This action might not be possible to undo. Are you sure you want to continue?
STRUCTURES AND RETRIEVAL OF
HOMOGENIZATION PARAMETERS
a thesis
submitted to the department of electrical and
electronics engineering
and the institute of engineering and sciences
of bilkent university
in partial fulfillment of the requirements
for the degree of
master of science
By
Erdin¸c Ircı
August 2007
I certify that I have read this thesis and that in my opinion it is fully adequate,
in scope and in quality, as a thesis for the degree of Master of Science.
Assist. Prof. Dr. Vakur B. Ert¨ urk(Supervisor)
I certify that I have read this thesis and that in my opinion it is fully adequate,
in scope and in quality, as a thesis for the degree of Master of Science.
Prof. Dr. Ayhan Altınta¸s
I certify that I have read this thesis and that in my opinion it is fully adequate,
in scope and in quality, as a thesis for the degree of Master of Science.
Assoc. Prof. Dr.
¨
Ozlem Aydın C¸ ivi
Approved for the Institute of Engineering and Sciences:
Prof. Dr. Mehmet Baray
Director of Institute of Engineering and Sciences
ii
ABSTRACT
WAVE PROPAGATION IN METAMATERIAL
STRUCTURES AND RETRIEVAL OF
HOMOGENIZATION PARAMETERS
Erdin¸c Ircı
M.S. in Electrical and Electronics Engineering
Supervisor: Assist. Prof. Dr. Vakur B. Ert¨ urk
August 2007
Electromagnetic wave propagation in metamaterial structures (metamaterial
slabs, metamaterial cylinders, metamaterial coated conducting cylinders etc.)
are investigated. Scattered and transmitted electromagnetic ﬁelds by these struc
tures due to electric line sources or plane wave illuminations are found. A generic
formulation of these wave propagation problems is done, enabling any kind of
metamaterial or conventional material to be used, having any sign combination
of constitutive parameters and having any electric and/or magnetic losses.
For one of these propagation problems i.e., metamaterial coated conducting
cylinders illuminated normally with plane waves, achieving transparency and
maximizing scattering are investigated thoroughly. It is found out that, rigorous
derivation of transparency and resonance (scattering maximization) conditions
for PEC core cylinder case under the subwavelength limitations yields the same
conditions of two electrically small concentric layers of conjugately paired cylin
ders, given in the literature (when the inner core layer is also taken to the PEC
limit). These transparency and resonance conditions are found to be heavily
iii
dependent on the permittivity of the metamaterial coating (for TE polariza
tion) and the ratio of coreshell radii. The relations between the permittivity
of the coating and the ratio of coreshell radii are investigated for achieving
transparency and scattering maximization. Numerical results show that these
analytical relations are quite successful and work better when the cylindrical
scatter is electrically very small.
A novel homogenization method for the retrieval of eﬀective constitutive pa
rameters of metamaterials is proposed and implemented. The method is based
on the simple idea that the total reﬂection coeﬃcient from a ﬁnite metamate
rial structure has to resemble the reﬂection from an homogeneous equivalent.
While implementing the method, 1, 2, . . ., 20 unit cells of the same metama
terial structure are stacked and their reﬂection coeﬃcients are collected. The
homogenization quality of the metamaterial is evaluated in terms of various fac
tors, which showed that the method is very successful to retrieve the eﬀective
constitutive parameters of the metamaterial.
Finally, another method has been proposed for the retrieval of surface wave
propagation constants on any periodic or nonperiodic grounded slab medium. As
a preliminary, the method is applied to grounded dielectric slabs. The numerical
results generally show good agreement with their theoretical counterparts.
Keywords: Metamaterials, Wave propagation, Scattering, Transmission,
Metamaterial cylinders, Metamaterial coated conducting cylinders, Transparency,
Resonance, Radar cross section, Homogenization, Parameter retrieval,
Surface waves, Grounded Slabs.
iv
¨
OZET
METAMALZEME YAPILARDA DALGA YAYINIMI VE
HOMOJENLES¸T
˙
IRME PARAMETRELER
˙
IN
˙
IN ELDE
ED
˙
ILMES
˙
I
Erdin¸c Ircı
Elektrik ve Elektronik M¨ uhendisli˘ gi B¨ol¨ um¨ u Y¨ uksek Lisans
Tez Y¨oneticisi: Yar. Do¸c. Dr. Vakur B. Ert¨ urk
A˘gustos 2007
Metamalzeme yapılarda (metamalzeme tabakalar, metamalzeme silindirler,
metamalzeme kaplı iletken silindirler vb.) elektromanyetik dalga yayınımı ince
lendi. C¸ izgisel elektrik kaynaklarından ya da d¨ uzlem dalga aydınlatmalarından
dolayı bu yapılardan sa¸cılan ve bunlara iletilen elektromanyetik alanlar bulundu.
Bu dalga yayınım problemlerinin genel form¨ ulasyonu, ortam parametrelerinin
i¸saretlerinin herhangi kombinasyonu i¸cin, herhangi elektrik/manyetik kayba da
sahip olabilecek ¸sekilde metamalzemeler ya da sıradan malzemeler i¸cin yapıldı.
Bu yayınım problemlerinden biri olan d¨ uzlem dalga ile dik aydınlatılmı¸s
metamalzeme kaplı iletken silindirler, saydamlık ve sa¸cılımın azamile¸stirilmesi
a¸cısından detaylıca incelendi. Saydamlık ve rezonans (sa¸cılım azamile¸stirmesi)
durumlarının dalgaboyualtı sınırında t¨ uretilmesi, literat¨ urdeki aynı eksenli, elek
triksel olarak k¨ u¸c¨ uk, ters i¸saretli olarak e¸sle¸stirilmi¸s silindirlerle aynı durumu
verdi (i¸c silindir iletken sınırına g¨ot¨ ur¨ uld¨ u˘g¨ unde).
Bu saydamlık ve rezonans durumlarının daha ¸cok metamalzeme kaplamanın
elektriksel ge¸cirgenli˘ gine (TE polarizasyonu i¸cin) ve ¸cekirdekkaplama yarı¸cap
v
oranına ba˘glı oldu˘gu bulundu. Saydamlık ve sa¸cılım azamile¸stirmesi i¸cin, kapla
manın elektrik ge¸cirgenli˘gi ile ¸cekirdekkaplama yarı¸cap oranı arasındaki ili¸skiler
incelendi. Sayısal sonu¸clar bu analitik ili¸skilerin olduk¸ca ba¸sarılı oldu˘gunu ve
silindirik sa¸cıcı elektriksel olarak ¸cok k¨ u¸c¨ ukken daha iyi ¸calı¸stı˘gını g¨osterdi.
Metamalzemelerin etkin ortam parametrelerinin elde edilmesi i¸cin yeni bir ho
mojenle¸stirme metodu ileri s¨ ur¨ uld¨ u ve uygulandı. Metod, sonlu bir metamalzeme
yapının toplam yansıma katsayısının homojen denginin yansımasına benzeyece˘ gi
ﬁkrine dayandırıldı. Metod uygulanırken metamalzemenin 1, 2, . . ., 20 ¨ unite
h¨ ucresi art arda sıralandı ve yansıma katsayıları kaydedildi. Metamalzemenin
homojenle¸stirme kalitesi de˘gi¸sik etkenler cinsinden incelendi ve metodun meta
malzemenin etkin ortam parametrelerinin elde edilmesi i¸cin ¸cok ba¸sarılı oldu˘gu
g¨oz¨ ukt¨ u.
Son olarak, bir ba¸ska metod da periyodik olan ya da olmayan herhangi bir
topraklanmı¸s tabaka ¨ uzerindeki y¨ uzey dalga yayınım katsayılarının elde edilmesi
i¸cin ileri s¨ ur¨ uld¨ u. Ba¸slangı¸c olarak metod topraklanmı¸s dielektrik tabakalara
uygulandı. Sayısal sonu¸clar genel olarak teorik kar¸sılıklarıyla iyi uyum sergiledi.
Anahtar Kelimeler: Metamalzemeler, Dalga yayınımı, Sa¸cılım,
˙
Iletim,
Metamalzeme silindirler, Metamalzeme kaplı iletken silindirler, Saydamlık,
Rezonans, Radar kesit alanı, Homojenle¸stirme, Parametre elde edimi,
Y¨ uzey dalgaları, Topraklanmı¸s tabakalar.
vi
ACKNOWLEDGMENTS
I would like to express my sincere gratitude to my supervisor Asst. Prof.
Vakur B. Ert¨ urk for his invaluable guidance, suggestions, encouragement and
support throughout the development of this thesis. He has always been a great
mentor and teacher to me.
I would like to thank Prof. Ayhan Altınta¸s and Assoc. Prof.
¨
Ozlem Aydın
C¸ ivi from METU for being in my jury, reading the thesis and commenting on it.
I would like to thank Prof. M.
˙
Ir¸sadi Aksun from Ko¸c University for allowing
us to collaborate in his research. Chapter 4 of this thesis is merely realization of
his ingenious ideas.
I would also like to thank The Scientiﬁc and Technological Research Council
of Turkey (T
¨
UB
˙
ITAK) for supporting me with the graduate scholarship during
my study.
Finally, I would like to thank my family, my friends Celal Alp Tun¸c, Onur
Bakır, Ayta¸c Alparslan, Burak G¨ uldo˘ gan and many others, whom I can’t all list
here, for their understanding, encouragement, friendship and support.
vii
Contents
1 INTRODUCTION 1
2 Wave Propagation in Metamaterial Structures 6
2.1 Wave Number, Index of Refraction and Wave Impedance of Meta
material Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Normal Incidence of Plane Waves on a Metamaterial Slab . . . . . 9
2.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3 Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 10
2.2.4 Solution of Boundary Conditions . . . . . . . . . . . . . . 11
2.3 Inﬁnite Length Metamaterial Cylinder Near an Inﬁnite Length
Electric Line Source: TM
z
Polarization . . . . . . . . . . . . . . . 13
2.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 13
2.3.3 Electric Line Source and Incident Electric Field . . . . . . 14
viii
2.3.4 Scattered and Transmitted Electric Fields . . . . . . . . . 14
2.3.5 Boundary Conditions for Electric Fields . . . . . . . . . . 15
2.3.6 Incident, Scattered and Transmitted Magnetic Fields . . . 15
2.3.7 Boundary Conditions for Magnetic Fields . . . . . . . . . . 16
2.3.8 Simultaneous Solution of the Boundary Conditions for
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 17
2.3.9 Calculation of the Radiation Patterns . . . . . . . . . . . . 18
2.3.10 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Normally Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Cylinder: TM
z
Polarization . . . . . . . . . . . . . 21
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.4.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 21
2.4.3 Uniform Plane Wave and Incident Electric Field . . . . . . 22
2.4.4 Scattered and Transmitted Electric Fields . . . . . . . . . 22
2.4.5 Boundary Conditions for Electric Fields . . . . . . . . . . 23
2.4.6 Incident, Scattered and Transmitted Magnetic Fields . . . 23
2.4.7 Boundary Conditions for Magnetic Fields . . . . . . . . . . 24
2.4.8 Simultaneous Solution of the Boundary Conditions for
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 24
2.4.9 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 25
ix
2.5 Normally Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Cylinder: TE
z
Polarization . . . . . . . . . . . . . . 27
2.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 27
2.5.3 Uniform Plane Wave and Incident Magnetic Field . . . . . 28
2.5.4 Scattered and Transmitted Magnetic Fields . . . . . . . . 28
2.5.5 Boundary Conditions for Magnetic Fields . . . . . . . . . . 28
2.5.6 Incident, Scattered and Transmitted Electric Fields . . . . 29
2.5.7 Boundary Conditions for Electric Fields . . . . . . . . . . 30
2.5.8 Simultaneous Solution of the Boundary Conditions for
Magnetic and Electric Fields . . . . . . . . . . . . . . . . . 30
2.5.9 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 31
2.6 Inﬁnite Length Metamaterial Coated Conducting Cylinder Near
an Inﬁnite Length Electric Line Source: TM
z
Polarization . . . . 32
2.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.6.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 32
2.6.3 Electric Line Source and Incident Electric Field . . . . . . 33
2.6.4 Scattered and Transmitted Electric Fields . . . . . . . . . 33
2.6.5 Boundary Conditions for Electric Fields . . . . . . . . . . 34
2.6.6 Incident, Scattered and Transmitted Magnetic Fields . . . 34
2.6.7 Boundary Conditions for Magnetic Fields . . . . . . . . . . 35
x
2.6.8 Simultaneous Solution of the Boundary Conditions for
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 35
2.6.9 Electric Line Source Inside the Metamaterial Coating . . . 36
2.6.10 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 38
2.7 Normally Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Coated Conducting Cylinder: TM
z
Polarization . . 40
2.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.7.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 40
2.7.3 Uniform Plane Wave and Incident Electric Field . . . . . . 41
2.7.4 Scattered and Transmitted Electric Fields . . . . . . . . . 43
2.7.5 Boundary Conditions for Electric Fields . . . . . . . . . . 43
2.7.6 Incident, Scattered and Transmitted Magnetic Fields . . . 44
2.7.7 Boundary Conditions for Magnetic Fields . . . . . . . . . . 44
2.7.8 Simultaneous Solution of the Boundary Conditions for
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 45
2.7.9 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 46
2.8 Normally Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Coated Conducting Cylinder: TE
z
Polarization . . 48
2.8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.8.2 Problem Geometry . . . . . . . . . . . . . . . . . . . . . . 48
2.8.3 Uniform Plane Wave and Incident Magnetic Field . . . . . 49
xi
2.8.4 Scattered and Transmitted Magnetic Fields . . . . . . . . 49
2.8.5 Boundary Conditions for Magnetic Fields . . . . . . . . . . 50
2.8.6 Incident, Scattered and Transmitted Electric Fields . . . . 50
2.8.7 Boundary Conditions for Electric Fields . . . . . . . . . . 50
2.8.8 Simultaneous Solution of the Boundary Conditions for
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 51
2.9 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Cylinder: TM
z
Polarization . . . . . . . . . . . . . 53
2.9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.9.2 Incident, Scattered and Transmitted Electric Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.9.3 Incident, Scattered and Transmitted Magnetic Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.9.4 φ Components of the Incident, Scattered and Transmitted
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 56
2.9.5 Boundary Conditions and Their Solution . . . . . . . . . . 57
2.9.6 Calculation of the Radar Cross Section . . . . . . . . . . . 59
2.10 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Cylinder: TE
z
Polarization . . . . . . . . . . . . . . 62
2.10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.10.2 Incident, Scattered and Transmitted Magnetic Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 62
xii
2.10.3 Incident, Scattered and Transmitted Electric Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.10.4 φ Components of the Incident, Scattered and Transmitted
Magnetic and Electric Fields . . . . . . . . . . . . . . . . . 63
2.10.5 Boundary Conditions and Their Solution . . . . . . . . . . 63
2.10.6 Calculation of the Radar Cross Section . . . . . . . . . . . 65
2.11 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Coated Conducting Cylinder: TM
z
Polarization . . 66
2.11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 66
2.11.2 Incident, Scattered and Transmitted Electric Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.11.3 Incident, Scattered and Transmitted Magnetic Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.11.4 φ Components of the Incident, Scattered and Transmitted
Electric and Magnetic Fields . . . . . . . . . . . . . . . . . 67
2.11.5 Boundary Conditions and Their Solution . . . . . . . . . . 68
2.11.6 Calculation of the Radar Cross Section . . . . . . . . . . . 71
2.12 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length
Metamaterial Coated Conducting Cylinder: TE
z
Polarization . . 72
2.12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.12.2 Incident, Scattered and Transmitted Magnetic Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 72
xiii
2.12.3 Incident, Scattered and Transmitted Electric Fields (z
components) . . . . . . . . . . . . . . . . . . . . . . . . . . 73
2.12.4 φ Components of the Incident, Scattered and Transmitted
Magnetic and Electric Fields . . . . . . . . . . . . . . . . . 73
2.12.5 Boundary Conditions and Their Solution . . . . . . . . . . 74
2.12.6 Calculation of the Radar Cross Section . . . . . . . . . . . 77
3 Achieving Transparency and Maximizing Scattering with Meta
material Coated Conducting Cylinders 78
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2 Transparency Condition . . . . . . . . . . . . . . . . . . . . . . . 79
3.3 Resonance (Scattering Maximization) Condition . . . . . . . . . . 84
3.4 Numerical Results and Discussion . . . . . . . . . . . . . . . . . . 87
4 Retrieval of Homogenization Parameters 101
4.1 Homogenization of Metamaterial Structures and Retrieval of Ef
fective Constitutive Parameters . . . . . . . . . . . . . . . . . . . 101
4.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.1.2 Homogenization of Metamaterials . . . . . . . . . . . . . . 103
4.1.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 121
4.1.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.2 Retrieval of Surface Wave Propagation Constants on a Grounded
Dielectric Slab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
xiv
4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.2.2 The TwoStep Method . . . . . . . . . . . . . . . . . . . . 129
4.2.3 Implementation: . . . . . . . . . . . . . . . . . . . . . . . . 144
4.2.4 Numerical Results . . . . . . . . . . . . . . . . . . . . . . 146
4.2.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5 CONCLUSIONS 160
APPENDICES 163
A Bessel Functions 163
B Derivation of the φ Components of Electric and Magnetic Fields:
TM
z
Polarization 166
C Derivation of the Transparency Condition 175
D Derivation of the Resonance Condition 180
xv
List of Figures
2.1 Uniform plane wave normally incident on a metamaterial slab. . . 10
2.2 Metamaterial cylinder near an electric line source. (a) Side view,
(b) Top view. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2 20
2.4 Uniform plane wave incident on a metamaterial cylinder: TM
z
Polarization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5 Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2 26
2.6 Uniform plane wave incident on a metamaterial cylinder: TE
z
Polarization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.7 Metamaterial coated conducting cylinder near an electric line
source (Cross section view). . . . . . . . . . . . . . . . . . . . . . 32
2.8 Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2 39
2.9 Plane wave normally incident on a metamaterial coated conduct
ing cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
xvi
2.10 Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2 47
2.11 Plane wave normally incident on a metamaterial coated conduct
ing cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.12 Uniform plane wave obliquely incident on a metamaterial cylinder:
TM
z
Polarization. . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.13 Longitudinal and transverse components of the incident and trans
mitted ﬁelds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.14 Uniform plane wave obliquely incident on a metamaterial coated
conducting cylinder: TM
z
Polarization. . . . . . . . . . . . . . . . 66
3.1 Normalized monostatic echo width of a metamaterial coated PEC
cylinder (a = 50mm, b = 70mm, f = 1GHz). Diamond marks
show the DPS and DNG coating cases in [1]. . . . . . . . . . . . 87
3.2 Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TE
z
polarization case, versus the corecoating ra
tio for coatings with diﬀerent constitutive parameters. The outer
radius of the coating is selected as (a)(d) b = λ
0
/100, (e)(h)
b = λ
0
/10. Dashed line shows the uncoated PEC case, with ra
dius a. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.3 Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TE
z
polarization case, versus the corecoating ra
tio for coatings with diﬀerent constitutive parameters. The outer
radius of the coating is selected as (a)(c) b = λ
0
/2, (d)(f) b = λ
0
.
Dashed line shows the uncoated PEC case, with radius a. . . . . 92
xvii
3.4 Normalized monostatic echo width of an ENG coated PEC cylin
der for the TE
z
polarization case, versus the corecoating ratio for
coatings with diﬀerent constitutive parameters. The outer radius
of the coating is selected as (a)(d) b = λ
0
/100, (e)(h) b = λ
0
/50.
Dashed line shows the uncoated PEC case, with radius a. . . . . 93
3.5 Normalized monostatic echo width of an ENG coated PEC cylin
der for the TE
z
polarization case, versus the corecoating ratio for
coatings with diﬀerent constitutive parameters. The outer radius
of the coating is selected as (a)(d) b = λ
0
/20, (e)(h) b = λ
0
/10.
Dashed line shows the uncoated PEC case, with radius a. . . . . 95
3.6 Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TM
z
polarization case, versus the coating perme
ability µ
c
for diﬀerent corecoating ratios. The outer radius of the
coating is b = λ
0
/100 and the coating permittivity is ε
c
= ε
0
. . . . 96
3.7 Eﬀects of ohmic losses on normalized monostatic echo width for
(a) DPS [transparency] (b) ENG [Scattering maximization] cases.
The outer radius of the coating is selected as b = λ
0
/100. . . . . . 97
3.8 Normalized bistatic echo widths for (a) DPS coated (b) ENG
coated PEC cylinder for the TE
z
polarization case. The outer
radius of the coating is selected as b = λ
0
/100. The angle of
incidence is φ
0
= 0
◦
. . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.9 Contour plots of axial component of the total magnetic ﬁeld (i.e.,
H
i
z
+H
s
z
) outside the PEC cylinder when there is (a) No coating,
(b) DPS coating, (c) ENG coating. Outer boundaries of the coat
ings are shown by dashed lines (a = λ
0
/200, b = λ
0
/100). Plane
wave illumination is along the +xaxis. . . . . . . . . . . . . . . . 99
xviii
3.10 Normalized monostatic echo widths for (a) DPS coated (b) ENG
coated PEC cylinder for the TE
z
polarization, oblique incidence
case. The outer radius of the coating is selected as b = λ
0
/100. . . 100
4.1 Metamaterial unit cell. . . . . . . . . . . . . . . . . . . . . . . . . 104
4.2 Direction of E and H ﬁelds for a unit cell. . . . . . . . . . . . . . 104
4.3 Direction of E and H ﬁelds for a unit cell. . . . . . . . . . . . . . 105
4.4 Alignment of unit cells inside the PECPMC waveguide. . . . . . 106
4.5 Problem geometry (crosssection view, for N
z
= 3). . . . . . . . . 107
4.6 E
x
 vs. z (f = 10GHz, N
z
= 1). . . . . . . . . . . . . . . . . . . 109
4.7 Fresnel reﬂection at (a) three layered media, (b) two layered media.112
4.8 Eﬀective homogenization parameters of the metamaterial over the
5GHz  15GHz frequency band, (a) ε
r
, (b) µ
r
. . . . . . . . . . . . 122
4.9 S
11
vs. frequency, obtained from the metamaterial and its homo
geneous equivalent. . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.10 Magnitude of Eﬁeld inside and outside the metamaterial medium
and its homogeneous equivalent at f = 5GHz. . . . . . . . . . . . 125
4.11 Magnitudes of Eﬁeld inside and outside the metamaterial medium
and its homogeneous equivalent at (a) f = 10.8GHz, (b) f =
15.0GHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.12 Geometry of a grounded dielectric slab. . . . . . . . . . . . . . . . 128
4.13 Geometry of the rectangular narrow patch and excitation. . . . . 130
4.14 Entire problem geometry in HFSS. . . . . . . . . . . . . . . . . . 131
xix
4.15 Magnitudes of complex function y(t) and its N uniform samples
y[k]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.16 Magnitudes of complex function y(t −t
0
) and its N uniform sam
ples y[k −k
0
]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.17 Magnitudes of
√
ρE
x
(ρ) and its N uniform samples y[k −k
0
]. . . . 136
4.18 Problem geometry for the Eline case. . . . . . . . . . . . . . . . . 145
4.19 Problem geometry for the Hline case. . . . . . . . . . . . . . . . . 146
4.20 Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.1λ
0
, ε
r
= 2.55,
ρ
start
= 5λ
0
, ρ
end
= 8λ
0
, N = 101) . . . . . . . . . . . . . . . . . . 155
4.21 Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.15λ
0
, ε
r
= 2.55,
ρ
start
= 5λ
0
, ρ
end
= 8λ
0
, N = 101) . . . . . . . . . . . . . . . . . . 156
4.22 Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.19λ
0
, ε
r
= 2.55,
ρ
start
= 5λ
0
, ρ
end
= 8λ
0
, N = 101) . . . . . . . . . . . . . . . . . . 157
4.23 Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.25λ
0
, ε
r
= 2.55,
ρ
start
= 4λ
0
, ρ
end
= 8λ
0
, N = 51) . . . . . . . . . . . . . . . . . . 158
xx
List of Tables
2.1 Arguments of µ, ε, k and η for Diﬀerent Types of Metamaterials . 9
3.1 Desired and Obtained γ for Achieving Transparency Using (3.10) 82
3.2 Desired and Obtained γ for Achieving Transparency Using (3.11) 83
4.1 Parameters of the GPOF approximation . . . . . . . . . . . . . . 124
4.2 Space Wave and Surface Wave Characteristics in the E and H
planes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.3 Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.1λ
0
, ε
r
= 2.55) . . . . . 150
4.4 Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.15λ
0
, ε
r
= 2.55) . . . . 151
4.5 Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.19λ
0
, ε
r
= 2.55) . . . . 152
4.6 Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.25λ
0
, ε
r
= 2.55) . . . . 153
4.7 Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.05λ
0
, ε
r
= 2.55) . . . . 154
xxi
Dedicated to my family.
Chapter 1
INTRODUCTION
Metamaterials are artiﬁcially engineered materials which can have negative eﬀec
tive electric permittivity and/or negative eﬀective magnetic permeability. Diﬀer
ent from conventional materials, which have both positive electric permittivity
and positive magnetic permeability [i.e., double positive (DPS)], metamaterials
show diﬀerent electromagnetic and optical properties. For instance, when elec
tric permittivity and magnetic permeability of the material are both negative
[i.e., double negative (DNG)], negative refraction happens and direction of phase
velocity is reversed. DNG metamaterials are also called Left Handed Materials
(LHM) because electric ﬁeld, magnetic ﬁeld and the direction of phase velocity
form a left handed coordinate system for these materials. On the other hand,
when only one of the constitutive parameters of the metamaterial is negative
[i.e., single negative (SNG)] evanescent waves appear.
In Chapter 2, electromagnetic scattering and transmission due to line sources
or plane waves from diﬀerent metamaterial structures is investigated. The meta
material structures are chosen from simple canonical geometries, such as metama
terial slabs, metamaterial cylinders and metamaterial coated conducting cylin
ders, which have exact eigenfunction solutions.
1
After a complex analysis, the correct complex branches for the wave number
and wave impedance of a metamaterial medium are selected. This choice of
complex branches is found to be valid for all kinds of materials, which can have
any combination of signs of constitutive parameters, or can have any electric
and/or magnetic losses.
Due to their aforementioned exceptional properties, metamaterials are being
investigated for many possible utilizations in diﬀerent scientiﬁc and engineering
applications, which otherwise cannot be easily accomplished with conventional
materials. Recently, reducing scattering from various structures, and in the limit
achieving transparency and building cloaking structures, have been investigated
by many researchers [1–7]. On the other hand, resonant structures aimed at
increasing the electromagnetic intensities, stored or radiated power levels have
also been studied extensively [7–14]. Similarly, metamaterial layers have been
proposed to enhance the power radiated by electrically small antennas [15–17].
While some of these studies are based on utilization of nonlinear metamater
ial structures, some of them rely on pairing slabs, spheres or cylinders with their
electromagnetic conjugates (e.g., pairing/coating DPS materials with DNG meta
materials or munegative (MNG) metamaterials with epsilonnegative (ENG)
metamaterials).
In Chapter 3, the transparency and resonance conditions for cylindrical struc
tures are extended to the case where the core cylinder is particularly PEC.
For achieving transparency or maximizing scattering, simple (i.e., homogeneous,
isotropic and linear) metamaterial coatings are used. For both transparency and
scattering maximization scenarios, the analytical relations between the ratio of
corecoating radii and the constitutive parameters of the metamaterial coating
are derived. These analytical relations are based on subwavelength approxi
mations and they are valid especially when the cylindrical scatterers (i.e., PEC
cylinders together with their metamaterial coatings) are electrically small. The
2
numerical simulations have showed the existence of transparency and resonance
conditions in good agreement with the analytical expectations.
Although Chapter 3 is based on the assumptions that the metamaterial coat
ing is homogeneous and isotropic, metamaterials currently produced are inho
mogeneous, anisotropic and highly dispersive. However, there are many research
eﬀorts to obtain homogeneous and isotropic metamaterials.
Meanwhile, another branch of these research eﬀorts is now focused on retrieval
of the eﬀective constitutive parameters of metamaterials, or in other words, ob
taining homogeneous equivalents for essentially inhomogeneous metamaterials.
The process of obtaining this homogeneous equivalent, with its all intermediate
steps, is called homogenization. The homogenization processes present in the
literature [18–22] are mainly based on utilization of transmission and reﬂection
characteristics of the metamaterial structures, or ﬁeld averaging. However, dur
ing these attempts for homogenization of metamaterials, usually transmission
and reﬂection properties of only one unit cell of the metamaterial is taken into
account. These methods are intrinsically unreliable since the unit cells, which
form the metamaterial, are made up of metallic inclusions, which cause very
strong electric and magnetic resonances. While using only one or two unit cells
of the metamaterial, one loses the valuable information of periodicity of unit cells
and their mutual interactions, therefore cannot represent the whole metamaterial
structure correctly.
As a remedy to these inadequate methods, in Chapter 4 we present a novel
method for the homogenization and parameter retrieval of metamaterials. If a
metamaterial slab can be successfully homogenized, its reﬂection characteristics
would mimic those of a homogeneous slab. Since total reﬂection from a homo
geneous slab is the sum of a direct reﬂection term and other multiple reﬂection
terms due to the waves bouncing inside the slab, with added phase delays, the
3
total reﬂection from the metamaterial slab can be written as a sum of expo
nentials. Also since the phase delays of the multiple reﬂections inside the slab
are dependent on the thickness of the slab, utilization of diﬀerent number of
unit cells will yield diﬀerent reﬂection results. Therefore, it becomes possible to
obtain the constitutive parameters of a homogeneous medium using the reﬂec
tion coeﬃcients of the metamaterial medium, made up of diﬀerent number of
unit cells. In our method, we have used 1 to 20 unit cells. After the constitu
tive parameters are retrieved, the electromagnetic behavior of the metamaterial
slab (e.g., its reﬂection and transmission properties, ﬁeld distributions inside and
outside the metamaterial) is compared with that of the homogeneous equivalent.
Our numerical results show very good agreement between these two.
Again in Chapter 4, we aim to present another method for the retrieval of
surface wave propagation constants on any periodic or nonperiodic grounded
slab medium. The method is basically based on the diﬀerence in spread factors
of space and surface waves propagating on the surface of the slab. Since space
wave contribution of the total electric ﬁeld on the surface of the slab decays faster,
multiplying the ﬁeld data with the proper power of the lateral distance mainly
leaves the surface wave contribution, for large lateral distances. The electric
ﬁeld data, then, can be approximated as a summation of complex exponentials,
from which one can deduce how many surface wave modes are present and what
their propagation constants are. At the present, the method is applied to a
dielectric slab, for which the theoretical surface wave propagation constants are
well known, and numerical results have shown good agreement to the theory.
In Chapter 5, conclusions of the thesis are drawn. Appendix A contains some
properties of Bessel functions. In Appendix B, φ components of the magnetic and
electric ﬁelds of Section 2.9 are derived from their z components. Derivations of
the transparency and resonance conditions of Chapter 3 are given in Appendix C
4
and Appendix D, respectively. Throughout this thesis, an e
jωt
time dependence
is assumed and suppressed.
5
Chapter 2
Wave Propagation in
Metamaterial Structures
In this chapter, electromagnetic wave propagation in diﬀerent metamaterial
structures is investigated. The metamaterial geometries are chosen from sim
ple canonical geometries, such that an exact analytical eigenfunction solution
can be obtained.
Metamaterials are artiﬁcial materials which can have negative eﬀective elec
tric permittivity (ε
eff
) and/or negative eﬀective magnetic permeability (µ
eff
).
The signs of the aforementioned eﬀective complex constitutive parameters are
based on the signs of their real parts, whereas their imaginary parts indicate the
presence of electric or magnetic losses, respectively. Therefore, metamaterials
form four groups, depending on their constitutive parameters:
• Double Positive (DPS): Re¦ε¦ > 0, Re¦µ¦ > 0
• Double Negative (DNG): Re¦ε¦ < 0, Re¦µ¦ < 0
• Mu Negative (MNG): Re¦ε¦ > 0, Re¦µ¦ < 0
• Epsilon Negative (ENG): Re¦ε¦ < 0, Re¦µ¦ > 0
6
DNG metamaterials are also called Left Handed Materials (LHM) due to
their unique electromagnetic/optical properties like negative refraction, negative
phase velocity and negative Doppler shift, which follow a left hand rule system.
MNG and ENG metamaterials are also called Single Negative (SNG) materials,
because of the obvious fact that they have either negative eﬀective magnetic
permeability or negative eﬀective electric permittivity, respectively.
2.1 Wave Number, Index of Refraction and
Wave Impedance of Metamaterial Structures
Without loss of generality, the wave number, index of refraction and wave im
pedance of a medium are given as
k = ω
√
µε, (2.1)
n =
√
µ
r
ε
r
, (2.2)
η =
µ
ε
, (2.3)
respectively, where ω = 2πf is the angular frequency, µ
r
= µ/µ
0
and ε
r
= ε/ε
0
are the relative constitutive parameters.
The square roots which appear in (2.1)(2.3) create controversy, especially
when DNG materials are considered. Since both constitutive parameters are
complex quantities with their real parts being negative, the wave number, index
of refraction and wave impedance of the medium heavily depend on which branch
of the complex roots is selected. This controversy appeared in the scientiﬁc
community after the idea of perfect lens [23] and discussions focused on validity
of negative refraction and negative phase velocity [24].
7
The complex electric permittivity and the complex magnetic permeability of
a metamaterial medium can be expressed in polar form, respectively, as
ε = [ε[e
jφ
ε
, (2.4)
µ = [µ[e
jφ
µ
. (2.5)
Similarly, the wave number and the wave impedance of the metamaterial coating
can be written as
k = ω
√
µε = [k[e
jφ
k
, (2.6)
η =
µ/ε = [η[e
jφ
η
, (2.7)
respectively, where
[k[ = ω
[µ[[ε[, (2.8)
[η[ =
[µ[/[ε[, (2.9)
with
φ
k
=
1
2
(φ
µ
+φ
ε
), (2.10)
φ
η
=
1
2
(φ
µ
−φ
ε
). (2.11)
The choice of branches for the square roots in (2.10)(2.11) is based on causal
ity in a linear dispersive medium, the wave directions associated with reﬂection
and transmission from the interfaces and the direction of electromagnetic power
ﬂow. This choice is given and examined in details in [25] for DNG metamateri
als, ﬁrst introducing inﬁnitesimal electric and magnetic losses (as in the case of
metamaterials approximated by Drude and Lorentz medium models [23, 25, 26])
and then deciding on which complex branch gives the physically correct solution.
A similar analysis for DPS, MNG and ENG metamaterials [11] show that, the
choice of branches for the square roots given in (2.10)(2.11) still remains valid
for these metamaterials. With the assumed e
jωt
time dependence in this thesis,
and considering only passive media, the arguments of µ, ε, k and η for diﬀerent
types of metamaterials are tabulated in Table 2.1.
8
Table 2.1: Arguments of µ, ε, k and η for Diﬀerent Types of Metamaterials
φ
µ
φ
ε
φ
k
φ
η
DPS
−
π
2
, 0
−
π
2
, 0
−
π
2
, 0
−
π
4
,
π
4
DNG
−π, −
π
2
−π, −
π
2
−π, −
π
2
−
π
4
,
π
4
MNG
−π, −
π
2
−
π
2
, 0
−
3π
4
, −
π
4
−
π
2
, 0
ENG
−
π
2
, 0
−π, −
π
2
−
3π
4
, −
π
4
0,
π
2
Examination of Table 2.1 shows that for lossless DPS medium, wave number is
real and positive. For lossless DNG medium, wave number is real and negative.
For lossless DPS and DNG media, wave impedance is real and positive. For
lossless MNG and ENG media, the wave number is negative and imaginary,
which shows the presence of evanescent waves.
Remark: It is worthwhile to mention that when any of the constitutive
parameters of a metamaterial medium is a negative real number, −π should
be selected as its argument instead of π, as shown in Table 2.1. This becomes
important when intrinsic functions in a programming environment are directly
used (e.g., ANGLE, ATAN2).
2.2 Normal Incidence of Plane Waves on a
Metamaterial Slab
2.2.1 Introduction
Let us assume that a TEM
z
plane wave is traveling in the +z direction. An
inﬁnite length metamaterial slab of thickness d is placed between the z = 0 and
z = d planes in free space, without loss of generality. Here we will investigate
the reﬂection and transmission properties of the metamaterial slab as well as
9
the waves traveling inside the metamaterial slab. The incident electric ﬁeld is
assumed to be in the +x direction and the incident magnetic ﬁeld is in +y
direction. The problem geometry is depicted in Fig. 2.1.
2.2.2 Problem Geometry
Figure 2.1: Uniform plane wave normally incident on a metamaterial slab.
2.2.3 Electric and Magnetic Fields
The total electric and magnetic ﬁelds in Medium 1 are
E
1
= ´a
x
E
+
1
e
−jk
0
z
+ E
−
1
e
jk
0
z
, (2.12)
H
1
= ´a
y
E
+
1
η
0
e
−jk
0
z
−
E
−
1
η
0
e
jk
0
z
, (2.13)
respectively, where k
0
= ω
√
µ
0
ε
0
and η
0
=
µ
0
/ε
0
.
10
The total electric and magnetic ﬁelds in Medium 2 are
E
2
= ´a
x
E
+
2
e
−jkz
+ E
−
2
e
jkz
, (2.14)
H
2
= ´a
y
E
+
2
η
e
−jkz
−
E
−
2
η
e
jkz
, (2.15)
respectively, where k = ω
√
µε and η =
µ/ε.
The electric and magnetic ﬁelds in Medium 3 are
E
3
= ´a
x
E
+
3
e
−jk
0
z
, (2.16)
H
3
= ´a
y
E
+
3
η
0
e
−jk
0
z
, (2.17)
respectively, where k
0
and η
0
are the same as in Medium 1.
2.2.4 Solution of Boundary Conditions
Boundary Conditions at z = 0:
E
+
1
+E
−
1
= E
+
2
+ E
−
2
, (2.18)
E
+
1
η
0
−
E
−
1
η
0
=
E
+
2
η
−
E
−
2
η
. (2.19)
Boundary Conditions at z = d:
E
+
2
e
−jkd
+E
−
2
e
jkd
= E
+
3
e
−jk
0
d
, (2.20)
E
+
2
η
e
−jkd
−
E
−
2
η
e
jkd
=
E
+
3
η
0
e
−jk
0
d
. (2.21)
Rearranging equations (2.18)  (2.21) we get:
−E
−
1
+ E
+
2
+ E
−
2
= E
+
1
, (2.22)
E
−
1
+
η
0
η
E
+
2
−
η
0
η
E
−
2
= E
+
1
, (2.23)
e
−jkd
E
+
2
+e
jkd
E
−
2
−e
−jk
0
d
E
+
3
= 0, (2.24)
e
−jkd
η
E
+
2
−
e
jkd
η
E
−
2
−
e
−jk
0
d
η
0
E
+
3
= 0, (2.25)
11
which can also be written in matrix form as follows:
−1 1 1 0
1
η
0
η
−
η
0
η
0
0 e
−jkd
e
jkd
−e
−jk
0
d
0
e
−jkd
η
−
e
jkd
η
−
e
−jk
0
d
η
0
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
E
−
1
E
+
2
E
−
2
E
+
3
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
=
E
+
1
E
+
1
0
0
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
. (2.26)
Using Symbolic Math Toolbox of MATLAB, the solution to this system of
equations can be found as:
E
−
1
=
j(η
2
−η
2
0
) sin kd
2ηη
0
cos kd + j(η
2
+ η
2
0
) sin kd
E
+
1
, (2.27)
E
+
2
=
η(η + η
0
)e
jkd
2ηη
0
cos kd + j(η
2
+ η
2
0
) sin kd
E
+
1
, (2.28)
E
−
2
=
η(η
0
−η)e
−jkd
2ηη
0
cos kd + j(η
2
+ η
2
0
) sin kd
E
+
1
, (2.29)
E
+
3
=
2ηη
0
e
jk
0
d
2ηη
0
cos kd + j(η
2
+ η
2
0
) sin kd
E
+
1
. (2.30)
Deﬁning
ζ =
η
η
0
=
µ
r
ε
r
, (2.31)
and using relation (2.31), equations (2.27)(2.30) can be reduced to:
E
−
1
=
j(ζ
2
−1) sin kd
2ζ cos kd + j(ζ
2
+ 1) sin kd
E
+
1
, (2.32)
E
+
2
=
(ζ
2
+ ζ)e
jkd
2ζ cos kd + j(ζ
2
+ 1) sin kd
E
+
1
, (2.33)
E
−
2
=
(ζ −ζ
2
)e
−jkd
2ζ cos kd + j(ζ
2
+ 1) sin kd
E
+
1
, (2.34)
E
+
3
=
2ζe
jk
0
d
2ζ cos kd + j(ζ
2
+ 1) sin kd
E
+
1
. (2.35)
Note that, the solutions (2.27)(2.30) or (2.32)(2.35) are valid for all four
types of metamaterials (i.e., DPS, DNG, MNG and ENG) provided that k and
η are calculated as in Section 2.1.
12
2.3 Inﬁnite Length Metamaterial Cylinder
Near an Inﬁnite Length Electric Line
Source: TM
z
Polarization
2.3.1 Introduction
An inﬁnite line of constant electric current is placed in the vicinity of a circular
metamaterial cylinder of inﬁnite length. The scattering and transmission by the
metamaterial cylinder is examined for TM
z
polarization. The problem geometry
is given in Fig. 2.2.
2.3.2 Problem Geometry
Figure 2.2: Metamaterial cylinder near an electric line source. (a) Side view, (b)
Top view.
13
2.3.3 Electric Line Source and Incident Electric Field
For the line source of constant electric current, I
e
, in Fig. 2.2, the electric ﬁeld
generated everywhere by the source in the absence of the cylinder is given as [27]
E
i
z
= −
k
2
0
I
e
4ωε
0
H
(2)
0
(k
0
[ ¯ ρ − ¯ ρ
[), (2.36)
which we will refer as the incident electric ﬁeld. Using the addition theorem
for Hankel functions [28], (2.36) can be written in the series expansion form as
follows [27]:
E
i
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
ρ)H
(2)
n
(k
0
ρ
)e
jn(φ−φ
)
ρ ≤ ρ
, (2.37)
E
i
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
ρ
)H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
ρ ≥ ρ
. (2.38)
2.3.4 Scattered and Transmitted Electric Fields
Similar to the incident ﬁeld expressions in (2.37) and (2.38), we will deﬁne the
scattered and transmitted electric ﬁelds in series expansion form, respectively as
E
s
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
a ≤ ρ ≤ ρ
, ρ ≥ ρ
, (2.39)
E
t
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
d
n
J
n
(kρ)e
jn(φ−φ
)
0 ≤ ρ ≤ a. (2.40)
For the scattered ﬁeld, our deﬁnition should include H
(2)
n
(k
0
ρ) term which
represents +´ ρ wave propagation. For the transmitted ﬁeld, our deﬁnition should
include J
n
(kρ) term which represents a standing wave and also avoids a blow up
at ρ = 0 (due to Y
n
). The ﬁelds are 2π periodic in φ, so e
jn(φ−φ
)
term is inserted
to show this and to be in accordance with the incident ﬁeld expressions and also
for convenience. The −
k
2
0
I
e
4ωε
0
terms are just for convenience in calculations, which
in fact could be included in c
n
and/or d
n
.
14
2.3.5 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous at the surface of
the cylinder, due to the boundary conditions. Therefore,
E
i
z
(ρ = a) + E
s
z
(ρ = a) = E
t
z
(ρ = a), (2.41)
−
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
e
jn(φ−φ
)
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
d
n
J
n
(ka)e
jn(φ−φ
)
, (2.42)
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a) = d
n
J
n
(ka), (2.43)
d
n
=
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.44)
2.3.6 Incident, Scattered and Transmitted Magnetic
Fields
The radial and tangential components of the magnetic ﬁelds are derived from
the electric ﬁelds using the Maxwell’s equation:
H = −
1
jωµ
∇E, (2.45)
E = ´a
z
E
z
, (2.46)
H = −
1
jωµ
´a
ρ
1
ρ
∂E
z
∂φ
−´a
φ
∂E
z
∂ρ
, (2.47)
H
ρ
= −
1
jωµ
1
ρ
∂E
z
∂φ
, (2.48)
H
φ
=
1
jωµ
∂E
z
∂ρ
. (2.49)
Since H
φ
is the only component of the magnetic ﬁelds we will utilize in bound
ary conditions, we are only interested in equation (2.49).
15
One important point where attention must be paid is the partial derivative of
E
z
with respect to ρ. Since in our E
z
deﬁnitions we have the Bessel and Hankel
functions, their derivatives should be taken with respect to the entire argument
of the corresponding Bessel and Hankel functions.
Let F(βρ) be a function representing the Bessel and Hankel functions. Then,
∂F(βρ)
∂ρ
=
∂(βρ)
∂ρ
∂F(βρ)
∂(βρ)
= β
∂F(βρ)
∂(βρ)
. (2.50)
Utilizing (2.49) and (2.50), and also keeping in mind that the derivatives
are all with respect to the entire arguments, the tangential components of the
magnetic ﬁelds are obtained as follows:
H
i
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
J
n
(k
0
ρ)H
(2)
n
(k
0
ρ
)e
jn(φ−φ
)
ρ ≤ ρ
, (2.51)
H
i
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
J
n
(k
0
ρ
)H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
ρ ≥ ρ
, (2.52)
H
s
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
a ≤ ρ ≤ ρ
, ρ ≥ ρ
, (2.53)
H
t
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
k
+∞
¸
n=−∞
d
n
J
n
(kρ)e
jn(φ−φ
)
0 ≤ ρ ≤ a. (2.54)
2.3.7 Boundary Conditions for Magnetic Fields
The tangential components of the magnetic ﬁelds are continuous at the surface
of the cylinder due to the boundary conditions. Therefore,
H
i
φ
(ρ = a) + H
s
φ
(ρ = a) = H
t
φ
(ρ = a), (2.55)
−
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
e
jn(φ−φ
)
= −
k
2
0
I
e
4ωε
0
1
jωµ
k
+∞
¸
n=−∞
d
n
J
n
(ka)e
jn(φ−φ
)
, (2.56)
16
k
0
µ
0
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
=
k
µ
d
n
J
n
(ka). (2.57)
Expressing
k
µ
=
k
0
µ
0
√
µ
r
ε
r
µ
r
=
k
0
µ
0
ε
r
µ
r
=
k
0
µ
0
1
ζ
, (2.58)
where ζ =
µ
r
/ε
r
as previously deﬁned in (2.31), and substituting (2.58) in
(2.57), we get:
ζ
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
= d
n
J
n
(ka), (2.59)
d
n
=
ζ
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.60)
2.3.8 Simultaneous Solution of the Boundary Conditions
for Electric and Magnetic Fields
Now we have two equations for d
n
: (2.44) and (2.60), which are derived from the
boundary conditions for the electric and magnetic ﬁelds, respectively. Our next
step will be to equate these equations:
d
n
=
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
=
ζ
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
,
(2.61)
J
n
(ka)
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
= ζJ
n
(ka)
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
, (2.62)
J
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
J
n
(ka)H
(2)
n
(k
0
a)
= ζJ
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
ζJ
n
(ka)H
(2)
n
(k
0
a), (2.63)
c
n
J
n
(ka)H
(2)
n
(k
0
a) −c
n
ζJ
n
(ka)H
(2)
n
(k
0
a)
= ζJ
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
) −J
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
), (2.64)
17
c
n
J
n
(ka)H
(2)
n
(k
0
a) −ζJ
n
(ka)H
(2)
n
(k
0
a)
= ζJ
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
) −J
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
), (2.65)
c
n
=
ζJ
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
) −J
n
(ka)J
n
(k
0
a)H
(2)
n
(k
0
ρ
)
J
n
(ka)H
(2)
n
(k
0
a) −ζJ
n
(ka)H
(2)
n
(k
0
a)
, (2.66)
c
n
=
ζJ
n
(ka)J
n
(k
0
a) −J
n
(ka)J
n
(k
0
a)
J
n
(ka)H
(2)
n
(k
0
a) −ζJ
n
(ka)H
(2)
n
(k
0
a)
H
(2)
n
(k
0
ρ
), (2.67)
where d
n
can be found from (2.44)
d
n
=
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
, (2.68)
or from (2.60)
d
n
=
ζ
J
n
(k
0
a)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.69)
Now, the incident, scattered and transmitted electric and magnetic ﬁelds can
be calculated using (2.37)(2.40) and (2.51)(2.54), respectively.
Remark: Note that, since the electric line source is placed outside the meta
material cylinder, when applying the boundary conditions for electric and mag
netic ﬁelds at ρ = a, (2.37) and (2.51) are used, respectively. If the line source
is placed inside the cylinder, boundary conditions should be written using (2.38)
and (2.52).
2.3.9 Calculation of the Radiation Patterns
To calculate the radiation patterns, the following large argument approximation
for Hankel functions of the second kind is used:
lim
k
0
ρ→∞
H
(2)
n
(k
0
ρ) =
2
πk
0
ρ
e
−j[k
0
ρ−π/4−n(π/2)]
. (2.70)
18
The total electric ﬁeld for ρ > ρ
can be written as:
E
r
z
(ρ, φ) = E
i
z
(ρ, φ) + E
s
z
(ρ, φ) = −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
[J
n
(k
0
ρ
) + c
n
] H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
.
(2.71)
Using (2.70),
lim
k
0
ρ→∞
E
r
z
(ρ, φ) = −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
[J
n
(k
0
ρ
) + c
n
]
2
πk
0
ρ
e
−j[k
0
ρ−π/4−n(π/2)]
e
jn(φ−φ
)
,
(2.72)
lim
k
0
ρ→∞
E
r
z
(ρ, φ) = −
k
2
0
I
e
4ωε
0
2
πk
0
ρ
e
−j(k
0
ρ−π/4)
+∞
¸
n=−∞
[J
n
(k
0
ρ
) + c
n
] e
jn(φ−φ
+π/2)
.
(2.73)
The radiation density is:
W
rad
(ρ, φ) = lim
k
0
ρ→∞
[E
r
z
(ρ, φ)[
2
2η
0
=
k
3
0
I
2
e
16η
0
ω
2
ε
2
0
πρ
+∞
¸
n=−∞
[J
n
(k
0
ρ
) + c
n
] e
jn(φ−φ
+π/2)
2
.
(2.74)
The radiation intensity is:
U(φ) = ρW
rad
(ρ, φ) =
k
3
0
I
2
e
16η
0
ω
2
ε
2
0
π
+∞
¸
n=−∞
[J
n
(k
0
ρ
) + c
n
] e
jn(φ−φ
+π/2)
2
. (2.75)
2.3.10 Numerical Results
Fig. 2.3 shows the magnitude of electric ﬁeld for diﬀerent choices of constitutive
parameters when f = 30GHz, λ
0
= 0.01m, a = λ
0
, ρ
= 1.5λ
0
, φ
= 0
◦
.
For DNG cases, focusing towards the line source and inside the metamaterial
cylinder is noticed. In Fig. 2.3 (a), this focusing occurs on the surface of the
cylinder. These unique focusing properties of DNG metamaterials are mainly
results of negative refraction.
19
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
2
4
6
8
10
12
14
x 10
4
(a)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
5
10
15
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(b)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
2
4
6
8
10
12
x 10
4
(c)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
2
4
6
8
10
12
14
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(d)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
2
4
6
8
10
12
x 10
4
(e)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
2
4
6
8
10
12
14
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(f)
Figure 2.3: Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2
20
2.4 Normally Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Cylinder:
TM
z
Polarization
2.4.1 Introduction
A uniform plane wave is normally incident on a metamaterial cylinder of inﬁnite
length. The plane wave travels in the −x direction. We will examine here the
scattering and transmission by the cylinder in the case the polarization of the
plane wave is TM
z
. For the −x propagation direction and TM
z
polarization,
electric ﬁeld is directed along the +z axis and magnetic ﬁeld is directed along
the +y axis. The problem geometry is given in Fig. 2.4.
2.4.2 Problem Geometry
Figure 2.4: Uniform plane wave incident on a metamaterial cylinder: TM
z
Po
larization.
21
2.4.3 Uniform Plane Wave and Incident Electric Field
Let us assume that a TM
z
polarized uniform plane wave is traveling in the −x
direction. This means electric ﬁeld is directed along the +z axis and magnetic
ﬁeld is directed along the +y axis. Therefore the electric ﬁeld can be written as
[27]
E
i
z
= E
0
e
jk
0
x
= E
0
e
jk
0
ρ cos φ
. (2.76)
By wave transformations and utilizing orthogonality condition [27,28], (2.76) can
be written in the series expansion form as follows [27]:
E
i
z
= E
0
+∞
¸
n=−∞
j
n
J
n
(k
0
ρ)e
jnφ
ρ ≥ a. (2.77)
2.4.4 Scattered and Transmitted Electric Fields
Similar to the incident ﬁeld expression in (2.77), we will deﬁne the scattered and
transmitted electric ﬁelds in series expansion form, respectively, as
E
s
z
= E
0
+∞
¸
n=−∞
j
n
c
n
H
(2)
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.78)
E
t
z
= E
0
+∞
¸
n=−∞
j
n
d
n
J
n
(kρ)e
jnφ
0 ≤ ρ ≤ a. (2.79)
For the scattered ﬁeld, our deﬁnition should include H
(2)
n
(k
0
ρ) term which
represents +´ ρ wave propagation. For the transmitted ﬁeld, our deﬁnition should
include J
n
(kρ) term which represents a standing wave and also avoids a blow up
at ρ = 0 (due to Y
n
). The ﬁelds are 2π periodic in φ, so e
jnφ
term is inserted
to show this and to be in accordance with the incident ﬁeld expressions and also
for convenience. The j
n
terms are just for convenience in calculations, which in
fact could be included in c
n
and/or d
n
.
22
2.4.5 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous at the surface of
the cylinder due to the boundary conditions. Therefore,
E
i
z
(ρ = a) + E
s
z
(ρ = a) = E
t
z
(ρ = a), (2.80)
E
0
+∞
¸
n=−∞
j
n
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
e
jnφ
= E
0
+∞
¸
n=−∞
j
n
d
n
J
n
(ka)e
jnφ
, (2.81)
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a) = d
n
J
n
(ka), (2.82)
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.83)
2.4.6 Incident, Scattered and Transmitted Magnetic
Fields
Utilizing (2.49) and (2.50), the tangential components of the magnetic ﬁelds are
obtained as
H
i
φ
= E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
n
J
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.84)
H
s
φ
= E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
n
c
n
H
(2)
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.85)
H
t
φ
= E
0
1
jωµ
k
+∞
¸
n=−∞
j
n
d
n
J
n
(kρ)e
jnφ
0 ≤ ρ ≤ a. (2.86)
The derivatives in (2.197)(2.199) are again with respect to the entire argu
ments.
23
2.4.7 Boundary Conditions for Magnetic Fields
The tangential components of the magnetic ﬁelds are continuous at the surface
of the cylinder due to the boundary conditions. Therefore,
H
i
φ
(ρ = a) + H
s
φ
(ρ = a) = H
t
φ
(ρ = a), (2.87)
E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
n
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
e
jnφ
= E
0
1
jωµ
k
+∞
¸
n=−∞
j
n
d
n
J
n
(ka)e
jnφ
,
(2.88)
k
0
µ
0
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
=
k
µ
d
n
J
n
(ka). (2.89)
Using (2.58) in (2.231),
ζ
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
= d
n
J
n
(ka), (2.90)
d
n
=
ζ
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.91)
2.4.8 Simultaneous Solution of the Boundary Conditions
for Electric and Magnetic Fields
Now we have two equations for d
n
: (2.83) and (2.91), which are derived from the
boundary conditions for the electric and magnetic ﬁelds, respectively. Our next
step will be to equate these equations:
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
=
ζ
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
, (2.92)
J
n
(ka)
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
= ζJ
n
(ka)
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
, (2.93)
J
n
(ka)J
n
(k
0
a) + c
n
J
n
(ka)H
(2)
n
(k
0
a) = ζJ
n
(ka)J
n
(k
0
a) + c
n
ζJ
n
(ka)H
(2)
n
(k
0
a),
(2.94)
24
c
n
J
n
(ka)H
(2)
n
(k
0
a) −c
n
ζJ
n
(ka)H
(2)
n
(k
0
a) = ζJ
n
(ka)J
n
(k
0
a) −J
n
(ka)J
n
(k
0
a),
(2.95)
c
n
J
n
(ka)H
(2)
n
(k
0
a) −ζJ
n
(ka)H
(2)
n
(k
0
a)
= ζJ
n
(ka)J
n
(k
0
a) −J
n
(ka)J
n
(k
0
a),
(2.96)
c
n
=
ζJ
n
(ka)J
n
(k
0
a) −J
n
(ka)J
n
(k
0
a)
J
n
(ka)H
(2)
n
(k
0
a) −ζJ
n
(ka)H
(2)
n
(k
0
a)
, (2.97)
where
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
, (2.98)
or
d
n
=
ζ
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.99)
2.4.9 Numerical Results
Fig. 2.5 shows the numerical results for f = 30GHz, λ
0
= 0.01m, a = λ
0
. In Fig.
2.5 (a), there are three foci close to the interface and inside the metamaterial
cylinder. In Fig. 2.5 (b), there is one dominant focus inside the cylinder, while
the other two diminish. Finally in Fig. 2.5 (c) there is one focus inside the
cylinder and another outside. Both foci are at the other side of the cylinder
(w.r.t plane wave illumination) as predicted for a DPS dielectric lens.
25
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
(a)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
0.5
1
1.5
2
2.5
3
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(b)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
3
3.5
4
(c)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(d)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
3
3.5
(e)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0.5
1
1.5
2
2.5
3
3.5
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(f)
Figure 2.5: Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2
26
2.5 Normally Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Cylinder:
TE
z
Polarization
2.5.1 Introduction
A uniform plane wave is normally incident on a metamaterial cylinder of inﬁnite
length, traveling in the −x direction, as in Section 2.4. We will examine here
the scattering and transmission by the cylinder in the case the polarization of
the plane wave is TE
z
. For the −x propagation direction and TE
z
polarization,
magnetic ﬁeld is directed along the +z axis and electric ﬁeld is directed along
the −y axis. The problem geometry is depicted in Fig. 2.6.
2.5.2 Problem Geometry
Figure 2.6: Uniform plane wave incident on a metamaterial cylinder: TE
z
Po
larization.
27
2.5.3 Uniform Plane Wave and Incident Magnetic Field
Let us assume that a TE
z
polarized uniform plane wave is traveling in the −x
direction, which means the magnetic ﬁeld is directed along the +z axis and the
electric ﬁeld is directed along the −y axis. Therefore, the magnetic ﬁeld can be
written as [27]
H
i
z
= H
0
e
jk
0
x
= H
0
e
jk
0
ρ cos φ
. (2.100)
By wave transformations and utilizing orthogonality condition [27,28], (2.100)
can be written in series expansion form as [27]
H
i
z
= H
0
+∞
¸
n=−∞
j
n
J
n
(k
0
ρ)e
jnφ
ρ ≥ a. (2.101)
2.5.4 Scattered and Transmitted Magnetic Fields
Similar to the incident ﬁeld expression in (2.101), we will deﬁne the scattered
and transmitted magnetic ﬁelds in series expansion form respectively as follows:
H
s
z
= H
0
+∞
¸
n=−∞
j
n
c
n
H
(2)
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.102)
H
t
z
= H
0
+∞
¸
n=−∞
j
n
d
n
J
n
(kρ)e
jnφ
0 ≤ ρ ≤ a. (2.103)
2.5.5 Boundary Conditions for Magnetic Fields
The tangential components of the magnetic ﬁelds are continuous at the surface
of the cylinder due to the boundary conditions. Hence,
H
i
z
(ρ = a) + H
s
z
() = H
t
z
(ρ = a), (2.104)
H
0
+∞
¸
n=−∞
j
n
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
e
jnφ
= H
0
+∞
¸
n=−∞
j
n
d
n
J
n
(ka)e
jnφ
, (2.105)
28
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a) = d
n
J
n
(ka), (2.106)
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
. (2.107)
2.5.6 Incident, Scattered and Transmitted Electric Fields
The radial and tangential components of the electric ﬁelds are derived from the
magnetic ﬁelds using the Maxwell’s equation:
E =
1
jωε
∇H, (2.108)
H = ´a
z
H
z
, (2.109)
E =
1
jωε
´a
ρ
1
ρ
∂H
z
∂φ
−´a
φ
∂H
z
∂ρ
, (2.110)
E
ρ
=
1
jωε
1
ρ
∂H
z
∂φ
, (2.111)
E
φ
= −
1
jωε
∂H
z
∂ρ
. (2.112)
Since E
φ
is the only component of the electric ﬁelds we will utilize in boundary
conditions, we are only interested in equation (2.112).
Utilizing (2.112), tangential components of the electric ﬁelds are obtained as
E
i
φ
= H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
n
J
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.113)
E
s
φ
= H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
n
c
n
H
(2)
n
(k
0
ρ)e
jnφ
ρ ≥ a, (2.114)
E
t
φ
= H
0
−1
jωε
k
+∞
¸
n=−∞
j
n
d
n
J
n
(kρ)e
jnφ
0 ≤ ρ ≤ a. (2.115)
29
2.5.7 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous at the surface of
the cylinder due to the boundary conditions. For this reason,
E
i
φ
(ρ = a) + E
s
φ
(ρ = a) = E
t
φ
(ρ = a), (2.116)
H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
n
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
e
jnφ
= H
0
−1
jωε
k
+∞
¸
n=−∞
j
n
d
n
J
n
(ka)e
jnφ
,
(2.117)
k
0
ε
0
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
=
k
ε
d
n
J
n
(ka). (2.118)
Expressing,
k
ε
=
k
0
ε
0
√
µ
r
ε
r
ε
r
=
k
0
ε
0
µ
r
ε
r
=
k
0
ε
0
ζ, (2.119)
and substituting (2.119) in (2.118),
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a) = d
n
ζJ
n
(ka), (2.120)
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
ζJ
n
(ka)
. (2.121)
2.5.8 Simultaneous Solution of the Boundary Conditions
for Magnetic and Electric Fields
Now we have two equations for d
n
: (2.107) and (2.121), which are derived from
the boundary conditions for the magnetic and electric ﬁelds, respectively. Our
next step will be to equate these equations:
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
ζJ
n
(ka)
, (2.122)
ζJ
n
(ka)
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
= J
n
(ka)
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
, (2.123)
ζJ
n
(ka)J
n
(k
0
a) + c
n
ζJ
n
(ka)H
(2)
n
(k
0
a) = J
n
(ka)J
n
(k
0
a) + c
n
J
n
(ka)H
(2)
n
(k
0
a),
(2.124)
30
c
n
ζJ
n
(ka)H
(2)
n
(k
0
a) −c
n
J
n
(ka)H
(2)
n
(k
0
a) = J
n
(ka)J
n
(k
0
a) −ζJ
n
(ka)J
n
(k
0
a),
(2.125)
c
n
ζJ
n
(ka)H
(2)
n
(k
0
a) −J
n
(ka)H
(2)
n
(k
0
a)
= J
n
(ka)J
n
(k
0
a) −ζJ
n
(ka)J
n
(k
0
a),
(2.126)
c
n
=
J
n
(ka)J
n
(k
0
a) −ζJ
n
(ka)J
n
(k
0
a)
ζJ
n
(ka)H
(2)
n
(k
0
a) −J
n
(ka)H
(2)
n
(k
0
a)
, (2.127)
where
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
J
n
(ka)
, (2.128)
or
d
n
=
J
n
(k
0
a) + c
n
H
(2)
n
(k
0
a)
ζJ
n
(ka)
. (2.129)
2.5.9 Numerical Results
Using duality, interchanging µ
r
and ε
r
, the same results in Fig. 2.5 can be
obtained (for magnitude of the magnetic ﬁeld).
31
2.6 Inﬁnite Length Metamaterial Coated Con
ducting Cylinder Near an Inﬁnite Length
Electric Line Source: TM
z
Polarization
2.6.1 Introduction
An inﬁnite line of constant electric current is placed in the vicinity of an inﬁ
nite length metamaterial coated conducting cylinder. The scattering and trans
mission by the metamaterial coated conducting cylinder is examined for TM
z
polarization. The problem geometry is shown in Fig. 2.7.
2.6.2 Problem Geometry
Figure 2.7: Metamaterial coated conducting cylinder near an electric line source
(Cross section view).
32
2.6.3 Electric Line Source and Incident Electric Field
For the line source of constant electric current, I
e
, in Fig. 2.7, the electric ﬁeld
generated everywhere by the source in the absence of the cylinder is given as [27]
E
i
z
= −
k
2
0
I
e
4ωε
0
H
(2)
0
(k
0
[ ¯ ρ − ¯ ρ
[), (2.130)
which is our incident electric ﬁeld. By the addition theorem for Hankel functions
[28], (2.130) can be written in the series expansion form as [27]
E
i
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
ρ)H
(2)
n
(k
0
ρ
)e
jn(φ−φ
)
ρ ≤ ρ
, (2.131)
E
i
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
ρ
)H
(2)
n
(k
0
ρ)e
jn(φ−φ
)
ρ ≥ ρ
. (2.132)
2.6.4 Scattered and Transmitted Electric Fields
Similar to the incident ﬁeld expressions in (2.131) and (2.132), we will deﬁne the
scattered and transmitted electric ﬁelds in series expansion form, respectively, as
E
s
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.133)
E
t
z
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
[a
n
J
n
(kρ) + b
n
Y
n
(kρ)] e
jn(φ−φ
0
)
. (2.134)
For the scattered ﬁeld, our deﬁnition should include H
(2)
n
(k
0
ρ) term which
represents +´ ρ wave propagation. For the transmitted ﬁeld, our deﬁnition should
include J
n
(kρ) and Y
n
(kρ) terms which represent standing waves. The ﬁelds are
2π periodic in φ, so e
jn(φ−φ
0
)
term is inserted to show this and to be in accordance
with the incident ﬁeld expressions and also for convenience. The −
k
2
0
I
e
4ωε
0
terms are
just for convenience in calculations, which in fact could be included in a
n
and/or
b
n
and/or c
n
.
33
2.6.5 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous on the outer
surface of the metamaterial coating, due to the boundary conditions. Also, on
the inner surface of the metamaterial coating (i.e., on the conducting cylinder
surface) tangential electric ﬁeld should vanish. Therefore,
E
t
z
(ρ = a) = 0, (2.135)
−
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
[a
n
J
n
(kρ) + b
n
Y
n
(kρ)] e
jn(φ−φ
0
)
ρ=a
= 0, (2.136)
a
n
J
n
(ka) + b
n
Y
n
(ka) = 0, (2.137)
E
i
z
(ρ = b) + E
s
z
(ρ = b) = E
t
z
(ρ = b), (2.138)
−
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
J
n
(k
0
b)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= −
k
2
0
I
e
4ωε
0
+∞
¸
n=−∞
[a
n
J
n
(kb) + b
n
Y
n
(kb)] e
jn(φ−φ
0
)
, (2.139)
J
n
(k
0
b)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
b) = a
n
J
n
(kb) + b
n
Y
n
(kb). (2.140)
2.6.6 Incident, Scattered and Transmitted Magnetic
Fields
Utilizing (2.49) and (2.50), the tangential components of the magnetic ﬁelds are
obtained as
H
i
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
J
n
(k
0
ρ)H
(2)
n
(k
0
ρ
)e
jn(φ−φ
0
)
, (2.141)
H
s
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.142)
34
H
t
φ
= −
k
2
0
I
e
4ωε
0
1
jωµ
k
+∞
¸
n=−∞
[a
n
J
n
(kρ) + b
n
Y
n
(kρ)] e
jn(φ−φ
0
)
. (2.143)
2.6.7 Boundary Conditions for Magnetic Fields
Tangential components of the magnetic ﬁelds are continuous on the outer surface
of the metamaterial coating, due to the boundary conditions. Therefore,
H
i
φ
(ρ = b) + H
s
φ
(ρ = b) = H
t
φ
(ρ = b), (2.144)
−
k
2
0
I
e
4ωε
0
1
jωµ
0
k
0
+∞
¸
n=−∞
J
n
(k
0
b)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= −
k
2
0
I
e
4ωε
0
1
jωµ
k
+∞
¸
n=−∞
[a
n
J
n
(kb) + b
n
Y
n
(kb)] e
jn(φ−φ
0
)
, (2.145)
k
0
µ
0
J
n
(k
0
b)H
(2)
n
(k
0
ρ
) + c
n
H
(2)
n
(k
0
b)
=
k
µ
[a
n
J
n
(kb) + b
n
Y
n
(kb)] . (2.146)
Substituting (2.58) in (2.146),
ζJ
n
(k
0
b)H
(2)
n
(k
0
ρ
) + c
n
ζH
(2)
n
(k
0
b) = a
n
J
n
(kb) + b
n
Y
n
(kb). (2.147)
2.6.8 Simultaneous Solution of the Boundary Conditions
for Electric and Magnetic Fields
Now we have three unknowns and three equations. Rearranging equations
(2.137), (2.140) and (2.147) we get:
J
n
(ka)a
n
+Y
n
(ka)b
n
= 0, (2.148)
J
n
(kb)a
n
+Y
n
(kb)b
n
−H
(2)
n
(k
0
b)c
n
= J
n
(k
0
b)H
(2)
n
(k
0
ρ
), (2.149)
J
n
(kb)a
n
+ Y
n
(kb)b
n
−ζH
(2)
n
(k
0
b)c
n
= ζJ
n
(k
0
b)H
(2)
n
(k
0
ρ
), (2.150)
35
which can also be written in matrix form as:
J
n
(ka) Y
n
(ka) 0
J
n
(kb) Y
n
(kb) −H
(2)
n
(k
0
b)
J
n
(kb) Y
n
(kb) −ζH
(2)
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
a
n
b
n
c
n
¸
¸
¸
¸
¸
¸
¸
¸
=
0
J
n
(k
0
b)H
(2)
n
(k
0
ρ
)
ζJ
n
(k
0
b)H
(2)
n
(k
0
ρ
)
¸
¸
¸
¸
¸
¸
¸
¸
. (2.151)
Using Symbolic Math Toolbox of MATLAB, the solution to this system of
equations can be found as:
a
n
=
ζY
n
(ka)
H
(2)
n
(k
0
b)J
n
(k
0
b) −H
(2)
n
(k
0
b)J
n
(k
0
b)
D
H
(2)
n
(k
0
ρ
), (2.152)
b
n
=
ζJ
n
(ka)
J
n
(k
0
b)H
(2)
n
(k
0
b) −J
n
(k
0
b)H
(2)
n
(k
0
b)
D
H
(2)
n
(k
0
ρ
), (2.153)
c
n
=
N
D
H
(2)
n
(k
0
ρ
), (2.154)
where
N = J
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)]
−ζJ
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)] , (2.155)
D = ζH
(2)
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)]
−H
(2)
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)] . (2.156)
Using the following Wronskian will further simplify a
n
and b
n
:
J
n
(x)H
(2)
n
(x) −J
n
(x)H
(2)
n
(x) =
−2j
πx
. (2.157)
2.6.9 Electric Line Source Inside the Metamaterial
Coating
When the electric line source is placed inside the metamaterial coating, the for
mulation given up to here has to be modiﬁed. This is mainly due to the electric
36
ﬁeld deﬁnition of the electric line source. Utilizing the previous procedure, it can
be seen that (2.137), (2.140) and (2.147) have to be modiﬁed as
J
n
(ka)H
(2)
n
(kρ
) + a
n
J
n
(ka) + b
n
Y
n
(ka) = 0, (2.158)
J
n
(kρ
)H
(2)
n
(kb) + a
n
J
n
(kb) + b
n
Y
n
(kb) = c
n
H
(2)
n
(k
0
b), (2.159)
J
n
(kρ
)H
(2)
n
(kb) + a
n
J
n
(kb) + b
n
Y
n
(kb) = c
n
ζH
(2)
n
(k
0
b), (2.160)
respectively. The system of equations can be written in matrix form as
J
n
(ka) Y
n
(ka) 0
J
n
(kb) Y
n
(kb) −H
(2)
n
(k
0
b)
J
n
(kb) Y
n
(kb) −ζH
(2)
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
a
n
b
n
c
n
¸
¸
¸
¸
¸
¸
¸
¸
=
−J
n
(ka)H
(2)
n
(kρ
)
−J
n
(kρ
)H
(2)
n
(kb)
−J
n
(kρ
)H
(2)
n
(kb)
¸
¸
¸
¸
¸
¸
¸
¸
. (2.161)
Using Symbolic Math Toolbox of MATLAB, the solution to this system of
equations can be found as
a
n
=
A
1
+ A
2
D
, (2.162)
b
n
=
B
1
+B
2
D
, (2.163)
c
n
=
C
1
+ C
2
+C
3
D
, (2.164)
where
A
1
=
H
(2)
n
(k
0
b)Y
n
(kb) −ζH
(2)
n
(k
0
b)Y
n
(kb)
J
n
(ka)H
(2)
n
(kρ
), (2.165)
A
2
=
ζH
(2)
n
(kb)H
(2)
n
(k
0
b) −H
(2)
n
(kb)H
(2)
n
(k
0
b)
Y
n
(ka)J
n
(kρ
), (2.166)
B
1
=
ζJ
n
(kb)H
(2)
n
(k
0
b) −J
n
(kb)H
(2)
n
(k
0
b)
J
n
(ka)H
(2)
n
(kρ
), (2.167)
B
2
=
H
(2)
n
(k
0
b)H
(2)
n
(kb) −ζH
(2)
n
(k
0
b)H
(2)
n
(kb)
J
n
(ka)J
n
(kρ
), (2.168)
37
C
1
= [J
n
(kb)Y
n
(kb) −J
n
(kb)Y
n
(kb)] J
n
(ka)H
(2)
n
(kρ
), (2.169)
C
2
= [Y
n
(ka)J
n
(kb) −Y
n
(kb)J
n
(ka)] J
n
(kρ
)H
(2)
n
(kb), (2.170)
C
3
= [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)] J
n
(kρ
)H
(2)
n
(kb), (2.171)
D = ζH
(2)
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)]
−H
(2)
n
(k
0
b) [J
n
(ka)Y
n
(kb) −J
n
(kb)Y
n
(ka)] . (2.172)
2.6.10 Numerical Results
Some of the numerical results are shown in Fig. 2.8 for f = 30GHz, λ
0
= 0.01m,
a = 0.5λ
0
, b = λ
0
, ρ
= 1.5λ
0
, φ
= 0
◦
. In Fig. 2.8 (a) there is a strong focus
on the outer surface of the metamaterial. In Fig. 2.8 (b), focus point moves
inside the cylinder. Finally in Fig. 2.8 (c) for the DPS case no focusing can be
observed.
