Wave propagation in photonic crystals and metamaterials: surface waves,

nonlinearity and chirality

by

Bingnan Wang

A dissertation submitted to the graduate faculty

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Condensed Matter Physics

Program of Study Committee:

Costas M. Soukoulis, Major Professor
Rana Biswas
Marshall Luban
Gary Tuttle
Marzia Rosati

Iowa State University

Ames, Iowa

2009

Copyright ⃝
c Bingnan Wang, 2009. All rights reserved.
 ii

TABLE OF CONTENTS

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

CHAPTER 1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Photonic crystals and metamaterials . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Introduction to photonic crystals . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2.1 A brief review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2.2 Wave propagation in photonic crystals . . . . . . . . . . . . . . . . . . . 4

1.2.3 Band structure of our photonic crystal . . . . . . . . . . . . . . . . . . . 7

1.3 Introduction to Metamaterials . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.3.1 Wave propagation in double-negative metamaterials . . . . . . . . . . . 9

1.3.2 A brief review of metamaterials . . . . . . . . . . . . . . . . . . . . . . . 11

1.4 Outline of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

CHAPTER 2. Surface modes and waveguiding of photonic crystals . . . . . 15

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.1.1 Surface waves in photonic crystals . . . . . . . . . . . . . . . . . . . . . 15

2.1.2 Line defects and waveguides in photonic crystals . . . . . . . . . . . . . 18

2.2 The excitation of surface modes . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.2.1 ATR method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.2.2 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.3 Beaming and enhance transmission through photonic crystal waveguide . . . . 25

 iii

2.3.1 Coupling to subwavelength apertures . . . . . . . . . . . . . . . . . . . . 26

2.3.2 The experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.3.3 Enhanced transmission . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.3.4 Directed emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.4 Experimental verification of quantized optical conductance in photonic crystal

waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.4.1 Introduction to quantized conductance . . . . . . . . . . . . . . . . . . . 38

2.4.2 Experimental design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

2.4.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

CHAPTER 3. Study of nonlinear split-ring resonators . . . . . . . . . . . . . 45

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.2 Varactor diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

3.3 Self-tuning of SRR with varactor . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.4 The simplified driven nonlinear resonator model . . . . . . . . . . . . . . . . . . 52

3.5 Nonlinear SRR with back-to-back varactors . . . . . . . . . . . . . . . . . . . . 55

3.6 Coupling between nonlinear SRRs . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

CHAPTER 4. Chiral metamaterials . . . . . . . . . . . . . . . . . . . . . . . . 59

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

4.2 Wave propagation in chiral media . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.3 Effective parameter retrieval of chiral media . . . . . . . . . . . . . . . . . . . . 63

4.4 Review of work on chiral metamaterials . . . . . . . . . . . . . . . . . . . . . . 65

4.4.1 Theoretical studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

4.4.2 Experimental realizations . . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.5 Nonplanar chiral metamaterial based on chiral SRRs . . . . . . . . . . . . . . . 71

4.5.1 The chiral SRR design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.5.2 Experimental method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

 iv

4.5.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4.6 A microwave absorber made from chiral metamaterial . . . . . . . . . . . . . . 77

4.6.1 Loss problem and metamaterial absorbers . . . . . . . . . . . . . . . . . 77

4.6.2 The chiral metamaterial absorber . . . . . . . . . . . . . . . . . . . . . . 79

4.6.3 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

CHAPTER 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
 v

LIST OF FIGURES

Figure 1.1 The 2D photonic crystal composed of periodically arranged alumina

rods. The alumina rods have square cross sections and are infinitely

long in z direction. The array extends infinitely in x and y directions. . 7

Figure 1.2 The band structure of the 2D photonic crystal is shown in the main fig-

ure. The solid lines show the bands of TM polarization and the dashed

lines show the bands of TE polarization. The gray area indicates the

bandgap of TM polarization. The primitive cell of the reciprocal lattice

is shown in the inset. Left figures indicate TM and TE polarizations

and the photonic crystal structure in xy plane. . . . . . . . . . . . . . . 8

Figure 1.3 (a) The split-ring resonator structure. (b) The first metamaterial with

negative index demonstrated by experiments [1]. . . . . . . . . . . . . . 12

Figure 2.1 (a) The projected band structure of the photonic crystal. Below the

figure shows the supercell and the arrow indicates that the supercell is

periodic in that direction. Please see text for the parameters. (b) The

projected band structure of the photonic crystal with surface modifica-

tions. The surface layer is composed of circular rods with diameter of

1.83 mm. Below the figure shows the supercell and the arrow indicates

that the supercell is periodic in that direction. . . . . . . . . . . . . . . 17

 vi

Figure 2.2 (a) The projected band structure of the photonic crystal calculated

with the supercell indicated in the inset. The arrow indicates that the

supercell is periodic in that direction. (b) The projected band structure

of the photonic crystal with line defect, calculated with the supercell

shown in the inset. The arrow indicates that the supercell is periodic

in that direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Figure 2.3 Experimental setup of the ATR measurement. The HP8510 network

analyzer and a pair of horn antennas are used to measure the reflection

spectrum of the prism. The photonic crystal is put behind the prism. . 20

Figure 2.4 The experiment (solid lines) and simulation (circles) results of reflection

at incident angle 53◦ for the photonic crystal with corrugation layer

of circular rods with D = 2.44 mm. From top to bottom shows the

reflection when the distance g of the air gap between the prism and the

surface of the structure increases. . . . . . . . . . . . . . . . . . . . . . 22

Figure 2.5 Experimental surface band structure of a 2D photonic crystal. Be-

tween the light gray areas exist the band gap of photonic crystal. Solid

lines and square dots with error bars are supercell band structure cal-

culation and experimental results of the surface band for (a) photonic

crystal with a corrugation layer of circular rods with D = 1.83 mm,

(b) photonic crystal with a corrugation layer of circular rods with

D = 2.44 mm, (c) a single layer of rods with D = 1.83 mm and (d)

a single layer of rods with D = 2.44 mm. . . . . . . . . . . . . . . . . . 24

Figure 2.6 The simulated electric field distribution with (a) a regular photonic

crystal waveguide formed by a line defect, (b) the same waveguide as

in (a) with a modified layer added and (c) the same waveguide as in

(b) with a grating layer added. The dashed lines in the figures show

the position of the structures. The structures are shown under the

corresponding field distributions. . . . . . . . . . . . . . . . . . . . . . 28

 vii

Figure 2.7 (a) The structure’s details and definition of the different boundaries

used. bnd0, bnd1, bnd2, bnd3 are the boundaries in the channel, at the

end of the photonic crystal, in the middle of the air part and at the end

of the air part respectively.(b) The shape of the surface layers in each

of the fifth different structures. . . . . . . . . . . . . . . . . . . . . . . 30

Figure 2.8 Simulated power flow versus frequency measured at boundary bnd0 for

the four structures which are only waveguide inside the photonic crystal

(structure1), the waveguide with the grating layers (structure2), the

waveguide with the modified layers (structure3), and the waveguide

with both the modified and the grating layers (structure4). . . . . . . . 32

Figure 2.9 Simulation and experimental power flow versus frequency for structure4,

which possesses both the gratings and the modified layers. . . . . . . 32

Figure 2.10 Power flow versus frequency for structure4 (simulation only) and struc-

ture5 (simulation and experiment). Notice that structure5 is similar to

structure4 except for the one grating missing in front and the end of

the waveguide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Figure 2.11 A 2d plot of the strength of the outgoing electric field for the case

of structure4. (a) and (b) are the experimental and the simulation

results respectively at frequency f = 11.85 GHz. (c) and (d) are the

experimental and the simulation results respectively at frequency f =

12.47 GHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Figure 2.12 The line plots of field intensity for structure4 at 11.85 GHz, corresponds

to the 2D distribution shown in Fig. 2.11. for (a) The field intensity

along the central line of the waveguide exit (horizontal direction in

Fig. 2.11(a)) for structure4 at 11.85 GHz. (b) The field intensity along

the interface of the waveguide end (vertical direction in Fig. 2.11(b)),

at the horizontal position indicated by the arrow in (a). The solid lines

are numerical results and dashed lines are experimental results. . . . . 35

 viii

Figure 2.13 A 2d plot of the strength of the electric field for structure5. (a) and

(b) are the experimental and the simulation results respectively at fre-

quency f = 11.85 GHz. (c) and (d) are the experimental and the

simulation results respectively at frequency f = 12.47 GHz. . . . . . . 36

Figure 2.14 Simulated 2d plot of the strength of the electric field of structure3 at

(a) f = 11.97 GHz and (b) f = 11.33 GHz. . . . . . . . . . . . . . . . . 37

Figure 2.15 An illustration of the experimental setup. . . . . . . . . . . . . . . . . 41

Figure 2.16 The experimental (circles) and numerical (solid line) results of optical

conductance of a photonic crystal waveguide at 10 GHz. . . . . . . . . 42

Figure 2.17 The experimental (circles) and numerical (solid line) results of optical

conductance of a metal channel for TM case (E field perpendicular to

the plane). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Figure 3.1 The schematic symbol of a varactor (left) and the equivalent circuit

model of a varactor (right). . . . . . . . . . . . . . . . . . . . . . . . . 46

Figure 3.2 Characteristic curves of the varactor Skyworks SMV1231-079. (a) The

capacitance of the varactor vs. the applied voltage. (b) The dissipative

current vs. the applied voltage. . . . . . . . . . . . . . . . . . . . . . . 47

Figure 3.3 (a) The fabricated SRR is shown on left and the SRR with a varactor

mounted into the gap is shown on right. (b) The loop antenna is posi-

tioned on top of the SRR and connected to the network analyzer (not

shown). The reflected signal is measured by the network analyzer. . . . 48

Figure 3.4 The reflection of the SRR loaded with one varactor, measured by a loop

antenna. The curves show the measurement results at input power from

-15 dBm to 9 dBm in 3 dBm steps. . . . . . . . . . . . . . . . . . . . 50

Figure 3.5 The hysteresis effect at high input power levels. The blue curves are

measured for forward sweep (the source frequency of the network ana-

lyzer is scanned from low to high) and the red dashed curves are mea-

sured for reverse sweep (frequency is scanned from high to low). . . . . 51

 ix

Figure 3.6 The circuit model of the coupling between the driven loop antenna and

the SRR loaded with one varactor. . . . . . . . . . . . . . . . . . . . . 52

Figure 3.7 The oscillation amplitude vs. source frequency calculated from the

simplified nonlinear oscillator model. The curves from bottom to top

correspond to excitation power from low to high. The blue arrow shows

the jump for forward sweep of the top curve and the red arrow shows

the jump for reverse sweep of the top curve. . . . . . . . . . . . . . . . 54

Figure 3.8 The reflection of the SRR loaded with two back-to-back varactors, mea-

sured by a loop antenna. The curves show the measurement results at

power levels from -14 dBm to 2 dBm, in 2 dBm steps. The inset shows

the sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

Figure 3.9 The reflection of (a) two linear SRRs and (b) two nonlinear SRRs at

input power of −15 dBm. The legend shows the distance between the

two SRRs. The distance between the first SRR and the antenna is fixed. 57

Figure 3.10 The 2d plot of the reflection of two coupled nonlinear SRRs. The x-

axis shows the distance between the two SRRs and the y-axis shows

the frequency. The four figures from left to right display the result at

input power from -15 dBm to 0 dBm. The SRRs and the antenna are

arranged such that they have the same axis and the planes of the rings

are parallel to each other. The distance between the loop antenna and

the first SRR is fixed and the second SRR is movable along its axis, as

seen from the inset. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

Figure 4.1 The transmission and reflection coefficients of a plane wave incident

upon a chiral slab from left. . . . . . . . . . . . . . . . . . . . . . . . . 64

Figure 4.2 Different designs of chiral metamaterials. (a) The twisted “swiss roll” [2];

(b) The canonical helix [3]; (c) The chiral SRR [4]; (d) The bi-layer

“swastika” [5]; (e) Dual strips with tilted bridge [6]; (f) Bi-layer rosettes [7] 66
 x

Figure 4.3 (a) Schematic representation of one unit cell of the cross-wire structure.

(b) Photograph of one side of a fabricated microwave-scale cross-wire

sample. The geometry parameters are given by ax = ay = 15mm,

l = 14 mm, w = 1 mm, s = 1.6 mm, ϕ0 = 45◦ , ϕ = 30◦ . (Adapted from

Ref. [8]) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Figure 4.4 The simulated current density distribution for the right circularly po-

larized EM wave at 6.5 GHz (a) and for the left circularly polarized EM

wave at 7.5 GHz (b). (Adapted from Ref. [8]) . . . . . . . . . . . . . . 70

Figure 4.5 (a) The structure of the chiral SRR. (b) A unit cell of the 2D chiral

metamaterial used in simulation. (c) The chiral SRRs are fabricated

on PCBs and cut into strips (d) The strips are interlocked to form the

chiral metamaterial slab. . . . . . . . . . . . . . . . . . . . . . . . . . . 72

Figure 4.6 The surface current distribution of the SRR near resonance. The cir-

culating current forms a magnetic dipole, causes the opposite charge

accumulation on the bottom SRR (+) and the top SRR (-) and forms

an extra electric dipole (see green straight arrow). . . . . . . . . . . . 73

Figure 4.7 The transmission and reflection coefficients of the chiral metamaterial

slab is measured by a pair of horn antennas served as transmitter and

receiver. The cables on the horn antennas are connected to a vector

network analyzer (not shown). . . . . . . . . . . . . . . . . . . . . . . 74

Figure 4.8 From top to bottom, the transmission and reflection amplitudes for

LCP and RCP waves, the azimuth rotation angle for linearly polarized

waves, and the ellipticity. Simulation results are shown on left and

experiment results are shown on right. . . . . . . . . . . . . . . . . . . 75

 xi

Figure 4.9 The retrieved effective parameters of the chiral metamaterial. The fig-

ures from top to bottom are the index of refraction for the two circularly

polarized waves n+ and n− , the effective index of refraction n together

with the chiral parameter κ, the effective relative permittivity ϵ and

relative permeability µ. Simulation results are shown on left and ex-

periment results are shown on right. . . . . . . . . . . . . . . . . . . . 76

Figure 4.10 (a) The chiral metamaterial slab. (b) The fabricated microwave absorber. 80

Figure 4.11 The simulation results of absorption at different angles θ for (a) TE

polarization and (b) TM polarization. Insets illustrate the two polar-

izations and the angle θ. . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Figure 4.12 The experimental results of absorption at different angles θ for (a) TE

polarization and (b) TM polarization. Insets illustrate the two polar-

izations and the angle θ. . . . . . . . . . . . . . . . . . . . . . . . . . . 82

 xii

ACKNOWLEDGEMENTS

I would like to take this opportunity to thank the people that helped me or inspired me

through my graduate studies at Iowa State. It would be difficult to list all the great people

I met here and all the details of their help to me. They made my experience at Iowa state

enjoyable and I am grateful to all of them.

First of all I would like to express my gratitude to Prof. Costas M. Soukoulis. I want

to thank him for advising my Ph.D. study at Iowa State University, for directing me to the

field of photonic crystals and metamaterials, for guiding me through my research projects, for

helping me with publications and this thesis, for the financial support through all my years at

Iowa State University.

Thank Prof. Rana Biswas, Prof. Marshall Luban, Prof. Gary Tuttle and Prof. Marzia

Rosati for serving in my program of study committee. The courses I have taken with Prof.

Marshall Luban and Prof. Rana Biswas are very helpful to my research. Thank Prof. Gary

Tuttle for offering me the opportunity to do experiments in microelectronic research center

and all the help.

I thank Dr. Thomas Koschny very much for the help through out my studies, for patiently

answering all my stupid questions, for all the guidance and suggestions on my research projects.

I learned a lot from him. I would like to thank the peer members in our group. Thank Dr.

Jiangfeng Zhou , Dr. Lei Zhang, Anan Fang and Weitao Dai for all the helpful discussions and

suggestions. I wish them the best in their career.

Most importantly, I would like to thank my parents for their consistent support and love.

Also I would like to express my gratitude to my fiancee Wenhong Liu. You made my years at

Iowa State unforgettable.

 xiii

ABSTRACT

Photonic crystals and metamaterials, both composed of artificial structures, are two in-

teresting areas in electromagnetism and optics. New phenomena in photonic crystals and

metamaterials are being discovered, including some not found in natural materials. This thesis

presents my research work in the two areas.

Photonic crystals are periodically arranged artificial structures, mostly made from dielectric

materials, with period on the same order of the wavelength of the working electromagnetic

wave. The wave propagation in photonic crystals is determined by the Bragg scattering of the

periodic structure. Photonic band-gaps can be present for a properly designed photonic crystal.

Electromagnetic waves with frequency within the range of the band-gap are suppressed from

propagating in the photonic crystal. With surface defects, a photonic crystal could support

surface modes that are localized on the surface of the crystal, with mode frequencies within the

band-gap. With line defects, a photonic crystal could allow the propagation of electromagnetic

waves along the channels. The study of surface modes and waveguiding properties of a 2D

photonic crystal will be presented in Chapter 1.

Metamaterials are generally composed of artificial structures with sizes one order smaller

than the wavelength and can be approximated as effective media. Effective macroscopic pa-

rameters such as electric permittivity ϵ, magnetic permeability µ are used to characterize the

wave propagation in metamaterials. The fundamental structures of the metamaterials affect

strongly their macroscopic properties. By designing the fundamental structures of the meta-

materials, the effective parameters can be tuned and different electromagnetic properties can

be achieved. One important aspect of metamaterial research is to get artificial magnetism.

Metallic split-ring resonators (SRRs) and variants are widely used to build magnetic meta-
 xiv

materials with effective µ < 1 or even µ < 0. Varactor based nonlinear SRRs are built and

modeled to study the nonlinearity in magnetic metamaterials and the results will be presented

in Chapter 3.

Negative refractive index n is one of the major target in the research of metamaterials.

Negative n can be obtained with a metamaterial with both ϵ and µ negative. As an alternative,

negative index for one of the circularly polarized waves could be achieved with metamaterials

having a strong chirality κ. In this case neither ϵ nor µ negative is required. My work on chiral

metamaterials will be presented in Chapter 4.

 1

CHAPTER 1. Overview

1.1 Photonic crystals and metamaterials

The discoveries in photonic crystals and metamaterials have largely broadened the vision

of electromagnetism and optics. Photonic crystals and metamaterials are both composed of

periodic structures and can mold the flow of electromagnetic (EM) waves in ways cannot

be achieved in natural materials. New phenomena, including but not limited to negative

refraction, artificial magnetism, near-field lensing and focusing, optical cloaking, have been

discovered and studied.

For homogeneous materials, the elemental structures (molecules for instance) are much

smaller than the propagating wavelength so that the propagating waves do not “feel” them.

On the contrary, both photonic crystals and metamaterials are composed of artificial structures

that are not negligible in size compared with the propagating wavelength. They are both

spatially dispersive media in this sense.

Photonic crystals are generally composed of periodic arrangements of structures with dif-

ferent permittivity. The period of the structures a is on the same order with the wavelength of

the propagating waves λ. Since the period of the structures is comparable with the wavelength,

EM waves will be diffracted. Thus photonic crystals generally are not considered as effective

media. Bloch waves and band structures are used to characterize the wave propagating proper-

ties. When designed properly, stop bands can exist in the band structure of a photonic crystal.

EM waves within the frequency range of the stop bands become evanescent inside the photonic

crystal.

Metamaterials are generally composed of artificial structures with sizes one order smaller

than the wavelength (λ/a ∼ 10). Although the ratio λ/a is not as large as natural materials, the
 2

metamaterials can already be considered as effective media. The wave propagating properties

of metamaterials can be characterized by effective macroscopic parameters and averaged fields.

