From
E
?
=
÷i
(.,c)
2
÷/
2
:
(/
:
\
?
1
:
+.j
0
\
?
H
:
e
:
)
=
÷i.j
0
(.,c)
2
÷/
2
:
\
?
H
:
e
:
,
we …nd the transverse electric …eld,
1
?
=
÷i.j
0
(.,c)
2
÷/
2
:
_
i:
j
H
:
c
j
÷
0H
:
0j
c
ç
_
.
From the condition that 1
ç
(j = a) = 0, it follows that
/
na|
=
_
_
.
na|
c
_
2
÷
_

1
_
2
=
r
0
na
a
,
where r
0
na
is the :-th root of J
0
n
(r) = 0. The resonance frequency is
.
na|
= c
_
_
r
0
na
a
_
2
+
_

1
_
2
.
The transverse magnetic …eld is
H
?
=
i/
:
(.,c)
2
÷/
2
:
\
?
H
:
=
i/
:
(.,c)
2
÷/
2
:
_
0H
:
0j
e
j
+
i:
j
H
:
e
ç
_
.
The TE
0a1
mode is commonly used in practical applications because the surface current on the end
walls is purely azimuthal (J
ç
only as can be seen from J
c
= n H) and thus a thin circular gap
20
does not signi…cantly perturb the …elds (and resonance frequency). The resonance frequency can
be controlled by varying the length 1 and a small but …nite gap cannot be avoided in moveable
plungers. The TE
0a1
is evidently independent of c and thus degeneracy problems between cos :c
and sin:c modes can also be avoided.
9.9 TM Mode in Circular Slow Waveguide
Waveguides used in linear electron accelerators must accommodate TM modes having a phase
velocity close to c. Waveguides with smooth inner walls can only accommodate modes having
phase velocities larger than c and thus cannot be used for this purpose. Modes must be TM
because TE modes have no electric …eld in the axial direction needed to accelerate electrons.
Figure 9-6: Circular slow waveguide of radius a with diaphragms of radius /. The spacing between
diaphragms is much smaller than the axial wavelength `
:
.
Slow waveguides have conductor diaphragms placed periodically along the axis as shown in
Fig.9-7. The purpose of the diaphragms is to increase the capacitance per unit length of the
waveguide which contributes to slowing down the phase velocity of electromagnetic waves. We
consider modes symmetric about the axis, : = 0. The waveguide has a radius a and diaphragms
have holes of radius /. In the region j < /, the Helmholtz equation for the axial electric …eld is
_
0
2
0j
2
+
1
j
0
0j
+
_
.
c
_
2
÷/
2
:
_
1
:
(j) = 0, j < /. (9.99)
Since we are interested in modes having an axial phase velocity slightly smaller than c, that is,
.
/
:
. c, (9.100)
21
the quantity (.,c)
2
÷/
2
:
must be negative, and thus solution for 1
:
(j) may be assumed to be
1
:
(j) = ¹1
0
(/j), j < / (9.101)
where
/
2
= /
2
:
÷
_
.
c
_
2
0. (9.102)
In the diaphragm region / < j < a, the electric …elds lines are essentially straight provided the
axial period of the diaphragms is su¢ciently smaller than the axial wavelength. We assume that
this condition is met. Then, the wave equation in the region / < j < a may be approximated by
_
0
2
0j
2
+
1
j
0
0j
+
_
.
c
_
2
_
1
:
(j) = 0, / < j < a. (9.103)
General solutions are
1
:
(j) = 1J
0
_
.
c
j
_
+C·
0
_
.
c
j
_
, / < j < a. (9.104)
The boundary conditions are:
1
:
(j = a) = 0, (9.105)
and
1
:
and H
ç
be continuous at j = /. (9.106)
These boundary conditions yield
1J
0
_
.
c
a
_
+C·
0
_
.
c
a
_
= 0, (9.107)
¹1
0
(//) = 1J
0
_
.
c
/
_
+C·
0
_
.
c
/
_
, (9.108)
¹
/
1
1
(//) =
c
.
_
1J
1
_
.
c
/
_
+C·
1
_
.
c
/
__
, (9.109)
where J
0
0
(r) = ÷J
1
(r), ·
0
0
(r) = ÷·
1
(r), 1
0
0
(r) = 1
1
(r) are noted. Eqs. (9.107) through (9.109)
give the following dispersion relation
c/
.
1
0
(//)
1
1
(//)
=
J
0
_
.
c
a
_
·
0
_
.
c
/
_
÷J
0
_
.
c
/
_
·
0
_
.
c
a
_
J
1
_
.
c
/
_
·
0
_
.
c
a
_
÷J
0
_
.
c
a
_
·
1
_
.
c
/
_. (9.110)
22
For the purpose of accelerating highly relativistic electrons, the axial phase velocity .,/
:
must be
close to c or / · 0. Then 1
0
(//) · 1, 1
1
(//) · //,2, and the LHS of Eq. (9.110) reduces to 2c,./.
For a given rf frequency . and the size of the waveguide a, the dispersion relation
2c
./
·
J
0
_
.
c
a
_
·
0
_
.
c
/
_
÷J
0
_
.
c
/
_
·
0
_
.
c
a
_
J
1
_
.
c
/
_
·
0
_
.
c
a
_
÷J
0
_
.
c
a
_
·
1
_
.
c
/
_, (9.111)
can be solved numerically to determine the aspect ratio a,/ of a slow wave circular waveguide.
Fig.9-7 shows the function
)(r) =
2a
r/
÷
J
0
(r) ·
0
_
/
a
r
_
÷J
0
_
/
a
r
_
·
0
(r)
J
1
_
/
a
r
_
·
0
(r) ÷J
0
(r) ·
1
_
/
a
r
_, r =
.a
c
, (9.112)
when a,/ = 2.5. The …rst root occurs at r · 3.89 and for a given rf frequency ., the outer radius
a can thus be determined.
x
4 3.9 3.8 3.7 3.6 3.5
1
0.5
0
-0.5
-1
Figure 9-7: Root of )(r) = 0 when a,/ = 2.5.
9.10 Dielectric Waveguides
An optical …ber can con…ne light waves because of total re‡ection at the surface. In contrast to
conductor waveguides, light waves in optical waveguides cannot be pure TE or TM modes. This
is because electromagnetic …elds outside, as well as inside, the optical …ber must be considered
23
simultaneously. Although the outer …elds are evanescent (otherwise waves cannot be con…ned), the
…elds near the surface do a¤ect those inside.
We …rst consider a simple case of step change in the index of refraction,
:(j) =
_
_
_
:, j < a,
1, j a.
Such an optical …ber is of no practical interest, for …bers used in optical communication all have
graded index of refraction with a gradual change with the radius j. The axial electric …eld 1
:
(r, t)
satis…es the wave equations in both regions,
_
0
2
0j
2
+
1
j
0
0j
+
1
j
2
0
2
0c
2
+
0
2
0.
2
+j
0
-.
2
_
1
<
:
(r) = 0, j < a, (9.113)
_
0
2
0j
2
+
1
j
0
0j
+
1
j
2
0
2
0c
2
+
0
2
0.
2
+j
0
-
0
.
2
_
1