38
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
2
4
6
8
10
12
14
x 10
4
(a)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
5
10
15
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(b)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
2
4
6
8
10
12
x 10
4
(c)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
2
4
6
8
10
12
14
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(d)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
1
2
3
4
5
6
7
8
9
10
11
x 10
4
(e)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
2
4
6
8
10
12
x 10
4
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(f)
Figure 2.8: Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2
39
2.7 Normally Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Coated
Conducting Cylinder: TM
z
Polarization
2.7.1 Introduction
A uniform plane wave is normally incident on a metamaterial coated conducting
cylinder of inﬁnite length. The plane wave travels in the direction which makes an
angle φ
0
with the +x axis. We will examine here the scattering and transmission
by the metamaterial coated conducting cylinder in the case the polarization of
the plane wave is TM
z
. In the numerical results of this section, the angle of
incidence φ
0
is selected as π. This corresponds to a plane wave traveling in the
−x direction, which is the case we have investigated in Section 2.4. The problem
geometry is depicted in Fig. 2.9.
2.7.2 Problem Geometry
x
y
ρ
σ = ∞
φ
φ
0
a
b
Plane Wave
ε
c
, µ
c
ε
0
, µ
0
Figure 2.9: Plane wave normally incident on a metamaterial coated conducting
cylinder.
40
2.7.3 Uniform Plane Wave and Incident Electric Field
Let us assume that a TM
z
polarized uniform plane wave is traveling in the
direction which makes an angle φ
0
with the +x axis. Electric ﬁeld is directed
along the +z axis. Referring to Figure 2.9 the electric ﬁeld can be written as
E
i
z
= E
0
e
−jk
0
(xcos φ
0
+y sin φ
0
)
, (2.173)
where
x = ρ cos φ , y = ρ sin φ. (2.174)
Therefore,
E
i
z
= E
0
e
−jk
0
(ρ cos φcos φ
0
+ρ sinφsin φ
0
)
,
= E
0
e
−jk
0
ρ(cos φcos φ
0
+sinφsinφ
0
)
,
= E
0
e
−jk
0
ρ cos(φ−φ
0
)
. (2.175)
The plane wave can be represented by an inﬁnite sum of cylindrical wave
functions:
E
i
z
= E
0
e
−jk
0
ρ cos(φ−φ
0
)
= E
0
+∞
¸
n=−∞
a
n
J
n
(k
0
ρ)e
jnφ
, (2.176)
since it must be 2π periodic in φ and ﬁnite at ρ = 0. Our next step is to ﬁnd the
coeﬃcients a
n
. Multiplying both sides of (2.176) by e
jmφ
, where m is an integer,
and integrating from 0 to 2π,
E
0
2π
0
e
−j(k
0
ρ cos(φ−φ
0
)+mφ)
dφ = E
0
2π
0
¸
+∞
¸
n=−∞
a
n
J
n
(k
0
ρ)e
j(n−m)φ
¸
dφ. (2.177)
Dropping ‘E
0
’s and interchanging the integration and summation, we have
2π
0
e
−j(k
0
ρ cos(φ−φ
0
)+mφ)
dφ =
+∞
¸
n=−∞
a
n
J
n
(k
0
ρ)
2π
0
e
j(n−m)φ
dφ. (2.178)
Utilizing the orthogonality condition of
2π
0
e
j(n−m)φ
dφ =
2π, n = m
0 , n = m
, (2.179)
41
the right hand side of (2.178) reduces to
+∞
¸
n=−∞
a
n
J
n
(k
0
ρ)
2π
0
e
j(n−m)φ
dφ = 2πa
m
J
m
(k
0
ρ). (2.180)
Using the integral of
2π
0
e
j(z cos φ+nφ)
dφ = 2πj
n
J
n
(z), (2.181)
and by a simple transformation ϕ = φ−φ
0
, the left side of (2.178) can be written
as
2π
0
e
−j(k
0
ρ cos(φ−φ
0
)+mφ)
dφ = e
−jmφ
0
2πj
−m
J
−m
(−k
0
ρ). (2.182)
Since
J
−m
(x) = (−1)
m
J
m
(x), (2.183)
and
J
m
(−x) = (−1)
m
J
m
(x), (2.184)
(2.182) can be written as
2π
0
e
−j(k
0
ρ cos(φ−φ
0
)+mφ)
dφ = e
−jmφ
0
2πj
−m
J
−m
(−k
0
ρ),
= e
−jmφ
0
2πj
−m
J
m
(k
0
ρ). (2.185)
Using (2.180) and (2.185) reduces (2.178) to
e
−jmφ
0
2πj
−m
J
m
(k
0
ρ) = 2πa
m
J
m
(k
0
ρ). (2.186)
Thus
a
m
= j
−m
e
−jmφ
0
. (2.187)
Therefore (2.176) can be written as
E
i
z
= E
0
e
−jk
0
ρ cos(φ−φ
0
)
,
= E
0
+∞
¸
n=−∞
a
n
J
n
(k
0
ρ)e
jnφ
,
= E
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ)e
jn(φ−φ
0
)
. (2.188)
42
2.7.4 Scattered and Transmitted Electric Fields
Similar to the incident ﬁeld expression in (2.188), we will deﬁne the scattered
and transmitted electric ﬁelds in series expansion form, respectively as follows:
E
s
z
= E
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.189)
E
t
z
= E
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
. (2.190)
2.7.5 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous on the outer
surface of the metamaterial coating, due to the boundary conditions. Also, on
the inner surface of the metamaterial coating (i.e., on the conducting cylinder
surface) tangential electric ﬁeld should vanish. Therefore,
E
t
z
(ρ = a) = 0, (2.191)
E
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
ρ=a
= 0, (2.192)
a
n
J
n
(k
c
a) + b
n
Y
n
(k
c
a) = 0, (2.193)
E
i
z
(ρ = b) + E
s
z
(ρ = b) = E
t
z
(ρ = b), (2.194)
E
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= E
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] e
jn(φ−φ
0
)
, (2.195)
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b) = a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b). (2.196)
43
2.7.6 Incident, Scattered and Transmitted Magnetic
Fields
Utilizing (2.49) and (2.50), the tangential components of the magnetic ﬁelds are
obtained as
H
i
φ
= E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.197)
H
s
φ
= E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.198)
H
t
φ
= E
0
1
jωµ
k
c
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
. (2.199)
2.7.7 Boundary Conditions for Magnetic Fields
Tangential components of the magnetic ﬁelds are continuous on the outer surface
of the metamaterial coating, due to the boundary conditions. Therefore,
H
i
φ
(ρ = b) + H
s
φ
(ρ = b) = H
t
φ
(ρ = b), (2.200)
E
0
1
jωµ
0
k
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= E
0
1
jωµ
k
c
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] e
jn(φ−φ
0
)
(2.201)
k
0
µ
0
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
=
k
c
µ
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] (2.202)
Using (2.58) in (2.231),
ζJ
n
(k
0
b) + c
n
ζH
(2)
n
(k
0
b) = a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b). (2.203)
44
2.7.8 Simultaneous Solution of the Boundary Conditions
for Electric and Magnetic Fields
Now we have three unknowns and three equations. Rearranging equations
(2.228), (2.222) and (2.232) we get
J
n
(k
c
a)a
n
+Y
n
(k
c
a)b
n
= 0, (2.204)
J
n
(k
c
b)a
n
+ Y
n
(k
c
b)b
n
−H
(2)
n
(k
0
b)c
n
= J
n
(k
0
b), (2.205)
J
n
(k
c
b)a
n
+ Y
n
(k
c
b)b
n
−ζH
(2)
n
(k
0
b)c
n
= ζJ
n
(k
0
b), (2.206)
or in matrix form
J
n
(k
c
a) Y
n
(k
c
a) 0
J
n
(k
c
b) Y
n
(k
c
b) −H
(2)
n
(k
0
b)
J
n
(k
c
b) Y
n
(k
c
b) −ζH
(2)
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
a
n
b
n
c
n
¸
¸
¸
¸
¸
¸
¸
¸
=
0
J
n
(k
0
b)
ζJ
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
. (2.207)
Using Symbolic Math Toolbox of MATLAB, the solution to this system of
equations can be found as:
a
n
=
ζY
n
(k
c
a)
H
(2)
n
(k
0
b)J
n
(k
0
b) −H
(2)
n
(k
0
b)J
n
(k
0
b)
D
, (2.208)
b
n
=
ζJ
n
(k
c
a)
J
n
(k
0
b)H
(2)
n
(k
0
b) −J
n
(k
0
b)H
(2)
n
(k
0
b)
D
, (2.209)
c
n
=
N
D
, (2.210)
where
N = J
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
− ζJ
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (2.211)
D = ζH
(2)
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
− H
(2)
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] . (2.212)
45
Using the following Wronskian will further simplify a
n
and b
n
:
J
n
(x)H
(2)
n
(x) −J
n
(x)H
(2)
n
(x) =
−2j
πx
. (2.213)
2.7.9 Numerical Results
Fig. 2.10 shows some of the numerical results when f = 30GHz, λ
0
= 0.01m,
a = 0.5λ
0
, b = λ
0
. In Fig. 2.10 (a) there are two foci: one of them is inside the
metamaterial and the other one is outside. In Fig. 2.10 (b) there are three foci
inside the metamaterial close to the conducting cylinder. In Fig. 2.10 (c) the
foci are distributed inside the cylinder. Though, one of them is stronger.
46
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
3
(a)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
0.5
1
1.5
2
2.5
3
3.5
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(b)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
3
3.5
4
(c)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(d)
x(m)
y
(
m
)
Magnitude of the Electric Field Inside and Outside the Cylinder
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
−0.02
−0.015
−0.01
−0.005
0
0.005
0.01
0.015
0.02
0.5
1
1.5
2
2.5
(e)
−0.02 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 0.02
0
0.5
1
1.5
2
2.5
3
Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis)
x(m)

E

(
V
/
m
)
(f)
Figure 2.10: Magnitude of the electric ﬁeld inside and outside the cylinder.
(a)(b) ε
r
= −1, µ
r
= −1, (c)(d) ε
r
= −2, µ
r
= −2, (e)(f) ε
r
= 2, µ
r
= 2
47
2.8 Normally Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Coated
Conducting Cylinder: TE
z
Polarization
2.8.1 Introduction
A uniform plane wave is normally incident on a metamaterial coated conducting
cylinder of inﬁnite length. The plane wave travels in the direction which makes an
angle φ
0
with the +x axis. We will examine here the scattering and transmission
by the metamaterial coated conducting cylinder in the case the polarization of
the plane wave is TE
z
. The problem geometry is depicted in Fig. 2.11.
2.8.2 Problem Geometry
x
y
ρ
σ = ∞
φ
φ
0
a
b
Plane Wave
ε
c
, µ
c
ε
0
, µ
0
Figure 2.11: Plane wave normally incident on a metamaterial coated conducting
cylinder.
48
2.8.3 Uniform Plane Wave and Incident Magnetic Field
Let us assume that a TE
z
polarized uniform plane wave is traveling in the direc
tion which makes an angle φ
0
with the +x axis. Magnetic ﬁeld is directed along
the +z axis. Referring to Figure 2.11 the magnetic ﬁeld can be written as
H
i
z
= H
0
e
−jk
0
(xcos φ
0
+y sinφ
0
)
, (2.214)
where
x = ρ cos φ , y = ρ sin φ. (2.215)
Therefore,
H
i
z
= H
0
e
−jk
0
(ρ cos φcos φ
0
+ρ sinφsinφ
0
)
,
= H
0
e
−jk
0
ρ(cos φcos φ
0
+sinφsin φ
0
)
,
= H
0
e
−jk
0
ρ cos(φ−φ
0
)
. (2.216)
Following the same procedure in Section 2.7, incident magnetic ﬁeld can be
written as
H
i
z
= H
0
e
−jk
0
ρ cos(φ−φ
0
)
,
= H
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ)e
jn(φ−φ
0
)
. (2.217)
2.8.4 Scattered and Transmitted Magnetic Fields
Similar to the incident ﬁeld expression in (2.217), we will deﬁne the scattered
and transmitted magnetic ﬁelds in series expansion form respectively as follows:
H
s
z
= H
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.218)
H
t
z
= H
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
. (2.219)
49
2.8.5 Boundary Conditions for Magnetic Fields
The tangential components of the magnetic ﬁelds are continuous on the outer
surface of the metamaterial coating, due to the boundary conditions. Therefore,
H
i
z
(ρ = b) + H
s
z
(ρ = b) = H
t
z
(ρ = b), (2.220)
H
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= H
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] e
jn(φ−φ
0
)
, (2.221)
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b) = a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b). (2.222)
2.8.6 Incident, Scattered and Transmitted Electric Fields
Utilizing (2.112) and (2.50), the tangential components of the electric ﬁelds are
obtained as
E
i
φ
= H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.223)
E
s
φ
= H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ)e
jn(φ−φ
0
)
, (2.224)
E
t
φ
= H
0
−1
jωε
k
c
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
. (2.225)
2.8.7 Boundary Conditions for Electric Fields
The tangential components of the electric ﬁelds are continuous on the outer
surface of the metamaterial coating, due to the boundary conditions. Also, on
the inner surface of the metamaterial coating (i.e., on the conducting cylinder
surface) tangential electric ﬁeld should vanish. Therefore,
E
t
φ
(ρ = a) = 0, (2.226)
50
H
0
−1
jωε
k
c
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ) + b
n
Y
n
(k
c
ρ)] e
jn(φ−φ
0
)
ρ=a
= 0, (2.227)
a
n
J
n
(k
c
a) + b
n
Y
n
(k
c
a) = 0, (2.228)
E
i
φ
(ρ = b) + E
s
φ
(ρ = b) = E
t
φ
(ρ = b), (2.229)
H
0
−1
jωε
0
k
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
e
jn(φ−φ
0
)
= H
0
−1
jωε
k
c
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] e
jn(φ−φ
0
)
(2.230)
k
0
ε
0
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b)
=
k
c
ε
[a
n
J
n
(k
c
b) + b
n
Y
n
(k
c
b)] (2.231)
Using (2.119) in (2.231),
J
n
(k
0
b) + c
n
H
(2)
n
(k
0
b) = a
n
ζJ
n
(k
c
b) + b
n
ζY
n
(k
c
b). (2.232)
2.8.8 Simultaneous Solution of the Boundary Conditions
for Electric and Magnetic Fields
Now we have three unknowns and three equations. Rearranging equations
(2.228), (2.222) and (2.232) we get
J
n
(k
c
a)a
n
+Y
n
(k
c
a)b
n
= 0, (2.233)
J
n
(k
c
b)a
n
+ Y
n
(k
c
b)b
n
−H
(2)
n
(k
0
b)c
n
= J
n
(k
0
b), (2.234)
ζJ
n
(k
c
b)a
n
+ζY
n
(k
c
b)b
n
−H
(2)
n
(k
0
b)c
n
= J
n
(k
0
b), (2.235)
51
or in matrix form
J
n
(k
c
a) Y
n
(k
c
a) 0
J
n
(k
c
b) Y
n
(k
c
b) −H
(2)
n
(k
0
b)
ζJ
n
(k
c
b) ζY
n
(k
c
b) −H
(2)
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
a
n
b
n
c
n
¸
¸
¸
¸
¸
¸
¸
¸
=
0
J
n
(k
0
b)
J
n
(k
0
b)
¸
¸
¸
¸
¸
¸
¸
¸
. (2.236)
Using Symbolic Math Toolbox of MATLAB, the solution to this system of
equations can be found as:
a
n
=
Y
n
(k
c
a)
H
(2)
n
(k
0
b)J
n
(k
0
b) −H
(2)
n
(k
0
b)J
n
(k
0
b)
D
, (2.237)
b
n
=
J
n
(k
c
a)
J
n
(k
0
b)H
(2)
n
(k
0
b) −J
n
(k
0
b)H
(2)
n
(k
0
b)
D
, (2.238)
c
n
=
N
D
, (2.239)
where
N = ζJ
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
−J
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (2.240)
D = H
(2)
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
−ζH
(2)
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] . (2.241)
Using the following Wronskian will further simplify a
n
and b
n
:
J
n
(x)H
(2)
n
(x) −J
n
(x)H
(2)
n
(x) =
−2j
πx
. (2.242)
52
2.9 Obliquely Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Cylinder:
TM
z
Polarization
2.9.1 Introduction
A uniform plane wave is obliquely incident on a metamaterial cylinder of inﬁnite
length. The plane wave travels in the direction which makes an angle φ
0
with
the +x axis and θ
0
with the −z axis. We will examine here the scattering and
transmission by the metamaterial cylinder in the case the polarization of the
plane wave is TM
z
. The problem geometry is depicted in Fig. 2.12.
φ
0
x
y
a
ε
0
, µ
0
ε , µ
z
θ
0
E
k
.
.
Figure 2.12: Uniform plane wave obliquely incident on a metamaterial cylinder:
TM
z
Polarization.
53
2.9.2 Incident, Scattered and Transmitted Electric Fields
(z components)
Referring to Fig. 2.12, the incident electric ﬁeld can be written as
E
i
= (´a
x
E
0
cos θ
0
cos φ
0
+´a
y
E
0
cos θ
0
sin φ
0
+´a
z
E
0
sin θ
0
)
.e
−jk
0
xsin θ
0
cos φ
0
e
−jk
0
y sinθ
0
sin φ
0
e
jk
0
z cos θ
0
, (2.243)
also since x = ρ cos φ and y = ρ sin φ, the z component of the electric ﬁeld can
be expressed as
E
i
z
= E
0
sin θ
0
e
−jk
0
ρ sinθ
0
(cos φcos φ
0
+sinφsinφ
0
)
e
jk
0
z cos θ
0
= E
0
sin θ
0
e
−jk
0
ρ sinθ
0
cos(φ−φ
0
)
e
jk
0
z cos θ
0
. (2.244)
In Section 2.7 Eqn. (2.188), we have previously derived that
E
0
e
−jk
0
ρ cos(φ−φ
0
)
= E
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ)e
jn(φ−φ
0
)
. (2.245)
Utilizing (2.245), (2.244) can be written as
E
i
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
. (2.246)
Since the cylinder is of inﬁnite length, the ﬁelds are periodic in the z direction
and vary according to the factor e
jk
0
z cos θ
0
[29]. The z components of the scattered
and transmitted electric ﬁelds are expressed similar to (2.246) as
E
s
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.247)
E
t
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.248)
The obliquely incident wave travels both in longitudinal and transverse direc
tions, as shown in Fig. 2.13. Due to phase matching, the propagation constant in
54
ε
0
, µ
0
z
θ
0
θ
1
k
0
k
z
= k
0
cosθ
0
k
t0
= k
0
sinθ
0
k
z
= k cosθ
1
k
t
= k sinθ
1
k
ε , µ
Figure 2.13: Longitudinal and transverse components of the incident and trans
mitted ﬁelds.
the longitudinal direction, k
z
, should be the same for free space and metamaterial
media. Therefore,
k
0
cos θ
0
= k cos θ
1
, (2.249)
cos θ
1
=
k
0
k
cos θ
0
, (2.250)
sin θ
1
=
1 −
k
0
k
2
cos
2
θ
0
[
k
0
k
cos θ
0
[ ≤ 1,
−j
k
0
k
2
cos
2
θ
0
−1 [
k
0
k
cos θ
0
[ > 1,
(2.251)
and
k
t
= k sin θ
1
(2.252)
is the transverse propagation constant in metamaterial medium. The arguments
of the Bessel and Hankel functions in (2.246)(2.248) basically include the trans
verse propagation constants.
Remark: Note that, since the metamaterial medium we consider here is not
limited to only DPS metamaterials, (2.252) should not be further simpliﬁed to:
k
t
=
k
2
−k
2
0
cos
2
θ
0
[
k
0
k
cos θ
0
[ ≤ 1,
−j
k
2
0
cos
2
θ
0
−k
2
[
k
0
k
cos θ
0
[ > 1.
(2.253)
55
As an example, consider a DNG medium with k = −k
0
at normal incidence
(θ
0
= π/2). Since the propagation is only in the transverse direction, (2.252)
gives k
t
= −k
0
, which is the correct solution. However, the aforementioned
simpliﬁcation in (2.253) would yield k
t
= k
0
, which is wrong.
2.9.3 Incident, Scattered and Transmitted Magnetic
Fields (z components)
Smooth perfectly conducting inﬁnite cylinders do not depolarize obliquely inci
dent waves. However, for scattering by dielectric or dielectric coated conducting
cylinders, depolarization is inevitable in order to satisfy the Maxwell’s equations
[27, 29]. Therefore, there exist longitudinal magnetic ﬁeld components for the
scattered and transmitted waves:
H
i
z
= 0, (2.254)
H
s
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.255)
H
t
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.256)
2.9.4 φ Components of the Incident, Scattered and Trans
mitted Electric and Magnetic Fields
The φ components of the incident, scattered and transmitted electric and mag
netic ﬁelds are derived from their z components, utilizing Maxwell’s Equations,
in Appendix B. They are found to be:
E
i
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.257)
56
E
s
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+jE
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.258)
E
t
φ
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+jE
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
, (2.259)
H
i
φ
= −j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.260)
H
s
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.261)
H
t
φ
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−j
E
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.262)
2.9.5 Boundary Conditions and Their Solution
Tangential components of the electric and magnetic ﬁelds should be continuous
on the surface of the metamaterial cylinder. Therefore,
E
i
z
(ρ = a) + E
s
z
(ρ = a) = E
t
z
(ρ = a), (2.263)
H
i
z
(ρ = a) + H
s
z
(ρ = a) = H
t
z
(ρ = a), (2.264)
E
i
φ
(ρ = a) + E
s
φ
(ρ = a) = E
t
φ
(ρ = a), (2.265)
H
i
φ
(ρ = a) + H
s
φ
(ρ = a) = H
t
φ
(ρ = a), (2.266)
57
which leads to
J
n
(k
0
a sin θ
0
) + c
n
H
(2)
n
(k
0
a sin θ
0
) = a
n
J
n
(ka sin θ
1
), (2.267)
¯c
n
H
(2)
n
(k
0
a sin θ
0
) = ¯a
n
J
n
(ka sin θ
1
), (2.268)
−
E
0
cos θ
0
k
0
a sin θ
0
nJ
n
(k
0
a sin θ
0
) −
E
0
cos θ
0
k
0
a sin θ
0
nc
n
H
(2)
n
(k
0
a sin θ
0
) + jE
0
η
0
¯c
n
H
(2)
n
(k
0
a sin θ
0
)
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
a sin
2
θ
1
na
n
J
n
(ka sin θ
1
) + jE
0
ζη
0
sin θ
0
sin θ
1
¯a
n
J
n
(ka sin θ
1
), (2.269)
−j
E
0
η
0
J
n
(k
0
a sin θ
0
) −
E
0
cos θ
0
k
0
a sin θ
0
n¯c
n
H
(2)
n
(k
0
a sin θ
0
) −j
E
0
η
0
c
n
H
(2)
n
(k
0
a sin θ
0
)
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
a sin
2
θ
1
n¯a
n
J
n
(ka sin θ
1
) −j
E
0
ζη
0
sin θ
0
sin θ
1
a
n
J
n
(ka sin θ
1
). (2.270)
As we have done in previous sections, the equations are converted into matrix
form and solved. The unknown coeﬃcients are found to be:
a
n
=
1
D
ζa
2
k
2
0
k
4
1
sin
2
θ
0
sin
3
θ
1
(2.271)
.
J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
) −J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
)
.
sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −ζ sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
,
¯a
n
= j
1
D
1
η
0
ζank
0
k
2
1
sin θ
0
sin
2
θ
1
cos θ
0
(2.272)
.
k
2
0
sin
2
θ
0
−k
2
sin
2
θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
.
J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
) −J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
)
,
c
n
=
a
n
J
n
(ka sin θ
1
) −J
n
(k
0
a sin θ
0
)
H
(2)
n
(k
0
a sin θ
0
)
, (2.273)
¯c
n
=
J
n
(ka sin θ
1
)
H
(2)
n
(k
0
a sin θ
0
)
¯a
n
, (2.274)
where
D = −(J
n
(ka sin θ
1
))
2
H
(2)
n
(k
0
a sin θ
0
)
2
cos
2
θ
0
n
2
ζ
k
2
0
sin
2
θ
0
−k
2
sin
2
θ
1
2
+
sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −ζ sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
.
ζ sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
. a
2
k
2
0
k
4
1
sin
2
θ
0
sin
2
θ
1
. (2.275)
58
2.9.6 Calculation of the Radar Cross Section
2D echo width can be found from either
σ = lim
ρ→∞
2πρ
[E
s
[
2
[E
i
[
2
, (2.276)
or
σ = lim
ρ→∞
2πρ
[H
s
[
2
[H
i
[
2
. (2.277)
Let us use the deﬁnition in (2.277). The magnitude of the scattered magnetic
ﬁeld is
[H
s
[ =
[H
s
ρ
[
2
+[H
s
φ
[
2
+[H
s
z
[
2
, (2.278)
and the magnitude of the incident magnetic ﬁeld is
[H
i
[ =
[E
0
[
η
0
. (2.279)
From Appendix B,
H
s
ρ
= −
E
0
sin θ
0
ωµ
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(2.280)
−
k
0
cos θ
0
ωµ
0
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
k
0
cos θ
0
ωµ
0
E
0
k
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
.
Note that, large argument forms of the Hankel functions and their deriva
tives have the spread factor of ρ
−1/2
. Therefore, the ﬁrst and second terms in
(2.280) decay with ρ
−3/2
, whereas the third term decays with ρ
−1/2
and becomes
dominant in the far zone. Therefore, when ρ →∞
H
s
ρ
≈ −
k
0
cos θ
0
ωµ
0
E
0
k
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
,
= −E
0
cos θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n−1
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
. (2.281)
59
Since when ρ →∞
H
(2)
n
(k
0
ρ sin θ
0
) ≈ −
2j
πk
0
ρ sin θ
0
j
n+1
e
−jk
0
ρ sin θ
0
, (2.282)
(2.281) becomes
H
s
ρ
≈ E
0
cos θ
0
e
jk
0
(z cos θ
0
−ρ sinθ
0
)
2j
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
. (2.283)
Therefore,
[H
s
ρ
[
2
≈ [E
0
[
2
cos
2
θ
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
. (2.284)
We will follow similar steps for H
s
φ
:
H
s
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+
E
0
jη
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
. (2.285)
When ρ →∞,
H
s
φ
≈
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n−1
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
,
≈ −
E
0
η
0
e
jk
0
(z cos θ
0
−ρ sinθ
0
)
2j
πk
0
ρ sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
, (2.286)
[H
s
φ
[
2
≈
[E
0
[
2
η
2
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
. (2.287)
H
s
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
. (2.288)
Using the large argument approximation
H
(2)
n
(k
0
ρ sin θ
0
) ≈
2j
πk
0
ρ sin θ
0
j
n
e
−jk
0
ρ sin θ
0
, (2.289)
60
when ρ →∞,
H
s
z
≈ E
0
sin θ
0
e
jk
0
(z cos θ
0
−ρ sinθ
0
)
2j
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
, (2.290)
[H
s
z
[
2
≈ [E
0
[
2
sin
2
θ
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
. (2.291)
[H
s
[
2
= [H
s
ρ
[
2
+[H
s
φ
[
2
+[H
s
z
[
2
,
≈ [E
0
[
2
cos
2
θ
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
+
[E
0
[
2
η
2
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+[E
0
[
2
sin
2
θ
0
2
πk
0
ρ sin θ
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
= [E
0
[
2
2
πk
0
ρ sin θ
0
1
η
2
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
. (2.292)
From (2.277),
σ =
4
k
0
sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+ η
2
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
,
=
2λ
0
π sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+ η
2
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
. (2.293)
The normalized (with respect to λ
0
) echo width is
σ/λ
0
=
2
π
1
sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+ η
2
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
, (2.294)
which at normal incidence special case becomes
σ/λ
0
=
2
π
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
. (2.295)
61
2.10 Obliquely Incident Plane Wave Scattering
by an Inﬁnite Length Metamaterial Cylinder:
TE
z
Polarization
2.10.1 Introduction
The solution for the TE
z
polarization case can be obtained from the TM
z
case
utilizing duality. The ﬁeld expressions and RCS calculations are similar to the
TM
z
case. Hence, in this section only key equations will be given for complete
ness of the problem.
2.10.2 Incident, Scattered and Transmitted Magnetic
Fields (z components)
H
i
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.296)
H
s
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.297)
H
t
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.298)
2.10.3 Incident, Scattered and Transmitted Electric
Fields (z components)
E
i
z
= 0, (2.299)
E
s
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.300)
E
t
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.301)
62
2.10.4 φ Components of the Incident, Scattered and
Transmitted Magnetic and Electric Fields
H
i
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.302)
H
s
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−j
H
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.303)
H
t
φ
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−j
H
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
, (2.304)
E
i
φ
= jH
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.305)
E
s
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+jH
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.306)
E
t
φ
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+jH
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
. (2.307)
2.10.5 Boundary Conditions and Their Solution
Tangential components of the electric and magnetic ﬁelds should be continuous
on the surface of the metamaterial cylinder. Therefore,
H
i
z
(ρ = a) + H
s
z
(ρ = a) = H
t
z
(ρ = a), (2.308)
63
H
i
z
(ρ = a) + H
s
z
(ρ = a) = H
t
z
(ρ = a), (2.309)
H
i
φ
(ρ = a) + H
s
φ
(ρ = a) = H
t
φ
(ρ = a), (2.310)
H
i
φ
(ρ = a) + H
s
φ
(ρ = a) = H
t
φ
(ρ = a), (2.311)
which leads to
J
n
(k
0
a sin θ
0
) + c
n
H
(2)
n
(k
0
a sin θ
0
) = a
n
J
n
(ka sin θ
1
), (2.312)
¯c
n
H
(2)
n
(k
0
a sin θ
0
) = ¯a
n
J
n
(ka sin θ
1
), (2.313)
−
H
0
cos θ
0
k
0
a sin θ
0
nJ
n
(k
0
a sin θ
0
) −
H
0
cos θ
0
k
0
a sin θ
0
nc
n
H
(2)
n
(k
0
a sin θ
0
) −j
H
0
η
0
¯c
n
H
(2)
n
(k
0
a sin θ
0
)
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
a sin
2
θ
1
na
n
J
n
(ka sin θ
1
) −j
H
0
ζη
0
sin θ
0
sin θ
1
¯a
n
J
n
(ka sin θ
1
), (2.314)
jH
0
η
0
J
n
(k
0
a sin θ
0
) −
H
0
cos θ
0
k
0
a sin θ
0
n¯c
n
H
(2)
n
(k
0
a sin θ
0
) + jH
0
η
0
c
n
H
(2)
n
(k
0
a sin θ
0
)
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
a sin
2
θ
1
n¯a
n
J
n
(ka sin θ
1
) + jH
0
ζη
0
sin θ
0
sin θ
1
a
n
J
n
(ka sin θ
1
). (2.315)
As we have done in previous sections, the equations are converted into matrix
form and solved. The unknown coeﬃcients are found to be:
a
n
=
1
D
a
2
k
2
0
k
4
1
sin
2
θ
0
sin
3
θ
1
(2.316)
.
J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
) −J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
)
.
ζ sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
,
¯a
n
= −j
1
D
ζη
0
ank
0
k
2
1
sin θ
0
sin
2
θ
1
cos θ
0
(2.317)
.
k
2
0
sin
2
θ
0
−k
2
sin
2
θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
.
J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
) −J
n
(k
0
a sin θ
0
)H
(2)
n
(k
0
a sin θ
0
)
,
c
n
=
a
n
J
n
(ka sin θ
1
) −J
n
(k
0
a sin θ
0
)
H
(2)
n
(k
0
a sin θ
0
)
, (2.318)
¯c
n
=
J
n
(ka sin θ
1
)
H
(2)
n
(k
0
a sin θ
0
)
¯a
n
, (2.319)
64
where
D = −(J
n
(ka sin θ
1
))
2
H
(2)
n
(k
0
a sin θ
0
)
2
cos
2
θ
0
n
2
ζ
k
2
0
sin
2
θ
0
−k
2
sin
2
θ
1
2
+
sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −ζ sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
.
ζ sin θ
1
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
) −sin θ
0
J
n
(ka sin θ
1
)H
(2)
n
(k
0
a sin θ
0
)
. a
2
k
2
0
k
4
1
sin
2
θ
0
sin
2
θ
1
. (2.320)
2.10.6 Calculation of the Radar Cross Section
Normalized echo width can be found using (2.276) as:
σ/λ
0
=
2
π
1
sin θ
0
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
+
1
η
2
0
+∞
¸
n=−∞
¯c
n
e
jn(φ−φ
0
)
2
, (2.321)
which at normal incidence case becomes
σ/λ
0
=
2
π
+∞
¸
n=−∞
c
n
e
jn(φ−φ
0
)
2
. (2.322)
65
2.11 Obliquely Incident Plane Wave Scatter
ing by an Inﬁnite Length Metamaterial
Coated Conducting Cylinder: TM
z
Polar
ization
2.11.1 Introduction
A uniform plane wave is obliquely incident on a metamaterial coated conducting
cylinder of inﬁnite length. The plane wave illumination and polarization is the
same with Section 2.9. The problem geometry is depicted in Fig. 2.14.
x
y
b
φ
0
ε
0
, µ
0
ε
c
, µ
c
z
θ
0
a
PEC
.
Figure 2.14: Uniform plane wave obliquely incident on a metamaterial coated
conducting cylinder: TM
z
Polarization.
Due to the conducting cylinder centered at the origin, the ﬁelds inside the
metamaterial coating are written not only in terms of Bessel functions of the ﬁrst
kind (i.e., J
n
(.)) but also in terms of Bessel functions of the second kind (i.e.,
66
Y
n
(.)), and their derivatives. This is the only diﬀerence in formulation, from
Section 2.9. Therefore, in this section only key equations are given.
2.11.2 Incident, Scattered and Transmitted Electric
Fields (z components)
E
i
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.323)
E
s
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.324)
E
t
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
.
(2.325)
2.11.3 Incident, Scattered and Transmitted Magnetic
Fields (z components)
H
i
z
= 0, (2.326)
H
s
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.327)
H
t
z
= E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
.
(2.328)
2.11.4 φ Components of the Incident, Scattered and
Transmitted Electric and Magnetic Fields
E
i
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.329)
67
E
s
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+jE
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.330)
E
t
φ
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
ρ sin
2
θ
1
e
jk
0
z cos θ
0
(2.331)
.
+∞
¸
n=−∞
nj
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
+jE
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
.
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
,
H
i
φ
= −j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.332)
H
s
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.333)
H
t
φ
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
ρ sin
2
θ
1
e
jk
0
z cos θ
0
(2.334)
.
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
−j
E
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
.
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
.
2.11.5 Boundary Conditions and Their Solution
Tangential components of the electric and magnetic ﬁelds are continuous on
the outer surface of the metamaterial coating. Also, on the inner surface of
68
the metamaterial coating (i.e., on the conducting cylinder surface) tangential
components of the electric ﬁeld should vanish. Therefore,
E
i
z
(ρ = b) + E
s
z
(ρ = b) = E
t
z
(ρ = b), (2.335)
H
i
z
(ρ = b) + H
s
z
(ρ = b) = H
t
z
(ρ = b), (2.336)
E
i
φ
(ρ = b) + E
s
φ
(ρ = b) = E
t
φ
(ρ = b), (2.337)
H
i
φ
(ρ = b) + H
s
φ
(ρ = b) = H
t
φ
(ρ = b), (2.338)
E
t
z
(ρ = a) = 0, (2.339)
E
t
φ
(ρ = a) = 0, (2.340)
which leads to
J
n
(k
0
b sin θ
0
) + c
n
H
(2)
n
(k
0
b sin θ
0
) = a
n
J
n
(k
c
b sin θ
1
) + b
n
Y
n
(k
c
b sin θ
1
), (2.341)
¯c
n
H
(2)
n
(k
0
b sin θ
0
) = ¯a
n
J
n
(k
c
b sin θ
1
) +
¯
b
n
Y
n
(k
c
b sin θ
1
), (2.342)
−
E
0
cos θ
0
k
0
b sin θ
0
nJ
n
(k
0
b sin θ
0
) −
E
0
cos θ
0
k
0
b sin θ
0
nc
n
H
(2)
n
(k
0
b sin θ
0
) + jE
0
η
0
¯c
n
H
(2)
n
(k
0
b sin θ
0
)
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
na
n
J
n
(k
c
b sin θ
1
) −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
nb
n
Y
n
(k
c
b sin θ
1
)
+jE
0
ζη
0
sin θ
0
sin θ
1
¯a
n
J
n
(k
c
b sin θ
1
) + jE
0
ζη
0
sin θ
0
sin θ
1
¯
b
n
Y
n
(k
c
b sin θ
1
), (2.343)
−j
E
0
η
0
J
n
(k
0
b sin θ
0
) −
E
0
cos θ
0
k
0
b sin θ
0
n¯c
n
H
(2)
n
(k
0
b sin θ
0
) −j
E
0
η
0
c
n
H
(2)
n
(k
0
b sin θ
0
)
= −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
n¯a
n
J
n
(k
c
b sin θ
1
) −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
n
¯
b
n
Y
n
(k
c
b sin θ
1
)
−j
E
0
ζη
0
sin θ
0
sin θ
1
a
n
J
n
(k
c
b sin θ
1
) −j
E
0
ζη
0
sin θ
0
sin θ
1
b
n
Y
n
(k
c
b sin θ
1
), (2.344)
a
n
J
n
(k
c
a sin θ
1
) + b
n
Y
n
(k
c
a sin θ
1
) = 0, (2.345)
−
E
0
k
0
sin θ
0
cos θ
0
k
2
c
a sin
2
θ
1
na
n
J
n
(k
c
a sin θ
1
) −
E
0
k
0
sin θ
0
cos θ
0
k
2
c
a sin
2
θ
1
nb
n
Y
n
(k
c
a sin θ
1
)
+jE
0
ζη
0
sin θ
0
sin θ
1
¯a
n
J
n
(k
c
a sin θ
1
) + jE
0
ζη
0
sin θ
0
sin θ
1
¯
b
n
Y
n
(k
c
a sin θ
1
) = 0. (2.346)
69
As we have done in previous sections, the equations are converted into matrix
form and solved. The unknown coeﬃcients are found to be:
a
n
= −
1
D
Y
n
(k
c
a sin θ
1
)ζk
2
0
k
4
1
b
2
sin
2
θ
0
sin
3
θ
1
(2.347)
.
J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
) −J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
)
.
sin θ
1
H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
−ζ sin θ
0
H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
¯a
n
= j
1
D
1
η
0
ζbnk
0
k
2
1
sin θ
0
sin
2
θ
1
cos θ
0
(2.348)
.
k
2
0
sin
2
θ
0
−k
2
c
sin
2
θ
1
Y
n
(k
c
a sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
) −J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
)]
.
J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
) −J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
)
,
b
n
= −
J
n
(k
c
a sin θ
1
)
Y
n
(k
c
a sin θ
1
)
a
n
, (2.349)
¯
b
n
= −
J
n
(k
c
a sin θ
1
)
Y
n
(k
c
a sin θ
1
)
¯a
n
, (2.350)
c
n
=
a
n
J
n
(k
c
b sin θ
1
) + b
n
Y
n
(k
c
b sin θ
1
) −J
n
(k
0
b sin θ
0
)
H
(2)
n
(k
0
b sin θ
0
)
, (2.351)
¯c
n
=
¯a
n
J
n
(k
c
b sin θ
1
) +
¯
b
n
Y
n
(k
c
b sin θ
1
)
H
(2)
n
(k
0
b sin θ
0
)
, (2.352)
where
D = D
1
+ D
2
, (2.353)
D
1
= ζ cos
2
θ
0
n
2
H
(2)
n
(k
0
b sin θ
0
)
2
k
2
0
sin
2
θ
0
−k
2
c
sin
2
θ
1
2
(2.354)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)] ,
70
D
2
= −b
2
k
2
0
k
4
1
sin
2
θ
0
sin
2
θ
1
(2.355)
.
J
n
(k
c
a sin θ
1
)
sin θ
1
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−ζ sin θ
0
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−Y
n
(k
c
a sin θ
1
)
sin θ
1
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−ζ sin θ
0
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
.
J
n
(k
c
a sin θ
1
)
ζ sin θ
1
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−sin θ
0
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−Y
n
(k
c
a sin θ
1
)
ζ sin θ
1
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−sin θ
0
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
2.11.6 Calculation of the Radar Cross Section
Calculation of the Radar Cross Section is the same as in Section 2.9. The nor
malized echo width is given in (2.294).
71
2.12 Obliquely Incident Plane Wave Scatter
ing by an Inﬁnite Length Metamaterial
Coated Conducting Cylinder: TE
z
Polar
ization
2.12.1 Introduction
A uniform plane wave is obliquely incident on a metamaterial coated conducting
cylinder of inﬁnite length. The plane wave illumination and polarization is the
same with Section 2.10. The problem geometry is as depicted in Fig. 2.14. As
in Section 2.11, due to the conducting cylinder centered at the origin, the ﬁelds
inside the metamaterial coating are written not only in terms of Bessel functions
of the ﬁrst kind (i.e., J
n
(.)) but also in terms of Bessel functions of the second
kind (i.e., Y
n
(.)), and their derivatives. This is the only diﬀerence in formulation,
from Section 2.10. Therefore, in this section only key equations are given.
2.12.2 Incident, Scattered and Transmitted Magnetic
Fields (z components)
H
i
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.356)
H
s
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.357)
H
t
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
.
(2.358)
72
2.12.3 Incident, Scattered and Transmitted Electric
Fields (z components)
E
i
z
= 0, (2.359)
E
s
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.360)
E
t
z
= H
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
.
(2.361)
2.12.4 φ Components of the Incident, Scattered and
Transmitted Magnetic and Electric Fields
H
i
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.362)
H
s
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−j
H
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.363)
H
t
φ
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
ρ sin
2
θ
1
e
jk
0
z cos θ
0
(2.364)
.
+∞
¸
n=−∞
nj
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
−j
H
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
.
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
,
E
i
φ
= jH
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.365)
73
E
s
φ
= −
H
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+jH
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
, (2.366)
E
t
φ
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
ρ sin
2
θ
1
e
jk
0
z cos θ
0
(2.367)
.
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(k
c
ρ sin θ
1
) +
¯
b
n
Y
n
(k
c
ρ sin θ
1
)
e
jn(φ−φ
0
)
+jH
0
ζη
0
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
.
+∞
¸
n=−∞
j
−n
[a
n
J
n
(k
c
ρ sin θ
1
) + b
n
Y
n
(k
c
ρ sin θ
1
)] e
jn(φ−φ
0
)
.
2.12.5 Boundary Conditions and Their Solution
Tangential components of the magnetic and electric ﬁelds are continuous on
the outer surface of the metamaterial coating. Also, on the inner surface of
the metamaterial coating (i.e., on the conducting cylinder surface) tangential
components of the electric ﬁeld should vanish. Therefore,
H
i
z
(ρ = b) + H
s
z
(ρ = b) = H
t
z
(ρ = b), (2.368)
E
i
z
(ρ = b) + E
s
z
(ρ = b) = E
t
z
(ρ = b), (2.369)
H
i
φ
(ρ = b) + H
s
φ
(ρ = b) = H
t
φ
(ρ = b), (2.370)
E
i
φ
(ρ = b) + E
s
φ
(ρ = b) = E
t
φ
(ρ = b), (2.371)
E
t
z
(ρ = a) = 0, (2.372)
E
t
φ
(ρ = a) = 0, (2.373)
which leads to
J
n
(k
0
b sin θ
0
) + c
n
H
(2)
n
(k
0
b sin θ
0
) = a
n
J
n
(k
c
b sin θ
1
) + b
n
Y
n
(k
c
b sin θ
1
), (2.374)
¯c
n
H
(2)
n
(k
0
b sin θ
0
) = ¯a
n
J
n
(k
c
b sin θ
1
) +
¯
b
n
Y
n
(k
c
b sin θ
1
), (2.375)
74
−
H
0
cos θ
0
k
0
b sin θ
0
nJ
n
(k
0
b sin θ
0
) −
H
0
cos θ
0
k
0
b sin θ
0
nc
n
H
(2)
n
(k
0
b sin θ
0
) −j
H
0
η
0
¯c
n
H
(2)
n
(k
0
b sin θ
0
)
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
na
n
J
n
(k
c
b sin θ
1
) −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
nb
n
Y
n
(k
c
b sin θ
1
)
−j
H
0
ζη
0
sin θ
0
sin θ
1
¯a
n
J
n
(k
c
b sin θ
1
) −j
H
0
ζη
0
sin θ
0
sin θ
1
¯
b
n
Y
n
(k
c
b sin θ
1
), (2.376)
jH
0
η
0
J
n
(k
0
b sin θ
0
) −
H
0
cos θ
0
k
0
b sin θ
0
n¯c
n
H
(2)
n
(k
0
b sin θ
0
) + jH
0
η
0
c
n
H
(2)
n
(k
0
b sin θ
0
)
= −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
n¯a
n
J
n
(k
c
b sin θ
1
) −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
b sin
2
θ
1
n
¯
b
n
Y
n
(k
c
b sin θ
1
)
+jH
0
ζη
0
sin θ
0
sin θ
1
a
n
J
n
(k
c
b sin θ
1
) + jH
0
ζη
0
sin θ
0
sin θ
1
b
n
Y
n
(k
c
b sin θ
1
), (2.377)
¯a
n
J
n
(k
c
a sin θ
1
) +
¯
b
n
Y
n
(k
c
a sin θ
1
) = 0, (2.378)
−
H
0
k
0
sin θ
0
cos θ
0
k
2
c
a sin
2
θ
1
n¯a
n
J
n
(k
c
a sin θ
1
) −
H
0
k
0
sin θ
0
cos θ
0
k
2
c
a sin
2
θ
1
n
¯
b
n
Y
n
(k
c
a sin θ
1
)
+jH
0
ζη
0
sin θ
0
sin θ
1
a
n
J
n
(k
c
a sin θ
1
) + jH
0
ζη
0
sin θ
0
sin θ
1
b
n
Y
n
(k
c
a sin θ
1
) = 0. (2.379)
As we have done in previous sections, the equations are converted into matrix
form and solved. The unknown coeﬃcients are found to be:
a
n
=
1
D
Y
n
(k
c
a sin θ
1
)k
2
0
k
4
1
b
2
sin
2
θ
0
sin
3
θ
1
(2.380)
.
J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
) −J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
)
.
ζ sin θ
1
H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
−sin θ
0
H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
¯a
n
= j
1
D
ζη
0
bnk
0
k
2
1
sin θ
0
sin
2
θ
1
cos θ
0
(2.381)
.
k
2
0
sin
2
θ
0
−k
2
c
sin
2
θ
1
Y
n
(k
c
a sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
. [J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
) −J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
)]
.
J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
) −J
n
(k
0
b sin θ
0
)H
(2)
n
(k
0
b sin θ
0
)
,
75
b
n
= −
J
n
(k
c
a sin θ
1
)
Y
n
(k
c
a sin θ
1
)
a
n
, (2.382)
¯
b
n
= −
J
n
(k
c
a sin θ
1
)
Y
n
(k
c
a sin θ
1
)
¯a
n
, (2.383)
c
n
=
a
n
J
n
(k
c
b sin θ
1
) + b
n
Y
n
(k
c
b sin θ
1
) −J
n
(k
0
b sin θ
0
)
H
(2)
n
(k
0
b sin θ
0
)
, (2.384)
¯c
n
=
¯a
n
J
n
(k
c
b sin θ
1
) +
¯
b
n
Y
n
(k
c
b sin θ
1
)
H
(2)
n
(k
0
b sin θ
0
)
, (2.385)
where
D = D
1
+ D
2
, (2.386)
D
1
= −ζ cos
2
θ
0
n
2
H
(2)
n
(k
0
b sin θ
0
)
2
k
2
0
sin
2
θ
0
−k
2
c
sin
2
θ
1
2
(2.387)
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)]
. [J
n
(k
c
b sin θ
1
)Y
n
(k
c
a sin θ
1
) −J
n
(k
c
a sin θ
1
)Y
n
(k
c
b sin θ
1
)] ,
D
2
= b
2
k
2
0
k
4
1
sin
2
θ
0
sin
2
θ
1
(2.388)
.
J
n
(k
c
a sin θ
1
)
sin θ
1
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−ζ sin θ
0
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−Y
n
(k
c
a sin θ
1
)
sin θ
1
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−ζ sin θ
0
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
.
J
n
(k
c
a sin θ
1
)
ζ sin θ
1
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−sin θ
0
Y
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−Y
n
(k
c
a sin θ
1
)
ζ sin θ
1
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
−sin θ
0
J
n
(k
c
b sin θ
1
)H
(2)
n
(k
0
b sin θ
0
)
76
2.12.6 Calculation of the Radar Cross Section
Calculation of the Radar Cross Section is the same as in Section 2.10. The
normalized echo width is given in (2.321).
77
Chapter 3
Achieving Transparency and
Maximizing Scattering with
Metamaterial Coated
Conducting Cylinders
3.1 Introduction
In this chapter, the electromagnetic interaction of plane waves with inﬁnitely long
metamaterial coated conducting cylinders is considered. Diﬀerent from “conju
gate” pairing of doublepositive (DPS) and doublenegative (DNG) or epsilon
negative (ENG) and munegative (MNG) concentric cylinders [5,7–12], achieving
transparency and maximizing scattering are separately achieved by covering per
fect electric conductor (PEC) cylinders with simple (i.e., homogeneous, isotropic
and linear) metamaterial coatings. As in the case of “conjugate” pairing, trans
parency and resonance are found to be heavily dependent on the ratio of core
coating radii, instead of the total size of the cylindrical structure.
78
In our work we show that, for TE polarization, the metamaterial coating
should have 0 < ε
c
< ε
0
as its permittivity to achieve transparency, whereas the
coating permittivity has to be in the −ε
0
< ε
c
< 0 interval for resonance so that
scattering maximization can be achieved. For both transparency and resonance
conditions, we derive the analytical relation between the ratio of corecoating
radii and the permittivity of the metamaterial coating in the TE polarization
case. The numerical results show the validity of these analytical relations, es
pecially when the cylindrical scatterers (i.e., PEC cylinders together with their
metamaterial coatings) are electrically small.
Besides, notice that because the core cylinder is PEC, unlike the aforemen
tioned “conjugate” pairing cases, the analytical relations we have derived for TE
polarization cannot be used for TM polarization by interchanging ε with µ (and
vice versa), unless the core cylinder is replaced with perfect magnetic conductor
(PMC). Yet, both transparency and resonant peaks can be achieved for TM po
larization. Here, we show numerically that for electrically small PEC cylinders
transparency can be obtained by covering them with metamaterial covers having
large [µ
c
[, whereas resonant peaks are observed when µ
c
< 0.
The theory and formulation for TM and TE polarizations have been previ
ously given in Sections 2.7 and 2.8, respectively.
3.2 Transparency Condition
The transparency condition for TE
z
polarization is derived in Appendix C by
setting the numerator of the scattering coeﬃcient c
TE
n
given in (2.240) to zero. In
the subwavelength limit, assuming [k
c
[a < [k
c
[b < 1, k
0
b < 1 and utilizing the
small argument forms of Bessel and Hankel functions, the following transparency
79
condition is obtained:
γ =
2n
ε
0
−ε
c
ε
0
+ ε
c
for n = 0. (3.1)
where γ = a/b is the ratio of coreshell radii, n is the index of series summation.
Alternatively, one can use the transparency condition for an electrically small
cylindrical scatterer, which is composed of two concentric layers of diﬀerent
isotropic materials, given in [5] for the TE
z
polarization as
γ =
2n
(ε
c
−ε
0
)(ε
c
+ ε)
(ε
c
−ε)(ε
c
+ ε
0
)
for n = 0, (3.2)
γ =
µ
c
−µ
0
µ
c
−µ
for n = 0, (3.3)
where (ε, µ) are constitutive parameters of the core cylinder and (ε
c
, µ
c
) are
constitutive parameters of the coating (shell) layer.
When the core cylinder is PEC, ε → −j∞ and µ = µ
0
. In this case (3.3)
becomes
γ =
µ
c
−µ
0
µ
c
−µ
0
= 1 for µ
c
= µ
0
, n = 0, (3.4)
which means there would be no coating. However, (3.2) can still be used in the
limiting case, yielding the same transparency condition in (3.1) as
γ →
2n
(ε
c
−ε
0
)(ε
c
−j∞)
(ε
c
+j∞)(ε
c
+ ε
0
)
=
2n
ε
0
−ε
c
ε
0
+ε
c
for n = 0. (3.5)
The root in (3.1) is of even degree of n (i.e., 2n), which implies that the
argument of the root must be positive. On the other hand, when there is a
coating γ should vary between 0 and 1. Therefore,
0 <
ε
0
−ε
c
ε
0
+ ε
c
< 1, (3.6)
which leads to
0 <
ε
0
−ε
c
ε
0
+ε
c
⇒−ε
0
< ε
c
< ε
0
, (3.7)
80
and
ε
0
−ε
c
ε
0
+ε
c
< 1 ⇒ε
c
< −ε
0
or 0 < ε
c
. (3.8)
From (3.7) and (3.8), the proper choice for ε
c
lies in
0 < ε
c
< ε
0
. (3.9)
As it can be seen from (3.1)(3.9), for the TE
z
case, the transparency condi
tion for the PEC cylinder is independent of the permeability of its metamaterial
coating. As a matter of fact, this is true when the cylindrical scatterer is elec
trically small and the scattering problem is consequently “quasielectrostatic”.
Simply we will choose µ
c
= µ
0
in the numerical experiments for convenience.
For a speciﬁc coating permittivity ε
c
, utilizing (3.1), one can analytically ﬁnd
the corecoating ratio γ at which transparency can be obtained. Similarly, one
can rewrite (3.1) as
ε
c
=
1 −γ
2n
1 + γ
2n
ε
0
, (3.10)
to ﬁnd the coating permittivity for a desired γ, again analytically. In the nu
merical experiments, the following procedure is applied to test the accuracy of
the transparency condition: for a desired γ value, we analytically ﬁnd what the
coating permittivity, ε
c
, should be. Then, using this coating permittivity, we
numerically ﬁnd at which γ value transparency is actually obtained.
In Table 3.1, for certain outer shell radii some γ values are selected where
transparency is desired to be observed. The permittivities of the metamaterial
coating corresponding to these γ values after (3.10) [by setting n = 1 in (3.10)] are
tabulated in Table 3.1. Based on numerical results, transparency is obtained at
diﬀerent γ values (reasonably below desired values), which are also tabulated in
Table 3.1. One way to explain this deviation (i.e., the diﬀerence between desired
and obtained γ values where transparency occurs) is when the subwavelength
limit assumptions are performed, expressions leading to ε
c
given in (3.10) [or
(3.1)] are overly simpliﬁed, particularly in terms of a and b. Interestingly, when
81
the core cylinder is replaced with a coredielectric, ε
c
given in (3.2) yields accurate
results as mentioned in [5] for electrically small cylinders. It is also observed
that as the electrical size of the cylindrical scatterer increases, deviation of the
obtained γ values from the the desired γ values increases. This is an expected
result since the accuracy of (3.10) decreases as the electrical size of the scatterer
increases.
Table 3.1: Desired and Obtained γ for Achieving Transparency Using (3.10)
b = λ
0
/100 b = λ
0
/10 b = λ
0
/5
Desired γ ε
c
/ε
0
Obtained γ Obtained γ Obtained γ
0.2 0.923 0.165 0.15 0.105
0.5 0.6 0.41 0.39 0.31
0.7 0.342 0.595 0.575 0.51
0.9 0.105 0.81 0.805 0.78
Based on Table 3.1 and noticing that the deviation between desired and
obtained γ values usually increases as the value of γ increases, we heuristically
modify (3.10) as
ε
c
=
1 −γ
(2n−γ)
1 + γ
(2n−γ)
ε
0
, (3.11)
to ﬁnd ε
c
for a desired γ value, analytically. In (3.11), the dependence of ε
c
to
a and b is more strongly pronounced. Similar to Table 3.1, desired γ values, the
corresponding ε
c
values and obtained γ values where transparency occurs after
(3.11) [again by setting n = 1 in (3.11)] are tabulated in Table 3.2. As it can be
seen from Table 3.2, our heuristic formula decreases the deviation successfully,
especially when b ≤ λ
0
/10.
The transparency condition for the initial cylindrical structure for the TM
z
polarization can be found from (3.2) and (3.3) utilizing duality:
γ =
2n
(µ
c
−µ
0
)(µ
c
+ µ)
(µ
c
−µ)(µ
c
+µ
0
)
for n = 0, (3.12)
82
Table 3.2: Desired and Obtained γ for Achieving Transparency Using (3.11)
b = λ
0
/100 b = λ
0
/10 b = λ
0
/5
Desired γ ε
c
/ε
0
Obtained γ Obtained γ Obtained γ
0.2 0.895 0.19 0.175 0.125
0.5 0.478 0.49 0.47 0.395
0.7 0.228 0.68 0.67 0.625
0.9 0.0579 0.875 0.875 0.86
γ =
ε
c
−ε
0
ε
c
−ε
for n = 0. (3.13)
After replacing the core cylinder with a PEC one, (3.12)(3.13) become
γ =
2n
(µ
c
−µ
0
)(µ
c
+µ
0
)
(µ
c
−µ
0
)(µ
c
+µ
0
)
= 1 for
µ
c
= µ
0
µ
c
= −µ
0
, n = 0, (3.14)
γ =
ε
c
−ε
0
ε
c
−ε
→
ε
c
−ε
0
ε
c
+ j∞
for n = 0. (3.15)
It can be deduced from (3.14)(3.15) that the transparency condition for the
TM
z
polarization does not lead to any reasonable outcome due to the core being
PEC. It is obvious that in DPSDNG or ENGMNG pairing no such diﬃculty
arises since duality can be simply applied. To be able to achieve transparency
for the TM
z
polarization utilizing similar transparency conditions we have de
rived for TE
z
polarization, the core should be PMC instead of PEC. Theoretical
analysis or simply duality shows that in such a case one can use the dual of
transparency condition for TE
z
polarization by interchanging any permittivity
with the corresponding permeability. Yet, even if the core cylinder is PEC, our
numerical investigations show that polarization can be obtained for electrically
small cylinders with metamaterial coatings having large [µ
c
[. Examples of this
situation are illustrated in Section 3.4 (Numerical Results and Discussion).
83
3.3 Resonance (Scattering Maximization) Con
dition
The resonance condition, which increases the scattering drastically for an elec
trically small cylindrical scatterer, is derived in Appendix D by setting the de
nominator of the scattering coeﬃcient c
TE
n
in (2.241) to zero, again in the sub
wavelength limit. This yields the following resonance condition:
γ =
2n
ε
0
+ ε
c
ε
0
−ε
c
for n = 0. (3.16)
Alternatively, one can use the resonance condition given in [8] for the TE
z
polarization
γ =
2n
(ε
c
+ ε
0
)(ε
c
+ε)
(ε
c
−ε
0
)(ε
c
−ε)
for n > 0. (3.17)
When the core cylinder is PEC, (3.17) becomes
γ →
2n
(ε
c
+ ε
0
)(ε
c
−j∞)
(ε
c
−ε
0
)(ε
c
+ j∞)
=
2n
ε
0
+ ε
c
ε
0
−ε
c
for n > 0. (3.18)
Since the root in (3.16) is of even degree of n (i.e., 2n) and 0<γ<1 should
be, then
0 <
ε
0
+ ε
c
ε
0
−ε
c
< 1, (3.19)
which leads to
0 <
ε
0
+ε
c
ε
0
−ε
c
⇒−ε
0
< ε
c
< ε
0
, (3.20)
and
ε
0
+ ε
c
ε
0
−ε
c
< 1 ⇒ε
c
< 0 or ε
c
> ε
0
. (3.21)
From (3.20) and (3.21), the proper choice for ε
c
lies in
−ε
0
< ε
c
< 0. (3.22)
Then, the ratio of coreshell radii γ, to maximize scattering from a metamaterial
coated PEC cylinder, can be found analytically from the permittivity of the
84
coating ε
c
utilizing (3.16), and vice versa:
ε
c
=
γ
2n
−1
γ
2n
+ 1
ε
0
. (3.23)
In our numerical experiments with scattering maximization, we follow the
same procedure as in the transparency condition (i.e., we ﬁnd the coating per
mittivity for a desired γ value analytically and then use it in the numerical
experiment). Our numerical experiments show that, for electrically small cylin
drical scatterers, (3.23) works quite well (by setting n = 1). Therefore, we do
not modify it as we have modiﬁed the analytical transparency relation.
Interestingly, comparison of (3.10) with (3.23) for a desired γ value shows
that, the permittivity of the coating to maximize scattering should be the neg
ative of the coating permittivity which makes the cylinder transparent. For the
TE
z
case, since the scattering maximization condition is independent of the per
meability of its coating and for electrically small cylindrical scatterers we are
dealing with the “quasielectrostatic” problem, we can safely choose µ
c
= µ
0
.
Therefore, coatings we use here for scattering maximization are ENG metama
terials (or plasmonic materials).
To understand how this resonance condition occurs, consider a PEC cylinder
which is illuminated by a TE
z
polarized plane wave. If the cylinder is electrically
small, the n = 0 term becomes dominant. However, the n = ±1 terms cannot
be neglected since they radiate more eﬃciently [28]. It has been shown in [28]
that the n = 0 term is equivalent to a zdirected magnetic line source, while
the n = ±1 terms, which are referred as dipolar terms in [11], correspond to a
ydirected electric dipole. Due to its electrically small size, this electric dipole
behaves like a capacitive element. If there is also an ENG coating present,
the coating will act like an inductive element. Therefore, the whole cylindrical
scatterer will form an inductorcapacitor (LC) resonator. A similar scenario
is investigated in [17] for electrically small antennas enclosed by metamaterial
85
shells. As the size of the scatterer increases, quadrupolar (i.e., n = 2), octopolar
(i.e., n = 3) and any higher order terms also emerge as resonant terms [11].
The resonance condition for the same cylindrical structure for the TM
z
po
larization, which can be derived from (3.17) utilizing duality, is given in [8] as
γ =
2n
(µ
c
+µ
0
)(µ
c
+ µ)
(µ
c
−µ
0
)(µ
c
−µ)
for n > 0. (3.24)
After replacing the core cylinder with a PEC one, (3.24) becomes
γ =
n
µ
c
+ µ
0
µ
c
−µ
0
for µ
c
= µ
0
, n > 0. (3.25)
Although (3.25) states a resonance relation between a desired γ value and
µ
c
for the TM
z
polarization, our numerical investigations show that µ
c
values
obtained via (3.25) (i.e., from the desired γ values) yield resonance (i.e., maxi
mum scattering) at γ values diﬀerent from the desired ones. On the other hand,
similar to the transparency condition, if PEC core is replaced by a PMC core,
then dual of (3.22) (i.e, −µ
0
< µ
c
< 0) yields a resonance at the desired γ value
for the TM
z
polarization.
Note that, all the formulations used for transparency and scattering maxi
mization conditions are independent of the electrical size of the cylindrical scat
terer (i.e., a and b). However, the formulations are expected to work well for
electrically very small cylinders (i.e., [k
c
[b < 1, k
0
b < 1), such that only a few
modes of the inﬁnite series summation is enough to represent the whole radar
cross section. Although the aforementioned theoretical analysis is based on elec
trically small cylinders and a few modes of the inﬁnite series is assumed to be
dominant, in the computation of the normalized echo widths we use suﬃciently
many modes to be accurate. In other words, our numerical results do not include
any assumption in this sense.
86
3.4 Numerical Results and Discussion
To assess the accuracy of our numerical routines, we have duplicated one of the
numerical results (normalized monostatic echo width of a metamaterial coated
PEC cylinder at 1GHz with PEC radius a = 50mm and coating radius b = 70mm)
in [1], which is shown in Fig. 3.1. In addition to the DPS and DNG coatings
investigated in [1], we also included ENG and MNG coatings. As seen in Fig.
3.1, we have excellent agreement with the results of [1]. Moreover, a perfect
continuation in the monostatic echo width values is observed (as expected) when
the coating medium becomes singlenegative (SNG) from a DPS or DNG coat
ing. In the previous sections, expanding the transparency condition given in
−10 −5 0 5 10
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
ε
c
/ε
0
σ
/
λ
0
TM
TE
TM (Li)
TE (Li)
(a) µ
c
= µ
0
−10 −5 0 5 10
0
1
2
3
4
5
6
ε
c
/ε
0
σ
/
λ
0
TM
TE
TM (Li)
TE (Li)
(b) µ
c
= −µ
0
−10 −5 0 5 10
0
1
2
3
4
5
6
7
µ
c
/µ
0
σ
/
λ
0
TM
TE
TM (Li)
TE (Li)
(c) ε
c
= 2.2ε
0
−10 −5 0 5 10
0
1
2
3
4
5
6
7
µ
c
/µ
0
σ
/
λ
0
TM
TE
TM (Li)
TE (Li)
(d) ε
c
= −2.2ε
0
Figure 3.1: Normalized monostatic echo width of a metamaterial coated PEC
cylinder (a = 50mm, b = 70mm, f = 1GHz). Diamond marks show the DPS
and DNG coating cases in [1].
[5], we have found that it is possible to make PEC cylinders transparent for
87
the TE
z
polarization by covering them with metamaterial covers which exhibit
the material property given by (3.9). By transparency we mean the signiﬁcant
reduction and minimization of scattering in the backscattering direction. As it
has been explained previously, the transparency condition is expected to work
well for electrically very small cylinders. Therefore, we start with an electrically
very small PEC cylinder (in the crosssectional sense) covered with our proposed
metamaterial coating such that the outer radius of the coating is b = λ
0
/100.
Then, for some γ values, where transparency is desired to be observed, the corre
sponding permittivities are analytically found using (3.11) as tabulated in Table
3.2. Finally, the normalized monostatic echo widths are calculated and depicted
in Figs. 3.2(a)3.2(d) for these permittivities. One can see that transparency is
indeed obtained for PEC cylinders almost at the desired γ values. The normal
ized monostatic echo widths for uncoated PEC cylinders (i.e., with radius a) are
shown with dashed lines to visualize the reduction in scattering when proposed
metamaterial coatings are used. Note that for the uncoated case small γ values
mean extremely small PEC cylinders. Naturally, as a goes to zero, no scattering
is supposed to take place. As the next step, we investigate what happens to the
transparency as the electrical size of the scatterer increases. For this purpose, we
gradually increase the outer radius of the cylindrical scatterer. The normalized
monostatic echo widths are calculated and depicted in Figs. 3.2(e)3.2(h), when
the outer radius of the scatterer is increased to b = λ
0
/10. From Figs. 3.2(e)
3.2(h) we see that increasing the electrical size of the cylindrical scatterer from
b = λ
0
/100 to b = λ
0
/10 increases the RCS considerably (e.g., the largest nor
malized monostatic echo width increases roughly from 40dB to 5dB). Despite
this huge increase in RCS, as it can be seen from Figs. 3.2(e)3.2(h) and Table
3.2, transparency can be achieved at the desired γ values. Similarly, we can still
achieve transparency close to desired γ values (as tabulated in Table 3.2) when
the outer radius of the scatterer is increased to b = λ
0
/5.
88
Fig. 3.2 and Table 3.2 show that as the permittivity of the coating is de
creased from ε
c
= ε
0
to ε
c
= 0, the corecoating ratio where transparency occurs
moves from γ = 0 to γ = 1. To explain this phenomenon, we can treat the
metamaterial coating as a cover which cancels out the electromagnetic response
of the PEC core. When the permittivity of the metamaterial coating is close
to ε
0
, this cancellation is quite weak (i.e., metamaterial cover behaves like free
space). In this case, the PEC core should be considerably small with respect to
the coating such that a full cancellation can occur. However, when the permit
tivity of the coating is decreased towards 0, the cancellation of the coating will
become stronger, which means that with even thinner coatings it becomes possi
ble to make larger PEC cores transparent. Note that a similar discussion is made
in [5] to explain the cancellation phenomenon for metamaterial coated dielectric
spheres. For both the dielectric core and the metamaterial cover, their polar
ization vectors are deﬁned, respectively as P = (ε − ε
0
)E and P
c
= (ε
c
− ε
0
)E.
The transparency condition is attributed to the cancellation of these antiparallel
polarization vectors, which happens when ε
c
< ε
0
. In our scenario, since the
core cylinder is PEC, the problem has a less degree of freedom and the analytical
solution shows that to achieve transparency 0 < ε
c
< ε
0
should be.
89
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−130
−120
−110
−100
−90
−80
−70
−60
−50
−40
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(a) ε
c
= 0.895ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−110
−100
−90
−80
−70
−60
−50
−40
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(b) ε
c
= 0.478ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−100
−90
−80
−70
−60
−50
−40
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(c) ε
c
= 0.228ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−100
−90
−80
−70
−60
−50
−40
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(d) ε
c
= 0.0579ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−80
−70
−60
−50
−40
−30
−20
−10
0
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(e) ε
c
= 0.895ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(f) ε
c
= 0.478ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−50
−45
−40
−35
−30
−25
−20
−15
−10
−5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(g) ε
c
= 0.228ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−45
−40
−35
−30
−25
−20
−15
−10
−5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(h) ε
c
= 0.0579ε
0
, µ
c
= µ
0
Figure 3.2: Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TE
z
polarization case, versus the corecoating ratio for coatings
with diﬀerent constitutive parameters. The outer radius of the coating is selected
as (a)(d) b = λ
0
/100, (e)(h) b = λ
0
/10. Dashed line shows the uncoated PEC
case, with radius a.
90
To see the limitations on the electrical size of the cylindrical scatterers for
achieving transparency, we will consider relatively larger scatterers. Since these
scatterers are electrically large, available analytical relations between γ and ε
c
do not hold any longer. Therefore, for these large scatterers we choose ε
c
in
a trial & error process. Figs. 3.3(a)3.3(c) show the results when the outer
radius of the scatterer is increased to b = λ
0
/2. In Figs. 3.3(d)3.3(f) this outer
radius is further increased to b = λ
0
. As it is seen in Fig. 3.3(a) and Fig.
3.3(d), the normalized monostatic echo width makes two dips at some γ. As
the permittivity of the coating is decreased towards 0, the dips move towards
γ = 1, destructively interfering with each other. Finally, the minimum value of
the normalized echo width (σ
TE
/λ
0
drops from 4dB to 25dB) is achieved when
the permittivity is very close to zero but positive, and γ being between 0.9 and
1. Therefore, larger cylinders require coatings having permittivities much closer
to zero. Since monostatic echo width is minimized in the 0.9 < γ < 1 region, the
PEC core can be quite large.