The fundamental structures of the metamaterials affect strongly the macroscopic properties.

By designing the fundamental structures of the metamaterials, the effective parameters such

as electric permittivity ϵ, magnetic permeability µ and chirality κ can be tuned and different

EM properties can be achieved.

1.2 Introduction to photonic crystals

1.2.1 A brief review

The boost of research on photonic crystals starts from 1987, when Yablonovitch [9] and

John [10] independently introduced the concept of photonic band-gap materials. Long before

that, one-dimensional (1D) periodic structures have been studied. It has been discovered more

than 100 year ago that 1D periodic dielectric stacks have a stop band, which is a spectral range

that has high reflectivity. This feature is used to make distributed Bragg reflectors (DBRs).

The term “photonic crystal” is generally referring to two-dimensional (2D) or three-dimensional

(3D) periodic structures, although DBRs are sometimes called 1D photonic crystals.

Since the work of Yablonivitch [9] and John [10], more and more efforts have been put

into photonic crystals by many researchers. The primary goal is to design and characterize

photonic crystals with a total band-gap in 3D, where the propagation of electromagnetic wave

is prohibited in all directions. In 1990, Ho et.al. demonstrated the existence of a full band-gap

in 3D with a diamond-like periodic structure, by numerical calculations [11]. In 1991, the

band-gap in 3D structures were proved by experiments in microwave regime [12]. The same

phenomenon is expected to happen in optical regime for the same structure scaled down in

geometry, due to the scaling law [13]. However, the diamond-like structure is too complicated

for microfabrication. Later on, the Iowa State group proposed a layer-by-layer “woodpile”

structure that give a 3D band-gap [14]. This structure is easier to fabricated and works over

a wide range of structural parameters.

Due to the existence of band-gap in properly designed photonic crystals, many interesting
 3

applications can be realized to manipulate the flow of electromagnetic waves and light [15]. For

example, by introducing point defects in photonic crystals, high-Q cavities can be fabricated.

By introducing line defects in photonic crystals, waveguides can be realized, where light is

guided along the line defects. The guiding mechanism of photonic crystal waveguides is different

than dielectric planar waveguides. For a dielectric waveguide, the dielectric constant of the

waveguide is higher than its surrounding materials, and light is guided in the dielectric due to

total internal reflection. For a photonic crystal waveguide, light is confined in the waveguide due

to the fact that there exists a band-gap and light can not penetrate into the photonic crystal.

The dielectric constant of the waveguide does not have to be larger than the surroundings.

Actually, the waveguiding region can be lower-indexed materials including air. This is an

important advantage of photonic crystal waveguides since that the propagation loss can be

significantly reduced. Also, photonic crystal waveguides can bend light at sharp angles with

lower loss. Photonic crystal waveguides have found applications in photonic crystal fibers and

telecommunications [16]. We will discuss photonic crystal waveguides in more details in the

next chapter.

Other than band-gaps, there are other intriguing features of photonic crystals that can be

used to control the flow of light. Photonic crystals have rich dispersion relations, i.e., the rela-

tion between angular frequency ω of allowed modes and wavevector ⃗k is not a straight line. The

equal-frequency surfaces (EFSs) of a photonic crystal are rather complicated. By definition,

the group velocity of light inside a photonic crystal, ⃗vg = ∇⃗k ω(⃗k), is perpendicular to the EFS

of the photonic crystal. Depending on the dispersion relation, the direction of group velocity

can be very different than the phase velocity, which is determined by wavevector ⃗k. Since

the group velocity governs the energy flow of light, peculiar refraction phenomena can occur

at the interface between the photonic crystal and its surrounding. Negative refraction, where

the energy flow is refracted to the “wrong” direction, can be realized in photonic crystals [17].

Due to the different diffraction behavior at different frequencies, superprism effect has been

demonstrated, where extremely large angular dispersion can be achieved for light with different

frequencies [18].
 4

1.2.2 Wave propagation in photonic crystals

The wave propagation properties of a photonic crystal are determined by the Bragg scat-

tering of the periodic structures and described by its photonic band structure. Starting from

Maxwell’s equations, the photonic band structure can be obtained by solving the eigenvalue

problem.

The source-free Maxwell’s equations are

∂B(r, t)
∇ × E(r, t) = − , (1.1)
∂t
∂D(r, t)
∇ × H(r, t) = , (1.2)
∂t
∇ · D(r, t) = 0, (1.3)

∇ · B(r, t) = 0, (1.4)

where E and H are the macroscopic electric and magnetic fields, respectively; D and B are

the displacement and magnetic induction fields, respectively. All the above four field vectors

are functions of both space r and time t.

Consider isotropic (ϵ and µ are scalars), linear (higher order responses to external fields are

negligible) and non-dispersive (ϵ and µ are not functions of frequency) media, the constitutive

relations can be written as

D(r, t) = ϵ0 ϵ(r)E(r, t), (1.5)

B(r, t) = µ0 µ(r)H(r, t), (1.6)

where ϵ0 and µ0 are the vacuum permittivity and permeability, respectively; ϵ(r) and µ(r) are

the relative permittivity and permeability of the medium, respectively. Most materials are

non-magnetic and the relative permeability µ(r) = 1.

Adopt the time-harmonic convention, the solutions of the electric and magnetic fields can

be constructed as follows

E(r, t) = E(r)e−iωt , (1.7)

H(r, t) = H(r)e−iωt , (1.8)

 5

where ω is the angular frequency. Then the two curl equations in the Maxwell’s equations can

be rewritten as

∇ × E(r) = iωµ0 H(r), (1.9)

∇ × H(r) = −iωϵ0 ϵ(r).E(r) (1.10)

Combining the above two equations, we can get a wave equation in the frequency domain

for the magnetic field

1 ω2
∇ × ∇ × E(r) = E(r), (1.11)
ϵ(r) c2
( )
1 ω2
∇× ∇ × H(r) = H(r). (1.12)
ϵ(r) c2
√
where c = 1/ ϵ0 µ0 is the speed of light in vacuum. When the permittivity ϵ(r) is known,

one of the above wave equations can be solved to obtain the electric field or the magnetic

field. Then the other field vector is solved accordingly with the relations determined by the

Maxwell’s equations.

i
E(r) = ∇ × H(r), (1.13)
ωϵ0 ϵ(r)
i
H(r) = − ∇ × E(r). (1.14)
ωµ0

For photonic crystals, the permittivity is a periodic function of space,

ϵ(r) = ϵ(r + R), (1.15)

∑d
where lattice vector R = i=1 mi ai is a combination of primitive lattice vectors ai , where

i = 1, . . . , d describes the dimension. Consider an indefinitely extended photonic crystal, the

lattice vector extends in d-dimensional space and mi can be any integers.

From the Floquet-Bloch theorem, the solution of the wave equation (1.12) is of the form

of a periodically modulated plane wave

Hn,k (r) = un,k (r)eik·r , (1.16)

un,k (r) = un,k (r + R), (1.17)

 6

where the subscript n is the index of the discrete solutions, k is the corresponding wave vector

and un,k (r) satisfies

( )
1 ω2
(∇ + ik) × (∇ + ik) × un,k (r) = un,k (r). (1.18)
ϵ(r) c2

The corresponding eigen-frequency of the n-th eigenmodes wave vector k is ωn (k) and is a

continuous function of k. A discrete band is the formed if we plot ωn (k) versus k. If we plot all

n-th bands, a full band structure of the photonic crystal is formed and all possible eigenmodes

are casted in band structure.

More importantly, we do not need to calculate the eigen-solutions of all possible wave

vectors, thanks to the periodicity of structure. Denote the reciprocal lattice vector associated

with the given Bravias lattice as G, then we have G · R = 2mπ, where m is an integer. The

reciprocal lattice vector G is a Bravias lattice too and is further described by combination of
∑
the primitive vectors bj in the j-th dimension in the reciprocal space, G = dj=1 nj bj . The

primitive vectors in the reciprocal space is connected to the primitive vectors in the real space

by ai bj = 2πδij . The solution at k is identical to the solution at k + G.

Hn,k (r) = Hn,k+G (r), (1.19)

ωn,k = ωn,k+G . (1.20)

The primitive cell of the reciprocal lattice is called the first Brillouin zone. Given the above

information, we then only need to calculate the eigen-solutions of k in the first Brillouin zone.

For simple 1D photonic crystals, the band structure can be obtained analytically. However,

for 2D and 3D photonic crystals, the computation of band structures must rely on numerical

tools, due to the complex photonic crystal structure. There are a number of well-developed

computational methods to calculate the band structure and wave propagation in photonic

crystals, including plane-wave expansion method, finite-difference time-domain method, finite-

element method.
 7

1.2.3 Band structure of our photonic crystal

Our focus will be on the surface modes and wave guiding properties of 2D photonic crystals.

A 2D photonic crystal is considered to be translation invariant in the third dimension. Although

2D photonic crystals do not have a full bandgap in all directions, they have found many

applications in light confinement, such as waveguides, microlasers and integrated photonic

circuits [13]. The 2D photonic crystal in our study is composed of alumina rods (permittivity

ϵ = 9.8) of square cross section (with side length of 3.18 mm), and arranged in a square lattice

(with lattice constant a = 11 mm). We assume the square array extends infinitely into x and

y directions and the rods are infinitely long in z direction (see Fig. 1.1). The numerically

computed band structure of the 2D photonic crystal is shown in Fig. 1.2. It is shown that

the 2D photonic crystal has a band-gap between 9.43 and 12.78 GHz for TM polarization

(electric field along the rods). In the following chapter, we study the surface modes and the

wave guiding properties of the 2D photonic crystal, both for TM polarization.

Figure 1.1 The 2D photonic crystal composed of periodically arranged alu-

mina rods. The alumina rods have square cross sections and are
infinitely long in z direction. The array extends infinitely in x
and y directions.
 8

Figure 1.2 The band structure of the 2D photonic crystal is shown in the
main figure. The solid lines show the bands of TM polarization
and the dashed lines show the bands of TE polarization. The
gray area indicates the bandgap of TM polarization. The prim-
itive cell of the reciprocal lattice is shown in the inset. Left fig-
ures indicate TM and TE polarizations and the photonic crystal
structure in xy plane.

1.3 Introduction to Metamaterials

Electromagnetic metamaterials are usually periodically arranged artificial structures that

show peculiar properties, such as negative refraction, which are not seen in natural materials

(for reviews of the metamaterial field, see [1, 19, 20]). The elemental structures of the metama-

terials are typically much smaller in size relative to the wavelength such that the metamaterials

can be approximated as homogeneous. Macroscopic effective parameters such as electric per-

mittivity ϵ and magnetic permeability µ can then be used to describe the electromagnetic (EM)

properties of the metamaterials [21]. The fascinating, but challenging work of scientists and

engineers in this area is to design new structures, or the photonic atoms, to achieve the EM

properties of metamaterials for specific applications. There are different types of metamateri-

als, depending on their effective parameters. When both ϵ and µ are negative, the metamaterial

is called “double-negative” In this case, the effective index of refraction n is negative. When

either ϵ or µ is negative but the other parameter is positive, the metamaterial is called “single-
 9

negative”. Metamaterials with small or even close to zero ϵ or µ are also interesting. When

both ϵ and µ are close to zero, the metamaterial has an effective index close to zero [22].

1.3.1 Wave propagation in double-negative metamaterials

The first and the most extensively studied type of metamaterials is “double-negative”

metamaterial. The work of Veselago [23] in 1958 is generally recognized as the first in this

field. In his pioneer work, Veselago discussed the exotic properties of a material with both ϵ

and µ negative. A “double-negative” media is interesting at first glance. Indeed, when both ϵ

and µ are negative, ϵµ still gives a positive value. From k 2 = ω 2 ϵµ, the wavevector k is real

in this case. Thus wave propagation is permitted in a double-negative medium. It is then

desirable to examine more rigorously if wave propagation in double-negative media is allowed

by physics laws. It is also interesting to find the differences of EM waves in a double-negative

medium and a regular (double-positive) medium.

Consider a monochromatic plane wave of the form

E(r, t) = E0 ei(k·r−ωt ) (1.21)

H(r, t) = H0 ei(k·r−ωt ) (1.22)

where E0 and H0 are the amplitudes of the electric and magnetic field respectively. Fit the

plane wave into the Maxwell’s equations (1.1)-(1.4), together with the constitutive relations

(1.5)-(1.6), we come to the following relations

k × E = ωµ0 µH (1.23)

k × H = −ωϵ0 ϵE (1.24)

For regular materials, ϵ > 0 and µ > 0, thus the triplet E, H and k form a right-handed triad.

However, for a double-negative medium, ϵ < 0 and µ < 0, the above relations can be rewritten

as

k × E = −ωµ0 |µ|H (1.25)

k × H = ωϵ0 |ϵ|E (1.26)

 10

and the triplet E, H and k new form a left-handed triad. For this reason, Veselago [23] called

a medium with both ϵ and µ negative a “left-handed” medium.

Meanwhile, the direction energy flux, is determined by the Poynting vector

S=E×H (1.27)

and is not affected by the values of ϵ and µ. The triplet E, H and S always form a right-handed

triad. Thus for double-negative media, the energy flux (determined by S) and the wavefronts

(determined by k) move in opposite directions. The index of refraction in a double-negative

√
medium is given by n = − ϵµ and and has a negative value. Thus a double-negative medium

is also called “negative-index” medium.

EM waves in double-negative media must not break the energy conservation law. Consider

the time-averaged energy density in a non-dispersive medium is given by [24]

1 1
U = ϵ0 ϵ|E|2 + µ0 µ|H|2 (1.28)
4 4

If a double-negative medium is non-dispersive, the energy density would be negative and is

unphysical. Thus the medium must be dispersive, i.e., ϵ and µ are both functions of ω. For a

quasi-monochromatic wave packet in a dispersive medium, the time-averaged energy density

is given by

1 ∂(ϵω) 2 1 ∂(µω)
U = ϵ0 |E| + µ0 |H|2 (1.29)
4 ∂ϵ 4 ∂µ

and the conditions ∂(ϵω)/∂ϵ > 0, and ∂(µω)/∂µ > 0, must be both satisfied. Actually when

ϵ < 0 and µ < 0 the medium must be highly dispersive, i.e., ∂ϵ/∂ω > |ϵ|/ω and ∂µ/∂ω >

|µ|/ω [25].

Most current realizations of double-negative metamaterials rely on highly resonant struc-

tures, where negative ϵ and negative µ are obtained around the corresponding resonances [19,

21]. Propagation losses of EM waves in metamaterials are inevitable. For passive media, it

is required by energy conservation that the imaginary parts Im ϵ > 0 and Im µ > 0. Also,

the values of ϵ and µ are not randomly chosen but are both limited by Kramers-Krönig rela-

tions [24].
 11

1.3.2 A brief review of metamaterials

Veselago’s left-handed medium [23] was not realized until more than three decades later,

due to the lack of natural materials satisfying the parameter requirements. Negative ϵ can

occur in plasmas and plasmonic materials such as noble metals for frequencies below the

plasma frequency. For bulk noble metals, the plasma frequency is typically in the infrared

regime. For a medium composed of thin metallic wires, the effective plasma frequency can be

tuned down to microwave frequencies [26]. However, negative µ is more difficult to achieve. In

natural materials, diamagnetic responses are usually weak and µ can only be slightly smaller

than unity. In 1999, Pendry [21] proposed a way to achieve negative µ with a kind of artificial

structures called split-ring resonators (SRRs). The SRRs are composed of metallic rings with

gaps (see Fig. 1.3(a)) to behave as resonators. Although made of non-magnetic materials,

the SRRs are magnetically active and can give strong responses to magnetic fields around the

resonance frequency. By tuning the geometric parameters of SRRs, a band of negative µ can

be obtained at the desired frequency range. If designed properly, the combination of these

two sets of resonators can give both negative ϵ and negative µ in the same frequency band.

This was realized with microwave experiments by Smith et.al. [27] in 2000. The same group

further demonstrated negative refraction by measuring the output signal of EM waves refracted

by a metamaterial in a wedge shape [28]. Fig. 1.3(b) shows the fabricated metamaterial

composed of SRRs and metallic wires. Since then, the SRR and metallic wire structures were

studied extensively by research groups all over the world, both theoretically and experimentally.

Applications of metamaterials, such as super-lens [1, 29], and invisible cloak [30, 31], were

proposed and studied.

By scaling down the SRR size and simplifying the structure, magnetic metamaterials were

fabricated to show magnetic responses from THz to infrared region [32–35]. However, for the

SRR structures, the magnetic field needs to have a component perpendicular to the plane to

obtain the magnetic resonance. Normal incidence of EM waves to the planar SRR structures

does not excite directly the magnetic resonance. Plus, the SRR and wire combination is

too complicated to fabricate in nanoscale, especially when three-dimensional (3D) medium is

 12

Figure 1.3 (a) The split-ring resonator structure. (b) The first metamate-
rial with negative index demonstrated by experiments [1].

needed.

New designs were proposed to achieve negative refraction towards the optical regime [36–

39]. Unlike the SRRs, these new designs are typically double-layered metallic structures, with

a dielectric substrate in the middle. The two metal layers on each side of the substrate are

identical and composed of short wires, metal plates, or fishnet-like structures. While the

parallel dipole resonance in each metal structure gives negative ϵ, the negative µ is provided

by the anti-parallel resonance in the metal structure pairs separated by the substrate. So

normal incidence of EM waves is supported. Bulk media can be fabricated by stacking the

layered structures. Recently, negative refraction was demonstrated at optical frequencies with

stacked fishnet-like structures [40, 41].

On the other hand, another type of metamaterials, the so-called indefinite metamaterials,

has also been studied. In contrast to isotropic metamaterials, whose ϵ and µ are the same in

all directions, an indefinite medium has anisotropic ϵ and µ tensors, with some components

having an opposite sign to the rest. In 2001, Lindell et.al. [42] showed that some indefinite

media can support backward waves. Smith and colleagues [43, 44] then analyzed the EM

wave propagating properties of indefinite media in detail [43] and showed that some types

of indefinite media can have the similar negative refraction effect as isotropic metamaterials.
 13

Partial focusing of a point source by an indefinite medium slab was also described. The wire

medium, which is composed of metal wire arrays in one direction only, is an important medium

of this kind. The wire medium is anisotropic and has a negative ϵ along the direction of the

wires below plasma frequency. The optical response of such a wire medium is not limited to

resonances, so that the medium has a higher tolerance to fabrication defects and the loss is

smaller than strongly resonant structures [45, 46]. Moreover, the wire array does not even has

to be periodic. Subwavelength imaging and negative refraction of the wire medium in infrared

and optical frequencies were studied theoretically and numerically [45, 46]. Recently, optical

negative refraction has been demonstrated experimentally with silver nanowires [47].

1.4 Outline of the thesis

In Chapter 2, the surface waves and waveguiding properties of a 2D photonic crystal will

be studied. The surface modes can be detected by attenuated total reflection (ATR) tech-

nique. The surface mode dispersion of photonic crystals with different surface defect layers

will be studied experimentally. Then the photonic crystal waveguides are studied. By mak-

ing use of surface modes, the waveguiding properties of photonic crystal waveguides can be

improved. Enhanced transmission and directed emission can be achieved through photonic

crystal waveguides of subwavelength aperture. The experimental work on optical conductance

of photonic crystal waveguides will also be presented. The experimental results show that the

optical conductance of photonic crystal waveguides are quantized.

