Kishk A. (Ed.) - Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic Waves

SOLUTIONS AND
APPLICATIONS OF
SCATTERING, PROPAGATION,
RADIATION AND EMISSION OF
ELECTROMAGNETIC WAVES
Edited by Ahmed Kishk

SOLUTIONS AND
APPLICATIONS OF
SCATTERING,
PROPAGATION,
RADIATION AND
EMISSION OF
ELECTROMAGNETIC
WAVES
Solutions and Applications of Scattering, Propagation, Radiation and Emission of
Electromagnetic Waves
http://dx.doi.org/10.5772/3369
Contributors
Gregorio José Molina-Cuberos, María J. Núñez, Angel J. García-Collado, Jose Margineda, Ernesto Martín, Irismar Da
Paz, Maria Carolina Nemes, Jose Geraldo Peixoto de Faria, Marcos D. R. Sampaio, Luis Cabral, José G. G. O. Junior, Saulo
Moreira, Gennadiy Vorobyov, Aleksandr Rybalko, Ivan Barsuk, Larissa Vietzorreck, Ada Vittoria Bosisio, Yuri Alvarez
Lopez, Fernando Las-Heras, Cebrián García González, Elena Vinogradova, Lyudmyla Velychko, Yuriy Sirenko, Jaime
Laviada-Martinez, Marcos Rodríguez-Pino, Ismael Barba, Ana Grande, Ana Cristina López Cabeceira, Álvaro Gómez
Gómez, José Represa, José Antonio Pereda, Hector Torres-Silva, Farhang Alijani, Nikolai Shul'Ga, Sergii Trofymenko,
Lutz Angermann, Vasyl Yatsyk
Published by InTech
Janeza Trdine 9, 51000 Rijeka, Croatia
Copyright © 2012 InTech

All chapters are Open Access distributed under the Creative Commons Attribution 3.0 license, which allows users to
download, copy and build upon published articles even for commercial purposes, as long as the author and publisher
are properly credited, which ensures maximum dissemination and a wider impact of our publications. After this work
has been published by InTech, authors have the right to republish it, in whole or part, in any publication of which they
are the author, and to make other personal use of the work. Any republication, referencing or personal use of the
work must explicitly identify the original source.
Notice
Statements and opinions expressed in the chapters are these of the individual contributors and not necessarily those
of the editors or publisher. No responsibility is accepted for the accuracy of information contained in the published
chapters. The publisher assumes no responsibility for any damage or injury to persons or property arising out of the
use of any materials, instructions, methods or ideas contained in the book.
Publishing Process Manager Dragana Manestar

Technical Editor InTech DTP team
Cover InTech Design team
First published November, 2012

Printed in Croatia
A free online edition of this book is available at www.intechopen.com

Additional hard copies can be obtained from [email protected]
Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic

Waves, Edited by Ahmed Kishk
p. cm.
ISBN 978-953-51-0838-2
Contents
Preface VII
Section 1 Waves in Chiral Media 1
Chapter 1 Electromagnetic Characterization of Chiral Media 3

J. Margineda, G.J. Molina-Cuberos, M.J. Núñez, A.J. García-Collado
and E. Martín
Chapter 2 Chiral Waves in Graphene Medium and Optical Simulation

with Metamaterial 25
H. Torres-Silva
Section 2 Numerical and Analytical Solutions 57
Chapter 3 Numerical Modeling of Electromagnetic Wave Propagation

Through Bi-Isotropic Materials 59
I. Barba, A. Grande, A.C.L. Cabeceira, A. Gómez, J.A. Pereda and J.
Represa
Chapter 4 Integral Equation Analysis with

Characteristic Basis Functions 97
Jaime Laviada, Fernando Las-Heras and Marcos R. Pino
Chapter 5 Analytical Grounds for Modern Theory of Two-

Dimensionally Periodic Gratings 123
L. G. Velychko, Yu. K. Sirenko and E. D. Vinogradova
Chapter 6 The Sources Reconstruction Method for Antenna Diagnostics

and Imaging Applications 159
Yuri Álvarez, Fernando Las-Heras and Cebrián García
Section 3 Microwave Applications 183

VI Contents
Chapter 7 Multi-Hole Waveguide Directional Couplers 185

Mahmoud Moghavvemi, Hossein Ameri Mahabadi and Farhang
Alijani
Chapter 8 The Diffraction of Electromagnetic Waves on the Periodic

Heterogeneities and Its Use for Realization of Practical
Technical and Electronic Devices of Millimeter and
Submillimeter Wavelength Range 209
Gennadiy Vorobyov, Larissa Vietzorreck, Ivan Barsuk and Aleksandr
Rybalko
Section 4 Scattering, Radiation and Propagation 233
Chapter 9 Radiative Transfer: Application to Indoor Propagation 235

Ada Vittoria Bosisio
Chapter 10 Electromagnetic Wave Packets in the Theory of Bremsstrahlung

and Transition Radiation by High-Energy Electrons 263
Nikolai Fyodorovich Shul’ga and Sergii Valeriyovich Trofymenko
Chapter 11 The Effect of Weak Fields at Multiple Frequencies on the

Scattering and Generation of Waves by
Nonlinear Layered Media 303
Lutz Angermann and Vasyl V. Yatsyk
Chapter 12 Matter Wave Interferometry, the Gouy Phase

and Complementarity Principle 333
Luis A. Cabral, Irismar G. da Paz, José G. G. de Oliveira Júnior, Saulo
Moreira, Marcos D. R. Sampaio and Maria C. Nemes
Preface
This book consists of four sections that are constructed from twelve chapters presented by
several international specialized authors with physics and engineering background in
different aspects of Electromagnetic waves. The reader may jump directly to the chapter of
his interest as each chapter is almost self-contained. Wide range of different topics related to
analytical as well as numerical solutions of problems related to scattering, propagation,
radiation, and emission in different medium are discussed. Design of several devices and
their measurements aspects are introduced. Topics related to microwave region as well as
Terahertz and quasi-optical region are considered. Bi-isotropic metamaterial in optical
region is investigated. Interesting numerical methods in frequency domain and time domain
for scattering, radiation, forward as well as reverse problems and microwave imaging are
summarized. Therefore, the book will satisfy different tastes for engineers interested for
example in microwave engineering, antennas, and numerical methods.
Chapter 1 presents practical methods to characterize the electrical properties of chiral media.
Detailed review of the propagation of electromagnetic waves in unbound chiral media is
revised and theoretical relations between constitutive parameters and the scattering
parameters are developed. A review for the different theoretical solutions, techniques and
experimental setups developed in order to characterize chiral media is presented. Special
attention to the retrieval algorithms is given.
Chapter 2 presents a short review on the chiral propagation of electron waves in monolayer
graphene and optical simulation with optical field in the negative-zero-positive index
metamaterial (NZPIM) and its close connection. The study verifies that the chiral field near
the Dirac point is a diffusive wave.
Chapter 3 is a review and modifications to several numerical techniques in both frequency
domain and time domain to deal with Bi-isotropic materials. The extension of the traditional
methods is necessary to deal with the constitutive parameters of the bi-isotropic materials,
as well as their inherently dispersive behavior. The extension is necessary especially when
dealing with time-domain methods, in which the frequency dispersion must be translated to
time-domain. Finite Differences Time Domain technique (FDTD), Transmission Line Matrix
(TLM) and Multiresolution in Time Domain (MRTD) have been modified to include the
frequency dispersion. In the frequency domain, an extended hybrid technique, based upon
the Coupled Mode Method (CMM), which is originally formulated for closed structures and
the mode matching method (MMM), are modified to deal with open structures, like free
VIII Preface
propagation inside a bi-isotropic medium. The MMM is included to characterize

discontinuities in the power propagation direction.
Chapter 4 presents the key aspects of the analysis of an electromagnetic problem with
characteristic basis functions (CBFs) that are a subtype of macrobasis functions. Thus, the
generation of the CBFs and their inclusion the conventional method of moments is
explained. In addition, several enhancements of the method such as the multilevel
generation of CBFs are detailed. The most important applications, where the method
becomes very useful, are covered. In addition, the recursive formulation for generating
nested CBFs is shown in order to improve the compression of the reduced matrix to show
that the reduced size of the system of equations enables the partial solution of electrically
large geometries.
Chapter 5 presents the necessary theoretical background to the numerical solution of initial
boundary value problems as applied to two-dimensionally periodic structures. Correct and
efficient truncation of the computational space in the problems describing spatial-temporal
electromagnetic wave transformation in two-dimensionally periodic structures is presented.
Some important characteristics and properties of transient and steady-state fields in
rectangular parts of the Floquet channel are discussed. The method of transformation
operators is described to show that the computational resources can be optimized when
calculating a multi-layered periodic structure or a structure on a thick substrate.
Chapter 6 presents a summary of the source reconstruction method (SRM) ), an integral
equation-based inverse technique, and its applications on antenna diagnostics and the recent
advances on imaging applications. Implementation and technical details, as well as a
compilation of SRM applications for antenna diagnostics and characterization, inverse
scattering and imaging applications are presented.
Chapter 7 focuses on the waveguide directional couplers and gives finalized designing
formulas and tables. A full generalized field theory for designing based on multi-hole
coupling structure is introduced. A recursive procedure is provided to adjust the required
directivity, coupling and frequency or waveguide type to provide the number of holes,
individual diameters, and their locations besides coupler length, length of matched load and
other sizes of structure for synthesis. Different methods such as binomial, Chebyshev are
used to achieve the wide bandwidth response.
Chapter 8 presents an experimental setup and procedure for measuring electrodynamic
characteristics of periodic metal and metal–dielectric structures, which can be used for
practical devices operating in millimeter, sub millimeter, and terahertz wavelength ranges.
A directional coupler quasioptical power dividers formed by grids of metal bars and a
dielectric waveguide are considered. Peculiarities of electromagnetic characteristics of
surface waves diffraction on the periodic heterogeneities are realized to create the sources
for radiation on the Smith- Purcell effect.
In chapter 9 the transfer theory is applied to investigate propagation within an indoor
environment to evaluate the performance of an indoor communication system. For
simplicity, a two-dimensional environment is considered and the assumption that the
Preface IX
propagation environment can be modelled as a layered parallel plane medium. This is a first
approximation of an office indoor environment made of a succession of tables where each
table is modelled as random medium containing randomly placed scatterers that are treated
as infinite circular cylinder so as to simplify the solution of the radiative transfer equation.
Chapter 10 studies electromagnetic field space-time evolution and its spectral-angular
characteristics in the processes of high-energy electron instantaneous scattering and its
traverse of thin metallic plate. It involves the scrupulous analysis of evolution of packets of
free electromagnetic waves by which the investigated field can be represented. Attention is
drawn to the fact that the state of electron with nonequilibrium field can manifest itself not
only in the case of ultra-high particle energies but also for electron energies of several tens
Mev in the millimeter region of radiated waves. The problem of relativistic electron’s
transition radiation on thin metallic ideally conducting plate is considered onward. The
space-time evolution of the electromagnetic field, which arises before and after passing of
the electron of the metallic plate, is discussed. The main attention here is drawn to the
effects in backward transition radiation in this process. The wave packets of the field
reflected from the plate in this case are the packets of free electromagnetic waves, which
gradually transform into transition radiation field.
Chapter 11 investigates the effects of weak fields at multiple frequencies on the scattering
and generation of waves by an isotropic, non-magnetic, linearly polarized (E-polarisation),
layered, cubically polarisable, dielectric structure. In the domain of resonance frequencies,
wave packets are consisting of both strong electromagnetic fields at the excitation frequency
of the nonlinear structure are considered leading to the generation of waves, and of weak
fields at the multiple frequencies, which do not lead to the generation of harmonics but
influence on the process of scattering and generation of waves by the nonlinear structure.
The electromagnetic waves for a nonlinear layer with a cubic polarisability of the medium
can be described by an infinite system of nonlinear boundary-value problems in the
frequency domain. The effect of weak fields at multiple frequencies on the scattering and
generation of waves is investigated numerically. The results indicate a possibility of
designing a frequency multiplier and nonlinear dielectrics with controllable permittivity.
Chapter 12 presents theoretical aspects of the Gouy phase for classical electromagnetic
waves. It shows how the phase can be measured from measuring quadratures, which are
related by the determinant of a covariance matrix just in the way it is done in usual
quantum mechanics. The geometrical construction is explicated and the relationship with
matter waves is established.
Dr. Ahmed Kishk

University of Mississippi
USA
Section 1
Waves in Chiral Media

Chapter 1
Electromagnetic Characterization of Chiral Media
J. Margineda, G.J. Molina-Cuberos, M.J. Núñez,

A.J. García-Collado and E. Martín
Additional information is available at the end of the chapter
1. Introduction
The electromagnetic characterization of materials is a fundamental problem in many re‐

search areas of Electromagnetism. The goal is to determine the electrical permittivity and
→ →
magnetic permeability, which make up the usual constitutive relations: D = ε0εr E ,
→ →
B = μ0μr H , where εr and μr are, for isotropic and homogeneous materials, two complex fre‐
quency-dependent parameters.
During recent decades, a great variety of novel and complex materials have been designed
with promising practical applications. It has been found, however, that in many cases the
relations between electric and magnetic fields cannot be described by standard constitutive
equations. Among these new materials we may mention, for example, chiral, nonreciprocal,
gyrotropic and negative refractive index media. Bi-isotropic media are the most general line‐
ar, homogeneous and isotropic materials, and they respond to electromagnetic excitation ac‐
cording to the following relations:
r r r
D= ò0òr E + ( c - jk ) ò0 m0 H
r r r
(1)
B = m0 mr H + ( c + jk ) ò0 m0 E
where κis the chirality, a dimensionless parameter for describing the handedness of the ma‐
terial, and χis the Tellegen parameter, a dimensionless quantity for the degree of inherent
nonreciprocity in the medium [1]. The reciprocal chiral medium, also known as Pasteur me‐
dium, is a class of bi-isotropic medium characterized by χ = 0. The electromagnetic behavior
of these materials reflects two effects: electromagnetic rotatory dispersion, which causes a
4 Electromagnetic Waves
Solutions and Applications of Scattering, Propagation, Radiation, and Emission of Electromagnetic Waves
rotation of the polarization direction for a linearly polarized wave, and circular dichroism (a
change in the polarization from linear to elliptical) due to the different absorption coeffi‐
cients of right- and left-handed circularly polarized waves.
Electromagnetic activity at optical frequencies was first observed in the XIX century, [2, 3].
Although nature presents many examples of chiral media at optical frequencies, there is, to
the best of our knowledge, no natural medium able to produce electromagnetic activity at
microwave frequencies. At these frequencies, the first man-made chirals were manufactured
by embedding a random distribution of conducting helices with the same handedness into a
host medium. The first experimental studies were developed using basic free-wave systems
with the aim of verifying electromagnetic activity at microwave frequencies. Lindman [4]
and later Tinoco and Freeman [5] found a critical wavelength for copper helix at λ ≅ 2L,
where L is the length of the wire in the helix; this constitutes a rule of thumb to select the
frequency range in which to search for interesting effects. By far, the most used structure for
chiral manufacture has been the metallic helix [6-10], although other elements such as metal‐
lic cranks have also been used [11]. Manufacturing techniques have evolved from the ran‐
dom inclusion of metallic particles to the presently used alternatives, which are based on
periodic distribution of planar or quasi-planar chiral particles printed on a circuit board or
even integrated circuits. The interested reader can find a review in [12].
The ordered distribution of resonant structures in a periodic lattice enhances the electromag‐
netic effects and is able to produce larger electromagnetic activity and circular dichroism.
These enhanced chirals are often known as chiral metamaterials. It has been found that a
material with very high electromagnetic activity can also possess a negative refractive index
[13]. During the last decade, a large variety of resonant structures have been designed and
analyzed in order to produce high values of optical activity, circular dichroism and negative
refractive index; for example, see [14-15] and references therein.
The presence of the chirality parameter in the constitutive equations prevents the use of
standard characterization procedures. Furthermore, the existence of resonant structures,
which finally produce strong variations in the characteristic parameters, makes the retrieval
algorithms more complicated. In this paper, the main techniques for chiral characterization
are reviewed. The second section is devoted to free-wave methods, which are the most
widely used systems for chiral characterization. The guided methods presented in the third
section, usually provide more accuracy than the free-wave methods, although the search for
electromagnetic rotatory dispersion limits the guided systems to those of axial symmetry. In
section four, resonator techniques are considered. Finally, the accuracy and feasibility of
each technique is discussed.
2. Electromagnetic fields in chiral media
The constitutive relations (1) for reciprocal media (χ = 0), are:

Electromagnetic Characterization of Chiral Media 5
r r r
D ò0òr E - jk / c H
=
r r r
(2)
B = m0 mr H + jk / c E
wherec = (ε0μ0)-1/2. The rotational Maxwell equations present a coupling between the electric
and magnetic fields. In frequency domain and for the sourceless case, the curl of the electric
and magnetic fields are:
r
æ jk r r
ö
Ñ ´ E = - jw ç E+ m H ÷
è c ø
(3)
r
æ r jk r ö
Ñ ´ H jw ç ò E -
= H÷
è c ø
It is possible to find a linear combination of the E and H fields, called normal modes and
→ →
written{En, Hn}n=1, 2 , that satisfy independent rotational Maxwell equations so that:
ì r n r nü
ïÑ ´ E = - jwm H ï
n
í n n ý (4)
ï Ñ ´ Hr = jwò n Er ï
î þn =1,2
These combinations of electromagnetic fields propagate through the chiral material as

through an isotropic non-chiral medium, with equivalent constitutive parameters:
{ε n, μ n}n=1,2.
The normal modes satisfying eq. (4) are:
→ → 1 → → → → j →
E1 ≡ E+ = (
2 E - jηH ), H1 ≡ H+ = η+ E+
→ → (5)
1 → → → → -j →
E2 ≡ E- = (
2 E + jηH ), H2 ≡ H- = η- E-
the medium parameters being:
ε± = ε (1 ± κr )
(6)
μ± = μ (1 ± κr )
where κr = κ / εr μr is the relative chirality, and:

η = η± = μ± / ε± = μ / ε (7)
2.1. Plane wave propagation in chiral media
A propagating wave can always be expressed as a linear combination of the propagation of

→ →
normal modes. The propagation constant for each mode {k +, k - } could be, in principle, dif‐
ferent:
→ → → → →
E ±( r ) = E ±exp ( - jk ± ∙ r )
→ → → → → (8)
H±(r ) = H ±exp ( - jk ± ∙ r )
If the wave is propagating in the u^ direction, the modes propagation constants are:
→
k ± = u^ k± (9)
wherek± = ω ε±μ± = ω εμ (1 ± κr ) = ω n (1 ± κr ) / c. The refraction indices for each mode are
n± = n (1 ± κr ) (10)
→ →
It can be deduced, from eq. (4, 5), that the vectorsu^ , E +and H + are mutually orthogonal. And
→ →
the same happens tou^ , E -andH -. Also, according to eq. (5) the propagation modes are circu‐
→ →
larly polarized electromagnetic waves to the right (RCP) {E +, H +}and to the left (LCP)
→ →
{E -, H -}.
3. Free-wave
3.1. Experimental systems
For regular materials where εand μare the only parameters to be retrieved, the experimental
determination of the complex scattering coefficients S11 and S21 are sufficient to solve the
problem. But there remains an extra parameter to be determined, the chiralityκ. To do this,
we are forced to experimentally measure one complex or two real additional quantities. This
is usually achieved in terms of the reflection coefficient and the two transmission coeffi‐
cients corresponding to the parallel and perpendicular direction to the incident wave. Sun et
al. [16] measured the complex reflection coefficients of short-circuit (backed with metal) and
open-circuit (not backed with metal) samples. For transmission, they measured the rotation
angle and the axial ratio.
The usual free-wave experimental setup consists of a pair of transmitting and receiving an‐
tennas, a sample holder, a vector network analyzer (to determine magnitude and phase of
the scattering parameters) and a calibration kit. In order to minimize the diffraction effects
at the edges of the sample it is necessary to decrease the size of the beam or place the sample
close to the antennas. But if the antenna is close to the sample, the conditions of plane wave
are not accomplished and the far-field retrieval algorithms cannot be applied. Moreover, the
effects of multiple reflections between the antenna and the sample are enhanced, which in‐
troduces extra errors into the measurements. The size of the beam can be reduced by placing
focusing devices, such us spot focusing lens in the antennas [17] or an ellipsoidal reflector
mirror after the transmitting antenna [18]. Behind the focus, the wave is not plane, but takes
the configuration of a Gaussian beam. In this case, the plane wave assumption does not in‐
troduce serious errors in the measurements [19].
Figure 1 shows a schematic diagram of the experimental system developed by Varadan et al.
[6, 17, 20]. It makes use of a lens in each antenna to focus the wave; the sample is located in
the common focal plane. The transmitting and receiving antennas are mounted on a carriage
and the distance between them can be adjusted for calibration. This setup was initially de‐
veloped by Ghodgaonkar [21] to measure permittivity and permeability, and later adapted
to determine rotation angle, ellipticity and power absorption. Making use of this system,
Guire et al. [20] measured the normal incidence reflection of linearly polarized waves of
metal-backed chiral composite samples at microwave frequencies. Ro et al. [6], determined
the ellipticity, rotation angle and power absorption for samples containing helix and ob‐
served that, at resonance, the angle of rotation rapidly increases or decreases or even
changes its sign, a phenomenon also known as the Cotton effect. They also reported that this
phenomenon is observed in the frequency region where maximum electromagnetic power
absorption occurs.
Figures 2 and 3 show a picture and a schematic diagram, respectively, of the free-wave ex‐
perimental system developed by the authors. The setup is based on a previous one for per‐
mittivity and permeability measurements [22], and adapted to measure electromagnetic
activity [10]. This system has been used to characterize chiral materials based on random
distribution of elements and periodical lattices [10-11, 23-25].
Figure 1. Setup for free-wave measurement, modified from [6, 17, 21].
The incident beam is focused by an ellipsoidal concave mirror, with the transmitting anten‐
na placed at one of the mirror foci and the sample at the other. The position of the receiving
antenna is not crucial in this configuration. Figure 2 shows a diagram where the focusing
mirror and sample holder are not shown for clarity. The incident wave is linearly polarized
in the x-axis direction. The rotation on the polarization angle, θ, is defined as the difference
between the polarization direction of the incident wave and the direction of the major axis of
the transmitted elliptically polarized wave.
Figure 2. Photo of the free-wave setup in the X-band developed by the authors.
Figure 3. Schematic diagram of figure (2), where the receiving antenna with a rotation angle of 90º to measure T CR .
The focusing mirror and sample holder are not shown for clarity.
The experimental system presented by Wang et al. [14] to characterize chiral metamaterials
with high electromagnetic activity makes use of a pair of standard gain horn antennas,
which are located close to the sample, without any focusing device. This configuration leads
to multiple reflections and border diffractions, which affect the experimental reflection and
transmission coefficients.
The scattering parameters are measured using a vector network analyzer. First, a two port
calibration should be made, for which purpose the TRL (Transmission, Reflection, Line)
technique is one of the most frequently used. The basic TRL calibration process is a two port
calibration widely used in non-coaxial environments, such as waveguide and free-wave sys‐
tems1. TRL uses a 12-term error model with three calibration standards: THRU, REFLECT
and LOAD. For the THRU step, the test ports are connected and the four S-parameters are
measured in both directions. For the REFLECT step, a highly reflective device is connected
to each port and the reflection coefficients are measured. In the LINE step, a short transmis‐
sion line is inserted and, again, the four S-parameters are measured.
The reference planes of the calibration can be located at the sample surface [17] or else‐
where, for example, at the two waveguide flanges where the antennas are connected [22]. In
the first case, Ghodgaonkar et al. [17, 21] implemented the TRL calibration for free-wave
systems: The THRU standard was configured by keeping the distance between the two an‐
tennas equal to twice the focal distance, without the sample. The LINE standard was ach‐
ieved by separating the focal plane of transmitting and receiving antennas by a distance
equal to a quarter of the free space wavelength at the center of the band. The REFLECT
standards for both ports were obtained by placing a metal plate in the focal planes of the
transmitting and receiving antenna, respectively.
The second option is the one used by the authors. In this case, it is possible to use standards for
calibration in rectangular waveguide: a short-circuit is located at the antenna flanges for the
REFLECT standard and a rectangular guide of λ/4 length for the LINE standard. A second cal‐
ibration is necessary to shift the reference planes to the sample surface. To do that, the reflec‐
tion coefficient, with a metal plate in sample position, and the transmission coefficient without
the sample were used as reference measurements. Then, the ratio between the measurement
with and without sample provides the magnitude of the scattering parameters of the sample.
The phase of the reflection coefficient must be corrected by a π factor introduced by metal
plate. For the phase of the transmission coefficient, it is necessary to subtract a quantity equals
tok0 L , where k0 is the vacuum propagation constant and L is the sample width.
At first sight, the second TRL calibration option seems more complicated; however, it has
the advantage of using LINE and THRU standards in rectangular waveguide, which are
more accurate and easier to use than standards obtained by moving the antennas.
A time domain (TD) transform can be used to filter out mismatches from the antennas, edge
diffraction effects and unwanted reflections from antennas, sample, mirrors or elsewhere.
The network analyzer makes measurements in the frequency domain, and the time domain
response is obtained by applying an inverse Fourier transform. The unwanted signals,
which have longer paths than the direct signal, are separated from the main signal in the TD
response and then can be removed by gating the signal. In order to obtain good filtering, it
is necessary to clearly identify the sample response in the time domain and to choose the
most suitable filter parameters [26]. The Fourier transform of the gated time-domain re‐
sponse provides the filtered signal. As will be shown later, this gating process greatly im‐
proves the experimental results.
1 Agilent Technologies Inc. Applying Error Correction to Network Analyzer Measurements, Application Note 2002,
1287-3, 5965-7709E.
3.2. Retrieval algorithm
The standard retrieval procedure for a normal medium relates the transmission and reflec‐
tion coefficients with the electromagnetic properties of the medium. As mentioned above,
for a chiral material a new experimental quantity is necessary. Here reflection coefficient, R,
and the transmission coefficients parallel (co-polar, T CO ) and perpendicular (cross-polar,
T CR ) to the polarization direction of the incident wave are the experimental data.
Consider a chiral slab of thickness L, which is illuminated by a normally incident linearly

polarized plane wave E0x^, Figure 4.
The incident wave may be decomposed as:
E0 E0
E0 x^ = 2
( x^ - j ^y ) + 2
( x^ + j ^y ) (11)
The reflected wave in z=0 is RE0x^, the co-polar and cross-polar transmitted fields in z=L are
T E x^ and T E y
CO 0
^, respectively.
CR 0
The reflected and transmitted electric fields are,
→ 1
E r = 2 RR E0( x^ - j ^y ) + 2 RL E0( x^ + j ^y ),
1
(12)
→ 1
E t = 2 T R E0( x^ - j ^y ) + 2 T L E0( x^ + j ^y ),
1
(13)
Figure 4. Schematic diagram of the experimental transmission and reflection coefficients and field of a chiral slab,
→ → →
where E i , E r and E t are the incident, reflected and transmitted fields, respectively.
where the subscript R (L) refers to right-hand (left-hand) circular polarization and subscript
r to the reflected wave. The magnetic field can be expressed in a similar way in terms of
wave impedance.
Applying the condition of continuity of tangential electric and magnetic fields at z=0 and
z=L, the reflection coefficients for RCP and LCP waves can be expressed as:
j sin (nk0 L )(1 - ηr2)

RR = R L = - 2 2ηr cos (nk0 L ) + j (ηr2 + 1)sin (nk0 L
(14)
)
The experimental reflection coefficient isR = R R + RL . The transmission coefficients for the
RCP and LCP waves are
- jκk 0 L
2ηr e
TR = (15)
2ηr cos (nk0 L ) + j (ηr2 + 1)sin (nk0 L )
+ jκk 0 L
2ηr e
TL = (16)
2ηr cos (nk0 L ) + j (ηr2 + 1)sin (nk0 L )
These are related with the experimental transmission coefficients T CO and T CR by:
T CO =
1
2
(T R + T L ) (17)
T CR =
-j
2
(T R - T L ) (18)
It is useful to consider the following additional parameters:
T 2 ≡ T RT L (19)
(T 2 - R 2 - 1) ± (T 2 - R 2 - 1)2 - 4R 2 ηr2 - 1
Γ≡ - = (η + 1)2 , (20)
2R r
-2 jnk0 L R-Γ
P 2= e = RΓ2 - Γ
(21)
We are now ready to compute the electromagnetic parameters. The relative impedance ηr is
calculated from equation (20), where the correct sign in square root is selected according to
Re(η) > 0.
The refractive index is obtained from equation (21) and is:
n = j 4πfL ln |P 2| - Phase(P 2) 4πfL -

c c c
2 fL m (22)
where m is an integer related to the branch index of the logarithm function. Only the imagi‐
nary part of n is unequivocally determined, but ambiguity arises for the real part because an
infinite number of solutions exist for different choices of m. The value of m can be deter‐
mined by applying Kramers-Kronig relations to the wavenumber [28] or by expanding
equation (22) in a Taylor series [29].
From nand ηr permittivity and permeability can be obtained:
n
εr = ηr (23)
μr = n ηr (24)
The chirality parameter is obtained from TL andT:
TL
T =e
j (κk0 L )
κ=
-j
k0 L ln | |+TL
T
1
k0 L Phase ( )+
TL
T
2π
k0 L p (25)
where pis an integer that can be determined by assuming null chirality far from resonance,
continuity conditions, and that both real and imaginary parts resonate at the same frequency.
The retrieval procedure may fail and problems arise when the thickness L of the effective
slab is not estimated accurately [30], when S11 and S21 are very small in magnitude [31] or
when the first boundary of the effective homogeneous slab is not well determined [29]
The rotation angle θand the ellipticity φof the transmitted wave can be calculated by:
1
θ = 2 arg ( ),
TL
TR
(26)
1
(| T R |2 - |T L
φ = 2 tan -1 |T |2 + |T |2
R L
|2
) (27)
3.3. Results
As already mentioned, periodical structures provide certain advantages over random distri‐
butions: for example, they are easier to manufacture and are more homogeneous and they
present extreme values for the rotation angle and chirality parameter. Figure 5 shows the re‐
flection and transmission coefficients, and the retrieved parameters εr , μr , andκr , for a typ‐
ical chiral slab composed of a periodical lattice of conductive cranks printed on a circuit
board substrate. The dimensions of the crank are similar to those proposed by [23, 24]. Ex‐
perimental measurements without and with TD filtering are shown in the first and second
columns, respectively. The third column shows the results obtained through simulations
based on the finite-difference time-domain method (FDTD). It can be seen that the TD filter‐
ing eliminates ripples in the measured transmission and reflection coefficients, which are
usually associated with multiple reflections or residual post calibration errors.
A clear resonance at around 11 GHz is present in all the plots, both measured and simulat‐
ed. Below the resonance, the retrieved parameters correspond to the blank FR4 board, which
does not present electromagnetic activity (εr ~ 4.2, μr ~ 1.0). As the frequency approaches the
resonance, all the parameters increase, with a maximum chirality in the order ofκ ~ 2.5. At
resonance, the chirality changes the sign and its imaginary part peaks. The frequency de‐
pendence of permittivity and permeability observed in the plot are the same as those in met‐
amaterials with a negative refractive index [14, 15, 24, 25]. The chirality parameter depends
on frequency, which follows a Condon model for a homogeneous chiral medium [32].
There is good agreement between the retrieved parameters obtained from experiments with
TD filtering and simulated data. It is important to highlight that the correct selection of TD
parameters is crucial for obtaining good results in the retrieval process. The values of the
time width (span) of the gating and the window shape affect the final results and, in this
respect, the use of experimental and numerical data improves the calculations. Two propa‐
gation modes with two different velocities inside the chiral slab can be observed in the sig‐
nal received by the receiving antenna when TD is applied, specially when the TCR
coefficient is measured. Two Gaussian curves arrive at two different times, which implies a
time span greater than that used for standard dielectric material.
4. Waveguides
4.1. Experimental systems

Waveguide techniques for characterizing chiral media are based on the same principles as
free-wave techniques, and it is necessary to measure three complex quantities. The experi‐
mental setup, therefore, must allow measurements in, at least, two different polarizations
for the transmission wave. All the experimental systems reported in the literature [7-9,
33-35] use circular waveguides because in rectangular waveguides the effects related to rota‐
tory dispersion cannot be observed. Lubkowski [36] proposed a square waveguide for meas‐
urements with SRR (Split Ring Resonators). There, the orthogonal modes TE10 and TE01 are
degenerated and, therefore, allow propagation for co-polar and cross-polar polarization.
This could be a good alternative for measurement involving chiral materials but, to the best
of our knowledge, such a system has not been implemented experimentally.
Figures 6 and 7 show a picture and a schematic diagram, respectively, of the experimental
setup using a circular waveguide developed by the authors. It is similar to that used by
Brewitt-Taylor [7] et al. and Liu et al. [34]. The circular measurement cell containing the chi‐
ral sample (2 in Figures 6, 7) is fed with a linearly polarized wave through a rectangular-to-
circular waveguide transition (1 in Figures 6, 7). The transmission wave passes to the
rectangular waveguide through another transition which can rotate around the longitudinal
axis (3 in Figures 6, 7), enabling measurements in different polarizations. The waveguide
transition contains a section of resistive film to absorb any field cross-polarized from the
fields being measured, which prevents them from being reflected from the waveguide tran‐
sition back on to the sample.
Figure 5. Transmission and reflection amplitudes (top panel), retrieved values for permittivity and permeability (mid‐
dle panel) and chirality (bottom panel). Experimental results without time domain filtering are shown in the left col‐
umn, with time domain filtering in the middle and simulation on the right.
S-parameters are measured using a network analyzer and the TRL technique is used for cali‐
bration as described for free-wave measurements. The REFLECT standard is achieved by
shorting with a metal plate the flanges where the measurement cell is connected, which de‐
fines the reference planes. The TRHU standard is achieved simply by connecting directly
these flanges without the measurement cell. Finally, the LINE standard is a section of the
circular waveguide quarter wavelength at midband.
Hollinger et al. [33] measured some effects of chiral materials including rotatory dispersion
and circular dichroism, but they did not study the inversion procedure. Liu et al. [9, 34] used
a slotted line instead of the network analyzer to measure the reflection coefficients of short-
circuited and open-circuited samples, completing the experimental measurements with the
rotation angle and axial ratio. Reinert et al. [8, 35] used coaxial-to-circular waveguide transi‐
tion to excite the fundamental mode in circular waveguide TE11, avoiding rectangular-to-cir‐
cular transitions.
Two aspects that can produce important errors must be taken into account in waveguide
measurements: the presence of local density variations in the sample, which causes fluctua‐
tions to the measurements, depending on sample orientation, and the existence of gaps be‐
tween the sample and the waveguide walls. In order to improve the accuracy of the results,
the samples are rotated and the measured scattering parameters for several positions are
averaged in an attempt to overcome the sample variations and inhomogeneities [8, 9, 33, 34].
The influence of the gap depends on sample manufacturing and can be overcome, for in‐
stance, by locating the chiral particles in non-conducting spheres [8], since the multiple air
gaps between the spheres are part of the sample, or by using an elastic material as dielectric
host for distributing the chiral particles [12].
Figure 6. Schematic diagram of the X-band circular waveguide experimental system developed by the authors.
Figure 7. Photo of the experimental diagram of Figure 6.
4.2. Modes in circular waveguides
The propagation of electromagnetic fields in circular waveguides filled with chiral media,
also known as chirowaveguides, can be better understood by using the normal modes for‐
mulation. Assuming, as usual, solutions of the type e -γz , the longitudinal component of the
electric field satisfies the Helmholtz equation [37],
(∇2t + K ±2)E±z = 0 (28)
where
K ± = (k±2 - γ 2)1/2 (29)
Once E±z are known, the transversal components can be obtained by:
→
(-γ∇t - k+^z × ∇t )E+z + (-γ∇t - k-^z × ∇t )E-z
1 1
Et (r ) = K +2 K -2
(30)
Similar expressions for the magnetic components can be deduced. Assuming an azimuthal
dependence ase - jmφ , the longitudinal component of the electric field is a linear combination
of the longitudinal components of the normal modes:
→ ∞ ∞ -γ z
E z (r ) = ∑ ∑ { Am,l ,+ J m( K l ,+ρ ) + Am,l ,- J m( K l ,-ρ )}e - jmφ e l (31)
m=-∞ l=1
where classification of the normal modes is described by {m, l } indices. Am,l ,±is the ampli‐
tude of the mode, J mis the m order Bessel function and J' m its derivate. If the boundary is
ideally conducting, the following eigenvalue equation is obtained:
mγl
a ( 1
K l 2,+
-
1
K l 2,-
)+ k+ J' m( K l ,+a)
K l 2,+ J m( K l ,+a)
+
k- J' m( K l ,-a)
K l 2,- J m( K l ,-a)
=0 (32)
where a is the waveguide radius. For a pair of solutions satisfying γl ' = - γl andm' = - m, the
parameters K + K - are coincident. The wave number is, in general, a complex quantity
γl = αl + jβl .
Figure 8 shows the dispersion diagram of a circular waveguide for X band (a = 1.19cm) filled
with a chiral material (εr = 3.0, μr = 1.0, κr = 0.8) for m = ± 1. This diagram is, clearly, much
more complex than the one corresponding to a non-chiral waveguide. A detailed discussion
on the modal expansion can be found in [37].
4.3. Retrieval algorithm
When several modes propagate in the chirowaveguide it is not possible to obtain inverse
equations. Only if just one mode propagates can a simple retrieval algorithm be developed.
However, this is only valid for low values of the chirality parameter.
Figure 8. Dispersion diagram of chirowaveguide (1.19 cm radius, εr = 3.0, μr = 1.0,κr = 0.8), for m = 1(solid line), m = - 1
(dotted line). The propagation constant at frequencies below the cut-off is shown by dash-dot lines.
Liu et al. [9, 34] assumed that only the fundamental mode, HE±11 (m = ± 1, j = 1), propagates in
the sample, and this is approximately treated in the same way as the TE11 mode, the funda‐
mental mode in the empty waveguide. In this case, the procedure for obtaining the inver‐
sion equations is similar to the one used in the free-wave technique. The above mentioned
authors used the following experimental quantities: Γ1andΓ2, the complex reflection coeffi‐
cients of the open circuit sample (sample in transmission situation) and the short circuit
sample (sample backed with metal plate) in the circular waveguide, respectively, θis the ro‐
tation angle and RA is the axial ratio, for the electromagnetic waves travelling through chiral
material in the circular waveguide.
The characteristic parameters of the material are calculated from the measured quantities by:
ωμ0
Z 0' = 1
(33)
ω 2ε0μ0 - kc2 2
( )
1
Γ1Γ2 + Γ1 + Γ2 + 1
Z ω = Z 0' 2 (34)
Γ1Γ2 - 3Γ1 + Γ2 + 1
A=
1
2d arccos ( Γ1Γ22 + Γ22 - 2Γ1Γ2 - Γ1 + 1
2(Γ1 - Γ2)
) (35)
B=
1
2d (2θ - jtanh -1
2R A
1 + R A2
) (36)
Zω A
μ= ω
(37)
B ( A 2 + kc2) 2
(38)
κ= aμω
( A 2 - B 2)μκ 2
ε= (39)
B2
where dis the thickness of the chiral sample and kc the cut-off constant for the fundamental
mode TE11 in the empty waveguide.
These equations are relatively easy to use but are only valid for media with small values of
the characteristic parameters. For example, Figure 8 shows two propagating modes in the X-
Band for a chirality relative constant κ r = 0.8, and, so, the previous retrieval algorithm would
not be useful in that case. Other authors consider more modes. For instance, Busse et al. [8,
35] used five modes but could not find inversion equations to calculate the permittivity, per‐
meability and chirality of the medium. Instead, they processed the direct problem numeri‐
cally, in an iterative way, to fit measured quantities: the reflection coefficient R, the co-polar
transmission coefficient T CO , and the cross-polar transmission coefficient T CR . Using the
expression for waveguide modes in the empty and filled regions and starting with approxi‐
mate values for permittivity, permeability and chirality, the material properties are obtained
by iterative fitting of the calculated S parameters to the experimental ones.
Brewitt-Taylor et al. [7] followed a similar procedure to calculate the characteristic parame‐
ters. More specifically, they used four modes and eight different points to perform the mo‐
dal analysis in the interfaces. In this case, mode matching was performed at the two
interfaces simultaneously. The number of iterations required varied from 5 to 30, although
the authors admit that high chirality samples could require more iterations.
Both Reinert et al. and Brewitt-Taylor et al. established the number of modes empirically,
this number increasing with the values of the characteristic parameters. For instance, the
samples used by Brewitt-Taylor et al. presented maximum values of 4 for the permittivity,
0.5 for the chirality and around 1 for the permeability, while the corresponding rotation an‐
gle was always less than 20º. Samples manufactured by the authors present values of up to 7
for permittivity, and 2 for chirality and permeability (see Figure 5), finding rotation angles
in the range of 90º (Figure 9). The number of modes which should be used is, probably,
more than five, increasing the computation requirements.
5. Resonators
Some authors propose using resonator techniques to characterize chiral media [38, 39].
Such techniques are usually the most suitable method for obtaining high precision meas‐
urements. However, there are several reasons why they are difficult to use with chiral me‐
dia. In particular, resonators work at discrete frequencies this is not useful to deal with
dispersive media; only two real quantities (resonant frequency and quality factor) can be
measured and, as mentioned above, three complex quantities have to be measured to ex‐
tract permittivity, permeability and chirality. Measurements using different resonant modes
and samples in several positions is a possibility, but, in this case, the experimental proce‐
dure becomes very complicated.
Figure 9. Rotation angle produced by several samples composed of four cranks in foam [12]. Labels indicate the size
(in mm) of the cranks.
Tretyakov et al. [38, 39] proposed a perturbation technique, although this gives rise to a new
problem. It is well known that perturbation techniques are only valid for small samples. But
small samples of chiral media will have a small number of inclusions, perhaps only one, and
so it makes no sense to treate them as a homogeneous material.
In our opinion, resonant techniques are not suitable for characterizing chiral media. In fact,
although Tretyakov et al. studied the proposed technique theoretically, none was imple‐
mented experimentally.
6. Discussion and conclusions
In this work, microwave techniques for characterizing chiral media have been revised, and
free-wave, waveguide and resonator techniques have been analyzed. Since chiral materials
are characterized by three complex parameters, permittivity, permeability and chirality, at

least, three experimental quantities have to be measured. All the techniques considered
present some difficulties, while free-wave techniques are, in our opinion, the most suitable:
i) They do not need any special component. ii) Quasi-plane wave propagation can be as‐
sumed in the sample, so that inversion equations for extracting characteristic parameters can
be developed without difficulty. Usually, the measured quantities are the reflection coeffi‐
cient, the co-polar and cross-polar transmission coefficients.
Free-wave methods present an inherent problem due to the finite size of the sample, which
should be large enough to avoid diffraction at the borders. Several techniques found in the
literature overcome this problem by using focusing systems, lenses or ellipsoidal mirrors.
They concentrate the propagated wave at a tiny spot, where the sample is situated and so
reasonably sized samples can be used. Many authors use time domain transformations to re‐
move unwanted reflections from the experimental measurements.
Waveguide measurements may be considered a possible alternative. The measuring princi‐

ple is similar to that of free-wave techniques, but they present the advantage of avoiding
many kinds of external interference and use only small samples, although good machining
is needed. However, they present one serious problem: usually several modes are propagat‐
ed in the sample and inversion equations cannot be developed. All the techniques found in
the literature use a circular waveguide because axial symmetry is necessary for measure‐
ments in different polarizations (only waves with polarization parallel to the short side can
be propagated in rectangular waveguide). Unfortunately many circular waveguide compo‐
nents are not commercially available. A common solution is to use a rectangular waveguide
and to feed the measurement cell, made in the circular waveguide, by means of two rectan‐
gular to circular waveguide transitions. The transition contains a resistive film to absorb any
fields cross-polarized from those being measured. Square waveguides could, in principle, be
another alternative for chiral characterization, although we have not found any theoretical
nor experimental study in the current literature.
Finally, it should be emphasized that resonator techniques are not suitable for characterizing
chiral media. Only one research group has proposed a resonator technique in a theoretical
study without experimental implementation. In fact, resonators work at a fixed frequency
and are therefore not useful for measurements in dispersive media.
Acknowledgements
This work was supported by the Dirección General de Investigación (TEC 2010-21496-
C03-02) of the Spanish Ministry of Education and Science and by Fundación Séneca
(11844/PI/09) Región de Murcia (Spain).
Author details
J. Margineda1*, G.J. Molina-Cuberos1, M.J. Núñez1, A.J. García-Collado1,2* and E. Martín1
1 Universidad de Murcia, Murcia, Spain
2 Universidad Católica San Antonio, Murcia, Spain
References
[1] Lindell, I. V., Sihvola, A. H., Tretyakov, S. A., & Vitanen, A. J. (1994). Electro-magnet‐
ic Waves in Chiral Media, Boston, USA: Artech House.
[2] Arago, D. F. (1811). Mémoire sur une Modification Remarquable Qu’éprouvent les
Rayons Lumineux dans leur Passage a Travers Certains Corps Diaphanes, et sur
Quelques Autres Nouveaux Phénoménes D’optiqueMémoires de la Classe des Sciences
Mathématiques et Physiques de l’Institut Impérial de France, 1-93.
[3] Biot, B. (1812). Mémoire sur un Nouveau Genre DÒscillation que les Molécules de la
Lumiére Éprouvent en Traversant Certains Cristaux. Mémoires de la Classe des Sciences
Mathématiques et Physiques de l’Institut Impérial de France, 1.
[4] Lindman, K. F. (1914). Om en Genom ett Isotropt System av Spiralformiga Reso‐
natorer Alstrad Rotationspolarisation av de Elektromagnetiska Vagorna. Öfversigt af
Finska Vetenskaps-Societetens förhandlingar, A. Matematik och naturvetenskaper LVII
(3) 1-32.
[5] Tinoco, I., & Freeman, M. P. (1957). The Optical Activity of Oriented Copper Helices.
I. Experimental, J Physical Chemistry, 61-1196.
[6] Ro, R., Varadan, V. V., & Varadan, V. K. (1992). Electromagnetic Activity and Ab‐
sorption in Microwave Composites, Proc. Inst. Elect. Eng. H, 139(5), 441-448.
[7] Brewitt-Taylor, C. R., Lederer, P. G., Smith, F. C., & Haq, S. (1999). Measurement and
Prediction of Helix-Loaded Chiral Composites, IEEE Transactions on Antennas and
Propagation, 47(4), 692-700.
[8] Busse, G., Reinert, J., & Jacob, A. F. (1999). Waveguide Characterization of Chiral Ma‐
terial: Experiments. IEEE Transactions on Microwave Theory and Techniques, 47(3),
297-301.
[9] Sun, G., Yao, K., & Liu, Z. (1999). Influence of Helix Parameters on the Reflectance of
Microwave by Chiral Composites, IEEE Transactions on Electromagnetic Compatibility,
411-350.
[10] Gómez, A., Lakhtakia, A., Margineda, J., Molina-Cuberos, G. J., Núñez, M. J., Ipiña,
S. A., & Vegas, A. (2008). Full-Wave Hybrid Technique for 3-D Isotropic-Chiral-Ma‐
terial Discontinuities in Rectangular Waveguides: Theory and Experiment, IEEE

Transactions on Microwave Theory and Techniques, 56(12), 2815-2824.
[11] Molina-Cuberos, G. J., García-Collado, A. J., Margineda, J., Núñez, M., & Martín, E.
(2009). Electromagnetic Activity of Chiral Media Based on Crank Inclusions, IEEE
Microwave and Wireless Components Letters, 19(5), 278-280.
[12] Barba, I., Cabeceira, A. C. L., García-Collado, A. J., Molina-Cubero, G. J., Margineda,
J., & Represa, J. (2011). Quasi-planar Chiral Materials for Microwave Frequencies. In:
Kishk A. (ed.) Electromagnetic Waves Propagation in Complex Matter, Intech,
978-9-53307-445-0, http://www.intechopen.com/books/electromagnetic-waves-propa‐
gation-in-complex-matter.
[13] Pendry, J. B. (2004). A Chiral Route to Negative Refraction, Science 306 1353--1355.
[14] Wang, B., Zhou, J., Koschny, T., Kafesaki, M., & Soukoulis, C. M. (2009). Chiral Meta‐
materials: Simulations and Experiments, J. Opt. A: Pure Appli. Opt. 11: 114003.
[15] Mackay, T. G., & Lakhtakia, A. (2012). Negatively Refracting Chiral Metamaterials: A
Review, SPIE Reviews 1: 018003.
[16] Sun, G., Yao, K., Liu, Z., & Huang, Q. (1998). A Study on Measuring the Electromag‐
netic Parameters of Chiral Materials. J. Phys. D: Appl. Phys., 31-2109.
[17] Chen, L. F., Ong, C. K., Neo, C. P., Varadan, V. V., & Varadan, V. K. (2004). Micro‐
wave Electronics Measurement and Materials Characterization, Chinchester, England
John Wiley & Sons.
[18] Lynch, A. C., & Simkin, D. (1990). Measurement of Permeability and Permittivity Of
Ferrites. Measurement Science Technology, 1-1162.
[19] Rickard, Petersson., & Smith, G. S. (2002). An Estimate of the Error Caused by the
Plane-Wave Approximation in Free-Space Dielectric Measurement Systems, IEEE
Transactions on Antennas Propagation, 50(6), 878-887.
[20] Guire, T., Varadan, V. K., & Varadan, V. V. (1990). Influence of Chirality on the Re‐
sponse of EM Waves by Planar Dielectric Slabs, IEEE Transactions on Electromagnetic
Compability, 32(4), 300-303.
[21] Ghodgaonkar, D. K., Varadan, V. V., & Varadan, V. K. (1989). A Free-Space Method
for Measurement of Dielectric Constants and Loss Tangent at Microwave Frequen‐
cies. IEEE Transactions on Instrumentation and Measurement, 38-789.
[22] Muñoz, J., Rojo, M., Parreño, A., & Margineda, J. (1998). Automatic Measurement of
Permittivity and Permeability at Microwave Frequencies Using Normal and Oblique
Free-Wave Incidence with Focused Beam, IEEE Transactions on Instrumentation and
Measurements, 47(4), 886-892.
[23] García-Collado, A. J., Molina-Cuberos, G. J., Margineda, J., Núñez, M. J., & Martín, E.
(2010). Isotropic and Homogeneous Behavior of Chiral Media Based on Periodical In‐
clusions of Cranks, IEEE Microwaves and Wireless Components Letters, 20(3), 176-177.
[24] Molina-Cuberos, G. J., García-Collado, A. J., Barba, I., & Margineda, J. (2011). Chiral
Metamaterials with Negative Refractive Index Composed by an Eight-Cranks Mole‐
cule, IEEE Antennas and Wireless Propagation Letters, 10-1488.
[25] García-Collado, A. J., Molina-Cuberos, G. J., Núñez, M. J., & Margineda, J. (2012).
Negative Refraction of Chiral Metamaterial Based on Four Crank Resonator, Journal
of Electromagnetic Waves and Applications, 26-986.
[26] Agilent Technologies Inc. (2012). Time Domain Analysis Using a Network Analyzer,
Application Note 1287-12, 5989-5723.
[27] Bohren, C. F. (1974). Light Scattering by an Optically Active Sphere. Chemical Physics
Letters, 29-458.
[28] Varadan, V. V., & Ro, R. (2007). Unique Retrieval of Complex Permittivity and Per‐
meability of Dispersive Materials from Reflection and Transmitted Fields By Enforc‐
ing Causality, IEEE Transactions on Microwave Theory and Techniques, 55(10),
2224-2230.
[29] Chen, X., Grzegorczyk, T. M., Wu-I, B., Pacheco, J., & Kong, J. A. (2004). Robust
Method to Retrieve the Constitutive Effective Parameters of Metamaterials. Physical
Review E.; 70 016608.
[30] Smith, D. R., Schultz, S., Markoš, P., & Soukoulis, C. M. (2002). Determination of Ef‐
fective Permittivity and Permeability of Metamaterials from Reflection and Trans‐
mission Coefficients. Physical Review B. 65 195104.
[31] Ziolkowski, R. W. (2003). Design, Fabrication, and Testing of Double Negative Meta‐
materials. IEEE Transactions on Antennas and Propagation, 7-1516.
[32] Condon, E. U. (1937). Theories of Optical Rotatory Power, Review Modern Physics,
9-432.
[33] Hollinger, R., Varadan, V. V., & Varadan, V. K. (1991). Eigenmodes in a Circular
Waveguide Containing an Isotropic Chiral Material, Radio Science, 26(5), 1335-1344.
[34] Liu, Z., Sun, G., Huang, Q., & Yao, K. (1999). A Circular Waveguide Method for
Measuring the Electromagnetic Parameters of Chiral Materials at Microwave Fre‐
quencies. Measurement Science Technolology, 10-374.
[35] Reinert, J., Busse, G., & Jacob, A. F. (1999). Waveguide Characterization of Chiral Ma‐
terial: Theory. IEEE Trans. Microwave Theory and Techniques, 47(3), 290-296.
[36] Lubkowski, G. (2009). Simulation of Electromagnetic Fields in Double Negative Met‐

amaterials, Technishen Universität Darmstadt, PhD Dissertation.
[37] Barybin, A. A. (1998). Modal Expansions and Orthogonal Complements in the Theo‐
ry of Complex Media Waveguide Excitation by External Sources for Isotropic, Aniso‐
tropic, and Bianisotropic Media, Progress in Electromagnetic Research, PIER, 19,
241-300.
[38] Tretyakov, S. A., & Viitanen, A. J. (1992). Perturbation Theory for a Cavity Resonator
with a Biisotropic Sample: Application to Measurement Techniques, Microwave and
Opt. Tech. Lett., 5(4), 174-177.
[39] Tretyakov, S. A., & Viitanen, A. J. (1995). Waveguide and Resonator Perturbation
Techniques Measuring Chirality and Nonreciprocity Parameters by Biisotropic Mate‐
rials, IEEE Transactions on Microwave Theory and Techniques, 43(1), 222-225.
Chapter 2
Chiral Waves in Graphene Medium and Optical

Simulation with Metamaterial
H. Torres-Silva
1. Introduction
In modern age Engineers have paved the way for a new generation of faster, more powerful
cell phones, computers and other electronics by developing a practical technique to replace
silicon with carbon on large surface. The capability of silicon, the material at the heart of
computer chips has been harnessed beyond its limits by engineers and carbon has come up
as an integrating replacement for the same. The material called“Graphene” which is a single
layer of atoms arranged in honeycomb lattice could let electronics to process information
and produce radio transmission 10 times better than silicon based devices.
For theorists, such a system is also of great interest because it provides a physical realization
of two-dimensional field theories with quantum anomalies. Indeed, the continuum limit of
the effective theory describing the electronic transport in graphene is that of two-dimension‐
al massless Dirac fermions. The reported and predicted phenomena include the Klein para‐
dox (the perfect transmission of relativistic particles through high and wide potential
barriers), the anomalous quantum Hall effect induced by Berry phases and its correspond‐
ing modified Landau levels and the experimental observation of a minimal conductivity.
From the point of view of its electronic properties, graphene is a two-dimensional zero-gap
semiconductor with the cone energy spectrum, and its low-energy quasiparticles are formal‐
ly described by the Dirac-like Hamiltonian [1, 2].
H 0 = − iℏvF σ ⋅ ∇
Where vF ≈ 106ms −1is the Fermi velocity and σ = (σx , σy ) are the Pauli matrices. The fact that
charge carriers are described by the Dirac-like equation, rather than the usual Schrödinger
equation, can be seen as a consequence of graphene’s crystal structure, which consists of two
equivalent carbon sublattices [1, 2]. Quantum mechanical hopping between the sublattices
leads to the formation of two cosine-like energy bands, and their intersection near the edges
of the Brillouin zone yields the conical energy spectrum. As a result, quasiparticles in gra‐
phene exhibit the linear dispersion relationEG = E = ℏkvF , as if they were massless relativistic
particles with momentum k (for example, photons) but the role of the speed of light is
played here by the Fermi velocityvF ≈ c / 300. Owing to the linear spectrum, it is expected that
graphene’s quasiparticles will behave differently from those in conventional metals and
semiconductors where the energy spectrum can be approximated by a parabolic (free-elec‐
tron-like) dispersion relation.
From a crystallographic point of view, the graphene is a triangular Bravais lattice with a dia‐
mond-shaped unit tile consisting of two sites so one gets the honeycomb structure. The very
unique feature of the graphene band structure is that the two lowest-energy bands, known
as the valence and the conduction bands, touch at two isolated points located at the corners
of the Brillouin zone. In the immediate vicinity of these degeneracy points, known as the
Dirac points, the band structure is a cone. In natural graphene samples, there is exactly one
electron per site, and thus, at zero temperature, all levels in the valence band are filled (a
situation known as half-filling). As a result, the energy of the last occupied level precisely
slices the band structure at the Dirac points. The low-energy excitations of this system are
then described by the massless two-dimensional Weyl-Dirac equation and their energy dis‐
persion relationω = vF k is that of relativistic massless fermions with particle-hole symmetry.
In graphene these massless fermions propagate with a velocityvF . The maths is simple but
the principles are deep. We will review the formulation of graphene’s massless Dirac Hamil‐
tonian, under the chiral electromagnetism approach, like a metamaterial media, hopefully
demystifying the material’s unusual chiral, relativistic, effective theory. The novel result
here is that in our theory we do not make c → vF [1, 2, 27-35], but we obtain vF as vF = c(1 − k0T )
ifk0T > 0 orvF = c(1 + k0T ) ifk0T < 0. These results are derived of the Chiral Electrodynamics
with T as the chiral parameter andk0 = ω / c when the electric wave Eis quasi parallel to the
magnetic waveH .
With this approach we can review different phenomena such: The Dirac point with a dou‐
ble-cone structure for optical fields, an optical analogy with Dirac fermions in graphene, can
be realized in optically homogenous metamaterials. The condition for the realization of Dir‐
ac point in optical systems is the varying of refractive index from negative to zero and then
to positive.
Also we give a support to the similitude of the band structure of a macroscopic photonic
crystal with the electronic band structure of grapheme, which is experimentally much more
difficult to access, allows the experimental study of various relativistic phenomena. With
our analytical and numerical analysis we hope to verify that, similar behaviors exist to elec‐
trons in graphene treated as mass-less particle, ie, electron wave propagation.
This chapter presents a short review on the chiral propagation of electron waves in mono‐
layer graphene and optical simulation with optical field in the negative-zero-positive index
metamaterial NZPIM and its close connection. Section II presents an enhanced vector dia‐
gram of Maxwell’s equations for chiral media with quasi parallel electromagnetic fields,
Chiral Waves in Graphene Medium and Optical Simulation with Metamaterial 27
E ∥ H . Chiral waves in graphene acting as metamaterial media are discussed in section III.
In section IV, two component equations and tunneling rate of Dirac electron in graphene are
derived. Section V is about Zitterbewegung of optical pulses near the Dirac point inside a
negative-zero-positive index metamaterial, and shows that the chiral field near the Dirac
point becomes a diffusive wave The last sections describe the theoretical description of mini‐
mal conductivity in graphene under chiral approach and absorption of light by quasi 2D
Dirac fermions.
2. A vector diagram of Maxwell’s equations for chiral media with E ∥ H
The idea of representing Maxwell’s time-harmonic equations in homogeneous isotropic me‐

dia by vector diagram as put forward by Wilton [3] and by S. Uckun [4] deserves considera‐
tion. All the common relations between field and potential quantities implied by Maxwell’s
equations can be represented by a diagram. It is started that the diagram not only illustrates
Maxwell’s equations, but also many of the methods for constructing diagram are based on
the formal similarity between many theorems of vector calculus and those of vector algebra.
An isotropic chiral medium is a macroscopically continuous medium composed of equiva‐

lent chiral objects that are uniformly distributed and randomly oriented. A chiral object is a
three-dimensional body that cannot be brought into agreement with its mirror image by
translation and rotation. An object of this sort has the property of handedness and must be
either left-handed or right-handed. An object that is not chiral is said to be achiral, and thus
all objects are either chiral o achiral. Due to their novel properties and wide applications in
microwave and radar engineering, chiral media has been undergoing extensive research
during the last years. That is why this study aims to cover chiral medium for the representa‐
tion of Maxwell’s equations in vector diagram form. In a chiral media a cross coupling be‐
tween electric and magnetic filed exists. Thus, the vector diagram has vectors along all three
coordinate axes whereas the vector diagram presented by Wilton [3] for achiral media has
vectors only in one plane with H vector normal to it.
2.1. Vector diagram construction
Assumingexp(iωt) time dependence, Maxwell’s time-harmonic equations [5, 6] for isotropic,

homogeneous, linear media are:
∇ × E = − jωB (1)
∇ × H = jωD + J (2)
∇ ⋅ B =0 (3)
∇ ⋅D =ρ (4)
Chirality is introduced into the theory by defining the following constitutive relations to de‐
scribe the isotropic chiral medium [5]
D = εE + jωεT B (5)
1 − ko2T 2
H = jωεT E + B (6)
μ
Where the chirality admittance − ωεT indicates the degree of chirality of the medium, and
the εandμare permittivity and permeability of the chiral medium, respectively. Since D and
Eare polar vectors and B and H are axial vectors, it follows that εand μare true scalars and
− ωεT is a pseudoscalar. This means that when the axes of a right-handed Cartesian coordi‐
nate system are reversed to form a left-handed Cartesian coordinate system, − ωεT changes
in sign whereas εand μremain unchanged.
For a graphical representation of the above relationships, following Wilton’s procedure [3],
let us assume vector-differential operator,∇ is an ordinary vector and treat the divergence
and curl operations in equations (1) to (4) as ordinary scalar (dot) and vector (cross) prod‐
ucts, respectively. Equation (3) implies that ∇ is perpendicular to Band the vector ∇ × B
must be perpendicular to both ∇ andB.
As shown in Figure 1, three transverse coordinate axes are chosen as∇ , from Equation (1)
B = − ∇ × E / ( jω) and, ∇ × B / ( jωεμ ') = ∇ × ∇ × E / k 2wherek 2 = ω 2εμ '.
Since ∇ ⋅ B = 0always, this conditions will hold identically if B is expressed as the curl of a
vector potential Asince the divergence of the curl of a vector is identically zero, thus
B =∇ × A (7)
andAmust be perpendicular to both ∇ and Band lie in ∇ and ∇ × Bplane. However, Ais not
unique since only its components perpendicular to ∇ contribute to the cross product. There‐
fore,∇ ⋅ A , the component of Aparallel to∇ , must be specified. The curl equation forE, as in
Equation (1), and Equation (7) give ∇ × (E + jωA) = 0 where the quantity in parentheses
should be parallel to ∇ and the curl of the gradient of a scalar function ϕ is identically zero;
so the above equation is E + jωA = − ∇ ϕ or
E = − jωA − ∇ ϕ (8)
That is shown in Figure 1.

με
Figure 1. Diagram of the full Maxwell System for a chiral media with Lorentz gauge = μ 'ε '.
1 − ko2T 2
Following the Uckun’s approach [5], we substitute Equation (7) into Equation (6) having
1 − ko2T 2
H= ∇ × A + jωεT E (9)
μ
Substituting Equation (9) and Equation (5) into Equation (2) gives
ωμεT ωμε ω 2μεT μ
∇ × ∇ × A + j 1 − k 2T 2 ∇ × E = j 1 − k 2T 2 E − 1 − ko2T 2 B+ 1 − ko2T 2 J
o o
2
ωμεT ωμε ω μεT μ
∇ × ∇ × A + j 1 − k 2T 2 ∇ × E = j 1 − k 2T 2 E − 1 − ko2T 2 B+ 1 − ko2T 2 J , playing the value of∇ × E from
o o
μω 2εT ωμε μ
Equation (1) into the above equation ∇ × ∇ × A + 2 1 − k 2T 2 B = j 1 − k 2T 2 E + 1 − ko2T 2 J and using
o o
2
the vector identity ∇ × ∇ × A = ∇ (∇ ⋅ A) − ∇ A enables us to write the above equation as
ω 2μεT ωμε μ
∇2 A + 2 1 − k 2T 2 ∇ × A = ∇ (∇ · A) − j 1 − k 2T 2 E − 1 − ko2T 2 J
o o
ω 2μεT ωμε μ
∇2 A + 2 1 − k 2T 2 ∇ × A = ∇ (∇ ⋅ A) − j 1 − k 2T 2 E − 1 − ko2T 2 J , so using Equation (8), we have
o o
k02 ω 2μεT ω 2μεTϕ μ

∇2 A + 1 − k02T 2 A + 2 1 − k 2T 2 (∇ × A) = ∇ (∇ ⋅ A − j 1 − k 2T 2 ) − 1 − ko2T 2 J (10)
o o
Here ∇ ⋅ Ais arbitrary, so in order to specify∇ ⋅ A, for uniqueA, we may choose
ωμεϕ
∇ ⋅ A= j (11)
1 − ko2T 2
And eliminate the term in parentheses in Equation (10). The choice in Equation (11) can be
known as a chiral Lorentz gauge. Then Equation (10) will be simplified to
k02 ω 2μεT μ
∇2 A + 1 − k02T 2 A + 2 1 − k 2T 2 (∇ × A) = − 1 − ko2T 2 J (12)
o
ωμε
Divide both sides of Equation (12) by ( j 1 − k 2T 2 ) and reorganize it to get
o
∇2 A 1
(1 − ko2T 2) = jωA + 2 jωT (∇ × A) − J (13)
jωμε jωε
( Figure 1, shows this vectorial equation).
The difference between our approach and the Uckun’s procedure [4], is that we take the chi‐
ral media characterized by D = ε(E + T ∇ × E) andB = μ(H + T ∇ × H ). In this form we can ob‐
tain the condition of quasi spatial parallel condition between Band Ewhere the main
equation is like a Beltrami equation which is important for the numerical simulation of gra‐
phene systems.
Placing the value of B from Equation (1) into Equation (5) D = ε(E + T ∇ × E)will be obtained.
Placing the value ofH , from Equation (6), and D into Equation (2) will give
1 − ko2T 2
∇ × ( jωεT E + μ B) = jω(εE + εT ∇ × E) + J
by rearranging this equation we obtain
∇ ×B J
(1 − ko2T 2) = E + 2T ∇ × E + (14)
jωμε jωε
will be obtained as shown in Figure 1. In this figure we putμ → μ / (1 − k02T 2).
Taking divergence of Equation (5) and using Equations (3) and (4) in it
ρ = ε∇ ⋅ E (15)
will be derived. To find the projection of E onto∇ , from Equation(15) ∇ ⋅ E = ρ / ε, take the
gradient of both sides and divide by scalar value ∇2 to normalize ∇ to a unit vector. So
∇ (∇ ⋅ E) ∇ ρ
= (16)
∇2 ε∇2
Similarly, getting gradient of both sides of Equation(11), using the vector identify
∇ × ∇ × A + ∇2 A for ∇ (∇ ⋅ A) and normalizing by ∇2 we have:
∇ ×∇ × A ∇ϕ
jw ∇2
+ jw A = k 2 ∇2
(17)
So it will be obtained as parallel component of A to ∇ coordinate.
By using Equations (8), (13), (14), (16), and (17) the vector diagram of Lorenz gauge can be
completed as shown in Figure 1, where all Maxwell’s relations and potential quantities appear.
Now let us examine derivation of some relations from the diagram. For example, it is seen
that the component of E and J / ( jωε) parallel to ∇ must be equal and opposite. By taking
the divergence of Equation (14) and using ∇ ⋅ E = ρ / ε can be shown that
1 ρ
∇ ⋅ E = − jωε ∇ ⋅ J = ε (18)
Taking the gradient in both sides of Equation (18) and dividing it by scalar value ∇2 will
give the same value as Equation (16) with opposite sign. From the right side of Equation (18)
it is seen that
1
ρ= − ∇ ⋅J (19)
jw
This is the known continuity equation. Since the divergence of the curl of any vector is iden‐
tically zero, the divergence of Equation (2) yields.0 = jω∇ ⋅ D + ∇ ⋅ J . Using Equation (4)
convert this immediately into continuity equation as, expected. Again, as seen in Figure 1,
2T ∇ × Eand − jω2T ∇ × Α are equal and opposite vectors. From Equation(8), taking curl of
both side and using the vector identity ∇ × ∇ ϕ = 0 will show that
2T ∇ × E = − jω2T ∇ × A (20)
as expected. By using the vector calculus a few possible equations from the vector diagram
can be written as follows
J ∇ ×B
2T ∇ × E + jωε − (1 − ko2T 2) jωμε − ∇ ϕ − jωA = 0 (21)
∇ ×∇ × A ∇ϕ
jω −k 2 2 − E −∇ϕ =0 (22)
∇2 ∇
∇ ×∇ × A ∇ϕ J ∇2 A
jω −k 2 2 + + (1 − ko2T 2) − 2 jωT ∇ × A = 0 (23)
∇ 2
∇ jωε jωμε
∇2 A ∇ ×B
(1 − ko2T 2) − 2 jωβ∇ × A + ∇ ϕ + (1 − ko2T 2) − 2T ∇ × E = 0 (24)
jωμε jωμε
For example, adding Equations (13) and (14) side by side and using Equation (8) will give
Equation (24) which shows the correctness of the equation derived from the diagram 1. In‐
stead of Lorenz gauge we can choose Coulomb’s gauge.
∇ ⋅ Ac = 0 (25)
In Equation (10) so that it will take the form

ω 2μεT ωμε μ
∇2 Ac + k 2 Ac − 2 1 − k 2T 2 (∇ × Ac ) = j 1 − k 2T 2 ∇ ϕ − 1 − ko2T 2 J
o o
where the subscript " c " is used it indicate Coulomb’s gauge. Using Equation (8) and (20) in
the above equation
∇2 Ac J
− (1 − ko2T 2) − 2T ∇ × E = E + (26)
jωμε jωε
will be obtained. Placing the values of Equations (5) and (6) into Equation (2) will give
(1 − ko2T 2)
∇ × ( jωεT E + μ B) = jω(εE + jωεT B) + J
and value of ∇ × E from Equation (1) will give
∇ ×B J
(1 − ko2T 2) jωμε + 2 jωT B = E + jωε
(31)
Combining these equations with Equation (26) and using Equation (7) we have
∇ ×B ∇2 Ac
(1 − ko2T 2) jωμε = − (1 − ko2T 2) jωμε (27)
By using the same coordinates axes∇ , Band (1 − ko2T 2)∇ × B / ( jωμε) and Equation (1), (8),
(16), (26) and (27) for the Coulomb gauge. It is clear from Equation (25) that the component
of the vector A parallel to∇ is equal to zero.
As seen in Figure 1, Lorenz gauge are the best choice because these make A either parallel or
perpendicular to any of the other vectors and simplify its relationship to those vectors. In
Figure 1a, if the chirality factorT , goes to zero, pointK , L and R approach pointM , P, and N
respectively, in which case the diagram will be the same as in Reference [3] for linear, homo‐
geneous, isotropic achiral medium. If (1 − ko2T 2) → 0 then E is quasi parallel to B, and parallel
toAso all vectors remain in an only plane. (see Figure 2).
Figure 2. Two dimensional condition when (1 − ko2T 2) → 0for a graphene system, the thickness is about one carbon
atom
This Beltrami condition is useful to numerical calculations in graphene. We apply this ap‐
proach to a two dimensional chiral graphene slab. This result cannot be obtained with the
Uckun’s approach [4]. In terms of chiral magnetic potentialAc withJ = 0, and∇ ρ = 0 we have
the wave equation
k02 ω 2μεT
∇2 Ac + Ac + 2 (∇ × Ac ) = 0 (28)
1 − k02T 2 1 − ko2T 2
Ask = kx2 + ky2 + kz2, ifkx = ksinθ, ky = 0, andkz = kcosθ, we have the matrix:
( − k 2(1 − k02T 2) + k02

2 jk02kT cosθ
0
−k
− 2 jk02kT cosθ
2
(1 − k02T 2)
2 jk02kT sinθ
+ k02
−k 2
0
− 2 jk02kT sinθ
(1 − k02T 2) + k02
)( )
Ax
Ay = 0
Az
(29)
The dispersion relation of the transversal wave is

2
(−k (1 − k02T 2) + k02)2 − 4k04k 2T 2(sin2θ 2
+ cos θ) = 0 ⇒ k = k± = ± k0 / (1 ± k0T ).
That is
/
k± = ± k0 (1 ± k0T ) = ±
ω
c (1 ± k0T )−1 (30)
So we have k± = ±
ω
c (1 /
± k0T )−1 = ω vF with vF = c(1 ± k0T )if we consider a bi dimensional gra‐
phene system.
The novel result here is that in our chiral theory we do not make c → vF but we obtain vF as
vF = c(1 − k0T ) ifk0T > 0 orvF = c(1 + k0T ) ifk0T < 0. These results are derived of the Chiral Elec‐
trodynamics with T as the chiral parameter andk0 = ω / c [5, 6, 7]. This situation corresponds
to a model of chiral electron where the E field is almost parallel to theH field, the movement
of electrons is helical in the valence band of the membrane graphene. Semi classically, for a
hydrogen atom, this condition was shown by Huang [36] for E wave perpendicular to H
wave with radiation and by Torres-Silva for E parallel toH , without radiation. See refs ([5,
37, 38]). Here, we extend this result to the electron bound to carbon atom.
In the next section we study the situation when the refractive index is negative.
3. Chiral waves in graphene acting as metamaterial media
Metamaterials are composite materials in which both permittivity and permeability possess
negative values at some frequencies has recently gained considerable attention [see e.g.,
[8-12]. This idea was originally initiated by Veselago in 1967, who theoretically studied
plane wave propagation in a material whose permittivity and permeability were assumed to
be simultaneously negative [11]. Recently Shelby, Smith, and Schultz constructed such a
composite medium for the microwave regime, and experimentally showed the presence of
anomalous refraction in this medium [10]. Previous theoretical study of electromagnetic
wave interaction with omega media using the circuit-model approach had also revealed the
possibility of having negative permittivity and permeability in omega media for certain
range of frequencies [9].
The anomalous refraction at the boundary of such a medium with a conventional medium,
and the fact that for a time-harmonic monochromatic plane wave the direction of the Poynt‐
ing vector is antiparallel with the direction of phase velocity, can lead to exciting features
that can be advantageous in design of novel devices and components. For instance, as a po‐
tential application of this material, compact cavity resonators in which a combination of a
slab of conventional material and a slab of metamaterial with negative permittivity and per‐
meability are possible. The problems of radiation, scattering, and guidance of electromag‐
netic waves in metamaterials with negative permittivity and permeability, and in media in
which the combined paired layers of such media together with the conventional media are
present, can possess very interesting features leading to various ideas for future potential
applications such as phase conjugators, unconventional guided-wave structures, compact
thin cavities, thin absorbing layers, high-impedance surfaces, to name a few. In this section,
we will first present a brief overview of electromagnetic properties of the media with nega‐
tive permittivity and permeability, and we will then discuss some ideas for potential appli‐
cations of these materials.
Such a medium is therefore termed left-handed medium [12]. In addition to this ‘‘left-hand‐
ed’’ characteristic, there are a number of other dramatically different propagation character‐
istics stemming from a simultaneous change of the signs of ε andμ, including reversal of
both the Doppler shift and the Cerenkov radiation, anomalous refraction, and even reversal
of radiation pressure to radiation tension. This configuration exhibit also chirality and a ro‐
tation of the polarization so the analysis of metamaterial presented by several authors pro‐
vides a good but not exact characterization of the metamaterial [9]. The evidence of chirality
behavior suggests that if it is included in the conditions to obtain a metamaterial behavior of
a medium futher progress will be obtained. In this short paper, we propose to investigate
the conditions to obtain a metamaterial having simultaneously negative ε and negative μ
and very low eddy current loss. As a initial point, we consider a media where the electric
polarization depends not only on the electric fieldE, and the magnetization depends not on‐
ly on the magnetic fieldH , and we may have, for example, constitutive relations given by
the Born-Federov formalism [12].
→ → →
D(r , ω) = ε(ω)(E(r , ω) + T (ω)∇ × E(r , ω)) (31)
→ → →
B(r , ω) = μ(ω)(H (r , ω) + T (ω)∇ × H (r , ω)) (32)
The pseudoscalar T represents the chirality of the material and it has length units. In the lim‐
itT → 0, the constitutive relations (31) and (32) for a standard linear isotropic lossless dielec‐
tric with permittivity ε and permeability μ are recovered.
According to Maxwell’s equations, electromagnetic waves propagating in a homogeneous

dielectric magnetic material are either positive or negative transverse circularly polarized
waves, and can be expressed as
→ ^ − j(k z−ω t )
E ±(r , t) = E 0(±)e ± 0 (33)
→ ^ − j(k z−ω t )
H ±(r , t) = H 0(±)e ± 0 (34)
→ → →
whereE0± = E0(x^ ± i ^y ), and∇ × E ±(r , t) = ∓ k±E ±, k± ≥ 0is the chiral wave number.
If the phase velocity and energy flow are in the same directions, and from Maxwell’s equa‐
→
tion, one can see that the electricE and magnetic field and H and the wave vector k will
form a right-handed triplet of vectors. This is the usual case for right-handed materials. In
→
contrast, if the phase velocity and energy flow are in opposite directions, andE, H , and k
will form a left-handed triplet of vectors. This is just the peculiar case for left handed materi‐
als where the effective permittivity ε(1 + T ∇ × ) and the effective permeability μ(1 + T ∇ × )
are simultaneously negative. So, for incident waves of a given frequencyω, we can deter‐
mine whether wave propagation in the composite is right handed or left handed through
the relative sign changes ofk.
To advance in our propose we considerer other more popular representation to describe a

chiral medium, [12] as
D = εP E + (χ + iκ)H , (35)
B = μP H + (χ − iκ)E, (36)
in which electromagnetic coupling terms are added to the basic terms. Bi-isotropy or biani‐
sotropy is used for calling such constitutive equations, according to the parameters to be
scalars or tensors. If κ = 0 andχ ≠ 0, it is the Tellegen medium; if κ ≠ 0andχ = 0, as the require‐
ment of reciprocity, it is the Pasteur medium:
D = εP E + iκH , (37)
B = μP H + iκE, (38)
There is a long dispute on strong chiral medium since it was introduced theoretically. Tradi‐
tional electromagnetic conclusions have limited us to understand strong chirality, i.e. k+T ≥ 1
orμP εP κ 2, until we see the fact that artificial Veselago’s medium [9] was successfully realiz‐
ed in certain frequency bands [3]. For the case, E ∥ H , we have to ask the following question:
can strong chiral medium exist?
In Ref. [14], the reason for traditional restriction of chirality parameters was concluded
as: 1) The wave vector of one eigenwave will be negative; 2) The requirement of a positive
definite matrix to keep positive energy:
( εP iκ
− iκ μP
) (39)
With the exploration of backward-wave medium, we know that negative wave vector, or
opposite phase and group velocities, are actually realizable. And there is an unfortunate
mathematical error in the second reason: in linear algebra, only if it is real and symmetric,
positive definite matrix is equivalent to that all eigenvalues should be positive. The matrix
(14) is a complex one, making the analysis on restriction of positive energy meaningless.
Actually, in a strong bi-isotropic medium with constitutive relations as Eqs. (1) and (2), the
energy can be drawn as
D⋅E B⋅H |E |2 |H |2
w = we + wm = + =ε +μ (40)
2 2 2 2
Mathematically, the amount of energy density propagated is proportional to the magnitude

