II. Electromagnetic Transients in Time-Varying Waveguides and Resonators

This page intentionally left blank



July 25, 2012 13:29



PSP Book - 9in x 6in



Chapter 5



An Electromagnetic Field in a Metallic

Waveguide with a Moving Medium



The first part of the book was devoted to the basic electromagnetic

transient phenomena caused by time-varying properties of a

homogeneous medium or a medium with very simple discontinuity

in the form of a plane boundary. The time variation can be

caused by changing time medium properties or by moving medium

boundaries. In the second part of the book the transients in

waveguide and resonator structures are investigated. Waveguides

with perfectly conducting walls, the dielectric waveguides and the

dielectric resonators will be considered.

The influence of a moving medium on a guided wave in a

rectangular waveguide with perfectly conducting walls, as well as

the wave evolution, is considered in this chapter. The transformation

of a guiding wave in the waveguide filled by a dielectric medium

which uniformly moves with relativistic velocity is examined. Based

on this investigation it is shown that it is possible to model various

phenomena concerned with the interaction of electromagnetic

waves with the boundary of a relativistic moving medium in the

presence of waveguide dispersion. Specifically, the interaction of the



Non-Stationary Electromagnetics

Alexander Nerukh, Nataliya Sakhnenko, Trevor Benson, and Phillip Sewell

Copyright c 2013 Pan Stanford Publishing Pte. Ltd.

ISBN 978-981-4316-44-6 (Hardcover), 978-981-4364-24-9 (eBook)

www.panstanford.com



05-Alexander-c05



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



348 An Electromagnetic Field in a Metallic Waveguide with a Moving Medium



electromagnetic field with a plasma “flashed” at zero moment of

time in the waveguide and then uniformly expanding is considered.

Investigations are made by using the evolution approach

developed in this book. A chain of evolution equations for a nonstationary electromagnetic field in a waveguide is derived and

solutions to these equations are obtained by virtue of the resolvent

method.



5.1 Expansion of an Electromagnetic Field by the

Non-Stationary Eigen-Functions of a Waveguide

The general integral equation (Eq. 1.3.3)

E = E0 −



1 ∂2G 0

1

∗χ

(P1 − Pex )

2

2

v ∂t

ε0 ε



∂G 0

∗ {curlχ μ0 μ(M1 − Mex ) + μ0 μ(χ j1 + iδ(S))} (5.1.1)

∂t

must be concretised for the case of a field in a rectangular metallic

waveguide, as the Green function in this case has the dyadic form

(Eq. 1.2.18)

−



Gˆ 0



ij



∂2

1 ∂2

− 2 2 δi k

∂ xi ∂ xk

v ∂t



−v 2

4π



= G ij =



˜f E

kj



(5.1.2)



The elements of this tensor are defined by the functions (1.2.23) for

the rectangular waveguide whose symmetry axis is directed along

the x axis and whose transverse dimensions are equal to a along the

y axis and b along the z axis:

˜f E = 8π v δ

ij

ij

ab



˜ f E,j f

mn mn



E,j

mn



(5.1.3)



m,n



where the coefficients

τ



t



˜ =

mn



dτ

−∞



dτ



mn



τ −t ,x −x



(5.1.4)



−∞



are determined by virtue of the functions

mn



(t, x) = J 0



ωmn



t2 −



x2

v2



θ



t−



|x|

v



(5.1.5)



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



Expansion of an Electromagnetic Field by the Non-Stationary



Here ωmn = vπ (m/a)2 + (n/b)2 is the eigen-frequency of the mn

mode.

Substitution of Eq. 5.1.2 into Eq. 5.1.1 gives the specific form of

the integral equation for waveguide

E i = E 0i +



1

ε



dt

−∞



∂

χ

∂t 2

2



×

+



∂

∂t



∞



∞



a



dx

−∞



Pj −



b



dy

0



dz

0



ε−1

Ej

4π



+ ce j kl



∂2

1 ∂2

− 2 2 δi j

∂ xi ∂ x j

v ∂t

∂2

∂ xk ∂t



Pl −



˜f E

ij



μ−1

Bl

4π μ



χ j j + i j δ (S)



(5.1.6)



where e j kl is the completely anti-symmetric tensor of rank 3.

