Chapter 5

Cylindrical Cavities and Waveguides

We shall consider an electromagnetic field propagating inside a hollow (in the present case
cylindrical) conductor. There are no sources inside the conductor, but we shall assume the
material is isotropic with electric permittivity , and magnetic permeability, . The speed of
the propagating wave is 1/  The direction of propagation will be along the cylindrical axis
which is the ẑ direction We shall assume that E(r, t) = E(r)e −it and B(r, t) = B(r)e −it .
Maxwell’s equations give:
[∇ 2 −  ∂ 2 ]E(r. t) = 0
2
5.1
∂t
[∇ 2 −  ∂ 2 ]B(r. t) = 0
2
5.2
∂t
[∇ 2 +  2 ]B(r) = 0 5.3
[∇ 2 +  2 ]E(r) = 0 5.4

Since the wave is propagating along the ẑ direction we shall further assume that:
∇ × E = iB(r); 5.5
∇ × B = −iE(r). 5.6

Since the wave is propagating along the ẑ direction we shall further assume that:
E(r) = E(x, y)e ±ikz 5.7a
B(r) = B(x, y)e ±ikz 5.7b
Thus Eq. (5.3 and 5.4) become
[∇ 2t +  2 − k 2 ]E(x, y) = 0 5.8b
[∇ 2t +  2 − k 2 ]B(x, y) = 0 5.8a
where
 ∇ 2t = ∂ 2 + ∂ 2
2 2

∂x ∂y
∇ t = x̂ ∇ x + ŷ∇ y

The expressions in Eqs. 5.5 and 5.6 then become;

∇ × E = [ẑ ∇ z + ∇ t ]×[ẑ E z + E t ]=iB(r) where

E t = E − ẑ E z = (ẑ × E) × ẑ
B t = B − ẑ B z = (ẑ × B) × ẑ
Then
∇ × E= ẑ ∇ z ×E t − ẑ × ∇ t E z + ∇ t ×E t = i(B t + ẑ B z )
∇ × B= ẑ ∇ z ×B t − ẑ × ∇ t B z + ∇ t ×B t = −i(E t + ẑ E z )
Thus,
ẑ ×∇ z E t − ẑ × ∇ t E z = −iẑ ×(ẑ × B)
∇ z E t + i(ẑ × B t ) = ∇ t E z 5.9
ẑ  (∇ t × E t ) = iB z 5.10
∇ z B t − i(ẑ × E t ) = ∇ t B z 5.11
ẑ  (∇ t × B t ) = −iE z 5.12
Also,
∇ t E t + ∇ z E z =0 5.13a
∇ t B t + ∇ z B z =0 5.13b

Finally, one can solve for E t and B t if E z and B z are known (and not both are zero).
ikE t = ∇ t E z − i(ẑ × B t )
ikB t = ∇ t B z + i(ẑ × E t )
ik(ẑ × B t ) = (ẑ × ∇ t B z ) + i(ẑ ×(ẑ × E t ))
and
ikE t = ∇ t E z − (i/ik)(ẑ × ∇ t B z ) + ( 2 /ik)(−E t )
( 2  − k 2 )E t = ik∇ t E z − i(ẑ × ∇ t B z )
E t = i( 2  − k 2 ) −1 [k∇ t E z − (ẑ × ∇ t B z )] likewise 5.14a
B t = i( 2  − k 2 ) −1 [k∇ t B z + (ẑ × ∇ t E z )] 5.14b

For waves in the opposite direction change k to −k.

 Transverse electromagnetic wave (TEM): E z and B z are zero
everywhere inside cylinder.
For TEM waves E TEM = E t :
∇ t × E TEM = 0 5.15a
∇ t E TEM = 0 5.15b
k = k 0 =   5.15c
B TEM = ±  (ẑ × E TEM ) 5.15d

Unfortunately, the TEM wave is not supported by a single hollow cylindrical conductor
(with infinite conductivity). The surface must be an equipotential surface and inside such a
conductor, the electric field vanishes. One needs two or more cylindrical surfaces (such as a
coaxial cable) to support a TEM wave.