→
of the Poynting vectorS , whereS = E × H . With the conditionE ∥ H , S(r, t) = 0so we find right
circularly polarized wave or left circularly polarized wave.
The concept of parallel fields is important in the theoretical formulation of: Space electro‐
magnetism and vacuum, the classical and quantum gravitational fields, the study of elemen‐
tary particles, operator and Dirac matrices, fields and chiral electrodynamics [15, 16]. If we
putE = αBand solving the Maxwell equations with the Born Fedorov relations we find
μ/2
α = i / με / 4 E = iηH = i H soε(ω), μ(ω)and T are transformed as:
ε/2
ε∥ =
ε
2
(
= εp 1 −
κ2
μp εp
) (41)
μ∥ =
μ
2
(
= μp 1 −
κ2
μp εp
) (42)
κ
T= (43)
ω(μP εP − κ 2)
In this case the total density energy is may be
w = we + wm = 0 (44)
In this special case where the energy propagated in one direction is equal to that propagated
in the opposite direction, there is no net energy flow in the medium and the sum of the two
TEM waves form what is generally known as a standing wave. The condition for a standing
wave is that the time average of Svanishes. This can be achieved if Sis zero all the time ev‐
erywhere in the region of space under consideration, i.e., S(r, t) = 0(see Fig. 4). Examination
of Eq. (39) shows thatS = 0 ifE ∥ H . In this last case, a particular solution of Eq. (31,3 2) is
whenk02T 2 = 1, where we have the conditionE ∥ H , andE = iηH , so we find the Beltrami force
free equation E + 2T ∇ × E = 0 and the vector Poynting S(r, t)vanishes [15-16].
In terms of Eqs. (41-42), ifμP εP < κ 2for whole frequency range, the energy will still keep positive
as long as the permittivity and permeability are positive, under the weak spatial dispersion
condition. Therefore, the real reason for traditional strong-chirality limitation is neither neg‐
ative wave vector nor energy conversation. Next we will point out two other important reasons.
First, with the assumption thatεP > 0, μP > 0, κ > 0andκ > μP εP , we easily show thatε∥, μ∥ and
T turn to negative from the transformation between Pasteur constitutive relations and BF re‐
lations shown in Eqs. (41)-(43). This is absolutely unacceptable before people realizing Vese‐
lago’s medium. Actually, strong chiral medium can be equivalent to Veselago’s medium for
the right circularly polarized wave [9-16]. The negativeε and μhave shown such a point.
Hence the negative sign in the BF model is not strange at all, since we realize effective dou‐
ble-negative with strong chirality parameter instead of simultaneously frequency resonan‐
ces. For a limiting case, the chiral nihility [10], in which μP → 0 and εP → 0 whileκ ≠ 0, the
parameters in DBF representation becomeε∥ → ∞,μ∥ → ∞ andT = − 1 / ωκ, remaining a finite
value after a simple mathematical analysis. There is no evidence that strong chirality cannot
exist in this aspect.
Second, it is the effectiveness of linear models. Similar to the case that linear optical and
electromagnetic models can no longer deal with very strong optical intensity and electro‐
magnetic field, we introduce nonlinear optics to take into account the higher order terms
of polarization.
If the spatial dispersion is strong enough, the higher order coupling terms cannot be neglect‐
ed as before. People used to mistake strong chirality with strong spatial dispersion, hence adding
a limitation to chirality parameter,κ μP εP . We believe that this is the most probable reason.
However, the strong spatial dispersion is embodied in the BF model, e.g. the value of T,
while the strong chirality is represented by the Pasteur model, e.g. the ratio of κto μP εP .
That is to say, strong chirality does not necessarily lead to strong spatial dispersion.
Based on Eqs. (41)-(43), we have computed T and ε∥ / εP or μ∥ / μP versusκ / μP εP , as shown
in Figs. 2 and 3. When κis very close to μP εP , the value of T is quite large, indicating a
strong spatial dispersion. Hence the singular point is the very point of traditional limitation.
However, withκcontinuously increasing, the spatial dispersion strength falls down very
quickly. Therefore, if κis not around μP εP , e.g. κ < 0.8 μP εP orκ > 1.2 μP εP , we need not take
nonlinear terms into consideration at all. Hence the strong spatial dispersion and nonlineari‐
ty cannot put the upper limitation to chirality parameters either.
Figure 3. The strength relationship of chirality and spatial dispersion. ωT / cversus κ / μP εP The point of κ / μP εP = 1 is
singularity, corresponding infinite spatial dispersion coefficient T. Whenκ / μP εP 1, T becomes negative for keeping the
positive rotation term coefficients with negative μ∥ andε∥. The 2-D grapheme system may be modeled when
ωT / c ∼ 0.9 − 1.1 and κ / μP εP ∼ 1
Figure 4. ε∥ / εp , μ∥ / μp versusκ / μP εP . With chirality strength increases, ε∥andμ∥ reduces quickly from εp and μp to − ∞.
The case 2-D graphene corresponds toκ / μP εP ∼ 1.
If ω | T | / c = 0.9927orω | T | / c = 1.0073, we have that the Fermi velocity is given by

vF = c(1 ± k0T ), so it appears a connection between the chiral electrodynamics and αthe fine
structure constant.
From Figure 4. we see that near of ε∥ / εp , μ∥ / μp ≈ 0the graphene have a linear behavior so the
Beltrami equation is
→ → →
∇ × E ±(r , t) = ∓ k±E ± (50)
Solving Beltrami's equation with boundary conditions, we can demonstrate the optical Zit‐
terbewegung effect by means of electromagnetic pulses propagating through a negative-zero-
positive index metamaterial (NZPIM) if we make
k± = ∓ k0 / (1 ± k0T ) = ∓ ω / vF with
vF = c(1 ± k0T ) (51)
Thus in optics, the Beltrami’s equation for electromagnetic waves can be reduced to the
Helmholtz equation.
In later sections we discuss some effects such as Zitterbewegung of optical pulses, diffusion
phenomenon and tunneling rate of dirac electron in graphene.
4. Two component equations and tunneling rate of dirac electron in

graphene
The usual choice of an orthogonal set of four plane-wave solutions of the free-particle Dirac
equation does not lend itself readily to direct and complete physical interpretation except in
low energy approximation. A different choice of solutions can be made which yields a direct
physical interpretation at all energies. Besides the separation of positive and negative energy
states there is a further separation of states for which the spin is respectively parallel or anti‐
parallel to the direction of the momentum vector. This can be obtained from the Maxwell’s
equation without charges and current in the E ∥ H configuration. Dirac's four-component
equation for the relativistic electron is [5-7]. (see our chapter “Chiral Transverse Electromag‐
netic standing waves with E ∥ H in the Dirac Equation and the spectra of the Hydrogen
Atom”, section 2,Intech 2011, Behaviour of electromagnetic waves in different media and
structures, edited by Ali Akdagli). Here we considerer a bidimensional graphene system so
the Dirac's four-component equation for the relativistic electron is:
∂ ^
iℏ ∂ t ψ = H D ψ (45)
where:
^ ^→
H D = vF (ᾱ ⋅ p ) + mvF 2β (46)
αk = ( )
0 σk
σk 0
, k = 1, 2, 3, (47)
β= ( )
I 0
0 −I
= σz (48)
and I is the two-by-two identity matrix and the Fermi velocity vF is deduced from the chiral
electrodynamics withvF = c(1 ± k0T ), where T is the chiral parameter in a metamaterial condi‐
tion. This result is capital to our approach because we fin a contact point between the gra‐
phene system and optical metamaterial makingvF = c(1 ± k0T ), no making c → vF as other
authors do it. In Figure 5 we have1.0073 > k0T > 0.9927.
The Hamiltonian commutes with the momentum vectorp ^ . In order to resolve this degenera‐
cy we seek a dynamical variable which commutes with both H andp ^ . Such a variable isσ^ ⋅ ^p ,
whereσ is the matrix Pauli. The eigenfunctions of the commuting variables p andσ^ ⋅ ^p are si‐
^
multaneous:
(σ^ ⋅ )2 = p 2 (49)
Thus for a simultaneous eigenstate of ^p andσ^ ⋅ ^p , the value of σ^ ⋅ ^p will be +p or –p, corre‐
sponding to states for which the spin is parallel or antiparallel, respectively, to the momen‐
tum vector like a graphene system.
A simultaneous eigenfunction of H and p will have the form of a plane wave
ψj = uj exp i (p ⋅ r − Et ) / h , j = 1, 2, 3, 4 (50)
where the ψj are the four components of the state function and uj four numbers to be deter‐
mined. Then E can have either of the two values.
E = ± ε = ± (m 2vF 4 + vF 2 p 2) 2
1
(51)
We now demand that ψj be also an eigenfunction of σ^ ⋅ belonging to one of the eigenvalues

pE , say, wherepE = ± p.
The eigenvalue equation is
σ^ ⋅ ψ = pE ψ (52)
Since W can be given either of the two values ±ε andpE , the two values± p, we have found
for given p four linearly independent plane wave solutions. It is easily verified that they are
mutually orthogonal.
Figure 5. Diagram where the graphene is located in the plane (x, y), the chiral photon is in either direction. The pontos
black figure on the left indicates the position of carbon atoms.
The physical interpretation of the solutions is now clear. Each solution represents a homoge‐
neous beam of particles of definite momentum p, of definite energy, either±ε, and with the
spin polarized either parallel or antiparallel to the direction of propagation. From here we
can obtain the well known equation for grapheneH 0 = − iℏvF σ ⋅ ∇ .
In this section, we study the tunneling rate of Dirac electrons in graphene through a barrier
with an intense electromagnetic field. A one transport phenomenon in graphene is the chi‐
ral tunneling [1, 2, 17, 18]. In mono layer graphene a perfect transmission through a poten‐
tial barrier in the normal direction is expected. This tunneling effect is due to the chirality of
the Dirac electrons, which prevents backscattering in general. This kind of reflectionless
transmission is independent of the strength of the potential, which limits the development of
graphene-based field-effect transistors (FET). The perfect transmission can be suppressed
effectively when the chiral symmetry of the Dirac electrons is broken by a laser field, when
the n-p junctions in graphene are irradiated by an electromagnetic field in the resonant condition
[19,20].We consider a rectangular potential barrier with heightH 0, width D in the X direc‐
tion, and infinite length in the Y direction. The Fermi level lies in the valence band in the
barrier region and in the conduction band outside the barrier. The gray filled areas indicate
the occupied states. The optical field under chiral condition is propagated perpendicular to
the layer surface and it is circularly polarized along the Z,Y direction with a detuningΔ0 = 2Eb − ℏω.
We choose Δ0 > 0to ensure that there is no inter-band absorption inside the barrier. Mean‐
while, 2Ek ≫ ℏωis used to guarantee that the influence of the optical field outside the barri‐
er can be neglected. Thus, neglecting the scattering between different valleys, the scattering
process of Dirac electrons in K point is described by the time-dependent Dirac equation, Eq.
(45) withm = 0, so ψ = ψg = (ψE , ψH )t where t means transposed wave function.
∂
iℏ ∂ t ψg (r, t) = H e + H 0I + H int ψg (r, t) (53)
where ψg (r, t) = ϕA(r, t), ϕB (r, t) t is the wave function, H e = vF σ^ ⋅ p = − iℏvF σ^ ⋅ ∇ is the unper‐
turbed Dirac Hamiltonian obtained from the chiral electrodynamics [17], σ^ = (σ σ )are the x y
Pauli matrices, vF ≈ 106ms −1is the Fermi velocity, H 0(r)is the height of the potential barrier,
I is the unit matrix, and H intis the interaction Hamiltonian with a chiral electromagnetic
potential.
H int = − ℏevF σ^ ⋅ A = ℏ ( 0
H 21
H 12
0
) (54)
where e is the electron charge and the chiral potential vector is A = Ax e iωt , i Ay e iωt with
i = − 1 and | Ax | = | Ay | so we have circular polarized wave of the electromagnetic field.
When the Dirac electrons perpendicularly collide with the barrier perpendicularly, we can
rewrite Eq. (53) as a set of partial differential equations
∂ ∂
iℏ ∂ t ϕA(x, t) = − ivF ∂x ϕB (x, t) + V 0ϕA(x, t) + H 12(t)ϕB (x, t) (55)
∂ ∂
iℏ ∂ t ϕB (x, t) = − ivF ∂x ϕA(x, t) + V 0ϕB (x, t) + H 21(t)ϕA(x, t) (56)
Since the tunneling time is of order of sub-picosecond and the potential H 12(t)and H 21(t)
vary as fast as the frequency of incident light beams, this scattering process is strongly
time-dependent.
In order to study such a strongly time-dependent scattering process, we employ the finite-
difference time-domain (FDTD) method to solve Eq. (47) and Eq. (48) numerically in the
time-domain [21]. In the traditional FDTD method, the Maxwell’s equations are discretized
by using central-difference approximations of the space and time partial derivatives. As a
time-domain technique, the FDTD method can demonstrate the propagation of electromag‐
netic fields through a model in real time. Similar to the discretization of Maxwell’s equations
in FDTD, we denote a grid point of the space and time as (i, k) = (iΔx, kΔt) [22]. For the no‐
dal variables we can apply the usual Lee discretization method in 2-D systems withk02T 2 ∼ 1
(See Figure 6).
→ → →
Figure 6. Lee discretization. Here H x = (∇ × A) x Ex = − (∂ (1 + T ∇ × )A / ∂ t) x and withJ = 0 (See [21, 22, 24]).
For any function of space and timeG(iΔx, kΔt) = G k (i), the first order in time or space partial
differential can be expressed as:
∂ G k (i + 1 / 2) − G k (i − 1 / 2) ∂ G (k +1/2(i) − G (k −1/2(i)
( ∂ x G(x, t)) x=iΔx ≈ Δx , ....., ( ∂ t G(x, t))t =kΔt ≈ Δt
These eqs. can be replaced by a finite set of finite differential equations like:
1 H 0(i) 1 H 0(i)
ϕAk +1/2(i) Δt − 2i = Δt + 2i ϕAk −1/2(i)
−
vF
Δx −
H 12k (i + 1 / 2)
2i ϕB k (i + 1 2) / (57)
(i − 1 / 2)
vF H 12k (i − 1 / 2)
+ Δx + 2i ϕB k
For computational stability, the space increment and the time increment need to satisfy the
relation Δx > vF Δt [21]. Furthermore, the space increment Δx must far smaller than the
wavelength of electronsΔx < λe / 10, and the time increment must be far smaller than the peri‐
od of the electromagnetic fieldT l .
At the boundary, one-dimensional Mur absorbing boundary conditions are used [21-22]. To
compare our results with [25] which use linear polarization for the vector potentialAwhere
the chiral parameter T is zero, we considerer at the input boundary, a normalized Gaussian
electronic wave packet, wheretg and τg denote the peak position and the pulse width, respec‐
tively. Thus, by solving Eq. (57) directly in the time domain we can demonstrate the propa‐
gation of a wave packet through a barrier in real time.
Numerical simulations are shown in Fig. 7. The following parameters are used in our calcula‐
tion: the peak position t0 = 1.5 ps, the pulse width τg = 1.0 ps, the space increment Δx = 0.1 nm,
the time incrementsΔt = 5 × 10−5 ps, and the height of the potential barrier H 0 = 500 meV.
Figure 7. a) Numerical simulations of a wave packet given byN = | A | 2 + | B | 2, tunneling through a barrier without
pump beams. Figures (b)-(d) show the time sequence of a wave packet tunneling through a barrier with pump intensi‐
tyI ω = 3MW / cm 2, Δ0= 5meV, and D = 350 nm. The light grey shows the barrier area.
When there is no pump beams, a perfect chiral tunneling can be found [see Fig. 7 (a)]. This
result is consistent with that of Ling et al. [25]. But when the sample is irradiated by an in‐
tense non resonant laser beam, a reflected wave packet appears [see Fig. 7 (d)]. The perfect
transmission is suppressed. By analyzing the transmitted wave packet and the reflected
wave packet, we can obtain the tunneling rate.
To explain the suppression of chiral tunneling, we first investigate the chiral potential wave
in the barrier within a rotating wave approximation [23, 24]. Figure 7 (a) shows the renor‐
malized band as a function of momentum k with intensityI ω = 3MW / cm 2.
Here, the important point is that we make±k0T ∼ 1 + α ≈ 1.0073, where α = e 2 / (4πε0ℏc)is the
fine structure constant, so we verify that our chiral theory is correct. Without external
beams, (fig. 7 (a)), we verify the known results on chiral tunneling [11]. Figures 7 (b)-(d)
show the time sequence of a wave packet tunneling through a barrier with pump intensity
I ω = 3MW / cm 2, Δ0= 5meV, and D = 350 nm.
Figure 8. Left: The reflectance R(circles) and the transmittance T (squares) of the barrier as a function of the detuning
for I ω = 3MW / cm 2 and D = 350 nm. Right: The transmittance versusI ω (MW / cm 2), having D as parameter. Here,
k02T 2 ≈ 1 , that is the wave electric field is almost parallel to the wave magnetic field within the graphene device.
Under intense light beams, the dressed states are strongly mixed with valence states and
conduction states. Therefore, the chiral symmetry of Dirac electrons in graphene can be bro‐
ken and perfect chiral tunneling is strongly suppressed. Numerical results are shown in Fig.
8 (left) with pump intensity I ω = 3MW / cm 2 and D = 350 nm. From Fig. 8(left) we can find
that the transmission is strongly suppressed, even with laser detuning (e.g., Δ0= 10 meV, the
transmittance is about 0.03).
Figure 8 (left), show that the reflectance decreases, and the transmittance increases as Δ0 in‐
creases. The strong laser field can enhance band mixing and reduce the transmittance. If D
increases we can see that the wide barrier can prolong the interaction time between elec‐
trons and photons, reduce the tunneling rate, and lower the threshold of the pump laser
power (Figure 8 right). Our results can be compared with [25] which use linear polarization
ofH int, however we think that inside of a plate of graphene, the eigenvectors of A have to be
characterized as chiral waves, because the electrons have a chiral nature in a graphene de‐
vice. Is for this reason that our numerical results correspond to half times the results shown
in [25].
5. Zitterbewegung of optical pulses near the Dirac point inside a negative-

zero-positive index metamaterial
This optical analog of the Zitterbewegung effect is similar to two-dimensional (2DPC) pho‐
tonic crystal.
The Zitterbewegung effect can appear in solids, because the electron at the touching point be‐
tween two interacting bands in a solid also obeys the massless Dirac equation [26]. In recent
years, a growing attention has been devoted to the simulations or demonstrations of the Zit‐
terbewegung effects in controllable physical systems, such as, ultra cold atoms, superconduc‐
tors, semiconductor nanostructures with spin-orbit coupling, a single trapped ion, and
graphene.
For a homogenous medium, when a light field is polarized in the z direction, the Beltrami’s
→ → →
equation∇ × E ±(r , t) = ∓ k±E ±can be transformed as a Helmholtz equation written as
∂ ∂ ∂ ∂
( ∂ x + i ∂ y )( ∂ x − i ∂ y )Ez (x, y, ω) = (ik(ω))2Ez (x, y, ω) (58)
with a wave number k(ω) This equation can be written in the form of the Dirac equation
( ∂
− i( ∂ x
0
+i
∂
∂y )
∂
0
∂
)(
− i( ∂ x − i ∂ y ) E z1(x, y, ω)
) (
E z2(x, y, ω)
= k(ω)
E z1(x, y, ω)
E z2(x, y, ω)
) (59)
In the case of grapheme near ofε∥ / εp , μ∥ / μp ≈ 0, there are two bands touch each other form‐
ing a double-cone structure with a linear dispersion, which can also be called Dirac Point.
Near the Dirac point the light transport obeys the massless Dirac equation if we make
vF = c(1 ± k0T ) = vD . Also naturally, it would be of great interest to find out what is the condi‐
tion to have the Dirac dispersion for the light field in a homogenous medium, the wave vec‐
tor of a medium, k(ω), can be expanded as k(ω) = k(ωD ) + (ω − ωD ) / vD + .........so ifk(ωD ) = 0,
then we have a linear dispersionk(ω) = (ω − ωD ) / vD , and the Dirac equation is
( ∂
− ivD ( ∂ x
0
+i
∂
∂y )
∂
0
∂
)(
− ivD ( ∂ x − i ∂ y ) E z1(x, y, ω)
E z2(x, y, ω)
) = (ω − ωD ) ( )
E z1(x, y, ω)
E z2(x, y, ω)
(60)
whereE z1(x, y, ω), E z2(x, y, ω)are two eigenfunctions of the electrical fields (polarized in
the z-direction) corresponding to the same frequencyω, and vD is the group velocity at the
Dirac point (ωD ). For media satisfying Eq. (58) withωD > 0, the wave number, k(ω), varies
from negative to zero and then to positive with the frequency, and so does the refractive in‐
dex of the medium with zero refractive index atωD .
Note that the positive and negative branches of the band structure coexist. Equation (58) is
the massless Dirac equation of the light fields in homogenous materials, which is the same
as that of electrons in graphene [27]. Therefore, for the NZPI metamaterial [satisfying Eq.
(58)], we will have the Dirac point with a double-cone structure for the light field at frequen‐
cyωD . It is expected that the propagation of the light field is analogous to that of the electron
in graphene. In fact, in the 2DPC [28], the effective refractive index varies from negative to
positive near this Dirac point. AtωD , Eq. (45) takes the form of diffusion equation,
∂ ∂ ∂ ∂
− ivD ( ∂ x − i ∂ y )E z1(x, y, ω) = 0, − ivD ( ∂ x + i ∂ y )E z2(x, y, ω) = 0 (60)
which are the same as that of the massless Dirac equation at zero energy. [29]. Therefore, it
can be predicted that the behavior of light fields at ωD has the diffusive properties inside the
medium of Eq. (2), like electrons at Dirac point of graphene [27-29]. At or nearωD , because of
k 2 = kx2 + ky2 ≈ 0, theky component becomes a pure imaginary number for any realkx , thus the
fields along they direction between a given interval L have the total energy transmittance
/
∞
∫
T all = (exp( − | kx | )2dkx = 1 L
−∞
(61)
which tells us that the propagation of light field at (or near) ωD exhibits the L / 1 scaling, a
main characteristic of the diffusion phenomenon. This effect can be obtained if
ε∥ / εp , μ∥ / μp → 0with low damping rates.
Here, the Dirac point with the double-cone structure for the light field can also be realized in
a homogenous negative-zero-positive index (NZPI) medium [26], in which the two-dimen‐
sional Helmholtz equation could be written as the two-dimensional massless Dirac equa‐
tion. The condition for the realization of the Dirac point in the homogenous optical medium
is the index varying from negative to zero and then to positive with frequency, and the light
field also obeys eq. (45) near the Dirac point.. Thus, in this work, we show that the Zitterbe‐
wegung effect with optical pulses appears near the Dirac point in NZPIM slabs. With the re‐
alization of the NZPIMs in experiments, we believe that it already has a great possibility to
observe Zitterbewegung effect with optical pulses in the GHz region.
6. Theoretical description of minimal conductivity in graphene under

chiral approach
The first results on the theoretical combination of graphene electron transport properties
and a characteristic property of Dirac chiral fermions were obtained by Katsnelson et al, [2].
The intrinsic nature of Dirac fermions gives rise to minimal conductivity even for an ideal
crystal, that is, without any scattering processes. The simplest way for the theoretical con‐
sideration is the Landauer approach [29]. Assuming that the sample is a ring of length L y
in y direction; author used the Landauer formula to calculate the conductance in the x
direction (Fig. 9).
Figure 9. Geometry of the sample. The direction of the current is parallel to the x axis [27].
The convenient boundary conditions are not physical, but to get finite transparency one
should chooseL x L y .
In the coordinate representation the Dirac equation (59) at zero energy takes the form
(Ψi = vD E zi )
∂ ∂ ∂ ∂
− i( ∂ x − i ∂ y )vD E z1(x, y, ω) = − i( ∂ x − i ∂ y )Ψ1 = 0 (62)
∂ ∂ ∂ ∂
− i( ∂ x + i ∂ y )vD E z2(x, y, ω) = − i( ∂ x + i ∂ y )Ψ2 = 0 (63)
The solutions of these equations areΨ1(x − iy),Ψ2(x + iy) , which are complex conjugated func‐
tions which are general solutions.
Due to periodicity in y direction both wave functions should be proportional toexp(iky y)

whereky = 2πn / L y ,, n = 0,±1,±2,... This means that the dependence on the x is also fixed:
the wave functions are proportional toexp(2πnx / L y ). The introduced boundary conditions
at the sample edges are x = 0 x = 0 andx = L x . The assumption is that the sample is doped
graphene with the Fermi energyEF = vkF = − V 0 . The wave functions in the sample are sup‐
posed to have the same y-dependence, that is,Ψ1,2(x, y) = Ψ1,2(x)exp(iky y) .
Requiring continuity of the each wave function at the edges of sample, one can find the
transmission coefficient:
cos2ϕ
Tn = cosh2(ky L x ).sin2ϕ
(64)
Wheresinϕ = ky / kF ,
kx = k 2 − ky2 (74)
Further, one should assume thatkF L x ≫ 1 and put ϕ = 0in equation. Thus, the trace of the
transparency which is just the conductance per valley per spin is
e2 2 ∞ 1 e2 L y
g= h tr( ∑ T n ) = eh ( ∑ )= h πL x (65)
n=−∞ cosh2(ky L x≈)
The conductance then equalsσL y / L x and the conductivity ise 2 / πh . Experimentally, it is

close to e 2/h [31], that is in πtimes larger than present estimation.
Zitterbewegung − circular motion of elementary particles caused by an interference between

positive and negative energy states − leads to the fluctuation of the position of an electron.
This relativistic “jittering” of an electron in graphene could be interpreted in terms of classi‐
cal physics as an interaction of electron with some potential caused by the presence of disor‐
der in crystal. Therefore, the Zitterbewegung plays a role of “intrinsic” disorder in the
system which appears in the presence of minimal conductivity of the ideal crystal (without
scattering) even at zero temperatures.
Under this chiral approach, we show that the optical transparency of suspended graphene can
be defined by the fine structure constant, α = e 2 / ℏc, the parameter that describes coupling
between light and relativistic electrons and is traditionally associated with quantum electro‐
dynamics rather than condensed matter physics. Despite being only one atom thick, gra‐
phene is found to absorb a significant (α = 0.78%toπα = 2.3%) fraction of incident white light,
which is a consequence of graphene’s unique electronic structure. This value translates direct‐
ly into universal dynamic conductivity G = e 2 / 4πℏwithin a few % accuracy.
There is a small group of phenomena in condensed matter physics, which are defined only
by the fundamental constants and do not depend on material parameters. Examples are the
resistivity quantum h / e 2that appears in a variety of transport experiments, including the
quantum Hall effect and universal conductance fluctuations, and the magnetic flux quan‐
tum h / e playing an important role in the physics of superconductivity (here h is the Planck
constant and e the elementary charge). By and large, it requires sophisticated facilities and
special measurement conditions to observe any of these phenomena. Here, we show that
such a simple observable as the visible transparency of graphene [1] is defined by the fine
structure constant,α. Our results are in agreement with the theory of ideal two-dimensional
(2D) Dirac fermions [2,3] and its recent extension into visible optics [4], which takes into ac‐
count the triangular warping and nonlinearity of graphene’s electronic spectrum, but our re‐
sult is π −1times the result of [2,3].
Optical properties of thin films are commonly described in terms of dynamic or optical con‐
ductivity G. For a 2D Dirac spectrum with a conical dispersion relation ε = ℏvF | k | (kthe
wavevector). G was theoretically predicted [32, 33] to exhibit a universal valueG = e 2 / 4ℏ, our
result is G0 = e 2 / 4πℏif the photon energy E is much larger than both temperature and Fermi
energyεF . Both conditions are stringently satisfied in our visible-optics experiments. The
universal value of G also implies that all optical properties of graphene (its transmittance T,
absorption Aabs and reflection R) can be expressed through fundamental constants only (T,
Aabs and R are unequivocally related to G in the 2D case). In particular, it was noted by Kuz‐
menko et al [34] that T trasp = (1 + 2πG0 / c)−2 ≈ (1 + 0.5πα) ≈ 1 − παfor the normal light incidence.
Our result is given byT trasp ≈ 1 − α. We emphasize that – unlike G – both T and R are observa‐
ble quantities that can be measured directly by using graphene membranes.
7. Absorption of light by quasi 2D Dirac fermions
Here, following [35], we show how the universal value of graphene’s opacity can be under‐
stood qualitatively, without calculating its dynamic conductivity. Let a light wave with elec‐
tric field E and frequency ω fall perpendicular to a graphene sheet of a unit area. The
incident energy flux is given by W i = c / 4π( | E | 2)
Taking into account the momentum conservationk for the initial | i and final | f states, on‐
ly the excitation processes pictured in Fig. 10 contribute to the light absorption. The absor‐
bed energy W a = nℏω is given by the number n of such absorption events per unit time and
can be calculated by using Fermi’s golden rule as n = (2π / ƛ)( | L | 2)D where L is the matrix
element for the interaction between light and Dirac fermions, and D is the density of states
at ε = E / 2 = ℏω / 2 (see Fig. A1). For quasi 2D Dirac fermions, D(ℏω / 2) = ℏω / πℏ2vF2and is a
linear function of ε.
The interaction between light and Dirac fermions is generally described by the Hamiltonian
H = vF σ^ ⋅ ^p = vF σ^ ⋅ (p
^ − e cA) = H + H
/ 0 A
where the first term is the standard Hamiltonian for 2D Dirac quasiparticles in graphene [1]
and
H A = − vF σ^ ⋅ (e / c)A = vF σ^ ⋅ (e / iω)Edescribes their interaction with electromagnetic field.
Here A = (ic / ω)E is the vector potential and σ the standard Pauli matrices. Averaging over
all initial and final states and taking into account the valley degeneracy, our calculations yield
L 2 = | f | v σ^ ⋅ (e iω)E | i | 2 = (1 8)e 2v 2E 2 ω 2
F / / F /
This results in W a = (e 2 / 4ℏ) | E | 2and, consequently, absorption Aabs = W a / W i = πe 2 / cℏ) = πα

both of which are independent of the material parameter vF that cancels out in the calcula‐
tions ofW a. Also note that the dynamic conductivity G = W a / | E | 2is equal toe 2 / 4ℏ. Our re‐
sult isAabs ≈ W a / W i = πe 2 / πcℏ) = α in correspondence with vF = c(1 − k0T ) Because graphene
practically does not reflect light (R<<1 as discussed above), its opacity (1 – T) is dominated
by the derived expression forAabs .
In the case of a zero-gap semiconductor with a parabolic spectrum (e.g., bilayer graphene at
low ε), the same analysis based on Fermi’s golden rule yieldsAabs = 2πα. This shows that the
fact that the optical properties of graphene are defined by the fundamental constants is re‐
lated to its 2D nature and zero energy gap and does not directly involve the chiral properties
of Dirac fermions.
On a more general note, graphene’s Hamiltonian H has the same structure as for relativistic
electrons (except for coefficient vF instead of the speed of light c). The interaction of light
with relativistic particles is described by a coupling constant, a.k.a. the fine structure con‐
stant. The Fermi velocity is only a prefactor for both HamiltoniansH 0 andH A and, according‐
ly, one can expect that the coefficient may not change the strength of the interaction, as
indeed our calculations show.
Figure 10. Excitation processes responsible for absorption of light in graphene.
Thus, we have found that the visible opacity of suspended graphene is given by π α within
a few percent accuracy and increases proportionally to the number of layers N for few-layer
graphene. Its dynamic conductivity at visible frequencies is remarkably close to the univer‐
sal value of(e 2 / 4ℏ)N . The agreement between the experiment and theory is particularly
striking because it was believed the universality could hold only for low energies (< 1eV)
beyond which the electronic spectrum of graphene becomes strongly warped and nonlinear
and the approximation of Dirac fermions breaks down.
Electrons from the valance band (bottom) are excited into empty states in the conduction
band with conserving their momentum and gaining the energy E= h ω.
The approximation of 2D Dirac fermions is valid for graphene only close to the Dirac point
and, for higher energies ε, one has to take into account such effects in graphene’s band struc‐
ture as triangular warping and nonlinearity.
Most theories suggest4e 2 / πh , which is about π times smaller than the typical values ob‐
served experimentally. The reason of disagreement is still an open question − “mystery of a
missing pie”. A value of conductivity ∼ e 2 / h at the Dirac point can emerge in case of pres‐
ervation of chiral symmetry by disorder or when the dominant disorder does not scatter
electrons between two valleys. Furthermore, macroscopic inhomogeneity (on the scale larg‐
er than the mean free path) is also important for measurements ofσmin.
8. Conclusions
In this chapter we presented a short review on the chiral propagation of electron waves in
monolayer graphene and optical simulation with optical field in the negative-zero-positive
index metamaterial NZPIM and its close connection. Section II presented an enhanced vec‐
tor diagram of Maxwell’s equations for chiral media with quasi parallel electromagnetic
fields,E ∥ H . Chiral waves in graphene acting as metamaterial media were discussed in
section III. In section IV, two equations and tunneling rate of Dirac electron in graphene
were derived. In Section V, Zitterbewegung of optical pulses near the Dirac point inside a
negative-zero-positive index metamaterial, showed that the chiral field near the Dirac point
becomes a diffusive wave. The last sections described the theoretical description of mini‐
mal conductivity in graphene under chiral approach and absorption of light by quasi 2D
Dirac fermions.
We reviewed the formulation of graphene’s massless Dirac Hamiltonian, under the chiral
electromagnetism approach, like a metamaterial medium, hopefully demystifying the mate‐
rial’s unusual chiral, relativistic, effective theory. The novel result here was that in our theo‐
ry we did not make c → vF [1, 2, 27-35], but we obtained vF as vF = c(1 − k0T ) ifk0T > 0 or
vF = c(1 + k0T ) ifk0T < 0. These results were derived of the Chiral Electrodynamics with T as
the chiral parameter andk0 = ω / c [5, 6, 7]. This situation corresponds to chiral electrons
where the Efield is almost parallel to theH field, and the movement of electron is helical in
the valence band of the membrane graphene. Semi classically, for an electron in hydrogen
atom, this condition was shown by Huang (2006) for E wave perpendicular to H wave with
radiation (E = H ) and by Torres-Silva for E parallel toH , without radiation (E = iH ), see refs
([5, 36, 37]). Here, we have extended this result to the electron bound to a carbon atom.
There is a small group of phenomena in condensed matter physics, which are defined only
by the fundamental constants and do not depend on material parameters. Examples are the
resistivity quantum h / e 2that appears in a variety of transport experiments, including the

quantum Hall effect and universal conductance fluctuations, and the magnetic flux quan‐
tum h / e playing an important role in the physics of superconductivity. In this chapter we
founded that under this chiral approach, the optical transparency of suspended graphene
can be defined by the fine structure constant,α = e 2 / ℏc , that isT trasp ≈ 1 − α, and
vF / c = α = (1 − k0T ).
Also we give a support to the similitude of the band structure of a macroscopic photonic
crystal with the electronic band structure of graphene, which is experimentally much more
difficult to access, allows the experimental study of various relativistic phenomena. With
our analytical and numerical analysis we hope to verify that, similar behaviors exist to elec‐
trons in graphene treated as mass-less particle, ie, electron wave discovered only 8 years ago
graphene is already one of the most studied carbon allotropes. But this material still poses a
lot of theoretical and experimental questions, which have to be answered.
Acknowledgements
I wish to thank to colleague Jorge Benavides Silva of the EIEE for many useful discussions
on particles and photons
Author details
H. Torres-Silva*
Address all correspondence to: [email protected]
Escuela de Ingeniería Eléctrica y Electrónica Universidad de Tarapacá- Arica, Chile
References
[1] Novoselov, S., et al. (2005). Nature, 438, 197-200.

[2] Katsnelson, M. I., et al. (2006). Nature Physics, 2, 620-624.
[3] Wilton, D. (1995). IEEE Antennas and Propagation Magazine., 37, 7-11.
[4] Uckun, S. (1998). Electrotechnical Conference. Melecon 98, 283-286.
[5] Torres-Silva, H. (2011). Chiral Transverse Electromagnetic standing waves with E‖H
in the Dirac Equation and the spectra of the Hydrogen Atom. ”, section 2. Intech, vol
Behavior of electromagnetic waves in different media and structures, edited by Ali
Akdagli
[6] Torres-Silva, H. (2000). Internal publication on Chiral Electrodynamics. (in spanish),

PCI:00.1036/97 UTA-CHILE
[7] Torres-Silva, H. (2008). Ingeniare, 16(1), pp. 43-47, 48-52.
[8] Shelby, R. A., et al. (2001). Science [5514], 292, 77-79.
[9] Veselago, V. G. (1968). Soviet Physics Uspekhi, 10(4), 509-514.
[10] Smith, D. R., et al. (2000). Phys. Rev. Lett, 84(18), 4184-4187.
[11] Monzon, C., & Forester, D.W. (2005). Phys. Rev. Lett, 95, 123904.
[12] Torres-Silva, H. (2008). Ingeniare, 16, 119-122.
[13] Lindell, I., Sihvola, A. H., Tretyakov, S. A., & Viitanen, y. A. J. (1994). Electromagnet‐
ic Waves in Chiral and Bi-Isotropic Media. Artech House, Boston.
[14] Plum, E., Zhou, J., Dong, J., Fedotov, V. A., Koschny, T., Soukolis, C. M., & Zheledev,
y. N. I. (2009). Physical Review B., 79(3), 0354071-0354076.
[15] Tretyakov, S., Nefedov, I., Sihvola, A., Maslovski, S., & Simovski, C. (2003). J. Electro‐
magn. Waves Appl, 17, 695.
[16] Zhou, J., Dong, J., Wang, B., Koschny, T., Kafesaki, M., & Soukoulis, C. (2009). Phys.
Rev. B, 79, 121104.
[17] Hannes, W. R., & Titov, M. (2010). Europhysics Letters, 89, 47007.
[18] Abanin, D. A., & Levitov, L. S. (2007). Science, 317, 641.
[19] Elias, D. C., et al. (2009). Science, 323, 610.
[20] Fistul, M. V., & Efetov, K. B. (2007). Phys. Rev. Lett ., 98, 256803.
[21] Yee, K. S. (1966). IEEE Transactions on Antennas and Propagation, ., 14, 302.
[22] Mur, G. (1981). IEEE Transactions on Electromagnetic Compatibility, EMC-23, 377.
[23] Fistul, M. V., & Efetov, K. B. (2007). Phys. Rev. Lett, 98, 256803.
[25] Ling, Jiang-Tao, Su, Fu-Hai, Huang, Hai, & Xin, Hai. (2007). Arxiv:1007.3430 , 3
[26] Wang, Li-Gang, et al. (2009). Opt. Lett, 34, 1510-1512.
[27] Nomura, K., & Mac Donald, A. H. (2007). Phys. Rev. Lett, 98, 076602.
[28] Notomi, M. (2000). Phys. Rev. B, 62, 10696.
[29] Katsnelson, M. I. (2006). Eur. Phys. J. B, 51, 157.
[30] Zhang, F., et al. (2008). J. Appl. Phys, 103, 084312.
[31] Novoselov, K. S., et al. (2005). Nature, 438, 197-200.

[32] Gusynin, V. P., et al. (2006). Phys. Rev. Lett., 96, 256802.
[33] Falkovsky, L. A., et al. (2007). Phys. Rev. B, 76, 153410.
[34] Kuzmenko, A. B., et al. (2008). Phys. Rev. Lett., 100, 117401.
[35] Nair, R., et al. (2008). Science, 320, 1308.
[36] Huang, X. Q. (2006). arxiv:physics/0601169.

Section 2
Numerical and Analytical Solutions

Chapter 3
Numerical Modeling of Electromagnetic Wave

Propagation Through Bi-Isotropic Materials
I. Barba, A. Grande, A.C.L. Cabeceira, A. Gómez,

J.A. Pereda and J. Represa
1. Introduction
As part of the present wave of interest in electromagnetic propagation in complex materials

(left-handed, chiral, nonreciprocal, nonlinear, etc.), a special mention may be made to bi-iso‐
tropic materials. After a period of relevance during the nineties [1-4], nowadays they enjoy a
renewed interest, because of new fabrication techniques that allow the use of mature tech‐
nologies (printed or integrated circuit) [5,6], and also the discovery of new properties associ‐
ated to the bi-isotropy, such us the presence of negative refraction index [6-11]. In order to
study electromagnetic wave propagation in such materials, different traditional numerical
methods, both in time and frequency domain, have been extended so that they may deal
with additional constitutive parameters, as well as their inherently dispersive. This last
point is especially important when dealing with time-domain methods, in which the fre‐
quency dispersion must be translated to time-domain.
In this work, we want to present a review of different techniques developed during the
last years:
• Time domain techniques: Finite Differences in the Time Domain (FDTD), Transmission
Line Matrix (TLM) and Multiresolution in Time Domain (MRTD). All of them have been
modified to treat with bi-isotropic media, and with several methods to include frequen‐
cy dispersion.
• In the frequency domain, the extension of the coupled mode method to cover the electro‐
magnetic wave propagation in closed structures containing these new media.
2. Bi-Isotropic materials. Constitutive Relations
In a general bi-isotropic material, the electric and magnetic fields are coupled. There are dif‐
ferent ways of formulating its constitutive relations in frequency domain [1]. Among them,
we use:
→ → →
D = εE + ξH
→ → → (1)
B = ζE + μH
→ → → →
In which we assume the medium is linear.D , E , B , and H are the electromagnetic field vec‐
tors.  and  are the permittivity and permeability,  and  would be the cross-coupling
parameters: the four parameters may depend on frequency (dispersive behavior). In a bi-
anisotropic medium, they have tensor form, and may contain up to 36 scalar material pa‐
rameters.  and  are usually expressed in the following way:
χ + jκ
ζ=
c
(2)
χ − jκ
ξ=
c
c being the light speed in vacuum. The parameter is known as “Pasteur” parameter, while
 is the “Tellegen” parameter; thanks to the introduction of c in the equation, both parame‐
ters are dimensionless; at the same time, in case of non-dispersive behavior, they are also
real [1]. Substituting these values into Eq. 2, it leads to:
→ → χ − jκ →
D = εE + H
c
(3)
→ → χ + jκ →
B = μH + E
c
Using these relations, instead of Eq. 1, the constitutive parameters have a more clear physi‐
cal meaning: the Pasteur parameter describes the “chirality” of the medium, i.e. if the medi‐
um is “non chiral” (that is, it is equivalent to its mirror image), then = 0. In other case,
the medium is said to be “chiral”. The Tellegen parameter represents the nonreciprocity of
the medium. A medium may be nonreciprocal due to other reasons (for example, a magne‐
tized ferrite), but only a cross coupling like the one described in Eq. 3 may be nonreciprocal
and isotropic (that is “bi-isotropic”) at the same time [2].
While reciprocal chiral materials have been extensively researched during the last years, and
different geometries and fabrication techniques have been designed [5,6], nonreciprocal ma‐
terials have stayed mainly in the plane of theoretical discussion. In [12,13], Lakhtakia and
Weiglhofer affirmed, using a covariance requirement deduced by Post (the so-called “Post
constraint” [14]), that a bi-isotropic medium must be reciprocal, so the Tellegen parameter
Numerical Modeling of Electromagnetic Wave Propagation Through Bi-Isotropic Materials 61
should always be 0. This opened a discussion on the theoretical possibility of the existence
of such materials that lasted more than a decade, mainly between Lakhtakia and Weiglhofer
[12,13,15-18] and Lindell, Sihvola, Tretyakov et al. [3,4,19-23]. Finally, Tretyakov et al. [24]
built a “Tellegen particle” and verified its non-reciprocal behavior. At the same time, it has
been found that some natural media (Fe2TeO6 and Cr2O3) though anisotropic, violate the
Post constraint [25], and it may be demonstrated that only in uniform and unbounded me‐
dia it may be stated that the Tellegen parameter vanishes [26-28], so “the Post constraint as a
general dogma should be buried with all due honors” [29]. Other authors, nevertheless, con‐
sider the question still open [30].
An alternative formulation of Eq. 3 was proposed by Kong [31]:

→
→ → 1 →
( dH
D = εE + c χH − κ dt )
→ (4)
→ → 1 →
(
B = μH + χE − κ
c
dE
dt
)
Note that this equation is formulated in time domain, instead of frequency domain, and that
the chirality parameter κ is not equivalent to the one present in Eq. 3. In addition, this equa‐
tion does not take into account the intrinsically dispersive behavior of this class of materials
[1,6,9-11]. To deal with that dispersion, already in 1937, E.U. Condon [32] proposed a sec‐
ond-order resonant model for optical chiral materials. Though in 1999 Kristensson showed
that such model is non-physical [33], it is still useful as an approximation: experimental
measurements show that the chirality parameter follows indeed a second order-resonant be‐
havior, that may be modeled as a combination of two orthogonal oscillators [34]. Electric
permittivity and magnetic permeability also follow a second-order resonant behavior (Lor‐
entz model [35]) with a resonant frequency virtually identical to ω 0 [11, 36]
Electromagnetic waves in general dispersive bi-isotropic media show some interesting be‐
havior [1]. First, the existence of the Pasteur parameter leads to:
• Optical/electromagnetic activity (OA): it is the turning of the plane of a linearly polarized
electromagnetic wave about the direction of motion as the wave travels through the mate‐
rial. This is caused by the fact that a chiral material transmits right- and left- handed cir‐
cularly polarized waves with different phase velocities.
• Optical/electromagnetic rotatory dispersion (ORD) is the variation of the optical/electro‐

magnetic activity with frequency. Usually, that dispersion leads to an OA that changes its
sign at a resonance frequency (Cotton effect).
• Circular dichroism (CD) is a modification of the nature of field polarization. It refers to

the different absorption of left- and right-handed circularly polarized electromagnetic
waves. As a consequence, linear polarization of a wave changes into an elliptical polariza‐
tion while travelling through a chiral material.
Second, the existence of the Tellegen parameter leads to:
• Non-orthogonallity of the electric and magnetic field vectors
• Non-reciprocal scattering parameters

• Isotropic polarization and magnetization induced in it by either a uniform electric or
magnetic field [19]
• Broken Time-reversal and space-inversion symmetries, that would lead with different
phenomena, still under discussion [30]
Finally, in certain cases, the cross-coupling relation may lead to a negative refraction index
[7, 8, 30], which has been confirmed experimentally in chiral media, as mentioned before
[9-11]
3. Time Domain Modelling
3.1. Finite Differences in Time Domain (FDTD)
The FDTD method is a time-domain numerical technique for solving Maxwell's equations.
Time-domain techniques produce wideband results with a single code execution and pro‐
vide deeper physical insight into the phenomena under study. In addition, nonlinear or
time-varying media are more easily handled in time-domain.
In the classical formulation of the FDTD approach the computational space is divided into
cells where the electric and magnetic field components are distributed according to the Yee
lattice [37]. Then, the space and time derivatives in Maxwell’s curl equations are approxi‐
mated by means of second-order central differences, obtaining a system of explicit finite-dif‐
ference equations that allows us compute the transient electric and the magnetic fields as
function of time. The resulting algorithm is a marching-in-time procedure that simulates the
actual electromagnetic waves in a finite spatial region.
The original FDTD scheme has been successfully extended to the modeling of different
kinds of materials [38]. However, the magnetoelectric coupling that characterizes the bi-iso‐
tropic media makes the extension of the FDTD method to incorporate bi-isotropic media a
challenging problem. First attempts can be found in [39-41]. In these works extensions of Be‐
renger’s PML for bi-isotropic and bi-anisotropic media are developed assuming that the
constitutive parameters are constant with frequency.
The schemes presented in [42-43] to incorporate bi-isotropic media are based on the field de‐
composition into the right- and the left-handed circularly polarized eigenwaves. These ei‐
genwaves are uncoupled and propagate in an equivalent isotropic media with effective
material parameters. Thus, the resulting sets of update equations do not differ in their gen‐
eral forms from the conventional FDTD update equations. Unfortunately, these schemes are
only valid in some particular cases, like normal incidence on a bi-isotropic medium or free
wave propagation, where the eigenwaves are not coupled. Problems like oblique incidence
on a chiral slab cannot be tackled using this technique.
The formulations developed in [44-45] consider the constitutive relations proposed by Kong
as governing equations. These approaches may be suitable for problems involving narrow‐
band signals. In [44] the algorithm is formulated for the particular case of Tellegen media. In
[45] the scheme is generalized for bi-isotropic media and a closed-form stability criterion,
with the same physical meaning than the conventional Courant’s condition, is derived for
Tellegen media.
The first formulation that considered the dispersive nature of the constitutive parameters
and allowed full transient modeling of general dispersive bi-isotropic media was presented
in [46-47]. In this formulation the frequency dependence of the permittivity and the permea‐
bility is represented by Lorentzian models while a single-resonance Condon model [1] [32]
is adopted to represent the dispersion in the chirality parameter. In [46] the dispersion is
handled using the piecewise linear recursive convolution (PLRC) technique, whereas in [47]
the auxiliary differential equation (ADE) method is employed.
One important feature of the schemes given in [46-47] is the use of a new FDTD lattice
developed to deal with the magnetoelectric coupling [44]. The strategy followed is to over‐
lap two Yee cells making the x-, y-, and z-components to coincide at the same point, respec‐
tively, and, therefore, defining x-, y-, and z-nodes staggered in space and time. The advantage
of this new mesh is that it simplifies the formulation since interpolation of the values of the
fields is not needed. However it presents drawbacks: the standard FDTD boundary condi‐
tions are not applicable and it requires implementing a specific interface with Yee’s origi‐
nal lattice.
Other FDTD models for dispersive chiral media have been developed in [48-49]. These
schemes use Yee’s conventional mesh and employ the Z-transform technique to convert fre‐
quency domain constitutive relations to Z-domain. However, these formulations lose the
second-order accuracy in time, since in order to decouple the updating equations backward
differences are used to approximate some time derivatives.
In this work we focus on an extension of the FDTD method for modeling frequency-disper‐
sive chiral media presented in [50]. The approach consists of discretizing Maxwell’s curl
equations according to Yee’s scheme and employing the Mobius transformation [51] to dis‐
cretize the constitutive relations. This scheme overcomes the main limitations found in other
approaches as it is a full-dispersive Model and is based on a modification of Yee’s original
cell. Moreover, the scheme preserves the second-order accuracy and the explicit nature of
the original FDTD formulation.
3.2. FDTD Formulation
A Differential model
In the Laplace domain, Maxwell’s curl equations can be written as

→ → → →
s B (r , s) = ∇ × E (r , s)
→ → → → (5)
sD (r , s) = ∇ × H (r , s)
and the constitutive relations for chiral media are

→ → → → 1 → →
D (r , s) = ε(s)E (r , s) − jκ(s)H (r , s)
c
(6)
→ → → → 1 → →
B (r , s) = μ(s)H (r , s) + jκ(s)E (r , s)
c
where the permittivity and permeability are characterized by a Lorentz model [35] and the
chirality parameter by a Condon model [32]
(εs − ε∞ )ωe2
ε(s) = ε∞ + (7)
ωe2 + 2δe s + s 2
(μs − μ∞ )ωh2
μ(s) = μ∞ + (8)
ωh2 + 2δh s + s 2
τκ ωκ2s
κ '(s) = jκ(s) = (9)
ωκ2 + 2δκ ωκ s + s 2
being ε∞ (μ∞ ) the permittivity (permeability) at infinite frequency, εs (μs ) the permittivity
(permeability) at zero frequency, ωe , ωh and ωκ the resonance frequencies, δe , δh and δκ the
loss factors and τκ a time constant. Introducing (6) into (5) we obtain
→ → → → → → → →
μ∞ sH (r , s) = − ∇ × E (r , s) − J hh (r , s) − J he (r , s)
→ → → → → → → → (10)
ε∞ sE (r , s) = ∇ × H (r , s) − J ee (r , s) − J eh (r , s)
→ → → →
where the terms J hh , J ee , J eh and J he are auxiliary current densities defined as
→ → → →
J hh (r , s) = σh (s)H (r , s) (11)
→ → → →
J eh (r , s) = ςh (s)H (r , s) (12)
→ → → →
J ee (r , s) = σe (s)E (r , s) (13)
→ → → →
J he (r , s) = ςe (s)E (r , s) (14)
→ → → →
with σh = s(μ − μ∞ ), ςh = − sκ^ / c, σe = s(ε − ε∞ ) andςe = − ςh . J hh , J eh and J ee , J he are evaluated at
→ →
the same point and at the same time step than H andE , respectively.
B Difference model
A discrete time-domain version of (10) is obtained by applying the property

sF (s) ↔ dF (t) / dt. Then, using central differences for the derivatives, average in time for
→ → → →
J hh , J ee and average in space forJ he , J eh , the following expressions are obtained [50]
r n+1 r n-1 Dt r
H 2 (=k + 12 ) H 2 ( k + 12 ) -
μ¥
{
[Ñ ´ E ]n ( k + 12 )
(15)
1 rn+1 rn-1 1 r r ü
+ é J hh 2 ( k + 12 ) + J hh 2 ( k + 12 ) ù + ëé J hen ( k ) + J hen ( k + 1) ûù ý
2 ë û 2 þ
r 1 r Dt ì r n +1
( k ) E n ( k ) + í [Ñ ´ H] 2 ( k )
E n +=
e¥ î
(16)
1 r r 1 é rn+1 æ 1 ö rn+1 æ 1 ö ùü
- éë Jeen +1 ( k ) + J nee ( k ) ùû - ê Jeh 2 ç k - ÷ + Jeh 2 ç k + ÷ ú ý
2 2ë è 2ø è 2 ø ûþ
Figure 1. Transformation of the constitutive equations into the discrete-time domain.
We should now derive discrete-time expressions for the constitutive relations (11)-(14). The
procedure is outlined in Figure 1. One attempt may consist on first transforming them into
the continuous-time domain by again using the propertysF (s) ↔ dF (t) / dt, and then approx‐
imating the resulting second-order differential equations by using finite differences. Howev‐
er, there are different ways to perform this discretization and each of them leads to different
algorithms with different numerical properties. In this study, the constitutive relations (11)-
(14) are discretized following an alternative way, first transforming them into the Z-domain
and then obtaining the discrete-time domain version. Thus, for (11) we obtain
→ → → →
J hh (r , Z ) = σh (Z )H (r , Z )
where σh (Z ) is the Z-domain magnetic conductivity. To obtain σh (Z ) we adopt Mobius

transformation technique [51]. This approach involves the following change of variable
2 1 − Z −1
s=
Δt 1 + Z −1
Applying this change to σh (Z ) yields
(σ ) (σ ) (σ )
→ → c0 h + c1 h Z −1 + c2 h Z −2 → →
J hh (r , Z ) = (σ ) (σ ) H (r , Z ) (17)
1 + d1 h Z −1 + d2 h Z −2
(σ ) (σ )
where cm h and dm h are real-valued coefficients. Now considering the Z-transform property
Z −m F (Z ) ↔ F n−m equation (17) can be written in difference form, as follows
→ 12 2 (σh ) → n+ 12 −m 2 (σh ) → n+ 12 −m
hh = ∑ cm H
J n+ − ∑ dm J hh
m=0 m=1
To reduce the memory requirements this expression is interpreted as an infinite-impulse re‐

sponse digital filter and is implemented by using the transposed direct form II. This type of
implementation is a canonical form, which guarantees the minimum number of delay ele‐
ments, thus the minimum number of additional storage variables. Applying this approach,
the resulting difference equations are
→ 12 → n− 12 (σh ) → n+ 2
1
J n+
hh = W hh ,1 + c0 H (18)
→ 12 → n− 12 (σh ) → n+ 2
1
(σ ) →
1
W n+
hh ,1 = W hh ,2 − d1 J hh + c1 h H n+ 2 (19)
→ 12 (σh ) → n+ 12 (σ ) →
1
W n+
hh ,2 = c2 H − d2 h J n+hh
2 (20)
These equations are evaluated atz = (k + (1 / 2))Δz. Note, that they are coupled to (15). Decou‐
→
pling them an eliminating W hh ,2 we obtain
r n + 1æ 1 ö 1 æ r n - 1æ 1 ö
H 2ç k+ ÷ = (s h ) ç
2m¥ H 2ç k+ ÷
è 2 ø 2 m¥ + Dtc0 è è 2ø
(21)
ì r æ 1 ö r n -1 æ 1 ö rn -1 æ 1 ö r r üö
-Dt í2 [Ñ É ] n ç k+ ÷ + Whh 2 ç k + ÷ + J hh 2 ç k + ÷ + Jhen ( k ) + J hen ( k +1) ý÷
î è 2ø è 2ø è 2ø þø
and
→ 12 → n− 12 (σh ) → n+ 2
1
J n+
hh = W hh + c0 H
(22)
→ 12 (σh ) → n− 12 (σ ) →
1
(σh ) → n+ 2
1
(σ ) →
1
W n+
hh = c2 H − d2 h J n−hh − d1
2
J hh + c1 h H n+ 2
→ →
whereW hh ≡ W hh ,1. Using the same procedure to discretize (12), we obtain the following
equation, which is evaluated at z = (k + (1 / 2))Δz
→ 12 → n− 12 (ξh ) → n+ 2
1
J n+
eh = W eh + c0 H
(23)
→ 12 (ξh ) → n− 12 (ξ ) →
1
(ξh ) → n+ 2
1
(ξ ) →
1
W n+
eh = c2 H − d2 h J n−eh − d1
2
J eh + c1 h H n+ 2
After that (13) is discretized in the same way than (11) y (12). The resulting equations are
coupled to (16). Decoupling them we get to the next expression
r 1 r
E n +1 ( k ) = (
2e ¥ + Dtc0(s e )
2e ¥ E n ( k )
(24)
ì r n+ 1 r r r n + 1 æ 1 ö r n+ 1 æ 1 ö ü ö
+Dt í2 [Ñ ´ H ] 2 ( k ) - Ween ( k ) - J nee ( k ) - Jeh 2 ç k - ÷ - J eh 2 ç k + ÷ ý÷
î è 2ø è 2 øþø
and
→ →n (σe ) → n+1
J n+1
ee = W ee + c0 E
→ (25)
(σe ) → n (σe ) → n (σe ) → n+1 (σe ) → n+1
W n+1
ee = c2 E − d2 J ee − d1 J ee + c1 E
Finally, (14) is discretized using the Mobius transformation, as was previously done for
(11)–(13), and implemented as
→ →n (ξe ) → n+1
J n+1
he = W he + c0 E
→ (26)
(ξe ) → n (ξe ) → n (ξe ) → n+1 (ξe ) → n+1
W n+1
he = c2 E − d2 J he − d1 J he + c1 E
Equations (25) and (26) are evaluated atz = kΔz.

Thus, the resulting computational procedure comprises the following calculations on each
time step:
→
1. H n+(1/2)is calculated by means of (21)
→ → →
2. J n+(1/2)
hh and W n+(1/2)
hh are updated using (22), where H n+(1/2) is known from step 1.
→ → →
3. J n+(1/2)
eh and W n+(1/2)
eh are obtained by using (23), where H n+(1/2) has been obtain in 1.
→
4. E n+1is evaluated by using (24)
→ → n+1 → n+1
5. J n+1
ee and W ee are calculated by means of (25), where E is known from step 4.
→ → n+1 → n+1
6. J n+1
he and W he are updated by using (26), where E has been obtained in 4.
This algorithm preserves the second-order accuracy and the explicit nature of the conven‐
tional FDTD formulation with only 4 additional back-stored variables per field component
and cell.
3.3. Results
Wave propagation in a chiral slab
FDTD schemes emulate the progression of the fields as they actually evolve in space and
time. This feature of time-domain simulators will allow us to visualize the noteworthy prop‐
erties of wave propagation in chiral media.
Figure 2. Evolution of the polarization of a wave traveling in a chiral slab: (a) lossless case: optical activity, (b) lossy case:
optical activity and circular dichroism.
Thus, in a first simulation we have considered a time-harmonic electric field that propagates
in the z positive direction and impinges on a slab of chiral material. Figure 2.a shows a snap‐
shot of the electric field in the entire simulation domain. It can be seen how, due to the opti‐
cal activity, the plane of polarization rotates as the wave travels through the chiral slab.
When losses are present in a chiral medium circular dichroism appears. The circular dichro‐
ism modifies the nature of the field polarization. To visualize this phenomenon we per‐
formed the same simulation than in the previous case but now considering a lossy chiral
slab. Figure 2.b shows how the polarization rotates as the wave propagates through the chi‐
ral slab and how the linearly polarized wave degenerates progressively into an elliptically
polarized wave.
Reflection and transmission from a chiral slab
Consider the propagation of an electromagnetic wave that travels in air and impinges on a
slab of chiral medium. In a chiral medium, the reflected and transmitted waves can be ob‐
tained through the following reflection and transmission dyadics
¯=
R̄ ( Rco
0
0
Rco
) T̄¯ = ( T co T cr
T cr T co
)
being Rco the copolarized reflection coefficient and T co , T cr the co- and crosspolarized trans‐
mission coefficients, respectively. The crosspolarized reflection coefficient equals zero.
The dispersive behavior of the constitutive parameters were assumed to follow the models
given in (7)-(9) with the following values εs = 4.2ε0, ε∞ = 3.3ε0, μs = 1.1μ0, μ∞ = μ0,
ωe = ωh = ωκ = 2π × 7.5GHz , δe = 0.05ωe , δh = 0.05ωh , δκ = 0.07and τκ = 1ps.
Figure 3. Reflection and transmission coefficients: theoretical values and relative errors.
A band limited Gaussian pulse was injected at a source point in the air region. The time step
was Δt = 1ps and the space discretizationΔz = 0.3mm. The thickness of the slab was d = 30Δz.
Figure 3 depicts the theoretical reflection and transmission coefficients [1] and the relative
error of the coefficients computed by the present FDTD formulation. These results demon‐
strate the accuracy and the full transient capability of this FDTD approach.
4. Multiresolution in Time Domain (MRTD)
Multiresolution in Time Domain (MRTD) technique is a new approach to reduce the simula‐
tion time of FDTD schemes (actually, it may be considered a generalization of FDTD [52]),
keeping the same degree of accuracy. In 1996, Krumpholz showed that Yee’s FDTD scheme
could be derived by applying the method of moments for the discretization of Maxwell’s
equations, using pulse basis functions for the expansion of the unknown fields [53]. In a
MRTD scheme the expansion is completed by means of a twofold expansion in scaling and
wavelet functions with respect to time/space. Scaling functions takes into account smoothly
varying fields. In regions or time periods characterized by strong field variations or field
singularities, higher resolution is enhanced by incorporating wavelet functions in the field
expansions. The major advantage of the use of MRTD in the time domain is the capability to
develop time and space adaptive grids. The scheme is, then, a generalization of Yee’s FDTD
and can extend this technique’s capabilities by improving computational efficiency and sub‐
stantially reducing computer resources.
The MRTD approach has been successfully applied to finite-difference time-domain (FDTD)
simulations of passive structures [53-56]. In those cases, multiresolution expansion has been
usually applied to the space dependence of the electromagnetic fields. In [57], the expansion
was applied to the time dependence of the fields. In this case, such expansion has the pur‐
pose of improving the resolution in the calculation of convolutions in dispersive media. As
we have seen previously, the chirality parameter, present in eq. (3) is intrinsically dispersive
(eqs. (7)-(9)), so we expand here a similar technique suitable for chiral media.
Figure 4. Haar wavelet functions: left, scale function Φ n(t). Right, first order wavelet function ψ n(t). Both cases are
particularized for n = 1.
To discretize the constitutive equations (2), we must develop all the components of the field
→ → → →
vectors (E , D , B , H ) in scale and wavelet functions:
F = ∑ ϕi (z )∑ (n F ϕ ϕn (t ) + n F ψ ψn (t )) (27)
i i
i n
F may be any component of any of the field vectors above mentioned. We have included
two levels of resolution in the time dependence of the function, while the space discretiza‐
tion is represented only in one dimension, assuming dependence on z. The extension to
three dimensions is straightforward: if the scale (Φ) and wavelet (ψ) functions are Haar
functions [58], as shown in Figure 4, the space discretization is equivalent to the FDTD dis‐
cretization explained in the previous section. That means that the discretization of Maxwell
curl equations will lead to two parallel Yee’s networks, one for scale coefficients (n F ϕ ), and
i
a second one for wavelet coefficients (n F ψ ).

i
The results for both networks are combined in the constitutive relations (Eq. (3)). If we dis‐
cretize them in time domain, using a recursive convolution method (in a similar way to
→ →
FDTD [46]), we obtain the following expressions (x component of D , B ):
D z, t ) = ∑
x(
i
{−∑ { ∫( 1 t
∑ ε (n E ix,ϕ ϕn (t ) + n E ix,ψ ψn (t ))
n
H x,ϕ ϕm(t − τ ) + m H x,ψ ψm(t − τ ))κ '(τ )dτ }ϕ z} i

( )
{ }
m
c 0 m' i i
(28)
∑ μ (n H ix,ϕ ϕn (t ) + n H ix,ψ ψn (t ))
n
B x (z, t ) = ∑
{ ∫( E }ϕ z
( )
1 t i
i +∑ x,ϕ ϕ (t − τ ) + m E x,ψ ψm(t − τ ))κ '(τ )dτ
m
c 0 m i m i
where κ’(τ) is the inverse Laplace transform of the chirality parameter in frequency domain
κ(ω). In this equationn E x,ϕ , for example, would represent the coefficient for the scale func‐
i
→
tion of the x-component of the electric fieldE , when t = n t and z = i κ z. Note that the equa‐
tion includes only the dispersion in the chirality response: dielectric and magnetic response
could be included as in [57]. If we are working with Haar functions, this equation may be
simplified [59], obtaining:
D z, t ) = ∑
x(
i,n
{ 1
− ϕn (t )∑
c
ε (n E x,ϕ ϕn (t ) + n E x,ψ ψn (t ))
i i
{n−m H ix,ϕκ'ϕm + n−m H ix,ψκ'ψm } } ϕi (z )
{ }
m
(29)
μ (n H x,ϕ ϕn (t ) + n H x,ψ ψn (t ))
i i
B x (z, t ) = ∑ 1 ϕi (z )
i,n + ϕn (t )∑
c m
{n−m E ix,ϕκ'ϕm + n−m E ix,ψκ'ψm }
∫ ∫ ∫
(n−m+1)Δt (n−m+1)Δt (n−m+1/2)Δt
whereκ'ϕn−m = (n−m)Δt
κ '(τ )dτ,κ'ψn−m = (n−m+1/2)Δt
κ '(τ )dτ − (n−m)Δt
κ '(τ )dτ.If we compare
these equations with (27), in order to separate scale and wavelet coefficients of D and B x, x
we find:
1
n Di
x,ϕ = ε E x,ϕ −
n i ∑
c0 m {n−m E ix,ϕκ'ϕm + n−m E ix,ψκ'ψm }
(30)
1
nB
x,ϕ = μ H x,ϕ +
i n i
∑
c0 m {n−m E
x,ϕ κ'ϕ +
i m n−m E
x,ψ κ'ψ
m i
}
n Di
x,ψ = ε ⋅ E x,ψ
n i
(31)
n Bi
x,ψ = μ ⋅ H x,ψ
n i
As we may see, the convolutions sums appear in eq. (30), i.e., they affect only the coefficients
→ →
of the scale functions of D and B (n D x,ϕ and n B x,ϕ ): if the wavelet functions give account of
i i
the faster variations of the fields, it is logic that they do not depend on the long term values
stored in the integral is reasonable.
So, we may follow a procedure similar to the one described in [57] for isotropic dispersive
materials, i.e. the algorithm would be composed by two different steps:
1. Propagation of the field, following the discretized Maxwell curl equations. We obtain
→ → → →
updated values of vectorsD andB , using the previous values of E andH .
→ →
2. Updating of the values of E andH , using the constitutive relations of the medium; we
should update first the wavelet (ψ) coefficients, following Eq. (31), because their value
is necessary to solve Eq. (30) and obtain the new values for the scale (ϕ) coefficients.
4.1. Results
To validate numerically the accuracy of our approximation, we have simulated the propaga‐
tion of a plane wave linearly polarized through a chiral medium, with ε =2ε0, μ=μ0, and κ
following the Condon model: ω 0=1 GHz, τ=3 ps and a damping factor δ=0 (medium without
losses). To simplify the results, we have not taken into account the dielectric and magnetic
dispersions (they do not affect the electromagnetic activity). The discretization is κ t = 1 ps
and κ x = 0.375 mm. We have introduced a harmonic source, with a frequency of 7 GHz, and
after 8000 time steps, we have calculated the rotation of the polarization plane in different
points: the result is compared with the one obtained using a FDTD scheme [46]. The results
are shown in Table 1: [59]
FDTD MRTD
Theoretical value
Value Error Value Error
1000Δz 11.55º 11.54º 0.09% 11.54º 0.09%
2000Δz 23.10º 23.09º 0.04% 23.08º 0.09%
3000Δz 34.64º 34.56º 0.2% 34.65º 0.03%
4000Δz 46.21º 46.11º 0.2% 46.04º 0.15%
Table 1. Rotation of the polarization plane during the propagation of a plane wave through a chiral medium [59]
5. Transmission Line Matrix (TLM)

The Transmission Line Matrix (TLM) method is a numerical technique that solves the wave
equation of a propagation phenomenon in time domain. As in the previous cases, the basic
TLM network [60] must be modified to include the cross terms emerging in the Maxwell
equations, thus in the wave equation, when dealing with bi-isotropic media. We consider
the propagation of waves in the xy–plane: although it is a bi-dimensional problem, the medi‐
um presents cross coupling on all the field components on every plane perpendicular to
propagation direction, so the number of unknown field components cannot be reduced less
than twelve.
After a simple manipulation, Maxwell equations can be arranged in four sets of equations as
shown in the Table 2, related by the constitutive relations described in equation (4).
where Φ is defined asΦ 2 = εμ − χ 2ε0μ0, and must have a non–zero value. This mathematical
condition, verified by all bi-isotropic media [1] means that any wave solution has associated
a finite value of the phase velocity.
∂ Ez ∂ By ∂ Ez ∂ Bx ∂ By ∂ Bx 2 ∂ Ez κ ∂2 Qmz
{1} = , =− , − = −
∂x ∂t ∂y ∂t ∂x ∂y ∂t c ∂t 2
∂ Hz ∂ Dy ∂ H z ∂ Dx ∂ Dy ∂ Dx ∂ H z κ ∂2 Qez
{2} =− , = , − = −Φ 2 −
∂x ∂t ∂y ∂t ∂x ∂y ∂t c ∂t 2
∂ Dz ∂ H y κ ∂2 Qey ∂ Dz ∂ H x κ ∂2 Qex ∂ H y ∂ H x ∂ Dz
{3} =Φ 2 + , = −Φ 2 − , − =
∂x ∂t c ∂t 2 ∂y ∂t c ∂t 2 ∂x ∂y ∂t
∂ Bz ∂ Ey κ ∂2 Qmy ∂ Bz ∂ Ex κ ∂2 Qmx ∂ Ey ∂ Ex ∂ Bz
{4} = −Φ 2 − , =Φ 2 + , − =−
∂x ∂t c ∂t 2 ∂y ∂t c ∂t 2 ∂x ∂y ∂t
Table 2. The Four Sets of Electromagnetic Field Vectors Equations [62]
To obtain a simplified representation, two next vectors, with no physical meaning, have
been introduced:
→ → χ→ → → → χ→ →
Q m = μH + E + B , Q e = εE + H + D (32)
c c
For every set of resulting equations, we may derive an independent wave equation describ‐
ing the propagation, and having the form:
∂2 A ∂2 A 2
2∂ A κ ∂3 ∂S
2 + 2 −ϕ 2 + c R+ =0 (33)
∂x ∂y ∂t ∂t 3 ∂t
A R S
Ez Qmz 0
Hz − Qez 0
χ
Dz εQmz − Qez Qez
c
χ
Bz Q − μQez Qmz
c mz
Table 3. SymbolsA, Rand S in wave equations [62]
The symbol A represents whatever z-component of the electromagnetic field vectors, and
→ →
each symbol R or S equals the z-component of a linear combination of Q m and Q e (both in‐
troduced in (32)) depending on the component A, as detailed in Table 3.
5.1. TLM Modeling
The classical TLM method works with interconnected transmission–lines along the coordi‐
nate axes establishing a network. The points at which the transmission–lines intersect are re‐
ferred to as “nodes”. Then, space and time are discretized, and voltage and current pulses
scatter from point to point in space, Δℓ, in a fixed time stepΔt, [60]. The starting point to
obtain any TLM model is to discretize the electromagnetic field equations (either the Max‐
well equations in Table 2 or the wave equations (33)), and to compare it with the equations
modeled by the TLM algorithm, that is, the equations relating the voltages and currents in
the transmission–line network. For a parallel connection of the transmission–line in the xy–
plane, the voltages and currents are related by the differential equations:
∂Vz ∂ Ix ∂Vz ∂ Iy ∂ Ix ∂ Iy ∂Vz

= −L = −L + = − 2C (34)
∂x ∂t ∂y ∂t ∂x ∂y ∂t
where L and C represent the inductance and the capacitance per unit length of the transmis‐
sion–lines, respectively. In these equations, the voltage V z, and the currents I x and I y sym‐
bolize the standard electric quantities in the transmission–line network [60].
A look at Table 2 allows us to implement the model of every set of equations with a classical
TLM mesh of shunt-connected nodes. Furthermore, the twelve field vectors components are
required in the new model due to the existing magneto–electric coupling, even in the 2D
case. In this way, a single set is not enough, but four meshes are needed to simulate each of
the twelve field components. Every TLM network will be labeled {1}, {2}, {3}, or {4}, follow‐
ing their equivalent set of Maxwell equations in Table 2. This leads to a specific analogy be‐
tween each TLM network electric quantity and its respective electromagnetic field vectors
component, as shown in Table 4 whereϕr = ϕ / c.
{1} {2} {3} {4}
V z ≡ ... Ez − μ0cH z Dz / ε0 − cBz
− I x ≡ ... By / μ0 cDy Φr2H y Φr2ε0cEy
I y ≡ ... Bx / μ0 cDx Φr2H x Φr2ε0cEx
Table 4. Analogies between TLM meshes quantities and field vectors [62].
∂Vz ∂ Ix ∂Vz ∂ Iy ∂ Ix ∂ Iy φ 2 ∂Vz

2
κ ∂ Qmz
{1} = − μ0 , = − μ0 , + =− +
∂x ∂t ∂y ∂t ∂x ∂y μ0 ∂ t η0 ∂ t 2
∂Vz ∂ Ix ∂Vz ∂ Iy ∂ Ix ∂ Iy φ 2 ∂Vz ∂2 Qez

{2} = − μ0 , = − μ0 , + =− +κ
∂x ∂t ∂y ∂t ∂x ∂y μ0 ∂ t ∂t 2
∂Vz ∂ Ix ∂2 Qey ∂Vz ∂ Iy ∂2 Qex ∂ Ix ∂ Iy ∂Vz

{3} = − μ0 + κη0 , = − μ0 − κη0 , + = − φr2ε0
∂x ∂t ∂t 2 ∂y ∂t ∂t 2 ∂x ∂y ∂t
∂Vz ∂ Ix ∂2 Qmy ∂Vz ∂ Iy ∂2 Qmx ∂ Ix ∂ Iy ∂Vz

{4} = − μ0 −κ , = − μ0 +κ , + = − φr2ε0
∂x ∂t ∂t 2 ∂y ∂t ∂t 2 ∂x ∂y ∂t
Table 5. The differential equations for the four TLM meshes [62].
→ →
The corresponding TLM equivalences of the vectors Q m andQ e , introduced in (32), are:
m0 ü 1 ü
Q= ém r I y {3} + cI y {4} + f2r I y {1}ûù ï Q
= ée r I y {4} + cI y {3} + f2r I y {2}ûù ï
fr2 ë fr2c ë
mx ex
ï ï
m ï 1 ï
Qmy = - 20 éëm r I x {3} + cI x {4} + f2r I x {1}ùû ï Qey = - 2 éëe r I x {4} + cI x {3} + fr2 I x {2}ûù ï
fr ý fr c ý (35)
ï ï
1
Qmz = - ëém rVz {2} + cVz {1} + Vz {4}ûù ï Qez= e0 ëée rVz {1} - cVz {2} + Vz {3}ùû ï
c ï ï
ï ï
þ þ
Again, these vectors have no special physical meaning. Please note that whatever the rela‐
tionship between field components and TLM variables is, these networks have the same
wave propagation characteristics of the classic TLM network, well studied in the literature,
[60]. Therefore, the numerical dispersion and consequently the velocity error of the bi-iso‐
tropic TLM network are well known.
To take into account the propagation velocity (the same in the four meshes following (3)), a
permittivity stub is shunt connected at all the nodes. The normalized admittance of these
stubs is yo = 4(ϕr2 − 1). Then, propagation velocities into each TLM network depend onϕ, dif‐
ferent from the velocity in the modeled bi-isotropic medium. The same transmission–lines
characteristics are imposed for all the meshes, and consequently the TLM meshes impedan‐
ces Z TLM equalηo / ϕr , where ηo denotes the vacuum intrinsic impedance.
Additionally, a new circuit element is needed at every node in order to incorporate the terms
in Table 2 with no equivalence in Table 4. The technique developed in [60] to model the waves
propagation in complex media, based on the voltage source connection, is used for this purpose.
It is important to remark that, since these voltages sources V sk will introduce the adequate
signal correction at any time iteration, we obtain a TLM algorithm that models the correct
propagation velocity.
The simulation begins with the excitation of one or several field vectors components, de‐
pending on the problem (symmetry, boundaries, output, …). Regarding the equivalences in
Table 4, voltage impulses are initially introduced into the system according to the desired
field excitation. The scattering and incidence processes of these voltage pulses follow the
classical TLM method procedures [60], but, for a bi-isotropic 2D–TLM model, the algorithm
is not yet completed, since at every time iteration k, the voltages at each node must be trans‐
ferred from a pair of networks to the other pair:
κ
V n {3} = εr V n {1} − χV n {2} + Δt Δt1V n {2}
κ
V n {4} = μr V n {2} − χV n {1} − Δ 1V {1}
Δt t n
1 κ (36)
Δt t ( r n
V n {1} = μ V {3} + χV n {4} − Δ 1 μ V {2} − χV n {1})
ϕr2 r n
1 κ
Δt t ( r n
V n {2} = ε V {4} + χV n {3} + Δ 1 ε V {1} − χV n {2})
ϕr2 r n
Here the index n denotes the branch number (n = 15) of the node, and the symbol Δtd repre‐
sents the d–order time variation of the pulses. The values of the new elements to be connect‐
ed at the TLM node equivalent circuit are obtained from the discretized form of wave
equations (33). The TLM method deals with the total voltage at the node, so it is more useful
to write the wave equations in terms of such voltages. For example, the wave equation for
the z- component of the electric field become for the mesh {1}:
∂2k V {1} ∂2k V {1}

T T
2 +
∂x ∂y2
(37)
∂2k V {1} κ ∂3 μ
T
− ϕr2ε0μ0 + − V {2} + χk V {1} −k V {4} = 0
∂t 2 c ∂ t 3 μ0 k T T T
The voltage of the four sources V sk at every time iteration k can be obtained directly from
these TLM wave equations. The resulting expressions for each TLM mesh are:
1 κ k
V sk +1{1} = V sk −1{1} − γ
ϕr2 Δt
1 κ k
V sk +1{2} = V sk −1{2} + ρ
ϕr2 Δt
(38)
1 κ κ 1 k
V sk +1{3} = V sk −1{3} − ε γ + χρ − Δ ρ
ϕr2 Δt r Δt t
1 κ κ 1 k
V sk +1{4} = V sk −1{4} + μ ρ + χγ + Δ γ
ϕr2 Δt r Δt t
where the symbols and indicate:
γ = Δt3 μrk V {2} − χk V {1} + k V {4} ρ = Δt3 εrk V {1} − χk V {2} + k V {3} (39)
T T T T T T
The sources are introduced into the algorithm by adding them to the voltage impulses, be‐
fore the scattering process, as follows:
k V {m} =
T
2
4 + y0 (∑ V {m} + y
4
n=1
k
i
n 0k 5
)
V i {m} + V sk {m} (40)
Finally, the information to get out any field vectors component, at any point of the network
and at any time k Δt, may be found in Table 4. The voltage V z at the node represents the
total voltage V T (10), and the currents I x and I y are proportional to the voltage pulses differ‐
ence. For example, the electric field strength components can be expressed in terms of the
incident voltage pulses upon the node:
k V i {4} −k V i {4} k V i {4} −k V i {4}

3 1 2 4
kE = kE =
x Δℓ 2ϕr2 y Δℓ 2ϕr2
4 (41)
∑ k V in{1} + yok V i5{1} V sk {1}
n=1
E = +
k z Δℓ2ϕr2 Δℓ
Finally, all these steps (pulse scattering, node connection, signal transfer between meshes
(36), voltage sources updating (38), voltage impulses summation (40), output of the fields
values (41)…) are repeated as many time steps as required.
5.2. Results
To validate the TLM model, we have modeled again the plane wave propagation in bi-iso‐
tropic media and compared the results with known theoretical ones [1]. In order to simulate
a plane wave propagating along the x–direction, the mesh boundaries are modeled with ab‐
sorbing conditions [62] for limits in x–direction to avoid not desired reflections, and magnet‐
ic walls for the y–direction. It is to be noted that in the four meshes (Table 4), the total
voltage at the node is either equivalent to an electric component ({1} and {3}) or to a re‐
versed–sign magnetic component ({2} and {4}) of the electromagnetic field, thus the same
boundary conditions in the y–direction have been modeled at all the meshes.
The propagation of a plane wave propagation in a bi-isotropic medium with relative permit‐
tivity and permeability εr = 5.0 and μr =1.5, Tellegen parameter χ = 1.2, and Pasteur parame‐
ter (Kong model) κ= –0.01 ps is now computed. In the simulation, the computational space
domain is n x = 5,000 nodes, and n y = 10 nodes along the x– and y–axes, respectively, with a
space step Δℓ = 1.00 mm. A monochromatic excitation of frequency 4.00 GHz is applied at
the plane x = 500Δℓ . The excitation is introduced with linear polarization in the z–direction
of the electric field strengthEz = E0sin(ωt ), then:
1
V {1} =k −1 V {1} + E sin(ωt ) forn = 1, ..., 5 (42)
k n n 2 0
Theoretical values TLM results Relative Error (%)
θEH 115.99º 116.09º −0.09
θDB 64.01º 64.12º −0.17
102 ηEH/o 54.77 54.63 +0.26
102 ηDB/o 54.77 54.92 −0.27
10 εr 44.94 44.87 +0.16
10μr 13.48 13.46 +0.15
Table 6. Angles, impedances and effective parameters [62]
In Table 6, the angle tilt between electric and magnetic field vectors θEH and θDB, the intrinsic
impedances ηEH and ηDB, and the effective relative permittivity and permeability are shown
after 10,000 time iterations (Δt= 2.36 ps) of the algorithm. In Table 7, the rotation of the angle
of the polarization plane at different distances from the source is presented. The respective
theoretical values [62] and relative errors are included also in these two tables of results:
they are in good agreement with the theoretical values, with relative errors less than 0.3 %.
Distance to the source (Δℓ) Theoretical values TLM results Relative Error (%)
500 90.60º 90.74º −0.15
1000 91.21º 91.36º −0.17
1500 91.81º 91.97º −0.18
2000 92.41º 92.60º −0.20
2500 93.02º 93.26º −0.26
Table 7. Rotation of the polarization plane [62]
6. Frequency Domain Modelling: The Coupled Mode Method (CMM)

The Coupled Mode Method is one of the available numerical methods in the frequency do‐
main used for analyzing the electromagnetic wave propagation inside waveguides that con‐
tain any isotropic, anisotropic or complex medium [63-69]. Originally, this method was
formulated by Schelkunoff [63] for uniform closed structures involving heterogeneous and
non-isotropic media. Afterwards, it was extended to cover other structures such us open di‐
electric guides [64], waveguides containing bi-isotropic media [65-69], and free propagation
in unbounded bi-isotropic media [70].
Moreover, the application of this numerical method is not only restricted to the analysis of
uniform structures. In fact, one of the main advantages of the coupled mode method is that
it can be combined with the Mode Matching Method (MMM) in order to analyze discontinu‐
ities between two waveguides that contain isotropic, anisotropic or complex media. Thus,
joining the coupled mode method and the mode matching method a hybrid numerical tech‐
nique capable of analyzing 3D structures implemented in a waveguide that presents discon‐
tinuities in the energy propagation direction is obtained [71-72]. Due to the mode matching
method properties, there is no limitation in the number of discontinuities that we can ana‐
lyze. Therefore 3D periodic structures can also be examined with this numerical tool.
6.1. Theory
In few words, the coupled mode method is a method of moments [63] which consists on ex‐
panding the electromagnetic field components inside a uniform waveguide partially filled
with any of the former media in terms of a set of base functions previously defined. Thus,
the general expression of any electromagnetic field component is expressed as
∞ ∞
G(x, y, z) = ∑ cn gn (x, y, z) = ∑ cn g̃ n (x, y)e −Γz (43)
n=1 n=1
where cn are the expansions coefficients, g̃ n the selected base functions and the propagation
constant of the proper modes, i.e., the modes of the partially filled waveguide. In many cases
these base functions are the electric and magnetic fields of the different modes supported by
the basis structure, i.e., the empty waveguide. Therefore, they are called basis modes. When
the guiding structure is a rectangular waveguide these basis modes correspond to the TE
and TM modes of the empty rectangular waveguide [63].
The obtention of the expansions coefficients and the propagation constants of the proper
modes constitutes the mathematical procedure of the coupled mode method. Mainly, it con‐
sists on selecting two electromagnetic field vectors as primary expansions1 and then substi‐
tuting them into the x- and y- components of the Maxwell Equations.
Next, the other two field vectors are expressed in terms of the primary ones through the
constitutive relations (3). Thus, the obtained equations only depend of the primary fields ex‐
pansions. In order to solve the problem, the Galerkin method is applied [73] and a set of dif‐
ferential equations, named the Generalized Telegraphist Equations, is obtained [63].
Finally, after some analytical work involving the z-components of the Maxwell Equations
[63,69] and the constitutive relations, we reach to an eigenvalue problem whose solution
provides the expansion coefficients and the propagation constants of the proper modes
{ A − Γ U } C = 0. (44)
In Eq. (44) [A] is a coupling matrix between the basis modes taken into account in the proce‐
dure, [U] is an unitary matrix, is a vector related to the propagation constants of the electro‐
magnetic field and [C] is a matrix containing the coefficients of the electromagnetic field
expansions. In following sections this procedure is analyzed in more detail.
6.1.1. A. The different formulations of the coupled mode method:
→ →
As it is know, the electromagnetic field is described by four field vectors H and, B which are
related among themselves through the constitutive relations. Thus, only the expansions of
→ → → →
two primary fields have to be proposed: one electric (E orH ) and another magnetic (E orB ).
The other two field vectors are obtained by substituting the primary fields into the constitu‐
tive relations. Hence, for the same problem, the coupled mode method can be formulated in
different ways giving rise to different formulations that produce similar solutions.
So far, in the literature two alternative formulations of the coupled mode method can be
→ →
found: the EH- and EB-formulations [67-69]. In the first one the primary fields are E andH ,
→ →
whereas in the second one the primary fields are E andB . As we will see later, the main dif‐
ference between the EH- and EB-formulations of the coupled mode method resides in the
satisfaction or not of all boundary conditions of the electromagnetic field components over
the perfect electric conductor that define the waveguide. In particular, the issue is related to
→ → → → →
1 As it will be shown later these primary fields can be E and H , D or B and E .
the normal component of the magnetic field over the walls in contact with the chiral materi‐
al, as they must be no null-valued [69].
EH-formulation of the coupled mode method.
In the EH-formulation, the primary expansions, particularized for a uniform rectangular

waveguide [63], are
nte ntm
∂T ( x, y ) ∂ T ( p )( x, y )
Ex ( x, y ) = ∑V q
q
∂y
+ ∑V ( p) ∂x
(45)
q=1 p=1
nte ntm
∂T ( x, y ) ∂ T ( p )( x, y )
Ey ( x, y ) = − ∑V q
q
∂x
+ ∑V ( p) ∂y
(46)
q=1 p=1
ntm
Ez ( x, y ) = ∑k z (
c ( p )V ( p )T ( p ) x, y) (47)
p=1
nte ntm
∂T ( x, y ) ∂ T ( p )( x, y )
H x ( x, y ) = ∑I q
q
∂x
− ∑I ( p) ∂y
(48)
q=1 p=1
nte ntm
∂T ( x, y ) ∂ T ( p )( x, y )
H y ( x, y ) = ∑I q
q
∂y
+ ∑I ( p) ∂x
(49)
q=1 p=1
nte
H z ( x, y ) = ∑k c Iz T
q q
( x, y ) + h 0 (50)
q
q=1
Here, V’s and I’s represent the electric and magnetic field’s expansion coefficients, and the
T-functions are the T-potentials that give the electromagnetic field components of the basis
modes TE and TM in the empty waveguide [63,68]. For rectangular waveguides they are
• TE modes
( nπa x) cos( mπb y),

η
n, m
T = cos
n, m kc ab (51)
n, m
n, m = 0, 1, 2, .... butnot n = m = 0
• TM modes
T (n, m) =
η(n, m)
kc(n, m) ab
( nπa x)sin( mπb y)
sin
(52)
n, m = 1, 2, 3, ....
with
η{n, m} = { 2 if n ≠ 0, m ≠ 0
2 if n = 0 or m = 0
(53)
Moreover, in Eqs. (45) -(52), k c(p) and k c[q] are the cutoff wavenumber of the basis modes; the
term h 0 is a constant that must be taken into account in some media in order to obtain the
correct solution of the electromagnetic field. Also, nte and ntm correspond, respectively, to
the number of basis modes TE and TM taken into account in the expansions. In all expres‐
sions, parentheses refer to TM-modes and brackets to TE-modes.
Following the procedure of the coupled mode method previously mentioned at the begin‐
ning of this section and detailed in [63,68], the Generalized Telegraphist Equations for the
EH-formulation turn out as:
dV( n ) æ ¶T( n ) ¶T( n ) ö

= jw ò ç Bx - By z
÷ ds + kc( n) V( n ) (54)
dz sç ¶y ¶x ÷ø
è
dV[m] æ ¶T[m] ¶T[m] ö

- jw ò ç Bx
= + By ÷ ds (55)
dz sç ¶x ¶y ÷ø
è
dI ( n ) æ ¶T( n ) ¶T( n ) ö
- jw ò ç Dx
= + Dy ÷ ds (56)
dz sç ¶x ¶y ÷ø
è
dI[m] æ ¶T[m] ¶T[m] ö

- jw ò ç Dx
= - Dy z
÷ ds + kc[m] I[m] (57)
dz sç ¶y ¶x ÷ø
è
Inspecting the set of equations (54)-(57) it can be seen that it is constituted by 4 equations
and 6 unknowns. Therefore, in order to find their solution, two additional expressions relat‐
ing the longitudinal coefficients with the transversal ones are needed. These relations are
provided by the longitudinal components of the Maxwell equations and the constitutive re‐
lations; they can be obtained following two alternative ways that give two different formula‐
tions of the EH-formulation.
EHD-formulation
The first presented way to obtain the additional relations between the longitudinal and
transversal coefficients it the one proposed by Ogusu [64]. Although it is not the original
procedure, it provides the more intuitive way to relate these coefficients. Therefore, it is
firstly introduced.
In this formulation the additional relations are obtained by rewriting the constitutive rela‐
tions (3) as
1 μr (χ − jκ )
Ez = D −c B (58)
ε0 μ ε − (χ 2 + κ 2) z μr εr − (χ 2 + κ 2) z
r r
(χ + jκ ) 1 εr
Hz = c (59)
(χ 2 + κ 2) − μr εr Dz − μ0 (χ 2 + κ 2) − μr εr Bz
Then, after introducing the longitudinal components of the Maxwell equations and the
expansions (45)-(50) into (58)-(59), and applying the Galerkin mode [73], we get the de‐
sired expressions
c nte ( c - jk )
kc( n) V( nz ) = åV kc2 kc2
jw j =1 [ j ] [ j] ( n) ò S é m re r - ( c 2 + k 2 ) ù [ j ] ( n )
T T ds
ë û
(60)
1 ntm mr
- å I kc2 kc2
jwe 0 i =1 (i ) (i) ( n) ò S é mre r - ( c 2 + k 2 ) ù (i ) ( n )
T T ds
ë û
1 nte er
kc[m] I[zm] = åV kc2 kc2
jwm0 j =1 [ j ] [ j] [m] ò S é( c 2 + k 2 ) - m re r ù [ j ] [m]
T T ds
ë û
(61)
c ntm ( c + jk )
- å I kc2 kc2
jw i =1 (i ) (i) [m] ò S é( c 2 + k 2 ) - mre r ù (i ) [m]
T T ds
ë û
Note that in (60)-(61) the longitudinal coefficients are directly related with the transversal
ones, i.e., no matrix inversion is required. Therefore we call it direct EH-formulation of the
coupled mode method or, simply, EHD-formulation.
Once the longitudinal and transversal coefficients are known, only the coefficient h 0 have to
be determined. Following a similar procedure, the value of h 0 can be obtained from
1 nte er c ntm ( c + jk ) T ds
=h0 ò ds
S åV kc2
jwm0 j =1 [ j ] [ j] ò S é( c 2 + k 2 ) - e r ù [ j ]
T ds - å I kc2
jw i =1 (i ) (i) ò S é( c 2 + k 2 ) - e r ù (i ) (62)
ë û ë û
EHI-formulation
The alternative process to obtain the relations between the longitudinal and transversal coeffi‐
cients is the original one formulated by Schelkunoff [63]. Now, the longitudinal components of
the Maxwell equations and the expansions (45)-(50) are directly introduced in the constitu‐
tive relations (3). After applying the Galerkin method [73], the desired equations are obtained
ntm
I ( n ) = - jwe 0 å kc(i) V(iz) ò e rT(i )T( n ) ds
S
i =1
w nte
-j å kc I z ( c - jk )T[ j]T( n)ds
c j =1 [ ] [ j ] ò S j
(63)
w
- j h0 ò ( c - jk ) T( n ) ds
c S
nte
- jwm0 å kc[ j] I[zj ] ò m rT[ j ]T[m]ds - jwm0 h0 ò m rT[m]ds
V[m] =
S S
j =1
(64)
w ntm
-j å kc V z ( c + jk )T(i )T[m]ds
c0 i =1 ( ) (i ) ò S
i
Inspecting (62) and (63) two main differences with the EHD-formulation can be seen. The
first one is that in order to isolate the longitudinal coefficients, matrix inversions are re‐
quired. Therefore, as the longitudinal coefficients cannot be directly expressed in terms of
the transversal ones, we call this formulation “Indirect EH-formulation” of the coupled
mode method or, simply, EHI-formulation. The second difference deals with the obtention
of the constant h 0. As it influences over the expansions coefficients, it must be taken into
account in the eigenvalue problem. Thus a third relation must be included. Following a sim‐
ilar procedure, the value of h 0 can be obtained from
nte
e0 ntm
- å kc[m] I[zm] ò m rT[m]ds -
h0 ò m r ds = å k ( )V( ) ò ( c + jk )T( )ds
c i
z
i i (65)
S
j =1
S m0 i =1
S
EB-formulation
We have seen that, in the EH formulation of the coupled mode method, both electric and
magnetic fields inside the partially filled waveguide are expressed in terms of the electric
and magnetic fields of the empty waveguide. Therefore they satisfy the same boundary con‐
ditions over the perfect electric conductor. As long as the media contained in the waveguide
allow the proper modes and the basis modes to satisfy the same boundary conditions, the
EH-formulation provides correct results. This happens when the media are dielectric, iso‐
tropic or anisotropic. The problem appears when the medium produces a discontinuity in
any component of the electromagnetic field that cannot be reproduced with the basis modes
of the waveguide. This occurs, for example, when a bi-isotropic medium fills the waveguide.
If we focus our attention on the normal component of the magnetic field over the perfect
electric conductor, we see that it must be not null valued, whereas the normal component of
the magnetic induction does. However, when the problem is analyzed with the coupled
mode method, it can be seen that, if the magnetic field satisfies the same boundary condi‐
tions than the basis modes, the normal component of the magnetic field will be zero over the
PEC, which is incorrect. In order to avoid this inconvenience we can choose as primary
fields those that satisfy the same boundary conditions over the perfect electric conductors
→ →
than the basis modes fields. These are E andB . Therefore, the proposed primary expansions
for these fields are
nte ntm
∂T ∂ T (n )
Ex = ∑V m ∂y
m
+ ∑V (n ) ∂x
(66)
m=1 n=1
nte ntm
∂T ∂ T (n )
Ey = − ∑V m ∂x
m
+ ∑V (n ) ∂y
(67)
m=1 n=1
ntm
Ez = ∑k z
c(n )V (n )T (n )
(68)
n=1
nte ntm
∂T ∂ T (n )
Bx = μ0 ∑b m ∂x
m
− μ0 ∑b (n ) ∂y
(69)
m=1 n=1
nte ntm
∂T ∂ T (n )
By = μ0 ∑b m ∂y
m
+ μ0 ∑b (n ) ∂x
(70)
m=1 n=1
nte
Bz = μ0 ∑k c bz T
m m
+ μ0b0 (71)
m
m=1
Following a similar procedure than the presented for the conventional formulation of the
coupled mode method, an eigenvalue problem of Eq. (54)-(57) type is obtained. As it hap‐
pened in the previous formulations, in order to solve this set of equations, two new equiva‐
lent relations between longitudinal and transversal coefficients and another third one to
obtain b 0 are needed. When the medium filling the waveguide is an isotropic chiral medi‐
um, the coefficient b 0 turns to be null.
Solving the Telegraphist equation eigenvalue problem, the coefficients expansions and the
→ →
propagation constants for H and H can be obtained in a way similar to the EH-formulation.
Magnetic Field
When only the knowledge of the electric field or the constant propagations of any proper
→ →
modes is required, the expansions of H and E obtained using the EB- formulation are
enough to obtain the desired solution. However, if we need to characterize a discontinuity,
→
an expansion for H is also necessary. As it was previously mentioned, the obtention of this
→ →
new expansion for E involves Eqs. (66)-(71) and the constitutive relations (3). Once the B
expansions are known, the mode matching method can be easily extended [71-72, 75] and
any discontinuity can be analyzed.
6.2. Results
In order to check the behavior of the different formulations of the coupled mode method
previously presented, the analysis of electromagnetic wave propagation inside a uniform
rectangular waveguide filled with an isotropic chiral medium is proposed (Figure 5). It is
important to note that there is no analytical solution available of this problem. Hence, a pri‐
ori one could not establish which formulation provides the most accurate results. However,
according to the conclusions extracted from the analysis of parallel-plate waveguide con‐
taining an isotropic chiral media [67], and the fact that the EB-formulation was developed to
ensure the satisfaction of all boundary conditions of the electromagnetic field over the per‐
fect electric conductor, we can infer that the EB- is the optimal formulation for analyzing this
problem. Thus, along this section this statement is verified with numerical results. For that
purpose we focus our attention in two properties of the electromagnetic field: the propaga‐
tion constant values of the proper modes and their magnetic field components profile. In the
first case it can be seen how the convergence behavior of each formulation is. In the second
one it is shown how the magnetic field obtained by the EB-formulation satisfies the boun‐
dary conditions of the normal component of the magnetic field whereas that extracted from
the EH-formulation does not.
Figure 6 shows the variation of the phase constant of the fundamental proper mode normal‐
ized to the vacuum wavenumber versus the number of basis TE and TM modes (conver‐
gence diagram) for the structure of Figure 5 provided by the mentioned formulations of the
coupled mode method. Here, the basis modes TE and TM have been taken by increasing
their cutoff frequency and the simulation data is displayed in the figure caption. An inspec‐
tion of this figure evidences that the final values of the propagation constant provided by
the EB- and EHI-formulations are very similar but slightly different than the EHD ones.
Moreover, the convergences of the EB- and EHI- are faster than the EHD- one as they need a
smaller number of basis modes TE and TM to reach to the final value. Therefore, with these
evidences, which can be extended to the rest of proper modes, we can infer that the EHI-
formulation gives more accurate results than the EHD-.
Figure 5. Cross section of a uniform rectangular waveguide fully loaded with an isotropic chiral media (ICM). P.E.C.
stands for perfect electric conductor
Figure 6. Convergence diagram of the fundamental mode for a uniform waveguide completely filled with an isotropic
chiral media. Dimensions in mm: a × b = 22.86 × 10.16;εr = 3, µr = 1, χ = 0 and κ= 1.5.
As it was previously mentioned, since the EHI-formulation implies a matrix inversion that
introduces a numerical error, one could think that the results provided by the EHD-formula‐
tion should be more accurate than those from the EHI. However, this statement is false and
the EHI- results are closer to the EB- than the EHD- ones. The explanation of this disagree‐
ment is related to the fact that in the EHD-formulation the continuous E z component is writ‐
ten as the product of discontinuous functions. This situation, which is right analytically, is
not correct when is obtained numerically, as it is impossible to reproduce a discontinuous
function in terms of a finite number of sines and cosines. In the other hand, in the EHI-for‐
mulation, although matrix inversions are required, a discontinuous function is expressed in

terms of the product of discontinuous and continuous functions, which is correct both ana‐
lytically and numerically. For this reason, the EHI-formulation results agree with the EB-for‐
mulation ones better than those obtained from the EHD-formulation.
Figure 7. Module of the magnetic field components of the fundamental mode, normalized to their maximum value,
as a function of the x- and y- coordinates provided by (a) EHI-formulation and (b) EB-formulation. Dimensions in mm:
a × b = 22.86 × 10.16. εr = 3, µr = 1, χ= 0 and κ= 0.75. Number of basis modes 120 for the EB-formulation and 150 for
the EH-.
Finally, in Figure 7 the module of the three components of the magnetic field normalized to
their maximum values, as a function of the x- and y- coordinates are presented. As it is clear
that the EHI-formulation is more appropriate that the EHD when analyzing this problem,
only results for EHI- and EB-formulations are shown. On inspecting Figure 7 it is easy to
prove that the magnetic field obtained with the EB-formulation does satisfy the boundary
conditions of the normal component of the magnetic field whereas that extracted from the
EH-formulation does not. It can be seen in Figure 7 how the normal component of the mag‐
netic field over the perfect electric conductor walls is null in the EHI-formulation meanwhile
it is not null in the EB- one.
7. Conclusions
A review of different numerical techniques for the modeling of electromagnetic wave propa‐
gation through bi-isotropic materials has been presented. In the time domain, different
FDTD models have been mentioned, with emphasis in one based on Mobius transformation,
as well as one MRTD approach, suitable to improve recursive convolution techniques and
one TLM model. In the frequency domain, two different formulations of the coupled mode
method, the EH- and the EB-formulations, have been presented. It has been seen that the EB-
formulation is the optimal for analyzing electromagnetic wave propagation inside a rectan‐
gular waveguide that contains an isotropic chiral media because the solution that it gives
verify all the boundary conditions of the electromagnetic field over the perfect electric walls
that constitute the waveguide.
Acknowledgements
Partial support for this work was provided by the Spanish MICINN through the
TEC2010-21496-C03-01, subprogram “Ramón y Cajal” (RYC-2010-06922) and CONSOLIDER
CSD2008-00066 projects.
Author details
I. Barba1*, A. Grande1, A.C.L. Cabeceira1, A. Gómez2, J.A. Pereda2 and J. Represa1
*Address all correspondence to: [email protected]
1 University of Valladolid,, Spain
2 University of Cantabria,, Spain

References
[1] Lindell, I. V., Sihvola, A. H., Tretyakov, S. A., & Viitanen, A. J. (1994). Electromagnet‐
ic waves in Chiral and Bi-Anisotropic Media. Boston, Artech House.
[2] Sihvola, A. H. (1994). Electromagnetic Modeling of Bi-Isotropic Media. Progress in

Electromagnetic Research, PIER, 9, 45-86.
[3] Sihvola, A. H., & Lindell, I. Material effects on bi-anisotropic electromagnetics. 1995,
IEICE Trans. Electron. (Tokyo) E78-C:, 1383-1390.
[4] Tretyakov, S. A., Sihvola, A. H., Sochava, A. A., & Simovski, C. R. (1998). Magneto‐
electric interactions in bi-anisotropic media. J. Electromag. Waves Appl. JEMWA, 12,
481-497.
[5] Soukoulis, C. M., & Wegener, M. (2011). Past achievements and future challenges in
the development of three-dimensional photonic metamaterials. Nature Photonics, 5,
523-530.
[6] Barba, I., Cabeceira, A. C. L., García-Collado, A. J., Molina-Cuberos, G. J., Margineda,
J., & Represa, J. (2011). Quasi-planar Chiral Materials for Microwave Frequencies. In:
Kishk A, editor. Electromagnetic Wave Propagation in Complex Matter. Rijeka: In‐
tech , 97-116.
[7] Pendry, J. B. (2004). A Chiral Route to Negative Refraction. Science, 306, 1353-1355.
[8] Tretyakov, S. A., Sihvola, A. H., & Jylhä, L. (2005). Backward-wave regime and nega‐
tive refraction in chiral composites. Photonics and Nanostructures: Fundamentals and
Applications,, 3, 107-115.
[9] Plum, E., Zhou, J., Dong, J., Fedotov, V. A., Koschny, T., Soukoulis, C. M., & Zhelu‐
dev, N. I. (2009). Metamaterial with negative index due to chirality,. Physical Review
B, 79, 035407-1.
[10] Zhou, J., Dong, J., Wang, W., Koschny, T., Kafesaki, M., & Soukoulis, C. M. (2009). ,
Negative refractive index due to chirality. Physical Review B , 79, 121104-1.
[11] Molina-Cuberos, G. J., García-Collado, A. J., Barba, I., & Margineda, J. (2011). Chiral
Metamaterials With Negative Refractive Index Composed by an Eight-Cranks Mole‐
cule. IEEE Antennas and Wireless Propagation Letters, 10, 1488-1490.
[12] Lakhtakia, A. (1994). Ten years past the Post. In McCall MW, Dewar G, editors. Com‐
plex Mediums V: Light and Complexity., Proc. SPIE (Int. Soc. for Optical Engineering),,
5508, 85-94.
[13] Weiglhofer, W. S. (1994). On a medium constraint arising directly from Maxwell’s

equations. J. Phys., A27:, L871-L874.
[14] Post, E. J. (1962). Formal Structure of Electromagnetics. Amsterdam, North-Holland.