Next, we turn our attention to investigate the validity of scattering maxi
mization condition. Hence, we follow a procedure similar to the one we have
done for the transparency condition. We again start with electrically very small
cylindrical scatterers and gradually increase their outer radii. We use the same
γ in Table 3.1 as our desired γ values, but this time to maximize scattering.
Hence, the coating permittivities are the negatives of coating permittivities tab
ulated in Table 3.1, as a result of (3.23). Figs. 3.4(a)3.4(d) show the normalized
monostatic echo widths for ENG coated PEC cylinders when the outer radius of
the scatterer is b = λ
0
/100. As it can be seen from the ﬁgures, RCS increases
drastically at the desired γ values, making peaks, depending on the permittivity
of the coating. This is mainly due to the resonance of dipolar terms which we
have explained previously. When the outer radius is b = λ
0
/50, the RCS peaks
can still be clearly seen in Figs. 3.4(e)3.4(h). But, this time the peaks are wider
and the peak centers deviate a little from their desired locations. Also note a
91
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−16
−14
−12
−10
−8
−6
−4
−2
0
2
4
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(a) ε
c
= 0.2ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−16
−14
−12
−10
−8
−6
−4
−2
0
2
4
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(b) ε
c
= 0.1ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−25
−20
−15
−10
−5
0
5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(c) ε
c
= 0.01ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−6
−4
−2
0
2
4
6
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(d) ε
c
= 0.03ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−6
−4
−2
0
2
4
6
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(e) ε
c
= 0.02ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−30
−25
−20
−15
−10
−5
0
5
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(f) ε
c
= 0.005ε
0
, µ
c
= µ
0
Figure 3.3: Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TE
z
polarization case, versus the corecoating ratio for coatings
with diﬀerent constitutive parameters. The outer radius of the coating is selected
as (a)(c) b = λ
0
/2, (d)(f) b = λ
0
. Dashed line shows the uncoated PEC case,
with radius a.
92
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(a) ε
c
= −0.923ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(b) ε
c
= −0.6ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(c) ε
c
= −0.342ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(d) ε
c
= −0.105ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(e) ε
c
= −0.923ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(f) ε
c
= −0.6ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(g) ε
c
= −0.342ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−80
−70
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(h) ε
c
= −0.105ε
0
, µ
c
= µ
0
Figure 3.4: Normalized monostatic echo width of an ENG coated PEC cylinder
for the TE
z
polarization case, versus the corecoating ratio for coatings with
diﬀerent constitutive parameters. The outer radius of the coating is selected as
(a)(d) b = λ
0
/100, (e)(h) b = λ
0
/50. Dashed line shows the uncoated PEC
case, with radius a.
93
second small peak which just emerges in Fig. 3.4(e) due to the quadrupolar
terms. These quadrupolar terms become more observable in Figs. 3.5(a)3.5(d)
where b = λ
0
/20. When the outer radius is increased to b = λ
0
/10, eﬀects of
other higher order terms can be observed from Figs. 3.5(e)3.5(h). In summary,
Figs. 3.43.5 suggest that as the electrical size of the scatterer increases the peak
due to the dipolar term becomes wider and moves towards γ = 1. Also, due to
the increased size, quadrupolar and higher order modes emerge. However, the
peak due to the dipolar term is much more dominant and can be safely used to
maximize RCS of objects.
To see whether any transparency or scattering maximization condition can
be obtained for the TM
z
polarization, we consider an electrically very small
cylindrical scatterer with outer radius b = λ
0
/100. For various γ values, we
calculate the monostatic echo widths when µ
c
/µ
0
is in the [−20 20] interval,
as shown in Fig. 3.6. For this “quasimagnetostatic” problem, we have chosen
ε
c
= ε
0
for convenience. Fig. 3.6 shows the existence of resonant modes which
maximize the RCS considerably, when µ
c
< 0. Transparency can be obtained
with coatings having large permeabilities in the absolute sense as seen in Figs.
3.6(a)3.6(c). For γ = 0.9, transparency is possible if µ
c
is positive and very
large.
As we have mentioned previously, the huge increase in the RCS of an ENG
coated PEC cylinder is due to high resonance. However, transparency we have
achieved using DPS coatings is not a result of such resonance, but simple can
cellation. This can be best observed from the changes in RCS with respect to γ,
when Figs. 3.23.3 are plotted in linear scale. In this case, it can be seen that
RCS is not very sensitive to γ near the transparency point. On the contrary, in
Fig. 3.4 we see high γ sensitivity. Since transparency condition is not a result
of resonation, we also expect it not to be very sensitive to ohmic losses. For the
ENG coated cases, however, there would be high sensitivity to ohmic losses near
94
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(a) ε
c
= −0.923ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(b) ε
c
= −0.6ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(c) ε
c
= −0.342ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(d) ε
c
= −0.105ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−45
−40
−35
−30
−25
−20
−15
−10
−5
0
5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(e) ε
c
= −0.923ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(f) ε
c
= −0.6ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−45
−40
−35
−30
−25
−20
−15
−10
−5
0
5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(g) ε
c
= −0.342ε
0
, µ
c
= µ
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−45
−40
−35
−30
−25
−20
−15
−10
−5
0
5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
(h) ε
c
= −0.105ε
0
, µ
c
= µ
0
Figure 3.5: Normalized monostatic echo width of an ENG coated PEC cylinder
for the TE
z
polarization case, versus the corecoating ratio for coatings with
diﬀerent constitutive parameters. The outer radius of the coating is selected as
(a)(d) b = λ
0
/20, (e)(h) b = λ
0
/10. Dashed line shows the uncoated PEC case,
with radius a.
95
−20 −15 −10 −5 0 5 10 15 20
−30
−25
−20
−15
−10
−5
0
5
µ
c
/µ
0
σ
T
M
/
λ
0
(
d
B
)
(a) γ = 0.2
−20 −15 −10 −5 0 5 10 15 20
−25
−20
−15
−10
−5
0
5
µ
c
/µ
0
σ
T
M
/
λ
0
(
d
B
)
(b) γ = 0.5
−20 −15 −10 −5 0 5 10 15 20
−18
−16
−14
−12
−10
−8
−6
−4
−2
0
2
µ
c
/µ
0
σ
T
M
/
λ
0
(
d
B
)
(c) γ = 0.7
−20 −15 −10 −5 0 5 10 15 20
−16
−14
−12
−10
−8
−6
−4
−2
0
2
4
µ
c
/µ
0
σ
T
M
/
λ
0
(
d
B
)
(d) γ = 0.9
Figure 3.6: Normalized monostatic echo width of a metamaterial coated PEC
cylinder for the TM
z
polarization case, versus the coating permeability µ
c
for
diﬀerent corecoating ratios. The outer radius of the coating is b = λ
0
/100 and
the coating permittivity is ε
c
= ε
0
.
the resonant modes. The eﬀects of small ohmic losses, as in the Drude or Lorentz
medium models, are shown in Fig. 3.7. As predicted, there is very little ohmic
sensitivity for transparency condition in Fig. 3.7(a). On the other hand, the high
sensitivity to ohmic losses can be seen clearly at the resonance location in Fig.
3.7(b). Again in Fig. 3.7(b), despite the decrease in the monostatic echo width
due to the ohmic losses, metamaterial coating provides at least approximately
65dB increase in the echo width at the resonance location, when compared with
the uncoated case. In the numerical results we have shown up to here, we have
considered the normalized monostatic echo widths (i.e., back scattering). To vi
sualize the farzone ﬁeld distribution in the xyplane, bistatic echo widths can be
calculated. Fig. 3.8 illustrates the bistatic scattering scenarios for transparency
96
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
x 10
−5
γ = a/b
σ
T
E
/
λ
0
ε
c
= 0.6ε
0
ε
c
= (0.6 − j0.1)ε
0
ε
c
= (0.6 − j0.2)ε
0
(a) ε
c
= 0.6ε
0
, µ
c
= µ
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−45
−40
−35
−30
−25
−20
−15
−10
−5
0
5
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
ε
c
= −0.6ε
0
ε
c
= (−0.6 − j0.01)ε
0
ε
c
= (−0.6 − j0.02)ε
0
(b) ε
c
= −0.6ε
0
, µ
c
= µ
0
Figure 3.7: Eﬀects of ohmic losses on normalized monostatic echo width for (a)
DPS [transparency] (b) ENG [Scattering maximization] cases. The outer radius
of the coating is selected as b = λ
0
/100.
and scattering maximization for the TE polarization considering a metamaterial
coated PEC cylinder with b = λ
0
/100. The angle of incidence is set to φ
0
= 0
◦
.
In Fig. 3.8(a), for the values of ε
c
= 0.6ε
0
, µ
c
= µ
0
and γ = 0.41, it is seen that
RCS increases gradually from backscattering direction (φ = 180
◦
) towards direc
tion of incidence (φ = 0
◦
). Therefore, while little portion of the incident wave is
reﬂected back, the much larger portion will continue traveling in the direction of
incidence. Indeed, this is the expected situation for transparency. In Fig. 3.8(b),
for ε
c
= −0.6ε
0
, µ
c
= µ
0
and γ = 0.505, RCS is maximized in the backscattering
and incidence directions, however it reduces towards φ = 90
◦
, ﬁnally becoming
eﬀectively zero in this direction. In other words, RCS is not only maximized in
the backscattering direction, but also in the direction of incidence. Fig. 3.9(a)
shows the contour plot of the axial component of the total magnetic ﬁeld (i.e.,
H
i
z
+ H
s
z
) in the presence of single PEC cylinder, with radius a = λ
0
/200. In
Fig. 3.9(b), the PEC cylinder is coated with a DPS metamaterial coating having
b = λ
0
/100, ε
c
= 0.6ε
0
and µ
c
= µ
0
. Comparison of Fig. 3.9(a) and Fig. 3.9(b)
shows the decrease in RCS with the proposed metamaterial coating, especially
in the backscattering direction. The case for the resonant ENG coating, for
b = λ
0
/100, ε
c
= −0.6ε
0
and µ
c
= µ
0
, which increases the RCS dramatically, is
97
0 20 40 60 80 100 120 140 160 180
0
1
2
3
4
5
6
7
8
x 10
−7
φ (Degrees)
σ
T
E
/
λ
0
(a) ε
c
= 0.6ε
0
, µ
c
= µ
0
, γ = 0.41
0 20 40 60 80 100 120 140 160 180
0
0.5
1
1.5
2
2.5
φ (Degrees)
σ
T
E
/
λ
0
(b) ε
c
= −0.6ε
0
, µ
c
= µ
0
, γ = 0.505
Figure 3.8: Normalized bistatic echo widths for (a) DPS coated (b) ENG coated
PEC cylinder for the TE
z
polarization case. The outer radius of the coating is
selected as b = λ
0
/100. The angle of incidence is φ
0
= 0
◦
.
shown in Fig. 3.9(c). The ﬁeld distribution conﬁrms the strong resonance in the
radiation of a ydirected electric dipole.
98
x/λ
0
y
/
λ
0
−0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06
−0.08
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
0.08
0.9985
0.999
0.9995
1
1.0005
1.001
(a) No coating
x/λ
0
y
/
λ
0
−0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06
−0.08
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
0.08
0.9984
0.9986
0.9988
0.999
0.9992
0.9994
0.9996
0.9998
1
1.0002
(b) ε
c
= 0.6ε
0
, µ
c
= µ
0
x/λ
0
y
/
λ
0
−0.08 −0.06 −0.04 −0.02 0 0.02 0.04 0.06
−0.08
−0.06
−0.04
−0.02
0
0.02
0.04
0.06
0.08
1
2
3
4
5
6
7
8
9
(c) ε
c
= −0.6ε
0
, µ
c
= µ
0
Figure 3.9: Contour plots of axial component of the total magnetic ﬁeld (i.e.,
H
i
z
+H
s
z
) outside the PEC cylinder when there is (a) No coating, (b) DPS coating,
(c) ENG coating. Outer boundaries of the coatings are shown by dashed lines
(a = λ
0
/200, b = λ
0
/100). Plane wave illumination is along the +xaxis.
99
Fig. 3.10 shows the preliminary results for the oblique incidence case. It
can be observed that, the angle of oblique incidence changes the transparency
condition, however, resonance condition is aﬀected very slightly.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−120
−110
−100
−90
−80
−70
−60
−50
−40
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
θ
0
= 90°
θ
0
= 60°
θ
0
= 30°
(a) ε
c
= 0.478ε
0
, µ
c
= µ
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
−60
−50
−40
−30
−20
−10
0
10
γ = a/b
σ
T
E
/
λ
0
(
d
B
)
θ
0
= 90°
θ
0
= 60°
θ
0
= 30°
(b) ε
c
= −0.6ε
0
, µ
c
= µ
0
Figure 3.10: Normalized monostatic echo widths for (a) DPS coated (b) ENG
coated PEC cylinder for the TE
z
polarization, oblique incidence case. The outer
radius of the coating is selected as b = λ
0
/100.
100
Chapter 4
Retrieval of Homogenization
Parameters
4.1 Homogenization of Metamaterial Struc
tures and Retrieval of Eﬀective Constitu
tive Parameters
4.1.1 Introduction
The physical properties of matter together with the underlying mathematics (e.g.,
Bloch’s Theorem, LyapunovFloquet Theorem) lead to extraordinary phenomena
when structures are aligned periodically. Photonic and electromagnetic band gap
materials, frequency selective surfaces and yet metamaterials are some artiﬁcial
structures which make use of the periodicity. With their unnatural behavior,
these structures are highly exploited for engineering purposes. In engineering,
the use of periodic alignment can also be seen in antennas, cMUTs, optical
101
gratings etc., to increase the overall performance of a system or to establish a
predeﬁned task.
For the analysis and design of periodic materials and structures, to be able
to practically incorporate them in larger systems, their overall equivalents have
to be calculated. As an example, for an array of antennas an array factor can
be deﬁned. The overall response of the antenna array can be calculated by
multiplying this array factor with the response of a single antenna. However,
most of the time the contribution of the interactions between these antennas
cannot be easily neglected. To obtain accurate results, full wave analysis of the
system is necessary. Similarly, for periodic materials an equivalent homogeneous
material can be deﬁned which exhibits the same properties with the material
of interest. The process of obtaining this homogeneous equivalent, with its all
intermediate steps, is called homogenization. For periodic materials, obtaining
the homogeneous equivalent from the basic building block of the material is
obviously simpler and more towards the design of actual material of interest.
Most of the time, the building block of the periodic material is not canonical
and a full wave analysis may be required to obtain the behavior of a building
block itself. Once the response of a single building block is obtained, the overall
structure can be modeled analytically from the results of the building block. The
homogenization processes present in the literature [18–22] are usually examples
of such processes.
However, the interactions between the many building blocks, which form
the periodic structure, must not be simply neglected. Especially for periodic
structures, the periodicity of the structure and therefore the presence of periodic
building blocks are of utmost importance. Hence, for accurate homogenization of
periodic structures, a rigorous method has to be formed to successfully represent
the whole periodic structure. The method we introduce in this work is intended
to accomplish this idea.
102
4.1.2 Homogenization of Metamaterials
In this section, we will focus on the homogenization of metamaterials. However,
the method we present here is not only restricted to metamaterials, therefore
it can be applied to any ﬁnite or semiinﬁnite periodic structure. Although
the method is quite versatile and applicable to oblique incidence scenarios, in
this work we will consider the normal incidence case for a three dimensional
metamaterial structure.
In general, metamaterials are inhomogeneous, anisotropic and highly disper
sive materials. With the homogenization process we obviously remove the inho
mogeneity, however the material maintains its anisotropic and dispersive state.
Therefore the homogenization process for metamaterials is inherently anisotropic
and dispersive. For this reason, the homogenization process can be applied at a
single direction (θ
0
, φ
0
) and at a single frequency, a time.
Metamaterial Geometry:
The building blocks for metamaterials are usually cubic cells, which are also
called unit cells. The unit cell is basically composed of a Split Ring Resonator
(SRR), a wire and a substrate on which the SRR and wire are mounted. The
SRR is formed by two circular or rectangular loops, one within another, with
gaps located at the opposite locations on these loops. A typical unit cell for a
metamaterial is shown in Fig. 4.1.
For metamaterials, the SRRs provide negative eﬀective magnetic permeabil
ity and the wires provide negative eﬀective electric permittivity. However, this
extraordinary behavior can be observed with the proper polarization of electric
and magnetic ﬁelds with respect to the SRR structure, such that the magnetic
ﬁeld should be perpendicular to the SRRs and the electric ﬁeld should reside in
the plane parallel to the SRRs. A typical case is shown in Fig. 4.2.
103
Figure 4.1: Metamaterial unit cell.
Figure 4.2: Direction of E and H ﬁelds for a unit cell.
104
In a right handed medium, for the TEM mode, the directions of E and H
ﬁelds in Fig. 4.2 suggest the direction of propagation to be in the +z direction.
However, in a left handed medium the phase velocity will be in the −z direction,
whereas the energy ﬂow will be again in the +z direction. The direction of
propagation, to avoid ambiguity, should refer to the direction of energy ﬂow (i.e.,
direction of the Poynting vector) for both right handed and left handed media.
Implementation of Boundary Conditions and Excitation:
Consider Fig. 4.3 where a metamaterial of thickness d is placed in air and a
plane wave is normally incident. The metamaterial medium (i.e., Medium 2) is
composed of metamaterial unit cells depicted in Fig. 4.2. Let N
x
, N
y
, and N
z
denote the number of unit cells stratiﬁed in the x, y and z directions, respectively.
The metamaterial medium is assumed to be of inﬁnite extent in the transverse
direction (i.e., N
x
→ ∞, N
y
→ ∞,). Practically this is the case for N
x
N
z
and N
y
N
z
.
Figure 4.3: Direction of E and H ﬁelds for a unit cell.
Let x
max
, x
min
, y
max
and y
min
denote the four surfaces of a unit cell, referring
to Fig. 4.1 and Fig. 4.2. If the four neighboring cells around a unit cell in the
105
transverse direction are considered, for their touching surfaces, x
max
, x
min
, y
max
and y
min
surfaces of the unit cell at the center are identical to x
min
, x
max
, y
min
and y
max
surfaces of the corresponding neighboring cells, respectively. This is
due to the fact that unit cells are indistinguishable in the transverse direction.
Therefore x
max
and x
min
surfaces of a unit cell are identical to each other, as y
max
and y
min
surfaces of the unit cell are identical to each other. Hence, a periodic
boundary condition for the x
max
and x
min
surfaces with zero phase and another
periodic boundary condition for the y
max
and y
min
surfaces with zero phase can
be used to simulate the periodicity in the transverse direction.
Figure 4.4: Alignment of unit cells inside the PECPMC waveguide.
Another, and computationally more eﬃcient, method has been suggested in
[30] for the simulation of SRR+wire metamaterial structures. Unit cells of the
metamaterial structure are placed in a PECPMC waveguide and stratiﬁed in the
z direction as seen in Fig. 4.4. The PMC walls are parallel to the SRR structure
and force the magnetic ﬁeld to be perpendicular to themselves and also to the
SRR structure. The PEC walls are perpendicular to the SRR structure such
106
that the electric ﬁeld becomes parallel to the SRR structure. The electric and
magnetic ﬁelds forced by the PECPMC waveguide are in full accordance with
the plane wave polarization seen in Fig. 4.3.
In our computer simulations, we used High Frequency Structure Simula
tor (HFSS) of Ansoft Inc., which is a Finite Element Method (FEM) based
electromagnetic simulator, and we implemented the aforementioned PECPMC
waveguide method. The problem geometry is depicted in Fig. 4.5.
Figure 4.5: Problem geometry (crosssection view, for N
z
= 3).
The metamaterial unit cell we use is the same with the symmetric unit cell
given in [21] except only that the SRR and wire structures are assumed to have
zero thicknesses and are applied Perfect Electric boundary condition (i.e., treated
as PEC) to reduce the required memory and the computation cost.
The thickness of the metamaterial structure d depends on the number of unit
cells stratiﬁed in the z direction N
z
, such that d = N
z
a, where a = 2.5mm is the
unit cell size.
With the coordinate system given in Fig. 4.5, the x = a/2 and x = −a/2
surfaces of Medium 1, Medium 2 and Medium 3 are applied the PEC boundary
condition, whereas their y = a/2 and y = −a/2 surfaces are applied the PMC
107
boundary condition, in accordance with the PECPMC waveguide method ex
plained previously. The two ends of the geometry (i.e., z = 0 and z = 2L + d
surfaces) are applied the Radiation Boundary Condition, where L = 3cm is the
length of Medium 1 and Medium 3.
The structure is illuminated with a plane wave which originates at z = 0
surface, with its polarization and propagation direction as shown in Fig. 4.5.
The plane wave has magnitude 1 and phase 0 at the z = 0 surface, where it
originates.
Homogeneous Equivalent :
If the metamaterial medium (Medium 2) can be successfully represented with
its homogeneous equivalent, we can deﬁne a Generalized Reﬂection Coeﬃcient
(GRC) at the interface between Medium 1 and Medium 2 as given in [31],
Γ
in
(z = L) = Γ
12
+ T
12
T
21
Γ
23
e
−j2k
z2
d
+ T
12
T
21
Γ
21
Γ
2
23
e
−j4k
z2
d
+ (4.1)
= Γ
12
+
T
12
T
21
Γ
23
e
−j2k
z2
d
1 −Γ
23
Γ
21
e
−j2k
z2
d
,
where k
z2
is the wave number in z direction in Medium 2, Γ
ij
and T
ij
are the
direct reﬂection and transmission coeﬃcients at the interface between layers i and
j, respectively. Note that, (4.1) is valid for the more general oblique incidence
case. For the normal incidence case, k
z2
= k
2
.
Extraction of Electric Field Data:
After the simulation, x component of the electric ﬁeld (i.e., E
x
) is measured
on the line passing through the centers of the unit cells. The begin and end points
of this line are (0, 0, 0) and (0, 0, 2L + d) respectively. A typical magnitude plot
of the x component of the electric ﬁeld is shown in Fig. 4.5. It should be noted
that the electric ﬁeld plotted in Fig. 4.5 is the scattered electric ﬁeld (i.e., [total
electric ﬁeld − incident electric ﬁeld] in Medium 1, 2 and 3). Since we use the
108
PECPMC waveguide method, the y and z components of the electric ﬁeld are
nearly zero.
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6

E
x

z (m)
E
x
 vs. z
Medium 1
Medium 3 Medium 2
GPOF Method is
implemented here
to find Γ
in
(z = 0)
Figure 4.6: [E
x
[ vs. z (f = 10GHz, N
z
= 1).
The scattered ﬁeld in Medium 1 is basically a plane wave traveling in the
−z direction. In Medium 3, the transmitted ﬁeld travels in the +z direction.
However, as seen in Fig. 4.5, there are transition regions near the boundaries
of metamaterial medium (i.e., Medium 2). This is mainly due to discontinuities
inside the PECPMC waveguide and mode conversions. However these nonTEM
modes decay fast. Therefore the scattered and transmitted ﬁelds away from the
metamaterial medium, in Medium 1 and Medium 3 respectively, are TEM waves.
Obtaining the Reﬂection Coeﬃcients:
GPOF Method [32] is one of the many methods used in approximating a
complex function in terms of complex exponentials. Other methods used for this
purpose are various forms of the Prony’s Method such as Least Square Prony’s
Method, Total Least Square Prony’s Method and Singular Value Decomposition
109
Prony’s Method. Another method is the Pencil of Function Method which forms
the basis of the Generalized Pencil of Function Method. In GPOF Method,
basically, a generalized eigenvalue problem is solved and subspace decomposition
is employed. GPOF Method is superior to aforementioned methods in its less
noise sensitivity and computational eﬃciency [33].
The reﬂection coeﬃcient Γ
in
(z = 0) in Medium 1 is found by applying the
Generalized Pencil of Function (GPOF) method to the E
x
ﬁeld component data
in z = [0, 2L/3] interval and ﬁtting it by 1 exponential. The propagation constant
of the scattered wave in Medium 1, obtained via GPOF method, is veriﬁed to be
−k
1
, where k
1
is the free space propagation constant in Medium 1 (propagating in
the +z direction). The z = [0, 2L/3] interval is selected by inspection because, for
all frequencies of interest, the electric ﬁeld data in this interval does not overlap
with the aforementioned transition region. Due to the nonuniform meshing of
the geometry and numerical noise, applying the GPOF method in this interval
is more reliable than simply dividing the scattered ﬁeld at z = 0 to the incident
ﬁeld at the same point (which is 1 + j0). GPOF method basically removes the
numerical noise in the data.
To ﬁnd Γ
in
(z = L) given in (4.1), which is the S
11
of the metamaterial
structure, we use the following relation:
S
11
= Γ
in
(z = L) = Γ
in
(z = 0)e
j2k
1
L
. (4.2)
To ﬁnd the sparameters of the metamaterial structure, we could only use
Medium 2 with two wave ports attached to its input and output surfaces to
set up the excitations. In this setup, we have run simulations with diﬀerent
number of modes for the waveguide (1, 2, . . .). However, S
11
results of these sim
ulations vary noticeably from our setup, maybe because of the aforementioned
fast decaying nonTEM modes are still existent. Our simulation setup seems
more reasonable and it is closer to a real life scenario. However, in the wave
ports setup, the voltages or powers are calculated over the entire surfaces. In
110
our setup, instead of measuring the electric ﬁeld in only one line, which passes
through the midpoints of unit cells, we could take diﬀerent parallel lines to this
line and average the results. As we will present with the results, another line
which passes through the edge of the unit cells gives the same electric ﬁeld dis
tribution in Mediums 1 and 3, while diﬀering from Medium 2 because it does
not go through the dielectric in the unit cell, but air.
Fresnel Reﬂection:
Γ
12
, the ﬁrst term in (4.1), is called the Fresnel reﬂection term. Fresnel re
ﬂection occurs when electromagnetic wave passes from one medium to a diﬀerent
medium. Therefore, Fresnel reﬂection term is obviously expected at the inter
face of two homogeneous and diﬀerent media. The nice thing about the Fresnel
reﬂection is its time causality. In other words, Fresnel reﬂection term is just a
result of discontinuity of the medium in which the wave travels, independent of
whatever the wave will experience in the future. If we consider Fig. 4.7, the
Fresnel reﬂections for (a) and (b) are the same. Therefore, to ﬁnd the Fresnel
reﬂection term at the interface of two media, the medium in which the wave is
transmitted can be taken as semiinﬁnite and all layers beyond this medium can
be neglected. All contributions of these layers will be present in the GRC, as
other terms except the Fresnel reﬂection term.
The Fresnel coeﬃcient can be easily calculated for Fig. 4.7 (b). However the
Fresnel reﬂection term will take two diﬀerent forms for the decoupled TE and
TM modes [31]:
Γ
TE
12
=
µ
2
k
z1
−µ
1
k
z2
µ
2
k
z1
+ µ
1
k
z2
, (4.3)
Γ
TM
12
=
ε
2
k
z1
−ε
1
k
z2
ε
2
k
z1
+ ε
1
k
z2
. (4.4)
For the TEM wave in Fig. 4.7, at normal incidence (k
z1
= k
1
and k
z2
= k
2
),
either (4.3) for E
x
or (4.4) for H
y
can be used. In our work we have checked
both methods and they gave the same result as expected.
111
Figure 4.7: Fresnel reﬂection at (a) three layered media, (b) two layered media.
Expressing S
11
as Summation of Complex Exponentials:
Let us rewrite the GRC equation in (4.1):
S
11
= Γ
12
+ T
12
T
21
Γ
23
e
−j2k
z2
d
+ T
12
T
21
Γ
21
Γ
2
23
e
−j4k
z2
d
+ . (4.5)
As seen in (4.5), S
11
is actually a function of d = N
z
a. In our method we vary
N
z
and record the S
11
s correspondingly. In other words, we have the S
11
vs.
N
z
response of the metamaterial medium and we express it as a summation of
complex exponentials:
S
11
(N
z
) ≈
M
¸
i=1
b
i
e
s
i
N
z
a
, N
z
= N
0
, N
0
+ 1, . . . , N
0
+ N −1 (4.6)
where b
i
’s are the complex residues, s
i
’s are the complex exponents, M is the
number of exponentials to represent the GRC with truncating the inﬁnite series,
N
0
is the initial number of unit cells stratiﬁed in the z direction and N is total
number of unit cells used. Fitting the GRC with M exponentials is done using
the GPOF Method.
112
We have obtained the reﬂection coeﬃcients, S
11
, for metamaterial stacks made
up of N
z
= 1, 2, . . . , 20 unit cells. For each stack, and each frequency from
5GHz to 15GHz (with 200MHz steps) S
11
are calculated as explained previously.
Referring to (4.6):
S
11
(N
0
) ≈ b
1
e
js
1
N
0
a
+b
2
e
js
2
N
0
a
+ + b
M
e
js
M
N
0
a
, (4.7)
S
11
(N
0
+ 1) ≈ b
1
e
js
1
(N
0
+1)a
+ b
2
e
js
2
(N
0
+1)a
+ +b
M
e
js
M
(N
0
+1)a
,
.
.
.
S
11
(N
0
+ N −1) ≈ b
1
e
js
1
(N
0
+N−1)a
+ b
2
e
js
2
(N
0
+N−1)a
+ + b
M
e
js
M
(N
0
+N−1)a
,
where N
0
≥ 1 but not necessarily N
0
= 1 and N
0
+ N − 1 = 20. (4.7) can also
be written as
S
11
(N
0
) ≈ b
1
e
js
1
0a
+ b
2
e
js
2
0a
+ + b
M
e
js
M
0a
, (4.8)
S
11
(N
0
+ 1) ≈ b
1
e
js
1
1a
+ b
2
e
js
2
1a
+ + b
M
e
js
M
1a
,
.
.
.
S
11
(N
0
+N −1) ≈ b
1
e
js
1
(N−1)a
+ b
2
e
js
2
(N−1)a
+ + b
M
e
js
M
(N−1)a
.
When we apply the GPOF method to a vector such as:
[S
11
(N
0
) S
11
(N
0
+ 1) . . . S
11
(N
0
+ N −1)], and ﬁt it with M exponentials, we
actually obtain b
i
and s
i
.a, the complex residues and exponentials in (4.8), re
spectively. This is because GPOF method treats the index of the ﬁrst entry in a
vector as zero, in our case as N
0
= 0. However, our aim is to ﬁnd b
i
and s
i
.a in
(4.7) where N
0
= 0 is the number of unit cells used as bias. Therefore we should
relate these complex residues and exponentials to each other. Comparison of
(4.7) with (4.8) shows that
b
i
= b
i
e
−js
i
N
0
a
, s
i
= s
i
. (4.9)
Now we have found b
i
and s
i
correctly, to approximate S
11
given in (4.5) as
a summation of complex exponentials expressed in (4.6). Comparison of two
113
equations term by term shows that:
b
1
= Γ
12
, s
1
= 0, (4.10)
b
2
= T
12
T
21
Γ
23
, s
2
= −j2k
z2
,
b
3
= T
12
T
21
Γ
21
Γ
2
23
, s
3
= −j4k
z2
,
.
.
.
.
.
.
which means b
1
is the Fresnel reﬂection term, s
1
is the complex exponential
corresponding to the Fresnel reﬂection term and should be zero, s
2
= −j2k
z2
,
s
3
= −j4k
z2
, . . . can be used to calculate k
z2
.
The propagation constant k
z2
, found from each of the complex exponential
terms s
i
, i ≥ 2, should be the same, such that the k
z2
.a product calculated using
each exponent is the same and remains in the reduced Brillouin zone [−π, π]. The
following derivations explain the reduction of the complex propagation constant,
k
z2
, to the reduced Brillouin zone.
Reduction to the Reduced Brillouin Zone:
Consider the second and third complex exponentials of the GPOF approxi
mation in (4.10), assuming M > 2.
e
−j2k
z2
a
= e
s
2
a
e
−j2πm
= e
s
2
a−j2πm
, (4.11)
e
−j4k
z2
a
= e
s
3
a
e
−j2πn
= e
s
3
a−j2πn
, (4.12)
where m, n are integers. Then,
−j2k
z2
a = s
2
a −j2πm, (4.13)
−j4k
z2
a = s
3
a −j2πn. (4.14)
−j2(Re¦k
z2
¦ + jIm¦k
z2
¦)a = (Re¦s
2
¦ + jIm¦s
2
¦)a −j2πm, (4.15)
−j4(Re¦k
z2
¦ + jIm¦k
z2
¦)a = (Re¦s
3
¦ + jIm¦s
3
¦)a −j2πn. (4.16)
114
2Im¦k
z2
¦a −j2Re¦k
z2
¦a = Re¦s
2
¦a + jIm¦s
2
¦a −j2πm, (4.17)
4Im¦k
z2
¦a −j4Re¦k
z2
¦a = Re¦s
3
¦a + jIm¦s
3
¦a −j2πn. (4.18)
Therefore,
−2Re¦k
z2
¦a = Im¦s
2
¦a −2πm → Re¦k
z2
¦a = −
1
2
Im¦s
2
¦a + πm, (4.19)
−4Re¦k
z2
¦a = Im¦s
3
¦a −2πn → Re¦k
z2
¦a = −
1
4
Im¦s
3
¦a +
π
2
n. (4.20)
2Im¦k
z2
¦a = Re¦s
2
¦a → Im¦k
z2
¦a =
1
2
Re¦s
2
¦a, (4.21)
4Im¦k
z2
¦a = Re¦s
3
¦a → Im¦k
z2
¦a =
1
4
Re¦s
3
¦a. (4.22)
The integers m, n in (4.19)(4.20) are selected such that the Re¦k
z2
¦a product
calculated using each exponential s
i
, i = 2, 3 is the same and remains in the
reduced Brillouin zone [−π, π]. The procedure is similar for exponentials s
i
,
i > 3.
Finding the Eﬀective Constitutive Parameters:
After Γ
12
and k
z2
have been found out, for TEM polarization at normal inci
dence, either (4.3) for TE polarized E
x
or (4.4) for TM polarized H
y
can be used
to ﬁnd the eﬀective relative µ
r
or ε
r
of the homogeneous medium respectively. In
our method we used E
x
component of the electric ﬁeld, therefore utilizing (4.3):
µ
r
=
(1 + Γ
12
)k
z2
(1 −Γ
12
)k
z1
, (4.23)
where k
z1
= k
1
= ω
√
µ
0
ε
0
is the wave number in Medium 1 (air).
Since
k
2
= ω
√
µ
2
ε
2
= 2πf
√
µ
r
ε
r
√
µ
0
ε
0
=
2πf
c
√
µ
r
ε
r
, (4.24)
ε
r
=
c
2πf
2
(k
2
t2
+ k
2
z2
)
µ
r
, (4.25)
115
where k
t2
= k
t1
= k
1
sin θ
i
is the wave number in transverse direction in Medium
2, sin θ
i
being the angle of oblique incidence. θ
i
= 0
◦
, k
z2
= k
2
for normal
incidence.
Replacing the Metamaterial with its Homogeneous Equivalent:
Once µ
r
and ε
r
of the homogeneous equivalent for the metamaterial structure
have been obtained, we can replace the metamaterial structure with its homoge
neous equivalent. Since the metamaterial structure is dispersive, its homogeneous
equivalent is also dispersive. Therefore, the obtained µ
r
and ε
r
values are fre
quency dependent and can be better written as µ
r
(ω) and ε
r
(ω). It is worthwhile
to mention that µ
r
(ω) and ε
r
(ω) are complex quantities.
In HFSS, electric permittivity and magnetic permeability of a material can
not be directly assigned complex numbers. However, relative permittivity (ε
r
)
and dielectric loss tangent (tan δ
d
) together with relative permeability (µ
r
) and
magnetic loss tangent (tan δ
m
) can be used alternatively.
ε
r
(ω) = ε
r
(ω) −jε
r
(ω) (4.26)
tan δ
d
(ω) =
ε
r
(ω)
ε
r
(ω)
(4.27)
µ
r
(ω) = µ
r
(ω) −jµ
r
(ω) (4.28)
tan δ
m
(ω) =
µ
r
(ω)
µ
r
(ω)
(4.29)
Both positive and negative real numbers can be assigned to relative permit
tivity, dielectric loss tangent, relative permeability and magnetic loss tangent.
Other properties of the material are left at their defaults:
Bulk Conductivity = 0 S/m, Magnetic Saturation = 0 T
Lande G factor = 2, Delta H = 0 A/m.
Speeding up the process: Although the homogeneous equivalent is much
simpler than the metamaterial structure, due to the size of problem, it takes
116
considerable amount of time to run the geometry in HFSS and to obtain the
electric ﬁeld data and reﬂection coeﬃcient (in the order of several hours). To
solve this problem, we used the slab problem of Section 2.2. Since the PECPMC
waveguide method enforces TEM wave propagation in the normal direction, using
a slab with constitutive parameters µ
r
and ε
r
, the reﬂection coeﬃcient of the
slab as well as electric ﬁelds in Medium 1, 2 and 3 are obtained, in very good
agreement with the HFSS results and in seconds. This allowed us to build an
eﬃcient optimization algorithm.
Optimization Algorithm:
Since metamaterials are highly dispersive, their homogeneous equivalents are
also expected to be highly dispersive. Hence, the eﬀective constitutive parame
ters of the homogeneous equivalent change with frequency, sometimes rapidly. If
a homogeneous slab is considered at a single frequency, the eﬀective constitutive
parameters of the slab play an important role on the number of exponentials to be
used, to approximate the generalized reﬂection coeﬃcient successfully, as in Dis
crete Complex Image Method (DCIM). Therefore, the number of exponentials to
successfully represent the generalized reﬂection coeﬃcient is expected to change
from one frequency to another, based on the eﬀective constitutive parameters of
the homogeneous equivalent at that frequency.
On the other hand, suppose the following vector is used in the GPOF method,
at a single frequency: [S
11
(N
0
) S
11
(N
0
+ 1) . . . S
11
(N
0
+ N −1)]. The variation
of physical length along this vector corresponds to (N−1)a. The electrical length
variation for the same vector should be suﬃciently large, such that the samples
of the vector do not reside very close to each other and hence cause singularity.
Usually, the electrical path length variation along the vector, k
z2
(N −1)a, is
expected to be larger than π for successful approximation with complex expo
nentials.
117
Another aspect of the homogenization problem in metamaterials is the highly
resonant properties of the unit cells which build the metamaterial structure.
The wire and split ring resonators inside the unit cell of a metamaterial cause
electric and magnetic resonances, which are very dominant near their resonant
frequencies. When the unit cells are stacked, the mutual interactions between
these unit cells are very high. Now, consider only one unit cell. If another unit
cell is added, the electromagnetic response of the metamaterial slab will change
abruptly. When a third unit cell is also added, the response is also expected
to change, but less abruptly. As unit cells are added, after some point, the
interaction of the newly added unit cell with the very ﬁrst unit cells (at the
other end of the stack) will be quite weak. Therefore, in the homogeneous state,
the interaction of a newly added unit cell is expected to be dominant only with
the unit cells in its neighborhood. Also, when another unit cell is added, the
interaction of the new unit cell with its neighbors should be at the same amount as
in the case of previously added unit cell. In summary, the metamaterial slab can
be said to be homogeneous when it has suﬃciently large number of unit cells. This
also means that if the metamaterial structure is not acting homogeneous, using
generalized reﬂection coeﬃcients for this structure will contaminate the retrieved
constitutive parameters. Therefore, the generalized reﬂection coeﬃcients used in
the GPOF vector should begin from a suﬃciently large number of unit cells,
which we have deﬁned previously as N
0
: number of unit cells used as bias.
To sum up, the two important parameters in the homogenization process are:
1. Number of unit cells used as bias: N
0
,
2. Number of complex exponentials: M.
Therefore, we have developed an optimization algorithm which ﬁnds the op
timum (N
0
, M) combinations using the S
11
data obtained from stacks made up
of N
z
= 1, 2, . . . , 20 unit cells. In our optimization code we follow these steps:
118
i. Select N
0
(from 1 to 17). This means that the length of the vector used in
the GPOF method is N = 20 −N
0
+ 1.
ii. Select M (from 2 to
N
2
 ).
iii. Apply GPOF method to the vector and ﬁt with M exponentials. Obtain
(b
i
, s
i
).
iv. Obtain (b
i
, s
i
) from (b
i
, s
i
) using (4.9).
v. Sort [s
i
[ in an increasing manner. Reindex s
i
and b
i
vectors in the same
sequence with the sorted [s
i
[.
vi. Obtain the Fresnel term from b
1
. (s
1
should be very close to zero, if ho
mogenization is successful.)
vii. Drop b
1
and s
1
from b
i
and s
i
vectors, respectively.
viii. Sort new [b
i
[ vector in a decreasing manner. Reindex b
i
and s
i
vectors in
the same sequence with the sorted [b
i
[. The entries of the s
i
vector now
correspond to s
2
, s
3
, . . .
ix. Obtain k
z2
from (4.19) and (4.21), selecting m such that the Re¦k
z2
¦a
product remains the in the reduced Brillouin zone [−π, π].
x. From Fresnel reﬂection coeﬃcient Γ
12
and wave number k
z2
, obtain the
eﬀective constitutive parameters µ
r
and ε
r
, using (4.23) and (4.25).
xi. Using (ε
r
, µ
r
) obtain the reﬂection coeﬃcients of the homogeneous equiv
alents (i.e., homogeneous slabs with thicknesses N
0
a, (N
0
+ 1)a, . . . ,
(N
0
+ N −1)a): [S
11
(N
0
) S
11
(N
0
+ 1) . . . S
11
(N
0
+ N −1)].
xii. Obtain the mean square error (MSE) for the last 4 stacks, which is a
heuristic choice for obtaining the error in homogenization:
119
MSE = [S
11
(N
0
+ N −4) −S
11
(N
0
+N −4)[
2
(4.30)
+[S
11
(N
0
+ N −3) −S
11
(N
0
+ N −3)[
2
+[S
11
(N
0
+ N −2) −S
11
(N
0
+ N −2)[
2
+[S
11
(N
0
+ N −1) −S
11
(N
0
+ N −1)[
2
xiii. Find the optimum (N
0
, M) pair in the least mean square (LMS) sense,
following the steps ixii for diﬀerent possible choices of N
0
and M.
After the optimization algorithm, the ﬁnal constitutive parameters of the ho
mogeneous equivalent are obtained. In the frequency band, at some frequencies,
both constitutive parameters are positive (i.e., DPS); at some frequencies, both
constitutive parameters are negative (i.e., DNG); and for the rest of the frequen
cies they have alternative signs (i.e., SNG). For the frequencies at which the
homogeneous equivalent is SNG, it is observed that two terms in the GPOF ap
proximation are dominant: 1) the ﬁrst term which gives the Fresnel term 2) the
second term which gives the propagation constant. The higher order terms of the
GPOF approximation are negligible. The propagation constant has a large and
negative imaginary part, and the wave inside the metamaterial structure decays
rapidly, which is related to the evanescent wave behavior of SNG metamaterials.
If more than 2 exponentials are used at these frequencies, the Fresnel term
more or less stays the same, but the second exponential term is aﬀected. In
terestingly, the optimization algorithm may erroneously select the number of
exponentials M to be more than 2, which may yield a better Fresnel coeﬃcient
for minimizing the MSE.
Here it should be noted that the optimization scheme and the LMS algorithm
are dependent only on the reﬂection data, and they are much more dependent
on Fresnel reﬂection than they are on the propagation constant. This is because
120
the Fresnel coeﬃcient is the most dominant factor in the generalized reﬂection
coeﬃcient, and therefore optimization, since the metamaterial unit cell and total
thickness of the metamaterial are very small.
For the SNG cases, because of the aforementioned reasons, we overrule the
ﬁndings of the optimization algorithm. For SNG cases, the number of bias unit
cells are selected as N
0
= 1, for the rapidly decaying wave to be able to bounce
back, and the number of exponentials is selected as M = 2, because there should
be 2 dominant terms in the GPOF approximation.
4.1.3 Numerical Results
Following the optimization procedure, and correction of the frequencies at which
the metamaterial is SNG, the eﬀective constitutive parameters of the homoge
neous equivalent are obtained as in Fig. 4.8.
After the homogeneous equivalent is obtained, its scattering parameters, S
11
,
are compared with those of the metamaterial structure, in Fig. 4.9. There is a
very good agreement between the scattering parameters over the frequency band.
The exponential approximations are tabulated in Table 4.1 as examples
of three diﬀerent situations: the SNG case at f = 5GHz, the DNG case at
f = 10.8GHz and the DPS case at f = 15GHz.
121
5 6 7 8 9 10 11 12 13 14 15
−35
−30
−25
−20
−15
−10
−5
0
5
Frequency (GHz)
E
f
f
e
c
t
i
v
e
P
e
r
m
i
t
t
i
v
i
t
y
Real part of permittivity
Imaginary part of permittivity
(a)
5 6 7 8 9 10 11 12 13 14 15
−4
−3
−2
−1
0
1
2
3
4
Frequency (GHz)
E
f
f
e
c
t
i
v
e
P
e
r
m
e
a
b
i
l
i
t
y
Real part of permeability
Imaginary part of permeability
(b)
Figure 4.8: Eﬀective homogenization parameters of the metamaterial over the
5GHz  15GHz frequency band, (a) ε
r
, (b) µ
r
.
122
5 6 7 8 9 10 11 12 13 14 15
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Frequency (GHz)

S
1
1

Metamaterial
Homogeneous equivalent
(a)
5 6 7 8 9 10 11 12 13 14 15
−4
−3
−2
−1
0
1
2
3
4
Frequency (GHz)
∠
S
1
1
Metamaterial
Homogeneous equivalent
(b)
Figure 4.9: S
11
vs. frequency, obtained from the metamaterial and its homoge
neous equivalent.
123
Table 4.1: Parameters of the GPOF approximation
f = 5GHz
i b
i
s
i
a Re¦k
2
¦a
1 −0.90 + j0.43 ≈ 0.0 + j0.0
2 −1.10 + j1.87 −3.73 −j1.46 0.7294
f = 10.8GHz
i b
i
s
i
a Re¦k
2
¦a
1 −0.35 −j0.09 ≈ 0.0 + j0.0
2 0.34 + j0.04 −0.11 + j0.87 −0.389
3 3.5 10
−3
−j0.1 −0.33 + j1.69 −0.423
4 2.4 10
−2
−j1.3 10
−3
−0.22 −j1.59 −0.782
5 −1.4 10
−2
+ j1.1 10
−3
−0.10 −j2.63 −0.456
f = 15GHz
i b
i
s
i
a Re¦k
2
¦a
1 −0.27 + j9.8 10
−3
≈ 0.0 + j0.0
2 0.25 −j0.015 −0.041 −j1.35 0.673
3 0.029 −j1.9 10
−4
−0.148 −j2.72 0.679
4 4.3 10
−3
−j5.4 10
−3
−0.206 + j2.27 0.669
5 −1.1 10
−3
+j1.8 10
−3
−0.025 −j2.09 1.047
6 −6.5 10
−4
−j1.7 10
−3
−0.053 + j1.28 0.500
As expected, s
1
≈ 0 + j0 for all of the frequencies given in Table 4.1. For
f = 10.8GHz, the Re¦k
2
¦a product obtained from exponents s
2
and s
3
, by re
ducing them to the reduced Brillouin zone, are close to each other. The products
obtained from the higher order exponents, however, may not yield close results,
since their corresponding coeﬃcients are relatively very small. Similarly, for
f = 15GHz the Re¦k
2
¦a product obtained from exponents s
2
, s
3
and s
4
are very
close to each other, while higher order exponents may give diﬀerent results, be
cause of the explained reason. In summary, comparison of the Re¦k
2
¦a products
obtained from the ﬁrst exponents of the GPOF approximation can be used as a
measure of quality of homogenization.
124
Note that the Re¦k
2
¦a product is in the [−π, 0] reduced Brillouin zone for
the DNG case at f = 10.8GHz, whereas Re¦k
2
¦a product is in the [0, π] reduced
Brillouin zone for the DPS case at f = 15GHz.
To better assess the quality of homogenization, the ﬁeld distribution along
the actual structure (i.e., metamaterial medium together with the air media
surrounding it) is compared with the case where metamaterial medium is replaced
with its homogeneous equivalent. The results for the frequencies of Table 4.1 are
given in Figs. 4.104.11 (when N
z
= 20 i.e., d = 5cm).
0 1 2 3 4 5 6 7 8 9 10 11
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
z (cm)
M
a
g
n
i
t
u
d
e
o
f
E
−
f
i
e
l
d
Metamaterial (Center)
Metamaterial (Edge)
Homogeneous equivalent
Figure 4.10: Magnitude of Eﬁeld inside and outside the metamaterial medium
and its homogeneous equivalent at f = 5GHz.
125
0 1 2 3 4 5 6 7 8 9 10 11
0
0.2
0.4
0.6
0.8
1
1.2
1.4
z (cm)
M
a
g
n
i
t
u
d
e
o
f
E
−
f
i
e
l
d
Metamaterial (Center)
Metamaterial (Edge)
Homogeneous equivalent
(a)
0 1 2 3 4 5 6 7 8 9 10 11
0
0.2
0.4
0.6
0.8
1
1.2
1.4
z (cm)
M
a
g
n
i
t
u
d
e
o
f
E
−
f
i
e
l
d
Metamaterial (Center)
Metamaterial (Edge)
Homogeneous equivalent
(b)
Figure 4.11: Magnitudes of Eﬁeld inside and outside the metamaterial medium
and its homogeneous equivalent at (a) f = 10.8GHz, (b) f = 15.0GHz.
126
The ﬁeld distributions show very good agreement between the metamaterial
and its homogeneous equivalent in Medium 1 and Medium 3. Especially there
is perfect agreement in Medium 1, since our method is based on reﬂection data
and our optimization process strongly forces homogeneous equivalent to mimic
the reﬂection properties of the metamaterial. The agreement in the transmitted
ﬁeld in Medium 3 is a sign of the success in homogenization. The ﬁeld inside
the metamaterial structure, passing through the centers of unit cells are close to
zero and they are seen like noise. This is mainly because they are in the vicinity
of metallic scatterers (i.e., the SRRs and the wire). For this reason, the ﬁeld
distributions are also recorded on another line, which passes from the edges of
the unit cells (through one of the PMC walls). Although the exact mechanism
inside the metamaterial region is not known, based on the unit cell geometry, a
major portion of the medium is air. Hence, if the line is taken from the edge, we
are at the furthest point from the SRR+wire combination, hence their coupling
eﬀects are minimized. Therefore, homogeneous equivalent is expected to resemble
to the ﬁelds sampled at the edge.
4.1.4 Conclusion
In this section, a simple and versatile method for retrieval of the homogenization
parameters of periodic structures is proposed. The method is tested with a
typical 3D metamaterial structure, present in the literature. The homogenization
quality of the metamaterial, compared with its homogeneous equivalent, is tested
in terms of agreement in sparameters, reduction of Re¦k
2
¦a products into the
reduced Brillouin zone and agreement in ﬁeld distributions. Numerical results
show that the method is very successful to retrieve the eﬀective constitutive
parameters of the metamaterial. As the future work, the method can be modiﬁed
to incorporate also the transmission data, so that the homogeneous equivalent
mimics the metamaterial more successfully in the transmission region.
127
4.2 Retrieval of Surface Wave Propagation
Constants on a Grounded Dielectric Slab
4.2.1 Introduction
In this work our aim is to build an eﬃcient and robust method to retrieve the
surface wave propagation constants corresponding to each TM and TE mode
that can propagate on the surface of a grounded dielectric slab, Fig. 4.12. The
twostep method we propose in this work consists of modeling and simulating
the problem geometry in a Finite Element Method (FEM) based electromagnetic
simulator and then processing the electric ﬁeld results obtained from the simula
tor to determine the surface wave propagation constants. The numerical results
determined using our method are compared with their theoretical counterparts.
Numerical results of the method are in good agreement with the theory, generally
achieving less than 2% error. However, there are some geometries and special
cases that the method requires improvement.
Figure 4.12: Geometry of a grounded dielectric slab.
The importance of our proposed method lies in its ability to be further gener
alized and applied to complex geometries. These complex geometries may include
multi layered structures or periodically aligned metamaterial structures to cre
ate an artiﬁcial medium which may have negative eﬀective electric permittivity
and/or negative eﬀective magnetic permeability. Surface waves, leaky waves and
128
evanescent waves related to metamaterials, electronic band gap (EBG) and pho
tonic band gap (PBG) structures have created a ﬂurry of interest among many
researchers [34–77]. To the best of our knowledge, an eﬃcient method to de
termine the propagation constants related to these waves has not been reported
yet. Expansion of our method to include these geometries and structures will
therefore meet an important need.
4.2.2 The TwoStep Method
High Frequency Structure Simulator (HFSS) Simulations
The ﬁrst step of our method is modeling and simulating the problem geometry
in a FEM based electromagnetic simulator. For this purpose we use the High
Frequency Structure Simulator of Ansoft Corporation.
In theory, the ground plane and the dielectric slab is assumed to be of inﬁnite
extent in the x and y directions. In our simulator, due to memory and com
putational restrictions, the inﬁnite geometry of the theory has to be truncated.
However, HFSS provides a very useful tool to take into account the truncated
parts of the geometry. HFSS allows the user to select radiation surfaces and im
pose Radiation Boundary Conditions (RBCs) on these surfaces/boundries, which
in turn allows the waves to radiate inﬁnitely far into space. For the accuracy
of simulations, HFSS recommends the radiation boundary to be located at least
onequarter of a wavelength away from a radiating structure.
To create surface waves on the dielectric slab, we use a rectangular narrow
patch at the surface of the slab and excite 1A constant current along the patch.
The length of the patch is L = 0.295λ
0
and the width of the patch is W = L/10 =
0.0295λ
0
, where λ
0
= 1cm at f = 30GHz. The geometry of the rectangular
narrow patch and constant current excitation is depicted in Fig. 4.13.
129
Figure 4.13: Geometry of the rectangular narrow patch and excitation.
To simulate the inﬁnite ground plane, we use a PEC plate just beneath the
dielectric substrate. We impose the Perfect Electric (PE) boundary condition on
this plate together with the Inﬁnite Ground Plane option being enabled.
As the dielectric substrate, we use a lossless dielectric material with dielectric
constant ε
r
= 2.55. The space above the dielectric slab is ﬁlled with ideal free
space (vacuum). The outer boundaries of the vacuum and the dielectric substrate
(excluding the inﬁnite ground plane) are the radiation surfaces where Radiation
Boundary Conditions are enforced. The outline of the entire problem geometry
in HFSS is as given in Fig. 4.14.
Deﬁnitions: At this point it is useful to make some deﬁnitions, to which we
will refer in the next sections.
• Eplane: xz plane (i.e., y = 0 plane).
• Hplane: yz plane (i.e., x = 0 plane).
• Eline: The line segment (−x width/2 ≤ x ≤ x width/2, y = 0, z = th).
• Hline: The line segment (x = 0, −y width/2 ≤ y ≤ y width/2, z = th).
130
Figure 4.14: Entire problem geometry in HFSS.
As their names imply, Eline and Hline reside in the Eplane and Hplane,
respectively. The Eplane and Hplane are deﬁned with respect to the orientation
of the narrow patch. Note that both Eline and Hline lie on the surface of the
dielectric slab (z = th plane).
Generalized Pencil of Function (GPOF) Method
Preliminaries:
The asymptotic expansion (for large lateral distances, ρ) of the Green’s func
tion of electric ﬁeld for ˆ xdirected ﬁlamentary microstrip dipoles, where the
source is taken to be the origin, is formed as [78]:
G
E
xx
(ρ)∼
Z
0
2π
¸
tan
2
(k
0
d
√
ε
r
−1)
ε
r
−1
sin
2
φ + cos
2
φ
e
−jk
0
ρ
ρ
2
−
Z
0
2k
0
ε
r
−1
ε
r
jRes
W
(β
TM
)
β
2
TM
2
cos
2
φ
H
(2)
2
(β
TM
ρ) −H
(2)
0
(β
TM
ρ)
−β
TM
sin
2
φ
ρ
H
(2)
1
(β
TM
ρ)
,
131
(4.31)
where Z
0
=
µ
0
/ε
0
is the intrinsic impedance of free space, d = th is the
thickness of the dielectric slab, β
TM
is the propagation constant of the single
proper TM pole and Res
W
(β
TM
) is the residue corresponding to the TM pole
(where the function W is as given in [78]). The direction of propagation makes
an angle φ with respect to the positive xaxis measured towards the positive
yaxis.
The ﬁrst term in (4.31) gives the space wave term:
Z
0
2π
¸
tan
2
(k
0
d
√
ε
r
−1)
ε
r
−1
sin
2
φ + cos
2
φ
e
−jk
0
ρ
ρ
2
. (4.32)
The second term in (4.31) shows the contribution from the TM surface wave:
−
Z
0
2k
0
ε
r
−1
ε
r
jRes
W
(β
TM
) (4.33)
β
2
TM
2
cos
2
φ
H
(2)
2
(β
TM
ρ) −H
(2)
0
(β
TM
ρ)
−β
TM
sin
2
φ
ρ
H
(2)
1
(β
TM
ρ)
.
In the Eplane G
E
xx
becomes:
G
E
xx
(ρ, φ = 0) ≈
Z
0
2π
e
−jk
0
ρ
ρ
2
(4.34)
−
Z
0
2k
0
ε
r
−1
ε
r
jRes
W
(β
TM
)
β
2
TM
2
H
(2)
2
(β
TM
ρ) −H
(2)
0
(β
TM
ρ)
.
In the Hplane G
E
xx
becomes:
G
E
xx
(ρ, φ =
π
2
) ≈
Z
0
2π
tan
2
(k
0
d
√
ε
r
−1)
ε
r
−1
e
−jk
0
ρ
ρ
2
(4.35)
+
Z
0
2k
0
ε
r
−1
ε
r
jRes
W
(β
TM
)
β
TM
ρ
H
(2)
1
(β
TM
ρ).
Hence for large ρ, G
E
xx
given in (4.35) and (4.36) have the following charac
teristics in the E and Hplanes, given in Table 4.2.
In our HFSS simulations we use an ˆ xdirected narrow patch with constant
current excitation to simulate the ﬁlamentary microstrip dipole mentioned in
132
Table 4.2: Space Wave and Surface Wave Characteristics in the E and Hplanes.
Space Wave Surface Wave
Plane Decay Prop. Const. Decay Prop. Const.
E ∼ ρ
−2
k
0
∼ ρ
−1/2
β
TM
H ∼ ρ
−2
k
0
∼ ρ
−3/2
β
TM
[78]. In accordance with G
E
xx
, we evaluate the E
x
component of the electric ﬁeld
on the surface of the dielectric substrate (along E and Hlines). The electric ﬁeld
data we have obtained in HFSS is then used in the GPOF method to determine
the surface wave propagation constants.
On the Generalized Pencil of Function Method for Surface Wave
Constant Determination:
Let y be a complex function, and y[0], y[1], . . . , y[N − 1] be the N uniform
samples of a real variable t, as shown in Fig. 4.15. These samples can be
represented by M complex exponentials as
y[k] =
M
¸
i=1
b
i
e
s
i
δtk
=
M
¸
i=1
b
i
z
k
i
, k = 0, 1, . . . , N −1 (4.36)
where z
i
= e
s
i
δt
and δt is the sampling interval. Uniform sampling relates the
samples k to the real variable t such that t = δtk. In (4.36), b
i
’s are called the
residues, s
i
’s are called the exponents.
Now consider the case shown in Fig. 4.16, where y(t) is shifted right by
t
0
= k
0
δt. The shifted function y(t − t
0
) is sampled with N samples, again δt
being the sampling interval, to form y[k −k
0
].
The shifted discrete complex signal sequence y[k − k
0
] can be can be repre
sented by M complex exponentials as in (4.36).
y[k −k
0
] =
M
¸
i=1
b
i
e
s
i
δt(k−k
0
)
=
M
¸
i=1
b
i
z
(k−k
0
)
i
, k −k
0
= 0, 1, . . . , N −1 (4.37)
or equivalently,
y[k−k
0
] =
M
¸
i=1
b
i
e
s
i
δt(k−k
0
)
=
M
¸
i=1
b
i
z
(k−k
0
)
i
, k = k
0
, k
0
+1. . . . , k
0
+N−1 (4.38)
133
Figure 4.15: Magnitudes of complex function y(t) and its N uniform samples
y[k].
Next sections will concentrate on the connection between the preliminaries
and the formal deﬁnition of the GPOF Method.
Application of the GPOF Method on the ELine:
As it is presented in the preliminaries section, for suﬃciently large lateral
distances ρ on the Eline, we expect the space wave term of the E
x
component
of the electric ﬁeld to decay with ρ
−2
whereas the surface wave term is expected
to decay with ρ
−1/2
. Now let us assume that there are M total propagating TM
and TE modes and they all decay with ρ
−1/2
. Therefore the E
x
component of
the electric ﬁeld on the Eline can be written as a function of lateral distance ρ
as follows
E
x
(ρ) =
A
0
ρ
2
e
−jβ
0
ρ
+
M
¸
i=1
A
i
√
ρ
e
−jβ
i
SW
ρ
, (4.39)
where β
0
is the space wave propagation constant, β
i
SW
are the surface wave
propagation constants, A
0
and A
i
are the complex amplitudes of the space and
134
Figure 4.16: Magnitudes of complex function y(t−t
0
) and its N uniform samples
y[k −k
0
].
surface wave terms excluding decay dependence. Multiplying (4.39) with
√
ρ
gives
√
ρE
x
(ρ) =
A
0
ρ
√
ρ
e
−jβ
0
ρ
+
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
. (4.40)
Now let us rewrite (4.40) as
√
ρE
x
(ρ) =
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
+ N
E
(ρ), (4.41)
where
N
E
(ρ) =
A
0
ρ
√
ρ
e
−jβ
0
ρ
, (4.42)
which comes from the space wave term contribution in (4.40).
For large lateral distances ρ, assuming [A
0
[ < [A
i
[, with proper choices of
sampling interval δρ and number of samples N, N
E
(ρ) can be assumed as a noise
term which can be discarded by the GPOF Method. In such a case, (4.41) can
135
be approximated as
√
ρE
x
(ρ) ·
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
, (4.43)
and can be represented with M complex exponentials utilizing the GPOF
Method.
As it has been explained previously, for (4.39) to be valid and for (4.43) to
be a correct approximation, the lateral distance ρ should be suﬃciently large.
Let us assume that the equations and approximations in (4.39)(4.43) are correct
when ρ ≥ ρ
0
, where ρ
0
is the starting value of the lateral distance to be used in
the GPOF Method.
Fig. 4.17 shows the resemblance between
√
ρE
x
(ρ) in (4.43) for ρ ≥ ρ
0
and
y(t −t
0
) of Fig. 4.16. Intuitively this suggests (4.38) to be used in representing
(4.43) with M complex exponentials.
Figure 4.17: Magnitudes of
√
ρE
x
(ρ) and its N uniform samples y[k −k
0
].
136
Therefore,
√
ρE
x
(ρ) ·
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
(4.44)
≡ y[k −k
0
] =
M
¸
i=1
b
i
e
s
i
δρ(k−k
0
)
, k = k
0
, k
0
+ 1, . . . , k
0
+ N −1.
where ρ = δρk. It is obvious that ρ
0
= δρk
0
in Fig. 4.17. The left hand side of
equivalence (4.44) is a continuous signal, whereas the right hand side is a discrete
signal. Expressing the RHS also as a continuous signal, we get
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
=
M
¸
i=1
b
i
e
s
i
(ρ−ρ
0
)
, (4.45)
A
i
e
−jβ
i
SW
ρ
= b
i
e
s
i
(ρ−ρ
0
)
, (4.46)
A
i
e
−jβ
i
SW
ρ
= b
i
e
−s
i
ρ
0
e
s
i
ρ
, (4.47)
which gives
A
i
= b
i
e
−s
i
ρ
0
, β
i
SW
= −Im¦s
i
¦ , Re ¦s
i
¦ = 0, (4.48)
Some Theoretical Examples (Eline):
Example 1: Consider the following case along Eline.
E
x
(ρ) =
A
0
ρ
2
e
−jβ
0
ρ
+
A
1
√
ρ
e
−jβ
1
SW
ρ
(4.49)
√
ρE
x
(ρ) =
A
0
ρ
√
ρ
e
−jβ
0
ρ
+ A
1
e
−jβ
1
SW
ρ
(4.50)
Assume A
0
= 0 , A
1
= −0.7 + j0.3 , β
0
= 6.2832 , β
1
SW
= 1.1β
0
= 6.9115.
When
√
ρE
x
(ρ) is used in the GPOF Method (in the [5λ
0
−8λ
0
] interval with
N = 51 samples) the following results are obtained.
b
1
= 0.7 −j0.3
s
1
= 0 −j6.9115
137
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.7 + j0.3
˜
β
1
SW
= −Im¦s
1
¦ = 6.9115
Since A
0
= 0, the noise term N
E
(ρ) becomes zero. Re¦s
1
¦ is zero as expected.
˜
A
1
and
˜
β
1
SW
are retrieved exactly.
Example 2:
Consider Example 1 again, with the following assumptions
A
0
= (−0.5 −j0.4) 10
−n
, A
1
= −0.7 + j0.3
β
0
= 6.2832 , β
1
SW
= 1.1β
0
= 6.9115.
When
√
ρE
x
(ρ) is used in the GPOF Method (in the [5λ
0
−8λ
0
] interval with
N = 51 samples) the following results are obtained.
n = 0 :
b
1
= 0.6628 −j0.3464
s
1
= 0.0223 −j6.8886
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.5537 + j0.3756
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6189 + j0.4198
˜
β
1
SW
= −Im¦s
1
¦ = 6.8886
n = 1 :
b
1
= 0.6963 −j0.3047
s
1
= 0.0024 −j6.9093
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6847 + j0.3087
138
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6928 + j0.3124
˜
β
1
SW
= −Im¦s
1
¦ = 6.9093
n = 2 :
b
1
= 0.6996 −j0.3005
s
1
= 0.0002 −j6.9113
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6985 + j0.3009
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6993 + j0.3012
˜
β
1
SW
= −Im¦s
1
¦ = 6.9113
n = 3 :
b
1
= 0.7 −j0.3
s
1
= 0 −j6.9115
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6998 + j0.3001
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6999 + j0.3001
˜
β
1
SW
= −Im¦s
1
¦ = 6.9115
n = 4 :
b
1
= 0.7 −j0.3
s
1
= 0 −j6.9115
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.7 + j0.3
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.7 + j0.3
˜
β
1
SW
= −Im¦s
1
¦ = 6.9115
139
In this particular example we see that when [A
0
[ is comparable to [A
1
[, the
contribution of the space wave term shows a noise eﬀect which cannot be easily
discarded. In such a case Re¦s
1
¦ = 0, which makes the theoretical
˜
A
1
= b
1
e
−s
1
ρ
0
complex amplitude retrieval formula wrong. When Re¦s
1
¦ = 0, the modiﬁed
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
formula yields better results for this purpose.
As n increases [A
0
[ <[A
1
[, which leads
˜
A
1
·
˜
A
1
· A
1
and
˜
β
1
SW
· β
1
SW
.
Application of the GPOF Method on the HLine:
The procedure for the Hline is very similar to the Eline case we have inves
tigated in the previous section. For suﬃciently large lateral distances ρ on the
Hline, we again expect the space wave term of the E
x
component of the electric
ﬁeld to decay with ρ
−2
whereas in this case the surface wave terms are expected
to decay with ρ
−3/2
. Let us again assume that there are M total propagating
TM and TE modes and they all decay with ρ
−3/2
. Therefore the E
x
component
of the electric ﬁeld on the Hline can be written as a function of lateral distance
ρ as follows
E
x
(ρ) =
A
0
ρ
2
e
−jβ
0
ρ
+
M
¸
i=1
A
i
ρ
√
ρ
e
−jβ
i
SW
ρ
. (4.51)
Multiplying (4.51) with ρ
√
ρ gives
ρ
√
ρE
x
(ρ) =
A
0
√
ρ
e
−jβ
0
ρ
+
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
. (4.52)
Now let us rewrite (4.52) as
ρ
√
ρE
x
(ρ) =
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
+ N
H
(ρ), (4.53)
where
N
H
(ρ) =
A
0
√
ρ
e
−jβ
0
ρ
, (4.54)
which comes from the space wave term contribution in (4.52).
140
Under the conditions presented in the previous section, N
H
(ρ) can be assumed
as a noise term which can be discarded by the GPOF Method. In such a case,
(4.53) can be approximated as
ρ
√
ρE
x
(ρ) ·
M
¸
i=1
A
i
e
−jβ
i
SW
ρ
, (4.55)
and can be represented with M complex exponentials utilizing the GPOF
Method. Again with the assumption that the equations and approximations
in (4.51)(4.55) are correct when ρ ≥ ρ
0
, one can apply the same procedure in
(4.44)(4.48) and ﬁnd out that
A
i
= b
i
e
−s
i
ρ
0
, β
i
SW
= −Im¦s
i
¦ , Re ¦s
i
¦ = 0. (4.56)
The only diﬀerence for the Hline case appears in (4.44) where
√
ρE
x
(ρ) has
to be replaced with ρ
√
ρE
x
(ρ).
On the other hand, N
E
(ρ) given in (4.42) decays with ρ
−3/2
whereas N
H
(ρ)
in (4.54) decays with ρ
−1/2
, which means application of GPOF Method on the
Eline is less noise sensitive compared to the Hline case.
Some Theoretical Examples (Hline):
Example 3:
Consider the following case along Hline:
E
x
(ρ) =
A
0
ρ
2
e
−jβ
0
ρ
+
A
1
ρ
√
ρ
e
−jβ
1
SW
ρ
, (4.57)
ρ
√
ρE
x
(ρ) =
A
0
√
ρ
e
−jβ
0
ρ
+ A
1
e
−jβ
1
SW
ρ
. (4.58)
Assume A
0
= 0 , A
1
= −0.7 + j0.3 , β
0
= 6.2832 , β
1
SW
= 1.1β
0
= 6.9115.
When ρ
√
ρE
x
(ρ) is used in the GPOF Method (in the [5λ
0
− 8λ
0
] interval
with N = 51 samples) the following results are obtained.
141
b
1
= 0.7 −j0.3
s
1
= 0 −j6.9115
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.7 + j0.3
˜
β
1
SW
= −Im¦s
1
¦ = 6.9115
Since A
0
= 0, the noise term N
H
(ρ) becomes zero. Re¦s
1
¦ is zero as expected.
˜
A
1
and
˜
β
1
SW
are retrieved exactly. Note that the same results with Example 1
are obtained.
Example 4:
Consider Example 3 again, with the following assumptions
A
0
= (−0.5 −j0.4) 10
−n
, A
1
= −0.7 + j0.3
β
0
= 6.2832 , β
1
SW
= 1.1β
0
= 6.9115.
When ρ
√
ρE
x
(ρ) is used in the GPOF Method (in the [5λ
0
− 8λ
0
] interval
with N = 51 samples) the following results are obtained.
n = 0 :
b
1
= 0.4925 −j0.5432
s
1
= 0.1149 −j6.7860
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.0449 + j0.4104
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.0798 + j0.7289
˜
β
1
SW
= −Im¦s
1
¦ = 6.7860
n = 1 :
b
1
= 0.6783 −j0.3249
142
s
1
= 0.0160 −j6.9008
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6091 + j0.3328
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6600 + j0.3606
˜
β
1
SW
= −Im¦s
1
¦ = 6.9008
n = 2 :
b
1
= 0.6978 −j0.3025
s
1
= 0.0017 −j6.9105
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6905 + j0.3036
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6962 + j0.3061
˜
β
1
SW
= −Im¦s
1
¦ = 6.9105
n = 3 :
b
1
= 0.6998 −j0.3002
s
1
= 0.0002 −j6.9114
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6990 + j0.3004
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.6996 + j0.3006
˜
β
1
SW
= −Im¦s
1
¦ = 6.9114
n = 4 :
b
1
= 0.7 −j0.3
s
1
= 0 −j6.9115
˜
A
1
= b
1
e
−s
1
ρ
0
= −0.6999 + j0.3
143
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
= −0.7 + 0.3001
˜
β
1
SW
= −Im¦s
1
¦ = 6.9115
In this particular example we see that when [A
0
[ is comparable to [A
1
[, the
contribution of the space wave term shows a noise eﬀect which cannot be easily
discarded. In such a case Re¦s
1
¦ = 0, which makes the theoretical
˜
A
1
= b
1
e
−s
1
ρ
0
complex amplitude retrieval formula wrong. When Re¦s
1
¦ = 0, the modiﬁed
˜
A
1
= b
1
e
−jIm{s
1
}ρ
0
formula yields better results for this purpose.
As n increases [A
0
[ <[A
1
[, which leads
˜
A
1
·
˜
A
1
· A
1
and
˜
β
1
SW
· β
1
SW
.
Also note that Example 2 for the Eline case works better than the Hline
case examined here, as expected.