:
(r) = 0, j a. (9.114)
The azimuthal dependence may be assumed to be c
inç
and the axial dependence c
iIz:
,
1(r) = 1(j)c
inç+iIz:
.
Then,
_
d
2
dj
2
+
1
j
d
dj
÷
:
2
j
2
+j
0
-.
2
÷/
2
:
_
1
<
:
(j) = 0, j < a, (9.115)
_
d
2
dj
2
+
1
j
d
dj
÷
:
2
j
2
+j
0
-
0
.
2
÷/
2
:
_
1

:
(j) = 0, j a, (9.116)
which admit the following bounded solutions,
1
<
:
(j) = ¹J
n
(/
1
j), j < a, (9.117)
1

:
(j) = 11
n
(/
2
j), j a. (9.118)
Here
/
1
=
_
j
0
-.
2
÷/
2
:
, /
2
=
_
/
2
:
÷j
0
-
0
.
2
. (9.119)
Note that the outer …eld should be evanescent for the waveguide to con…ne light waves.
24
Similarly, the axial magnetic …eld H
:
(r) may be assumed to be
H
<
:
(j) = CJ
n
(/
1
j), j < a, (9.120)
H

:
(j) = 11
n
(/
2
j), j a. (9.121)
The transverse …elds E
?
and H
?
can then be calculated by referring to Eqs. (9.41) and (9.42).
The azimuthal components of the …elds are
1
<
ç
(j) =
÷i
/
2
1
_
/
:
i:
j
¹J
n
(/
1
j) ÷.j
0
/
1
CJ
0
n
(/
1
j)
_
, (9.122)
1

ç
(j) =
i
/
2
2
_
/
:
i:
j
1J
n
(/
2
j) ÷.j
0
/
2
11
0
n
(/
2
j)
_
, (9.123)
H
<
ç
(j) =
i
/
2
1
_
/
:
i:
j
CJ
n
(/
1
j) +.-/
1
¹J
0
n
(/
1
j)
_
, (9.124)
and
H