[15] Lakhtakia, A., & Weiglhofer, W. S. (1995). On a constraint on the electromagnetic

constitutive relations of nonhomogeneous linear media. IMA J. of Appl. Math., 54,
301-306.
[16] Lakhtakia, A., & Weiglhofer, W. S. (1996). Lorentz covariance, Occam’s razor, and a
constraint on linear constitutive relations. Phys. Lett. A Erratum A222: 459., 213,
107-111.
[17] Weiglhofer, W. S., & Lakhtakia, A. (1998). The Post constraint revisited Arch. Elek‐
tron. Uebertrag., 52, 276-279.
[18] Lakhtakia, A. (2004). On the genesis of the Post constraint in modern electromagnet‐
ism. Optik-International Journal of Light and Electron Optics, 115, 151-158.
[19] Raab, R. E., & Sihvola, A. H. (1997). On the existence of linear non-reciprocal bi-iso‐
tropic (NRBI) media. JPhys. A: Math. Gen. , 30, 1335-1344.
[20] Lindell, I. V., & Sihvola, A. H. (2004). Perfect electromagnetic conductor. Journal of
Electromagnetic waves and Applications JEMWA, 19, 861-869.
[21] Lindell, I. V., & Wallén, K. H. (2004). Differential-form electromagnetics and bi-aniso‐
tropic Q-media. J. Electromagn. Waves and Appl. JEMWA, 18, 957-968.
[22] Lindell, I. V., & Wallén, K. H. (2004). Generalized Q-media and field decomposition
in differential-form approach. J. Electromagn. Waves and Appl. JEMWA, 18, 1045-1056.
[23] Lindell, I. V., & Sihvola, A. H. (2005). Transformation method for problems involving
perfect electromagnetic conductor (PEMC). IEEE Trans.Antennas Propag., 53,
3005-3011.
[24] Tretyakov, S. A., Maslovski, S. I., Nefedov, I. S., Viitanen, A. J., Belov, P. A., & San‐
martin, A. (2003). Artificial Tellegen particle,. Electromagnetics, 23, 665-680.
[25] de Lange, O. L., & Raab, R. E. (2001). Post’s constraint for electromagnetic constitu‐
tive relations,. J. Optics A3, L23-L26.
[26] Serdyukov, A. N., Semchenko, I. V., Tretyakov, S. A., & Sihvola, A. H. (2001). Electro‐
magnetics of Bi-anisotropic Materials: Theory and Applications. Amsterdam, Gordon
and Breach.
[27] Canto, J. R., Paiva, C. R., & Barbosa, A. M. (2011). Modal Analysis of Bi-isotropic H-
Guides. Progress In Electromagnetics Research PIER, 111, 1-24.
[28] Matos, S. A., Paiva, C. R., & Barbosa, A. M. (2009). A Spacetime Algebra Approach to
Moving Bi-isotropic Media. Proc IEEE AP-S/URSI International Symp. APSURSI 09.
[29] Hehl, F. W., & Obukhov, Y. N. (2005). Linear media in classical electrodynamics and
the Post constraint. Phys. Lett. A, 334, 249-259.
[30] Kamenetskii, E. O., Sigalov, M., & Shavit, R. (2009). Tellegen particles and magneto‐
electric metamaterials. Journal of Applied Physics, 105, 013537-1.
[31] Kong, J. A. (2008). Electromagnetic Wave Theory. Cambridge, USA EMW Publish‐
ing., 1016.
[32] Condon, E. U. (1937). Theories of optical rotatory power. Rev. Modern. Phys., 9,
432-457.
[33] Kristensson, G. (1999). Condon’s model on optical rotatory power and causality- a
scientific trifle. In: Eds. A. Karlsson and G. Kristensson. Festskrift till Staffan Ström:
Lund, Sweden, KF Sigma. Available: http://www.lunduniversity.lu.se/o.o.i.s?
id=12683&postid=530495. Accesed 2012 May 30., 97-119.
[34] Rogacheva, A. V., Fedotov, V. A., Schwanecke, A. S., & Zheludev, N. I. (2006). Giant
Gyrotropy due to Electromagnetic-Field Coupling in a Bilayered Chiral Structure.
Physical Review Letters, 97, 177401-1.
[35] Balanis, C. A. (1989). Advanced Engineering Electromagnetics. New York, USA, Wi‐
ley.
[36] Bahr, A. J., & Clausing, K. R. (1994). An approximate model for artificial chiral mate‐
rial. IEEE Trans. Antennas Propag., 42, 1592-1599.
[37] Yee, K. S. (1966). Numerical solution of initial boundary value problems involving
Maxwell’s equation is isotropic media. IEEE Trans. Antennas Propag., 14, 302-307.
[38] Taflove, A. (1995). Computational Electrodynamics: The Finite-Difference Time-Do‐
main Method. Norwood, MA, Artech House.
[39] García, S., Perez, I., Martín, R., & Olmedo, B. (1998). Extension of Berenger’s PML for
bi-anisotropic media. IEEE Microwave Guided Wave Lett., 8, 297-299.
[40] Teixeira, F., & Chew, W. (1998). General closed-form PML constitutive tensors to
match arbitrary bi-anisotropic and dispersive linear media. IEEE Microwave Guided
Wave Lett., 8, 223-225.
[41] Garcia, S., Perez, I., Martin, R., & Olmedo, B. (1999). BiPML: A PML to match waves
in bi-anisotropic media. Microwave Opt. Technol. Lett., 20, 44-48.
[42] Akyurtlu, A., & Werner, D. H. (2004). BI-FDTD: A novel finite-difference time-do‐
main formulation for modeling wave propagation in bi-isotropic media. IEEE Trans.
Antennas Propag., 52, 416-425.
[43] Akyurtlu, A., & Werner, D. H. (2004). A novel dispersive FDTD formulation for mod‐
eling transient propagation in chiral Metamaterials. IEEE Trans. Antennas Propag., 52,
2267-2276.
[44] Barba, I., Grande, A., Cabeceira, A. C. L., & Represa, J. (2003). Time domain modeling
of electromagnetic wave propagation in Tellegen media. Microwave Opt. Technol.
Lett., 38, 167-168.
[45] Grande, A., Barba, I., Cabeceira, A. C. L., Represa, J., Kärkkäinen, K., & Sihvola, A. H.
(2004). Analyzing the stability of the new FDTD technique for non-dispersive bi-iso‐
tropic media using the von Neumann method. VIII ACES, Symp. Syracuse., New
York.
[46] Grande, A., Barba, I., Cabeceira, A. C. L., Represa, J., So, P. M. P., & Hoefer, W. J. R.
(2004). FDTD modeling of transient microwave signals in dispersive and lossy bi-iso‐
tropic media. IEEE Trans. Microwave. Theory Tech., 52, 773-784.
[47] Grande, A., Barba, I., Cabeceira, A. C. L., Represa, J., Kärkkäinen, K., & Sihvola, A. H.
(2005). Two dimensional extension of a novel FDTD technique for modeling disper‐
sive lossy bi-isotropic media using the auxiliary differential equation method. IEEE
Microwave Wireless Components Lett., 15, 375-377.
[48] Demir, V., Elsherbeni, A. Z., & Arvas, E. (2005). FDTD formulation for dispersive chi‐
ral media using the Z transform method. IEEE Trans. Antennas Propag., 53, 3374-3384.
[49] Nayyerir, V., Soleimani, M., Rashed-Mohassel, J., & Dehmollaian, M. (2011). FDTD
modeling of dispersive bianisotropic media using Z-transform method. IEEE Trans.
Antennas Propag., 59, 2268-2278.
[50] Pereda, J. A., Grande, A., Gonzalez, O., & Vegas, A. (2006). FDTD modeling of chiral
media by using the Mobius transformation technique. IEEE Antennas Wireless Propag.
Lett., 5, 327-330.
[51] Pereda, J. A., Vegas, A., & Prieto, A. (2002). FDTD modeling of wave propagation in
dispersive media by using the Mobius transformation technique. IEEE Trans. Micro‐
wave Theory Tech., 506, 1689-1695.
[52] Bushyager, N., & Tentzeris, E. M. (2006). MRTD (Multi Resolution Time Domain) Meth‐
od in Electromagnetics (Synthesis Lectures in Computational Electromagnetics)., San Fran‐
cisco, USA:, Morgan & Claypool,, 116.
[53] Krumpholz, M., & Katehi, L. P. B. (1996). MRTD: New time-domain schemes based
on multiresolution analysis. IEEE Trans. Microwave Theory Tech., 44, 555-571.
[54] Tentzeris, E. M., Cangellaris, A., Katehi, L. P. B., & Harvey, J. (2002). Multiresolution
time-domain (MRTD) adaptive schemes using arbitrary resolutions of wavelets,.
IEEE Trans. Microwave Theory Tech., 50, 501-516.
[55] Fujii, M., & Hoefer, W. J. R. (1998). A three-dimensional Haar-wavelet-based multire‐

solution analysis similar to the FDTD method-Derivation and application. IEEE
Trans. Microwave Theory Tech., 46, 2463-2475.
[56] Grivet-Talocia, S. (2000). On the accuracy of Haar-based multiresolution time- do‐

main schemeS. IEEE Microwave Guided Wave Lett., 10, 397-399.
[57] Barba, I., Represa, J., Fujii, M., & Hoefer, W. J. R. (2003). Multiresolution model of
electromagnetic wave propagation in dispersive materials. International Journal of Nu‐
merical Modelling, 16, 239-247.
[58] Daubechies, I. (1992). Ten Lectures on Wavelets.Philadelphia, USA, SIAM, 357.

[59] Barba, I., Grande, A., Cabeceira, A. C. L., & Represa, J. (2006). A Multiresolution
Model of Transient Microwave Signals in Dispersive Chiral Media. IEEE Transactions
on Antennas & Propagation, 54, 2808-1812.
[60] Hoefer, W. J. R. (1985). The Transmission-Line Matrix method- Theory and Applica‐
tions. IEEE Trans. Microwave Theory Tech., 33, 882-893.
[61] Barba, I., Cabeceira, A. C. L., Represa, J., & Panizo, M. (1996). Modelling dispersive
dielectrics for HSCN TLM method. Int. J. Numer. Modeling, 14, 15-30.
[62] Cabeceira, A. C. L., Barba, I., Grande, A., & Represa, J. (2006). A Time-Domain Mod‐
elling for EM wave propagation in Bi-Isotropic media based on the TLM methoD.
IEEE Trans. Microwave Theory Tech., 54, 2780-2789.
[63] Schelkunoff, S. A. (1952). Generalized telegraphist’s equations for waveguides. Bell

Syst. Tech. J., 3, 784-801.
[64] Ogusu, K. (1977). Numerical analysis of rectangular dielectric waveguide and its
modifications. IEEE Trans Microwave Theory Tech MTT-25, 441-446.
[65] Xu, Y., & Bosisio, R. G. (1995). An efficient method for study of general bi-anisotropic
waveguides. IEEE Trans Microwave Theory Tech MTT-43, 873-879.
[66] Xu, Y., & Bosisio, R. G. (1997). A study on the solutions of chirowaveguides and bia‐
nisotropic waveguides with the use of coupled-mode analysis. Microwave Opt Technol
Lett, 14, 308-311.
[67] Gómez, Á., Vegas, A., & Solano, M. A. (2004). A brief discussion on the different for‐
mulations of the coupled mode method in chiral media: Application to the parallel-
plate chirowaveguide. Microwave and Optical Technology Letters, 42, 181-185.
[68] Gómez, Á., Vegas, A., & Solano, M. A. (2006). On two different formulations of the
Coupled Mode Method: Application to 3D rectangular chirowaveguides. AEU-Inter‐
national Journal of Electronics and Communications, 60, 690-704.
[69] Gómez, Á., Lakhtakia, A., Vegas, A., & Solano, M. A. (2010). Hybrid technique for
analysing metallic waveguides containing isotropic chiral materials. IET Microwaves,
Antennas and Propagation, 4, 305-315.
[70] Gómez, Á., Barba, I., Cabeceira, A. C. L., Represa, J., Vegas, A., & Solano, M. A.
(2008). Application of Schelkunoff’s method for simulating isotropic chiral free prop‐
agation: Clarifying some common errors. Journal of Electromagnetic Waves and Applica‐
tions, 22, 861-871.
[71] Solano, M. A., Vegas, A., & Prieto, A. (1994). Numerical analysis of discontinuities in
a rectangular waveguide loaded with isotropic or anisotropic obstacles by means of
the coupled-mode method and the mode matching method. Int J Numer Model Elec‐
tron Networks Devices Fields, 7, 433-452.
[72] Solano, M. A., Vegas, A., & Prieto, A. (1998). Theoretical and experimental study of
two ports structures composed of different size rectangular waveguides partially fil‐
led with isotropic dielectrics. Int J Electron, 84, 521-528.
[73] Harrington, R. F. (1993). Field computation by moment methods. IEEE Press Series on
Electromagnetic Waves, New Jersey, Chapter 1.
[74] Solano, M. A., Vegas, A., & Gómez, A. (2006). Comments on "a comprehensive study
of discontinuities in chirowaveguides". IEEE Trans. on Microwave Theory and Techni‐
ques, 54, 1297-1298.
[75] Gómez, Á., Lakhtakia, A., Margineda, J., Molina-Cuberos, G. J., Núñez, M. J., Saiz,
Ipiña. J. A., Vegas, A., & Solano, M. A. (2008). Full-wave hybrid technique for 3-D
isotropic-chiral-material discontinuities in rectangular waveguides: Theory and ex‐
periment. IEEE Trans. on Microwave Theory and Techniques, 56, 2815-2825.
Chapter 4
Integral Equation Analysis with

Characteristic Basis Functions
Jaime Laviada, Fernando Las-Heras and

Marcos R. Pino
1. Introduction
Full-wave electromagnetic analysis algorithms dependence on the number of unknowns N.

The complexities ranges from O( N log N ) for certain iterative schemes such as the fast mul‐
tipole method [8,42] or the adaptive integral method [5] to O( N 3) for the conventional meth‐
od of moments (MoM) [20].
Thus, the reduction of the number of unknowns results very attractive since it reduces the
demanded computational resources in terms of memory and CPU time. In the method of
moments context, the choice of the basis functions has a strong impact on the degrees of
freedom that are required to solve the problem and, therefore, strategies for unknowns re‐
duction are based on choosing the basis functions as good as possible.
Conventional piecewise basis functions such as Rao-Wilto-Glisson (RWG) [40] or rooftops
[15], involve an expansion that is able to model almost any kind of current. Thus, it could be
convenient to limit the set of basis functions so that is is only made of that ones which can
model currents with a physical meaning and, therefore, to reduce the number of degrees of
freedom of the problem under analysis. It is well-known that entire-domain basis functions
enable a very efficiently modeling of the currents. Nonetheless, the computation of a set of
entire-domain basis functions it is only possible for certain canonical geometries [44,31,4] or
via numerical techniques which involve solving the entire problem [19,18] with the corre‐
sponding computational cost.
The basis functions to be used for solving our problem should be able to model any kind of
current that could be induced in the geometry under analysis. In a general analysis, it is not
possible to obtain a finite and complete set of basis functions unless the conditions of the
problem are strongly restricted. Nonetheless, it will be shown in this book chapter that it is
possible to compute a set of basis functions that can be progressively increased to reach an
accuracy similar to the accuracy that would be achieved by means of low-level basis func‐
tions (RWG, rooftops, etc.).
The entire-domain basis functions are usually expressed as linear combinations of low-level
basis functions. This definition enables an easy and efficient computation of the system of
equations matrix since it is possible to exploit the linear properties of the integral equations
to carry out the computation.
Among the pioneering works on this field, we find the theory ofcharacteristic modes that has
been carry out by Harrington [19,18]. This modes are computed by solving the following
generalized eigenvalue problem:
¯ Ī = λ R̄
X̄ ¯
n n Ī n (1)
where the matrices R̄¯ and X̄¯ are the real and imaginary part of the impedance matrix, re‐
spectively. The impedance matrix should have been computed from the EFIE with the Ga‐
lerkin's method. Thus, the eigenvalues, λn, and eigenvectors, In, will be real due to the
symmetries. The choice of the eigenvectors as basis functions yields a diagonal matrix with
the corresponding computational advantages. If the geometry under analysis is modeled
with N conventional low-level basis functions (e.g., RWG or rooftop), the number of eigen‐
values will be also N. This eigenvalues are a complete base that can model any current with
the same accuracy as the one that could be achieved with the N low-level basis functions.
Nonetheless, the eigenvector with highest eigenvalues will have a more significant contribu‐
tion to the radiation so that the eigenvectors under a certain threshold can be discarded as
basis functions. Thus, it is possible to find a set of basis functions with very good accuracy
whose size is smaller than N.
During the eighties and nineties, the research on this kind of basis functions was diffused.
Nevertheless, the interest on these techniques has reemerged due to the possibilities in the
application to the design of antennas [6]. In this research line, the physical meaning of the
characteristic modes is exploited for the design of antennas with certain characteristics. The
convergence of the accuracy as a function of the number of basis functions is improved by
including a source mode that is computed by exciting the antenna and attaching the induced
current with the eigenvectors.
Computation of characteristic modes involves the solution of an eigenvalue problem that

limits the size of the elements where the basis functions can be defined. For this reason, this
technique has been limited to the analysis of arrays with elements that are not connected
each other. In addition, once the basis functions has been computed for an element, they can
be copied for the rest of the elements and, therefore, to save CPU time. An example of this
technique has been shown in [32] where the authors analyze a 3 × 3 antenna array by defin‐
ing basis functions for each element. In particular, they consider three kinds of elements de‐
Integral Equation Analysis with Characteristic Basis Functions 99
pending on the position of the element (center, corner or border) and they reuse the same
basis function for each kind of element.
A similar problematic happens in the adaptive basis function method that has been pro‐
posed in [45]. In this method, a diagonally dominant moment matrix is achieved by building
macrobasis functions for a certain body. This macrobasis functions are chosen so that they
are orthogonal to the testing functions on the body. This method was also extended in
[14,12,13] for the solution of arrays of equally-spaced disconnected elements.
A special case of basis functions defined as aggregation of low-level basis functions is the
Multilevel Method of Moments [22]. The authors developed an iterative system based on the
generation of hierarchical basis functions for the analysis of two-dimensional scattering
problems. In this scheme, the solution of the problem was computed by iteratively solving it
with basis functions of different sizes so that it enables to refine the solution. Although the
aim of these basis functions was not an efficient modeling of the current, it was one of the
pioneering on nesting basis functions which will be one of the key point of the techniques
presented in this book chapter.
The previous scheme would be modified in [38] for the analysis of printed circuits. In this
case, the technique was modified to work with high level basis functions without the need
of going back to the low-level basis functions in the intermediate stages. In this method, the
circuits were split into blocks and artificial ports were introduced at the end of the blocks.
The basis functions were obtained by feeding the natural and the artificial ports and consid‐
ering the continuity equation for the currents. The extension of this technique for geometries
different from printed circuits (e.g. a three-dimensional scatterer) was not accomplished to
the authors' best knowledge.
A technique similar to the previous one is the Subdomain Multilevel Approach (SMA) [43] for
the analysis of certain kind of microstrip circuits. In this method, the geometry is decom‐
posed into several subdomains (e.g. the feeding network and the antennas) and these subdo‐
mains are fed by a generator. The induced currents are used as the basis functions which are
referred to as macrobasis functions. The rest of the geometry, which is not decomposed into
subdomains, is analyzed with conventional basis functions (weak compression) or through
new macrobasis functions computed by feeding each port (strong compression). The accura‐
cy of the method is improved by decomposing each macrobasis functions into two basis
functions corresponding to the real and imaginary parts. In addition, the joint between mac‐
robasis functions is improved by including rootops between them. The main drawback of
this method is also the difficulty to be extended to arbitrary geometries.
The characteristic basis function method (CBFM) [39] laid the foundations to generate macroba‐
sis functions for three-dimension arbitrary-shaped scattering and radiation problems. In this
first work, the generation of the characteristic basis functions (CBFs) yields two kinds of ba‐
sis functions primary and secondary ones. This method will be later presented and com‐
pared with some new techniques for generating CBFs.
Another method which shares many points in common with the CBFM is the Synthetic-Func‐
tions eXpansion (SFX) [35]. The different versions of the CBFM [33,11] as well as the SFX can
be considered as a generalization of the Multilevel method of moments and SMA to three-

dimensional problems where plane-waves or external dipoles are used instead of artificial
ports. This point will be deeply discussed later in section 2.2.
This chapter is structured as follows. Firstly, the theory of the analysis with characteristic ba‐
sis functions is presented. This first section includes the general basis of any method based
on macrobasis functions defined as linear combinations of low-level basis functions, the the‐
oretical consideration of generating CBFs with the external sources method, a comparison of
methods for the computation of the reduced matrix and the recursive application of the
characteristic basis functions. In the next section, several applications of the CBFM are de‐
scribed including monostatic radar cross section computation, synthesis of phased arrays
and analysis of partial modifications on large structures. Finally, the conclusions of the pre‐
vious sections are summarized.
2. Analysis with characteristic basis functions
2.1 General basis
The characteristic basis function method pursues to find a set of efficient basis functions so
that a smaller number of degrees of freedom is required to solve the problem with the meth‐
od of moments. The aforementioned basis functions are defined on domains grouping sev‐
eral contiguous low-level basis functions (e.g., RWG or rooftops). In the CBFM context, the
domains of definitions of the characteristics basis functions are referred to as blocks. Thus,
the first step in order to apply the CBFM is to split the geometry under analysis into B
blocks (see Figure 1).
Figure 1. Block decomposition for an airplane-like geometry.
On each block, a set of basis functions, which are referred to as characteristic basis functions,
are defined. Therefore, the current is expressed as follows:
where Ki is the number of CBFs in the i-th block; is the n-th CBF in the i-th block; and,
the coefficients Ii,n are the weights of the CBFs in order to express the final solution as a line‐
ar combination of the new basis functions. This coefficients are the unknowns to be solved
by the method of moments. The CBFs are defined as linear combination of the low-level ba‐
sis functions:
where Ni is the number of low-level basis functions in the i-th block; and f i,m is the m-th
low-level basis functions belonging to the i-th block. The definition of the CBFs as a linear
combination enables a more simple computation of certain terms such as the reaction terms
in the impedance matrix and, therefore, it enables an easy adaptation of previously imple‐
mented method of moments codes.
In order to achieve a compact notation, it is useful to define matrices grouping the coeffi‐
cients of the CBFs for each block. This matrix is defined for the i-th block as follows:
2.2 External illumination method
The generation of the CBFs is usually accomplished by means of one of the two following
techniques: a) primary and secondary CBFs method; b) external illumination method.
The first method was proposed in the pioneering technique about the CBFM [39] and it was
mainly designed for the analysis of arrays of M elements not connected between them. The
primary basis functions were obtained by computing the current on each isolated element as
a consequence of feeding the port connected to it. Once the primary CBFs are computed for
each element, the secondary basis functions for a given element are computed by consider‐
ing the current induced due to the radiation of the primary CBFs on the rest of the elements.
Thus, one primary CBF and M − 1 secondary basis functions are defined for each element.
Although the primary and secondary CBFs method has shown an excellent behavior for the
analysis of arrays, the accuracy is not so good for scattering problems where the primary
CBFs are generated by solving the isolated blocks for the illumination under analysis. In ad‐
dition, the CBFs generated by this method depend on the employed incident field and,
therefore, they have to be recomputed again if the excitations are changed.
The external illumination method was introduced in [33,11] to solve the previous problems. As
it will be next shown, the CBFs computed by this technique are a complete set, what has not
been proved to be true for the primary and secondary CBFs method.
Let us consider the electric field integral equation for perfect electric conductors (PECs) in
terms of the currents induced on each block:
where Jn is the current induced in the n-th block and Ei is the impressed field. The linear
operator L( ⋅ ) relates the currents and the electric field radiated by them and it is defined
in [7].
If the currents on all the blocks but the m-the are known, then the integral equation can be
expressed as:
the solution of (6) yields the current on the block. It is important to observe that if the super‐
position principle is considered, then the final current on the block can be obtained as a line‐
ar combination of the currents induced due to the field radiated by the rest of the blocks and
the impressed field.
The field outside a certain block can be expressed as a linear combination of a set of modes
which are a complete base for the electromagnetic field. The most common expansions of
the field are spherical-wave expansion and the plane-wave expansion. Another possible expan‐
sion is possible by enclosing the block with a surface supporting equivalent magnetic and
electric currents which model the radiation due to the sources outside the block. This last
expansion is the one used in the SFX [35] where the RWG basis functions are used to model
the equivalent currents.
If it is assumed that any field impinging the block can be expanded as a discrete and finite
set of waves (or as the field radiated by a discretized equivalent current enclosing the
block[35]), what can be carried out with arbitrary accuracy by using spherical-or plane-
waves (or low-level basis functions), then any current induced in the block can be expressed
as a linear combination of the currents induced by each wave of the aforementioned set (or
by the field radiated by each low-level basis functions).
The approach presented in [33,11] is based on approximating the external illumination by

plane-waves. In order to understand this approach, it is important to consider the well-
known expressions for plane-wave expansion. These expressions enable us to expand the
electric field under a certain plane z = z0 as [17, chapter 3]:
∫∫
∞ ∞
E(r) = T(k x , k y )e − jk⋅rdk x dk y z ≥ z0, (7)
−∞ −∞
where the vector T(k x , k y ) is the Plane Wave Spectrum (PWS); and the propagation vector is
defined as k = k x x + k y y + k z z with:
It is also possible to use analogous expressions for the cases of sources above or under other
planes (e.g., x = x0). Since it is possible to enclose the block by a surface with flat faces, then
the field outside can be expanded by employing a plane-wave expansion for each face of the
enclosing surface.
In order to illustrate it, let us consider the two-dimensional case depicted in Figure 2. In this
example, the body is enclosed by a rectangle so that the sources can be decomposed into
four groups. The first and second groups are the sources above and under the rectangle, re‐
spectively, whereas the third and four groups are the sources at the right and left of the rec‐
tangle which do not belong to any of the previous groups. Thus, it is possible to carry out
expansions similar to the one in (7) for each plane and, therefore, to express the field inside
the rectangle as a superposition of plane-waves.
Figure 2. Decomposition of the outer sources into regions for the expansion into plane-waves.
The PWS is infinite and unbounded as inferred from (7). Hence, several numerical approxima‐
tions must be carried out. The first one consist on sampling the PWS; guidelines to choose
the number of plane-waves depending on the electrical size of the block are given in [28].
In order to truncate the PWS, it is important to observe that it can be split into two regions.
The visible region corresponds to the area where the propagation vector is real and, there‐
fore, the wave propagates in that direction. On the other hand, the invisible region corre‐
sponds to propagation vectors where at least one of the components is complex and,
therefore, this kind of waves is attenuated when it travels away from the sources. Thus, if
we assume that the block is far enough from the sources, then it is possible to neglect the
contribution from the invisible spectrum. Similar approximations for the PWS were carried
out for the analysis of open cavities in [41].
If the blocks are close (or touching each other), then the approximation of neglecting the in‐
visible spectrum can slow down the convergence of the solution in terms of the number of
the CBFs. In order to overcome it, when illuminating with plane-waves, the blocks are not
considered isolated. Instead of that, they are solved together with a small portion of the con‐
tinuous blocks. Typical block extensions are between 0.1λ and λ.
Taking into account these two approximations, the coefficients for the CBFs can be comput‐
ed as:
¯
˜ J̄˜¯ ' = V̄
Z̄ ¯ by i = 1, ..., M
˜ (9)
ii i i
where is the matrix containing the field tested by the basis functions in the extended i-th
block due to each plane waves. Each column of the matrix corresponds to an excitation vec‐
tor and, therefore, its dimensions are Ñ i × N PW , where Ñ i is the number of low-level basis
functions in the extended block and NPW is the number of considered plane-waves.
After discarding the currents on the extensions, a complete set of basis functions, J̄˜¯ 'i , is ach‐
ieved. Nonetheless, the linear dependency of these CBFs is not warranted and, hence, if they
were used as basis functions, it would yield a MoM matrix that could be ill-conditioned or
even singular. In order to detect the number of degrees of freedom in the matrix J̄¯˜ ' , a singu‐ i
lar value decomposition (e.g., see [16]) is used. This factorization yields the following result:
J̄¯ 'i = Ū
¯ D̄¯ ¯
i i W̄ i (10)
where W̄ ¯ and Ū
¯ are orthogonal matrices with dimensions N × N and N × N , where as
i i i i PW PW
¯
D̄ i is a diagonal matrix with dimensions Ni× NPW that contains the singular values σ. It will
¯ are arranged from the highest one
be assumed that the singular values in the diagonal of D̄ i
to the smaller one without loss of generality.
¯ are orthogonal each other and, therefore, they are a complete base of the
The columns of Ū i
¯ . In addition, the ratio σ /σ is a
vector subspace generated by the column vectors in Ū i k 1
¯ with respect to the first k − 1 first
measurement of the dependency of the k-column of Ū i
columns. Hence, if the singular values under a given threshold are approximated by zero,
¯ with the singular values over the previ‐

then it is only necessary to retain the columns in Ū
ous threshold. Finally, the matrix containing the coefficients by columns of the CBFs is:
J̄¯ i = Ū
¯ (:, 1: K )
i i (11)
where Ki is the number of singular values normalized by σ1 that are over a given threshold
tsvd. The Matlab [36] notation U i ( : , 1 : K i ) denotes all the columns in Ui from the first one to
the Ki-th one.
The inclusion of the SVD provides an efficient way to compute CBFs that are independent
each other and, furthermore, it enables an adjust of the quality of the solution by means of
the threshold tsvd. On the contrary, one of the drawbacks of employing a SVD is that the
CBFs lose the physical meaning since they do not correspond anymore to the currents in‐
duced by each plane-wave but they are a linear combination of them.
2.3 Reduced matrix computation
In the CBFM, the coefficients of the matrix of the system of equations are the reaction terms
between CBFs. Thus, considering the EFIE, the Galerkin's method and a symmetric product
for the testing, then the reaction term between the m-th basis function in the i-th block and
the n-th basis function in the j-th block is given by:
where the index (1) denotes that the matrix contains the interaction between the CBFs in‐
stead of the interaction between low-level basis functions. This superindex will be deeply
exploited when considering the multilevel version of the CBFM.
If (3) is introduced in (12):
¯ (0) is the matrix that contains the reaction terms between the low-level basis func‐
where Z̄ ij
tions in the i-th observation block and the j-th source block.
Equation (13) enables us to express the reaction terms between the CBFs of two blocks as a
simple product between matrices:
¯ (1) = J̄¯ T Z̄
Z̄ ¯ (0) J̄¯ (14)
ij i ij j
¯ (1) is the matrix containing the reaction terms between the CBFs in the i-th observa‐
where Z̄ ij
tion block and j-th source block.
The matrix of the system of equations for the CBFM, Z̄¯ 1 , is usually referred to as the reduced
( )
matrix and it can be computed by means of matrix-matrix products involving the coefficients
of the CBFs and the reaction terms of the low-level basis functions:
It is important to remark that the computation of the reduced matrix involves the entire
MoM matrix. Nonetheless, it is not necessary to store the entire matrix in memory at the
same time. Instead of that, it is only required to store the block of the original MoM matrix
that is employed for the computation of each submatrix in the reduced matrix.
Similar steps can be followed in order to computed the feeding vector (or feeding matrix if
multiple excitations are considered) by means of the following expression:
where V i (0) is the feeding matrix containing the low-level feeding vectors of the i-th block.
2.3.1 Fast computation of the reduced matrix
Since the reduced matrix involves all the terms in the MoM matrix, the computational times
can be considerably high for electrically large problems. Several approaches are available in
the literature to mitigate this drawback. One of the most successful approaches has been the
one presented in [10] for the fast computation of reaction terms between the CBFs of an ar‐
ray of repeated elements. This method is based on a multipolar expansion of the free space
Green's function as in the FMM.
Another method is the one presented in [34] based on the adaptive cross approximation (ACA)
[2,3]. In this method, the matrices containing the reaction terms between two blocks are
modeled by a matrix product of two matrices that can be efficiently computed. In addition,
this efficiency improves with the distance between blocks.
A third technique has been proposed in [30]. In this approach, the field radiated by the CBFs
of a source block is computed on the observation block by carrying out an interpolation so
that it avoids integrating the currents on the source block for each point of the observa‐
tion block.
Table 1 contains a comparison of the main characteristic of these methods. The points that
have been considered are the following ones:
ACA Multipolar expansion Interpolation

Setup time ✓ X ✓
Additional memory ✓ X ✓
requirements
Dependency on the ✓ X ✓
Green''s function
Distance adaptation ✓ X ✓
Continuous blocks – X ✓
Arbitrarily-shaped blocks ✓ ✓ X
Table 1. Comparison of several fast methods for computing the reaction terms between CBFs.
• Setup time: the multipolar expansion requires the precomputation of certain parameters.
On the hand, the interpolation scheme and the ACA does not require this step.
• Additional memory requirements: the precomputations carried out by the multipolar ex‐
pansion on the setup stage must be kept in memory so that introduces an additional pen‐
alty that does not exist in the other methods.
• Dependency on the Green's function: the ACA and interpolation scheme does not depend
on the Green's function that is considered (e.g., free space, multilayer, periodic structure,
etc.). Nevertheless, different formulations of the multipolar scheme have to be considered
depending on the Green''s function.
• Distance adaptation: the ACA becomes faster depending on the distance between the
blocks. Similarly, the interpolation scheme reduces the number of evaluation points so
that it also becomes faster if the distance between source and observation blocks is in‐
creased. The conventional multipolar expansion does not provide a mechanism to take
advantage of the distance between the blocks.
• Contiguous blocks: multipolar expansion can only be applied to blocks in the far-field so
that it is not used for contiguous blocks. Although, the interpolation scheme and the
adaptive cross approximation can work for contiguous blocks, it results not so efficient
for these blocks due to a considerable slowdown of the computations.
• Arbitrary-shaped blocks: the interpolation scheme is designed for planar blocks and,
therefore, it only works for two-dimensional blocks.
According to the previous considerations, the ACA seems to be the most complete and ver‐
satile method for the combination of it with the CBFM. Thus, it will be used in the rest of the
paper for the computation of the reduced matrix unless otherwise indicated.
2.4 Solution of the system of equations
Once the matrix and feeding vector have been computed, the following system of equations
must be solved:
¯ (1)Ī (1) = V̄ (1)

Z̄ (17)
where I(1) is the vector containing the unknowns for the expansion of the currents in (2):
and Ii(1) are the coefficients of the expansions of the CBFs in the i-th block:
( )
Ī 11
( )
Ī 21
(1)
Ī = (18)
⋮
( )
Ī B1
and Ii(1) are the coefficients of the expansions of the CBFs in the i-th block:
( )
I1 1
( )
( )
I2 1
Ī i 1 = (19)
⋮
Ii, N i
Once (17) has been solved, it is possible to retrieved the currents vector for the low-level ba‐
sis functions by expanding the coefficients of the CBFs:
J¯1I 1 1
( )
J¯2I 2 1
( )
Ī = (20)
⋮
J̄¯ B Ī B1
( )
(1)
If multiple excitations are considered, then Ṽ becomes a matrix a previously dis‐
(1)
cussed. As a consequence, Ĩ become also a matrix containing the solution for each ex‐
citation on each column.
2.5 Nested characteristic basis functions
The most important feature of the CBFM is the capacity to solve a problem with a number of
unknowns smaller than the number required by the conventional MoM. It has been empiri‐
cally proved that the compression rate of the CBFM (i.e., the ratio of the number of low-level
basis functions to CBFs) increases if the size of the blocks is increased. Figure 3 shows that
the number of CBFs that survives to the SVD filtering does not have a linear behavior with
the surface and, therefore, it is always recommended to consider blocks as electrically large
as possible to improve the compression.
Figure 3. Number of required CBFs for different surfaces as a function of the electrical size.
However, the main drawback of increasing the size of the blocks is that the computation of
the CBFs requires more computational resources to solve (9). In order to alleviate this task, a
possible alternative is to split the blocks into sublocks and, thus, applied the CBFM to solve
(9). Hence, it results into a recursive scheme where the CBFs can be expressed in terms of
CBFs from a lower level.
The first step to apply the multilevel CBFM (ML-CBFM) is to hierarchically decompose the
geometry into blocks until reaching the low-level basis functions. A decomposition example
is shown in Figure 4 for a two level decomposition of the NASA almond.
Figure 4. Hierarchical decomposition of the geometry.
In order to mathematically express the multilevel formulation, it is useful to define the nota‐
tion for a sequence of blocks {seq} = {i1, i2, … , il } where the values ik are the indexes of the
blocks from the highest level to the lower one. Thus, {seq} = {2, 5} is used to denote the fifth
block, which belongs to the first level, of the second block, which belongs to the second and
highest level.
Thus, the n-th CBF inside the block {seq} is given by:
where B{seq} is the number of block inside the block pointed by N {seq,b} is the number of CBFs
(n)
in the block I {seq,b},m are the coefficients for the expansion in terms of the CBFs on the sup‐
porting sublocks. The CBFs from the first level, i.e., the level before the low-level basis func‐
tions, are defined as in the single-level CBFM in terms of the low-level basis functions:
The current is as usually expressed as a linear combination of the highest level CBFs:
where B is the number of blocks and I {b},m are the coefficients to be computed by solving the
system of equations of the MoM.
The computation of the coefficients of the CBFs for each level is carried out in a similar fash‐
ion to the monolevel CBFM. Firstly, the first level blocks are illuminated by multiple plane-
waves and the CBFs are obtained on them by computing the currents and applying the SVD
filter. Next, the second level blocks are again illuminated by multiple plane-waves and the
induced currents (expressed in terms of the first level CBFs) are computed and filtered out.
The procedure is repeated until reaching the highest level. It is important to note that the
number of plane-waves and the SVD threshold can be different for each level depending on
the required accuracy.
Once the CBFs for each block in the highest level have been computed, then it is possible to
compute the system of equations matrix containing the reaction terms by applying again a
nested scheme. The reaction terms between the CBFs of the (I - 1)-th level belonging to the
blocks {d} and {e} of the (I-1)-th level level can be expressed as:
¯ (l −1)
Z̄ ¯ (l −1) ¯ (l −1)
{d ,1},{e,1} Z̄ {d ,1},{e,2} ⋯ Z̄ {d ,1},{e,1}
¯ (l −1)
Z̄ ¯ (l −1) ¯ (l −1)
{d ,2},{e,1} Z̄ {d ,2},{e,2} ⋯ Z̄ {d ,2},{e,B(e )}
¯ (l −1) =
Z̄ , (24)
{d },{e }
⋮ ⋮ ⋱ ⋮
¯ (l −1)
Z̄ ¯
Z̄
(l −1) ¯ ( )
⋯ Z̄ l −1
{d ,B{d }},{e,1} {d ,1},{e,1} {d ,B{d }},{e,B(e )}
where each submatrix can be computed by considering the coefficients of the CBFs defini‐
tion and the reaction terms from the CBFs of the previous level:
¯ (l ) = J̄¯ T J̄¯ (l −1) J̄¯ .

Z̄ (25)
{d },{e } {d } {d },{e } {e }
Similar steps can be carried out to compute the feeding vector (or matrix in the case of multi‐
ple excitations) so that the following system of equations can be solved in order to find the
unknown coefficients in (23):
¯ ( L )Ī ( L ) = V̄ ( L ).
Z̄ (26)
Once the system of equations has been solved, it is possible to express the CBFs in terms of
the CBFs of the next level by means of the following operation:
( )
Ī {lseq,1
−1
}
( )
Ī {lseq,2
−1
}
= J̄¯ {seq }Ī {lseq }
()
⋮ (27)
( −1)
Ī {lseq,B{seq }}
this formulation can be applied to all the blocks and repeated until reaching the lowest level
so that the solution is expressed in terms of the conventional low-level basis functions.
In the case of considering multiple excitations, the vectors V (L ), and the equivalent vectors
for each level, becomes matrices as discussed in the single-level CBFM
3. Applications
As it has been shown in the previous sections, the CBFM enables to comprise the MoM ma‐
trix. Since the solution of the system of equations is the most time consuming stage for the
analysis of electrically large problems with the MoM, a reduction of the number of un‐
knowns will significantly decrease the total solution time. Furthermore, the factor limiting the
size of the problems that can be solved with the conventional MoM is size of the impedance
matrix which must fit in the available RAM memory. Thus, the CBFM compression results
on an extension of the application range of the MoM. Considering these two facts, the CBFM
can be considered a speed-up technique for the MoM. In addition, the CBFM is not based on an
iterative scheme so it will not suffer from converge problems.
Although the use of the CBFM to compute an iteration-free of relatively large problems may
be the most direct application of the MoM, it will be shown in the following sections that the
compression of the MoM matrix can be exploited for many other purposes.
3.1 Monostatic Radar Cross Section Computation
The compressed system of equations described by (17) and (26) for the monolevel- and mul‐
tilevel-CBFM, respectively, can be solved by using conventional direct solutions schemes
such as a LU decomposition for problems electrically larger than the ones that can be solved
by the MoM. These schemes can handle a relatively large number of RHS with a negligible
time increment with respect of the case with a single RHS. As a consequence, many authors
(e.g., [33,11]) have benefited from this advantage to compute monostatic radar cross sections
(RCS) where there are as many RHS as incident angles.
This application is illustrated for the computation of the monostatic RCS of a 4λ cube. The
cube is solved with a single level CBFM by decomposing the geometry into 24 blocks of size
2λ × 2λ (see Figure 5). The number of plane waves to solve the problem is 400 and the SVD
threshold is set to 10-3. The original problem was discretized into 31728 RWG unknowns that
were reduced to only 3004 after applying the CBFM.
The results for the θ − θ RCS for 91 points in the cut ϕ = 0 are shown in Figure 5 togeth‐
er with the reference results from the MoM. Excellent agreement is found between both
of them.
Figure 5. Monostatic RCS for a 4λ cube computed by the MoM and the CBFM.
The CPU times are shown in Table 2. In this case, the ACA has not been used to speed-up
the matrix filling time which is the most time consuming stage. The CPU times for the
CBFM simulation have been computed on a PC equipped with an Intel® Core™ at 2.4 GHz
(only one core was used) and 4GB RAM memory whereas the MoM simulation was run on a
computational server in order to fit the impedance matrix on the RAM memory. This server
is equipped with 8 Dual Core AMD Opteron™ 880 at 2.4GHz (only one core was used) and
56GB of RAM.
CBFs generation Matrix filling Solving time Total time

time
MoM - 1468s 14198s 15666s
CBFM 510s 1614s 14s 2138s
Table 2. Computational times for solving a 4λ cube with the MoM and the CBFM.
Although the size of the previous problem is relatively electrically small, it shows how the
compression of CBFM enables to solve problems larger than the ones solved by the conven‐
tional MoM. The solution of larger problems can be efficiently carried out by combining the
multilevel scheme together with an appropriate parallelization scheme [29].
3.2 Fast evaluation of phased arrays

In addition to the monostatic RCS computation, there are other cases where the simultane‐
ous solution of multiple RHS is highly recommended. One important example is the synthe‐
sis of phased arrays. In this problem, the optimal combination of voltages at the input of
each element of the array has to be found in order to achieved the desired radiation pattern.
¯ , that relates the samples of the electromag‐
It is possible to show [1] that there is a matrix, Ḡ
netic field on a given set of points (e.g., a circular cut of points in the far-field) with the vol‐
tages by the following matrix-vector product:
¯ ῡ
Ē = Ḡ (28)
Once this matrix has been obtained, then the computation of the radiated field for a certain
choice of voltages can be computed with a very small computational effort. Nonetheless, the
computation of this matrix involves solving the geometry of the entire problem (the array
and the surrounding elements) for as many RHS as elements [1]. Therefore, the computation
¯ can be very time-consuming when the number of elements in the array is high and the
of Ḡ
structure to be solved is electrically large. As in the previous example, the CBFM can help to
mitigate the computational cost due to its ability to solve simultaneously multiple RHS.
In order to illustrate this example, a secondary surveillance radar (SSR) is considered. In this
kind of problems, it is very important to have a low level of secondary lobes to avoid false
interrogations. The radar under analysis consist on an array of 10×33 elements that is sur‐
rounded by the supporting structure, a primary radar, a circular supporting platform and
several lighting rods (see Figure 6). The working frequency is 1030 MHz so that the geome‐
try is discretized into 139, 762 RWGs.
Figure 6. First level decomposition for a secondary surveillance radar.
A two level CBFM is used to efficiently compute the matrix in (28). The first level blocks are
shown in Figure 6. After applying an SVD threshold equal to 10-3, the number of first-level
CBFs is only 20259.
A second level CBFM is applied on this first-level CBFs by grouping the blocks into five
macro blocks. The first one contains the SSR and the supporting structure, and the remain‐
ing four blocks contain the rest of the structure. The number of degrees of freedom after this
second level is reduced to 13312. In both cases, the number of plane waves to illuminates the
first- and second-level blocks is 6720.
In this example, a separable distribution of the feedings is assumed and, therefore, the eleva‐
tion and azimuth radiation patterns are assumed to be controlled by the feeding distribution
on the vertical and horizontal distribution. The vertical distribution is designed as in [25]
which considers the isolated structure and computes the vertical feeding to provide a cose‐
cant squared pattern. On the other hand, the azimuth elevation pattern, which must suit the
masks depicted in Figure 7, is expected to be affected by the surrounding environment and,
in special, by the lighting rods. Thus, the feedings computed in [25] are not expected to be
valid and they must be synthesized for the new structure. Hence, the matrix G has been
computed to relates the 33 voltages of the horizontal distribution with 181 sample points of
the azimuth radiation pattern.
Once the matrix is computed, many synthesis schemes can be efficiently applied. In this
book chapter, a genetic algorithm has been used to generated the proper radiation pattern
shown in Figure 7 but other optimization techniques could also be used [26]. The evaluation
of each generation of the genetic algorithm, which involves computing a hundred of radia‐
tion pattern, is accomplished in only 0.22 seconds. In opposite, the radiation pattern of a
usual hamming window distribution is also shown revealing that it is out of the mask.
Figure 7. Normalized radiation patterns for a hamming window feeding and a feeding computed by a genetic algo‐
rithm.
3.3 Partial modifications for the analysis of on-board antennas

The quick expansion of iterative schemes such as the fast multipole method has enabled us
to solve problems electrically larger than in the past. Nonetheless, some enhancements of
the MoM cannot be easily applied on these iterative schemes. Since the CBFM is an extend‐
ed version of the MoM in which the conventional low-level basis functions are replaced by
CBFs, then most of MoM enhancements can be easily adapted.
The partial solution of geometries [23,37,21,9] is one of these approaches that can be easily
applied to the CBFM. In this book chapter, we will focus on the partial solution based on the
Sherman-Morrison formula as described in [24].
In this technique, the structure is splitted into two parts. The first one is assumed to be fixed
and it will not be modified. The other part is assumed to be smaller than the first one and it
is assumed that it needs to be modified several times during the analysis. Thus, once the
fixed geometry is known, it is possible to carry out most of the steps to compute the inverse
of the entire impedance matrix so that the analysis of each modification can be resumed
from that point saving computational effort. Since the inverse of the entire matrix is in‐
volved, then it can only be applied to problems where this matrix fits in the available RAM
memory and, hence, the CBFM extends the range of applicability of the method to geome‐
tries electrically larger than the structures analyzed by the conventional MoM.
An example of application of these techniques could be the analysis of an antenna on an air‐

plane affected by the turbines rotation. In this problem, the airplane must be solved for dif‐
ferent rotation angles of the turbines in order to check the impact of them in the radiation
characteristics of it. In this case, the fixed structure would be the airplane without the tur‐
bines what comprises most of the total geometry, and the turbines areas are the parts that
must be modified.
In this book chapter, the technique is illustrated for a problem similar to the previously
described where the analysis of the radiation pattern of an monopole placed on an heli‐
copter affected by the blades rotation is accomplished. The geometry is shown in Fig‐
ure 8 and the monopole is placed at the beginning of the tail and parallel to it. The
problem is solved at 400, MHz so that the geometry is discretized into 96, 851 low-lev‐
el basis functions. After the application of the CBFM with 400 plane waves and a SVD
threshold equal to 10-3, the number of degrees of freedom for the total problem is 9837
with only 830 belonging to the blades.
The analysis of the structure with the CBFM involves 10, 200 s. On the other hand, the re‐
quired precomputations with the technique described in this section requires 10, 026 s. After
that, the analysis of each new blades position can be accomplished in only 865s. Thus, the
analysis with this technique becomes quickly more efficient than the use of the conventional
CBFM. Times are measured in a machine equipped with four Quad-Core AMD Opteron™
8378 at 2.4 GHz (only one core was used) and 128 GB of RAM. The radiation patterns for
several rotation angles of the main rotor are shown in Figure 8.
Finally, it is important to remark that other approaches can be used for similar purposes. For
example, antenna placement is considered in [27] by combining the CBFM with an incom‐
plete Gauss-Jordan elimination.
Figure 8. Gain on the cut = 90∘ for different rotation positions in the main rotor. The inset depicts the helicopter with
the blades at the zero angle.
4. Conclusions
The characteristic basis function method has been deeply described in this book chapter.
New perspectives have been provided to understand the generation of the characteristic ba‐
sis functions based on plane-waves revealing that it yields a complete set of basis functions
if the approximation of external field is accurate enough. Several speeds up have been com‐
pared for the case of filling the impedance matrix and the adaptive cross approximation
seems to be the most balanced one for a generic analysis.
As a consequence, the CBFM has been proved to be a rigorous method that achieved an iter‐
ation-free solution for problems electrically larger than the ones that can be analyzed with
the conventional method of moments. In addition, the compression of the matrix that the
CBFM carries out is compatible with many of the MoM enhancements. The most direct ap‐
plication is the solution of problems where multiple excitations have to be considered such
as monostatic radar cross section computations or characterization of antenna arrays. In ad‐
dition, techniques for efficiently analysis of small modifications on large structures have
been shown to be compatible with the CBFM.
5. Acknowledgments
This work has been supported by the Ministerio de Ciencia e Innovación of Spain /FEDER
under projects TEC2011-24492, CSD2008-00068; by the Gobierno del Principado de Asturias
(PCTI)/FEDER-FSE under project PC10-06; contract FUO-EM-221-10, and by Cátedra Tele‐
fónica-Universidad de Oviedo.
Author details
Jaime Laviada*, Fernando Las-Heras and Marcos R. Pino
Escuela Politécnica de ingeniería de Gijón, Universidad de Oviedo, Electrical Engineering

Department, Gijón, España
References
[1] Ayestarán, R.G., Laviada, J., & Las-Heras, F. (2009). Realistic antenna array synthesis
in complex environments using a MoM-SVR approach. J. of Electromagn. Waves and
Appl., 23, 97-108.
[2] Bebendorf, M. (June 2000). Approximation of boundary element matrices. Numen.

Math, 86(4), 565-589.
[3] Bebendorf, M., & Rjasanow, S. (March 2003). Adaptive low-rank approximation of
collocation matrices. Computing, 70(1), 1-24.
[4] Bilotti, F., & Vegni, C. (December 2001). Rigorous and efficient full-wave analysis of
trapezoidal patch antennas. IEEE Trans. Antennas Propagat., 49(12), 1773-1776.
[5] Bleszynski, E., Bleszynski, M., & Jaroszewicz, T. (Sep.-Oct. 1996). AIM: adaptive inte‐
gral method for solving large-scale electromagnetic scattering and radiation prob‐
lems. Radio Sci., 31(5), 1225-1251.
[6] Cabedo-Fabres, M., Antonino-Daviu, E., Valero-Nogueira, A., & Bataller, M. F. (Octo‐
ber 2007). The theory of characteristic modes revisited: A contribution to the design
of antennas for modern applications. IEEE Antennas Propagat. Mag., 49(5), 52-68.
[7] Chew, W. C., Jin, J. M., Michielssen, E., & Song, J. (2001). Fast and efficient algorithms
in computational electromagnetics. Artech House, Norwood, Massachusetts 02062,
USA.
[8] Coifman, R., Rokhlin, V., & Wandzura, S. (June 1993). The fast multipole method for
the wave equation: A pedestrian prescription. IEEE Trans. Antennas Propagat. Mag.,
53(3), 7-12.
[9] Cormos, D., Loison, R., & Gillard, R. (June 2005). Fast optimization and sensitivity
analysis of nonintuitive planar structures. IEEE Trans. Microwave Theory Tech., 53(6),
2019-2025.
[10] Craeye, Christophe. (October 2006). A fast impedance and pattern computation
scheme for finite antenna arrays. IEEE Trans. Antennas Propagat., 54(10), 3030-3034.
[11] Delgado, Carlos, Felipe Cátedra, Manuel, & Mittra, Raj. (March 2008). Application of
the characteristic basis function method utilizing a class of basis and testing func‐
tions defined on NURBS patches. IEEE Antennas Propagat., 56(3), 784-791.
[12] Eshrah, Islam A., & Kishk, Ahmed A. (June 2004). Application of the adaptive basis
functions/diagonal moment matrix technique to arrays of dielectric scatterers. Proc.
IEEE AP-S Int. Symp., 1, 647-650.
[13] Eshrah, Islam A., & Kishk, Ahmed A. (May 2004). Implementation of the adaptive
basis functions/diagonal moment matrix technique to array of identical elements.
Proc. URSI International Symposium on Electromagnetic Theory, 837-839.
[14] Eshrah, Islam A., & Kishk, Ahmed A. (March 2005). Analysis of linear arrays using
the adaptive basis functions/diagonal moment matrix technique. IEEE Trans. Anten‐
nas Propagat., 53(3), 1121-1125.
[15] Glisson, A. W., & Wilton, D. R. (September 1980). Simple and efficient numerical
methods for problems of electromagnetic radiation and scattering from surfaces.
IEEE Trans. Antennas Propagat., AP-28(5), 593-603.
[16] Golub, G. H., & Loan, C. F. van. (1989). Matrix computations. John Hopkins Univ.
Press, Baltimore, MD.
[17] Hansen, T. B., & Yaghjian, Arthur D. (1999). Plane-Wave Theory of time domain
fields. IEEE press, New York, NY.
[18] Harrington, R. F., & Mautz, J. R. (September 1971). Computation of characteristic
modes for conducting bodies. IEEE Trans. Antennas Propagat., AP-19, 629-639.
[19] Harrington, R. F., & Mautz, J. R. (September 1971). Theory of characteristic modes for
conducting bodies. IEEE Trans. Antennas Propagat., AP-19, 622-628.
[20] Harrington, R.F. (1993). Field Computation by Moment Method. IEEE press, New
York, NY.
[21] Johnson, J. Michael, & Rahmat-Samii, Yahya. (October 1999). Genetic algorithms and
method of moments ({GA/MOM}) for the design of integrated antennas. IEEE Trans.
Antennas Propagat., 47(10), 1606-1614.
[22] Kalbasi, Khalil, & Demarest, Kenneth R. (May 1993). A multilevel formulation of the
method of moments. IEEE Trans. Antennas Propagat., 41(5), 589-599.
[23] Kastner, R. (March 1989). An ‘add-on’ method for the analysis of scattering from
large planar structures. IEEE Trans. Antennas Propagat., 37(3), 353-361.
[24] Kastner, R. (January 1990). Analysis of microstrip antenna structures using the ‘add-
on’ technique. IEEE Trans. Antennas Propagat., 38(1), 114-117.
[25] Las-Heras, F., Galocha, B., & Besada, J. L. (July 1997). Equivalent source modelling
and reconstruction for antenna measurement and synthesis. Proc. IEEE AP-S Int.
Symp., 1, 156-159, Montreal, Canada.
[26] Laviada, J., Ayestarán, R.G., Pino, M. R., Las-Heras, F., & Mittra, R. (2009). Synthesis
of phased arrays in complex environments with the multilevel characteristic basis
function method. Progress In Electromagnetics Research, 92, 347-360.
[27] Laviada, J., Gutiérrez-Meana, J., Pino, M. R., & Las-Heras, F. (October 2010). A partial
solution of MoM matrices based on characteristic basis functions and its application
to on-board antennas positioning. Applied Computational Electromagnetics Society
(ACES) Journal, 25(10), 830-840.
[28] Laviada, J., Pino, M. R., Las-Heras, F., & Mittra, Raj. (October 2009). Solution of elec‐
trically large problems with multilevel characteristic basis functions. IEEE Trans. An‐
tennas Propagat., 57(10), 3189-3198.
[29] Laviada, J., Pino, M. R., Mittra, R., & Las-Heras, Fernando. (December 2009). Parallel‐
ized multilevel characteristic basis function method for solving electromagnetic scat‐
tering problems. Microwave Optical and Technology Letters, 51(12), 2963-2969.
[30] Laviada, J., Pino, Marcos R., Las-Heras, F., & Mittra, R. (August 2009). Interpolation
scheme for fast calculation of reaction terms in the characteristic basis function meth‐
od. Microwave Optical and Technology Letters, 51(8), 1818-1824.
[31] Losada, V., Boix, R. R., & Horno, M. (April 1999). Resonant modes of circular micro‐
strip patches in multilayered substrates. IEEE Trans. Antennas Propagat., 47, 488-498.
[32] Wei Bing, Lu, Cui, Tie Jun, Qian, Zhi Guo, Yin, Xiao Xing, & Hong, Wei. (November
2004). Accurate analysis of large-scale periodic structures using an efficient sub-en‐
tire-domain basis function method. IEEE Trans. Antennas Propagat., 52(11), 3078-3085.
[33] Lucente, Eugenio, Monorchio, Agostino, & Mittra, Raj. (April 2008). An iteration-free
MoM approach based on excitation independent characteristic basis functions for
solving large multiscale electromagnetic scattering problems. IEEE Antennas Propa‐
gat., 56(4), 999-1007.
[34] Maaskant, R., Mittra, R., & Tijhuis, A. (November 2008). Fast analysis of large anten‐
na arrays using the characteristic basis function method and the adaptive cross ap‐
proximation algorithm. IEEE Trans. Antennas Propagat., 56(11), 3440-3451.
[35] Matekovits, L., Laza, V. A., & Vechhi, G. (September 2007). Analysis of large complex
structures with the synthetic-functions approach. IEEE Trans. Antennas Propagat.,
55(9), 2509-2521.
[36] MathWorks. (2009). Matlab Getting Started Guide. http://www.mathworks.com.
[37] Misra, P., & Naishadham, K. (December 1996). Order-recursive gaussian elimination
(ORGE) and efficient CAD of microwave circuits. IEEE Trans. Microwave Theory Tech.,
44(12), 2166-2173.
[38] Ooms, S., & DeZutter, D. (March 1998). A new iterative diakoptics-based multilevel
moments method for planar circuits. IEEE Trans. Microw. Theory Tech., MTT-46(3),
280-291.
[39] Prakash, V.V.S., & Mittra, R. (January 2003). Characteristic basis function method: A
new technique for efficient solution of method of moments matrix equation. Micro‐
wave Opt. Technol. Lett., 36(2), 95-100.
[40] Rao, S. M., Wilton, D. R., & Glisson, A. W. (May 1982). Electromagnetic scattering by
surfaces of arbitrary shape. IEEE Trans. Antennas Propagat., AP-30, 409-418.
[41] Rousseau, P. R., & Burkholder, R. J. (October 1995). A hybrid approach for calculat‐
ing the scattering from obstacles within large, open cavities. IEEE Trans. Antennas
Propagat., 43(10), 1068-1075.
[42] Song, J. M., Lu, C. C., & Chew, W. C. (October 1997). Multilevel fast multipole algo‐
rithm for electromagnetic scattering by large complex objects. IEEE Trans. Antennas
Propagat., 45(10), 1488-1493.
[43] Suter, Eric, & Mosig, Juan R. (August 2000). A subdomain multilevel approach for
the efficient MoM analysis of large planar antennas. Microwave Opt. Technol. Lett.,
26(4), 270-277.
[44] Vegni, L., Toscano, A., & Popovski, B. (1997). Electromagnetic plane wave scattering
by large and finite strip array on dielectric slab. Ann. Telecommun., 52(3-4), 209-218.
[45] Waller, M. L., & Rao, S. M. (October 2002). Application of adaptive basis functions
for a diagonal moment matrix solution of arbitrarily shaped three-dimensional con‐
ducting body problems. IEEE Trans. Antennas Propagat., 50(10), 1445-1452.
Chapter 5
Analytical Grounds for Modern Theory of Two-

Dimensionally Periodic Gratings
L. G. Velychko, Yu. K. Sirenko and E. D. Vinogradova
1. Introduction
Rigorous models of one-dimensionally periodic diffraction gratings made their appearance

in the 1970s, when the corresponding theoretical problems had been considered in the con‐
text of classical mathematical disciplines such as mathematical physics, computational
mathematics, and the theory of differential and integral equations. Periodic structures are
currently the objects of undiminishing attention. They are among the most called-for disper‐
sive elements providing efficient polarization, frequency and spatial signal selection. Fresh
insights into the physics of wave processes in diffraction gratings are being implemented in‐
to radically new devices operating in gigahertz, terahertz, and optical ranges, into new ma‐
terials with inclusions ranging in size from micro- to nanometers, and into novel circuits for
in-situ man-made and natural material measurements.
However, the potentialities of classical two-dimensional models [1-7] are limited. Both theo‐
ry and applications invite further investigation of three-dimensional, vector models of peri‐
odic structures in increasing frequency. In our opinion these models should be based on
time-domain (TD) representations and implemented numerically by the mesh methods [8,9].
It follows from the well-known facts: (i) TD-approaches are free from the idealizations in‐
herent in the frequency domain; (ii) they are universal owing to minimal restrictions im‐
posed on geometrical and material parameters of the objects under study; (iii) they allow
explicit computational schemes, which do not require inversion of any operators and call for
an adequate run time when implementing on present-day computers; (iv) they result in data
easy convertible into a standard set of frequency-domain characteristics. To this must be
added that in recent years the local and nonlocal exact absorbing conditions (EAC) have
been derived and tested [6,7]. They allow one to replace an open initial boundary value
problem occurring in the electrodynamic theory of gratings with a closed problem. In addi‐
tion, the efficient fast Fourier transform accelerated finite-difference schemes with EAC for
characterizing different resonant structures have been constructed and implemented [10].
It is evident that the computational scheme solving a grating problem must be stable and
convergent, computational error must be predictable, while the numerical results are bound
to be unambiguously treated in physical terms. To comply with these requirements, it is im‐
portant to carry out theoretical analysis at each stage of the modeling (formulation of boun‐
dary value and initial boundary value problems, determination of the correctness classes for
them, study of qualitative characteristics of singularities of analytical continuation for solu‐
tions of model boundary value problems into a domain of complex-valued frequencies, etc.).
In the present work, we present a series of analytical results providing the necessary theoret‐
ical background to the numerical solution of initial boundary value problems as applied to
two-dimensionally periodic structures. Section 1 is an Introduction. In Section 2 we give
general information required to formulate a model problem in electrodynamic theory of gra‐
tings. Sections 3 and 4 are devoted to correct and efficient truncation of the computational
space in the problems describing spatial-temporal electromagnetic wave transformation in
two-dimensionally periodic structures. Some important characteristics and properties of
transient and steady-state fields in regular parts of the rectangular Floquet channel are dis‐
cussed in Sections 5 and 7. In Section 6, the method of transformation operators (the TD-ana‐
log of the generalized scattering matrix method) is described; by applying this method the
computational resources can be optimized when calculating a multi-layered periodic struc‐
ture or a structure on a thick substrate. In Section 8, elements of spectral theory for two-di‐
mensionally periodic gratings are given in view of its importance to physical analysis of
resonant scattering of pulsed and monochromatic waves by open periodic resonators.
2. Fundamental Equations, Domain of Analysis, Initial and Boundary

Conditions
Space-time and space-frequency transformations of electromagnetic waves in diffraction

gratings, waveguide units, open resonators, radiators, etc. are described by the solutions of
initial boundary value problems and boundary value problems for Maxwell’s equations. In
this chapter, we consider the problems of electromagnetic theory of gratings resulting from
the following system of Maxwell’s equations for waves propagating in stationary, locally in‐
homogeneous, isotropic, and frequency dispersive media [9,11]:
→ →
→ ∂ E (g, t ) + χε (g, t ) ∗ E (g, t ) → →
rotH (g, t ) = η0−1 + χσ (g, t ) ∗ E (g, t ) + j (g, t ), (1)
∂t
→ →
→ ∂ H (g, t ) + χμ (g, t ) ∗ H (g, t )
rotE (g, t ) = − η0 , (2)
∂t
where
Analytical Grounds for Modern Theory of Two-Dimensionally Periodic Gratings 125
g = { x, y, z } is the point in a three-dimensional spaceR 3;
x, y, and z are the Cartesian coordinates;

→ →
E (g, t ) = {Ex , Ey , Ez } and H (g, t ) = {H x , H y , H z } are the electric and magnetic field vectors;
η0 = (μ0 / ε0)1/2 is the intrinsic impedance of free space;
ε0 and μ0 are permittivity and permeability of free space;

→
j (g, t ) is the extraneous current density vector;
χε (g, t ), χμ (g, t ), and χσ (g, t ) are the electric, magnetic, and specific conductivity susceptibil‐
∫
ities; f 1(t ) ∗ f 2(t ) = f 1(t − τ ) f 2(τ )dτstands for the convolution operation.
We use the SI system of units. From here on we shall use the term “time” for the parametert,
which is measured in meters, to mean the product of the natural time and the velocity of
light in vacuum.
With no frequency dispersion in the domainG ⊂ R 3, for the points g ∈ G we have
χε (g, t ) = δ (t ) ε (g ) − 1 , χμ (g, t ) = δ (t ) μ (g ) − 1 , χσ (g, t ) = δ (t )σ (g ),
where δ (t ) is the Dirac delta-function;ε (g ), μ (g ), and σ (g ) are the relative permittivity, rela‐
tive permeability, and specific conductivity of a locally inhomogeneous medium, respective‐
ly. Then equations (1) and (2) take the form:
→
→ ∂ E (g, t ) → →
rotH (g, t ) = η0−1ε (g ) + σ (g )E (g, t ) + j (g, t ), (3)
∂t
→
→ ∂ H (g, t )
rotE (g, t ) = − η0μ (g ) . (4)
∂t
In vacuum, where ε (g ) = μ (g ) = 1 andσ (g ) = 0, they can be rewritten in the form of the follow‐
ing vector problems [6]:
∂2 → → ∂ → →
Δ − grad div − ∂ t2
E (g, t ) = F E (g, t ), ∂t H (g, t ) = − η0−1rot E (g, t ) ,
(5)
→ ∂ →
F E (g, t ) = η0 ∂ t j (g, t )
or
∂2 → → ∂ → → →
Δ − ∂ t 2 H (g, t ) = F H (g, t ), η0−1 ∂ t E (g, t ) = rotH (g, t ) − j (g, t ) ,
→ → (6)
F H (g, t ) = − rot j (g, t ) .
→
By Δ we denote the Laplace operator. As shown in [6], the operator grad divE can be omit‐
ted in (5) from the following reasons. By denoting the volume density of induced and exter‐
→
nal electric charge through ρ1(g, t ) and ρ2(g, t ), we can write grad divE = ε0−1grad(ρ1 + ρ2). In
homogeneous medium, where ε and σ are positive and non-negative constants, we have
ρ1(g, t ) = ρ1(g,0)exp( − tσ / ε ), and if ρ1(g,0) = 0, then ρ1(g, t ) = 0 for anyt > 0. The remaining term
ε0−1gradρ2 can be moved to the right-hand side of (5) as a part of the function defining cur‐
rent sources of the electric field.
To formulate the initial boundary value problem for hyperbolic equations (1)-(6) [12], ini‐
tial conditions at t = 0 and boundary conditions on the external and internal boundaries of
the domain of analysis Q should be added. In 3-D vector or scalar problems, the domain Q
is a part of the R 3-space bounded by the surfaces S that are the boundaries of the domains
¯ . In the so-called open problems, the do‐
intS, filled with a perfect conductor:Q = R 3 \ intS
main of analysis may extend to infinity along one or more spatial coordinates.
The system of boundary conditions for initial boundary value problems is formulated in the
following way [11]:
• on the perfectly conducting surface S the tangential component of the electric field vector
is zero at all times t
Εtg (g, t ) | g ∈S = 0 for t ≥ 0; (7)
the normal component of the magnetic field vector on S is equal to zero (H nr (g, t ) | g ∈S = 0),
and the function H tg (g, t ) | g ∈S defines the so-called surface currents generated on S by the
external electromagnetic field;
• on the surfacesS ε,μ,σ , where material properties of the medium have discontinuities, as
well as all over the domainQ, the tangential components Etg (g, t ) and H tg (g, t ) of the
electric and magnetic field vectors must be continuous;
• in the vicinity of singular points of the boundaries ofQ, i.e. the points where the tangents
and normals are undetermined, the field energy density must be spatially integrable;
→ →
• if the domain Q is unbounded and the field {E (g, t ), H (g, t )} is generated by the sources
having bounded supports in Q then for any finite time interval (0,T ) one can construct a
closed virtual boundary M ⊂ Q sufficiently removed from the sources such that
→ →
{E (g, t ), H (g, t )} | g ∈M ,t ∈(0,T ) = 0. (8)
The initial state of the system is determined by the initial conditions att = 0. The reference
→ → →
states E (g,0) and H (g,0) in the system (1), (2) or the system (3), (4) are the same as E (g,0)
→ → →
and ∂ E (g, t ) / ∂ t | t =0 (H (g,0)and ∂ H (g, t ) / ∂ t | t =0) in the differential forms of the second
→ →
order (in the terms oft), to which (1), (2) or (3), (4) are transformed if the vector H (vectorE )
is eliminated (see, for example, system (5), (6)). Thus, (5) should be complemented with the
initial conditions
→ →
Ε (g,0) = φ (g ),
∂ →
∂t
Ε (g, t ) | t =0
→
= ψ (g), g ∈ Q̄. (9)
→ → →
The functionsφ (g ), ψ (g ), and F (g, t ) (called the instantaneous and current source functions)
usually have limited support in the closure of the domainQ. It is the practice to divide cur‐
rent sources into hard and soft [9]: soft sources do not have material supports and thus they
are not able to scatter electromagnetic waves. Instantaneous sources are obtained from the
→ → →
pulsed wave U i (g, t ) exciting an electrodynamic structure: φ (g ) = U i (g,0)and
→( ) →i →i
ψ g = ∂ U (g, t ) / ∂ t | t =0. The pulsed signal U (g, t ) itself should satisfy the corresponding
wave equation and the causality principle. It is also important to demand that the pulsed
signal has not yet reached the scattering boundaries by the momentt = 0.
The latter is obviously impossible if infinite structures (for example, gratings) are illuminat‐
ed by plane pulsed waves that propagate in the direction other than the normal to certain
infinite boundary. Such waves are able to run through a part of the scatterer’s surface by
any moment of time. As a result a mathematically correct modeling of the process becomes
impossible: the input data required for the initial boundary value problem to be set are de‐
fined, as a matter of fact, by the solution of this problem.
3. Time Domain: Initial Boundary Value Problems
The vector problem describing the transient states of the field nearby the gratings whose ge‐
ometry is presented in Figure 1 can be written in the form
r r
ì r ¶ é E ( g , t ) + ce ( g , t ) * E ( g , t ) ùû r r
=ïrotH ( g , t ) h0 -1 ë + cs ( g , t ) * E ( g , t ) + j ( g , t ) ,
ï ¶t
ï r r
ï r ¶ é H ( g , t ) + c m ( g , t ) * H ( g , t ) ùû
írotE ( g , t ) = -h0 ë , g= { x, y, z} Î Q, t > 0 (10)
ïr ¶t
r r r
=ï Ε ( g ,0 ) j E ( g ) , = H ( g ,0 ) j H ( g ), g Î Q
ï
ï Εtg ( g , t=
) 0, H nr ( g , = t ) gÎS 0, t ³ 0 .
î g ÎS
Here, Q̄is the closure ofQ, χε,μ,σ (g, t )are piecewise continuous functions and the surfaces S
are assumed to be sufficiently smooth. From this point on it will be also assumed that the
continuity conditions for tangential components of the field vectors are satisfied, if required.
¯ occupies a great deal of the R 3-space. The problem for‐
The domain of analysis Q = R 3 \ intS
mulated for that domain can be resolved analytically or numerically only in two following
cases.
Figure 1. Geometry of a two-dimensionally periodic grating.
• The problem (10) degenerates into a conventional Cauchy problem (intS ¯ = ∅ , the medium
→ →
is homogeneous and nondispersive, while the supports of the functionsF (g, t ), φ (g ), and
→
ψ (g) are bounded). With some inessential restrictions for the source functions, the classi‐
cal and generalized solution of the Cauchy problem does exist; it is unique and is descri‐
bed by the well-known Poisson formula [12].
→ → →
• The functionsF (g, t ), φ (g ), and ψ (g) have the same displacement symmetry as the period‐
ic structure. In this case, the domain of analysis can be reduced to
Q N = {g ∈ Q : 0 < x < lx ; 0 < y < ly }, by adding to problem (10) periodicity conditions [7] on
lateral surfaces of the rectangular Floquet channelR = {g ∈ R 3 : 0 < x < lx ; 0 < y < ly }.
The domain of analysis can also be reduced to Q N in a more general case. The objects of
analysis are in this case not quite physical (complex-valued sources and waves). However,
by simple mathematical transformations, all the results can be presented in the customary,
physically correct form. There are several reasons (to one of them we have referred at the
end of Section 3) why the modeling of physically realizable processes in the electromagnetic
theory of gratings should start with the initial boundary value problems for the images
f N
(g, t, Φx , Φy ) of the functions f (g, t ) describing the actual sources:
∫ ∫ f̃ (z, t, Φ , Φ )exp(2πiΦ lx )exp(2πiΦ ly )dΦ dΦ

∞ ∞
f (g, t ) = x y x y x y
x y
−∞ −∞
∞ ∞
(11)
= ∫∫f
−∞ −∞
N
(g, t, Φx , Φy )dΦx dΦy
From (11) it follows that
f N { ∂∂f x }(x + l , y, z, t, Φ , Φ ) = e
N
x x y
2πiΦx
f N { ∂∂f x }(x, y, z, t, Φ , Φ ),
N
x y
f N { ∂∂f y }(x, y + l , z, t, Φ , Φ ) = e
N
y x y
2πiΦy
f N { ∂∂fy }(x, y, z, t, Φ , Φ )
N
x y
or, in other symbols,
D f N
( x + lx , y ) = e2πiΦx D f N ( x, y ),D f N
( x, y + ly ) = e2πiΦy D f N ( x, y ).
The use of the foregoing conditions truncates the domain of analysis to the domainQ N , which
is a part of the Floquet channelR, and allows us to rewrite problem (10) in the form
∞ ∞ ∞ ∞
→
E (g, t ) = ∫∫
−∞ −∞
→ →
E N (g, t, Φx , Φy )dΦx dΦy , H (g, t ) = ∫ ∫ H→
−∞ −∞
N
(g, t, Φx , Φy )dΦx dΦy (12)
and
r r
ì r ¶ é E N ( g , t ) + ce ( g , t ) * E N ( g , t ) ùû r r
=ïrotH N ( g , t ) h0 -1 ë + cs ( g , t ) * E N ( g , t ) + j N ( g , t ) ,
ï ¶t
ï r r
ïrotEr N g , t = ¶ é H N ( g , t ) + c m ( g , t ) * H N ( g , t ) ùû
( ) -h0 ë , g Î QN , t > 0
ï ¶t
ïï r N rN r r
= í Ε ( g ,0 ) j E ( g ) , = H N ( g ,0 ) j HN ( g ), g Î QN (13)
ï rN r N rN r N
= ( ) 2 i F
( )
ï D éë E H ùû ( lx , y ) e x D éë E H ùû ( 0, y ) , 0 £ y £ l y
p
ï r r r r
ï ë û ( ) ë û ( )
ï D é E N H N ù ( x, l y ) = e 2p iF y D é E N H N ù ( x,0 ) , 0 £ x £ lx
ïΕ N ( g,=t ) 0, H N
( g , t
= ) 0, t ³0.
îï tg g ÎS nr g ÎS
It is known [6-8] that initial boundary value problems for the above discussed equations
can be formulated such that they are uniquely solvable in the Sobolev spaceW 21(Q T ),
where Q T = Q × (0,T ) and0 ≤ t ≤ T . On this basis we suppose in the subsequent discussion
that the problem (13) for all t ∈ 0,T has also a generalized solution from the space
W 21(Q N ,T ) and that the uniqueness theorem is true in this space. Here symbol × stands
for the operation of direct product of two sets, (0,T )and 0,T are open and closed inter‐
→
vals, W mn (G )is the set of all elements f (g ) from the space L m(G ) whose generalized deriv‐
atives up to the order n inclusive also belong toL m(G ). L m(G )is the space of the functions
→
f (g ) = { f x , f y , f z } (forg ∈ G) such that the functions | f ...(g ) | m are integrable on the do‐
mainG.
4. Exact Absorbing Conditions for the Rectangular Floquet Channel
In this section, we present analytical results relative to the truncation of the computational
space in open 3-D initial boundary value problems of the electromagnetic theory of gratings.
In Section 3, by passing on to some special transforms of the functions describing physically
realizable sources, the problem for infinite gratings have been reduced to that formulated in
the rectangular Floquet channel R or, in other words, in the rectangular waveguide with
quasi-periodic boundary conditions. Now we perform further reduction of the domain Q N
to the region QLN = {g ∈ Q N : | z | < L } (all the sources and inhomogeneities of the Floquet
channel R are supposedly located in this domain). For this purpose the exact absorbing con‐
ditions [6,7,10,13,14] for the artificial boundaries L ± (z = ± L ) of the domain QLN will be con‐
structed such that their inclusion into (13) does not change the correctness class of the
→ →
problem and its solutionE N (g, t ),H N (g, t ).
From here on we omit the superscripts N in (13). By applying the technique similar to that
→
described in [13,14], represent the solution E (g, t ) of (13) in the closure of the domains
A = {g ∈ R : z > L } and B = {g ∈ R : z < − L } in the following form:
→ ∞
→
E (g, t ) = ∑ u ±nm(z, t )μnm( x, y ), { x, y } ∈ R̄ z , t ≥ 0, (14)
n,m=−∞
where the superscript ‘+ ’ corresponds to z ≥ L and ‘− ’ to z ≤ − L and the following notation

is used:
Rz = (0 < x < lx ) × (0 < y < ly );
{μnm( x, y )} (n, m = 0, ± 1, ± 2,...) is the complete in L 2(Rz ) orthonormal system of the functions
μnm( x, y ) = (lx ly )−1/2exp(iαn x )exp(iβm y );
αn = 2π (Φx + n ) / lx , βm = 2π (Φy + m) / ly , andλnm