4.2.3 Implementation:
To determine the surface wave propagation constants accurately, the lateral dis
tance ρ should be suﬃciently large such that the space wave contribution to the
electric ﬁeld can be neglected compared with the surface wave contributions. On
the other hand, the size of the problem geometry cannot exceed a threshold size
which is predeﬁned by the HFSS as a restriction. We want to investigate the
cases where 0 ≤ ρ ≤ 10λ
0
, which means the size of the dielectric substrate must
be 20λ
0
20λ
0
th. However this size is impossible to implement due to the
aforementioned size restriction of HFSS. To solve this problem, we use diﬀerent
geometries for Eline and Hline cases. For the Eline case, dimensions of the sub
strate are x width y width th, where x width < 20λ
0
and y width = 20λ
0
.
For the Hline case, dimensions of the substrate are x width y width th,
where x width = 20λ
0
and y width < 20λ
0
. The problem geometries for the E
and Hline cases are shown in Fig. 4.18 and Fig. 4.19, respectively.
144
Figure 4.18: Problem geometry for the Eline case.
When the simulation is complete, the electric ﬁeld data along the Eline or H
line are exported to a ﬁle (i.e., ¦Re(E
x
), Im(E
x
), Re(E
y
), Im(E
y
), Re(E
z
), Im(E
z
)¦).
This ﬁle is processed and the necessary E
x
component of the electric ﬁeld is
formed. For the Eline case E
x
is multiplied with
√
ρ, whereas for the Hline case
E
x
is multiplied with ρ
√
ρ.
Theoretically for both Eline and Hline cases we might use the
[ρ
0
≤ ρ ≤ 10λ
0
] interval in the GPOF Method, where ρ
0
can be taken 2λ
0
−3λ
0
intuitively. However on the Eline or Hline ρ = 10λ
0
corresponds to two of the
surfaces where Radiation Boundary Conditions are set. In the simulation there
will be small reﬂections from these surfaces, which will contaminate the results
in their neighborhood. Therefore the logical interval to be used in the GPOF
Method will be [2λ
0
−3λ
0
≤ ρ ≤ 8λ
0
−9λ
0
].
145
Figure 4.19: Problem geometry for the Hline case.
4.2.4 Numerical Results
Surface wave propagation constants retrieved from the E and Hline data are
compared with their analytical counterparts, which are calculated by solving
the transcendental surface wave equations [79]. In the simulations, frequency is
selected as f = 30GHz, therefore λ
0
= 1cm. The slab has a dielectric constant
of ε
r
= 2.55. The retrieval process is repeated for various thicknesses of the
dielectric slab.
In the GPOF Method, HFSS results in the [ρ
start
− ρ
end
] interval with N
number of samples are used. In Table 4.3, for diﬀerent intervals and using diﬀer
ent number of samples, the surface wave propagation constant, β
1
SW
, is retrieved
from the E and Hline data, when the thickness of the slab is th = 0.1λ
0
.
The percentage error of the numerical results for TM
0
mode is calculated as
%Error =
β
1
SW
−β
TM
0
β
TM
0
100, (4.59)
where β
TM
0
= 680.87 = 1.084β
0
is the analytically found surface wave propaga
tion constant for the TM
0
mode when th = 0.1λ
0
.
For some of the cases given in Table 4.3, percentage errors are calculated:
146
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 8λ
0
¦ → %Error = 1.32
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 9λ
0
¦ → %Error = 1.41
¦Hline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 8λ
0
¦ → %Error = 0.02
¦Hline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 9λ
0
¦ → %Error = 0.53
As it is observed from Table 4.3 and percentage errors, both Eline and H
line cases give good results for most of the intervals and number of samples.
Especially Hline case gives better results for this particular thickness of the
dielectric substrate.
For th = 0.15λ
0
, β
TM
0
= 749.32 = 1.193β
0
and the results are tabulated in
Table 4.4. For two of the intervals and number of samples the percentage errors
can be found to be:
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 8λ
0
¦ → %Error = 0.04
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 9λ
0
¦ → %Error = 0.22
Examination of Table 4.4 shows that, Eline case gives very successful results.
But for the Hline case the results are unsatisfactory.
When the thickness of the slab is increased to th = 0.19λ
0
, the theoretical
surface wave propagation constant increases to β
TM
0
= 805.81 = 1.282β
0
. The
retrieved propagation constants are given in Table 4.5. The errors for two cases
can be found as:
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 8λ
0
¦ → %Error = 0.63
¦Eline , N = 51 , ρ
start
= 5λ
0
, ρ
end
= 9λ
0
¦ → %Error = 0.67
As it is seen from Table 4.5, Eline case gives acceptable results. But for the
Hline case the results are again unsatisfactory.
147
For th = 0.25λ
0
, except from the TM
0
mode, another mode emerges, which
is TE
1
. The analytically found surface wave propagation of these modes for
th = 0.25λ
0
are β
TM
0
= 868.94 = 1.383β
0
and β
TE
1
= 669.06 = 1.065β
0
, and the
retrieved ones are tabulated in Table 4.6. For the TE
1
mode, percentage error is
calculated as in (4.59).
¦Eline , N = 51 , ρ
start
= 4λ
0
, ρ
end
= 8λ
0
¦ → %Error(TM
0
) = 0.60
%Error(TE
1
) = 0.13
¦Eline , N = 51 , ρ
start
= 4λ
0
, ρ
end
= 9λ
0
¦ → %Error(TM
0
) = 0.89
%Error(TE
1
) = 0.15
¦Eline , N = 21 , ρ
start
= 4λ
0
, ρ
end
= 9λ
0
¦ → %Error(TM
0
) = 0.69
%Error(TE
1
) = 4.20
From Table 4.6 it is observed that Eline case gives acceptable results for
the TM
0
mode. The results for this mode for the given intervals and number of
samples yield approximately the same results. Along the Eline there are some
problems for the TE
1
mode. For this mode, the best results are obtained in the
[4λ
0
− 8λ
0
] and [4λ
0
− 9λ
0
] intervals with N = 51 samples. For the worst case,
among the given intervals and number of samples, the percentage error is 4.20
for this mode. Along the Hline case the results are again unsatisfactory.
We have also simulated a case, where the dielectric substrate is very thin,
th = 0.05λ
0
. In this example, the surface propagation constant, β
TM
0
= 640.63 =
1.02β
0
, is very close to the space wave propagation constant. The results tabu
lated in Table 4.7 are far away from being successful for both E and Hline cases.
One thing to note is that, the surface wave propagation constants determined
148
using the GPOF Method are even smaller than the space wave propagation con
stant. The possible reason of failure for this particular thin case is propagation
constants of space and surface wave terms being very close to each other.
Finally, the complex coeﬃcients and exponentials found in the GPOF approx
imation are used for generating the electric ﬁeld distribution along the Eline. In
Figs. 4.204.23, the ﬁrst two subplots of the ﬁgures show the ﬁeld distribution
inside the interval used for GPOF approximation. The last two subplots of the
ﬁgures show the extrapolation of electric ﬁeld distribution using the previously
found complex coeﬃcients and exponentials. HFSS data and GPOF approxima
tion are in very good agreement. Also notice that GPOF method removes the
noise in the HFSS data very well.
149
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
670.72 1.067
101 5λ
0
8λ
0
671.84 1.069
51 4λ
0
8λ
0
671.16 1.068
51 5λ
0
8λ
0
671.89 1.069
26 4λ
0
8λ
0
669.33 1.065
26 5λ
0
8λ
0
674.08 1.073
21 4λ
0
8λ
0
668.92 1.065
21 5λ
0
8λ
0
671.92 1.069
101 4λ
0
9λ
0
671.08 1.068
101 5λ
0
9λ
0
670.08 1.066
51 4λ
0
9λ
0
670.58 1.067
51 5λ
0
9λ
0
671.30 1.068
26 4λ
0
9λ
0
669.00 1.065
26 5λ
0
9λ
0
673.39 1.072
21 4λ
0
9λ
0
671.68 1.069
21 5λ
0
9λ
0
673.32 1.072
(a)
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
669.37 1.065
101 5λ
0
8λ
0
681.69 1.085
51 4λ
0
8λ
0
668.81 1.064
51 5λ
0
8λ
0
680.98 1.084
26 4λ
0
8λ
0
666.77 1.061
26 5λ
0
8λ
0
677.92 1.079
21 4λ
0
8λ
0
667.77 1.063
21 5λ
0
8λ
0
678.47 1.080
101 4λ
0
9λ
0
670.70 1.067
101 5λ
0
9λ
0
677.51 1.078
51 4λ
0
9λ
0
671.06 1.068
51 5λ
0
9λ
0
677.25 1.078
26 4λ
0
9λ
0
668.35 1.064
26 5λ
0
9λ
0
674.21 1.073
21 4λ
0
9λ
0
668.66 1.064
21 5λ
0
9λ
0
672.99 1.071
(b)
Table 4.3: Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.1λ
0
, ε
r
= 2.55)
150
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
750.59 1.195
101 5λ
0
8λ
0
749.60 1.193
51 4λ
0
8λ
0
750.92 1.195
51 5λ
0
8λ
0
749.61 1.193
26 4λ
0
8λ
0
751.75 1.196
26 5λ
0
8λ
0
749.92 1.194
21 4λ
0
8λ
0
751.25 1.196
21 5λ
0
8λ
0
750.15 1.194
101 4λ
0
9λ
0
750.46 1.194
101 5λ
0
9λ
0
750.47 1.194
51 4λ
0
9λ
0
750.03 1.194
51 5λ
0
9λ
0
750.97 1.195
26 4λ
0
9λ
0
750.94 1.195
26 5λ
0
9λ
0
750.15 1.194
21 4λ
0
9λ
0
752.29 1.197
21 5λ
0
9λ
0
750.63 1.195
(a)
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
678.15 1.079
101 5λ
0
8λ
0
551.77 0.878
51 4λ
0
8λ
0
692.83 1.103
51 5λ
0
8λ
0
552.05 0.879
26 4λ
0
8λ
0
714.73 1.138
26 5λ
0
8λ
0
554.49 0.882
21 4λ
0
8λ
0
703.84 1.120
21 5λ
0
8λ
0
547.86 0.872
101 4λ
0
9λ
0
505.62 0.805
101 5λ
0
9λ
0
494.96 0.788
51 4λ
0
9λ
0
515.59 0.821
51 5λ
0
9λ
0
495.13 0.788
26 4λ
0
9λ
0
528.81 0.842
26 5λ
0
9λ
0
495.33 0.788
21 4λ
0
9λ
0
541.64 0.862
21 5λ
0
9λ
0
494.29 0.787
(b)
Table 4.4: Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.15λ
0
, ε
r
= 2.55)
151
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
812.92 1.294
101 5λ
0
8λ
0
814.56 1.296
51 4λ
0
8λ
0
812.40 1.293
51 5λ
0
8λ
0
815.77 1.298
26 4λ
0
8λ
0
810.91 1.291
26 5λ
0
8λ
0
815.51 1.298
21 4λ
0
8λ
0
813.97 1.295
21 5λ
0
8λ
0
815.93 1.299
101 4λ
0
9λ
0
811.20 1.291
101 5λ
0
9λ
0
809.44 1.288
51 4λ
0
9λ
0
811.55 1.292
51 5λ
0
9λ
0
811.24 1.291
26 4λ
0
9λ
0
811.14 1.291
26 5λ
0
9λ
0
810.97 1.291
21 4λ
0
9λ
0
811.81 1.292
21 5λ
0
9λ
0
808.01 1.286
(a)
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
573.13 0.912
101 5λ
0
8λ
0
602.71 0.959
51 4λ
0
8λ
0
573.08 0.912
51 5λ
0
8λ
0
602.31 0.959
26 4λ
0
8λ
0
574.13 0.914
26 5λ
0
8λ
0
600.82 0.956
21 4λ
0
8λ
0
573.48 0.913
21 5λ
0
8λ
0
603.75 0.961
101 4λ
0
9λ
0
604.62 0.962
101 5λ
0
9λ
0
627.22 0.998
51 4λ
0
9λ
0
604.91 0.963
51 5λ
0
9λ
0
625.78 0.996
26 4λ
0
9λ
0
602.64 0.959
26 5λ
0
9λ
0
624.15 0.993
21 4λ
0
9λ
0
603.83 0.961
21 5λ
0
9λ
0
625.02 0.995
(b)
Table 4.5: Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.19λ
0
, ε
r
= 2.55)
152
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
β
2
SW
β
2
SW
/β
0
101 4λ
0
8λ
0
874.37 1.392 661.72 1.053
101 5λ
0
8λ
0
874.33 1.392 642.69 1.023
51 4λ
0
8λ
0
874.17 1.391 668.20 1.063
51 5λ
0
8λ
0
874.59 1.392 641.47 1.021
26 4λ
0
8λ
0
872.81 1.389 660.68 1.052
26 5λ
0
8λ
0
877.74 1.397 651.17 1.036
21 4λ
0
8λ
0
875.05 1.393 664.86 1.058
21 5λ
0
8λ
0
878.32 1.398 592.63 0.943
101 4λ
0
9λ
0
876.05 1.394 666.87 1.061
101 5λ
0
9λ
0
878.67 1.398 659.38 1.049
51 4λ
0
9λ
0
876.68 1.395 668.06 1.063
51 5λ
0
9λ
0
878.97 1.399 656.69 1.045
26 4λ
0
9λ
0
876.14 1.394 664.88 1.058
26 5λ
0
9λ
0
881.39 1.403 649.90 1.034
21 4λ
0
9λ
0
874.90 1.392 640.98 1.020
21 5λ
0
9λ
0
877.24 1.396 662.16 1.054
(a)
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
β
2
SW
β
2
SW
/β
0
101 4λ
0
8λ
0
669.71 1.066 531.31 0.846
101 5λ
0
8λ
0
677.75 1.079 450.48 0.717
51 4λ
0
8λ
0
666.46 1.061 508.42 0.809
51 5λ
0
8λ
0
679.28 1.081 501.52 0.798
26 4λ
0
8λ
0
668.97 1.065 528.33 0.841
26 5λ
0
8λ
0
680.98 1.084 587.81 0.936
21 4λ
0
8λ
0
663.85 1.057 516.06 0.821
21 5λ
0
8λ
0
673.33 1.072 422.74 0.673
101 4λ
0
9λ
0
638.59 1.016 652.13 1.038
101 5λ
0
9λ
0
693.72 1.104 683.90 1.088
51 4λ
0
9λ
0
632.50 1.007 619.51 0.986
51 5λ
0
9λ
0
695.36 1.107 677.21 1.078
26 4λ
0
9λ
0
631.59 1.005 628.70 1.001
26 5λ
0
9λ
0
691.11 1.100 656.84 1.045
21 4λ
0
9λ
0
630.60 1.004 652.51 1.038
21 5λ
0
9λ
0
686.37 1.092 678.40 1.080
(b)
Table 4.6: Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.25λ
0
, ε
r
= 2.55)
153
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
606.04 0.965
101 5λ
0
8λ
0
605.96 0.964
51 4λ
0
8λ
0
600.33 0.955
51 5λ
0
8λ
0
605.39 0.964
26 4λ
0
8λ
0
600.82 0.956
26 5λ
0
8λ
0
603.34 0.960
21 4λ
0
8λ
0
608.13 0.968
21 5λ
0
8λ
0
604.81 0.963
101 4λ
0
9λ
0
609.92 0.971
101 5λ
0
9λ
0
604.67 0.962
51 4λ
0
9λ
0
604.11 0.961
51 5λ
0
9λ
0
603.40 0.960
26 4λ
0
9λ
0
608.04 0.968
26 5λ
0
9λ
0
602.33 0.959
21 4λ
0
9λ
0
602.19 0.958
21 5λ
0
9λ
0
607.50 0.967
(a)
N ρ
start
ρ
end
β
1
SW
β
1
SW
/β
0
101 4λ
0
8λ
0
610.13 0.971
101 5λ
0
8λ
0
598.61 0.953
51 4λ
0
8λ
0
609.25 0.970
51 5λ
0
8λ
0
599.69 0.954
26 4λ
0
8λ
0
604.90 0.963
26 5λ
0
8λ
0
600.32 0.955
21 4λ
0
8λ
0
606.37 0.965
21 5λ
0
8λ
0
597.78 0.951
101 4λ
0
9λ
0
614.84 0.979
101 5λ
0
9λ
0
598.74 0.953
51 4λ
0
9λ
0
616.50 0.981
51 5λ
0
9λ
0
599.72 0.954
26 4λ
0
9λ
0
606.43 0.965
26 5λ
0
9λ
0
601.48 0.957
21 4λ
0
9λ
0
614.57 0.978
21 5λ
0
9λ
0
600.26 0.955
(b)
Table 4.7: Surface wave propagation constants retrieved from (a) Eline,
(b) Hline. (f = 30GHz, λ
0
= 1cm, th = 0.05λ
0
, ε
r
= 2.55)
154
5 5.5 6 6.5 7 7.5 8
0
5
10
15
20
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
5 5.5 6 6.5 7 7.5 8
−200
−100
0
100
200
Phase
ρ/λ
0
2 3 4 5 6 7 8 9
0
20
40
60
80
100
120
140
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
2 3 4 5 6 7 8 9
−200
−100
0
100
200
Phase
ρ/λ
0
Figure 4.20: Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.1λ
0
, ε
r
= 2.55, ρ
start
= 5λ
0
,
ρ
end
= 8λ
0
, N = 101)
155
5 5.5 6 6.5 7 7.5 8
50
100
150
200
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
5 5.5 6 6.5 7 7.5 8
−200
−100
0
100
200
Phase
ρ/λ
0
2 3 4 5 6 7 8 9
0
100
200
300
400
500
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
2 3 4 5 6 7 8 9
−200
−100
0
100
200
Phase
ρ/λ
0
Figure 4.21: Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.15λ
0
, ε
r
= 2.55, ρ
start
= 5λ
0
,
ρ
end
= 8λ
0
, N = 101)
156
5 5.5 6 6.5 7 7.5 8
200
250
300
350
400
450
500
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
5 5.5 6 6.5 7 7.5 8
−200
−100
0
100
200
Phase
ρ/λ
0
2 3 4 5 6 7 8 9
100
200
300
400
500
600
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
2 3 4 5 6 7 8 9
−200
−100
0
100
200
Phase
ρ/λ
0
Figure 4.22: Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.19λ
0
, ε
r
= 2.55, ρ
start
= 5λ
0
,
ρ
end
= 8λ
0
, N = 101)
157
4 4.5 5 5.5 6 6.5 7 7.5 8
200
250
300
350
400
450
500
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
+ b
2
e
s
2
(ρ−ρ
0
)
4 4.5 5 5.5 6 6.5 7 7.5 8
−200
−100
0
100
200
Phase
ρ/λ
0
2 3 4 5 6 7 8 9 10
200
250
300
350
400
450
500
550
Magnitude
ρ/λ
0
HFSS E
x
√ρ
GPOF b
1
e
s
1
(ρ−ρ
0
)
+ b
2
e
s
2
(ρ−ρ
0
)
2 3 4 5 6 7 8 9 10
−200
−100
0
100
200
Phase
ρ/λ
0
Figure 4.23: Comparison of GPOF approximation with HFSS data and its Ex
trapolation . (f = 30GHz, λ
0
= 1cm, th = 0.25λ
0
, ε
r
= 2.55, ρ
start
= 4λ
0
,
ρ
end
= 8λ
0
, N = 51)
158
4.2.5 Conclusions
Numerical results and comparisons with theoretical calculations show that our
method works quite successfully along the Eline. The only exception occurs in
the th = 0.05λ
0
case. But along the Hline, successful results are obtained only
in the th = 0.1λ
0
case.
In the implementation of the GPOF Method, the [4λ
0
− 8λ
0
], [5λ
0
− 8λ
0
],
[4λ
0
−9λ
0
] and [5λ
0
−9λ
0
] intervals give good results. The reasons are: ρ should
be suﬃciently large and reﬂections from the radiation boundaries should decay
not to contaminate the solutions.
The number of samples (N) is also important in the GPOF Method. Taking
too many samples makes the system of equations solved in the method more
linearly dependent, on the other hand taking not enough number of samples give
inaccurate results because the behavior of the complex function to be approxi
mated cannot be tracked correctly. In our method, we use N = 51 or N = 101
in the GPOF Method.
Another factor that aﬀects the accuracy of the results is the ratio of the
surface wave propagation constant to space wave propagation constant. When
this ratio is very close to 1, the diﬀerentiation of the space wave and surface wave
terms becomes diﬃcult. This phenomena can be observed at the th = 0.05λ
0
case
where surface wave propagation constant is very close to space wave propagation
constant. As the thickness of the dielectric slab increases β
TM
0
increases too. This
increases the accuracy of results to determine β
TM
0
. However when thickness is
suﬃciently large and TE
1
mode emerges, β
TE
1
cannot be determined as accurate
as β
TM
0
is determined. For each mode that emerges recently, as the thickness
increases, the results will not be very satisfactory at ﬁrst. But as the thickness
continues to increase the results will be more successful for this mode.
159
Chapter 5
CONCLUSIONS
In this thesis, electromagnetic scattering and transmission from metamaterial
structures, such as metamaterial slabs, metamaterial cylinders and metamaterial
coated conducting cylinders, are investigated. These structures are illuminated
by electric line sources or plane waves. The formulation of these wave propagation
problems is done in such a way that it remains valid for any kind of material
used, having any sign combination of constitutive parameters and having any
electric and/or magnetic losses.
For one of these propagation problems i.e., metamaterial coated conducting
cylinders illuminated normally with plane waves, achieving transparency and
maximizing scattering are investigated thoroughly. It is found out that, rig
orous derivation of transparency and resonance conditions for PEC core cylin
der case under the subwavelength limitations yields the same conditions of two
electrically small concentric layers of conjugately paired cylinders, given in the
literature (when the inner core layer is also taken to the PEC limit). These
transparency and resonance conditions heavily depend on the permittivity of
the metamaterial coating (for TE polarization) and the ratio of coreshell radii.
The relations between the permittivity of the coating and the ratio of coreshell
160
radii are investigated for achieving transparency and scattering maximization.
Numerical results show that these analytical relations are quite successful and
work better when the cylindrical scatter is electrically very small. As the future
work, similar analytical transparency and resonance conditions can be derived
and tested for obliquely incident plane waves. Our preliminary numerical re
sults for oblique incidence scenarios show the existence of such transparency and
resonance conditions. As another future work, the inﬁnite length metamaterial
coated conducting cylinder can be truncated, while keeping other geometry and
material parameters the same, and can be simulated in full wave simulators to see
whether such transparency or resonance conditions exist for reallife geometries.
A novel homogenization method for the retrieval of eﬀective constitutive pa
rameters of metamaterials is proposed and implemented. The method is based
on the simple idea that the total reﬂection coeﬃcient from a ﬁnite metamate
rial structure has to resemble the reﬂection from an homogeneous equivalent.
While implementing the method, 1, 2, . . ., 20 unit cells of the same metamaterial
structure are stacked and their reﬂection coeﬃcients are collected. The homog
enization quality of the metamaterial is evaluated in terms of various factors,
which showed that the method is very successful to retrieve the eﬀective consti
tutive parameters of the metamaterial. Since the method is merely dependent
on reﬂection, the homogeneous equivalent characterizes the reﬂection property of
the metamaterial best. As the future work, the method can be modiﬁed to incor
porate also the transmission data, so that the homogeneous equivalent mimics
the metamaterial more successfully in the transmission region. Another future
work can be homogenization in the oblique incidence case, since the method is
already capable for this, if an eﬃcient oblique incidence implementation scheme
can be formed for the simulation of metamaterial.
Finally, another method has been proposed for the retrieval of surface wave
propagation constants on any periodic or nonperiodic grounded slab medium. As
161
a preliminary, the method is applied to grounded dielectric slabs. The numerical
results generally show good agreement with their theoretical counterparts.
162
APPENDIX A
Bessel Functions
In cylindrical coordinate system, while solving the wave equation, Bessel’s dif
ferential equation arises, which can be written as
x
2
d
2
y
dx
2
+ x
dy
dx
+
x
2
−p
2
y = 0. (A.1)
Since Bessel’s equation in (A.1) is a second order diﬀerential equation, it has
two linearly independent solutions:
y(x) = A
1
J
p
(x) + B
1
J
−p
(x) p = 0 or integer, (A.2)
y(x) = A
2
J
n
(x) + B
2
Y
n
(x) p = n = 0 or integer, (A.3)
where J
p
(x) is referred to as the Bessel function of the ﬁrst kind of order p and
Y
p
(x) as the Bessel function of the second kind of order p (or sometimes as the
Neumann function).
When p = n = integer,
J
−n
(x) = (−1)
n
J
n
(x), (A.4)
J
n
(−x) = (−1)
n
J
n
(x). (A.5)
163
Small Argument Forms:
When the argument of the Bessel functions is small (i.e., x →0),
for p = 0:
J
0
(x) · 1, (A.6)
Y
0
(x) ·
2
π
ln
γx
2
, (A.7)
γ = 1.781, (A.8)
for p > 0:
J
p
(x) ·
1
p!
x
2
p
, (A.9)
Y
n
(x) · −
(p −1)!
π
2
x
p
. (A.10)
Large Argument Forms:
When the argument of the Bessel functions is large (i.e., x →∞),
J
p
(x) ·
2
πx
cos
x −
π
4
−
pπ
2
, (A.11)
Y
p
(x) ·
2
πx
sin
x −
π
4
−
pπ
2
. (A.12)
From electromagnetic point of view, these cosine and sine functions in Bessel
functions of the ﬁrst and second kinds represent standing waves. For wave prop
agation, it becomes more convenient to deﬁne Hankel functions:
H
(1)
p
(x) = J
p
(x) + jY
p
(x), (A.13)
H
(2)
p
(x) = J
p
(x) −jY
p
(x), (A.14)
where H
(1)
p
(x) is the Hankel function of the ﬁrst kind of order p and H
(2)
p
(x) is
the Hankel function of the second kind of order p.
164
For large arguments (i.e., x →∞):
H
(1)
p
(x) ·
2
πx
e
j[x−p(π/2)−π/4]
, (A.15)
H
(2)
p
(x) ·
2
πx
e
−j[x−p(π/2)−π/4]
. (A.16)
With the assumed e
jωt
time dependence, Hankel functions of the ﬁrst kind
represent inward propagating waves, whereas Hankel functions of the second kind
represent outward propagating waves.
For the derivatives of Bessel and Hankel functions, the following recurrence
relation can be used:
dF
p
(x)
dx
=
1
2
[F
p−1
(x) −F
p+1
(x)] , (A.17)
where F(x) represents any kind of Bessel or Hankel function. Other alternative
forms of the recurrence relations and many other properties of Bessel and Hankel
functions of integer and noninteger orders can be found in [80].
165
APPENDIX B
Derivation of the φ Components
of Electric and Magnetic Fields:
TM
z
Polarization
To ﬁnd the φ components of the electric and magnetic ﬁelds, we will use Maxwell’s
Equations:
∇E = −jωµH → H = −
1
jωµ
∇E, (B.1)
H
ρ
= −
1
jωµ
1
ρ
∂E
z
∂φ
−
∂E
φ
∂z
, (B.2)
H
φ
= −
1
jωµ
∂E
ρ
∂z
−
∂E
z
∂ρ
, (B.3)
H
z
= −
1
jωµ
1
ρ
¸
∂
∂ρ
(ρE
φ
) −
∂E
ρ
∂φ
. (B.4)
∇H = jωεE → E =
1
jωε
∇H, (B.5)
E
ρ
=
1
jωε
1
ρ
∂H
z
∂φ
−
∂H
φ
∂z
, (B.6)
E
φ
=
1
jωε
∂H
ρ
∂z
−
∂H
z
∂ρ
, (B.7)
E
z
=
1
jωε
1
ρ
¸
∂
∂ρ
(ρH
φ
) −
∂H
ρ
∂φ
. (B.8)
166
In the following derivations, all derivatives of Bessel and Hankel functions are
taken with respective to their entire arguments.
H
i
φ
:
H
i
φ
= −
1
jωµ
0
∂E
i
ρ
∂z
−
∂E
i
z
∂ρ
(B.9)
H
i
φ
= −
1
jωµ
0
¸
∂E
i
ρ
∂z
−
∂
∂ρ
E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
¸
(B.10)
H
i
φ
= −
1
jωµ
0
∂E
i
ρ
∂z
−E
0
k
0
sin
2
θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.11)
E
i
ρ
=
1
jωε
0
1
ρ
∂H
i
z
∂φ
−
∂H
i
φ
∂z
(B.12)
Since H
i
z
= 0,
E
i
ρ
= −
1
jωε
0
∂H
i
φ
∂z
(B.13)
We also know that all ﬁeld variations in the z direction are in the form of
e
jk
0
z cos θ
0
. Therefore,
E
i
ρ
= −
k
0
cos θ
0
ωε
0
H
i
φ
(B.14)
Substituting (B.14) in (B.11),
H
i
φ
= −
1
jωµ
0
−
k
0
cos θ
0
ωε
0
∂H
i
φ
∂z
(B.15)
−E
0
k
0
sin
2
θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
H
i
φ
= −
1
jωµ
0
−
jk
2
0
cos
2
θ
0
ωε
0
H
i
φ
(B.16)
−E
0
k
0
sin
2
θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
H
i
φ
=
k
2
0
cos
2
θ
0
ω
2
µ
0
ε
0
H
i
φ
+
E
0
k
0
sin
2
θ
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.17)
167
Since k
0
= ω
√
µ
0
ε
0
,
(1−cos
2
θ
0
)H
i
φ
= sin
2
θ
0
H
i
φ
=
E
0
k
0
sin
2
θ
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.18)
H
i
φ
=
E
0
k
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
= −j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.19)
E
i
φ
:
E
i
φ
=
1
jωε
0
∂H
i
ρ
∂z
−
∂H
i
z
∂ρ
(B.20)
Since H
i
z
= 0 and
∂
∂z
= jk
0
cos θ
0
,
E
i
φ
=
k
0
cos θ
0
ωε
0
H
i
ρ
(B.21)
H
i
ρ
= −
1
jωµ
0
1
ρ
∂E
i
z
∂φ
−
∂E
i
φ
∂z
(B.22)
H
i
ρ
= −
1
jωµ
0
¸
1
ρ
∂
∂φ
E
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
∂E
i
φ
∂z
¸
(B.23)
H
i
ρ
= −
1
jωµ
0
¸
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
∂E
i
φ
∂z
¸
(B.24)
Substituting (B.21) in (B.24)
H
i
ρ
= −
1
jωµ
0
¸
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
jk
2
0
cos
2
θ
0
ωε
0
H
i
ρ
(B.25)
H
i
ρ
= −
E
0
sin θ
0
ωµ
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+
k
2
0
cos
2
θ
0
ω
2
µ
0
ε
0
H
i
ρ
(B.26)
168
(1−cos
2
θ
0
)H
i
ρ
= sin
2
θ
0
H
i
ρ
= −
E
0
sin θ
0
ωµ
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.27)
H
i
ρ
= −
E
0
ωµ
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.28)
Substituting (B.28) in (B.21) gives
E
i
φ
= −
E
0
k
0
cos θ
0
ω
2
µ
0
ε
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.29)
E
i
φ
= −
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
J
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.30)
E
s
φ
:
E
s
φ
=
1
jωε
0
∂H
s
ρ
∂z
−
∂H
s
z
∂ρ
(B.31)
E
s
φ
=
1
jωε
0
jk
0
cos θ
0
H
s
ρ
(B.32)
−E
0
k
0
sin
2
θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
E
s
φ
=
k
0
cos θ
0
ωε
0
H
s
ρ
−
E
0
k
0
sin
2
θ
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.33)
H
s
ρ
= −
1
jωµ
0
1
ρ
∂E
s
z
∂φ
−
∂E
s
φ
∂z
(B.34)
H
s
ρ
= −
1
jωµ
0
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−jk
0
cos θ
0
E
s
φ
(B.35)
H
s
ρ
= −
E
0
sin θ
0
ωµ
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+
k
0
cos θ
0
ωµ
0
E
s
φ
(B.36)
169
Substituting (B.36) in (B.33) gives
E
s
φ
=
k
0
cos θ
0
ωε
0
−
E
0
sin θ
0
ωµ
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+
k
0
cos θ
0
ωµ
0
E
s
φ
−
E
0
k
0
sin
2
θ
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.37)
E
s
φ
=−
E
0
k
0
sin θ
0
cos θ
0
ω
2
µ
0
ε
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.38)
+
k
2
0
cos
2
θ
0
ω
2
µ
0
ε
0
E
s
φ
−
E
0
k
0
sin
2
θ
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
E
s
φ
=−
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
E
0
k
0
jωε
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.39)
E
s
φ
=−
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+jE
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.40)
H
s
φ
:
H
s
φ
= −
1
jωµ
0
∂E
s
ρ
∂z
−
∂E
s
z
∂ρ
(B.41)
H
s
φ
= −
1
jωµ
0
jk
0
cos θ
0
E
s
ρ
(B.42)
−E
0
k
0
sin
2
θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
H
s
φ
= −
k
0
cos θ
0
ωµ
0
E
s
ρ
+
E
0
k
0
sin
2
θ
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.43)
170
E
s
ρ
=
1
jωε
0
1
ρ
∂H
s
z
∂φ
−
∂H
s
φ
∂z
(B.44)
E
s
ρ
=
1
jωε
0
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−jk
0
cos θ
0
H
s
φ
(B.45)
E
s
ρ
=
E
0
sin θ
0
ωε
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
k
0
cos θ
0
ωε
0
H
s
φ
(B.46)
Substituting (B.46) in (B.43) gives
H
s
φ
=−
k
0
cos θ
0
ωµ
0
E
0
sin θ
0
ωε
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−
k
0
cos θ
0
ωε
0
H
s
φ
+
E
0
k
0
sin
2
θ
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.47)
H
s
φ
=−
E
0
k
0
sin θ
0
cos θ
0
ω
2
µ
0
ε
0
ρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.48)
+
k
2
0
cos
2
θ
0
ω
2
µ
0
ε
0
H
s
φ
+
E
0
k
0
sin
2
θ
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
H
s
φ
=−
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
+
E
0
k
0
jωµ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.49)
H
s
φ
=−
E
0
cos θ
0
k
0
ρ sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
−j
E
0
η
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
c
n
H
(2)
n
(k
0
ρ sin θ
0
)e
jn(φ−φ
0
)
(B.50)
E
t
φ
:
E
t
φ
=
1
jωε
∂H
t
ρ
∂z
−
∂H
t
z
∂ρ
(B.51)
171
E
t
φ
=
1
jωε
jk
0
cos θ
0
H
t
ρ
−E
0
k sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.52)
E
t
φ
=
k
0
cos θ
0
ωε
H
t
ρ
−
E
0
k sin θ
0
sin θ
1
jωε
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.53)
H
t
ρ
= −
1
jωµ
1
ρ
∂E
t
z
∂φ
−
∂E
t
φ
∂z
(B.54)
H
t
ρ
= −
1
jωµ
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−jk
0
cos θ
0
E
t
φ
(B.55)
H
t
ρ
= −
E
0
sin θ
0
ωµρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+
k
0
cos θ
0
ωµ
E
t
φ
(B.56)
Substituting (B.56) in (B.53) gives
E
t
φ
=
k
0
cos θ
0
ωε
−
E
0
sin θ
0
ωµρ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+
k
0
cos θ
0
ωµ
E
t
φ
−
E
0
k sin θ
0
sin θ
1
jωε
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.57)
E
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
ω
2
µερ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+
k
2
0
cos
2
θ
0
ω
2
µε
E
t
φ
−
E
0
k sin θ
0
sin θ
1
jωε
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.58)
Since
k
2
0
cos
2
θ
0
ω
2
µε
=
k
2
0
cos
2
θ
0
k
2
= cos
2
θ
1
(B.59)
E
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−
E
0
k sin θ
0
jωε sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.60)
172
E
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+jE
0
η
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.61)
H
t
φ
:
H
t
φ
= −
1
jωµ
∂E
t
ρ
∂z
−
∂E
t
z
∂ρ
(B.62)
H
t
φ
=−
1
jωµ
jk
0
cos θ
0
E
t
ρ
(B.63)
−E
0
k sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
H
t
φ
= −
k
0
cos θ
0
ωµ
E
t
ρ
+
E
0
k sin θ
0
sin θ
1
jωµ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.64)
E
t
ρ
=
1
jωε
1
ρ
∂H
t
z
∂φ
−
∂H
t
φ
∂z
(B.65)
E
t
ρ
=
1
jωε
1
ρ
jE
0
sin θ
0
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−jk
0
cos θ
0
H
t
φ
(B.66)
E
t
ρ
=
E
0
sin θ
0
ωερ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−
k
0
cos θ
0
ωε
H
t
φ
(B.67)
Substituting (B.67) in (B.64) gives
H
t
φ
=−
k
0
cos θ
0
ωµ
E
0
sin θ
0
ωερ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−
k
0
cos θ
0
ωε
H
t
φ
+
E
0
k sin θ
0
sin θ
1
jωµ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.68)
H
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
ω
2
µερ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.69)
+
k
2
0
cos
2
θ
0
ω
2
µε
H
t
φ
+
E
0
k sin θ
0
sin θ
1
jωµ
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
173
Since
k
2
0
cos
2
θ
0
ω
2
µε
=
k
2
0
cos
2
θ
0
k
2
= cos
2
θ
1
(B.70)
H
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
+
E
0
k sin θ
0
jωµsin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.71)
H
t
φ
=−
E
0
k
0
sin θ
0
cos θ
0
k
2
ρ sin
2
θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
nj
−n
¯a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
−j
E
0
η
sin θ
0
sin θ
1
e
jk
0
z cos θ
0
+∞
¸
n=−∞
j
−n
a
n
J
n
(kρ sin θ
1
)e
jn(φ−φ
0
)
(B.72)
174
APPENDIX C
Derivation of the Transparency
Condition
For achieving transparency with metamaterial coated conducting cylinders at
normal incidence and TE
z
polarization, numerator of the scattering coeﬃcients,
c
n
, given in (2.240) should be zero:
num¦c
n
¦ = ζJ
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] (C.1)
−J
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
= 0.
Let T
1
and T
2
be the the two terms of the numerator of c
n
:
T
1
= ζJ
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (C.2)
T
2
= −J
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (C.3)
such that T
1
+T
2
= 0.
Using the small argument approximations and the recurrence relation,
T
1
= ζ
1
n!
k
0
b
2
n
¸
1
(n−1)!
k
c
a
2
n−1
−
1
(n+1)!
k
c
a
2
n+1
2
(C.4)
175
¸
¸
−
(n−2)!
π
2
k
c
b
n−1
+
n!
π
2
k
c
b
n+1
2
−
¸
1
(n−1)!
k
c
b
2
n−1
−
1
(n+1)!
k
c
b
2
n+1
2
¸
¸
−
(n−2)!
π
2
k
c
a
n−1
+
n!
π
2
k
c
a
n+1
2
¸
¸
¸
,
T
2
= −
¸
1
(n−1)!
k
0
b
2
n−1
−
1
(n+1)!
k
0
b
2
n+1
2
(C.5)
¸
1
(n−1)!
k
c
a
2
n−1
−
1
(n+1)!
k
c
a
2
n+1
2
−
(n −1)!
π
2
k
c
b
n
−
1
n!
k
c
b
2
n
¸
¸
−
(n−2)!
π
2
k
c
a
n−1
+
n!
π
2
k
c
a
n+1
2
¸
¸
¸
,
T
1
= ζ
1
n!
k
0
b
2
n
1
4π
(C.6)
¸
−
1
(n −1)
a
b
n−1
+ n
a
b
n−1
4
(k
c
b)
2
+
1
(n + 1)n(n −1)
a
b
n−1
(k
c
a)
2
4
−
1
(n + 1)
a
b
n+1
−
−
1
(n −1)
b
a
n−1
+ n
b
a
n−1
4
(k
c
a)
2
+
1
(n + 1)n(n −1)
b
a
n−1
(k
c
b)
2
4
−
1
(n + 1)
b
a
n+1
¸¸
,
T
2
= −
1
4π
1
(n −1)!
k
0
b
2
n−1
−
1
(n + 1)!
k
0
b
2
n+1
(C.7)
¸
−
a
b
n−1
2
k
c
b
+
1
(n + 1)n
a
b
n
k
c
a
2
+
1
(n + 1)n
b
a
n−1
k
c
b
2
−
b
a
n
2
k
c
a
¸
.
176
Since k
c
a <1 and k
c
b <1, we can keep only the dominant terms in T
1
and
T
2
(i.e., the terms where k
c
a and k
c
b are in the denominator):
T
1
= ζ
1
n!
k
0
b
2
n
1
4π
¸
n
a
b
n−1
4
(k
c
b)
2
−n
b
a
n−1
4
(k
c
a)
2
¸
, (C.8)
T
2
=
1
4π
1
(n −1)!
k
0
b
2
n−1
−
1
(n + 1)!
k
0
b
2
n+1
(C.9)
¸
a
b
n−1
2
k
c
b
+
b
a
n
2
k
c
a
.
Writing
1
(n −1)!
k
0
b
2
n−1
=
1
n!
k
0
b
2
n
n
2
k
0
b
, (C.10)
1
(n + 1)!
k
0
b
2
n+1
=
1
n!
k
0
b
2
n
1
(n + 1)
k
0
b
2
, (C.11)
1
(n −1)!
k
0
b
2
n−1
−
1
(n + 1)!
k
0
b
2
n+1
=
1
n!
k
0
b
2
n
¸
n
2
k
0
b
−
1
(n + 1)
k
0
b
2
·
1
n!
k
0
b
2
n
n
2
k
0
b
, (C.12)
T
2
becomes
T
2
=
1
n!
k
0
b
2
n
1
4π
¸
n
a
b
n−1
4
k
0
k
c
b
2
+ n
b
a
n
4
k
0
k
c
ab
, (C.13)
T
1
+ T
2
=
1
n!
k
0
b
2
n
1
4π
(C.14)
¸
ζn
a
b
n−1
4
k
2
c
b
2
−ζn
b
a
n−1
4
k
2
c
a
2
+ n
a
b
n−1
4
k
0
k
c
b
2
+n
b
a
n
4
k
0
k
c
ab
= 0.
177
Therefore,
ζ
a
b
n−1
1
k
2
c
b
2
−ζ
b
a
n−1
1
k
2
c
a
2
+
a
b
n−1
1
k
0
k
c
b
2
+
b
a
n
1
k
0
k
c
ab
= 0 . (C.15)
Note that,
ζ =
η
c
η
0
=
µ
c
ε
c
µ
0
ε
0
=
µ
c
ε
0
µ
0
ε
c
, (C.16)
k
c
= ω
√
µ
c
ε
c
, (C.17)
k
0
= ω
√
µ
0
ε
0
. (C.18)
Hence,
0 =
a
b
n−1
µ
c
ε
0
µ
0
ε
c
1
ω
2
µ
c
ε
c
1
b
2
−
b
a
n−1
µ
c
ε
0
µ
0
ε
c
1
ω
2
µ
c
ε
c
1
a
2
(C.19)
+
a
b
n−1
1
ω
2
√
µ
0
ε
0
√
µ
c
ε
c
1
b
2
+
b
a
n
1
ω
2
√
µ
0
ε
0
√
µ
c
ε
c
1
ab
,
or dividing each term by ω
2
ζ,
0 =
a
b
n−1
1
µ
c
ε
c
1
b
2
−
b
a
n−1
1
µ
c
ε
c
1
a
2
(C.20)
+
a
b
n−1
µ
0
ε
c
µ
c
ε
0
1
√
µ
0
ε
0
µ
c
ε
c
1
b
2
+
b
a
n
µ
0
ε
c
µ
c
ε
0
1
√
µ
0
ε
0
µ
c
ε
c
1
ab
,
0 =
a
b
n−1
1
µ
c
ε
c
1
b
2
−
b
a
n−1
1
µ
c
ε
c
1
a
2
+
a
b
n−1
1
µ
c
ε
0
1
b
2
+
b
a
n
1
µ
c
ε
0
1
ab
=
a
n
b
n
b
a
1
µ
c
ε
c
1
b
2
−
b
n
a
n
a
b
1
µ
c
ε
c
1
a
2
+
a
n
b
n
b
a
1
µ
c
ε
0
1
b
2
+
b
n
a
n
1
µ
c
ε
0
1
ab
=
1
µ
c
ab
¸
a
n
b
n
1
ε
c
−
b
n
a
n
1
ε
c
+
a
n
b
n
1
ε
0
+
b
n
a
n
1
ε
0
. (C.21)
Denoting γ = a/b,
γ
n
ε
c
−
γ
−n
ε
c
+
γ
n
ε
0
+
γ
−n
ε
0
= 0, (C.22)
178
γ
n
1
ε
c
+
1
ε
0
= γ
−n
1
ε
c
−
1
ε
0
, (C.23)
γ
2n
=
1
ε
c
−
1
ε
0
1
ε
c
+
1
ε
0
=
ε
0
−ε
c
ε
0
+ ε
c
, (C.24)
γ =
2n
ε
0
−ε
c
ε
0
+ ε
c
. (C.25)
179
APPENDIX D
Derivation of the Resonance
Condition
For scattering maximization (i.e., resonance) with metamaterial coated conduct
ing cylinders at normal incidence and TE
z
polarization, denominator of the
scattering coeﬃcients, c
n
, given in (2.241) should be zero:
den¦c
n
¦ = −ζH
(2)
n
(k
0
b) [J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] (D.1)
+H
(2)
n
[J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)]
= 0.
Let R
1
and R
2
be the the two terms of the denominator of c
n
:
R
1
= −ζH
(2)
n
[J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (D.2)
R
2
= H
(2)
n
[J
n
(k
c
a)Y
n
(k
c
b) −J
n
(k
c
b)Y
n
(k
c
a)] , (D.3)
such that R
1
+R
2
= 0.
Using the small argument approximations and the recurrence relation,
R
1
= −ζ
1
n!
k
0
b
2
n
+j
(n −1)!
π
2
k
0
b
n
(D.4)
180
¸
1
(n−1)!
k
c
a
2
n−1
−
1
(n+1)!
k
c
a
2
n+1
2
¸
¸
−
(n−2)!
π
2
k
c
b
n−1
+
n!
π
2
k
c
b
n+1
2
−
¸
1
(n−1)!
k
c
b
2
n−1
−
1
(n+1)!
k
c
b
2
n+1
2
¸
¸
−
(n−2)!
π
2
k
c
a
n−1
+
n!
π
2
k
c
a
n+1
2
¸
¸
¸
,
R
2
=
¸
¸
1
(n−1)!
k
0
b
2
n−1
−
1
(n+1)!
k
0
b
2
n+1
2
−j
−
(n−2)!
π
2
k
0
b
n−1
+
n!
π
2
k
0
b
n+1
2
¸
1
(n−1)!
k
c
a
2
n−1
−
1
(n+1)!
k
c
a
2
n+1
2
−
(n −1)!
π
2
k
c
b
n
−
1
n!
k
c
b
2
n
¸
¸
−
(n−2)!
π
2
k
c
a
n−1
+
n!
π
2
k
c
a
n+1
2
¸
¸
¸
, (D.5)
R
1
= −ζ
1
n!
k
0
b
2
n
+ j
(n −1)!
π
2
k
0
b
n
1
4π
(D.6)
¸
−
1
(n −1)
a
b
n−1
+n
a
b
n−1
4
(k
c
b)
2
+
1
(n + 1)n(n −1)
a
b
n−1
(k
c
a)
2
4
−
1
(n + 1)
a
b
n+1
−
−
1
(n −1)
b
a
n−1
+n
b
a
n−1
4
(k
c
a)
2
+
1
(n + 1)n(n −1)
b
a
n−1
(k
c
b)
2
4
−
1
(n + 1)
b
a
n+1
¸¸
,
R
2
=
1
2π
¸
¸
1
(n−1)!
k
0
b
2
n−1
−
1
(n+1)!
k
0
b
2
n+1
2
−j
−
(n−2)!
π
2
k
0
b
n−1
+
n!
π
2
k
0
b
n+1
2
¸
−
a
b
n−1
2
k
c
b
+
1
(n + 1)n
a
b
n
k
c
a
2
+
1
(n + 1)n
b
a
n−1
k
c
b
2
−
b
a
n
2
k
c
a
¸
. (D.7)
181
Since k
c
a <1 and k
c
b <1, we can keep only the dominant terms in R
1
and
R
2
(i.e., the terms where k
c
a and k
c
b are in the denominator):
R
1
= −jζ
(n −1)!
π
2
k
0
b
n
1
4π
¸
n
a
b
n−1
4
(k
c
b)
2
−n
b
a
n−1
4
(k
c
a)
2
¸
, (D.8)
R
2
= j
1
4π
−
(n −2)!
π
2
k
0
b
n−1
+
n!
π
2
k
0
b
n+1
¸
a
b
n−1
2
k
c
b
+
b
a
n
2
k
c
a
. (D.9)
Writing
−
(n −2)!
π
2
k
0
b
n−1
= −
(n −1)!
π
2
k
0
b
n
1
(n −1)
k
0
b
2
, (D.10)
n!
π
2
k
0
b
n+1
= −
(n −1)!
π
2
k
0
b
n
(−n)
2
k
0
b
, (D.11)
−
(n −2)!
π
2
k
0
b
n−1
+
n!
π
2
k
0
b
n+1
= −
(n −1)!
π
2
k
0
b
n
¸
1
(n −1)
k
0
b
2
−n
2
k
0
b
·
(n −1)!
π
2
k
0
b
n
n
2
k
0
b
, (D.12)
R
2
becomes
R
2
= j
1
4π
(n −1)!
π
2
k
0
b
n
n
2
k
0
b
¸
a
b
n−1
2
k
c
b
+
b
a
n
2
k
c
a
, (D.13)
R
1
+ R
2
= −j
1
4π
n!
π
2
k
0
b
n
(D.14)
¸
ζ
a
b
n−1
4
(k
c
b)
2
−ζ
b
a
n−1
4
(k
c
a)
2
−
2
k
0
b
a
b
n−1
2
k
c
b
−
2
k
0
b
b
a
n
2
k
c
a
= 0.
182
Therefore,
ζ
a
b
n−1
1
k
2
c
b
2
−ζ
b
a
n−1
1
k
2
c
a
2
−
a
b
n−1
1
k
0
k
c
b
2
−
b
a
n
1
k
0
k
c
ab
= 0 . (D.15)
Note that,
ζ =
η
c
η
0
=
µ
c
ε
c
µ
0
ε
0
=
µ
c
ε
0
µ
0
ε
c
, (D.16)
k
c
= ω
√
µ
c
ε
c
, (D.17)
k
0
= ω
√
µ
0
ε
0
. (D.18)
Hence,
0 =
a
b
n−1
µ
c
ε
0
µ
0
ε
c
1
ω
2
µ
c
ε
c
1
b
2
−
b
a
n−1
µ
c
ε
0
µ
0
ε
c
1
ω
2
µ
c
ε
c
1
a
2
(D.19)
−
a
b
n−1
1
ω
2
√
µ
0
ε
0
√
µ
c
ε
c
1
b
2
−
b
a
n
1
ω
2
√
µ
0
ε
0
√
µ
c
ε
c
1
ab
,
or dividing each term by ω
2
ζ,
0 =
a
b
n−1
1
µ
c
ε
c
1
b
2
−
b
a
n−1
1
µ
c
ε
c
1
a
2
(D.20)
−
a
b
n−1
µ
0
ε
c
µ
c
ε
0
1
√
µ
0
ε
0
µ
c
ε
c
1
b
2
−
b
a
n
µ
0
ε
c
µ
c
ε
0
1
√
µ
0
ε
0
µ
c
ε
c
1
ab
,
0 =
a
b
n−1
1
µ
c
ε
c
1
b
2
−
b
a
n−1
1
µ
c
ε
c
1
a
2
−
a
b
n−1
1
µ
c
ε
0
1
b
2
−
b
a
n
1
µ
c
ε
0
1
ab
=
a
n
b
n
b
a
1
µ
c
ε
c
1
b
2
−
b
n
a
n
a
b
1
µ
c
ε
c
1
a
2
−
a
n
b
n
b
a
1
µ
c
ε
0
1
b
2
−
b
n
a
n
1
µ
c
ε
0
1
ab
=
1
µ
c
ab
¸
a
n
b
n
1
ε
c
−
b
n
a
n
1
ε
c
−
a
n
b
n
1
ε
0
−
b
n
a
n
1
ε
0
. (D.21)
Denoting γ = a/b,
γ
n
ε
c
−
γ
−n
ε
c
−
γ
n
ε
0
−
γ
−n
ε
0
= 0, (D.22)
183
γ
n
1
ε
c
−
1
ε
0
= γ
−n
1
ε
c
+
1
ε
0
, (D.23)
γ
2n
=
1
ε
c
+
1
ε
0
1
ε
c
−
1
ε
0
=
ε
0
+ ε
c
ε
0
−ε
c
, (D.24)
γ =
2n
ε
0
+ ε
c
ε
0
−ε
c
. (D.25)
184
Bibliography
[1] C. Li and Z. Shen, “Electromagnetic scattering by a conducting cylinder
coated with metamaterials,” Progress In Electromagnetics Research, vol. 42,
pp. 91–105, 2003.
[2] X. Zhou and G. Hu, “Design for electromagnetic wave transparency with
metamaterials,” Phys. Rev. E, vol. 74, pp. 026607/1–8, Aug. 2006.
[3] J. B. Pendry, D. Schurig, and D. R. Smith, “Controlling electromagnetic
ﬁelds,” Science, vol. 312, pp. 1780–1782, June 2006.
[4] S. A. Cummer, B. I. Popa, D. Schurig, D. R. Smith, and J. Pendry, “Full
wave simulations of electromagnetic cloaking structures,” Phys. Rev. E,
vol. 74, pp. 036621/1–5, Sep. 2006.
[5] A. Al` u and N. Engheta, “Achieving transparency with plasmonic and meta
material coatings,” Phys. Rev. E, vol. 72, pp. 016623/1–9, July 2005.
[6] A. Al` u and N. Engheta, “Plasmonic materials in transparency and cloaking
problems: mechanism, robustness, and physical insights,” Optics Express,
vol. 15, pp. 3318–3332, March 2007.
[7] A. Al` u and N. Engheta, “Pairing an epsilonnegative slab with a munegative
slab: Resonance, tunneling and transparency,” IEEE Trans. Antennas Prop
agat., vol. 51, pp. 2558–2571, Oct. 2003.
185
[8] A. Al` u and N. Engheta, “Resonances in subwavelength cylindrical struc
tures made of pairs of doublenegative and doublepositive or epsilon
negative and munegative coaxial shells,” in Proc. Int. Conf. on Electro
magnetics in Advanced Applications, (Torino, Italy), pp. 435–438, Sep. 812
2003.
[9] A. Al` u and N. Engheta, “Subwavelength resonant structures containing
doublenegative (DNG) or singlenegative (SNG) media: Planar, cylindri
cal and spherical cavities, waveguides, and open scatterers,” in Progress in
Electromagnetic Research Symp., (Waikiki, HI), p. 12, Oct. 1316 2003.
[10] A. Al` u and N. Engheta, “Peculiar radar cross section properties of double
negative and singlenegative metamaterials,” in Proc. of the 2004 IEEE
Radar Conf., (Philadelphia, PA, USA), Apr. 2629 2004.
[11] A. Al` u and N. Engheta, “Polarizabilities and eﬀective parameters for col
lections of spherical nanoparticles formed by pairs of concentric double
negative (DNG), singlenegative (SNG) and/or doublepositive (DPS) meta
material layers,” J. Appl. Phys, vol. 97, pp. 094310/1–12, Apr. 2005.
[12] A. Al` u, N. Engheta, and R. W. Ziolkowski, “FDTD simulation of tunneling
and “growing exponential” in a pair of epsilonnegative and munegative
slabs,” in USNCURSI Nat. Radio Science Meeting Dig., (Monterey, CA),
p. 18, Jun. 2026 2004.
[13] S. Arslanagic and O. Breinbjerg, “Electriclinesource illumination of a cir
cular cylinder of lossless doublenegative material: an investigation of near
ﬁeld, directivity, and radiation resistance,” IEEE Antennas Propagat. Mag.,
vol. 48, pp. 38–54, June 2006.
[14] S. Arslanagic, R. W. Ziolkowski, and O. Breinbjerg, “Excitation of an elec
trically small metamaterialcoated cylinder by an arbitrarily located line
186
source,” Microwave and Optical Technology Letters, vol. 48, pp. 2598–2606,
Dec. 2006.
[15] R. W. Ziolkowski and A. D. Kipple, “Application of double negative mate
rials to increase the power radiated by electrically small antennas,” IEEE
Trans. Antennas Propagat., vol. 51, pp. 2626–2640, Oct. 2003.
[16] A. Erentok and R. W. Ziolkowski, “Eﬃcient electrically small antenna de
sign using an electric dipole in a multilayered ENG metamaterial shell,” in
Int. Workshop on Antenna Technology: Small Antennas and Novel Meta
materials, (White Plains, New York), pp. 400–403, March 68 2006.
[17] R. W. Ziolkowski and A. Erentok, “Metamaterialbased eﬃcient electrically
small antennas,” IEEE Trans. Antennas Propagat., vol. 54, pp. 2113–2130,
July 2006.
[18] D. R. Smith, D. C. Vier, N. Krol, and S. Schultz, “Direct calculation of
permeability and permittivity for a lefthanded metamaterial,” Appl. Phys.
Lett., vol. 77, pp. 2246–2248, Oct. 2000.
[19] D. R. Smith and J. B. Pendry, “Homogenization of metamaterials by ﬁeld
averaging,” J. Opt. Soc. Am. B, vol. 23, pp. 391–403, Mar. 2006.
[20] D. R. Smith, S. Shultz, P. Markos, and C. M. Soukoulis, “Determination of
eﬀective permittivity and permeability of metamaterials from reﬂection and
transmission coeﬃcients,” Phys. Rev. B, vol. 65, p. 195104, 2002.
[21] D. R. Smith, D. C. Vier, T. Koschny, and C. M. Soukoulis, “Electromag
netic parameter retrieval from inhomogeneous metamaterials,” Phys. Rev.
E, vol. 71, p. 036617, 2005.
[22] T. Koschny, M. Kafesaki, E. N. Economou, and C. M. Soukoulis, “Eﬀective
medium theory of lefthanded materials,” Phys. Rev. Lett, vol. 93, p. 107402,
2004.
187
[23] J. B. Pendry, “Negative refraction makes a perfect lens,” Phys. Rev. Lett.,
vol. 85, pp. 3966–3969, Oct. 2000.
[24] P. M. Valanju, R. M. Walser, and A. P. Valanju, “Wave refraction in
negativeindex media: Always positive and very inhomogeneous,” Phys. Rev.
Lett., vol. 88, p. 187401, Apr 2002.
[25] R. W. Ziolkowski and E. Heyman, “Wave propagation in media having nega
tive permittivity and permeability,” Phys. Rev. E, vol. 64, pp. 056625/1–15,
Nov. 2001.
[26] D. R. Smith and N. Kroll, “Negative refractive index in lefthanded mate
rials,” Phys. Rev. Lett., vol. 85, pp. 2933–2936, Oct. 2000.
[27] C. A. Balanis, Advanced Engineering Electromagnetics. New York: Wiley,
1989.
[28] R. F. Harrington, TimeHarmonic Electromagnetic Fields. New York:
McGrawHill, 1961.
[29] J. R. Wait, Electromagnetic Radiation from Cylindrical Structures. London:
Peter Peregrinus Ltd., 1988.
[30] R. W. Ziolkowski, “Design, fabrication and testing of double negative meta
materials,” IEEE Trans. Antennas Propagat., vol. 65, pp. 1516–1529, July
2003.
[31] N. Kinayman and M. I. Aksun, Modern Microwave Circuits. Artech House,
Norwood, MA, 2005.
[32] Y. Hua and T. K. Sarkar, “Generalized penciloffunction method for ex
tracting poles of an em system from its transient response,” IEEE Trans.
Antennas Propagat., vol. 37, pp. 229–234, Feb. 1989.
188
[33] C¸ a˘gatay Tokg¨ oz, “Derivation of closedform Green’s functions for cylindri
cally stratiﬁed media,” Master’s thesis, Middle East Technical University,
1997.
[34] S. Maci, M. Caiazzo, A. Cucini, and M. Casaletti, “A polezero matching
method for ebg surfaces composed of a dipole fss printed on a grounded
dielectric slab,” IEEE Trans. Antennas Propagat., vol. 53, pp. 70–81, Jan.
2005.
[35] N. Engheta and R. W. Ziolkowski, “A positive future for doublenegative
metamaterials,” vol. 53, pp. 1535–1556, Apr. 2005.
[36] X. S. Rao and C. K. Ong, “Ampliﬁcation of evanescent waves in a lossy
lefthanded material slab,” Phys. Rev. B, Condens. Matter, vol. 68, no. 11.
Paper 113103.
[37] C. Caloz and T. Itoh, “Array factor approach of leakywave antennas and ap
plication to 1d/2d composite right/lefthanded (crlh) structures,” vol. 14,
pp. 274–276, June 2004.
[38] A. P. Feresidis, G. Apostolopoulos, N. Serfas, and J. C. Vardaxoglou,
“Closely coupled metallodielectric electromagnetic bandgap structures
formed by doublelayer dipole and tripole arrays,” IEEE Trans. Antennas
Propagat., vol. 52, pp. 1149–1158, May 2004.
[39] A. D. Boardman, P. Egan, L. Velasco, and N. King, “Control of planar
nonlinear guided waves and spatial solitons with a lefthanded medium,” J.
Opt. A: Pure Appl. Opt., vol. 7, pp. 57–67, Jan. 2005.
[40] S. Maci, M. Casaletti, M. Caiazzo, and A. Cucini, “Dispersion analysis
of printed periodic structures by using a polezero network synthesis,” in
17th International Conference on Applied Electromagnetics and Communi
cations, ICECom 2003, (Dubrovnik, Croatia), pp. 300–303, Oct. 2003.
189
[41] S. Maci, M. Camletti, M. Caiazzo, and C. Boﬀa, “Dispersion properties
of periodic grounded structures via equivalent network synthesis,” in IEEE
Antennas and Propagation Society International Symposium, 2003, vol. 1,
pp. 493–496, June 2003.
[42] D. Artigas and L. Torner, “Dyakonov surface waves in photonic metamate
rials,” Phys. Rev. Lett., vol. 94, Jan. 2005. Paper 013901.
[43] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, P. Lampariello, G. Lo
vat, and S. Paulotto, “Eﬀects of leakywave propagation in metamaterial
grounded slabs excited by a dipole source,” vol. 53, pp. 32–44, Jan. 2005.
[44] Y. Guo, G. Goussetis, A. P. Feresidis, and J. C. Vardaxoglou, “Eﬃcient
modeling of novel uniplanar lefthanded metamaterials,” vol. 53, pp. 1462–
1468, Apr. 2005.
[45] F. D. M. Haldane, “Electromagnetic Surface Modes at Interfaces with Neg
ative Refractive Index make a “NotQuitePerfect” Lens,” ArXiv Condensed
Matter eprints, June 2002.
[46] C. Caloz, C. J. Lee, D. R. Smith, J. B. Pendry, and T. Itoh, “Existence
and properties of microwave surface plasmons at the interface between a
righthanded and a lefthanded media,” in IEEE Antennas and Propagation
Society International Symposium, 2004, vol. 3, pp. 3151–3154, June 2004.
[47] R. Ruppin, “Extinction properties of a sphere with negative permittivity
and permeability,” Solid State Commun., vol. 116, pp. 411–415, Aug. 2000.
[48] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, P. Lampariello, G. Lo
vat, and S. Paulotto, “Fundamental modal properties of surface waves on
metamaterial grounded slabs,” vol. 53, pp. 1431–1442, Apr. 2005.
[49] F. Capolino, D. R. Jackson, and D. R. Wilton, “Fundamental properties of
the ﬁeld at the interface between air and a periodic artiﬁcial material excited
190
by a line source,” IEEE Trans. Antennas Propagat., vol. 53, pp. 91–99, Jan.
2005.
[50] K. Y. Kim, Guided and Leaky Modes of Circular Open Electromagnetic
Waveguides Dielectric, Plasma, and Metamaterial Columns. PhD thesis,
The Graduate School, Kyungpook National University, Dec. 2004.
[51] R. S. Kshetrimayum and L. Zhu, “Guidedwave characteristics of waveguide
based periodic structures loaded with various fss strip layers,” IEEE Trans.
Antennas Propagat., vol. 53, pp. 120–124, Jan. 2005.
[52] A. Alu, F. Bilotti, N. Engheta, and L. Vegni, “How metamaterials may
signiﬁcantly aﬀect the wave transmission through a subwavelength hole
in a ﬂat perfectly conducting screen,” in IEE Workshop on Metamaterials
for Microwave and (Sub)millimetre Wave Applications: Photonic Bandgap
and Double Negative Designs, Components and Experiments, (London, UK),
Nov. 2003.
[53] X. Yina, L. Hesselink, Z. Liu, N. Fang, and X. Zhang, “Large positive
and negative lateral optical beam displacements due to surface plasmon
resonance,” Appl. Phys. Lett., vol. 85, pp. 372–374, July 2004.
[54] C. Caloz, S. Lim, C. Allen, and T. Itoh, “Leakage phenomena from negative
refractive index structures,” in URSI Int. Electromagnetic Theory Symp.,
(Pisa, Italy), May 2004.
[55] A. A. Oliner and D. R. Jackson, “Leaky surfaceplasmon theory for dra
matically enhanced transmission through a subwavelength aperture, part i:
Basic features,” in IEEE Antennas and Propagation Society International
Symposium, 2003, vol. 2, pp. 1091–1094, Aug. 2003.
[56] C. A. Allen, C. Caloz, and T. Itoh, “Leakywaves in a metamaterial
based twodimensional structure for a conical beam antenna application,”
191
in Microwave Symposium Digest, 2004 IEEE MTTS International, vol. 1,
pp. 305–308, June 2004.
[57] A. K. Iyer and G. V. Eleftheriades, “Leakywave radiation from planar
negativerefractiveindex transmissionline metamaterials,” in IEEE An
tennas and Propagation Society International Symposium, 2004, vol. 2,
pp. 1411– 1414, June 2004.
[58] A. Alu, F. Bilotti, N. Engheta, and L. Vegni, “Metamaterial covers over
a small aperture.” Submitted to IEEE Trans. Antennas Propagat. under
review.
[59] A. J. Viitanen and S. A. Tretyakov, “Metawaveguides formed by arrays of
small resonant particles over a ground plane,” J. Opt. A: Pure Appl. Opt.,
vol. 7, pp. 133–140, Jan. 2005.
[60] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, P. Lampariello, G. Lovat,
and S. Paulotto, “Modal properties of surface and leaky waves propagating
at arbitrary angles along a metal strip grating on a grounded slab,” IEEE
Trans. Antennas Propagat., vol. 53, pp. 36–46, Jan. 2005.
[61] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, P. Lampariello, G. Lovat,
and S. Paulotto, “New dispersion characteristics and surfacewave suppres
sion in doublenegative metamaterial grounded slabs,” in URSI Int. Elec
tromagnetic Theory Symp., (Pisa, Italy), pp. 379–381, May 2004.
[62] I. V. Shadrivov, A. A. Sukhorukov, Y. S. Kivshar, A. A. Zharov, A. D.
Boardman, and P. Egan, “Nonlinear surface waves in lefthanded materials,”
Phys. Rev. E, vol. 69, Jan. 2004. Paper 016617.
[63] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, G. Lovat, and S. Paulotto,
“Novel analytical representations of the continuousspectrum current in mul
tilayer stripline structures,” vol. 47, pp. 17–27, Feb. 2005.
192
[64] A. Alu and N. Engheta, “Pairing an epsilonnegative slab with a munegative
slab: Resonance, tunneling and transparency,” IEEE Trans. Antennas Prop
agat., vol. 51, pp. 2558–2571, Oct. 2003.
[65] S. A. Ramakrishna, “Physics of negative refractive index materials,” Rep.
Prog. Phys., vol. 68, pp. 449–521, Jan. 2005.
[66] T. Itoh, “Invited paper: Prospects for metamaterials,” Electron. Lett.,
vol. 40, pp. 972–973, Aug. 2004.
[67] M. Beruete, M. Sorolla, I. Campillo, and J. S. Dolado, “Subwavelength
slotted corrugated plate with enhanced quasioptical millimeter wave trans
mission,” vol. 15, pp. 286–288, Apr. 2005.
[68] R. Ruppin, “Surface polaritons and extinction properties of a lefthanded
material cylinder,” J. Phys.: Condens. Matter, vol. 16, pp. 5991–5998, Aug.
2004.
[69] C. J. Tang and L. Gao, “Surface polaritons and imaging properties of
a multilayer structure containing negativerefractiveindex materials,” J.
Phys.: Condens. Matter, vol. 16, pp. 4743–4751, June 2004.
[70] R. Ruppin, “Surface polaritons of a lefthanded material slab,” J. Phys.:
Condens. Matter, vol. 13, pp. 1811–1819, Jan. 2001.
[71] R. Ruppin, “Surface polaritons of a lefthanded medium,” Phys. Rev. A,
vol. 277, pp. 61–64, Oct. 2000.
[72] I. V. Shadrivov, A. A. Sukhorukov, Y. S. Kivshar, A. D. Boardman, and
A. A. Zharov, “Surface polaritons of nonlinear lefthanded materials,” in
Quantum Electronics Conference, 2003, p. 55, June 2003.
[73] S. Ancey, Y. Decanini, A. Folacci, and P. Gabrielli, “Surface polaritons on
lefthanded cylinders: A complex angular momentum analysis,” Phys. Rev.
B, vol. 72, Aug.
193
[74] P. Baccarelli, P. Burghignoli, G. Lovat, and S. Paulotto, “Surfacewave sup
pression in a doublenegative metamaterial grounded slab,” vol. 2, pp. 269–
272, 2003.
[75] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, P. Lampariello, G. Lovat,
and S. Paulotto, “The nature of radiation from leaky waves on single and
doublenegative metamaterial grounded slabs,” in Microwave Symposium
Digest, 2004 IEEE MTTS International, vol. 1, pp. 309–312, June 2004.
[76] P. Baccarelli, P. Burghignoli, F. Frezza, A. Galli, G. Lovat, and D. R. Jack
son, “Uniform analytical representation of the continuous spectrum excited
by dipole sources in a multilayer dielectric structure through weighted cylin
drical leaky waves,” IEEE Trans. Antennas Propagat., vol. 52, pp. 653–665,
Mar. 2004.
[77] G. GomezSantos, “Universal features of the time evolution of evanescent
modes in a lefthanded perfect lens,” Phys. Rev. Lett., vol. 90, Feb. 2003.
Paper 077401.
[78] P. R. Haddad and D. M. Pozar, “Anomalous mutual coupling between mi
crostrip antennas,” IEEE Trans. Antennas Propagat., vol. 42, pp. 1545–
1549, Nov. 1994.
[79] D. M. Pozar, Microwave Engineering. Wiley, New York, 1998.
[80] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions with
Formulas, Graphs, and Mathematical Tables. New York: Dover, 1964.
194
I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.
Assist. Prof. Dr. Vakur B. Ert¨rk(Supervisor) u
I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.
Prof. Dr. Ayhan Altınta¸ s
I certify that I have read this thesis and that in my opinion it is fully adequate, in scope and in quality, as a thesis for the degree of Master of Science.
¨ Assoc. Prof. Dr. Ozlem Aydın Civi ¸
Approved for the Institute of Engineering and Sciences:
Prof. Dr. Mehmet Baray Director of Institute of Engineering and Sciences
ii
ABSTRACT WAVE PROPAGATION IN METAMATERIAL STRUCTURES AND RETRIEVAL OF HOMOGENIZATION PARAMETERS
Erdin¸ Ircı c M.S. in Electrical and Electronics Engineering Supervisor: Assist. Prof. Dr. Vakur B. Ert¨rk u August 2007
Electromagnetic wave propagation in metamaterial structures (metamaterial slabs, metamaterial cylinders, metamaterial coated conducting cylinders etc.) are investigated. Scattered and transmitted electromagnetic ﬁelds by these structures due to electric line sources or plane wave illuminations are found. A generic formulation of these wave propagation problems is done, enabling any kind of metamaterial or conventional material to be used, having any sign combination of constitutive parameters and having any electric and/or magnetic losses. For one of these propagation problems i.e., metamaterial coated conducting cylinders illuminated normally with plane waves, achieving transparency and maximizing scattering are investigated thoroughly. It is found out that, rigorous derivation of transparency and resonance (scattering maximization) conditions for PEC core cylinder case under the subwavelength limitations yields the same conditions of two electrically small concentric layers of conjugately paired cylinders, given in the literature (when the inner core layer is also taken to the PEC limit). These transparency and resonance conditions are found to be heavily iii
dependent on the permittivity of the metamaterial coating (for TE polarization) and the ratio of coreshell radii. The relations between the permittivity of the coating and the ratio of coreshell radii are investigated for achieving transparency and scattering maximization. Numerical results show that these analytical relations are quite successful and work better when the cylindrical scatter is electrically very small. A novel homogenization method for the retrieval of eﬀective constitutive parameters of metamaterials is proposed and implemented. The method is based on the simple idea that the total reﬂection coeﬃcient from a ﬁnite metamaterial structure has to resemble the reﬂection from an homogeneous equivalent. While implementing the method, 1, 2, . . ., 20 unit cells of the same metamaterial structure are stacked and their reﬂection coeﬃcients are collected. The homogenization quality of the metamaterial is evaluated in terms of various factors, which showed that the method is very successful to retrieve the eﬀective constitutive parameters of the metamaterial. Finally, another method has been proposed for the retrieval of surface wave propagation constants on any periodic or nonperiodic grounded slab medium. As a preliminary, the method is applied to grounded dielectric slabs. The numerical results generally show good agreement with their theoretical counterparts.