ç
(j) =
÷i
/
2
1
_
/
:
i:
j
11
n
(/
2
j) +.-
0
/
2
11
0
n
(/
1
j)
_
. (9.125)
The continuity of 1
:
, 1
ç
, H
:
and H
ç
yields
¹J
n
(/
1
a) = 11
n
(/
2
a), (9.126)
CJ
n
(/
1
a) = 11
n
(/
2
a), (9.127)
÷
1
/
2
1
_
/
:
i:
a
¹J
n
(/
1
a) ÷.j
0
/
1
CJ
0
n
(/
1
a)
_
=
1
/
2
2
_
/
:
i:
a
11
n
(/
2
a) ÷.j
0
/
2
11
0
n
(/
2
a)
_
, (9.128)
1
/
2
1
_
/
:
i:
a
CJ
n
(/
1
a) +.-/
1
¹J
0
n
(/
1
a)
_
= ÷
1
/
2
2
_
/
:
i:
a
11
n
(/
2
a) +.-
0
/
2
11
0
n
(/
1
a)
_
. (9.129)
Then the determinantal dispersion relation is
_
1
/
2
1
+
1
/
2
2
_
2
_
:/
:
a
_
2
=
_
.
/
1
c
1
_
2
_
J
0
n
(/
1
a)
J
n
(/
1
a)
_
2
+
_
.
/
2
c
2
_
2
_
1
0
n
(/
2
a)
1
n
(/
2
a)
_
2
+
.
2
/
1
/
2
_
1
c
2
1
+
1
c
2
2
_
J
0
n
(/
1
a)1
0
n
(/
2
a)
J
n
(/
1
a)1
n
(/
2
a)
. (9.130)
Figure 9-8 shows the dispersion relation, namely, axial wavenumber /
:
normalized by /
0
= .,c
as a function of the normalized frequency /
0
a = .a,c when : = 1 and : = 1.1. It can be seen that
25
a cuto¤ occurs at .a,c · 1.386. Fig.9-9 shows the case when : = 2. The cuto¤ frequency increases
to .a,c · 5.43.
x
30 25 20 15 10 5 0
1.1
1.08
1.06
1.04
1.02
1
Figure 9-8: c/
:
,. vs. .a,c when : = 1.1, : = 1. The cuto¤ frequency is .
c
a,c · 1.386.
x
30 25 20 15 10 5 0
1.1
1.08
1.06
1.04
1.02
1
Figure 9-9: c/
:
,. vs. .a,c when : = 1.1, : = 2. The cuto¤ frequency is .
c
a,c · 5.43.
26
In optical …bers used in practical communication, the index of refraction is designed to have gradual,
rather than step, variation with the radius. Quadratic variation is commonly employed,
:(j) = :
0
_
1 ÷c
2
j
2
¸
, c = constant (9.131)
because the electromagnetic …elds are then well con…ned with a Gaussian pro…le c
o
2
j
2
. The
corresponding permittivity is
-(j) = -
0
:
2
0
_
1 ÷c
2
j
2
¸
2
. (9.132)
Since
\ [-(r)E] = -(r)\ E+E \- = 0, (9.133)
the wave equation
\\E = .
2
j
0
-(r)E, (9.134)
reduces to
\
2
E+.
2
j
0
-(r)E+\
_
E \-
-
_
= 0. (9.135)
If the change in the permittivity is small, the last term can be ignored in the lowest order approx-
imation, and we obtain a simple wave equation with an inhomogeneous permittivity,
\
2
E+.
2
j
0
-(r)E · 0. (9.136)
In the cylindrical geometry, a cartesian component of the transverse electric …eld satis…es
_
0
2
0j
2
+
1
j
0
0j
+
1
j
2
0
2
0c
2
+
0
2
0.
2
+.
2
j
0
-(j)
_
1
i
= 0. (9.137)
For weak variation of -(j),
-(j) = -
0
:
2
0
(1 ÷c
2
j
2
)
2
· -
0
:
2
0
_
1 ÷2c
2
j
2
_
,
27
and axially symmetric mode 0,0c = 0, Eq. (9.137) reduces to
_
d
2
dj
2
+
1
j
d
dj
÷/
2
:
+/
2
0
_
1 ÷2c
2
j
2
_
_
1(j) = 0, (9.138)