Boundary conditions at the surface

The existence of surface charge densities, , and surface current densities, K, at the
interface provide the following boundary conditions:
n̂  (D − D c )| S = 
n̂  (B − B c )| S = 0
n̂ × (E − E c )| S = 0
n̂ × (H − H c )| S = K
In the conductor the electric field, E c , (and for time varying electric fields B c ) is zero.
Thus, inside the hollow cylinder the boundary conditions can only be satisfied at the interface
when
(n̂ × E)| S = 0 516a
n̂  B| S = 0. 5.16b
That is, the component of the electric field tangent to the interface (E z ) must equal zero at
the surface:
Ez|S = 0 5.17
The corresponding condition on B z is (see Eq. (5.11)):
n̂  ∇B z | S = ∇ t B z | S = 0

Since we can not have both E z and B z equal to zero everywhere inside the cylinder, there
are two simple cases which satisfy the boundary conditions:
 Transverse Magnetic (TM) wave: Bz = 0 everywhere and E z | S = 0 5.18
E t = i( 2  − k 2 ) −1 k∇ t E z 5.18b
B t = i( 2  − k 2 ) −1 (ẑ × ∇ t E z ) 5.18c
0 = [∇ 2t +  2 − k 2 ]E z (x, y) 5.18d
Transverse Electric (TE) wave: Ez = 0 everywhere and ∂B z | S = 0 5.19
∂n
2 −1
B t = i(  − k ) k∇ t B z
2
5.19b
E t = i( 2  − k 2 ) −1 [−(ẑ × ∇ t B z )] 5.19c
0 = [∇ 2t +  2 − k 2 ]B z (x, y) 5.19d
The differential equations (5.18d) for E z and (5.19d) for B z and the boundary conditions
(5.18) and (5.19) give rise to eigenvalues of k (dependent on ) for which the propagation is
allowed. Since the boundary conditions for E z and B z are different, the eigenvalues are also
different. The allowed TE and TM waves (and the TEM wave, if it exists) provide a complete
set of waves from which one can construct an arbitrary electromagnetic disturbance in the
waveguide or cavity.

Modes in a rectangular waveguide:

We shall determine the TE modes in a rectangular waveguide with dimensions a in x and b

in y (with a > b) as shown in Fig. 8.5.

Note that this means E z = 0 everywhere (E is transverse). E t will be found from B z . So,
first one must solve Eq. 5.8 for B z :
[∇ 2t +  2 − k 2 ]B z (x, y) = 0
[ ∂ 2 + ∂ 2 + k ′  k ′ ]B z (x, y) = 0
2 2

∂x ∂y
The general solution is:
B z (x, y) = C 1 e +ik r +C 2 e −ik r
′
t
′
t
where
r t = xx̂ + yŷ
 The form for B z (x, y) which is non-zero when x = y = 0 is :
B z (x, y) = B o cos(k ′x x) cos(k ′y y) 5.20
The boundary conditions are
∂ B| =0
∂n z S
∂ B cos(k ′ x) cos(k ′ y)| x=0,a = 0
∂x o x y

∂ B cos(k ′ x) cos(k ′ y)| y=0,b = 0

∂y o x y

These give:
sin(k ′x a) = sin(k ′x 0) = 0 or 5.21
k ′x = m/a; m = 0, 1, 2, . . .
sin(k ′y b) = sin(k ′y 0) = 0 or 5.21b
k ′y = n/b; n = 0, 1, 2,
B z (x, y) = B o cos(mx/a) cos(ny/b) 5.22a
( m 2 n 2
a ) + ( b ) =  − k ; m, n = 0, 1, 2, 3. . .
2 2

k 2 =  2 − ( m n 2
a ) +( b ) > 0
2
5.22d
1/2
> 1 ( ma ) 2 + ( n ) 2 5.22e
 b
 min =  ; m = 1, n = 0 5.22f
a 

 mn =  (m n 2 1/2
a ) +(b)
2


For a non-trivial solution, m and n can not both be zero. Equation (5.22d) provides a
cutoff on the wave vector, k, since for k 2 < 0 the factor e +ikz becomes e ±kz and the wave
would not propagate. The full solution for each TE mn mode is:

B t = i( 2  − k 2 ) −1 k∇ t B z e i(kz−t) ;
−1
B t (m, n) = −ik ( m n 2
a ) +( b )
2
B o,m,n [x̂ (m/a) sin(mx/a) cos(ny/b) 5.23a
+ ŷ(n/b) cos(mx/a) sin(ny/b)]e i(kz−t)
E t = −  (ẑ × B t )
k
k 2 =  2 − ( m n 2
a ) +( b )
2

The solution for m = 1, n = 0 is:

 B t = −i ka  B o x̂ sin(x/a) exp i(kz − t) 5.24a
ẑ B z = ẑ B o cos(x/a) exp i(kz − t) 5.24b
E t = i a B o ŷ sin(x/a) exp i(kz − t) 5.24c

k = k 10 =   2 −  1 (  )2 5.24d
a
There is no propagation for
1 (  )2
 <  a
Note the 90 degree phase difference between the B x and B z arising from the −i = e −i/2 factor.
The B t and E t are 180 degrees out of phase.