Keywords:
Metamaterials, Wave propagation, Scattering, Transmission,
Metamaterial cylinders, Metamaterial coated conducting cylinders, Transparency, Resonance, Radar cross section, Homogenization, Parameter retrieval,
Surface waves, Grounded Slabs.
iv
Do¸. ters i¸aretli olarak e¸le¸tirilmi¸ silindirlerle aynı durumu uc ¨ s s s s verdi (i¸ silindir iletken sınırına g¨t¨r¨ld¨˘unde). s c Bu yayınım problemlerinden biri olan d¨zlem dalga ile dik aydınlatılmı¸ u s metamalzeme kaplı iletken silindirler. Cizgisel elektrik kaynaklarından ya da d¨zlem dalga aydınlatmalarından ¸ u dolayı bu yapılardan sa¸ılan ve bunlara iletilen elektromanyetik alanlar bulundu. Saydamlık ve rezonans (sa¸ılım azamile¸tirmesi) c c s durumlarının dalgaboyualtı sınırında t¨retilmesi. saydamlık ve sa¸ılımın azamile¸tirilmesi c s a¸ısından detaylıca incelendi. herhangi elektrik/manyetik kayba da s c sahip olabilecek ¸ekilde metamalzemeler ya da sıradan malzemeler i¸in yapıldı. literat¨rdeki aynı eksenli. Ert¨rk o c u A˘ustos 2007 g Metamalzeme yapılarda (metamalzeme tabakalar.¨ OZET METAMALZEME YAPILARDA DALGA YAYINIMI VE ˙ ˙ HOMOJENLESTIRME PARAMETRELERININ ELDE ¸ ˙ ˙ ˙ EDILMESI Erdin¸ Ircı c Elektrik ve Elektronik M¨hendisli˘i B¨l¨m¨ Y¨ksek Lisans u g ou u u Tez Y¨neticisi: Yar. eleku u triksel olarak k¨¸uk.) elektromanyetik dalga yayınımı incelendi. metamalzeme silindirler. Vakur B. metamalzeme kaplı iletken silindirler vb. c Bu dalga yayınım problemlerinin genel form¨lasyonu. ortam parametrelerinin u i¸aretlerinin herhangi kombinasyonu i¸in. Dr. c o u u ug ¨ Bu saydamlık ve rezonans durumlarının daha ¸ok metamalzeme kaplamanın c elektriksel ge¸irgenli˘ine (TE polarizasyonu i¸in) ve ¸ekirdekkaplama yarı¸ap c g c c c v .
Ba¸langı¸ olarak metod topraklanmı¸ dielektrik tabakalara c uu u s c s uygulandı. Radar kesit alanı. Metamalzeme silindirler. Metamalzemenin u homojenle¸tirme kalitesi de˘i¸ik etkenler cinsinden incelendi ve metodun metas gs malzemenin etkin ortam parametrelerinin elde edilmesi i¸in ¸ok ba¸arılı oldu˘u c c s g g¨z¨kt¨. Metod uygulanırken metamalzemenin 1. .oranına ba˘lı oldu˘u bulundu. Dalga yayınımı. c ˙ Iletim. Y¨zey dalgaları. Sayısal sonu¸lar bu analitik ili¸kilerin olduk¸a ba¸arılı oldu˘unu ve c s c s g silindirik sa¸ıcı elektriksel olarak ¸ok k¨¸ukken daha iyi ¸alı¸tı˘ını g¨sterdi. u s vi . Metamalzeme kaplı iletken silindirler. Sayısal sonu¸lar genel olarak teorik kar¸ılıklarıyla iyi uyum sergiledi. . o u u Son olarak. Saydamlık ve sa¸ılım azamile¸tirmesi i¸in. Metod. 2. Homojenle¸tirme. 20 unite ¨ h¨cresi art arda sıralandı ve yansıma katsayıları kaydedildi. kaplag g c s c manın elektrik ge¸irgenli˘i ile ¸ekirdekkaplama yarı¸ap oranı arasındaki ili¸kiler c g c c s incelendi. Rezonans. Topraklanmı¸ tabakalar. sonlu bir metamalzeme s uu u yapının toplam yansıma katsayısının homojen denginin yansımasına benzeyece˘i g ﬁkrine dayandırıldı. bir ba¸ka metod da periyodik olan ya da olmayan herhangi bir s topraklanmı¸ tabaka uzerindeki y¨zey dalga yayınım katsayılarının elde edilmesi s ¨ u i¸in ileri s¨r¨ld¨. c c uc ¨ c s g o Metamalzemelerin etkin ortam parametrelerinin elde edilmesi i¸in yeni bir hoc mojenle¸tirme metodu ileri s¨r¨ld¨ ve uygulandı. c s Anahtar Kelimeler: Metamalzemeler. . Sa¸ılım. Saydamlık.. s Parametre elde edimi.
Prof.ACKNOWLEDGMENTS I would like to express my sincere gratitude to my supervisor Asst. Ozlem Aydın s Civi from METU for being in my jury. ¨ I would like to thank Prof. Onur c Bakır. Prof. I would also like to thank The Scientiﬁc and Technological Research Council ¨ ˙ of Turkey (TUBITAK) for supporting me with the graduate scholarship during my study. M. encouragement and u support throughout the development of this thesis. friendship and support. Finally. ¸ ˙ s I would like to thank Prof. Chapter 4 of this thesis is merely realization of his ingenious ideas. He has always been a great mentor and teacher to me. Ayhan Altınta¸ and Assoc. encouragement. for their understanding. vii . Ir¸adi Aksun from Ko¸ University for allowing c us to collaborate in his research. whom I can’t all list c u g here. Ert¨rk for his invaluable guidance. Burak G¨ldo˘an and many others. suggestions. I would like to thank my family. Vakur B. Ayta¸ Alparslan. my friends Celal Alp Tun¸. reading the thesis and commenting on it.
. . . . . . .3. . . . .4 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 7 9 9 10 10 11 Inﬁnite Length Metamaterial Cylinder Near an Inﬁnite Length Electric Line Source: T M z Polarization . . . .2.2. . . . . . . . .2 Normal Incidence of Plane Waves on a Metamaterial Slab . . . . .2 2. . . . . . . . . . . . . .3 Introduction . . . . . . . . . . 2. . .2 2. . .3 Introduction . . . Problem Geometry . . . . . . . . . . . . . . Index of Refraction and Wave Impedance of Metamaterial Structures . . . . 13 13 13 14 viii . . Electric and Magnetic Fields . . . . . . . . . . 2.Contents 1 INTRODUCTION 1 2 Wave Propagation in Metamaterial Structures 2. .2. .3. . . Problem Geometry . . 2. . .2. . . Solution of Boundary Conditions . . . .1 2. . . . . . . .1 Wave Number. . . . . . . . .1 2. . . . . . . .3 2. . . . . . . .3. . . . . . . . . . . . . . . . . Electric Line Source and Incident Electric Field . .
. . . . . 24 25 21 21 21 22 22 23 23 24 ix . . . . . . . . . .9 Calculation of the Radiation Patterns . .1 2.4 2. . . . . . . . . . . . . . Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields . . . .3. .3.4 2. . .3 2. . . 14 15 15 16 17 18 19 2. . . . .3.4. 2. . . . . . . .8 Scattered and Transmitted Electric Fields . . 2. . . . Boundary Conditions for Magnetic Fields . . . . . . . . . . . . .4. . . . . .3. . . . . . .4. . . .3.5 2. . . .2 2. . . Uniform Plane Wave and Incident Electric Field . . .4. .4. . . . .5 2. . Boundary Conditions for Magnetic Fields . . . . . . . . . . . . . . . . . Scattered and Transmitted Magnetic Fields . . Problem Geometry . . . Scattered and Transmitted Electric Fields . . . . . . .9 Numerical Results .4. . . . . .6 2. . . . . .4. . . . . . Boundary Conditions for Electric Fields . . . . . . . .8 Introduction . . . . . . . . . . .10 Numerical Results 2. . . . . .3. . . . . . . . . 2. .7 2.4 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T M z Polarization .4. . . . .7 2. Incident. Scattered and Transmitted Magnetic Fields . . . . . .3. . .6 2. . . . . . . . . . . . . Incident. . . . . . . . . . . .2. . . . .4. . . . Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields . Boundary Conditions for Electric Fields . . . .
.6. . . . . . . . .6. . .6.1 2. . . . . . . . . . . . . 2. . . . . . . . . . . . . . .5. . . . Incident. . . . .6. . 2. . . . . . . . Electric Line Source and Incident Electric Field . . . . .2 2. . . . Problem Geometry . x 32 32 32 33 33 34 34 35 .3 2.3 2.6 2. . . . .5 2. . . . . . .6.6 Inﬁnite Length Metamaterial Coated Conducting Cylinder Near an Inﬁnite Length Electric Line Source: T M z Polarization . . .4 2. .2. . . . . . . . .2 2. . .6. . . . Scattered and Transmitted Electric Fields . . . . . . . . . . . . . . . . . . .5 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T E z Polarization . . .5. .9 Numerical Results . . . . .5. . . . . . . . Scattered and Transmitted Magnetic Fields . Problem Geometry . .4 2. .6. . . . . . . . . . . . . .5. . Scattered and Transmitted Electric Fields .5 2. . . . . . Incident. . . . . . Simultaneous Solution of the Boundary Conditions for Magnetic and Electric Fields . .8 Introduction . .7 Introduction . . . . . . . .6 2. . . . . Uniform Plane Wave and Incident Magnetic Field . . . . . 2. Boundary Conditions for Electric Fields . . . . . .5. . . . . . . . . . .5. . . . . . . . Scattered and Transmitted Magnetic Fields . . . . .1 2. . . . . Boundary Conditions for Magnetic Fields .5. . . . .5. .7 2.5. . Boundary Conditions for Electric Fields . . . . . . . . Boundary Conditions for Magnetic Fields . . . . . . 30 31 27 27 27 28 28 28 29 30 2. . .
.3 Introduction . . . . Boundary Conditions for Electric Fields .10 Numerical Results 2. . . .6 2. .9 Electric Line Source Inside the Metamaterial Coating . . . 45 46 40 40 40 41 43 43 44 44 2. . . . . . Scattered and Transmitted Electric Fields . 48 48 48 49 xi . . . . .2.7. . . .5 2. . . . .8. . . . . . . . 2. . . . . 2. . . . 35 36 38 2. .7. .8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields . . . . .9 Numerical Results . . . . . .8.7.7. . Problem Geometry . . . . . . . . . . . .1 2. . . .4 2. . . . . . . . Uniform Plane Wave and Incident Electric Field . . . . . . . Uniform Plane Wave and Incident Magnetic Field . . . . .7. . .6. . .7. . .7 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T M z Polarization . . . . . . . . . .6.8. . Incident. .6.1 2. . . 2. . . . . .2 2. .7 2.8 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T E z Polarization . . . . . . . . . . . .8 Introduction . . . . . . . .7. . . . . . . . . . . . . . . . . . . . . . . . . . . . Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields . . . . . . . . . Scattered and Transmitted Magnetic Fields . . . . . . . . . . .3 2. . . . Problem Geometry . . . . . . . . Boundary Conditions for Magnetic Fields . . . . . . . .7. . . .2 2. . . . . . . .7. . . . . . . . . 2. . .
9. . . . . .8.8. . Boundary Conditions for Electric Fields . Incident.2 Incident. . .10 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T E z Polarization . .5 2. .9. . . . . . . . . 2. . . . . . .4 φ Components of the Incident. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Boundary Conditions and Their Solution . .9. . . . . . . . . .9. . . . . . . 2. . . .2. . . . . . . 49 50 50 50 51 2. .10. . . . . . . . . . . .7 2. 2. . . . . . . . . .8. . . . . . . . .8. . . .3 Incident. . . . . . .9 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T M z Polarization . . . . . . . . 56 57 59 56 54 53 53 2. . . Scattered and Transmitted Magnetic Fields (z components) . . . . . 2. . . . . . . . . . .6 2. . . . .2 Introduction . . . . Scattered and Transmitted Electric Fields . 2. . . . Boundary Conditions for Magnetic Fields .8. Calculation of the Radar Cross Section .4 2. . .9. . . . . . . 2. . . . . . . . . . .9. Scattered and Transmitted Magnetic Fields (z components) . . . . . . . .10. . . .8 Scattered and Transmitted Magnetic Fields . . . . . Incident. . . . Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields . . . . . . . . . . . . . . . . . . . Scattered and Transmitted Electric and Magnetic Fields . . 62 62 62 xii . Scattered and Transmitted Electric Fields (z components) . . . . . . . . . . . . . . . . .1 2. . . .5 2.1 Introduction . . . . . . .
. . . . . . . .6 Calculation of the Radar Cross Section . .6 Calculation of the Radar Cross Section . . . . . . . . . . .2 Incident. . .10. .10. . . . . . . . . . .1 Introduction . . . . . . . .1 Introduction . . . . 2. . . . . . . . . . . . 2. 2. Scattered and Transmitted Electric and Magnetic Fields . . . . . .11. . . .3 Incident. . . . . . . 2. . Scattered and Transmitted Magnetic Fields (z components) . . . Scattered and Transmitted Magnetic and Electric Fields . .11. . . . . .5 Boundary Conditions and Their Solution . . . 2. . . . . . .10. . . . . . . . . . . . .4 φ Components of the Incident. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12. . . . 2.12. .11. .12 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T E z Polarization . . . . . . . . Scattered and Transmitted Magnetic Fields (z components) . .5 Boundary Conditions and Their Solution . . .4 φ Components of the Incident. 2. . . . . . . . . . . . .11. .2. 72 72 72 67 68 71 67 67 66 66 63 63 65 62 xiii . Scattered and Transmitted Electric Fields (z components) . .11. . . . . . . . . . . . . . . . . . . . . . . . . . . 2. 2. . . . 2. . . . . . . .2 Incident. .3 Incident. . .11 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T M z Polarization . . . . 2.10. . . . . 2. . . . . .11. . . . . . . . . 2. . . . . . . . Scattered and Transmitted Electric Fields (z components) . .
. . . . . . . . . . . .3 4. 128 xiv . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 74 77 73 3 Achieving Transparency and Maximizing Scattering with Metamaterial Coated Conducting Cylinders 3. . .2 Introduction . . . . . . . 101 Homogenization of Metamaterials . 121 Conclusion . . . 78 78 79 84 87 4 Retrieval of Homogenization Parameters 4. . . . . . . . . . . . . . . . . . . . Scattered and Transmitted Magnetic and Electric Fields . . . . . .1. . . . . . . . . . . . . . . . . . . . . . . . . . . .1. . . 2.2 4. . . . . . .12. . . . .12. . .1 Homogenization of Metamaterial Structures and Retrieval of Ef 101 fective Constitutive Parameters . . . . .2. Numerical Results and Discussion . . .1 4. . .4 Introduction . . . . .4 4. . . . . . . . . . . . . . . .3 3. . . .1. . . . .5 Boundary Conditions and Their Solution . . . Scattered and Transmitted Electric Fields (z components) .4 φ Components of the Incident. . . . . . . . 127 Retrieval of Surface Wave Propagation Constants on a Grounded Dielectric Slab . . . . . . . . . 2. . . . . .1. .12. . . . . . . . . . . . . . . . Transparency Condition . . . . . . . . . . . . . . . . . . . . 2. . . .12. . . . . . . . Resonance (Scattering Maximization) Condition . . . . . 101 4. . . . . . . . . . . . . . . . . . . . . . . .2 3.3 Incident. . . . . . . .6 Calculation of the Radar Cross Section . . 103 Numerical Results . . . . . . . . . . . .1 3. . . . . . . . .
2. . . . . 128 The TwoStep Method . .1 4. . . . . . . . . . . . . . . . . . . .4. . . . . .2. . . . . . .2 4. . . . . . . . . . . . 129 Implementation: . . . . 146 Conclusions . . . . . . . . . . . .2. . .3 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 4. .2. . . 159 5 CONCLUSIONS 160 APPENDICES 163 A Bessel Functions 163 B Derivation of the φ Components of Electric and Magnetic Fields: T M z Polarization 166 C Derivation of the Transparency Condition 175 D Derivation of the Resonance Condition 180 xv . 144 Numerical Results . . . . . . . .2. . . .5 Introduction . . . . . . .
4 Uniform plane wave incident on a metamaterial cylinder: T M z Polarization. . . µr = 2 20 2. . . . . . . . . . . . . . . . . . µr = −1. . . . .7 Metamaterial coated conducting cylinder near an electric line source (Cross section view). . . . . 2. . . . µr = −1. .9 Plane wave normally incident on a metamaterial coated conducting cylinder. . . . . 2. . 2. . .8 Magnitude of the electric ﬁeld inside and outside the cylinder. .1 2. . . . . . (a) Side view. . . . . . . (c)(d) εr = −2. . . . . . .List of Figures 2. . . . µr = −1. . . µr = −2. . . . (e)(f) εr = 2. . . (e)(f) εr = 2. (c)(d) εr = −2. . . µr = −2. . .6 Uniform plane wave incident on a metamaterial cylinder: T E z Polarization. . . . . . . . . 2. . . . . (e)(f) εr = 2. . (c)(d) εr = −2. . .5 Magnitude of the electric ﬁeld inside and outside the cylinder. (a)(b) εr = −1. . . . . . µr = 2 39 2. (a)(b) εr = −1. . . .3 Magnitude of the electric ﬁeld inside and outside the cylinder. .2 Uniform plane wave normally incident on a metamaterial slab. . . . . . . . . . . . . (a)(b) εr = −1. (b) Top view. 40 32 27 21 13 10 xvi . . . . . . . . . . . . . . . . . µr = −2. . . . . . . . . . . . . Metamaterial cylinder near an electric line source. µr = 2 26 2. . . . . . . . .
. . . . . . . . . 2. . . . . . . . . . . . . . . . 92 xvii . . 90 3. . . . . versus the corecoating ratio for coatings with diﬀerent constitutive parameters. .12 Uniform plane wave obliquely incident on a metamaterial cylinder: T M z Polarization. . (e)(h) b = λ0 /10. f = 1GHz). . versus the corecoating ratio for coatings with diﬀerent constitutive parameters. . . . . . µr = 2 47 2. . . . . . . . 2. . .3 Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T E z polarization case. . . . . . . . . . . . . . . . . µr = −1. . . (c)(d) εr = −2. . Dashed line shows the uncoated PEC case. . (e)(f) εr = 2. The outer radius of the coating is selected as (a)(c) b = λ0 /2. . . .2 Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T E z polarization case. . . . . . Diamond marks show the DPS and DNG coating cases in [1]. . . . . . . . . . .13 Longitudinal and transverse components of the incident and transmitted ﬁelds. . . . . . . The outer radius of the coating is selected as (a)(d) b = λ0 /100. b = 70mm.1 Normalized monostatic echo width of a metamaterial coated PEC cylinder (a = 50mm. 2. . . . . . . .10 Magnitude of the electric ﬁeld inside and outside the cylinder. . . (d)(f) b = λ0 . . . µr = −2. . . . . . with radius a. . .2. . . . . . . 87 3. . . . . . . . . . . . . with radius a. . . . . . . . . . Dashed line shows the uncoated PEC case. 66 55 53 48 3. . . . . .14 Uniform plane wave obliquely incident on a metamaterial coated conducting cylinder: T M z Polarization. . . . (a)(b) εr = −1. . . .11 Plane wave normally incident on a metamaterial coated conducting cylinder. . .
. The outer radius of the coating is selected as b = λ0 /100. The outer radius of the coating is selected as (a)(d) b = λ0 /20. . The outer radius of the coating is selected as (a)(d) b = λ0 /100. . . . b = λ0 /100). . . . . (e)(h) b = λ0 /10. (e)(h) b = λ0 /50. . . . .5 Normalized monostatic echo width of an ENG coated PEC cylinder for the T E z polarization case. . Dashed line shows the uncoated PEC case. versus the coating permeability µc for diﬀerent corecoating ratios. . 95 3. versus the corecoating ratio for coatings with diﬀerent constitutive parameters. . . .3.. . versus the corecoating ratio for coatings with diﬀerent constitutive parameters. 97 3. s i Hz + Hz ) outside the PEC cylinder when there is (a) No coating.6 Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T M z polarization case. (b) DPS coating. 98 3. .8 Normalized bistatic echo widths for (a) DPS coated (b) ENG coated PEC cylinder for the T E z polarization case. . 96 3. 93 3. . . . . . . . Plane wave illumination is along the +xaxis. . . . . . .9 Contour plots of axial component of the total magnetic ﬁeld (i. 99 xviii .e. . . . . . . . . . Outer boundaries of the coatings are shown by dashed lines (a = λ0 /200.7 Eﬀects of ohmic losses on normalized monostatic echo width for (a) DPS [transparency] (b) ENG [Scattering maximization] cases. The angle of incidence is φ0 = 0◦ . The outer radius of the coating is b = λ0 /100 and the coating permittivity is εc = ε0 . Dashed line shows the uncoated PEC case. with radius a. . . . (c) ENG coating. . .4 Normalized monostatic echo width of an ENG coated PEC cylinder for the T E z polarization case. with radius a. . . The outer radius of the coating is selected as b = λ0 /100. . . . . . .
15GHz frequency band. . . (b) two layered media. . . . . . . .2 4. . 130 4. . . . . . . . . . . . . . . . . . . . . . . .11 Magnitudes of Eﬁeld inside and outside the metamaterial medium and its homogeneous equivalent at (a) f = 10. . 106 Problem geometry (crosssection view. 131 xix .3 4. . .4 4.7 4. . . . . . . . .3. . . . . . . . . . . 109 Fresnel reﬂection at (a) three layered media. . . . obtained from the metamaterial and its homogeneous equivalent. . . . . . . . . . . . . . . . . 122 4. . . . . . .1 4. . . . . . . .8GHz.5 4. . . . . . . . .6 4.12 Geometry of a grounded dielectric slab. . . (a) εr . . . . . . .0GHz. .13 Geometry of the rectangular narrow patch and excitation. . 104 Direction of E and H ﬁelds for a unit cell. . . . . Nz = 1).8 Metamaterial unit cell. . . . . . for Nz = 3). . .112 Eﬀective homogenization parameters of the metamaterial over the 5GHz . z (f = 10GHz. . . . . . . . . .10 Normalized monostatic echo widths for (a) DPS coated (b) ENG coated PEC cylinder for the T E z polarization. . . . (b) µr . . . . . . . . . . . . . . . . . .14 Entire problem geometry in HFSS. . . . . . . . . . . . 104 Direction of E and H ﬁelds for a unit cell. . 128 4. (b) f = 15. . . . . . . . . oblique incidence case. . .10 Magnitude of Eﬁeld inside and outside the metamaterial medium and its homogeneous equivalent at f = 5GHz. 100 4. . . . . . . The outer radius of the coating is selected as b = λ0 /100. 107 Ex  vs. . 126 4. . . . . . 123 4. . frequency. 105 Alignment of unit cells inside the PECPMC waveguide. . . . .9 S11 vs. . . . . . . . 125 4. . . . . . . . . .
4.15 Magnitudes of complex function y(t) and its N uniform samples y[k]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.16 Magnitudes of complex function y(t − t0 ) and its N uniform samples y[k − k0 ]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 4.17 Magnitudes of √ ρEx (ρ) and its N uniform samples y[k − k0 ]. . . . 136
4.18 Problem geometry for the Eline case. . . . . . . . . . . . . . . . . 145 4.19 Problem geometry for the Hline case. . . . . . . . . . . . . . . . . 146 4.20 Comparison of GPOF approximation with HFSS data and its Extrapolation . (f = 30GHz, λ0 = 1cm, th = 0.1λ0 , εr = 2.55, ρstart = 5λ0 , ρend = 8λ0 , N = 101) . . . . . . . . . . . . . . . . . . 155 4.21 Comparison of GPOF approximation with HFSS data and its Extrapolation . (f = 30GHz, λ0 = 1cm, th = 0.15λ0 , εr = 2.55, ρstart = 5λ0 , ρend = 8λ0 , N = 101) . . . . . . . . . . . . . . . . . . 156 4.22 Comparison of GPOF approximation with HFSS data and its Extrapolation . (f = 30GHz, λ0 = 1cm, th = 0.19λ0 , εr = 2.55, ρstart = 5λ0 , ρend = 8λ0 , N = 101) . . . . . . . . . . . . . . . . . . 157 4.23 Comparison of GPOF approximation with HFSS data and its Extrapolation . (f = 30GHz, λ0 = 1cm, th = 0.25λ0 , εr = 2.55, ρstart = 4λ0 , ρend = 8λ0 , N = 51) . . . . . . . . . . . . . . . . . . 158
xx
List of Tables
2.1 Arguments of µ, ε, k and η for Diﬀerent Types of Metamaterials . 9
3.1 3.2
Desired and Obtained γ for Achieving Transparency Using (3.10) Desired and Obtained γ for Achieving Transparency Using (3.11)
82 83
4.1 4.2
Parameters of the GPOF approximation . . . . . . . . . . . . . . 124 Space Wave and Surface Wave Characteristics in the E and Hplanes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.3
Surface wave propagation constants retrieved from (a) Eline, (b) Hline. (f = 30GHz, λ0 = 1cm, th = 0.1λ0 , εr = 2.55) . . . . . 150
4.4
Surface wave propagation constants retrieved from (a) Eline, (b) Hline. (f = 30GHz, λ0 = 1cm, th = 0.15λ0 , εr = 2.55) . . . . 151
4.5
Surface wave propagation constants retrieved from (a) Eline, (b) Hline. (f = 30GHz, λ0 = 1cm, th = 0.19λ0 , εr = 2.55) . . . . 152
4.6
Surface wave propagation constants retrieved from (a) Eline, (b) Hline. (f = 30GHz, λ0 = 1cm, th = 0.25λ0 , εr = 2.55) . . . . 153
4.7
Surface wave propagation constants retrieved from (a) Eline, (b) Hline. (f = 30GHz, λ0 = 1cm, th = 0.05λ0 , εr = 2.55) . . . . 154
xxi
Dedicated to my family.
Chapter 1 INTRODUCTION
Metamaterials are artiﬁcially engineered materials which can have negative eﬀective electric permittivity and/or negative eﬀective magnetic permeability. Diﬀerent from conventional materials, which have both positive electric permittivity and positive magnetic permeability [i.e., double positive (DPS)], metamaterials show diﬀerent electromagnetic and optical properties. For instance, when electric permittivity and magnetic permeability of the material are both negative [i.e., double negative (DNG)], negative refraction happens and direction of phase velocity is reversed. DNG metamaterials are also called Left Handed Materials (LHM) because electric ﬁeld, magnetic ﬁeld and the direction of phase velocity form a left handed coordinate system for these materials. On the other hand, when only one of the constitutive parameters of the metamaterial is negative [i.e., single negative (SNG)] evanescent waves appear. In Chapter 2, electromagnetic scattering and transmission due to line sources or plane waves from diﬀerent metamaterial structures is investigated. The metamaterial structures are chosen from simple canonical geometries, such as metamaterial slabs, metamaterial cylinders and metamaterial coated conducting cylinders, which have exact eigenfunction solutions.
1
On the other hand.g. which otherwise cannot be easily accomplished with conventional materials. While some of these studies are based on utilization of nonlinear metamaterial structures.. metamaterial layers have been proposed to enhance the power radiated by electrically small antennas [15–17]. PEC cylinders together with their metamaterial coatings) are electrically small. Recently. the analytical relations between the ratio of corecoating radii and the constitutive parameters of the metamaterial coating are derived.e. some of them rely on pairing slabs. simple (i. which can have any combination of signs of constitutive parameters. In Chapter 3. have been investigated by many researchers [1–7]. or can have any electric and/or magnetic losses. isotropic and linear) metamaterial coatings are used..After a complex analysis.e.. These analytical relations are based on subwavelength approximations and they are valid especially when the cylindrical scatterers (i. metamaterials are being investigated for many possible utilizations in diﬀerent scientiﬁc and engineering applications. The 2 . the correct complex branches for the wave number and wave impedance of a metamaterial medium are selected. resonant structures aimed at increasing the electromagnetic intensities. For achieving transparency or maximizing scattering. This choice of complex branches is found to be valid for all kinds of materials. pairing/coating DPS materials with DNG metamaterials or munegative (MNG) metamaterials with epsilonnegative (ENG) metamaterials). For both transparency and scattering maximization scenarios. homogeneous. stored or radiated power levels have also been studied extensively [7–14]. and in the limit achieving transparency and building cloaking structures. the transparency and resonance conditions for cylindrical structures are extended to the case where the core cylinder is particularly PEC. spheres or cylinders with their electromagnetic conjugates (e. Due to their aforementioned exceptional properties. Similarly. reducing scattering from various structures.
usually transmission and reﬂection properties of only one unit cell of the metamaterial is taken into account. with its all intermediate steps. or ﬁeld averaging. As a remedy to these inadequate methods. are made up of metallic inclusions. another branch of these research eﬀorts is now focused on retrieval of the eﬀective constitutive parameters of metamaterials. which cause very strong electric and magnetic resonances. However. These methods are intrinsically unreliable since the unit cells. during these attempts for homogenization of metamaterials. If a metamaterial slab can be successfully homogenized. with added phase delays. in Chapter 4 we present a novel method for the homogenization and parameter retrieval of metamaterials. The process of obtaining this homogeneous equivalent. therefore cannot represent the whole metamaterial structure correctly. metamaterials currently produced are inhomogeneous. the 3 . Meanwhile.numerical simulations have showed the existence of transparency and resonance conditions in good agreement with the analytical expectations. or in other words. However. anisotropic and highly dispersive. Since total reﬂection from a homogeneous slab is the sum of a direct reﬂection term and other multiple reﬂection terms due to the waves bouncing inside the slab. obtaining homogeneous equivalents for essentially inhomogeneous metamaterials. there are many research eﬀorts to obtain homogeneous and isotropic metamaterials. one loses the valuable information of periodicity of unit cells and their mutual interactions. its reﬂection characteristics would mimic those of a homogeneous slab. While using only one or two unit cells of the metamaterial. The homogenization processes present in the literature [18–22] are mainly based on utilization of transmission and reﬂection characteristics of the metamaterial structures. is called homogenization. Although Chapter 3 is based on the assumptions that the metamaterial coating is homogeneous and isotropic. which form the metamaterial.
can be approximated as a summation of complex exponentials. At the present.total reﬂection from the metamaterial slab can be written as a sum of exponentials. utilization of diﬀerent number of unit cells will yield diﬀerent reﬂection results. for large lateral distances. then. it becomes possible to obtain the constitutive parameters of a homogeneous medium using the reﬂection coeﬃcients of the metamaterial medium. conclusions of the thesis are drawn. and numerical results have shown good agreement to the theory. In Appendix B. made up of diﬀerent number of unit cells. In our method. the method is applied to a dielectric slab. from which one can deduce how many surface wave modes are present and what their propagation constants are. ﬁeld distributions inside and outside the metamaterial) is compared with that of the homogeneous equivalent. Appendix A contains some properties of Bessel functions. we aim to present another method for the retrieval of surface wave propagation constants on any periodic or nonperiodic grounded slab medium. The method is basically based on the diﬀerence in spread factors of space and surface waves propagating on the surface of the slab. Derivations of the transparency and resonance conditions of Chapter 3 are given in Appendix C 4 . Again in Chapter 4. After the constitutive parameters are retrieved. for which the theoretical surface wave propagation constants are well known. its reﬂection and transmission properties.9 are derived from their z components. Our numerical results show very good agreement between these two. Also since the phase delays of the multiple reﬂections inside the slab are dependent on the thickness of the slab. The electric ﬁeld data. φ components of the magnetic and electric ﬁelds of Section 2. Since space wave contribution of the total electric ﬁeld on the surface of the slab decays faster. Therefore.g. the electromagnetic behavior of the metamaterial slab (e. multiplying the ﬁeld data with the proper power of the lateral distance mainly leaves the surface wave contribution. In Chapter 5.. we have used 1 to 20 unit cells.
and Appendix D. respectively. 5 . Throughout this thesis. an ejωt time dependence is assumed and suppressed.
Chapter 2 Wave Propagation in Metamaterial Structures
In this chapter, electromagnetic wave propagation in diﬀerent metamaterial structures is investigated. The metamaterial geometries are chosen from simple canonical geometries, such that an exact analytical eigenfunction solution can be obtained. Metamaterials are artiﬁcial materials which can have negative eﬀective electric permittivity (εef f ) and/or negative eﬀective magnetic permeability (µef f ). The signs of the aforementioned eﬀective complex constitutive parameters are based on the signs of their real parts, whereas their imaginary parts indicate the presence of electric or magnetic losses, respectively. Therefore, metamaterials form four groups, depending on their constitutive parameters: • Double Positive (DPS): Re{ε} > 0, Re{µ} > 0
• Double Negative (DNG): Re{ε} < 0, Re{µ} < 0 • Mu Negative (MNG): Re{ε} > 0, Re{µ} < 0
• Epsilon Negative (ENG): Re{ε} < 0, Re{µ} > 0 6
DNG metamaterials are also called Left Handed Materials (LHM) due to their unique electromagnetic/optical properties like negative refraction, negative phase velocity and negative Doppler shift, which follow a left hand rule system. MNG and ENG metamaterials are also called Single Negative (SNG) materials, because of the obvious fact that they have either negative eﬀective magnetic permeability or negative eﬀective electric permittivity, respectively.
2.1
Wave Number, Index of Refraction and Wave Impedance of Metamaterial Structures
Without loss of generality, the wave number, index of refraction and wave impedance of a medium are given as √ k = ω µε, n= √ µr εr , µ , ε (2.1) (2.2) (2.3)
η=
respectively, where ω = 2πf is the angular frequency, µr = µ/µ0 and εr = ε/ε0 are the relative constitutive parameters. The square roots which appear in (2.1)(2.3) create controversy, especially when DNG materials are considered. Since both constitutive parameters are complex quantities with their real parts being negative, the wave number, index of refraction and wave impedance of the medium heavily depend on which branch of the complex roots is selected. This controversy appeared in the scientiﬁc community after the idea of perfect lens [23] and discussions focused on validity of negative refraction and negative phase velocity [24].
7
The complex electric permittivity and the complex magnetic permeability of a metamaterial medium can be expressed in polar form, respectively, as ε = εejφε , µ = µejφµ . (2.4) (2.5)
Similarly, the wave number and the wave impedance of the metamaterial coating can be written as √ k = ω µε = kejφk , η= respectively, where k = ω η = with 1 φk = (φµ + φε ), 2 1 φη = (φµ − φε ). 2 (2.10) (2.11) µε, µ/ε, (2.8) (2.9) µ/ε = ηejφη , (2.6) (2.7)
The choice of branches for the square roots in (2.10)(2.11) is based on causality in a linear dispersive medium, the wave directions associated with reﬂection and transmission from the interfaces and the direction of electromagnetic power ﬂow. This choice is given and examined in details in [25] for DNG metamaterials, ﬁrst introducing inﬁnitesimal electric and magnetic losses (as in the case of metamaterials approximated by Drude and Lorentz medium models [23, 25, 26]) and then deciding on which complex branch gives the physically correct solution. A similar analysis for DPS, MNG and ENG metamaterials [11] show that, the choice of branches for the square roots given in (2.10)(2.11) still remains valid for these metamaterials. With the assumed ejωt time dependence in this thesis, and considering only passive media, the arguments of µ, ε, k and η for diﬀerent types of metamaterials are tabulated in Table 2.1.
8
Table 2.1: Arguments of µ, ε, k and η for Diﬀerent Types of Metamaterials φµ DPS DNG MNG ENG −π, 0 2 −π, − π 2 −π, − π 2 −π, 0 2 φε −π, 0 2 −π, − π 2 −π, 0 2 −π, − π 2 φk −π, 0 2 −π, − π 2 − 3π , − π 4 4 − 3π , − π 4 4 φη −π, π 4 4 −π, π 4 4 −π, 0 2 0, π 2
Examination of Table 2.1 shows that for lossless DPS medium, wave number is real and positive. For lossless DNG medium, wave number is real and negative. For lossless DPS and DNG media, wave impedance is real and positive. For lossless MNG and ENG media, the wave number is negative and imaginary, which shows the presence of evanescent waves. Remark: It is worthwhile to mention that when any of the constitutive parameters of a metamaterial medium is a negative real number, −π should be selected as its argument instead of π, as shown in Table 2.1. This becomes important when intrinsic functions in a programming environment are directly used (e.g., ANGLE, ATAN2).
2.2
Normal Incidence of Plane Waves on a Metamaterial Slab
2.2.1
Introduction
Let us assume that a T EM z plane wave is traveling in the +z direction. An inﬁnite length metamaterial slab of thickness d is placed between the z = 0 and z = d planes in free space, without loss of generality. Here we will investigate the reﬂection and transmission properties of the metamaterial slab as well as
9
2.3 Electric and Magnetic Fields The total electric and magnetic ﬁelds in Medium 1 are + − E1 = ax E1 e−jk0 z + E1 ejk0 z . The problem geometry is depicted in Fig. 2.2. 2.1: Uniform plane wave normally incident on a metamaterial slab. .2.13) H1 = ay + − E1 −jk0 z E1 jk0 z e − e . The incident electric ﬁeld is assumed to be in the +x direction and the incident magnetic ﬁeld is in +y direction.1. η0 η0 √ respectively. (2.12) (2.the waves traveling inside the metamaterial slab. where k0 = ω µ0 ε0 and η0 = 10 µ0 /ε0 .2 Problem Geometry Figure 2.
20) (2.21) we get: + − + − −E1 + E2 + E2 = E1 . 2. e−jkd + ejkd − e−jk0 d + E2 − E − E3 = 0. H3 = ay η0 (2.4 Solution of Boundary Conditions Boundary Conditions at z = 0: + − + − E1 + E1 = E2 + E2 . where k0 and η0 are the same as in Medium 1. + − E2 −jkz E2 jkz e − e .2.(2.16) (2. η η (2.17) respectively.The total electric and magnetic ﬁelds in Medium 2 are + − E2 = ax E2 e−jkz + E2 ejkz .18) .18) (2. − + + E2 −jkd E2 jkd E3 −jk0 d e − e = e .25) η0 + η0 − + E2 − E2 = E1 .22) (2. where k = ω µε and η = µ/ε.21) Rearranging equations (2.14) (2. − E1 + (2. η0 η0 η η (2.15) H2 = ay √ respectively. η η η0 (2.19) Boundary Conditions at z = d: + − + E2 e−jkd + E2 ejkd = E3 e−jk0 d .23) (2. η η 2 η0 11 . η η + − + e−jkd E2 + ejkd E2 − e−jk0 d E3 = 0.24) (2. The electric and magnetic ﬁelds in Medium 3 are + E3 = ax E3 e−jk0 z . + E3 −jk0 z e . + − + − E1 E1 E2 E2 − = − .
32)(2.26) Using Symbolic Math Toolbox of MATLAB.1.30) + E2 = − E2 = + E3 = η(η + η0 )ejkd E +. DPS.27)(2. 2 + η 2 ) sin kd 2ηη0 cos kd + j(η 0 (2. 2ζ cos kd + j(ζ 2 + 1) sin kd 1 (2. equations (2.35) are valid for all four types of metamaterials (i.27)(2. DNG.35) + E2 = − E2 = + E3 = Note that.30) can be reduced to: − E1 = j(ζ 2 − 1) sin kd E +. εr (2. the solutions (2..31) and using relation (2.34) (2. MNG and ENG) provided that k and η are calculated as in Section 2. 2 2ηη0 cos kd + j(η 2 + η0 ) sin kd 1 η(η0 − η)e−jkd E +. the solution to this system of equations can be found as: − E1 2 j(η 2 − η0 ) sin kd + = E1 . 2ζ cos kd + j(ζ 2 + 1) sin kd 1 (ζ − ζ 2 )e−jkd E +. 0 0 (2.e.27) (2.31). 2ζ cos kd + j(ζ 2 + 1) sin kd 1 (ζ 2 + ζ)ejkd E +.32) (2.30) or (2. 2 2ηη0 cos kd + j(η 2 + η0 ) sin kd 1 Deﬁning ζ= η = η0 µr .29) (2.which can also be written in matrix form as follows: − E1 −1 1 1 0 + η0 η0 −η 0 1 E2 η = 0 e−jkd ejkd −e−jk0 d E − 2 0 e−jkd − ejkd − e−jk0 d E + η η η0 3 + E1 + E1 .33) (2. 2ζ cos kd + j(ζ 2 + 1) sin kd 1 2ζejk0 d E +. 2 2ηη0 cos kd + j(η 2 + η0 ) sin kd 1 2ηη0 ejk0 d E +. 12 .28) (2.
2. 13 .1 Introduction An inﬁnite line of constant electric current is placed in the vicinity of a circular metamaterial cylinder of inﬁnite length. (b) Top view. (a) Side view.2.3 Inﬁnite Length Metamaterial Cylinder Near an Inﬁnite Length Electric Line Source: T M z Polarization 2. The scattering and transmission by the metamaterial cylinder is examined for T M z polarization. The problem geometry is given in Fig.2 Problem Geometry Figure 2.2: Metamaterial cylinder near an electric line source.3. 2. 2.3.
For the transmitted ﬁeld. which 0 (2) k2 I in fact could be included in cn and/or dn . (2.40) For the scattered ﬁeld.38).3 Electric Line Source and Incident Electric Field For the line source of constant electric current. respectively as s Ez 2 k0 Ie =− 4ωε0 +∞ (2) cn Hn (k0 ρ)ejn(φ−φ ) n=−∞ +∞ a≤ρ≤ρ .4 Scattered and Transmitted Electric Fields Similar to the incident ﬁeld expressions in (2. (2. ρ≥ρ. in Fig. the electric ﬁeld generated everywhere by the source in the absence of the cylinder is given as [27] i Ez = − 2 k0 Ie (2) H (k0 ¯ − ρ ). 14 . The − 4ωεe terms are just for convenience in calculations.37) and (2.3. we will deﬁne the scattered and transmitted electric ﬁelds in series expansion form.3.38) 2. The ﬁelds are 2π periodic in φ.37) i Ez ρ≥ρ. Using the addition theorem for Hankel functions [28]. (2.36) can be written in the series expansion form as follows [27]: i Ez k 2 Ie =− 0 4ωε0 k 2 Ie =− 0 4ωε0 +∞ (2) Jn (k0 ρ)Hn (k0 ρ )ejn(φ−φ ) n=−∞ +∞ (2) Jn (k0 ρ )Hn (k0 ρ)ejn(φ−φ ) n=−∞ ρ≤ρ.39) t Ez k 2 Ie =− 0 4ωε0 dn Jn (kρ)ejn(φ−φ ) n=−∞ 0 ≤ ρ ≤ a. 2. our deﬁnition should include Hn (k0 ρ) term which represents +ρ wave propagation. our deﬁnition should include Jn (kρ) term which represents a standing wave and also avoids a blow up at ρ = 0 (due to Yn ). ρ ¯ 4ωε0 0 (2. Ie .2.2. so ejn(φ−φ ) term is inserted to show this and to be in accordance with the incident ﬁeld expressions and also 0 for convenience. (2. (2.36) which we will refer as the incident electric ﬁeld.
48) (2. jωµ ρ ∂φ 1 ∂Ez Hφ = .44) 2. t s i Ez (ρ = a) + Ez (ρ = a) = Ez (ρ = a). Therefore.49) E = az Ez .49). we are only interested in equation (2.5 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous at the surface of the cylinder.43) dn = Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) . n=−∞ (2.6 Incident.41) k 2 Ie − 0 4ωε0 =− dn Jn (ka)ejn(φ−φ ) . 15 .2. H=− 1 jωµ aρ ∂Ez 1 ∂Ez − aφ ρ ∂φ ∂ρ Hρ = − 1 1 ∂Ez .3.45) (2.47) (2.42) (2) (2) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) = dn Jn (ka). +∞ (2) (2) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) ejn(φ−φ ) n=−∞ 2 k0 Ie 4ωε0 +∞ (2. Jn (ka) (2) (2) (2.46) . jωµ ∂ρ Since Hφ is the only component of the magnetic ﬁelds we will utilize in boundary conditions. Fields Scattered and Transmitted Magnetic The radial and tangential components of the magnetic ﬁelds are derived from the electric ﬁelds using the Maxwell’s equation: H=− 1 jωµ × E. (2. due to the boundary conditions.3. (2. (2.
7 Boundary Conditions for Magnetic Fields The tangential components of the magnetic ﬁelds are continuous at the surface of the cylinder due to the boundary conditions.54) 2.52) s Hφ k 2 Ie 1 (2) =− 0 k0 cn Hn (k0 ρ)ejn(φ−φ ) 4ωε0 jωµ0 n=−∞ +∞ t Hφ a≤ρ≤ρ . 4ωε0 jωµ n=−∞ +∞ +∞ (2.51) i Hφ ρ≥ρ. (2. (2. ∂F (βρ) ∂(βρ) ∂F (βρ) ∂F (βρ) = =β .50) Utilizing (2. ∂ρ ∂ρ ∂(βρ) ∂(βρ) (2. Then. (2.One important point where attention must be paid is the partial derivative of Ez with respect to ρ. t s i Hφ (ρ = a) + Hφ (ρ = a) = Hφ (ρ = a).3. their derivatives should be taken with respect to the entire argument of the corresponding Bessel and Hankel functions.55) − 2 k0 Ie 1 (2) (2) k0 J (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) ejn(φ−φ ) 4ωε0 jωµ0 n=−∞ n 2 k0 Ie 1 =− k dn Jn (ka)ejn(φ−φ ) . the tangential components of the magnetic ﬁelds are obtained as follows: i Hφ k 2 Ie 1 (2) k0 Jn (k0 ρ)Hn (k0 ρ )ejn(φ−φ ) =− 0 4ωε0 jωµ0 n=−∞ k 2 Ie 1 (2) Jn (k0 ρ )Hn (k0 ρ)ejn(φ−φ ) =− 0 k0 4ωε0 jωµ0 n=−∞ +∞ +∞ +∞ ρ≤ρ.56) 16 . (2.53) k 2 Ie 1 =− 0 k dn Jn (kρ)ejn(φ−φ ) 4ωε0 jωµ n=−∞ 0 ≤ ρ ≤ a. Therefore.49) and (2. (2.50). Since in our Ez deﬁnitions we have the Bessel and Hankel functions. Let F (βρ) be a function representing the Bessel and Hankel functions. and also keeping in mind that the derivatives are all with respect to the entire arguments. ρ≥ρ.
µ0 µ Expressing k0 k = µ µ0 where ζ = √ (2.60). which are derived from the boundary conditions for the electric and magnetic ﬁelds.64) 17 .44) and (2.63) (2) (2) cn Jn (ka)Hn (k0 a) − cn ζJn (ka)Hn (k0 a) (2) (2) = ζJn (ka)Jn (k0 a)Hn (k0 ρ ) − Jn (ka)Jn (k0 a)Hn (k0 ρ ).58) µr /εr as previously deﬁned in (2. we get: (2) (2) ζ Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) = dn Jn (ka). Our next step will be to equate these equations: (2) (2) ζ Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) = dn = .60) 2.59) dn = ζ Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) Jn (ka) (2) (2) . (2.58) in (2. respectively.31). µr µ0 ζ (2. (2. (2. Jn (ka) Jn (ka) (2.57). (2. (2.57) µr εr k0 = µr µ0 k0 1 εr = .k k0 (2) (2) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) = dn Jn (ka).61) (2) (2) (2) (2) Jn (ka) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) (2) (2) = ζJn (ka) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) .3.62) (2) (2) Jn (ka)Jn (k0 a)Hn (k0 ρ ) + cn Jn (ka)Hn (k0 a) (2) (2) = ζJn (ka)Jn (k0 a)Hn (k0 ρ ) + cn ζJn (ka)Hn (k0 a).8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields Now we have two equations for dn : (2. and substituting (2.
44) dn = or from (2. Jn (ka) (2) (2) (2.60) dn = ζ Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) Jn (ka) (2) (2) Jn (k0 a)Hn (k0 ρ ) + cn Hn (k0 a) .38) and (2.70) 18 . (2. the incident.69) Now. boundary conditions should be written using (2. scattered and transmitted electric and magnetic ﬁelds can be calculated using (2.51) are used. the following large argument approximation for Hankel functions of the second kind is used: (2) lim Hn (k0 ρ) = k0 ρ→∞ 2 −j[k0 ρ−π/4−n(π/2)] e .54).40) and (2.66) cn = − (2) ζJn (ka)Hn (k0 a) (2. (2) (2) (2) (2) .(2) (2) cn Jn (ka)Hn (k0 a) − ζJn (ka)Hn (k0 a) (2) (2) = ζJn (ka)Jn (k0 a)Hn (k0 ρ ) − Jn (ka)Jn (k0 a)Hn (k0 ρ ).3. respectively. 2. since the electric line source is placed outside the metamaterial cylinder.68) .65) cn = ζJn (ka)Jn (k0 a)Hn (k0 ρ ) − Jn (ka)Jn (k0 a)Hn (k0 ρ ) Jn (ka)Hn (k0 a) − ζJn (ka)Hn (k0 a) ζJn (ka)Jn (k0 a) − Jn (ka)Jn (k0 a) (2) Jn (ka)Hn (k0 a) (2) Hn (k0 ρ ). (2. If the line source is placed inside the cylinder.67) where dn can be found from (2.51)(2.37)(2. (2.52).37) and (2.9 Calculation of the Radiation Patterns To calculate the radiation patterns. when applying the boundary conditions for electric and magnetic ﬁelds at ρ = a. Remark: Note that. respectively. πk0 ρ (2. (2.
πk0 ρ n=−∞ (2. φ) = 16η0 ω 2 ε2 π 0 +∞ 2 [Jn (k0 ρ ) + cn ] e n=−∞ jn(φ−φ +π/2) . φ) = lim = k0 ρ→∞ 2η0 16η0 ω 2 ε2 πρ 0 +∞ 2 [Jn (k0 ρ ) + cn ] e n=−∞ jn(φ−φ +π/2) .72) +∞ k0 ρ→∞ lim r Ez (ρ. a = λ0 . ρ = 1. φ) = − 2 k0 Ie 4ωε0 +∞ (2) [Jn (k0 ρ ) + cn ] Hn (k0 ρ)ejn(φ−φ ) .01m.70). focusing towards the line source and inside the metamaterial cylinder is noticed. φ = 0◦ . These unique focusing properties of DNG metamaterials are mainly results of negative refraction. For DNG cases. φ) = Ez (ρ. 19 . n=−∞ (2. lim r Ez (ρ.3 (a). φ)2 k0 Ie Wrad (ρ.71) Using (2. this focusing occurs on the surface of the cylinder. In Fig.3 shows the magnitude of electric ﬁeld for diﬀerent choices of constitutive parameters when f = 30GHz. 2. φ) k0 ρ→∞ k 2 Ie =− 0 4ωε0 +∞ [Jn (k0 ρ ) + cn ] n=−∞ 2 −j[k0 ρ−π/4−n(π/2)] jn(φ−φ ) e e . λ0 = 0.73) The radiation density is: r 3 2 Ez (ρ.The total electric ﬁeld for ρ > ρ can be written as: r i s Ez (ρ. 2. πk0 ρ (2.3.74) The radiation intensity is: 3 2 k0 Ie U (φ) = ρWrad (ρ. φ) + Ez (ρ. (2. (2. φ) 2 k0 Ie =− 4ωε0 2 −j(k0 ρ−π/4) e [Jn (k0 ρ ) + cn ] ejn(φ−φ +π/2) .10 Numerical Results Fig.5λ0 .75) 2.
01 0.015 0. µr = −2.015 4 (d) 14 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 12 0.005 0.02 x 10 12 0.005 0.005 8 y(m) 0 6 −0.005 4 −0.015 −0.02 −0.02 (c) Magnitude of the Electric Field Inside and Outside the Cylinder 0.015 12 0.005 4 −0.Magnitude of the Electric Field Inside and Outside the Cylinder 0.01 −0.02 −0.005 0 x(m) 0.005 0 x(m) 0.005 0 x(m) 0. (c)(d) εr = −2.015 0.01 10 0.02 −0.02 −0.015 0.015 0.02 4 −0.02 −0.01 −0.02 0 −0.02 −0.02 (a) Magnitude of the Electric Field Inside and Outside the Cylinder 0.005 0 x(m) 0. (a)(b) εr = −1.01  E  (V/m) 8 y(m) 6 4 −0.01 0.015 2 2 −0.015 −0.015 2 −0. µr = 2 20 .01 −0. µr = −1.02 −0.015 −0.005 0.02 −0.01 −0.01 −0. (e)(f) εr = 2.005 0 x(m) 0.01 0.02 x 10 12 4 (b) 14 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 0.005 0.005 12 10 0 6  E  (V/m) 8 y(m) 6 −0.005 8 0 6 −0.02 (e) (f) Figure 2.01 0.3: Magnitude of the electric ﬁeld inside and outside the cylinder.015 4 2 −0.015 −0.02 0 −0.01 −0.01 10 10 0.015 0.01 8 0.015 0.005 0.005 10  E  (V/m) 5 0 −0.01 0.015 −0.01 2 −0.015 10 0.02 −0.01 −0.005 0.01 0.015 −0.02 x 10 14 4 15 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 0.005 0 x(m) 0.
2. electric ﬁeld is directed along the +z axis and magnetic ﬁeld is directed along the +y axis.4 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T M z Polarization 2.2.4: Uniform plane wave incident on a metamaterial cylinder: T M z Polarization. For the −x propagation direction and T M z polarization.1 Introduction A uniform plane wave is normally incident on a metamaterial cylinder of inﬁnite length. 2.4. The plane wave travels in the −x direction. 21 . The problem geometry is given in Fig.4.2 Problem Geometry Figure 2. We will examine here the scattering and transmission by the cylinder in the case the polarization of the plane wave is T M z .4.
respectively. For the transmitted ﬁeld. The j n terms are just for convenience in calculations. (2. (2. so ejnφ term is inserted to show this and to be in accordance with the incident ﬁeld expressions and also for convenience. Therefore the electric ﬁeld can be written as [27] i Ez = E0 ejk0 x = E0 ejk0 ρ cos φ . (2. (2.77). our deﬁnition should include Jn (kρ) term which represents a standing wave and also avoids a blow up at ρ = 0 (due to Yn ).76) can be written in the series expansion form as follows [27]: +∞ i Ez = E0 n=−∞ j n Jn (k0 ρ)ejnφ ρ ≥ a.4.77) 2. (2. as +∞ s Ez = E0 n=−∞ +∞ (2) j n cn Hn (k0 ρ)ejnφ ρ ≥ a.4 Scattered and Transmitted Electric Fields Similar to the incident ﬁeld expression in (2. which in fact could be included in cn and/or dn .3 Uniform Plane Wave and Incident Electric Field Let us assume that a T M z polarized uniform plane wave is traveling in the −x direction.76) By wave transformations and utilizing orthogonality condition [27. This means electric ﬁeld is directed along the +z axis and magnetic ﬁeld is directed along the +y axis.28]. our deﬁnition should include Hn (k0 ρ) term which represents +ρ wave propagation. we will deﬁne the scattered and transmitted electric ﬁelds in series expansion form. (2) 22 .4.78) t Ez = E0 n=−∞ j n dn Jn (kρ)ejnφ 0 ≤ ρ ≤ a.79) For the scattered ﬁeld.2. The ﬁelds are 2π periodic in φ.
83) 2.2. (2.6 Incident.82) dn = Jn (k0 a) + cn Hn (k0 a) .86) The derivatives in (2.50).85) t Hφ 0 ≤ ρ ≤ a.4.4. Fields Scattered and Transmitted Magnetic Utilizing (2. (2) Jn (k0 a) + cn Hn (k0 a) = dn Jn (ka). Therefore. 23 . (2. the tangential components of the magnetic ﬁelds are obtained as i Hφ 1 = E0 k0 j n Jn (k0 ρ)ejnφ jωµ0 n=−∞ +∞ +∞ ρ ≥ a.49) and (2. Jn (ka) (2) (2.197)(2.81) E0 n=−∞ j n Jn (k0 a) + (2) cn Hn (k0 a) e jnφ = E0 n=−∞ j n dn Jn (ka)ejnφ . (2. i s t Ez (ρ = a) + Ez (ρ = a) = Ez (ρ = a).80) (2.199) are again with respect to the entire arguments.84) s Hφ 1 (2) = E0 k0 j n cn Hn (k0 ρ)ejnφ jωµ0 n=−∞ 1 k j n dn Jn (kρ)ejnφ = E0 jωµ n=−∞ +∞ ρ ≥ a. +∞ +∞ (2. (2.5 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous at the surface of the cylinder due to the boundary conditions.
88) k k0 (2) Jn (k0 a) + cn Hn (k0 a) = dn Jn (ka). Our next step will be to equate these equations: (2) ζ Jn (k0 a) + cn Hn (k0 a) Jn (k0 a) + cn Hn (k0 a) .4.87) E0 1 1 (2) k0 j n Jn (k0 a) + cn Hn (k0 a) ejnφ = E0 k j n dn Jn (ka)ejnφ .92) (2) (2) Jn (ka) Jn (k0 a) + cn Hn (k0 a) = ζJn (ka) Jn (k0 a) + cn Hn (k0 a) . (2) ζ Jn (k0 a) + cn Hn (k0 a) = dn Jn (ka).2.94) 24 .89) µ0 µ Using (2. (2.8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields Now we have two equations for dn : (2.93) (2) (2) Jn (ka)Jn (k0 a) + cn Jn (ka)Hn (k0 a) = ζJn (ka)Jn (k0 a) + cn ζJn (ka)Hn (k0 a). (2.90) dn = ζ Jn (k0 a) + cn Hn (k0 a) Jn (ka) (2) .58) in (2. = dn = Jn (ka) Jn (ka) (2) (2. which are derived from the boundary conditions for the electric and magnetic ﬁelds.91). respectively.231).4. +∞ +∞ (2. Therefore. (2.91) 2. jωµ0 n=−∞ jωµ n=−∞ (2. (2.83) and (2.7 Boundary Conditions for Magnetic Fields The tangential components of the magnetic ﬁelds are continuous at the surface of the cylinder due to the boundary conditions. i s t Hφ (ρ = a) + Hφ (ρ = a) = Hφ (ρ = a). (2.
5 (b). Jn (ka) or dn = Jn (ka) ζ Jn (k0 a) + cn Hn (k0 a) (2) (2) (2) (2) .99) 2. In Fig. Finally in Fig.01m. 2.97) (2. 25 . (2. Both foci are at the other side of the cylinder (w.t plane wave illumination) as predicted for a DPS dielectric lens.96) cn = where ζJn (ka)Jn (k0 a) − Jn (ka)Jn (k0 a) Jn (ka)Hn (k0 a) − ζJn (ka)Hn (k0 a) Jn (k0 a) + cn Hn (k0 a) dn = . (2. 2.95) (2) (2) cn Jn (ka)Hn (k0 a) − ζJn (ka)Hn (k0 a) = ζJn (ka)Jn (k0 a) − Jn (ka)Jn (k0 a). λ0 = 0.5 shows the numerical results for f = 30GHz. 2. a = λ0 .r. 2.9 Numerical Results Fig.4.(2) (2) cn Jn (ka)Hn (k0 a) − cn ζJn (ka)Hn (k0 a) = ζJn (ka)Jn (k0 a) − Jn (ka)Jn (k0 a).5 (c) there is one focus inside the cylinder and another outside. (2.5 (a). (2. In Fig. there are three foci close to the interface and inside the metamaterial cylinder.98) . there is one dominant focus inside the cylinder. while the other two diminish.
005 0.015 0.015 −0. µr = −2.005 0.015 3.02 0.01 2 0.005 1 −0.015 0.5 2 1.5 1.01 −0.01 0.005 0.5 0.02 −0.5 4 0.01 −0.5 −0.02 0 −0.5 2.005 0 x(m) 0.015 0.02 (e) (f) Figure 2.015 1 0.01 0.02 −0.5 3.01 0.5 −0.005 0.02 3 0.005 0 x(m) 0.015 0.005 0 x(m) 0. (a)(b) εr = −1.01 0.015 0.02 −0.015 0.005 0 x(m) 0.02 −0.5 −0.5 0.5 1 −0.01 −0.015 0.01 2.5 0.5 3  E  (V/m) 2.5 −0.01 1 1. (c)(d) εr = −2.5: Magnitude of the electric ﬁeld inside and outside the cylinder.015 −0.5 −0.02 −0.005 3.02 −0.015 −0.005 0.5 1.02 4 (b) Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 4.01 −0.01 −0.5 0.02 3.005 3 0 2 −0.02 −0.02 −0.01 −0.015 2.005 0 x(m) 0.5 y(m) 2.005 0 x(m) 0.5 2 −0.Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) Magnitude of the Electric Field Inside and Outside the Cylinder 0.015 3 0.5 −0. (e)(f) εr = 2.005 2  E  (V/m) y(m) 0 1.005 0. µr = 2 26 . µr = −1.015 −0.01 0.005 2 y(m) 0 1.02 0 −0.015 0.5 −0.02 −0.5 (d) Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 4 0.02 (c) Magnitude of the Electric Field Inside and Outside the Cylinder 0.015 −0.02 (a) Magnitude of the Electric Field Inside and Outside the Cylinder 0.01 0.01 1 1 −0.5 −0.005  E  (V/m) 2.015 −0.01 3 0.01 0.
27 . We will examine here the scattering and transmission by the cylinder in the case the polarization of the plane wave is T E z . as in Section 2.4. traveling in the −x direction.5 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T E z Polarization 2. For the −x propagation direction and T E z polarization.1 Introduction A uniform plane wave is normally incident on a metamaterial cylinder of inﬁnite length.2 Problem Geometry Figure 2. magnetic ﬁeld is directed along the +z axis and electric ﬁeld is directed along the −y axis.6: Uniform plane wave incident on a metamaterial cylinder: T E z Polarization.5.6. The problem geometry is depicted in Fig.5. 2.2. 2.
(2. which means the magnetic ﬁeld is directed along the +z axis and the electric ﬁeld is directed along the −y axis. (2.104) (2. we will deﬁne the scattered and transmitted magnetic ﬁelds in series expansion form respectively as follows: +∞ s Hz = H0 n=−∞ +∞ (2) j n cn Hn (k0 ρ)ejnφ ρ ≥ a.28].5. i s t Hz (ρ = a) + Hz () = Hz (ρ = a).5.4 Scattered and Transmitted Magnetic Fields Similar to the incident ﬁeld expression in (2. the magnetic ﬁeld can be written as [27] i Hz = H0 ejk0 x = H0 ejk0 ρ cos φ . +∞ +∞ (2. (2.102) t Hz = H0 n=−∞ j n dn Jn (kρ)ejnφ 0 ≤ ρ ≤ a.3 Uniform Plane Wave and Incident Magnetic Field Let us assume that a T E z polarized uniform plane wave is traveling in the −x direction.100) By wave transformations and utilizing orthogonality condition [27. (2.5.105) H0 n=−∞ j n Jn (k0 a) + (2) cn Hn (k0 a) e jnφ = H0 n=−∞ j n dn Jn (ka)ejnφ .100) can be written in series expansion form as [27] +∞ i Hz = H0 n=−∞ j n Jn (k0 ρ)ejnφ ρ ≥ a. Therefore. Hence.5 Boundary Conditions for Magnetic Fields The tangential components of the magnetic ﬁelds are continuous at the surface of the cylinder due to the boundary conditions.2. 28 . (2.101) 2.101).103) 2.
106) dn = Jn (k0 a) + cn Hn (k0 a) . we are only interested in equation (2. Scattered and Transmitted Electric Fields The radial and tangential components of the electric ﬁelds are derived from the magnetic ﬁelds using the Maxwell’s equation: E= 1 jωε × H.114) t Eφ 0 ≤ ρ ≤ a. (2. (2. (2. E= 1 jωε aρ 1 ∂Hz ∂Hz − aφ ρ ∂φ ∂ρ 1 1 ∂Hz . jωε ρ ∂φ 1 ∂Hz . jωε ∂ρ Eρ = Eφ = − Since Eφ is the only component of the electric ﬁelds we will utilize in boundary conditions. (2.(2) Jn (k0 a) + cn Hn (k0 a) = dn Jn (ka). (2.108) (2.5. Jn (ka) (2) (2. tangential components of the electric ﬁelds are obtained as i Eφ −1 k0 j n Jn (k0 ρ)ejnφ = H0 jωε0 n=−∞ +∞ +∞ ρ ≥ a.112) H = az Hz .110) (2. (2.107) 2.112).113) s Eφ −1 (2) = H0 k0 j n cn Hn (k0 ρ)ejnφ jωε0 n=−∞ −1 = H0 k j n dn Jn (kρ)ejnφ jωε n=−∞ +∞ ρ ≥ a.112). Utilizing (2.109) .111) (2.115) 29 .6 Incident.
5.118). (2) (2) ζJn (ka)Jn (k0 a) + cn ζJn (ka)Hn (k0 a) = Jn (ka)Jn (k0 a) + cn Jn (ka)Hn (k0 a). k k0 = ε ε0 √ µr ε r k0 = εr ε0 k0 µr = ζ. +∞ +∞ (2. (2. εr ε0 (2.117) k k0 (2) Jn (k0 a) + cn Hn (k0 a) = dn Jn (ka).5. respectively. For this reason. (2) Jn (k0 a) + cn Hn (k0 a) = dn ζJn (ka). jωε0 n=−∞ jωε n=−∞ (2.124) 30 . (2.121) 2.123) (2) (2) ζJn (ka) Jn (k0 a) + cn Hn (k0 a) = Jn (ka) Jn (k0 a) + cn Hn (k0 a) . = n Jn (ka) ζJn (ka) (2) (2) (2.118) ε0 ε Expressing.7 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous at the surface of the cylinder due to the boundary conditions.2.116) −1 −1 (2) H0 k0 j n Jn (k0 a) + cn Hn (k0 a) ejnφ = H0 k j n dn Jn (ka)ejnφ .119) and substituting (2.119) in (2. Our next step will be to equate these equations: dn = J (k0 a) + cn Hn (k0 a) Jn (k0 a) + cn Hn (k0 a) .8 Simultaneous Solution of the Boundary Conditions for Magnetic and Electric Fields Now we have two equations for dn : (2.120) J (k0 a) + cn Hn (k0 a) dn = n .122) (2. ζJn (ka) (2) (2.107) and (2.121). which are derived from the boundary conditions for the magnetic and electric ﬁelds. i s t Eφ (ρ = a) + Eφ (ρ = a) = Eφ (ρ = a). (2.
interchanging µr and εr .126) cn = where Jn (ka)Jn (k0 a) − ζJn (ka)Jn (k0 a) ζJn (ka)Hn (k0 a) − Jn (ka)Hn (k0 a) Jn (k0 a) + cn Hn (k0 a) dn = . (2.127) (2.128) (2.129) 2. 2.9 Numerical Results Using duality.5.125) (2) (2) cn ζJn (ka)Hn (k0 a) − Jn (ka)Hn (k0 a) = Jn (ka)Jn (k0 a) − ζJn (ka)Jn (k0 a). (2. ζJn (ka) (2) (2) (2) (2) . Jn (ka) or J (k0 a) + cn Hn (k0 a) dn = n . the same results in Fig. 31 .(2) (2) cn ζJn (ka)Hn (k0 a) − cn Jn (ka)Hn (k0 a) = Jn (ka)Jn (k0 a) − ζJn (ka)Jn (k0 a). (2.5 can be obtained (for magnitude of the magnetic ﬁeld).
7.6. 2.7: Metamaterial coated conducting cylinder near an electric line source (Cross section view).6. The scattering and transmission by the metamaterial coated conducting cylinder is examined for T M z polarization.2 Problem Geometry Figure 2.2. 32 .1 Introduction An inﬁnite line of constant electric current is placed in the vicinity of an inﬁnite length metamaterial coated conducting cylinder.6 Inﬁnite Length Metamaterial Coated Conducting Cylinder Near an Inﬁnite Length Electric Line Source: T M z Polarization 2. 2. The problem geometry is shown in Fig.
133) t Ez k 2 Ie =− 0 4ωε0 [an Jn (kρ) + bn Yn (kρ)] ejn(φ−φ0 ) .6.4 Scattered and Transmitted Electric Fields Similar to the incident ﬁeld expressions in (2. in Fig. 33 . we will deﬁne the scattered and transmitted electric ﬁelds in series expansion form. By the addition theorem for Hankel functions [28]. ρ ¯ 4ωε0 0 (2. which in fact could be included in an and/or bn and/or cn .7. The ﬁelds are 2π periodic in φ.132). n=−∞ (2.2. so ejn(φ−φ0 ) term is inserted to show this and to be in accordance 0 with the incident ﬁeld expressions and also for convenience. The − 4ωεe terms are 0 k2 I just for convenience in calculations.6. 2.134) For the scattered ﬁeld.130) can be written in the series expansion form as [27] i Ez = − 2 k0 Ie 4ωε0 +∞ (2) Jn (k0 ρ)Hn (k0 ρ )ejn(φ−φ ) n=−∞ +∞ (2) Jn (k0 ρ )Hn (k0 ρ)ejn(φ−φ ) n=−∞ ρ≤ρ. (2.131) i Ez k 2 Ie =− 0 4ωε0 ρ≥ρ.3 Electric Line Source and Incident Electric Field For the line source of constant electric current. n=−∞ (2) (2. Ie . as s Ez k 2 Ie =− 0 4ωε0 +∞ +∞ (2) cn Hn (k0 ρ)ejn(φ−φ0 ) . our deﬁnition should include Jn (kρ) and Yn (kρ) terms which represent standing waves. our deﬁnition should include Hn (k0 ρ) term which represents +ρ wave propagation. respectively.131) and (2.132) 2. the electric ﬁeld generated everywhere by the source in the absence of the cylinder is given as [27] i Ez = − 2 k0 Ie (2) H (k0 ¯ − ρ ).130) which is our incident electric ﬁeld. (2. For the transmitted ﬁeld. (2.
(2. on the inner surface of the metamaterial coating (i. on the conducting cylinder surface) tangential electric ﬁeld should vanish. due to the boundary conditions.6 Incident.139) (2) (2) Jn (k0 b)Hn (k0 ρ ) + cn Hn (k0 b) = an Jn (kb) + bn Yn (kb).135) = 0.50). (2.49) and (2.5 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous on the outer surface of the metamaterial coating.140) 2..6.138) k 2 Ie − 0 4ωε0 +∞ (2) (2) Jn (k0 b)Hn (k0 ρ ) + cn Hn (k0 b) ejn(φ−φ0 ) n=−∞ 2 k0 Ie 4ωε0 +∞ =− [an Jn (kb) + bn Yn (kb)] ejn(φ−φ0 ) . Fields Scattered and Transmitted Magnetic Utilizing (2.6. t Ez (ρ = a) = 0.141) k 2 Ie 1 (2) =− 0 k0 cn Hn (k0 ρ)ejn(φ−φ0 ) .142) .e. the tangential components of the magnetic ﬁelds are obtained as i Hφ k 2 Ie 1 (2) k0 J (k0 ρ)Hn (k0 ρ )ejn(φ−φ0 ) .2. =− 0 4ωε0 jωµ0 n=−∞ n s Hφ +∞ (2. (2. 4ωε0 jωµ0 n=−∞ 34 +∞ (2. (2.137) i s t Ez (ρ = b) + Ez (ρ = b) = Ez (ρ = b). Also.136) an Jn (ka) + bn Yn (ka) = 0. n=−∞ (2. ρ=a k 2 Ie − 0 4ωε0 +∞ [an Jn (kρ) + bn Yn (kρ)] ejn(φ−φ0 ) n=−∞ (2. Therefore.