2
= αn2 + βm2.
→
The space-time amplitudes u ±nm(z, t ) satisfy the equations
{ →
−
∂2
+
∂2
u ±n m(z,0) = 0,
→
− λ 2 u ± (z, t ) = 0, t > 0
∂t 2 ∂z 2 n m n m
∂ →±
u (z, t )
∂t n m
| t =0
=0
, { z≥L
z≤ −L
} . (15)
Equations (14) and (15) are obtained by separating variables in the homogeneous boundary
→
value problems for the equation Δ − ∂2 / ∂ t 2 E (g, t ) = 0 (see formula (5)) and taking into ac‐
→ →
count that in the domains A and B we have grad div E (g, t ) = 0 andF E (g, t ) = 0. It is also as‐
sumed that the field generated by the current and instantaneous sources located in QL has
not yet reached the boundaries L ± by the moment of timet = 0.
→
The solutions u ±nm(z, t ) of the vector problems (15), as well as in the case of scalar problems
[13,14], can be written as
t
→
∫ →
u ±nm(± L , t ) = ∓ J 0 λnm(t − τ ) u ±nm ′(± L , τ )dτ, t ≥ 0 .
0
(16)
→
The above formula represents nonlocal EAC for the space-time amplitudes of the field E (g, t )
in the cross-sections z = ± L of the Floquet channelR. The exact nonlocal and local absorb‐
→
ing conditions for the field E (g, t ) on the artificial boundaries L ± follow immediately from
(16) and (14):
→
E (x, y, ± L , t)
{∫ →
| }
t lx ly
∫ ∫ ∂ E (x̃,∂ỹ,z z, τ)
∞
=∓ ∑ J 0 λnm(t − τ ) ∗(
μnm x̃, ỹ )d x̃d ỹ dτ μnm( x, y ), (17)
n,m=−∞ z=±L
0 00
{ x, y } ∈ R̄ z , t ≥0
and
r
2 ¶ WE ( x, y, t ,j )
p 2
r
=E ( x, y , ± L, t ) ò dj , { x, y} Î Rz , t³0
p 0 ¶t
r
ìé ¶ 2 2 æ ¶
2
¶ 2 öù r ¶E ( g , t )
ï ê 2 - sin j ç 2 + 2 = ÷ ú WE ( x, y, t ,j ) m , { x, y} Î Rz , t >0
ïë ¶t è ¶ x ¶ y øû ¶z z =± L
ï r (18)
ïï r ¶WE ( x, y, t ,j )
íWE ( x,=y, t ,j )
t =0 ¶t
= 0, { x, y} Î Rz
ï t =0
=ï D éWr ù ( l , y ) e 2p iF x D éWr ù ( 0, y ) , 0 £ y £ l
ï ë E û x ë E û y
ï ér ù r
= D W
îï ë E û ( x , l y ) e
2p i F y
D éW
ë Eû ù ( x ,0 ) , 0 £ x £ l x, t³0 .
→ →
Here, u ±nm ′(± L , τ ) = ∂ u ±nm(z, τ ) / ∂ z | z=±L , J 0(t )is the zero-order Bessel function, the super‐
→
script ‘∗ ’ stands for the complex conjugation operation, W E ( x, y, t, φ )is some auxiliary
function, where the numerical parameter φ lies in the range0 ≤ φ ≤ π / 2.
→
It is obvious that the magnetic field vector H (g, t ) of the pulsed waves
→ → →
U (g, t ) = {E (g, t ), H (g, t )}outgoing towards the domains A and B satisfies similar boundary
→ →
conditions onL ±. The boundary conditions for E (g, t ) and H (g, t ) (nonlocal or local) taken
together reduce the computational space for the problem (13) to the domain QL (a part of
the Floquet channelR) that contains all the sources and obstacles.
→ → →
Now suppose that in addition to the sources j (g, t ), φ E (g ), andφ H (g ), there exist sources
→A → →
j (g, t ), φ AE (g ), and φ AH(g ) located in A and generating some pulsed wave
→i → →
U (g, t ) = {E i (g, t ), H i (g, t )} being incident on the boundary L + at timest > 0. The field
→
U i (g, t ) is assumed to be nonzero only in the domainA. Since the boundary conditions (17),
(18) remain valid for any pulsed wave outgoing through L ± towards z = ± ∞ [13,14], then the
→ →
total field {E (g, t ), H (g, t )}is the solution of the initial boundary value problem (13) in the
domain QL with the boundary conditions (17) or (18) on L − and the following conditions on
the artificial boundaryL +:
→
E s (x, y, L , t)
{∫ →s
| }
t lx ly
∫ ∫ ∂ E (x̃,∂ zỹ, z, τ)
∞
=− ∑ J 0 λnm(t − τ ) ∗(
μnm x̃, ỹ )d x̃d ỹ dτ μnm( x, y ), (19)
n,m=−∞ z=L
0 00
{ x, y } ∈ R̄ z , t ≥0
or
r
2 ¶ WE ( x, y, t ,j )
p 2
r
E s ( x, y , L, t ) { x, y} Î Rz ,
p ò0
= dj , t³0
¶t
r
ìé ¶ 2 2 æ ¶
2
¶ 2 öù r ¶E s ( g , t )
ï ê 2 - sin j ç 2 + 2 ÷ ú WE ( x, y, t ,j ) = - , { x, y} Î Rz , t >0
ï ë ¶t è ¶x ¶y ø û ¶z z=L
ï r (20)
ïï r ¶WE ( x, y, t ,j )
íWE ( x,=y, t ,j )
t =0
= 0, { x, y} Î Rz
¶t
ï t =0
r r
ï D éW ù ( l , y ) e 2p iF x D éW ù ( 0, y ) , 0 £ y £ l
=
ï ë Eû x ë Eû y
ï ér ù r
ïî D ëWE û ( x, l y ) e D éëWE ùû ( x,0 ) , 0 £ x £ lx , t ³ 0 .
2p i F y
=
→ → → → →
Here U s (g, t ) = {E s (g, t ), H s (g, t )} = U (g, t ) − U i (g, t ) (g ∈ A,t > 0) is the pulsed wave outgo‐
→
ing towardsz = + ∞. It is generated by the incident wave U i (g, t ) (‘reflection’ from the virtual
→ → →
boundaryL +) and the sources j (g, t ), φ E (g ), andφ H (g ).
5. Some Important Characteristics of Transient Fields in the Rectangular

Floquet Channel
For numerical implementation of the computational schemes involving boundary condi‐

→
tions like (19) or (20), the function U i (g, t ) for t ∈ 0,T and its normal derivative with re‐
→
spect to the boundary L +are to be known. To obtain the required data for the wave U i (g, t )
→ → →
generated by a given set of sources j A(g, t ), φ AE (g ), andφ AH(g ), the following initial boundary
value problem for a regular hollow Floquet channelR are to be solved:
r rA rA
ìé ¶ 2 ù ïì E i ïü ïìh0¶j ¶t + e0 grad r2 ïü ïì FE ïü
-1 A
ï ê - 2 + D ú= r
í iý í rA = ý í r A ý= , g { x, y, z} Î R, t > 0
ï ë ¶t û îï H þï îï- rot j þï îï FH þï
ï r r r
ïì E i ( g ,0 ) ïü ïìjr A üï ì¶E i ( g , t ) ¶t = h0 rot H i ( g ,0 ) üï ïìyr A ïü
ïïí r = E
,
ï t =0
=
E
ý í r A ý, g Î R
í H i ( g ,0 ) ý íïjr A ïý í ri
¶ H ( g , t ) ¶ t =- h -1
rot
ri
E ( g ,0 ) ïy ï (21)
ïîï ï
þ î H þ ï
î t =0 0 ï
þ î Hþ
ï ri r i ri r i
= ( )
ï D éë E H ùû ( lx , y ) e
2p i F x
ë ( )
D é E H ù ( 0, y ) , 0 £ y £ l y
û
ï ri r i r r
=
î ë ( )
ï D é E H ù ( x, l y ) e
û
2p i F y
ë ( )
D é E i H i ù ( x,0 ) , 0 £ x £ lx , t ³ 0 .
û
The function ρ2A(g, t ) here determines the volume density of foreign electric charge.
→ →
First we determine the longitudinal components Ezi and H zi of the field {E i , H i } at all points
gof the domain R for all timest > 0. Let us consider the scalar initial boundary value prob‐
lems following from (21):
ìé ¶ 2 ù ìï E üï ìï Fz , E üï
i A
ïê- 2 = + D ú í zi ý í A ý , g Î R , t >0
ïë ¶ t û îï H z þï îï Fz , H þï
ï
ïì E i ( g ,0 ) ü ìj A ü ì¶Ezi ( g , t ) ¶t ü ìy A ü
ï=ï z ï ï z,E ï ï t =0 ï ï z,E ï
í
í H i g ,0 ý í ý , í = ý í ý, g Î R (22)
ïîï z ( )þï îïj zA, H þï îï¶H zi ( g , t ) ¶t t = 0 þï îïy zA, H þï
ï
ï D éë Ez ( H z ) ùû ( lx , y ) = e x D éë Ez ( H z ) ùû ( 0, y ) , 0 £ y £ l y
i i 2p i F i i
ï
ï D é Ezi ( H zi ) ù ( x, l y ) e y D é Ezi ( H zi ) ù ( x,0 ) , 0 £ x £ lx , t ³ 0 .
=
2 p i F
î ë û ë û
By separating of the transverse variables x and y in (22) represent the solution of the prob‐
lem as
{ Ezi (g, t )
H zi (g, t)
} {= ∑
∞
n,m=−∞
vnm(z,E )(z, t )
vnm(z,H )(z, t )
} μnm( x, y ) (23)
To determine the scalar functions vnm(z,E )(z, t ) andvnm(z,H )(z, t ), we have to invert the follow‐
ing Cauchy problems for the one-dimensional Klein-Gordon equations:
ìé 2 ìvnm ( z , E ) ( z , t ) üï ìï FnmA ( z , E ) üï
ï ê - ¶ + ¶ - lnm ùú ïí=
2
ý í A ý , t > 0, - ¥ < z < ¥

ï ë ¶t 2 ¶z 2 û ïîvnm ( z , H ) ( z , t ) ïþ ïî Fnm ( z , H ) ïþ
ï
ï ìv
ïï nm ( z , E ) ( z ,0 ) ïü ìïj nm ( z , E ) üï ¶ ïìvnm ( z , E ) ( z , t ) ïü ìïy nm ü
A A
(z,E) ï
íí ý=í A ý, í ý =í A ý, (24)
ïïîvnm ( z , H ) ( z ,0 ) ïþ ïîj nm ( z , H ) ïþ ¶ t ïvnm ( z , H ) ( z , t ) ï
î þ t = 0 ïîy nm ( z , E ) ïþ
ï
ï - ¥ < z < ¥, n, m= 0, ±1, ±2,... .
ï
ï
î
A A A A A A
HereF nm ( z,E ), φnm( z,E ), ψnm( z,E )andF nm( z,H ), φnm( z,H ), ψnm( z,H )are the amplitudes of the Fourier
A
transforms of the functionsF z,E A
, φz,E A
, ψz,E A
andF z,H A
, φz,H A
, ψz,H in the basic set{μnm( x, y )}.
A A
Let us continue analytically the functionsvnm(z,E )(z, t ), vnm(z,H )(z, t )andF nm ( z,E ), F nm( z,H )by
zero on the semi-axis t0 and pass on to the generalized formulation of the Cauchy problem
(24) [12]:
B (λnm)
vnm(z,E )(z, t )
vnm(z,H )(z, t )
≡ −
∂2
∂t 2 +
∂2
∂z
2
2 − λnm v {
vnm(z,E )(z, t )
( )
nm( z,H ) z, t
}
= { } { } { }{ }
A
F nm
A
F nm
( z,E )
( z,H )
( )
− δ 1 (t )
A
φnm
A
φnm
( z,E )
( z,H )
− δ (t )
A
ψnm
A
ψnm
( z,E )
( z,H )
=
f nm(z,E )
f nm(z,H )
,
(25)
− ∞ < z < ∞, − ∞ < t < ∞, n, m = 1, ± 2, ± 3,... ,
( )
where δ (t ) and δ m (t ) are the Dirac delta-function and its derivative of the orderm. Taking
into account the properties of the fundamental solution
G (z, t, λ ) = − (1 / 2)χ (t − | z | ) J 0(λ t 2 − z 2) of the operator B(λ) [6,13,14] (χ (t )is the Heaviside
step function), the solutions vnm(z,E )(z, t ) and vnm(z,H )(z, t ) of equations (25) can be written as
{ vnm(z,E )(z, t )
vnm(z,H )(z, t )
} = G (z, t, λnm) ∗ { f nm(z,E )
f nm(z,H )
} =
({ } { } { })
t−|z−ω | ∞ A A A (26)
∫ ∫ J (λ
F nm φnm ψnm
−
1
2 0 nm
(t − τ )2 − (z − ω )2 ) A
( z,E )
( )
− δ 1 (t ) A
( z,E )
− δ (t ) A
( z,E )
dωdτ,
−∞ −∞
F nm ( z,H ) φnm ( z,H ) ψnm ( z,H )
− ∞ < z < ∞, t ≥ 0, n, m = 1, ± 2, ± 3,... .
Relations (23) and (26) completely determine the longitudinal components of the field
{E→ i , H
→i
}.
Outside the bounded domain enclosing all the sources, in the domainG ⊂ R, where the
waves generated by these sources propagate freely, the following relations [6,14] are valid:
{ →
Ei =
η H =(
0
→
(
∂2 U E ∂2 U H →
∂ x∂ z
i∂ U
−
∂ y∂ t
∂ y∂ t
+
∂ U2
) (
x+
∂ x∂ z
∂2 U E ∂2 U H →
∂ y∂ z
E
+
∂ x∂ t
2 H
y+
∂z2
) (
∂2 U E ∂2 U E →
−
∂t 2
z
)x→ + ( − ∂∂ x∂U t + ∂∂ y∂U z ) y→ + ( ∂ ∂Uz

2 E 2 H 2
2
H
−
)
∂2 U H →
∂t 2
z )
(27)
in which
∞
U E ,H (
g, t ) = ∑ E ,H (
unm z, t )μnm( x, y ) (28)
n,m=−∞
are the scalar Borgnis functions such that Δ − ∂2 / ∂ t 2 ∂ U E ,H (g, t ) / ∂ t = 0. Equations (23),
→ →
(26)-(28) determine the field {E i , H i } at all points g of the domain G for all timest > 0. Really,
since at the time point t = 0 the domain G is undisturbed, then we have Δ − ∂2 / ∂ t 2 U E ,H
=0
(g ∈ G,t > 0). Hence, in view of (27), (28), it follows:
Ez =
∂2 U E ∂2 U E
∂z 2 −
∂t 2 =−
∂x2
+
∂y2
(
∂2 U E ∂2 U E ∞
= ∑ λnm
n,m=−∞
2 E
unm μnm, )
η0H z =
∂2 U H ∂2 U H
∂z 2 −
∂t 2 =−
∂2 U H ∂2 U H
∂x2
+
∂y2
(∞
= ∑ λnm
n,m=−∞
2 H
unm μnm )
and (see representation (23))
E (
unm z, t ) = (λnm)−2vnm(z,E )(z, t ), unH (z, t ) = η0(λnm)−2vnm(z,H )(z, t ). (29)
→ →
Hence the functions U E ,H (
g, t ) as well as the transverse components of the field {E i , H i } are
determined.
The foregoing suggests the following important conclusion: the fields generated in the re‐
flection zone (the domainA) and transmission zone (the domainB) of a periodic structure are
uniquely determined by their longitudinal (directed along z-axis) components and can be
represented in the following form (see also formulas (14) and (23)). For the incident wave we
have
{ Ezi (g, t )
H zi (g, t)
} { = ∑
∞
n,m=−∞
vnm(z,E )(z, t )
vnm(z,H )(z, t )
}μnm( x, y ), g ∈ Ā, t ≥ 0, (30)
→ → →
for the reflected wave U s (g, t )(which coincides with the total field U (g, t ) ifU i (g, t ) ≡ 0) we
have
{ Ezs (g, t )orEz (g, t )

H zs (g, t )orH z (g, t )
} {
= ∑
∞
n,m=−∞
+
unm
+
unm
( z,E )( z, t )
( z,H )( z, t )
} μnm( x, y ), g ∈ Ā, t ≥ 0 (31)
→
and for the transmitted wave (coinciding in the domain B with the total fieldU (g, t )) we can
write
{ Ez (g, t )
H z (g, t )
} = ∑
∞
n,m=−∞
{ −
unm
−
unm
( z,E )( z, t )
( z,H )( z, t )
}
μnm( x, y ), g ∈ B̄, t ≥ 0. (32)
In applied problems, the most widespread are situations where a periodic structure is excit‐
ed by one of the partial components of TE-wave (withEzi (g, t ) = 0) or TM -wave (with
H zi (g, t ) = 0) [7]. Consider, for example, a partial wave of orderpq. Then we have
→ →
U i (g, t ) = U ipq (H )(g, t ) : H zi (g, t ) = v pq (z,H )(z, t )μ pq ( x, y )
or
→ →
U i (g, t ) = U ipq (E )(g, t ) : Ezi (g, t ) = v pq (z,E )(z, t )μ pq ( x, y ).
The excitation of this kind is implemented in our models in the following way. The time
function v pq (z,H )( L , t ) or v pq (z,E )( L , t )is defined on the boundaryL +. This function deter‐
→
mines the width of the pulseU i (g, t ), namely, the frequency range K 1, K 2 such that for all
frequencies k from this range (k = 2π / λ, λis the wavelength in free space) the value
| ṽ pq (z,H orE )( L , k ) |
γ=
max | ṽ pq (z,H orE )( L , k ) |
k∈ K 1;K 2
where ṽ pq (z,H orE )( L , k ) is the spectral amplitude of the pulsev pq (z,H orE )( L , t ), exceeds some
given valueγ = γ0. All spectral characteristics f̃ (k ) are obtainable from the temporal charac‐
teristics f (t ) by applying the Laplace transform
∞ iα+∞
∫ ikt
f̃ (k ) = f (t )e dt ↔ f (t ) =
0
1
2π ∫
iα−∞
f̃ (k )e−ikt dk, 0 ≤ α ≤ Imk. (33)
For numerical implementation of the boundary conditions (19) and (20) and for calculating
→
space-time amplitudes of the transverse components of the wave U i (g, t ) in the cross-sec‐
tion z = L of the Floquet channel (formulas (27) and (29)), the function (v pq (z,H orE ))′( L , t ) are
to be determined. To do this, we apply the following relation [7,14]:
t
→
∫ →
v pq (H orE )( L , t ) = J 0 λ pq (t − τ ) (v pq (H orE ))′( L , t )dτ, t ≥ 0.
0
(34)
→
which is valid for all the amplitudes of the pulsed wave U i (g, t ) outgoing towards z = − ∞
and does not violate the causality principle.
6. Transformation Operator Method
6.1. Evolutionary basis of a signal and transformation operators

Let us place an arbitrary periodic structure of finite thickness between two homogeneous di‐
electric half-spaces z1 = z − L > 0 (withε = ε1) and z2 = − z − L > 0 (withε = ε2). Let also a local co‐
ordinate system gj = { xj , yj , zj } be associated with each of these half-spaces (Figure 2).
Assume that the distant sources located in the domain A of the upper half-space generate a
→ → →
primary wave U i1(g, t ) = {E i1(g, t ), H i1(g, t )} being incident on the artificial boundary L + (on
the planez1 = 0) as viewed fromz1 = ∞.
→ → →
Denote by U sj(g, t ) = {E sj(g, t ), H sj(g, t )} the fields resulting from scattering of the primary
→i
wave U 1(g, t ) in the domains A (where the total field is
→ → → → → → →
U (g, t ) = {E (g, t ), H (g, t )} = U s1(g, t ) + U i1(g, t )) and B (whereU (g, t ) = U s2(g, t )). In Section 5,
we have shown that the fields under consideration are uniquely determined by their longi‐
tudinal components, which can be given, for example, as:
{ Ezi (g, t )
H zi (g, t )
} {
=
∞
∑
n,m=−∞
vnm(1,E )(z1, t )
vnm(1,H )(z1, t )
} μnm( x, y ), z1 ≥ 0, t ≥ 0 (35)
{ Ezs (g, t )
H zs (g, t)
} {
=
∞
∑
n,m=−∞
unm( j,E )(zj , t )
unm( j,H )(zj , t )
} μnm( x, y ), zj ≥ 0, t ≥ 0, j = 1,2 (36)
(see also formulas (30)-(32)). Here, as before, {μnm( x, y )}∞n,m=−∞ is the complete (inL 2 (Rz )) or‐
thonormal system of transverse eigenfunctions of the Floquet channel R (see Section 4),
while the space-time amplitudes unm( j,E )(zj , t ) and unm( j,H )(zj , t ) are determined by the solu‐
tions of the following problems (see also problem (15)) for the one-dimensional Klein-Gor‐
don equations:
{ ∂2
− εj ∂ t 2 +
∂2
∂ zj 2
unm( j,E or H )(zj ,0) = 0,

− λn2 m unm( j,E or H )(zj , t ) = 0, t > 0
∂
∂t unm( j,E or H )(zj , t ) | t =0 = 0
, j = 1,2 , n, m = 0, ± 1, ± 2,... . (37)
Compose from the functionsvnm(1,E )(z1, t ), vnm(1,H )(z1, t ), unm( j,E )(zj , t ), unm( j,H )(zj , t )and the
eigenvalues λnm (n, m = 0, ± 1, ± 2,...) the setsv(1)(z1, t ) = {v p (1)(z1, t )}∞p=−∞ ,
u( j )(zj , t ) = {u p ( j )(zj , t )}∞p=−∞ , and {λp }∞p=−∞ such that their members are defined according to the
rules depicted in Figure 3. The sets v(1)(z1, t ) and u( j )(zj , t ) are said to be evolutionary bases
→ →
of signals U i1(g, t ) andU sj(g, t ). They describe completely and unambiguously transforma‐
tion of the corresponding nonsine waves in the regular Floquet channels A and Bfilled with
dielectric.
Figure 2. A two-dimensionally periodic grating between two dielectric half-spaces as element of a multi-layered struc‐
ture.
Let us introduce by the relations
| =∫ ∑
t
∂ ∞
∂ zj p ( j )( j )
AA( BA(
u p( j ) ′(0,t ) ≡ u z, t S pq t − τ )δj1 + S pq t − τ )δj2 vq (1)(0,τ )dτ,
zj =0 q=−∞
(38)
0
t ≥ 0, p = 0, ± 1, ± 2,..., j = 1,2
u( j )′(0,t) = {u p ( j )′(0,t)} p = S AAδj1 + S BAδj2 v(1)(0,τ) , t ≥ 0, j = 1,2 (39)
the boundary (on the boundarieszj = 0) transformation operators S AA and S BA of the evolu‐
→
tionary basis v(1)(z1, t ) of the wave U i1(g, t ) incoming from the domainA. Here δmn stands for
XY
the Kronecker delta, the operators’ elements Snm specify the space-time energy transforma‐
tion from the domain Y into the domain X and from the mode of order m into the mode of
ordern.
It is evident that the operators S AA and S BA working in the space of evolutionary bases are
intrinsic characteristics of the periodic structure placed between two dielectric half-spaces.
They totalize an impact of the structure on elementary excitations composing any incident
→
signalU i1(g, t ). Thus forvq (1)(0,t ) = δqr δ (t − η ), where r is an integer andη > 0, we have
u p (1)′(0,t ) = S prAA(t − η ) andu p (2)′(0,t ) = S prBA(t − η ). We use this example with an abstract non‐
physical signal by methodological reasons in order to associate the transformation opera‐
tors’ components S prAA(t − τ ) and S prBA(t − τ ) with an ‘elementary excitation’.
Figure 3. Construction of sets of the valuesv p (1), u p ( j ), and λp (p = 0, ± 1, ± 2,...) from sets of the valuesvnm(1,E ), unm( j,E ),
vnm(1,H ), unm( j,H ), and λnm (m, n = 0, ± 1, ± 2,...): (a)p = 0,1,2,...; (b)p = − 1, − 2, − 3,....
The operators S AA and S BA determine all the features of transient states on the upper and
bottom boundaries of the layer enclosing the periodic structure. Secondary waves outgoing
from these boundaries propagate freely in the regular Floquet channels A and B therewith
undergoing deformations (see, for example, [6]). The space-time amplitudes u p ( j )(zj , t) of the
partial components of these waves (the elements of the evolutionary bases of the signals
→
U sj(g, t )) vary differently for different values of p and j. These variations on any finite sec‐
tions of the Floquet channels A и B are described by the diagonal transporting operators
A B
Z 0→z1
and Z 0→z2
acting according the rule:
u( j )(zj , t) = {u p ( j )(zj , t)} = Z 0→z

A
1 j
B
δ 1 + Z 0→z δ 2 u( j )′(0,τ) , j = 1,2.
2 j
(40)
The structure of the operators given by (40) can be detailed by the formula
u p ( j )(zj , t) = −
1
εj∫0
J 0 λp
(t − τ )2
εj
− zj2 χ (
t −τ
εj
)
− zj u p ( j )′(0,τ )dτ,
(41)
t ≥ 0, zj ≥ 0, p = 0, ± 1, ± 2,..., j = 1,2,
which reflects general properties of solutions of homogeneous problems (37), i.e. the solu‐
tions that satisfy zero initial conditions and are free from the components propagating in the
direction of decreasingzj . The derivation technique for (41) is discussed at length in [6,13,14].
6.2. Equations of the operator method in the problems for multilayer periodic structures
The operators S AA and S BA completely define properties of the periodic structure excited
from the channelA. By analogy with (38) we can determine transformation operators S BB
and S AB for evolutionary basis v(2)(z2, t ) = {v p (2)(z2, t )}∞p=−∞ of the wave
→i →i →i
U 2(g, t ) = {E 2(g, t ), H 2(g, t )} incident onto the boundary z2 = 0 from the channelB:
∫∑
∞
AB ( BB (
u p( j ) ′(0,t ) = S pq t − τ )δj1 + S pq t − τ )δj2 vq (2)(0,τ )dτ,
m=−∞
(42)
0
t ≥ 0, p = 0, ± 1, ± 2,..., j = 1,2.
Let us construct the algorithm for calculating scattering characteristics of a multilayer struc‐
ture consisting of two-dimensionally periodic gratings, for which the operatorsS AA, S BA,
AB BB
S pq , and S pq are known. Consider a double-layer structure, whose geometry is given in
Figure 4. Two semi-transparent periodic gratings I and II are separated by a dielectric layer
of finite thickness M (hereε = ε2(I) = ε1(II)) and placed between the upper and the bottom die‐
lectric half-spaces with the permittivity ε1(I) andε2(II), respectively. Let also a pulsed wave
like (35) be incident onto the boundary z1(I) = 0from the Floquet channelA.
Retaining previously accepted notation (the evident changes are conditioned by the pres‐
ence of two different gratings I and II), represent the solution of the corresponding initial
boundary value problem in the regular domainsA, B, and C in a symbolic form
∞
U ( A) = ∑ v p (1)(z1(I), t ) + u p (1)(z1(I), t ) μp ( x, y ),
p=−∞
∞
U (B) = ∑ u p (2)(z2(I), t ) + u p (1)(z1(II), t ) μp ( x, y ),
p=−∞
∞
U (C ) = ∑ u p (2)(z2(II), t )μp ( x, y ).
p=−∞
The first terms in the square brackets correspond to the waves propagating towards the do‐
mainC, while the second ones correspond to the waves propagating towards the domain A
(Figure 4). The set {μp ( x, y )}∞p=−∞ is formed from the functionsμnm( x, y ), (n, m = 0, ± 1, ± 2,...),
while the set {λp }∞p=−∞ is composed from the valuesλnm, (n, m = 0, ± 1, ± 2,...) (Figure 3).
Figure 4. Schematic drawing of a double-layered structure.
By denoting
u( j )′(I) ≡
∂
∂ zj (I) ( j )( j
u z (I), t ) | ,
zj (I)=0
u( j )(I) = {u p ( j )(zj (I), t )}| z (I)=0,
j
according to formulas (38)-(42), we construct the following system of operator equations:
ìu ¢ ( I ) = S AA ( I ) év ( I ) ù + S AB ( I ) Z B éu ¢ II ù
ï (1) ë (1) û z1 ( II )= 0 ® M ê (1) ( ) ú
ë û
ï ¢ ¢
ïu( 2) ( I ) = S ( I ) éëv(1) ( I ) ùû + S ( I ) Z z1 ( II )= 0 ® M éêu(1) ( II ) ùú
BA BB B
ï ë û
í
¢ ¢
(43)
é
ïu(1) ( II ) = S ( II ) Z z ( I )= 0 ® M u( 2) ( I )
BB B ù
ï 2 ëê ûú
ï ¢ éu ¢ ( I ) ù .
u
ïî ( 2) ( II ) = S CB
( II ) Z B
z2 ( I )= 0 ® M ê ( 2 )
ë ûú
Equations (43) clearly represent step-by-step response of the complex structure on the exci‐
→
tation by the signal U i1(g, t ) with the evolutionary basis v(1)(z1(I), t ) = {v p (1)(z1(I), t )}∞p=−∞ (or
simplyv(1)(I)). Тhus, for example, the first equation can be interpreted as follows. A signal
u(1)(I) (the secondary field inA) is a sum of two signals, where the first signal is a result of the
reflection of the incident signal v(1)(I) by the gratingI, while another one is determined by the
signal u(1)(II)being deformed during propagation in the channel Band interaction with the
gratingI.
By method of elimination the system (43) is reduced to the operator equation of the second
kind
u(2)′(I) = S BA(I) v(1)(I) + S BB (I)Z zB1(II)=0→M S BB (II)Z zB2(I)=0→M u(2)′(I) (44)
and some formulas for calculating the electromagnetic field components in all regions of the
two-layered structure. The observation time t for the unknown function u(2)′(I) from the left-
hand side of equation (44) strictly greater of any moment of time τ for the function u(2)′(I) in
the right-hand side of the equation (owing to finiteness of wave velocity). Therefore equa‐
tion (44) can be inverted explicitly in the framework of standard algorithm of step-by-step
progression through time layers. Upon realization of this scheme and calculation of the
boundary operators by (38), (42), the two-layered structure can be used as ‘elementary’ unit
of more complex structures.
Turning back to (38)-(42), we see that the operators entering these equations act differently
that their analogues in the frequency domain, where the boundary operators relate a pair
A B
‘field → field’. Reasoning from the structure of the transport operators Z 0→z1
and Z 0→z2
(for‐
mulas (40) and (41)), we relate a pair ‘field → directional derivative with respect to the
propagation direction’ to increase numerical efficiency of the corresponding computational
algorithms.
7. Some Important Properties of Steady-State Fields in the Rectangular

Floquet Channel
7.1. Excitation by a TM -wave
Let a grating (Figure 1) be excited form the domain A by a pulsed TM -wave

→ →
U i (g, t ) = U ipq (E )(g, t ) : Ezi (g, t ) = v pq (z,E )(z, t )μ pq ( x, y ) and the region QL is free from the sour‐
→ → →
ces j (g, t ), φ E (g ), andφ H (g ). The field generated in the domains A and B is determined com‐
pletely by their longitudinal components. They can be represented in the form of (31), (32).
→ →
˜
˜ (g, k ), H
Define steady-state fields {E (g, k )} (see formula (33) withImk = 0) corresponding to
→i →i →s →s → →
{ } { }
the pulsed fields E , H , E , H in A and the pulsed field {E , H } inB, by their z-compo‐
nents:
{ Ẽ iz(g, k )
H̃ iz(g, k)
}{=
ṽ pq (z,E )(k )
0
} e
−iΓ pq ( z−L )
μ pq ( x, y ), g ∈ Ā (45)
{ Ẽ sz(g, k )
H̃ sz(g, k )
} { = ∑
∞
n,m=−∞
ũ +nm(z,E )(k )
ũ +nm(z,H )(k )
} e
iΓnm( z−L )
μnm( x, y ), g ∈ Ā (46)
{ Ẽ z (g, k )
H̃ z (g, k )
} = ∑
∞
−{
ũ −nm(z,E )(k )
n,m=−∞ ũ nm ( z,H )(k )
} e
−iΓnm( z+L )
μnm( x, y ), g ∈ B̄ (47)
where the following notation is used:ṽ pq (z,E )(k ) ↔ v pq (z,E )( L , t ),

ũ ±nm(z,E orH )(k ) ↔ unm
±
( z,E orH )(± L , t ), Γnm = (k 2 2 1/2
− λnm , ) ReΓnmRek ≥ 0, ImΓnm ≥ 0[7].
The amplitudes ũ ±nm(z,E orH )(k ) form the system of the so-called scattering coefficients of the
grating, namely, the reflection coefficients
( ) ũ +nm( z ,H )(k ) ( ) ũ +nm( z ,E )(k )

nm H
R pq (E ) = ṽ pq ( z ,E )(k ) ,
nm E
R pq (E ) = ṽ pq ( z ,E )(k ) , n, m = 0, ± 1, ± 2,... (48)
specifying efficiency of transformation of pq-th harmonic of a monochromatic TM -wave in‐

→ →
to of order nm-th harmonics of the scattered field {E ˜ s } in the reflection zone, and the
˜ s, H
transmission coefficients
( ) ũ −nm( z ,H )(k ) ( ) ũ −nm( z ,E )(k )

nm H
T pq (E ) = ṽ pq ( z ,E )(k ) ,
nm E
T pq (E ) = ṽ pq ( z ,E )(k ) , n, m = 0, ± 1, ± 2,... (49)
determining the efficiency of excitation of the transmitted harmonics in the domainB.

These coefficients are related by the energy balance equations
∑
∞
n,m=−∞
1
2
λnm
(| R pqnm(E(E) ) | 2 + | T pqnm(E(E) ) | 2) ± η02(| R pqnm(E(H) ) | 2 + | T pqnm(E(H) ) | 2) { }
ReΓnm
ImΓnm
{ } { }
( ) (50)
1 ReΓ pq + 2ImΓ pq ImR pqpq(EE)
1 W1
= 2 ( E ) ∓ ε W , p, q = 0, ± 1, ± 2, … ,
λ pq ImΓ pq − 2ReΓ pq ImR pq (E )
pq
0 2
ε0η0
∫ σ(g, k ) | E˜→ (g, k ) | dg,
2
W1=
k
QL
(51)
W2=
QL
∫ →
μ0μ (g, k ) | H 0
→
˜ (g, k ) | 2 − ε ε (g, k ) | E
˜ (g, k ) | 2 dg
They follow from the complex power theorem (Poynting theorem) in the integral form [11]
∮ ( E˜→ × H˜→
SL
∗
→
⋅ ds ) = ∫ div E˜→ × H→˜
QL
∗
→
dg = ikη0 μ | H
QL
∫
˜ | 2dg − ik
η0 ∫ ε | E˜→ | dg − ∫ σ | E˜→ | dg
QL
2
QL
2
(52)
whereε (g, k ) − 1 = χ̃ ε (g, k ) ↔ χε (g, t ), μ (g, k ) − 1 = χ̃ μ (g, k ) ↔ χμ (g, t ),

→
σ (g, k ) = χ̃ σ (g, k ) ↔ χσ (g, t ), ds is the vector element of the surface SL bounding the domain
QL . Equations (50)-(52) have been derived starting from the following boundary value
problem for a diffraction grating illuminated by a plane TM -wave
→i
˜ i(
U pq ( E )
( g, k ) : Ẽ z g, k ) = exp − iΓ pq
( z − L ) μ pq
( x, y ):
ì r r
ïh0 rotH% ( g , k ) = -ike ( g , k ) E% ( g , k ) ,
ï r r
=ïrotE% ( g , k ) ikh0 m ( g , k ) H% ( g , k ) , g Î QL
ï
=
ï é r% r% ù
ï ë û ( )
2p i F é r% r% ù
í D ê E H ú ( lx , y ) e x D ê E H ú ( 0, y ) , 0 £ y £ l y ,
ë û ( ) z <L (53)
ï é r% r% ù
( ) ( )
r% r% ù
ï D ëê E H ûú ( x, l y ) e
=
2p i F y é
D E H ( x,0 ) , 0 £ x £ lx , z <L
ëê ûú
ï
ï Ε% tg ( g , k ) = 0, H% nr ( g , k ) =0
î g ÎS g ÎS
{ Ẽ z (g, k )
= e
H̃ z (g, k ) 0
} {}
1 −iΓ pq (z−L )
μ pq ( x, y ) + ∑
∞
nm( E )( )
R pq (E ) k
nm( H )( )
n,m=−∞ R pq ( E ) k
{
iΓ ( z−L )
e nm }
μnm( x, y ), g ∈ Ā,
} ∑{ }
(54)
{
nm( E )( )
Ẽ z (g, k ) ∞ T pq (E ) k −iΓnm( z+L )
= e μnm( x, y ), g ∈ B̄.
H̃ z (g, k ) n,m=−∞
nm( H )( )
T pq (E ) k
When deriving (50), (51) we have also used the equations relating z-components of the ei‐
genmode of the Floquet channel
→
˜
U (g, k ) : Ẽ z (g, k ) = Ae±iΓz μ ( x, y ) and H̃ z (g, k ) = Be±iΓz μ ( x, y ) (55)
(subscripts nm are omitted) with its longitudinal components:
βkη0 αΓ αkη0 βΓ
Ẽ x = − 2 H̃ z ∓ 2 Ẽ z , Ẽ y = H̃ z ∓ 2 Ẽ z ,
λ λ λ2 λ
(56)
αΓ βk βΓ αk
H̃ x = ∓ H̃ + Ẽ , H̃ y = ∓ 2 H̃ z − Ẽ .
λ 2 z η0λ 2 z λ η0λ 2 z
Here, ε̄ (g, k ) = ε (g, k ) + iη0σ (g, k ) / k, μ ( x, y ) = (lx ly )−1/2exp(iαx )exp(iβy ), Γ = k 2 − λ 2,

λ 2 = α 2 + β 2.
→ → → →
According to the Lorentz lemma [11], the fields {E ˜ (1)} and {E
˜ (1), H ˜ (2)} resulting from the
˜ (2), H
interaction of a grating with two plane TM -waves
→ i (1)
˜
U ( ) i (1)(g, k ) = exp − iΓ pq (Φx , Φy )(z − L ) μ pq ( x, y, Φx , Φy ) and
pq ( E ) g, k : Ẽ z
→ i (2)
˜
U ( ) i (2)(g, k ) = exp − iΓ−r ,−s ( − Φx , − Φy )(z − L ) μ−r ,−s ( x, y, − Φx , − Φy ),
−r ,−s ( E ) g, k : Ẽ z
satisfy the following equation
∮ (( E→˜
SL
(1) → (2)
˜
×H
→ (2) →
˜
− E ×H
→
˜ (1) ) ⋅ ds ) = 0. (57)
From (57), using (54) and (56), we obtain
( ) ( )
( E )(Φx , Φy )λ p,q (Φx , Φy ) ,−s ( E ) ( − Φx , − Φy )λ−r ,−s ( − Φx , − Φy )
rs E
R pq 2
R−r− p,−q E 2
= ,
Γ pq (Φx , Φy ) Γ−r ,−s ( − Φx , − Φy ) (58)
p, q, r, s = 0, ± 1, ± 2,...
– the reciprocity relations, which are of considerable importance in the physical analysis of
wave scattering by periodic structures as well as when testing numerical algorithms for
boundary problems (53), (54).
→ i (1)
˜
Assume now that the first wave U ( )
pq ( E ) g, k :
i (1)(
→
˜ i (1) (
: Ẽ z g, k = exp − iΓ pq (Φx , Φy ) z − L μ pq ( x, y, Φx , Φy ) = U pq (E ) g, k, A be incident on the
) ( ) )
grating from the domainA, as in the case considered above, while another wave
→ i (2)
˜
U ( ) i (2)(g, k, B ) = exp iΓ−r ,−s ( − Φx , − Φy )(z + L ) μ−r ,−s ( x, y, − Φx , − Φy ) is incident
−r ,−s ( E ) g, k : Ẽ z
fromB. Both of these waves satisfy equation (57), whence we have
( ) ( )
( E )(Φx , Φy , A)λ p,q (Φx , Φy ) ,−s ( E ) ( − Φx , − Φy , B )λ−r ,−s ( − Φx , − Φy )
rs E
T pq 2
T −r− p,−q E 2
= ,
Γ pq (Φx , Φy ) Γ−r ,−s ( − Φx , − Φy ) (59)
p, q, r, s = 0, ± 1, ± 2,...
7.2. Excitation by a TE-wave
Let a grating be excited form the domain A by a pulsed TE-wave

→ →
U i (g, t ) = U ipq (H )(g, t ) : H zi (g, t ) = v pq (z,H )(z, t )μ pq ( x, y ) and the region QL is free from the
→ → →
sources j (g, t ), φ E (g ), andφ H (g ). The field generated in the domains A and B is determined
completely by their longitudinal components. They can be represented in the form of (31),
→ → → →
˜ (g, k ), H
(32). Define steady-state fields {E ˜ (g, k )} corresponding to the pulsed fields{E i , H i },
→s →s → →
{E , H }in A and the pulsed field {E , H } inB, by their z-components as was done for the
nm E nm H ( ) ( )
TM -case (see equations (45)-(47)). Introduce the scattering coefficientsR pq ( H ) , R pq ( H ) ,
nm( E ) nm( H )
T pq (H ) , and T pq (H ) by the relations like (48). These coefficients can be determined from the
problems
ì r r
ï r r
ï
=
ï é r% r% ù
ï ë û ( )
2p i F é r% r% ù
í D ê E H ú ( lx , y ) e x D ê E H ú ( 0, y ) , 0 £ y £ l y ,
ë û ( ) z <L (60)
ï é r% r% ù
( ) ( )
r% r% ù
=
2p i F y é
D E H ( x,0 ) , 0 £ x £ lx , z <L
ëê ûú
ï
ï Ε% tg ( g , k ) = 0, H% nr ( g , k ) =0,
î g ÎS g ÎS
{ Ẽ z (g, k )
= e
H̃ z (g, k ) 1
} {}
0 −iΓ pq (z−L )
μ pq ( x, y ) + ∑
∞
nm( E )( )
R pq (H ) k
nm( H )( )
n,m=−∞ R pq ( H ) k
{
iΓ ( z−L )
e nm }
μnm( x, y ), g ∈ Ā,
} ∑{ }
(61)
{
( )
nm E ( )
Ẽ z (g, k ) ∞ T pq (H ) k −iΓnm( z+L )
= e μnm( x, y ), g ∈ B̄
H̃ z (g, k ) n,m=−∞
nm( H )( )
T pq (H ) k
and satisfy the following relations, which are corollaries from the Poynting theorem and the
Lorentz lemma:
∑
∞
n,m=−∞
1
2
λnm
(| R pqnm(H(H) ) | 2 + | T pqnm(H(H) ) | 2) ± 1
η0
nm( E ) 2 nm( E ) 2
{ } ReΓnm
2 (| R pq ( H ) | + | T pq ( H ) | ) ImΓ
nm
=
{ } { }
pq H ( ) (62)
1 ReΓ pq + 2ImΓ pq ImR pq (H ) 1 W1
= 2 − , p, q = 0, ± 1, ± 2, …
λ pq ImΓ pq − 2ReΓ pq ImR pqpq((HH)) μ0 W 2
and
( ) ( )
( H )(Φx , Φy )λ p,q (Φx , Φy ) ,−s ( H ) ( − Φx , − Φy )λ−r ,−s ( − Φx , − Φy )
rs H
R pq 2
R−r− p,−q H 2
= ,
Γ pq (Φx , Φy ) Γ−r ,−s ( − Φx , − Φy ) (63)
p, q, r, s = 0, ± 1, ± 2,...
( ) ( )
( H )(Φx , Φy , A)λ p,q (Φx , Φy ) ,−s ( H ) ( − Φx , − Φy , B )λ−r ,−s ( − Φx , − Φy )
rs H
T pq 2
T −r− p,−q H 2
= ,
Γ pq (Φx , Φy ) Γ−r ,−s ( − Φx , − Φy ) (64)
p, q, r, s = 0, ± 1, ± 2,...
7.3. General properties of the grating’s secondary field
Let now k be a real positive frequency parameter, and let an arbitrary semi-transparent gra‐
ting (Figure 1) be excited from the domain A by a homogeneous TM - or TE-wave
→i
pq ( E or H ) g, k : { Ẽ z g, k H̃ iz(g, k )}=e
˜ i( −iΓ pq ( z−L )
U ( ) ) or μ pq ( x, y ) , p, q : ImΓ pq = 0 . (65)
The terms of infinite series in (54) and (61) are z-components of nm-th harmonics of the scat‐
nm E orH nm E orH ( ) ( )
tered field for the domains A andB. The complex amplitudes R pq ( E orH ) and T pq ( E orH ) are
the functions ofk, Φx , Φy , as well as of the geometry and material parameters of the grating.
Every harmonic for which ImΓnm = 0 and ReΓnm > 0 is a homogeneous plane wave propagat‐
→
ing away from the grating along the vectork nm:kx = αn , ky = βm, kz = Γnm(inA; Figure 5) or
±
kz = − Γnm (inB). The frequencies ksuch that Γnm(k ) = 0 (k = knm = ± | λnm | ) are known as thresh‐
old frequency or sliding points [1-6]. At those points, a spatial harmonic of order nm with
ImΓnm > 0 are transformed into a propagating homogeneous pane wave.
→
It is obvious that the propagation directions k nm of homogeneous harmonics of the secon‐
dary field depends on their ordernm, on the values of k and on the directing vector of the
→
incident wavek ipq :kxi = αp , kyi = βq ,kzi = − Γ pq . According to (50) and (62), we can write the follow‐
ing formulas for the values, which determine the ‘energy content’ of harmonics, or in other
words, the relative part of the energy directed by the structure into the relevant spatial radi‐
ation channel:
2
ReΓnm λ pq
pq = (| R pq ( E ) | + η0 | R pq ( E ) | )
(WR )nm nm( E ) 2 nm( H ) 2 (E )
)nm(H )
2
= (WR )nm (
pq ( E ) + WR pq ( E ) ,
λ 2 Γ pq
nm
(66)
2
ReΓnm λ pq
(WT )np = (| T nm( E ) 2
pq ( E ) | + η 2
| T nm( H ) 2
pq ( E ) | ) = (WT )np
(E )
( )np (H )
pq ( E ) + WT pq ( E )
pq 0
λ 2 Γ pq nm
(for TM -case) and
Figure 5. On determination of propagation directions for spatial harmonics of the field formed by a two-dimensional‐
ly periodic structure.
(| )
2
1 ReΓnm λ pq
(WR )nm
pq =
nm H
R pq
(
(H )
)
|2 + 2 | R pq ( H ) |
nm( E ) 2
= (WR )nm
(H )
( )nm(E )
pq ( H ) + WR pq ( H ),
η0 λnm Γ pq
2
(67)
(| |)
2
1 ReΓnm λ pq
(WT )np
pq =
nm( H )
T pq (H ) | 2
+ 2 | T pq
nm( E )
(H )
2
= (WT )nm
(H )
( )nm(E )
pq ( H ) + WT pq ( H )
η0 λ 2 Γ pq
nm
(for TE-case). The channel corresponding to the nm-th harmonic will be named ‘open’ if
ImΓnm = 0. The regime with a single open channel (nm = pq) will be called the single-mode re‐
gime.
→ →
Since | k ipq | = | k nm | = k, the nm-th harmonic of the secondary field in the reflection zone
propagates in opposition to the incident wave only if αn = − αp and βm = − βq or, in other nota‐
tion, if
n = − 2Φx − p and m = − 2Φy − q (68)

Generation of the nonspecularly reflected mode of this kind is termed the auto-collimation.
nm E orH ( ) nm E orH( )
The amplitudes R pq ( E orH ) or T pq ( E orH ) are not all of significance for the physical analysis. In
the far-field zone, the secondary field is formed only by the propagating harmonics of the
orders nm such thatReΓnm ≥ 0. However, the radiation field in the immediate proximity of the
grating requires a consideration of the contribution of damped harmonics (nm : ImΓnm > 0).
Moreover, in some situations (resonance mode) this contribution is the dominating one [6].
7.4. The simplest corollaries of the reciprocity relations and the energy conservation law
Let us formulate several corollaries of the relations (50), (58), (59), and (62)-(64) basing on the
results presented in [3] and [7] for one-dimensionally periodic gratings and assuming that
ε (g, k ) ≥ 0, μ (g, k ) ≥ 0, andσ (g, k ) ≥ 0.
• The upper lines in (50) and (62) represent the energy conservation law for propagating
waves. IfImΓ pq = 0, the energy of the scattered field is clearly related with the energy of the
→i
˜
incident wave. The energy of the wave U (g, k ) is partially absorbed by the grating
pq ( E orH )
(only ifW 1 ≠ 0), and the remaining part is distributed between spatial TM - and TE-har‐
monics propagating in the domains A and B (the wave is reradiating into the directions
z = ± ∞). If a plane inhomogeneous wave be incident on a grating (ImΓ pq > 0), the total en‐
( )
ergy is defined by the imaginary part of reflection coefficientR pqpq(EEorH
orH
) , which in this case
is nonnegative.
• The relations in the bottom lines in (50), (62) limit the values of
∑ ∞
n,m=−∞ | R pqnm(E(E) ) | 2λnm
−2
ImΓnm, ∑∞n,m=−∞ | T pq ( E ) | λnmImΓnm, etc. and determine thereby the
nm( E ) 2 −2
class of infinite sequences
{
l̄ 2 = a = {anm}∞nm=−∞ :
∞
∑
nm=−∞
| anm | 2
n2 + m2
∞ } (69)
nm E nm E ( ) ( )
or energetic space, to which amplitudes of the scattered harmonicsR pq ( E ) , T pq ( E ) , etc. be‐
long.
• It follows from (58), (59), (63), and (64) that for all semi-transparent and reflecting gratings
we can write
( E orH )
(WR )00 ( )00(E orH )
00( E orH )(Φx , Φy ) = WR 00( E orH )( − Φx , − Φy ),
( E orH )
(70)
(WT )00 ( )00(E orH )
00( E orH )(Φx , Φy , A) = WT 00( E orH )( − Φx , − Φy , B ).
The first equation in (70) proves that the efficiency of transformation of the TM - or TE-wave
into the specular reflected wave of the same polarization remains unchanged if the grating is
rotated in the plane x0y about z-axis through180 ° . The efficiency of transformation into the
principal transmitted wave of the same polarizations does not also vary with the grating ro‐
→
tation about the axis lying in the plane x0y and being normal to the vector k 00 (Figure 5).
• When r = s = p = q = 0 we derive from (58), (59), (63), and (64) that
( ) ( )
R0000(EEorH ) (Φx , Φy ) = R00( E orH ) ( − Φx , − Φy ),
orH 00 E orH
( ) ( )
(71)
T 0000(EEorH ) (Φx , Φy , A) = T 00( E orH ) ( − Φx , − Φy , B ).
orH 00 E orH
That means that even if a semi-transparent or reflecting grating is non symmetric with re‐
spect to the any planes, the reflection and transmission coefficients entering (71) do not de‐
pend on the proper changes in the angles of incidence of the primary wave.
• Relations (50), (58) allow the following regularities to be formulated for ideal (σ (g, k ) ≡ 0)
asymmetrical reflecting gratings. Let the parametersk, Φx , and Φy be such that
ReΓ00(Φx , Φy ) > 0 and ReΓnm(Φx , Φy ) = 0 forn, m ≠ 0. If the incident wave is an inhomogene‐
→i
˜
ous plane waveU ± p,±q ( E )( g, k, ± Φx , ± Φy ), then
ReΓ00(±Φx , ± Φy )
(| R±00p,±q
(E )
( E )(±Φx , ± Φy ) | 2 + η02 | R±00p,±q
( H )
( E )(±Φx , ± Φy ) | )
2
=
λ002(±Φx , ± Φy )
(72)
( )
ImΓ± p,±q (±Φx , ± Φy )
( E ) (±Φx , ± Φy )
p,±q E
= 2ImR±±p,±q .
λ±2p,±q (±Φx , ± Φy )
( ) ( )
SinceR pqpq(EE) (Φx , Φy ) = R−−p,−q ( E ) ( − Φx , − Φy ), we derive from (72)
p,−q E
| R p,q ( E )(Φx , Φy ) | + η0 | R p,q ( E ) (Φx , Φy ) | =

00( E ) 2 2 00( H ) 2
(73)
= | R−00p,−q ( E )( − Φx , − Φy ) | + η0 | R− p,−q ( E )( − Φx , − Φy ) | 2.
(E ) 2 2 00( H )
It is easy to realize a physical meaning of the equation (73) and of similar relation for TE-
case, which may be of interest for diffraction electronics. If a grating is excited by a damped
harmonic, the efficiency of transformation into the unique propagating harmonic of spatial
spectrum is unaffected by the structure rotation in the plane x0y about z-axis through180 ° .
The above-stated corollaries have considerable utility in testing numerical results and mak‐
ing easier their physical interpretation. The use of these corollaries may considerably reduce
amount of calculations.
8. Elements of Spectral Theory for Two-Dimensionally Periodic Gratings
The spectral theory of gratings studies singularities of analytical continuation of solutions of

boundary value problems formulated in the frequency domain (see, for example, problems
(53), (54) and (60), (61)) into the domain of complex-valued (nonphysical) values of real pa‐
rameters (like frequency, propagation constants, etc.) and the role of these singularities in
resonant and anomalous modes in monochromatic and pulsed wave scattering. The funda‐
mental results of this theory for one-dimensionally periodic gratings are presented in [4,6,7].
We present some elements of the spectral theory for two-dimensionally periodic structures,
which follow immediately form the results obtained in the previous sections. The frequency
k acts as a spectral parameter; a two-dimensionally periodic grating is considered as an
open periodic resonator.
8.1. Canonical Green function
{
Let a solution G̃ 0(g, p, k ) of the scalar problem
Δg + k 2 G̃ 0(g, p, k ) = δ (g − p ), g = { xg , yg , zg } ∈ R, p = { xp , yp , zp } ∈ QL
2πiΦx
D G̃ 0 (lx , yg ) = e D G̃ 0 (0,yg ), 0 ≤ yg ≤ ly , | zg | ≤ L
D G̃ 0 ( xg , ly ) = e
2πiΦy
D G̃ 0 ( xg ,0), 0 ≤ xg ≤ lx , | zg | ≤ L (74)
G̃ 0(g, p, k ) =
∞
∑
n,m=−∞
{ Anm( p, k )
Bnm( p, k )
} e
±iΓnm( zg ∓L )
μnm( xg , yg ), g ∈ {}
Ā
B̄
is named the canonical Green function for 2-D periodic gratings. In the case of the elementa‐
ry periodic structure with the absence of any material scatterers, the problems of this kind
but with arbitrary right-hand parts of the Helmholtz equation are formulated for the mono‐
chromatic waves generated by quasi-periodic current sources located in the region|z| < L .
Let us construct G̃ 0(g, p, k ) as a superposition of free-space Green functions:
1
∞ exp ik | g − pnm | 2πinΦ 2πimΦ
G̃ 0(g, p, k ) = − 4π ∑ | g − pnm | e x
e y
, pnm = { xp + nlx , yp + mly , zp }. (75)
n,m=−∞
By using in (75) the Poisson summation formula [15] and the tabulated integrals [16]
exp(ip x 2 + a 2) ( 2 2)
∞ ∞
∫
−∞ x2 + a2
eibx dx = πiH 0 1 (a
( )
∫
| p 2 − b 2 | ) and H 0(1)( p x 2 + a 2)e ibx dx = 2 exp ia p − b ,
−∞ p −b2 2
( )
where H 0 1 ( x ) is the Hankel function of the first kind, we obtain
i
∞
i αn ( xg −xp )+βm( yg − yp )
exp i | zg − zp | Γnm
G̃ 0(g, p, k ) = − 2lx ly ∑ e
Γnm
(76)
n,m=−∞
The surface K of analytic continuation of the canonical Green function (76) into the domain
of complex-valued k is an infinite-sheeted Riemann surface consisting of the complex planes
k ∈ C with cuts along the lines(Rek )2 − (Imk )2 − λnm
2
= 0, n, m = 0, ± 1, ± 2,..., Imk ≤ 0(Figure 6).
The first (physical) sheet Ck of the surface K is uniquely determined by the radiation condi‐
tions for G̃ 0(g, p, k ) in the domains A andB, i.e. by the choice of ReΓnmRek ≥ 0 and ImΓnm ≥ 0
on the axisImk = 0. On this sheet, in the domain0 ≤ argk < π, we haveImΓnm > 0, while ReΓnm ≥ 0
for 0argk ≤ π / 2 and ReΓnm ≤ 0 forπ / 2 ≤ argkπ. In the domain 3π / 2 ≤ argk < 2π for finite num‐
ber of functions Γnm(k ) (with n and m such that(Rek )2 − (Imk )2 − λnm
2
> 0), the inequalities
ImΓnm < 0 and ReΓnm > 0 hold; for the rest of these functions we have ImΓnm > 0 andReΓnm ≤ 0.
In the domainπ < argk ≤ 3π / 2, the situation is similar only the signs of ReΓnm are opposite.
On the subsequent sheets (each of them with its own pair{k; Γnm(k )}), the signs (root branch‐
es) of Γnm(k ) are opposite to those they have on the first sheet for a finite number of n andm.
±
The cuts (solid lines in Fig. 6) originate from the real algebraic branch pointsknm = ± | λnm | .
Figure 6. Natural domain of variation of the spectral parameterk: the first sheet of the surfaceK .
In the vicinity of some fixed point K ∈ K the function G̃ 0(g, p, k ) can be expanded into a
Loran series in terms of the local variable [17]
κ= { k − K ; K ∉ {knm
±
k − K ; K ∈ {knm
±
}
}
.
Therefore, this function is meromorphic on the surfaceK . Calculating the residuals

ResG̃ 0(g, p, k ) at the simple polesk̄ ∈ {knm
±
}, we obtain nontrivial solutions of homogeneous
k=k̄
→i
˜ ¯ = ∅ ) problems (53), (54) and (60), (61):
(U (g, k ) ≡ 0) canonical (ε̄ (g, k ) ≡ 1, μ (g, k ) ≡ 1,intS
→
˜
E (g, knm
±
) = {Ẽ x , Ẽ y , Ẽ z } ; Ẽ x, y or z
→ → (77)
= ax, y or z exp i (αn x + βm y ) and H ˜ (g, knm ±
) = (iknm
±
η0)−1rotE
˜ (g, k ± )
nm
where ax, y or z are the arbitrary constants. These solutions determine free oscillations in the
space stratified by the following conditions:
→ →
˜ →
˜ ) ( x + l , y ) = e2πiΦx D E → → ˜
˜ ) ( x, y ), D E
˜ (H ˜ → → ˜
˜ →
D E (H x
(H ) ( x, y + ly ) = e2πiΦy D E (H ) ( x, y ). (78)
8.2. Spectrum qualitative characteristics
Let a set Ωk of the points {k̄ }

j j∈K such that for all k ∈ {k̄ } the homogeneous (spectral)
j j
problem
ì r r
ï r r
ï
=
ï é r% r% ù
ï ë û ( ) 2p i F é r% r% ù
ë û ( )
í D ê E H ú ( lx , y ) e x D ê E H ú ( 0, y ) , 0 £ y £ l y , z <L (79)
ï é r% r% ù
( ) ( )
r% r% ù
=
2p i F y é
D E H ( x,0 ) , 0 £ x £ lx , z <L
ëê ûú
ï
ï Ε% tg ( g , k ) = 0, H% nr ( g , k ) =0 ,
î g ÎS g ÎS
{ Ẽ z (g, k )
H̃ z (g, k )
} = ∑
∞
{
n,m=−∞
Anm(E )(k )
Anm(H )(k )
}
e
iΓnm( z−L )
μnm( x, y ), g ∈ Ā
{ } ∑ { }
(80)
Ẽ z (g, k ) ∞ Bnm(E )(k )
−iΓnm( z+L )
= e μnm( x, y ), g ∈ B̄
H̃ z (g, k ) n,m=−∞ Bnm(H )(k )
→
˜ → →
has a nontrivial (not necessarily unique) solution U (g, k̄ j ) = {E˜ (g, k̄ j ), H
˜ (g, k̄
j )}be called the
point spectrum of the grating. It is obvious that these solutions characterize the so-called
free oscillations, whose field pattern, structure of their spatial harmonics and behavior of
these harmonics for large | z | and t are determined by the value of k̄ j = Rek̄ j + iImk̄ j and by
→
˜
a position of the point k̄ (the eigen frequency associated with a free oscillationU
j (g, k̄ )) on j
the surface K [4,6,7]. By continuing analytically the problems (53), (54) and (60), (61) togeth‐
→
˜ →
˜ →
˜
er with their solutions U (g, k ) = {E (g, k ), H (g, k )} into the domain K of the complex-valued
→
˜ (g, k )at the pointsk = k̄ . In the vicinity of these poles, the
k, we detect poles of the function U j
desired solutions can be represented by the Loran series in terms of the local on K variable κ
[17]. The analytical findings of this kind may form the basis for detailed study of physical
features of resonant wave scattering by one-dimensionally and two-dimensionally periodic
structures [4,6,7,18,19].
Derive now the conditions that constrain existence of nontrivial solutions of the problem
(79), (80). These conditions can be considered as uniqueness theorems for the problems (53),
(54) and (60), (61) formulated for different domains of the surfaceK . Notice that the study of
the uniqueness allows one to estimate roughly a domain where elements of the set Ωk are
localized and simplify substantially the subsequent numerical solution of spectral problems
owing to reduction of a search zone of the eigen frequencies. The uniqueness theorems serve
also as a basis for application of the ‘meromorphic’ Fredholm theorem [20] when construct‐
ing well grounded algorithms for solving diffraction problems as well as when studying
qualitative characteristics of gratings’ spectra [4,7].
Assume that grating scattering elements are nondispersive (ε (g, k ) = ε (g ), μ (g, k ) = μ (g ), and
σ (g, k ) = σ (g )). In this case, the analytical continuation of the spectral problem (79), (80) into
the domain of complex-valued k are simplified considerably. From the complex power theo‐
→
˜
rem in the integral form formulated for the nontrivial solutions U (g, k̄ j ) like
∮ ( E˜→ × H˜→
SL
∗
→
⋅ ds ) = ∫ div E˜→ × H→˜
QL
∗
dg
(81)
∫→
˜ | 2dg −
= ikη0 μ | H
QL
ik ∗
η0 ∫
QL
→ 2
˜
ε|E | dg −
QL
∫ →
˜ | 2dg
σ|E
the following relations result:
∑
∞
n,m=−∞
2
λnm
{
1 (ReΓnmRek + ImΓnmImk )
(ImΓnmRek − ReΓnmImk )
} (| Anm(E ) | 2 + | Bnm(E ) | 2)
{ }
(82)
1 − Imk (V 3 + V 2) − V 1
±η02 ( | Anm(H ) | 2 + | Bnm(H ) | 2 ) =
ε0 Rek (V 3 − V 2)
→ →
˜ =E
˜
Notation:k = k̄ j , E (g, k̄ j ), Γnm = Γnm(k̄ j ), Anm(E ) = Anm(E )(k̄ j ), etc., and
∫
QL
→ 2
˜
V 1 = ε0η0 σ | E | dg, V 2 =
QL
∫ ε ε | E→˜ | dg,V = ∫ μ μ | H˜→ | dg.
0
2
3
QL
0
2
No free oscillations exist whose amplitudes do not satisfy equations (82). From this general
statement, several important consequences follow. Below some of them are formulated for
gratings withε (g ) > 0, μ (g ) > 0, andσ (g ) ≥ 0.
• There are no free oscillations whose eigen frequencies k̄ j are located on the upper half-
plane (Imk0) of the first sheet of the surfaceK . This can be verified by taking into account
the upper relation in (82), the function Γnm(k ) onCk , and the inequalitiesV 1 ≥ 0, V 2 > 0,V 3 > 0.
• If σ (g ) ≡ 0 (the grating is non-absorptive), no free oscillations exist whose eigen frequen‐

cies k̄ j are located on the bottom half-plane (Imk < 0) of the sheet Ck between the cuts cor‐
±
responding to the least absolute values ofknm . In Figure 6, this region of the first sheet of K
and the above-mentioned domain are shaded by horizontal lines.
• If σ (g ) > 0 on some set of zero-measure pointsg ∈ QL , then there are no elements k̄ j of gra‐
ting’s point spectrum Ωk that are located on the real axis of the planeCk .
Investigation of the entire spectrum of a grating, i.e. a set of the pointsk ∈ K , for which the
diffraction problems given by (53), (54) and (60), (61) are not uniquely solvable, is a compli‐
cated challenge. Therefore below we do no more than indicate basic stages for obtaining
well grounded results. The first stage is associated with regularization of the boundary val‐
ue problem describing excitation of a metal-dielectric grating by the currents
→ →
J˜ (g, k ) ↔ J (g, t ) located in the domainQL :
ì r r% r%
ïh0 rotH% ( g , k ) = -ike ( g , k ) E ( g , k ) + J ( g , k ) ,
ï r r
ï
=
ï é r% r% ù
ï ë û ( ) 2p i F é r% r% ù
ë û ( )
í D ê E H ú ( lx , y ) e x D ê E H ú ( 0, y ) , 0 £ y £ l y , z <L (83)
ï é r% r% ù
( ) ( )
r% r% ù
= D
ï ëê E H
ûú
( y)
x , l e
2p i F y
D é
ëê
E H
ûú
( x,0 ) , 0 £ x £ lx , z <L
ï
ï Ε% ( g , k ) = 0, H% nr ( g , k ) =0 ,
î tg g ÎS g ÎS
{ Ẽ z (g, k )
H̃ z (g, k )
} = ∑
∞
{
n,m=−∞
Anm(E )(k )
Anm(H )(k )
}
e
iΓnm( z−L )
μnm( x, y ), g ∈ Ā,
{ } ∑ { }
(84)
Ẽ z (g, k ) ∞ Bnm(E )(k )
−iΓnm( z+L )
= e μnm( x, y ), g ∈ B̄.
H̃ z (g, k ) n,m=−∞ Bnm(H )(k )
By regularization is meant (see, for example, [7]) a reduction of the boundary value electro‐
dynamic problem to the equivalent operator equation of the second kind
E + B (G̃ 0, S, ε̄, μ, k ) X = Y , EX = X (85)
with a compact (in some space W of vector fields) finite-meromorphic (in local on K varia‐
blesκ) operator-function B (G̃ 0, S, ε̄, μ, k ) [20,21]. If the problem given by (83), (84) is consid‐
ered separately for metal gratings (intS ¯ ≠ ∅ and S are sufficiently smooth surfaces;
( ) ( ) ¯ = ∅ , ε̄ (g, k ) = ε (g )and μ (g, k ) = μ (g ) are suffi‐
ε̄ g, k = μ g, k ≡ 1) and dielectric gratings (intS
ciently smooth functions), then its regularization can be performed by applying the poten‐
tial theory methods [4,7,22].
In the second stage, the following statements should be proved: (i) the resolvent E + B (k ) −1
(k ∈ K ) of the problem in (85) is a finite-meromorphic operator-function; (ii) its poles are lo‐
cated at the points k = k̄ j ( j = 1,2,3,...); (iii) the entire spectrum coincides with its point spec‐
trumΩk ; (iv) Ωk is nothing more than a countable set without finite accumulation points. All
these statements are corollaries of the ‘meromorphic’ Fredholm theorem [4,20,21] and the
uniqueness theorem proved previously.
By inverting homogeneous operator equation (85), we can construct a numerical solution of

the spectral problem given by (79), (80) [4,6], in other words, calculate the complex-valued
→
˜ → →
eigen frequencies k̄ and associated eigen waves U
j j
˜ (g, k̄ )} or free oscil‐
(g, k̄ ) = {E˜ (g, k̄ ), H j j
lations of an open two-dimensionnaly periodic resonator. Commonly, this operation is re‐

duced to an approximate solution of the characteristic equation like:
det C (k ) = 0. (86)
Here C (k ) is some infinite matrix-function; the compactness of the operator B (k ) ensures (i)
existence of the determinant det C (k ) and (ii) the possibility to approximate the solutions k̄
of equation (86) by the solutions k̄ ( N ) of the equation det C (k, N ) with the matrix C (k, N )
reduced to dimensionN × N .
Let k̄ be a root of characteristic equation (86) that do not coincide with any pole of the opera‐
tor-functionB (k ). The multiplicity of this root determines the multiplicity of the eigen value
k̄ of homogeneous operator equation (85), i.e. the value M = M (1) + M (2) + ... + M (Q ) [21].
→ (q )
˜
Here, Qis the number of linearly-independent eigen functionsU (g, k̄ ); q = 1,2,...,Q(the
number of free oscillations) corresponding to the eigen value (eigenfrequency)k̄, while
→ (q )
˜
M (q ) − 1 is the number of the associated functionsU (m)( g, k̄ );m = 1,2,...,M (q ) − 1. The order of
−1
pole of the resolvent E + B (k ) (and of the Green function G̃ (g, p, k ) of the problem in
(83), (84)) for k = k̄is determined by a maximal value ofM (q ).
9. Conclusion
The analytical results presented in the chapter are of much interest in the development of
rigorous theory of two-dimensionally periodic gratings as well as in numerical solution of
the associated initial boundary value problems. We derived exact absorbing boundary con‐
ditions truncating the unbounded computational space of the initial boundary value prob‐
lem for two-dimensionally periodic structures to a bounded part of the Floquet channel.
Some important features of transient and steady-state fields in rectangular parts of the Flo‐
quet channel were discussed. The technique for calculating electrodynamic characteristics of
multi-layered structure consisting of two-dimensionally periodic gratings was developed by
introducing the transformation operators similar to generalized scattering matrices in the
frequency domain. In the last section, the elements of spectral theory for two-dimensionally
periodic gratings were discussed.
Author details
L. G. Velychko1, Yu. K. Sirenko1 and E. D. Vinogradova2*
1 Usikov Institute of Radiophysics and Electronics, National Academy of Sciences of Uk‐

raine, Kharkov, Ukraine
2 Macquarie University, Department of Mathematics, North Ryde, Australia
References
[1] Shestopalov, V. P., Litvinenko, L. N., Masalov, S. A., & Sologub, V. G. (1973). Wave
Diffraction by Gratings. Kharkov: Kharkov State Univ. Press.
[2] Petit, R. editor (1980). Electromagnetic Theory of Gratings. New York: Springer.
[3] Shestopalov, V. P., Kirilenko, A. A., Masalov, S. A., & Sirenko, Y. K. (1986). Reso‐
nance Wave Scattering. Vol. 1. Diffraction Gratings. Kiev: Naukova Dumka; (in Rus‐
sian).
[4] Shestopalov, V. P., & Sirenko, Y. K. (1989). Dynamic Theory of Gratings. Kiev: Nau‐
kova Dumka; (in Russian).
[5] Neviere, M., & Popov, E. (2003). Light Propagation in Periodic Media: Differential
Theory and Design. New York: Marcel Dekker.
[6] Sirenko, Y. K., Strom, S., & Yashina, N. P. (2007). Modeling and Analysis of Transient
Processes in Open Resonant Structures. New Methods and Techniques. New York:
Springer.
[7] Sirenko, Y. K., & Strom, S. editors (2010). Modern Theory of Gratings. Resonant Scat‐
tering: Analysis Techniques and Phenomena. New York: Springer.
[8] Ladyzhenskaya, O. A. (1985). The Boundary Value Problems of Mathematical Phys‐
ics. New York: Springer-Verlag.
[9] Taflove, A., & Hagness, S. C. (2000). Computational Electrodynamics: the Finite-Dif‐
ference Time-Domain Method. Boston: Artech House.
[10] Sirenko, K., Pazynin, V., Sirenko, Y., & Bagci, H. (2011). An FFT-Accelerated FDTD
Scheme with Exact Absorbing Conditions for Characterizing Axially Symmetric Res‐
onant Structures. Progress in Electromagnetics Research, 111, 331-64.
[11] Rothwell, E. J., & Cloud, M. J. (2001). Electromagnetics. New York: CRC Press.
[12] Vladimirov, V. S. (1971). Equations of Mathematical Physics. New York: Dekker.
[13] Sirenko, K. Y., & Sirenko, Y. K. (2005). Exact’Absorbing’ Conditions in the Initial
Boundary-Value Problems of the Theory of Open Waveguide Resonators. Computa‐
tional Mathematics and Mathematical Physics, 45, 490-506.
[14] Kravchenko, V. F., Sirenko, Y. K., & Sirenko, K. Y. (2011). Electromagnetic Wave
Transformation and Radiation by the Open Resonant Structures. Modelling and
Analysis of Transient and Steady-State Processes. Moscow: FizMathLit; (in Russian).
[15] Titchmarsh, E. (1948). Introduction to the Theory of Fourier Integrals. Oxford: Clar‐
endon Press.
[16] Gradshteyn, I. S., & Ryzhik, I. M. (1994). Table of Integrals, Series, and Products.
New York: Academic.
[17] von Hurwitz, A., & von Courant, R. (1964). Allgemeine Funktionentheorie und Ellip‐
tische Funktionen: Geometrische Funktionentheorie. Berlin: Springer-Verlag; (in Ger‐
man).
[18] Sirenko, Y. K., Velychko, L. G., & Erden, F. (2004). Time-Domain and Frequency-Do‐
main Methods Combined in the Study of Open Resonance Structures of Complex Ge‐
ometry. Progress in Electromagnetics Research, 44, 57-79.
[19] Velychko, L. G., Sirenko, Y. K., & Shafalyuk, O. S. (2006). Time-Domain Analysis of
Open Resonators. Analytical Grounds. Progress in Electromagnetics Research, 61, 1-26.
[20] Reed, M., & Simon, B. (1978). Methods of Modern Mathematical Physics. IV: Analysis
of Operators. New York: Academic Press.
[21] Hokhberg, I. Z., & Seagul, Y. I. (1971). Operator Generalization of the Theorem about
Logarithmic Residue and the Rouche Theorem. Matematicheskii Sbornik, 84, 607-629,
(in Russian).
[22] Colton, D., & Kress, R. (1983). Integral Equation Methods in Scattering Theory. New
York: Wiley-Interscience.
Chapter 6
The Sources Reconstruction Method for Antenna

Diagnostics and Imaging Applications
Yuri Álvarez, Fernando Las-Heras and

Cebrián García
1. Introduction
The advances on communications systems require more challenging antenna designs, in‐
creasing the complexity of design, manufacturing, and testing stages. Different sources of er‐
ror may affect the antenna final prototype, degrading its performance, and then requiring a
re-design step to identify and correct these manufacturing errors. In general, this re-design
step requires invasive techniques based on trial-error procedure, whose technical and eco‐
nomical costs are directly related to the antenna complexity. For this reason, the develop‐
ment of fast and accurate non-invasive methods for antenna faults detection (antenna
diagnostics) has been of great interest in the last decade.
This chapter describes the Sources Reconstruction Method (SRM), an integral equation-based
inverse technique, and its applications on antenna diagnostics, near-field to far-field transfor‐
mation, and the recent advances on imaging applications. The method provides an electro‐
magnetic model of the problem-under-study based on an equivalent currents distribution,
which can be used to evaluate the field radiated at any point of the space. The equivalent mod‐
el can be used for antenna diagnostics applications in the field of antenna measurements, as
well as for geometry reconstruction purposes in the case of scattering acquisitions.
1.1. Antenna Diagnostics
Antenna diagnostics comprises these techniques based on the recovery of the extreme near-
field of the Antenna-Under-Test (AUT) from the field radiated by it. Extreme near-field is
defined as the field in the proximity of the AUT (i.e. a distance < 1 wavelength (λ), or on the
AUT aperture plane). The field radiated by the AUT can be acquired in the near or far field
region (Yaghian, 1986).
The extreme near-field is strongly related to the physical currents (in the case of metallic sur‐
faces) or aperture field distribution of the AUT. In consequence, an accurate retrieval of this
field provides a useful mapping of the fields and currents distribution on the manufactured
antenna, which can be compared to the initial design to identify which parts of the antenna
have faults or differences with respect to the designed one.
1.1.1. Backpropagation techniques
It is well-known that the AUT radiation pattern can be calculated as the Fourier Transform
of the fields on the AUT aperture plane (Yaghian, 1986). In consequence, this calculation can
be reverted in order to recover the AUT aperture fields from the radiation pattern. The sim‐
plest case relates a planar field acquisition domain with the AUT aperture plane. For exam‐
ple, in (Kaplan et al., 1979; Rahmat-Samii, 1984) the fields acquired on a plane are
backpropagated to the aperture plane by using the inverse Fast Fourier transform (iFFT). Po‐
lar and bi-polar measurement setups are also suitable for this kind of technique (Gatti &
Rahmat-Samii, 1988; Yaccarino et al., 1994), which has been extended to phaseless measure‐
ments (Yaccarino & Rahmat-Samii, 1999; Razavi & Rahmat-Samii, 2007). Equivalent current
techniques have been also developed (Petre & Sarkar, 1992), showing similar performance to
the iFFT-based methods.
Next step was the aperture field calculation from the field acquired in spherical domains,
which is one of the most extended antenna measurement systems (Yaghian, 1986). A formu‐
lation relating spherical wave modes with the AUT plane wave spectrum has been devel‐
oped in (Cappellin et al., 2008). Also equivalent current methods have been extended to
this measurement setup, providing the same accuracy as wave mode based techniques (Ál‐
varez et al., 2008c).
It is out of doubt that fastest diagnostics results are given by iFFT-based methods. So for
many years, the computational cost of equivalent current techniques, derived from a kind of
inverse Method-of-Moments (MoM), was prohibitive for practical antenna diagnostics.
However, the increase of antenna design complexity requires the extreme near-field to be re‐
covered not only in the AUT aperture plane, but also in a surface fitting the AUT geometry.
The need of this requirement, together with the improvement of calculation resources fos‐
tered the development of equivalent current techniques.
1.1.2. Three-dimensional techniques
Wave mode-based techniques are limited to canonical geometries (mainly planar, cylindri‐
cal, and spherical), for both reconstruction domain (e.g. extreme near-field) and field acquis‐
ition domain. Equivalent current methods for antenna diagnostics are able to overcome this
limitation, so that arbitrary geometry domains can be defined. The only limitation is that the
extreme near-field surface must enclose all the radiating elements.
Equivalent current methods provide an accurate electromagnetic model of the AUT, that has
been mainly used for antenna diagnostics, as for example, detection of faulting elements in
antenna arrays (Álvarez et al., 2009b), reflector antenna distortions (Álvarez et al., 2008c; Las-
The Sources Reconstruction Method for Antenna Diagnostics and Imaging Applications 161
Heras et al., 2009), leakage identification (Foged et al., 2011a), and parasitic currents in anten‐
na elements (Araque & Vecchi, 2009). In addition to the equivalent current techniques, it has
been proposed the use of electric and magnetic infinitesimal dipoles that model the AUT elec‐
tromagnetic behaviour (Sijher & Kishk, 2005; Mikki & Kishk, 2007). The orientation and exci‐
tation of these infinitesimal dipoles is determined from the acquired near-field distribution.
Figure 1. General scheme showing the equivalent current (or sources reconstruction) method for antenna diagnostics
and geometry reconstruction.
Apart from antenna diagnostics, these techniques have potential applications in other anten‐
na measurement related topics: near-field to far-field transformation (that is, antenna radia‐
tion pattern calculation) (Las-Heras et al., 2006; Álvarez et al., 2008a; Álvarez et al., 2007;
Álvarez et al., 2012b), analysis of the AUT interaction with scatterers (Las-Heras, 2001; Gar‐
cía et al., 2011), filtering of parasitic elements (Araque & Vecchi, 2009; Cano et al., 2010) and
electromagnetic emissions analysis (Laviada et al., 2010; Hernando et al., 2008; Persson &
Gustafsson, 2005; Eibert & Schmidt, 2009). Due to its wide field of applications, some anten‐
na measurement companies have even released their own antenna diagnostics software, as
for example (Foged et al., 2011b) and (Cappellin et al., 2011).
Equivalent current methods have been extended to cope with phaseless measurements (Las-
Heras & Sarkar, 2002; Álvarez et al., 2010c; Álvarez et al., 2012b) which are of special interest
on sub-mm and THz frequency bands, where phase measurements require expensive equip‐
ment (Hellicar et al., 2009; Hislop & Hellicar, 2009).
1.2. Inverse scattering methods based on equivalent Currents
Equivalent currents provide an equivalent model of the AUT that can be also extrapolated to
scattering problems. In this case, the scatterer is illuminated by an incident field, inducing a
set of currents on the object. Thus, if the incident and scattered fields are known, it would be
possible to recover an equivalent currents distribution related to the scatterer geometry. For
metallic objects the idea is simple: the equivalent currents intensity will be related to the re‐
gions of the space where the physical induced currents are flowing.
Several inverse scattering techniques are based upon this idea (Habashy et al., 1990; Caorsi
et al., 1990; Qin & Ciric, 1993; Van den Berg, Kleimann, 1997; Lin & Kiang, 1996; Álvarez et
al., 2010b). The main limitation is the need of volumetric equivalent currents domains,
which significantly increases the calculation time. Computational acceleration schemes can
be applied to reduce it.
To sum up this section, general schemes of the equivalent current techniques, also referred
as Sources Reconstruction Method (SRM), is presented in Fig. 1, together with its applica‐
tions to antenna diagnostics and characterization, near-field to far-field transformation, and
profile reconstruction.
2. The Sources Reconstruction Method
The concept of the Sources Reconstruction Method (SRM) is illustrated in Fig. 2. On the left,
the direct radiation problem is plotted: the field radiated by the AUT (Erad) is collected on the
acquisition domain Sobs. Now, the SRM goal is to recover the extreme near-field on a surface
S’ enclosing the AUT. The field samples (Erad) are backpropagated from Sobs to S’ (Fig. 2, right)
obtaining an electromagnetic equivalent currents distribution, Jeq, Meq. These equivalent cur‐
rents radiate the same electromagnetic fields as the AUT outside the domain enclosed by S’;
in other words, the currents provide an equivalent electromagnetic model of the AUT.
2.1. The electromagnetic equivalence principle
The SRM is based on the electromagnetic Equivalence Principle (Harrington, 2001; Stratton,
1941; Balanis, 1997): consider the fields (En, Hn) and currents (Jn, Mn) defined in Fig. 3 (a)
where two media are present: an inner medium (n= 1) with constitutive parameters ε1, μ1, σ1,
defined in the volume V’, and bounded by S’; and the outer medium (n=2), with constitutive
parameters ε2, μ2, σ2.
y y
Antenna Probe
Under Test
r r
Erad Erad S'
x x
r r r r
J eq , M eq
Erad Erad r
Sobs Radiated field Sobs Erad
acquisition domain
Direct radiation problem Inverse radiation problem
Figure 2. Direct and inverse radiation problem.
Figure 3. Electromagnetic Equivalence Principle.
In Fig. 3 (b), a set of equivalent electric (Jeq) and magnetic currents (Meq) is defined on the
boundary between the two media (S’). These currents are defined as the gap between the
tangential components of the fields in media 1,2. Note that no assumption about the fields
and currents inside S’ is considered, just imposing the currents to radiate the same fields
outside S’ as in Fig. 3 (a). In consequence, Jeq and Meq characterize the electromagnetic be‐
haviour outside S’ of the fields (E1, H1) and currents (J1, M1) enclosed by S’.
Several SRM implementations (Foged et al., 2011a; Araque & Vecchi, 2009; Cappellin et al.,
2011; Araque & Vecchi, 2010) make use of the Love’s Equivalence Principle: in this case, the
fields inside S’ are forced to be zero (Fig. 3 (c)), so that the equivalent currents on S’ are di‐
rectly the field tangential components. In the case of antenna diagnostics, they correspond to
the extreme near-fields.
2.2. Integral Equations
The Sources Reconstruction Method (SRM) is based on the integral equations relating the
electromagnetic field (Erad) observed in Sobs with their sources. In the case of the SRM, the
sources are the equivalent currents Jeq and Meq defined on S’ (Fig. 2 (b)).
The integral equations relating the electric field with an equivalent electric Jeq(r’) and mag‐
netic Meq(r’) currents distribution defined on a surface S’ are given by Eqs. (1) and (2) re‐
spectively (Harrington, 2001; Balanis, 1997), so that the total electric field is the sum of the
electric field due to electric currents Erad,Jeq(r) and the electric field due to magnetic currents
Erad,Meq(r).
r r r r
r r jh ìï 2 e - jk0 R( r ;r ') r r æ æ e - jk0 R( r ;r ') r r ö ö üï
Erad , J eq (r ) = - 0
4p k0 ò R(r ; r ')
í k 0 r r J eq ( r ') + Ñ çç Ñ × ç r r J eq (r ') ÷ ÷÷ ý dS ' (1)
S' ï
î è è R(r ; r ') ø ø ïþ
r r
r r 1 æ e - jk0 R( r ;r ') r r ö
4p Sò'
Erad , Meq (r ) = - Ñ ´ ç r r M eq (r ') ÷ dS ' (2)
è R(r ; r ') ø
where k0 is the wavenumber, η0 the intrinsic impedance, r the position vector of the points
where the field is acquired (surface Sobs), and r’ the position vector of the points where the
currents are reconstructed, which belong to the S’ domain. R(r ; r’) is defined as |r - r’|.
2.3. Numerical solution
Integral equations (1), (2) have to be discretized for a numerical solution. The field observa‐
tion domain (surface Sobs) is sampled in N points. The equivalent currents domain (S’) is dis‐
cretized in M planar facets (e.g. triangular facets or Rao-Wilton-Glisson basis functions (Rao
et al., 1982) when modelling arbitrary 3D geometries).
Equivalent currents vectors are expressed as a linear combination of two perpendicular

components defined on each facet ΔS’m. These components are referred to an orthogonal co‐
ordinate system defined on each facet using the unitary vectorsu^ m, v^ m, n^ m(see Fig. 4).
The definition of the components is given by Eq. (3), assuming that the equivalent currents
are constant on each facet. This approach is, in general, valid, provided that ΔS’m<< (λ)2.
→
J eq,m = J eq,umu^ m + J eq,vmv^ m
→ (3)
M eq,m = M eq,umu^ m + M eq,vmv^ m
The discretized integral equations as a function of indexes m, n, are described by (4), (5). All
the constants and variables are known except the complex coefficients Jeq,um, Jeq,vm, Meq,um,
Meq,vm.
ìæ e - jk0 R( rrn ;rrm ') æ æ e - jk0 R( rn ;rm ') ö ö ö ü

r r
jh M ïïç k0 r r + Ñ çç Ñ × çç r r ÷÷ ÷÷ ÷ ×ïï
2
r r
Erad , J eq (rn ) = - 0 å íçè R(rn ; rm ') è è R(rn ; rm ') ø ø ø ý
÷ (4)
4p k0 m=1 ï ï
(
ïî J eq ,um uˆ m + J eq , vm vˆ m DSm ') ïþ
r r
r r 1 M ìï æ e - jk0 R( rn ;rm ') ö üï
Erad , Meq (rn ) = -
4p
å íÑ ´ ç r r ( )
M eq ,um uˆ m + M eq , vm vˆ m ÷ DSm 'ý (5)
ï è R(rn ; rm ')
m=1 î ø þï
Observation domain
Equivalent currents domain, S’ (field acquisition domain)
nˆm
uˆ m Jr (rr ' )
eq m
r r
M eq (rm' )
vˆ m r r
DSm' rn - rm'
r r
Erad (rn )
r
r rn ( xn , yn , zn )
rm' ( xm' , ym' , zm' )
z
O
y
x
Field samples
Figure 4. Definition of the observation and equivalent currents domain, and how they are discretized.
The discretized equations can be rewritten in a more compact form, resulting in the matrix
system (6). The unknown coefficients of the equivalent currents correspond to the right col‐
umn vector of (6), the field samples correspond to the independent column vector, and the
rest of variables and coefficients of Eqs. (4) and (5) are grouped in a matrix called Z-matrix.
æ J eq ,u ö
ç ÷
æ E ö æ ZErad
t1 t1
; J eq ,u
ZEt1 ; J eq ,v
ZEt1 ; M eq ,u
ZEt1
rad ; M eq ,v
ö
ç J eq , v ÷
ç
rad
÷ = ç rad rad
÷ (6)
è E ø çè ZErad
t2
ZEt2 ZEt2 ZEt2 ; M ÷ ç M eq ,u ÷
eq ,v ø
rad t2
; J eq ,u rad ; J eq ,v rad ; M eq ,u rad çç ÷÷
è M eq , v ø
The field samples on Sobs can be expressed by means of 3 orthogonal components: Erad =
Erad,t1t1 + Erad,t2t2+ E rad,nn. The majority of antenna measurement systems only acquire the tan‐
gential field components, as the normal component decay much faster than the tangential
ones. Thus, only the tangential field components (Erad,t1, Erad,t2) are used to recover the equiv‐
alent currents. These tangential components can be referred to the corresponding canonical
coordinate systems. For example, in the case of cylindrical acquisition system, t1 = ϕ, t2 = z.
In spherical system: t1 = θ, t2 = φ.
The resulting linear system of equations has 2xN equations and 4xM unknowns. Sometimes,
it is possible to use just one type of equivalent currents if the reconstruction domain S’ is a
flat surface (Petre & Sarkar, 1992; Las-Heras et al., 2006; Sarkar & Taaghol, 1999). In this
case, the observation domain is restricted to one of the two halves of the space divided by
the virtual infinite plane containing S’.
2.4. Numerical solvers
Different techniques can be used to solve the linear system of equations (6). In general, the
number of equations (Ne) and unknowns (Mu) may be different; Ne ≠ Mu, i.e. the system of
equations may not be square. In addition, depending on the Z-matrix rank and the number
of unknowns, the system can be overdetermined (Rank(Z-matrix)>Mu) or undetermined
(Rank(Z-matrix)< Mu). The numerical solvers considered in this section seek for the least
mean squares solution, that is, the equivalent currents distribution that minimizes the error
(ε) between the acquired field samples and the field radiated by the reconstructed equiva‐
lent currents evaluated on the field sampling points. The issue of the solution accuracy is
also discussed in detail in (Araque & Vecchi, 2010), analyzing the system conditioning and
discussing several regularization techniques.
Table 1 summarizes the solvers that have been tested. Regarding computational cost consid‐
erations, N ≈ M ≈ Ne ≈ Mu is assumed1.
1 Notation: N denotes the number of field samples, and Ne the number of equations. M defines the number of equiva‐
lent currents facets, and Mu the number of unknowns.
Singular Value Decomposition (SVD) calculates the singular values of the Z-matrix (Álvarez
et al., 2009b; Araque & Vecchi, 2010), providing a valuable analysis about its conditioning: if
the problem is ill-conditioned (high condition number) (Sarkar et al., 1981b), that means that
the solution of the problem (equivalent currents) is quite sensitive to small changes in the
inputs (i.e. the field samples). Lower singular values can be neglected (Truncated-SVD) (Ál‐
varez et al., 2009b; Araque & Vecchi, 2010), improving the Z-matrix conditioning, and then
stabilizing the solution. The main drawback of this technique is the moderate computational
cost (O(N3)) as the SVD requires the calculation of the pseudoinverse of the Z-matrix.
Method Type Time O(N) Memory O(N)
Singular Value Decomposition (SVD) Matrix inversion O(N )

3
O(N2)
Conjugate Gradient (GC) Iterative O(N2) O(N2)
Memory Saving Technique (MST) (1)

Iterative (1)
O(N )
2
O(N)
Fast Multipole Method (FMM) (1)

Iterative (1)
O(N )
1.5
O(N1.5)
Adaptive Cross Iterative (1) O(N4/3 logN) O(N4/3 logN)

Approximation algorithm (ACA) (1)
(1)
These techniques are implemented over the CG method.
Table 1. Numerical techniques and computational cost.
A more affordable implementation in terms of calculation time relies on iterative solvers.