Usually one designs the wave guide so that the m = 1, n = 0 mode is the dominant TE
mode. One can define the general k mn as follows:
k mn =   2 −  2mn 5.25a

 mn =  (m n 2 1/2
a ) +(b)
2
5.25b

For each mode, the k mn varies with frequency  >  mn . The  mn is the cutoff frequency for
the mode. In Fig. 8.4 from Jackson is a plot of k mn /(  ) as a function of , where k  =k mn .
It is often convenient to choose the dimensions of the waveguide so that at the operating
frequency only the lowest mode can occur. Since the wave number, k mn , is always less than
the "free space’ value,  , the wavelength in the waveguide is always larger than the free
space wavelength.

For the TM modes:

 E z = E 0 sin(mx/a) sin(ny/b)e i(kz−t) 5.26a
E z | S = 0 at x = 0, a, and y = 0, b
E t = i( 2  − k 2 ) −1 k∇ t E z
= i( 2  − k 2 ) −1 kE 0 [x̂ (m/a) cos(mx/a) sin(ny/b) 5.26b
+ ŷ(n/b) sin(mx/a) cos(ny/b)]e i(kz−t)


Bt = (ẑ × E t ) 5.26c
k
k 2 =  2 − ( m n 2
a ) +( b )
2

In the TM modes if n = 0 or if m = 0, E z = 0. Hence E t and B t are also zero. The next

possible mode is n = m = 1 with
 11 =  (1
a ) 2
+ ( 1 ) 2 1/2 5.27
 b
 min TM =  11 >  min TE = 
a 
Thus TE 10 mode has the smallest cutoff frequency.

Higher order modes:

The following shows the E t for some TE modes (E z = 0). (Taken from N. Stoyanov,
Department of Chemistry, MIT, Ph.D. thesis, 2003)
Summary of TE and TM
TE modes : E z = 0; ∂B z | = 0 5.28
∂n S
B z (x, y) = B o cos(mx/a) cos(ny/b)e i(kz−t)
B t = ik ( m n 2 −1 ∇ t B z
a ) +( b )
2

E t = −  (ẑ × B t )
k
k 2 =  2 − ( m n 2
a ) +( b )
2

TM modes: B z = 0; Ez|S = 0 5.29

E z (x, y) = E o sin(mx/a) sin(ny/b)e i(kz−t)

−1
E t = ik ( m
a ) +( b )
2 n 2 ∇tEz #

Bt = (ẑ × E t )
k

Energy Flow in the Waveguide for TE Modes

The time averaged flux of energy is given by the real part of the following expression
 S= 1
2
(E × H ∗ )
= 1
2
(E t ×(B t + ẑ B z ) ∗ )
= −  (ẑ × B t ) ×(B t + ẑ B z ) ∗
2k
=  [ẑ B t B ∗t −B t B ∗z ] 5.30
2k
ẑ  S =  −k 2 ∇ t B z  ∇ t B ∗z
2k ( m ) 2 + ( n ) 2 2
a b

= C(, m, n)∇ t B z  ∇ t B ∗z
Now we can integrate this over the cross section of the waveguide to find the power:
P TE = ∫ ∫ ẑ  SdA
= C(, m, n) ∫ ∫ ∇ t B z  ∇ t B ∗z dA
= C(, m, n) ∫ ∫[∇ t  (B z ∇ t B ∗z ) − B z ∇ 2t B ∗z ]dA

= C(, m, n)[  B z ∇ t B ∗z  dr − ∫ ∫ B z ∇ 2t B ∗z dA]

boundary

= C(, m, n)[ B z ∇ t B ∗z  dr − ∫ ∫ B z ∇ 2t B ∗z dA]

= C(, m, n)[ B z (x̂ ∂ + ŷ ∂ )B ∗z  dr − ∫ ∫ B z ∇ 2t B ∗z dA]

∂x ∂y
∗
∂B z ∂B ∗
= C(, m, n)[ B z (dxx̂ +dyŷ z ) − ∫ ∫ B z ∇ 2t B ∗z dA]
∂x ∂y
= 0 + C(, m, n) ( m a ) + ( b ) ∫ ∫ B z B z dA
2 n 2 ∗

=  k
∫ ∫ B z B ∗z dA
2 ( ma ) +( b )
2 n 2

  2 −  2mn
=  ∫ ∫ B z B ∗z dA
2 ( a )
m 2
+( )
n 2
b

1 −  2mn / 2
=  
2

2  ( a )
m 2
+( )
n 2 ∫ ∫ B z B ∗z dA 5.31
b
The corresponding value for TM modes is
2 1 −  2mn / 2
P TM =
2  ( m
a ) +( b )
2 n 2 ∫ ∫ E z E ∗z dA 5.32

Thus at fixed frequency, the power is inversely proportional to the  2mn and the smallest mode
numbers correspond to maximal power.