6.146) (2. (2) (2) Jn (kb)an + Yn (kb)bn − ζHn (k0 b)cn = ζJn (k0 b)Hn (k0 ρ ). t i s Hφ (ρ = b) + Hφ (ρ = b) = Hφ (ρ = b).147) 2.147) we get: Jn (ka)an + Yn (ka)bn = 0.6. 35 . µ0 µ Substituting (2.143) 2. (2.8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields Now we have three unknowns and three equations. (2. +∞ +∞ (2. 4ωε0 jωµ n=−∞ k0 k (2) (2) Jn (k0 b)Hn (k0 ρ ) + cn Hn (k0 b) = [an Jn (kb) + bn Yn (kb)] .146). 4ωε0 jωµ n=−∞ +∞ (2.150) (2) (2) Jn (kb)an + Yn (kb)bn − Hn (k0 b)cn = Jn (k0 b)Hn (k0 ρ ).t Hφ k 2 Ie 1 =− 0 k [an Jn (kρ) + bn Yn (kρ)] ejn(φ−φ0 ) .140) and (2. due to the boundary conditions.149) (2.7 Boundary Conditions for Magnetic Fields Tangential components of the magnetic ﬁelds are continuous on the outer surface of the metamaterial coating. Rearranging equations (2. (2) (2) ζJn (k0 b)Hn (k0 ρ ) + cn ζHn (k0 b) = an Jn (kb) + bn Yn (kb).58) in (2.145) (2. Therefore.148) (2.137). (2.144) k 2 Ie 1 (2) (2) − 0 k0 J (k0 b)Hn (k0 ρ ) + cn Hn (k0 b) ejn(φ−φ0 ) 4ωε0 jωµ0 n=−∞ n k 2 Ie 1 =− 0 k [an Jn (kb) + bn Yn (kb)] ejn(φ−φ0 ) .
152) bn = (2) Hn (k0 ρ ). (2.155) (2. the formulation given up to here has to be modiﬁed.151) Using Symbolic Math Toolbox of MATLAB.153) cn = where (2.6. (2) D = ζHn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] (2) −Hn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] .154) N = Jn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] −ζJn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] . This is mainly due to the electric 36 . D n (2) (2) (2) (2) (2) Hn (k0 ρ ). the solution to this system of equations can be found as: an = ζYn (ka) Hn (k0 b)Jn (k0 b) − Hn (k0 b)Jn (k0 b) D ζJn (ka) Jn (k0 b)Hn (k0 b) − Jn (k0 b)Hn (k0 b) D N (2) H (k0 ρ ). (2.which can also be written in matrix form as: 0 0 Jn (ka) Yn (ka) an (2) (2) Jn (kb) Yn (kb) −Hn (k0 b) bn = Jn (k0 b)Hn (k0 ρ ) (2) (2) Jn (kb) Yn (kb) −ζHn (k0 b) cn ζJn (k0 b)Hn (k0 ρ ) . (2. πx (2.156) Using the following Wronskian will further simplify an and bn : (2) (2) Jn (x)Hn (x) − Jn (x)Hn (x) = −2j . (2.9 Electric Line Source Inside the Metamaterial Coating When the electric line source is placed inside the metamaterial coating.157) 2.
160) respectively.159) (2.137).168) 37 . D B1 + B2 . (2.164) (2) (2) (2) A1 = Hn (k0 b)Yn (kb) − ζHn (k0 b)Yn (kb) Jn (ka)Hn (kρ ). (2.140) and (2. (2. (2.147) have to be modiﬁed as (2) Jn (ka)Hn (kρ ) + an Jn (ka) + bn Yn (ka) = 0. The system of equations can 0 Jn (ka) Yn (ka) an (2) Jn (kb) Yn (kb) −Hn (k0 b) bn (2) cn Jn (kb) Yn (kb) −ζHn (k0 b) be written in matrix form as (2) −Jn (ka)Hn (kρ ) (2) = −Jn (kρ )Hn (kb) . D (2.ﬁeld deﬁnition of the electric line source. (2. Utilizing the previous procedure. (2) (2) Jn (kρ )Hn (kb) + an Jn (kb) + bn Yn (kb) = cn Hn (k0 b).167) (2) (2) (2) (2) B2 = Hn (k0 b)Hn (kb) − ζHn (k0 b)Hn (kb) Jn (ka)Jn (kρ ).158) (2.163) cn = where C1 + C2 + C3 . D (2.165) (2) (2) (2) (2) A2 = ζHn (kb)Hn (k0 b) − Hn (kb)Hn (k0 b) Yn (ka)Jn (kρ ).166) (2) (2) (2) B1 = ζJn (kb)Hn (k0 b) − Jn (kb)Hn (k0 b) Jn (ka)Hn (kρ ). (2. (2. the solution to this system of equations can be found as an = A1 + A2 .162) bn = (2. it can be seen that (2.161) (2) −Jn (kρ )Hn (kb) Using Symbolic Math Toolbox of MATLAB. (2) (2) Jn (kρ )Hn (kb) + an Jn (kb) + bn Yn (kb) = cn ζHn (k0 b).
2. λ0 = 0.172) 2.(2) C1 = [Jn (kb)Yn (kb) − Jn (kb)Yn (kb)] Jn (ka)Hn (kρ ). focus point moves inside the cylinder. 2.8 (c) for the DPS case no focusing can be observed.5λ0 . 2. a = 0. In Fig. (2.170) (2) C3 = [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] Jn (kρ )Hn (kb). (2.01m. In Fig. 38 . ρ = 1. (2.8 (b).8 (a) there is a strong focus on the outer surface of the metamaterial. b = λ0 . Finally in Fig. φ = 0◦ .5λ0 . (2.10 Numerical Results Some of the numerical results are shown in Fig.8 for f = 30GHz. 2.169) (2) C2 = [Yn (ka)Jn (kb) − Yn (kb)Jn (ka)] Jn (kρ )Hn (kb).6.171) (2) D = ζHn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] (2) −Hn (k0 b) [Jn (ka)Yn (kb) − Jn (kb)Yn (ka)] .
02 −0.015 4 (b) 14 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 12 0.01 −0.005 0 x(m) 0.005 4 −0.005 0. µr = −1.005 8 y(m) 0 6 −0.005 7 y(m) 0 6 5 −0. µr = 2 39 .02 (e) (f) Figure 2.02 −0.02 0 −0.01 9 8 0.015 12 0.015 0.02 x 10 11 0.015 2 −0.005 0 x(m) 0.015 0.005 0 x(m) 0.02 (c) Magnitude of the Electric Field Inside and Outside the Cylinder 0.02 −0.01 −0.02 −0.005 0.02 x 10 12 0.015 0. (c)(d) εr = −2.01 0.015 −0.02 x 10 14 4 15 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 0.01 0.005 10  E  (V/m) 5 0 −0.01 0.01 −0.02 −0.015 10 0. (e)(f) εr = 2.005 0.015 0.01 −0.015 2 2 −0.01 0.005 0.02 −0.015 −0.01 3 2 −0.02 −0.01 0.005 8 0 6 −0.015 −0.8: Magnitude of the electric ﬁeld inside and outside the cylinder.015 −0.015 0.005 0 x(m) 0.02 −0.01 0.015 −0.01 10 10 0.015 0.Magnitude of the Electric Field Inside and Outside the Cylinder 0.01 10 0. µr = −2.01 4 −0.005 0.015 1 −0.02 0 −0.01  E  (V/m) 8 y(m) 6 4 −0.02 −0.005 0.005 0 x(m) 0.01 −0.02 4 (d) 12 x 10 4 Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 10 8  E  (V/m) 6 4 2 −0.01 −0. (a)(b) εr = −1.005 0 x(m) 0.02 (a) Magnitude of the Electric Field Inside and Outside the Cylinder 0.005 4 −0.015 −0.
µ0 φ0 Plane Wave Figure 2. In the numerical results of this section.9.2.7 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T M z Polarization 2. 40 .2 Problem Geometry y ρ σ=∞ φ a b x εc .4. The plane wave travels in the direction which makes an angle φ0 with the +x axis. the angle of incidence φ0 is selected as π.9: Plane wave normally incident on a metamaterial coated conducting cylinder. which is the case we have investigated in Section 2.7. The problem geometry is depicted in Fig. 2. We will examine here the scattering and transmission by the metamaterial coated conducting cylinder in the case the polarization of the plane wave is T M z .1 Introduction A uniform plane wave is normally incident on a metamaterial coated conducting cylinder of inﬁnite length.7. 2. µc ε0 . This corresponds to a plane wave traveling in the −x direction.
Referring to Figure 2. (2. (2. (2. (2.177) Dropping ‘E0 ’s and interchanging the integration and summation.175) The plane wave can be represented by an inﬁnite sum of cylindrical wave functions: i Ez +∞ = E0 e −jk0 ρ cos(φ−φ0 ) = E0 n=−∞ an Jn (k0 ρ)ejnφ . Electric ﬁeld is directed along the +z axis. (2. Our next step is to ﬁnd the coeﬃcients an .9 the electric ﬁeld can be written as i Ez = E0 e−jk0 (x cos φ0 +y sin φ0 ) . Multiplying both sides of (2. (2. and integrating from 0 to 2π. where m is an integer.179) .3 Uniform Plane Wave and Incident Electric Field Let us assume that a T M z polarized uniform plane wave is traveling in the direction which makes an angle φ0 with the +x axis. 2π E0 0 e −j(k0 ρ cos(φ−φ0 )+mφ) 2π +∞ dφ = E0 0 n=−∞ an Jn (k0 ρ)ej(n−m)φ dφ. 2π ej(n−m)φ dφ = 0.7. = E0 e−jk0 ρ cos(φ−φ0 ) . y = ρ sin φ.2.174) = E0 e−jk0 ρ(cos φ cos φ0 +sin φ sin φ0 ) .176) by ejmφ .173) where x = ρ cos φ .178) Utilizing the orthogonality condition of 2π. i Ez = E0 e−jk0 (ρ cos φ cos φ0 +ρ sin φ sin φ0 ) . (2.176) since it must be 2π periodic in φ and ﬁnite at ρ = 0. 0 41 n=m n=m . we have 2π 0 +∞ e−j(k0 ρ cos(φ−φ0 )+mφ) dφ = n=−∞ 2π an Jn (k0 ρ) 0 ej(n−m)φ dφ. Therefore.
the right hand side of (2. (2.182) Since J−m (x) = (−1)m Jm (x).188) 42 . the left side of (2.186) (2.178) can be written as 2π 0 e−j(k0 ρ cos(φ−φ0 )+mφ) dφ = e−jmφ0 2πj −m J−m (−k0 ρ). +∞ (2. = e−jmφ0 2πj −m Jm (k0 ρ).182) can be written as 2π 0 (2.185) reduces (2. (2.183) (2.185) Using (2. n=−∞ = E0 (2. and Jm (−x) = (−1)m Jm (x).180) and (2. (2.178) reduces to +∞ 2π an Jn (k0 ρ) n=−∞ 0 ej(n−m)φ dφ = 2πam Jm (k0 ρ). Therefore (2. (2.184) e−j(k0 ρ cos(φ−φ0 )+mφ) dφ = e−jmφ0 2πj −m J−m (−k0 ρ). j −n Jn (k0 ρ)ejn(φ−φ0 ) . Thus am = j −m e−jmφ0 .187) = E0 n=−∞ +∞ an Jn (k0 ρ)ejnφ . (2.176) can be written as i Ez = E0 e−jk0 ρ cos(φ−φ0 ) .178) to e−jmφ0 2πj −m Jm (k0 ρ) = 2πam Jm (k0 ρ).180) Using the integral of 2π 0 ej(z cos φ+nφ) dφ = 2πj n Jn (z).181) and by a simple transformation ϕ = φ − φ0 .
192) an Jn (kc a) + bn Yn (kc a) = 0.e.194) E0 n=−∞ (2) j −n Jn (k0 b) + cn Hn (k0 b) ejn(φ−φ0 ) +∞ = E0 n=−∞ j −n [an Jn (kc b) + bn Yn (kc b)] ejn(φ−φ0 ) .196) 43 . t Ez (ρ = a) = 0. (2..189) t Ez = E0 n=−∞ j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) . on the conducting cylinder surface) tangential electric ﬁeld should vanish. on the inner surface of the metamaterial coating (i.195) (2) Jn (k0 b) + cn Hn (k0 b) = an Jn (kc b) + bn Yn (kc b). ρ=a E0 n=−∞ j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) (2. we will deﬁne the scattered and transmitted electric ﬁelds in series expansion form. (2.193) t s i Ez (ρ = b) + Ez (ρ = b) = Ez (ρ = b).190) 2. (2.7.191) = 0. +∞ (2.7. Also. (2.188).4 Scattered and Transmitted Electric Fields Similar to the incident ﬁeld expression in (2. due to the boundary conditions. Therefore.5 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous on the outer surface of the metamaterial coating. +∞ (2. (2.2. respectively as follows: +∞ s Ez = E0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ)ejn(φ−φ0 ) .
7.231).201) k0 kc (2) Jn (k0 b) + cn Hn (k0 b) = [an Jn (kc b) + bn Yn (kc b)] µ0 µ Using (2.198) t Hφ 1 = E0 kc j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) . Therefore. jωµ0 n=−∞ +∞ (2.202) (2. (2) ζJn (k0 b) + cn ζHn (k0 b) = an Jn (kc b) + bn Yn (kc b).7 Boundary Conditions for Magnetic Fields Tangential components of the magnetic ﬁelds are continuous on the outer surface of the metamaterial coating. jωµ n=−∞ (2. jωµ0 n=−∞ +∞ +∞ (2. i s t Hφ (ρ = b) + Hφ (ρ = b) = Hφ (ρ = b).199) 2.197) s Hφ 1 (2) = E0 k0 j −n cn Hn (k0 ρ)ejn(φ−φ0 ) .50).58) in (2.2. Fields Scattered and Transmitted Magnetic Utilizing (2.49) and (2. the tangential components of the magnetic ﬁelds are obtained as i Hφ 1 = E0 k0 j −n Jn (k0 ρ)ejn(φ−φ0 ) .200) 1 (2) E0 k0 j −n Jn (k0 b) + cn Hn (k0 b) ejn(φ−φ0 ) jωµ0 n=−∞ = E0 1 j −n [an Jn (kc b) + bn Yn (kc b)] ejn(φ−φ0 ) kc jωµ n=−∞ +∞ +∞ (2. (2.6 Incident. (2.203) 44 . due to the boundary conditions.7.
(2.212) 45 .210) N = Jn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] − ζJn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] .204) (2.222) and (2. (2) D = ζHn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] (2) − Hn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] .7. (2) Jn (kc b)an + Yn (kc b)bn − ζHn (k0 b)cn = ζJn (k0 b).208) bn = . the solution to this system of equations can be found as: an = ζYn (kc a) Hn (k0 b)Jn (k0 b) − Hn (k0 b)Jn (k0 b) D ζJn (kc a) Jn (k0 b)Hn (k0 b) − Jn (k0 b)Hn (k0 b) D N .211) (2.206) . (2) Jn (kc b)an + Yn (kc b)bn − Hn (k0 b)cn = Jn (k0 b).2.228). or in matrix form 0 Jn (kc a) Yn (kc a) (2) Jn (kc b) Yn (kc b) −Hn (k0 b) (2) Jn (kc b) Yn (kc b) −ζHn (k0 b) 0 an bn = Jn (k0 b) cn ζJn (k0 b) (2. (2.205) (2.209) cn = where (2. D (2) (2) (2) (2) . Rearranging equations (2. (2. (2. (2.232) we get Jn (kc a)an + Yn (kc a)bn = 0.207) Using Symbolic Math Toolbox of MATLAB.8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields Now we have three unknowns and three equations.
01m. b = λ0 .Using the following Wronskian will further simplify an and bn : (2) (2) Jn (x)Hn (x) − Jn (x)Hn (x) = −2j . one of them is stronger. In Fig. λ0 = 0. 2.10 (c) the foci are distributed inside the cylinder. 2. 2.10 (a) there are two foci: one of them is inside the metamaterial and the other one is outside.213) 2.10 shows some of the numerical results when f = 30GHz. In Fig.5λ0 . πx (2.10 (b) there are three foci inside the metamaterial close to the conducting cylinder. Though.7.9 Numerical Results Fig. In Fig. 46 . a = 0. 2.
01 0.015 −0.005 0.015 0.5 −0.005 0 x(m) 0.5 0.02 0 −0.015 0. µr = 2 47 .02 (a) Magnitude of the Electric Field Inside and Outside the Cylinder 0.02 −0.02 −0.01 0.5 −0.015 −0.005 0.015 0.5 −0.Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) Magnitude of the Electric Field Inside and Outside the Cylinder 0.01 −0.10: Magnitude of the electric ﬁeld inside and outside the cylinder.01 0.5 0  E  (V/m) y(m) 1.005 0.015 −0.01 −0.005 1 −0.01 0.02 0 −0. µr = −1.015 0.005 0.02 −0.5 2 1.5 4 3.02 −0.02 −0. µr = −2.005 0.5 −0.02 −0.02 (e) (f) Figure 2.02 0 −0.005 0 x(m) 0. (c)(d) εr = −2.015 −0.5 −0.01 −0.015 0.01  E  (V/m) 2 y(m) 2 1.01 0.02 (d) Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 3 0.02 3.015 3.02 (c) Magnitude of the Electric Field Inside and Outside the Cylinder 0.01 1  E  (V/m) 2. (e)(f) εr = 2.5 1 −0.5 1 −0.01 0.5 3 0 2 −0.02 −0.5 0 1.01 −0.01 2 2 0.005 0 x(m) 0.015 −0.5 0.005 0 x(m) 0.5 0.015 −0.01 3 0.005 0 x(m) 0.5 0.015 0.005 2.01 −0.005 0 x(m) 0.02 −0.5 2.5 −0.005 1 1 −0.01 2.015 3 3 0.5 y(m) 2.005 1.01 0.005 1.5 0.5 −0. (a)(b) εr = −1.015 0.005 0.005 (b) Magnitude of the Electric Field Inside and Outside the Cylinder (along x axis) 4.02 4 0.02 −0.015 2.015 0.01 −0.015 0.5 0.
µc ε0 .11. µ0 φ0 Plane Wave Figure 2.8 Normally Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T E z Polarization 2.2. We will examine here the scattering and transmission by the metamaterial coated conducting cylinder in the case the polarization of the plane wave is T E z . 2. The problem geometry is depicted in Fig.11: Plane wave normally incident on a metamaterial coated conducting cylinder.2 Problem Geometry y ρ σ=∞ φ a b x εc . 48 .1 Introduction A uniform plane wave is normally incident on a metamaterial coated conducting cylinder of inﬁnite length. The plane wave travels in the direction which makes an angle φ0 with the +x axis.8. 2.8.
2. y = ρ sin φ.8. = H0 e−jk0 ρ cos(φ−φ0 ) .3 Uniform Plane Wave and Incident Magnetic Field Let us assume that a T E z polarized uniform plane wave is traveling in the direction which makes an angle φ0 with the +x axis.217).11 the magnetic ﬁeld can be written as i Hz = H0 e−jk0 (x cos φ0 +y sin φ0 ) . Referring to Figure 2.217) 2. incident magnetic ﬁeld can be written as i Hz = H0 e−jk0 ρ cos(φ−φ0 ) . (2. (2.8.4 Scattered and Transmitted Magnetic Fields Similar to the incident ﬁeld expression in (2. (2.215) = H0 e−jk0 ρ(cos φ cos φ0 +sin φ sin φ0 ) . we will deﬁne the scattered and transmitted magnetic ﬁelds in series expansion form respectively as follows: +∞ s Hz = H0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ)ejn(φ−φ0 ) .214) where x = ρ cos φ . +∞ = H0 n=−∞ j −n Jn (k0 ρ)ejn(φ−φ0 ) .218) t Hz = H0 n=−∞ j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) . Therefore. (2.7. Magnetic ﬁeld is directed along the +z axis. (2. (2.219) 49 . i Hz = H0 e−jk0 (ρ cos φ cos φ0 +ρ sin φ sin φ0 ) .216) Following the same procedure in Section 2.
jωε0 n=−∞ +∞ +∞ (2. on the inner surface of the metamaterial coating (i. (2.2. due to the boundary conditions. jωε n=−∞ (2.225) 2. +∞ (2. Scattered and Transmitted Electric Fields Utilizing (2.221) (2) Jn (k0 b) + cn Hn (k0 b) = an Jn (kc b) + bn Yn (kc b).222) 2.e. Therefore. jωε0 n=−∞ +∞ (2.. t s i Hz (ρ = b) + Hz (ρ = b) = Hz (ρ = b). the tangential components of the electric ﬁelds are obtained as i Eφ −1 = H0 k0 j −n Jn (k0 ρ)ejn(φ−φ0 ) . (2.8.223) s Eφ −1 (2) = H0 k0 j −n cn Hn (k0 ρ)ejn(φ−φ0 ) . on the conducting cylinder surface) tangential electric ﬁeld should vanish.7 Boundary Conditions for Electric Fields The tangential components of the electric ﬁelds are continuous on the outer surface of the metamaterial coating. Also.220) H0 n=−∞ (2) j −n Jn (k0 b) + cn Hn (k0 b) ejn(φ−φ0 ) +∞ = H0 n=−∞ j −n [an Jn (kc b) + bn Yn (kc b)] ejn(φ−φ0 ) .6 Incident.50).112) and (2.8.224) t Eφ = H0 −1 kc j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) .226) 50 . t Eφ (ρ = a) = 0. due to the boundary conditions.5 Boundary Conditions for Magnetic Fields The tangential components of the magnetic ﬁelds are continuous on the outer surface of the metamaterial coating. Therefore. (2.8.
230) k0 kc (2) Jn (k0 b) + cn Hn (k0 b) = [an Jn (kc b) + bn Yn (kc b)] ε0 ε Using (2.232) 2. 51 .8.227) (2. (2.231) (2. (2) Jn (k0 b) + cn Hn (k0 b) = an ζJn (kc b) + bn ζYn (kc b).8 Simultaneous Solution of the Boundary Conditions for Electric and Magnetic Fields Now we have three unknowns and three equations.−1 H0 kc j −n [an Jn (kc ρ) + bn Yn (kc ρ)] ejn(φ−φ0 ) jωε n=−∞ an Jn (kc a) + bn Yn (kc a) = 0. (2) ζJn (kc b)an + ζYn (kc b)bn − Hn (k0 b)cn = Jn (k0 b).231).222) and (2.232) we get Jn (kc a)an + Yn (kc a)bn = 0.228).229) −1 (2) k0 j −n Jn (k0 b) + cn Hn (k0 b) ejn(φ−φ0 ) H0 jωε0 n=−∞ = H0 −1 kc j −n [an Jn (kc b) + bn Yn (kc b)] ejn(φ−φ0 ) jωε n=−∞ +∞ +∞ (2. Rearranging equations (2.119) in (2. (2. ρ=a (2.234) (2. (2.233) (2.235) (2) Jn (kc b)an + Yn (kc b)bn − Hn (k0 b)cn = Jn (k0 b).228) i s t Eφ (ρ = b) + Eφ (ρ = b) = Eφ (ρ = b). (2. +∞ = 0.
or in matrix form 0 Jn (kc a) Yn (kc a) (2) Jn (kc b) Yn (kc b) −Hn (k0 b) (2) ζJn (kc b) ζYn (kc b) −Hn (k0 b) . (2.237) bn = .239) N = ζJn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] −Jn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] . 0 an bn = Jn (k0 b) cn Jn (k0 b) (2. (2.241) Using the following Wronskian will further simplify an and bn : (2) (2) Jn (x)Hn (x) − Jn (x)Hn (x) = −2j .240) (2) D = Hn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] (2) −ζHn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] .238) cn = where (2. D (2) (2) (2) (2) .242) 52 . (2.236) Using Symbolic Math Toolbox of MATLAB. (2. πx (2. the solution to this system of equations can be found as: an = Yn (kc a) Hn (k0 b)Jn (k0 b) − Hn (k0 b)Jn (k0 b) D Jn (kc a) Jn (k0 b)Hn (k0 b) − Jn (k0 b)Hn (k0 b) D N .
9. k θ0 x ε. z a E y . µ . We will examine here the scattering and transmission by the metamaterial cylinder in the case the polarization of the plane wave is T M z . The plane wave travels in the direction which makes an angle φ0 with the +x axis and θ0 with the −z axis. 2.2. 53 .9 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T M z Polarization 2. µ0 Figure 2.12: Uniform plane wave obliquely incident on a metamaterial cylinder: T M z Polarization. The problem geometry is depicted in Fig.1 Introduction A uniform plane wave is obliquely incident on a metamaterial cylinder of inﬁnite length. φ0 ε0 .12.
2. Due to phase matching. (2. (2.243) also since x = ρ cos φ and y = ρ sin φ. 2.244) In Section 2.247) t Ez = E0 sin θ0 e jk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) . the ﬁelds are periodic in the z direction and vary according to the factor ejk0 z cos θ0 [29]. Scattered and Transmitted Electric Fields (z components) Referring to Fig.9.7 Eqn.13. as shown in Fig.2. the z component of the electric ﬁeld can be expressed as i Ez = E0 sin θ0 e−jk0 ρ sin θ0 (cos φ cos φ0 +sin φ sin φ0 ) ejk0 z cos θ0 = E0 sin θ0 e−jk0 ρ sin θ0 cos(φ−φ0 ) ejk0 z cos θ0 . (2.245). (2. the incident electric ﬁeld can be written as Ei = (ax E0 cos θ0 cos φ0 + ay E0 cos θ0 sin φ0 + az E0 sin θ0 ) .246) Since the cylinder is of inﬁnite length. (2. (2. (2. (2. we have previously derived that +∞ E0 e −jk0 ρ cos(φ−φ0 ) = E0 n=−∞ j −n Jn (k0 ρ)ejn(φ−φ0 ) .e−jk0 x sin θ0 cos φ0 e−jk0 y sin θ0 sin φ0 ejk0 z cos θ0 . the propagation constant in 54 .188).246) as +∞ s Ez = E0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .248) The obliquely incident wave travels both in longitudinal and transverse directions.12.2 Incident.245) Utilizing (2.244) can be written as +∞ i Ez = E0 sin θ0 e jk0 z cos θ0 n=−∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) . The z components of the scattered and transmitted electric ﬁelds are expressed similar to (2.
k cos2 θ0 (2.246)(2. k  k0 cos θ0  > 1.251) cos2 θ0 − 1 and kt = k sin θ1 (2.253) 2 −j k0 cos2 θ0 − k 2  k0 cos θ0  > 1. (2.13: Longitudinal and transverse components of the incident and transmitted ﬁelds. k k0 cos θ0 .250)  k0 cos θ0  ≤ 1. Therefore. µ0 ε.249) (2. cos θ1 = sin θ1 = −j 1− k0 2 k k0 2 k (2. Remark: Note that. µ Figure 2.248) basically include the transverse propagation constants. The arguments of the Bessel and Hankel functions in (2. kz . k 55 . should be the same for free space and metamaterial media. k0 cos θ0 = k cos θ1 . since the metamaterial medium we consider here is not limited to only DPS metamaterials.252) should not be further simpliﬁed to: 2 k 2 − k0 cos2 θ0  k0 cos θ0  ≤ 1.252) is the transverse propagation constant in metamaterial medium. the longitudinal direction.z kt0 = k0sinθ0 kz = k0cosθ0 k0 θ0 k = k sinθ t 1 θ1 k kz = k cosθ1 ε0 . k kt = (2.
256) 2.4 φ Components of the Incident. for scattering by dielectric or dielectric coated conducting cylinders. in Appendix B. Therefore. there exist longitudinal magnetic ﬁeld components for the scattered and transmitted waves: i Hz = 0. (2.252) gives kt = −k0 . depolarization is inevitable in order to satisfy the Maxwell’s equations [27. k0 ρ sin θ0 n=−∞ +∞ (2. scattered and transmitted electric and magnetic ﬁelds are derived from their z components. Scattered and Transmitted Magnetic Fields (z components) Smooth perfectly conducting inﬁnite cylinders do not depolarize obliquely incident waves. which is the correct solution. However.3 Incident. (2.255) = E0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . t Hz = E0 sin θ0 e jk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) . the aforementioned simpliﬁcation in (2. They are found to be: i Eφ E0 cos θ0 jk0 z cos θ0 =− e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .9. Scattered and Transmitted Electric and Magnetic Fields The φ components of the incident. which is wrong.253) would yield kt = k0 . 2.254) (2. utilizing Maxwell’s Equations. +∞ s Hz (2. consider a DNG medium with k = −k0 at normal incidence (θ0 = π/2). However.9. 29].As an example. Since the propagation is only in the transverse direction.257) 56 .
266) 57 .260) s Hφ E0 cos θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) =− k0 ρ sin θ0 n=−∞ E0 (2) −j ejk0 z cos θ0 j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .265) (2.258) t Eφ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) =− k 2 ρ sin2 θ1 n=−∞ sin θ0 jk0 z cos θ0 +jE0 ζη0 e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) . i s t Eφ (ρ = a) + Eφ (ρ = a) = Eφ (ρ = a). (2.5 Boundary Conditions and Their Solution Tangential components of the electric and magnetic ﬁelds should be continuous on the surface of the metamaterial cylinder. i s t Ez (ρ = a) + Ez (ρ = a) = Ez (ρ = a). (2.262) 2.259) E0 jk0 z cos θ0 e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .263) (2. Therefore. t s i Hφ (ρ = a) + Hφ (ρ = a) = Hφ (ρ = a).261) t Hφ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 =− e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) k 2 ρ sin2 θ1 n=−∞ E0 sin θ0 jk0 z cos θ0 −j e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) .264) (2. i s t Hz (ρ = a) + Hz (ρ = a) = Hz (ρ = a). ζη0 sin θ1 n=−∞ +∞ (2.s Eφ E0 cos θ0 jk0 z cos θ0 (2) =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +∞ +jE0 η0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . sin θ1 n=−∞ i Hφ = −j +∞ +∞ (2.9. η0 n=−∞ +∞ +∞ (2. η0 n=−∞ +∞ +∞ (2.
ζ sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) 2 4 .271) (2) (2) . Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) − Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) (2. sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − ζ sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) . an = j 1 1 2 ζank0 k1 sin θ0 sin2 θ1 cos θ0 D η0 2 (2) . (2. (2. the equations are converted into matrix form and solved.which leads to (2) Jn (k0 a sin θ0 ) + cn Hn (k0 a sin θ0 ) = an Jn (ka sin θ1 ). Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) − Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) . (2) cn Hn (k0 a sin θ0 ) = an Jn (ka sin θ1 ). 2 a sin2 θ sin θ1 k 1 − E0 E0 cos θ0 E0 (2) (2) Jn (k0 a sin θ0 ) − ncn Hn (k0 a sin θ0 ) − j cn Hn (k0 a sin θ0 ) η0 k0 a sin θ0 η0 E0 k0 sin θ0 cos θ0 E0 sin θ0 (2. (2. k0 sin2 θ0 − k 2 sin2 θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) (2.274) where (2) D = − (Jn (ka sin θ1 ))2 Hn (k0 a sin θ0 ) 2 cos2 θ0 n2 ζ k0 sin2 θ0 − k 2 sin2 θ1 2 (2) (2) + sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − ζ sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) (2) (2) .267) (2.272) (2) (2) . a2 k0 k1 sin2 θ0 sin2 θ1 .275) 58 . (2.269) =− nan Jn (ka sin θ1 ) + jE0 ζη0 an Jn (ka sin θ1 ).268) E0 cos θ0 E0 cos θ0 (2) (2) nJn (k0 a sin θ0 ) − ncn Hn (k0 a sin θ0 ) + jE0 η0 cn Hn (k0 a sin θ0 ) k0 a sin θ0 k0 a sin θ0 E0 k0 sin θ0 cos θ0 sin θ0 (2. cn = an Jn (ka sin θ1 ) − Jn (k0 a sin θ0 ) Hn (k0 a sin θ0 ) cn = Jn (ka sin θ1 ) Hn (k0 a sin θ0 ) 2 (2) (2) . The unknown coeﬃcients are found to be: an = 1 2 2 4 2 ζa k0 k1 sin θ0 sin3 θ1 D (2) (2) .270) =− nan Jn (ka sin θ1 ) − j an Jn (ka sin θ1 ).273) an . 2 a sin2 θ ζη0 sin θ1 k 1 −j As we have done in previous sections.
ωµ0 jωε0 n=−∞ Note that.279) From Appendix B. large argument forms of the Hankel functions and their derivatives have the spread factor of ρ−1/2 .280) decay with ρ−3/2 .9. η0 (2. the ﬁrst and second terms in (2. when ρ → ∞ s Hρ ≈ − +∞ k0 cos θ0 E0 k0 jk0 z cos θ0 (2) e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . whereas the third term decays with ρ−1/2 and becomes dominant in the far zone.277).277) Let us use the deﬁnition in (2. Therefore. (2.2.6 Calculation of the Radar Cross Section 2D echo width can be found from either σ = lim or σ = lim ρ→∞ ρ→∞ 2πρ Es 2 Ei 2 Hs 2 Hi 2 . ωµ0 jωε0 n=−∞ +∞ jk0 z cos θ0 n=−∞ (2) j −n−1 cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . (2. s Hρ = − E0 sin θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 ρ n=−∞ +∞ +∞ (2. The magnitude of the scattered magnetic ﬁeld is Hs  = s s s Hρ 2 + Hφ 2 + Hz 2 .280) k0 cos θ0 E0 cos θ0 jk0 z cos θ0 (2) − e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 k0 ρ sin θ0 n=−∞ k0 cos θ0 E0 k0 jk0 z cos θ0 (2) − e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . +∞ = −E0 cos θ0 e (2.278) and the magnitude of the incident magnetic ﬁeld is Hi  = E0  .281) 59 . Therefore. (2.276) 2πρ .
n=−∞ (2. (2. 2 s Hρ 2 ≈ E0 2 cos2 θ0 πk0 ρ sin θ0 s We will follow similar steps for Hφ : +∞ 2 cn ejn(φ−φ0 ) . n=−∞ (2. s Hφ ≈ E0 jk0 z cos θ0 (2) e j −n−1 cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .282) (2.Since when ρ → ∞ (2) Hn (k0 ρ sin θ0 ) ≈ − 2j j n+1 e−jk0 ρ sin θ0 .289) .285) When ρ → ∞. η0 n=−∞ 2j πk0 ρ sin θ0 +∞ +∞ +∞ E0 ≈ − ejk0 (z cos θ0 −ρ sin θ0 ) η0 s Hφ 2 cn ejn(φ−φ0 ) . n=−∞ 2 (2.287) +∞ s Hz = E0 sin θ0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . πk0 ρ sin θ0 60 (2. jη0 n=−∞ +∞ +∞ (2.288) Using the large argument approximation (2) Hn (k0 ρ sin θ0 ) ≈ 2j j n e−jk0 ρ sin θ0 .283) Therefore. πk0 ρ sin θ0 (2.281) becomes s Hρ ≈ E0 cos θ0 e jk0 (z cos θ0 −ρ sin θ0 ) 2j πk0 ρ sin θ0 +∞ cn ejn(φ−φ0 ) .286) E0 2 2 ≈ 2 η0 πk0 ρ sin θ0 cn e n=−∞ jn(φ−φ0 ) .284) s Hφ E0 cos θ0 jk0 z cos θ0 (2) =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ E0 jk0 z cos θ0 (2) + e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . (2.
(2. n=−∞ 2 (2. n=−∞ (2. s Hz ≈ E0 sin θ0 e jk0 (z cos θ0 −ρ sin θ0 ) 2j πk0 ρ sin θ0 +∞ cn ejn(φ−φ0 ) . cn e n=−∞ jn(φ−φ0 ) + 2 η0 n=−∞ cn ejn(φ−φ0 ) 2 .295) 61 .291) s s s Hs 2 = Hρ 2 + Hφ 2 + Hz 2 . (2.292) +∞ 2 cn ejn(φ−φ0 ) n=−∞ +∞ . 2 ≈ E0 2 cos2 θ0 πk0 ρ sin θ0 E0 2 2 + 2 η0 πk0 ρ sin θ0 2 2 +∞ +∞ 2 cn ejn(φ−φ0 ) n=−∞ 2 cn e n=−∞ jn(φ−φ0 ) 2 cn ejn(φ−φ0 ) +E0  sin θ0 πk0 ρ sin θ0 n=−∞ 2 1 +∞ 2 jn(φ−φ0 ) 2 cn e = E0  + 2 πk0 ρ sin θ0 η0 n=−∞ From (2.277).293) The normalized (with respect to λ0 ) echo width is 2 2 +∞ +∞ 2 1 2 .294) cn ejn(φ−φ0 ) .when ρ → ∞.290) s Hz 2 2 ≈ E0  sin θ0 πk0 ρ sin θ0 2 2 +∞ cn e n=−∞ jn(φ−φ0 ) . (2. σ= 4 k0 sin θ0 2λ0 = π sin θ0 +∞ 2 2 cn ejn(φ−φ0 ) + η0 n=−∞ +∞ 2 +∞ 2 +∞ 2 cn ejn(φ−φ0 ) n=−∞ . cn ejn(φ−φ0 ) cn ejn(φ−φ0 ) + η0 σ/λ0 = π sin θ0 n=−∞ n=−∞ which at normal incidence special case becomes 2 σ/λ0 = π +∞ 2 (2.
2. (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .301) 62 .296) s Hz = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2. in this section only key equations will be given for completeness of the problem. The ﬁeld expressions and RCS calculations are similar to the T M z case.299) (2. t Ez = H0 sin θ0 e jk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) . Scattered and Transmitted Magnetic Fields (z components) +∞ i Hz = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .298) 2.3 Incident. (2. Hence. +∞ s Ez (2.1 Introduction The solution for the T E z polarization case can be obtained from the T M z case utilizing duality. (2.300) = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .10.10 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Cylinder: T E z Polarization 2.2.2 Incident. (2.297) t Hz = H0 sin θ0 ejk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) . Scattered and Transmitted Electric Fields (z components) i Ez = 0.10.10.
302) s Hφ = − H0 (2) −j ejk0 z cos θ0 j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .2.4 φ Components of the Incident.5 Boundary Conditions and Their Solution Tangential components of the electric and magnetic ﬁelds should be continuous on the surface of the metamaterial cylinder. η0 n=−∞ +∞ t Hφ = − (2. (2.304) = jH0 η0 e jk0 z cos θ0 n=−∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .10.307) 2. t s i Hz (ρ = a) + Hz (ρ = a) = Hz (ρ = a). +∞ (2. Therefore. k0 ρ sin θ0 n=−∞ H0 cos θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +∞ +∞ (2.306) t Eφ H0 k0 sin θ0 cos θ0 jk0 z cos θ0 e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) =− k 2 ρ sin2 θ1 n=−∞ sin θ0 jk0 z cos θ0 +jH0 ζη0 e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) .305) s Eφ H0 cos θ0 jk0 z cos θ0 (2) =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +jH0 η0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . Scattered and Transmitted Magnetic and Electric Fields i Hφ H0 cos θ0 jk0 z cos θ0 =− e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) . (2.303) H0 k0 sin θ0 cos θ0 jk0 z cos θ0 e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) k 2 ρ sin2 θ1 n=−∞ +∞ H0 sin θ0 jk0 z cos θ0 e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) .10.308) 63 . −j ζη0 sin θ1 n=−∞ +∞ i Eφ (2. sin θ1 n=−∞ +∞ +∞ (2.
(2.319) . the equations are converted into matrix form and solved.315) 2 a sin2 θ sin θ1 k 1 As we have done in previous sections. ζ sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) . cn = an Jn (ka sin θ1 ) − Jn (k0 a sin θ0 ) Hn (k0 a sin θ0 ) cn = Jn (ka sin θ1 ) Hn (k0 a sin θ0 ) 64 (2) (2) . i s t Hφ (ρ = a) + Hφ (ρ = a) = Hφ (ρ = a). (2. (2) cn Hn (k0 a sin θ0 ) = an Jn (ka sin θ1 ). (2.310) (2. (2.i s t Hz (ρ = a) + Hz (ρ = a) = Hz (ρ = a). i s t Hφ (ρ = a) + Hφ (ρ = a) = Hφ (ρ = a). an = −j 1 2 ζη0 ank0 k1 sin θ0 sin2 θ1 cos θ0 D (2) 2 .314) 2 a sin2 θ ζη0 sin θ1 k 1 − jH0 η0 Jn (k0 a sin θ0 ) − H0 cos θ0 (2) (2) ncn Hn (k0 a sin θ0 ) + jH0 η0 cn Hn (k0 a sin θ0 ) k0 a sin θ0 H0 k0 sin θ0 cos θ0 sin θ0 =− nan Jn (ka sin θ1 ) + jH0 ζη0 an Jn (ka sin θ1 ).309) (2. k0 sin2 θ0 − k 2 sin2 θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) (2. Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) − Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) .318) an . (2.313) H0 cos θ0 H0 H0 cos θ0 (2) (2) nJn (k0 a sin θ0 ) − ncn Hn (k0 a sin θ0 ) − j cn Hn (k0 a sin θ0 ) k0 a sin θ0 k0 a sin θ0 η0 H0 k0 sin θ0 cos θ0 H0 sin θ0 =− nan Jn (ka sin θ1 ) − j an Jn (ka sin θ1 ).311) which leads to (2) Jn (k0 a sin θ0 ) + cn Hn (k0 a sin θ0 ) = an Jn (ka sin θ1 ). (2. Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) − Jn (k0 a sin θ0 )Hn (k0 a sin θ0 ) (2. The unknown coeﬃcients are found to be: an = 1 2 2 4 2 a k0 k1 sin θ0 sin3 θ1 D (2) (2) .317) (2) (2) .312) (2.316) (2) (2) .
321) cn e n=−∞ jn(φ−φ0 ) . π sin θ0 n=−∞ η0 n=−∞ which at normal incidence case becomes 2 σ/λ0 = π +∞ 2 (2.where (2) D = − (Jn (ka sin θ1 ))2 Hn (k0 a sin θ0 ) 2 2 cos2 θ0 n2 ζ k0 sin2 θ0 − k 2 sin2 θ1 2 (2) (2) + sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − ζ sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) (2) (2) .6 Calculation of the Radar Cross Section Normalized echo width can be found using (2.322) 65 . ζ sin θ1 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) − sin θ0 Jn (ka sin θ1 )Hn (k0 a sin θ0 ) 2 4 . (2. (2. a2 k0 k1 sin2 θ0 sin2 θ1 .276) as: 2 2 +∞ +∞ 2 1 1 jn(φ−φ0 ) jn(φ−φ0 ) σ/λ0 = cn e cn e + 2 .10.320) 2.
11 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T M z Polarization 2.9. 66 . 2. Due to the conducting cylinder centered at the origin. µc . Jn (..)) but also in terms of Bessel functions of the second kind (i.e. The plane wave illumination and polarization is the same with Section 2.14.e.2. The problem geometry is depicted in Fig. the ﬁelds inside the metamaterial coating are written not only in terms of Bessel functions of the ﬁrst kind (i.11.14: Uniform plane wave obliquely incident on a metamaterial coated conducting cylinder: T M z Polarization. z PEC a b y θ0 x εc . φ0 ε0 . µ0 Figure 2..1 Introduction A uniform plane wave is obliquely incident on a metamaterial coated conducting cylinder of inﬁnite length.
11.2 Incident. (2. 2.324) t Ez = E0 sin θ0 ejk0 z cos θ0 n=−∞ j −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) . t Hz = E0 sin θ0 e jk0 z cos θ0 n=−∞ j −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) . k0 ρ sin θ0 n=−∞ +∞ (2.329) 67 . Scattered and Transmitted Electric and Magnetic Fields i Eφ E0 cos θ0 jk0 z cos θ0 =− e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .325) 2.11.323) s Ez = E0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2. (2.326) (2. and their derivatives.4 φ Components of the Incident. in this section only key equations are given. from Section 2.328) 2. (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . +∞ (2. Therefore. Scattered and Transmitted Electric Fields (z components) +∞ i Ez = E0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .9.11.327) s Hz = E0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .3 Incident.)).Yn (. This is the only diﬀerence in formulation. Scattered and Transmitted Magnetic Fields (z components) i Hz = 0. (2.
η0 n=−∞ +∞ +∞ (2. η0 n=−∞ t Hφ = − (2. n=−∞ nj −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) E0 sin θ0 jk0 z cos θ0 e ζη0 sin θ1 +∞ −j . n=−∞ nj −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) sin θ0 jk0 z cos θ0 e sin θ1 j −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) . n=−∞ +jE0 ζη0 +∞ .333) E0 k0 sin θ0 cos θ0 jk0 z cos θ0 e 2 kc ρ sin2 θ1 +∞ (2.330) t Eφ = − E0 k0 sin θ0 cos θ0 jk0 z cos θ0 e 2 kc ρ sin2 θ1 +∞ (2. n=−∞ 2.332) s Hφ = − E0 cos θ0 jk0 z cos θ0 (2) nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) e k0 ρ sin θ0 n=−∞ +∞ E0 (2) −j ejk0 z cos θ0 j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .s Eφ E0 cos θ0 jk0 z cos θ0 (2) =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +∞ +jE0 η0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .5 Boundary Conditions and Their Solution Tangential components of the electric and magnetic ﬁelds are continuous on the outer surface of the metamaterial coating.331) . (2.11.334) . i Hφ = −j E0 jk0 z cos θ0 e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) . Also. on the inner surface of 68 . j −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) .
341) (2) cn Hn (k0 b sin θ0 ) = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ). i s t Hz (ρ = b) + Hz (ρ = b) = Hz (ρ = b). (2.336) (2. on the conducting cylinder surface) tangential components of the electric ﬁeld should vanish. (2.the metamaterial coating (i.337) (2.344) ζη0 sin θ1 ζη0 sin θ1 an Jn (kc a sin θ1 ) + bn Yn (kc a sin θ1 ) = 0..340) which leads to (2) Jn (k0 b sin θ0 ) + cn Hn (k0 b sin θ0 ) = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ).345) E0 k0 sin θ0 cos θ0 E0 k0 sin θ0 cos θ0 nan Jn (kc a sin θ1 ) − nbn Yn (kc a sin θ1 ) 2 a sin2 θ 2 kc kc a sin2 θ1 1 sin θ0 sin θ0 +jE0 ζη0 an Jn (kc a sin θ1 ) + jE0 ζη0 bn Yn (kc a sin θ1 ) = 0. i s t Eφ (ρ = b) + Eφ (ρ = b) = Eφ (ρ = b). (2.e. i s t Ez (ρ = b) + Ez (ρ = b) = Ez (ρ = b).335) (2. t Ez (ρ = a) = 0.339) (2.342) E0 cos θ0 E0 cos θ0 (2) (2) nJn (k0 b sin θ0 ) − ncn Hn (k0 b sin θ0 ) + jE0 η0 cn Hn (k0 b sin θ0 ) k0 b sin θ0 k0 b sin θ0 E0 k0 sin θ0 cos θ0 E0 k0 sin θ0 cos θ0 =− nan Jn (kc b sin θ1 ) − nbn Yn (kc b sin θ1 ) 2 b sin2 θ 2 kc kc b sin2 θ1 1 sin θ0 sin θ0 an Jn (kc b sin θ1 ) + jE0 ζη0 bn Yn (kc b sin θ1 ). (2. (2. − (2. Therefore. i s t Hφ (ρ = b) + Hφ (ρ = b) = Hφ (ρ = b).338) (2. t Eφ (ρ = a) = 0.346) sin θ1 sin θ1 69 .343) +jE0 ζη0 sin θ1 sin θ1 − −j E0 E0 cos θ0 E0 (2) (2) Jn (k0 b sin θ0 ) − ncn Hn (k0 b sin θ0 ) − j cn Hn (k0 b sin θ0 ) η0 k0 b sin θ0 η0 E0 k0 sin θ0 cos θ0 E0 k0 sin θ0 cos θ0 nan Jn (kc b sin θ1 ) − nbn Yn (kc b sin θ1 ) =− 2 b sin2 θ 2 kc kc b sin2 θ1 1 E0 sin θ0 E0 sin θ0 −j an Jn (kc b sin θ1 ) − j bn Yn (kc b sin θ1 ). (2.
70 . The unknown coeﬃcients are found to be: an = − 1 2 4 Yn (kc a sin θ1 )ζk0 k1 b2 sin2 θ0 sin3 θ1 (2. (2) (2) (2.348) bn = − Jn (kc a sin θ1 ) an . [Jn (kc a sin θ1 )Yn (kc b sin θ1 ) − Jn (kc b sin θ1 )Yn (kc a sin θ1 )] (2) (2) . [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] 1 1 2 ζbnk0 k1 sin θ0 sin2 θ1 cos θ0 D η0 2 2 (2) .347) D (2) (2) . k0 sin2 θ0 − kc sin2 θ1 Yn (kc a sin θ1 )Hn (k0 b sin θ0 ) . Yn (kc a sin θ1 ) .353) 2 (2) D1 = ζ cos2 θ0 n2 Hn (k0 b sin θ0 ) 2 2 2 k0 sin2 θ0 − kc sin2 θ1 (2. (2. Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) − Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) (2) . Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) − Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) . (2.350) cn = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ) − Jn (k0 b sin θ0 ) Hn (k0 b sin θ0 ) cn = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ) Hn (k0 b sin θ0 ) D = D1 + D2 . [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] (2) −ζ sin θ0 Hn (k0 b sin θ0 ) . Yn (kc a sin θ1 ) Jn (kc a sin θ1 ) an . [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] .351) .349) bn = − (2. the equations are converted into matrix form and solved.352) where (2. sin θ1 Hn (k0 b sin θ0 ) .As we have done in previous sections. [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] .354) . an = j (2.
71 .9.294).11.6 Calculation of the Radar Cross Section Calculation of the Radar Cross Section is the same as in Section 2.355) 2.2 4 D2 = −b2 k0 k1 sin2 θ0 sin2 θ1 (2) . The normalized echo width is given in (2. Jn (kc a sin θ1 ) ζ sin θ1 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) − sin θ0 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −Yn (kc a sin θ1 ) ζ sin θ1 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) − sin θ0 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2. Jn (kc a sin θ1 ) sin θ1 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −ζ sin θ0 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −Yn (kc a sin θ1 ) sin θ1 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −ζ sin θ0 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) .
1 Introduction A uniform plane wave is obliquely incident on a metamaterial coated conducting cylinder of inﬁnite length. from Section 2. and their derivatives. The problem geometry is as depicted in Fig.2 Incident. due to the conducting cylinder centered at the origin.)). (2. in this section only key equations are given. the ﬁelds inside the metamaterial coating are written not only in terms of Bessel functions of the ﬁrst kind (i. Jn (. As in Section 2. Yn (. This is the only diﬀerence in formulation.14.e.356) s Hz = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2. (2.e. 2.. 2.12 Obliquely Incident Plane Wave Scattering by an Inﬁnite Length Metamaterial Coated Conducting Cylinder: T E z Polarization 2.12.. The plane wave illumination and polarization is the same with Section 2.357) t Hz = H0 sin θ0 e jk0 z cos θ0 n=−∞ j −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) . Therefore.10.12.)) but also in terms of Bessel functions of the second kind (i.2. Scattered and Transmitted Magnetic Fields (z components) +∞ i Hz = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .10. (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) .358) 72 .11.
359) (2.361) 2.362) s Hφ = − −j (2.4 φ Components of the Incident. (2.360) = H0 sin θ0 e jk0 z cos θ0 n=−∞ +∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . t Ez = H0 sin θ0 ejk0 z cos θ0 n=−∞ j −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) . η0 n=−∞ +∞ +∞ +∞ (2.364) . (2. Scattered and Transmitted Electric Fields (z components) i Ez = 0.3 Incident.2.12. +∞ s Ez (2. k0 ρ sin θ0 n=−∞ H0 cos θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ H0 jk0 z cos θ0 (2) e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . j −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) .365) 73 . n=−∞ +∞ i Eφ = jH0 η0 ejk0 z cos θ0 n=−∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .12. n=−∞ nj −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) H0 sin θ0 jk0 z cos θ0 e ζη0 sin θ1 +∞ −j . Scattered and Transmitted Magnetic and Electric Fields i Hφ = − H0 cos θ0 jk0 z cos θ0 e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) .363) t Hφ = − H0 k0 sin θ0 cos θ0 jk0 z cos θ0 e 2 kc ρ sin2 θ1 +∞ (2.
370) (2.371) (2. i s t Hz (ρ = b) + Hz (ρ = b) = Hz (ρ = b).s Eφ H0 cos θ0 jk0 z cos θ0 (2) =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +∞ +jH0 η0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) . (2.374) (2) cn Hn (k0 b sin θ0 ) = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ). n=−∞ +jH0 ζη0 +∞ . (2.12.367) . (2.373) which leads to (2) Jn (k0 b sin θ0 ) + cn Hn (k0 b sin θ0 ) = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ).e. i s t Hφ (ρ = b) + Hφ (ρ = b) = Hφ (ρ = b).369) (2.366) t Eφ = − H0 k0 sin θ0 cos θ0 jk0 z cos θ0 e 2 kc ρ sin2 θ1 +∞ (2.368) (2.. 2.375) 74 . on the conducting cylinder surface) tangential components of the electric ﬁeld should vanish. Also. i s t Eφ (ρ = b) + Eφ (ρ = b) = Eφ (ρ = b). t Eφ (ρ = a) = 0. (2. Therefore.372) (2. i s t Ez (ρ = b) + Ez (ρ = b) = Ez (ρ = b). on the inner surface of the metamaterial coating (i. n=−∞ nj −n an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 ) ejn(φ−φ0 ) sin θ0 jk0 z cos θ0 e sin θ1 j −n [an Jn (kc ρ sin θ1 ) + bn Yn (kc ρ sin θ1 )] ejn(φ−φ0 ) .5 Boundary Conditions and Their Solution Tangential components of the magnetic and electric ﬁelds are continuous on the outer surface of the metamaterial coating. t Ez (ρ = a) = 0.
380) D n (2) (2) . (2.381) 75 . [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] an = j 1 2 ζη0 bnk0 k1 sin θ0 sin2 θ1 cos θ0 D (2) 2 2 .379) +jH0 ζη0 sin θ1 sin θ1 As we have done in previous sections. k0 sin2 θ0 − kc sin2 θ1 Yn (kc a sin θ1 )Hn (k0 b sin θ0 ) .377) sin θ1 sin θ1 an Jn (kc a sin θ1 ) + bn Yn (kc a sin θ1 ) = 0. Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) − Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) (2) . the equations are converted into matrix form and solved. The unknown coeﬃcients are found to be: an = 1 2 4 Y (kc a sin θ1 )k0 k1 b2 sin2 θ0 sin3 θ1 (2. [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] (2) − sin θ0 Hn (k0 b sin θ0 ) . − (2. Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) − Jn (k0 b sin θ0 )Hn (k0 b sin θ0 ) . (2. ζ sin θ1 Hn (k0 b sin θ0 ) . ζη0 sin θ1 ζη0 sin θ1 jH0 η0 Jn (k0 b sin θ0 ) − H0 cos θ0 (2) (2) ncn Hn (k0 b sin θ0 ) + jH0 η0 cn Hn (k0 b sin θ0 ) k0 b sin θ0 H0 k0 sin θ0 cos θ0 H0 k0 sin θ0 cos θ0 =− nan Jn (kc b sin θ1 ) − nbn Yn (kc b sin θ1 ) 2 2 2 kc b sin θ1 kc b sin2 θ1 sin θ0 sin θ0 +jH0 ζη0 an Jn (kc b sin θ1 ) + jH0 ζη0 bn Yn (kc b sin θ1 ). (2.376) an Jn (kc b sin θ1 ) − j bn Yn (kc b sin θ1 ).378) H0 k0 sin θ0 cos θ0 H0 k0 sin θ0 cos θ0 nan Jn (kc a sin θ1 ) − nbn Yn (kc a sin θ1 ) 2 2 2 kc a sin θ1 kc a sin2 θ1 sin θ0 sin θ0 an Jn (kc a sin θ1 ) + jH0 ζη0 bn Yn (kc a sin θ1 ) = 0.− H0 cos θ0 H0 cos θ0 H0 (2) (2) nJn (k0 b sin θ0 ) − ncn Hn (k0 b sin θ0 ) − j cn Hn (k0 b sin θ0 ) k0 b sin θ0 k0 b sin θ0 η0 H0 k0 sin θ0 cos θ0 H0 k0 sin θ0 cos θ0 nan Jn (kc b sin θ1 ) − nbn Yn (kc b sin θ1 ) =− 2 2 kc b sin2 θ1 kc b sin2 θ1 H0 sin θ0 H0 sin θ0 −j (2. [Jn (kc a sin θ1 )Yn (kc b sin θ1 ) − Jn (kc b sin θ1 )Yn (kc a sin θ1 )] (2) (2) .
385) where (2. Jn (kc a sin θ1 ) sin θ1 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −ζ sin θ0 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −Yn (kc a sin θ1 ) sin θ1 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −ζ sin θ0 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) . Yn (kc a sin θ1 ) (2. [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] . Yn (kc a sin θ1 ) Jn (kc a sin θ1 ) an . (2.387) .383) cn = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ) − Jn (k0 b sin θ0 ) Hn (k0 b sin θ0 ) cn = an Jn (kc b sin θ1 ) + bn Yn (kc b sin θ1 ) Hn (k0 b sin θ0 ) D = D1 + D2 . 2 4 D2 = b2 k0 k1 sin2 θ0 sin2 θ1 (2) . (2.384) . (2) (2) .bn = − Jn (kc a sin θ1 ) an . [Jn (kc b sin θ1 )Yn (kc a sin θ1 ) − Jn (kc a sin θ1 )Yn (kc b sin θ1 )] .382) bn = − (2. Jn (kc a sin θ1 ) ζ sin θ1 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) − sin θ0 Yn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) −Yn (kc a sin θ1 ) ζ sin θ1 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2) − sin θ0 Jn (kc b sin θ1 )Hn (k0 b sin θ0 ) (2.388) 76 .386) 2 (2) D1 = −ζ cos2 θ0 n2 Hn (k0 b sin θ0 ) 2 2 2 k0 sin2 θ0 − kc sin2 θ1 (2.
2.6 Calculation of the Radar Cross Section Calculation of the Radar Cross Section is the same as in Section 2. 77 .10. The normalized echo width is given in (2.12.321).
achieving transparency and maximizing scattering are separately achieved by covering perfect electric conductor (PEC) cylinders with simple (i.7–12]. isotropic and linear) metamaterial coatings.1 Introduction In this chapter.. transparency and resonance are found to be heavily dependent on the ratio of corecoating radii.Chapter 3 Achieving Transparency and Maximizing Scattering with Metamaterial Coated Conducting Cylinders 3. 78 . instead of the total size of the cylindrical structure. the electromagnetic interaction of plane waves with inﬁnitely long metamaterial coated conducting cylinders is considered. As in the case of “conjugate” pairing. homogeneous.e. Diﬀerent from “conjugate” pairing of doublepositive (DPS) and doublenegative (DNG) or epsilonnegative (ENG) and munegative (MNG) concentric cylinders [5.
Here. Besides. unlike the aforementioned “conjugate” pairing cases. In n the subwavelength limit. The theory and formulation for T M and T E polarizations have been previously given in Sections 2.7 and 2. unless the core cylinder is replaced with perfect magnetic conductor (PMC).8. PEC cylinders together with their metamaterial coatings) are electrically small.2 Transparency Condition The transparency condition for T E z polarization is derived in Appendix C by setting the numerator of the scattering coeﬃcient cT E given in (2.e. 3. we derive the analytical relation between the ratio of corecoating radii and the permittivity of the metamaterial coating in the T E polarization case. we show numerically that for electrically small PEC cylinders transparency can be obtained by covering them with metamaterial covers having large µc . whereas the coating permittivity has to be in the −ε0 < εc < 0 interval for resonance so that scattering maximization can be achieved.In our work we show that. respectively.240) to zero. for T E polarization.. The numerical results show the validity of these analytical relations. both transparency and resonant peaks can be achieved for T M polarization. whereas resonant peaks are observed when µc < 0. k0 b 1 and utilizing the small argument forms of Bessel and Hankel functions. notice that because the core cylinder is PEC. For both transparency and resonance conditions. especially when the cylindrical scatterers (i. Yet. the following transparency 79 . assuming kc a < kc b 1. the analytical relations we have derived for T E polarization cannot be used for T M polarization by interchanging ε with µ (and vice versa). the metamaterial coating should have 0 < εc < ε0 as its permittivity to achieve transparency.
ε0 + εc (3. n = 0.1) is of even degree of n (i.e. In this case (3. Alternatively. ε → −j∞ and µ = µ0 . However.3) where (ε.3) becomes γ= µc − µ0 = 1 for µc = µ0 . When the core cylinder is PEC.1) as γ→ 2n (εc − ε0 )(εc − j∞) = (εc + j∞)(εc + ε0 ) 2n ε0 − εc ε0 + εc for n = 0. (3.4) which means there would be no coating.condition is obtained: γ= 2n ε0 − εc ε0 + εc for n = 0.5) The root in (3.2) can still be used in the limiting case. (3. one can use the transparency condition for an electrically small cylindrical scatterer. (3. ε0 + εc (3. which implies that the argument of the root must be positive. given in [5] for the T E z polarization as γ= 2n (εc − ε0 )(εc + ε) (εc − ε)(εc + ε0 ) µ c − µ0 µc − µ for n = 0.1) where γ = a/b is the ratio of coreshell radii.. which is composed of two concentric layers of diﬀerent isotropic materials. when there is a coating γ should vary between 0 and 1. µ) are constitutive parameters of the core cylinder and (εc . µc ) are constitutive parameters of the coating (shell) layer.2) γ= for n = 0. 2n). Therefore. µc − µ0 (3. yielding the same transparency condition in (3. (3. 0< which leads to 0< ε0 − εc ⇒ −ε0 < εc < ε0 . (3. On the other hand.6) 80 . n is the index of series summation.7) ε0 − εc < 1.
Simply we will choose µc = µ0 in the numerical experiments for convenience. transparency is obtained at diﬀerent γ values (reasonably below desired values). which are also tabulated in Table 3. Based on numerical results.9).8). Similarly. expressions leading to εc given in (3. again analytically.1. the transparency condition for the PEC cylinder is independent of the permeability of its metamaterial coating. the following procedure is applied to test the accuracy of the transparency condition: for a desired γ value.1)] are overly simpliﬁed.and ε0 − εc < 1 ⇒ εc < −ε0 or 0 < εc . εc . this is true when the cylindrical scatterer is electrically small and the scattering problem is consequently “quasielectrostatic”. Then. when 81 .9) (3.7) and (3. for certain outer shell radii some γ values are selected where transparency is desired to be observed. As a matter of fact. utilizing (3. 1 + γ 2n (3.e.1) as εc = 1 − γ 2n ε0 . using this coating permittivity. (3. ε0 + εc From (3.10)] are tabulated in Table 3. the proper choice for εc lies in 0 < ε c < ε0 . the diﬀerence between desired and obtained γ values where transparency occurs) is when the subwavelength limit assumptions are performed. In the numerical experiments. Interestingly.10) [or (3.. The permittivities of the metamaterial coating corresponding to these γ values after (3.10) to ﬁnd the coating permittivity for a desired γ. for the T E z case. we analytically ﬁnd what the coating permittivity. One way to explain this deviation (i. particularly in terms of a and b.1). should be.8) As it can be seen from (3.1)(3. In Table 3. For a speciﬁc coating permittivity εc . one can analytically ﬁnd the corecoating ratio γ at which transparency can be obtained.1.10) [by setting n = 1 in (3. one can rewrite (3.1. we numerically ﬁnd at which γ value transparency is actually obtained.
11).the core cylinder is replaced with a coredielectric.2 0.165 0.31 0.2) and (3. This is an expected result since the accuracy of (3.923 0.10) as εc = 1 − γ (2n−γ) ε0 .575 0.11) to ﬁnd εc for a desired γ value.3) utilizing duality: γ= 2n (µc − µ0 )(µc + µ) (µc − µ)(µc + µ0 ) for n = 0.9 εc /ε0 0.2.41 0.2) yields accurate results as mentioned in [5] for electrically small cylinders.6 0.1 and noticing that the deviation between desired and obtained γ values usually increases as the value of γ increases.342 0.11)] are tabulated in Table 3.39 0.12) 82 .5 0. especially when b ≤ λ0 /10. The transparency condition for the initial cylindrical structure for the T M z polarization can be found from (3. As it can be seen from Table 3.1: Desired and Obtained γ for Achieving Transparency Using (3. 1 + γ (2n−γ) (3. we heuristically modify (3. our heuristic formula decreases the deviation successfully.7 0. the corresponding εc values and obtained γ values where transparency occurs after (3.81 b = λ0 /10 Obtained γ 0.15 0.1.51 0.10) b = λ0 /100 Desired γ 0. In (3.2. Table 3.10) decreases as the electrical size of the scatterer increases. It is also observed that as the electrical size of the cylindrical scatterer increases.105 Obtained γ 0. (3.105 0. εc given in (3. analytically.595 0. desired γ values.11) [again by setting n = 1 in (3. deviation of the obtained γ values from the the desired γ values increases. the dependence of εc to a and b is more strongly pronounced. Similar to Table 3.78 Based on Table 3.805 b = λ0 /5 Obtained γ 0.
67 0.7 0. (µc − µ0 )(µc + µ0 ) µc = −µ0 εc − ε0 → εc − ε εc − ε0 εc + j∞ for n = 0.68 0. To be able to achieve transparency for the T M z polarization utilizing similar transparency conditions we have derived for T E z polarization. the core should be PMC instead of PEC.86 γ= for n = 0. (3.5 0. (3.875 b = λ0 /5 Obtained γ 0.19 0.125 0. n = 0.175 0. It is obvious that in DPSDNG or ENGMNG pairing no such diﬃculty arises since duality can be simply applied. our numerical investigations show that polarization can be obtained for electrically small cylinders with metamaterial coatings having large µc .9 εc /ε0 0.4 (Numerical Results and Discussion).228 0.13) become γ= 2n µc = µ 0 (µc − µ0 )(µc + µ0 ) = 1 for .0579 Obtained γ 0. even if the core cylinder is PEC.2: Desired and Obtained γ for Achieving Transparency Using (3.Table 3. Yet.478 0. Theoretical analysis or simply duality shows that in such a case one can use the dual of transparency condition for T E z polarization by interchanging any permittivity with the corresponding permeability.14)(3.49 0.14) γ= (3.11) b = λ0 /100 Desired γ 0. (3.625 0. 83 . Examples of this situation are illustrated in Section 3.15) that the transparency condition for the T M z polarization does not lead to any reasonable outcome due to the core being PEC.47 0.12)(3.2 0.895 0.15) It can be deduced from (3.13) After replacing the core cylinder with a PEC one.395 0.875 εc − ε0 εc − ε b = λ0 /10 Obtained γ 0.
then 0< which leads to 0< and ε0 + εc < 1 ⇒ εc < 0 or εc > ε0 .16) Alternatively. (3. (3. ε0 − εc (3. the proper choice for εc lies in −ε0 < εc < 0.e.20) ε0 + εc < 1. This yields the following resonance condition: γ= 2n ε0 + εc ε0 − εc for n = 0. ε0 − εc From (3. the ratio of coreshell radii γ. ε0 − εc (3..18) Since the root in (3.16) is of even degree of n (i.3 Resonance (Scattering Maximization) Condition The resonance condition.20) and (3. again in the subn wavelength limit.17) When the core cylinder is PEC.17) becomes γ→ 2n (εc + ε0 )(εc − j∞) = (εc − ε0 )(εc + j∞) 2n ε0 + εc ε0 − εc for n > 0. to maximize scattering from a metamaterial coated PEC cylinder. one can use the resonance condition given in [8] for the T E z polarization γ= 2n (εc + ε0 )(εc + ε) (εc − ε0 )(εc − ε) for n > 0.21) ε0 + εc ⇒ −ε0 < εc < ε0 .22) (3. (3. (3. (3. which increases the scattering drastically for an electrically small cylindrical scatterer.3. 2n) and 0<γ<1 should be.241) to zero.19) Then. is derived in Appendix D by setting the denominator of the scattering coeﬃcient cT E in (2. can be found analytically from the permittivity of the 84 .21).
while the n = ±1 terms. It has been shown in [28] that the n = 0 term is equivalent to a zdirected magnetic line source. the permittivity of the coating to maximize scattering should be the negative of the coating permittivity which makes the cylinder transparent. we ﬁnd the coating permittivity for a desired γ value analytically and then use it in the numerical experiment). Due to its electrically small size. which are referred as dipolar terms in [11]. Our numerical experiments show that. we follow the same procedure as in the transparency condition (i. consider a PEC cylinder which is illuminated by a T E z polarized plane wave. (3. coatings we use here for scattering maximization are ENG metamaterials (or plasmonic materials). Interestingly. the whole cylindrical scatterer will form an inductorcapacitor (LC) resonator. comparison of (3.coating εc utilizing (3.23) works quite well (by setting n = 1).16).. Therefore.23) for a desired γ value shows that. this electric dipole behaves like a capacitive element. for electrically small cylindrical scatterers. If the cylinder is electrically small. the n = 0 term becomes dominant. However. the n = ±1 terms cannot be neglected since they radiate more eﬃciently [28].10) with (3.e. Therefore. and vice versa: εc = γ 2n − 1 ε0 . correspond to a ydirected electric dipole. the coating will act like an inductive element. A similar scenario is investigated in [17] for electrically small antennas enclosed by metamaterial 85 . For the T E z case. To understand how this resonance condition occurs.23) In our numerical experiments with scattering maximization. since the scattering maximization condition is independent of the permeability of its coating and for electrically small cylindrical scatterers we are dealing with the “quasielectrostatic” problem. γ 2n + 1 (3. we can safely choose µc = µ0 . Therefore. we do not modify it as we have modiﬁed the analytical transparency relation. If there is also an ENG coating present.
is given in [8] as γ= 2n (µc + µ0 )(µc + µ) (µc − µ0 )(µc − µ) for n > 0. octopolar (i.. −µ0 < µc < 0) yields a resonance at the desired γ value for the T M z polarization. a and b). However.25) Although (3.17) utilizing duality. Although the aforementioned theoretical analysis is based on electrically small cylinders and a few modes of the inﬁnite series is assumed to be dominant. our numerical investigations show that µc values obtained via (3.e.24) becomes γ= n µc + µ0 µc − µ0 for µc = µ0 ..shells. k0 b 1). from the desired γ values) yield resonance (i. then dual of (3.. As the size of the scatterer increases.e. all the formulations used for transparency and scattering maximization conditions are independent of the electrical size of the cylindrical scatterer (i. such that only a few modes of the inﬁnite series summation is enough to represent the whole radar cross section.. On the other hand.25) (i.e. (3.25) states a resonance relation between a desired γ value and µc for the T M z polarization. n = 2). our numerical results do not include any assumption in this sense.e. kc b 1. which can be derived from (3.. n = 3) and any higher order terms also emerge as resonant terms [11]. if PEC core is replaced by a PMC core.e.e. similar to the transparency condition.24) After replacing the core cylinder with a PEC one. The resonance condition for the same cylindrical structure for the T M z polarization.e. In other words.. the formulations are expected to work well for electrically very small cylinders (i. quadrupolar (i.22) (i. maximum scattering) at γ values diﬀerent from the desired ones. Note that. in the computation of the normalized echo widths we use suﬃciently many modes to be accurate. n > 0. 86 . (3. (3.
a perfect continuation in the monostatic echo width values is observed (as expected) when the coating medium becomes singlenegative (SNG) from a DPS or DNG coating. In the previous sections. we have found that it is possible to make PEC cylinders transparent for 87 . we have excellent agreement with the results of [1].2ε0 (d) εc = −2. Moreover.5 4 3. Diamond marks show the DPS and DNG coating cases in [1]. which is shown in Fig.5 0 −10 −5 0 ε /ε 5 10 0 −10 −5 c 0 0 εc/ε0 5 10 (a) µc = µ0 7 6 5 4 3 2 1 0 −10 TM TE TM (Li) TE (Li) 7 6 5 4 3 2 1 0 −10 TM TE TM (Li) TE (Li) (b) µc = −µ0 σ/λ0 −5 0 µc/µ0 5 10 σ/λ0 −5 0 µc/µ0 5 10 (c) εc = 2.4 Numerical Results and Discussion To assess the accuracy of our numerical routines.3. expanding the transparency condition given in 4. we also included ENG and MNG coatings. b = 70mm.1: Normalized monostatic echo width of a metamaterial coated PEC cylinder (a = 50mm. f = 1GHz). As seen in Fig.5 2 1.5 3 TM TE TM (Li) TE (Li) 6 5 4 σ/λ0 3 2 1 TM TE TM (Li) TE (Li) σ/λ 0 2. [5]. we have duplicated one of the numerical results (normalized monostatic echo width of a metamaterial coated PEC cylinder at 1GHz with PEC radius a = 50mm and coating radius b = 70mm) in [1].2ε0 Figure 3.5 1 0. In addition to the DPS and DNG coatings investigated in [1].1.1. 3. 3.