The next two chapters are studies on metamaterials. In the introductions above, we have

assumed that the responses of the metamaterial to EM waves are linear, i.e., effective ϵ and

µ of the metamaterial do not change with the intensity of EM fields. On the other hand, it

is interesting to study the transmission properties of metamaterials in nonlinear regime. One

interesting idea is that, if the responses of metamaterials depends on the intensity of EM fields,

the parameters ϵ/µ can then be tuned by external fields [48]. It is possible, for example, to

switch µ from negative to positive in a particular frequency band thus make a double-negative

metamaterial single-negative. Other phenomena in nonlinear metamaterials such as higher

 14

harmonics generation [49], discrete breathers [50] have also been studied by other research

groups. SRRs are key building blocks of metamaterials and it is thus important to study the

nonlinear responses of SRRs. In Chapter 3, our study on nonlinear SRRs will be presented.

Other than double-negative metamaterials, chirality as another route to negative index of

refraction has also been proposed and studied. Proposed by Pendry [2], chiral metamaterials

have been shown to possess negative index for one of the circularly polarized waves. Instead

of requiring both ϵ < 0 and µ < 0 in conventional metamaterials, chiral metamaterials need

neither ϵ nor µ to be negative to get negative refractive index. The negative index in chiral

metamateirals is due to the strong chirality parameter, which comes from the cross coupling

between electric and magnetic fields. In Chapter 4, we will discuss the mechanism of chiral

metamaterials in details. Our study on chiral metamaterials, including designs, simulations

and experiments, will also be presented in Chapter 4.

 15

CHAPTER 2. Surface modes and waveguiding of photonic crystals

Due to the existence of photonic band gaps, EM wave localization in photonic crystals

can be achieved by introducing defects. Photonic crystals with proper surface defects have

been shown to support surface waves, which are confined at the interfaces between photonic

crystals and their surroundings. Photonic crystal waveguides can be formed by introducing

line defects in photonic crystals. In this chapter, we will study the surface mode dispersion of

photonic crystals with different surface defect layers. We will show that the dispersion curve

of the surface modes can be traced experimentally. Then we will show that, by making use

of surface modes, the waveguiding properties of photonic crystal waveguides can be modified

drastically. Enhanced transmission and directed beaming can be achieved through photonic

crystal waveguides of subwavelength aperture. Next we will present our experimental work on

optical conductance of photonic crystal waveguides. We will show that the optical conductance

of photonic crystal waveguides are quantized.

2.1 Introduction

2.1.1 Surface waves in photonic crystals

The most well-known feature of Photonic crystals (PCs) is the photonic band gap (PBG),

inside which the electromagnetic waves are prohibited to propagate in all directions [15, 51].

However, when appropriately terminated, photonic crystals can support surface modes with

frequencies lying inside the PBG [13]. Surface modes are easily obtained for 1D photonic

crystals (stacks of alternating continuous layers) and their dispersion relation has been obtained

experimentally and theoretically [52].

For a 2D photonic crystal the existence of surface waves has been shown theoretically (in
 16

band structure calculations for semi-infinite photonic crystal) [53, 54] and experimentally by

attenuated total reflection (ATR) measurements [55]. Further numerical work has shown that

the existence and dispersion of surface waves at a 2D photonic crystal are very sensitive to the

surface termination [54,56]. When terminated improperly, both the poor impedance matching

and the interaction with surface modes can disrupt the electromagnetic wave propagation and

are bad to the performance of photonic devices such as PC waveguides and PC cavities [56].

Surface waves can be helpful for certain applications when designed properly. For example,

photonic crystals have been used to fabricate subwavelength imaging devices. The excitation

of surface waves in this case is the key to get subwavelength resolution [57]. Recent studies

also show that when the photonic crystal is appropriately periodically corrugated at the sur-

face, enhanced transmission and beaming of light can be achieved through a photonic crystal

waveguide with a subwavelength width [58–60] . This effect is due to the excitation of surface

modes and constructive interference at the axis of the waveguide. It is thus important to study

the dispersion relation of surface modes. In the following, we will study the surface dispersion

with numerical simulations, and trace the surface band experimentally by attenuated total

reflection (ATR) method.

The 2D photonic crystal we study is a square array of square alumina rods, 21 layers in

x direction and 15 layers in y direction. The parameters of the rods are given in Chapter 1.

With the help of the supercell technique, the band structure of the finite-size photonic crystal

can be calculated using the plane wave expansion method [61]. We studied TM (electric field

along the rods) modes through out the work since only TM modes can give a full band gap

for a 2D photonic crystal [13]. As shown, for example, in Fig.2.1(a), the PC has a band gap

indicated by the gray area. There are no modes inside the band gap and below the light line.

The uncorrugated photonic crystal does not support surface waves (in the lowest total

band gap); to observe surface modes we have to corrugate the surface layer of the photonic

crystal, rendering it different from the air on one side and the photonic crystal bulk on the

other side. The actual shape of the corrugation layer elements is not very important. We

use cylindrical rods of different diameters, but also cut semi-cylinders [55], rectangular cross-
 17

section rods different from the ones comprising the bulk photonic crystal [62] or geometrically

identical rods with different dielectric constant might be used. However, there is no guarantee

to obtain surface modes for a particular surface termination; the surface layer parameters have

to be chosen carefully for the corrugated PC to support surface modes in the band gap of the

bulk photonic crystal. In order to support surface modes, we add a new layer at the surface

Figure 2.1 (a) The projected band structure of the photonic crystal. Be-
low the figure shows the supercell and the arrow indicates that
the supercell is periodic in that direction. Please see text for
the parameters. (b) The projected band structure of the pho-
tonic crystal with surface modifications. The surface layer is
composed of circular rods with diameter of 1.83 mm. Below
the figure shows the supercell and the arrow indicates that the
supercell is periodic in that direction.

of the photonic crystal with circular alumina rods as a periodic corrugation (Fig. 2.3). The

band structure of the corrugated PC was calculated using the plane wave expansion and the

supercell technique. Fig. 2.1(b) shows the calculated band structure of the photonic crystal

with surface rods of diameter 1.83 mm. It is clearly shown in the figure that a new band is

enabled, which lies inside the bandgap and under the light line. Modes in this band can neither
 18

penetrate into the photonic crystal nor propagate into the air. Thus these modes are bounded

to the surface region of the photonic crystal and are called surface modes.

2.1.2 Line defects and waveguides in photonic crystals

A line defect in a photonic crystal can be created for example by changing the geometry

or remove one row of the structures. In this case, a defect in an otherwise perfect periodicity

is created and the translational symmetry is destroyed. Eigenmodes can be created in the

frequency range of the bandgap. These new eigenmodes allow the propagation of EM waves

inside the defect region. Meanwhile, the EM waves are still prohibited from propagating inside

the bulk photonic crystal, since the frequencies are inside the bandgap. Thus the EM waves

are confined in the defect region. So a waveguide is created in this way.

Photonic crystal waveguides guide EM waves in a mechanism different than traditional

dielectric waveguides and have certain advantages over the latter [13]. Traditional dielectric

waveguides guide EM waves based on total internal reflection. The dielectric constant of the

waveguide must be higher than the surroundings. Propagation loss is inevitable since the EM

waves are propagating inside the higher dielectric region. Also, there will be significant loss

when the waveguide is bent since the waveguiding mechanism requires a large incident angle to

get total internal reflection. It is difficult to bend the EM wave in a sharp angle by a dielectric

waveguide. Photonic crystal waveguides on the other hand guide EM waves based on photonic

bandgaps. The EM waves are confined in the defect region and higher dielectric constant is

not required. It is thus possible to guide the EM waves inside a lower dielectric region or even

air. Lower propagation loss can thus be achieved with photonic crystal waveguides. Another

merit of photonic crystal waveguides is that sharp bending can be created with relatively high

transmission and low scattering loss. Photonic crystal waveguides have found applications in

photonic crystal fibers and integrated photonic circuits [16]. The dispersion of the waveguide

modes depends on how the defect is constructed. There could also be single-mode and multi-

mode waveguides. In our 2D photonic crystal, a waveguide is created by removing one row

of the dielectric rods. The dispersion of the waveguide modes can be computed by numerical
 19

Figure 2.2 (a) The projected band structure of the photonic crystal cal-
culated with the supercell indicated in the inset. The arrow
indicates that the supercell is periodic in that direction. (b)
The projected band structure of the photonic crystal with line
defect, calculated with the supercell shown in the inset. The
arrow indicates that the supercell is periodic in that direction.

tools. For comparison, Fig. 2.2(a) shows the projected band structure of the photonic crystal

calculated with the supercell indicated in the inset. The photonic crystal is finite in the vertical

direction but extends infinitely in the horizontal direction, indicated by the arrow in Fig. 2.2(a).

It is clear that no mode is allowed in the bandgap region. Fig. 2.2(b) shows the band structure

of the photonic crystal with the central row of rods removed, as shown in the inset. It is seen

that a new band is allowed, which is located inside the bandgap. Modes in this band can not

penetrate into the bulk photonic crystal since their frequencies are within the bandgap. Thus

these modes are confined in the line defect region and are called waveguide modes.

2.2 The excitation of surface modes

In this section, we study experimentally and numerically the dispersion relation of surface

modes for different surface corrugations that support surface waves. In contrast to previous

work, we determine experimentally the complete dispersion relation of the surface modes within

the photonic band gap of the photonic crystal [63].

 20

2.2.1 ATR method

Since surface modes lies to the right of the light line, they cannot be excited by propagating

waves directly. To observe a surface mode experimentally, we use the ATR method (Otto

configuration) [64].

d
a

y
x

Network
Analyzer

Figure 2.3 Experimental setup of the ATR measurement. The HP8510

network analyzer and a pair of horn antennas are used to mea-
sure the reflection spectrum of the prism. The photonic crystal
is put behind the prism.

Consider a dielectric prism with index of refraction n. If the incident angle θ at the reflecting

surface is larger than the critical angle θC , θ > θC = sin−1 (1/n), total internal reflection occurs

at the outgoing interface. When there is nothing behind, the incident wave would be totally

reflected. However, apart from the reflected wave, total reflection also involves a transmitted

evanescent wave which travels along the interface and decays exponentially away from the

surface. That is because the wavevector component parallel to the interface,

kx = k sin θ = nω sin θ/c > ω/c, (2.1)

 21

is conserved across the interface. So outside the prism, the wavevector component perpen-

dicular to surface, ky , is imaginary. When we put the corrugated photonic crystal close to

the prism, the tail of this evanescent wave may couple to a surface mode at the surface of

the photonic crystal. By changing the incident angle, we can change the value of kx and get

coupling with different surface modes. When the evanescent wave is coupled to a surface mode,

there is strong field at the surface and the energy is dissipated through the coupling. Thus

the reflected field intensity is smaller when the coupling happens. If the reflection spectrum is

plotted, there would be a dip at the coupling frequency. The excited surface mode is then iden-

tified. By changing the incident angle, the wavevector of the evanescent wave is changed and a

new surface mode could be excited if the wavevector is matched. By plotting the experimental

results at different incident angles, the surface band can be obtained.

The first ATR experiment on surface modes of photonic crystal was reported more than

10 years ago [55]. However, it was not straightforward to see the coupling of surface waves

from the experimental data and a dispersion relation was not obtained experimentally. In our

experiment, an HP-8510 network analyzer was used to measure the S parameters so that the

reflection spectrum can be immediately seen. A pair of horn antennas serves as transmitter

and receiver (Fig. 2.3). The dielectric prism used in our experiments is an isosceles-right-

triangle-shaped wedge with side length of 15 cm and with index of refraction n = 1.61. The

horn antennas were placed in such a way that the E field is polarized along the dielectric rods.

For a given incident angle, Snell’s law was used to identify the angle and position of the horn

antennas such that the setup is symmetric along the dashed line.

The ATR method requires the structure to be set close to the prism to get good coupling

effect because the evanescent waves decay fast; however, while we are making use of the prism,

the prism itself modifies the surface modes especially when it is very close to the structure. To

get a satisfying experimental result, we need to optimize the airgap size to be able to record

the mode while keeping the structure as far away from the prism as possible. In Fig. 2.4 the

solid curves show the coupling effect at different air gap sizes at incident angle 53◦ for the

photonic crystal with corrugation layer of circular rods with diameter D = 2.44 mm. You can
 22

-10

g=2mm
-20
0
Reflection (dB)

-10

g=4mm
-20
0

-10

g=6mm
-20
0

-10
g=8mm

-20
0

-10
g=10mm
-20
0.35 0.40 0.45

frequency ( a/2 c)

Figure 2.4 The experiment (solid lines) and simulation (circles) results of
reflection at incident angle 53◦ for the photonic crystal with
corrugation layer of circular rods with D = 2.44 mm. From top
to bottom shows the reflection when the distance g of the air gap
between the prism and the surface of the structure increases.
 23

see clearly in Fig. 2.4 that when the surface waves are excited, there is a very well defined

dip in the reflection data. This is well shown for the distance of g = 10 mm. The prism

disturbs the coupling and the observed reflection dip position when the air gap is small; the

dip converges to a constant frequency as the air gap increases. At larger incident angles, the

coupling of the evanescent waves to surface waves is smaller and the dip position converges

faster. This is because larger incident angle provides larger kx and thus larger absolute value

of imaginary ky ; so the evanescent wave decays faster. For the surface corrugated photonic

crystal, the ATR dip is still strong enough to be observed when the air gap between the prism

and the surface of the structure is 10 mm and is already converged except at incident angles

near the critical angle. For a single layer of corrugation rods, the coupling is not as strong as

for the full structure (photonic crystal and corrugation layer) case. To get good coupling, for

the single corrugation layer, the air gap size is chosen to be 7 mm.

2.2.2 Experimental results

We studied the surface band of the 2D photonic crystal with different surface modifications.

The diameter of the corrugation rods needs to be chosen properly so that the surface dispersion

lies within the band gap. When the surface mode goes out of the band gap, the mode can extend

into the photonic crystal and is no longer a surface mode. The solid lines in Figures 2.5(a)

and 2.5(b) show the dispersion relation of the surface modes when the diameter of the surface

rods is D = 1.83 mm and D = 2.44 mm, respectively. Actually, a single layer of only dielectric

rods surrounded by air can support surface modes. The solid line in Fig. 2.5 (c) and (d)

shows the surface band of a single layer of circular alumina rods with diameter D = 1.83 mm

and D = 2.44 mm, respectively. These two surface dispersion curves shown in Figures 2.5(c)

and 2.5(d) are different from the two (shown in Fig. 2.5(a) and (b)) which also involve the

photonic crystal. This can be explained from the fact that the presence of the photonic crystal

modifies the surface mode of a single layer.

Fig. 2.5 shows the experimental results of the surface bands together with the calculation

results for two different corrugation layers with and without photonic crystal. The solid points
 24

Figure 2.5 Experimental surface band structure of a 2D photonic crystal.

Between the light gray areas exist the band gap of photonic
crystal. Solid lines and square dots with error bars are super-
cell band structure calculation and experimental results of the
surface band for (a) photonic crystal with a corrugation layer
of circular rods with D = 1.83 mm, (b) photonic crystal with
a corrugation layer of circular rods with D = 2.44 mm, (c) a
single layer of rods with D = 1.83 mm and (d) a single layer of
rods with D = 2.44 mm.
 25

are the averaged dip frequencies over three measurements while the error bar on each point

comes from the half width of the reflection dip, averaged over three different measurements.

We can see that the experimental results are in good agreement with the calculated surface

bands. However, at smaller incident angles, especially close to the critical angle, relatively

large error bars are obtained. This is because the finite-width incident beam might not be

fully reflected; some plane wave components propagate through the interface and change the

coupling to the surface mode. Instead of a nice and sharp coupling dip, the coupling now is

over a larger frequency range. That makes the half width of the dip larger. Also the evanescent

wave decays slower at smaller kx and the coupling may not have converged yet at the largest

air gap. So the overall experimental error is larger.

Simulations with a similar setup were done in FEMLAB. In the simulations, the source

is a Gaussian beam normally incident at one side of a right-angled dielectric wedge. The

rod array is set behind the hypotenuse of the wedge. The reflected power is obtained by

integrating over the other side of the wedge. To get a coupling at different angles, the two

acute angles of the wedge are changed so that the Gaussian beam is always normally incident

on the wedge. Compared with experiments (Fig. 2.4), the reflection spectrum in simulations is

always smoother. Also, the coupling can still be seen when the air gap is several centimeters

large.

2.3 Beaming and enhance transmission through photonic crystal

waveguide

In the previous section, we showed that photonic crystals can support surface waves with

proper surface modifications. The surface mode dispersion can be traced by numerical tools

as well as experiments. In this section, we demonstrate one important application of surface

modes. We will show that by exciting the corresponding surface modes, enhanced transmission

and directed beaming can be obtained through photonic crystal waveguides of subwavelength

width [65].

The effect of the corrugation and grating layers on the beaming of the electromagnetic (EM)
 26

waves from a photonic crystal are presented experimentally. The role of the two components

of the beaming, the transmission of the channel and the directionality, are explicitly analyzed.

It is demonstrated that a small change, which consists of taking away one grating from the

grating layers, can improve the beaming by 20% and opens up a new widow of frequencies with

a second beaming in our photonic crystal. These experimental observations can be explained

by the derived dispersion relation of the surface waves.

2.3.1 Coupling to subwavelength apertures

As we have discussed, a photonic crystal waveguide can be created by introducing a line

defect in the bulk photonic crystal. EM waves can be guided along the waveguide and will not

escape into the bulk photonic crystal due to the existence of the bandgap. The waveguide is

essentially a local disturbance that breaks the translational symmetry at the interface between

the photonic crystal and its surrounding. Consequently, it couples different wavevectors com-

ponents and allows an incoming normally incident propagating wave with a wavevector k∥ = 0
ω
to couple to the surface modes with k∥ > c where ω is the frequency and c is the vacuum speed

of light. Many waveguide structures based on photonic crystal have been proposed [58, 59, 66].

Qualitative [60] and quantitative [67, 68] explanation of the physics behind the phenomena

have been reported. However, in most cases, the apertures of the waveguide openings are

smaller than the working wavelength. The problem of a poor coupling between the incident

wave and the waveguide, which is not specific to the photonic crystal structures, lead to a low

transmission efficiency. Moreover, the outgoing wave from a subwavelength waveguide diffracts

in all directions and do not form a directed beam.

It is desirable to get better coupling into and out of the photonic crystal waveguides. For

example in optical telecommunications, it is important to get good coupling between photonic

crystal waveguides and dielectric waveguides and optical fibers. Due to the different waveguid-

ing mechanism of photonic crystal waveguides and conventional dielectric waveguides, signifi-

cant reflection and diffraction are expected at the interfaces. A relatively straightforward way

is to use tapered structures. Either the photonic crystal waveguide opening, or the dielectric
 27

waveguide end, or both, can be designed to be tapered [69]. It shown in Ref. [70] that, when a

self-collimation photonic crystal is added to the photonic crystal waveguide opening, a better

directed outgoing beam from the waveguide can be obtained. Directional emission from pho-

tonic crystal waveguides are also demonstrated using defected-mode coupling [71, 72]. Defects

are introduced near the waveguide opening which serve as scatterers and the directed emission

comes from the diffraction from the scatterers.

In parallel to the research on photonic crystals, a lot of discoveries has been made on

subwavelength apertures on metals. In 1998, Ebbesen et. al. [73] observed an extraordinary

optical transmission through subwavelength hole arrays in optically opaque metal films. It

has been shown that this enhancement of transmission is due to the excitation of surface

plasmons on the metal-dielectric interfaces [74]. Later on, several experimental works [75, 76]

have shown the possibility of enhancing and channeling a beam through metallic structures.

The key feature was to corrugate the surface surrounding the aperture allowing the excited

surface plasmons at the metallic interface to couple to the outgoing beam (propagating modes)

leading to a well directed beam. The potential of applications of such kind of structures has

been mentioned in several publications [77–79].