As the integrals with respect to t and x represent convolutions,

the differentiation by these variables can be brought from the

field functions to Green’s function. After differentiation and some

manipulations we express Eq. 5.1.6 in the form

E = E0 +



8π v

ab



dx

m,n



1

ε



2

E

E 1 ∂

κ∗)κ − ˆfmn

( ˆfmn

v 2 ∂t2



ε−1

E + ˜ mn (χ ji n + iδ (S))

4π

μ−1

E

E μ ∂

B

− ˆfmn

κ × mn χ fˆ mn Mi n −

c ∂t

4π μ



×



mn χ



fˆ



E

mn



Pi n −



(5.1.7)



E,j

E

E ,i j

E,j

where ˆfmn

is the tensor with the elements fmn

= δi j mn fmn

f mn

,

κ = (∂/∂ x1 , −κm , −κn ), κ∗ = (∂/∂ x1 , κm , κn ), κm = mπ/a, κn =

nπ/b and the product ˆf A is a product of a diagonal tensor ˆf with a

vector A.

The kernel in Eq. 5.1.7 is a series of waveguide eigen-functions. It

is natural to represent all field values in the form of an expansion

in terms of waveguide eigen-functions which have to be vectorial

ones. It is common to derive these functions with the assumption

that fields can be represented in the form of the expansion on

modes which are monochromatic plane inhomogeneous waves with

a certain frequency and a certain wave number. In the general case

of non-monochromatic fields with inseparable dependence on time

and spatial coordinates the expressions for vector eigen-functions



349



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



350 An Electromagnetic Field in a Metallic Waveguide with a Moving Medium



have to be corrected. The derivation of these eigen-functions is given

in this sub-section following the scheme for the investigation of

fields in waveguides, as in Ref. 1.

It is well known that in a regular waveguide all waves are divided

into two classes: E- or TM-waves (transverse magnetic waves)

and H- or TE-waves (transverse electric waves). For the TM-waves

the longitudinal components of the electric field satisfy the first

boundary value problem which consists of the equation and the

boundary condition

∂2 E

∂2 E

+

+ κ2 E = 0

∂ y2

∂z2



(5.1.8)



E |L = 0

where L is the rectangular contour of the waveguide cross-section.

The eigen-functions of this problem are

E x = sin κm y sin κn z κm = mπ/a



κn = nπ/b



(5.1.9)



Expansion for the transverse electric components is given by the

vector eigen-functions emn⊥ which are expressed through the scalar

function (Eq. 5.1.9) by the following

E ,22

E ,33

e y + κn fmn

emn⊥ = ∇⊥ E x = κm fmn



ez = κm cos κm y sin κn ze y + κn sin κm y cos κn zf3E ez (5.1.10)

Here the scalar functions are given by (Eq. 1.2.20)

⎫

E ,11

fmn

= sin κm y sin κn z ⎬

E ,22

= cos κm y sin κn z

fmn

⎭

E ,33

= sin κm y cos κn z

fmn



(5.1.11)



The axes are directed as shown in Fig. 5.1 where the coordinates

x, y,and z will be numbered as 1, 2, and 3, respectively. It is easy

to check that these functions satisfy the first vector boundary value

problem

2

emn⊥ = 0

∇⊥2 emn⊥ + κmn

2

κmn



κm2



emnt | L = 0



div⊥ emnt | L = 0



κn2



where

=

+ are the eigen-values.

The vectors (Eq. 5.1.10) give the system of the vector functions

for a magnetic field

E ,33

E ,22

eˆ y + κm fmn

eˆ z

e∗mn = [ex , ∇⊥ E x ] = −κn fmn



(5.1.12)



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



Expansion of an Electromagnetic Field by the Non-Stationary



Figure 5.1. Geometry of the problem.



The system of vector functions for an electric field is given by

t



the Maxwell’s equation E = curl Bdt. Omitting integration by time

(which is not essential in this case) we obtain

∂

∂

2

E ,11

E ,22

E ,33

eˆ x + κm fmn

eˆ y

eˆ z .

fmn

emn = κmn

+ κn fmn

(5.1.13)

∂x

∂x

Thereby the electric and magnetic fields of the TM-waves are

represented by series of the systems of the vector eigen-functions

(Eqs. 5.1.12 and 5.1.13)

E=



emn E mn (t, x)



(5.1.14)



e∗mn Bmn (t, x)



(5.1.15)



m,n



B=

m,n



The vector eigen-functions for TE-waves generated by the scalar

functions

H x = cos κm y cos κn z



(5.1.16)



are equal to

B,22

B,33

eˆ y − κn fmn

bmn⊥ = ∇⊥ H x = −κm fmn



eˆ z = −κm sin κm y cos κn zeˆ y − κn cos κm y sin κn zeˆ z (5.1.17)

where

B,11

= cos κm y cos κn z

fmn



B,22

E ,33

fmn

= fmn



B,33

E ,22

fmn

= fmn

(5.1.18)