The normalized monostatic echo widths for uncoated PEC cylinders (i. we start with an electrically very small PEC cylinder (in the crosssectional sense) covered with our proposed metamaterial coating such that the outer radius of the coating is b = λ0 /100. transparency can be achieved at the desired γ values. with radius a) are shown with dashed lines to visualize the reduction in scattering when proposed metamaterial coatings are used. we gradually increase the outer radius of the cylindrical scatterer. no scattering is supposed to take place. when the outer radius of the scatterer is increased to b = λ0 /10. the largest normalized monostatic echo width increases roughly from 40dB to 5dB).2(d) for these permittivities. 3. as it can be seen from Figs. the normalized monostatic echo widths are calculated and depicted in Figs.the T E z polarization by covering them with metamaterial covers which exhibit the material property given by (3.2) when the outer radius of the scatterer is increased to b = λ0 /5. 3. where transparency is desired to be observed. 88 . we investigate what happens to the transparency as the electrical size of the scatterer increases. we can still achieve transparency close to desired γ values (as tabulated in Table 3. Finally. For this purpose.2(h). Similarly..9). Therefore. 3. By transparency we mean the signiﬁcant reduction and minimization of scattering in the backscattering direction. Then.2(e)3. One can see that transparency is indeed obtained for PEC cylinders almost at the desired γ values.2(h) and Table 3. Despite this huge increase in RCS. Naturally.2(h) we see that increasing the electrical size of the cylindrical scatterer from b = λ0 /100 to b = λ0 /10 increases the RCS considerably (e.g. as a goes to zero.2.11) as tabulated in Table 3.2(a)3. the transparency condition is expected to work well for electrically very small cylinders. As the next step. Note that for the uncoated case small γ values mean extremely small PEC cylinders. As it has been explained previously.. the corresponding permittivities are analytically found using (3. The normalized monostatic echo widths are calculated and depicted in Figs.e. From Figs. for some γ values.2(e)3.2. 3.2(e)3.
To explain this phenomenon. 89 . However.. The transparency condition is attributed to the cancellation of these antiparallel polarization vectors. the cancellation of the coating will become stronger. the corecoating ratio where transparency occurs moves from γ = 0 to γ = 1. which happens when εc < ε0 . respectively as P = (ε − ε0 )E and Pc = (εc − ε0 )E. In our scenario. metamaterial cover behaves like free space).Fig. In this case. when the permittivity of the coating is decreased towards 0. 3.2 show that as the permittivity of the coating is decreased from εc = ε0 to εc = 0. which means that with even thinner coatings it becomes possible to make larger PEC cores transparent. this cancellation is quite weak (i. we can treat the metamaterial coating as a cover which cancels out the electromagnetic response of the PEC core. When the permittivity of the metamaterial coating is close to ε0 . the PEC core should be considerably small with respect to the coating such that a full cancellation can occur. since the core cylinder is PEC. their polarization vectors are deﬁned.e. the problem has a less degree of freedom and the analytical solution shows that to achieve transparency 0 < εc < ε0 should be. For both the dielectric core and the metamaterial cover.2 and Table 3. Note that a similar discussion is made in [5] to explain the cancellation phenomenon for metamaterial coated dielectric spheres.
8 0.9 1 (c) εc = 0.6 0.895ε0 .8 0.9 1 TE 0.8 0. µc = µ0 −40 −40 (b) εc = 0. µc = µ0 0 −10 −10 −20 −20 0 (d) εc = 0.2 0.1 −60 0.2 0.9 1 0 0 −70 −80 −80 −90 −90 −100 0.4 0.9 1 (e) εc = 0.3 0.1 σ /λ0 (dB) −20 −25 −30 −35 −40 −45 0.7 0.8 0.2 0.7 0. with radius a.7 0.6 0.4 0.8 0.228ε0 .2 0.5 γ = a/b 0. µc = µ0 Figure 3.895ε0 . µc = µ0 σ /λ0 (dB) −40 −50 −60 σ /λ0 (dB) −30 −30 TE TE −40 −50 −70 −80 0. (e)(h) b = λ0 /10.478ε0 .4 0.2 0.7 0.2 0.9 1 −80 −90 −100 −70 −80 −90 0 −110 −120 −130 0.1 0.3 0.1 γ = a/b −100 0.−40 −50 −60 −40 −50 −60 −70 σTE/λ (dB) σTE/λ (dB) 0.8 0.9 1 (a) εc = 0.1 0 γ = a/b 0.2 0.228ε0 .3 0.7 0.9 1 (g) εc = 0.4 0.1 0.6 0.7 0. The outer radius of the coating is selected as (a)(d) b = λ0 /100. Dashed line shows the uncoated PEC case.8 0.5 γ = a/b 0.3 0.2 0.4 0.7 0.6 0.1 TE 0.4 0.3 0. µc = µ0 −5 −10 −15 −20 −5 −10 −15 (f) εc = 0.5 0.3 0.0579ε0 .3 0. 90 .478ε0 . µc = µ0 σ /λ0 (dB) −25 −30 −35 −40 −45 −50 0.4 0.8 0.6 0.6 0.1 −100 −110 0.2: Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T E z polarization case.5 0.5 γ = a/b 0.9 1 0.7 0. µc = µ0 (h) εc = 0.3 0.5 γ = a/b 0.6 0.5 γ = a/b 0.5 γ = a/b 0. versus the corecoating ratio for coatings with diﬀerent constitutive parameters.0579ε0 .6 0.4 0. µc = µ0 −50 −50 −60 −60 σTE/λ (dB) −70 σTE/λ (dB) 0.
To see the limitations on the electrical size of the cylindrical scatterers for achieving transparency.3(f) this outer radius is further increased to b = λ0 . Finally. for these large scatterers we choose εc in a trial & error process.4(a)3. this time the peaks are wider and the peak centers deviate a little from their desired locations.9 < γ < 1 region. Therefore. depending on the permittivity of the coating. Therefore.4(e)3. 3.3(a) and Fig. Figs. making peaks.23). the minimum value of the normalized echo width (σ T E /λ0 drops from 4dB to 25dB) is achieved when the permittivity is very close to zero but positive. We again start with electrically very small cylindrical scatterers and gradually increase their outer radii. As it can be seen from the ﬁgures.4(d) show the normalized monostatic echo widths for ENG coated PEC cylinders when the outer radius of the scatterer is b = λ0 /100. and γ being between 0. we turn our attention to investigate the validity of scattering maximization condition. 3. This is mainly due to the resonance of dipolar terms which we have explained previously. available analytical relations between γ and εc do not hold any longer. Hence. the RCS peaks can still be clearly seen in Figs. but this time to maximize scattering. the coating permittivities are the negatives of coating permittivities tabulated in Table 3. In Figs.3(a)3. As the permittivity of the coating is decreased towards 0.3(d)3. the dips move towards γ = 1. we follow a procedure similar to the one we have done for the transparency condition. Figs.4(h). 3. 3. destructively interfering with each other. Hence. RCS increases drastically at the desired γ values.3(c) show the results when the outer radius of the scatterer is increased to b = λ0 /2. the PEC core can be quite large.9 and 1. 3. larger cylinders require coatings having permittivities much closer to zero. the normalized monostatic echo width makes two dips at some γ. But.1 as our desired γ values. 3. We use the same γ in Table 3. As it is seen in Fig. Next.1. Since these scatterers are electrically large. Since monostatic echo width is minimized in the 0.3(d). we will consider relatively larger scatterers. When the outer radius is b = λ0 /50. Also note a 91 . as a result of (3.
02ε0 .9 1 (e) εc = 0.8 0.5 γ = a/b 0. µc = µ0 6 10 5 0 (d) εc = 0.1 0.005ε0 .1 0.1 0.2ε0 .7 0.2 0.2 0. µc = µ0 (f) εc = 0.1 TE 0.2 0.3 0.7 0.03ε0 .9 1 −30 0.8 0.2 0.3 0.9 1 (c) εc = 0.2 0.4 0.5 0. Dashed line shows the uncoated PEC case.4 0.3: Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T E z polarization case. µc = µ0 Figure 3. with radius a.3 0.7 0. µc = µ0 4 2 σTE/λ0 (dB) 0 σ /λ0 (dB) −5 −10 −15 −20 −2 −4 −25 −6 0.5 0.4 0.7 0. 92 .3 0.7 0.6 0.1ε0 .7 0.01ε0 .4 0.6 0.4 2 0 −2 4 2 0 −2 σ /λ0 (dB) −4 −6 −8 −10 −12 −14 −16 0.9 1 −6 0.1 0.8 0.4 0.8 0.3 0. µc = µ0 6 0 4 −5 2 σ /λ0 (dB) σTE/λ0 (dB) −10 0 TE −15 −2 −20 −4 −25 0.5 γ = a/b 0.2 0.8 0.8 0. versus the corecoating ratio for coatings with diﬀerent constitutive parameters.6 0. µc = µ0 5 (b) εc = 0.3 0.6 0.4 0. The outer radius of the coating is selected as (a)(c) b = λ0 /2.9 1 σ /λ0 (dB) −4 −6 −8 −10 −12 −14 −16 0.1 0.6 0.9 1 TE γ = a/b TE γ = a/b (a) εc = 0.5 γ = a/b 0.5 γ = a/b 0.6 0. (d)(f) b = λ0 .
9 1 TE γ = a/b TE γ = a/b (c) εc = −0.7 0.4: Normalized monostatic echo width of an ENG coated PEC cylinder for the T E z polarization case.7 0.3 0. with radius a.923ε0 .2 0.923ε0 . Dashed line shows the uncoated PEC case.9 1 TE 0.7 0.4 0.5 γ = a/b 0.9 1 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 −90 0. µc = µ0 10 0 −10 −20 10 0 −10 −20 (f) εc = −0.8 0. µc = µ0 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 0.8 0.5 γ = a/b 0.6 0.8 0.7 0.6 0. 93 . µc = µ0 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 −90 0.8 0.7 0.1 TE 0.3 0.3 0.4 0.105ε0 .9 1 (e) εc = −0.6 0.1 0.342ε0 .1 0.7 0.8 0.5 0.7 0. versus the corecoating ratio for coatings with diﬀerent constitutive parameters.2 0.2 0.2 0.9 1 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 −90 0.5 0. µc = µ0 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 0.9 1 TE 0.1 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 0.105ε0 .5 γ = a/b 0. The outer radius of the coating is selected as (a)(d) b = λ0 /100. µc = µ0 (h) εc = −0.6ε0 .6 0.4 0.2 0.8 0.3 0.4 0.2 0.2 0.10 0 −10 −20 10 0 −10 −20 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 −90 0.1 σ /λ0 (dB) −30 −40 −50 −60 −70 −80 0.6 0.3 0. µc = µ0 10 0 −10 −20 10 0 −10 −20 (d) εc = −0. µc = µ0 Figure 3.4 0.3 0.1 0.6 0.1 TE 0.5 0.9 1 (g) εc = −0. µc = µ0 10 0 −10 −20 10 0 −10 −20 (b) εc = −0.6 0.6 0.1 0.6ε0 .4 0.3 0.4 0.342ε0 .4 0.7 0.8 0.9 1 TE γ = a/b TE γ = a/b (a) εc = −0. (e)(h) b = λ0 /50.5 0.2 0.5 γ = a/b 0.8 0.3 0.
4(e) due to the quadrupolar terms. When the outer radius is increased to b = λ0 /10. quadrupolar and higher order modes emerge. However. 3.9. For this “quasimagnetostatic” problem. In this case. 3. For various γ values. Transparency can be obtained with coatings having large permeabilities in the absolute sense as seen in Figs. 3.5(h). Also. To see whether any transparency or scattering maximization condition can be obtained for the T M z polarization. 3. On the contrary.5(d) where b = λ0 /20. due to the increased size. as shown in Fig. the huge increase in the RCS of an ENG coated PEC cylinder is due to high resonance.6(a)3. As we have mentioned previously. Figs. when µc < 0. For the ENG coated cases. the peak due to the dipolar term is much more dominant and can be safely used to maximize RCS of objects. 3.23. 3. but simple cancellation.5(e)3. when Figs. eﬀects of other higher order terms can be observed from Figs. In summary. we also expect it not to be very sensitive to ohmic losses. Fig.4 we see high γ sensitivity. there would be high sensitivity to ohmic losses near 94 .43. For γ = 0. Since transparency condition is not a result of resonation. 3. 3. However. we have chosen εc = ε0 for convenience. in Fig.6 shows the existence of resonant modes which maximize the RCS considerably. it can be seen that RCS is not very sensitive to γ near the transparency point. transparency we have achieved using DPS coatings is not a result of such resonance. These quadrupolar terms become more observable in Figs. This can be best observed from the changes in RCS with respect to γ. 3.5 suggest that as the electrical size of the scatterer increases the peak due to the dipolar term becomes wider and moves towards γ = 1.6(c). we calculate the monostatic echo widths when µc /µ0 is in the [−20 20] interval.5(a)3.3 are plotted in linear scale. however.second small peak which just emerges in Fig.6. transparency is possible if µc is positive and very large. we consider an electrically very small cylindrical scatterer with outer radius b = λ0 /100.
4 0.6 0.2 0.2 0.8 0. µc = µ0 10 0 −10 10 0 −10 (b) εc = −0.9 1 (c) εc = −0.8 0.3 0.6 0. versus the corecoating ratio for coatings with diﬀerent constitutive parameters. 95 .2 0.8 0. µc = µ0 σ /λ0 (dB) −20 −30 −40 −50 −60 0.4 0. µc = µ0 5 0 −5 −10 5 0 −5 −10 (f) εc = −0.6 0.1 0.7 0. µc = µ0 −10 σ /λ0 (dB) −15 −20 −25 −30 −35 −40 −45 0.4 0.7 0.6 0. The outer radius of the coating is selected as (a)(d) b = λ0 /20.1 TE 0.9 1 σ /λ0 (dB) −20 TE TE −30 −40 γ = a/b −50 0.3 0.105ε0 .9 1 TE 0.105ε0 .1 σ /λ0 (dB) −20 −30 −40 −50 −60 0.5 γ = a/b 0.7 0.9 1 (e) εc = −0.5 γ = a/b 0.6ε0 .1 TE 0. µc = µ0 5 0 −5 −10 0 10 (d) εc = −0.2 0.9 1 TE 0.2 0.2 0.5: Normalized monostatic echo width of an ENG coated PEC cylinder for the T E z polarization case.3 0.7 0.8 0.6 0.1 0. with radius a.923ε0 . Dashed line shows the uncoated PEC case.3 0.6ε0 . µc = µ0 (h) εc = −0.4 0.9 1 (a) εc = −0.3 0.5 γ = a/b 0.923ε0 .5 γ = a/b 0.4 0.1 σ /λ0 (dB) −20 −30 −40 −50 −60 0.5 0.4 0.6 0.4 0.3 0.8 0.2 0.8 0. (e)(h) b = λ0 /10.5 γ = a/b 0.9 1 TE γ = a/b TE γ = a/b (g) εc = −0.6 0.5 0.7 0.1 0.10 0 −10 10 0 −10 σ /λ0 (dB) −20 −30 −40 −50 −60 0.342ε0 .3 0.3 0.7 0.1 0.6 0.8 0.7 0.5 0.7 0.4 0. µc = µ0 Figure 3.9 1 σ /λ0 (dB) −15 −20 −25 −30 −35 −40 −45 0.342ε0 .2 0. µc = µ0 σ /λ0 (dB) −15 −20 −25 −30 −35 −40 −45 0.8 0.
are shown in Fig. the high sensitivity to ohmic losses can be seen clearly at the resonance location in Fig. Fig.7(b).7.7(a).5 0 −5 5 0 −5 /λ0 (dB) −10 −15 −20 −25 −30 −20 /λ0 (dB) −10 TM σ σ −15 −20 −25 −20 −15 −10 −5 µc/µ0 0 5 10 15 20 TM −15 −10 −5 µc/µ0 0 5 10 15 20 (a) γ = 0. 3. As predicted. 3.6: Normalized monostatic echo width of a metamaterial coated PEC cylinder for the T M z polarization case.5 /λ0 (dB) −6 −8 −10 −12 −14 −16 −18 −20 −15 −10 −5 0 5 10 15 20 /λ0 (dB) −4 −6 −8 −10 −12 −14 −16 −20 −15 −10 −5 0 5 10 15 20 TM σ µc/µ0 σ TM µc/µ0 (c) γ = 0.2 2 0 −2 −4 4 2 0 −2 (b) γ = 0. 3.7 (d) γ = 0.8 illustrates the bistatic scattering scenarios for transparency 96 . To visualize the farzone ﬁeld distribution in the xyplane. bistatic echo widths can be calculated. the resonant modes.. versus the coating permeability µc for diﬀerent corecoating ratios. On the other hand. we have considered the normalized monostatic echo widths (i. there is very little ohmic sensitivity for transparency condition in Fig. despite the decrease in the monostatic echo width due to the ohmic losses. when compared with the uncoated case. as in the Drude or Lorentz medium models. metamaterial coating provides at least approximately 65dB increase in the echo width at the resonance location.9 Figure 3. The eﬀects of small ohmic losses.7(b). In the numerical results we have shown up to here.e. back scattering). The outer radius of the coating is b = λ0 /100 and the coating permittivity is εc = ε0 . Again in Fig. 3. 3.
especially in the backscattering direction. µc = µ0 (b) εc = −0. which increases the RCS dramatically.9 1 −45 0 0. 3.6 − j0. µc = µ0 and γ = 0.8 0. εc = 0.6 − j0. Therefore. is 97 . this is the expected situation for transparency.2)ε0 5 0 −5 −10 εc = −0. µc = µ0 and γ = 0.7: Eﬀects of ohmic losses on normalized monostatic echo width for (a) DPS [transparency] (b) ENG [Scattering maximization] cases. the much larger portion will continue traveling in the direction of incidence. 3.9(b).9(a) and Fig. s i Hz + Hz ) in the presence of single PEC cylinder.3 0.1)ε0 εc = (0. In Fig.6ε0 εc = (−0. The angle of incidence is set to φ0 = 0◦ .6ε0 . µc = µ0 Figure 3.9 1 (a) εc = 0.6 − j0. 3. 3.01)ε0 εc = (−0. for b = λ0 /100.6 0.5 0.6ε0 5 εc = (0.5 0. RCS is maximized in the backscattering and incidence directions.6ε0 .7 0.3 0. The outer radius of the coating is selected as b = λ0 /100.8(b).9(a) shows the contour plot of the axial component of the total magnetic ﬁeld (i. it is seen that RCS increases gradually from backscattering direction (φ = 180◦ ) towards direction of incidence (φ = 0◦ ). In other words.. while little portion of the incident wave is reﬂected back.505.6ε0 . for the values of εc = 0.6ε0 and µc = µ0 .8 0.2 0.6ε0 and µc = µ0 .8(a). and scattering maximization for the T E polarization considering a metamaterial coated PEC cylinder with b = λ0 /100.6 0.6ε0 .9(b) shows the decrease in RCS with the proposed metamaterial coating. Comparison of Fig.02)ε0 4 σ /λ0 (dB) σ /λ0 −15 −20 −25 −30 TE 3 2 −35 −40 1 0 TE 0 0. however it reduces towards φ = 90◦ . 3.41. the PEC cylinder is coated with a DPS metamaterial coating having b = λ0 /100. Indeed. The case for the resonant ENG coating. εc = −0.2 0. 3.1 0.1 0.7 0. but also in the direction of incidence.4 γ = a/b 0.4 γ = a/b 0.e. for εc = −0. RCS is not only maximized in the backscattering direction.6 − j0.6 x 10 −5 εc = 0. In Fig. In Fig. Fig. ﬁnally becoming eﬀectively zero in this direction. with radius a = λ0 /200.
6ε0 . γ = 0. µc = µ0 . shown in Fig.41 (b) εc = −0. 98 .8: Normalized bistatic echo widths for (a) DPS coated (b) ENG coated PEC cylinder for the T E z polarization case. The outer radius of the coating is selected as b = λ0 /100. The angle of incidence is φ0 = 0◦ .5 TE TE 4 3 2 σ /λ0 1 0. µc = µ0 .9(c).8 7 6 5 x 10 −7 2.505 Figure 3.5 2 σ /λ0 1. The ﬁeld distribution conﬁrms the strong resonance in the radiation of a ydirected electric dipole. 3.6ε0 .5 0 0 1 0 0 20 40 60 φ (Degrees) 80 100 120 140 160 180 20 40 60 φ (Degrees) 80 100 120 140 160 180 (a) εc = 0. γ = 0.
Outer boundaries of the coatings are shown by dashed lines (a = λ0 /200.06 −0.04 −0.08 0.9984 0 x/λ0 0.08 0.e.04 −0.02 y/λ0 0 −0.08 0.999 0.6ε0 .02 0.02 0.06 −0.02 y/λ0 0 −0.06 −0.04 0.02 0.08 −0.06 1.04 −0. µc = µ0 Figure 3. b = λ0 /100).6ε0 .9: Contour plots of axial component of the total magnetic ﬁeld (i.04 0.0.999 −0.06 −0.04 −0.06 0.06 0.9998 (b) εc = 0.08 −0.06 −0.9986 −0.08 −0.0002 1 0.04 0. (c) ENG coating.9994 0.08 −0.08 −0.0005 0.04 0.06 (c) εc = −0.06 −0.02 −0.04 0.08 −0. 99 .02 9 8 7 6 5 4 3 2 1 0 x/λ0 0. s i Hz +Hz ) outside the PEC cylinder when there is (a) No coating.9995 1 (a) No coating 0. (b) DPS coating.02 0.9985 0 x/λ0 0. µc = µ0 0.02 −0.04 0.06 0.04 0..02 −0. Plane wave illumination is along the +xaxis.9996 0.9992 0.06 0.04 0.9988 1.001 1.02 0.02 y/λ0 0 −0.
4 0.1 0. µc = µ0 Figure 3. however.6 0. 3. µc = µ0 10 0 −10 σTE/λ (dB) −20 0 −30 −40 θ0 = 90° −50 θ0 = 60° θ0 = 30° −60 0 0.7 0.Fig.10: Normalized monostatic echo widths for (a) DPS coated (b) ENG coated PEC cylinder for the T E z polarization.3 0.1 0. oblique incidence case.10 shows the preliminary results for the oblique incidence case.6ε0 . −40 −50 −60 −70 σ /λ0 (dB) −80 TE −90 −100 θ0 = 90° θ0 = 60° θ0 = 30° 0 0.6 0.5 0. The outer radius of the coating is selected as b = λ0 /100.9 1 −110 −120 γ = a/b (a) εc = 0. 100 .5 0.7 0.8 0.3 0.478ε0 . resonance condition is aﬀected very slightly.4 γ = a/b 0. It can be observed that.8 0.9 1 (b) εc = −0. the angle of oblique incidence changes the transparency condition.2 0.2 0.
g..1 Homogenization of Metamaterial Structures and Retrieval of Eﬀective Constitutive Parameters 4. With their unnatural behavior.Chapter 4 Retrieval of Homogenization Parameters 4. Bloch’s Theorem. In engineering. these structures are highly exploited for engineering purposes. Photonic and electromagnetic band gap materials. the use of periodic alignment can also be seen in antennas. frequency selective surfaces and yet metamaterials are some artiﬁcial structures which make use of the periodicity. optical 101 .1 Introduction The physical properties of matter together with the underlying mathematics (e.1. LyapunovFloquet Theorem) lead to extraordinary phenomena when structures are aligned periodically. cMUTs.
As an example. for an array of antennas an array factor can be deﬁned. most of the time the contribution of the interactions between these antennas cannot be easily neglected. For the analysis and design of periodic materials and structures. However. For periodic materials. a rigorous method has to be formed to successfully represent the whole periodic structure. with its all intermediate steps. Once the response of a single building block is obtained. for accurate homogenization of periodic structures. the interactions between the many building blocks. their overall equivalents have to be calculated.gratings etc. to increase the overall performance of a system or to establish a predeﬁned task. which form the periodic structure. Similarly. obtaining the homogeneous equivalent from the basic building block of the material is obviously simpler and more towards the design of actual material of interest. The overall response of the antenna array can be calculated by multiplying this array factor with the response of a single antenna. To obtain accurate results. Especially for periodic structures. the building block of the periodic material is not canonical and a full wave analysis may be required to obtain the behavior of a building block itself. for periodic materials an equivalent homogeneous material can be deﬁned which exhibits the same properties with the material of interest. The process of obtaining this homogeneous equivalent. 102 . the overall structure can be modeled analytically from the results of the building block. Most of the time. full wave analysis of the system is necessary. must not be simply neglected. The homogenization processes present in the literature [18–22] are usually examples of such processes.. Hence. the periodicity of the structure and therefore the presence of periodic building blocks are of utmost importance. The method we introduce in this work is intended to accomplish this idea. to be able to practically incorporate them in larger systems. However. is called homogenization.
1. metamaterials are inhomogeneous. the homogenization process can be applied at a single direction (θ0 . A typical unit cell for a metamaterial is shown in Fig. However. Although the method is quite versatile and applicable to oblique incidence scenarios. Therefore the homogenization process for metamaterials is inherently anisotropic and dispersive. In general. one within another. 4. For metamaterials. a time. with gaps located at the opposite locations on these loops. φ0 ) and at a single frequency. For this reason. we will focus on the homogenization of metamaterials. anisotropic and highly dispersive materials.2 Homogenization of Metamaterials In this section. With the homogenization process we obviously remove the inhomogeneity. However. a wire and a substrate on which the SRR and wire are mounted. The unit cell is basically composed of a Split Ring Resonator (SRR). in this work we will consider the normal incidence case for a three dimensional metamaterial structure.1. A typical case is shown in Fig. the method we present here is not only restricted to metamaterials.2. 103 . The SRR is formed by two circular or rectangular loops. 4. such that the magnetic ﬁeld should be perpendicular to the SRRs and the electric ﬁeld should reside in the plane parallel to the SRRs. therefore it can be applied to any ﬁnite or semiinﬁnite periodic structure.4. this extraordinary behavior can be observed with the proper polarization of electric and magnetic ﬁelds with respect to the SRR structure. however the material maintains its anisotropic and dispersive state. the SRRs provide negative eﬀective magnetic permeability and the wires provide negative eﬀective electric permittivity. which are also called unit cells. Metamaterial Geometry: The building blocks for metamaterials are usually cubic cells.
104 . Figure 4.1: Metamaterial unit cell.Figure 4.2: Direction of E and H ﬁelds for a unit cell.
3 where a metamaterial of thickness d is placed in air and a plane wave is normally incident. Ny . y and z directions. The metamaterial medium (i. Let xmax . Let Nx . to avoid ambiguity.e. direction of the Poynting vector) for both right handed and left handed media. should refer to the direction of energy ﬂow (i. whereas the energy ﬂow will be again in the +z direction.3: Direction of E and H ﬁelds for a unit cell. and Nz denote the number of unit cells stratiﬁed in the x..2. Nx → ∞. for the TEM mode. 4. 4. respectively. 4..2.2 suggest the direction of propagation to be in the +z direction.e. 4. xmin . Nz Figure 4.. Medium 2) is composed of metamaterial unit cells depicted in Fig. 4. The metamaterial medium is assumed to be of inﬁnite extent in the transverse direction (i. Implementation of Boundary Conditions and Excitation: Consider Fig. However.In a right handed medium. Practically this is the case for Nx and Ny Nz . in a left handed medium the phase velocity will be in the −z direction. Ny → ∞. referring to Fig.e. the directions of E and H ﬁelds in Fig. ymax and ymin denote the four surfaces of a unit cell.). The direction of propagation.1 and Fig. If the four neighboring cells around a unit cell in the 105 .
ymin and ymax surfaces of the corresponding neighboring cells. and computationally more eﬃcient. Therefore xmax and xmin surfaces of a unit cell are identical to each other. Another. method has been suggested in [30] for the simulation of SRR+wire metamaterial structures.4. for their touching surfaces. Figure 4. Unit cells of the metamaterial structure are placed in a PECPMC waveguide and stratiﬁed in the z direction as seen in Fig. 4. This is due to the fact that unit cells are indistinguishable in the transverse direction. The PMC walls are parallel to the SRR structure and force the magnetic ﬁeld to be perpendicular to themselves and also to the SRR structure.transverse direction are considered. Hence. xmax .4: Alignment of unit cells inside the PECPMC waveguide. ymax and ymin surfaces of the unit cell at the center are identical to xmin . xmax . xmin . The PEC walls are perpendicular to the SRR structure such 106 . a periodic boundary condition for the xmax and xmin surfaces with zero phase and another periodic boundary condition for the ymax and ymin surfaces with zero phase can be used to simulate the periodicity in the transverse direction. respectively. as ymax and ymin surfaces of the unit cell are identical to each other.
whereas their y = a/2 and y = −a/2 surfaces are applied the PMC 107 . The electric and magnetic ﬁelds forced by the PECPMC waveguide are in full accordance with the plane wave polarization seen in Fig. the x = a/2 and x = −a/2 surfaces of Medium 1.. The problem geometry is depicted in Fig. and we implemented the aforementioned PECPMC waveguide method.3.. In our computer simulations. The metamaterial unit cell we use is the same with the symmetric unit cell given in [21] except only that the SRR and wire structures are assumed to have zero thicknesses and are applied Perfect Electric boundary condition (i.5. 4. 4. such that d = Nz a. With the coordinate system given in Fig.5mm is the unit cell size. The thickness of the metamaterial structure d depends on the number of unit cells stratiﬁed in the z direction Nz . 4. Medium 2 and Medium 3 are applied the PEC boundary condition.5.5: Problem geometry (crosssection view. treated as PEC) to reduce the required memory and the computation cost. for Nz = 3). Figure 4. we used High Frequency Structure Simulator (HFSS) of Ansoft Inc. which is a Finite Element Method (FEM) based electromagnetic simulator. where a = 2.that the electric ﬁeld becomes parallel to the SRR structure.e.
2L + d) respectively. [total electric ﬁeld − incident electric ﬁeld] in Medium 1. 4. It should be noted that the electric ﬁeld plotted in Fig.5. Since we use the 108 . we can deﬁne a Generalized Reﬂection Coeﬃcient (GRC) at the interface between Medium 1 and Medium 2 as given in [31]. 1 − Γ23 Γ21 e−j2kz2 d −j2kz2 d (4.1) where kz2 is the wave number in z direction in Medium 2. respectively. kz2 = k2 . in accordance with the PECPMC waveguide method explained previously. (4. x component of the electric ﬁeld (i.5 is the scattered electric ﬁeld (i. Note that. where it originates. 2 and 3). 0) and (0.. 0. The two ends of the geometry (i. Ex ) is measured on the line passing through the centers of the unit cells. 4.. z = 0 and z = 2L + d surfaces) are applied the Radiation Boundary Condition. For the normal incidence case. Γin (z = L) = Γ12 + T12 T21 Γ23 e−j2kz2 d + T12 T21 Γ21 Γ2 e−j4kz2 d + · · · 23 = Γ12 + T12 T21 Γ23 e .e. Extraction of Electric Field Data: After the simulation. A typical magnitude plot of the x component of the electric ﬁeld is shown in Fig. The begin and end points of this line are (0. where L = 3cm is the length of Medium 1 and Medium 3.. Homogeneous Equivalent : If the metamaterial medium (Medium 2) can be successfully represented with its homogeneous equivalent. 4. 0.e. The structure is illuminated with a plane wave which originates at z = 0 surface. The plane wave has magnitude 1 and phase 0 at the z = 0 surface.boundary condition.1) is valid for the more general oblique incidence case. Γij and Tij are the direct reﬂection and transmission coeﬃcients at the interface between layers i and j.5. with its polarization and propagation direction as shown in Fig.e.
05 0.5.06 0.6 Medium 1 1.01 0.03 z (m) 0. the transmitted ﬁeld travels in the +z direction. Nz = 1).8 0. Total Least Square Prony’s Method and Singular Value Decomposition 109 . are TEM waves. However. the y and z components of the electric ﬁeld are nearly zero.e.04 0.2 1 Ex GPOF Method is implemented here to find Γin (z = 0) 0.6 0. However these nonTEM modes decay fast. there are transition regions near the boundaries of metamaterial medium (i.02 0.6: Ex  vs. z (f = 10GHz. Obtaining the Reﬂection Coeﬃcients: GPOF Method [32] is one of the many methods used in approximating a complex function in terms of complex exponentials. The scattered ﬁeld in Medium 1 is basically a plane wave traveling in the −z direction. In Medium 3.07 Figure 4.4 0. Therefore the scattered and transmitted ﬁelds away from the metamaterial medium.4 Medium 2 Medium 3 1.2 0 0. Other methods used for this purpose are various forms of the Prony’s Method such as Least Square Prony’s Method. z 1. This is mainly due to discontinuities inside the PECPMC waveguide and mode conversions. 4.. Medium 2). in Medium 1 and Medium 3 respectively.PECPMC waveguide method. as seen in Fig. Ex vs.
Our simulation setup seems more reasonable and it is closer to a real life scenario. 2L/3] interval is selected by inspection because. The z = [0. In 110 . in the wave ports setup. In GPOF Method.1). for all frequencies of interest. GPOF Method is superior to aforementioned methods in its less noise sensitivity and computational eﬃciency [33]. the electric ﬁeld data in this interval does not overlap with the aforementioned transition region. However. maybe because of the aforementioned fast decaying nonTEM modes are still existent.). we have run simulations with diﬀerent number of modes for the waveguide (1. we use the following relation: S11 = Γin (z = L) = Γin (z = 0)ej2k1 L . . However.Prony’s Method. GPOF method basically removes the numerical noise in the data. Another method is the Pencil of Function Method which forms the basis of the Generalized Pencil of Function Method. Due to the nonuniform meshing of the geometry and numerical noise. basically. is veriﬁed to be −k1 . applying the GPOF method in this interval is more reliable than simply dividing the scattered ﬁeld at z = 0 to the incident ﬁeld at the same point (which is 1 + j0). obtained via GPOF method. 2L/3] interval and ﬁtting it by 1 exponential. . we could only use Medium 2 with two wave ports attached to its input and output surfaces to set up the excitations. S11 results of these simulations vary noticeably from our setup. The reﬂection coeﬃcient Γin (z = 0) in Medium 1 is found by applying the Generalized Pencil of Function (GPOF) method to the Ex ﬁeld component data in z = [0. To ﬁnd Γin (z = L) given in (4. where k1 is the free space propagation constant in Medium 1 (propagating in the +z direction). 2. a generalized eigenvalue problem is solved and subspace decomposition is employed.2) To ﬁnd the sparameters of the metamaterial structure. The propagation constant of the scattered wave in Medium 1. In this setup. . which is the S11 of the metamaterial structure. (4. the voltages or powers are calculated over the entire surfaces.
The Fresnel coeﬃcient can be easily calculated for Fig. as other terms except the Fresnel reﬂection term.7. In our work we have checked both methods and they gave the same result as expected.4) for Hy can be used. As we will present with the results. independent of whatever the wave will experience in the future. 4. In other words. but air. Fresnel reﬂection term is obviously expected at the interface of two homogeneous and diﬀerent media. Fresnel reﬂection occurs when electromagnetic wave passes from one medium to a diﬀerent medium.7.3) for Ex or (4. to ﬁnd the Fresnel reﬂection term at the interface of two media. ε2 kz1 + ε1 kz2 (4. the medium in which the wave is transmitted can be taken as semiinﬁnite and all layers beyond this medium can be neglected. µ2 kz1 + µ1 kz2 ε2 kz1 − ε1 kz2 . However the Fresnel reﬂection term will take two diﬀerent forms for the decoupled TE and TM modes [31]: ΓT E = 12 ΓT M = 12 µ2 kz1 − µ1 kz2 .3) (4. All contributions of these layers will be present in the GRC. is called the Fresnel reﬂection term. the ﬁrst term in (4.7 (b). 4. while diﬀering from Medium 2 because it does not go through the dielectric in the unit cell. instead of measuring the electric ﬁeld in only one line.our setup. The nice thing about the Fresnel reﬂection is its time causality. Therefore. 111 . another line which passes through the edge of the unit cells gives the same electric ﬁeld distribution in Mediums 1 and 3. at normal incidence (kz1 = k1 and kz2 = k2 ). Fresnel Reﬂection: Γ12 . either (4. Fresnel reﬂection term is just a result of discontinuity of the medium in which the wave travels. the Fresnel reﬂections for (a) and (b) are the same. which passes through the midpoints of unit cells. 4. If we consider Fig.1).4) For the TEM wave in Fig. Therefore. we could take diﬀerent parallel lines to this line and average the results.
Fitting the GRC with M exponentials is done using the GPOF Method. S11 is actually a function of d = Nz a.7: Fresnel reﬂection at (a) three layered media.5).Figure 4. Expressing S11 as Summation of Complex Exponentials: Let us rewrite the GRC equation in (4. N0 + N − 1 (4. 112 .5) As seen in (4. N0 + 1. . si ’s are the complex exponents. M is the number of exponentials to represent the GRC with truncating the inﬁnite series. Nz response of the metamaterial medium and we express it as a summation of complex exponentials: M S11 (Nz ) ≈ i=1 bi esi Nz a . In other words. . (b) two layered media.6) where bi ’s are the complex residues. we have the S11 vs. .1): S11 = Γ12 + T12 T21 Γ23 e−j2kz2 d + T12 T21 Γ21 Γ2 e−j4kz2 d + · · · . N0 is the initial number of unit cells stratiﬁed in the z direction and N is total number of unit cells used. 23 (4. Nz = N0 . In our method we vary Nz and record the S11 s correspondingly. .
a in (4.6): S11 (N0 ) ≈ b1 ejs1 N0 a + b2 ejs2 N0 a + · · · + bM ejsM N0 a . Therefore we should relate these complex residues and exponentials to each other. This is because GPOF method treats the index of the ﬁrst entry in a vector as zero.8) shows that bi = bi e−jsi N0 a . S11 (N0 + 1) ≈ b1 ejs1 1a + b2 ejs2 1a + · · · + bM ejsM 1a . S11 (N0 + N − 1) ≈ b1 ejs1 (N −1)a + b2 ejs2 (N −1)a + · · · + bM ejsM (N −1)a . in our case as N0 = 0. and each frequency from 5GHz to 15GHz (with 200MHz steps) S11 are calculated as explained previously.8). However. S11 (N0 + N − 1)]. .9) Now we have found bi and si correctly. and ﬁt it with M exponentials. S11 (N0 + N − 1) ≈ b1 ejs1 (N0 +N −1)a + b2 ejs2 (N0 +N −1)a + · · · + bM ejsM (N0 +N −1)a . for metamaterial stacks made up of Nz = 1. For each stack. respectively. . si = si . Comparison of two 113 . (4. 2.6). Referring to (4.a. . S11 . Comparison of (4.8) (4.7) When we apply the GPOF method to a vector such as: [S11 (N0 ) S11 (N0 + 1) .5) as a summation of complex exponentials expressed in (4.7) where N0 = 0 is the number of unit cells used as bias. we actually obtain bi and si . . . . the complex residues and exponentials in (4.7) can also be written as S11 (N0 ) ≈ b1 ejs1 0a + b2 ejs2 0a + · · · + bM ejsM 0a . .We have obtained the reﬂection coeﬃcients. our aim is to ﬁnd bi and si . (4.7) with (4. . where N0 ≥ 1 but not necessarily N0 = 1 and N0 + N − 1 = 20. (4. . 20 unit cells. . to approximate S11 given in (4. . . S11 (N0 + 1) ≈ b1 ejs1 (N0 +1)a + b2 ejs2 (N0 +1)a + · · · + bM ejsM (N0 +1)a .
equations term by term shows that: b1 = Γ12 . 114 (4. s1 = 0.12) −j2(Re{kz2 } + jIm{kz2 })a = (Re{s2 } + jIm{s2 })a − j2πm. such that the kz2 . kz2 . The following derivations explain the reduction of the complex propagation constant.14) (4. . . Reduction to the Reduced Brillouin Zone: Consider the second and third complex exponentials of the GPOF approximation in (4. s2 = −j2kz2 .13) (4. (4.a product calculated using each exponent is the same and remains in the reduced Brillouin zone [−π. to the reduced Brillouin zone. . e−j4kz2 a = es3 a e−j2πn = es3 a−j2πn .16) .11) (4.15) (4. should be the same. π]. −j4(Re{kz2 } + jIm{kz2 })a = (Re{s3 } + jIm{s3 })a − j2πn. −j2kz2 a = s2 a − j2πm. s3 = −j4kz2 .10). . . b3 = T12 T21 Γ21 Γ2 . b2 = T12 T21 Γ23 . 23 . s1 is the complex exponential corresponding to the Fresnel reﬂection term and should be zero. (4. e−j2kz2 a = es2 a e−j2πm = es2 a−j2πm . . s3 = −j4kz2 . found from each of the complex exponential terms si . Then. where m. . . can be used to calculate kz2 . i ≥ 2. −j4kz2 a = s3 a − j2πn.10) which means b1 is the Fresnel reﬂection term. s2 = −j2kz2 . The propagation constant kz2 . assuming M > 2. n are integers.
1 −2Re{kz2 }a = Im{s2 }a − 2πm → Re{kz2 }a = − Im{s2 }a + πm. π]. In our method we used Ex component of the electric ﬁeld.3): µr = (1 + Γ12 )kz2 . 2 1 π −4Re{kz2 }a = Im{s3 }a − 2πn → Re{kz2 }a = − Im{s3 }a + n. either (4.19) (4. n in (4. i = 2.18) (4.25) .20) (4. (4. k2 = ω µ2 ε2 = 2πf µr εr µ0 ε0 = c εr = c 2πf 2 2 2 (kt2 + kz2 ) (4. therefore utilizing (4.3) for TE polarized Ex or (4. 4 2 1 2Im{kz2 }a = Re{s2 }a → Im{kz2 }a = Re{s2 }a. 2 1 4Im{kz2 }a = Re{s3 }a → Im{kz2 }a = Re{s3 }a. 4Im{kz2 }a − j4Re{kz2 }a = Re{s3 }a + jIm{s3 }a − j2πn.24) µr 115 .4) for TM polarized Hy can be used to ﬁnd the eﬀective relative µr or εr of the homogeneous medium respectively.21) (4. Since 2πf √ √ √ √ µr ε r . Therefore.17) (4.22) The integers m.2Im{kz2 }a − j2Re{kz2 }a = Re{s2 }a + jIm{s2 }a − j2πm. for TEM polarization at normal incidence.20) are selected such that the Re{kz2 }a product calculated using each exponential si . i > 3. 4 (4. Finding the Eﬀective Constitutive Parameters: After Γ12 and kz2 have been found out.19)(4. The procedure is similar for exponentials si . 3 is the same and remains in the reduced Brillouin zone [−π. (1 − Γ12 )kz1 (4.23) √ where kz1 = k1 = ω µ0 ε0 is the wave number in Medium 1 (air).
Speeding up the process: Although the homogeneous equivalent is much simpler than the metamaterial structure. relative permeability and magnetic loss tangent.28) (4. due to the size of problem. Therefore. Delta H = 0 A/m. Magnetic Saturation = 0 T Lande G factor = 2. It is worthwhile to mention that µr (ω) and εr (ω) are complex quantities.where kt2 = kt1 = k1 sin θi is the wave number in transverse direction in Medium 2.27) (4. kz2 = k2 for normal incidence. electric permittivity and magnetic permeability of a material cannot be directly assigned complex numbers. the obtained µr and εr values are frequency dependent and can be better written as µr (ω) and εr (ω). dielectric loss tangent. we can replace the metamaterial structure with its homogeneous equivalent. Since the metamaterial structure is dispersive. Other properties of the material are left at their defaults: Bulk Conductivity = 0 S/m. In HFSS. sin θi being the angle of oblique incidence. θi = 0◦ . its homogeneous equivalent is also dispersive. However.29) µr (ω) = µr (ω) − jµr (ω) tan δm (ω) = Both positive and negative real numbers can be assigned to relative permittivity.26) (4. relative permittivity (εr ) and dielectric loss tangent (tan δd ) together with relative permeability (µr ) and magnetic loss tangent (tan δm ) can be used alternatively. it takes 116 . εr (ω) = εr (ω) − jεr (ω) tan δd (ω) = εr (ω) εr (ω) µr (ω) µr (ω) (4. Replacing the Metamaterial with its Homogeneous Equivalent: Once µr and εr of the homogeneous equivalent for the metamaterial structure have been obtained.
On the other hand. The variation of physical length along this vector corresponds to (N −1)a. their homogeneous equivalents are also expected to be highly dispersive. If a homogeneous slab is considered at a single frequency. to approximate the generalized reﬂection coeﬃcient successfully.considerable amount of time to run the geometry in HFSS and to obtain the electric ﬁeld data and reﬂection coeﬃcient (in the order of several hours). suppose the following vector is used in the GPOF method. such that the samples of the vector do not reside very close to each other and hence cause singularity. . Optimization Algorithm: Since metamaterials are highly dispersive. the number of exponentials to successfully represent the generalized reﬂection coeﬃcient is expected to change from one frequency to another. Therefore. we used the slab problem of Section 2.2. in very good agreement with the HFSS results and in seconds. kz2 (N − 1)a. sometimes rapidly. the eﬀective constitutive parameters of the slab play an important role on the number of exponentials to be used. . This allowed us to build an eﬃcient optimization algorithm. the reﬂection coeﬃcient of the slab as well as electric ﬁelds in Medium 1. Usually. The electrical length variation for the same vector should be suﬃciently large. as in Discrete Complex Image Method (DCIM). To solve this problem. Since the PECPMC waveguide method enforces TEM wave propagation in the normal direction. using a slab with constitutive parameters µr and εr . 117 . at a single frequency: [S11 (N0 ) S11 (N0 + 1) . Hence. the eﬀective constitutive parameters of the homogeneous equivalent change with frequency. based on the eﬀective constitutive parameters of the homogeneous equivalent at that frequency. 2 and 3 are obtained. S11 (N0 + N − 1)]. the electrical path length variation along the vector. is expected to be larger than π for successful approximation with complex exponentials.
. When a third unit cell is also added. Number of unit cells used as bias: N0 . the electromagnetic response of the metamaterial slab will change abruptly. If another unit cell is added. Therefore. In our optimization code we follow these steps: 118 . in the homogeneous state.Another aspect of the homogenization problem in metamaterials is the highly resonant properties of the unit cells which build the metamaterial structure. Also. 2. the interaction of the newly added unit cell with the very ﬁrst unit cells (at the other end of the stack) will be quite weak. When the unit cells are stacked. we have developed an optimization algorithm which ﬁnds the optimum (N0 . the generalized reﬂection coeﬃcients used in the GPOF vector should begin from a suﬃciently large number of unit cells. The wire and split ring resonators inside the unit cell of a metamaterial cause electric and magnetic resonances. the interaction of the new unit cell with its neighbors should be at the same amount as in the case of previously added unit cell. Number of complex exponentials: M . using generalized reﬂection coeﬃcients for this structure will contaminate the retrieved constitutive parameters. the interaction of a newly added unit cell is expected to be dominant only with the unit cells in its neighborhood. . In summary. the mutual interactions between these unit cells are very high. which are very dominant near their resonant frequencies. Now. which we have deﬁned previously as N0 : number of unit cells used as bias. 2. Therefore. the two important parameters in the homogenization process are: 1. but less abruptly. M ) combinations using the S11 data obtained from stacks made up of Nz = 1. after some point. consider only one unit cell. . Therefore. 20 unit cells. To sum up. the metamaterial slab can be said to be homogeneous when it has suﬃciently large number of unit cells. As unit cells are added. the response is also expected to change. . This also means that if the metamaterial structure is not acting homogeneous. when another unit cell is added.
Sort si  in an increasing manner. .19) and (4. .9). Reindex si and bi vectors in the same sequence with the sorted si . Drop b1 and s1 from bi and si vectors. Obtain kz2 from (4. respectively. vi. iv. Apply GPOF method to the vector and ﬁt with M exponentials. (s1 should be very close to zero. .i. v. Obtain (bi . µr ) obtain the reﬂection coeﬃcients of the homogeneous equivalents (i. si ). From Fresnel reﬂection coeﬃcient Γ12 and wave number kz2 . Reindex bi and si vectors in the same sequence with the sorted bi . x. iii. si ) using (4.) vii. Using (εr .23) and (4. . if homogenization is successful. π].e. homogeneous slabs with thicknesses N0 a. Sort new bi  vector in a decreasing manner. The entries of the si vector now correspond to s2 . which is a heuristic choice for obtaining the error in homogenization: 119 . . ix. . viii. Select N0 (from 1 to 17). si ) from (bi . . Obtain the Fresnel term from b1 .25). . obtain the eﬀective constitutive parameters µr and εr . Select M (from 2 to N 2 ). xi. .21). Obtain (bi .. xii. (N0 + 1)a. ii. This means that the length of the vector used in the GPOF method is N = 20 − N0 + 1. S11 (N0 + N − 1)]. (N0 + N − 1)a): [S11 (N0 ) S11 (N0 + 1) . s3 . Obtain the mean square error (MSE) for the last 4 stacks. using (4. selecting m such that the Re{kz2 }a product remains the in the reduced Brillouin zone [−π.
it is observed that two terms in the GPOF approximation are dominant: 1) the ﬁrst term which gives the Fresnel term 2) the second term which gives the propagation constant.e. DPS). The higher order terms of the GPOF approximation are negligible. The propagation constant has a large and negative imaginary part.MSE = S11 (N0 + N − 4) − S11 (N0 + N − 4)2 +S11 (N0 + N − 3) − S11 (N0 + N − 3)2 +S11 (N0 + N − 2) − S11 (N0 + N − 2)2 +S11 (N0 + N − 1) − S11 (N0 + N − 1)2 (4. If more than 2 exponentials are used at these frequencies. For the frequencies at which the homogeneous equivalent is SNG. After the optimization algorithm. DNG). both constitutive parameters are positive (i. which is related to the evanescent wave behavior of SNG metamaterials. In the frequency band. and the wave inside the metamaterial structure decays rapidly.. the optimization algorithm may erroneously select the number of exponentials M to be more than 2.. and for the rest of the frequencies they have alternative signs (i. which may yield a better Fresnel coeﬃcient for minimizing the MSE. Interestingly.e. Here it should be noted that the optimization scheme and the LMS algorithm are dependent only on the reﬂection data.30) xiii.. but the second exponential term is aﬀected. Find the optimum (N0 . following the steps ixii for diﬀerent possible choices of N0 and M .e. at some frequencies. the Fresnel term more or less stays the same. SNG). both constitutive parameters are negative (i. M ) pair in the least mean square (LMS) sense. This is because 120 . and they are much more dependent on Fresnel reﬂection than they are on the propagation constant. the ﬁnal constitutive parameters of the homogeneous equivalent are obtained. at some frequencies.
because of the aforementioned reasons. S11 .3 Numerical Results Following the optimization procedure. There is a very good agreement between the scattering parameters over the frequency band.1 as examples of three diﬀerent situations: the SNG case at f = 5GHz. and therefore optimization. because there should be 2 dominant terms in the GPOF approximation.9. The exponential approximations are tabulated in Table 4. the number of bias unit cells are selected as N0 = 1.8.the Fresnel coeﬃcient is the most dominant factor in the generalized reﬂection coeﬃcient. 4. are compared with those of the metamaterial structure. After the homogeneous equivalent is obtained. and the number of exponentials is selected as M = 2. the eﬀective constitutive parameters of the homogeneous equivalent are obtained as in Fig. 4. For the SNG cases.8GHz and the DPS case at f = 15GHz. for the rapidly decaying wave to be able to bounce back. 4. and correction of the frequencies at which the metamaterial is SNG. since the metamaterial unit cell and total thickness of the metamaterial are very small.1. its scattering parameters. For SNG cases. we overrule the ﬁndings of the optimization algorithm. 121 . in Fig. the DNG case at f = 10.
5 0 −5 Effective Permittivity −10 −15 −20 −25 −30 Real part of permittivity Imaginary part of permittivity −35 5 6 7 8 9 10 11 Frequency (GHz) (a) 12 13 14 15 4 3 2 Effective Permeability 1 0 −1 −2 −3 Real part of permeability Imaginary part of permeability −4 5 6 7 8 9 10 11 Frequency (GHz) (b) 12 13 14 15 Figure 4. (b) µr . 122 .8: Eﬀective homogenization parameters of the metamaterial over the 5GHz . (a) εr .15GHz frequency band.
8 0. frequency.3 0. obtained from the metamaterial and its homogeneous equivalent.1 0 5 Metamaterial Homogeneous equivalent 6 7 8 9 10 11 Frequency (GHz) 12 13 14 15 (a) 4 3 2 1 ∠ S11 0 −1 −2 −3 Metamaterial Homogeneous equivalent −4 5 6 7 8 9 10 11 Frequency (GHz) (b) 12 13 14 15 Figure 4.2 0.6 S11 0. 123 .5 0.4 0.1 0.7 0.9: S11 vs.9 0.
8 × 10 −0.782 −2 −3 −1.206 + j2.1: Parameters of the GPOF approximation f = 5GHz i bi si a 1 −0.8 × 10 ≈ 0. by reducing them to the reduced Brillouin zone.5 × 10 − j0.28 0. because of the explained reason.5 × 10 − j1.11 + j0. since their corresponding coeﬃcients are relatively very small. For f = 10.0 + j0. while higher order exponents may give diﬀerent results.047 −4 −3 −6. s1 ≈ 0 + j0 for all of the frequencies given in Table 4.69 −0.389 −3 3.Table 4.1 −0.4 × 10 + j1.7294 i 1 2 3 4 5 f = 10.90 + j0.679 −3 −3 4.041 − j1.0 0.8GHz bi si a Re{k2 }a −0.423 −2 −3 2.025 − j2. are close to each other. 124 .673 −4 0.27 + j9.35 0.09 ≈ 0.1 × 10 + j1.456 f = 15GHz bi si a Re{k2 }a −3 −0.87 −3.27 0.029 − j1.63 −0.148 − j2.500 i 1 2 3 4 5 6 As expected.43 ≈ 0.3 × 10 − j5.0 0.0 2 −1.22 − j1.4 × 10 − j1. The products obtained from the higher order exponents.015 −0.0 + j0. for f = 15GHz the Re{k2 }a product obtained from exponents s2 .46 Re{k2 }a 0. however.0 + j0. may not yield close results.10 + j1.59 −0.25 − j0.053 + j1.35 − j0. s3 and s4 are very close to each other.10 − j2.73 − j1.9 × 10 −0. Similarly.1.33 + j1.72 0.7 × 10 −0.34 + j0.4 × 10 −0. the Re{k2 }a product obtained from exponents s2 and s3 .04 −0.1 × 10 −0.87 −0. comparison of the Re{k2 }a products obtained from the ﬁrst exponents of the GPOF approximation can be used as a measure of quality of homogenization.3 × 10 −0. In summary.09 1.669 −3 −3 −1.8GHz.
2 1. 0] reduced Brillouin zone for the DNG case at f = 10. 4. the ﬁeld distribution along the actual structure (i.8 0. whereas Re{k2 }a product is in the [0.8GHz..6 0..Note that the Re{k2 }a product is in the [−π.e.10: Magnitude of Eﬁeld inside and outside the metamaterial medium and its homogeneous equivalent at f = 5GHz.2 1 0.e.104.4 0.2 0 0 1 2 3 4 5 6 z (cm) 7 8 9 10 11 Metamaterial (Center) Metamaterial (Edge) Homogeneous equivalent Figure 4.4 Magnitude of E−field 1. metamaterial medium together with the air media surrounding it) is compared with the case where metamaterial medium is replaced with its homogeneous equivalent.8 1. To better assess the quality of homogenization. The results for the frequencies of Table 4.1 are given in Figs. π] reduced Brillouin zone for the DPS case at f = 15GHz.6 1.11 (when Nz = 20 i. d = 5cm). 125 .
4 Metamaterial (Center) Metamaterial (Edge) Homogeneous equivalent 1. 126 .2 0 0 1 2 3 4 5 6 z (cm) (a) 7 8 9 10 11 1.6 0.2 0 0 1 2 3 4 5 6 z (cm) (b) 7 8 9 10 11 Figure 4.6 0.8 0.4 Metamaterial (Center) Metamaterial (Edge) Homogeneous equivalent 1. (b) f = 15.2 1 Magnitude of E−field 0.0GHz.4 0.4 0.1.8GHz.11: Magnitudes of Eﬁeld inside and outside the metamaterial medium and its homogeneous equivalent at (a) f = 10.8 0.2 1 Magnitude of E−field 0.
The ﬁeld inside the metamaterial structure. a major portion of the medium is air.4 Conclusion In this section.. a simple and versatile method for retrieval of the homogenization parameters of periodic structures is proposed. present in the literature. the SRRs and the wire). The agreement in the transmitted ﬁeld in Medium 3 is a sign of the success in homogenization. based on the unit cell geometry. hence their coupling eﬀects are minimized. since our method is based on reﬂection data and our optimization process strongly forces homogeneous equivalent to mimic the reﬂection properties of the metamaterial. so that the homogeneous equivalent mimics the metamaterial more successfully in the transmission region. As the future work. compared with its homogeneous equivalent. is tested in terms of agreement in sparameters. Therefore. Although the exact mechanism inside the metamaterial region is not known. passing through the centers of unit cells are close to zero and they are seen like noise. homogeneous equivalent is expected to resemble to the ﬁelds sampled at the edge. 4. Especially there is perfect agreement in Medium 1.e.The ﬁeld distributions show very good agreement between the metamaterial and its homogeneous equivalent in Medium 1 and Medium 3. For this reason. the ﬁeld distributions are also recorded on another line. reduction of Re{k2 }a products into the reduced Brillouin zone and agreement in ﬁeld distributions.1. the method can be modiﬁed to incorporate also the transmission data. 127 . This is mainly because they are in the vicinity of metallic scatterers (i. Hence. if the line is taken from the edge. which passes from the edges of the unit cells (through one of the PMC walls). The homogenization quality of the metamaterial. Numerical results show that the method is very successful to retrieve the eﬀective constitutive parameters of the metamaterial. The method is tested with a typical 3D metamaterial structure. we are at the furthest point from the SRR+wire combination.
2 Retrieval of Surface Wave Propagation Constants on a Grounded Dielectric Slab 4.12: Geometry of a grounded dielectric slab.12. 4. generally achieving less than 2% error. Fig. These complex geometries may include multi layered structures or periodically aligned metamaterial structures to create an artiﬁcial medium which may have negative eﬀective electric permittivity and/or negative eﬀective magnetic permeability. Numerical results of the method are in good agreement with the theory.1 Introduction In this work our aim is to build an eﬃcient and robust method to retrieve the surface wave propagation constants corresponding to each TM and TE mode that can propagate on the surface of a grounded dielectric slab. leaky waves and 128 . However. The twostep method we propose in this work consists of modeling and simulating the problem geometry in a Finite Element Method (FEM) based electromagnetic simulator and then processing the electric ﬁeld results obtained from the simulator to determine the surface wave propagation constants. The numerical results determined using our method are compared with their theoretical counterparts.4. there are some geometries and special cases that the method requires improvement. The importance of our proposed method lies in its ability to be further generalized and applied to complex geometries.2. Figure 4. Surface waves.
The length of the patch is L = 0. To the best of our knowledge. electronic band gap (EBG) and photonic band gap (PBG) structures have created a ﬂurry of interest among many researchers [34–77]. To create surface waves on the dielectric slab.13.evanescent waves related to metamaterials.0295λ0 .2 The TwoStep Method High Frequency Structure Simulator (HFSS) Simulations The ﬁrst step of our method is modeling and simulating the problem geometry in a FEM based electromagnetic simulator.2. HFSS recommends the radiation boundary to be located at least onequarter of a wavelength away from a radiating structure. The geometry of the rectangular narrow patch and constant current excitation is depicted in Fig.295λ0 and the width of the patch is W = L/10 = 0. the ground plane and the dielectric slab is assumed to be of inﬁnite extent in the x and y directions. HFSS provides a very useful tool to take into account the truncated parts of the geometry. For this purpose we use the High Frequency Structure Simulator of Ansoft Corporation. In theory. 4. we use a rectangular narrow patch at the surface of the slab and excite 1A constant current along the patch. where λ0 = 1cm at f = 30GHz. the inﬁnite geometry of the theory has to be truncated. For the accuracy of simulations. due to memory and computational restrictions. Expansion of our method to include these geometries and structures will therefore meet an important need. In our simulator. which in turn allows the waves to radiate inﬁnitely far into space. HFSS allows the user to select radiation surfaces and impose Radiation Boundary Conditions (RBCs) on these surfaces/boundries. 4. an eﬃcient method to determine the propagation constants related to these waves has not been reported yet. 129 . However.
130 . The space above the dielectric slab is ﬁlled with ideal free space (vacuum). Deﬁnitions: At this point it is useful to make some deﬁnitions. The outline of the entire problem geometry in HFSS is as given in Fig.13: Geometry of the rectangular narrow patch and excitation.Figure 4.14. As the dielectric substrate. z = th). x = 0 plane). y = 0 plane). −y width/2 ≤ y ≤ y width/2. we use a PEC plate just beneath the dielectric substrate. To simulate the inﬁnite ground plane. to which we will refer in the next sections. We impose the Perfect Electric (PE) boundary condition on this plate together with the Inﬁnite Ground Plane option being enabled. • Eplane: xz plane (i. 4. z = th).55.. • Hplane: yz plane (i. y = 0.e. • Eline: The line segment (−x width/2 ≤ x ≤ x width/2.e. The outer boundaries of the vacuum and the dielectric substrate (excluding the inﬁnite ground plane) are the radiation surfaces where Radiation Boundary Conditions are enforced. • Hline: The line segment (x = 0. we use a lossless dielectric material with dielectric constant εr = 2..
Generalized Pencil of Function (GPOF) Method Preliminaries: The asymptotic expansion (for large lateral distances. where the ˆ source is taken to be the origin. The Eplane and Hplane are deﬁned with respect to the orientation of the narrow patch. ρ) of the Green’s function of electric ﬁeld for xdirected ﬁlamentary microstrip dipoles. is formed as [78]: Z0 GE (ρ)∼ xx √ e−jk0 ρ Z0 εr − 1 tan2 (k0 d εr − 1) 2 sin φ + cos2 φ − jResW (βT M ) 2 2π εr − 1 ρ 2k0 εr 2 sin2 φ (2) βT M (2) (2) cos2 φ H2 (βT M ρ) − H0 (βT M ρ) − βT M H1 (βT M ρ) .14: Entire problem geometry in HFSS. · 2 ρ 131 . Eline and Hline reside in the Eplane and Hplane. respectively.Figure 4. Note that both Eline and Hline lie on the surface of the dielectric slab (z = th plane). As their names imply.
34) 2π ρ2 Z0 εr − 1 β2 (2) (2) − jResW (βT M ) T M H2 (βT M ρ) − H0 (βT M ρ) .and Hplanes. given in Table 4. GE given in (4.33) 2k0 εr 2 βT M sin2 φ (2) (2) (2) 2 × cos φ H2 (βT M ρ) − H0 (βT M ρ) − βT M H1 (βT M ρ) .35) Hence for large ρ.31) where Z0 = µ0 /ε0 is the intrinsic impedance of free space. 2 ρ In the Eplane GE becomes: xx GE (ρ.(4. φ = 0) ≈ xx Z0 e−jk0 ρ (4. 2k0 εr ρ (4.2.35) and (4. The ﬁrst term in (4. sin φ + cos2 φ 2π εr − 1 ρ2 (4. φ xx √ π Z0 tan2 (k0 d εr − 1) e−jk0 ρ = )≈ 2 2π εr − 1 ρ2 Z0 ε r − 1 βT M (2) + jResW (βT M ) H1 (βT M ρ).36) have the following characxx teristics in the E.32) The second term in (4. The direction of propagation makes an angle φ with respect to the positive xaxis measured towards the positive yaxis. d = th is the thickness of the dielectric slab. βT M is the propagation constant of the single proper TM pole and ResW (βT M ) is the residue corresponding to the TM pole (where the function W is as given in [78]).31) shows the contribution from the TM surface wave: − Z0 εr − 1 jResW (βT M ) (4. In our HFSS simulations we use an xdirected narrow patch with constant ˆ current excitation to simulate the ﬁlamentary microstrip dipole mentioned in 132 .31) gives the space wave term: √ Z0 tan2 (k0 d εr − 1) 2 e−jk0 ρ . 2k0 εr 2 In the Hplane GE becomes: xx GE (ρ.
k0 +1. where y(t) is shifted right by t0 = k0 δt. N − 1 (4. . we evaluate the Ex component of the electric ﬁeld xx on the surface of the dielectric substrate (along E.and Hlines).2: Space Wave and Surface Wave Characteristics in the E. In (4.36) where zi = esi δt and δt is the sampling interval. The shifted function y(t − t0 ) is sampled with N samples. Uniform sampling relates the samples k to the real variable t such that t = δtk. In accordance with GE . −2 −1/2 E ∼ρ k0 ∼ρ βT M −2 −3/2 H ∼ρ k0 ∼ρ βT M [78]. N − 1 (4. These samples can be represented by M complex exponentials as M M y[k] = i=1 bi e si δtk = i=1 bi zik . .Table 4. Now consider the case shown in Fig. 4. . si ’s are called the exponents. M M y[k−k0 ] = i=1 bi esi δt(k−k0 ) = i=1 bi zi (k−k0 ) .and Hplanes. . k0 +N −1 (4. y[N − 1] be the N uniform samples of a real variable t. bi ’s are called the residues. . .38) 133 . k − k0 = 0.15. as shown in Fig. The shifted discrete complex signal sequence y[k − k0 ] can be can be represented by M complex exponentials as in (4. y[1]. . Space Wave Surface Wave Plane Decay Prop. Decay Prop. . 1. . . 1.37) or equivalently. Const. . . The electric ﬁeld data we have obtained in HFSS is then used in the GPOF method to determine the surface wave propagation constants.36). M M y[k − k0 ] = i=1 bi e si δt(k−k0 ) = i=1 bi zi (k−k0 ) . . Const. . . . On the Generalized Pencil of Function Method for Surface Wave Constant Determination: Let y be a complex function. k = k0 . k = 0. and y[0]. to form y[k − k0 ].16. again δt being the sampling interval.36). 4.
39) i where β0 is the space wave propagation constant.15: Magnitudes of complex function y(t) and its N uniform samples y[k]. we expect the space wave term of the Ex component of the electric ﬁeld to decay with ρ−2 whereas the surface wave term is expected to decay with ρ−1/2 .Figure 4. A0 and Ai are the complex amplitudes of the space and 134 . ρ (4. Now let us assume that there are M total propagating TM and TE modes and they all decay with ρ−1/2 . Next sections will concentrate on the connection between the preliminaries and the formal deﬁnition of the GPOF Method. Application of the GPOF Method on the ELine: As it is presented in the preliminaries section. βSW are the surface wave propagation constants. Therefore the Ex component of the electric ﬁeld on the Eline can be written as a function of lateral distance ρ as follows A0 Ex (ρ) = 2 e−jβ0 ρ + ρ M i=1 Ai i √ e−jβSW ρ . for suﬃciently large lateral distances ρ on the Eline.
NE (ρ) can be assumed as a noise term which can be discarded by the GPOF Method.Figure 4. In such a case.16: Magnitudes of complex function y(t− t0 ) and its N uniform samples y[k − k0 ]. i=1 i (4. surface wave terms excluding decay dependence.41) can 135 .42) sampling interval δρ and number of samples N . Multiplying (4.39) with gives √ A0 ρEx (ρ) = √ e−jβ0 ρ + ρ ρ M √ ρ Ai e−jβSW ρ . (4.40) Now let us rewrite (4.40) as √ M ρEx (ρ) = i=1 Ai e−jβSW ρ + NE (ρ). assuming A0  Ai .41) where A0 NE (ρ) = √ e−jβ0 ρ . ρ ρ which comes from the space wave term contribution in (4. i (4. with proper choices of (4.40). For large lateral distances ρ.
17 shows the resemblance between √ ρEx (ρ) in (4. Let us assume that the equations and approximations in (4.17: Magnitudes of √ ρEx (ρ) and its N uniform samples y[k − k0 ].16. Figure 4. Fig. where ρ0 is the starting value of the lateral distance to be used in the GPOF Method.39)(4.be approximated as √ M ρEx (ρ) i=1 Ai e−jβSW ρ .43) are correct when ρ ≥ ρ0 .39) to be valid and for (4.38) to be used in representing (4. for (4.43) to be a correct approximation. i (4. 4.43) with M complex exponentials.43) and can be represented with M complex exponentials utilizing the GPOF Method. 4. 136 . the lateral distance ρ should be suﬃciently large. Intuitively this suggests (4. As it has been explained previously.43) for ρ ≥ ρ0 and y(t − t0 ) of Fig.
(4. whereas the right hand side is a discrete signal.46) (4. √ M ρEx (ρ) i=1 Ai e−jβSW ρ M i (4.47) Ai e−jβSW ρ = bi esi (ρ−ρ0 ) . k0 + N − 1. Expressing the RHS also as a continuous signal. βSW = 1. .9115 137 . which gives Ai = bi e−si ρ0 .2832 . k = k0 .7 − j0. i βSW = −Im {si } i i . . ≡ y[k − k0 ] = i=1 where ρ = δρk.45) (4. (4. Re {si } = 0.44) is a continuous signal.50) A0 1 ρEx (ρ) = √ e−jβ0 ρ + A1 e−jβSW ρ ρ ρ 1 Assume A0 = 0 . Ex (ρ) = √ A0 −jβ0 ρ A1 −jβSW ρ 1 e +√ e ρ2 ρ (4.48) Some Theoretical Examples (Eline): Example 1: Consider the following case along Eline. Ai e−jβSW ρ = bi e−si ρ0 esi ρ .17.3 s1 = 0 − j6.44) bi esi δρ(k−k0 ) . It is obvious that ρ0 = δρk0 in Fig. . A1 = −0. k0 + 1. b1 = 0.7 + j0.9115.1β0 = 6.Therefore. 4. .3 . β0 = 6. The left hand side of equivalence (4. we get M M Ai e i=1 i −jβSW ρ = i=1 bi esi (ρ−ρ0 ) . When √ ρEx (ρ) is used in the GPOF Method (in the [5λ0 − 8λ0 ] interval with N = 51 samples) the following results are obtained.49) (4.
3756 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0. βSW = 1. When √ ρEx (ρ) is used in the GPOF Method (in the [5λ0 − 8λ0 ] interval with N = 51 samples) the following results are obtained.0223 − j6.˜ A1 = b1 e−s1 ρ0 = −0.4198 ˜1 βSW = −Im{s1 } = 6. with the following assumptions A0 = (−0.3 1 β0 = 6. ˜1 ˜ A1 and βSW are retrieved exactly. n=0: b1 = 0.3047 s1 = 0.6628 − j0.2832 .0024 − j6.7 + j0.9115 Since A0 = 0.8886 ˜ A1 = b1 e−s1 ρ0 = −0.3 ˜1 βSW = −Im{s1 } = 6.6847 + j0.9115.3464 s1 = 0.6189 + j0.3087 138 . the noise term NE (ρ) becomes zero.9093 ˜ A1 = b1 e−s1 ρ0 = −0.5 − j0.4) × 10−n .8886 n=1: b1 = 0.7 + j0. Example 2: Consider Example 1 again. A1 = −0.6963 − j0.1β0 = 6.5537 + j0. Re{s1 } is zero as expected.
3 ˜1 βSW = −Im{s1 } = 6.9115 n=4: b1 = 0.9115 ˜ A1 = b1 e−s1 ρ0 = −0.3012 ˜1 βSW = −Im{s1 } = 6.3001 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.7 − j0.3005 s1 = 0.6993 + j0.6928 + j0.3001 ˜1 βSW = −Im{s1 } = 6.7 + j0.9113 ˜ A1 = b1 e−s1 ρ0 = −0.9113 n=3: b1 = 0.˜ A1 = b1 e−j Im{s1 }ρ0 = −0.6985 + j0.6999 + j0.6996 − j0.3 s1 = 0 − j6.9115 ˜ A1 = b1 e−s1 ρ0 = −0.7 − j0.6998 + j0.7 + j0.3009 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.3 s1 = 0 − j6.3124 ˜1 βSW = −Im{s1 } = 6.3 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.0002 − j6.9115 139 .9093 n=2: b1 = 0.