Inspired by the work on metals, people started to apply the idea of exciting surface waves

on photonic crystals [58–60, 66–68]. As we have discussed, photonic crystals can support

surface waves if their operational frequency is within the photonic band gap. In contrast to

metallic surfaces that easily excite surface modes, photonic crystals do not support surface

waves unless they are appropriately terminated [63]. Similar to the metallic counterpart, once

the surface waves are excited, a grating-like structure can be applied to couple them into

the outgoing propagating waves. Under specific conditions, the grating layer destructively

interfere the waves everywhere except about the axis of the waveguide resulting to a beaming

of the propagating modes in the forward direction. Compare to metallic structures, photonic

crystal structures can be advantageous in terms of performance. The Ohmic losses in metallic

structures limit their performance in exciting surface plasmons, which is the key feature for

the enhanced transmission and the high directionality of the beam. The photonic crystal
 28

structures on the other hand are composed of dielectric materials and generally have lower

loss, particularly at higher frequencies.

Figure 2.6 The simulated electric field distribution with (a) a regular pho-
tonic crystal waveguide formed by a line defect, (b) the same
waveguide as in (a) with a modified layer added and (c) the same
waveguide as in (b) with a grating layer added. The dashed lines
in the figures show the position of the structures. The struc-
tures are shown under the corresponding field distributions.

In the following we call the layer on the surface of the bulk photonic crystal used to

support surface waves modified layer and the extra layer which acts like a grating to couple

the surface waves into outgoing beam grating layer. Fig. 2.6 is an illustration of the principle.

In Fig. 2.6(a), we see clearly that the field propagates inside the waveguide and diffracts at the

subwavelength waveguide opening. In Fig. 2.6(b), a modified layer is added to the waveguide.

we see that strong field is localized at the surface layer of the waveguide, due to the excitation of

a surface mode. However, the field does not couple into propagating wave due to the wavevector

mismatch. The outgoing beam still diffracts in all directions. In Fig. 2.6(c), an extra grating
 29

layer is added to the surface-modified waveguide. The grating layer couples the surface mode

into propagating wave and interference destructively except in the normal direction. Thus we

see a strong main beam along the direction of the waveguide in Fig. 2.6(c). Next we will study

the effect of the modified and the grating layers added to the photonic crystal on the surface

waves and their coupling to the outgoing modes. By studying the dispersion of surface modes,

we are able to predict which conditions lead to higher exited surface waves and consequently

to a better beaming.

2.3.2 The experimental setup

The 2D photonic crystal consists of a 2D square array of square alumina rods in air,

with parameters of the rods same as before. The dimensions of the structure is 21 layers in

the lateral direction and 10 layers in the propagation direction. A schematic picture of the

photonic crystal structure and the arrangement for both experiments and simulations is shown

in Fig. 2.7. In order to create a waveguide, we omit the ten rods forming the middle row in

the structure to channel the beam through it. The actual aperture’s width of the waveguide is

approximately 1.9 cm measured from the outer-part rods of the photonic crystal. Therefore,

for all the frequencies of the incident plane wave (between 10 and 13 GHz), the aperture is

smaller than the wavelength of the incident beam. Consequently, we expect the aperture to

transmit very poorly and diffract light in all directions. For the modified layer, we add a row

of circular rods of diameter 1.83 mm at the beginning and the end of the photonic crystal. As

a grating layer a row of square rods of width 3.18 mm is added to the structure with a lattice

constant twice the lattice constant of the photonic crystal (see Fig. 2.7).

A finite plane wave source with width of 7.85 cm is placed at a distance 3 cm from the outer

interface of the structure. A commercial finite element method software Femlab (COMSOL)

was used to calculate the field distribution of the different structures under study. In all the

simulations we performed, perfect matched layer (PML) [80,81] boundary conditions were used

to surround the computational domain. For both experimental and numerical results a TM

polarization ( the electric field E is parallel to the axis of the rods) is used. Transmission
 30

Figure 2.7 (a) The structure’s details and definition of the different bound-
aries used. bnd0, bnd1, bnd2, bnd3 are the boundaries in the
channel, at the end of the photonic crystal, in the middle of the
air part and at the end of the air part respectively.(b) The shape
of the surface layers in each of the fifth different structures.

measurements were performed to verify and test the enhancement as well as the beaming.

The experimental set-up consists of an HP 8510C network analyzer, a horn antenna as the

transmitter, and a monopole antenna as the receiver. The monopole antenna is mounted on a

motor-controlled 2D table and the position can be controlled by the computer. Each time the

position of the monopole antenna is changed, the new transmission data is measured by the

network analyzer and recorded to the computer. The field distribution can then be obtained

by scanning the area after the photonic crystal aperture.

To differentiate between the portion of the beaming coming from the transmission of the

channel and the one coming from the directionality, the following forms of structures have been

studied: the waveguide by itself is referred as structure1, the waveguide with the grating layers

is referred as structure2, the waveguide with the modified layers is referred as structure3,

and the waveguide with both the modified and grating layers is referred as structure4 (see

Fig. 2.7(b)). The power flow which is the time-average of the Pointing vector for each of

these structures is examined in different boundaries in the propagation direction in numerical

 31

simulations, as shown in Fig. 2.7(a). The first, second and third boundaries located at the end

of the photonic crystal, the middle of the air part and the end of the air part are referred as

bnd1, bnd2, bnd3 respectively. The boundary, bnd0, located in the middle of the photonic

crystal is the boundary used for normalization.

2.3.3 Enhanced transmission

We will first examine the transmission of the channel, Fig. 2.8 displays the power flow

versus the frequency for the four cases measured at boundary (bnd0). The simulated power

flow is frequency dependent as it is clearly shown from Fig. 2.8. This figure shows that the

transmission of the waveguide depends on the frequency and especially for the structures that

grating and both grating and modified layers are added. Second, depending on the termination

of the basic structure (structure1), the transmission of the channel is very sensitive to what is

at its ends or termination. Notice that for case of structure1, i.e., only the waveguide without

modified and grating layers, the transmission for all the range of frequencies is very low. The

addition of the grating layer changes waveguide termination and the coupling of EM waves and

thus the transmission spectrum (structure2, dashed line in Fig. 2.8). In some frequency region

the transmission is increased compare to structure1. However the enhancement is difficult to

control and is not what we are looking for. For structure3 (dash-dotted line in Fig. 2.8), a

modified layer is added to the waveguide, the transmission does not change much compared to

structure1. This is because the surface waves are bounded to the surface region and can not

coupling into propagating waves. However, when the grating and the modified layers are added

(structure4, dotted line in Fig. 2.8), the power flow is more than five times higher and reaches

its maximum of 50% at the frequency of 11.85 GHz that correspond to the beaming frequency.

We experimentally checked the transmission of structure4 and the result is plotted in Fig. 2.9.

A very good agreement is found between the simulation results shown as solid line and the

experimental results shown as dashed line. For both the simulation and the experiment the

peak of enhanced transmission in the forward direction occurs at 11.85 GHz.

To show how the transmission of the waveguide depends strongly on the channel termi-
 32

Figure 2.8 Simulated power flow versus frequency measured at boundary

bnd0 for the four structures which are only waveguide inside
the photonic crystal (structure1), the waveguide with the grat-
ing layers (structure2), the waveguide with the modified layers
(structure3), and the waveguide with both the modified and the
grating layers (structure4).

Figure 2.9 Simulation and experimental power flow versus frequency for
structure4, which possesses both the gratings and the modified
layers.
 33

Figure 2.10 Power flow versus frequency for structure4 (simulation only)
and structure5 (simulation and experiment). Notice that
structure5 is similar to structure4 except for the one grating
missing in front and the end of the waveguide.

nation. The forth structure is slightly modified by taking away one grating at the middle of

both grating layers. This modified structure will be referred to as structure5 (see Fig. 2.7).

Fig. 2.10 displays the simulation and the experimental results of the power flow for structure5

in comparison with those of structure4. First, the most striking change is that the transmis-

sion enhancement by almost 20% at 11.85 GHz by just removing only one grating in front of

the channel. Second, a very strong transmission peak (more than 80%) is observed at 12.5

GHz making this structure a good candidate to have two strong transmission at two different

frequencies. It can be inferred that the back and forth reflection inside the waveguide is very

sensitive to any small changes introduced into the structure. Consequently, it plays a major

factor in determining the strength of the transmission and therefore the quality of the beaming

if it is achieved together with a high directionality. The agreement between the experiment

and the simulation is quite good except at low frequency where the experimental results show

some structure that can be attributed to the normalization procedures used in the experiment.
 34

2.3.4 Directed emission

Figure 2.11 A 2d plot of the strength of the outgoing electric field for the
case of structure4. (a) and (b) are the experimental and the
simulation results respectively at frequency f = 11.85 GHz.
(c) and (d) are the experimental and the simulation results
respectively at frequency f = 12.47 GHz.

We next studied the field distribution for different structures both in numerical simulation

and experimental measurements. This is a straightforward way to see how the outgoing beam

is directed. For further insight on the strength of the field at both beaming and non-beaming

frequencies, we plot in Fig. 2.11 the experimental and the simulation results for structure4. The

figure is a 2d plot of the strength of the electric field for both experimental (Fig. 2.11(a) and

Fig. 2.11(c)) and simulation result (Fig. 2.11(b) and Fig. 2.11(d)). At the beaming frequency

(i.e. 11.85 GHz, with experimental result shown in Fig. 2.11(a) and simulation result shown

in Fig. 2.11(b)) structure4 generates a very nice beaming that goes further away from the

photonic crystal with a noticeable intensity. However, at the non beaming frequency (i.e.

12.47 GHz, with experimental result shown in Fig. 2.11(c) and simulation result shown in

Fig. 2.11(d)), the beaming effect disappears, and both experimental and simulation result

show a non pronounced field outside the photonic crystal that split in two directions and

vanishes afterward. The directed beam is characterized by the line plots shown in Fig. 2.12

for the beaming frequency 11.85 GHz, corresponding to the 2D plots shown in Figs. 2.11(a)
 35

Figure 2.12 The line plots of field intensity for structure4 at 11.85 GHz,
corresponds to the 2D distribution shown in Fig. 2.11. for (a)
The field intensity along the central line of the waveguide exit
(horizontal direction in Fig. 2.11(a)) for structure4 at 11.85
GHz. (b) The field intensity along the interface of the waveg-
uide end (vertical direction in Fig. 2.11(b)), at the horizontal
position indicated by the arrow in (a). The solid lines are
numerical results and dashed lines are experimental results.

and 2.11(b). Fig. 2.12(a) shows the field intensity along the central line of the waveguide axis,

starting from the waveguide exit. The small oscillations close to the waveguide exit are the

Fresnel oscillations due to the diffraction effect at the interface. The smooth peak indicated

by the arrow is the intermediate field pattern that form the directed beam shown in the 2D

plots. At the position of the peak indicated by the arrow in Fig. 2.12(a), the line plot along

the waveguide end (vertical direction in Fig. 2.11(b)) is plotted in Fig. 2.12(b). This shows

the profile of the main beam. The full width at half maximum (FWHM) is about ±5◦ . Again

the simulation and experimental results agree very well with each other. The directed beam

is because a surface mode is excited and coupled to the propagating wave. The grating layer

help to build constructive interference along the central line of the waveguide.

In Fig. 2.13, we examine structure5 as we did with structure4 in Fig. 2.11. The field pattern

at 11.85 GHz is shown in Figs. 2.13(a) and (b). We see that a well directed beam is formed

and is almost identical to the pattern of structure4 (shown in Figs. 2.11(a) and (b)). This is
 36

Figure 2.13 A 2d plot of the strength of the electric field for structure5.
(a) and (b) are the experimental and the simulation results
respectively at frequency f = 11.85 GHz. (c) and (d) are the
experimental and the simulation results respectively at fre-
quency f = 12.47 GHz.

because the same surface mode is excited as in structure4. The result at 12.47 GHz is shown in

Figs. 2.13(c) and (d) and looks different than the structure4 results (Figs. 2.11(c) and (d)). The

outgoing beam from structure5 is stronger than structure4. This is because the modification of

the waveguide termination changes the transmission property, as we have discussed. However,

compared to the 11.85 GHz case, the beam at 12.47 decays and diffracts faster when it is away

from the waveguide. Again the simulation and experimental results agree well.

We have noticed that at 11.85 GHz, the transmission from the waveguide has a peak value

and a nice beam is formed. This is not a coincidence but both because of the excitation

of the surface mode. We can find the corresponding surface modes from the surface mode

dispersion relation. The band structure of the photonic crystal with modified layer which

extends to infinity in the lateral direction is shown in Fig. 2.1, where a surface band can

be seen inside the band gap and under the light line. Since the waves coming out from the

photonic crystal waveguide tend to diffract in all directions, they have all k// components.

Thus for any frequency in the range of the surface band, there exists some k// that satisfies the

dispersion relation of the surface band. So, the surface wave excitation should be continuous
 37

Figure 2.14 Simulated 2d plot of the strength of the electric field of struc-
ture3 at (a) f = 11.97 GHz and (b) f = 11.33 GHz.

over the spectrum for the infinite photonic crystal. However, in our study the photonic crystal

has a finite size with 21 rods in the lateral direction. The surface wave excitation in this case

is discrete, rather than continuous. This is because at this finite and discrete surface layer,
2π
the possible values of k// are discrete. In the first Brillouin zone, k// can be N a m, where

N = 21 is the total number of unit cells along the surface and m is any integer between

−10 and 10. From the dispersion curve, when m = 10, the first surface mode is excited at

f = 11.97 GHz (Fig. 2.14(a)); when m = 9, the second surface mode is excited at f = 11.33

GHz (Fig. 2.14(b)). The first mode is expected to be the strongest since it has the largest k//

so, the electromagnetic energy is most confined in the surface layer. However, in Fig. 2.14(b)

in which k// is smaller, the surface waves are not as strong as for the case of Fig. 2.14(a). Also,

the grating has a period of 2a, which corresponds to the strongest surface mode at 11.97 GHz.

Thus the surface mode at 11.97 GHz can be coupled to the propagating wave and a higher

transmission and directed beam can be obtained. The actually peak frequency 11.85 GHz

differs a little from 11.97 GHz because the addition of the grating layer disturbs the surface

mode dispersion and shifts the surface mode frequency.

 38

2.4 Experimental verification of quantized optical conductance in photonic

crystal waveguides

Electron conductance has been proved and verified to be quantized in low-dimensional

structures. The optical conductance can be defined in a similar way as electron conductance.

It has also been shown by experiment in infrared regime that the optical conductance of a

metal waveguide is also quantized. It is interesting to observe this effect in photonic crystal

waveguides, which have more complicated guiding mechanism than metal waveguides. It is

also interesting to observe the classical analog of the quantization effect in microwave regime.

In this work, we measure the optical conductance of a photonic crystal waveguide in microwave

frequencies and verify that the conductance is indeed quantized [82].

2.4.1 Introduction to quantized conductance

One of the interesting phenomena in science is the quantization of different physical quanti-

ties. Quantization occurs because of the wave nature of particles. The quantization effects are

getting stronger when the size of the system is getting smaller. One quantity that gives strong

quantization is the conductance G. Conductance is the inverse of the resistance R = ρL/A,

where ρ is the resistivity, L is the length of the sample and A is the area of the cross section.

It is very difficult to calculate analytically the transport properties of small devices. Landauer

was the first one to make the connection between the conductance and the transmission co-

efficient [83, 84]. In particular, Landauer showed that G = (e2 /~π)T /(1 − T ), where T is the

transmission coefficient. If one has a perfect metal T = 1, and then G → ∞ and therefore

R → 0 as expected for a perfect metal. In 1981, Soukoulis and Economou [85] proved, by using

the Kubo-Greenwood formula, that for 1D system G = (e2 /~π)T . There was a big controversy

in 1980’s which formula was correct (for a historical account of the controversy from two dif-

ferent perspectives, see [86, 87]), since the Economou-Soukoulis formula gave a finite value for

the resistance for the perfect metal. The experiments [88–90] resolved this issue and indeed it

was found that G = (e2 /~π)T . Extensions to higher dimension G [91, 92] were achieved and it
 39

was shown that for a multichannel wire

e2 ∑
Nc
G= |tnm |2 (2.2)
~π
n,m=1

where |tnm |2 is the transmission coefficient between incident mode n and output mode m. If

|tnm |2 = δnm , the the conductance G = Nc e2 /~π is quantized in units of e2 /~π. Since the con-

ductance quantum e2 /~π contains only fundamental constants, the conductance quantization

must occur not only in metals, but in semiconductors and superconductors.

The key idea of Landauer was to relate the conduction of electrons with its transmission. So

all the idea of quantization of conductance can be also obeyed by all the waves, electromagnetic

waves, acoustic and elastic waves. It is amazing that the only experiment [93] that has been

done with waves is the transmission of a slit of variable width for a given wavelength of

λ = 1.55µm. Transmission steps of equal height occur whenever the slit width W equals half

the wavelength. Similar to its electronic counterpart, the optical conductance of a structure

is described as the total light transmitted through the structure from a normalized diffuse

illumination. In the experiment of Montie et al. [93] a two dimensional (2D) diffuser was used

to achieve diffuse illumination. The diffuser was essentially a 2D random array of scatterers

through which the normally incident plane wave scatters diffusively and isotropically in plane.

The diffused light passed through a metal slit and the transmitted light was collected. The

result showed that the optical conductance increases in a staircase fashion with a step of λ/2

in slit width. A new stair occurs when the slit width W = nλ/2 (n = 1, 2, 3, ...) i.e., a new

mode is enabled in the slit.

Photonic crystals can be designed to have a bandgap that prohibits wave propagation in

a certain frequency range. Line defects of photonic crystals can confine light with frequencies

within the bandgap inside the channel and act as waveguides. photonic crystal waveguides have

been studied extensively and many applications in industry have been proposed and realized.

However, the concept of optical conductance was not applied to photonic crystal waveguides

until recently [94,95] and no experimental work has been reported to our knowledge. We study

the optical conductance of photonic crystal waveguides, both numerically and experimentally.

We will show that the optical conductance of a 2D photonic crystal waveguide has the similar
 40

staircase effect as the metal slit.

2.4.2 Experimental design

We design the working frequency of the waveguide at microwave region, which makes the

photonic crystal easy to fabricate. One of the main difficulties to observe the quantization

of the conductance for microwaves frequencies is the need of diffusive waves. This is easy to

achieved at optical wavelengths through a diffusor consisting of a random array of particles.

For microwaves, a large enough random array of scatters are needed to act as a diffusor to

get isotropically diffusive illumination out of a plane source. Since the wavelength is now

of the order of centimeters, the diffusor must be really big in size and the intensity of the

transmitted waves will be very weak to be used efficiently in microwave experiments. So our

idea is to apply Equ. (2.2) to experiments on the conductance of EM waves. We will measure

the transmitted power for each particular incident plane wave (wavevector k) with different

incident angle separately and sum them up to obtain the conductance of EM waves. The plane

wave is provided by a horn antenna and the incident angle can be changed to get different k

components.

The photonic crystal we study is a 2D square array of square alumina rods. The lattice

constant is a = 11 mm and the square rods are of dimension d = 3.18 mm with relative

permittivity 9.8 and height h = 15 cm. Band structure of the photonic crystal shows a

bandgap between 9.43 GHz and 12.78 GHz for TM modes (Electric field parallel to the rods).

The waveguide is formed by two closely separated pieces of photonic crystal slabs with a size

of 25a × 4a each. The width of the waveguide can be varied by changing the distance between

the two photonic crystal slabs. In experiments, the two photonic crystal slabs are fixed on two

motorized linear stages and the position can be precisely controlled. HP8510B network analyzer

and a pair of horn antennas are used to measure the transmission through the waveguide. Both

the transmitter antenna and the receiver antenna are mounted on motorized rotational stages

such that both the incident and outgoing angle can be controlled. The antennas can move

along a semicircle with a radius of 30 cm which is centered at the middle of the near end of
 41

Figure 2.15 An illustration of the experimental setup.

the waveguide ( See Fig. 2.15 for details).