Among different techniques, the Conjugate Gradient (CG) (Sarkar & Arvas, 1981a) has been
found to provide a good trade-off between calculation time O(N 2) and accuracy, and in par‐
ticular, the implemented CG algorithm (Wang & Hwang, 1995) has a monotonically de‐
crease of the error (ε). Iterative solvers require the definition of stopping rules or criteria. In
this case, the first rule is the definition of a maximum number of iterations, K. The second
rule is the variation of the error (Δε) between two consecutive iterations: Δε(k) = ε(k-1)-ε(k).
If this variation is smaller than a fixed threshold, the CG finishes. Typical threshold values
can be Δε = 0.001 – 0.01 (Álvarez et al., 2009b; López et al., 2012). In order to provide a refer‐
ence, convergence is usually reached in no more than K = 20 iterations with ε < 0.05 for the
given Δε threshold values in practical antenna measurements.
Electrically large antenna diagnostics may require a significant amount of calculation time.
To reduce it, different computational acceleration schemes have been applied over the CG
formulation. The first one was the Fast Multipole Method (FMM) (Eibert & Schmidt, 2009;
Engheta et al., 1992; Coifman et al., 1993): in brief, the idea is to approximate field and
source samples which belong to a certain region in terms of plane and spherical waves. In
general, FMM is applied to problems in which all the regions belong to the same physical
domain, so interactions between adjacent regions have to be analyzed without the FMM ap‐
proach. In the case of the SRM, the source (equivalent currents) and field samples belong to
different physical domains (Fig. 2) (Eibert & Schmidt, 2009; Álvarez et al., 2008b), avoiding
the presence of adjacent regions.
Another tested acceleration technique was the Adaptive Cross Approximation (ACA) algo‐
rithm. While the FMM takes advantage of the physical properties of the problem (i.e. the
fields and currents approach in terms of plane and spherical waves), the ACA explodes the
algebraic properties of the Z-matrix (Álvarez et al., 2009a). Assuming that the Z-matrix is
rank deficient, then, it can be approximated as Z-matrix = U x V. Considering that the Z-
matrix has Ne rows and Mu columns, then U has Ne rows and R columns, and V, R rows and
Mu columns, where R is defined as the effective rank of the Z-matrix (Zhao et al., 2005). ACA
is an efficient acceleration technique both in time and memory provided that (Ne+Mu) R<<
Nex Mu (Zhao et al., 2005; Álvarez et al., 2009a).
Finally, the inner CG implementation allows a quite efficient memory saving technique at
the expense of increasing the calculation time. This method relies on the fact that the Z-ma‐
trix elements can be calculated using mathematical expressions (4), (5) if the indexes m,n are
known. In consequence, instead of calculating and storing the entire Z-matrix, the Z-matrix
elements are calculated every time they are needed, reducing the memory consumption
from O(N2) to O(N). As an example, consider a problem with similar number of field sam‐
ples and facets, Ne ~ Mu ~ 100,000. The Z-matrix storage would require 150 GB of memory.
The memory saving technique requires less than 100 MB of memory (Álvarez et al., 2007;
López et al., 2012).
Graphics Processing Units (GPU) is the ultimate technology for hardware acceleration in
those problems demanding intensive calculations. The above mentioned memory saving
technique fulfils these requirements for GPU implementation, providing a speed-up greater
than 70 times with respect to previous CPU implementation (Álvarez et al., 2009b; Álvarez
et al., 2007). As an example, an antenna diagnostics problem with ~20,000 field samples and
~20,000 equivalent currents facets is solved in just 30 s (López et al., 2012).
2.5. Extension to inverse scattering problems
2.5.1. Inverse scattering methods for geometry reconstruction
The SRM extension to inverse scattering problems can be classified as an inverse source
problem technique (Habashy et al., 1990; Caorsi et al., 1990; Qin & Ciric, 1993; Van den Berg,
Kleimann, 1997; Álvarez et al., 2010b), that pursuits the geometry and constitutive parame‐
ters estimation by retrieving a set of equivalent sources. While the geometry reconstruction
problem yields a linear system of equations, constitutive parameters retrieval involves a
non-linear system of equations, increasing the inverse problem complexity.
Focusing on the geometry reconstruction case, other techniques have been developed. One
is the Linear Sampling Method (Colton et al., 2003), which determines if a point on the re‐
gion of interest belongs to the object-under-test or not. Its interest lies on its simplicity, as
just a linear system of equations has to be solved.
Other methods are based on contour parameterization: starting from an initial guess, the
parameterized geometry is modified until one that satisfies an imposed boundary condition
(e.g. zero tangential field in case of perfect electric conductors (PEC) or tangential field con‐
tinuity on dielectric surfaces) is found. The resulting system of equations relating the con‐
tour coordinates and the scattered field is non-linear, and it is necessary to use non-linear
optimization methods to solve it (Lin & Kiang, 1996; Farmahini, 2009; Martínez et al., 2011;
Hajihashemi & El-Shenawee, 2011). As the entire contour is optimized at the same time,
these inversion methods usually require a remarkable amount of calculation time. Addition‐
ally, as the cost function to be optimized is non-linear, the inversion process can easily get
trapped in local minima.
2.5.2. The Volumetric SRM
As indicated before, the analysis of the extreme near-field provides information about the
physical currents flowing on the metallic structure of the antenna. For example, consider the
case depicted in Fig. 5: the equivalent currents recovered on the radome enclosing the Yagi-
Uda antenna elements provide information about their position inside the radome.
Taking into account this fact, it has been considered the possibility of extending the SRM for
inverse scattering problems (Volumetric SRM) (Álvarez et al., 2010b); in particular, for met‐
allic objects geometry reconstruction. The idea is to retrieve the electric currents induced by
an incident field. As in the case of the Yagi-Uda antenna in Fig. 5, it is expected that the
highest intensity of the reconstructed equivalent currents will be related to the placement of
the physical induced currents.
The direct and inverse scattering problem using equivalent currents are depicted in Fig. 6. The
main difference with respect to the SRM for antenna diagnostics is the use of a volumetric
equivalent currents distribution (Jeq,V): now, the reconstruction domain is a volume V’ enclos‐
ing the metallic scatterers, so the field radiated by the reconstructed equivalent currents Jeq,V
is the same as the field scattered by the original geometry (Volumetric Equivalence Princi‐
ple (Livesay & Chen, 1974)). The integral equation relating the scattered field (Escatt) with the
volumetic equivalent currents (Jeq,V) is (7) (Álvarez et al., 2010a; Álvarez et al., 2010b).
r r r r
r r jh ìï e - jk0 R( r ;r ') r r æ æ e - jk0 R( r ;r ') r r ö ö üï
Escatt (r ) = - 0 ò í k02 r r J eq ,V (r ') + Ñ çç Ñ × ç r r J eq ,V (r ') ÷ ÷÷ ý dV ' (7)
4p k0 V ' îï R(r ; r ') è è R(r ; r ') ø ø þï
One of the main drawbacks of the Volumetric SRM is the problem dimensionality, which is
increased from O(N2) to O(N3). For electrically large scatterers, the domain V’ may be too
large to be analyzed in a reasonable time even with a powerful workstation. To overcome
this drawback, an inverse Fast Multipole Method (iFMM) has been developed in (Álvarez et
al., 2011) (2D) and (Álvarez et al., 2012a) (3D) allowing a significant speed-up (up to 500
times) without compromising the accuracy of the retrieved geometry. Even more, the use of
a Dyadic formulation (7) provides an advantage with respect to scalar formulations (López-
Sánchez & Fortuny-Guash, 2000) as polarization information can be efficiently used to im‐
prove the quality of the reconstruction.
Figure 5. Antenna diagnostics of a 3-element Yagi-Uda antenna covered by a radome.
y y
Scatterer r
(metallic
Einc r Probe
object) Escatt
r
Escatt
x r x
r r
Escatt
V' J eq ,V
Escatt r
Sobs Scattered field Sobs Escatt
acquisition domain
Direct scattering problem Inverse scattering problem
Figure 6. Direct and inverse scattering problems using equivalent currents.
3. Application examples
3.1. Antenna array diagnostics
The first application example is devoted to clearly show the antenna diagnostics capabilities
of the SRM. For this purpose, a 3x3 antenna array of λ/2 dipoles has been designed and si‐
mulated. Nominal excitations are plotted in Fig. 7. In order to simulate a failure in two an‐
tenna array elements, dipoles 2 and 4 are distorted, reducing their feeding from 0 dB to -6
dB and -30 dB respectively. A Method-of-Moments code is used to analyze the antenna and
to evaluate the radiation pattern, which is depicted in Fig. 7: it is clear that the distortion
introduced in a two elements array antenna degrades the radiation pattern, but no informa‐
tion about the malfunctioning elements can be extracted from the pattern.
The SRM is applied to retrieve an equivalent currents distribution on the antenna array
aperture plane. The reconstructed equivalent currents will be related to the extreme near-
field and hence, to the nominal excitations of the antenna, so any failure in the array ele‐
ments can be identified.
The working frequency is f = 300 MHz. The antenna size is defined in terms of the minimum
sphere enclosing the array (Rmin), which is Rmin = ~1.5 m, so according to (Hansen, 1988; Ya‐
ghian, 1986) the minimum number of radiation pattern samples for this antenna is N=441: θ
= [-90° +90°], with Δθ = 9°, and φ = [0° 180°], with Δφ = 9°. The equivalent currents are recon‐
structed in a planar domain S’ = (3 x 3 m)2, discretized in M = 441 facets (ΔS’ = (0.15 x 0.15
m)2 ) in order to have, for this particular case, the same number of field samples and equiva‐
lent currents.
(2)
Normalized amplitude, in dB
(4)
Normalized amplitude, in dB
Figure 7. Array of dipoles. Top left: nominal excitations, in which dipoles 2 and 4 are distorted. Top right: reconstruct‐
ed Meq,x. Bottom: array of dipoles radiation pattern.
The application of Image Theory for this problem allows neglecting equivalent electric cur‐
rents in the formulation, using just magnetic currents. The integral equation relating the
spherical field tangential components Eθ, Eφ, with the magnetic currents components Meq,u,
Meq,v (Meq,x, Meq,y, for the particular case) can be expressed as a linear system of equations to
be solved (8):
q
æ Erad ö æ ZEqrad ; Meq ,x ZEq ö æ M eq ,x ö
; M eq ,y
ç j ÷=ç ÷ç
rad
÷ (8)
è Erad ø çè ZEjrad ; Meq ,x ZEj ; M ÷ è M eq , y ø
rad eq ,y ø
The system has Ne=441 x 2 equations, and Mu=441 x 2 unknowns. As indicated before, one of
the main drawbacks of inverse problems is the intrinsic ill-conditioning. The condition num‐
ber of the Z-matrix in (8) has been calculated, and it is 3.9 x 1025, or in other words, this is the
relation between the highest and lowest singular values, which are plotted in Fig. 8.
The linear system of equations is solved using the Truncated SVD (T-SVD) (Álvarez et al.,
2009b; Araque & Vecchi, 2010) and the Conjugate Gradient (CG) (Álvarez et al., 2009b; Ál‐
varez et al., 2007), both of them providing the same results. In the case of the T-SVD, differ‐
ent thresholds are selected: the idea is that, by filtering the lowest singular values, the Z-
matrix condition number decreases, then improving the matrix conditioning. With respect
to the CG, the metric to evaluate the solution accuracy is based on the error between the si‐
mulated field and the field radiated by the reconstructed equivalent currents. The stopping
criterion is set to Δε = ε (k-1) – ε (k) < 0.001. For this problem, convergence was reached after
19 iterations (ε = 0.019), requiring just one minute of calculation in a conventional laptop.
To conclude this example, the reconstructed equivalent magnetic currents (Meq,x) are plotted
in Fig. 7. It is observed that the 9 elements (including the 2 distorted dipoles) are correctly
identified, predicting also the amplitude on each.
Figure 8. Singular values of the Z-matrix appearing in (8).

3.2. Phaseless measurements
The following application example is focused on antenna diagnostics and near-field to far-
field transformation using phaseless information. The AUT is a logarithmic-periodic anten‐
na working in the 400 MHz-3 GHz frequency band (Álvarez et al., 2010c). This antenna has
been measured at the spherical range in anechoic chamber of the University of Oviedo, at
two different distances: R1 = 3.20 m, and R2 = 5.25 m (see Fig. 9), and at the working frequen‐
cy of f = 1,030 MHz. The field has been sampled on a full sphere: θ = [-180° 180°), Δθ = 4°,
and φ = [0° 180°], Δφ = 4°, resulting in N=4,140 field samples on each acquisition sphere.
y z
Figure 9. Antenna measurement at spherical range in anechoic chamber at two different distances.
The equivalent currents reconstruction domain is chosen to be a surface S’ fitting the log-
periodic antenna radome (see Fig. 10, (a),(b)). This surface is discretized into M=3,298 trian‐
gular facets. In consequence, taking into account the number of field samples on each sphere
and the number of facets, the number of equations is Ne=4,140 x 2 field components x 2
measurement distances = 16,560, and the number of unknowns, Mu = 3,298 x 2 type of cur‐
rents x 2 current components = 13,192.
Two different kinds of problems are now solved. First, the system of equations relating the
field samples (amplitude and phase) with the equivalent currents (9) is solved. The retrieved
currents are plotted in Fig. 10 (d),(f): it is observed that for the selected working frequency,
the resonant elements are placed in the middle of the log-periodic antenna (Álvarez et al.,
2010c; Álvarez et al., 2009b). As the elements are x-polarized (Fig. 10 (b)), only the Jeq,x com‐
ponent is plotted. Next, from the reconstructed equivalent currents, the field at any point of
the space can be calculated, and hence, far field radiation pattern of the log-periodic antenna
which is plotted in Fig. 11 (dark and gray solid lines).
æZ q ZEq ZEq ZEq ö

æ Eqrad , R1 ö ç Erad ,R1 ; J eq ,u ÷ æ J eq ,u ö
rad ,R1 ; J eq ,v rad ,R1 ; M eq ,u rad ,R1 ; M eq ,v
ç j ÷ çZ j ZEj ZEj ZEj ; M ÷ ç J ÷

ç Erad , R1 ÷ = ç Erad ,R1 ; J eq ,u rad ,R1 ; J eq ,v rad ,R1 ; M eq ,u rad ,R1 eq ,v
÷ ç eq , v ÷
(9)
ç Eqrad , R ÷ ç Z q ZEq ZEq ZEq ; M ÷ ç M eq ,u ÷
ç j 2 ÷ ç Erad ,R2 ; J eq ,u rad ,R2 ; J eq ,v rad ,R2 ; M eq ,u rad ,R2 eq ,v
÷ çç ÷÷
ç Erad , R ÷
è 2 ø ç ZEj ; J eq ,u ZEj ZEj ZEj ; M ÷ è M eq , v ø
è rad ,R2 rad ,R2 ; J eq ,v rad ,R2 ; M eq ,u rad ,R2 eq ,v ø
(d) (e)
Jx - Normalized amplitude (dB) – Distances in m

y z x y z x
(f) (g)
Jx - Phase (degrees) – Distances in m
y z x y z x
Figure 10. Photos: (a) log-periodic antenna array with radome. (b) Without radome. (c) Measurement at spherical
range in anechoic chamber. Reconstructed equivalent currents, Jeq,x component. (d),(f) using amplitude and phase in‐
formation. (e),(g) using amplitude-only information.
The second problem to be solved neglects the phase information of the field samples. A non-
linear system of equations, relating the amplitude of the field samples with the amplitude of
the field radiated by the equivalent currents (10), has to be solved.
æ Eqrad , R ö æ ZEqrad ,R ; J eq ,u ZEq ZEq ZEq ö

÷ æ J eq ,u ö
; J eq ,v ; M eq ,u ; M eq ,v
ç 1
÷ ç 1 rad ,R1 rad ,R1 rad ,R1
ç Erad , R
j ÷ ç ZEj ; J eq ,u ZEj ; J eq ,v
ZEj ; M eq ,u
ZEj ; M ÷ ç J ÷
ç 1
÷= ç rad ,R1 rad ,R1 rad ,R1 rad ,R1 eq ,v
÷ ç eq , v ÷
(10)
ç Eqrad , R ÷ ç ZEqrad ,R ; J eq ,u ZEq ZEq ZEq ; M ÷ ç M eq ,u ÷
ç 2
÷ ç 2 rad ,R2 ; J eq ,v rad ,R2 ; M eq ,u rad ,R2 eq ,v
÷ çç ÷÷
ç Erad , R
j ÷ Z
ç Ej ; J eq ,u ZEj ZEj ZEj ; M ÷ è M eq , v ø
è 2 ø è rad ,R2 rad ,R2 ; J eq ,v rad ,R2 ; M eq ,u rad ,R2 eq ,v ø
The problem of antenna diagnostics using phaseless field samples has been addressed by
different approaches. For example (Yaccarino & Rahmat-Samii, 1999; Razavi & Rahmat-Sa‐
mii, 2007) introduces a plane-to-plane iterative backpropagation method for phase retrieval,
that can be used for both near-field to far-field transformation and antenna diagnostics ap‐
plications. Phaseless SRM is described in (Las-Heras & Sarkar, 2002): in this case, an equiva‐
lent magnetic currents distribution is calculated by minimizing a cost function which relates
the amplitude of the measured field and the contribution due to the equivalent currents,
similarly to Eq. (10). However, these techniques are restricted to flat reconstruction surfaces
S’, whereas the problem to be solved comprises a 3D reconstruction domains S’. Thus, in or‐
der to solve Eq. (10), the two-stage algorithm described in (Álvarez et al., 2010c; Álvarez et
al., 2012b) is applied.
Figure 11. Log-periodic antenna far field pattern. Cuts φ = 0° and φ = 90°. Comparison of the results when consider‐
ing amplitude and phase field samples, and phaseless field samples.
The reconstructed equivalent currents from the phaseless field samples are plotted in Fig. 10
(e),(g), observing a good agreement with the ones retrieved from amplitude and phase infor‐
mation. It is important to remark that not only the amplitude, but also the phase of the
equivalent currents (Fig. 10, (f),(g)) are in good agreement, proving the phaseless SRM for
phase retrieval.
The latest step is the calculation of the log-periodic antenna radiation pattern from the re‐
constructed equivalent currents. The comparison between the radiation pattern from ampli‐
tude and phase information (Fig. 11, solid lines) and using phaseless information (Fig. 11,
dashed lines) shows an acceptable agreement in the copolar component (black line). With
regard to the crosspolar component, the amplitude levels are in the same order of magni‐
tude (20-25 dB below the copolar component).
3.3. Geometry reconstruction applications
The latest example illustrates the SRM application for geometry reconstruction of metallic
scatterers. A 2D problem is considered (Álvarez et al., 2010b) in which the object-under-test
(depicted in Fig. 12, left plot) is supposed to have translation symmetry along one axis. The
object is illuminated by 12 TM-polarized plane waves coming from 0° to 330° in 30°-steps.
Four working frequencies are considered: 300, 450, 700 and 1,000 MHz. The scattered field is
observed on a circumference of R = 10 m, sampled every 6°, which results in 60 field sam‐
ples. Finally, noise is added according to a Signal-to-Noise (SNR) radio of 30 dB.
12 incidences and 4
One incidence (180º) and 12 incidences and one frequencies (all the
one frequency (300 MHz) frequency (300 MHz) available information)
Equivalent electric currents (normalized amplitude, dB)
Figure 12. Geometry reconstruction application example. The solid line on the left plot indicates the true profile of
the object-under-test.
The reconstruction domain V’ is a square region of 6 x 6 m discretized into 121 x 121 square
subdomains. Thus, for each incident wave and frequency, the system of equations to be
solved has Ne=60 equations, and Mu=14,641 unknowns. The Conjugate Gradient algorithm is
used, requiring about K=9-11 iterations to reach convergence (ε < 0.01, Δε < 0.001).
First, the inverse problem for one frequency (300 MHz) and one incidence (180°) is solved.
The reconstructed equivalent electric currents are plotted in Fig. 12 (left). The highest inten‐
sity of the reconstructed equivalent currents is located on the boundary of the object-under-
test and, in particular, the part of the contour illuminated by the plane wave.
Combination of multiple incidences yields to a complete retrieval of the profile of the object-
under-test. For example, the center plot of Fig. 12 represents the combination of the equiva‐
lent currents reconstructed for every incident plane wave at 300 MHz. The object-under-test
profile can be now guessed. Higher resolution is provided by combining multifrequency in‐
formation: the right plot of Fig. 12 clearly shows a well-defined profile.
4. Conclusion
The Sources Reconstruction Method for antenna diagnostics, near-field to far-field transfor‐
mation, and geometry reconstruction applications has been described. The interest on this
non-invasive technique is supported by several features which, in summary, are:
i) Establishment of an accurate equivalent electromagnetic model of the AUT that can be

used to evaluate the electromagnetic field at any point of the space.
ii) The capability of handling arbitrary field acquisition and extreme near-field reconstruc‐
tion domains, overcoming other diagnostics techniques limitation to canonical domains.
iii) Reliability and accuracy as the method uses full-wave integral equation formulation.
iv) The possibility of implementation on low cost GPU cards, enabling fast calculation even
in conventional computers as those ones that can be found in antenna measurement facili‐
ties. Low memory footprint is also kept thanks to a developed memory saving technique
(López et al., 2012).
v) Extension to phaseless measurements, which is of special interest for sub-mm and THz
systems in which phase measurements are especially complex and expensive.
Acknowledgements
This work has been supported by the European Union under COST Action IC1102 (VISTA);
by the “Ministerio de Ciencia e Innovación” of Spain/FEDER under projects CONSOLIDER-
INGENIO CSD2008-00068 (TERASENSE), TEC2011-24492/TEC (iSCAT),
IPT-2011-0951-390000 (TECNIGRAF), and under PhD grant BES-2009-024060; and by the
“Gobierno del Principado de Asturias” PCTI-FEDER under project PC10-06 (FLEXANT).
Author details
Yuri Álvarez1*, Fernando Las-Heras1 and Cebrián García1
1 Área de Teoría de la Señal y Comunicaciones – Universidad de Oviedo, Spain
References
[1] Álvarez, Y., Las-Heras, F., & Pino, M. R. (2007). Reconstruction of Equivalent Cur‐
rents Distribution Over Arbitrary Three-Dimensional Surfaces Based on Integral
Equation Algorithms. IEEE Transactions on Antennas and Propagation, 54, 3460-3468.
[2] Álvarez, Y., Las-Heras, F., & Pino, M. R. (2008). On the Comparison between the
Spherical Wave Expansion and the Sources Reconstruction Method. IEEE Transac‐
tions on Antennas and Propagation, 56(10), 3337-3341.
[3] Alvarez, Y., Las-Heras, F., Pino, M. R., & López, J. (2008). Acceleration of the Sources
Reconstruction Method via the Fast Multipole Method. 2008 IEEE International Sym‐
posium on Antennas and Propagation (APS’08), San Diego, CA (USA) 5-12 July 2008.
[4] Alvarez, Y., Cappellin, C., Las-Heras, F., & Breinbjerg, O. (2008). On the comparison
of the Spherical Wave Expansion-to-Plane Wave Expansion and the Sources Recon‐
struction Method for Antenna Diagnostics. Progress In Electromagnetics Research, PIER
87, 245-262.
[5] Alvarez, Y., Las-Heras, F., & Pino, M. R. (2009). Application of the Adaptive Cross Ap‐
proximation Algorithm to the Sources Reconstruction Method. 3rd European Confer‐
ence on Antennas and Propagation (EUCAP’09), Berlin (Germany), 23-27 March 2009, 1-5.
[6] Álvarez, Y., Las-Heras, F., Pino, M. R., & Sarkar, T. K. (2009). An Improved Super-
Resolution Source Reconstruction Method. IEEE Transactions on Instrumentation and
Measurement, 58(11), 3855-3866.
[7] Álvarez, Y., Casas, B. A., García, C., & Las-Heras, F. (2010). The Volumetric Sources
Reconstruction Method and its Applications to Geometry Reconstruction. 4th Europe‐
an Conference on Antennas and Propagation (EUCAP’10), Barcelona (Spain), 12-16 April
2010, 1-5.
[8] Álvarez, Y., Casas, B. A., García, C., & Las-Heras, F. (2010). Geometry Reconstruction
of Metallic Bodies Using the Sources Reconstruction Method. IEEE Antennas and
Wireless Propagation Letters, 9, 1197-1200.
[9] Álvarez, Y., Las-Heras, F., & Pino, M. R. (2010). The Sources Reconstruction Method
for Amplitude-Only Field Measurements. IEEE Transactions on Antennas and Propaga‐
tion, 58(8), 2776-2881.
[10] Álvarez, Y., Martínez, J. A., Las-Heras, F., & Rappaport, C. (2011). An Inverse Fast
Multipole Method for Imaging Applications. IEEE Antennas and Wireless Propagation
Letters, 10, 1259-1262.
[11] Álvarez, Y., Martínez, J. A., Las-Heras, F., & Rappaport, C. (2012). An inverse Fast
Multipole Method for geometry reconstruction using scattered field information.
IEEE Transactions on Antennas and Propagation, 60(7), 1-9, In press.
[12] Álvarez, Y., Las-Heras, F., & Pino, M. R. (2012). Antenna Diagnostics Using Phaseless
NF Information. AUTOMATIKA Journal for Control, Measurement, Electronics, Comput‐
ing and Communications, 53(1), 49-55.
[13] Araque, J., & Vecchi, G. (2009). Removal of unwanted structural interactions from an‐
tenna measurements. Antennas and Propagation Society International Symposium, 2009,
APSURSI’09. Charleston, South Caroline, USA. June 2009.
[14] Araque, J., & Vecchi, G. (2010). Field And Source Equivalence In Source Reconstruc‐
tion On 3D Surfaces. Progress In Electromagnetics Research, PIER 103, 67-100.
[15] Balanis, C. A. (1997). Antenna Theory. Analysis and Design, John Wiley & Sons.
[16] Cano, F. J., Sierra-Castañer, M., Burgos, S., & Besada, J. L. (2010). Applications of
sources reconstruction techniques: Theory and practical results. Proceedings of the
Fourth European Conference on Antennas and Propagation (EuCAP 2010), Barcelona,
Spain. 12-16 April 2010, 1-5.
[17] Caorsi, S., Gragnani, G. L., & Pastorino, M. (1990). Two-dimensional microwave
imaging by a numerical inverse scattering solution. IEEE Transactions on Microwave
Theory and Techniques, 38(8), 981-989.
[18] Cappellin, C., Frandsen, A., & Breinbjerg, O. (2008). Application of the SWE-toPWE
antenna diagnostics technique to an offset reflector antenna. IEEE Antennas and Prop‐
agation Magazine, 50(5), 204-213.
[19] Cappellin, C., & Meincke, P. (2011). Diatool by TICRA. Brochure: http://
www.ticra.com/products/software/diatool-antenna-diagnostics.
[20] Coifman, R., Rokhlin, V., & Wandzuraz, S. (1993). The Fast Multipole Method for the
Wave Equation: A Pedestrian Prescription. IEEE Antennas and Propagation Magazine,
35(3), 7-11.
[21] Colton, D., Haddar, H., & Piana, M. (2003). The Linear Sampling Method in Inverse
Electromagnetic Scattering Theory. Inverse Problems, 19, 105-137.
[22] Eibert, T. F., & Schmidt, C. (2009). Multilevel fast multipole accelerated inverse equiva‐
lent current method employing Rao-Wilton-Glisson discretization of electric and mag‐
netic surface currents. IEEE Transactions on Antennas and Propagation, 57(4), 1178-1185.
[23] Engheta, N., Murphy, W. D., Rokhlin, V., & Vassiliou, M. S. (1992). The Fast Multi‐
pole Method (FMM) for Electromagnetic Scattering Problems. IEEE Transactions on
Antennas and Propagation, 40(6), 634-641.
[24] Farmahini, M. (2009). Fast and Accurate Cascaded Particle Swarm Optimization
Method for solving 2-D Inverse Scattering Problems. Journal of Applied Computational
Electromagnetic Society, ACES, 24(5), 1-4.
[25] Foged, L. J., Scialacqua, L., Saccardi, F., Sabbadini, M., Araque, J. L., & Vecchi, G.
(2011). Practical Application of the Equivalent Source Method as an Antenna Diag‐
nostics Tool. 2011 Antenna Measurement Techniques Conference (AMTA 2011), Denver,
Colorado (USA). 17-20 October, 1-5.
[26] Foged, L. J., Scialacqua, L., & Saccardi, . (2011). Microwave Vision Group, Brochure:,
http://www.microwavevision.com/sites/www.microwavevision.com/files/files/
plaquette_insight_BD.pdf.
[27] García, C., Álvarez, Y., Casas, B. A., & Las-Heras, F. (2011). Characterization of An‐
tenna Interaction with Scatterers by means of Equivalent Currents. Progress In Elec‐
tromagnetics Research, PIER 116, 185-202.
[28] Gatti, M. S., & Rahmat-Samii, Y. (1988). FFT applications to plane-polar near-field an‐
tenna measurements. IEEE Transactions on Antennas and Propagation, 36, 781-791.
[29] Habashy, T. M., Chow, E. Y., & Dudley, D. G. (1990). Profile inversion using the re‐
normalized source-type integral equation approach. IEEE Transactions on Antennas
and Propagation, 38(5), 668-682.
[30] Hajihashemi, M. R., & El-Shenawee, M. (2011). Inverse Scattering of Three-Dimen‐
sional PEC objects using the level-set method. Progress In Electromagnetics Research
PIER, 116, 23-47.
[31] Hansen, J. E. (1988). Spherical Near-field Antenna Measurements, IEE Electromagnetic
Waves Series 26, London: Peter Peregrinus Ltd., 10.1049/PBEW026E.
[32] Harrington, R. F. (2001). Time-Harmonic Electromagnetic Fields, John Wiley & Sons,
10.1109/9780470546710.
[33] Hellicar, A. D., Hanham, S. M., Hislop, G., & Du, J. (2009). Terahertz Imaging with
Antenna Coupled Detectors. Proceedings on 3rd European Conference on Antennas and
Propagation (EuCAP’09), Berlin, 23-27 March 2009, 1-5.
[34] Hernando, M., Fernández, A., Arias, M., Rodríguez, M., Álvarez, Y., & Las-Heras, F.
(2008). EMI Radiated Noise Measurements Using the Sources Reconstruction Techni‐
que. IEEE Transactions on Industrial Electronics, 55(9), 3258-3265.
[35] Hislop, G., Li, L., & Hellicar, A. (2009). Phase Retrieval for Millimeter- and Submillim‐
eter-Wave Imaging. IEEE Transactions on Antennas and Propagation, 57(1), 286-290.
[36] Kaplan, L., Hanfling, J. D., & Borgiotti, G. V. (1979). The backward transform of the
near field for reconstruction of aperture field. IEEE Antennas and Propagation Society
Symposium Digest, 764-767.
[37] Las-Heras, F. (2001). Using Equivalent Currents to Analyze Antennas in Complex
Environments. Microwave and Optical Technology Letters, 31(1), 62-65.
[38] Las-Heras, F., Pino, M. R., Loredo, S., Alvarez, Y., & Sarkar, T. K. (2006). Evaluating
near field radiation patterns of commercial antennas. IEEE Transactions on Antennas
and Propagation, 54, 2198-2207.
[39] Las-Heras, F., Galocha, B., & Álvarez, Y. (2009). On The Sources Reconstruction
Method Application For Array And Aperture Antennas Diagnostics. Microwave and
Optical Technology Letters, 51(7), 1664-1668.
[40] Las-Heras, F., & Sarkar, T. K. (2002). A direct optimization approach for source re‐
construction and NF-FF transformation using amplitude-only data. IEEE Transactions
on Antennas and Propagation, 50(4), 500-510.
[41] Laviada, J., Álvarez, Y., & Las-Heras, F. (2010). Efficient determination of the near-
field in the vicinity of an antenna for the determination of its safety perimeter. Prog‐
ress In Electromagnetics Research, PIER 103, 371-391.
[42] Lin, C. Y., & Kiang, Y. W. (1996). Inverse scattering for conductors by the equivalent
source method. IEEE Transactions on Antennas and Propagation, 44(3), 310-316.
[43] Livesay, D. E., & Chen, K.-M. (1974). Electromagnetic fields induced inside arbitrarily
shaped biological bodies. IEEE Transactions on Microwave Theory and Techniques,
22(12), 1273-1280.
[44] López, J. A., López, M., Álvarez, Y., García, C., Martínez, D., & Las-Heras, F. (2012).
Fast Antenna Characterization Using The Sources Reconstruction Method on Graph‐
ics Processors. Progress In Electromagnetics Research, PIER 126, 185-201.
[45] Lopez-Sanchez, J. M., & Fortuny-Guash, J. (2000). 3-D Radar Imaging Using Range
Migration Techniques. IEEE Transactions on Antennas and Propagation, 48(5), 728-737.
[46] Martinez, J. A., Gonzalez, B., Rappaport, C., Meana, J. G., & Pino, A. G. (2011). Re‐
constructing Distortions on Reflector Antennas With the Iterative-Field-Matrix Meth‐
od Using Near-Field Observation Data. IEEE Trans. on Antennas and Propagation,
59(6), 2434-2437.
[47] Mikki, S., & Kishk, A. (2007). Theory and Applications of Infinitesimal Dipole Mod‐
els for Computational Electromagnetics. IEEE Transactions on Antennas and Propaga‐
tions, 55(5), 1325-1337.
[48] Persson, K., & Gustafsson, M. (2005). Reconstruction of equivalent currents using a
near-field data transformation- with radome applications. Progress In Electromagnetics
Research, PIER 54, 179-198.
[49] Petre, P., & Sarkar, T. K. (1992). Planar near-field to far-field transformation using an
equivalent magnetic current approach. IEEE Transactions on Antennas and Propagation,
40(11), 1348-1356.
[50] Qin, Y. M., & Ciric, I. R. (1993). Inverse scattering solution with current modeling and
Tikhonov regularization. Proceedings on IEEE International Antennas and Propagation
Symposium, June 1993, 1, 492-495.
[51] Rahmat-Samii, Y. (1984). Surface diagnosis of large reflector antennas using micro‐
wave holography metrology. Radio Science, 19, 1205-1217.
[52] Rao, S. M., Wilton, D. R., & Glisson, A. W. (1982). Electromagnetic Scattering by Sur‐
faces of Arbitrary Shape. IEEE Transactions on Antennas and Propagation, AP-30(3),
409-418, May 1982.
[53] Razavi, S. F., & Rahmat-Samii, Y. (2007). A new look at phaseless planar near-field
measurements: limitations, simulations, measurements, and a hybrid solutions. IEEE
Antennas and Propagation Magazine, 49(2), 170-178.
[54] Sarkar, T. K., & Arvas, E. (1981). On the class of finite step iterative methods (Conju‐
gate Directions) for the solution of an operator equation arising in electromagnetics.
IEEE Transactions on Antennas and Propagation, 29, 370-373.
[55] Sarkar, T. K., Weiner, D., & Jain, K. V. (1981). Some Mathematical Considerations in
Dealing with the Inverse Problem. IEEE Transactions on Antennas and Propagation,
AP-29(2), March 1981.
[56] Sarkar, T. K., & Taaghol, A. (1999). Near-field to near/far-field transformation for ar‐
bitrary near-field geometry utilizing an equivalent current and MoM. IEEE Transac‐
[57] Sijher, T. S., & Kishk, A. (2005). Antenna modeling by infinitesimal dipoles using ge‐
netic algorithms. Progress In Electromagnetics Research, PIER 52, 225-254.
[58] Stratton, J. (1941). Electromagnetic theory. McGraw-Hill, New York.
[59] Van den Berg, P. M., & Kleinman, R. E. (1997). A contrast source inversion method.
Inverse Problems, 13, 1607-1620.
[60] Wang, H.-C., & Hwang, K. (1995). Multicoloring of grid-structured PDE solvers for
parallel execution on shared-memory multiprocessors. IEEE Transactions on Parallel
and Distributed Systems, 6(11), 1195-1025.
[61] Yaccarino, R. G., & Rahmat-Samii, Y. (1999). Phaseless bi-polar planar nearfield
measurements and diagnostics of array antennas. IEEE Transactions on Antennas and
Propagation, 47(3), 574-583.
[62] Yaccarino, R. G., Rahmat-Samii, Y., & Williams, L. I. (1994). The bi-polar planar near-
field measurement technique, part II: near-field to far-field transformation and holo‐
graphic imaging methods. IEEE Transactions on Antennas and Propagation, 42, 196-204.
[63] Yaghjian, A. D. (1986). An overview of near-field antenna measurements. IEEE Trans‐

actions on Antennas and Propagation, 34(1), 30-45.
[64] Zhao, K., Vouvakis, M. N., & Lee, J.-F. (2005). The Adaptive Cross Approximation
Algorithm for Accelerated Method of Moments Computations of EMC Problems.
IEEE Transactions on Electromagnetic Compatibility, 47(4), 763-773.
Section 3
Microwave Applications
Chapter 7
Multi-Hole Waveguide Directional Couplers
Mahmoud Moghavvemi,
Hossein Ameri Mahabadi and Farhang Alijani
1. Introduction
The directional couplers are inherently assumed as four-port devices, which consisted of
two transmission lines that are electromagnetically coupled to each other. The first port is
named as input, and the second one as output or transmitted, the third one as sampling or
coupled and the fourth one as isolated or terminated. By using a special design the input
power is divided between output and coupled port in a certain ratio named coupling factor.
The required value for coupling factor P1/P3 defines the range of applications for directional
couplers. Based on the application, coupling factor could be any value like 3, 6, 10, 20 dB
and even more. The performance of the directional coupler is usually evaluated by its direc‐
tivity between port 3 and 4.The directivity is a calculated parameter from isolation and cou‐
pling factor and shows how the two components of wave cancel each other at port 4.
Though we prefer to have high value for directivity as much as possible, but in real situation
this could be happened only around center frequency of designing band. The waveguide di‐
rectional couplers have a good directivity compared to microstrip or stripline couplers and
in spite of their bulky size, give us a low loss, high power handling, good characteristics and
low cost due to use of just a simple waveguide.
Nowadays the numerical methods are widely used for simulation and optimization of high
frequency structures. Some of them such as HFSS and FEKO, are well commercialized and
used widely by researchers and engineers.. But for designing procedure and for starting
point we need an initialization value to input into simulator and then optimize the parame‐
ters by its internal routines.
In this chapter we focus on the waveguide directional couplers and we try to give a good
reference as well as finalized designing formulas in closed form and tables to be used indi‐
vidually or as initial values for numerical software. The full generalized field theory and its
equations for designing based on multi-hole coupling structure will be introduced, though
more detailed basic information could be found in given references.
Moreover, by solving the equations, the recursive procedure is employed in a computer pro‐
gram to adjust the required directivity, coupling and frequency or waveguide type to speci‐
fy the number of coupling holes, individual diameters and locations of holes in waveguide’s
coordinate. Besides of those parameters, the length of coupler, matched load and other sizes
of structure for fabricating, will be defined too. By using the different methods like bino‐
mial, Chebyshev, and super imposed to calculate the coupling of each hole, the wide band‐
width response is achieved. At the end, a number of books and papers are given as good
references for further study.
Figure 1. An Ordinary Broad-Wall Waveguide Directional Coupler and its ports
1.1. Definitions
As mentioned, couplers are considered as 4-port passive devices in which, a part of input
wave reaches to output port 2 and the remained would be coupled to the coupled port 3.
Port 4 usually internally is matched to damp the residual internally reflected waves from
port 2 and port 3. Ideally there is no wave reach port 4. Port 4 usually is terminated by a full
band load as shown in Figure1 and Figure2.
In waveguide couplers, the coupling method is done by putting a waveguide on top of anoth‐
er one and by making some aperture holes in their common wall a determined portion of wave
would be leaked into the other waveguide. Though the waveguides axis and coupling aper‐
tures can be chosen arbitrary [2, 3], but for adequate specific usage and for easy derivation of
design equations, we consider that two waveguide’s are lay exactly on each other. Here the
broad-wall coupling configuration is more interested and concentrated. Though, for side-wall
the derivation of the design equations are so simple but the bandwidth is limited in spite of
higher power handling.
Multi-Hole Waveguide Directional Couplers 187
Figure 2. An equivalent 4-Port configuration for a directional coupler
Since 1945, extensive studies have been conducted on the issue and many researchers have
tried to optimize the designing equations to make the result more accurate. For instance, dif‐
ferent slot shapes had been introduced to increase the bandwidth and modifying the specifi‐
cations. [12, 25, 27]
In order to start the calculations, there are three major parameters that we need to define for
each coupler:
• Coupling factor “C” in dB, that represents the power received by port 3 as:
Pi Pf
C = 10log Pf or C = - 10log Pi
(1)
The coupling factor shows the ratio of power at port 3 to input power at port 1. Typically
we prefer to have one of the 3, 6, 10, 20 or 30 dB as standard values but for specific appli‐
cation it also can be defined freely.
• Directivity “D”, is the ratio of output power at port 3 to received power at port 4. Since
we prefer to eliminate the power at port 4, therefore the high values for “D” is more inter‐
ested. The ”D” in terms of dB is defined as:
Pf Pb
D = 10log Pb or D = - 10log Pf
(2)
• Bandwidth “BW”, which depends on directivity. By increasing the number of coupling

apertures, the order of coupler increases ( similar to the order of filter) hence the directivi‐
ty is increased. Meanwhile, higher bandwidth is also achieved. Therefore, by choosing the
required minimum directivity, D min, the available bandwidth is calculated.
For a 10 dB coupling or having a 0.1 of input power to port 3, we would have:
Pf 1
10log Pb = 10 → P f = 10 Pi (3)
In the same way, for a 3dB coupling, half of input power will receive to port 3:
Pi 1
10log Pf =3 → P f = 2 Pi (4)
And if we consider D = 40 dB for directivity:
Pf
10log Pb = 40 → P f = 10000Pb (5)
It is the adequate value for designing a good directional coupler.
A number of references, which have studied the couplers and have given the relationship
between number of aperture holes “n” and directivity “D” are listed in references [12, 25,
23]. In addition to number of aperture holes”n”, in the designing procedure for directional
couplers, certain parameters should be well defined as:
• Distances between the holes
• Distances between holes to side-wall (holes center offset from waveguide axis)
• The holes dimensions (diameter of holes for circular holes).
It has been shown that to have an optimum coupling around a certain frequency, the criteria
(6) should be kept in which “x” is the distance between the holes centers to the side-wall and
“a” is the broad wall size of waveguide: [24]
x
a ≤ 0.25 (6)
Furthermore, by precise study, the best design value for ratio of (6) is given as [23]
x
a = 0.203 (7)
Figure 3. Cross section of the directional coupler and coupling holes
The distances between holes should be about λg/4 however a question remains, what is the
proper value of λg when the bandwidth is limited to the λg1 to λg2interval? To answer the
question, there are three definitions used for λg:
1. λg is the average of wavelength of lower band λg1 and upper band λg2 so:
λg 1 + λg 2 λg 1
λg = 2 → 4 = 8 (λg 1 + λg 2) (8)
2. λg can be considered as geometric mean between λg 1 and λg 2 :
λg 1
λg = λg 1.λg 2 → 4 = 4 λg 1.λg 2 (9)
3. λg can be considered as mean value between λg 1 and λg 2:
2 1 1 2λg 1λg 2 λg λg 1λg 2

λg = λg 1 + λg 2 → λg = λg 1 + λg 2 → 4 = 2(λg + λg ) (10)
1 2
The best choice for defining the centers of two holes is the 3rd definition since it has been
practically approved too [23]. So the wavelength would be derived from (10). Therefore, in
order to define the dimensions of each hole (or diameter in case of circular hole type), the
each hole’s coupling should be calculated first, and the hole’s diameter would be derived
consequently.
1.2. Fields Equations
In order to calculate the coupling of each hole and by using the required Dmin that we need‐
here, the number of holes “n” will be derived in two different ways:
i- The coupling coefficient mapped to coefficients of nth order of Chebyshev polynomial.
ii- The coupling coefficient mapped to coefficients of nth order of Binomial polynomial.
By assuming the same order for polynomials “i” and “ii” and by noticing that the directivity
slope in case of “i” is higher, we expect to have higher bandwidth in comparizon to “ii” and
limited ripple in pass-band. In case of “ii” though there is no ripple in pass-band but the
slope of directivity is lower than “i” with same order of polynomial, therefore the band‐
width is lower than “i”. For years many of manufacturers chose the “i” method and consid‐
ering the number of holes n = 20. Here, the “i” method is chosen, however the number of
holes “n” would be defined from Dmin and it will be not fixed anymore.
In fabricating the couplers, any arbitrary shape for holes can be used but the circular; elliptic
and rounded-edge rectangle has been widely studied, simulated and used in research reports.
[30] Here, the circular holes have been adopted. The circular holes can be aligned in one, two or
three parallel rows, but in our case, 2-rows are used. In order to calculate the coupling coeffi‐
cients and related field equations the “Bethe’s small-hole coupling theory” is used as the main
computational method. Further, by using a correcting function, the theory is expanded to use
big-size holes as well [12, 26, 27]. In that way, Levi’s work would be followed to find the effect
of wall-thickness “t” and also the relationship between variations of directivity “D” and cou‐
pling error “ΔC”. [27]. Levi showed that if “D” increases, “ΔC” will decrease.
In special case, if we require high directivity “D”, like “D = 50 dB” for small bandwidth like
8.9 to 9 GHz, only 2 holes are needed to synthesis the coupler.
πx λ0
For calculating the distance of holes’ centers to side-wall “x”, the equation sin a = is used
6a
in which “a” is broad-side of waveguide and λ0 is the wavelength in the middle of the band. [2]
The coupled wave equations for incident wave A1 and reflected B1by assuming the same
amplitude for waves are as follows:
2π ( ) ( ) ( ) ( ) ( ) ( )
A1 = j abλg M x H x 1 H x 2 + M z H z 1 H z 2 - P E y1 E y2 (11)
2π ( ) ( ) ( ) ( ) ( ) ( )
B1 = j abλg -M x H x 1 H x 2 + M z H z 1 H z 2 - P E y1 E y2 (12)
In which, “a” and “b” are the waveguide dimensions, M x and M z are the magnetic polariza‐
( )
tion components in “x” and “z” axis and P is electrical polarization. The H x 1 is the ampli‐
( )
tude’s wave component in the first waveguide and H x 2 is for second and so on. If two
( ) ( ) ( ) ( )
waveguides are identical then H x 1 = H x2 and H z 1 = H z2 . Then the fields’ components are ex‐
pressed as:
πx
H x = - Sin a e - jγz (13)
λg πx
H z = j 2a Cos a e - jγz (14)
λg πx
Ey = λ Sin a e - jγz (15)
The field equations are given separately for “Narrow wall” and “Broad wall” cases. Here,
we briefly introduce them and give the relations for our interested one (i.e., Broad wall):
1.2.1. Narrow wall
By referring to Figure 5, since x = 0, the equations could be simplified as:
jλg M z
A1 = A2 = - (16)
2a 3b
In which, M z is independent from frequency. In other words, “Narrow wall” coupling has
significant difference comparing to “Broad wall” coupling.
Figure 4. The geometry of field equations and waveguide
Figure 5. Narrow wall coupling
1.2.2. Broad wall
This case is shown in Fig.6 and equations are expressed as following:
A1 =
j2π
abλg
{M - ( ) .P }sin
x
λg 2
λ
2 πx
a + Mz ( ) cos
λg 2
2a
2 πx
a
(17)
B1 =
j2π
abλg
{M + ( ) .P }sin
x
λg 2
λ
2 πx
a - Mz ( ) cos
λg 2
2a
2 πx
a
(18)
Figure 6. Broad-wall coupling
If the holes are circular with a diameter “d” the parameters “P, M x and M z ” are cal‐
culated as:
d3
Mx =Mz = 6
(19)
d3
P= 12
(20)
By putting (19 & 20) in (17 &18) we will have simplified form:
jπd 3
2a λg aλg
A1 = λg + 2a - (21)
12a 2b λ2
jπd 3
2a λg aλg
B1 = λg - 2a + (22)
12a 2b λ2
1 1 1
And if we put λ2
= λ2g
- (2a)2 we will have:
jπd 3
2a λg
A1 = λg + 2a
(23)
24a 2b
jπd 3
6a λg
B1 = λg - 2a
(24)
24a 2b
By putting λg = 2a then minimum value for A1 is calculated as:

3
jπd
A10 = (25)
12a 2b
And coupling variation to frequency changing is:
2a λg
+
λg 2a (26)
ΔC = 20log 2
On the other hand, variation of directivity to frequency is given by:
2a λg
A1 λg + 2a
D = 20log B1 = 20log 6a λg (27)
λg - 2a
The solved equations for R100 waveguide is given in Fig.7
Figure 7. Theoretical coupling variation and directivity for transverse arrangements of 1, 2 or 3 holes in common
broad-wall of R100 waveguide.[14]
When a number of single holes are aligned in a row, the array’s directivity will be added to
basic value of directivity.
As it has been shown earlier, the Broad-wall couplers have better characteristics in wide fre‐
quency spans comparing to Narrow-wall couplers while the power handling characteristic
of Narrow-wall couplers are better.

1.3. Coupling by big holes
The equations given in section 1.3 were valid for small holes. By considering the wall-thick‐
ness “t” and big hole’s by surface size of “A”, the equations should be corrected. This has
already done and the results will be used now. [15]
1.3.1. Corrections for wall thickness “t”
When the coupling holes have thickness “t”, it equals to a short-length circular waveguide,
which is working under its cut-off frequency (when frequency is lower than resonance fre‐
quency). Therefore the amplitude of the coupled wave attenuates by following term:
exp { -2πtA
λc 1-( )
λc 2
λ
1
2 } (28)
In which, the λc is the cut-off wavelength of hole and λ is the wavelength in operating fre‐
quency, “t ” is the wall-thickness and “A” is the correction factor which includes the interac‐
tions of adjacent fields and is defined practically. [15] For elliptic holes (Narrow slot) the
“A” factor is in 3rd order but for circular holes it is close to first order. If the fields have multi
components, correction factor (28) should be applied on all of the Bethe’s equations. It
means that in equations (17) and (18) the magnetic field correction factor for coupling have a
parameter λc that is due to excitation by TE 11 mode so for this mode λc → λH = 1.705d and
correction factor for electric field coupling uses a λc , which relates to TM 01 mode and equals
to λc → λE = 1.305d.
1.3.2. Corrections for big size holes
A good study for such big size holes have been done by an equivalent circuit. [15] In this
way, the hole’s effect will be simulated by a two port network in parallel or in series be‐
tween two waveguides which have mutual coupling. This network is a lossless and should
be defined by Foster’s reactance theorem. Here, the impedance will be defined by locations
of its zeroes and poles in addition to the multiplications coefficients.
In simple expression of Bethe’s small sized coupling hole theory, this impedance has a sim‐
ple form. For example the reactance of a small hole in a thin diaphragm at the cross section
of a rectangular waveguide, working in TE10 is:
4πM Z 0
X= abλg
(29)
In which, the “a” and “b” are the waveguide dimensions that have been shown in Fig.4
The term λg is the guided wavelength, “M” magnetic polarization, Z0 characteristic impe‐
λg
dance of the waveguide. The Z0 has a direct relation to term λ , which shows that lumped
reactance “X” has a direct relation to frequency (X ∝ f) too.
Therefore the small hole coupling theory assumes that the “X” would be a constant reac‐
tance but it is not true. Because, there are a few numbers of unwanted resonances occurred
in the frequency band. For this reason the (29) would be a good definition when the operat‐
ing frequency is somehow lower than the first resonance. For considering the resonance ef‐
fect in equation (29), the corrected “M” would be expressed by introducing a new term that
M
considers the effect of cut-off wavelength
1- ( ) and the result is as follows:
f 2
f c2
X 4πM
=
Z0
( ( ))
abλg 1 -
f 2
f c2
(30)
From measurements, it has been shown that the above correction factor gives a good ap‐
proximation.
The attenuation definition (28) can be combined to (30) to give us a general correction factor
for big size holes:
exp
{ - 2πtA
λc
( )
1-
f 2
f c2
1
}
2
(31)
1-( )f 2
f c2
1.4. Multi holes coupling
λg
A longitudinal coupling consists of a series of holes by center distance of 4 that has a great
coupling in forward and weak coupling in backward direction.
The slight coupling for a single hole has been studied and the directivity introduced by:
A1
Directivity (dB ) = 20log B1
(32)
In which: [15]
L /2
A1 ∫ ϕ( x )dx
B1 = L /2
-L /2
(33)
∫ ϕ( x )exp ( - 2 jβx )dx
-L /2
In the Fig.8 a series of “n” holes in one row is shown. The coupling voltage of the series is
named a1, a2,...,an.
Figure 8. The cross section of n-Hole array and coupling coefficients
All the hole’s center distances and electrical length are the same and are considered in the
middle of the band. If the input wave to port 1 has constant amplitude and matched to other
3 ports, the reflected wave can be expressed by:
B1 = a1 + a2Exp (-2 jϕ ) + a3Exp (-4 jϕ ) + … + an Exp (-2(n - 1) jϕ ) (34)
The interesting and useful case is when the coefficients of the series being symmetrical from
center. Therefore:
a1 = an , a2 = an-1, aK = an-K +1 (35)
So by putting the values (35) into (34):
B1 = { 2a1cos (n - 1)ϕ + 2a2cos (n - 2)ϕ + … + 2an/2cos ϕ e
2a1cos (n - 1)ϕ + 2a2cos (n - 2)ϕ + … +

an+1
2 e j (n-1)ϕ
j (n-1)ϕ
n even
n odd
(36)
The direct coupled wave at port 3 will be:
j (n-1)ϕ
A1 = ∑nr =1 ar e (37)
And the directivity “D” is calculated by normalizing B1 to A1 in (36) by dividing the sum of
each coupling voltages. In special case if there are “n” identical holes, therefore:
B1 sin nϕ
A1 = n sin ϕ
(38)
2. Design methods based on arrays
2.1. Chebyshev Array
If the minimum voltage over the full bandwidth to reach a good directivity “D” is needed,
the Chebyshev polynomial can be used for distribution function of each hole’s voltage. Such
coefficients are derived by putting the B1 in (36) by considering the following equal ripple’s
directivity function as following:
B1 = amT n-1 ( cos ϕ

cos ϕ0 ) (39)
In which the am is the maximum of B1 over the coupling bandwidth that given by following:
ϕ0 ≤ ϕ ≤ π - ϕ0 (40)
The am is calculated by putting the ϕ = 0 in (39):
n
∑ ar
| A1|
am = (41)
)=T (
r =1
T n-1 ( 1
cos ϕ0
1
n-1 cos ϕ0
)
In (36) if we put ϕ = 0:
n
B1 = ∑ ar (42)
r=1
Therefore the minimum directivity over the bandwidth would be:
Dmin = 20log T n-1 ( 1

cos ϕ0 ) (43)
Comparing this method to method of Binomial polynomial is very informative that has been
done by Levi. In this case we should have: [16]
B1 = am ' ( cos ϕ n-1

cos ϕ0 ) (44)
In which:
n
∑ ar
| A1|
am ' =
r =1
= (45)
( 1
cos ϕ0
n-1
) ( 1
cos ϕ0
n-1
)
The minimum directivity at the edge of the band for this case is:
Dmin = 20log ( 1
cos ϕ0 )
n-1
(46)
Obviously the (46) always is significantly lower than the value for Chebyshev case (43).
The coupling equation for Chebyshev case is derived by putting the identical coefficients of
cos ϕ in (36). Young gave such coefficients for 3 ≤ n ≤ 8 [3]. But here, the generalized case is
obtained by a computer program for 1 ≤ n ≤ 25.
For coupling C = 0 these coefficients are changed into Pascal’s triangle that for C = 0 the infin‐
ite directivity over a zero bandwidth obtained.
The hole’s size is derived by coupling of each hole in dB. That relation for r th
hole
is as follows:
( )
n
∑ ar
Cr = 20log
r =1 (47)
ar
n
Since ∑ Cr = 1, all the theoretically given hole couplings, transferred all power by assum‐
r=1
ing the 0dB in the formula. Therefore in order to design a “C dB” coupler the “C” is
added to C r in (47). The entire hole sizes by this way and by given theory for small
size holes (or if we need by using the correction coefficient curves given by referen‐
ces) can be computed. [17, 18]
In addition to both mentioned series for calculating the coefficients (Chebyshev and Bi‐
nomial), there is another method that actually derived from them. It is named “Super
Imposed Arrays”.
2.2. Super Imposed Array
When the strong coupling is needed, i.e. 3dB or 6dB, it is not possible to use the one row of
holes (single array), since diameter of holes will be increased. Therefore it is more conven‐
ient to have approximately same diameter for all to get good coupling quality. For this case
the super imposed array is used. As first step, we need the coefficient series in which the
holes get bigger. It would be happened when n > 4. For starting we can use Chebyshev or
binomial coefficient series in one line. Then the same series should be written in second line
but in shifted position. It means, first coefficient of line 2 in under the 4th coefficient of line 1
and so on. By adding the two lines we would have a new series that its coefficients (or holes’
diameters) alternately are the same. For example by a 6-element binomial series, we can
make a 9-element super imposed series:
As it has shown, the elements in new series are alternately identical. This can be done by
any other number of elements or polynomials. If we wanted to add more number of
holesTable 1, the same way is chosen:
1 5 10 10 5 1
1 5 10 10 5 1
1 5 10 11 10 11 10 5 1
Table 1. A 9-element super imposed array based on 6-element binomial array
1 5 10 11 10 11 10 5 1
1 5 10 10 5 1
1 5 10 11 10 11 11 10 11 10 5 1
Table 2. Added more number of elements to table 1.
For special characteristics of Chebyshev series, hereinafter, it will be used as basic polyno‐
mial for further design. In super imposed method the number of elements can be any value
and we would have longer coupler. If we chose other two methods the the holes diameter
get bigger and bigger and it may exceeds the broad-wall size of waveguide. By considering
n
the ∑ Cr = 1it is cleared that each C rmust be less than 1 and in super imposed this would be
r=1
happened.
For strong couplings and the holes with the same shape the following relation is proposed
by Cohn: [14]
Am = sin ( mπ
4m3dB ) (48)
In (48) the m3dB is the number of series for obtaining the 3 dB coupling. For example, if a 6-
element series is needed for 3dB coupling, therefore the coupling for one series will be:
A1 = sin ( 24π ) = 0.1305 (49)
That equals to 17.68 dB. The directivity for Chebyshev-based super imposed array is greater
than the single array. The reason can be explained by this fact that returned waves are add‐
ed in phase and amplitude, so the maximum amplitude for returned wave will not exceed
from single array in any case and it will damped soon.
2.3. Transverse groups of holes
It maybe concluded that by using two or three rows of holes in broad wall, the stronger cou‐
pling would be obtained. It is true if there are two rows and the distance of center of holes to
side walls being equal to x = 0.25a but for three rows the result is not good. For two rows the
coupling and directivity are derived from (23) and (27). See Fig. 9
Figure 9. Groups of two or three rows holes.
The test results of multi-hole couplers are as following:
A 6dB coupler by 1 " × 1 / 2 " waveguide in 8.2~12.4GHz band is designed by 21-elements su‐
per imposed array based on 6-element Chebyshev in two rows (totally 42 holes). The dis‐
tance from side-wall for circular-holes is x = 0.25a = a / 4 and distance between hole centers is
λg / 4. The obtained directivity “D” is more than 40dB and coupling deviation ∆ C is about
±0.5dB. [19]
Shelton has tested multi-rows couplers and has given the coupling curve in terms of holes
diameters for X and Ku bands waveguides. His efforts by using 3 rows were not successful.
[20] For 1 " × 1 / 2 " waveguide, Cohn used rounded rectangular holes in two rows. His re‐
search approved that 2-rows is better than 3-rows and for shortening the length it is not pos‐
sible to use 3-rows holes.
It was also declared that in case of 3-rows, the resonance in upper band will happen. The
reason is, where the electric vector is in parallel to broad wall, the even mode is excited and
coupler acts as side wall coupler. Such case is not happened for 2-rows couplers. By reduc‐
ing the height of “b” it is possible to put the resonance frequency of even modes out of oper‐
ating frequency band. Only a slight reduction in “b” is needed since the resonance is
occurred when the coupling region has length of λg / 2 and it is near cut-off for even modes.
Indeed, as an example, the “b” should be reduced from 0.4” to 1/16”. This reduction in “b”
increases the coupling of each hole. See (17). So a few number of holes needed to make 3dB
coupling. As an example, a 2nd order Chebyshev transformer has a theoretical VSWR of 1.01
over its bandwidth. For better matching the waveguide height should be reduced at the two
ends. Each series of holes at each side of coupler should be located inside of transformer in
the way that it does not change the length. The final length would be 3.9 inches consisting of
2-rows of 10 holes that gives a coupling of 3 ± 0.5dB and Isolation of more than 30dB for X-
band 8.2~12.4GHz.
3. Practical designing
3.1. A real sample
After reviewing the basics of directional coupler, we start to design a coupler practically. First
of all it is better to introduce the abbreviations that we use. They are listed in following table:
C = Coupling in dB
λg = Guided wavelength at the lower end of the required bandwidth (mm)

1
λg = Guided wavelength at the upper end of the required bandwidth (mm)

2
λg = Mean guided wavelength

mid
N = number of coupling elements in basic array
Dmin = minimum directivity (dB)
/
λg
1
= 180 1 + λg (deg)
2
λg = Guided wavelength (at the center frequency of the wave-guide bandwidth) (mm)
X = Axis across broad dimensional of waveguide
A = Broad dimension of a waveguide wall (mm)
B = Narrow dimension of waveguide wall (mm)
d = Diameter of hole in millimeter (mm)
A' =1 - ( λ0 )
1.71d 2
Term giving correction of resonance phenomena
λ0 = free space wavelength (mm)
T = wall thickness (mm)
= 32 ( dt ) 1 - ( 1.71d
λ ) 1/2 term giving correction to the attenuation effect on a finite wall
A'' 0
thickness
X0 = 1 / cos
Table 3. The terms and abbreviations that used in design procedure.
As it is mentioned before in (10):
λg λg 1λg 2
4
mid
= 2(λg + λg ) (50)
1 2
The number of holes can be defined by minimum directivity Dmin as: [23]
Dmin
20
n = 1 + cosh -1 10 (51)
cosh-1 ( cos1 ϕ )
The starting coefficient in Chebyshev argument is calculated as:
( )
1 1
X0= cos ϕ = 180
cos λg
1
(52)
1 + λ
g2
For X-band that we have λg1 = 6.089 Cm and λg2 = 2.489 Cmthe X 0 = 1.853 is obtained. Next step
is to find the Chebyshev polynomial coefficients by computer program that gives:
{40.507, 172.277, 355.449, 445.373} (Notice that, only a half of the coefficients are enough due
to symmetric specification of Chebyshev polynomial).
Then the coefficients are normalized to least element that gives following table:
A B C D
1.0 4.253 8.775 10.995
Table 4. Normalizing the Chebyshev coefficients.
Therefore the whole structure of the holes will be as follows:
A B C D C B A
Table 5. Sequence of holes and its related Chebyshev coefficients for coupler synthesis.
Now we add them all together:
2( A + B + C ) + D = 39.051 (53)
The coupling for each hole will be defined in dB as follows:

39.051
Coupling for Holes A = 20log 1 dB = 31.832 dB
39.051
Coupling for Holes B = 20log 4.253 dB = 19.259 dB
39.051
Coupling for Holes C = 20log 8.775 dB = 13.968 dB
39.051
Coupling for Holes D = 20log 10.051 dB = 11.776 dB
Now, consider that we want to design a 10dB coupler, so we add a 10dB to each coefficient:
CA=41.832 dB, Holes CB=29.259 dB, Holes CC=23.968 dB, Holes CD=21.776 dB
Finally the achieved numbers should be inserted into Bethe’s formula for small size holes: [12]
C = 20log { 12a 2b
πd 3 1- ( λ0 ) } + 20log {32( ) 1 - (
1.71d 2 t
d
1.71d
λ0 ) }
1/2
(54)
Now we solve the above equation (for each hole) by iteration method and the diameter of
each hole would be determined. By considering the distance of circle centers to side wall as
x = 0.203a (7) following values for diameters would be obtained:
A=0.234 inch, B=0.343 inch, C=0.397 inch, D=0.421 inch
Note that the solved example is for single array. If we wanted to have the double rows we
should put the (C+6) dB instead of C dB (that we considered 10 dB in above example).
Notice: an approximation way to define the number of holes “n” is using the Dmin in equal
to maximum coupling between holes plus 3 ~ 5 dB. For instance in the solved example, the
maximum coupling is belonged to “A” that was C A=41.832 dB. So:
Dmin = C A + 5 = 47 dB (55)
And the number of holes would be:
47
20
n = 1 + cosh-1
10
-1 ≈ 6.9 → n = 7 (56)
cosh X0
Therefore if we wanted to have a good directivity, a directivity higher than 47dB then we
should have 7 holes in the coupler.
In practice, for eliminating the effect of wall thickness “t”, it is possible to remove one broad
wall of a waveguide and mill- the next wall to have half thickness between to waveguides. [23]
The real designed 20 dB coupler by R70 waveguide and 14 holes in two rows (each row has
7 holes) is fabricated and tested. The results are given in Fig.10
In Fig. 11 and 12, another directional coupler for C = 10 dB is sketched. The diameter of cir‐
cles for its five categories of holes, are:
Holes number 4 =4.16 mm
For further information see Fig. 11 and 12.
Figure 10. The test result for a multi-hole waveguide R70, directional coupler C=20dB
Figure 11. The R70 directional coupler, C = 10 dB by two rows of 9-elements
Figure 12. The circle centers and distance to side walls are the same.
Figure 13. Cross section for a 38-holes (double rows of 19-elements) directional coupler. The lossy load maching using
ferro-based materials (courtesy H. Mottaghi: [email protected])
The port-4 in the couplers is matched by a conical or pyramidal load. To obtain the lowest
reflection from this port, the tapered structure is used to make a slight change in characteris‐
tic impedance of the waveguide along its length. Since the power loss at this port, naturally
is not significant, the medium or low power absorber or load is enough.
Author details
Mahmoud Moghavvemi, Hossein Ameri Mahabadi and Farhang Alijani
Department of Electrical Engineering, University of Malaya (UM), Malaysia

References
[1] Gandhi, O. P. (1981). Microwave Engineering and Applications, Pergamon Press.
[2] Collin, R. E. (1966). Foundations for Microwave Engineering, Mc. Graw Hill.
[3] Young, Leo. (1966). Advances in Microwave, volume I, Academic Press.
[4] Miller, S. E. (1954). Coupled Wave Theory and Waveguide Applications. Bell system
Tech. J., 33, 661.
[5] Riblet, H. J. (1954). The short- slot hybrid Junction. Proc, IRE, 40, 180.
[6] Tomiyasu, K., & Cohn, S. B. (1953). The Transvar Directional Coupler. Proc. IRE, 41,
922.
[7] Cook, J. S. (1955). Tapered velocity couplers. Bell system Tech. J., 34, 807-822.
[8] Fox, A. G. (1955). Wave coupling by warped normal modes. Bell System Tech. J., 34,
823.
[9] Louisell, W. H. (1955). Analysis of a single tapered mode. Bell System Tech. J., 34, 853.
[10] Miller, S. E., & Mumford, W. W. (1952). Multi-element directional couplers. Proc. IRE,
40, 1070.
[11] Bethe, H. A. (1943). Theory of side windows wave guides. Mass. Inst. Technology Radi‐
ation Lab Report [43].
[12] Bethe, H. A. (1944). Theory of diffraction by small holes. Phy. Rev. , 66, 163.
[13] Surdin, M. (1946). Directive couplers in wave guides. J. Inst. Elec. Engrs (London) ,
93(IIIA), 725.
[14] Cohn, S. B., Knootz, R. H. S., & Wehn, S. L. (1962). Microwave hybrid coupler study
program. Rantec- Corporation Report [61], 361-363.
[15] Cohn, S. B. (1952). Microwave coupling by large aperture. Proc IRE, 40, 697.
[16] Levy, R. H. (1959). Guide to the practical application of Chebyshev functions to the
design of microwave components. Proc. Inst. Elec. Engrs. (London) C, 106, 193.
[17] Harrison, J. R. (1945). Design considerations for directional couplers. MIT Radiation
Lab. Rept. [724].
[18] Saad, T. (1964). The Microwave Engineers handbook. 109 Boston, Massachusetts.
[19] Hensperger, E. S. (1959). The design of multi-hole coupling arrays. Microwave J., 2, 38.
[20] Shelton, W. (1961). Compact multi-hole waveguide directional couplers. Microwave J.,
4, 89.
[21] Geranpaye. (1985). Waveguide standards. ITRC, Tehran, Iran.

[22] Alison, W. B. W. A. (1987). A handbook for the mechanical tolerance of waveguide

components. Artech house.
[23] Hancock, K. E. (1967). The design and manufacture of waveguide Tchebyshev direc‐
tional-coupler. Electron. Eng , 39, 292.
[24] Liang, C. H., & Cheng, D. K. (1982). On performance limitations of aperture coupling
between rectangular waveguide. MTT-30 [5].
[25] Levy, R. (1968). Analysis and synthesis of waveguide multi-aperture directional cou‐
plers. MTT-16 [12].
[26] Mc Donald, N. A. (1972). Electric and Magnetic coupling through small apertures in
shield wall of any thickness. MTT-20 [10].
[27] Levy, R. (1980). Improved single and multi- aperture waveguide coupling theory in‐
cluding explanation of mutual interactions. MTT-28 [4].
[28] Riblet, H. J. (1947). A Mathematical theory of differential coupler. Proc. IRE, 35, 1307.
[29] Crompton, J. W. (1957). A contribution to the design of multi-element directional

couplers. Proc. IEE, 104C(6TD), 398-402.
[30] Matthaei, G. L. (1964). Microwave Filters impedance matching networks and coupling
structures, Mc Graw Hill.
[31] Wheeler, H. A. (1964). Coupling holes between resonant cavities or waveguides eval‐
uation terms of volume ratios. MTT, 231.
[32] Sangster, A. J. (1973). Slot Coupling between uniform rectangular waveguides.

MTT-27 [7], 705.
[33] Pandhuri, Pande. V. H., & Das, B. W. (1979). comments on coupling of waveguides
through large aperture. MTT-27 [7], 702.
[34] Pandhuri, Pande. V. H., & Das, B. W. (1978). coupling of waveguide through large
aperture. MTT-26 [3], 209.
[35] Schmiedel, H., & Arndt, F. (1986). Field Theory Design of Rectangular Waveguide
Multiple-Slot Narrow-Wall couplers. MTT-34 [7], 791.
[36] Oliner, A. H. (1960). Equivalent circuits for small symmetrical longitudinal aperture
and obstacles, . IRE MTT, 72.
[37] Hakkak, M., & Barakat, F. (1984). Waveguide bend characteristics. ITRC publication‐
sTehran, Iran.
[38] Mottaghi, H. (2010). Measuring the Load absorption parameters in waveguide.

IT&mmwaves SDN BHD, Kuala Lumpur.
Chapter 8
The Diffraction of Electromagnetic Waves on the

Periodic Heterogeneities and Its Use for Realization of
Practical Technical and Electronic Devices of Millimeter
and Submillimeter Wavelength Range
Gennadiy Vorobyov, Larissa Vietzorreck,

Ivan Barsuk and Aleksandr Rybalko
1. Introduction
Among the open structures, which are used in millimeter and submillimeter (MSM) wave en‐
gineering, diffraction gratings (DG) made in different modifications (periodic metal and met‐
al-dielectric structures (MDS)) are of primary importance along with open cavities and open
waveguides. Such systems are basic in the design of electromagnetic oscillation sources and
electronic components of different instrumentation of such wavelength range. If there is a dif‐
fraction of electromagnetic fields by DG, “two-act” wave transformation usually takes place.
When homogeneous plane wave falls on the plane one-dimensionally periodic grating, scat‐
tered field can be considered as a spectrum of homo- and heterogeneous plane waves. In this
case body (incident) plane wave is transformed into body (scattered) homogeneous plane and
heterogeneous (surface) waves and, thus, “two-act” transformation occurs. This type of the
boundary-value problems has been thoroughly studied in the work [1] and partly realized in
the experiment [2]. In addition, processes of surface wave transformation of distributed sour‐
ces into body waves by periodic heterogeneities are of special interest. Such phenomenon can
be watched when an electron beam (EB) moves uniformly near the metal DG or periodic MDS.
In this case self-surface field of the EB is scattered by DG and at least one of its harmonics is
transformed into body wave of the diffraction radiation or Cherenkov radiation. It should be
noted, that transformation of the surface wave of EB by DG into the diffraction radiation is also
an example of the “two-act” diffraction process. In addition, phenomena, connected with the
transformation of DG of the surface waves of a dielectric waveguide (DW), play a great role in
MSM engineering. In this case surface waves of a DW are transformed by the DG into surface
waves of the DW or into body waves separated from them.
Multilinked quasi-optical systems in different modifications are very perspective in design‐

ing fundamentally new electronic devices and electronic components of instrumentation of
MSM wavelength range including infrared waves. One should mentioned, that they are
partly investigated by numerically-analytic methods in the approximation of constant cur‐
rent [1] and using the experimental modelling. It was determined that systems with MDS
included in the open cavities structure can have qualitatively new properties [3] that makes
it possible to propose high-frequency filters, frequency stabilizers, energy output devices
from the volume of open cavities and new types of semiconductor oscillators on its basis. A
possibility of the development of a Smith-Parcell amplifier was considered in paper [4].
Huge number of the previously proposed systems isn’t accompanied with its complex re‐
search. This fact, therefore, makes the process of practical realization of new modifications
of electronic device schemes and electronic components of MSM wavelength range on its ba‐
sis very slow. So far existed numerical methods of optimization of the three-dimensional su‐
perhigh frequency structures [5, 6] enable to analyze effectively electrodynamic
characteristics only of some multilinked quasi-optical system elements such as reflecting
MDS. Meanwhile, a comprehensive solving problem of optimization of such structures re‐
quires the huge expenditure of computer time and computational power with usually am‐
biguous results in the end. That is why problems of development of the universal
experimental facility, general technique of experimental modelling of the electromagnetic
phenomena in multilinked quasi-optical systems and its realization for studying electromag‐
netic effects in complex quasi-optical systems using so far existed numerically-analytic
methods, are very topical.
In the current work general electrodynamic characteristics of the multilinked quasi-optical

system coupling elements have been determined on the basis of previous theoretical and ex‐
perimental results in research of the simplest types of radiating system, formed by single-
row DG, and using experimental modelling of transformation of the DW surface waves into
body waves by two-row DG of different modifications. Schemes of practical microwave en‐
gineering devices have been suggested on basis of these characteristics.
2. An experimental setup and measuring technique
The use of periodic MDS formed by a strip metal grating on a dielectric layer is promising
for both constructing extremely high-frequency electric vacuum devices (optical coupler and
diffraction radiation generator [7, 8]) and integrated forms of various functional assemblies
and devices, operating in MSM ranges, including the terahertz range [9]. In contrast to re‐
flecting metal gratings, the MDS have a number of specific features related to a possibility of
exciting spatial waves of Cherenkov, normal, and abnormal diffraction radiation, when an
The Diffraction of Electromagnetic Waves 211
EB moves along the MDS [1]. Therefore, it is necessary to have exhaustive information on
electrodynamic characteristics of the DG and MDS, when they are used in various devices.
Paper [1] presents a method for identifying properties of waves of the space charge of an EB
traveling along a periodic structure, to the surface wave of the DW. This method has found
a wide utility in simulating electrodynamic characteristics of various radiation modes of
spatial waves on periodic MDS and on normal metal gratings. The setup for studying con‐
versions of DW surface waves into body waves on normal metal gratings is described in
[10]. However, its application for studying the MDS calls for modernization of the measur‐
ing part of the setup, taking into account special features of the analyzed object (possibility
of the presence of body waves both in dielectric and outside).
Below, a universal setup that can be used for analyzing electrodynamic characteristics of
both the MDS and traditional metal periodic structures is described. Results of test measure‐
ments are compared to the numerical analysis.
The complex experimental test bench for determining electrodynamic characteristics of peri‐
odic heterogeneities (MDS, strip and reflecting metal gratings), when they are excited by the
DW surface wave, consists of two main modules (figure 1): (1) module for measuring wave‐
guide characteristics (standing wave ratios (SWR), attenuation constants, etc.), and (2) mod‐
ule for measuring spatial characteristics of periodic structures (directional radiation patterns
in the far-field zone and amplitude distributions of fields in the near-field zone). The mod‐
ule for measuring spatial characteristics includes the studied object, which is generally die‐
lectric prism 1 with the strip DG 2 imposed on its side surface. Prism 1 is attached to the
special adjusting unit intended to spatially orient it in the x-, y-, and z-directions with an er‐
ror of ±0.1 mm. DW 3 joins in matching junctions 4, which through waveguides of specified
sections, determined by the studied wavelength range, are mated with the waveguide char‐
acteristics recording unit and matched load 5. Depending on characteristics of the DW
(phase velocity of the surface wave νph) and MDS (strip grating 2 with period l and dielectric
permittivity ε of prism 1), there are two possible modes of transforming the DW surface
waves into body waves, which are excited with indices n = 0, ±1, ±2, … [1]. The main ones
are shown schematically in figure 1 as directional radiation patterns (γ are the angles of ra‐
diation of the MDS-DW system, and ϕ are the angles of corresponding radiation harmonics
in the free space after transmission through the dielectric prism).
The module for measuring spatial characteristics of the object consists of two versatile horn
antennas 6, rotation axes of which pass in the E-plane through the radiation aperture and
are brought into coincidence with the vertical axis of grating 2, and the rotation axis in the
H-plane coincides with the DW longitudinal axis 3. This is intended to record radiation an‐
gles in interval ϕ = 10°–170° with error Δϕ = ±0.25°, when the horns are mounted on special
precision movable units and installed in the far-field zone, which is determined by the
known ratio z = a m /λ, where a m is the maximal size of the antenna aperture, and λ is the
radiation wavelength. In the course of measuring radiation patterns, the signal received by
horn 6 arrived at input Y of the two-coordinate GP, and input X of the GP was connected to
the rate-of-turn sensor of the receiving horn. Thus, when the movable antenna travels, the
directional radiation pattern is fixed on the GP, and, after conversion into digital form, it is
used for the computer processing of the data received.
Figure 1. Block diagram of the experimental setup: (SFG) sweep-frequency generator, (APR) automatic power regula‐
tor, (ADC) analog-to-digital converter, (PC) personal computer, (GP) two-coordinate graph plotter. 1 ― dielectric
prism, 2 ― strip DG, 3 ― DW, 4 ― matching junctions, 5 ― matched load, 6 ― horn antenna, 7 ― probe, 8 ― (DC)
directional couplers, γ ― radiation angles of the MDS-DW system, φ ― angles of corresponding radiation harmonics in
the free space after transmission through the dielectric prism, x, y, and z ― spatial coordinates, Δ ― thickness of the
dielectric layer, α ― distance between the DW and strip grating 2 (aiming parameter).
The amplitude field distributions along the axes of the radiating system are studied in the
near-field zone (z ≈ λ) by probe 7, made as a dielectric wedge (ε = 2.05), conjugated with the
standard waveguide measuring section via the matching junction. Upon detecting, the sig‐
nal from probe 7 arrives at the GP and is processed by the computer. Typical sizes of the
probe of ~ (0.1–0.2)λ ensured minimal distortions of the fields during measurements. The
surface field indication system was installed on the transportation carriage, and, in this case,
samples in the x-, y-, and z-directions were obtained with an error of ~ 0.1 mm.
On the whole, all indication systems of spatial characteristics and studied object were placed
on a single solid laboratory platform, illuminated by special precision adjusting units. This
allowed us to orient the regular part of the DW with respect to the MDS plane with an error
of ~ 0.1 mm and ensure the corresponding monitoring of the coordinates of measuring ele‐
ments of the setup and studied object.
The module for measuring waveguide characteristics (figure 1) is based on a standard pan‐
oramic SWR and attenuation constant measurer, consisting of the SFG, SWR and attenuation
constant indicator, and DC 8 with detector sections connected to corresponding connectors
of the SWR indicator. Depending on the method of bringing the couplers into the measuring
line, plots of either transmission gain or SWR were determined in the specified frequency
range. The obtained results were processed by the designed ADC and arrived via the USB
bus at the PC for further processing. The constant power level at the input to the studied
object was kept by the APR, being the part of the panoramic measurer, and the minimal re‐
flections at the DW output were ensured by inserting matched load 5 into the measuring
section (when absolute power levels were measured in the transmission line, standard watt‐
meters inserted directly into the measuring section instead of matched load 5 were used).
Procedure of performing measurements on the setup (figure 1):
(1) The matching of DW 3 with the module for measuring waveguide characteristics, which
consists in reaching 1.1- to 1.2 SWR values for the specified frequency band by optimizing
parameters of matching junctions 4. (2) Determination of the relative velocity of DW surface
wave βw = νph/c (c is the velocity of light) and stray radiation level from matching elements 4.
For this purpose, the calibration reflecting DG, characteristics of which are intended for ra‐
diation in the normal direction (angle γ–1ν in figure 1) in accordance with the procedure [10],
and indication system of fields in the far-field zone (horn antennas 6) are used. (3) Determi‐
nation according to the procedure [1] of the optimal value of the aiming parameter a from
the minimal distortion of directional lobes at the central frequency. (4) Determination of
MDS and DW characteristics for the specified excitation modes of body waves in accordance
with relationships of [1]. (5) Replacement of the calibration reflecting grating by the studied
MDS. (6) Measurements of spatial MDS characteristics with the simultaneous automatic con‐
trol of its waveguide characteristics.
The described experimental setup (figure 1) is implemented for a 53- to 80-GHz band, and
this fact determined the selection, as a panoramic SWR and attenuation constant measurer,
of a corresponding device and waveguide sections with a 3.6×1.8-mm2 cross section. The flu‐
oroplastic DW with a 5.2×2.6-mm2 cross section permitted us to obtain relative phase veloci‐
ties of the surface wave in interval βw = 0.78–0.81 and to excite on the fluoroplastic MDS two
main spatial radiation harmonics, namely, n = 0 (l = 1.17 mm) and n = –1 (l=3.07 mm). The
performed test measurements of SWR, directional radiation patterns, and amplitude field
distributions gave satisfactory results, as compared with the numerical analysis by the finite
difference method [5]. Thus, in particular, directional radiation patterns of the studied MDS
are shown in figure 2.
Figure 2. Typical directional radiation patterns of spatial waves on the MDS: (a) Cherenkov and abnormal-diffraction
modes; (b) diffraction-Cherenkov mode. 1 ― experiment, 2 ― numerical simulation.
MDS parameters implemented in the experiment for three main operating modes of free-
space wave excitation at the central frequency are given in the Table 1.
No Operating mode of exciting spatial waves l (mm) d (mm) k u β
Cherenkov
1 1,17 0,39 0,30 +0,5 0,788
n = 0 ― in dielectric
diffraction-Cherenkov
2 n = 0, -1, -2 ― in dielectric 3,07 1,535 0,79 0 0,788
n = -1 ― in free space
abnormal diffraction
3 1,36 1,084 0,35 -0,8 0,598
n = -1 ― in dielectric
Table 1. Parameters of the investigated MDS
The experimentally obtained data has been compared to the results of the numerical experi‐
ment based on solving Maxwell equations in a form of partial derivative using finite-differ‐
ence approach and taking into account constitutive relations.
It follows from the given plots that the experiment satisfactorily correlates with the numerical
analysis, and this, in turn, confirms serviceability of the described setup for studying electrody‐
namic characteristics of periodic structures belonging to a new class, i.e., planar MDS, which
can find application in producing devices of MSM and terahertz wavelength ranges.
3. Practical devices of microwave technology and electronics
3.1. A quasi-optical directional coupler
The general principle of designing DC is to use two energy transmission lines coupled to
each other [11, 12], along one of which the main power flow is transmitted; the auxiliary line
is intended for interference and separation of forward and backward waves.
To date, depending on the imposed requirements, a great number of DC modifications are