where
/
2
0
= .
2
-
0
j
0
:
2
0
. (9.139)
Assuming
1(j) = 1
0
c
o
2
j
2
,
we …nd
a
2
=
1
_
2
/
0
c, (9.140)
/
2
:
= /
2
0
÷
4
_
2
/
0
c. (9.141)
The electric …eld is con…ned with a Gaussian pro…le in the radial direction. The c-folding radial
distance is
n =
1
a
=
4
_
2
_
/
0
c
, (9.142)
which is called beam radius. A constant beam radius is maintained only for appropriate injection
of light wave at the input end. If not, the beam radius varies with the axial distance accompanied
by periodic focusing and defocusing as intuitively expected from the picture of repeated total
re‡ections.
28
Problems
1. A waveguide has a semicircular cross section of radius a. Determine the lowest cuto¤ frequency.
2. Prove Eqs. (??) and (??).
3. Each corner of a rectangular waveguide having a cross section a / is smoothed as shown
with c ¸a, /. Find the change in the cuto¤ frequency of the TE
10
mode.
4. The cuto¤ frequency of a ridge waveguide shown may be estimated from equivalent capaci-
tance and inductance,
C
|
· -
0
d
q
(F/m), 1 | =
1
2
j
0
a/ (H m)
.
c
=
1
_
1C
= c
_
2q
a/d
.
Explain.
5. Thin diaphragms along the wide walls of a rectangular waveguide act as an e¤ective ca-
pacitance, while diaphragms along the narrow walls act as an e¤ective inductance. Give
qualitative explanations.
6. In the magic T shown, TE
10
mode entering port 3 (or 4) is equally divided between ports
1 and 2 but does not come out of port 4 (or 3). Explain in terms of cuto¤ of higher order
modes.
7. Two waveguides are coupled through two small holes drilled in the wider walls separated by
a quarter wavelength in each direction as shown. TE
10
mode entering port 1 comes out of
ports 2 and 4, but not port 3. Explain.
8. Show that the dispersion relation of TM modes in a coaxial cable having inner and outer radii
a and / is given by
J
n
(/a)·
a
(//) ÷J
n
(//)·
n
(/a) = 0,
where
/
2
=
_
.
c
_
2
÷/
2
:
.
When a = 1 mm, / = 4 mm, what are cuto¤ frequencies? Consider : = 0 and 1.
29
9. Repeat the preceding problem for TE modes to show that the dispersion relation in this case
is
J
0
n
(/a)·
0
a
(//) ÷J
0
n
(//)·
0
n
(/a) = 0.
10. Wave propagation along a conducting helix with a pitch angle c is expected to be characterized
by the axial propagation speed,
.
/
:
· c sinc,
because electromagnetic waves are guided along the spiral path of the helix. Show that an
approximate dispersion relation is given by
_
.
c
_
2
cot
2
c = /
2
1
0
(/a)1
0
(/a)
1
1
(/a)1
1
(/a)
,
where
/
2
= /
2
:
÷
_
.
c
_
2
.
30