2.4.3 Results and discussion

The experimental result of the optical conductance of the photonic crystal waveguide is

shown together with numerical simulations in Fig. 2.16. We see that the optical conductance

increases in a staircase manner, in both experiments and simulations, with the increase of the

waveguide width. The stairs has basically the same height and a new stair appears every λ/2

when a new mode is introduced in the waveguide. The first waveguide mode is enabled at the

onset of the first stair and this position is determined by the band structure of the photonic

crystal waveguide. From Fig. 2.16 we can see that the onset of the stairs in experiments

and simulations are in good agreement. However, the oscillations at the onset of stairs in

simulations, which are proved to be Fabry-Pérot resonances, are missing in the experiments.

The major reason is that, while in simulations the model is perfectly 2D and all the transmitted

power can be collected without scattering into z direction, we do not have an ideal 2D system.

The source we get from a radiating horn antenna is a confined beam. When the antenna is
 42

3 exp
sim
Conductance G

0.0 0.5 1.0 1.5

W/λ

Figure 2.16 The experimental (circles) and numerical (solid line) results
of optical conductance of a photonic crystal waveguide at 10
GHz.
 43

far away from the waveguide and the waveguide width is small, the beam can be considered

to be uniform in x direction over the waveguide entrance. However in z direction, the incident

beam is confined but not uniform. When it encounters the photonic crystal structure, while

the main part is propagating in xy plane the beam also scatters out of the plane. Since the

receiving antenna only collects the transmitted power in the xy plane where it sits, not all the

transmitted power is obtained due to the out of plane scattering. At the onset of the stairs,

more power is lost due to the multiple reflection. So the experimental curve is lower than the

one obtained by simulations and the Fabry-Pérot resonances are missing.

3 exp
sim
Conductance G

0
0.5 1.0 1.5
w/λ

Figure 2.17 The experimental (circles) and numerical (solid line) results of
optical conductance of a metal channel for TM case (E field
perpendicular to the plane).

We also studied the optical conductance of a metal channel, which serves as a simplified

model of the optical waveguides. Analytical calculations starting from Maxwell’s equations

are performed to understand the mechanism behind the stairs. Neglect the reflections at the

two ends of the metal channel, a smooth curve of optical conductance is obtained without the
 44

Fabry-Pérot resonances. Numerical simulations are also executed in which the reflections are

also removed by applying a PML layer. The result is in good agreement with the analytical

curve. Experimental measurements have also been done and the results are shown together

with the numerical results in Fig. 2.17 for TM (E in z direction) case.

2.5 Conclusion

In conclusion, we studied the surface modes and wave guiding properties of 2D photonic

crystals. Photonic crystals with proper surface defects can support surface modes and the

dispersion relation of the surface modes can be obtained by numerical calculations and ex-

perimental techniques. Attenuated total reflection experiments were done to trace the surface

mode dispersion of the photonic crystal and the results agree with numerical calculations. We

then studied how the surface modes can improve the transmission properties of photonic crystal

waveguides. With surface modes excited and coupled to propagating waves, the transmission

through the photonic crystal waveguide, which is a subwavelength aperture, can be largely

enhanced. Meanwhile, the outgoing beam is shown to be focused and directed to a small an-

gular range, instead of diffracting in all directions. The photonic crystal waveguide is further

studied in terms of optical conductance, which is, shown by experimental measurements, to be

quantized. The quantization effect is due to the discrete waveguide modes that are allowed to

propagate inside the waveguide.

 45

CHAPTER 3. Study of nonlinear split-ring resonators

3.1 Introduction

While most of the research in metamaterials is in the linear regime, where the electromag-

netic responses are independent of the external fields, some effort has been made to study

the nonlinear effects of the metamaterials, especially the nonlinear tunability of the split-ring

resonators (SRRs) [48, 96–99]. The SRRs are essentially LC resonators and the resonance fre-

quency is determined by the geometry of the rings. To tune the magnetic responses of the

SRRs, extra components or materials need to be introduced into the SRRs.

In Ref. [100], ferroelectric films are added to the substrate of the SRRs and the magnetic

tunability is achieved by controlling the electric permittivity of the ferroelectric films with

the change of temperature. In Ref. [101], low-doped semiconductors are photo-doped within

the slits of the SRRs and the magnetic response is tuned by varying the conductivity of the

semiconductors with an external light source. In Ref. [102], ferrite rods are introduced to

ambient the SRR unit cells and the magnetic resonance is modulated by magnetically tuning

the inductance of the ferrite rods by an external magnetic field. Compared to these methods,

the use of varactors is more feasible in microwave applications in that the tunability can either

be realized by a small DC bias voltage [96, 98, 103] or self-tuned by the intensity of the applied

electromagnetic fields without biasing [97,104]. Tunable metamaterials, based on the nonlinear

SRRs with varactors, have been tested experimentally in both transmission line form [98, 99]

and bulk form [103].

To study the properties of nonlinear metamaterials, it is important to learn the behaviors

of a single SRR and the mutual coupling between SRRs. In this chapter, experiments are

completed to analyze the properties of the SRRs loaded with linear and nonlinear elements.
 46

The bistable resonance [97,99] of nonlinear SRRs with one varactor is analyzed. The nonlinear

properties of SRRs with one varactor and two varactors are compared. The coupling effect of

two nonlinear SRRs are also discussed.

3.2 Varactor diodes

A varactor is a special kind of diode whose capacitance is a variable of the voltage applied

to its terminals. The variable capacitance is due to the change of the depletion layer thickness

at the p-n junction inside the varactor diode. When there is no external voltage applied,

the depletion layer is formed and eventually stabilized due to the balance of electron-hole

recombination and the build-in potential across the junction. The capacitance of the varactor

is then determined by the thickness of the depletion layer. When the varactor is reverse biased,

the applied voltage has the same polarity as the build-in potential and the increased potential

makes the depletion layer even wider, thus the capacitance gets smaller. When the varactor

is forward biased, the applied voltage is opposite to the build-in potential. The depletion

layer gets thinner and the capacitance gets larger. When the forward voltage gets larger, the

significant electron-hole recombination process causes current through the diode. The basic

working mechanism of a varactor diode is the same as a regular diode. A regular diode makes

use of the on/off switching behavior and people tries to fabricate diodes with the voltage-

dependence effect of the capacitance as small as possible. On the other hand, a varactor diode

is fabricated to exploit the voltage-dependence effect and increase the capacitance tunability.

Figure 3.1 The schematic symbol of a varactor (left) and the equivalent
circuit model of a varactor (right).

The schematic symbol of a varactor and its equivalent circuit model are shown in Fig. 3.1.
 47

C(V ) is the voltage-dependent depletion layer capacitance, ID (V ) is the dissipative current

through the p-n junction and R(V ) is the effective nonlinear resistance due to the dissipation,

which is also voltage-dependent. Li and Ri are the intrinsic linear inductance and resistance

of the diode, respectively.

Thus the current through the diode I is the sum of the change rate of the accumulated

charge on the depletion layer capacitance dQ/dt, and the dissipative current ID (V ).

For the Skyworks SMV1231-079 varactor used in our study, the following equations, with

the parameters VD , Vp , I0 and M provided by the manufacturer, can be used to describe its

performance.

C(VD ) = C0 (1 − VD /Vp )−M , (3.1)

( )
ID (VD ) = I0 e(VD /VT ) − 1 , (3.2)

where C0 = 2.2 pF is the DC rest capacitance, Vp = 1.5 V is the intrinsic potential and

M = 0.8. VT = kB T /e is the thermal voltage, where kB is the Boltzmann constant, T is the

temperature, and e is the electron charge. Fig. 3.2 shows the voltage-dependent capacitance

and the dissipative current plotted with the above equations.

20 10

8
15

6
ID (mA)
C (pF)

10
4

5
2

0 0
−1 −0.5 0 0.5 1 −1 −0.5 0 0.5 1
V (V) V (V)

(a) (b)

Figure 3.2 Characteristic curves of the varactor Skyworks SMV1231-079.

(a) The capacitance of the varactor vs. the applied voltage. (b)
The dissipative current vs. the applied voltage.
 48

3.3 Self-tuning of SRR with varactor

Varactors are most commonly used as voltage-controlled capacitors under reverse biasing.

The magnetic metamaterials with varactors have also been studied by other groups [96,98,103].

They use external DC bias to tune the capacitance of the varactor, and the resonance frequency

of the SRRs. Since an SRR is essentially an LC resonator, the tunability of its resonance with

a tunable capacitor is expected. However, this tunability by DC biasing is different from the

nonlinear effect we want to study. For a nonlinear metamaterial, the response is a function of

the applied electromagnetic field. In other words, the response of a nonlinear metamaterial is

self-tuned by the applied electromagnetic field. In general, the nonlinear response of an SRR

is negligible unless the field intensity is very strong. It is not

In this chapter, as a model system, we study the self-tuning of nonlinear SRR with the help

of varactors. The study of the response of individual SRRs and the coupling between SRRs

would be helpful in the design of nonlinear metamaterials.

(a)

(b)

Figure 3.3 (a) The fabricated SRR is shown on left and the SRR with a
varactor mounted into the gap is shown on right. (b) The loop
antenna is positioned on top of the SRR and connected to the
network analyzer (not shown). The reflected signal is measured
by the network analyzer.
 49

In our study, single-ring SRR with a single gap is used for simplication. The ring has an

outer diameter of 7 mm, an inner diameter of 6 mm and a slit of width 0.7 mm The SRR

is fabricated on a PC board substrate of 0.8 mm thick, with a picture shown in Fig. 3.3(a).

The magnetic resonance frequency of the SRR is around 4 GHz. To tune the resonance of

the SRR, a varactor is mounted across the gap of the ring, the picture of the tunable SRR is

also shown in Fig. 3.3(a). The resonance of the SRR is captured by a loop antenna, as shown

in Fig. 3.3(b). The loop antenna is made by a semi-rigid coaxial cable and connected to the

port of the network analyzer. The loop antenna has a diameter of 8 mm and is not resonant

over the frequency region we are measuring. Most of the power is reflected back to the cable

unless the loop antenna is coupled to the resonance of the SRR. In the reflection spectrum,

the resonance is shown as a dip in the flat background. By changing the input power, we can

see the resoance change of the SRR by looking at the reflection spectrum of the loop antenna.

For comparison, an SRR with a high-Q capacitor (2pF ) mounted in the gap is also studied

as the corresponding linear system. The capacitance of the capacitor is chosen so that it is

approximately the same as the rest capacitance of the varactor diode used in this study. With

the capacitor or diode, the capacitance of the SRR is greatly increased and the resonance

frequency is moved down to 0.9 GHz. The resonance of the SRR with high-Q capacitor does

not shift with input power levels, as expected. Note that the insertion of the capacitor boosts

the ratio of wavelength and structure size to about 50. This can be useful to miniaturize the

size of potential microwave devices [105].

The measured reflection with the SRR loaded with a varactor at different input power levels

is shown in Fig. 3.4. When the SRR is off resonance, the reflection spectrum is flat, which

means the signal of the loop antenna is not affected by the SRR. At the resonance of the SRR,

the loop antenna is coupled to the SRR and energy from the antenna is dissipated during the

process. So we observe a dip on the reflection spectrum. When the incident power is low,

the reflection dip is sharp and narrow, which means the loss at resonance is relatively low.

When the incident power is increased, the resonance frequency is shifted to lower frequencies.

Meanwhile, the reflection dip gets shallower and wider, which means the loss is higher. When
 50

5
−15dBm
0 −12dBm
−5 −09dBm
−06dBm
−10 −03dBm
+00dBm
|S | (dB) −15 +03dBm
+06dBm
11
−20
+09dBm
−25
−30
−35
−40
0.6 0.7 0.8 0.9 1
Frequency (GHz)

Figure 3.4 The reflection of the SRR loaded with one varactor, measured
by a loop antenna. The curves show the measurement results
at input power from -15 dBm to 9 dBm in 3 dBm steps.

the incident power gets even larger, the response of the reflection starts to show discontinuous

jumps. Also, at lower power, the frequency shift is relatively small; at higher power, the

frequency shifts drastically.

When the input power gets higher, the capacitance gets larger. Thus the resonance fre-
√
quency, determined by 1/ LC, is smaller. The different behaviors at low power and higher

power is because the sources of the nonlinearity are different. At lower power, the induced

voltage on the varactor diode is relatively small and the only source of nonlinearity is the

depletion layer capacitance. In this case the dissipative current is not involved yet so that the

loss during the process is relatively small. Also the capacitance change due to the change of

power is small so that the frequency shift is not so obvious. At higher power, the induced

voltage on the varactor diode gets higher and the dissipative current sets in. The nonlinearity

is now from the combined effect of the nonlinear capacitance and the nonlinear current. Thus

the loss during the resonance is higher and the reflection dip gets shallower and wider. Also

the nonlinearity is much stronger and the frequency shift due to the change of incident power
 51

is much larger.

Figure 3.5 The hysteresis effect at high input power levels. The blue curves
are measured for forward sweep (the source frequency of the net-
work analyzer is scanned from low to high) and the red dashed
curves are measured for reverse sweep (frequency is scanned
from high to low).

We now take a closer look at the reflection spectrum at high incident powers. To get the

reflection spectrum, the reflection of the loop antenna at different excitation frequencies is

recorded. The network analyzer records the reflection at one frequency and then move to the

next frequency until all the frequencies are recorded. The frequencies can be swept in two

ways: from low frequency to high frequency, or from high frequency to low frequency. We

call the sweep from low frequency to high frequency forward sweep, and the sweep from high

frequency to low frequency reverse sweep. The two sweep methods are both obtained and

plotted for higher input powers, as shown in Fig. 3.5. We see discontinuous jumps in both

forward sweep and reverse sweep. Moreover, the reflection curve jumps at different frequencies

for forward sweep and reverse sweep. The higher the input power, the larger the discrepancy.

This is the behavior of a bistable resonance. The resonance frequency is multivalued and

which value it chooses depends on the history of the coupling between the antenna and the
 52

SRR. The discrepancy between forward sweep and reverse sweep curves is because they pick

different branches. When the input power is higher, the multivalued frequency region gets

larger thus the discrepancy between the forward sweep and reverse sweep gets larger. The

above experimental results can be explained by a nonlinear resonator model shown below.

3.4 The simplified driven nonlinear resonator model

Below, we present some analytical arguments to explain the tunability and the hysteresis

effects of the nonlinear SRRs observed experimentally.

Figure 3.6 The circuit model of the coupling between the driven loop an-
tenna and the SRR loaded with one varactor.

The system of the driven loop antenna and the nonlinear SRR can be modeled by an

equivalent circuit, as shown in Fig. 3.6. The signal source, shown on the left side, is connected

to the loop antenna; the antenna is coupled to the nonlinear SRR via mutual inductance. Ie

and I represent the current in the loop antenna and the SRR; R0 and RS are the resistance of

the antenna and the SRR, respectively. Then the driven RLC circuit can be expressed by the

following voltage equation

dI
−L − RS − VD = ε(t), (3.3)
dt
where L is the inductance of the resonator determined by the ring geometry, and ε is the driven

term provided by the loop antenna.

As we mentioned before, both the depletion layer capacitance and the dissipative current

are voltage dependent, which makes the analysis of the system complicated. However, from
 53

Fig. 3.2, we can see that the dissipative current can be neglected for small signal excitations. For

simplication, the following analysis assumes small signal excitation and considers the nonlinear

capacitance only. The current through the diode is then determined only by the change rate

of the accumulated charge.

For small excitations, ID can be neglected so the current can be estimated by I ≈ dQD / dt.
dQD
From C(VD ) = dVD , we can determine the time-dependent charge,

C0 Vp [ ]
QD = 1 − (1 − VD /Vp )1−M . (3.4)
1−M

Assume VD < Vp , then the voltage across the diode can be expressed by the charge
[ ( ) 1 ]
1 − M 1−M
VD (q) = Vp 1 − 1 − q , (3.5)
Vp

where the renormalized voltage is defined as q = QD /C0 .

The equation of motion is now

d2 q dq
2
+γ + ω02 VD = −ω02 ε(t), (3.6)
dt dt

where ω02 = 1/(LC0 ) and γ = ω02 RS C0 .

Expand the restoring term VD by the Taylor series for small oscillations (the oscillation

amplitude satisfies (1 − M )|q| < Vp ) and omit the higher order terms,

M 2 M (2M − 1) 3
VD (q) ≈ q − q + q . (3.7)
2Vp 6Vp2

Assume harmonic excitation so that ε(t) = f cos(ωt), where f is the excitation amplitude and

ω is the excitation frequency, the equation of motion is further estimated by

d2 q dq
2
+γ + ω02 q + αq 2 + βq 3 = −ω02 f cos(ωt), (3.8)
dt dt
ω02 M ω02 M (2M −1)
where α = − 2Vp , β= 6Vp2
. This is now a nonlinear driven oscillator problem [106].

The driven frequency can be written as ω = ω0 + δ. When δ is small, the driven frequency

is close to the resonance frequency. Without the q 2 and q 3 term, the oscillator is linear and

the amplitude of oscillation, b, is given by

( )
b2 δ 2 + γ 2 /4 = ω02 f 2 /4. (3.9)
 54

Figure 3.7 The oscillation amplitude vs. source frequency calculated from
the simplified nonlinear oscillator model. The curves from bot-
tom to top correspond to excitation power from low to high.
The blue arrow shows the jump for forward sweep of the top
curve and the red arrow shows the jump for reverse sweep of
the top curve.
 55

The nonlinear q 2 and q 3 terms make the eigen-frequency amplitude dependent, ω0 → ω0 + κb2 ,
3β 5α2
where κ = 8ω0 − 12ω03
. Then δ → δ − κb2 , and the oscillation amplitude satisfies the equation
[( )2 ]
b2 δ − κb2 + γ 2 /4 = ω02 f 2 /4. (3.10)

This is a cubic equation about b2 and the real roots give the amplitude of oscillations. When

the external excitation is small, the oscillation amplitude is also small and the higher orders

of b may be neglected and the oscillation can be considered to be linear. When the excitation

power is larger, the curve is distorted and the resonance shifts to a lower frequency, since, in

our case, κ is negative. When the excitation power is large enough, there are three real roots

of b2 and the curve is folded over, see Fig. 3.5. The branch in the middle is unstable and the

oscillation tends to go to the other two branches. In experiment, the oscillation follows the lower

branch until it jumps to the higher branch for forward sweep, and follows the higher branch

until it jumps to the lower branch. So, the hysteresis effect is observed in our experimental

measurements. Note, when the voltage on the varactor is larger than 0.5 V, the nonlinear

DC dissipative current sets in and this increases the loss on the SRR and the reflection dip

measured is not as strong as the small oscillation case. See Fig. 3.4, the reflection minimum

increases from around -40 dB to -3 dB when the input power is increased from -15 dBm to 9

dBm. This is not covered in the simplified model of the nonlinear oscillator model.

3.5 Nonlinear SRR with back-to-back varactors

To remove the nonlinear DC current and obtain a better self-tuning effect, a new SRR is

fabricated. The SRR ring is of the same geometry, but with another identical cut on the other

side of the ring and a varactor is mounted onto each of the cuts, see inset in Fig. 3.8. The

varactors are arranged back-to-back such that no DC current can circulate in the SRR and

the effective capacitance characteristics C(V ) of the two varactors is now symmetric. This

configuration has similar effect of one heterostructure barrier varator (HBV) diode [104]. The

two varactors can now be regarded as two tunable capacitors connected in a series. The total

capacitance is smaller than a single varactor at the same power level, so that the resonance
 56

frequency shifts to a higher region (see Fig. 3.8). When the input power increases, the ef-

fective capacitance decreases and the resonance frequency increases. When the input power

is changed from -14 dBm to 2dBm, the resonance frequency is shifted upward by 20%. Also

see from Fig. 3.8 that the resonance strength and the quality factor are almost the same for

different input power levels. These nice features make this configuration a perfect candidate

for nonlinear tuning in metamaterials. The only problem with this configuration is that the

varactors discharge very slowly, due to the lack of a circulating current.