used in MSM wavelength measuring circuits.
The systems based on DW or dielectric planar waveguides [13] with distributed coupling, local
coupling, and reemission into the secondary channel are the most close to the proposed DC.
In this work, the design of the DC based on two diffraction-coupled transmission lines with
distributed radiation sources is studied, the sources being formed by periodic structures and
DW placed along them. Figure 3 shows a general DC drawing and two possible reflector
configurations, namely, plane-parallel and plane-cylindrical.
The basic DC section is formed from periodic structure 1, along the longitudinal axis of
which DW 2 is placed at distance a. The second section is made similarly; it also consists of
periodic structure 3 and DW 4. The periodic structures are applied on surfaces of flat and
cylindrical mirrors with aperture D, the axes of which are placed in parallel at distance H
and form a quasi-optical transmission line producing a beam of spatial waves. One of the
radiators of the system is installed so that it can move smoothly along axis z, changing dis‐
tance H and correcting the phase characteristics of waves.
The principle of operation of the diffraction-coupled DC is based on exciting in-phase and
antiphase spatial waves. As a result of the propagation of these waves along the longitudi‐
nal DC axis, the radiators exchange energies and some power is directed into the secondary
channel. Results of theoretical and experimental studies of transformations of DW waves in‐
to spatial waves and of spatial waves into surface waves on periodic structures are descri‐
bed in detail in [1].
Figure 3. Schematic drawing of the design of the DC based on diffraction-coupled transmission lines: (a) cross section
of components along the longitudinal axis; (b, c) cross section of components in the A–A plane at the (a) plane-paral‐
lel and (b) plane-cylindrical configurations of reflectors. 1, 3 ― periodic structures, 2, 4 ― DW, I–IV ― arms of the DC,
P1― power at the DC input, P3 ― power of the backward derived wave, P2 , P4 ― transmitted powers.
Let us dwell on the special features of the wave processes in the proposed quasi-optical DC
(figure 3). When the microwave signal is applied to input I, the delayed wave propagates in
DW 2 and is scattered at periodic structure 1. In this case, a diffraction field arises, which is a
superposition of plane waves. Some waves go into the DC volume at angles α as spatial
waves, and the remaining ones are localized near the grating as slow harmonics propagat‐
ing to the output of waveguide 2. The radiation angle of the spatial waves is determined by
the relation [1]:
α = arccos(1 / βw + n / k ) (1)
where βw = νw/c is the relative wave velocity in the DW; νw is the phase velocity of the wave;
n = -1, -2,… is the number of the spatial radiation harmonic; and k = l/λ is the wave number
(lis the period of the grating, and λ is the radiation wavelength). It is possible to obtain prev‐
alence of one or another type of waves by selecting the radiator characteristics, namely, the
period of the structure, the wave velocity in the waveguide, and impact parameter α.
The second excitation stage of the system in figure 3 is the incidence of the spatial wave
formed at angle α on structure 3. As a result of diffraction, the complete field over the peri‐
odic structure consists of the incident and spatial harmonics of the scattered field. If l<λ/
(1+sinα), the zero spatial harmonic is only reflected from structure 3 at angle α0=90°-α, while
the remaining field is the superposition of surface spatial harmonics propagating along the
grating with the phase velocities
νf = kc / (n + ksinα ) (2)
When νf ≈ νw, the delayed wave is also excited in DW 4 and is transformed into a spatial
wave on periodic structure 3. When surface waves are successively transformed into spatial
ones and spatial waves are transformed into surface ones, in-phase and antiphase waves are
formed along the axis of reflectors of the DC and, as a result of their interference, part of the
incident and reflected power is branched off to the secondary section.
The geometric sizes of the considered DC are selected from the inequalities [6] that specify
the fulfillment of laws of ray optics in the double-mirror quasi-optical systemml:
d2
Hλ ≪ ( Hd )2; l / λ ≥ 10 (3)
where H and L are the distance between the mirrors and their length, respectively, and
d=D/2 is the aperture radius of the radiator mirrors (distance from the longitudinal axis of
the system to the mirror periphery).
The fulfillment of the first inequality allows one to represent the field of the studied system
as paraxial wave beams (figure 3), which are in many ways similar to a plane wave [14]. The
second inequality minimizes the resonance phenomenon display along the longitudinal axis
y. The structure period l is selected from (1) for specified values of λ, n, βw, and α. The values
of impact parameter α < λ are corrected experimentally by studying spatial characteristics of
radiating systems, based on the criterion of the minimum parasitic DC effect on the field of
the periodic structure.
When cylindrical mirrors with quadratic correction are used, the optimal values of their cur‐
vature radii, R, are in interval 2/R = 4÷6. This is attributed to the fact that the focusing action
of the mirrors decreases at 2R/λ > 6 and, for 2R/λ < 4, the mirror aperture intercepts only
part of the radiation of the periodic system–DW system.
We consider an example of selecting parameters of the DC and its components in frequency
band f = 30÷37 GHz. As a source of the surface wave, a polystyrene waveguide with a cross
section of 7.2×3.4 mm2 was used. It provided for relative velocity βw ≈ 0.9 at the wavelength
λ = 9 mm. The main lobe of the radiation pattern (n = -1) of the spatial wave was formed at
angle α ≈ 70°, and, from relation (1), it corresponded to l = 11.5mm. The waveguide length L
= 150 mm and radii of radiating apertures d = 30 mm ensured the fulfillment of conditions
(3), and value R = 25 mm ensured the optimal focusing of the quasi-optical waveguide beam
along the DC axis, when one of the mirrors was cylindrical (figure 3 (c)).
The experimental studies of the DC prototype were performed on the described above setup
by measuring directional diagrams of the radiating systems and their near-zone fields and
also by measuring waveguide characteristics of both separate DC components and the sys‐
tem as a whole. The mechanical part of the setup allowed one to move DC radiators in three
planes with an accuracy of ±0.1 mm and, by varying over wide ranges values a and H, to
adjust the coupler according to optimal output parameter values.
As an example, the characteristics of the above design of the DC for two distances between
the radiating mirrors are shown in figure 4.
It follows from these characteristics that, for the specified frequency interval, at distances be‐
tween radiators H = 65 mm, the attenuation constant (P4/P2) values are in a 3- to 12-dB inter‐
val and the directivity factors (P3/P4) are in a 15- to 25-dB interval. It is possible to improve
the DC output characteristics by correcting the phases of propagating waves, while H is var‐
ied; this is demonstrated by curves 2 in figure 4 (P4/P2 ≈ 3–7 dB, and P3/P4 ≈ 30 dB).
Figure 4. Characteristics of the coupler at (1) H = 65 and (2) 80 mm.
Figure 5. Relative power levels in the secondary channel as function of the distance between the mirrors. P4max is the
maximal power arriving at the secondary channel.
It is determined that the quasi-optical wave behavior of the studied system remains un‐
changed for H = (5–10) λ, for which inequality (3) is true.
Figure 5, in particular, illustrates the relative dependences of power levels P4/P4max = f(H) and
P3/P4max = f(H), which are detected in arms IV and III, respectively (figure 3). It can be seen
from the plots that the optimal (from the viewpoint of decoupling the branched and reflect‐
ed signals) values are H = 60–90 mm. At H > 10λ, the diffraction loss increases and the sys‐
tem becomes critical to the mirror alignment.
By comparing the characteristics of the described DC with couplers on coupled DW [15], the
following conclusions can be drawn. The DC characteristics in the working frequency band
∆f = 5 GHz are comparable with those of the coupler on the effect of the directional signal
reradiation from the section of a sharp DW bend [15], the DC design being significantly sim‐
pler. In addition, this DC has two variants for correcting the transient attenuation (by chang‐
ing impact parameter a and distance between radiators H), thus decreasing process
requirements for manufacturing this system in the MSM wavelength range.
3.2. Quasioptical power dividers
In this subsection the results of studying the power divider (PD) design based on a two-row
semi-transparent periodic structure with a distributed radiation source (main section) in the
form of the DW, which is placed along its longitudinal axis are presented. In this system, the
surface wave of the DW is transformed on periodic grids into spatial harmonics of body
waves [16]. Figure 6 shows two PD modifications with longitudinal (Δd) and angular (ϕ)
shifts of grid bars.
Figure 6. Embodiment scheme of the PD on two-row periodic structures: (a) with longitudinal Δd; (b) with angular φ
shifts of grid bars. 1 ― main section (DW), 2 ― matching junctions, 3, 4 ― grids, 5 ― radiation receivers.
The PD contains DW 1, which is embedded into the main microwave section through
matching junctions 2, grids made of bars 3 and 4, and radiation receivers 5. Grid 3 is fixed in
position with respect to the y axis, and grid 4 is placed on the positioner that ensures either
its linear movement along the y axis with accuracy Δd = ±0.01 mm or angular movement in
the plane of the x, y axes with accuracy Δϕ = ±1°. The radiation receivers operating in MSM
wavelength ranges can be made, e.g., as horn or lens antennas and are intended for detect‐
ing the power diverted from the main section.
The principle of operation of these PD is based on transforming the DW surface wave into
the spatial (body) wave, which is excited on the grids. The radiation power level of this
wave can be regulated by changing phase relations of slow harmonics of waves excited on
the grids of the two-row structure when its bars are shifted or when the rotation angle with
respect to the DW axis is changed.
We consider general features of wave processes in the described quasi-optical PD (see fig‐
ure6). When the microwave signal with power Р1 is applied to the input of DW 1, the de‐
layed wave propagates over its surface and scatters on grids 3 and 4. In this case, the
diffraction field is formed, which is the superposition of plane waves. The part of these
waves goes away into the free space (powers Р 3 and Р4 extracted from the main power sec‐
tion) at angles γn in the form of spatial waves, and the remaining waves are localized near
the grids in the form of slow harmonics propagating toward DW output 1 (power Р2 . The
angle and conditions of emission of the spatial waves in the DW-grid system are determined
from the relation [10]:
1
| cosγn | = k α − l( 2πn
)≤1 (4)
where γn is the emission angle of the harmonic with index n < 0, 1 is the period of the grid, k
= 2π/λ, and α = k – (2πn)/l is the propagation constant.
From equation (4), it follows that the waves with n ≥ 0 and |cosγn| > 1 are a spectrum of
heterogeneous plane waves existing near the grid and propagating along the longitudinal
axis of PD system with phase velocities νph< c, where the νph is the phase velocity of the
wave, and с is the speed of light. For all remaining waves with indexes n < 0, the condition
of their emission into the ambient space is met, which characterizes them as spatial (body)
waves. By selecting appropriate parameters of the electrodynamic system and the surface
wave propagating along the DW, it is possible to attain the predominance of one or another
type of waves.
The geometrical sizes of the considered PD are selected from equation (4) when the funda‐
mental (n = –1) spatial wave is emitted at angle γ–1 = 90° to the plane of grids 3 and 4 and the
latter are mirrored with respect to the DW axis (Δd = 0 and Δϕ = 0°). In this case, it is neces‐
sary to take into account the influence of bar thickness 2h on the radiation intensity [17]. The
maximum bar thickness is determined as:
λ 4 πθ
2h = N 2 + d π lnsin 2 , ( N = 1,2,...) (5)
where d is the bar width, and θ is the relative distance between the grid bars.
In the grid with the optimal profile, somewhat smaller than the integer number of half-
waves should be fitted in bar thickness 2h. The correction for half-wave thickness
4
(
Δh = d π ln sin
πθ
2
) is equal for all harmonics and determined only by the geometrical param‐
eters of the grid regardless of the wavelength and angle of emission. This correction is at
maximum (Δhmax = –0.23l), when the slot width θ ≅ 0.3.
In addition, the intensity of the spatial waves in the considered systems can be regulated by
the duty factor of the semi-transparent grids made of bars u = cos(πd/l) and sighting param‐
eter а, which is the distance from the DW to the surface of the grids and determines the de‐
gree of relationship between the surface waves and heterogeneous waves of the grids.
The interval of the optimal value region of parameter Δd for the first PD modification (see
figure 6(a)) was determined experimentally due to the absence of the theory of these sys‐
tems. For the second PD modification (see figure 6(b)), the ranges of angles ϕ within which
grid 4 is shifted with respect to grid 3 can be estimated from the limiting (critical) values of
the angle of rotation of the single grid [10]:
ϕcr = arccos
1
2
( k βw
n −
nβw
k −
k
nβw
) (6)
Hence, the surface wave mode occurs in the open structure of the DW and grid 4 at ϕ = ϕcr,
and the further increase in ϕ results in the Bragg diffraction, for which
N B βw
ϕBr = arccos , N B = 1,2,3,... is the order of diffraction. In this mode, the heterogeneous
2k
surface wave of the DW is transformed into the surface wave of the grid. The formed new
surface wave propagates at an angle of (180°–ϕ) with the direction of the original heteroge‐
neous wave. The degree of transformation of one wave into the other is determined by the
longitudinal dimensions of the grid and the coupling between the DW and grid field.
As an example, let us select the parameters of the PD and their main elements in the four-milli‐
meter wavelength range (f = 60–80 GHz). As a source of the surface wave, a fluoro-plastic
waveguide with a 5.2×2.6 mm2 cross section is used, which ensures the relative wave velocity
βw = 0.788 at wavelength λ = 4 mm. The main lobes of the radiation pattern (n=–1) of the spatial
waves were formed at angle γ = 90°. According to equations (4) and (5), this corresponded to
values l = 3 mm, u = 0, and 2h = 1 mm. The grid aperture D = 60 mm and their length along the
DW axis L = 54 mm ensured the meeting of the main quasioptic conditions [15].
The experimental studies of the PD prototypes (figures 6 (a), (b)) were performed in accord‐
ance with the described above procedure.
As an example, the amplitude (relative values Р3 / Р3max ) and waveguide (transmission coef‐
ficients Ktr = Р2 / Р1 ) characteristics of the above PD are shown in figure 7. They indicate that
it is possible to extract 80% of the power from the main section (Р2 = 0.2Р1) and divide it
equally when the grids are placed symmetrically with respect to the DW (Р3 = Р4 = 0.4Р1) for
the specified sighting parameter а = 0.7 mm.
The analysis of the plots in figure 7 shows that the power in the arm Р3 can be smoothly
regulated either by shifting the bars or changing the rotation angle of grid 4 with respect to
axis of grid 3. In this case, for the PD modification in figure 6 (a), it is expedient to select the
Δd/λ values in an interval of 0.2–0.4, in which the dependence of the radiated power is close
to the linear one and has no resonance effects.
Figure 7. Amplitude and waveguide characteristics of the PD with (a) longitudinal and (b) angular shifts of grid bars
with respect to the axis of the waveguide. 1 ― relative power values P3, 2 ― K tr values, 3 ― calculated curve.
From the viewpoint of regulating and calibrating the emitted power Р3, the second PD mod‐
ification is preferable (figure 6 (b)). Its amplitude characteristic is virtually linear for the se‐
lected angle interval ϕ = 0–35°. For both PD modifications, the standing-wave ratio values
were 1.1–1.3 in frequency range f = 60–80 GHz. The characteristics shown in figure 7 were
numerically simulated by the finite difference method [5], and the calculated results are il‐
lustrated in figure 7 by dotted lines 3. The comparison of the experimental results and nu‐
merical analysis indicates their correlation.
By analyzing the characteristics of the described PD, it is possible to make the following con‐
clusions. In contrast to the waveguide analogs and PD on DW [15], the described PD possess
wider functional capabilities, allowing one to divide the power coming from the main sec‐
tion into equal parts between two channels and smoothly regulate it in one of the output
arms of the divider. In this case, the powers arriving at the output arms of the divider can be
regulated by changing the sighting distance of the DW with respect to the grid surfaces. The
presented PD can be also manufactured in the planar form if periodic MDS are used [11].
3.3. Possible variants of implementation of MSM radiation sources
Nowadays MSM microwave devices (backward-wave tubes, taveling-wave tubes, klystrons)

with high level of output power are widely used in transmitting equipment of communication
systems, radars and radio countermeasures devices. Thus, there is the rapid development in
production of traveling-wave tubes with slow-wave structures as a chain of coupled resona‐
tors in different geometrical modifications [8], creation amplifiers based on the multielectron-
beam tubes [18] etc. Axially symmetric EB as the main working element of such devices
virtually defines their basic working parameters. Therefore, special attention is given to the
improvement and optimization of electron-optical systems of MSM microwave devices.
As a rule, an optimization of electron-optical system parameters bases on the information
about EB characteristics. Currently, because of the rapid development in computational re‐
sources and approaches, different techniques of electromagnetic simulation based on vari‐

ous numerical algorithms are mostly used to get this information. Such approach allows
avoiding carrying out time-consuming and expensive experiment. Usually modelling tech‐
niques consist of two stages: computation of electromagnetic fields of the investigated sys‐
tem and subsequent trajectory analysis of charged particles in these fields.
In the current subsection the example of the model of the three-electrode electron gun which
is used in real microwave devices such as travelling-wave tube has been presented, the opti‐
mization problems of its operating modes and axially symmetric EB characteristics have
been outlined.
Axially-symmetric EB, as a rule, is formed by three-electrode electron gun with converging

optics and introduced into slowing-down system where it is focused by periodic magnetic
field. The cathode is usually produced in the form of a core made of tungsten-rhenium
blend with activated surface. The typical electrode configuration of axially symmetric elec‐
tron-optical system is demonstrated in figure 8.
Figure 8. Scheme of electron-optical system of axially symmetric electron gun: 1 ― hot cathode (U k ), 2 ― focusing
electrode (Uf), 3 ― first anode (Ua1), 4 ― second anode (Ua2).
The guns of such type allow forming EB with diameter of about 0.3mm in the crossover, the
beam current of 1–25 mA with accelerating voltage 2000–6000 V. For numerical simulations
the finite integration technique (in literature is known as FIT [19]) was chosen as the optimal
numerical algorithm for analyzing the above mentioned systems.
The combination of electrode potentials, described in [20], was taken as the initial parame‐
ters with a beam perveance P = 0.045 mkA/V3/2, the number of emitted particles N = 2965,
average kinetic energy of electrons Ek = 5 eV with energy spread dEk = ±0.33 Ek, maximal an‐
gular deflection α = ±10° from the axis of the system, that corresponds to the real guns of
such type. Meanwhile, the emitting area has been defined as a perfectly plane surface with
the uniform current density distribution.
Figure 9. Coefficient of the EB passage with the different potential values Ua1 and Ua2.
When simulation of the electron-optical system is carried out, the main parameter of opti‐
mality is the coefficient of EB passage. That is why in the first step of the optimization of the
electron gun operating modes the research of a beam passage coefficient K = Icat/I col was con‐
ducted, where Icat and Icol are cathode and collector currents respectively. The simulation has
shown that too high potential of focusing electrode (Uf ≈ -40V) impairs the passage of EB
and causes unsatisfactory crossover position between two anodes of electron-optical system
but not after the second anode. In addition, as the focusing electrode is placed in the imme‐
diate vicinity to the cathode surface the partial EB blocking effect is created and the kinetic
energy of the emitted electrons is not high enough to overcome thus formed potential barri‐
er. This blocking effect decreases considerably an EB passage ratio and disturbs beam lami‐
narity. Changing the potential of the focusing electrode might partly solve mentioned
problems. When Uf is decreased to zero potential the beam crossover is shifted to the region
of the second anode transit channel. In this case the diameter of the crossover decreases and
the passage of EB with Uf = -10÷0V might reach 93-95%. Thus, to avoid any impact of the
focusing electrode on the cathode edge Uf = 0V was taken as the most appropriate for the
experimental research, filament current of the tungsten- rhenium (W-Re) cathode for the
whole set of measurements was constant ― 0.7 A.
Experimental measuring results have shown that the coefficient of the EB passage increases
with increasing Ua2 nd riches the value of K = 0.934 with Ua2 = 3000V, that corresponds to the
current value I = 9.1 mA. However, further increasing of the Ua1 and Ua2 potentials stimulates
electron emission and causes rapid growth of the cathode current (Icat > 15 mА). This, in turn,
leads to the strong heating of the gun elements, bombarded by electrons, which considerably
disturb the consistency of the device operation and causes different measurement errors.
Figure 9 shows that the coefficient K reaches its maximum value at the accelerating poten‐
tials of Ua1 = 80 V and Ua2 = 3000 V. Thus, such electron-optical system operating mode can
be considered as the optimal for the passage of the EB.
Besides the information about the electron deposition on the gun electrodes, which allows
carrying out a preliminary analysis of its operating modes, the EB quality parameters (such
as type of a particle distribution in the cross-section, laminarity, spread of the velocity trans‐
verse components etc.) are also very important. In the current work root-mean-square emit‐
tance (statistical emittance) was used to describe such characteristics:
ε =( x 2 ⋅ x ′ − x ⋅ x ′ ) =( x − x ⋅ (x ′ − x ′ )2 − ( x − x ) ⋅ ( x ′ − x ′ ) 2)1/2
2 2 1/2 2 (7)
where
νx N
1
x ′= ν⊥ , x = N ∑ xi (8)
n=1
It should be noted, that formula of this type in contrast to other definitions makes it possible
to express an EB emittance in a simple numerical form. In addition, we do not find it reason‐
able to consider a normalized beam emittance in this case because of short drift duct and
low accelerating potentials of the system (EB is not relativistic) [21].
The calculated emittance with the maximal passage of the EB at a distance of 10 mm from
the cathode made up εx = 1.71 mrad·mm and εy = 1.85 mrad mm in x- and y-directions re‐
spectively, the beam diameter ― D = 0.8 mm.
However, as the numerical experiments to study transversal electron dynamics (figure10

(a)) has shown, value of the emittance can be decreased by variations of the potential Ua1.
The analysis of two-dimensional diagrams of the transversal emittance has demonstrated that
the increase of the Ua1 to 110 V considerably disturbs the beam laminarity (figure10(b)) while
its decrease improves the EB quality (type of the diagram corresponds to the perfectly diver‐
gent beam) with slight growth of its diameter (D ≈ 0.8 mm). On the basis of the analysis of ex‐
perimental and numerical simulation results the optimal operating mode with the potentials
Ua1 = 60 V и Ua2 = 3000 V has been chosen. It allows getting the passage of the EB at the level of К
≈ 0.85 with transversal emittance values εx = 1.67 mrad·mm and εy=1.76 mrad·mm.
Development of high-resolution sensitive elements for terahertz frequency bandwidth is an

actual problem due to a number of existing international projects of radio astronomy as well
as the projects with the objective of studying the Earth atmosphere. The basic issues of ar‐
rangement of receivers in the given bandwidth are solved by application of solid-state heter‐
odyne oscillation sources and the mixers based on the effect of electron heating in the
superconductor (Hot Electron Bolometer) because such mixers have no competition analogs
in this bandwidth. By now these mixers have been successfully realized at the frequencies of
the order of 3 GHz. However, several research projects are related to development of the
mixers optimized for higher frequencies; e.g. within the framework of the SOFIA project it is
developed a heterodyne receiver for 4.8 ТHz. Principal opportunity for creation of the mixer
for the given bandwidth is shown in the paper [22] where the method of electron and photo
lithography is applied for its realization. As an example, figure 11 provides photos of the
central part of the mixer obtained with the scanning electron microscope.
Figure 10. Dependence of EB emittance on the Ua1 potential: (a) numeral values in x- and y-directions; (b), (c) two-
dimensional diagrams for Ua1 = 110 V and Ua1 = 40 V, respectively.
The mixers were made of NbN films with the thickness of 2 to 3.5 nm upon a silicon sub‐
strate with a MgO buffer sublayer and possessed the temperature of superconducting transi‐
tion of 9 to 11 К. The best value of the noise temperature of the receiver on the basis of the
electron heated mixer amounted to 1300 К and 3100 К at the heterodyne frequencies of 2.5
and 3.8 ТHz correspondingly.
In the radio astronomy devices and tools the preference is given to solid state heterodyne
sources within the terahertz band due to their small dimensions, low weight and power con‐
sumption requirements, despite low output power level, which is not exceeding 1 μW at the
frequency of 2 ТHz that to a significant extent complicates the problem of development of
mixers for a low level of output power. Thus, there still remains actual the problem of reali‐
zation of low-voltage electrovacuum oscillation sources within the terahertz wavelength
band possessing higher values of power levels as compared to those of solid state oscillators.
By the present time this problem can be solved by means of using in the vacuum electronics
of planar periodic MDS [1, 11]. Block diagrams of such devices that could be realized on the
basis of the above described technologies and new types of dielectrics with larger values of
dielectric permeability and low loss values at high frequencies [23], are provided in figure12.
Block diagram of the oscillator in figure 12(а) suggests modulation of the EB upon a back‐
ward wave of the periodic structure (position 2) with subsequent excitation of the Cheren‐
kov oscillation harmonic in MDS (position 3). Within the oscillator in figure 12(b) it is
applied the mode of abnormal diffraction oscillation, which is realized at substantially less
values of accelerating voltages of the electron flux compared to those of the Cherenkov oscil‐
lation harmonic.
(a) (b)
Figure 11. Image of the helix (a) and of the central part of the helix with the NbN bridge (b) obtained with the scan‐
ning electron microscope [22].
Figure 12. Block diagrams of terahertz frequency band electrovacuum oscillation sources.1 ― EB, 2 ― periodic struc‐
ture, 3 ― MDS with diffraction strip-like grating, 4 ― screen (mirror), 5 ― energy output device.
At practical realization of the above block diagrams of sources of oscillation it is necessary to

solve a set of problems related to the technology of manufacturing of main units of the de‐
vice – the electrodynamic system, the low-voltage source of electrons and the focusing mag‐
netic system. The above considered technological processes – like, for instance, nano die
forming – eliminate all problems related to manufacturing of both reflecting and ribbon (ap‐
plied upon the dielectric layer) DG with the micron period.
Oscillator schemes presented in figure 10 are the ideological continuation of the so far devel‐
oped low-voltage backward-wave tube with multi-row slow-wave structures [8, 24]. A sche‐
matic view of a low-voltage orotron based on parallel MDS and three EB are shown in figure 10
(a). The calculated MDS parameters have been determined for EB radiation at the angle of 90°.
The diagram of a backward-wave tube based on an anomalous diffraction radiation for MDS is
shown in figure13 (b). The arrows show the direction of wave radiation on the MDS.
Figure 13. Schematic view of low voltage vacuum devices employing multi-linked MDS:(a) orotron, (b) backward
wave oscillator. 1 ― cathode, 2 ― anode, 3 ― EB, 4 ― planar periodic MDS, 5 ― collector, 6 ― mirror, 7 ― output
port.
Figure 14. Microphotographic images of the field cathode surface [27] (a) and block diagram of the “slot” L-cathode
device [28] (b).
The analysis of parameters of the discussed electrodynamic systems [11] and devices dem‐
onstrated in figure 13 shows that the DG/MDS or multi-row systems on basis of MDS at ac‐
celeration voltages U ≈ 1000 V and εr = 100 (ceramics based on titanium oxide) can be
realized for a wave length range from 1mm to 0.1 mm and periods lfrom 64 μm to 4 μm.
Currently the fabrication technology for such structures is mature and has been used for fab‐
rication of multiple parallel slow-wave structures in backward-wave tube. The fabrication
technologies include electro-erosion machining, cold forming, photolithography, electron
and x-ray lithography, vacuum and plasma deposition. The fabrication of slow-wave struc‐
tures of vacuum electron tubes requiring nanoscale precision is performed by ion beam lith‐
ography [25] in combination with nanoforming [26].
The research conducted in [1] indicates that in order to realize the described devices in sub‐
millimeter and infrared wave ranges, an EB should be as thin as 0.04 mm. Nowadays the
practical realization of such EB is possible using array type or slot type L-cathode, which al‐
low getting uniformly distributed and stable electron emission with high current density at
comparatively low field intensity.
The matrices of field emission cathodes possess the preset geometrical dimensions (diameter,
step), they allow elimination of the screening effect and obtaining of homogeneous upon the
surface and stable in time electron emission with the average current value of 40 μА from a sep‐
arate cathode at relatively low electric field intensity values. The cathode represented in figure
14 (b) includes the cylinder container filled with a stock of the substance that decreases the out‐
put operation of the operating surface, which is represented by a continuous row of micro-ele‐
ments forming up a “slot” L-cathode. Experimental results obtained while pilot testing of
those cathodes at IRE NAS of Ukraine proved the opportunity for obtaining of such electron
fluxes with high-value current density at not high values of accelerating voltages.
4. Conclusions
In the current work the results of development of the experimental setup and procedure for
measuring electrodynamic characteristics of planar periodic MDS, which can be used for
manufacturing practical equipment operating in MSM and terahertz wavelength ranges are
presented. Serviceability of the setup is checked by comparing the waveguide and spatial
characteristics obtained experimentally and by numerical methods in a 4-mm wavelength
range. A circuit and principle of operation of DC based on diffraction-coupled transmission
lines, the emitting apertures of which are formed from periodic structures and DW has been
proposed on the basis of the conducted research.
The experimental studies of the coupler prototype in a 30- to 37-GHz band have shown that the
coupling on spatial waves allows one to obtain transient attenuation values in a 3- to 20-dB in‐
terval at a ~ 30 dB directivity, this being roughly in conformity with similar characteristics of a
coupler on DW. The main advantage of the described DC is that it is possible to correct its char‐
acteristics over wide ranges by changing the distance between the emitting apertures.
Circuits and principle of operation of quasi-optical PD based on two-row periodic struc‐

tures, formed by grids of metal bars and a DW placed along their longitudinal axis, have
been proposed and described as the second example of realization of described above char‐
acteristics. The experimental studies of PD prototypes in a frequency range of 60–80 GHz
have shown a possibility of regulating the emitted power level in the main divider arm by
changing longitudinal and angular coordinates of the two-row periodic structure, which can
be used for designing quasi-optical attenuators.
The example of the model of the three-electrode electron gun of travelling-wave tube has
been described, the optimization problems of its operating modes and axially symmetric EB
characteristics have been outlined.
In addition, studied properties of electrodynamic diffraction characteristics of the surface

waves on periodic heterogeneities can be realized in implementation of radiation sources on
the Smith-Parcell effect.
Work is supported by the governmental programme No0112U001379.

Author details
Gennadiy Vorobyov1*, Larissa Vietzorreck2, Ivan Barsuk1 and Aleksandr Rybalko1
1 Sumy State University, Sumy, Ukraine
2 Technische Universität München, München, Germany
References
[1] Shestopalov, V.P. (1991). Diffraction radiation oscillators, Kyiv, Naukova Dumka.
[2] Vorobjov, G., Shulga, Yu., & Zhurbenko, V. (2011). Quasi-optical Systems Based on
Periodic Structures. Zhurbenko V. (ed.) Electromagnetic Waves, Rijeka, InTech, 257-282.
[3] Vorobjov, G. S., Petrovskii, M. V., & Krivets, A. S. (2006). Possible applications of
quasioptical open resonant metal-dielectric structures in EHF electronics. Radioelec‐
tronics and Communications Systems, 49(7), 38-42.
[4] Vorobjov, G. S., Tsvyk, A. I., Krivets, A. S., Shmatko, А.А., & Petrovsky, M. V. (2003).
The Smith-Pursell Effect Amplification of the Electromagnetic Waves in a Open
Waveguide with a Matal-Dielectric Layer. Telecommunications and Radio Engineering.
[5] Bankov, S.Е. (2004). Analysis and optimization of microwave three-dimensional structures
with HFSS., Moscow, Solon-Press.
[6] Sirenko, Yu. K., Strom, S., & Yashina, N. P. (2007). Modeling and Analysis of Transi‐
ent Processes in Open Resonant Structures: New Methods and Techniques. New
York, Springer.
[7] Bratman, V. L., Glyavin, M., Yu, , Kalynov, Yu. K., Litvak, A. G., Luchinin, A. G., Sa‐
vilov, A. V., & Zapevalov, V. E. (2010). Terahertz Gyrotrons at IAP RAS: Status and
New Designs. Journal of Infrared, Millimeter and Terahertz Waves [8], 8-934.
[8] Yakovenko, V.М. (2007). Vacuum devices millimeter waves, Sevastopol, Weber.
[9] Federici, J., & Moeller, L. (2010). Review of terahertz and subterahertz wireless com‐
munications. Journal of Applied Physics [107], 107-1063.
[10] Shestopalov, V.P. (1985). Physical basis for millimeter- and submillimeter-wave equipmen.
1Open-type structures), Kyiv, Naukova Dumka.
[11] Vorobyov, G. S., Petrovsky, M. V., Zhurba, V. O., Ruban, A. I., Belous, O. I., & Fisun,
A. I. (2007). Perspectives of application of new modifications of resonant quasi-opti‐
cal structures in EHF equipment and electronics. Telecommunications and Radio Engi‐
neering, 66(20), 1839-1862.
[12] Xiao-Ping, Chen., & Ke, Wu. (2008). Substrate Integrated Waveguide Cross-Coupled
Filter With Negative Coupling Structure. Microwave Theory and Techniques, IEEE
Transactions on Microwave Theory and Techniques, 56(1), 142-149.
[13] Demydchik, V.I. Microwave electrodynamics Minsk Universitetskoe (1992). .
[14] Weinstein, L.A. (1966). Open resonators and open waveguides, Moscow, Sov. Radio.
[15] Valitov, R.A. (1969). Submillimeter-wave equipment, Moscow, Sov. Radio.
[16] Quan, X., Kang, X., Yinghua, R., & Jun, T. (2011). 3dB power splitter design based on
coupled cavity waveguides. International Journal for Light and Electron Optic, 122(2),
156-158.
[17] Shestopalov, V.P. (1976). Diffraction electronics, Kharkiv, Kharkiv UniversityPublish‐
ers.
[18] Sinicin, N. I., Zakharchenko, Y. F., & Gulyaev, Y. V. (2009). A new class of high-pow‐
er low voltage multipath TWT elated to chains multiple-Resonators with a trans‐
versely-extendedtype of interaction for airborne radar and communication systems
the short-millimeter waves. Jornal of radioelectronics, 50(10), 46-51.
[19] Weiland, T.A. (1977). Discretization Method for the Solution of Maxwell’s Equations
for Six-Component Fields.Electron. Commun. (AEU), 31(3), 116-120.
[20] Belousov, Ye. V., Vorobyov, G. S., Korg, V. G., Pushkarev, K. A., & Chaban, V. Y.
(1992). Experimental investigation of the static parameters axisymmetrical electron
beams of small diameter. Recent developments in applied physics, 2(1), 87-100.
[21] Braun, Ian. G. (1998). The physics and technology of ion sources, New York, A Wiley and
Interscience Publication.
[22] Finkel, М. I., Маslennikov, S. N., & Goltsman, G. N. (2005). Superheterodyne receiv‐
ers with superconducting THz mixer on electronic heating. Trans. Higher Education.
Radio Physics.
[23] Nanasheva, Ye.А., Тrubitsina, О. N., Каrtenko, N. F., & Usov, О.А. (1999). Ceramic
materials for microwave electronics. Solid State Physics, 41(5), 882-884.
[24] (2002). Vacuum Microwave Electronics: Collection of reviews, Nizhny Novgorod, Insti‐
tute of Applied Physics.
[25] Watt, F., Bettiol, A. A., van Kan, J. A., Teo, E. J., & Breese, M. B. (2005). Ion beam lith‐
ography and nanofabrication. International Journal of Nanoscience, 4(3), 269-286.
[26] Ansari, K., van Kan, J. A., Bettiol, A. A., & Watt, F. (2006). Stamps for nanoimprint
lithography fabricated by proton beam writing and nickel electroplating. J. Micro‐
mech. Microeng., 16-1967.
[27] Solovey, D. V., Sakharuk, V.N., & Novitsky, А.М. (2009). 19th International Crimean
Conference “Microwave Equipment and Telecommunication Technologies CriMiCo 2009
September, 14-18, Sevastopol National University, Sevastopol, Ukraine.
[28] Belousov, Ye. V., Zavertannyi, V. V., & Nesterenko, А. V. (2006). Diode electron gun
with a slot L-cathode. Radio Physics and Electronics, 11(2), 275-280.

Section 4
Scattering, Radiation and Propagation

Chapter 9
Radiative Transfer: Application to Indoor Propagation
Ada Vittoria Bosisio
1. Introduction
In recent years, with the increasing interest in indoor wireless communications systems, the
development of appropriate tools for modeling the propagation within an indoor environ‐
ment is becoming of utmost importance. A versatile technique for studying propagation in
such a complex scenario is ray tracing [1-2]. Through this approach, a number of paths,
stemming from the transmitter, are traced along their way to the receiver, accounting for re‐
flection over the obstacles within the scenario. Other mechanisms of interaction between the
wave and the environment, such as diffraction, can be accommodated in ray tracing proce‐
dures by appropriate generalization of the basic theory [2].
While this method is purely deterministic, in actual environments with many randomly
placed scatterers of size comparable to the wavelength (Mie scattering), statistical characteri‐
zation of the multipath channel [3-6] may be the only viable approach in order to have an
accurate model of the propagation [7-8]. Statistical modeling built on iteration of ray tracing
results suffers of its inner computational intensity. As for the radio channel design what it is
of interest are the fluctuations about a mean value of the received power, one can use a sim‐
pler and efficient method to take into account variations due to randomly placed obstacles
in the propagation environment. Radiative transfer theory seems to be appropriate, as it
deals with the wave propagation within a random medium characterized by randomly
placed scatterers. Based on a phenomenological description of the transfer of energy, the ba‐
sic equation (referred to as radiative transfer equation, RTE) simply states the conservation
of energy in terms of the specific intensity I (r, ^s ), i. e. the power per unit area and per unit
solid angle propagating along ^s , and which is a function of position r in the random medi‐
um [9]. The development of the theory is heuristic and does not entail any information
about the phase of the wave. Chandrasekhar first studied the RTE within the context of as‐
trophysics [10]. Later, the same formulation was employed successfully for many other ap‐
plications, such as propagation modeling in the atmosphere or in forested environments,
heat transfer through insulating materials, neutron scattering, and power absorption within
biological tissues [11-14].
In this chapter, the author reports the RTE results for the evaluation of the power fluctua‐
tions in an indoor environment described as a homogeneous medium filled with scatterers
arbitrarily placed. The radiative transfer outcome is compared with the ray tracing predic‐
tions to assess its limits of applicability.
For the sake of simplicity and without introducing inessential complications in our analysis,
the investigation is limited to a propagation environment that can be modeled as a layered
parallel plane medium. This modeling is a first approximation of an indoor environment in‐
cluding people, benches or, for instance, a row of chairs in an auditorium, whose location is
not fixed. The random medium layer accounts for the average condition out of many possi‐
ble spatial configurations, given or assumed the number density of scatterers of a succession
of layers where each layer is modeled as random medium containing randomly placed scat‐
terers. Moreover, we consider infinite-length circular cylinder as scatterers so as to simplify
the solution of the RTE. It should be clear that while the ray tracing approach can be in prin‐
ciple used for any geometry and provide information about the phase of the wave, RT is in
practice only applicable to simple geometries and can only yield information about the sec‐
ond order statistics of the wave. The scatterer’s number density is chosen so as interference
and interaction between scatterers could be neglected. RT reliability under this condition
has been thoroughly investigated [15-17]; here, it is shown how it can be a useful and simple
tool for indoor propagation analysis regarding the spatial correlation as defined in Section 5.
The chapter is organized as follows. A review of the radiative transfer is presented in Sec‐
tion 2, together with the definition of main quantities. Section 3 and 4 are devoted to the de‐
scription of the RTE and of the numerical techniques for its solution. Iterative procedure and
its limits of applicability are discussed. Section 5 reports the numerical results in two study
cases in a 2D geometry with focus on the impact of the system parameters on the specific
intensity and spatial correlation. The results of the comparison between the radiative trans‐
fer results and the predictions obtained through ray tracing are also reported. Section 6 is
devoted to comments and conclusions.
2. Radiative Transfer Theory: physical background
Two basic theories have been developed in order to approach the study of wave propaga‐
tion within a random medium characterized by randomly placed scatterers. The first is the
analytical theory, where taking into account the scattering and absorption characteristic of
the particles solves the Maxwell equations. This approach is mathematically and physically
rigorous since in principle the effects of the mechanisms involved in multiple scattering, dif‐
fraction and interference can be appropriately modeled. However, in practice, various ap‐
proximations have to be made in order to obtain feasible solutions (see [9] or [10] for an
overview). Recent developments [18] adopt the random medium as paradigm to describe
the propagation channel. Stochastic Green’s functions are computed to obtain the channel
Radiative Transfer: Application to Indoor Propagation 237
transfer matrix T in MIMO applications. On the other hand, the transport theory is based on
a phenomenological description of the transfer of energy. The basic equation simply states
the conservation of energy expressed in terms of the specific intensity and it is equivalent to
the Boltzmann equation used in kinetic theory of gases [9]. The development of the theory is
heuristic and it does not entail any information about the phase of the wave.
In this chapter, the basic concepts and the quantities of interest when dealing with radiative
transfer theory are reviewed, while the reader is referred to literature for a deeper insight.
As the problem is defined in a two-dimensional domain, the classic theoretical formulation
is adapted to this framework. Accordingly, the chosen reference system is in cylindrical co‐
ordinates, as shown in Figure 1. The analysis deals with monochromatic signals with fre‐
quency f and the phasor notation is used.
Figure 1. Coordinate system.
To study the propagation of a wave in presence of randomly distributed particles, the main
results related to scattering and absorption of a single particle in vacuum are reviewed. Let
us consider an elliptically polarized incident plane wave Ei(ρ) in the direction t i described
by the azimuth angle ϕ i:
^
t i = cosϕi x^ + sinϕi ^y (1)
that is:
jki ρ jki ⋅ρ
Ei = Ei0e ⋅ 0
= Ev,i v + Eh0,i h e (2)
being (E0v,i, E0h,i) the field amplitudes for the vertical (z) and the horizontal (unit vector h ly‐
ing on the xy plane) components, respectively. k is the wave number vector:
ki = k0ti = ω μ0ε0ti (3)
and (μ0, ε0) represent the vacuum dielectric constant and its permeability. As the propaga‐
^
tion takes place in the xy plane, the observed directions are denoted by a unit vector t or by
the corresponding azimuth angle ϕ according to (1). The particle is characterized by a com‐
plex dielectric relative constant:
ε
εr = = ε ' + jεr'' (4)
ε0 r
and it is assumed for simplicity to be homogeneous with:
σ
εr'' = − (5)
ωε0
where σ is the conductivity [Sm-1].
In far field, the scattered field behaves like a cylindrical wave:
1 (
j kρ−
π
)
Es = F (ϕi , ϕ )Ei0e 4
(6)
r
where F(ϕi,ϕ) is the 2×2 scattering matrix accounting for the amplitude, the phase and the
polarization of the scattered wave in direction ϕ when illuminated by a plane wave propa‐
gating in direction ϕι. For a 2D problem the polarizations are independent and the scattering
matrix is diagonal:
F vv (ϕi , ϕ) 0
F (ϕi , ϕ) = (7)
0 F hh (ϕi , ϕ)
Therefore, for each polarization p (p = v, h) we can write:
1 (
j kρ−
π
)
E p,s = F pp (ϕi , ϕ )E p,i
0
e 4
(8)
r
In the following, where not stated otherwise the subscript p is dropped for simplicity of no‐
tation.
2.1. Scattering and absorption cross widths
The power dPs scattered along direction ϕ within the differential width dl subtended by the
differential angle dϕ,dl = r dϕ is:
d Ps = Ss dl
where Ss [Wm-1] is the amplitude of the Poynting vector of the scattered wave,
Ss = | ES | 2 / 2ηand η = μ0 / ε0 is the vacuum electromagnetic impedance. It follows from (8)
that:
d Ps = | F (ϕi , ϕ) | 2Si dϕ (9)
and integrating over all angles:

2π 2π
∫
Ps = d Ps =
0
∫ | F (ϕ , ϕ) | S dϕ = S σ
0
i
2
i i s
where the scattering cross width σs is defined as:
2π
σs =
Ps
Si
= ∫ | F (ϕ , ϕ) | dϕ
0
i
2 (10)
This quantity represents the equivalent width that would produce the amount Ps of scat‐
tered power if illuminated by a wave with power density Si.
The geometric cross width σg [m] of a particle is its geometric width projected onto a plane
that is perpendicular to the direction of the incident wave ti. The relationship between the
geometric and scattering widths can be investigated in two regimes. If the size of the object
D (maximum distance between two points inside the object) is much smaller than the wave‐
length λ, it follows that:
σs
< < 1 (D < < λ)
σg
According to Rayleigh scattering theory [19], it means meaning that in this regime the pow‐
er scattered by the particle is much smaller than the product of geometric cross width and
the amplitude of the Poynting vector. Besides, in the high frequency regime D>>λ:
σs
→1
σg
which is known as geometric optics limit.

Similarly to (10), the absorption cross width can be defined as the ratio between the absor‐
bed power Pa and the incident Poynting vector Si. From Ohm’s law [20]:
1
2 ∫
Pa = ω dρεr'' | Eint (ρ) | 2
A
(11)
where Eint(ρ) is the internal field within the particle and A is the particle’s area.
2.2. Extinction cross width and albedo
The extinction cross-section σext of a particle is defined as:
σext = σs + σa (12)
and it represents the total power loss from the incident wave due to scattering and absorp‐
tion. The fraction of scattering over the extinction cross width is defined as albedo α:
σs
α= ≤1 (13)
σext
The computations of the extinction cross width (12) can be carried out from the knowledge
of the scattering matrix, as stated by the forward scattering theorem (also known as optical
theorem):
π
2k0 (
σext = − 4 Re F (ϕi , ϕ))
This result can be proved either undertaking the explicit computation of (11) [15] or comput‐
ing the received power over a given width and relating this quantity to the geometric di‐
mensions [19]. Besides, it can be shown that for the high frequency regime [9]:
σa
→1 (14)
σg
σext
→2 (15)
σg
Equation (15) is also known as extinction paradox.
2.3. Example: scattering from a circular cylinder
Figure 2 shows the scattering, the absorption and the extinction cross widths are computed
for a circular cylinder of radius a [21]. In figure 2 these cross width values are normalized
over the geometric cross width σg=2a and they are plotted versus the radius a normalized
over the wavelength λ. The particle is characterized by εr ' = 4 and σ=10−3. One could notice
that as the size of the particle increases the absorption and scattering cross widths tend to σg
width; whereas the extinction cross width tends to 2σg (extinction paradox, (15)). Also, Fig‐
ure 2 distinguishes three different regions: Rayleigh scattering (a<< λ), Mie scattering (a ≈λ)
and optical region (a>> λ).
Figure 2. Scattering, absorption and extinction cross widths normalized over the geometric cross width σg =2a versus
a/λ (εr ' = 4and σ= 10−3).
2.4. Specific intensity
Transport theory deals with the propagation of energy in a medium containing randomly
placed particles. For a given point ρ and a given direction specified by vector
^
t = cosϕx^ + sinϕ ^y (or equivalently by the azimuth angle ϕ), the power flux density within a
unit frequency band centered at frequency f within a unit angle is defined as specific intensi‐
ty and denoted by I(ρ,ϕ) [W m−1rad−1Hz-1]. Hence, the amount of power dP flowing along
direction ϕ within an angle dϕ through an elementary width dl with normal that forms an
angle Δϕ with ϕ (see Figure 3) in a frequency interval (f, f+df) is:
dP = I (ρ, ϕ)cosΔϕdldϕdf (16)
The specific intensity I(ρ,ϕ) as it appears in (16) could be related either to the power emitted
from a surface or to the power received by the unit width. As far as the single particle of §
2.3 is concerned, the specific intensity carried by the incident plane wave (2) is:
I i (ρ, ϕ) = Si δ(ϕ − ϕ) (17)
in which δ(.) is the Dirac Delta function, whereas, recalling (9), the specific intensity for the
scattered wave can be written as:
I s (ρ, ϕ) = | F (ϕi , ϕ) | 2Si = | F (ϕi , ϕ) | 2I i (ρ, ϕi ) (18)

Hence, the square modulus of the scattering function F(ϕi,ϕ) relates the incident specific in‐
tensity with the scattered specific intensity. In a random medium, the specific intensity is
computed as the ensemble average of the power per unit angle, frequency and area over the
distribution of the random scatterers.
Figure 3. Geometry used for the definition of specific intensity: the amount of power dP flows along direction ^within
an angle d through an elementary width dl whose normal n ^forms an angle Δ with the direction of propagation^.
3. Radiative transfer equation
The radiative transfer equation is an integro-differential equation that governs the propaga‐
tion of specific intensity within a random medium. Let us assume that the random medium
is a made of uniform slabs in the x-direction (see Figure 4). Therefore, specific intensity is a
function of position in space only through y. In the following the derivation of the radiative
transfer equation is briefly reminded. The reader is referred to [15, 17] for further details.
Figure 4. Geometry of the propagation environment.

3.1. Reduced intensity
Let us consider an area with side 1 in the x-direction and dy in the y-direction The area dy
contains Ndy particles, where N is the number of particle per unit area. The particles are uni‐
formly distributed in space. Each particle absorbs the power Pa = σaI and scatters the power
Ps = σsI, so that the decrease of specific intensity due to area dy is (recall (12)):
sinϕ ⋅ dI (y, ϕ) = − dy N σt I (y, ϕ) = m − dy ⋅ kext I (y, ϕ) (19)
where kest = Nσext [m-1] is the extinction coefficient. Notice that if there are different kinds of
particles (say m), each with possibly different orientation, density Nj and extinction cross
width σext,j [9]:
m
kext = ∑ N j < σext , j > (20)
j=1
where < > represents the ensemble average over the distribution of particles orientations.
Equation (19) defines the so-called reduced intensity since it only takes into account the ex‐
tinction of the incident wave [9].
3.2. Independent scattering and limits of the transfer theory
Scattering of waves impinging on the area dy from all directions ϕi increases the intensity
along direction ϕ according to (18). As the scatterers are assumed to be independent, the
specific intensities due to different particles can be added. This is strictly true under some
conditions that are discussed in the following as this point entails the limits of applicability
of the radiative transfer theory [15]. As the Maxwell equations are linear, the total scattered
field E can be written as the sum of the Ej fields scattered from each particle:
E =∑ E j (21)
j
The specific intensity is proportional to the square modulus of the electric field averaged
over the distribution of scatterers:
< | E | 2 > = ∑ < | E j | 2 > + ∑ ∑ < E j E l* >

j j l≠ j
jαj
Now, letEj = | Ej | e , we get that:
j (αj −αl )
< E j E l* > = < | E j | | E l | e
The phase difference (αj− αl) depends on the distance between the particles djl through the
product k0djl. If the distribution of particle separation is not much smaller than the wave‐
length, i.e. the standard deviation (S.D.) of djl satisfies:
λ
S.D.(d jl ) ≥ (22)
4
then (αj− αl) is approximately uniformly distributed within [0, 2π] so that:
< E j E l* > = 0
From the discussion above, the assumption underlying the radiative transfer equation of in‐
dependent scattering limits the applicability of the transfer theory to cases where the dis‐
tance between particles is large enough (see (22)) so as to make negligible the near far
interactions between particles. Experimental studies confirm this conclusion: for the radia‐
tive transfer to be applicable the spacing between scatterers must be larger than λ/3 and
0.4D where λ is the wavelength of the host medium and D the diameter of the scatterers
[10]. Theoretical studies on the relationship between the radiative transfer approach and the
wave approach using Maxwell’s equations can be found in [22-23]. Moreover, the limits of
the transfer theory are investigated for a two-dimensional problem similar to the one con‐
sidered here in [16] through comparison with the wave approach.
Assuming independent scattering, the increase on the specific intensity along direction ϕ
due to scattering within the area dy is:
2π 2π
∫
dy N | F (ϕi , ϕ) |
0
2 I (ρ,
∫
ϕi )dϕi = dy p(ϕi ϕ)I (ρ, ϕi )dϕi
0
(23)
wherep(ϕi ϕ) = N F (ϕi , ϕ) is the phase function. Notice that if the random medium contains
particles of different kinds, the overall phase function is defined as [9]:
m
p(ϕi , ϕ) = ∑ N j < pj (ϕi , ϕ) > (24)
j=1
3.3. Progressive and regressive intensity
The radiative transfer equation is obtained by combining (19) and (23):
2π
sinϕ
∂ I (y, ϕ)
∂y
= − kext I (y, ϕ) + ∫ p(y, ϕ , ϕ)I (y, ϕ )dϕ
0
i i i
(25)
In this formulation, the extinction coefficient and the phase function are considered function
of the position y. A first step toward the solution of scalar radiative transfer problem is con‐
verting equation (25) into two coupled integro-differential equations by introducing the pro‐
gressive intensity I+, that corresponds to propagating directions 0 < ϕ < π, and the regressive
intensity I-, that accounts for the propagating directions π < ϕ < 2π. Τhe two specific intensi‐
ties are defined as:
I +(y, ϕ) = I (y, ϕ)
I −(y, ϕ) = I (y, ϕ + π)
where the azimuth direction ϕ ranges within 0 < ϕ< π. The scalar radiative transfer equation
(25) can be equivalently stated as:
sinϕ
∂ I +(y, ϕ)
∂y ∫
= − kext I +(y, ϕ) + p(y, ϕi , ϕ)I +(y, ϕi )dϕi +
0
π
∫
+ p(y, ϕi + π, ϕ)I −(y, ϕi )dϕ
0
π
(26)
− sinϕ
∂ I −(y, ϕ)
∂y ∫
= − kext I −(y, ϕ) + p(y, ϕi , ϕ + π)I +(y, ϕi )dϕi +
0
π
∫
+ p(y, ϕi + π, ϕ + π)I −(y, ϕi )dϕ
0
This equivalent formulation of (25) makes it easier to set the boundary conditions as ex‐
plained in §3.3.1 and §3.3.2. For a uniform distribution of the particles over the random me‐
dium and circular cylindrical particles, the phase matrix becomes a function ϕ-ϕi only and
the extinction matrix becomes independent on ϕ [17]. Therefore, in this case the scalar radia‐
tive transfer equations (26a-26b)can be written as:
sinϕ
∂ I +(y, ϕ)
∂y ∫
= − kext I +(y, ϕ) + p +(y, ϕ − ϕi )I +(y, ϕi )dϕi +
0
π
(27)
∫ −
+ p (y, ϕ − ϕi )I (y, ϕi )dϕ
0
−
− sinϕ
∂ I −(y, ϕ)
∂y ∫
= − kext I −(y, ϕ) + p −(y, ϕ − ϕi )I +(y, ϕi )dϕi +
0
π
(28)
∫
+ p +(y, ϕ − ϕi )I −(y, ϕi )dϕ
0
where p +(y, ϕ − ϕi ) = p(y, ϕi , ϕ) andp −(y, ϕ − ϕi ) = p(y, ϕi , ϕ + π) = p(y, ϕ + πi , ϕ), based on

the 2π periodicity of both p+ and p- [17].
3.3.1. Boundary conditions on the specific intensity
The radiative transfer equation has to be solved by imposing appropriate boundary condi‐
tions. Here, the boundary conditions that the specific intensity must satisfy on a plane boun‐
dary between two media with indices of refraction n1 and n2 are considered. The expression
for the transmission T and reflection coefficient R for the two polarizations are known in lit‐
erature [20]. The reflected specific intensity Ir is related to the incident specific intensity Ii as:
I r = | R | 2I i
As far as the transmitted specific intensity is concerned, we can write the conservation of
power on a segment dl belonging to the boundary (recall (16)):
I i cosϕ1dϕ1dl = I r cosϕ1dϕ1dl + I t cosϕ2dϕ2dl (29)
As by the Snell’s lawn1sinϕ1 = n2dsinϕ2, it ends:
I t n2
= (1 − | R | 2)
I i n1
4. Solution of the radiative transfer equation through numerical

quadrature
The radiative transfer equation is an integro-differential equation whose solution in analyti‐

cal form is very difficult, if not impossible. However, efficient numerical solutions can be de‐
vised. A comprehensive treatment of the main techniques can be found in [15, 24].
The case under study concerns a random medium where relevant scattering occurs. An ap‐
proximate solution can be obtained by computing the integrals in (26) by numerical quadra‐
ture as firstly proposed in [15]. The continuum of propagation directions ϕ is discretized
into a set of n directions ϕiI = (i-1) Δϕ, where Δϕ = π/n and i = 1,..., n, and the corresponding
n×1 vectors gathering the progressive and regressive intensities:
I +(h , ϕ1)
I +(h , ϕ2)
I +(r) = (30)
⋮
I +(h , ϕn )
I −(h , ϕ1)
I −(h , ϕ2)
I −(r) = (31)
⋮
I −(h , ϕn )
The two integro-differential equations (27) can be approximate as follows:
d I +(y)
= − K (y)I +(y) + P +(y)I +(y) + P −(y)I −(y) (32)
dy
d I −(y)
= − K (y)I −(y) − P −(y)I +(y) − P +(y)I −(y) (33)
dy
where then ×n matrices P+ and P- are defined as:
1
P ±(y) i, j = p ±(y, ϕi − ϕj )Δϕ
sinϕi
so that:
π
1
∫
n
p ±(y, ϕ − ϕ ')I ±(y, ϕ ')dϕ ' ≈ ∑ P ±(y) i, j I
±
(y, ϕj ) (34)
sinϕi j=1
0
The n×n K(y) matrix is defined as:
1
K (y) i, j = k (y)i = j
sinϕi ext (35)
= 0i ≠ j
Now, defining the 2n×1 vectorI (y) = I +(y)T I −(y)T , one obtains the system of first order lin‐
ear equation:
d I (y)
= G(y)I (y) (36)
dy
where
− K (y) + P +(y) P −(y)

G(y) =
− P −(y) K (y) − P +(y)
4.1. Discrete ordinate Eigen analysis
If the space contains slabs of homogeneous random (and non-random) media, the linear dif‐
ferential system (36) within each slab (say the l th) can be written as
d I (y)
= Gl I (y) (37)
dy
being matrix Gl a constant. This follows from the fact that, the extinction coefficient and the
phase function within each slab are constant. In this case, the system (37) can be solved by
the discrete ordinate analysis method as follows. Let I (y) = βexp( − λg)be a tentative solution
where β is a 2n×1, one get by substitution in (36) - dropping l for simplicity:
(λU + G)β = 0
As a result, (β,λ) represent pair of eigenvectors/eigenvalues of the matrix –G. Notice that
given the symmetry relations:
Gl (n+i))n+i)
= − Gl i,i
Gl (n+i))n+ j)
= − Gl iji
i, j = 1, ⋯ , n
by ordering the eigenvalues in increasing order (λi ≥ λi−1) it follows:
λ(n+i) = − λi i = 1, ⋯ , n
and the corresponding eigenvectors satisfy the condition:

β(n+i) (n+)
= βi j i = 1, ⋯ , n
The solution of (36) can be written as the linear combination
2n
I (y) = ∑ βi e
(−λi y)ci
= BD(y)c (38)
i=1
whereB = β1 ⋯ β2n , c = c1 ⋯ c2n T , and D(y) is a diagonal matrix with elements

(−λ1 y) (−λ2n y)
e ⋯e .
The vector of unknown constant c can be computed from the knowledge of I(y) for one val‐
ue of y, sayȳ, as:
c = ( BD(ȳ))−1I (ȳ) (39)
4.1.1. Setting the boundary conditions for a single slab

Given geometry depicted in figure 4, with only one slab of random medium (ranging within
0 ≤ y ≤d), the semi-infinite medium along the positive direction of y is a dielectric one with
given dielectric constant. The boundary conditions read:
I +(0) = Ī +0 (40)
I −(d) = AR Ī +(d) (41)
where Ī +0is the incident specific intensity and AR is the n×n reflection matrix relative to the
interface at y = d between the random medium and the last slab. This can be easily computed
from the results in §2.4. The value of the unknown constant vector (39) for the specific inten‐
sity within the random medium is obtained through the computation of the specific intensi‐
tyI (ȳ). To do this, as first step one need to calculate from (38) and (39):
I (d ) = BD(d )B -1 I ( y ) I (0)
14243 (42)
T
Then, using (40) and (41), and partitioning the matrix T as:
T 11 T 12
T= (43)
T 21 T 22
in which Tij is n×n, one gets:
I (0)− = ( AR T 12 − T 22)−1(T 21 − AR T 11)Ī +0
The substitution of this finding in (39) is used for the computation of the vector c.
4.1.2. Setting the boundary conditions for a multi-layer medium
In the multi-layered geometry of figure 4, each slab has depth dl and the specific intensity is
(38). The unknown is the constant vector cl and the subscript l runs over the different slabs, l
= 1,..., L. Notice that slab can contain either a random medium characterized by its matrix Gl
or vacuum (matrix Gl=0 for vacuum). The semi-infinite medium along the positive direction
of y is a general dielectric with given dielectric constant. The boundary conditions (40) and
L
(41) still hold withd = ∑ dl .
l=1
The specific intensity Il,2 - at the interface between the l and the (l + 1)th slab - is related to the
specific intensity Il,1 at the interface between the (l – 1) and l th slab as:
I l ,2 = Bl D(dl )B l -1 I ( y ) I l ,1
14 4244 3 (44)
Tl
Therefore, similarly to (42) we can write:
Proceeding as in the previous Section, the specific intensity I(0) is now calculated and used
in (39) to obtain the constant vector c1 for the first slab. Then, equation (44) can be iteratively
applied so as to compute the specific intensity at the beginning of each slab. The latter al‐
lows through (39) the computation of cl for each slab.
5. Numerical results
In this section two examples illustrate how radiative transfer theory could be employed to
study the beam broadening and the corresponding spatial correlation for an indoor environ‐
ment within the context of a communication system.
The main assumption is that the propagation environment can be modeled as a layered par‐
allel plane medium. This situation is a first approximation of an open space office made of a
succession of tables where each table is modeled as random medium containing randomly
placed scatterers. These are modeled as circular cylinder so as to simplify the solution of the
radiative transfer equation.
5.1. Case study I: one table
In this example, the geometry under study is depicted in Figure 5. The whole xy plane is
characterized by the vacuum dielectric constant ε0. One slab ranging within 1 ≤ y ≤ 1.8 con‐
tains N [m−2] uniform randomly distributed circular cylinders with radius a = 6 cm. Where
not stated otherwise, frequency of operation is f = 5.2 GHz, according to standard wireless
local area networks (WLAN) such as IEEE802.11x and Hyperlan/2.
Figure 5. Geometry of the propagation environment (case study I).
5.1.1. Beam broadening
Figure 6 shows the specific intensity I+(y,ϕ) for N=10, ε’r =4, σ=10−3 Sm-1 and vertical polariza‐
tion. Specifically, the image represents the beam broadening of a plane wave propagating in
the direction ϕi = 90° with specific intensity I 0+= 1 [W m−1rad−1Hz].
The specific intensities are assumed to be normalized with respect to I 0+and thus are shown
in dB. Right after the entrance into the table region the progressive beam I+(y, ϕ) broadens
since the energy is scattered in all directions by the cylinders. As expected the regressive beam
I−(y, ϕ) is zero on the right side of the table and almost uniform in the yϕ plane (not shown).
Figure 6. Specific intensity I+(y,) versus y (vertical polarization, ε’r = 4, σ = 10−3 Sm-1, N = 10).
In order to get a quantitative insight into the beam broadening discussed above, Figure 7
shows the specific intensity I+(y, ϕ) for y = 1.9 m (on the right of the table) for different val‐
ues of the density N=2, 10 (vertical polarization). The reduced or line of sight (LOS) contri‐
bution as shown in the box decreases for increasing object densities. However, a larger
density of scatterers entails a more relevant contribution of the diffuse energy (i.e., I+(y, ϕ)
for ϕ ≠ 90°).
Figure 7. Specific intensity I+(y, ) for y = 1.9 m (on the right of the table) for different values of the density N = 2, 10
(vertical polarization, ε’r = 4, σ=10−3 Sm-1).
Figure 8. Specific intensity I+(y, ) for y = 1.9 m (on the right of the table) for different values of the frequency f = 5.2, 52
GHz (vertical polarization, N = 6, ε’r = 4, σ = 10−3 Sm-1).
The effect of an increase in the carrier frequency, envisioned for next generation wireless
LAN, is shown in Figure 8. A larger carrier frequency (f = 52 GHz) yields a less consistent
contribution of the LOS direction (see box) and a greater number of interference fringes. Em‐
ploying the horizontal polarization instead of the vertical one yields qualitatively similar re‐
sults as shown in Figure 9 for N=6. Therefore, in the following only the vertical polarization
is considered.
Figure 9. Specific intensity I+(y, ) for y = 1.9 m (on the right of the table) for horizontal and vertical polarization fre‐
quency (f = 5.2 GHz, N = 6, ε’r = 4, σ = 10−3 Sm-1).
5.1.2. Spatial correlation
In a communication link, it is of great interest to assess the degree of correlation between the
signals received by different antennas as a function of their inter-spacing Δ [m]. In fact,
wireless links can capitalize on the uncorrelation between the received signals to increase
the degree of diversity of the system that in turn rules the asymptotic performance of the
link in terms of probability of error [25].
The correlation ry(Δ) of the signals received by antennas separated by Δ can be expressed in
terms of the probability density function py(ϕ) of the direction of arrivals of the waves im‐
pinging on the receivers. This function can be obtained by interpreting (after appropriate
scaling) the power received over a certain direction ϕ as a measure of the probability that a
signal is received through direction ϕ:
I +(y, ϕ)
py (ϕ) =
∫I (y, ϕ)dϕ
+
(45)
Then, recalling that the ratio between signals received at two points separated by Δ in the
direction x isexp( − j2π / λcosϕΔ):
∫ (− )
j 2π
cosϕΔ (46)
ry (Δ) = py (ϕ)e λ
dϕ
0
The correlation (46) is evaluated for the example at hand under the same assumptions as in
Figure 7.
Figure 10. Correlation ry(Δ) versus Δ/ λ for y = 1.9 m (on the right of the table) for different values of the density N = 2,
10 (vertical polarization, ε’r = 4, σ = 10−3 Sm-1).
The results are shown in figure 10. The correlation decreases with increasing object density.
Thus, from the perspective of communication system performance, increasing the object
density is beneficial in terms of degree of diversity. For instance, for Δ = λ the correlation
decreases from around 0.9 to around 0.7.
5.2. Case study II: two tables with interface
In this example the environment has a more complicated geometry, where two tables are
followed by a semi-infinite dielectric slab (dielectric constant εr,I) as shown in figure 11. The
simulation parameters follow the setting described for Case study I, in particular the radius
of the scatterers is a=6 cm and the carrier frequency is f=5.2 GHz. The dielectric constant of
the slab is chosen as εr,I =3.
Figure 11. Geometry of the propagation environment (case study II).
5.2.1. Beam broadening
Figure 12 shows the specific intensity I+(y, ϕ) for N = 6, ε’r = 4, σ = 10−3 Sm-1 and ϕi=90° and
vertical polarization. Again, the specific intensity is normalized with respect to I 0+ and thus
are shown in dB. As discussed for case study I, right after the entrance into the table regions
the progressive beam I+(y, ϕ) broadens since the energy is scattered in all directions by the
cylinders. Moreover, the refraction over the interface at y=4.6 m focuses the beam within the
semi-infinite dielectric, hence reducing its angular spread. The refraction and the reflection
of the reduced (or LOS) component are apparent in Figures 12 and 13, respectively. The inci‐
dence angle is chosen equal to 45°. Refracted and reflected angles satisfy the Snell’s law.
Figure 12. Specific intensity I+(y, ) versus y (vertical polarization, ε’r = 4, σ = 10−3 Sm-1, N = 6, i=90°).
Figure 13. Specific intensity I+(y, ) versus y (vertical polarization, ε’r = 4, σ = 10−3 Sm-1, N = 6, i=45°).
Figure 14. Specific intensity I-(y, ) versus y (vertical polarization, ε’r = 4, σ = 10−3 Sm-1, N = 6, i=45°).
5.2.2. Spatial correlation
The spatial correlation ry(Δ) is shown for the same parameters of the previous example in
Figure 15. It is computed for y = 1.9 m (just on the right of the first table) and y = 3.7 m (just
on the right of the second table). As expected, the increased scattering contribution due to
the random scatterers of the second table decreases the spatial correlation. In particular, for
inter-antenna spacing around Δ = 5λ the correlation is decreased from 0.75 to around 0.61.
Figure 15. Correlation ry(Δ) versus Δ/ λ for y = 1.9 m (on the right of the first table) and y = 3.7 m (on the right of the
second table) for N = 6, vertical polarization, ε’r = 4, σ=10−3 Sm-1, i=90°.
5.3. Limits and validity of radiative transfer predictions
The validation of numerical results presented in §5.1 and §5.2 relies on a discussion, as there
is no empirical evidence (measurements) to compare with. The two chosen cases are not re‐
alistic but they represent a possible benchmark to be employed against the numerical results
that issue from analytical methods. Among these, the T-matrix approach [26, 27] seems to be
the best suited to compute scattering from a random distribution of cylinders and to com‐
pare the radiated fields on a realization-by-realization basis with ray tracing. At this stage of
development, the only comparison that the author is able to provide to assess the validity of
a radiative transfer approach is against a ray tracing technique based upon the beam tracing
method [26]. The numerical code used for the simulations was developed at the Politecnico
di Milano [29] and it has already been used for different purposes related to the indoor
propagation [30].
To make the beam tracing procedure suitable for the study of electromagnetic propagation,
it is necessary to include the reflection coefficient associated with the interactions of the path
with the environment. In particular, the signal received from each path has to be scaled by
the product of the reflection coefficients corresponding to the bounces each path goes
through when propagating from the transmitter to the receiver. To compare the outcome of
the ray tracing simulation with the results of radiative transfer theory some approximations
were made. Case I was reproduced in such a way to replicate the same geometrical and ra‐
dio electrical conditions. The reader is referred to [31] for further details about the numerical
computation and the comparison results, while here the author recalls the general approach
and the basic information. The transmitted plane wave was approximated by building a lin‐
ear antenna array of NT elements with inter-element spacing ΔT lying parallel to the y axis.
To fulfill the requirement of far field regime, the transmitting array was placed at a range
distance R > ( N ΔT )2 / λ [9].
The circular scatterers were approximated by polygons of NP sides. The quantities of interest
were averaged over NI realizations of the random medium. The signal is received at the de‐
sired points by a linear antenna array of NR elements with inter-element spacing ΔR lying
parallel to the y axis. Figure 16 illustrates the geometry together with the transmitting and
the receiving antenna arrays.
Figure 16. Geometry of the experiment studied by means of ray tracing.
The specific numerical values for the ray-tracing algorithm were selected after careful em‐
pirical investigation to yield negligible approximation errors. The transmitting array has NT=
100 elements and inter-element distance ΔT=λ/2, while the receiving array has NR= 80 ele‐
ments with inter-element spacing ΔR = λ/4, thus resulting in an angular resolution of approx‐
imately Δϕ=6°. The key parameter that discriminates the reliability of the radiative transfer
approach was identified as the fraction of area that is effectively occupied by the scatterers.
Let Aext = π(σext / 2)2 be the effective area occupied by each particle. The fraction of effective
area that is occupied by the scatterers isη = N Aext . The numerical investigation proved that
for scatterers that yield a relatively low value of η the radiative transfer provides a solution
that closely matches the second order statistics (i.e. power fluctuations) as given by the ray
tracing results.
Figure 17 shows the progressive specific intensity I +(y,ϕ) computed according to radiative
^
theory compared with I (y, ϕ) from the ray tracing procedure. To compare the two ap‐
proaches, the effect of limited angular resolution is accounted for by processing the outcome
of the radiative transfer I+(y,ϕ) and the number of averaging iterations for ray tracing is NI =
100 [31]. By increasing the density N, and consequently the fraction η of effective area occu‐
pied by the scatterers, the difference between the prediction of radiative transfer and those
from ray tracing increases.
Figure 18 shows the comparison between the spatial correlation r y(Δ) computed by the radi‐
ative transfer theory and the same quantity ^r y (Δ)from ray tracing versus Δ/λ for y = 1.9 m
(on the right of the table) and two values of the density N=6,10.
Figure 17. Specific intensity I+(y,) computed according to the radiative theory with and without limited resolution and
^
I (y, ) from the ray tracing procedure for ε’r = 4, σ = 10−3 Sm-1, y = 1.9 m (on the right of the random medium), N = 10 m
−2
and vertical polarization.
Figure 18. Spatial correlation ry(Δ) computed by the radiative transfer theory and ^r y (Δ) from ray tracing versus Δ/λ for
y = 1.9 (on the right of the random medium) and different values of the density N = 6, 10 m−2 (vertical polarization and
ε’ r = 4, σ = 10−3 Sm-1).
6. Conclusions
In this work, the use of radiative transfer theory to study the propagation in an indoor envi‐
ronment was reported. A particular focus was made on two specific aspects related to the
channel performances, such as the beam broadening and the spatial correlation, as they are
parameters of interest when deploying a network. The pretty good agreement, although un‐
der well-controlled conditions, encourages proceeding along two distinct directions. From
one side, an improvement in the modeling and in the sketching of actual environments is
mandatory if one wants to use the radiative transfer technique as a possible approach to an‐
alyze indoor channel performances in practical scenarios. From the other side, validation
should be performed either against measurements or - at least - against appropriate analyti‐
cal solution without limitations or approximations in the frequency domain (or wavelength
scale) of interest.
Acknowledgements
The author would like to gratefully acknowledge the work of Osvaldo Simeone (now Assis‐
tant Professor at the New Jersey Institute of Technology), who contributed with ideas, nu‐
merical developments and simulations to this research during his doctoral studies in Milano
at the Politecnico, Faculty of Telecommunication Engineering.
Author details
Ada Vittoria Bosisio*
Address all correspondence to: [email protected]
National Research Council of Italy, Istituto di Elettronica ed Ingegneria dell’Informazione e

delle Telecomunicazioni, Italy
References
[1] Falsafi, A., Pahlavan, K., & Ganning, Y. (1996). Transmission techniques for radio
LAN’s-a comparative performance evaluation using ray tracing. Journal on Selected
Areas in Communications, 14(3), 477-491.
[2] Ji, Z., Li, B., Wang, Chem, H., & Sarkar, T. K. (2001). Efficient ray-tracing methods for
propagation H. prediction for indoor wireless communications. Antennas and Propa‐
gation Magazine, 43(2), 41-49.
[3] Stuber, G. (2001). Principles of Mobile Communication. Dordrecht (The Nether‐

lands), Kluwer Academic Publishers.
[4] Ullmo, D., & Baranger, H. U. (1999). Wireless propagation in buildings: A statistical
scattering approach. IEEE Transactions on Vehicular Technology, 48(3), 947-955.
[5] Iskander, M. F., & Yun, Z. (2002). Propagation prediction models for wireless com‐
munication systems. IEEE trans. Microwave Theory and Techniques, 50(3), 662-673.
[6] Valenzuela, R. A. (1994). Ray tracing prediction for indoor radio propagation in: Pro‐
ceedings of 5th IEEE International Symposium on Personal, Indoor and Mobile Ra‐
dio Communication, PIMRC1994, 18-23 September 1994, The Hague, The
Netherlands, 1(3), 140-144.
[7] Muller, A. (1994). Monte-Carlo multipath simulation of ray tracing channel models
in: Proceedings of IEEE Global Telecommunications Conference GLOBECOM’94, 28
November - 2 December 1994, San Francisco (CA), 3, 1446-1450.
[8] Hassan-Ali, M., & Pahlavan, K. (2002). A new statistical model for site-specific indoor
radio propagation prediction based on geometric optics and geometric probability.
IEEE Trans on Wireless Communications, 1(1), 112-124.
[9] Ishimaru, A. (1978). Wave propagation and scattering in random media. Chap 7 and
11, New York, Academic Press, 1.
[10] Chandrasekhar, S. (1960). Radiative transfer theory, New York, Dover Publications.
[11] Kim, A. D., & Ishimaru, A. (1998). Optical and millimeter wave pulse propagation
through fog and rain layers in Proceedings of International Geoscience and Remote
Sensing Symposium IGARSS98, 6-10 July 1998, Seattle (WA), 1, 33-35.
[12] Caldeirinha, R. F. S., & Al-Nuaimi, M. O. (2002). Propagation modeling of bistatic

scattering of isolated trees for micro- and millimetre wave urban microcells in Pro‐
ceedings of 13th IEEE International Symposium on Personal, Indoor and Mobile Ra‐
dio Communication PIMRC’02, 15-18 September 2002, Lisbon, 1(3), 135-139.
[13] Al-Nuaimi, M. O., & Hammoudeh, A. M. (1994). Measurements and predictions of

attenuation and scatter of microwave signal by trees. IEEE Proceedings on Microwave,
Antennas and Propagation, 141(2), 70-76.
[14] Hawley, M. S., Conway, J., & Anderson, A. P. (1998). The influence of tissue layering
on microwave thermographic measurements. International Journal of Hyperthermia,
4(4), 427-435.
[15] Ulaby, F. T., Moore, R. K., & Fung, A. K. (1986). Microwave Remote Sensing. Active
and Passive, Vol. III: From Theory to Applications. Norwood (MA), Artech House.
[16] Roux, L., Mareschal, P., Vukadinovic, N., Thibaud, J. B., & Greffet, J. J. (2001). Scatter‐
ing by a slab containing randomly located cylinders: comparison between radiative
transfer and electromagnetic simulation. Journal of the Optical Society of America A,
18(2), 374-384.
[17] Sarabandi, K., Tavakoli, A., & Ulaby, F. T. (1992). Propagation in a two-dimensional
periodic random medium with inhomogeneous particle distribution. IEEE Transac‐
[18] Ishimaru, A., Jaruwatanadilok, S., Ritcey, J. A., & Kuga, Y. (2010). A MIMO Propaga‐
tion Channel Model in a Random Medium. IEEE Transactions on Antenna and Propaga‐
tion, 58(1), 178-186.
[19] Van de Hulst, H. C. (1957). Light scattering by small particles, place, J. Wiley & Sons.
[20] Ramo, S., Whinnery, J. R., & Van Duzer, T. (1994). Fields and waves in communica‐
tion electronics, place John Wiley and sons.
[21] Karam, M. A., Fung, A. K., & Antar, Y. H. M. (1998). Electromagnetic wave scattering
from some vegetation samples. IEEE Transactions on Geoscience and Remote Sensing,
26(6), 799-808.
[22] Fante, R. L. (1981). Relationship between radiative transfer theory and Maxwell’s
equations in dielectric media. Journal of the Optical Society of America, 71(4), 460-468.
[23] Ishimaru, A., & Kuga, Y. (1982). Attenuation constant of a coherent field in a dense
distribution of particles. Journal of the Optical Society of America, 72(10), 1317-1320.
[24] Tsang, L., Kong, J. A., & Shin, R. (1984). Radiative transfer theory for active remote
sensing of a layer of nonspherical particles. Radio Science, 19(2), 629-642.
[25] Acampora, A., & Winters, J. (1987). System applications for wireless indoor commu‐
nications. IEEE Communications Magazine, 25(8), 11-20.
[26] Siqueira, P. R., & Sarabandi, K. (2000). T-Matrix Determination of Effective Permittiv‐
ity for Three-Dimensional Dense Random Media. IEEE Transactions on Antennas and
Propagation, 48(2), 317-327.
[27] Chew, W. C., Wang, Y-M., Otto, G., Wagner, R. L., & Liu, Q. H. (1992). A Generalized
Recursive Algorithm for Wave-Scattering Solutions in Two Dimensions. IEEE Trans‐
actions on Microwave Theory and Techniques, 40(4), 716-723.
[28] Funkhouser, T., Carlbom, I., Elko, G., Pingali, G., Sondhi, M., & West, J. (1998). A
beam approach to acoustic modeling for interactive virtual environment in ACM
Computer Graphics Proceedings of SIGGRAPH’98; Orlando (F), 19-24 July, 21-32.
[29] Foco, M., Polotti, P., Sarti, A., & Tubaro, S. (2003). Sound spatialization based on fast
beam tracing in the dual space in Proceedings of the International Conference
DAFx-03, London (UK), 8-11 September.
[30] Bosisio, A. V. (2011). Performances of an RSSI-based positioning and tracking algo‐

rithm in Proceedings of International Conference on Indoor Positioning and indoor
Navigation IPIN2011, Guimaraes (P), 21-23 September, doi:10.1109/IPIN.2011.6071952.
[31] Bosisio, A. V., & Simeone, O. (2007). Comparison between radiative transfer and ray
tracing for indoor propagation in Proceedings of International Conference on Elec‐
tromagnetics in Advanced Applications ICEAA07, Torino (I), 17-21 Sept., 395-398.
Chapter 10
Electromagnetic Wave Packets in the Theory of

Bremsstrahlung and Transition Radiation by High-
Energy Electrons
Nikolai Fyodorovich Shul’ga and