351



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



352 An Electromagnetic Field in a Metallic Waveguide with a Moving Medium



This system of functions satisfies the second vector boundary

value problem

2

∇⊥2 bmn⊥ + κmn

bmn⊥ = 0



(bmn⊥ , n)| L = 0



(curl⊥ , bmn )⊥



L



=0



The function (Eq. 5.1.16) gives the system of the functions for an

electric field

B,33

B,22

b∗mn = [ex , ∇⊥ H x ] = κn fmn

eˆ y − κm fmn

eˆ z



and the Maxwell’s equation B = −curl

functions for a magnetic field

2

B,11

B,22

eˆ x − κm fmn

eˆ y

fmn

bmn = κmn



(5.1.19)



t



Edt gives the system of



∂

∂

B,33

− κn fmn

eˆ z

∂x

∂x



(5.1.20)



Thereby the fields for the TE-waves are given by the series

E=



b∗mn E mn (t, x)



(5.1.21)



bmn Bmn (t, x)



(5.1.22)



m,n



B=

m,n



The series for the fields (Eqs. 5.1.15, 5.1.21 and 5.1.22) allow one

to reduce the vector integral Eq. 5.1.17 to the system of the scalar

integral equations with respect to the expansion coefficients. If a

material object in the waveguide has an arbitrary shape then the

field contains the waves of both classes and the system of integral

equations is very intricate. As the main interest in this book is

non-stationary behaviour of electrodynamics processes we confine

ourselves further to the case where the material object is a planeparallel insertion filling the whole waveguide cross-section. In this

case the object characteristic function χ = χ (t, x) depends on one

spatial coordinate only.

Let us consider the case when the initial field consists of H- or

TE-waves. It is evident that the geometric properties of the object

eliminate the appearance of the waves of the class other than the

initial one. So, the vectors E, P and j can be represented in the form

of Eq. 5.1.21 and the vectors B and M in the form of Eq. 5.1.22. These

expansions allow integration over the transverse cross-section of



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



Expansion of an Electromagnetic Field by the Non-Stationary



the waveguide in the integral Eq. 5.1.7

∞



∞



b∗mn



E = E0 − 2π v

m,n



dt

−∞



1 ∂2

εv 2 ∂t2



dx

−∞



ε−1

E mn + ˜ mn (χ jmn + i mn δ (S))

Pmn −

mn χ

4π

μ−1

μ ∂ ∂ mn ∂

2

− κmn

+

M mn −

Bmn (5.1.23)

mn χ

c ∂t

∂x ∂x

4π μ

Equating coefficients of the terms with the same vector-functions

reduces Eq. 5.1.23 to the temporal and 1D spatial integral scalar

equation for an arbitrary mode of the electric field whose index is

omitted

2π

E = E0 −

εv



∞



∞



dt

−∞



dx

−∞



+ ˜ k (χ ji n + i δ (s)) + c

Mi n −



×

where

k



⎛

= J 0 ⎝ωk



∂2

∂t2

∂

∂t



kχ



Pi n −



ε−1

E

4π



∂ k ∂

− κk2

∂x ∂x



μ

B

4π μ



k



χ

(5.1.24)



⎞

2

(x

)

−

x

⎠θ

(t − t )2 −

v2



t−t −



x−x

v

(5.1.25)



t



˜k =



dτ



k



τ −t ,x −x



ωk = vκk



(5.1.26)



−∞



and the index k designates the oscillation mode considered, κk =

κmn . The differentiation operations in this equation can be brought

out as the integrals as the latter are convolutions.

If the background medium is not magnetic (μ = 1) and the

object consists of such a medium that the magnetisation vector M

has transverse components only then the integral equation can be

simplified. In this case the magnetic field B does not enter directly in

the equation and the magnetisation M can be expanded in a series of

the transverse eigen-vectors

B,22

B,33

eˆ y − κn fmn

eˆ z

bmn⊥ = −κm fmn



(5.1.27)



353



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



354 An Electromagnetic Field in a Metallic Waveguide with a Moving Medium



in which, in comparison with vectors (Eq. 5.1.20), the derivative

∂/∂ x is brought into the coefficients M mn . In this case the coefficients

¯ mn (t, x) are connected with the

bmn⊥ M

of the new expansion M =

m,n



¯ mn = ∂ M mn . One has to take into account

old ones by the relation M

∂x

also that the equality κmn = 0 would be stated in Eq. 5.1.20. After

manipulations, the Eq. 5.1.24 takes the form

∞



∞



E = E 0 − 2π



dt

−∞



dx

−∞



1 ∂2

εv ∂t2



1 ∂2

+ ˜ k (χ ji n + i δ (s)) + √

ε ∂t∂ x



k



ε−1

E

4π



Pi n −



k χ Mi n



¯



.