Figure 3.8 The reflection of the SRR loaded with two back-to-back varac-
tors, measured by a loop antenna. The curves show the mea-
surement results at power levels from -14 dBm to 2 dBm, in 2
dBm steps. The inset shows the sample.

3.6 Coupling between nonlinear SRRs

To make nonlinear metamaterials out of the nonlinear SRRs, the mutual coupling of the

varactor loaded SRRs must to be studied. Since the mutual coupling of two coplane SRRs

are very weak, we studied the case of two parallel SRRs with the same axis (solenoid case).

The loop antenna is also parallel to the two SRRs and has the same axis. The distance of the

antenna and the first SRR is fixed, and the second SRR is moved away from the first SRR.
 57

Reflection is measured by the loop antenna for different distances between the two SRRs. In

Fig. 3.9(a), we present the frequency-dependence of the reflection coefficient for two linear

SRRs, as the distance between the two SRRs increases. As one can see from Fig. 3.9(a), if the

distance between the SRRs is large, only one reflection resonance is observed and as the two

SRRs move closer, the mutual coupling becomes stronger and the reflection splits. The closer

the two SRRs, the wider the split. For two linear SRRs, the mutual coupling can be calculated

analytically with a simple LC model.

Figure 3.9 The reflection of (a) two linear SRRs and (b) two nonlinear
SRRs at input power of −15 dBm. The legend shows the dis-
tance between the two SRRs. The distance between the first
SRR and the antenna is fixed.

In Fig. 3.9(b), we present the frequency-dependence of the reflection coefficient for two

nonlinear SRRs, as the distance between the two SRRs increases. The incident power is −15

dBm, which is relatively low and the results presented in Fig. 3.9(b) are almost equivalent to

the linear SRRs presented in Fig. 3.9(a), except that the reflection dip is not as deep as the

linear case, due to a higher loss in the varactor than the high-Q capacitor. When the input

power is low, the nonlinear SRRs behaves like the linear ones and the splitting is nicely seen in

the -15 dBm case of the 2d plot in Fig. 3.10. When the input power gets higher, the resonance

of the SRR gets broadened, due to higher loss and the splitting becomes worse, as seen in

Fig. 3.10. One can clearly see from Fig. 3.10 that the hybridization gets very weak as the
 58

incident power increases.

Figure 3.10 The 2d plot of the reflection of two coupled nonlinear SRRs.
The x-axis shows the distance between the two SRRs and the
y-axis shows the frequency. The four figures from left to right
display the result at input power from -15 dBm to 0 dBm.
The SRRs and the antenna are arranged such that they have
the same axis and the planes of the rings are parallel to each
other. The distance between the loop antenna and the first
SRR is fixed and the second SRR is movable along its axis, as
seen from the inset.

3.7 Conclusions

In conclusion, we have demonstrated experimentally dynamic tunability, hysteresis, and

bistable behavior in nonlinear SRRs. The nonlinear SRR has a typical SRR design and we

have soldered in the gap of the SRR a commercial varactor diode. Tunability of the resonance

frequency was completed dynamically by increasing the incident power of the vector network

analyzer. We have introduced different nonlinear designs and observed experimentally that the

resonance frequency can decrease or increase by increasing the incident power. This way, we

are able to change the sign of the nonlinearity. A theoretical model was given that explained

the observed nonlinear effects. Finally, we study the hybridization effects of the linear and

nonlinear SRR.
 59

CHAPTER 4. Chiral metamaterials

Conventional metamaterials require both negative ϵ and negative µ to achieve negative in-

dex of refraction. Chiral metamaterial is a new class of metamaterials offering a simpler route

to negative index and consequently negative refraction. In this chapter, we briefly review the

history of metamaterials and the developments on chiral metamaterials. We study the wave

propagation properties in chiral metamaterials and show that negative index can be realized in

chiral metamaterials with a strong chirality, with neither ϵ nor µ negative required. We have

developed a retrieval procedure, adopting uniaxial bi-isotropic model to calculate the effective

parameters n± , κ, ϵ and µ of the chiral metamaterials. Our work on the design, numerical cal-

culations and experimental measurements of chiral metamaterials, will be introduced. Strong

chiral behaviors such as optical activity and circular dichroism are observed and negative re-

fraction is obtained for circularly polarized waves in these chiral metamaterials. The potential

applications of chiral metamaterials, including resonant absorbers, will be discussed.

4.1 Introduction

A chiral medium is composed of particles that can not be superimposed on their mirror

images [107]. A chiral medium has different responses for a left circularly polarized (LCP)

wave and a right circularly polarized (RCP) wave due to the intrinsic chiral asymmetry of the

medium. Also, there is cross-coupling between electric field and magnetic field going through

a chiral medium. A dimensionless chirality parameter κ is used to describe this cross-coupling

effect. The refractive indices of RCP and LCP waves become different due to the existence of

non-zero κ. These will be explained in detail in the following sections.

There is a long history in the study of chiral media. In the early 19th century, the optical
 60

rotation in quartz crystals as well as some liquids and gases has already been discovered by Biot

and others [108]. Biot also suggested that the phenomenon has a root in the molecules. The

handedness nature of the molecules in optically active materials is confirmed by Pasteur in the

1840’s [107]. The discoveries turned out to be useful in areas such as analytical chemistry and

pharmaceutics. In 1873, Lord Kelvin first used the word “chirality” to describe the handedness

in his lectures [109]. In 1910’s Lindeman managed to bring optical activity phenomenon

in visible lights to radio waves, with a collection of helical coils served as artificial chiral

“molecules” [107]. The studies of chiral media in microwave region have found applications in

many areas such as antennas, polarizators, and waveguides [110].

In 2003, Tretyakov et.al. [111] discussed the possibility to realize negative refraction by

chiral nihility. The authors first proposed the idea to fabricate a metamaterial composed

of chiral particles, such as helical wires. To get negative refraction for one of the circular
√
polarizations, κ needs to be larger than ϵµ. In natural materials such as quartz and sugar
√
solutions, κ is generally much smaller than 1, while ϵµ is generally larger than 1. So negative

refraction is not possible in natural chiral materials. However, with chiral metamaterials, the

macroscopic parameters can be designed. The idea of chiral nihility is that, when ϵ and µ of a

chiral medium are small and very close to zero, the chirality can make the refractive index for

one circular polarization to become negative, even when κ is small.

The metamaterial based on chiral nihility is a special case of chiral metamaterials. In

2004, Pendry [2] discussed in general the possibility to achieve negative refraction in chiral

metamaterials. He analyzed the conditions to realize negative refraction in chiral metamaterials

and showed that they are simpler than regular metamaterials, which require both electric and

magnetic resonances to get negative ϵ and negative µ. In chiral metamaterials, neither ϵ nor

µ needs to be negative. As long as the chiral parameter κ is large enough, negative refraction

can be obtained in chiral metamaterials.

 61

4.2 Wave propagation in chiral media

Chiral media belong to a wider range of bi-isotropic (BI) media. BI media [107] are char-

acterized by the following constitutive relations

√
D = ϵ0 ϵE + (χ + iκ) µ0 ϵ0 H (4.1)
√
B = µ0 µH + (χ − iκ) µ0 ϵ0 E (4.2)

where ϵ is the relative permittivity of the medium, ϵ0 is the permittivity of vacuum, µ is the

relative permeability of the medium and µ0 is the permeability of vacuum. The difference

between BI media and regular isotropic media lies in the extra terms of the constitutive rela-

tions. χ is the dimensionless magneto-electric parameter, and describes the reciprocity of the

material. Materials with χ ̸= 0 are non-reciprocal. κ is the dimensionless chirality parameter

of the material. Materials with κ ̸= 0 are chiral. The imaginary unit i in the formula is from

the time-harmonic convention e−iωt .

Depending on the values of χ and κ, the BI media can be classified as

• General BI medium: non-reciprocal (χ ̸= 0), chiral (κ ̸= 0);

• Tellegen medium: non-reciprocal (χ ̸= 0), non-chiral (κ = 0);

• Pasteur medium: reciprocal (χ = 0), chiral (κ ̸= 0).

We are interested in the Pasteur medium, or the reciprocal chiral medium.

Consider the plane wave propagation in an isotropic chiral medium. Combine the above

constitutive relations with χ = 0 and the frequency-domain source-free Maxwell’s equations,

the following wave equation can be obtained for the electric field E:
( )
k × (k × E) = −k02 ϵµ − κ2 E − 2iκk0 (k × E) (4.3)

where k is the wave vector in the chiral medium, k0 = ω/c is the free-space wave vector and
√
c = ϵ0 µ0 is the speed of light in vacuum.

For simplicity, and without loss of generality, we assume k = kẑ. Then the wave equation

is simplified and k is solved:

k± = k0 (n ± κ) (4.4)
 62

√
where n = ϵµ is the index of refraction of the medium without chirality. From this equation
k+ −k−
we can write the chiral parameter κ as κ = 2k0 .

The eigen-vectors, or the allowed solutions of plane waves in the chiral medium can then

be obtained from E(r) = (E0x x̂ + E0y ŷ) eikz . Then we get the following relations:

E0y k 2 (n2 − κ2 ) − k±
2
= 0 = ±i (4.5)
E0x 2ik0 κk±

So, the wave vector solution k+ corresponds to the eigen-vector of RCP wave and k−

corresponds to the eigen-vector of LCP wave. Here the handedness is defined as seen from

source, or as looking in the direction of propagation. Define the index of refraction of RCP/LCP

waves as n± , from the relation k± = n± k0 , we can get

n± = n ± κ (4.6)

The polarization plane of a linearly polarized light will rotate when it passed through a

chiral medium. This polarization effect is called optical activity and is characterized by the

polarization azimuth rotation angle of elliptically polarized light

1 1
θ = δ = [arg(T++ ) − arg(T−− )] (4.7)
2 2

The first subscript in T++ and T−− indicates the initial polarization and the second subscript

indicates the transmitted polarization. So T++ and T−− are the transmission coefficients for

RCP and LCP waves, respectively.

Due to the chiral nature of the medium and the circularly polarized waves, the LCP and

RCP light interacts with the particles of the chiral medium differently. This causes the differ-

ence in absorption and distortion of the two polarizations going through the medium, which

is called circular dichroism. Since the impedance of the chiral medium is the same for the two

different polarizations, the reflections of the two polarizations are the same. Then the circular

dichroism is characterized by the ellipticity, which is defined from the difference in transmitted

power of the two polarizations.

( )
1 |T++ |2 − |T−− |2
η = sin−1 (4.8)
2 |T++ |2 + |T−− |2
 63

4.3 Effective parameter retrieval of chiral media

In the study of metamaterials, effective parameter retrieval is an important technique to

characterize the EM properties of the effective media and is used extensively by researchers,

with both numerical and experimental methods to guide the design of new metamaterials and

identify the negative refractive behavior of metamaterials.

Effective parameter retrieval is the procedure of obtaining the macroscopic parameters of

a medium based on the transmission and reflection coefficients (S parameters) from a planar

slab of this medium. For a homogeneous material, ϵ, µ and the refractive index n are intrinsic

properties of the material and are independent of the thickness of the slab. Thus one can

choose a slab as thin as possible to do the measurements, and thus obtain the parameters

without ambiguity. However for a metamaterial slab, the structures are inhomogeneous and

the smallest slab thickness is limited by the unit cell size of the metamaterial [112–114]. The

retrieval solution in this case is generally multi-branched. The branches need to be chosen

carefully to obey energy conservation rules.

For a chiral slab, the retrieval process is analog to that for regular metamaterials, but a little

more complicated. The refractive indices for the two eigen-solutions (RCP and LCP waves)

need to be calculated [115]. Consider a circularly polarized plane wave normally incident upon

a homogeneous one-dimensional (1D) chiral slab of thickness d in vacuum, with refractive

index n± and impedance Z for RCP/LCP waves (see Fig. 4.1). The incident wavevector k0

is in z-direction and the wavevector in the chiral slab for RCP/LCP wave is k± . The electric

and magnetic field vectors are all tangential to the interfaces in this scenario.

Suppose the incident RCP/LCP electric field has unit amplitude, the reflection coefficient

is given by R± and the transmission coefficient after the second interface is given by T± .

Multireflection happens inside the chiral slab and the transmitting and reflecting waves inside
′ for RCP and LCP waves (see Fig. 4.1).
the chiral slab are represented by coefficients T±′ and R±

Note that the polarization state is reversed after reflection. Define normalized impedance of
√
the chiral slab as z = Z/Z0 , where Z0 = µ0 /ϵ0 is the free space impedance.
 64

Figure 4.1 The transmission and reflection coefficients of a plane wave in-
cident upon a chiral slab from left.

The tangential electric field and magnetic field are continuous at the first interface (x = 0).

1 + R± = T±′ + R±
′
(4.9)
T±′ − ′
R±
1 − R± = (4.10)
z

Similarly at the second interface (x = d),

T±′ eik± d + R±
′ −ik∓ d
e = T± (4.11)
T±′ eik± d − ′ e−ik∓ d
R±
= T± (4.12)
z

Note that k+ + k− = 2nk0 , we can get the transmission and reflection coefficients from the

above equations.

4zeik± d
T± = (4.13)
(1 + z)2 − (1 − z)2 e2ink0 d
(1 − z 2 )(e2ink0 d − 1)
R± = (4.14)
(1 + z)2 − (1 − z)2 e2ink0 d
 65

R+ and R− are equal since the impedance for RCP and LCP waves is the same. If we

denote T and R as the transmission and reflection coefficients for the κ = 0 medium, we have

R± = R (4.15)

T± = T e±iκk0 d (4.16)

The impedance and refractive index can be calculated from the above coefficients.

√
(1 + R)2 − T+ T−
z = ± (4.17)
(1 − R)2 − T+ T−
{ [ ( )] }
i 1 z−1
n± = log 1− R ± 2mπ (4.18)
k0 d T± z+1

where m can be any integer.

The sign of the square root in Eq. 4.17 and the multi-branches in Eq. 4.18 need to be

chosen carefully according to the energy conservation principle, i.e. the real part of impedance

z must be positive, as well as the imaginary part of n.

Once z and n± are fixed, the other parameters can be identified subsequently. κ = (n+ −

n− )/2, n = (n+ + n− )/2, µ = nz and ϵ = n/z.

4.4 Review of work on chiral metamaterials

Since the idea was first proposed [2,111], a lot of work has been devoted toward the theoret-

ical understanding, numerical modeling and experimental realization of chiral metamaterials.

4.4.1 Theoretical studies

In Pendry’s pioneering work [2], he proposed a design to realize strong chiral metamaterials.

The design is a twisted version of the “swiss roll” structure, as shown in Fig. 4.2(a). Realistic

parameters of the structure were estimated by theoretical calculations. Very recently, this

design has been shown by experiments to have strong chirality and negative refractive index

in microwave regime [116].

 66

Figure 4.2 Different designs of chiral metamaterials. (a) The twisted “swiss
roll” [2]; (b) The canonical helix [3]; (c) The chiral SRR [4]; (d)
The bi-layer “swastika” [5]; (e) Dual strips with tilted bridge [6];
(f) Bi-layer rosettes [7]
 67

The idea of making use of helical inclusions in the pioneering work of Tretyakov et.al. [111]

was further studied [3]. An isotropic chiral metamaterial composed of randomly distributed

helical inclusions (shown in Fig. 4.2(b)) was theoretically modeled. The chirality of the pro-

posed metamaterial can be very high. The possibilities of negative refraction and “perfect

lensing” [29] were also discussed in this paper. In an isotropic strong chiral medium, the re-

fractive index of one of the circularly polarized waves can be negative. With such a chiral

medium slab, it is shown that the circularly polarized wave can be focused with subwavelength

resolution [3,117], which works in the similar way as Pendry’s “perfect lens” for linearly polar-

ized waves [29]. Later on, the idea of building a bulk chiral metamaterial with a negative index

of refraction, by manipulating the parameters of the helical inclusions, was discussed [118].

The refractive properties of uniaxially anisotropic chiral media composed of helices were stud-

ied theoretically [119]. The conditions to obtain negative refraction was shown to be easier in

chiral metamaterials than in regular metamaterials.

Other than negative refraction, another interesting behavior in strong chiral medium is

pointed out [120]. Due to the different refractive indices of LCP and RCP waves, the reflection

at the interface between a chiral medium and a nonchiral medium is worth to study. Generally,

for one circularly polarized wave, both RCP and LCP waves will be reflected back to the chiral

medium. A particularly interesting case is the reflection of a circularly polarized wave at the

interface between a strong chiral medium and a perfect electric conductor (PEC). It is shown

that one of the reflected wave will be on the same side of the interface normal, or negatively

reflected [120]. Based on this property, a circularly polarized wave can be partially focused

inside the chiral medium when it is bounded with a PEC mirror.

The reflection between the boundary of vacuum and an isotropic chiral medium is studied

theoretically [121]. If the parameters of the chiral medium can be designed such that the wave

impedance and the wave vector of the medium are identical to vacuum, the chiral medium

could be transparent to one of the circularly polarized waves, i.e., no reflection or refraction

is expected at the boundary. In this case, the other circularly polarized wave can be mostly

reflected. Based on this idea, a circular polarizing beam splitter is proposed [121].
 68

As we discussed above, the conditions to get negative refraction in a chiral medium are:

small ϵ and µ, large κ at resonance. Relatively high losses are associated with the strong

resonances. New ideas have been proposed to get negative refraction in different ways. Qiu

et.al. studied theoretically the wave propagating properties in gyrotropic chiral media, where

both ϵ and µ are described by gyrotropic tensors and κ is described by a scalar [122]. Their

study showed that negative refraction can be realized with fewer restrictions and all parameters

in the ϵ and µ tensors, as well as κ, can be positive when negative refraction occurs. The same

group also analyzed the behavior of another kind of chiral media, where the chirality parameter

κ is describe by a tensor with non-zero off-diagonal elements while ϵ and µ are described by

scalars [123]. The theoretical model showed that negative refraction can be achieved without

the requirement of very strong κ due to the gyrotropic parameters.

So far, most of the studies on isotropic chiral metamaterials based on helical inclusions are

theoretical. No experimental result on isotropic chiral metamaterials has been reported yet.

This is partly due to the difficulty in the fabrication and homogenization of 3D structures.

4.4.2 Experimental realizations

On the other hand, researches on planar chiral metamaterials have also been reported by

many groups, both in theories and experiments [5, 7, 124–128]. Planar structures are easier to

fabricate than bulk media and there are very interesting behaviors and potential applications

in planar structures, such as strong optical activity and circular dichroism in a thin film.

Zheludev and coworkers at University of Southampton first reported the optical activity of a

planar chiral structure with experiments done in optical regime [124]. The chiral sample is

composed of a single layer of a two-dimensional square array of rosette-shaped wires. The

patterns are closely arranged and there is coupling between neighboring patterns. A very large

rotation of polarization azimuth (greater than 30◦ ) is observed. The rotation is found to be

associated with both the arrangement of the patterns and the chirality of the pattern itself.

Later on, the same group studied a bi-layer rosette-shaped structure (see Fig. 4.2(f)) in

microwave regime [125]. The patterns are physically separated on each layer and the patterns
 69

on the second layer are rotated by an angle, instead of identical to the first layer. This turned

out to give an extremely strong rotation, which, in terms of rotary power per wavelength, is

five orders of magnitude larger than a gyrotropic quartz crystal. More recently, this bi-layer

structure was proved by experiments to have a negative index of refraction for one of the

circularly polarized waves [7].