Sergii Valeriyovich Trofymenko
1. Introduction
A lot of high-energy physical processes develop within large domains of space along the di‐
rection of particle motion (see, for example monographs [1-3] and references in them). In the
case of electromagnetic processes the size of these domains can substantially exceed some‐
times not only interatomic distances of substance but the size of experimental facility (detec‐
tors) as well [1,2,4-13]. Essential in this case is the fact that interaction of particles with atoms
and experimental facility situated within such domains and outside them can substantially
differ. Such situation arises, for example, when considering long-wave radiation in process‐
es of bremsstrahlung and transition radiation by ultra relativistic electrons. Therefore, it is
necessary to know what happens within such regions and what the peculiarities of evolu‐
tion of such processes in space and time are. The present chapter is dedicated to the consid‐
eration of different aspects of this problem, which concern the behavior of high-energy wave
packets, which take place in processes of bremsstrahlung and transition radiation.
We begin with consideration of the behavior of localized high-energy wave packets of a sca‐
lar massive particle in wave mechanics [2,14], of Gaussian packets of free electromagnetic
waves and of packets, which take place in the equivalent photon method [15]. It is shown
that in all cases high energies make the stabilizing effect upon the packet motion. Some pe‐
culiarities of dispersion of such packets and their reconstruction into the packets of diverg‐
ing waves are considered as well.
Further we show that the discussed wave packets naturally arise in processes of bremsstrah‐
lung and transition radiation by high-energy electrons. For this purpose, firstly, the proc‐
ess of bremsstrahlung at an ultra relativistic electron instantaneous scattering to a large
angle is considered [16,17]. The consideration is made on the basis of classical electrodynam‐
ics. In this case the moving electron is considered as a charge with its own coulomb field
moving together with it. At the instantaneous scattering the perturbation of this field oc‐
curs. This perturbation is treated here as appearance of a packet of free plane electromag‐
netic waves, which reconstructs then into a packet of diverging waves. For ultra relativistic
particles, however, this does not happen at once. The length within which this process de‐
velops has a name of the coherence length of the radiation process [1,2]. It is 2γ 2 times larg‐
er than the length λ of the considered radiated waves (γis here the electron's Lorentz-
factor). We show that within this length the field around the electron substantially differs
from the coulomb one. This leads, in particular, to the fact that the bremsstrahlung charac‐
teristics in this case substantially depend on both the detector’s size and its position rela‐
tive to the scattering point.
It is shown further that analogous effects take place in the process of transition radiation by
an ultra relativistic electron during its traverse of thin ideally conducting plate as well
[16-18]. The picture of evolution in space and time of the electromagnetic field, which arises
before and after the electron traverse of thin metallic plate is considered here. The main at‐
tention is paid to effects in the process of backward transition radiation. In this case the
wave packets of the field reflected from the plate are the packets of free waves, which recon‐
struct into the field of transition radiation. We show that the structure of these packets is in
many respects analogous to the structure of the packets, which take place at instantaneous
scattering of the particle to a large angle. This fact explains the presence of analogous effects
in transition radiation and bremsstrahlung in the considered cases.
The special attention is drawn to the transition radiation by a scattered electron, which own
field is not totally reconstructed after the scattering [17-19]. During a long period of time
in this case the electron is in ‘half-bare’ state, which is the state in which some Fourier har‐
monics in the field around the electron are suppressed compared to the equilibrium cou‐
lomb field. The large values of distances, which the electron covers in this state allow us
to place the plate within these distances and to consider the transition radiation by such elec‐
tron on this plate. We show that in this case characteristics of backward transition radia‐
tion substantially differ from transition radiation characteristics in the case when the target
is situated on large distances from the scattering point. The effect of transition radiation sup‐
pression and the effect of oscillatory dependence of transition radiation characteristics on
the distance between the plate and the scattering point take place in this case. The causes
of such effects are discussed.
Electromagnetic Wave Packets in the Theory of Bremsstrahlung and Transition Radiation by High-Energy Electrons 265
2. High energy wave packets
2.1. Dispersion of relativistic wave packets
The general solution of the wave equation can be presented in the form of a wave packet,
which spatially disperses in course of time. In semiclassical approximation such packet does
not disperse. It moves according to the laws of classical mechanics (see, for example [2, 20]).
It is going beyond the semiclassical approximation that leads to the packet dispersion. The
high-energy wave packets are of special interest because the speed of their dispersion de‐
creases with the increase of their energy. Let us pay attention to some peculiarities of disper‐
sion of such packets. Significant here is the fact that characteristic features of this dispersion
are similar for all fields. Therefore it is sufficient to consider just scalar field.
The general solution of the wave equation
( ∂2
∂t 2
→
)
− ∇2 − m 2 ϕ(r , t) = 0 (1)
for a scalar particle with the mass m can be written in the following form of the expansion of
→
the field ϕ(r , t)over plane waves:
→
ϕ(r , t) = ∫
d 3κ i(κ→ r→ −ω t ) →
(2π)3
e Cκ (2)
→
where ω = κ 2 + m 2 and Cκ→ - are the expansion coefficients. Here and further we will use the
system of units in which the speed of light c and the Plank constant ℏ equal unit.
Let us consider the dispersion of the wave packet, which at the initial moment of time coin‐
cides with the Gaussian packet modulated by the plane wave with large value of the mo‐
→
mentum p [2,14]. Moreover we will assume that the initial widths of the packet a|| and a⊥
→
parallel and perpendicular to the particle momentum p are different. For such packet at the
→
initial moment of time the field ϕ(r , t) has the following form:
→
→→ z2 ρ2
→ i p r − 2a 2 − 2a 2 (3)
ϕ(r , t) = e || ⊥
→ →
where z and ρ are the coordinates parallel and orthogonal top . At the moment of time t this
packet will be defined by the relation (2) with
( p−κz ) 2a||2 κ⊥2a⊥2
Cκ→ = (2π)3/2a||a⊥2e
− 2 − 2 (4)
→
We can write the obtained expression for the field ϕ(r , t) in the form
→ →→ →
ϕ(r , t) = Ae i( p r −ε t )I (r , t) (5)
→
in which A = a||a⊥2, ε = p 2 + m 2and
∫
( p−κz ) 2a||2 κ⊥2a⊥2
→ → →
→ i(κ − p )r − − −i(ωk −ε)t (6)
I (r , t) = (2π)−3/2 d 3κe 2 2
→ → →
Having made in this expression the variable substitution κ = p + q we find that
∫
qz2a||2 q⊥2a⊥2
→→
→ iq r − − −i(ω →p +q→ −ε)t (7)
I (r , t) = (2π)−3/2 d 3qe 2 2
→
In the case of large energies it is possible to make the expansion over | q | / p in the quantity
(ω p→ +q→ − ε) in (7). Having preserved the quadratic terms of expansion we obtain
qz2 q⊥2
ω p→ +q→ − ε ≈ vqz + + (8)
2εγ 2 2ε
−1/2
where v = p / ε andγ = (1 − v 2) . Substituting this expression into (6) after simple calculations
we obtain
→
I (r , t) =
1
a||2 + i
1
t a2 + i t
εγ 2
⊥ ε
{
exp −
(z − vt)2
(
2 a||2 + i
t
εγ 2
−
ρ2
2 a⊥2 + i
t
ε ) ( ) } (9)
The formula (9) can be written in the following form as well:
→ →
{
I (r , t) = A(t)exp iα(r , t) −
(z − vt)2
−
ρ2
2Δ||(t) 2Δ⊥2(t)
2 } (10)
in which A(t)is a slowly changing quantity
1
A(t) =
( a⊥2 + i
t
ε
)a 2
|| +i
t
εγ 2
(11)
→
α(r , t)is the real phase
→ (z − vt)2 t / εγ 2 ρ2 t /ε
α(r , t) = + (12)
2 4 (
a|| + t / εγ 2)2 2 a⊥
4 (
+ t / ε )2
Δ||(t) and Δ⊥(t) are the longitudinal and transverse widths of the packet at the moment of
time t
Δ||2(t) = a||2 + ( t
a||εγ 2
2
) ( )
, Δ⊥2(t) = a⊥2 +
t
a⊥ε
2
(13)
In the case of a|| = a⊥ the obtained above formulae coincide with the corresponding result of
the paper [14].
The formulae (13) show that in longitudinal and transverse directions the squares of the widths
of the packet Δ||2(t) and Δ⊥2(t) grow with time proportionally to t 2m 4 / ε 6 andt 2 / ε 2. In nonre‐
lativistic case these quantities do not depend on the particle energy (t 2 / ε 2 = t 2 / m 2). In relativ‐
istic case the quantities t 2m 4 / ε 6 and t 2 / ε 2 are substantially smaller than the corresponding
values for nonreativistic particles. Let us note that the additional factor m 4 / ε 4 exists for lon‐
gitudinal direction inΔ||2(t). It leads to the substantial decrease of the speed of the packet dis‐
persion in this direction compared to the speed of the packet dispersion in transverse direction.
Thus the relativistic effects do the stabilizing influence upon the wave packets.
In conclusion let us note that while deriving the formula (9) we neglected the terms propor‐
tional to tq 4 / ε 3 in the exponential factor in (7). Therefore the formula (9) is valid during the
interval of time, which satisfies the condition
tqeff4 / ε 3 ≪ 1 (14)
→
where qeff are the characteristic values of the variable q in (7), which make contribution to
this integral. For a|| ≈ a⊥the inequality (14) can be written in the form
t 1
≪1 (15)
a⊥2ε (a⊥ε)2
Thus the formula (9) is valid for the time interval t, which satisfies the inequality (15). For ul‐
tra relativistic particles this interval rapidly grows with the increase of the particle’s energy.
2.2. Dispersion of a high-energy packet of electromagnetic waves
Now let us consider high energy packets of free electromagnetic waves. Scalar and vector
potentials of such packets are the solutions of the wave equation (1) with m = 0. Therefore in
order to analyze the peculiarities of dispersion of such packets we can use the formulae of
the previous section assuming that all the terms in them containing the Lorentz-factor γ
equal zero. In this case for scalar potential we find that
→ →
→ →
ϕ(r , t) = Ae i(k r −ω t )I (r , t) (16)
→
where k and ω are the wave vector and the frequency of the electromagnetic wave and
→
{ →
I (r , t) = A(t)exp iαk (r , t) −
(z − t)2
−
ρ2
2Δ||(t) 2Δ⊥2(t)
2 } (17)
Here
1 → ρ2 t /ω
A(t) = , α(r , t) =
(
a|| a⊥2 + it / ω ) 2 a 4 + (t / ω )2
⊥
(18)
and
Δ||2(t) = a||2, Δ⊥2(t) = a⊥2 + (t / a⊥ω )2
The obtained formulae show that the initially Gaussian packet does not disperse in the di‐
→
rection parallel to the k vector. In transverse direction the square of the packet widths grows
proportionally to(t / ω)2. Thus the speed of the packet dispersion decreases with the increase
of the wave frequencyω.
When considering a process of radiation by relativistic electrons it is often necessary to deal

with packets, which are constructed of plane waves with wave vectors, which directions are
→
close to the direction of a given vectork . Such wave packets differ somehow from the ones
considered above. Let us consider some peculiarities of dispersion of such packets assuming
for simplicity that at the initial moment of tome t = 0 the distribution of the waves over the
→
wave vectors is Gaussian relative to the given vector k [15]. For such distribution in the ini‐
→
tial moment of time the scalar potential ϕk (r ,0) has the following form:
∫
→→
→ 1 2 −ϑ 2/Δ̄ 2ϑ ik r
ϕk (r ,0) = 2 d ϑe e (19)
πΔ̄ ϑ
→
where ϑ is the angle between the packet wave vector and the wave vectork , Δ̄ 2ϑ is the aver‐
age value of the square of the angleϑ,Δ̄ 2ϑ ≪ 1.
The coefficients Cq→ of the Fourier expansion (2) for such initial packet have the following
form
∫ πΔ̄
2
d ϑ −ϑ 2/Δ̄ 2ϑ → →
Cq→ = (2π)3 e 2 δ(k − q ) (20)
ϑ
→ →
in which δ(k − q ) is the delta-function. As a result we come to the following expression for
the scalar potential
(kρ/2) 2Δ̄ 2ϑ
→ 1 ik (z−t )− 1+ikz Δ̄ 2 2
ϑ/ (21)
ϕk (r , t) = e
1 + ikzΔ̄ 2ϑ / 2
→ →
where z and ρ are the coordinates parallel and orthogonal tok .
Let us note that the considered case corresponds to the wave packet, which consists of the
plane waves the directions of the wave vectors of which have some small scatter around the
z axis. If the initial packet has some small variation in frequencyω, the formula (21) should
be averaged over this variation of the waves in frequency. Assuming for simplicity that the
distribution of the waves over frequency has the Gaussian form with the average value of
the square of the frequency variation Δ̄ 2ω we find that
(z −t ) 2Δ̄ 2ω (ωρ/2) 2Δ̄ 2ϑ
→ 1 iω(z−t )− 4 − 1+i ω z Δ̄ 2
ϑ 2 / (22)
ϕk (r , t) = e
1 + iωzΔ̄ 2ϑ / 2
The given expression for the wave packet has the same structure as the corresponding ex‐
pression for the packet (16). If the substitutions (t / a⊥2ε) → (ω zΔ̄ 2ϑ / 2) and a||2 → 4Δ̄ 2ω are made
in the latter expression the both formulae for the wave packet will become identical.
The formula (22) shows that for ω zΔ̄ 2ϑ / 2 ≪ 1
→ {
ϕk (r , t) ≈ exp iω(z − t) −
(z − t)2Δ̄ 2ω ωρ 2
4
−
2
Δ̄ 2ϑ ( ) } (23)
and for ω zΔ̄ 2ϑ / 2 ≫ 1
→
ϕk (r , t) ≈ −
2i
2
ω zΔ̄ ϑ
exp iω(z {
− t) −
(z − t)2Δ̄ 2ω
4
+ iω
ρ2
2z
ρ2
− 2 2
z Δ̄ ϑ
} (24)
For z ≫ ρ the latter formula can be written in the form of a diverging wave
→
ϕk (r , t) ≈ −
2i
2
ω rΔ̄ ϑ
exp iω(r − {
t) −
(z − t)2Δ̄ 2ω
4
ρ2
− 2 2
z Δ̄ ϑ
} (25)
wherer = z 2 + ρ 2 ≈ z + ρ 2 / 2z. Thus on distances z from the center of the initial packet, which
satisfy the condition
ω zΔ̄ 2ϑ / 2 ≪ 1 (26)
the form of the packet (22) coincides with the form of the packet att = 0. Only on the distan‐
ces, which satisfy the condition
ω zΔ̄ 2ϑ / 2 ≫ 1 (27)
the transformation of the packet (22) to the packet of spherical diverging waves occurs.
Let us note that in the theory of radiation of electromagnetic waves by a moving electron the
spatial region in which the formation of spherical diverging waves occurs has a name of the
wave zone (see for example [21]). In particular, for nonrelativistic charged particles the wave
zone begins on distances from the radiation region, which exceed the length of the radiated
waveλ. However, the condition (27) shows that for Δ̄ 2ϑ ≪ 1 the wave zone formation occurs
not on distances z ≫ λ as in the case of a nonrelativistic particle but on distances
z ≫ 2λ / Δ̄ 2ϑ (28)
which are much larger than the wave lengthλ = 1 / ω. For sufficiently small values of Δ̄ 2ϑ the
length z = 2λ / Δ̄ 2ϑ can reach macroscopic size.
2.3. Wave packets in the equivalent photon method
The problem of dispersion of wave packets naturally arises in the equivalent photon meth‐
od (or the method of virtual photons) in which at the certain moment of time (t = 0) the elec‐
tron’s coulomb field is substituted by a packet of free electromagnetic waves (see [22] and
references there). Let us consider some peculiarities of wave packets behavior in this meth‐
od [15].
For this purpose we write the scalar potential of the coulomb field of the electron moving
→
along the z axis with the velocity v in the form of the following Fourier-expansion:
→
ϕC (r , t) = Re ∫d 3k i(k→ r→ −k→ v→ t ) C
(2π)3
e Ck (29)
in which
8πeθ(kz )
CkC = (30)
k⊥2 + kz2 / v 2γ 2
Here eis the electron’s charge, γis the Lorentz-factor, kz and k⊥ are the components of vector
→
k parallel and orthogonal to z axis, θ(kz )is the Heaviside step function.
In the equivalent photon method it is assumed that for t = 0 the packet (2), which consists of
free electromagnetic waves coincides with the packet (29). It corresponds to the Fourier ex‐
pansion (2) with the coefficientsCκ→ = CkC . Let us consider the behavior of such packet in ultra
relativistic case in whichγ ≫ 1.
→
For γ ≫ 1 the main contribution to (2) are made by the values of κ with directions close to
→
the direction of the electron’s velocityv . Taking this fact into account we can present the
packet (2) in the following form:
∞
→ →
∫
ϕ(r , t) = Re dkϕk (r , t)
0
(31)
where
∞
→ 2
∫
ϕk (r , t) = e ik (z−t ) 2
π
0
ϑ
ϑdϑ
+ γ −2 J 0(kρϑ)e
−ikzϑ 2/2 (32)
→ →
Here ϑ is the angle between k and v (ϑ ≪ 1) and J 0(x) is the Bessel function.
→
The function ϕk (r , t) has the same structure as the function (21) corresponding to the Gaus‐
→
sian distribution of vectors k over the anglesϑ. Namely, ifkzϑ 2 / 2 ≪ 1, the main contribution
to the integral (32) is made by the values ϑ ≈ γ −1 and
→ 2
ϕk (r , t) ≈ K 0(kρ γ)e ik (z−t )
π / (33)
→
where K 0(x) is the modified Hankel function. In this case after integration over k in (31) we
find that
→ e
ϕ(r , t) = (34)
2 −2
ρ γ + (z − t)2
The main contribution to (31) is made by the valuesk ≈ γ / ρ, hence expression (34) is valid in
the range of coordinates ρ and z that satisfy the conditionz < γρ. In this range of coordinates
the packet under consideration moves with the velocity of light in the z axis direction.
So on distances z < 2γ 2λ the considered wave packet practically coincides with the initial
one (att = 0). Substantial transformation of the packet would happen only on distances
z > 2γ 2λ (35)
In this case for the evaluation of the integral in (32) over ϑ one could apply the method of
stationary phase. As a result of using of this method we find that
→ 2i 1 1 ik (r −t )
ϕk (r , t) = − e (36)
π θ 2 + γ −2 kr
0
where r ≈ z + ρ 2 / 2z and ϑ0 = ρ / z is the point of stationary phase of the integral (32). We see
that the components (36) of our packet have in the case under consideration the form of di‐
verging spherical waves. Under this condition the angle ϑ0 corresponds to the direction of ra‐
diation, and the function before the diverging wave describes the angular distribution of the
radiation. So, the condition (35) draws out the wave zone in application to given problem.
The value 2γ 2λ presenting in the condition (35) is known in the theory of radiation by ultra‐
relativistic particles as the formation length or the coherence length [1, 2].
3. The bremsstrahlung at an electron instantaneous scattering
3.1. The electromagnetic field structure at an electron instantaneous scattering. The ‘half-
bare’ electron
The electromagnetic wave packets similar to the ones considered above arise, for example,
in the processes of bremsstrahlung by relativistic electron at its instantaneous scattering to a
large angle and in the process of transition radiation during an electron traverse of thin met‐
allic plate in vacuum. The present section is dedicated to the analysis of evolution of electro‐
magnetic wave packets and peculiarities of formation of radiation by relativistic electron in
the wave and the pre wave zones in the first process, while the next one – to the analogous
questions concerning the second one.
→
Let a relativistic electron move along the z axis with the velocity v and at the moment of
→ →
time t = 0 in the point z = 0 change it abruptly from v to v ' (Figure 1).
Figure 1. The picture of total field after instantaneous scattering of an electron to a large angle.
Scalar and vector potentials of the total electromagnetic field, which takes place in such
process can be defined from inhomogeneous Maxwell equations
∂2 ϕ →
Δϕ − = − 4πρ(r , t),
∂t 2
→ (37)
→ ∂2 A → →
ΔA − = − 4π j (r , t)
∂t 2
→
in which ρ and j are respectively the charge and the current density of the particle, which in
the case of a single point particle motion should be taken as
→ → →
ρ(r , t) = eδ(r − r (t))
→ → → → → (38)
j (r , t) = ev (t)δ(r − r (t))
→ → →
where δ(r )is the delta-function and r (t)and v (t) are respectively the particle’s trajectory and
its velocity. The solution of the set of equations (37) has the physical sense if it has the form
of the retarded potentials. In our case of a point particle the required form of the solution of
(37) and (38) is the following [21]:
+∞
→
{ A, ϕ } = e ∫ | r→ −dtr→ (t' ') | {v→ (t '), 1}δ(t ' − t + | r→ − r→ (t ') | )
−∞
(39)
→
In the case of a uniform particle motion with the velocity v along the z axis (39) leads to the
following explicit expressions for the potentials:
→
→ → ev → e
Av (r , t) = , ϕv (r , t) = (40)
2 −2 2 2 −2
ρ γ + (z − vt) ρ γ + (z − vt)2
which are the coulomb potentials of the moving particle. Here γ = E / m is the particle’s Lor‐
→
entz-factor and ρ is the absolute value of the coordinate ρ of the observation point in the
→
plane orthogonal to the z axis. The formula (40) for ϕ(r , t)shows that the equipotential sur‐
faces of the field of the particle, which moves uniformly straightforward are the ellipsoids
flattened in the direction of particle motion.
In order to obtain the solution of (37) for the considered case of the particle instantaneous
scattering it is convenient to express the potentials in the form of Fourier-integrals. Let us
consider, for example, the vector potential:
∫
→ → →→
1 →→ ik r 3
A(r , t) = 3 Ak (t)e d k (41)
(2π)
In order to obtain the expansion in the form of the retarded potential we should calculate
→
the Fourier-component Ak→ (t) with the use of the inverse Fourier-transform of the expression
(39) for the vector potential:
∫
→ → → →→
Ak→ (t) = A(r , t)e −ik r d 3r =
+∞
→
∫ ∫
→→
v (t ') → →
= e dt ' d 3re −ik r | → → δ(t ' − t + | r − r (t ') | )
r − r (t ') |
−∞
→ → →
Making here the substitution R = r − r (t ') and integrating the expression over d 3R it is possi‐
ble finally to present the Fourier-expansion (41) of the vector potential in the following form:
∫
d 3k −i(kt −k→ r→ )
∫
→ → →→
e →
A(r , t) = − 2 Im k e dt 'v (t ')e i(kt '−k r (t ')) (42)
2π −∞
→ → →
Substituting into (42)v (t ') = v θ( − t ') + v 'θ(t '), where θ(x)is the Heaviside theta function, we
have for t < 0 (which is before the scattering moment):
→
→ →
A(r , t) =
e
2π
d 3k
∫
v → i(k→ r→ −k→ v→ t )
→
2 Re k ck − k v e
(43)
→
As integration over k may show, (43) is nothing else than the Fourier-expansion of the own
coulomb field (40) of a uniformly moving particle.
The different and rather interesting situation takes place after the scattering moment (for
t > 0). In this case from (42) we obtain:
∫ { }
→ →
→ → e d 3k v→' →→ →→
v→ −ikt ik→ r→
A(r , t) = 2 Re k
→ 1 − e −i(k −k v ')t e −ik v 't + →e e (44)
2π k −kv ' k −kv
→
The integration over k gives in this case:
→ → → → → →
A(r , t) = Av→ '(r , t)θ(t − r) + Av→ (r , t)θ(r − t) (45)
→ →
where Av→ and Av→ 'are respectively the coulomb potentials of the particles, which uniformly
move in the directions of z and z ' axes. The analogous expressions for the scalar potential
can be easily obtained by the same method.
The expression (43) shows that before the scattering moment the total field around the elec‐
→
tron coincides with its own coulomb field, which moves with the velocity v together with the
electron. After the scattering, according to (44) and (45), the total field around the electron
breaks into two parts represented by two items in the considered expressions.
The first item in braces in (44) corresponds to the nonequilibrium field, which the scattered
electron has already managed to rebuild around itself by the moment of timet. This field
consists of the own coulomb field of the electron, which moves along the z ' axis with the
→
velocity v ' (the first item in square brackets in (44) corresponds to it) and the packet of free
waves, which moves in the same direction (it is described by the second item in square
brackets). There is substantial interference between these fields due to which (as the first
item in (45) shows) the total field vanishes in the regionr > t, which the signal about the elec‐
tron scattering at the moment of time t = 0 has not yet reached. Inside the sphere of radius
r = t the total field coincides with the equilibrium coulomb field of the electron.
The second item in braces in (44) describes the field, which as though `tears away' from the
electron at the scattering moment. It is a packet of free electromagnetic waves, which moves
→
in the direction of the initial electron's velocity v and gradually transforms into bremsstrah‐
lung. This field is different from zero outside the sphere r = t and vanishes inside it.
The equipotential surfaces of the scalar potential of the field around the electron after its
scattering to a large angle are presented on Figure 1.
The behavior of certain Fourier-components of the nonequilibrium field of the electron after
its scattering and of the field ‘torn away’ from the electron at its scattering is of special inter‐
→
est. According to (44), the values ofk , which make contribution to each item of the nonequili‐
brium field of the electron has directions close to the direction of the electron’s final velocity
→ →→
v '. During the period of timet < (k − k v ')−1, however, both of these items substantially cancel
each other due to the interference. Therefore the coulomb field, which corresponds to the
→→
scattered electron, does not appear instantly. Namely, during the period of time t ≤ (k − k v ')−1
→
the Fourier components with wave vector k are nearly absent in the field around the scat‐
→
tered electron. As the main contribution to the coulomb field is made by k with directions
→
close to the direction of the velocity v ' this period of time is defined by the relationt ≤ 2γ 2 / k.
Figuratively speaking we can say that after the scattering during such period of time the
electron is in ‘half-bare’ state, which means without considerable part of its field. Such state
of electron with nonequilibrium field manifests itself during further scatterings of the elec‐
tron, which causes different effects of bremsstrahlung suppression during an electron's mo‐
tion in substance, such as Landau-Pomeranchuk-Migdal effect [23-25], the effect of radiation
suppression in thin layer of substance (TSF-effect [26-28]), etc. Recently the series of detailed
experimental investigations of these effects at ultra high energies was performed on SLAC
[29,30] and CERN[31,32] accelerators, which confirmed the main theoretical predictions.
The notion of a ‘half-bare’ electron was introduced in the papers of E.L. Feinberg [33,34]
who studied the time evolution of the state vector of the system ‘electron + photon’ after the
scattering of a fast electron to a large angle on atom. The classical theory of this effect was
given in [2, 27, 35].
Let us note that during the period of time t = 2γ 2 / ω the electron covers the distance
lC = 2γ 2v / ω (46)
In the theory of radiation by relativistic electrons the length lC has a name of the coherence
length of the radiation process [1,2]. Within this length the interference effects in radiation
are significant. For ultra relativistic electrons in the region of small frequencies of radiated
waves the length lC can have macroscopic size, which exceeds not only interatomic distance
of the substance but the size of the experimental facility as well. Indeed, in the millimeter
range of the waves radiated by electrons with the energy of 50Mev the coherence length is
lC = 20m. It means that within such length after scattering the electron is in ‘half-bare’ state,
which means that the Fourier-components of the corresponding wavelengths are absent in
the field around the electron.
The field, which ‘tears away’ from the electron at its scattering (the second item in braces in
(44)) has the structure similar to the one, which has the packet of free waves considered
above in the equivalent photons method. Therefore the main peculiarities of the reconstruc‐
tion of the ‘torn away’ field to the field of radiation will be the same as the considered above
peculiarities of the wave packets evolution. Let us consider this process in detail.
3.2. The problem of measurement of bremsstrahlung characteristics
The results presented above show that for ultra relativistic electrons the radiation formation
process develops on large distances along the initial and final directions of the electron mo‐
tion, which can be of macroscopic size. In this case a detector, which registers the radiation
characteristics can be situated both in the wave zone (which means on large distances r → ∞
from the scattering point) and in the pre wave zone (which means on within the coherence
length lC of the radiation process). Let us show that the results of measurements can sub‐
stantially differ in these cases. For this purpose let us consider the peculiarities of the forma‐
tion of radiation from the ‘torn-away’ field on different distances from the scattering point
[17]. The consideration will be made for a point detector situated in the point with coordi‐
→ → →
natesr = (ρ , z). Here the z axis is parallel to the vector of the electron’s initial velocity and ρ
are the coordinates in the orthogonal plane. In this case by the point detector we mean the
detector, which measures the characteristics of electromagnetic waves (wave packet), which
→ →
fall on a small element of surface situated in the point with the coordinates r = (ρ , z) and
seen at solid angle do from the scattering point.
Making in the second item in (44) the variable substitution kz → kby kz = k 2 − q 2(kz and qare
→
respectively the components of the wave vector k along the z axis and orthogonal to it) and
denotingk = ω, it is possible to present the Fourier-expansion of the `torn-away' field scalar
potential in the following form:
+∞ |ω |
→
ϕ(r , t) =
e
πv 2 −∞
∫
dω e −iω t ∫ dq q
0
2
qJ 0(qρ)
+ ω 2 / v 2γ 2
Q(z) (47)
where
ωcos( ω 2 − q 2z )
Q(z) = + ivsin( ω 2 − q 2z ) (48)
ω 2−q 2
→
and ρ = | ρ | .
In (47) and (48) the square root ω 2 − q 2 is considered to be a single-valued branch of the
analytical function, which is equal to | ω 2 − q 2 | for ω > q and − | ω 2 − q 2 | forω < − q.
In ultra relativistic case (γ ≫ 1) the range ofq, which make the main contribution to the inte‐
gral (47) is q ≤ ω / γ ≪ ω and it is possible to expand the square roots ω 2 − q 2 in (48) in the
small factorq / ω. Let us leave the items proportional to the second power of q / ωin the argu‐
ments of sine and cosine, while in the other parts of the expression (48) neglect them. More‐
over the integration over qcan be extended to the region0 < q < ∞. This leads to the following
expression for the `torn-away' field potential in ultra relativistic case:
+∞ +∞
→
ϕ(r , t) =
e
π ∫ ∫
−∞
dω dq 2
0
q
qJ 0(qρ)
+ ω 2
/ v 2 2e
γ
iω(z−t )−iq 2 z /2ω (49)
The equations (37) are presented in Lorentz gauge

→ ∂ϕ
div A + =0 (50)
∂t
→
If we knowϕ(r , t), we can derive the vector potential component Az from this equation
→
(A⊥ = 0in this case). In the considered case the vector potential equals to the scalar one up
to the neglected items of the order of(q / ω )2.
→ →
In order to determine the total energy radiated in the direction of a small area ds = n r 2do
→ →
situated in the point r = (ρ , z) we can calculate the flux of the Poynting vector through this
area during the whole time of the particle motion
dε =
r 2do
4π ∫ → → →
dt (E × H )n
→ → → →
where do is the element of solid angle in the direction of radiationn = r / r. The fields E and H
in this expression are the fields of the packet of free electromagnetic waves (the field ‘torn-
→
away’ from the electron at its scattering). They are related to scalar ϕ and vector A poten‐
tials of the radiation field by
→
→ ∂A → →
E=− − ∇ ϕ, H = ∇ × A (51)
∂t
→
Proceeding to the Fourier-expansions of the fields ϕ and A over frequency ω we obtain the
following expression for the radiation spectral-angular density:
dε r2 → → → → →
= E (r ) × H −ω (r ) n (52)
dωdo 4π 2 ω
in whichω ≥ 0.
→ →
With the use of the Maxwell equationH ω = ( − i / ω)∇ × E ω , the radiation spectral-angular
density can be expressed in the terms of Fourier-component of the electric field alone:
dε ir 2 → → → →
= E ω × (∇ × E −ω (r )) ⋅ n (53)
dωdo ω
Let us note that the formulae (52) and (53) are valid for arbitrary distances from the scatter‐
ing point. Therefore they can be used for radiation consideration both in the wave and the
pre wave zones. In the wave zone (which means in the regionr → ∞) they can be considera‐
bly simplified. Therefore, firstly, let us dwell on the consideration of this case.
On large distances from the scattering point (r → ∞) a Fourier-component of the field of the
→ →
radiation waves E ω (r ) reconstructs into a packet of diverging waves, which amplitude is
proportional to the factorr −1exp(iωr). The action of the Hamilton operator ∇ in (53) upon
→
E −ω in this case can be applied only to the factorexp(iωr):
→ → →
∇ × E −ω = − iω n × E −ω (54)
As a result we obtain the following expression for the radiation spectral-angular density for
r → ∞:
dε r2 → → 2
dωdo 4π 2 | ω |
= E (r ) (55)
Let us note that the formula (55) is valid on large distances from the scattering point (r → ∞).
→→
While proceeding from (53) to (55) we took into account thatn E ω = 0. This relation directly
comes from the expression of Fourier-components of electric and magnetic fields in terms of
→ → →
potentials ϕω (r ) and Aω (r )
→ → → → → → →
E ω = − ∇ ϕω (r ) + iω Aω (r ), H ω = ∇ × Aω (r ) (56)
→ → →
where ϕω (r ) and Aω (r ) are defined from the expression (49).
In the case of ultrarelativistic particle in the region of characteristic for this process small ra‐
diation angles ϑ ≪ 1 the ‘torn-away’ electric field can be considered as transverse having on‐
ly E⊥ component orthogonal to z axis. In this case the formula (55) for the radiation spectral-
angular density has the following form:
dε r2 →
dωdo 4π 2 | ω⊥ |
= E (r ) 2 (57)
Using (49) we can derive the electric field Fourier-component orthogonal to z axis. Substitut‐
ing it into (57) for the spectral-angular distribution of bremsstrahlung we achieve:
( ) ∫ dq q| | 2
∞
dε ez 2 q 2 J 1(qρ) −i
q 2z
= e 2ω (58)
dωdo π
0
2
+ ω 2 / v 2γ 2
For large distances from the scattering point, namely in the wave zone of the radiation proc‐
ess (z ≫ 2γ 2 / ω) the integral in (58) can be calculated with the use of stationary phase meth‐
od [36]. It leads to the well known expression for radiation distribution from the
bremsstrahlung theory [1,4,37]:
dε e2 ϑ2
= (59)
dωdo π 2 (ϑ 2 + γ −2)2
where ϑ = ρ / zis the angle between the direction of radiation and the z axis. As we can see
from (59) in the wave zone the radiation is mainly concentrated within characteristic angles
ϑ ≈ 1 / γ. The schematic form of the bremsstrahlung angular distribution in the wave zone is
presented on the Figure 2 by the solid curve.
Figure 2. Bremsstrahlung angular distribution given by a point detector in the wave zone (solid curve) and in the pre-
wave zone (dashed curve) for a certain frequency ω
For ultra high energies of the radiating particle in the region of characteristic small angles of
radiation the expression (57) (and hence the expression (58)) is valid for the description of
radiation spectral-angular density on small distances from the scattering point as well (in
particular, in the pre-wave zone (z ≪ 2γ 2 / ω) of the radiation process). Indeed, as the elec‐
→ →
tric field can be considered transverse in this case thanE ω = − ∇ ϕω = − ∂ ϕω / ∂ ρ . The equality
→ →
of ϕω (r ) and Aω (r ) leads to the following expression for the magnetic field:
→ → →
H ω = ∇ × Aω = ∇ ϕω × e z
→
where e z is a unit vector in the direction of z axis. Taking into account the orthogonality of
→ →
∇ ϕω and e z for the absolute value of the magnetic field we obtain | H ω | = | ∇ ϕω | , which is
equal to the absolute value of the electric field. Moreover, these fields are orthogonal
→ → → → →
(E ω H ω = ∇ ϕω (∇ ϕω × e z ) = 0). Taking into account the fact that the magnetic field H (r , t) is
→ →
the real function, which leads to the relationH −ω = H *ω , we can present the general expression
(52) in the following form:
dε r2 → → →
= E E * (r ) e z n
dωdo 4π 2 ω⊥ ω⊥
→ →
which coincides with (57) at small angles between e z andn . Now let us consider radiation in
the pre-wave zone on the basis of this formula.
In the pre-wave zone (z ≪ 2γ 2 / ω) of the radiation process it is not possible to use the sta‐
tionary phase method for the analysis of radiation characteristics. Here, making the substitu‐
tions q = ω x / γandρ = zϑ, we can present the integral (58) in the form:
dε
dωdo
= ( ) |I −I |
eω z
πγ
2
1 2
2 (60)
where
+∞
∫
ωz
−i x2
I 1 = dx J 1(ω zγ −1 xϑ)e 2γ 2
0
+∞
∫ J 1(ω zγ −1 xϑ) ωz
−i 2γ 2
x2
I 2 = dx 2 e
0
x +1
In the case γ ≫ 1 the absolute value of the integral I 2 is negligibly small comparing to the
corresponding value of I 1 and for spectral-angular density of bremsstrahlung in the pre-
wave zone we obtain:
dε
dωdo
=
eω z
π
( ) | I | = 4eπ
2
1
2
2
2
ϑ
1
2 sin
(
2 ω zϑ
4
2
) (61)
From (61) we can conclude that in the pre-wave zone the radiation is mainly concentrated
within anglesϑ ≈ 2 / ω z, which exceed the characteristic angles ϑ ≈ 1 / γof the wave zone.
Therefore in the pre-wave zone (z ≪ 2γ 2 / ω) the point detector gives broader angular distri‐
bution of radiation (dashed curve on Figure 2) than in the wave zone (z ≫ 2γ 2 / ω). More‐
over this distribution depends on the frequency ω of the radiated waves.
By the point detector we mean here the detector of the smaller size δρthan the transversal
radiation length of the processlT ≈ γ / ω, which is the characteristic transversal distance on
which at the moment of time t = 0the Fourier harmonics of frequency ω are concentrated in
the wave packet (49). Such detector registers the radiation of frequencyω, which falls on a
small domain of space, where the detector is situated.
The measurements, however, can be made by the extended detector of the larger size than
the characteristic transversal length of the radiation process, so thatδρ ≫ lT . Such detector
registers not only the waves of frequencyω, which fall on the small element of surface with
→
coordinates ρ andz, as the point detector does, but all the electromagnetic waves of frequen‐
→ →
cyω, which propagate in the direction of wave vector k ( | k | = ω). In order to calculate the
bremsstrahlung spectral-angular distribution, which is registered by an extensive detector,
which is a plate of large size, we need to integrate the expression (58) over the entire consid‐
ered plate and express the obtained result in the form of an integral over the directions of
wave vectors of radiated waves. The integrand in this case will be nothing else than the re‐
quired distribution. In our case after performing the procedures described above we can
present the expression (58) in the following form:
dε e2 ϑγ 2
= 2 (62)
dωdoγ π (ϑ 2 + γ −2)2
γ
→
where ϑγ = q / ω is the angle between the direction of the wave vector k and the z axis. Hence
the bremsstrahlung spectral-angular distribution obtained by the extended detector coin‐
cides with the one (59) obtained by the point detector in the wave zone. But unlike the case
with point detector this distribution does not depend on the distance from the scattering
point and is the same both in the wave and the pre-wave zones.
4. Transition radiation by relativistic electron on thin metallic plate
4.1. Scalar and vector potentials of transition radiation field
The electromagnetic wave packets of the structure analogous to the one considered in the
process of an electron instantaneous scattering take place also in the process of relativistic
electron traverse of thin conducting plate. Let us consider a problem about transition radia‐
tion that arises during normal traverse of thin ideally conducting plate, situated in the plane
z = 0, by an electron, which moves along the z axis from z = − ∞to z = + ∞ (Figure 3). Let us
investigate the structure of electromagnetic fields that take place before and after the elec‐
tron's traverse of the plate in vacuum [16-18].
Figure 3. The electron normal traverse of thin metallic plate.

Scalar and vector potentials of the electromagnetic field, which is generated by an electron
moving in vacuum are the solutions of inhomogeneous Maxwell equations (37). The equa‐
tions set (37) in the considered problem should be supplemented by a boundary condition,
which corresponds to the fact that on the plate's surface the tangential component of the to‐
tal electric field equals zero. The general solutions of the equations (37) for the electron,
→
which moves uniformly with the velocity v can be represented in the form of the following
→ → →
Fourier-expansions of the potentials ϕ(r , t)andA(r , t):
→
∫
ϕ(r , t) =
d 3kdω i(k→ r→ −ω t ) →C
(2π) 4 e
⇀→
ϕk ,ω δ(ω − k v ) + ϕk→ f,ω δ(k 2 − ω 2)
(63)
A(r , t) = ∫
3 →→
→ → d kdω i(k r −ω t )
→ ⇀→ →
e 4 AkC→ ,ω δ(ω − k v ) + Ak→f ,ω δ(k 2 − ω 2)
(2π)
The first items in (63) are the Fourier-expansion of the electron's Coulomb field, for which
8π 2e → C→ →
ϕk→C,ω = − , Ak ,ω = v ϕk→C,ω (64)
ω 2−k 2
The second items in (63) are the Fourier-expansion of the field of induced surface currents
→
on the plate (we will name it the free field), for which in vacuum | k | = | ω | . On large dis‐
tances from the region in which the transformation of the surrounding electron field takes
place this items form the transition radiation field. It can be derived from the boundary con‐
→ → →
dition for the total electric field E = E C + E f on the surface of the plate:
→ → → →
E C⊥(ρ , z = 0,t) + E ⊥f (ρ , z = 0,t) = 0 (65)
→ →
Here E C is the electron's Coulomb field and E f is the field of induced surface currents on the
plate. Moreover it is required that the free field produced by the plate propagates on the left
and on the right of the plate respectively in the negative and positive directions of z axis. The
scalar and the vector potentials define electric and magnetic fields by the relation (51). Due to
→
the symmetry of the problem the vector potential is directed along the particle velocityv :
→ → → →
A(r , t) = v A(r , t) (66)
and the relation
→ →
A C (r , t) = ϕ C (r , t) (67)
for the particle's field in vacuum is valid. In general case such relation between potentials is
not valid.
→
Using Fourier expansion (63) it is possible to derive the potential ϕ(r , t) from the condition
(65). For this let us perform the integration over the component kz in (63). Taking also into
account the relation between potentials and fields (51) we obtain the following expression
for transversal component of the electric field:
∫ |
2 →
→ → ∂ d k⊥dω i(k ⊥ρ→ +kz z−ω t ) →C
| (ϕk ,ω
1
E ⊥(r , t) = − → e ϕk ,ω + →f
| +
∂ ρ (2π)4
kz =
ω
v
2 | 2
ω − k⊥2 kz = ω 2−k⊥2
(68)
+ϕk→ f,ω |
kz =− ω 2−k⊥2
) .
Forz = 0, according to (65) this component should equal zero. From this we find that
− ϕk→C,ω | kz =ω /v
=
2 |
1
ω 2 − k⊥2 | (ϕk ,ω
→f
|
kz = ω 2−k⊥2
+ ϕk→ f,ω |
kz =− ω 2−k⊥2
) (69)
The values kz = ± ω 2 − k⊥2 satisfy the dispersion relationω 2 = kz2 + k⊥2, which is defined by the
respective δ- function in (63). The sign before the square root ω 2 − k⊥2determines the direc‐
tion of propagation of plane waves (Fourier components) with given values ofωand | k⊥ | [1,
4]. Indeed, the equation of a plane wave constant phase along the zaxis iskz z − ω t = const. The
plane waves, which the free field produced by the plate consists of, should propagate away
from the plate. Hence for ω > 0 on the right of the plate (which meansz > 0) it is necessary to
take into account only Fourier-components with positive sign before the root ω 2 − k⊥2 in
(62), while for ω < 0 and z > 0- only Fourier-components with negative sign before this root in
(68). In the region z < 0 for ω > 0 and ω < 0 in (68) we should take into account the items with
opposite signs before the root ω 2 − k⊥2 relatively to the case forz > 0. The value of the square
root itself is considered either positive or to belong to the upper complex half plane.
Thus, taking into account all requirements mentioned above we can write the scalar poten‐
tial of the free field in the following form:
∞
i ( zω )
∫ ∫
→ →
→ e 1 1−k⊥2/ω 2−ω t +k ⊥ρ
f
ϕ (r , t) = − d 2k⊥ dω 2 2e
(70)
2π 2v −∞
k ⊥ + ω 2
/ p
where p = vγ (γ– electron's Lorentz-factor). It is a packet of free electromagnetic waves,

which gradually turns into the field of transition radiation in such way that each harmonic
with frequency ω reconstructs into diverging spherical wave on distancez > lC ≈ 2γ 2 / ω,
which is the formation length of the radiation process.
Let us note that the value of | k⊥ | in (63) is arbitrary. Therefore it is necessary to perform
the integration in (63) not only over travelling wavesk⊥2 < ω 2, but over surface ones k⊥2 > ω 2 as
well.
Using (70) and the Lorentz gauge (50) in which the equations (37) are presented we can de‐
rive the vector potential:
∞
i ( zω )
∫ ∫
→ →
→ e 1 1 1−k⊥2/ω 2−ω t +k ⊥ρ
A f (r , t) = − d 2k⊥ dω 2 e (71)
2π 2 −∞
k ⊥ + ω 2
/ p 2
1 − k⊥ / ω
2 2
Making in (70) the substitution | k⊥ | = | ω | x and separating the contributions to the poten‐
tial by the free field of travelling and surface waves, we can write the potential of this field
in the following form:
→ → →
ϕ f (r , t) = Ф1(r , t) + Ф2(r , t) (72)
where
1 ∞
→
Ф1(r , t) = −
2e
∫ xdx
πv x 2 + p −2
0 0
∫
dω J 0(ωxρ)cos ω ( | z | 1 − x 2 − t ) (73)
∞ ∞
→
Ф2(r , t) = −
2e
∫
xdx
πv x 2 + p −2
1 0
∫ | |
dω J 0(ωxρ)cos(ωt)e − z ω x 2−1 (74)
Deriving (72), we performed in (70) integration over azimuth angle between k⊥ and ρand
proceeded from integration over ω along the interval − ∞ < ω < + ∞ to integration over only
positive values of this variable.
The corresponding expressions for vector potentialA f = A1 + A2, according to (71), differ
from (73) and (74) only by additional factor 1 / 1 − x 2 in the integrands.
4.2. The structure of transition radiation field
Let us discuss the structure of the fields that arise during the electron traverse of thin ideally
conducting plate. Firstly, let us consider the structure of this field along the z axis for ρ = 0
[16]. This case is interesting for the fact that the calculation of all the integrals in (73) and (74)
essentially simplifies forρ = 0. Nevertheless, in this case all the main peculiarities of free
waves formation in the considered process remain intact.
As a result of rather simple calculations (see Appendix) we obtain the following expression
for the field produced by the plate on the z axis forρ = 0:
e e
ϕ f (z, t) = − | | | θ( | z | − t) − | | | θ(t − | z | ) (75)
z − vt | z + vt |
The total field produced by the electron and the plate can be obtained by addition of the ex‐
pression (75) and the electron’s own coulomb field on the z axis:
ϕ C (z, t) = e / | z − vt |
The obtained results show that for t < 0the total field produced by the electron and the plate in
the region z < 0 is the electron’s own coulomb field, which moves towards the plate and the
field, which coincides with the field of electron’s image inside the plate. By the field of elec‐
tron's image we assume the field, which is created by an imaginary particle with a charge of
the opposite sign, which is situated on the opposite side of the plate and moves symmetrical‐
ly to the electron relatively to the plate. In the region z > 0the total field equals zero fort < 0.
For t > 0 the picture of the total field distribution is the following. In the region z < 0 for
| z | > t the total field is defined by the electron’s field in this coordinate region (the electron
in this case is situated on the right of the plate) and the field of its ‘image’, which moves in
the direction opposite to the z axis. In the region z < 0 for | z | < t the total field equals zero.
In the region of positive z values for z > t the total field equals zero while for z < t this field is
the sum of the field of the electron, which is situated in this coordinate region for t > 0 and
the field of its ‘image’ situated on the left of the plate.
The integrals in (73) and (74) can be analytically calculated as well. After rather long calcula‐
→
tions we finally obtain the following expression forϕ f (r , t):
→ e e
ϕ f (r , t) = − θ(r − t) − θ(t − r) (76)
2 −2 2 2 −2
ρ γ + ( | z | − vt) ρ γ + ( | z | + vt)2
The electron’s own coulomb field has the following form:
→ e
ϕ C (r , t) = (77)
ρ 2γ −2 + (z − vt)2
The structure of the expressions for scalar potential is the same as the structure of these ex‐
pressions forρ = 0. If t < 0 then, according to (76) and (77), the total field in the left half-space is
equal to the sum of the electron’s coulomb field in this region and the field of its image:
→ e e
ϕ(r , t) = −
2 −2 2 2 −2
ρ γ + (z − vt) ρ γ + ( | z | − vt)2
In the right half-space the total field equals zero fort < 0.
After electron's traverse of the plate, which means fort > 0, according to (76) and (77), the to‐
tal field in the left half-space is defined by the formula
→ e e
ϕ(r , t) = − θ(r − t) (78)
ρ 2γ −2 + (z − vt)2 ρ 2γ −2 + ( | z | − vt)2
In the right half-space, where the electron is situated after the traverse of the plate, the total
field has the following form:
→ e e
ϕ(r , t) = − θ(t − r) (79)
2 −2 2 2 −2
ρ γ + (z − vt) ρ γ + (z + vt)2
Thus for t > 0 the picture of the total field, which is created by the electron-plate system is as
following. In the left half-space in the coordinate region r > t it is a sum of the electron's cou‐
lomb field of the opposite sign reflected from the plate and the own field of electron, which is
→
situated on the right of the plate. The reflected field in this case moves with velocity − v in the
direction opposite to the direction of electron's motion. In the coordinate regionr < t, which the
signal about the electron's traverse of the plate at t = 0 has already reached, the total field equals
zero. In the right half-space for t > 0 and r < t the total field equals the sum of the fields of the
electron and its ìmage' on the left of the plate. For r > t the total field equals zero.
The analogous expressions can be obtained for vector potential as well. Namely, for t > 0 the
total field vector potential (the sum of particle's coulomb field and radiation potentials) has
the following form:
→ → → e e
A(r , t) = v + θ(r − t) (80)
2 −2 2 2 −2
ρ γ + (z − vt) ρ γ + ( | z | − vt)2
The expression in square brackets in (80) differs from the same expression for scalar poten‐
tial (78) only by the sign of second item. The reason of this can be understood from the fol‐
lowing reasoning. For z < 0 in the region r > t the field (78) is the difference between two
coulomb fields, the sources of which are the electron and its image. The vector potential of
→ → →
the coulomb field is related to its scalar potential byA = v ϕ, where v is the velocity of the
→ →
field source. As electron moves with velocity v and its image - with velocity − v , their vector
→ → → →
potentials respectively equal Ae = v ϕe andAi = − v ϕi . It is the presence of the `minus' sign in
→
the expression for Ai that causes the discussed difference in the signs of the items in square
brackets in (78) and (80).
In the region z > 0 for t > 0 the vector potential has the following form:
→ → → e e
A(r , t) = v + θ(t − r) (81)
2 −2 2 2 −2
ρ γ + (z − vt) ρ γ + (z + vt)2
The obtained results are valid for arbitrary electron velocities. The case of an ultra relativis‐
tic particle is of special interest because for such particles the reconstruction of the total field,
created by the plate and the electron after its traverse of the plate, into the field of radiation
occurs on large distances. The results obtained in this case are illustrated by Figure 4. Here
the equipotential surfaces of the scalar potential of the field reflected to the left half-space
and the field around the electron on the right of the plate are presented fort > 0.
Figure 4. The total field in ultra relativistic case fort > 0.
In order to understand what occurs with the electric field on the surface of the sphere of ra‐
dius r = t with the center in the point of the electron traverse of the plate (θ - sphere) let us
consider the structure of the force lines of the total field for t > 0 and, for example,z < 0.
While building the force lines it is necessary to take into account that they should originate
or end either on charges or in the infinity and not to cross each other. In the case of an infin‐
ite plate all the field lines originate and end either on the surface charges of the plate or on
the electron, which traverses it.
Thus each force line of the total field (78), which originates on a surface charge of the plate
somewhere in the area ρ > t and stretches through the space region r > t to the θ - sphere, should
be refracted and stretch further along the surface of the sphere, ending on another surface
charge of the plate at ρ = t(Figure 5). It is the force lines, which overlap each other on the θ -
sphere are the force lines of the transition radiation field. Indeed, the field on the θ - sphere
propagates in the radial direction with the speed of light and is perpendicular to this direction.
Moreover, as will be shown in further discussion, this field decreases with the distance as1 / r.
It is necessary to note that the given picture of the force lines indicates the necessity of the
existence of the field (78) outside the θ - sphere along with the radiation field on it. It is only
in this case that the force lines of the radiation field, which originate (end) on the surface
charges of the plate and stretch along the θ - sphere can proceed to the region r > t forming
the lines of the field (78) and not intersect or break on the θ - sphere.
Figure 5. The picture of the total field force lines fort > 0,z < 0.
4.3. The transition radiation field and its intensity
In ultra relativistic case (γ ≫ 1) the range ofk⊥, which make the main contribution to the in‐
tegral (70), which represents the transition radiation field is k⊥ ≤ ω / γ ≪ ω and it is possible
to make in (70) the same expansion of the square root ω 2 − k⊥2in the small factor k⊥ / ω as we
did in (47) and (48) for the bremsstrahlung field. Such expansion represents the fact that
transition radiation is considered at small angles around the particle’s trajectory
ϑ = k⊥ / ω ≈ 1 / γ within which the most part of the radiation is concentrated in ultra relativis‐
tic case. In the result of the expansion of (70) we obtain the expression for the transition radi‐
ation field, which coincides with the analogous expression (49) for the bremsstrahlung field.
This means that the packets of free electromagnetic waves, which arise in the processes of
electron instantaneous scattering and the particle traverse of thin metallic plate have similar
(but not totally identical) structure and the radiation effects, which take place in these proc‐
esses should be analogous.
Let us consider the radiation, which arises during an electron normal traverse of thin ideal‐
ly conducting plate [16-18]. In this case by radiation we mean the part of the electromagnet‐
ic energy, which belongs to the frequency interval (ω, ω + dω) and falls into the small detector,
which is placed on different distances r from the point of electron traverse of the plate. The
→
detector’s position relative to this point is defined by the transverse ρ and longitudinal z co‐
ordinates (Figure 4). The radiation direction ϑ is then defined by the ratio of these coordinates:
ρ
ϑ = arctg | z | (82)
In further discussion we will mainly concentrate our attention on the peculiarities of the ra‐
diation formation process in the left half-space, which is the region of negative z. Naturally,
the general expression (55) for the radiation spectral-angular density on large distances from
the target is valid in this case as well. The discussed structural similarity of the fields in the
processes of electron instantaneous scattering and electron traverse of thin metallic plate al‐
so gives us the possibility to use the expression (57) for the description of the transition radi‐
ation field in the range of small angles ϑ ≈ 1 / γ characteristic to this process on arbitrary
distances z from the plate. Firstly, let us consider the transition radiation in the wave zone.
Let us show that in the region of large distances from the target (r → ∞) the formula (55)
gives the well known result from the theory of transition radiation by electron on metallic
plate [1, 4, 37]. Indeed, in differentiating of (78) and (80) the items proportional to deriva‐
tives of the square brackets will give the total field outside the θ- sphere, which decreases
with the distance r faster thanr −1. The item proportional to the derivative of the θ- function
is on large distances the diverging wave, which amplitude is proportional tor −1. This item
defines all the transition radiation characteristics.
f
→
The scalar ϕ and vector A f potentials on the θ- sphere can be written as
e cosϑ → → ϕ
ϕ = − 2v θ(r − t), A = − v (83)
r 1 − v 2cos2ϑ cosϑ
→
where ϑ is the angle between r and − v . So the transition radiation electric field is totally de‐
fined by the spatial derivative of the scalar potential:
( )
→
→ → e δ(r − t) v →
E f (r , ϑ, t) = 2v + n cosϑ (84)
r 1 − v cos ϑ v
2 2
The Fourier-component of this expression has the following form:
( )
→
→ → e 1 v →
E fω(r , ϑ) = 2v + n cosϑ e iω r (85)
r 1 − v 2cos2ϑ v
→→ →→
Let us note that the relation n E ω = 0 is valid for (81) asn v = − vcosϑ.
Substituting (85) into (55) we obtain the well known expression for spectral-angular density
of backward transition radiation [1, 4, 5]:
dε e 2v 2 sin2ϑ
= 2 (86)
dωdo π (1 − v 2cos2ϑ )2
The presented derivation of the formula (86) for the transition radiation spectral-angular
density is based on the analysis of the electron’s field reflected from the plate on large dis‐
tances r → ∞ from the plate. In this case by large distances we mean the distances from the
target, which substantially exceed the coherence length of the radiation process [1, 2]
ω −1
lC (ϑ) = (87)
1 − vcosϑ
For ultrarelativistic electrons the transition radiation is mainly concentrated inside a narrow
cone with the opening angleϑ ≈ 1 / γ. In this region of angles ϑ the coherence length (87)
quickly increases with the increase of the electron’s Lotentz-factor:
2γ 2 / ω
lC (ϑ) = (88)
1 + γ 2ϑ 2
For large γ and small ω this length can have ultra large values, which allow, for example, a
detector to be situated within this length. The analysis of the transition radiation process in
this case requires the development of the methods, which allow to consider the radiation
process not only on large distances from the target comparing to the coherence length, lC ,
but on the small ones as well. One of such methods is based on the application in the consid‐
→
ered problem the Fourier-expansions of scalar, ϕ f , and vector, A f , potentials of the free
field, which is generated by the plate during its traverse by the electron. Primarily, it is nec‐
essary to note that these fields are the wave packets, which consist of plane electromagnetic
waves. At large values of the time interval after the interaction of the electron with the target
these wave packets reconstruct into spherical diverging waves of radiation. On small distan‐
ces from the target the spatial structure of these packets is close to the structure of the packet
formed by the electron’s own field, which is reflected from the plate.
Firstly, let us consider on the base of this method the transition radiation on large distances
from the plate | z | ≫ 2γ 2 / ω. In this case the integrals over xin (72) can be calculated with the
use of the stationary phase method. Using for this purpose the asymptotic expression for the
→
Bessel function J 0(ωxρ) for large values of its argument we can write ϕωf (r ) in the next way:
1
→ 2e
ϕωf (r ) = − Re 2∫ xdx 2
−2 πωρ x e
{
i (ωρ x − 4 +| z |ω
π
)
1−x 2
+e
(
i ωρ x −
π
4
)}
−| z |ω 1−x 2 (89)
v x +p
0
The first item in the braces gives the stationary phase point
ρ / |z|
x0 = = sinϑ (90)
ρ2/z2 + 1
where the ratio ρ / | z | is defined by the relation (82). As stationary phase point is absent in
the second item in (89) in the considered region of the variablex, we can neglect the contri‐
bution of this item to the integral overx. For the same reason we neglect the contribution to
→
ϕωf (r ) of the item in (72), which contains the integration over the valuesx > 1. In the result for
→
ϕωf (r ) we obtain:
→ 2e cos2ϑ
ϕωf (r ) = − e i(ω r −π/2) (91)
ω vr sin2ϑ + p −2
We took into account the fact that in spherical coordinate system ρ = rsinϑ and | z | = rcosϑ.
→ → ⇀
With the use of (56) the expression (E ω × H −ω )n in (52) can be written in the form:
→ → ⇀ → →
(E ω × H −ω )n = ( − ∂ z ϕωf + iω Afω)n ⊥ ⋅ ∇⊥ A−ωf − cos∇⊥ ϕωf ⋅ ∇⊥ A−ωf (92)
In the point of stationary phase for r → ∞ Aωf = − ϕωf / cosϑ. Taking into account
∂ ∂
ϕ = iωsinϑϕω and ϕ = iωcosϑϕω we find that
∂ρ ω ∂z ω
→ → ⇀
(E ω × H −ω )n = ω 2t g 2ϑϕωf ϕ−ωf (93)
Substituting the asymptotic (91) for the potential ϕω into this expression we obtain the for‐
mula (86) for the radiation spectral-angular density.
For ultra relativistic particles characteristic values of the radiation angles ϑ ≈ γ −1 are much
less than unit. The radiation spectral-angular density (86) in this case has the following form:
dε e2 ϑ2
= 2 −2 (94)
dωdo π (γ + ϑ 2)2
In this region of radiation angles the characteristic values of the variablex, which make con‐
tribution to the integral over x in (72) are smallxeff ≈ γ −1. The stationary phase method is val‐
id for calculation of this integral if
In other words, it is required that the distance between the target and the detector should
substantially exceed the coherence length of the radiation process, which means the radia‐
tion should be considered in the wave zone.
As it was pointed out, the length lC can have macroscopic values and the detector can be
situated in the pre-wave zone, which is on smaller distances that the coherence length of the
radiation process. The stationary phase method is not applicable for calculation of the inte‐
gral over x in (72) in this case. For this purpose the expansion of the phase over small values
of x in the exponential factor in (72) can be made. Neglecting the items proportional to high‐
er powers than x 2 we find that
ϕωf =−
2e
πv
Re 2
0
x
∫
xdx
+ p −2 J 0(ωxρ)e
i(ω | z |−ω | z | x 2/2) (95)
In the region of small radiation angles for whichϑ 2 | z | ω ≤ 1, the convergence of the integral
(95) is determined by the valuesx ≈ 1 / | z | ω, therefore the items proportional to x 2 should
be preserved in the exponent phase. With the same accuracy in the region of small angles
the vector potential Aωf is related ϕωf by
→ → ⇀
In this case (E ω × H −ω )n = ∇⊥ ϕωf ⋅ ∇⊥ ϕ−ωf and the radiation spectral-angular density has the
following form:
dε
dωdo
=
πv
( ) | B(z) |
2e 2 2
2 (96)
where
B(z) = ω | z | x 2dx∫
0
J 1(xω | z | ϑ)
x 2 + γ −2
e −iω
| z | x 2/2 (97)
Let us note that during the derivation of the expression (97) we only took into account the fact
that the consideration of the radiation process was made in the region of small angles of radi‐
ation. Therefore the formula (97) is valid both for large (z ≫ lC ) and small (z ≪ lC ) distances
between the detector and the target. If | z | ≫ lC , the calculation of the integral B(z) can be
made with the use of the stationary phase method. As a result formula (97) gives the corre‐
sponding result of the transition radiation theory (see formula (94)). In the region | z | < lC the
formula (96) leads to the broader radiation angular distribution than the one defined by the
expression (94). The possibility of existence of such effect was considered in the paper [10]. Its
theory was elaborated in the paper [8]. According to it, in particular, in the region of small an‐
gles on distances | z | < lC the radiation spectral-angular density is defined by the formula:
dε
=
4e 2 1
dωdo π 2 ϑ 2
sin2
4
(
ω |z |ϑ 2
) (98)
Such modification of the radiation angular distribution for | z | < lC comparing to the case of
| z | ≫ lC is caused by the fact that in the region | z | < lC only part of the waves reflected
from the plate fall into the point detector (see [8]). It is necessary to note that for | z | < lC the
radiation angular distribution depends on the frequency of the radiated electromagnetic
wave. If the measurements are performed by the detector of the larger size than the charac‐
teristic transversal length of the radiation process the radiation spectral-angular density is
defined by the expression (94) on arbitrary distances from the plate and the discussed above
pre-wave zone effects disappear.
The obtained results show that the effects analogous to the ones which take place in the
process of bremsstrahlung at the instantaneous scattering of the electron to a large angle (the
broadening of the radiation angular distribution and its dependence on the frequency of the
registered photon in the measurements performed by a point detector on small distances
from the scattering point) take place also for backward transition radiation in the process of
an electron traverse of metallic plate. For ultra relativistic particles, according to (94) and
(98), the radiation is mainly concentrated in the region of small anglesϑ ≪ 1: ϑeff ≈ 1 / γin the
wave zone, in the pre-wave zoneϑeff ≈ 2 / ω | z | .
5. Transition radiation by ‘half-bare’ electron
The wave packets, which arise at ultra relativistic electron instantaneous scattering to a
large angle, reconstruct into radiation field on distances along the initial and final direc‐
tions of the electron’s velocity, which are of the order of the coherence length of the radia‐
tion process. For large energies of the electron and low frequencies of the radiated waves,
as was stated above, this length can be of macroscopic size. In this case the possibility of
investigating of the evolution of such wave packets in space and time by macroscopic de‐
vices appears. In [15] one of such possibilities, which concerns the reflection of wave pack‐
ets from an ideally conducting plate situated on different distances from the scattering point,
was discussed. In this case if the plate is situated in the direction of motion of the scat‐
tered electron perpendicular to its velocity the reflected field is the backward transition ra‐
diation. However, the peculiarity of this process lies in the fact that unlike the ordinary
backward transition radiation the considered one is the radiation by the particle with non‐
eqiulibrium field. The plate in this case can be considered as an element of the radiation
detector. Let us obtain the formulae, which describe the given process and discuss some of
its peculiarities on their basis [17].
Figure 6. Normal incidence of the scattered ‘half-bare’ electron on thin metallic plate.
Let the ideally conducting plate be situated in the plane z ' = z'0 (Figure 6). The Fourier-ex‐
pansion of the field around the scattered electron
→ → → →→
→
ϕ(r , t) =
e
2π 2 Re k∫
d 3k e ik (r→−v 't ) e ik r −ikt
→ −
k − k v ' ck − k v '
→→ (99)
consists of two parts, the first of which describes the equilibrium coulomb field of the elec‐
→
tron, which moves with the velocity v ' along the direction of scattering, while the second part
is the nonequilibrium field, which is structurally equal to the ‘torn away’ field (it is equal to
equilibrium coulomb field outside the θ- sphere and vanishes inside it). Hence, the second
→ →
part of the field (99) can be presented in the form (47) with a mere substitutionv → v '. The first
part of the field (99) can be presented in the analogous form by making the substitution kz → k
from k = kz2 + q 2 and denotingkz v ' = ω. From the expression for scalar potential obtained by
the considered transformations we can derive the expression for the Fourier-component of the
electric field perpendicular to z axis, which in ultra relativistic case is:
+∞ ∞
∫ ∫ q 2 J 1(qρ) ω z'
→ i
ω z'
−i (γ −2+q 2v'2/ω 2)
E⊥(r , ω) = 2e dω e v ' dq 1−e 2v ' 2 (100)
−∞ 0
2
q +ω 2
/ v' γ
2 2
From (100) it follows that the rebuilding of the field around the electron occurs in such way
that each Fourier-harmonic of frequency ω0 totally reconstructs and becomes the harmonic
of equilibrium coulomb field on the distance from the scattering point, which coincides with
radiation formation length ( | z ' | ≈ 2γ 2 / ω0) for thisω0. It is possible to place the plate quite
close to the scattering point so that at the moment of electron's traverse of the plate the Four‐
ier-harmonics of certain frequencies ω < ω0 will have not yet reconstructed. In other words it
is possible to place the plate in the pre-wave zone for these frequencies. In this case the inci‐
dent electron will be ‘half-bare’ and its transition radiation should differ from such radiation
by electron with equilibrium field.
The total field of the electron-plate system consists of the field of 'half-bare' electron E⊥ and
the field E⊥f of currents induced on the surface of the plate. Applying the boundary condi‐
tion for electric field on the surface of the plate
we can find the expression for the Fourier-harmonic of the field of induced surface currents:
iωR
→ e ϑ →
E⊥f (r , ω) = 2e F (r ) − 1 (101)
R ϑ 2 + γ −2 ω
{ }
2 −2 2
→ 1 ρ' + γ (z ' − 2z'0 ) iω z'0 1 ρ'2
where F ω (r ) = exp 2 2 + (z ' − z' )(z ' − 2z' ) ,
v ρ'2 + γ −2(z ' − z' )2 2 v γ 0 0
0
Ris the distance between the point of the electron's traverse of the plate and the point where
→
the field is considered, R ≈ z'0 − z ' + ρ'2 / 2(z ' − z'0 )and ϑ is counted from the direction of − v '.
This field gradually transforms into backward transition radiation.
The expression (101) can be simplified for − z ' ≫ 2γ 2 / ω. In this case:
z ' 0ω
iωR
→ e ϑ 1 i (1+γ 2ϑ 2)
E⊥f (r , ω) = 2e e 2γ 2
−1 (102)
R ϑ 2 + γ −2 v
and using (57) for spectral-angular density of transition radiation by 'half-bare' electron we
obtain:
dε
=
e2 ϑ2
dωdo π 2 (ϑ 2 + γ −2)2
2 1 − cos {
ω z'0
2
(γ −2 + ϑ 2) } (103)
The expression (103) differs from the corresponding expression for transition radiation by
electron with equilibrium field by the interference factor inside the braces and the coefficient
two in front of them. As we can see from (103), when the distance z'0 between the scattering
point and the plate is much less than the radiation formation length (lC ≈ 2γ 2 / ω) the radia‐
tion is highly suppressed. For larger values of z'0 the dependence of the radiation intensity
on z'0 has the oscillation type with the period of the order of the formation length:
4π
Λ= (104)
ω(ϑ 2 + γ −2)
Due to the nonzero frequency resolution Δω of the detector it is possible to observe such os‐
cillations only in the area limited by the condition
2π
z'0 < (105)
Δω(ϑ 2 + γ −2)
Also due to the nonzero size and, therefore, angular resolution of the detector the oscilla‐
tions can be observed only inside the region
π
z'0 < (106)
ωϑ Δϑ
For large distances z'0 ≫ Λ the considered oscillations disappear and the detector registers
an incoherent sum of contributions to transition radiation by electron's own field reflected
from the plate and by the field of bremsstrahlung in this direction.
6. Conclusion
The behavior of localized high-energy electromagnetic wave packets, which take place in
processes of transition radiation and bremsstrahlung by relativistic electrons has been con‐
sidered. It was shown that with the increase of the energy the stabilization of characteristics
of motion of such packets takes place, which consists in substantial decrease of the speed of
their dispersion. Essential here is the fact that at high energies the lengths, on which the re‐
construction of the form of such packets into packets of diverging waves takes place, can
reach macroscopic size, which can exceed the size of experimental facility. In this case both
the size of the used detector and its position relative to the region of the wave packet forma‐
tion become essential for measurements.
Such situation takes place, for example, after the sharp scattering of an electron to a large
angle. It was shown that as a result of such scattering the electron’s own coulomb field tears
away from it and turns into a localized packet of free electromagnetic waves, which trans‐
forms into a packet of diverging waves on large distance from the scattering point. For ultra
relativistic electrons such transformation of certain Fourier-harmonics of the packet field
takes place within the coherence length of the radiation process, which substantially exceeds
the length of the considered wave of radiation. In the case of low-frequency radiation this
coherence length can be macroscopic. It gives birth to the problem of bremsstrahlung char‐
acteristics measurement by different detectors, which consists in the dependence of the re‐
sults of measurement on the detector’s size and its position relative to the scattering point.
In the final direction of the electron motion (after the scattering) certain Fourier-components
of the field around it do not appear at once. The regeneration of these Fourier-components
occurs within the coherence length of the radiation processlC ≈ 2γ 2 / ω. In this case during a
long period of time the ultra relativistic electron is in ‘half-bare’ state without a certain part
of Fourier-components of its field. Such state of electron manifests itself, for example, during
further collisions of the electron with atoms of a substance, which causes different effects of
bremsstrahlung suppression (Landau-Pomeranchuk-Migdal effect, the effect of radiation
suppression in thin layer of substance – TSF-effect, etc.). We have shown that the ‘half-bare’
state of the scattered electron should manifest itself in the process of further transition radia‐
tion by such electron as well. This manifestation consists in the fact that characteristics of
transition radiation by such electron substantially depend on the position of the metallic
plate relative to the scattering point. In this case both the transition radiation suppression
and the oscillatory dependence of its characteristics on the distance between the plate and
the scattering point take place.
The transition radiation formation process also develops within distances of the order of the
coherence length of the radiation process,lC . Such situation takes place both in the direction
of the electron motion and in the opposite one. In the case of transition radiation, as in the
case of bremsstrahlung, these distances can be macroscopic. The analysis of development of
this process in space and time has shown that the packet of the waves reflected from the
plate has the same structure as the wave packet torn away from the electron at its instanta‐
neous scattering to a large angle. The structure of the field around the electron after its tra‐
verse of thin metallic plate is similar to the structure of the field, which forms around the
electron after its scattering to a large angle. In other words, in this case the electron can be in
the ‘half-bare’ state during a long period of time: a certain part of the Fourier components in
the field around it is suppressed. The mentioned analogies cause the existence of the similar
effects in transition radiation and bremsstrahlung. Thus the possibility of long existence of
electron in the ‘half-bare’ state and of different manifestations of such state of electron can
be investigated on the basis of the process of transition radiation by such electron.
Appendix
→
According to (72) the potential ϕ f (r , t) of the free field of electromagnetic waves is the sum
→ →
of the contributions to it by the propagating Ф1(r , t) and the surface Ф2(r , t) waves. While
deriving these functions we will pay special attention to the reasons of the step functions ap‐
pearance in (75).
→
In order to calculate Ф1(r , t) we will present this function in the form:
1 ∞
→ 2e
Ф1(r , t) = − lim 2
π
η→0 0
∫x
xdx
+ p −2 0
∫dω cosω ( | z | 1 − x 2 − t )e −ηω (107)
whereη > 0. Then after the variable substitution y = 1 − x 2 and integration over ω it is easy to
obtain that
1
→
Ф1(r , t) = −
2e
πv ∫ ydy
lim 2 −2 2
η→0 0 v − y
η
η + (y | z | − t)2
(108)
Integration over yin this expression leads to
→
Ф1(r , t) = −
2e
πv
lim
η→0
2( 2
η
1−v η + t + z / v
2 2 2)2 / ( 2 2)
4z t
− {
v 2(η 2 + t 2 + z 2 / v 2)
4z 2t 2
lnγ
v v 2(η 2 + t 2 + z 2 / v 2) 1 η 2 + ( | z | − t )2
− lnγ(1 + v) − 2 ln
(109)
2zt 4t 2
2z η2 + t2
( + arctg ) + ( + arctg )}
t |z | −t t v (η + t ) 1
2 2 2 |z | −t t
+ arctg arctg
2
z η η η 2zt zη η η
Taking into account that for η →0 only items proportional to

(arctg ( | z | − t) / η + arctg t / η ) remain in (109) and that for these items
|z | −t t π π
arctg + arctg = sign( | z | − t) + sign(t) = θ( | z | − t)θ(t)
η η 2 2
→
we obtain the following expression forФ1(r , t):
→
Ф1(r , t) =
e
( 1
−
1
2 | z | + vt | z | − vt
)
θ( | z | − t)θ(t) (110)
→
Thus the appearance of the step function θ( | z | − t) in Ф1(r , t) is connected with the limiting
procedure η → 0 in (109).
→
The calculation of Ф2(r , t) does not represent any difficulties as the integral over ω in it con‐
verges. As a result of the variable substitution u = 1 − x 2 and elementary integration we find

that:
→
Ф2(r , t) =
e
( 1
−
1
2 | z | − vt | z | + vt
)
sign(t) −
e 1
2 | z | − vt
(+ | |
1
z + vt
= )
(111)
1 1
=− | | θ( − t) − | | θ(t).
z − vt z + vt
Substituting the obtained expressions (110) and (111) for Ф1 and Ф2 into (72) we obtain the
expression (75) for the potential of the free waves.
Author details
Nikolai Fyodorovich Shul’ga* and Sergii Valeriyovich Trofymenko
*Address all correspondence to: shulga@ kipt.kharkov.ua
Akhiezer Institute for Theoretical Physics of National Science Centre “Kharkov Institute of
Physics and Technology” Kharkov, Ukraine
References
[1] Ter-Mikaelyan, M. L. (1972). High-Energy Electromagnetic Processes in media. New

York: Wiley.
[2] Akhiezer, A. I., & Shul’ga, N. F. (1996). High Energy Electrodynamics in Matter. Am‐
sterdam: Gordon and Breach Publ., 388 p.
[3] Dokshitzer, Yu. L., Khoze, V. A., Mueller, A. H., & Troyan, S. I. (1991). Basics of Per‐
turbative QCD. Gif sur Yvette Cedex: Editions Frontiers. 274p.
[4] Garibyan, G. M., & Shi, Y. (1983). X-Ray transition radiation. Yerevan: Publ. of Acad.
of Sc. of Arm. SSR, 320p. (in Russian)
[5] Ginzburg, V. L., & Tsytovich, V. N. (1984). Transition Radiation and Transition Scat‐
tering. Bristol: Adam Hilger.
[6] Rullhusen, P., Artru, X., & Dhez, P. (1998). Novel radiation sources using relativistic
electrons. Singapore: World Scientific Publ. 202p.
[7] Shibata, Y., Hasebe, S., Ishiki, K., et al. (1995). Observation of coherent diffraction ra‐
diation from bunched electrons passing through a circular aperture in the millimeter-
and submillimeter- wave length regions. Phys. Rev., 52 E, 6737.
[8] Dobrovolsky, S. N., & Shul’ga, N. F. (2003). Transversal spatial distribution of transi‐
tion radiation by relativistic electron in formation zone by the dotted detector. Nucl.
Instrum. Methods B, 201, 123-132.
[9] Gorham, P., et al. (2000). Radio-frequency measurements of coherent transition and
Cherenkov radiation: implication for high energy neutrino detection. Phys. Rev., 62 E,
8590-8605.
[10] Verzilov, V. A. (2000). Transition radiation in the pre-wave zone. Phys. Lett., 273 A,
135-140.
[11] Akhiezer, A. I., Shul’ga, N. F., & Fomin, S. P. (2005). Landau-Pomeranchuk-Migdal

Effect. Physics Reviews, Edited by I. M. Khalatnikov. Cambridge Sci. Publ., Printed in
UK, 22, 1-215.
[12] Potylitsyn, A. P. (2011). Electromagnetic Radiation of Electrons in Periodic Struc‐

tures. Berlin: Springer. 213p. 10.1007/978-3-642-19248-7.
[13] Naumenko, G. N., Potylitsyn, A. P., Sukhih, L. G., et al. (2009). Macroscopic effect of
shadow of electromagnetic field of relativistic electrons. JETP Letters, 90(2), 105-110.
[14] Blokhintsev, D. I. (1967). In the book: High-Energy Physics and Elementary Particles
Theory. Kiev: Nauk. Dumka. p. 778.
[15] Shul’ga, N. F., Syshchenko, V. V., & Shul’ga, S. N. (2009). On the motion of high-ener‐
gy wave packets and transition radiation by “half-bare” electron. Phys. Lett., 374 A,
331-334.
[16] Shul’ga, N. F., Trofymenko, S. V., & Syshchenko, V. V. (2010). The Space-Time Evolu‐
tion of the Process of Transition Radiation by Relativistic Electron. Journal of Kharkiv
National University, No. 916, phys. series “Nuclei, Particles, Fields”, 3(47), 23-41.
[17] Shul’ga, N. F., Trofymenko, S. V., & Syshchenko, V. V. (2011). On transition radiation
and bremsstrahlung by relativistic electron with nonequilibrium field. JETP Letters,
93(1), 3-7.
[18] Shul’ga, N. F., Trofymenko, S. V., & Syshchenko, V. V. (2011). On transition radiation
by relativistic electron. Il Nuovo Cimento, 34 C(4), 327-334.
[19] Shul’ga, N. F., Trofymenko, S. V., & Syshchenko, V. V. (2012). The prewave zone ef‐
fect in transition radiation and bremsstrahlung by relativistic electron. Problems of
atomic science and technology, No. 1 series: Nuclear Physics Investigations, 57, 134-138.
[20] Miller, W. (1974). Classical Limit Quantum Mechanics and the Theory of Molecular
Collisions. New York: Wiley, 69-177.
[21] Landau, L. D., & Lifshitz, E. M. (1987). The Classical Theory of Fields. Oxford: Perga‐
mon.
[22] Jackson, J. D. (1999). Classical Electrodynamics. New York: Wiley. 808p.
[23] Landau, L. D., & Pomeranchuk, I. Ya. (1953). Electron-cascade processes at ultra-high
energies. Dokl. Acad. Nauk. SSSR, 92, 735.
[24] Migdal, A. B. (1956). Bremsstrahlung and pair production in condensed media at

high energies. Phys. Rev., 103, 1811.
[25] Akhiezer, A. I., & Shul’ga, N. F. (1987). Influence of multiple scattering on the radia‐
tion of relativistic particles in amorphous and crystalline media. Sov. Phys. Usp., 30,
197-219.
[26] Ternovsky, F. F. (1961). On the theory of radiative processes in piece wise homogene‐
ous media. Sov. Phys. JETP, 12, 123.
[27] Shul’ga, N. F., & Fomin, S. P. (1978). Suppression of radiation in an amorphous medi‐
um and a crystal. JETP Letters, 27, 117-120.
[28] Fomin, S. P., & Shul’ga, N. F. (1986). On space-time evolution of the process of ultra
relativistic electron radiation in thin layer of substance. Phys. Lett., 114 A(3), 148-152.
[29] Anthony, P., et al. (1997). Bremsstrahlung suppression due to LPM and dielectric ef‐
fects in a variety of targets. Phys. Rev., 12 D, 1286.
[30] Klein, S. (1998). Suppression of bremsstrahlung and pair production due to environ‐
mental factors. Rev. of Mod. Phys., 71, 1501.
[31] Thomsen, H. D., Esberg, J., Kirsebom, K., et al. (2009). On the Macroscopic Formation
Length for GeV Photons. Phys. Lett., 672 B, 323-327.
[32] Thomsen, H. D., Esberg, J., Andersen, K. K., et al. (2010). Distorted Coulomb field of
the scattered electron. Phys. Rev., 81 D, 18.
[33] Feinberg, E. L. (1966). Consecutive interactions at high energies. Sov Phys. JETP, 23,
132.
[34] Feinberg, E. L. (1972). The problems of theoretical physics: paper collection dedicated
to the memory of I.Y. Tamm. Moscow, Nauka, p. 248 (in Russian).
[35] Akhiezer, A. I., & Shul’ga, N. F. (1982). Radiation of relativistic particles in single
crystals. Sov. Phys. Usp., 25, 541-564.
[36] Bleistein, N., & Handelsman, R. (1986). Asymptotic Expansions of Integrals. New
York: Holt, Dover Publications. 425p.
[37] Bolotovsky, B. M., & Serov, A. V. (2009). Peculiarities of transition radiation field.
Sov. Phys. Usp., 179, 517-524, (in Russian).
Provisional chapter
Chapter 11
The Effect of Weak Fields at Multiple Frequencies

The Effect of Weak Fields at Multiple Frequencies on
on the Scattering and Generation of Waves by
the Scattering and Generation of Waves by Nonlinear
Nonlinear Layered Media
Layered Media
LutzAngermann
Lutz Angermannandand
VasylVasyl V. Yatsyk
V. Yatsyk
Additionalinformation
Additional informationis is available
available at the
at the endend of the
of the chapter
chapter
http://dx.doi.org/10.5772/ 50679
1. Introduction
The chapter deals with the effects of weak fields at multiple frequencies on the scattering and
generation of waves by an isotropic, nonmagnetic, linearly polarised (E-polarisation), layered,
cubically polarisable, dielectric structure. In the domain of resonance frequencies we consider
wave packets consisting of both strong electromagnetic fields at the excitation frequency
of the nonlinear structure, leading to the generation of waves, and of weak fields at the
multiple frequencies, which do not lead to the generation of harmonics but influence on the
process of scattering and generation of waves by the nonlinear structure. The electromagnetic
waves for a nonlinear layer with a cubic polarisability of the medium can be described by
an infinite system of nonlinear boundary-value problems in the frequency domain. As has
been shown in previous articles of the authors ([3], [4]), in the study of particular nonlinear
effects it is possible to restrict this system to a finite number of equations. If the classical
formulation of the problem is supplemented by the condition of phase synchronism, we
arrive at a self-consistent formulation of a system of boundary-value problems with respect
to the components of the scattered and generated fields. It is known that this system is
equivalent to a system of nonlinear boundary-value problems of Sturm-Liouville type and
also to a system of one-dimensional nonlinear Fredholm integral equations of the second
kind.
The solution of the system of integral equations is approximated by the help of numerical
algorithms. Those include the application of suitable quadrature rules and iterative
procedures to solve the resulting nonlinear algebraic problems. Since in each iteration step
the solution of linear algebraic systems is required, the approximate solution of the nonlinear
problems is described by means of solutions of linear problems with an induced nonlinear
permittivity.
In continuation of our results from previous works ([3], [4]), where we only considered
one excitation field at the basic frequency, here results of calculations of characteristics of the
scattered and generated fields of plane waves are presented, taking into account the influence
of weak fields at multiple frequencies on the cubically polarisable layer. We restrict ourselves
to the investigation of the third harmonic generated by layers with a positive value of the
cubic susceptibility of the medium.
Within the framework of the self-consistent system which is given by a system of nonlinear
integral equations, we show the following. The variation of the imaginary parts of the
permittivities of the layer at the multiple frequencies can take both positive and negative
values along the height of the nonlinear layer. It is induced by the nonlinear part of the
permittivities and is caused by the loss of energy in the nonlinear medium which is spent
for the generation of the electromagnetic fields. The magnitudes of these variations are
determined by the amplitude and phase characteristics of the fields which are scattered and
generated by the nonlinear layer.
Layers with negative and positive values of the coefficient of cubic susceptibility of the
nonlinear medium have fundamentally different scattering and generation properties. In
the case of negative values of the susceptibility, a decanalisation of the electromagnetic field
can be detected. The maximal portion of the total energy generated in the third harmonic
is observed in the direction normal to the structure and nearly amounts to 4% of the total
dissipated energy. For a layer with a positive value of the susceptibility an effect of energy
canalisation is observed (see [4]). Increasing intensities of the incidents fields lead to an
increase of the angle of transparancy which increasingly deviates from the direction normal
to the layer. In this case, the maximal portion of energy generated in the third harmonic is
observed near the angle of transparency of the nonlinear layer. In the numerical experiments
there have been reached intensities of the excitation field of the layer such that the relative
portion of the total energy generated in the third harmonic is about 35% (see [4]).
In the chapter the effect of weak fields at multiple frequencies on the scattering and
generation of waves is investigated numerically. The results indicate a possibility of
designing a frequency multiplier and nonlinear dielectrics with controllable permittivity.
2. The mathematical model

We consider a layered nonlinear medium which is located in an infinite plate of thickness
4πδ, where δ > 0 is a given parameter: {r = ( x, y, z)⊤ ∈ R3 : |z| ≤ 2πδ}.
It is assumed that the vector of the polarisation moment P can be expanded in terms of the
electric field intensity E as follows:
P = χ(1) E + (χ(2) E)E + ((χ(3) E)E)E + . . . , (1)
where χ(1) , χ(2) , χ(3) are the media susceptibility tensors of rank two, three and four, with
(1) (2) (3)
components {χij }3i,j=1 , {χijk }3i,j,k=1 and {χijkl }3i,j,k,l =1 , respectively (see [5]). In the case of
isotropic media, the quadratic term disappears.
It is convenient to split P into its linear and nonlinear parts as P = P( L) + P( NL) := χ(1) E +
P( NL) . Similarly, with ε := I + 4πχ(1) and D( L) := εE, where I denotes the identity in C3 ,
The Effect of Weak Fields at Multiple Frequencies on the Scattering and Generation of Waves... 305
The Effect of Weak Fields at Multiple Frequencies on the Scattering and Generation of Waves by Nonlinear Layered Media 3
Figure 1. The nonlinear dielectric layered structure
the displacement field D can be decomposed as
D = D( L) + 4πP( NL) . (2)
ε is the linear term of the permittivity tensor. Furthermore, if the media under consideration
are nonmagnetic, isotropic and transversely inhomogeneous w.r.t. z, i.e. ε = ε( L) I with a
scalar, possibly complex-valued function ε( L) = ε( L) (z), if the wave is linearly E-polarised,
i.e.
E = ( E1 , 0, 0)⊤ , H = (0, H2 , H3 )⊤ , (3)

and if the electric field E is homogeneous w.r.t. the coordinate x, i.e. E(r, t) =
( E1 (t; y, z), 0, 0)⊤ , then the Maxwell’s equations together with (2) are simplified to
ε ( L ) ∂2 4π ∂2
∇2 E − 2 2
E − 2 2 P( NL) = 0, (4)
c ∂t c ∂t
where ∇2 reduces to the Laplacian w.r.t. y and z, i.e. ∇2 := ∂2 /∂y2 + ∂2 /∂z2 .