Let us again assume that there are M total propagating TM and TE modes and they all decay with ρ−3/2 . which leads A1 ˜ A1 ˜1 A1 and βSW 1 βSW . ρ which comes from the space wave term contribution in (4.53) 140 .52). When Re{s1 } = 0.In this particular example we see that when A0  is comparable to A1 . As n increases A0  ˜ A1 .51) Ai e−jβSW ρ . we again expect the space wave term of the Ex component of the electric ﬁeld to decay with ρ−2 whereas in this case the surface wave terms are expected to decay with ρ−3/2 . √ e ρ ρ (4. the modiﬁed ˜ A1 = b1 e−j Im{s1 }ρ0 formula yields better results for this purpose. (4. the contribution of the space wave term shows a noise eﬀect which cannot be easily ˜ discarded. Therefore the Ex component of the electric ﬁeld on the Hline can be written as a function of lateral distance ρ as follows A0 Ex (ρ) = 2 e−jβ0 ρ + ρ √ Multiplying (4.52) as √ ρ ρEx (ρ) = where A0 NH (ρ) = √ e−jβ0 ρ . In such a case Re{s1 } = 0.54) M M M i=1 Ai −jβSW ρ i .51) with ρ ρ gives A0 ρ ρEx (ρ) = √ e−jβ0 ρ + ρ √ Now let us rewrite (4. which makes the theoretical A1 = b1 e−s1 ρ0 complex amplitude retrieval formula wrong. i=1 i (4. For suﬃciently large lateral distances ρ on the Hline. i=1 i (4. Application of the GPOF Method on the HLine: The procedure for the Hline is very similar to the Eline case we have investigated in the previous section.52) Ai e−jβSW ρ + NH (ρ).
Under the conditions presented in the previous section. Some Theoretical Examples (Hline): Example 3: Consider the following case along Hline: Ex (ρ) = A0 −jβ0 ρ A1 1 e + √ e−jβSW ρ .57) (4. one can apply the same procedure in (4.44) where √ to be replaced with ρ ρEx (ρ).3 .58) A0 √ 1 ρ ρEx (ρ) = √ e−jβ0 ρ + A1 e−jβSW ρ . √ When ρ ρEx (ρ) is used in the GPOF Method (in the [5λ0 − 8λ0 ] interval with N = 51 samples) the following results are obtained.2832 .1β0 = 6. Re {si } = 0. ρEx (ρ) has On the other hand.42) decays with ρ−3/2 whereas NH (ρ) in (4.53) can be approximated as √ ρ ρEx (ρ) M Ai e−jβSW ρ .55) are correct when ρ ≥ ρ0 . √ (4.55) and can be represented with M complex exponentials utilizing the GPOF Method.44)(4. which means application of GPOF Method on the Eline is less noise sensitive compared to the Hline case. NH (ρ) can be assumed as a noise term which can be discarded by the GPOF Method. i βSW = −Im {si } . (4. ρ 1 Assume A0 = 0 . βSW = 1.9115.54) decays with ρ−1/2 . 141 .7 + j0. Again with the assumption that the equations and approximations in (4.48) and ﬁnd out that Ai = bi e−si ρ0 . A1 = −0. NE (ρ) given in (4. i=1 i (4.56) The only diﬀerence for the Hline case appears in (4. β0 = 6.51)(4. 2 ρ ρ ρ (4. In such a case.
n=0: b1 = 0.3 s1 = 0 − j6.7860 n=1: b1 = 0. A1 = −0. Example 4: Consider Example 3 again.5 − j0. ˜1 ˜ A1 and βSW are retrieved exactly.4) × 10−n .7 − j0. Note that the same results with Example 1 are obtained.6783 − j0.4925 − j0.9115 ˜ A1 = b1 e−s1 ρ0 = −0.1β0 = 6.7860 ˜ A1 = b1 e−s1 ρ0 = −0.7 + j0. Re{s1 } is zero as expected.5432 s1 = 0. with the following assumptions A0 = (−0.3 1 β0 = 6. βSW = 1. the noise term NH (ρ) becomes zero.7 + j0.7289 ˜1 βSW = −Im{s1 } = 6.9115.0449 + j0.9115 Since A0 = 0.1149 − j6.4104 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.0798 + j0.2832 .3 ˜1 βSW = −Im{s1 } = 6.3249 142 . √ When ρ ρEx (ρ) is used in the GPOF Method (in the [5λ0 − 8λ0 ] interval with N = 51 samples) the following results are obtained.b1 = 0.
3328 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.6978 − j0.3025 s1 = 0.6091 + j0.3004 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.9105 n=3: b1 = 0.3036 ˜ A1 = b1 e−j Im{s1 }ρ0 = −0.6990 + j0.9008 n=2: b1 = 0.s1 = 0.0017 − j6.3002 s1 = 0.6998 − j0.9115 ˜ A1 = b1 e−s1 ρ0 = −0.9114 ˜ A1 = b1 e−s1 ρ0 = −0.3 143 .6962 + j0.3606 ˜1 βSW = −Im{s1 } = 6.3061 ˜1 βSW = −Im{s1 } = 6.6996 + j0.7 − j0.9008 ˜ A1 = b1 e−s1 ρ0 = −0.9114 n=4: b1 = 0.3 s1 = 0 − j6.6999 + j0.9105 ˜ A1 = b1 e−s1 ρ0 = −0.0160 − j6.0002 − j6.3006 ˜1 βSW = −Im{s1 } = 6.6905 + j0.6600 + j0.
18 and Fig. We want to investigate the cases where 0 ≤ ρ ≤ 10λ0 . which means the size of the dielectric substrate must be 20λ0 × 20λ0 × th. we use diﬀerent geometries for Eline and Hline cases. which leads A1 ˜ A1 ˜1 A1 and βSW 1 βSW . dimensions of the substrate are x width × y width × th. As n increases A0  ˜ A1 . as expected. 4.19.2. For the Hline case. The problem geometries for the Eand Hline cases are shown in Fig. Also note that Example 2 for the Eline case works better than the Hline case examined here. On the other hand. When Re{s1 } = 0. For the Eline case. where x width < 20λ0 and y width = 20λ0 . respectively. which makes the theoretical A1 = b1 e−s1 ρ0 complex amplitude retrieval formula wrong. 4. the lateral distance ρ should be suﬃciently large such that the space wave contribution to the electric ﬁeld can be neglected compared with the surface wave contributions. 144 . where x width = 20λ0 and y width < 20λ0 . the modiﬁed ˜ A1 = b1 e−j Im{s1 }ρ0 formula yields better results for this purpose. In such a case Re{s1 } = 0. the contribution of the space wave term shows a noise eﬀect which cannot be easily ˜ discarded. 4.˜ A1 = b1 e−j Im{s1 }ρ0 = −0. To solve this problem.7 + 0. However this size is impossible to implement due to the aforementioned size restriction of HFSS. the size of the problem geometry cannot exceed a threshold size which is predeﬁned by the HFSS as a restriction.3 Implementation: To determine the surface wave propagation constants accurately. dimensions of the substrate are x width × y width × th.9115 In this particular example we see that when A0  is comparable to A1 .3001 ˜1 βSW = −Im{s1 } = 6.
{Re(Ex ). where ρ0 can be taken 2λ0 − 3λ0 intuitively. whereas for the Hline case √ Ex is multiplied with ρ ρ.e.18: Problem geometry for the Eline case. the electric ﬁeld data along the Eline or Hline are exported to a ﬁle (i. Re(Ez ).. Im(Ey ). 145 . Theoretically for both Eline and Hline cases we might use the [ρ0 ≤ ρ ≤ 10λ0 ] interval in the GPOF Method. When the simulation is complete. For the Eline case Ex is multiplied with ρ. However on the Eline or Hline ρ = 10λ0 corresponds to two of the surfaces where Radiation Boundary Conditions are set. Therefore the logical interval to be used in the GPOF Method will be [2λ0 − 3λ0 ≤ ρ ≤ 8λ0 − 9λ0 ]. Im(Ez )}). Im(Ex ). Re(Ey ).Figure 4. which will contaminate the results in their neighborhood. This ﬁle is processed and the necessary Ex component of the electric ﬁeld is √ formed. In the simulation there will be small reﬂections from these surfaces.
1λ0 . In the simulations. the surface wave propagation constant. which are calculated by solving the transcendental surface wave equations [79]. for diﬀerent intervals and using diﬀer1 ent number of samples.59) where βT M0 = 680. 4.2. βT M0 (4. βSW .3. The retrieval process is repeated for various thicknesses of the dielectric slab.Figure 4. therefore λ0 = 1cm.19: Problem geometry for the Hline case. The percentage error of the numerical results for TM0 mode is calculated as %Error = 1 βSW − βT M0 × 100.3. For some of the cases given in Table 4.084β0 is the analytically found surface wave propagation constant for the TM0 mode when th = 0.and Hline data are compared with their analytical counterparts. percentage errors are calculated: 146 . when the thickness of the slab is th = 0. HFSS results in the [ρstart − ρend ] interval with N number of samples are used.and Hline data. is retrieved from the E.55.4 Numerical Results Surface wave propagation constants retrieved from the E. In Table 4. In the GPOF Method. frequency is selected as f = 30GHz. The slab has a dielectric constant of εr = 2.1λ0 .87 = 1.
3 and percentage errors.22 Examination of Table 4. βT M0 = 749.04 {Eline .63 {Eline . ρstart = 5λ0 . ρend = 9λ0 } → %Error = 0. Especially Hline case gives better results for this particular thickness of the dielectric substrate. The retrieved propagation constants are given in Table 4.15λ0 . N = 51 .4. The errors for two cases can be found as: {Eline . ρend = 8λ0 } → %Error = 0. ρstart = 5λ0 . N = 51 .81 = 1. N = 51 . N = 51 . ρstart = 5λ0 .{Eline .5. But for the Hline case the results are again unsatisfactory. When the thickness of the slab is increased to th = 0.53 As it is observed from Table 4. ρstart = 5λ0 . Eline case gives very successful results. ρstart = 5λ0 . For two of the intervals and number of samples the percentage errors can be found to be: {Eline . N = 51 .32 {Eline . For th = 0. N = 51 . Eline case gives acceptable results.4 shows that. N = 51 .19λ0 . ρend = 9λ0 } → %Error = 0.193β0 and the results are tabulated in Table 4.32 = 1. both Eline and Hline cases give good results for most of the intervals and number of samples. ρstart = 5λ0 . ρstart = 5λ0 .5. ρend = 8λ0 } → %Error = 0. ρstart = 5λ0 .67 As it is seen from Table 4. ρend = 8λ0 } → %Error = 1. ρend = 8λ0 } → %Error = 0. ρend = 9λ0 } → %Error = 1. But for the Hline case the results are unsatisfactory. 147 .41 {Hline . N = 51 .282β0 .02 {Hline . the theoretical surface wave propagation constant increases to βT M0 = 805. ρend = 9λ0 } → %Error = 0.
One thing to note is that. the best results are obtained in the [4λ0 − 8λ0 ] and [4λ0 − 9λ0 ] intervals with N = 51 samples. For the TE1 mode.6. βT M0 = 640.20 for this mode. ρend = 8λ0 } → %Error(TM0 ) = 0. The analytically found surface wave propagation of these modes for th = 0.94 = 1.and Hline cases. which is TE1 . and the retrieved ones are tabulated in Table 4. We have also simulated a case. the surface wave propagation constants determined 148 . except from the TM0 mode. the surface propagation constant.15 {Eline .69 %Error(TE1 ) = 4. ρstart = 4λ0 . percentage error is calculated as in (4.7 are far away from being successful for both E.For th = 0. In this example.20 From Table 4. {Eline . N = 51 .02β0 .13 {Eline . ρstart = 4λ0 .25λ0 . The results for this mode for the given intervals and number of samples yield approximately the same results.6 it is observed that Eline case gives acceptable results for the TM0 mode. Along the Eline there are some problems for the TE1 mode. where the dielectric substrate is very thin. The results tabulated in Table 4. N = 21 .60 %Error(TE1 ) = 0. ρend = 9λ0 } → %Error(TM0 ) = 0.05λ0 .25λ0 are βT M0 = 868.383β0 and βT E1 = 669. For the worst case.63 = 1. the percentage error is 4. For this mode. ρend = 9λ0 } → %Error(TM0 ) = 0.59).06 = 1.89 %Error(TE1 ) = 0. ρstart = 4λ0 . is very close to the space wave propagation constant.065β0 . Along the Hline case the results are again unsatisfactory. another mode emerges. N = 51 . th = 0. among the given intervals and number of samples.
HFSS data and GPOF approximation are in very good agreement.using the GPOF Method are even smaller than the space wave propagation constant. 4. The possible reason of failure for this particular thin case is propagation constants of space and surface wave terms being very close to each other. Finally. the complex coeﬃcients and exponentials found in the GPOF approximation are used for generating the electric ﬁeld distribution along the Eline. Also notice that GPOF method removes the noise in the HFSS data very well.204. The last two subplots of the ﬁgures show the extrapolation of electric ﬁeld distribution using the previously found complex coeﬃcients and exponentials.23. 149 . the ﬁrst two subplots of the ﬁgures show the ﬁeld distribution inside the interval used for GPOF approximation. In Figs.
47 670.085 1.068 1. λ0 = 1cm.35 674.77 677.77 678.061 1.32 1 βSW /β0 1.068 1.92 671.92 671.067 1.68 673.51 671.84 671.08 670.69 668.067 1.069 1.30 669.72 671. (b) Hline.81 680.078 1.067 1.079 1.072 1 βSW 669.06 677.N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (a) ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (b) 1 βSW 670.065 1.080 1.99 1 βSW /β0 1.066 1.89 669.064 1.078 1.064 1.66 672.069 1.25 668.068 1. th = 0.072 1.073 1.069 1. εr = 2.16 671.08 670.55) 150 .068 1.39 671.37 681.00 673.70 677.065 1.071 Table 4.58 671.3: Surface wave propagation constants retrieved from (a) Eline.08 668.1λ0 .069 1.064 1.073 1. (f = 30GHz.33 674.98 666.084 1.21 668.065 1.92 667.063 1.065 1.
195 1.94 750.15 752.842 0.120 0.60 750.194 1.788 0.194 1.788 0.49 703.73 554.84 547.882 1.195 1.46 750.47 750.193 1.92 751.13 528.15 750.805 0.787 Table 4.862 0. εr = 2.96 515. (f = 30GHz.05 714.N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (a) ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (b) 1 βSW 750.194 1.4: Surface wave propagation constants retrieved from (a) Eline.195 1.63 1 βSW /β0 1.29 750.92 749.83 552.079 0.821 0. (b) Hline.77 692.61 751.194 1.195 1.15 551.33 541. th = 0.872 0.97 750.195 1 βSW 678.59 749.55) 151 .196 1.194 1.878 1.03 750.81 495.197 1.196 1.64 494.194 1.879 1.788 0.62 494.59 495.29 1 βSW /β0 1.193 1.25 750.75 749.86 505.138 0. λ0 = 1cm.103 0.15λ0 .
959 0. th = 0.298 1.N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (a) ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (b) 1 βSW 812.55 811.996 0.24 811.292 1.291 1.40 815.956 0. (f = 30GHz.97 811.97 815.75 604.02 1 βSW /β0 0.08 602.914 0.31 574.293 1.92 814.71 573.962 0. λ0 = 1cm.298 1.294 1.296 1.913 0.961 0.998 0.291 1.963 0.83 625.56 812.81 808.291 1.292 1.93 811.959 0.55) 152 .912 0.13 600.299 1.91 625.15 603. εr = 2.20 809.14 810.912 0.78 602.295 1.77 810.01 1 βSW /β0 1.993 0.291 1.291 1.961 0.19λ0 .22 604.5: Surface wave propagation constants retrieved from (a) Eline.288 1.959 0.286 1 βSW 573. (b) Hline.62 627.82 573.995 Table 4.64 624.44 811.91 815.13 602.48 603.51 813.
943 1.05 1.97 1.21 628.37 1.85 1.391 874.396 (a) 1 1 βSW βSW /β0 669.16 2 βSW /β0 1.11 1.936 0.986 1.058 1.061 1.394 881.045 1.072 638.46 1.080 Table 4.809 0.51 678.058 0.821 0.97 1.841 0.42 501.061 679.48 508.72 642.17 664.90 619.395 878.063 1.021 1.038 1.88 649.32 1.052 1.393 878.67 1.078 1.392 872.057 673.081 668.06 656.392 877.33 1.673 1.81 516.020 1.06 422.084 663.74 1.36 1.59 1.049 1.053 1.52 528.392 874. (f = 30GHz.69 664.079 666.045 1.50 1.33 587.59 1.24 1.392 874.71 1.51 677.39 1.28 1.001 1.98 662.87 659.74 652.60 1.90 1.717 0.81 1.065 680.798 0.063 1.70 656.399 876.054 2 βSW 531.004 686.90 640.403 874.38 668.63 666.86 592. λ0 = 1cm.68 1.25λ0 .398 876.016 693.394 878.100 630.104 632.107 631.007 695.005 691.088 0.066 677.038 1.14 1.33 1.59 1.72 1.13 683. εr = 2.69 668.37 1.31 450.40 2 βSW /β0 0.036 1.397 875. th = 0.84 652.023 1.68 651.47 660.846 0.55) 153 .98 1.6: Surface wave propagation constants retrieved from (a) Eline.N ρstart 101 4λ0 101 5λ0 51 4λ0 51 5λ0 26 4λ0 26 5λ0 21 4λ0 21 5λ0 101 4λ0 101 5λ0 51 4λ0 51 5λ0 26 4λ0 26 5λ0 21 4λ0 21 5λ0 N ρstart 101 4λ0 101 5λ0 51 4λ0 51 5λ0 26 4λ0 26 5λ0 21 4λ0 21 5λ0 101 4λ0 101 5λ0 51 4λ0 51 5λ0 26 4λ0 26 5λ0 21 4λ0 21 5λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 1 1 βSW βSW /β0 874.034 1.092 (b) 2 βSW 661.398 876.05 1. (b) Hline.75 1.17 1.389 877.20 641.
55) 154 . λ0 = 1cm.04 605.964 0.82 603.968 0.955 0.33 605.961 0. (b) Hline.958 0.953 0. εr = 2.33 602.962 0. (f = 30GHz.965 0.963 0.13 598.32 606.960 0.964 0.05λ0 .34 608.81 609.953 0.25 599.968 0.74 616.67 604.7: Surface wave propagation constants retrieved from (a) Eline.19 607.981 0.978 0.959 0.971 0.90 600.955 0.69 604.61 609.39 600.955 Table 4.11 603.57 600.50 599.72 606.78 614.971 0.965 0.957 0.967 1 βSW 610.84 598.50 1 βSW /β0 0.92 604.956 0.954 0.963 0.26 1 βSW /β0 0.48 614.37 597.04 602.40 608.979 0.43 601.965 0.96 600.951 0.960 0.13 604. th = 0.954 0.N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 N 101 101 51 51 26 26 21 21 101 101 51 51 26 26 21 21 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρstart 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 4λ0 5λ0 ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (a) ρend 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 8λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 9λ0 (b) 1 βSW 606.970 0.
5 6 6. ρend = 8λ0 .55.5 8 200 100 0 −100 −200 5 5.5 ρ/λ0 Magnitude 7 7.5 8 140 120 100 80 60 40 20 0 2 3 4 5 6 7 8 9 HFSS Ex √ρ GPOF b1es1(ρ−ρ0 ) ρ/λ0 Phase 200 100 0 −100 −200 2 3 4 5 ρ/λ0 6 7 8 9 Figure 4. (f = 30GHz.1λ0 . λ0 = 1cm.5 ρ/λ0 Phase 7 7.Magnitude 20 HFSS Ex √ρ 15 GPOF b1es1(ρ−ρ0 ) 10 5 0 5 5. N = 101) 155 .20: Comparison of GPOF approximation with HFSS data and its Extrapolation . ρstart = 5λ0 . εr = 2.5 6 6. th = 0.
th = 0.5 ρ/λ0 Phase 7 7.55.5 ρ/λ0 Magnitude 7 7.21: Comparison of GPOF approximation with HFSS data and its Extrapolation . (f = 30GHz. ρstart = 5λ0 .5 8 500 HFSS Ex √ρ 400 300 200 100 0 GPOF b1es1(ρ−ρ0 ) 2 3 4 5 ρ/λ0 Phase 6 7 8 9 200 100 0 −100 −200 2 3 4 5 ρ/λ0 6 7 8 9 Figure 4. N = 101) 156 . ρend = 8λ0 .5 8 200 100 0 −100 −200 5 5. εr = 2.15λ0 .5 6 6.5 6 6. λ0 = 1cm.Magnitude 200 HFSS Ex √ρ GPOF b1es1(ρ−ρ0 ) 150 100 50 5 5.
19λ0 .55. ρend = 8λ0 .22: Comparison of GPOF approximation with HFSS data and its Extrapolation .5 6 6.5 8 600 HFSS Ex √ρ 500 400 300 200 100 GPOF b1es1(ρ−ρ0 ) 2 3 4 5 ρ/λ0 Phase 6 7 8 9 200 100 0 −100 −200 2 3 4 5 ρ/λ0 6 7 8 9 Figure 4. ρstart = 5λ0 .5 ρ/λ0 Magnitude 7 7.5 6 6. th = 0. εr = 2.5 ρ/λ0 Phase 200 7 7. λ0 = 1cm.Magnitude 500 450 400 350 300 250 200 5 5.5 8 HFSS Ex √ρ GPOF b1es1(ρ−ρ0 ) 100 0 −100 −200 5 5. (f = 30GHz. N = 101) 157 .
εr = 2.23: Comparison of GPOF approximation with HFSS data and its Extrapolation . ρend = 8λ0 .5 6 ρ/λ0 Phase 200 6.5 5 5.Magnitude 500 450 400 350 300 250 200 4 4. λ0 = 1cm.25λ0 . (f = 30GHz. ρstart = 4λ0 .5 6 ρ/λ0 Magnitude 6. N = 51) 158 .5 7 7.5 5 5.5 7 7.55.5 8 550 500 450 400 350 300 250 200 2 3 4 5 6 ρ/λ0 Phase 200 7 8 9 10 HFSS Ex √ρ GPOF b1es1(ρ−ρ0 ) + b2es2(ρ−ρ0 ) 100 0 −100 −200 2 3 4 5 6 ρ/λ0 7 8 9 10 Figure 4.5 8 HFSS Ex √ρ GPOF b1es1(ρ−ρ0 ) + b2es2(ρ−ρ0 ) 100 0 −100 −200 4 4. th = 0.
159 .05λ0 case. As the thickness of the dielectric slab increases βT M0 increases too. When this ratio is very close to 1. the diﬀerentiation of the space wave and surface wave terms becomes diﬃcult. This phenomena can be observed at the th = 0.5 Conclusions Numerical results and comparisons with theoretical calculations show that our method works quite successfully along the Eline. For each mode that emerges recently. on the other hand taking not enough number of samples give inaccurate results because the behavior of the complex function to be approximated cannot be tracked correctly. Another factor that aﬀects the accuracy of the results is the ratio of the surface wave propagation constant to space wave propagation constant. The only exception occurs in the th = 0. [4λ0 − 9λ0 ] and [5λ0 − 9λ0 ] intervals give good results. In the implementation of the GPOF Method.2. This increases the accuracy of results to determine βT M0 . [5λ0 − 8λ0 ]. Taking too many samples makes the system of equations solved in the method more linearly dependent. But as the thickness continues to increase the results will be more successful for this mode. successful results are obtained only in the th = 0. The reasons are: ρ should be suﬃciently large and reﬂections from the radiation boundaries should decay not to contaminate the solutions. as the thickness increases. But along the Hline. the [4λ0 − 8λ0 ]. βT E1 cannot be determined as accurate as βT M0 is determined. However when thickness is suﬃciently large and TE1 mode emerges. In our method. the results will not be very satisfactory at ﬁrst. The number of samples (N ) is also important in the GPOF Method. we use N = 51 or N = 101 in the GPOF Method.1λ0 case.4.05λ0 case where surface wave propagation constant is very close to space wave propagation constant.
are investigated. These transparency and resonance conditions heavily depend on the permittivity of the metamaterial coating (for TE polarization) and the ratio of coreshell radii. metamaterial cylinders and metamaterial coated conducting cylinders. The formulation of these wave propagation problems is done in such a way that it remains valid for any kind of material used. such as metamaterial slabs. given in the literature (when the inner core layer is also taken to the PEC limit). electromagnetic scattering and transmission from metamaterial structures.Chapter 5 CONCLUSIONS In this thesis. The relations between the permittivity of the coating and the ratio of coreshell 160 .e. having any sign combination of constitutive parameters and having any electric and/or magnetic losses. metamaterial coated conducting cylinders illuminated normally with plane waves.. These structures are illuminated by electric line sources or plane waves. achieving transparency and maximizing scattering are investigated thoroughly. It is found out that. rigorous derivation of transparency and resonance conditions for PEC core cylinder case under the subwavelength limitations yields the same conditions of two electrically small concentric layers of conjugately paired cylinders. For one of these propagation problems i.
. A novel homogenization method for the retrieval of eﬀective constitutive parameters of metamaterials is proposed and implemented. 20 unit cells of the same metamaterial structure are stacked and their reﬂection coeﬃcients are collected. . Numerical results show that these analytical relations are quite successful and work better when the cylindrical scatter is electrically very small. The homogenization quality of the metamaterial is evaluated in terms of various factors. As 161 . the method can be modiﬁed to incorporate also the transmission data. 2. As another future work. another method has been proposed for the retrieval of surface wave propagation constants on any periodic or nonperiodic grounded slab medium. similar analytical transparency and resonance conditions can be derived and tested for obliquely incident plane waves. while keeping other geometry and material parameters the same. . While implementing the method.radii are investigated for achieving transparency and scattering maximization. the inﬁnite length metamaterial coated conducting cylinder can be truncated. Since the method is merely dependent on reﬂection. As the future work. 1. and can be simulated in full wave simulators to see whether such transparency or resonance conditions exist for reallife geometries. As the future work. since the method is already capable for this. Another future work can be homogenization in the oblique incidence case. so that the homogeneous equivalent mimics the metamaterial more successfully in the transmission region. which showed that the method is very successful to retrieve the eﬀective constitutive parameters of the metamaterial. Finally. The method is based on the simple idea that the total reﬂection coeﬃcient from a ﬁnite metamaterial structure has to resemble the reﬂection from an homogeneous equivalent. . the homogeneous equivalent characterizes the reﬂection property of the metamaterial best. if an eﬃcient oblique incidence implementation scheme can be formed for the simulation of metamaterial. Our preliminary numerical results for oblique incidence scenarios show the existence of such transparency and resonance conditions.
162 . The numerical results generally show good agreement with their theoretical counterparts.a preliminary. the method is applied to grounded dielectric slabs.
Jn (−x) = (−1)n Jn (x).1) is a second order diﬀerential equation. J−n (x) = (−1)n Jn (x). (A.1) Since Bessel’s equation in (A. 2 dx dx (A.4) (A.2) (A. When p = n = integer. while solving the wave equation.5) 163 . Bessel’s differential equation arises.3) where Jp (x) is referred to as the Bessel function of the ﬁrst kind of order p and Yp (x) as the Bessel function of the second kind of order p (or sometimes as the Neumann function). p = n = 0 or integer. which can be written as x2 d 2y dy + x + x2 − p2 y = 0. (A. it has two linearly independent solutions: y(x) = A1 Jp (x) + B1 J−p (x) y(x) = A2 Jn (x) + B2 Yn (x) p = 0 or integer.APPENDIX A Bessel Functions In cylindrical coordinate system.
(1) (A.13) (A.10) (p − 1)! − π (A. 2 x p (A. for p > 0: Jp (x) Yn (x) Large Argument Forms: When the argument of the Bessel functions is large (i. ln π 2 γ = 1.11) (A. it becomes more convenient to deﬁne Hankel functions: (1) Hp (x) = Jp (x) + jYp (x).. x → ∞). πx 4 2 2 π pπ sin x − − . x → 0).8) 2 γx . these cosine and sine functions in Bessel functions of the ﬁrst and second kinds represent standing waves.Small Argument Forms: When the argument of the Bessel functions is small (i. πx 4 2 1 x p! 2 p .. (A. Jp (x) Yp (x) 2 π pπ cos x − − .7) (A. (2) Hp (x) = Jp (x) − jYp (x). (A. for p = 0: J0 (x) Y0 (x) 1.e. For wave propagation.e.781.6) (A.14) (2) where Hp (x) is the Hankel function of the ﬁrst kind of order p and Hp (x) is the Hankel function of the second kind of order p.12) From electromagnetic point of view. 164 .9) .
17) where F (x) represents any kind of Bessel or Hankel function.For large arguments (i.16) (2) Hp (x) With the assumed ejωt time dependence. πx 2 −j[x−p(π/2)−π/4] e .e. For the derivatives of Bessel and Hankel functions. the following recurrence relation can be used: dFp (x) 1 = [Fp−1 (x) − Fp+1 (x)] . dx 2 (A. 165 . Hankel functions of the ﬁrst kind represent inward propagating waves.15) (A. Other alternative forms of the recurrence relations and many other properties of Bessel and Hankel functions of integer and noninteger orders can be found in [80].. x → ∞): (1) Hp (x) 2 j[x−p(π/2)−π/4] e . whereas Hankel functions of the second kind represent outward propagating waves. πx (A.
2) (B. jωε ρ ∂ρ ∂φ 166 .1) (B. .6) (B. × H = jωεE → E = Eρ = 1 jωε 1 jωε 1 ∂Hz ∂Hφ − ρ ∂φ ∂z ∂Hρ ∂Hz − ∂z ∂ρ Eφ = Ez = 1 1 ∂ ∂Hρ (ρHφ ) − . × E.3) (B.4) (B. jωµ ρ ∂ρ ∂φ 1 jωε × H. . (B.8) 1 ∂Ez ∂Eφ − ρ ∂φ ∂z ∂Eρ ∂Ez − ∂z ∂ρ Hφ = − Hz = − ∂Eρ 1 1 ∂ (ρEφ ) − . we will use Maxwell’s Equations: × E = −jωµH → H = − Hρ = − 1 jωµ 1 jωµ 1 jωµ . .7) (B.APPENDIX B Derivation of the φ Components of Electric and Magnetic Fields: T M z Polarization To ﬁnd the φ components of the electric and magnetic ﬁelds.5) (B.
1 jωε0 i Eρ = − i i 1 ∂Hz ∂Hφ − ρ ∂φ ∂z (B. i Eρ = − k0 cos θ0 i Hφ ωε0 (B. all derivatives of Bessel and Hankel functions are taken with respective to their entire arguments.14) Substituting (B. Therefore.17) 167 .16) j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) +∞ −E0 k0 sin θ0 e i Hφ = 2 k0 cos2 θ0 i E0 k0 sin2 θ0 jk0 z cos θ0 H + e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω 2 µ0 ε0 φ jωµ0 n=−∞ (B.11).15) j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) −E0 k0 sin θ0 e 1 jωµ0 i Hφ = − − 2 jk0 cos2 θ0 i Hφ ωε0 +∞ 2 jk0 z cos θ0 n=−∞ (B.9) i Hφ E0 sin θ0 e jk0 z cos θ0 n=−∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) (B.13) We also know that all ﬁeld variations in the z direction are in the form of ejk0 z cos θ0 .11) i Eρ = i Since Hz = 0.14) in (B.10) i Hφ 1 =− jωµ0 +∞ i ∂Eρ 2 jk0 z cos θ0 − E0 k0 sin θ0 e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) ∂z n=−∞ (B. i Hφ = − 1 jωµ0 − i k0 cos θ0 ∂Hφ ωε0 ∂z +∞ 2 jk0 z cos θ0 n=−∞ (B.In the following derivations. i Hφ : i Hφ i ∂Eρ 1 ∂ =− − jωµ0 ∂z ∂ρ 1 =− jωµ0 i i ∂Eρ ∂Ez − ∂z ∂ρ +∞ (B.12) i 1 ∂Hφ jωε0 ∂z (B.
22) i ∂Eφ ∂z (B.20) = jk0 cos θ0 .√ Since k0 = ω µ0 ε0 .26) 168 . (1−cos 2 i θ0 )Hφ = sin 2 i θ 0 Hφ E0 k0 sin2 θ0 jk0 z cos θ0 = e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) jωµ0 n=−∞ (B. i Eφ = k0 cos θ0 i Hρ ωε0 (B.21) in (B.25) cos θ0 i Hρ ωε0 +∞ i Hρ E0 sin θ0 jk0 z cos θ0 k 2 cos2 θ0 i e H =− nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) + 0 2 ωµ0 ρ ω µ0 ε 0 ρ n=−∞ (B.21) i Hρ 1 =− jωµ0 i i 1 ∂Ez ∂Eφ − ρ ∂φ ∂z +∞ (B.24) Substituting (B.18) +∞ +∞ i Hφ E0 k0 jk0 z cos θ0 e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) = jωµ0 n=−∞ = −j E0 jk0 z cos θ0 e j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) η0 n=−∞ +∞ (B.24) i Hρ 1 1 =− jωµ0 ρ − +∞ jE0 sin θ0 e 2 jk0 2 jk0 z cos θ0 n=−∞ nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) (B.19) i Eφ : i Eφ = i Since Hz = 0 and ∂ ∂z 1 jωε0 i i ∂Hρ ∂Hz − ∂z ∂ρ (B.23) i Hρ = − 1 1 ∂ jωµ0 ρ ∂φ E0 sin θ0 ejk0 z cos θ0 n=−∞ j −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) − i Hρ = − 1 1 jωµ0 ρ +∞ jE0 sin θ0 ejk0 z cos θ0 n=−∞ nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) − i ∂Eφ ∂z (B.
31) s Eφ = 1 s jk0 cos θ0 Hρ jωε0 +∞ (B.27) +∞ +∞ i Hρ = − E0 ejk0 z cos θ0 nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 ρ sin θ0 n=−∞ (B.28) in (B.34) s Hρ 1 =− jωµ0 1 (2) jE0 sin θ0 ejk0 z cos θ0 nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ρ n=−∞ s −jk0 cos θ0 Eφ +∞ (B.36) 169 .28) Substituting (B.(1−cos 2 i θ0 )Hρ = sin 2 i θ0 Hρ E0 sin θ0 jk0 z cos θ0 =− e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 ρ n=−∞ (B.32) jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) −E0 k0 sin θ0 e 2 s Eφ k0 cos θ0 s E0 k0 sin2 θ0 jk0 z cos θ0 (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) = Hρ − e ωε0 jωε0 n=−∞ (B.21) gives i Eφ E0 k0 cos θ0 jk0 z cos θ0 =− 2 e nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω µ0 ε0 ρ sin θ0 n=−∞ E0 cos θ0 jk0 z cos θ0 e =− nj −n Jn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ +∞ +∞ (B.29) i Eφ (B.30) s Eφ : s Eφ = 1 jωε0 s s ∂Hρ ∂Hz − ∂z ∂ρ (B.33) +∞ s Hρ = − 1 jωµ0 s s ∂Eφ 1 ∂Ez − ρ ∂φ ∂z +∞ (B.35) s Hρ E0 sin θ0 jk0 z cos θ0 k0 cos θ0 s (2) =− e Eφ nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) + ωµ0 ρ ωµ0 n=−∞ (B.
33) gives s k0 Eφ = cos θ0 ωε0 E0 sin θ0 jk0 z cos θ0 (2) − e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 ρ n=−∞ + k0 cos θ0 s Eφ ωµ0 +∞ +∞ E0 k0 sin2 θ0 jk0 z cos θ0 (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) e − jωε0 n=−∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω 2 µ0 ε 0 ρ n=−∞ +∞ +∞ (B.42) jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) −E0 k0 sin θ0 e 2 s Hφ = − k0 cos θ0 s E0 k0 sin2 θ0 jk0 z cos θ0 (2) Eρ + e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωµ0 jωµ0 n=−∞ (B.41) s Hφ = − 1 s jk0 cos θ0 Eρ jωµ0 +∞ (B.43) +∞ 170 .39) +jE0 η0 e jk0 z cos θ0 n=−∞ (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) (B.37) s Eφ =− (B.Substituting (B.40) s Hφ : s Hφ = − 1 jωµ0 s s ∂Eρ ∂Ez − ∂z ∂ρ (B.36) in (B.38) 2 k0 cos2 θ0 s E0 k0 sin2 θ0 jk0 z cos θ0 (2) + 2 Eφ − e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω µ0 ε 0 jωε0 n=−∞ +∞ E0 cos θ0 jk0 z cos θ0 s (2) Eφ =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ E0 k0 jk0 z cos θ0 (2) − e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) jωε0 n=−∞ +∞ E0 cos θ0 jk0 z cos θ0 (2) s e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) Eφ =− k0 ρ sin θ0 n=−∞ +∞ +∞ (B.
48) k 2 cos2 θ0 s E0 k0 sin2 θ0 jk0 z cos θ0 (2) + 02 H + e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω µ0 ε0 φ jωµ0 n=−∞ +∞ E0 cos θ0 jk0 z cos θ0 s (2) Hφ =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) k0 ρ sin θ0 n=−∞ E0 k0 jk0 z cos θ0 (2) e + j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) jωµ0 n=−∞ +∞ E0 cos θ0 jk0 z cos θ0 (2) s e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) Hφ =− k0 ρ sin θ0 n=−∞ +∞ (B.49) E0 (2) j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) −j ejk0 z cos θ0 η0 n=−∞ t Eφ : t Eφ = +∞ (B.s Eρ = 1 jωε0 s s ∂Hφ 1 ∂Hz − ρ ∂φ ∂z +∞ (B.47) (B.46) Substituting (B.45) s Eρ E0 sin θ0 jk0 z cos θ0 k0 cos θ0 s (2) = Hφ e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) − ωε0 ρ ωε0 n=−∞ (B.44) s Eρ = 1 jωε0 1 (2) jE0 sin θ0 ejk0 z cos θ0 nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ρ n=−∞ s −jk0 cos θ0 Hφ +∞ (B.51) 171 .50) 1 jωε t t ∂Hρ ∂Hz − ∂z ∂ρ (B.46) in (B.43) gives k0 s Hφ =− cos θ0 ωµ0 E0 sin θ0 jk0 z cos θ0 (2) e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ωε0 ρ n=−∞ − k0 cos θ0 s Hφ ωε0 +∞ +∞ E0 k0 sin2 θ0 jk0 z cos θ0 (2) + e j −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) jωµ0 n=−∞ +∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 s (2) Hφ =− e nj −n cn Hn (k0 ρ sin θ0 )ejn(φ−φ0 ) ω 2 µ0 ε 0 ρ n=−∞ +∞ (B.
58) Since 2 k0 cos2 θ0 k 2 cos2 θ0 = 0 2 = cos2 θ1 ω 2 µε k +∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 t e Eφ =− nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) k 2 ρ sin2 θ1 n=−∞ (B.53) +∞ t Hρ = − 1 jωµ +∞ t t 1 ∂Ez ∂Eφ − ρ ∂φ ∂z (B.60) .56) in (B.59) E0 k sin θ0 jk0 z cos θ0 e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) − jωε sin θ1 n=−∞ 172 +∞ (B.57) t Eφ =− E0 k0 sin θ0 cos θ0 jk0 z cos θ0 k 2 cos2 θ0 t e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) + 0 2 Eφ ω 2 µερ ω µε n=−∞ E0 k sin θ0 sin θ1 jk0 z cos θ0 e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) jωε n=−∞ +∞ − (B.52) t Eφ = k0 cos θ0 t E0 k sin θ0 sin θ1 jk0 z cos θ0 Hρ − e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) ωε jωε n=−∞ (B.54) t Hρ 1 =− jωµ 1 t jE0 sin θ0 ejk0 z cos θ0 nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) − jk0 cos θ0 Eφ ρ n=−∞ (B.53) gives k0 t Eφ = cos θ0 ωε E0 sin θ0 jk0 z cos θ0 − e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) ωµρ n=−∞ + k0 cos θ0 t Eφ ωµ +∞ +∞ E0 k sin θ0 sin θ1 jk0 z cos θ0 − e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) jωε n=−∞ +∞ (B.55) +∞ k0 cos θ0 t E0 sin θ0 jk0 z cos θ0 t e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) + Eφ Hρ = − ωµρ ωµ n=−∞ (B.56) Substituting (B.t Eφ 1 = jωε +∞ t jk0 cos θ0 Hρ − E0 k sin θ0 sin θ1 e jk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) (B.
65) t Eρ 1 t jE0 sin θ0 ejk0 z cos θ0 nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) − jk0 cos θ0 Hφ ρ n=−∞ (B.64) t Eρ = 1 jωε +∞ t t 1 ∂Hz ∂Hφ − ρ ∂φ ∂z (B.66) +∞ k0 cos θ0 t E0 sin θ0 jk0 z cos θ0 t Eρ = e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) − Hφ (B.64) gives k0 t Hφ =− cos θ0 ωµ − E0 sin θ0 jk0 z cos θ0 e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) ωερ n=−∞ k0 cos θ0 t Hφ ωε +∞ +∞ 1 = jωε E0 k sin θ0 sin θ1 jk0 z cos θ0 + e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) jωµ n=−∞ E0 k0 t Hφ =− sin θ0 cos θ0 jk0 z cos θ0 e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) ω 2 µερ n=−∞ +∞ +∞ (B.62) (B.67) ωερ ωε n=−∞ Substituting (B.+∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 t Eφ =− e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) 2 ρ sin2 θ k 1 n=−∞ sin θ0 jk0 z cos θ0 +jE0 η e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) sin θ1 n=−∞ t Hφ : t Hφ = − +∞ (B.67) in (B.61) 1 jωµ t t ∂Eρ ∂Ez − ∂z ∂ρ (B.69) k 2 cos2 θ0 t E0 k sin θ0 sin θ1 jk0 z cos θ0 Hφ + e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) + 0 2 ω µε jωµ n=−∞ 173 .63) t Hφ =− 1 t jk0 cos θ0 Eρ jωµ +∞ −E0 k sin θ0 sin θ1 e jk0 z cos θ0 n=−∞ j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) +∞ t Hφ = − k0 cos θ0 t E0 k sin θ0 sin θ1 jk0 z cos θ0 j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) Eρ + e ωµ jωµ n=−∞ (B.68) (B.
72) 174 .Since 2 k0 cos2 θ0 k 2 cos2 θ0 = 0 2 = cos2 θ1 ω 2 µε k +∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 t Hφ =− e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) k 2 ρ sin2 θ1 n=−∞ (B.71) E0 sin θ0 jk0 z cos θ0 −j e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) η sin θ1 n=−∞ +∞ (B.70) E0 k sin θ0 jk0 z cos θ0 + e j −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) jωµ sin θ1 n=−∞ +∞ E0 k0 sin θ0 cos θ0 jk0 z cos θ0 t Hφ =− e nj −n an Jn (kρ sin θ1 )ejn(φ−φ0 ) 2 ρ sin2 θ k 1 n=−∞ +∞ (B.
4) n! 2 2 (C.APPENDIX C Derivation of the Transparency Condition For achieving transparency with metamaterial coated conducting cylinders at normal incidence and T E z polarization.3) (C.2) (C.240) should be zero: num{cn } = ζJn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] −Jn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] = 0. 1 kc a n−1 1 kc a n+1 n − (n+1)! 2 1 k0 b (n−1)! 2 T1 = ζ (C. such that T1 + T2 = 0. Let T1 and T2 be the the two terms of the numerator of cn : T1 = ζJn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] . Using the small argument approximations and the recurrence relation. numerator of the scattering coeﬃcients. cn . given in (2. T2 = −Jn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] .1) 175 .
6) 1 a n−1 a n−1 4 +n (n − 1) b b (kc b)2 1 a n−1 (kc a)2 1 a + − (n + 1)n(n − 1) b 4 (n + 1) b 1 (n − 1) b a n−1 n+1 − − + +n b a b a n−1 n−1 4 (kc a)2 b a n+1 1 (n + 1)n(n − 1) 1 (kc b)2 − 4 (n + 1) . n 2 1 T1 = ζ n! × k0 b 2 − n 1 4π (C. (n−2)! − π × 2 kc b n−1 + n! π 2 kc b n+1 2 − 2 1 (n+1)! kc b n+1 2 − × 1 (n−1)! kc b n−1 2 (n−2)! − π 2 kc a n−1 + n! π 2 kc a n+1 . 2 T2 = − 1 (n−1)! k0 b n−1 2 − 2 1 (n+1)! k0 b n+1 2 (C.5) − (n − 1)! π + n! π 2 kc a × 1 (n−1)! kc a n−1 2 − 2 1 (n+1)! kc a n+1 2 − 1 n! kc b 2 n (n−2)! − π 2 kc a n−1 2 kc b n+1 .7) + a 1 (n + 1)n b kc b 2 − b a 1 + (n + 1)n n−1 2 kc a . 1 T2 = − 4π × − 1 (n − 1)! a b n−1 k0 b 2 2 kc b b a n−1 1 − (n + 1)! n k0 b 2 kc a 2 n n+1 (C. 176 .
k0 kc ab (C. (kc a)2 (C.8) 1 T2 = 4π × 1 (n − 1)! a b n−1 k0 b 2 2 kc b + n−1 1 − (n + 1)! n k0 b 2 .11) 1 − (n + 1)! 1 = n! 1 n! k0 b 2 k0 b 2 k0 b 2 n n n n 2 1 − k0 b (n + 1) 2 .9) b a 2 kc a Writing 1 (n − 1)! 1 (n + 1)! 1 (n − 1)! k0 b 2 k0 b 2 n−1 k0 b 2 n+1 n−1 1 = n! k0 b 2 k0 b 2 n n n 2 k0 b .13) T1 + T2 = 1 n! 1 4π n−1 (C. (C. n+1 (C.10) 1 = n! 1 (n + 1) n+1 k0 b 2 .14) 4 − ζn 2 b2 kc 4 +n k0 kc b2 b a n−1 × ζn +n = 0.12) T2 becomes 1 T2 = n! k0 b 2 n 1 a n 4π b k0 b 2 a b a b n n−1 4 +n k0 kc b 2 b a n 4 . k0 b k0 b 2 (C.e.Since kc a 1 and kc b 1. (C.. the terms where kc a and kc b are in the denominator): T1 = ζ 1 n! k0 b 2 n a 1 n 4π b n−1 4 −n (kc b)2 b a n−1 4 . 4 2 kc a2 n−1 b a n 4 k0 kc ab 177 . we can keep only the dominant terms in T1 and T2 (i.
εc εc ε0 ε0 178 (C.16) √ k c = ω µc ε c . √ √ 2 µ ε ω 0 0 µc εc ab (C. ηc = ζ= η0 µc εc µ0 ε0 = µc ε0 . (C. √ µc ε0 µ0 ε0 µc εc ab 0= 1 1 b a n−1 1 1 1 1 − + + µc ε c b 2 a µc εc a2 b µc ε0 b2 an b 1 1 bn a 1 1 an b 1 1 bn 1 1 = n − n + n + n b a µc εc b2 a b µc εc a2 b a µc ε0 b2 a µc ε0 ab 1 bn 1 an 1 bn 1 an 1 = − n + n + n . Note that.17) (C. µ0 ε c (C.18) Hence.20) n µ0 ε c 1 1 .21) Denoting γ = a/b.22) . ζ a b n−1 1 −ζ 2 b2 kc b a n−1 1 a + 2 a2 kc b n−1 1 + k0 kc b 2 b a n 1 =0 k0 kc ab (C. √ k0 = ω µ0 ε0 . 0= a b n−1 a + b n−1 µc ε 0 1 b 1 − 2 µ ε b2 µ0 εc ω c c a b 1 1 + √ √ 2 µ ε 2 ω a 0 0 µ c εc b n−1 n µ c ε0 1 1 2 µ ε a2 µ0 εc ω c c 1 1 . 1 1 b 1 1 0= − 2 µc ε c b a µc εc a2 b a n−1 µ0 εc 1 1 + + √ 2 b µc ε0 µ0 ε0 µc εc b a a b n−1 n−1 (C. µc ab bn εc a εc b ε0 a ε0 a b n−1 n−1 b a n 1 1 µc ε0 ab (C.Therefore. γ n γ −n γ n γ −n − + + = 0.19) or dividing each term by ω 2 ζ.15) .
(C.γn 1 1 + εc ε0 1 εc 1 εc = γ −n 1 1 − εc ε0 ε0 − εc .25) 179 .24) γ= 2n ε0 − εc .23) γ 2n = − + 1 ε0 1 ε0 = (C. ε0 + εc . ε0 + εc (C.
APPENDIX D Derivation of the Resonance Condition
For scattering maximization (i.e., resonance) with metamaterial coated conducting cylinders at normal incidence and T E z polarization, denominator of the scattering coeﬃcients, cn , given in (2.241) should be zero:
(2) den{cn } = −ζHn (k0 b) [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] (2) +Hn [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)]
(D.1)
= 0. Let R1 and R2 be the the two terms of the denominator of cn :
(2) R1 = −ζHn [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] , (2) R2 = Hn [Jn (kc a)Yn (kc b) − Jn (kc b)Yn (kc a)] ,
(D.2) (D.3)
such that R1 + R2 = 0. Using the small argument approximations and the recurrence relation, R1 = −ζ 1 n! k0 b 2
n
+j
(n − 1)! π
2 k0 b
n
(D.4)
180
×
1 (n−1)!
kc a n−1 2
− 2
1 (n+1)!
kc a n+1 2
(n−2)! − π
2 kc b
n−1
+
n! π
2 kc b
n+1
2
2 kc a n−1
−
1 (n−1)!
kc b n−1 2
− 2
1 (n+1)!
kc b n+1 2
(n−2)! − π
+
n! π
2 kc a
n+1
,
2
R2 =
1 (n−1)! k0 b n−1 2
− 2
1 (n+1)!
k0 b n+1 2
−j
− (n−2)! π
2 k0 b
n−1
+
n! π
2 k0 b
n+1
2 − (n − 1)! π +
n! π 2 kc a
×
1 (n−1)!
kc a n−1 2
− 2
1 (n+1)!
kc a n+1 2
−
1 n!
kc b 2
n
(n−2)! − π
2 kc a
n−1
2 kc b n+1 ,
n
2
(D.5)
R1 = −ζ ×
k0 b 2 1 (n − 1)! +j 2 π k0 b 4π n−1 n−1 a a 4 1 +n − (n − 1) b b (kc b)2 1 a n−1 (kc a)2 1 a + − (n + 1)n(n − 1) b 4 (n + 1) b − − + 1 (n − 1) b a
n−1
1 n!
n
n
(D.6)
n+1
+n b a
b a
n−1
n−1
4 (kc a)2 b a
n+1
1 (n + 1)n(n − 1)
(kc b)2 1 − 4 (n + 1)
,
R2 = 1 2π × − +
1 (n−1)! k0 b n−1 2
− 2
1 (n+1)!
k0 b n+1 2
−j
n
− (n−2)! π
2 k0 b
n−1
+
n! π
2 k0 b
n+1
2 kc a 2
a b
n−1
2 kc b b a
+
a 1 (n + 1)n b kc b 2 − b a
1 (n + 1)n
n−1
n
2 kc a
.
(D.7)
181
Since kc a
1 and kc b
1, we can keep only the dominant terms in R1 and
R2 (i.e., the terms where kc a and kc b are in the denominator): R1 = −jζ (n − 1)! π 2 k0 b
n
a 1 n 4π b
n−1
4 −n (kc b)2
b a
n−1
4 , (D.8) (kc a)2
1 R2 = j 4π × a b
(n − 2)! − π
n−1
2 k0 b +
n−1
n! + π 2 kc a
2 k0 b .
n+1
2 kc b
b a
n
(D.9)
Writing (n − 2)! − π n! π (n − 2)! − π 2 k0 b 2 k0 b
n−1
(n − 1)! =− π
2 k0 b 2 k0 b
n
n
1 (n − 1) 2 k0 b
k0 b 2
,
(D.10)
n+1
(n − 1)! =− π 2 k0 b
n+1
(−n)
,
(D.11)
2 k0 b
n−1
n! + π
2 1 (n − 1)! =− π k0 b (n − 1) n (n − 1)! 2 2 n , π k0 b k0 b R2 becomes 1 (n − 1)! R2 = j 4π π 2 k0 b
n
n
k0 b 2
−n
2 k0 b (D.12)
n
2 k0 b 2 k0 b
n
a b
n−1
2 kc b
+
b a
n
2 kc a
, (D.13)
R1 + R2 = −j
1 n! 4π π a b
(D.14) b a 2 kc b
n−1
× ζ − = 0.
n−1
4 −ζ (kc b)2 a b
n−1
4 (kc a)2 2 k0 b b a
n
2 k0 b
−
2 kc a
182
Therefore, ζ a b
n−1
1 −ζ 2 b2 kc
b a
n−1
1 a − 2 a2 kc b
n−1
1 − k0 kc b 2
b a
n
1 =0 k0 kc ab
(D.15) .
Note that, ηc = ζ= η0
µc εc µ0 ε0
=
µc ε0 , µ0 ε c
(D.16)
√ k c = ω µc ε c , √ k0 = ω µ0 ε0 .
(D.17)
(D.18)
Hence, 0= a b
n−1
a − b
n−1
µc ε0 1 b 1 − 2 µ ε b2 µ0 ε c ω c c a b 1 1 − √ √ 2 µ ε 2 ω a 0 0 µ c εc b
n−1
n
µc ε0 1 1 2 µ ε a2 µ0 ε c ω c c 1 1 , √ √ 2 µ ε ω 0 0 µc εc ab
(D.19)
or dividing each term by ω 2 ζ, 1 1 b 1 1 0= − 2 µc εc b a µc εc a2 b a n−1 µ0 εc 1 1 − − √ 2 b µc ε0 µ0 ε0 µc εc b a a b
n−1 n−1
(D.20)
n
µ0 εc 1 1 , √ µc ε0 µ0 ε0 µc εc ab
0=
1 1 b a n−1 1 1 1 1 − − − µc εc b2 a µc ε c a 2 b µc ε0 b2 an b 1 1 bn a 1 1 an b 1 1 bn 1 1 = n − n − n − n b a µc εc b2 a b µc εc a2 b a µc ε0 b2 a µc ε0 ab 1 bn 1 an 1 bn 1 an 1 = − n − n − n . µc ab bn εc a εc b ε0 a ε0 a b
n−1
n−1
b a
n
1 1 µc ε0 ab
(D.21)
Denoting γ = a/b, γ n γ −n γ n γ −n − − − = 0, εc εc ε0 ε0 183 (D.22)
23) γ 2n = + − 1 ε0 1 ε0 = (D.γn 1 1 − εc ε0 1 εc 1 εc = γ −n 1 1 + εc ε0 ε0 + εc . ε0 − εc .24) γ= 2n ε0 + εc .25) 184 . ε0 − εc (D. (D.
[4] S. B. vol. 3318–3332. “Pairing an epsilonnegative slab with a munegative u slab: Resonance. Smith. 51. Oct. “Controlling electromagnetic ﬁelds. R. Antennas Propagat. Shen. Pendry.” Phys. vol. 2003. Sep. Schurig. pp. 312. 1780–1782. E. 2006. “Achieving transparency with plasmonic and metau material coatings. R.” IEEE Trans. Al` and N. 91–105. “Fullwave simulations of electromagnetic cloaking structures. tunneling and transparency. Popa. Smith. March 2007. E. 185 . 74. Engheta. 74.” Phys. 2558–2571. 42. Aug. pp. 016623/1–9. Rev. Engheta. July 2005. “Design for electromagnetic wave transparency with metamaterials. Hu. Rev. [5] A.” Progress In Electromagnetics Research. robustness. Cummer. 72. and D. Li and Z. vol.Bibliography [1] C. A. June 2006. [3] J. 036621/1–5. vol. Al` and N. and J. vol. Engheta. “Electromagnetic scattering by a conducting cylinder coated with metamaterials. Al` and N. D. [6] A. pp. E.. I. Zhou and G. and physical insights.” Phys. Rev. [7] A. vol. 2006.” Science. “Plasmonic materials in transparency and cloaking u problems: mechanism. pp. Pendry. 026607/1–8. pp. 2003. 15. D. pp. B. Schurig. pp. D.” Optics Express. vol. [2] X.
[8] A. Al` and N. Engheta, “Resonances in subwavelength cylindrical strucu tures made of pairs of doublenegative and doublepositive or epsilonnegative and munegative coaxial shells,” in Proc. Int. Conf. on Electromagnetics in Advanced Applications, (Torino, Italy), pp. 435–438, Sep. 812 2003. [9] A. Al` and N. Engheta, “Subwavelength resonant structures containing u doublenegative (DNG) or singlenegative (SNG) media: Planar, cylindrical and spherical cavities, waveguides, and open scatterers,” in Progress in Electromagnetic Research Symp., (Waikiki, HI), p. 12, Oct. 1316 2003. [10] A. Al` and N. Engheta, “Peculiar radar cross section properties of doubleu negative and singlenegative metamaterials,” in Proc. of the 2004 IEEE Radar Conf., (Philadelphia, PA, USA), Apr. 2629 2004. [11] A. Al` and N. Engheta, “Polarizabilities and eﬀective parameters for colu lections of spherical nanoparticles formed by pairs of concentric doublenegative (DNG), singlenegative (SNG) and/or doublepositive (DPS) metamaterial layers,” J. Appl. Phys, vol. 97, pp. 094310/1–12, Apr. 2005. [12] A. Al`, N. Engheta, and R. W. Ziolkowski, “FDTD simulation of tunneling u and “growing exponential” in a pair of epsilonnegative and munegative slabs,” in USNCURSI Nat. Radio Science Meeting Dig., (Monterey, CA), p. 18, Jun. 2026 2004. [13] S. Arslanagic and O. Breinbjerg, “Electriclinesource illumination of a circular cylinder of lossless doublenegative material: an investigation of near ﬁeld, directivity, and radiation resistance,” IEEE Antennas Propagat. Mag., vol. 48, pp. 38–54, June 2006. [14] S. Arslanagic, R. W. Ziolkowski, and O. Breinbjerg, “Excitation of an electrically small metamaterialcoated cylinder by an arbitrarily located line
186
source,” Microwave and Optical Technology Letters, vol. 48, pp. 2598–2606, Dec. 2006. [15] R. W. Ziolkowski and A. D. Kipple, “Application of double negative materials to increase the power radiated by electrically small antennas,” IEEE Trans. Antennas Propagat., vol. 51, pp. 2626–2640, Oct. 2003. [16] A. Erentok and R. W. Ziolkowski, “Eﬃcient electrically small antenna design using an electric dipole in a multilayered ENG metamaterial shell,” in Int. Workshop on Antenna Technology: Small Antennas and Novel Metamaterials, (White Plains, New York), pp. 400–403, March 68 2006. [17] R. W. Ziolkowski and A. Erentok, “Metamaterialbased eﬃcient electrically small antennas,” IEEE Trans. Antennas Propagat., vol. 54, pp. 2113–2130, July 2006. [18] D. R. Smith, D. C. Vier, N. Krol, and S. Schultz, “Direct calculation of permeability and permittivity for a lefthanded metamaterial,” Appl. Phys. Lett., vol. 77, pp. 2246–2248, Oct. 2000. [19] D. R. Smith and J. B. Pendry, “Homogenization of metamaterials by ﬁeld averaging,” J. Opt. Soc. Am. B, vol. 23, pp. 391–403, Mar. 2006. [20] D. R. Smith, S. Shultz, P. Markos, and C. M. Soukoulis, “Determination of eﬀective permittivity and permeability of metamaterials from reﬂection and transmission coeﬃcients,” Phys. Rev. B, vol. 65, p. 195104, 2002. [21] D. R. Smith, D. C. Vier, T. Koschny, and C. M. Soukoulis, “Electromagnetic parameter retrieval from inhomogeneous metamaterials,” Phys. Rev. E, vol. 71, p. 036617, 2005. [22] T. Koschny, M. Kafesaki, E. N. Economou, and C. M. Soukoulis, “Eﬀective medium theory of lefthanded materials,” Phys. Rev. Lett, vol. 93, p. 107402, 2004.
187
[23] J. B. Pendry, “Negative refraction makes a perfect lens,” Phys. Rev. Lett., vol. 85, pp. 3966–3969, Oct. 2000. [24] P. M. Valanju, R. M. Walser, and A. P. Valanju, “Wave refraction in negativeindex media: Always positive and very inhomogeneous,” Phys. Rev. Lett., vol. 88, p. 187401, Apr 2002. [25] R. W. Ziolkowski and E. Heyman, “Wave propagation in media having negative permittivity and permeability,” Phys. Rev. E, vol. 64, pp. 056625/1–15, Nov. 2001. [26] D. R. Smith and N. Kroll, “Negative refractive index in lefthanded materials,” Phys. Rev. Lett., vol. 85, pp. 2933–2936, Oct. 2000. [27] C. A. Balanis, Advanced Engineering Electromagnetics. New York: Wiley, 1989. [28] R. F. Harrington, TimeHarmonic Electromagnetic Fields. McGrawHill, 1961. [29] J. R. Wait, Electromagnetic Radiation from Cylindrical Structures. London: Peter Peregrinus Ltd., 1988. [30] R. W. Ziolkowski, “Design, fabrication and testing of double negative metamaterials,” IEEE Trans. Antennas Propagat., vol. 65, pp. 1516–1529, July 2003. [31] N. Kinayman and M. I. Aksun, Modern Microwave Circuits. Artech House, Norwood, MA, 2005. [32] Y. Hua and T. K. Sarkar, “Generalized penciloffunction method for extracting poles of an em system from its transient response,” IEEE Trans. Antennas Propagat., vol. 37, pp. 229–234, Feb. 1989. New York:
188
[33] Ca˘atay Tokg¨z, “Derivation of closedform Green’s functions for cylindri¸ g o cally stratiﬁed media,” Master’s thesis, Middle East Technical University, 1997. [34] S. Maci, M. Caiazzo, A. Cucini, and M. Casaletti, “A polezero matching method for ebg surfaces composed of a dipole fss printed on a grounded dielectric slab,” IEEE Trans. Antennas Propagat., vol. 53, pp. 70–81, Jan. 2005. [35] N. Engheta and R. W. Ziolkowski, “A positive future for doublenegative metamaterials,” vol. 53, pp. 1535–1556, Apr. 2005. [36] X. S. Rao and C. K. Ong, “Ampliﬁcation of evanescent waves in a lossy lefthanded material slab,” Phys. Rev. B, Condens. Matter, vol. 68, no. 11. Paper 113103. [37] C. Caloz and T. Itoh, “Array factor approach of leakywave antennas and application to 1d/2d composite right/lefthanded (crlh) structures,” vol. 14, pp. 274–276, June 2004. [38] A. P. Feresidis, G. Apostolopoulos, N. Serfas, and J. C. Vardaxoglou, “Closely coupled metallodielectric electromagnetic bandgap structures formed by doublelayer dipole and tripole arrays,” IEEE Trans. Antennas Propagat., vol. 52, pp. 1149–1158, May 2004. [39] A. D. Boardman, P. Egan, L. Velasco, and N. King, “Control of planar nonlinear guided waves and spatial solitons with a lefthanded medium,” J. Opt. A: Pure Appl. Opt., vol. 7, pp. 57–67, Jan. 2005. [40] S. Maci, M. Casaletti, M. Caiazzo, and A. Cucini, “Dispersion analysis of printed periodic structures by using a polezero network synthesis,” in 17th International Conference on Applied Electromagnetics and Communications, ICECom 2003, (Dubrovnik, Croatia), pp. 300–303, Oct. 2003.
189
53. Frezza. 53. Paper 013901. 2005. 411–415. G.” Phys. Camletti. Pendry. 2005. [45] F. Jan. Artigas and L. 3. “Electromagnetic Surface Modes at Interfaces with Negative Refractive Index make a “NotQuitePerfect” Lens. “Existence and properties of microwave surface plasmons at the interface between a righthanded and a lefthanded media. Paulotto. C. 1431–1442. 2005. R. Burghignoli.” vol. C. P. D. Baccarelli. Caloz. P. Galli. A. P. A. J. Apr. [44] Y.” in IEEE Antennas and Propagation Society International Symposium. Jan. Frezza. and S. Feresidis.” Solid State Commun. D.” in IEEE Antennas and Propagation Society International Symposium. pp. “Dyakonov surface waves in photonic metamaterials. J. [42] D. 116.” vol. Rev.[41] S. 53. G. vol. Vardaxoglou. Caiazzo. 2000. 3151–3154. Guo. 2004. F. and D. Boﬀa. A. Paulotto. B. 493–496. Capolino. Lampariello. 2005. pp. June 2003. Lovat. and C. vol. vol. [43] P.” vol. 32–44. [49] F. “Fundamental modal properties of surface waves on metamaterial grounded slabs. June 2004. Goussetis. Maci. 1. R. 94. Aug. Baccarelli. “Eﬀects of leakywave propagation in metamaterial grounded slabs excited by a dipole source. and J. M. Lee.. M.” ArXiv Condensed Matter eprints. 2003. D. Galli. [46] C. R. Burghignoli. June 2002. [48] P. Smith. Torner. Lett. vol. pp. F. G. pp. “Extinction properties of a sphere with negative permittivity and permeability. P.. 1462– 1468. Ruppin. and S. Wilton. M. pp. Jackson. “Eﬃcient modeling of novel uniplanar lefthanded metamaterials. and T. “Fundamental properties of the ﬁeld at the interface between air and a periodic artiﬁcial material excited 190 . Itoh. “Dispersion properties of periodic grounded structures via equivalent network synthesis. P. Apr. pp. Lampariello. Haldane. Lovat. [47] R.
pp. 2005. Zhang. R. Aug. “Leakage phenomena from negative refractive index structures. Y. S. 91–99. Bilotti. [54] C. and T. UK).. 85. 2003. [53] X. “Leaky surfaceplasmon theory for dramatically enhanced transmission through a subwavelength aperture. Fang. Lett. [51] R. July 2004. Vegni. 53.. “Guidedwave characteristics of waveguide based periodic structures loaded with various fss strip layers. vol. Oliner and D. Nov. Plasma. and L. A. The Graduate School. Itoh. May 2004. vol. part i: Basic features. Alu. and Metamaterial Columns. C. Phys. S. [56] C.. [52] A.” Appl. F. Components and Experiments. Zhu. “Large positive and negative lateral optical beam displacements due to surface plasmon resonance. Caloz. vol. [50] K. A. Jan. N. (Pisa. Caloz. pp. 372–374. N. Lim. Electromagnetic Theory Symp.” in URSI Int. Antennas Propagat. Engheta. PhD thesis. 120–124.” 191 . Guided and Leaky Modes of Circular Open Electromagnetic Waveguides Dielectric. 2005. “Leakywaves in a metamaterialbased twodimensional structure for a conical beam antenna application. Kshetrimayum and L. 2004. Hesselink. Yina. and X. and T. Allen. pp. vol.” in IEEE Antennas and Propagation Society International Symposium. pp. Jan. Z.by a line source. 2003. Itoh.” IEEE Trans.” IEEE Trans. Dec. (London. 2003. Jackson. 2.. [55] A. 53. Antennas Propagat. Italy). Allen. “How metamaterials may signiﬁcantly aﬀect the wave transmission through a subwavelength hole in a ﬂat perfectly conducting screen. Kyungpook National University.” in IEE Workshop on Metamaterials for Microwave and (Sub)millimetre Wave Applications: Photonic Bandgap and Double Negative Designs. Liu. 1091–1094. Kim. L. C.
Jan. pp. 1. Y. Engheta. 379–381. A. P. Paulotto.” J. P. 2004. Iyer and G. P. 2004 IEEE MTTS International. P. A. Burghignoli. 1411– 1414. Jan. E. V. Lampariello. Shadrivov. “New dispersion characteristics and surfacewave suppression in doublenegative metamaterial grounded slabs. Antennas Propagat. A. A. S. Rev. and S. Burghignoli. and L. 133–140. 69.” in URSI Int. Lovat. F. [58] A. Paulotto. Opt. 192 . Lovat.. “Metawaveguides formed by arrays of small resonant particles over a ground plane. P. (Pisa. 36–46. Italy). Tretyakov. A. vol. June 2004. [63] P. under review. 17–27. vol. G. Feb. Sukhorukov. Baccarelli. Egan. Baccarelli. [62] I. [60] P. Galli. D. N. “Metamaterial covers over a small aperture. F. Bilotti. “Novel analytical representations of the continuousspectrum current in multilayer stripline structures. vol. 2004.” vol. Paulotto. Lovat. [57] A. Galli. vol. vol. Lampariello. Boardman. 2005. and S. Paper 016617. pp. A. “Leakywave radiation from planar negativerefractiveindex transmissionline metamaterials. 2.” IEEE Trans. 2005. Zharov.. K. G.. Vegni. 53. 7.” Submitted to IEEE Trans. Baccarelli. Alu. A. Viitanen and S. pp. and S. Electromagnetic Theory Symp. “Nonlinear surface waves in lefthanded materials. and P. Kivshar. J. A. 305–308. 2005. Galli. Antennas Propagat.” Phys.” in IEEE Antennas and Propagation Society International Symposium. 47. May 2004. A: Pure Appl. Burghignoli. Frezza. pp.in Microwave Symposium Digest. “Modal properties of surface and leaky waves propagating at arbitrary angles along a metal strip grating on a grounded slab. Frezza. V. Frezza. [59] A. Opt. F. pp. A. June 2004. Eleftheriades. Jan. F. [61] P. pp. G.
5991–5998.” vol.: Condens. 16. June 2003.. pp.” J. pp. Jan. Kivshar. A. and A. “Surface polaritons and extinction properties of a lefthanded material cylinder. Rev. [66] T. 2005. pp. [73] S. Zharov. Y. 1811–1819. [68] R. [67] M. Ruppin.” J. A. pp. vol. [65] S.” J. B. Phys.: Condens. 51. 2004. pp. p.” Phys. 2558–2571. vol. Gabrielli. Y. [71] R. Shadrivov. Engheta. Oct. 15. Matter. pp. 972–973.[64] A. Prog. J. Ruppin. Aug. 2004. V. 193 . Aug. A. vol. “Surface polaritons of a lefthanded material slab. Apr. 277.” Rep.” in Quantum Electronics Conference. vol. 286–288. Ruppin. 40. Matter. Phys. vol. tunneling and transparency. I. A. A. S. “Surface polaritons on lefthanded cylinders: A complex angular momentum analysis. Lett. [69] C. and P. Beruete. A. 61–64. Phys. Antennas Propagat. [72] I. pp. 55. 72. 2001. M. Gao. Decanini. June 2004. vol. “Subwavelength slotted corrugated plate with enhanced quasioptical millimeter wave transmission. Sorolla. 449–521. “Surface polaritons of a lefthanded medium. vol. pp. “Physics of negative refractive index materials. vol. 4743–4751. Itoh.” Phys. Aug.. Campillo. Boardman. “Invited paper: Prospects for metamaterials. 2003. Rev. 2003. 13. “Pairing an epsilonnegative slab with a munegative slab: Resonance. Ramakrishna. Jan. A. D. 68. S. Folacci. Tang and L. “Surface polaritons and imaging properties of a multilayer structure containing negativerefractiveindex materials. Alu and N. and J.” IEEE Trans. [70] R. Dolado. 16. Ancey. Matter.: Condens. Phys. 2005. “Surface polaritons of nonlinear lefthanded materials. Oct. Sukhorukov..” Electron. 2000.
90. Paulotto. 1998. Lovat. [78] P. “Uniform analytical representation of the continuous spectrum excited by dipole sources in a multilayer dielectric structure through weighted cylindrical leaky waves. Burghignoli. Burghignoli. vol. New York. A.. and D. [79] D. F..” vol. 1964. Pozar. Baccarelli. 194 . Antennas Propagat. Lett. Stegun. Mar. R. [77] G. 269– 272. Baccarelli. G. F. Frezza. Lovat. 653–665. A. M. P. pp. 1545– 1549.and doublenegative metamaterial grounded slabs. vol. and S. June 2004. Graphs. Jackson.” Phys. vol. G. Rev.” in Microwave Symposium Digest. 1. Lampariello. and Mathematical Tables. Nov. 1994. Microwave Engineering. Feb. vol. “Surfacewave suppression in a doublenegative metamaterial grounded slab.” IEEE Trans. [80] M. “Universal features of the time evolution of evanescent modes in a lefthanded perfect lens. GomezSantos. 52. 309–312. Frezza.” IEEE Trans. Antennas Propagat. “The nature of radiation from leaky waves on single. 2. 2004 IEEE MTTS International. Wiley. A. P. Galli. 2004. pp. Handbook of Mathematical Functions with Formulas. [75] P. G. Burghignoli. pp. Haddad and D. Pozar. Paulotto. pp. P. P. 2003. [76] P. “Anomalous mutual coupling between microstrip antennas.. New York: Dover. and S. R. Abramowitz and I. Galli. Baccarelli.[74] P. 2003. Paper 077401. 42. M. Lovat.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.