Meanwhile, the circular dichroism and polarization rotation properties of planar chiral

structures, composed of variants of the rosettes, have also been studied by many other groups,

both theoretically [126] and experimentally [5,126–128]. Fig. 4.2(d) shows the bi-layer “swastika”

design used to realized strong circular dichroism [5]. More recently, negative refractive index

of a planar chiral metamaterial, built with arrangements of dual strips connected by a tilted

bridge (shown in Fig. 4.2(e)), was demonstrated by parameter retrieval in terahertz frequen-

cies [6].

Actually, the structure on each layer of the bi-layer film does not even have to be chi-

ral to achieve negative refraction. Zhou et.al. [8] recently demonstrated strong rotation of

polarization azimuth and negative refraction on a bi-layer structures composed of non-chiral

cross-wires.

The layout of the proposed structure is shown in Fig. 4.3. A 18 × 14 array cross-wires is

patterned on a double side copper-clad FR-4 board. The transmission and reflection coefficients

are measured in the same manner as the chiral SRR experiments. Due to the asymmetric

geometry along the propagating direction, the transmission responses for RCP and LCP split

into two curves. At the two resonance frequencies of 6.3 GHz and 7.3 GHz, the azimuth

rotation and ellipticity reach their maximum values. In the region between two resonance

peaks (around 6.9 GHz), which is also the region with low loss and nearly zero dichroism,

rotation of polarization can achieve −50◦ with η ≈ 0 [8].

The mechanism of the resonances for the bi-layer cross-wire pair design can be understood

by studying the current density distribution as shown in and 4.4. At the magnetic resonance,

the anti-parallel current exists on the top and bottom layer (Fig. 4.4(a)), which is an asym-

metric resonance mode. In Fig. 4.4(b), parallel current flows on the two layers, which is a
 70

symmetric resonance mode. The current distribution shows that the bi-layer structures can be

regarded as a chiral version of the short wire pairs [37, 38, 129], which has the similar current

distributions in the symmetric and asymmetric resonance modes.

The chirality of the design comes from the rotation of the second layer of the cross wires

with respect to the first. With an appropriate twist angle, strong chirality can be obtained

and the refractive index can go to negative for RCP wave at one resonance and for LCP wave

at another resonance. When the twist angle is zero, there would be no chiral behavior, even

though the resonances can still be seen [8].

Figure 4.3 (a) Schematic representation of one unit cell of the cross-wire
structure. (b) Photograph of one side of a fabricated mi-
crowave-scale cross-wire sample. The geometry parameters are
given by ax = ay = 15mm, l = 14 mm, w = 1 mm, s = 1.6 mm,
ϕ0 = 45◦ , ϕ = 30◦ . (Adapted from Ref. [8])

(a) (b)

Figure 4.4 The simulated current density distribution for the right circu-
larly polarized EM wave at 6.5 GHz (a) and for the left cir-
cularly polarized EM wave at 7.5 GHz (b). (Adapted from
Ref. [8])
 71

4.5 Nonplanar chiral metamaterial based on chiral SRRs

Although many proposed models for chiral metamaterials are three-dimensional (3D), most

of the experimental results reported so far are planar structures (Fig. 4.2(d)-(f)). Planar

structures are easier to fabricate and they have strong optical activity and circular dichro-

ism. Bi-layer rosette-shaped [7, 125] and cross wires [8] structures were recently studied in the

microwave regime. While the thicknesses of the planar chiral metamaterials in the wave prop-

agation direction are much smaller than the working wavelengths, the size of the patterns in

plane is usually comparable with half-wavelength [7,8]. Moreover, in effective medium theory,

homogeneity of a metamaterial is assumed to calculate the macroscopic parameters. However,

some of the proposed planar chiral metamaterials are not symmetric in the propagation di-

rection [6]. This causes differences in transmission and reflection from opposite directions and

ambiguity in the calculation of effective parameters. So these planar chiral metamaterials are

not suitable to build 3D bulk chiral metamaterials.

In this section we demonstrate experimentally and numerically the negative refraction in a

non-planar chiral metamaterials. We observe strong cross coupling between the magnetic and

electric fields by the two coupled split-ring resonators (SRRs) on two sides of the substrates

at the chiral resonance frequency. The negative refractive index of chiral metamaterials arise

from this strong chirality. Compared with other planar chiral structures, our non-planar chiral

metamaterialsis symmetric along the propagation direction and much smaller than the working

wavelength in all dimensions. So it is a good candidate for isotropic 3D chiral metamateri-

als [130].

4.5.1 The chiral SRR design

The chiral structure of our studies is formed by two identical SRRs separated by a dielectric

substrate and interconnected by vias, as shown in Fig. 4.5(a). The chiral SRRs are arranged

in square lattices to form a 2D slab. Fig. 4.5(b) shows a unit cell of such a 2D array. In

this arrangement, each of the chiral SRRs is shared by the two neighboring cells. When the

structure is excited by external field, both magnetic dipole and electric dipole exist and they
 72

Figure 4.5 (a) The structure of the chiral SRR. (b) A unit cell of the 2D
chiral metamaterial used in simulation. (c) The chiral SRRs
are fabricated on PCBs and cut into strips (d) The strips are
interlocked to form the chiral metamaterial slab.

are both parallel to the axis of the SRRs. While the induced current around in the SRRs

gives the magnetic dipole, electric charges accumulate on the two SRRs with opposite signs,

which introduce a strong electric field between the top and bottom SRRs and give the electric

dipole in the same direction as the magnetic dipole(see Fig. 4.6). A electric dipole can excite

a magnetic dipole in a similar way.

This chiral SRR design (shown in Fig. 4.2(c)) was proposed [4, 131, 132] to develop 3D

isotropic chiral metamaterials [4]. By arranging the chiral SRRs in cubic (fcc) Bravais lat-

tices, a bi-isotropic medium can be obtained. By calculating the susceptibility parameters,

including the chiral parameter κ, by Lorentz local field theory, the medium is shown to provide

negative refraction over a frequency band. We study a single layer of the CM made of chiral

SRRs and demonstrate both numerically and experimentally the strong optical activity and

circular dichroism, and negative refraction. This is an important step toward the design and

characterizations of 3D isotropic chiral metamaterials.

 73

Figure 4.6 The surface current distribution of the SRR near resonance.
The circulating current forms a magnetic dipole, causes the op-
posite charge accumulation on the bottom SRR (+) and the top
SRR (-) and forms an extra electric dipole (see green straight
arrow).

4.5.2 Experimental method

The chiral structures are fabricated on the two sides of FR-4 printed circuit boards (PCBs),

with a relative dielectric constant of ϵr = 3.76 with loss tangent 0.0186 and thickness of 1.6

mm. The metal structures are built on copper with a thickness of 36 µm. The SRRs are 2-gap

split rings with an identical gap width of 0.3 mm. The inner radius of the rings is 1.25 mm

and the outer radius is 2.25 mm. The SRRs on opposite sides of the board are connected by

vias with a diameter of 0.5 mm. The distance between adjacent rings is 8 mm. The sample is

fabricated by 50 strips with 26 cells each, which are then interlocked to form a slab of size 208

mm × 208 mm of the chiral metamaterial. A fabricated strip of SRRs is shown in Fig. 4.5(c)

and the chiral metamaterial slab is shown in Fig. 4.5(d).

The transmission and reflection measurements are done with a vector network analyzer

(Agilent E8364B) in our lab. A pair of standard gain horn antennas are used as transmitter

and receiver, see Fig. 4.7. The signals from these horn antennas are linearly polarized, so the

linear transmission coefficients, Txx , Txy , Tyx and Tyy are measured, where the first subscript

indicates the transmitted field polarization (x− or y−polarized) and the second subscript

indicates the incident field polarization. The circular transmission coefficients, T++ , T−+ , T+−

and T−− , where the first subscript indicates the transmitted field polarization (±, RCP/LCP)
 74

Figure 4.7 The transmission and reflection coefficients of the chiral meta-
material slab is measured by a pair of horn antennas served as
transmitter and receiver. The cables on the horn antennas are
connected to a vector network analyzer (not shown).

and the second subscript indicates the incident field polarization, are converted from the linear

transmission coefficients using the following equation.

 
 T++ T+− 
 
T−+ T−−
 
 (Txx + Tyy ) + i(Txy − Tyx ) (Txx − Tyy ) − i(Txy + Tyx ) 
= 12   (4.19)
(Txx − Tyy ) + i(Txy + Tyx ) (Txx + Tyy ) − i(Txy − Tyx )

A similar expression exists for the circular reflection coefficients R++ , R−+ , R+− and R−− .

4.5.3 Results and discussion

Fig. 4.8(a) and (b) shows the simulated and measured transmission coefficients T++ and

T−− , as well as the reflection coefficients R+− and R−+ , as a function of frequency. There

is obvious difference (about 5 dB) in T++ and T−− around the resonance. The reflection

coefficients R+− = R−+ = R, since the impedances for RCP and LCP waves are identical.

The cross coupling transmission T−+ and T+− are negligible and they are not shown. While

the simulation curves are smooth off the resonance, there are a lot of noise in the experimental

curves. This is mainly due to the scattering from the surroundings and multiple reflections
 75

5 5
T−− T−−
0 0
T++ T++
−5 R+− −5 R+−

T & R (dB)

T & R (dB)
−10 R−+ −10 R−+
−15 −15
−20 −20
−25 −25
−30 −30
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(a) (b)
100 100

50 50
θ (degree)

θ (degree)
0 0

−50 −50

−100 −100
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(c) (d)
5 5
0 0
−5 −5
η (degree)

η (degree)
−10 −10
−15 −15
−20 −20
−25 −25
−30 −30
−35 −35
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(e) (f)

Figure 4.8 From top to bottom, the transmission and reflection amplitudes
for LCP and RCP waves, the azimuth rotation angle for lin-
early polarized waves, and the ellipticity. Simulation results
are shown on left and experiment results are shown on right.

between the horn antennas and the sample. Also, the resonance frequency in experiment is

about 0.1 GHz lower than the resonance frequency in simulation. This is not surprising since

the parameters, particularly the dielectric constant of the PCB, may not be accurate in our

frequency range. The azimuth rotation θ, and the ellipticity η, are calculated based on the

transmission data and presented in Fig. 4.8(c)-(f). At the resonance, the azimuth rotation

reaches a maximum of around −100◦ . Meanwhile, the ellipticity is also the largest (−17◦ ),

meaning a linearly polarized wave is strongly distorted and becomes elliptical.

The retrieved effective parameters of the chiral metamaterial are shown in Fig. 4.9. We

see that the effective index of refraction for RCP wave (Fig. 4.9(a) and (b)) shows strong

response at the resonance and goes negative above the resonance while the index of refraction
 76

6 6
Re(n ) Re(n )
+ +
Im(n+) Im(n+)
4 4

n+

n+
2 2

0 0

−2 −2
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(a) (b)
6 6
Re(n ) Re(n−)
−
Im(n−) Im(n−)
4 4
n−

n−
2

0 0

−2 −2
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(c) (d)
3 Re(n) 3 Re(n)
Re(κ) Re(κ)
2 2
n&κ

n&κ

1 1

0 0

−1 −1

−2 −2
6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(e) (f)
Re(ε) Re(ε)
4 4
Im(ε) Im(ε)

2 2
ε

0 0

−2 −2

6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(g) (h)
Re(µ) Re(µ)
4 4
Im(µ) Im(µ)

2 2
µ
µ

0 0

−2 −2

6 7 8 9 6 7 8 9
Frequency (GHz) Frequency (GHz)

(i) (j)

Figure 4.9 The retrieved effective parameters of the chiral metamaterial.

The figures from top to bottom are the index of refraction for
the two circularly polarized waves n+ and n− , the effective index
of refraction n together with the chiral parameter κ, the effective
relative permittivity ϵ and relative permeability µ. Simulation
results are shown on left and experiment results are shown on
right.
 77

for LCP wave (Fig. 4.9(c) and (d)) changes only slightly and is positive in all the frequency

range. The negative index here is due to a relatively strong chirality κ, and a small value of

n at the resonance (Fig. 4.9(e) and (f)). Although there is some discrepancies in simulated

and experimental ϵ and µ (Fig. 4.9(g)-(j)), it is obvious that, ϵ and µ are not both negative,

which is the signature of conventional negative index metamaterials. The retrieval results

show that the negative refraction comes from a strong chirality of the structure, instead of the

double-negative (ϵ and µ) in conventional metamaterials. The negative index has a minimum

of around 1.7 with a figure of merit about 2. The peak of the imaginary part of the index of

refraction comes from the strong absorption in the PCB substrate at the resonance. The loss

could be reduced significantly when a low-loss substrate can be used.

In summary, the electromagnetic properties of the non-planar chiral metamaterial composed

of chiral SRRs have been studied. Very strong optical activity, as well as circular dichroism

can be achieved with the metamaterial. Moreover, negative index of refraction is obtained

by both numerical simulations and experimental measurement. A 3D bulk medium can be

fabricated by stacking layers of this chiral metamaterial together. Eventually 3D isotropic

chiral metamaterials, with negative refraction due to chirality, can be obtained.

4.6 A microwave absorber made from chiral metamaterial

4.6.1 Loss problem and metamaterial absorbers

Since most proposed metamaterials are metallic resonant structures and rely on strong

resonances, losses are inevitable. The existence of losses deteriorates the performance of po-

tential devices such as superlenses [29]. Different ways to reduce the losses have been studied,

including the use of low-loss materials, the optimization of designs [133] and the use of gain

materials to compensate losses [134].

Instead of trying to reduce the losses, very recently, ideas have been proposed to build

resonant absorbers with metamaterials [135–138], as well as other metallic nanostructures [139].

The absorption is defined as A(ω) = 1 − R(ω) − T (ω), where A(ω), R(ω), and T (ω) are the

absorption, the reflection, and the transmission as functions of frequency ω, respectively. It

 78

is straightforward to get the two design principles to make the absorption as close to unity

as possible: minimize R(ω) and minimize T (ω). To minimize the reflection, we can tune the

parameters of the metamaterial to get the effective ϵ and µ matched so that the impedance of
√
the metamaterial z = µ/ϵ is equal to one and matched to the free space. Thus, the reflection

can in principle be eliminated. To minimize the transmission, the metamaterial needs to be

designed so that the imaginary parts of ϵ and µ are as large as possible since they correspond

to the loss in the metamaterial. Although it is difficult to get the transmission eliminated

by one single layer of the metamaterial, there are ways to achieve unit absorption. The first

is to use multiple layers of such metamaterial films to eliminate the transmission [135, 138];

the second is to use a ground plane to reflect the transmitted wave back [137, 139]. These

two approaches have their advantages and disadvantages. The first approach can obtain in

principle perfect absorption at the resonance peak, with the sacrifice of increasing thickness.

The second approach can give a very thin absorber, but the absorption may not be perfect.

Most of the proposed metamaterial absorbers are composed of conducting electric res-

onators on two sides of a dielectric substrate [135, 136, 138]. The electric response can be

obtained from the excitation of the electric resonators by the electric field, and the magnetic

response is provided by the anti-parallel currents on the two sides of the substrate [40, 129].

These absorbers depend strongly on the polarization of the incident waves [135, 136] as well as

the incident angle. They work only for one polarization at normal incidence and the absorption

drops rapidly for off-normal-incidence cases. The absorber in Ref. [138], which is composed

of paired metallic rods symmetric on two sides of a substrate, covers a wider angle but only

works for one polarization. The design of periodic metallic strips with a ground plate [139]

can have an absorption of over 80% at 70◦ , but is still limited to one particular polarization.

In Ref. [137], the metamaterial absorber, composed of split-ring-shaped electric resonators,

operates over a wide range of incidence angles for both transverse electric (TE) and transverse

magnetic (TM) waves. However, the electric resonators in the absorber are not symmetric and

still have azimuthal dependence.

 79

4.6.2 The chiral metamaterial absorber

Here we propose a different type of resonant absorbers that are made of chiral metamate-

rials. The chiral metamaterial absorber is shown, by both numerical simulations and experi-

mental measurements, to be angle and polarization independent. Near-perfect absorption can

be achieved at the resonance. Moreover, the proposed absorber has a thickness of merely 1/5

of the working wavelength λ. Although the modeling and experiments are done in microwave

frequencies, the absorber can be scaled to find applications at higher frequency regimes [140].

Similar to conventional metamaterials, chiral metamaterials require strong resonances to get

a large κ, thus losses are also associated with chiral metamaterials. In microwave experiments,

the major loss is shown to be a dielectric loss due to the lossy substrates used. The loss can

be reduced significantly by using low-loss substrates. On the other hand, resonant absorbers

can be made with chiral metamaterials.

The chiral metamaterial in this study is based on the chiral SRR structures [130]. The

chiral metamaterial has been shown to have negative refraction due to strong chirality at the

resonance frequency [130]. Meanwhile, significant loss is associated with the resonance, which,

shown by numerical simulations, is mainly due to dielectric loss in the lossy FR-4 board. The

loss can thus be reduced by choosing substrates with a smaller dissipation factor. On the other

hand, a metamaterial absorber can be made out of the current structure.

A chiral metamaterial absorber can be build from the chiral metamaterial slab, backed

with a ground copper plate, and covered with a dielectric plate, see Fig. 4.10. For normal

incidence, the incident wave is rotated by an angle θ when it reaches the ground plane; when

the wave is reflected back to the chiral metamaterial, it is rotated by the same angle θ, but in

the opposite direction. Therefore, the polarization is preserved after the reflection, due to the

reciprocity of the metamaterial. In the case of oblique incidence, part of the reflected wave is

transformed. Simulations have shown that the polarization transformation is less than 3% an

even the incident angle is 85◦ off normal.

 80

Figure 4.10 (a) The chiral metamaterial slab. (b) The fabricated mi-
crowave absorber.

(a) (b)

Figure 4.11 The simulation results of absorption at different angles θ for

(a) TE polarization and (b) TM polarization. Insets illustrate
the two polarizations and the angle θ.
 81

4.6.3 Results and discussion

The numerical simulations are done in CST Microwave studio. With the help of unit-cell

boundaries, an infinitely large slab of the absorber is simulated. With the ground plane in the

back, the transmission is eliminated and the absorption is calculated by AE = 1 − REE − RHE

for TE polarization and AH = 1 − RHH − REH for TM polarization, where the subscript E

indicates TE polarization and H indicates TM polarization. The term REE means the reflection

coefficient of a TE wave from an incident TE wave and RHE means the reflection coefficient

of a TM wave from an incident TE wave. The absorption is calculated for different incident

angles, as shown in Fig. 4.11(a) for TE polarization and Fig. 4.11(b) for TM polarization.

For TE polarization, the absorption is almost unity at normal incidence and remains above

98% until 60◦ . The peak absorption drops for large incident angles but is still above 90%

even when the incidence angle is 70◦ . When taking into account the small shift in the center

frequency (1.25% from 0◦ to 70◦ ), the absorption at the peak frequency of normal-incidence

case is still more than 40% (60%) at 70◦ (60◦ ). For TM polarization, the absorption is above

90% for all incident angles and the center frequency shift is less than 1%. Moreover, since the

metamaterial slab is uniaxial in plane, the absorber has no azimuthal dependence for either

polarization. All these results show that the proposed design can be used as a near perfect

absorber.

Experiments have also been done to test the absorption behavior of the fabricated sample.