A stationary electromagnetic wave (with the oscillation frequency ω > 0) propagating in
a nonlinear dielectric structure gives rise to a field containing all frequency harmonics (see
[20], [22], and other authors). Therefore, representing E, P( NL) via Fourier series
1 1
E(r, sω )e−isωt , P( NL) (r, t) = P( NL) (r, sω )e−isωt ,
2 s∑ 2 s∑
E(r, t) = (5)
∈Z ∈Z
we obtain from (4) an infinite system of coupled equations w.r.t. the Fourier amplitudes. In
the case of a three-component E-polarised electromagnetic field (3), this system reduces to a
system of scalar equations w.r.t. E1 :
ε( L) (sω )2 4π (sω )2 ( NL)

∆E1 (r, sω ) + 2
E1 (r, sω ) + P1 (r, sω ) = 0, s ∈ N. (6)
c c2
In writing equation (6), the property E1 (r; jω ) = E1 (r; − jω ) of the Fourier coefficients and
the lack of influence of the static electric field E1 (r, sω )|s=0 = 0 on the nonlinear structure
were taken into consideration.
We assume that the main contribution to the nonlinearity is introduced by the term
P( NL) (r, sω ) (cf. [25], [14], [8], [1], [24], [10], [19], [11]), and we take only the
lowest-order terms in the Taylor series expansion of the nonlinear part P( NL) (r, sω ) =
�⊤
( NL)
�
P1 (r, sω ), 0, 0 of the polarisation vector in the vicinity of the zero value of the electric
field intensity, cf. (1). In this case, the only nontrivial component of the polarisation vector
is determined by susceptibility tensor of the third order χ(3) . In the time domain, this
component can be represented in the form (cf. (1) and (5)):
( NL) · 1 (3)
P1 (r, t) = ∑ χ1111 (sω; nω, mω, pω ) E1 (r, nω ) E1 (r, mω ) E1 (r, pω )e−isωt , (7)
8�
n,m,p,s∈Z \{0}
n+m+ p=s
·
where the symbol = means that higher-order terms are neglected. Applying to (7) the Fourier
transform w.r.t. time, we obtain an expansion in the frequency domain:
( NL) 1 (3)
3χ1111 (sω; jω, − jω, sω )| E1 (r, jω )|2 E1 (r, sω )
4 j∑
P1 (r, sω ) =
∈N
1 (3)
4  n,m,p∑
+ χ1111 (sω; nω, mω, pω ) E1 (r, nω ) E1 (r, mω ) E1 (r, pω ).
∈Z \{0}
(8)

 n�=−m, p=s


m�=− p, n=s
 n�=− p, m=s



n+m+ p=s
We see that, under the above assumptions, the electromagnetic waves in a nonlinear medium
with a cubic polarisability are described by an infinite system (6) and (8) of coupled nonlinear
equations ([25], [14], [8], [2]).
Here we study nonlinear effects involving the waves at the first three frequency components
of E1 only. That is, we further neglect the influence of harmonics of order higher than 3.
Then it is possible to restrict the examination of the system (6) and (8) to three equations,
and also to leave particular terms in the representation of the polarisation coefficients. Taking
into account only the nontrivial terms in the expansion of the polarisation coefficients and
(3)
using the so-called Kleinman’s rule (i.e. the equality of all the coefficients χ1111 at the multiple
frequencies, [7], [9]), we arrive at the following system:
∆E1 (r, nκ ) + (nκ )2 ε nκ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ ))

= −δn1 κ 2 α(z) E12 (r, 2κ ) E1 (r, 3κ ) (9)
� �
1 3
− δn3 (3κ )2 α(z) E1 (r, κ ) + E12 (r, 2κ ) E1 (r, κ ) , n = 1, 2, 3,
3

1, |z| > 2πδ, (1)
with κ := ω
c = 2π
λ , ε nκ := ( NL) and ε( L) := 1 + 4πχ11 ,
ε( L) + ε nκ , |z| ≤ 2πδ,
( NL)
= α(z) | E1 (r, κ )|2 + | E1 (r, 2κ )|2 + | E1 (r, 3κ )|2

ε nκ
2 (10)
E1 (r, κ ) E (r, 2κ )
E (r, 3κ ) + δn2 1

+ δn1 E (r, κ ) E1 (r, 3κ ) ,
E1 (r, κ ) 1 E1 (r, 2κ ) 1
(3)
where α(z) := 3πχ1111 (z) is the so-called function of cubic susceptibility and δnj denotes
Kronecker’s symbol. For transversely inhomogeneous media (a layer or a layered structure),
the linear part ε( L) of the permittivity is described by a piecewise smooth or even a piecewise
constant function. Similarly, the function of the cubic susceptibility α = α(z) is also a
piecewise smooth or even a piecewise constant function. This assumption allows us to
investigate the scattering and generation characteristics of a nonlinear layer and of a layered
structure (consisting of a finite number of linear and nonlinear dielectric layers) within one
and the same mathematical model.
Here and in what follows we use the following notation: (r, t) are dimensionless
spatial-temporal coordinates such that the thickness of the layer is equal to 4πδ. The
time-dependence is determined by the factors exp(−inωt), where ω = κc is the
dimensionless circular frequency and κ is a dimensionless frequency parameter. This
parameter characterises the ratio of the true thickness h of the layer to the free-space
wavelength λ, i.e. h/λ = 2κδ. c = (ε 0 µ0 )−1/2 denotes a dimensionless parameter, equal
to the absolute value of the speed of light in the medium containing the layer (Im c = 0). ε 0
and µ0 are the material parameters of the medium. The absolute values of the true variables
r′ , t′ , ω ′ are given by the formulas r′ = hr/4πδ, t′ = th/4πδ, ω ′ = ω4πδ/h.
The scattered and generated field in a transversely inhomogeneous, nonlinear dielectric
layer excited by a plane wave is quasi-homogeneous along the coordinate y, hence it can be
represented as
(C1) E1 (r, nκ ) =: E1 (nκ; y, z) := U (nκ; z) exp(iφnκ y), n = 1, 2, 3.
Here U (nκ; z) and φnκ := nκ sin ϕnκ denote the complex-valued transverse component of the
Fourier amplitude of the electric field and the value of the longitudinal propagation constant
(longitudinal wave-number) at the frequency nκ, respectively, where ϕnκ is the given angle
of incidence of the exciting field of frequency nκ (cf. Fig. 1).
Furthermore we require that the following condition of the phase synchronism of waves is
satisfied:
(C2) φnκ = nφκ , n = 1, 2, 3.
Then the permittivity of the nonlinear layer can be expressed as

ε nκ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) = ε nκ (z, α(z), U (κ; z), U (2κ; z), U (3κ; z))
= ε( L) (z) + α(z) |U (κ; z)|2 + |U (2κ; z)|2 + |U (3κ; z)|2

+ δn1 |U (κ; z)||U (3κ; z)| exp {i [−3arg(U (κ; z)) + arg(U (3κ; z))]} (11)
+ δn2 |U (κ; z)||U (3κ; z)| exp {i [−2arg(U (2κ; z)) + arg(U (κ; z)) + arg(U (3κ; z))]} ,
n = 1, 2, 3.
A more detailed explanation of the condition (C2) can be found in [3, Sect. 3]. In the
considered case of spatially quasi-homogeneous (along the coordinate y) electromagnetic
fields (C1), the condition of the phase synchronism of waves (C2) reads as
sin ϕnκ = sin ϕκ , n = 1, 2, 3.
Consequently, the given angle of incidence of a plane wave at the frequency κ coincides with
the possible directions of the angles of incidence of plane waves at the multiple frequencies
nκ, i.e. ϕnκ = ϕκ , n = 1, 2, 3. The angles of the wave scattered by the layer are equal to ϕscat
nκ =
− ϕnκ in the zone of reflection z > 2πδ and ϕscat nκ = π + ϕnκ and in the zone of transmission
of the nonlinear layer z < −2πδ, where all angles are measured counter-clockwise in the
(y, z)-plane from the z-axis (cf. Fig. 1).
The conditions (C1), (C2) allow a further simplification of the system (9). Before we do so,
we want to make a few comments on specific cases which have already been discussed in
the literature. First we mention that the effect of a weak quasi-homogeneous electromagnetic
field (C1) on the nonlinear dielectric structure such that harmonics at multiple frequencies are
not generated, i.e. E1 (r, 2κ ) = 0 and E1 (r, 3κ ) = 0, reduces to find the electric field component
E1 (r, κ ) determined by the first equation of the system (9). In this case, a diffraction problem
for a plane wave on a nonlinear dielectric layer with a Kerr-type nonlinearity ε nκ = ε( L) (z) +
α(z)| E1 (r, κ )|2 and a vanishing right-hand side is to be solved, see [1, 8, 11, 14, 19, 24, 25]. The
generation process of a field at the triple frequency 3κ by the nonlinear dielectric structure
is caused by a strong incident electromagnetic field at the frequency κ and can be described
by the first and third equations of the system (9) only, i.e. n = 1, 3. Since the right-hand side
of the second equation in (9) is equal to zero, we may set E1 (r, 2κ ) = 0 corresponding to the
homogeneous boundary condition w.r.t. E1 (r, 2κ ). Therefore the second equation in (9) can
be completely omitted, see [2] and the numerical results in [3], [4].
A further interesting problem consists in the investigation of the influence of a packet of
waves on the generation of the third harmonic, if a strong incident field at the basic frequency
κ and, in addition, weak incident quasi-homogeneous electromagnetic fields at the double
and triple frequencies 2κ, 3κ (which alone do not generate harmonics at multiple frequencies)
excite the nonlinear structure. Then we have to take into account all three equations of
system (9). This is caused by the inhomogeneity of the corresponding problem, where a
weak incident field at the double frequency 2κ (or two weak fields at the frequencies 2κ and
3κ) excites (resp. excite) the dielectric medium.
So we consider the problem of scattering and generation of waves on a nonlinear, layered,
cubically polarisable structure, which is excited by a packet of plane waves consisting of a
strong field at the frequency κ (which generates a field at the triple frequency 3κ) and of
weak fields at the frequencies 2κ and 3κ (having an impact on the process of third harmonic
generation due to the contribution of weak electromagnetic fields)
3
E1inc (r, nκ ) := E1inc (nκ; y, z) := ainc
nκ exp i φnκ y − Γnκ ( z − 2πδ ) , z > 2πδ , (12)
n =1
with amplitudes ainc

nκ and angles of incidence ϕnκ , | ϕ | < π/2 (cf. Fig. 1), where φnκ : =
nκ sin ϕ�
nκ are the longitudinal propagation constants (longitudinal wave-numbers) and
Γnκ := (nκ )2 − φnκ

2 are the transverse propagation constants (transverse wave-numbers).
In this setting, if a packet of plane waves excites a nonmagnetic, isotropic, linearly polarised
(i.e.
1 ∂E1 (nκ; y, z) 1 ∂E1 (nκ; y, z) ⊤

� �
E(r, nκ ) = ( E1 (nκ; y, z), 0, 0)⊤ , H(r, nκ ) = 0, ,−
inωµ0 ∂z inωµ0 ∂y
(1)
(E-polarisation)), transversely inhomogeneous ε( L) = ε( L) (z) = 1 + 4πχ11 (z) dielectric layer
( NL)
(see Fig. 1) with a cubic polarisability P( NL) (r, nκ ) = ( P1 (nκ; y, z), 0, 0)⊤ of the medium,
the complex amplitudes of the total fields
E1 (r, nκ ) =: E1 (nκ; y, z) := U (nκ; z) exp(iφnκ y) := E1inc (nκ; y, z) + E1scat (nκ; y, z)
satisfy the system of equations (cf. (9) – (10))
∇2 E1 (r, κ ) + κ 2 ε κ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) E1 (r, κ )




= −α(z)κ 2 E12 (r, 2κ ) E1 (r, 3κ ),





∇2 E1 (r, 2κ ) + (2κ )2 ε 2κ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) E1 (r, 2κ ) = 0,

(13)
∇2 E1 (r, 3κ ) + (3κ )2 ε 3κ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) E1 (r, 3κ )




 �1 �
= −α(z)(3κ )2 E13 (r, κ ) + E12 (r, 2κ ) E1 (r, κ )



3
together with the following conditions, where Etg (nκ; y, z) and Htg (nκ; y, z) denote
the tangential components of the intensity vectors of the full electromagnetic field
{E(nκ; y, z)}n=1,2,3 , {H(nκ; y, z)}n=1,2,3 :
(C1) E1 (nκ; y, z) = U (nκ; z) exp(iφnκ y), n = 1, 2, 3

(the quasi-homogeneity condition w.r.t. the spatial variable y introduced above),
(C2) φnκ = nφκ , n = 1, 2, 3,
(the condition of phase synchronism of waves introduced above),
(C3) Etg (nκ; y, z) and Htg (nκ; y, z) (i.e. E1 (nκ; y, z) and H2 (nκ; y, z))
are continuous at the boundary layers of the nonlinear structure,
� scat �
anκ
(C4) E1scat (nκ; y, z) = scat exp (i ( φnκ y ± Γnκ ( z ∓ 2πδ ))) , z < ± 2πδ ,
> n = 1, 2, 3
bnκ
(the radiation condition w.r.t. the scattered and generated fields).
The condition (C4) provides a physically consistent behaviour of the energy characteristics
of scattering and generation. It guarantees the absence of waves coming from infinity (i.e.
z = ±∞), see [12]. We study the scattering and generation properties of the nonlinear layer,
where in (C4) we always have
Im Γnκ = 0, Re Γnκ > 0. (14)
Note that (C4) is also applicable for the analysis of the wave-guide properties of the layer,
where Im Γnκ > 0, Re Γnκ = 0. The desired solution of the scattering and generation problem
(13) under the conditions (C1) – (C4) can be represented as follows:
E1 (nκ; y, z) = U (nκ; z) exp(iφnκ y)

 ainc scat
nκ exp(i ( φnκ y − Γnκ ( z − 2πδ ))) + anκ exp(i ( φnκ y + Γnκ ( z − 2πδ ))), z > 2πδ,
= U (nκ; z) exp(iφnκ y), |z| ≤ 2πδ, (15)
 scat
bnκ exp(i (φnκ y − Γnκ (z + 2πδ))), z < −2πδ,
n = 1, 2, 3.
Substituting this representation into the system (13), the following system of nonlinear
ordinary differential equations results, where “ ′ ” denotes the differentiation w.r.t. z:
U ′′ (nκ; z) + �Γ2nκ − (nκ )2 [1 − ε nκ (z, α(z)�

� �
, U (κ; z), U (2κ; z), U (3κ; z))] U (nκ; z)
��
2 2 1 3 2
= −(nκ ) α(z) δn1 U (2κ; z)U (3κ; z) + δn3 U (κ; z) + U (2κ; z)U (κ; z) , (16)
3
|z| ≤ 2πδ, n = 1, 2, 3.
The boundary conditions follow from the � continuity �of the tangential
� components
� of the
full fields of scattering and generation Etg (nκ; y, z) n=1,2,3 , Htg (nκ; y, z) n=1,2,3 at the
boundaries z = ±2πδ of the nonlinear layer (cf. (C3)). According to (C3) and the
representation of the electrical components of the electromagnetic field (15), at the boundary
of the nonlinear layer we obtain:
U (nκ; 2πδ) = ascat inc ′

� scat inc
�
nκ + anκ , U ( nκ; 2πδ ) = iΓnκ anκ − anκ , (17)
scat
U (nκ; −2πδ) = bnκ , ′ scat
U (nκ; −2πδ) = −iΓnκ bnκ , n = 1, 2, 3.
Eliminating in (17) the unknown values of the complex amplitudes ascat

� �
nκ n=1,2,3 ,
� scat �
bnκ n=1,2,3 of the scattered and generated fields and taking into consideration that ainc
nκ =
U inc (nκ; 2πδ), we arrive at the desired boundary conditions for the problem (16):
iΓnκ U (nκ; −2πδ) + U ′ (nκ; −2πδ) = 0,

(18)
iΓnκ U (nκ; 2πδ) − U ′ (nκ; 2πδ) = 2iΓnκ ainc
nκ , n = 1, 2, 3.
The system of ordinary differential equations (16) and the boundary conditions (18) form
a semi-linear boundary-value problem of Sturm-Liouville type, see also [2, 14, 15, 25]. The
existence and uniqueness of a weak solution of the nonlinear boundary-value problem (16)
and (18) have been demonstrated in [4, Sect. 4].
The problem (13), (C1) – (C4) can also be reduced to finding solutions of one-dimensional
nonlinear integral equations w.r.t. the components U (nκ; z), n = 1, 2, 3, z ∈ [−2πδ, 2πδ] , of
the fields scattered and generated in the nonlinear layer (cf. [3], [2], [15], [25], [14], [8]):
i (nκ )2 2πδ
�
U (nκ; z) + exp(iΓnκ |z − z0 |) ×
2Γnκ −2πδ
× [1 − ε nκ (z0 , α(z0 ), U (κ; z0 ) , U (2κ; z0 ) , U (3κ; z0 ))] U (nκ; z0 )dz0
i (nκ )2 2πδ
�
= δn1 exp(iΓnκ |z − z0 |)α(z0 )U 2 (2κ; z0 )U (3κ; z0 )dz0 (19)
2Γnκ −�2πδ
i (nκ )2 2πδ
� �
1 3
+ δn3 exp(iΓnκ |z − z0 |)α(z0 ) U (κ; z0 ) + U 2 (2κ; z0 )U (κ; z0 ) dz0
2Γnκ −2πδ 3
+ U inc (nκ; z), |z| ≤ 2πδ, n = 1, 2, 3.
Here U inc (nκ; z) = ainc

nκ exp [−iΓnκ ( z − 2πδ )] , n = 1, 2, 3.
The solution of the original problem (13), (C1) – (C4), represented as (15), can be obtained
from (19) using the formulas
U (nκ; 2πδ) = ainc scat

nκ + anκ ,
scat
U (nκ; −2πδ) = bnκ , n = 1, 2, 3, (20)
(cf. (C3)). The system (19) can be regarded as an integral representation of the desired
solution of (13), (C1) – (C4) (i.e. solutions of the form E1 (nκ; y, z) = U (nκ; z) exp (iφnκ y), n =
1, 2, 3, see (15)) for points located outside the nonlinear layer: {(y, z) : |y| < ∞, |z| > 2πδ} .
Indeed, given the solution of nonlinear integral equations (19) in the region |z| ≤ 2πδ, the
substitution into the integrals of (19) leads to explicit expressions of the desired solutions
U (nκ; z) for points |z| > 2πδ outside the nonlinear layer at each frequency nκ, n = 1, 2, 3.
In the case of a linear system (19), i.e. if α ≡ 0, the problem of existence and uniqueness
of solutions has been investigated in [16], [12]. Sufficient conditions for the existence and
uniqueness of solutions U (nκ; ·) ∈ L2 (−2πδ, 2πδ) to the nonlinear system (19) have been
given in [4, Sect. 6].
3. A self-consistent approach to the numerical analysis of the nonlinear

integral equations
According to [3], [2], the application of suitable quadrature rules to the system of nonlinear
integral equations (19) leads to a system of complex-valued nonlinear algebraic equations of
the second kind:
 (I − Bκ (Uκ , U2κ , U3κ ))Uκ = Cκ (U2κ , U3κ ) + Uinc


κ ,
(I − B2κ (Uκ , U2κ , U3κ ))U2κ = Uinc
2κ , (21)
(I − B3κ (Uκ , U2κ , U3κ ))U3κ = C3κ (Uκ , U2κ ) + Uinc
3κ ,

where {zl }lN=1 is a discrete set of nodes such that −2πδ =: z1 < z2 < ... < zl < ... < z N =:
2πδ.
Unκ := {Ul (nκ )}lN=1 ≈ {U (nκ; zl )}lN=1 denotes the vector of the unknown approximate
solution values corresponding to the frequencies nκ, n = 1, 2, 3. The matrices are of the
form
N
Bnκ (Uκ , U2κ , U3κ ) = { Am Klm (nκ, Uκ , U2κ , U3κ )}l,m =1
with entries
i (nκ )2 � �
Klm (nκ, Uκ , U2κ , U3κ ) := − exp (iΓnκ |zl − zm |) 1 − ε( L) (zm )
� 2Γnκ
+ α(zm ) |Um (κ )|2 + |Um (2κ )|2 + |Um (3κ )|2 (22)
+ δn1 |Um (κ )| |Um (3κ )| exp {i [−3arg Um (κ ) + arg Um (3κ )]} ��
+ δn2 |Um (κ )| |Um (3κ )| exp {i [−2arg Um (2κ ) + arg Um (κ ) + arg Um (3κ )]} .
N
The numbers Am are the coefficients determined by the quadrature rule, I := {δlm }l,m =1 is
the identity matrix, and δlm is Kronecker’s symbol.
The right-hand side of (21) is defined by
N
Uinc inc
nκ : = { anκ exp[−iΓnκ ( zl − 2πδ )]}l =1 ,
� iκ 2 N �N
2
Cκ (U2κ , U3κ ) :=
2Γκ m=1 ∑ Am exp(iΓκ |zl − zm |)α(zm )Um (2κ )U m (3κ )
l =1
,
� i (3κ )2 N �1 �� N
3 2
C3κ (Uκ , U2κ ) :=
2Γ3κ ∑ Am exp(iΓ3κ |zl − zm |)α(zm )
3
Um (κ ) + Um (2κ )U m (κ )
l =1
.
m =1
Given a relative error tolerance ξ > 0, the solution of (21) is approximated by means of the
following iterative method:
 ��
( s −1) (S (q)) (S (q))
��
(s)
Q

 I − Bκ Uκ , U2κ2 , U3κ3 Uκ 

 
S ( q ) : η ( S ( q ))< ξ
 
(S2 (q)) (S3 (q)) 1 1 1
 � � � 

 = C U , U + U inc 


 κ 2κ 3κ κ



 s =1 
 �� S2 (q):η2 (S2 (q))<ξ  
(S1 (q)) ( s −1) (S3 (q)) (s)
� ��
I − B2κ Uκ , U2κ , U3κ U2κ = U2κ inc (23)

 s =1 

(S1 (q)) (S2 (q)) ( s −1) (s)
 �� 
I − B U , U , U U
 
3κ κ
 


 2κ 3κ 3κ 

 �S3 (q):η3 (S3 (q))<ξ  
(S1 (q)) (S2 (q))
 � � 

 = C3κ Uκ , U2κ + U3κinc 

s =1 q =1
(s) ( s −1) (s)

with ηn (s) := |Unκ − Unκ �/�Unκ , n = 1, 2, 3, where the terminating index Q ∈ N is
defined by the requirement
max {η1 ( Q), η2 ( Q), η3 ( Q)} < ξ .

4. Eigen-modes of the linearised problems of scattering and generation of

waves on the cubically polarisable layer
The solution of the system of nonlinear equations (19) is approximated by the solution of a
linearised system, for given values of the induced dielectric permittivity and of the source
functions at the right-hand side of the system. The solution can be found by the help of
algorithm (23), where at each step a system of linearised nonlinear complex-valued algebraic
equations of the second kind is solved iteratively. The analytic continuation of the linearised
nonlinear problems into the region of complex values of the frequency parameter allows
us to switch to the analysis of spectral problems. That is, the eigen-frequencies and the
corresponding eigen-fields of the homogeneous linear problems with an induced nonlinear
permittivity are to be determined. The results of the development of a spectral theory of
linear problems for structures with noncompact boundaries can be found in [23], [13], [16],
[12], [18], [17].
As mentioned above, the classical formulation of the problem of scattering and generation
of waves, described by the system of boundary value problems (13), (C1) – (C4), can be
reformulated as a set of independent spectral problems in the following way:

Find the eigen-frequencies κn and the corresponding eigen-functions E1 (r, κn ) (i.e. κn ∈
3
Ωnκ ⊂ Hnκ , E1 (r, κn ) n=1 , where Ωnκ are the discrete sets of eigen-frequencies lying on the
two-sheeted Riemann surfaces Hnκ , see Fig. 2 and the more detailed explanations below)
satisfying the equations
∇2 E1 (r, κn ) + κn2 ε nκ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) E1 (r, κn ) = 0, n = 1, 2, 3, (24)
together with the following conditions:
(CS1) E1 (κn ; y, z) = U (κn ; z) exp(iφnκ y), n = 1, 2, 3

(the quasi-homogeneity condition w.r.t. the spatial variable y),
(CS2) φnκ = nφκ , n = 1, 2, 3
(the condition of phase synchronism of waves),
(CS3) Etg (κn ; y, z) and Htg (κn ; y, z) (i.e. E1 (κn ; y, z) and H2 (κn ; y, z)) are
continuous at the boundary layers of the structure with the induced permittivity ε nκ for
κ := κ inc , n = 1, 2, 3,

aκ n
(CS4) E1 (κn ; y, z) = exp (i (φnκ y ± Γκn (κn , φnκ )(z ∓ 2πδ))) , z> < ± 2πδ , n = 1, 2, 3
bκn
(the radiation condition w.r.t. the eigen-field).
For real values of the parameters κn and φnκ , the condition (CS4) meets the physically
reasonable requirement of the absence of radiation fields of waves coming from infinity
z = ±∞:
Im Γκn (κn , φnκ ) ≥ 0, Re Γκn (κn , φnκ ) Re κn ≥ 0 for Im φnκ = 0, Im κn = 0, n = 1, 2, 3.

(25)
Figure 2. The geometry of the two-sheeted Riemann surfaces Hnκ
The nontrivial solutions (eigen-fields) of problem (24), (CS1) – (CS4) can be represented as

 aκn exp(i (φnκ y + Γκn (κn , φnκ )(z − 2πδ))), z > 2πδ,
E1 (κn ; y, z) = U (κn ; z) exp(iφnκ y) = U (κn ; z) exp(iφnκ y), |z| ≤ 2πδ,
bκn exp(i (φnκ y − Γκn (κn , φnκ )(z + 2πδ))), z < −2πδ,

κn ∈ Ωnκ ⊂ Hnκ , n = 1, 2, 3,
(26)
where κ := κ inc is a given constant equal to the value of the excitation frequency
of the nonlinear structure, Γκn (κn , φnκ ) := (κn2 − φnκ 2 )1/2 are the transverse propagation
functions depending on the complex frequency spectral variables κn , φnκ := nκ sin( ϕnκ )
denote the given real values of the longitudinal propagation constants, ε nκ =
ε nκ (z, α(z), E1 (r, κ ), E1 (r, 2κ ), E1 (r, 3κ )) are the induced dielectric permittivities at the
frequencies nκ corresponding to the parameter κ := κ inc , Ωnκ are the sets of eigen-frequencies
and Hnκ are two-sheeted Riemann surfaces (cf. Fig. 2), n = 1, 2, 3. The range of the spectral
parameters κn ∈ Ωnκ is completely determined by the boundaries of those regions in
which the analytic continuation (consistent with the condition (25)) of the canonical Green’s
functions
i
G0 (κn ; q, q0 ) = exp {i [φnκ (y − y0 ) + Γκn (κn , φnκ ) |z − z0 |]} /Γκn (κn , φnκ ), n = 1, 2, 3,
4Y
into the complex space of the spectral parameters κn of the unperturbed problems (24), (CS1)
– (CS4) (i.e. for the special case ε nκ ≡ 1, n = 1, 2, 3) is possible. These complex spaces are
two-sheeted Riemann surfaces Hnκ (see Fig. 2) with real algebraic branch points of second
order κn± : Γκn (κn± , φnκ ) = 0 (i.e. κn± = ±|φnκ |, n = 1, 2, 3) and with cuts starting at these
points and extending along the lines
(Re κn )2 − (Im κn )2 − φnκ

2
= 0, Im κn ≤ 0, n = 1, 2, 3 . (27)
The first, “physical” sheets (i.e. the pair of values {κn , Γκn (κn , φnκ )}) on each of the surfaces
Hnκ , n = 1, 2, 3, are completely determined by the condition (25) and the cuts (27). At the first
sheets of Hnκ the signs of the pairs {κn , Re Γκn } and {κn , Im Γκn } are distributed as follows:
Im Γκn > 0 for 0 < arg κn < π, Re Γκn ≥ 0 for 0 < arg κn < π/2 and Re Γκn ≤ 0 for
π/2 ≤ arg κn < π. For points κn with 3π/2 ≤ arg κn ≤ 2π, the function values (where
(Re κn )2 − (Im κn )2 − φnκ2 > 0) are determined by the condition Im Γκn < 0, Re Γκn > 0, for
the remaining points κn the function Γκn (κn , φnκ ) is determined by the condition Im Γκn > 0,
Re Γκn ≤ 0. In the region π < arg κn < 3π/2 the situation is similar to the previous one
up to the change of the sign of Re Γκn . The second, “unphysical” sheets of the surfaces Hnκ ,
n = 1, 2, 3 are different from the “physical” ones in that, for each κn , the signs of both Re Γκn
and Im Γκn are reversed.
The qualitative analysis of the eigen-modes of the linearised problems (24), (CS1) – (CS4) is
carried out using the equivalent formulation of spectral problems for the linearised nonlinear
integral equations (19). It is based on the analytic continuation of (19) into the space of
spectral values κn ∈ Ωnκ ⊂ Hnκ , n = 1, 2, 3.
The spectral problem reduces to finding nontrivial solutions U (κn ; z) of a set of homogeneous
(i.e. with vanishing right-hand sides), linear (i.e. linearised equations (19)) integral equations
with the induced dielectric permittivity at the frequencies nκ of excitation and generation:
iκn2 2πδ
�
U (κ n ; z ) + exp(iΓκn (κn , φnκ )|z − z0 |)
2Γκn (κn , φnκ ) −2πδ
(28)
× [1 − ε nκ (z0 , α(z0 ), U (κ; z0 ), U (2κ; z0 ), U (3κ; z0 ))] U (κn ; z0 )dz0 = 0;
|z| ≤ 2πδ, κ := κ inc , κn ∈ Ωnκ ⊂ Hnκ , n = 1, 2, 3.
The solution of the problem (24), (CS1) – (CS4) can be obtained from the solution of the
equivalent problem (28), where – according to (CS3) – in the representation of the eigen-fields
(26) the following formulas are used:
U (κn ; 2πδ) = aκn , U (κn ; −2πδ) = bκn , n = 1, 2, 3. (29)
The qualitative analysis of the spectral characteristics allows to develop algorithms for
solving the spectral problems (24), (CS1) – (CS4) by reducing them to the equivalent spectral
problem of finding nontrivial solutions of the integral equations (28), see [13], [23]. The
solvability of (28) follows from an analysis of the basic qualitative characteristics of the
spectra. Applying to the integral equations (28) appropriate quadrature formulas, we
obtain a set of independent systems of linear algebraic equations of second kind depending
nonlinearly on the spectral parameter: (I − Bnκ (κn ))Uκn = 0, where κn ∈ Hnκ , κ := κ inc , n =
1, 2, 3. Consequently, the spectral problem of finding the eigen-frequencies κn ∈ Ωnκ ⊂ Hnκ
and the corresponding eigen-fields (i.e. the nontrivial solutions of the integral equations (28))
reduces to the following algorithm:

 f nκ (κn ) := det(I − Bnκ (κn )) = 0,
(I − Bnκ (κn ))Uκn = 0, (30)
κ := κ inc , κn ∈ Ωnκ ⊂ Hnκ , n = 1, 2, 3.

Here we use a similar notation to that in Section 3. κn are the desired eigen-frequencies, and
Uκn = {U (κn ; zl )}lN=1 := {Ul (κn )}lN=1 are the vectors of the unknown approximate solution
values corresponding to the frequencies κn . The matrices are of the form
N
Bnκ (κn ) := Bnκ (κn ; Uκ , U2κ , U3κ ) = { Am Klm (κn , Uκ , U2κ , U3κ )}l,m =1 (31)
with given values of the vectors of the scattered and generated fields Unκ = {U (nκ; zl )}lN=1 :=
{Ul (nκ )}lN=1 , n = 1, 2, 3. The numbers Am are the coefficients determined by the quadrature
rule, and the entries Klm (κn , Uκ , U2κ , U3κ ) are calculated by means of (22), where the first
argument nκ is replaced by κn . The eigen-frequencies κn ∈ Ωnκ ⊂ Hnκ , n = 1, 2, 3, i.e.
the characteristic numbers of the dispersion equations of the problem (30), are obtained by
solving the corresponding dispersion equations f nκ (κn ) := det(I − Bnκ (κn )) = 0 by the help
of Newton’s method or its modifications. The nontrivial solutions Uκn of the homogeneous
systems of linear algebraic equations (30) corresponding to these characteristic numbers are
the eigen-fields (26) of the linearised nonlinear layered structures with an induced dielectric
constant. Since the solutions Uκn are unique up to multiplication by an arbitrary constant,
we require U (κn ; 2πδ) = aκn := 1 (cf. (26)). According to (31), the matrix entries in (30)
depend on the dielectric permittivities. The latter are defined by the scattered and generated
fields Uκ , U2κ , U3κ of the problem (19) by means of the algorithm (23). This defines the basic
design of the implemented numerical algorithm. The investigation of the eigen-modes of
the linearised nonlinear structures (30) should always precede the solution of the nonlinear
scattering and generation problem in the self-consistent formulation (23). Note that, in
the analysis of the linear structures, the problem of excitation (scattering) and the spectral
problem can be solved independently.
5. Numerical examples
Consider the excitation of the nonlinear structure by a strong incident field at the basic
frequency κ and, in addition, by weak incident quasi-homogeneous electromagnetic fields at
the double and triple frequencies 2κ, 3κ (see (12)), i.e.
0 < max{| ainc inc inc

2κ |, | a3κ |} ≪ | a1κ |. (32)
The desired solution of the scattering and generation problem (13), (C1) – (C4) (or of the
equivalent problems (16) and (19)) are represented as in (15).According to (17) we determine
the values of complex amplitudes ascat scat
nκ , bnκ : n = 1, 2, 3 in (15) for the scattered and
generated fields by means of the formulas (20). The solution of (21) is obtained by means of
successive approximations using the self-consistent approach based on the iterative algorithm
(23).
In order to describe the scattering and generation properties of the nonlinear structure in the
zones of reflection z > 2πδ and transmission z < −2πδ, we introduce the following notation:
3 3
Rnκ := | ascat 2
∑ |ainc 2 scat 2
∑ |ainc 2

nκ | nκ | and Tnκ := |bnκ | nκ | , n = 1, 2, 3.
n =1 n =1
The quantities Rnκ , Tnκ are called reflection, transmission or generation coefficients of the waves
w.r.t. the total intensity of the incident packet.
We note that, for nonabsorbing media with Im ε( L) (z) = 0, the energy balance equation

3
∑

Rnκ + Tnκ = 1 (33)
n =1
is satisfied. This equation generalises the law of conservation of energy which has been
treated in [12], [21] for the case of a single incident field and a single equation. If we define
by
Wnκ := | ascat 2 scat 2

nκ | + | bnκ | (34)
the total energy of the scattered and generated fields at the frequencies nκ, n = 1, 2, 3, then
the energy balance equation (33) can be rewritten as
3 3
∑ Wnκ = ∑ |ainc 2
nκ | .
n =1 n =1
In the numerical experiments, the quantities W3κ /Wκ (which characterises the portion of
energy generated in the third harmonic in comparison to the energy scattered in the nonlinear
layer) and
3
W (Error) := 1 − ∑

Rnκ + Tnκ (35)
n =1
(which characterises the numerical violation of the energy balance) are of particular interest.
We emphasize that in the numerical simulation of scattering and generation processes
without any weak fields, i.e. ainc inc
2κ = a3κ = 0, the residual of the energy balance equation
(33) does not exceed the value |W (Error) | < 10−8 . However, taking into consideration the
impact of weak fields in the numerical simulation of the same scattering and generation
processes, i.e. ainc
nκ � = 0, n = 2, 3, the error in the balance equation (33) can reach up to
several percent. This indicates that the intensities of the exciting weak fields are sufficiently
large such that these fields become also sources for the generation of oscillations. For such
situations, the presented mathematical model (13), (C1) – (C4) and (24), (CS1) – (CS4) should
take into account the complex Fourier amplitudes of the oscillations at the frequencies nκ for
numbers n > 3. Furthermore we observe, on the one hand, situations in which the influence
of a weak field ainc
2κ � = 0 on the scattering and generation process of oscillations leads to small
errors in the energy balance equation (33) not exceeding 2% (that is |W (Error) | < 0.02), and,
on the other hand, situations in which the error can reach 10% (that is |W (Error) | < 0.1).
The scattering, generating, energetic, and dielectric properties of the nonlinear layer are
illustrated by surfaces in dependence on the parameters of the particular problem. The
bottom chart depicts the surface of the value of the residual W (Error) of the energy balance
equation (see (35)) and its projection onto the top horizontal plane of the figure. In particular,
by the help of these graphs it is easy to localise that region of parameters of the problem,
where the error of the energy balance does not exceed a given value, that is |W (Error) | < const.
The spectral characteristics of the linearised nonlinear problems (24), (CS1) – (CS4) with
the induced dielectric permittivity (11) at the frequencies nκ, n = 1, 2, 3, of excitation and
generation were calculated by means of the algorithm (30). In the graphical illustration of
the eigen-fields Uκn in the representation (26) we have set aκn := 1 for κn ∈ Ωnκ ⊂ Hnκ ,
n = 1, 2, 3. Finally we mention that the later-used classification of scattered, generated or
eigen-fields of the dielectric layer by the Hm,l,p -type is identical to that given in [12], [13],
[23]. In the case of E-polarisation, see (3), Hm,l,p (or TEm,l,p ) denotes the type of polarisation
of the wave field under investigation. The subscripts indicate the number of local maxima
of | E1 | (or |U |, as | E1 | = |U |, see (15), (26)) along the coordinate axes x, y and z (see Fig.
1). Since the considered waves are homogeneous along the x-axis and quasi-homogeneous
along the y-axis, we study actually fields of the type H0,0,p (or TE0,0,p ), where the subscript p
is equal to the number of local maxima of the function |U | of the argument z ∈ [−2πδ, 2πδ].
In what follows we present and discuss results of the numerical analysis of scattering and
generation properties as well as the eigen-modes of the dielectric layer with a positive value of
the cubic susceptibility of the medium. In more detail, we consider a single-layered structure
( NL)
with a dielectric permittivity ε nκ (z, α(z), U (κ; z), U (2κ; z), U (3κ; z)) = ε( L) (z) + ε nκ , n =
( NL)
1, 2, 3, where ε( L) (z) := 16 and α(z) := 0.01 for z ∈ [−2πδ, 2πδ], ε nκ is given by (11),
δ := 0.5, κ inc := κ := 0.375, and ϕκ ∈ [0◦ , 90◦ ). The Figs. 3 – Fig. 13 illustrate the following
cases of the incident fields:
1 inc inc
ainc
2κ = 3 aκ , a3κ = 0 ... graphs labeled by “1/3”,
2 inc inc
ainc
2κ = 3 aκ , a3κ = 0 ... graphs labeled by “2/3”,
ainc
2κ = a inc = 0
3κ ... graphs labeled by “0”.
1 inc 2 inc
Figure 3. The portion of energy generated in the third harmonic: ainc inc
2κ = 3 aκ (left), a2κ = 3 aκ (right)
Figure 4. The portion of energy generated in the third harmonic: #1 . . . ainc inc inc
κ = 1, #2 . . . aκ = 9.93, #3 . . . aκ = 14, #4 . . .
ainc inc inc inc
κ = 19 for a2κ = 0 (top left), the dependence of W3κ /Wκ on ϕκ , a2κ for aκ = 14 (top right), some graphs describing the
properties of the non-linear layer for ϕκ = 60◦ , ainc inc 1 inc inc 2 inc
κ = 14 and a2κ = 3 aκ (bottom left), a2κ = 3 aκ (bottom right): #0.0 . . .
ε( L) , #1 . . . |U (κ; z)|, #2 . . . |U (2κ; z)|, #3 . . . |U (3κ; z)|, # n.1 . . . Re(ε nκ ), # n.2 . . . Im(ε nκ )
The results shown in Fig. 3 allow us to track the dynamic behaviour of the quantity W3κ /Wκ
characterising the ratio of the generated and scattered energies. Fig. 3 shows the dependence
of W3κ /Wκ on the angle of incidence ϕκ and on the amplitude ainc κ of the incident field
for different relations between ainc inc
2κ and aκ . It describes the portion of energy generated in
the third harmonic by the nonlinear layer when a plane wave at the excitation frequency
κ and with the amplitude ainc κ is passing the layer under the angle of incidence ϕκ . It can
be seen that the weaker incident field at the frequency 2κ leads to an increase of W3κ /Wκ in
comparison with the situation where the structure is excited only by a single field at the basic
frequency κ. For example, in Fig. 3 the maximum value of W3κ /Wκ and the value W (Error) are
reached at the following parameters [ ainc inc
κ , a2κ , ϕκ ]: W3κ /Wκ = 0.08753, W
(Error) = −1.98292 ·
− 9 inc inc ◦
10 , [ aκ = 9.93, a2κ = 0, ϕκ = 53 ] . . . graph #0 and, taking into consideration the weak
field at the double frequency, W3κ /Wκ = 0.13903, W (Error) = −0.01692, [ ainc inc
κ = 9.93, a2κ =
1 inc
a , ϕ = 53 ◦ ] . . . graph #1/3 (top left); W /W = 0.03265, W (Error) = −8.53239 · 10−9 ,
3 κ κ 3κ κ
[ ainc inc ◦
κ = 8, a2κ = 0, ϕκ = 42 ] . . . graph #0 and, taking into consideration the weak field at the
double frequency, W3κ /Wκ = 0.1864, W (Error) = −0.04625, [ ainc inc 2 inc ◦
κ = 8, a2κ = 3 aκ , ϕκ = 50 ]
. . . graph #2/3 (top right); W3κ /Wκ = 0.25054, W (Error) − 10 inc
= −9.29243 · 10 , [ aκ = 14, a2κ =inc
0, ϕκ = 60◦ ] . . . graph #0 (bottom left and right) and, taking into consideration the weak
field at the double frequency, W3κ /Wκ = 0.26589, W (Error) = −0.00578, [ ainc inc
κ = 14, a2κ =
1 inc
a , ϕ = 60 ◦ ] . . . graph #1/3 (bottom left); W /W = 0.42662, W (Error) = −0.06037 [ ainc =
3 κ κ 3κ κ κ
13, ainc 2 inc ◦
2κ = 3 aκ , ϕκ = 60 ] . . . graph #2/3 (bottom right).
The numerical analysis of the processes displayed in Fig. 4 (top left) by the curves #3 in
the range of angles ϕκ ∈ (66◦ , 79◦ ) and #4 in the range of angles ϕκ ∈ (62◦ , 82◦ ) did not
lead to the convergence of the computational algorithm. The value W3κ /Wκ = 0.3558 for
ainc inc ◦
κ = 14, a2κ = 0 and ϕκ = 66 (see the graph #3 in Fig. 4 (top left)) indicates that W3κ
is 35.58% of Wκ . This is the maximal value of W3κ /Wκ that has been achieved in the case
of a single incident field at the basic frequency κ. Fig. 4 (top right) illustrates the portion of
energy generated in the third harmonic in dependence on the angle of incidence ϕκ and on
the amplitude ainc 2κ of the incident field at the double frequency. Here the maximum value
of W3κ /Wκ is reached at [ ainc = 14, ainc ◦
κ 2κ = 8, ϕκ = 60 ]. If the structure is excited by a
single field at the basic frequency κ only, then the portion of energy generated in the third
harmonic is ≈ 25.05%, i.e. for [ ainc inc ◦
κ = 14, a2κ = 0, ϕκ = 60 ] we have W3κ /Wκ = 0.25054
and W (Error) − 10
= −9.29243 · 10 . If we take into consideration the weak exciting field at
the double frequency, then we get for [ ainc inc ◦
κ = 14, a2κ = 8, ϕκ = 60 ] the following results:
W3κ /Wκ = 0.35084 and W (Error) = −0.0377 (which corresponds to a relative error of 3.78% in
the energy balance).
The bottom diagrams in Fig. 4 display some graphs characterising the scattering and
generation properties of the nonlinear structure. Graph #0.0 illustrates the value of the
linear part ε( L) = 16 of the permittivity of the nonlinear layered structure. Graphs # n.1
and # n.2 show the real and imaginary parts of the permittivities at the frequencies nκ,
n = 1, 2, 3. The figure also shows the absolute values |U (κ; z)|, |U (2κ; z)| of the amplitudes
of the full scattered fields at the frequencies of excitation κ, 2κ (graphs #1, #2) and |U (3κ; z)|
of the generated field at the frequency 3κ (graph #3). The values |U (nκ; z)| are given in the
nonlinear layered structure (|z| ≤ 2πδ) and outside it (i.e. in the zones of reflection z > 2πδ
and transmission z < −2πδ). Here W (Error) = −5.782328 · 10−3 , i.e. the error in the energy
balance is less than 0.58% (bottom left) and W (Error) = −4.567534 · 10−2 , i.e. the error in the
energy balance is less than 4.57% (bottom right).
Figs. 5 and 6 show the numerical results obtained for the scattered and the generated fields
in the nonlinear structure and for the residual W (Error) of the energy balance equation (33)
for an incident angle ϕκ = 60◦ in dependence on the amplitudes ainc inc
κ and a2κ of the plane
incident waves at the basic frequency κ and at the double frequency 2κ, resp. The figures
show the graphs of Unκ ainc inc

κ , a2κ , z , n = 1, 2, 3, demonstrating the dynamic behaviour

of the scattered and the generated fields |U (nκ; z)| in the nonlinear layered structure in
dependence on increasing amplitudes ainc inc ◦
κ and a2κ for an incident angle ϕκ = 60 of the
inc inc 1 inc
plane waves. We mention that, in the range aκ ∈ (0, 22] and a2κ = 3 aκ (see Fig. 5) or
2 inc
ainc
2κ = 3 aκ (see Fig. 6) of the amplitudes of the incident fields and for an incident angle
ϕκ = 60◦ of the plane waves, the scattered field has the type H0,0,4 at the frequency κ and
H0,0,7 at the frequency 2κ. The generated field, observed in the range ainc κ ∈ [5, 22], is of the
type H0,0,10 at the frequency 3κ, see Figs. 5, 6 (bottom left).
( NL)
The nonlinear parts ε nκ of the dielectric permittivity at each frequency nκ depend on
the values Unκ := U (nκ; z), n = 1, 2, 3, of the fields. The variation of the nonlinear parts
( NL)
ε nκ of the dielectric permittivity for increasing amplitudes ainc and ainc of the incident
inc inc κ 2κ
fields are illustrated by the behaviour of Re ε nκ aκ , a2κ , z and Im ε nκ ainc inc

κ , a2κ , z at
1 inc inc = 2 ainc ). The quantities
the frequencies nκ in Fig. 7 (case ainc
2κ = 3 a κ ) and Fig. 8 (case a 2κ 3 κ
Im (ε nκ ) take both positive and negative values along the height of the nonlinear layer (i.e.
Figure 5. Graphs of the scattered and generated fields in the nonlinear layered structure in dependence on ainc inc

κ , a2κ , z for
ϕκ = 60◦ and ainc
2κ = 1 inc
a
3 κ : | Uκ | (top left), | U2κ | (top right) | U3κ | (bottom left), and the residual W (Error)
of the energy balance
equation (bottom right)
in the interval z ∈ [−2πδ, 2πδ]), see Figs. 7, 8 (right). For given amplitudes ainc inc
κ and a2κ ,
the graph of Im ε nκ ainc inc , z

κ , a 2κ characterises the loss of energy in the nonlinear layer at the
excitation frequencies nκ, n = 1, 2, caused by the generation of the electromagnetic field of
the third harmonic. Such a situation arises because of the right-hand side of (13) at the triple
frequency and the generation which is evoked by the right-hand side of (13) at the basic
frequency. In our case Im ε( L) (z) = 0 and Im [α (z)] = 0, therefore, according to (11),
Im(εnκ (z, α(z), U (κ; z), U (2κ; z), U (3κ; z)))

= α(z) δn1 |U (κ; z)||U (3κ; z)|Im (exp {i [−3arg(U (κ; z)) + arg(U (3κ; z))]})
+ δn2 |U (κ; z)||U (3κ; z)|Im (exp {i [−2arg(U (2κ; z)) + arg(U (κ; z)) + arg(U (3κ; z))]}) ,
n = 1, 2, 3.
(36)
From Figs. 7, 8 (right) we see that small values of ainc inc

κ and a2κ induce a small amplitude of
inc
the function Im(ε nκ ), i.e. |Im(ε nκ )| ≈ 0. The increase of aκ corresponds to a strong incident
field and leads to the generation of a third harmonic field U (3κ; z), and the increase of ainc2κ
changes the behaviour of ε nκ (compare the surface #0 with the surfaces #1/3 and #2/3 in
Figs. 7, 8). Figs. 7, 8 (right) show the dynamic behaviour of Im(ε nκ ). It can be seen that
Im(ε 3κ ) = 0, whereas at the same time the values of Im(ε nκ ), n = 1, 2, may be positive or
negative along the height of the nonlinear layer, i.e. in the interval z ∈ [−2πδ, 2πδ], see (36).
Figure 6. Graphs of the scattered and generated fields in the nonlinear layered structure in dependence on ainc inc

κ , a2κ , z for
◦ inc 2 inc
ϕκ = 60 and a2κ = 3 aκ : |Uκ | (top left), |U2κ | (top right) |U3κ | (bottom left), and the residual W (Error)
(bottom right)
The zero values of Im(ε nκ ), n = 1, 2, are determined by the phase relations between the
scattered and the generated fields in the nonlinear layer, namely at the basic frequency κ by
the phase relation between U (κ; z), U (3κ; z), and at the double frequency 2κ by the phases of
{U (nκ; z)}n=1,2,3 , see (36):
δn1 [−3arg(U (κ; z)) + arg(U (3κ; z))]

+ δn2 [−2arg(U (2κ; z)) + arg(U (κ; z)) + arg(U (3κ; z))] = pπ, p = 0, ±1, . . . , n = 1, 2.
We mention that the behaviour of both the quantities Im(ε nκ ) and
Re(εnκ (z, α(z), U (κ; z), U (2κ; z), U (3κ; z)) − ε 3κ (z, α(z), U (κ; z), U (2κ; z), U (3κ; z)))
= α(z) δn1 |U (κ; z)||U (3κ; z)|Re (exp {i [−3arg(U (κ; z)) + arg(U (3κ; z))]})
+ δn2 |U (κ; z)||U (3κ; z)|Re (exp {i [−2arg(U (2κ; z)) + arg(U (κ; z)) + arg(U (3κ; z))]}) ,
n = 1, 2,
(37)
plays an essential role in the process of third harmonic
generation. Figs.
inc7, inc
8 (bottom)
show the graphs describing the behaviour of Re ε κ ainc inc , z − ε

κ , a 2κ 3κ a κ , a 2κ , z and
inc inc inc inc
Re ε 2κ aκ , a2κ , z − ε 3κ aκ , a2κ , z .
Figure 7. Graphs characterising the nonlinear dielectric permittivity in dependence on ainc inc ◦ inc

κ , a2κ , z for ϕκ = 60 and a2κ =
1 inc
a
3 κ : Re ( ε κ ) (top left), Im ( ε κ ) (top right), Re ( ε 2κ ) (second from top left), Im ( ε 2κ ) (second from top right), Re (ε 3κ )
(second to the last left), Im (ε 3κ ) (second to the last right), Re (ε κ − ε 3κ ) (bottom left), Re (ε 2κ − ε 3κ ) (bottom right)
We mention that the impact of a strong electromagnetic field with an amplitude ainc κ even in
the absence of a weak field ainc
2κ = 0 (where U (2κ; z ) = 0, see (11) and the surface #0 in Fig. 5
Figure 8. Graphs characterising the nonlinear dielectric permittivity in dependence on ainc inc ◦ inc

κ , a2κ , z for ϕκ = 60 and a2κ =
2 inc
3 aκ : Re ( ε κ ) (top left), Im ( ε κ ) (top right), Re ( ε 2κ ) (second from top left), Im ( ε 2κ ) (second from top right), Re ( ε 3κ )
(second to the last left), Im (ε 3κ ) (second to the last right), Re (ε κ − ε 3κ ) (bottom left), Re (ε 2κ − ε 3κ ) (bottom right)
and Fig. 6 (top right)) induces a nontrivial component of the nonlinear dielectric permittivity
at the frequency 2κ. Figs. 7, 8 (second from top) show that the existence of nontrivial values

Re (ε 2κ ) �= Re ε( L) and Im (ε 2κ ) �= 0 is caused by the amplitude and phase characteristics
of the fields U (κ; z) and U (3κ; z), see (11) taking into account U (2κ; z) = 0. Moreover, the
nonlinear component of the dielectric permittivity, which is responsible for the variation of
Re (ε nκ − ε 3κ ) and Im (ε nκ ) , does not depend on the absolute value of the amplitude of the
field at the double frequency |U (2κ; z)|, see (37) and (36). Thus, even a weak field (see (11),
#1/3 in Fig. 5 and #2/3 in Fig. 6 (top right)) includes a mechanism for the redistribution of
the energy of the incident wave packet which is consumed for the scattering process and the
generation of waves, cf. the dynamics of the surfaces #0 with #1/3 and #2/3 in Figs. 5 – 8.
1 inc
Figure 9. The scattering and generation properties of the nonlinear structure in dependence on ϕκ , ainc inc inc

κ , a2κ for a2κ = 3 aκ :
Rκ , Tκ (top), R2κ , T2κ (second from top), R3κ , T3κ (bottom)
The scattering and generation properties of the nonlinear structure in the ranges ϕκ ∈
[0◦ , 90◦ ), ainc
κ ∈ [1, 9.93] , ainc 1 inc ◦
2κ = 3 aκ and ϕκ ∈ [0 , 90 ), aκ
◦ inc ∈ [1, 8] of the parameters
of the incident field are presented inincFigs. 9 – 11. Theincgraphs show

theincdynamics of the
scattering (Rκ ϕκ , ainc inc inc inc inc

κ , a2κ , Tκ ϕκ , aκ , a2κ , R2κ ϕκ , aκ , a2κ , T2κ ϕκ , aκ , a2κ , see Figs.
2 inc

κ , a2κ for a2κ = 3 aκ :
Rκ , Tκ (top), R2κ , T2κ (second from top), R3κ , T3κ (bottom)
9, 10 (top 2)) and generation (R3κ ϕκ , ainc inc inc inc

κ , a2κ , T3κ ϕκ , aκ , a2κ , see Figs. 9, 10 (bottom))
properties of the structure. Fig.
11 (top) showscrossinc
sections
of the surfaces depicted in Fig.
9 and of the graph of W3κ ϕκ , ainc inc inc
κ , a2κ /Wκ ϕκ , aκ , a2κ (see Fig. 3 (left)) by the planes
ϕκ = 60◦ and ainc
κ = 9.93.
In Figs. 11 – 13, a slightly more detailed illustration for the situation of a single incident
field (i.e. ainc
2κ = 0) is given, cf. also the graphs labeled by “0” in Fig. 9. In the resonant
range of wave scattering and generation frequencies, i.e. κ scat := κ inc = κ and κ gen = 3κ,
resp., the dynamic behaviour of the characteristic quantities depicted in Figs. 11 – 13 has the
following causes. The scattering and generation frequencies are close to the corresponding
eigen-frequencies of the linear (α = 0) and linearised nonlinear (α �= 0) spectral problems
(24), (CS1) – (CS4). Furthermore, the distance between the corresponding eigen-frequencies
of the spectral problems with α = 0 and α �= 0 is small. Thus, the graphs in Fig. 11 (top)
Figure 11. The curves Rκ (#1), Tκ (#2), R2κ (#3), T2κ (#4), R3κ (#5), T3κ (#6), W3κ /Wκ (#7) for ϕκ = 60◦ (top left) and ainc
κ = 9.93
( L)
(top right), the curves κ := κ inc := 0.375 (#1), 3κ = κ gen = 3κ inc = 1.125 (#2), the complex eigen-frequencies Re(κ1 ) (#3.1),
( L) ( L) ( L) ( NL) ( NL)
Im(κ1 ) (#3.2), Re(κ3 ) (#4.1), Im(κ3 ) (#4.2) of the linear problem (α = 0) and Re(κ1 ) (#5.1), Im(κ1 ) (#5.2),
( NL) ( NL)
Re(κ3 ) (#6.1), Im(κ3 ) (#6.2) of the linearised nonlinear problem (α = +0.01) for ϕκ = 60◦ (second from top left) and
ainc ◦ inc
κ = 9.93 (second from top right), and the graphs of the eigen-fields of the layer for ϕκ = 60 , aκ = 14. The linear problem
( L) ( L) ( L) ( L)
(α = 0, bottom left ): |U (κ1 ; z)| with κ1 = 0.3829155 − i 0.01066148 (#1), |U (κ3 ; z)| with κ3 = 1.150293 − i 0.01062912
( NL) ( NL)
(#2), the linearised nonlinear problem (α = +0.01, bottom right): |U (κ1 ; z)| with κ1 = 0.3705110 − i 0.01049613 (#1),
( NL) ( NL)
|U (κ3 ; z)| with κ3 = 1.121473 − i 0.009194824 (#2)
can be compared with the dynamic behaviour of the branches of the eigen-frequencies of the
spectral problems presented in Fig. 11 (second from top). The graphs of the eigen-fields
corresponding to the branches of the considered eigen-frequencies are shown in Fig. 11
(bottom).
Fig. 11 (second from top) illustrates the dispersion characteristics of the linear (α = 0) and
( NL
the linearised nonlinear (α = +0.01) layer ε nκ = ε( L) + ε nκ , n = 1, 2, 3. The nonlinear
components of the permittivity at the scattering (excitation) frequencies κ scat := κ inc = κ
and the generation frequencies κ gen := 3κ depend on the amplitude aincκ and the angle of
Figure 12. The scattering and generation properties of the nonlinear structure in dependence on ϕκ , ainc inc ◦

κ , a2κ for ϕκ = 60 :
Rκ , Tκ (#1.1, #1.2 top left), R2κ , T2κ (#2.1, #2.2 top right), W3κ /Wκ , R3κ , T3κ (#3, #3.1, #3.2 bottom left), W (Error) (bottom
right)
incidence ϕκ of the incident field. This is reflected in the dynamics of the behaviour of the
complex-valued eigen-frequencies of the linear and the linearised nonlinear layer.
We start the analysis of the results of our calculations with the comparison of the dispersion
relations given by the branches of the eigen-frequencies (curves #3.1, #3.2 and #5.1, #5.2) near
the scattering frequency (curve #1, corresponding to the excitation frequency) and (curves
#4.1, #4.2, #6.1, #6.2) near the oscillation frequency (line #2) in the situations presented in Fig.
11 (second from top). The graph #5.1 lies below the graph #3.1 and the graph #6.1 below the
graph #4.1. That is, canalising properties (properties of transparency) of the nonlinear layer
occur if α > 0. This case is characterised by the increase of the angle of transparency of the
nonlinear structure at the excitation frequency with an increasing amplitude of the incident
field (see Fig. 9 (top left), there where the reflection coefficient is close to zero). The analysis
of the eigen-modes of Fig. 11 (second from top) allows us to explain the mechanisms of the
canalisation phenomena (transparency) (see Fig. 9 (top left), Fig. 11 (top, graph #1)) and wave
generation (see Fig. 9 (bottom), Fig. 11 (top, graphs #5, #6)).
Comparing the results shown in Fig. 11 (top) and Fig. 11 (second from top) we note the
following. The intersection of the curves #1 and #5.1 in Fig. 11 (second from top) defines
certain parameters, in the neighborhood of which the canalisation effect (transparency) of
the nonlinear structure can be observed in Fig. 11 (top). For example, in Fig. 11 (second from
top left) the curves #1 and #5.1 intersect at ainc
κ = 9.5, also here the curve #5.2 achieves a local
maximum. Near this value, we see the phenomenon of canalisation (transparency) of the
layer in Fig. 11 (top left). If we compare the Figs. 11 (top right) and 11 (second from top right),
we detect a similar situation. The intersection of the curves #1 and #5.1 defines the parameter
ϕκ = 64◦ , near which we observe the canalisation effect in Fig. 11 (second from top right).
The same is true – to some extent – for the description of the wave generation processes. For
example, for similar values of the imaginary parts of the branches of the eigen-frequencies
#5.2 and #6.2 in Fig. 11 (second from top right), the intersection of the curves #2 and #6.1
defines the parameter ϕκ = 45◦ . Near this value, stronger generation properties of the layer
can be observed, see Fig. 11 (top) and Fig. 9 (bottom), at ϕκ = 45◦ . Let us also consider the
situation in Fig. 11 (second from top left). Here, at the point of intersection of the curves #2
and #6.1, the graph #5.2 starts to decrease monotonically in some interval. The intersection of
the curves #2 and #6.1 defines the parameter ainc κ = 12.6, which falls into the range [9.5, 13.6]
of values of the amplitudes at which the curve #5.2 is monotonically decreasing. This leads
to a shift in the imaginary part of the eigen-frequency of the scattering structure (graph #5.2)
with respect to the eigen-frequency of the generating structure (graph #6.2). The magnitude
of the shift depends on the distance between the curves of #6.2 and #5.2 at the given value
ainc
κ . The maximal distance between the graphs #6.2 and #5.2 is achieved at the local minimum
of the graph #5.2 at ainc inc
κ = 13.6. Right from this point, i.e. with an increasing amplitude aκ ,
the distance between the graphs #6.2 and #5.2 shows no significant change. The maximum
value of the generation is achieved at an amplitude close to the intersection of curves #2 and
#6.1, but shifted to the point of the local minimum of the curve #5.2, see R3κ , T3κ , W3κ /Wκ in
Fig. 11 (top left), Fig. 9 (bottom) and Fig. 3 (top right).

κ , a2κ for aκ = 9.93:
Rκ , Tκ (#1.1, #1.2 top left), R2κ , T2κ (#2.1, #2.2 top right), W3κ /Wκ , R3κ , T3κ (#3, #3.1, #3.2 bottom left), W (Error) (bottom
right)
Fig. 11 (bottom) presents the characteristic distribution of the eigen-fields corresponding to

the branches of the eigen-frequencies under consideration. The graphs of the eigen-fields of
type H0,0,4 are labeled by #1, the graphs of the eigen-fields of type H0,0,10 by #2.
Figs. 12, 13 show the same dependencies as in Fig. 11 (top) but with the additional parameter
ainc
2κ . Here we can track the dynamics of the scattering, generation and energy characteristics
of the nonlinear layer under the influence of the wave package. The incident package consists
of a strong and a weak magnetic field with amplitudes ainc inc
κ and a2κ , resp.
The numerical results presented in this chapter were obtained using an approach based on
the description of the wave scattering and generation processes in a nonlinear, cubically
polarisable layer by a system of nonlinear integral equations (19), and of the corresponding
spectral problems by the nontrivial solutions of the integral equations (28). We have
considered an excitation of the nonlinear layer defined by the condition (32). For this case
we passed from (19) to (21) and from (28) to (30) by the help of Simpson’s quadrature rule.
The numerical solution of (21) was obtained using the self-consistent iterative algorithm (23).
The problem (30) was solved by means of Newton’s method. In the investigated range of
parameters, the dimension of the resulting systems of algebraic equations was N = 301, and
the relative error of calculations did not exceed ξ = 10−7 .
6. Conclusion
We presented a mathematical model of resonance scattering and generation of waves on
an isotropic nonmagnetic nonlinear layered dielectric structure excited by a packet of plane
waves in a self-consistent formulation, where the analysis is performed in the domain of
resonance frequencies [3, 4, 15]. Here, both the radio [6] and optical [9] frequency ranges are
of interest. The wave packets consist of both strong electromagnetic fields at the excitation
frequency of the non-linear structure (leading to the generation of waves) and of weak
fields at the multiple frequencies (which do not lead to the generation of harmonics but
influence on the process of scattering and generation of waves by the non-linear structure).
The model reduces to a system of nonlinear boundary-value problems of Sturm-Liouville
type or, equivalently, to a system of nonlinear Fredholm integral equations. The solution
of these nonlinear problems was obtained rigorously in a self-consistent formulation and
without using approximations of the given field, slowly varying amplitudes etc.
The approximate solution of the nonlinear problems was obtained by means of solutions
of linear problems with an induced nonlinear dielectric permeability. The analytical
continuation of these linear problems into the region of complex values of the frequency
parameter allowed us to switch to the analysis of spectral problems. In the frequency domain,
the resonant scattering and generation properties of nonlinear structures are determined
by the proximity of the excitation frequencies of the nonlinear structures to the complex
eigen-frequencies of the corresponding homogeneous linear spectral problems with the
induced nonlinear dielectric permeability of the medium.
We presented a collection of numerical results that describe interesting properties of the
nonlinear permittivities of the layers as well as their scattering and generation characteristics.
The results demonstrate the possibility to control the scattering and generating properties of
a nonlinear structure via the intensities of its excitation fields. They also indicate a possibility
of designing a frequency multiplier and other electrodynamic devices containing nonlinear
dielectrics with controllable permittivity.
Author details
Lutz Angermann1 and Vasyl V. Yatsyk2
1 Institut für Mathematik, Technische Universität Clausthal, Clausthal-Zellerfeld, Germany
2 Usikov Institute of Radiophysics and Electronics, National Academy of Sciences of Ukraine,
Kharkov, Ukraine
References
[1] Angermann, L. & Yatsyk, V. [2008]. Numerical simulation of the diffraction of weak
electromagnetic waves by a Kerr-type nonlinear dielectric layer, Int. J. Electromagnetic
Waves and Electronic Systems 13(12): 15–30.
[2] Angermann, L. & Yatsyk, V. [2010]. Mathematical models of the analysis of processes of
resonance scattering and generation of the third harmonic by the diffraction of a plane
wave through a layered, cubically polarisable structure, Int. J. Electromagnetic Waves and
Electronic Systems 15(1): 36–49. In Russian.
[3] Angermann, L. & Yatsyk, V. [2011a]. Generation and resonance scattering of waves on
cubically polarisable layered structures, in L. Angermann (ed.), Numerical Simulations –
Applications, Examples and Theory, InTech, Rijeka/Vienna, Croatia/Austria, pp. 175–212.
[4] Angermann, L. & Yatsyk, V. [2011b]. Resonance properties of scattering and generation
of waves on cubically polarisable dielectric layers, in V. Zhurbenko (ed.), Electromagnetic
Waves, InTech, Rijeka/Vienna, Croatia/Austria, pp. 299–340.
[5] Butcher, P. [1965]. Nonlinear optical phenomena, Bulletin 200, Ohio State University,
Columbus.
[6] Chernogor, L. [2004]. Nonlinear Radiophysics, V.N. Karazin Kharkov National University,
Kharkov.
[7] Kleinman, D. [1962]. Nonlinear dielectric polarization in optical media, Phys. Rev.
126(6): 1977–1979.
[8] Kravchenko, V. & Yatsyk, V. [2007]. Effects of resonant scattering of waves by

layered dielectric structure with Kerr-type nonlinearity, Int. J. Electromagnetic Waves and
Electronic Systems 12(12): 17–40.
[9] Miloslavsky, V. [2008]. Nonlinear Optics, V.N. Karazin Kharkov National University,
Kharkov.
[10] Schürmann, H. W., Serov, V. & Shestopalov, Y. [2001]. Reflection and transmission of a
TE-plane wave at a lossless nonlinear dielectric film, Physica D 158: 197–215.
[11] Serov, V., Schürmann, H. & Svetogorova, E. [2004]. Integral equation approach to
reflection and transmission of a plane te-wave at a (linear/nonlinear) dielectric film
with spatially varying permittivities, J. Phys. A: Math. Gen. 37: 3489–3500.
[12] Shestopalov, V. & Sirenko, Y. [1989]. Dynamical Theory of Gratings, Naukova, Dumka,
Kiev.
[13] Shestopalov, V. & Yatsyk, V. [1997]. Spectral theory of a dielectric layer and the Morse
critical points of dispersion equations, Ukrainian J. of Physics 42(7): 861–869.
[14] Shestopalov, Y. & Yatsyk, V. [2007]. Resonance scattering of electromagnetic waves by a

Kerr nonlinear dielectric layer, Radiotekhnika i Elektronika (J. of Communications Technology
and Electronics) 52(11): 1285–1300.
[15] Shestopalov, Y. & Yatsyk, V. [2010]. Diffraction of electromagnetic waves by a layer filled
with a Kerr-type nonlinear medium, J. of Nonlinear Math. Physics 17(3): 311–335.
[16] Sirenko, Y., Shestopalov, V. & Yatsyk, V. [1985]. Elements of the spectral theory of
gratings, Preprint 266, IRE NAS Ukraine, Kharkov.
[17] Sirenko, Y. & Ström, S. (eds) [2010]. Modern Theory of Gratings. Resonant Scattering:
Analysis Techniques and Phenomena, Springer-Verlag, New York. Springer Series in
Optical Sciences, Vol. 153.
[18] Sirenko, Y., Ström, S. & Yashina, N. [2007]. Modeling and Analysis of Transient Processes
in Open Resonant Structures. New Methods and Techniques., Springer-Verlag, New York.
[19] Smirnov, Y., Schürmann, H. & Shestopalov, Y. [2005]. Propagation of TE-waves in

cylindrical nonlinear dielectric waveguides, Physical Review E 71: 0166141–10.
[20] Sukhorukov, A. P. [1988]. Nonlinear wave interactions in optics and radio physics, Nauka,
Moskva. (In Russian).
[21] Vainstein, L. [1988]. Electromagnetic Waves, Radio i Svyas, Moscow. In Russian.
[22] Vinogradova, M., Rudenko, O. & Sukhorukov, A. [1990]. Wave Theory, Nauka, Moscow.
[23] Yatsyk, V. [2000]. A constructive approach to construction of local equations of irregular

dispersion and evolution of fields in a quasi-homogeneous electrodynamic structure,
Usp. Sovr. Radioelektroniki 10: 27–44. Translated in: Telecommunications and Radio
Engineering, 56(8&9): 89-113, 2001.
[24] Yatsyk, V. [2006]. Diffraction by a layer and layered structure with positive and negative
susceptibilities of Kerr-nonlinear media, Usp. Sovr. Radioelektroniki 8: 68–80.
[25] Yatsyk, V. [2007]. About a problem of diffraction on transverse non-homogeneous

dielectric layer of Kerr-like nonlinearity, Int. J. Electromagnetic Waves and Electronic
Systems 12(1): 59–69.
Provisional chapter
Chapter 12
Matter Wave Interferometry,

Matter Wave Interferometry, the Gouy Phase and
the Gouy Phase and Complementarity Principle
Complementarity Principle
Luis A. Cabral, Irismar G. da Paz,

Luis A. Cabral, Irismar G. da Paz,
José G. G. de Oliveira Júnior, Saulo Moreira,
José G.D.
Marcos G.R.de Oliveira
Sampaio andJúnior,
Maria Saulo Moreira,
C. Nemes
Marcos D. R. Sampaio and Maria C. Nemes
1. Introduction
Over a hundred years ago L. G. Gouy described an anomalous behavior of the phase of a
converging diffracted spherical wave as it passes through a focus [1, 2]. He also performed
the experiment which confirmed his theoretical analysis. He considered a point source and
let it impinge onto two mirrors with different curvature radius, one concave of other plane,
parallel to each other. The observational plane was transverse to the so generated beams.
Their interference yielded circular interference fringes. Since then many observations of
such phase anomaly have been reported, some of them in [3–12].
It turns out that the Gouy phase has found several important applications. As one
example we quote the influence of this anomalous phase as being of critical importance
for any application as ultrashort laser pulses, including high-harmonics and attosecond
pulse generation [13]. It also has important consequences in the optical range of the
electromagnetic spectrum such as phase matching in high order harmonic generation
governed by atomic response and the Gouy phase of the fundamental radiation [14–16].
On the theoretical side numerous authors have treated the problem an explanation for such
phase. One of the first authors to consider the question was P. Debye in 1909, who found an
exact solution to the wave equation for all space treating the boundary condition in an elegant
way, avoiding its mathematical complication [17]. The anomalous propagation through the
focus has been treated by several authors since then. In refs [18, 19] one also finds early
rather involved theories explaining the phase anomaly.
Interesting enough even nowadays several authors still consider the Gouy phase as "the most
poorly understood subjects in physical optics" [20], although several simple explanation have
been offered in terms of Berry geometrical phase [4], the uncertainty principle [10] and so
one [5, 6, 21–25].
A successful description of optical waves crossing an aperture is the so called paraxial
approximation to the Helmholtz equation (explained below). A physically appealing aspect
of this equation is its complete correspondence to Schrödinger equation for matter waves.
However in the latter case the corresponding phase is a time dependent overall phase and
therefore omitted in quantum mechanics text books. In spite of that it is manifest and in
fact the agent responsible for the growth of the width gaussian wave packet under a free
evolution. One of the objectives of the present contribution is to unveil the similarities
between these to apparently independent subjects, which, however bear a strong conceptual
basis, which has been recently show [9, 10]. Also an experimental proposal was put forth
for the observation of the Gouy phase of matter waves in the optical regime using Ramsey
interferometry with Rydberg atoms [12].
In fact the basis for the interpretation of the anomalous phase of matter waves is the
Schrödinger Robertson uncertainty relation which can be immediately carried over for
laser light. As will be shown here the Gouy phase in both cases is related to the x − p
cross correlations, i.e., the expectation value of the position-symmetrized-momentum. It
becomes thus clear that the phase is intimately connected to the correlations generated by
the uncertainty relation.
Our second purpose, which is a rather novel investigation relates to the role of the Gouy
phase in interference phenomena. Here in particular we restrict ourselves to two slit
experiments which, again, are equally both for light or matter.
The first theoretical to be faced is the bridge between the complementarity principle as
usually constructed using q-bits and their equivalent for continuous systems. We define
and explore the constructed definition for visibility and predictability, showing, in particular
that the Gouy phase will be manifest in the number of fringes of a given arrangement. Thus
by changing the slit aperture of the grating and measuring the corresponding change in the
number of fringes on the screen gives an indirect evidence of this phase. Again the idea that
the Gouy phase is at the root of this phenomena is not exclusive of matter or light waves,
since it has only to do with wave nature.
In section 2 we make the analogy between the paraxial wave equation for light and
the Schrödinger’s equation for a free particle. In section 3 we discuss the generalized
complementarity principle within a double slit experiment. We quantify the wave-particle
duality through the definition of quantities called visibility and predictability which
characterize the wave and particle behavior respectively [26, 32]. Section 4 is reserved to
a wave quantitative analogies of interferometry of matter waves in double slit experiment
showing that the Gouy phase is related to number of fringes. Moreover we show that
the Gouy phase can be indirectly extracted by analyzing the change in the interference
fringes with the variation of the slit width or the distance of the detector. The results
that we will obtain here for matter waves can be equally applied for classical and quantum
electromagnetic waves in the paraxial regime. In section 5 we make our conclusions.
2. From Maxwell to matter: Waves the Gouy phase

We proceed to exploring the analogies between properties of electromagnetic waves in the
paraxial approximation and matter waves. In Fig. 1 we compare plane, spherical and
Gaussian wavefronts. The latter is obtained when a converging spherical wave is obstructed
by a circular aperture. Notice that a Gaussian wavefront acquires an anomalous phase shift
as it passes through the focal region, i.e., their wavefront are slighted advanced in relation to
the others. This phase shift is the Gouy phase.
Matter Wave Interferometry, the Gouy Phase and Complementarity Principle 335
Figure 1. Wavefronts: a) plane wave, b) spherical wave, c) Gaussian beam. Extracted from [ref. B. E. A. Saleh and M. Teich,
Fundamentals of Photonics (John Wiley Sons, New York, 1991).]
Consider a stationary electric field in vacuum
E(r ) = A(r ) exp(ikz), (1)
where k is the wave number. In the paraxial approximation we assume that the modulus
of the wave vector |k| ≈ k z , i.e., k z ≫ k x , k y . This amounts to saying that the function A(r )
which modulates the field varies slowly compared with the wavelength λ = 2π/k. Under
this condition the Helmholtz equation for the electric field is [25]
∂2 ∂2

1 ∂
+ + i4π A ( x, y, z) = 0, (2)
∂x2 ∂y2 λ L ∂z
where λ L is the light wavelength.

On the other hand the two-dimensional Schrödinger equation for a free particle of mass m
reads
2
∂2

∂ m ∂
+ + 2i ψ( x, y, t) = 0. (3)
∂x2 ∂y2 h̄ ∂t
Notice that snapshots of ψ( x, y, ti ) is identical to slices for fixed zi of A( x, y, zi ). Therefore

we can interpret the time evolution of ψ( x, y, t) in terms of parameter z related to time as
t = z/vz , where vz can be interpreted in terms of the de Broglie wavelength λ P = h/mvz as
we assume p ∼ pz .
Based on this discussion Equation (3) can formally written as [7–9]
∂2 ∂2

1 ∂
2
+ 2 + i4π ψ ( x, y, t = z/vz ) = 0. (4)
∂x ∂y λ P ∂z
Next we show that the Gouy phase is intimately connected to the generalized
Schrödinger-Robertson uncertainty relation [7–10] which provides for a physically appealing
interpretation of this phase which can be carried over to light waves in paraxial
approximation.
The solution of Equation (2) for a transverse Gaussian beam collimated in the z direction is
x2 kx2

w0
AG ( x, z) = A0 exp − + i + iξ ( z ) (5)
w(z) w ( z )2 2R(z)
where we dropped the y component for brevity. In the equation above, w(z) is the beam
width, R(z) is the curvature radius, ξ (z) is the Gouy phase and z0 is the Rayleigh range
given by
2 21 z 2
z
w ( z ) = w0 1 + , R(z) = z 1 + 0 , (6)
z0 z
kw02

1 z
ξ (z) = − arctan , z0 = . (7)
2 z0 2
The total variation of the Gouy phase as we go from z = −∞ to z = +∞ is π/2 as we can
immediately verify from (7). This change is abrupt and effectively takes place in the so called
Rayleigh range z0 . On the other hand, solution of Equation (4) for an initial condition given
by a Gaussian wave packet is given by [7–9]
x2 mx2

1
ψ ( x, t) = √ exp − + i + iµ ( t ) . (8)
πB (t) 2B(t)2 2h̄R(t)
From the solutions (5) and (8) we can make the following identifications
2 12
t
w (z) −→ B (t) = b0 1 + , (9)
τ0
τ 2
R (z) −→ R (t) = t 1 + 0 , (10)
t

1 t
ξ (z) −→ µ (t) = − arctan , (11)
2 τ0
and
mb02
z0 −→ τ0 = . (12)
h̄
To get a better insight inter the role played by τ0 (or equivalently the Rayleigh range z0 ) is
convenient to rewrite (8) as
ψ( x, t) = ϕ̃( x, t) exp[iS( x, t)], (13)

where
x2

1
ϕ̃( x, t) = √ exp − , (14)
πB(t) 2B(t)2
and
x2 t
S( x, t) = + µ ( t ). (15)
2B(t)2 τ0
Notice that the position dependence phase S contains the time scale τ0 . the ratio t/τ0 will
determine the importance of this x-dependent phase to the interference pattern. In the
experimental setups using fullerene molecules [36] t/τ0 ≈ 104 which is also the condition
for Frauhoffer diffraction (see ref. [27]). The time scale τ0 is fundamentally determined √ by
Heisenberg’s uncertainty relation, given the initial position dispersion
√ ∆x ( 0 ) = σ0 / 2. In
fact, the corresponding momentum dispersion is ∆p = h̄/(σ0 2). Because the momentum
is a constant of motion this momentum spread will be preserved in time. Both ∆x and ∆p
constitute intrinsic properties of the initial wave packet, in terms of which the time scale τ0 is
expressed as
∆x (0)
τ0 = . (16)
(∆p)/m
The numerator in the above relation represents the spatial dimensions of the initial wave
packet, whilst the denominator stands for the scale of velocity difference enforced by the
uncertainty principle. Therefore the time scale τ0 corresponds essentially to the time during
which a distance of the order of the wave packet extension is traversed with a speed
corresponding to the dispersion in velocity. It can therefore be viewed as a characteristic
time for the "aging“ of the initial state, which consists in components with larger velocities
(relatively to the group velocity of the wave packet) concentrating at the frontal region of the
packet. This can be seen explicitly by deriving the velocity field associated with the phase S
in Equation (15), which reads
h̄ ∂S tx
v( x, t) = = 2 . (17)
m ∂x τ0 + t2
This expression shows that for t > 0 the initial velocity field v( x, 0) = 0 varies linearly with
respect to the distance from center of the wave packet (x = 0).
Next we relate quantitatively this "aging" effect to position-momentum correlations. This
is readily achieved using the generalized uncertainty relation devised by Schrödinger [28],
which is expressed in this case in terms of the determinant of the covariance matrix
h̄2

σxx σxp
det ∑ ≡ det ≥ . (18)
σxp σpp 4
For the minimum uncertainty wave packet of Equation (8) we obtain, at all times,
h̄2

σxx σxp
det = . (19)
σxp σpp 4
Therefore we establish a direct relation between the Gouy phase and the position-momentum
correlation for matter waves,
h̄t h̄
σxp = = − tan 2µ(t). (20)
2τ0 2
For light waves a similar interpretation applies,
z 1
σxk x = = − tan 2ζ (z). (21)
2z0 2
The formulation of the generalized uncertainty relation for light waves in terms of operators
can be found in [29]
3. The complementarity principle in double slit experiment

Wave-particle duality as first conceived by [30] stated that in a given experimental set up
one observes either wave or particle properties. Using his words: "we are presented with a
choice of either tracing the path of a particle or observing interference effects". In double slit
experiments the wave nature of the object is reflected by the interference pattern exhibited
on a screen. We know however that once the knowledge of the object’s path (i.e., through
which slit it crossed) is obtained, the interference pattern is completely destroyed. A natural
question that can arise is how partial knowledge of path affects such pattern. This discussion
was first addressed by [31] and later by [32]. The quantification of which-way information
(predictability) and visibility was established by [33] and the relation between both given by
[34].
We proceed to define visibility and predictability in two slit experiments. Consider a double
slit experiment as depicted in the Fig. 2. B and C denote the slits, A the position from which
the particles are emitted and D the detector which covers the whole extension of the screen
and counts the numbers of particles. Notice that A need not be placed on the symmetry axis
Ox.
Figure 2. Illustration of the double slit with a displaced source of particles.
Suppose N particles are sent to the grating. Consider that (NB ) NC particles cross slit (B) C
leaving slit (C) B closed with relative probability (PB ) PC .
The probability amplitude at the screen is

ψ ( x0 , y ) = PB ψB ( x0 , y) + PC ψC ( x0 , y). (22)
The probability density of finding a particle at the point ( x0 , y) on the screen knowing that it
crossed slit B is obviously
PB (y) = PB |ψB ( x0 , y)|2 . (23)

Analogously
PC (y) = PC |ψC ( x0 , y)|2 , (24)

for slit C.
Now the quantification of which way information, also known as predictability is

P ( y ) − PC ( y )
P (y) = B . (25)
P B ( y ) + PC ( y )
The visibility V is defined through the expression for the intensity |ψ( x0 , y)|2 on the screen
I (y) = PB |ψB ( x0 , y)|2 + PC |ψC ( x0 , y)|2 + 2|

PB PC ψB ( x0 , y)ψC ( x0 , y)| cos φ, (26)
with φ = arg ψB ( x0 , y) − arg ψC ( x0 , y).
We can rewrite I (y) as
√
PB PC ψB ( x0 , y)ψC ( x0 , y)

I (y) = [PB (y) + PC (y)] 1 + 2 cos φ , (27)
P B ( y ) + PC ( y )
from which the visibility [37]
Imax − Imin
V= , (28)
Imax + Imin
can be read off as
√
2| PB PC ψB ( x0 , y)ψC ( x0 , y)|
V= . (29)
P B ( y ) + PC ( y )
We can immediately verify that [32, 34]
P 2 + V 2 = 1, (30)
for all y. The relation above is known as the complementarity relation.
4. Double slit interference: Complementarity principle and the Gouy

phase
In this section we discuss double-slit interference of matter waves, for instance neutrons
or fullerene molecules [35], to illustrate the ideas that we discussed so far. Consider the
experimental set up illustrated in the Fig. 3. b0 is the characteristic width of a wavepacket,
assumed to be coherent, i.e., not affected by indeterminacies inherent to the production
process, σ0 is the slit widths, x0 the distance between the slit grating and screen and d is the
distance between the slits. The particles travel in the x direction with speed v, assuming that
the center of mass moves classically, that is ∆p x ≪ p x .
The intensity on the screen reads
Figure 3. Experimental set up.
2
I (y, T = x0 /v) = PB ψB (y, T = x0 /v) + PC ψC (y, T = x0 /v) (31)
where
(y + d2 )2 (1 − i τT0 )

1
ψB (y, T ) = √ exp − + iµ( T ) , (32)
B( T ) π 2B( T )2
(y − 2d )2 (1 − i τT0 )

1
ψC (y, T ) = √ exp − + iµ( T ) , (33)
B( T ) π 2B( T )2
mσ2
and τ0 = h̄ 0 . An analytical expression for PB and PC can in principle be derived in terms
of the parameters as illustrated in Fig. 3 as performed in [35] but it is not essential to the
purposes of our discussion.
It is easy to show that I (y, T ) can be written as
√
2 PB PC
I (y) = F (y) 1 + − cos( βy) , (34)
PBeαy+ PC e αy
where
PB eαy + PC e−αy
F (y) = I0 e−γ(y) √ , (35)
2 PB PC
y2 + ( d2 )2
γ(y) = , (36)
B ( T )2
d 1 d
α= = , (37)
B ( T )2 σxx 2
and
T d σxp d
β= = , (38)
τ0 B( T )2 σxx h̄
where hereafter we write I (y, T ) = I (y) for sake of brevity.

Recalling the expression for predictability and visibility (25), (29) we get
√
2 PB PC
V (y) = , (39)
PB eαx + PC e−αy
P eαy − PC e−αy

P (y) = B αy , (40)
PB e + PC e−αy
and
P 2 (y) + V 2 (y) = 1. (41)
Let p = PPCB . From expression (39) we can find the point on the screen where the visibility
reaches its maximal value,
√
ln p
ymax = . (42)
α
At this point the which way information is minimum, that is to say P (ymax ) = 0 and
V (ymax ) = 1. Notice that it. In Fig. 4 we illustrate this with p = 1, 0.15 and 0.015.
For p = 1, illustrated in Fig. 4, we have
Figure 4. Intensity, visibility and predictability for p = 1, p = 0.15 and p = 0.015, respectively.
cos( βy)

I (y) = F (y) 1 + , (43)
cosh(αy)
1
V (y) = , (44)
cosh(αy)
and

P (y) = tanh(αy). (45)
The peaks of intensity and visibility occurs at y = 0 where predictability is minimum. In

terms of the complementarity principle we have the minimum of knowledge about from
which slit the particle cames from. Note also that we have intermediate situations where
we have interference fringes with 0 < P < 1 and 0 < V < 1, signalizing the existence of
situations where the wave like property does not preclude the particle like property.
An important quantity is the ratio R between α and β,
α h̄
R= = . (46)
β 2σxp
It can be used to estimate the effective number of fringes (contrast) ν before the visibility
decreases by a factor of 1/e [26]
0.264
ν= , (47)
R
for instance, R = 1 for ν = 0.264.
It is also related to the Gouy phase. From Equations (20) and (46) we get

1 1
µ( T ) = − arctan . (48)
2 R
In Fig. 5 we plot µ against the slit width σ0 . We immediately verify that for σ0 between 0
and 0.02µm the phase varies abruptly from −0.05 to π4 rad. For values of the Gouy phase
within this range we associated a noticeable contrast in the intensity curves as shown in Fig.
5. When the Gouy phase tends to zero the interference fringes tends to disappear because
the geometrical optical regime is attained.
Figure 5. Gouy phase and intensity contrast according Gouy phase.

5. Conclusion
Our results show how partial knowledge of path affects the interference pattern. The position
of the peaks of intensity and visibility are affected by the probabilities of the particles cross
the slit B (PB ) or C (PC ). When PB = PC these peaks are centered in the position ymax = 0
and when PB �= PC they are displaced from this position. In positions y �= ymax we have
the situations where V and P are both different of zero, i.e., the wave like property not
exclude the particle like property. The crucial point here is the connection between the
complementarity principle and the Gouy phase of matter waves. We see that this phase
appears in the number of interference fringes exhibiting the wave behavior of the matter. We
see that the number of fringes tends to disappear when the Gouy phase tends to zero.
Acknowledgments
This work was in part supported by the Brazilian agencies CNPq, Capes, Fapepi and
Fapemig.
Author details
Luis A. Cabral1 , Irismar G. da Paz2 ,
José G. G. de Oliveira Júnior3 , Saulo Moreira4 ,
Marcos D. R. Sampaio4 and Maria C. Nemes4
1 Departamento de Física, Universidade Federal do Tocantins, Brazil
2 Departamento de Física, Universidade Federal do Piauí, Brazil
3 Departamento de Ciências Exatas e Tecnológicas, Universidade Estadual de Santa Cruz,
Brazil
4 Departamento de Física, Universidade Federal de Minas Gerais, Brazil
References
[1] Gouy, L. G. (1890). Sur une propriete nouvelle des ondes lumineuses, Acad. Sci. Paris
110, 1251.
[2] Gouy, L. G. (1891). Sur la propagation anomale des ondes, Ann. Chim. Phip. Ser. 6 24,
145.
[3] Berry, M. V. (1984). Quantal phase factors accompanying adiabatic changes, Proc. R.
Soc. London A 392, 45.
[4] Simon, R., and Mukunda, N. (1993). Bargman Invariant and the Geometry of the Gouy
Effect, Phys. Rev. Lett. 70, 880.
[5] Feng, S., and Winful, H. G. (2001). Physical origin of the Gouy phase shift, Opt. Lett.
26, 485.
[6] Yang, J., and Winful, H. G. (2006) Generalized eikonal treatment of the Gouy phase
shift, Opt. Lett. 31, 104.
[7] da Paz, I. G. (2006). Matter Waves and Paraxial Propagation of Light, Department of
Physics, UFMG, MG, Brazil.
[8] da Paz, I. G. (2011). Gouy phase in matter waves: from pure and mixed Gaussian states,
Department of Physics, UFMG, MG, Brazil.
[9] da Paz, I. G., Nemes, M. C., and Peixoto de Faria, J. G. (2011). Electromagnetic Waves -
Chapter four - Edited by Vitaliy Zhurbenko, InTech, Janeza Trdine 9, 51000 Rijeka, Croatia.
[10] da Paz, I. G., Monken, C. H., Padua, S., Nemes, M. C., and Peixoto de Faria, J. G.
(2010). Indirect evidence for the Gouy phase for matter waves, Phys. Lett. A 374, 1660.
[11] da Paz, I. G., Nemes, M. C., and Peixoto de Faria, J. G. (2007). Gouy phase and matter
waves, J. Phys.: Conference Series 84, 012016.
[12] da Paz, I. G., Saldanha, P. L., Nemes, M. C., and Peixoto de Faria, J. G. (2011).
Experimental proposal for measuring the Gouy phase of matter waves, New Journal
of Physics 13, 125005.
[13] Lindner, F., Paulus, G. G., Walther, H., Batuska, A., Goulielmakis, E., Lezius, M., and
Krausz, F. (2004). Gouy Phase Shift for Few-Cycle Laser Pulses, Phys. Rev. Lett. 92,
113001.
[14] Balcou, P., and LŠHuillier, A. (1993). Phase-matching effects in strong-field harmonic
generation, Phys. Rev. A 47, 1447.
[15] Lewenstein, M., Salières, P., and LŠHuillier, A. (1995). Phase of the atomic polarization
in high-order harmonic generation, Phys. Rev. A 52, 4747.
[16] Lindner, F., Stremme, W., Schätzel, M. G., Grasbon, F., Paulus, G. G., Walther, H.,
Hartmann, R.,and Strüder, L. (2003). High-order harmonic generation at a repetition
rate of 100 kHz, Phys. Rev. A 68, 013814.
[17] Debye, P. (1905). Ann. Phys. 30, 755.
[18] Rubinowicz, A. (1938). On the Anomalous Propagation of Phase in the Focus, Phys.
Rev. 54, 931.
[19] Linfoot, E. H. (1956). Proc. Phys. Soc. London 69, 823.
[20] Visser, T. D., and Wolf, E. (2010). The origin of the Gouy phase anomaly and its
generalization to astigmatic wavefields, Opt. Comm. 283, 3371.
[21] Boyd, R. W. (1980). Intuitive explanation of the phase anomaly of focused light beams,
J. Opt. Soc. Am. 70, 877.
[22] Hariharan, P., and Robinson, P. A. (1996). The Gouy phase shift as a geometrical
quantum effect, J. Mod. Opt. 43, 219.
[23] Born, M., and Wolf, E. (1999). Principles of Optics, Cambridge University Press,
Cambridge.
[24] Mandel, L., and Wolf, E. (1995). Optical Coherence and Quantum Optics, Cambridge, New
York.
[25] Saleh, B. E. A., and Teich, M. C. (1991). Fundamentals of Photonics, John Wiley et Sons,
New York.
[26] Bramon, A., Garbarino.,and Hiesmayr, B. C. (2004). Quantitative complementarity in

two-path interferometry, Phys. Rev. A 69, 022112.
[27] Viale, A., Vicari, M., and Zanghì, N. (2003). Analysis of the loss of coherence in
interferometry with macromolecules, Phys. Rev. A 68, 063610.
[28] Schrödinger, E. (1930). Zum Heisenbergschen Unsch"arfeprinzip,Proceedings of the

Prussian Academy of Sciences XIX pp. 296Ű303.
[29] Stoler, D. (1981). Operator methods in physical optics, J. Opt. Soc. Am. 71, 334.
[30] Bohr, N. (1928). The quantum postulate and the recent development of atomic theory,
Nature 121, 580.
[31] Wootters, W. K., and Zurek, W. K. (1979). Complementarity in the double-slit

experiment: Quantum nonseparability and a quantitative statement of Bohr’s
principle, Phys. Rev. D 19, 473.
[32] Greenberger, D. M., and Yasin, A. (1988). Simultaneous wave and particle knowledge
in a neutron interferometer, Phys. Lett. A 128, 391.
[33] Mandel, L. (1991). Coherence and indistinguishability, Opt. Lett. 16, 1882.
[34] Englert, B. G. (1996). Fringe Visibility and Which-Way Information: An Inequality,

Phys. Rev. Lett. 77, 2154.
[35] Glionna, G., Blin, A. H., Nemes, M. C., Sampaio, M., and Toledo Piza, A. F. R. (2008).
Dispersion and uncertainty in multislit matter wave diffraction, Physica A 387, 1485.
[36] Nairz, O., Arndt, M., and Zeilinger, A. (2002). Experimental verification of the
Heisenberg uncertainty principle for fullerene molecules, Phys. Rev. A 65, 032109.
[37] Scully, M. O., and Zubairy, M. S. (1997). Quantum Optics, Cambridge University Press,
Cambridge.

Kishk A. (Ed.) - Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic Waves

Uploaded by

Copyright:

Available Formats

Kishk A. (Ed.) - Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic Waves

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Kishk A. (Ed.) - Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic Waves

Uploaded by

Copyright:

Available Formats

SOLUTIONS AND

Edited by Ahmed Kishk

Copyright © 2012 InTech

Publishing Process Manager Dragana Manestar

First published November, 2012

A free online edition of this book is available at www.intechopen.com

Solutions and Applications of Scattering, Propagation, Radiation and Emission of Electromagnetic

Section 1 Waves in Chiral Media 1

Chapter 1 Electromagnetic Characterization of Chiral Media 3

Chapter 2 Chiral Waves in Graphene Medium and Optical Simulation

Section 2 Numerical and Analytical Solutions 57

Chapter 3 Numerical Modeling of Electromagnetic Wave Propagation

Chapter 4 Integral Equation Analysis with

Chapter 5 Analytical Grounds for Modern Theory of Two-

Chapter 6 The Sources Reconstruction Method for Antenna Diagnostics

Section 3 Microwave Applications 183

Chapter 7 Multi-Hole Waveguide Directional Couplers 185

Chapter 8 The Diffraction of Electromagnetic Waves on the Periodic

Section 4 Scattering, Radiation and Propagation 233

Chapter 9 Radiative Transfer: Application to Indoor Propagation 235

Chapter 10 Electromagnetic Wave Packets in the Theory of Bremsstrahlung

Chapter 11 The Effect of Weak Fields at Multiple Frequencies on the

Chapter 12 Matter Wave Interferometry, the Gouy Phase

propagation inside a bi-isotropic medium. The MMM is included to characterize

Dr. Ahmed Kishk

Waves in Chiral Media

Electromagnetic Characterization of Chiral Media

J. Margineda, G.J. Molina-Cuberos, M.J. Núñez,

Additional information is available at the end of the chapter

The electromagnetic characterization of materials is a fundamental problem in many re‐

2. Electromagnetic fields in chiral media

The constitutive relations (1) for reciprocal media (χ = 0), are:

These combinations of electromagnetic fields propagate through the chiral material as

The normal modes satisfying eq. (4) are:

the medium parameters being:

where κr = κ / εr μr is the relative chirality, and:

2.1. Plane wave propagation in chiral media

A propagating wave can always be expressed as a linear combination of the propagation of

wherek± = ω ε±μ± = ω εμ (1 ± κr ) = ω n (1 ± κr ) / c. The refraction indices for each mode are

3.1. Experimental systems

3.2. Retrieval algorithm

Consider a chiral slab of thickness L, which is illuminated by a normally incident linearly

The incident wave may be decomposed as:

The reflected and transmitted electric fields are,

j sin (nk0 L )(1 - ηr2)

It is useful to consider the following additional parameters:

The refractive index is obtained from equation (21) and is:

n = j 4πfL ln |P 2| - Phase(P 2) 4πfL -

From nand ηr permittivity and permeability can be obtained:

The chirality parameter is obtained from TL andT:

4.1. Experimental systems

Figure 7. Photo of the experimental diagram of Figure 6.

4.2. Modes in circular waveguides

(∇2t + K ±2)E±z = 0 (28)

K ± = (k±2 - γ 2)1/2 (29)

4.3. Retrieval algorithm

6. Discussion and conclusions

are characterized by three complex parameters, permittivity, permeability and chirality, at

Waveguide measurements may be considered a possible alternative. The measuring princi‐

J. Margineda1, G.J. Molina-Cuberos1, M.J. Núñez1, A.J. García-Collado1,2 and E. Martín1