χ

(5.1.28)



The integrals in this equation represent convolutions, so the

differentiation operators can be brought out of the integrals

according to properties of generalised functions.

Thereby the integral equation for the electric field in the

rectangular metallic waveguide containing a dielectric object filling

the whole cross-section of the waveguide and restricted by the

plane-parallel boundaries which are normal to the waveguide’s

longitudinal axis is given by Eq. 5.1.28.

Equation 5.1.28 is the starting one for the further investigation

of non-stationary behaviour in the waveguide. It is the Volterra

integral equation and it breaks up into a chain of evolutionary linked

equations which can be solved by the resolvent method as in the case

of an unbounded medium considered in the previous chapters. The

existence of waveguide dispersion gives effects of theoretical and

practical interest in the case of such simple dynamics as uniform

movement of a restricted dielectric, even if the object inside the

waveguide does not change its properties. Such phenomena will be

considered in the next sections.



5.2 Equations for a Field in the Waveguide with a

Non-Stationary Insertion

The influence of the waveguide dispersion on the electromagnetic

transients is revealed in the stationary movement of the medium or



July 25, 2012 13:29



PSP Book - 9in x 6in



Equations for a Field in the Waveguide with a Non-Stationary Insertion



Figure 5.2. Movement of the insertion in the waveguide.



its boundary [3–6]. The stationary movement is understood as the

movement which does not change its character since the infinitely

remote past. This investigation is interesting because it reveals

peculiarities that are brought about by the waveguide dispersion.

The results can be compared with results for strictly non-stationary

phenomena, for example, when the movement begins at some finite

moment of time.

We consider a dielectric non-stationary insertion filling the

whole cross-section of the waveguide and moving along it. This

insertion is restricted by planes x1 (t) and x2 (t) which are perpendicular to the waveguide’s longitudinal axis (Fig. 5.2). The waveguide

itself is filled by a motionless background non-magnetic, μ = 1,

medium. A guided wave of a given mode E 0 is falling on the layer

from the side x < x1 (t). One must distinguish two cases: (i)

the medium inside the insertion is motionless, and (ii) it moves

together with its boundary. In both cases we consider only the

most practical case when the movement is directed along the

waveguide.

If the medium when at rest is non-conducting and non-magnetic

and has a dielectric polarisability αi j then the general constituent

relations [3] give the expressions for the polarisation vectors for the

medium uniformly moving with the velocity u



05-Alexander-c05



355



July 25, 2012 13:29



PSP Book - 9in x 6in



05-Alexander-c05



356 An Electromagnetic Field in a Metallic Waveguide with a Moving Medium



Pi = γ 2 γi j α j m (γmn E n + emns βn B S ) ,



(5.2.1)



M i = −γ 2 ei j k β j αkn (emns βn Bs + γmn E n ) .



(5.2.2)



Here β = u/c, γ 2 = 1/ 1 − β 2 , γi j = δi j + 1−γ

β β and emns

γβ 2 i j

is the complete anti-symmetric tensor of the third rank. For an

isotropic medium the polarisability is a scalar one, αi j = α δi j ,

so the constituent relations (5.2.1) and (5.2.2) are simplified as

follows:

P = α γ 2 {E − β (β E) + [βB]}



M = [Pβ] .



(5.2.3)



These relations provide transversality for all the field values E, P and

M in the case of the TE-waves as the mode transformation is absent

under the chosen geometry of the insertion.

The stationary nature of the problem allows one to represent all

field values in the form of a Fourier transform with respect to the

time variable. In this representation the fields inside the insertion

consist of two waves

2



∞



dν ˜

E s (ν) ei νt−i ks (ν)x

2π



E =

s=1 −∞



ε−1

E =

P−

4π



2



2



∞



s=1 −∞

∞



¯ =

M

s=1 −∞



(5.2.4)



dν ¯˜

Ps (ν) ei νt−i ks (ν)x

2π



dν ˜¯

M s (ν) ei νt−i ks (ν)x

2π



If the kernel (Eq. 5.1.25) is represented as Fourier transform also

∞

k



=

−∞



dω i ω(t−t )

1

e

e−i k0 (ω)|x−x |

2π

i vk0 (ω)



(5.2.5)



then the integral Eq. 5.1.24 in the previous section, after integration

by x over the interval [x1 (t), x2 (t)] with x belonging to this interval,