A vector network analyzer (Agilent E8364B) and a pair of standard gain horn antennas are

used to measure the reflection coefficient from the sample. The EM wave from the horn antenna

is linearly polarized. By changing the orientation and the angle of the antennas, the reflection

coefficient for both TE and TM polarizations at different incident angles can be measured. The

reflection from a metal plate without the absorber is used for normalization. The absorption is

then calculated and shown in Figs. 4.12(a) and 4.12(b). The absorption peaks at resonance are

smaller than the simulation results. This is partly due to the fact that the actual dissipation

factor of the PC boards may be different from the one used in simulations. The fabrication

imperfection contributes to the discrepancies as well. The measured absorption off resonance

 82

(a) (b)

Figure 4.12 The experimental results of absorption at different angles θ for

(a) TE polarization and (b) TM polarization. Insets illustrate
the two polarizations and the angle θ.

is generally higher than the simulation results and has many small oscillations. This is caused

by the scattering from imperfections in the fabricated structure and the scattering between

the horn antennas due to their relatively large aperture. In spite of these discrepancies, the

experimental measurements agree with the simulation results that high absorption can be

obtained at the resonance for a wide range of incident angles for different polarizations. With

these nice absorption properties, the proposed metamaterial absorber is still relatively thin,

with a thickness of 1/5 of the working wavelength. The strong chirality of the metamaterial

comes from the cross coupling between electric and magnetic field, as has been pointed out. Due

to this cross-coupling effect and the nonplanar arrangement of the resonators, the resonance

can be excited by different polarizations at a wide range of incident angles.

The idea to fabricate microwave absorbers with chiral media can be dated back to 1980s [141].

The interest is due to the extra macroscopic parameter κ other than ϵ and µ. The au-

thors [141] believed that κ can give extra flexibility in the design of phase-matched microwave

absorbers and thus provide improvement to the performance. A debate started soon after

these publications, concerning whether κ can really affect the impedance and enhance the

absorption [142, 143]. Meanwhile, there is not much experimental support for the superiority

of chiral absorbers. A discussion at the conference Bianisotropics’97 concluded that chirality

 83

does not lead to superior absorption [144]. Later experimental results with chiral and achiral

resonant structures show that, resonance structures, no matter chiral or not, in a lossy host can

significantly enhance the absorption at the resonance frequency and chirality does not play a

role in the enhancement of absorption [144]. The superior absorption properties of our design

are due to the very strong resonance that can be excited by both electric and magnetic fields,

and the compact size of the resonators.

In summary, the absorption properties of a resonant absorber made from chiral metama-

terials are studied. The absorber is shown, by both numerical simulations and experimental

measurements, to be working for a very wide angle. The absorber is also independent of polar-

izations and azimuthal angles. Due to the compact size of the resonators, the proposed absorber

has a very small thickness (about 1/5 of the center wavelength). All those features make the

proposed design a perfect absorber. By scaling the parameters, the design in microwave can

be applied to higher frequency regimes.

4.7 Conclusion

In conclusion, we have briefly reviewed the history of metamaterials, from theoretical pro-

posals to experimental realizations in GHz, THz and optical regimes. While the conventional

metamaterials require negative ϵ and µ simultaneously to achieve negative index of refraction,

chiral metamaterials offer an alternative and simpler route. We have studied the wave prop-

agation properties in chiral metamaterials and showed that negative index and consequently

negative refraction can be realized in chiral metamaterials with a strong chirality while nei-

ther ϵ nor µ negative is required. We have developed a retrieval procedure, adopting uniaxial

bi-isotropic model to calculate the effective parameters of the chiral metamaterials. We have

reviewed the progresses in the chiral metamaterials field, both theoretically and experimentally.

We have introduced our new designs, numerical calculations and experimental measurements

of chiral metamaterials. We show that very strong chiral behaviors such as optical activity and

circular dichroism can be observed. Negative refraction for circularly polarized waves can be

obtained with various designs. Losses are associated with the highly resonant structures. We
 84

have demonstrated with experimental studies that chiral metamaterials can be used to create

resonant absorbers with very good performance.

 85

CHAPTER 5. Conclusions

The research in photonic crystals and metamaterials communities is still growing rapidly

and new phenomena are being discovered. These discoveries have potential applications in

many areas from radio frequencies, THz to visible frequencies. This thesis describes my grad-

uate studies in the field.

As the first part, surface modes and wave guiding properties of 2D photonic crystals are

studied. With surface defects, a photonic crystal could support surface modes that are lo-

calized on the surface of the crystal, with mode frequencies within the band-gap. With line

defects, a photonic crystal could allow the propagation of electromagnetic waves along the

channels. Photonic crystals with different surface layers were designed to support surface

modes. Attenuated total reflection technique was used to detect the surface modes. The sur-

face mode dispersion was traced experimentally. Photonic crystal waveguides with different

terminations were studied. It was shown that the waveguiding properties can be improved

with the help of surface modes. Enhanced transmission and directed beaming from a photonic

crystal waveguide of subwavelength aperture was demonstrated. The optical conductance of

a photonic crystal waveguide was also studied experimentally. Due to the discrete waveguide

modes that are allowed to propagate inside the waveguide, the optical conductance changes

with the waveguide width in a staircase manner.

The last decade has seen great progress in metamaterial research. Metamaterials are gen-

erally composed of artificial structures with sizes one order smaller than the wavelength and

can be approximated as effective media. Effective macroscopic parameters such as electric

permittivity ϵ, magnetic permeability µ are used to characterize the wave propagation in

metamaterials. The fundamental structures of the metamaterials affect strongly their macro-
 86

scopic properties. By designing the fundamental structures of the metamaterials, the effective

parameters can be tuned and different electromagnetic properties can be achieved.

One important aspect of metamaterial research is to get artificial magnetism. Metallic

split-ring resonators (SRRs) and variants are widely used to build magnetic metamaterials

with effective µ < 1 or even µ < 0. In the next part, varactor based nonlinear SRRs were built

and modeled to study the nonlinearity in magnetic metamaterials. The nonlinear SRR has a

typical SRR design and we have soldered in the gap of the SRR a commercial varactor diode.

Dynamic tunability, hysteresis, and bistable behavior in nonlinear SRRs were demonstrated

experimentally. Tunability of the resonance frequency was completed dynamically by increasing

the incident power of the vector network analyzer. We have introduced different nonlinear

designs and observed experimentally that the resonance frequency can decrease or increase

by increasing the incident power. A theoretical model was given to explain the observed

nonlinear effects. The mutual coupling between two SRRs at different input power levels were

also studied. The magnetic response of a single SRR and the mutual coupling between SRRs

are important in understanding and modeling nonlinear magnetic metamaterials.

Negative refractive index n is one of the major target in the research of metamaterials.

Negative n can be obtained with a metamaterial with both ϵ and µ negative. As an alternative,

negative index for one of the circularly polarized waves could be achieved with metamaterials

having a strong chirality κ. A retrieval procedure, adopting uniaxial bi-isotropic model has

been developed to calculate the effective parameters of the chiral metamaterials. Designs,

numerical calculations and experimental measurements of chiral metamaterials were presented.

Very strong chiral behaviors such as optical activity and circular dichroism can be observed.

Negative index for circularly polarized waves can be obtained with various designs. Chiral

SRR structures were studied in particular and negative index was achieved. It is also shown,

with chiral SRR design, that chiral metamaterials can be used to create resonant absorbers

with very good performance.

 87

BIBLIOGRAPHY

[1] D. R. Smith, J. B. Pendry, and M. C. K. Wiltshire. Metamaterials and Negative Refrac-

tive Index. Science, 305(5685):788–792, 2004.

[2] J. B. Pendry. A Chiral Route to Negative Refraction. Science, 306(5700):1353–1355,

2004.

[3] S. Tretyakov, A. Sihvola, and L. Jylh. Backward-wave regime and negative refraction in

chiral composites. Photonics Nanostruct.: Fundam. Appl., 3(2-3):107 – 115, 2005.

[4] L. Jelinek, R. Marqués, F. Mesa, and J. D. Baena. Periodic arrangements of chiral scatter-

ers providing negative refractive index bi-isotropic media. Phys. Rev. B, 77(20):205110,

2008.

[5] M. Decker, M. W. Klein, M. Wegener, and S. Linden. Circular dichroism of planar chiral

magnetic metamaterials. Opt. Lett., 32(7):856–858, 2007.

[6] S. Zhang, Y. Park, J. Li, X. Lu, W. Zhang, and X. Zhang. Negative refractive index in

chiral metamaterials. Phys. Rev. Lett., 102(2):023901, 2009.

[7] E. Plum, J. Zhou, J. Dong, V. A. Fedotov, T. Koschny, C. M. Soukoulis, and N. I.

Zheludev. Metamaterial with negative index due to chirality. Phys. Rev. B, 79(3):035407,

2009.

[8] J. Zhou, J. Dong, B. Wang, T. Koschny, M. Kafesaki, and C. M. Soukoulis. Negative

refractive index due to chirality. Phys. Rev. B, 79(12):121104, 2009.

[9] E. Yablonovitch. Inhibited spontaneous emission in solid-state physics and electronics.

Phys. Rev. Lett., 58(20):2059–2062, May 1987.

 88

[10] S. John. Strong localization of photons in certain disordered dielectric superlattices.

Phys. Rev. Lett., 58(23):2486–2489, Jun 1987.

[11] K. M. Ho, C. T. Chan, and C. M. Soukoulis. Existence of a photonic gap in periodic

dielectric structures. Phys. Rev. Lett., 65(25):3152–3155, Dec 1990.

[12] E. Yablonovitch, T. J. Gmitter, and K. M. Leung. Photonic band structure: The face-

centered-cubic case employing nonspherical atoms. Phys. Rev. Lett., 67(17):2295–2298,

Oct 1991.

[13] J. D. Joannopoulos, R. D. Meade, and J. N. Winn. Photonic Crystals: Molding the Flow

of Light. Princeton University Press, 1995.

[14] K.M. Ho, C.T. Chan, C.M. Soukoulis, R. Biswas, and M. Sigalas. Photonic band gaps

in three dimensions: New layer-by-layer periodic structures. Solid State Commun.,

89(5):413 – 416, 1994.

[15] C. M. Soukoulis, editor. Photonic Band Gap Materials. Springer, 1996.

[16] P. Russell. Photonic Crystal Fibers. Science, 299(5605):358–362, 2003.

[17] E. Cubukcu, K. Aydin, E. Ozbay, S. Foteinopoulou, and C. M. Soukoulis. Electromag-

netic waves: Negative refraction by photonic crystals. Nature, 423:604–605, 2003.

[18] H. Kosaka, T. Kawashima, A. Tomita, M. Notomi, T. Tamamura, T. Sato, and

S. Kawakami. Superprism phenomena in photonic crystals. Phys. Rev. B, 58(16):R10096–

R10099, Oct 1998.

[19] C. M. Soukoulis, M. Kafesaki, and E. N Economou. Negative-index materials: New

frontiers in optics. Adv. Mat., 18:1941–1952, 2006.

[20] C. M. Soukoulis, S. Linden, and M. Wegener. PHYSICS: Negative Refractive Index at

Optical Wavelengths. Science, 315(5808):47–49, 2007.

 89

[21] J. B. Pendry, A. J. Holden, D. J. Robbins, and W.J. Stewart. Magnetism from conductors

and enhanced nonlinear phenomena. IEEE Trans. Microwave Theory Tech., 47:2075, Nov

1999.

[22] A. Alu, N. Engheta, A. Erentok, and R.W. Ziolkowski. Single-negative, double-negative,

and low-index metamaterials and their electromagnetic applications. IEEE Antennas

and Propagation Magazine, 49(1):23–36, 2007.

[23] V. G. Veselago. The electrodynamics of substances with simultaneously negative values

of ϵ and µ. Sov. Phys. Usp., 10:509–514, 1968.

[24] J. D. Jackson. Classical electrodynamics. Wiley; 3rd edition, 1998.

[25] R. Marqus, F. Martn, and M. Sorolla. Metamaterials with Negative Parameters: Theory,

Design and Microwave Applications. Wiley, 2008.

[26] J B Pendry, A J Holden, D J Robbins, and W J Stewart. Low frequency plasmons in

thin-wire structures. J. Phys.: Condens. Matter, 10(22):4785–4809, 1998.

[27] D. R. Smith, Willie J. Padilla, D. C. Vier, S. C. Nemat-Nasser, and S. Schultz. Composite

medium with simultaneously negative permeability and permittivity. Phys. Rev. Lett.,

84(18):4184–4187, May 2000.

[28] R. A. Shelby, D. R. Smith, and S. Schultz. Experimental verification of a negative index

of refraction. Science, 292(5514):77–79, 2001.

[29] J. B. Pendry. Negative refraction makes a perfect lens. Phys. Rev. Lett., 85(18):3966–

3969, Oct 2000.

[30] J. B. Pendry, D. Schurig, and D. R. Smith. Controlling Electromagnetic Fields. Science,

312(5781):1780–1782, 2006.

[31] D. Schurig, J. J. Mock, B. J. Justice, S. A. Cummer, J. B. Pendry, A. F. Starr, and

D. R. Smith. Metamaterial Electromagnetic Cloak at Microwave Frequencies. Science,

314(5801):977–980, 2006.
 90

[32] T. J. Yen, W. J. Padilla, N. Fang, D. C. Vier, D. R. Smith, J. B. Pendry, D. N.

Basov, and X. Zhang. Terahertz Magnetic Response from Artificial Materials. Science,

303(5663):1494–1496, 2004.

[33] S. Linden, C. Enkrich, M. Wegener, J. Zhou, T. Koschny, and C. M. Soukoulis. Magnetic

Response of Metamaterials at 100 Terahertz. Science, 306(5700):1351–1353, 2004.

[34] N. Katsarakis, G. Konstantinidis, A. Kostopoulos, R. S. Penciu, T. F. Gundogdu,

M. Kafesaki, E. N. Economou, T. Koschny, and C. M. Soukoulis. Magnetic response

of split-ring resonators in the far-infrared frequency regime. Opt. Lett., 30(11):1348–

1350, 2005.

[35] C. Enkrich, M. Wegener, S. Linden, S. Burger, L. Zschiedrich, F. Schmidt, J. F. Zhou,

T. Koschny, and C. M. Soukoulis. Magnetic metamaterials at telecommunication and

visible frequencies. Phys. Rev. Lett., 95(20):203901, Nov 2005.

[36] S. Zhang, W. Fan, N. C. Panoiu, K. J. Malloy, R. M. Osgood, and S. R. J. Brueck.

Experimental demonstration of near-infrared negative-index metamaterials. Phys. Rev.

Lett., 95(13):137404, 2005.

[37] G. Dolling, C. Enkrich, M. Wegener, J. F. Zhou, C. M. Soukoulis, and S. Linden. Cut-

wire pairs and plate pairs as magnetic atoms for optical metamaterials. Opt. Lett.,

30(23):3198–3200, 2005.

[38] V. M. Shalaev, W. Cai, U. K. Chettiar, H Yuan, A. K. Sarychev, V. P. Drachev, and

A. V. Kildishev. Negative index of refraction in optical metamaterials. Opt. Lett.,

30(24):3356–3358, 2005.

[39] G. Dolling, C. Enkrich, M. Wegener, C. M. Soukoulis, and S. Linden. Simultaneous

Negative Phase and Group Velocity of Light in a Metamaterial. Science, 312(5775):892–

894, 2006.

[40] G. Dolling, M. Wegener, C. M. Soukoulis, and S. Linden. Negative-index metamaterial

at 780 nm wavelength. Opt. Lett., 32(1):53–55, 2007.

 91

[41] J. Valentine, S. Zhang, T. Zentgraf, E. Ulin-Avila, D. A. Genov, G. Bartal, and

X. Zhang. Three-dimensional optical metamaterial with a negative refractive index.

Nature, 455:376–379, 2008.

[42] I. V. Lindell, S. A. Tretyakov, K. I. Nikoskinen, and S. Ilvonen. BW media - media with

negative parameters, capable of supporting backward waves. Microwave Opt. Technol.

Lett., 31:129–133, 2001.

[43] D. R. Smith and D. Schurig. Electromagnetic wave propagation in media with indefinite

permittivity and permeability tensors. Phys. Rev. Lett., 90(7):077405, Feb 2003.

[44] D. R. Smith, P. Kolinko, and D. Schurig. Negative refraction in indefinite media. J. Opt.

Soc. Am. B, 21(5):1032–1043, 2004.

[45] R. Wangberg, J. Elser, E. E. Narimanov, and V. A. Podolskiy. Nonmagnetic nanocom-

posites for optical and infrared negative-refractive-index media. J. Opt. Soc. Am. B,

23(3):498–505, 2006.

[46] M. G. Silveirinha, P. A. Belov, and C. R. Simovski. Subwavelength imaging at infrared

frequencies using an array of metallic nanorods. Phys. Rev. B, 75(3):035108, 2007.

[47] J. Yao, Z. Liu, Y. Liu, Y. Wang, C. Sun, G. Bartal, A. M. Stacy, and X. Zhang. Optical

Negative Refraction in Bulk Metamaterials of Nanowires. Science, 321(5891):930–, 2008.

[48] A. A. Zharov, I. V. Shadrivov, and Y. S. Kivshar. Nonlinear properties of left-handed

metamaterials. Phys. Rev. Lett., 91(3):037401, Jul 2003.

[49] I. V. Shadrivov, A. A. Zharov, and Y. S. Kivshar. Second-harmonic generation in non-

linear left-handed metamaterials. J. Opt. Soc. Am. B, 23(3):529–534, 2006.

[50] N. Lazarides, M. Eleftheriou, and G. P. Tsironis. Discrete breathers in nonlinear magnetic

metamaterials. Phys. Rev. Lett., 97(15):157406, 2006.

[51] C. M. Soukoulis, editor. Photonic Crystals and Light Localization in the 21st Century.

Springer, 2001.
 92

[52] W. M. Robertson and M. S. May. Surface electromagnetic wave excitation on one-

dimensional photonic band-gap arrays. Appl. Phys. Lett., 74(13):1800–1802, 1999.

[53] J. M. Elson and P. Tran. Coupled-mode calculation with the r-matrix propagator for the

dispersion of surface waves on a truncated photonic crystal. Phys. Rev. B, 54(3):1711–

1715, Jul 1996.

[54] F. Ramos-Mendieta and P. Halevi. Surface electromagnetic waves in two-dimensional

photonic crystals: Effect of the position of the surface plane. Phys. Rev. B, 59(23):15112–

15120, Jun 1999.

[55] W. M. Robertson, G Arjavalingam, R. D. Meade, K. D. Brommer, A. M. Rappe, and

J. D. Joannopoulos. Observation of surface photons on periodic dielectric arrays. Opt.

Lett., 18(7):528–530, 1993.

[56] Y. A. Vlasov, N. Moll, and S. J. McNab. Observation of surface states in a truncated

photoniccrystal slab. Opt. Lett., 29(18):2175–2177, 2004.

[57] R. Moussa, S. Foteinopoulou, Lei Zhang, G. Tuttle, K. Guven, E. Ozbay, and C. M.

Soukoulis. Negative refraction and superlens behavior in a two-dimensional photonic

crystal. Phys. Rev. B, 71(8):085106, 2005.

[58] P. Kramper, M. Agio, C. M. Soukoulis, A. Birner, F. Müller, R. B. Wehrspohn, U. Gösele,

and V. Sandoghdar. Highly directional emission from photonic crystal waveguides of

subwavelength width. Phys. Rev. Lett., 92(11):113903, Mar 2004.

[59] E. Moreno, F. J. Garcı́a-Vidal, and L. Martı́n-Moreno. Enhanced transmission and

beaming of light via photonic crystal surface modes. Phys. Rev. B, 69(12):121402, Mar

2004.

[60] I. Bulu, H. Caglayan, and E. Ozbay. Beaming of light and enhanced transmission via

surface modes of photonic crystals. Opt. Lett., 30(22):3078–3080, 2005.

 93

[61] K. M. Ho, C. T. Chan, and C. M. Soukoulis. Existence of a photonic gap in periodic

dielectric structures. Phys. Rev. Lett., 65(25):3152–3155, Dec 1990.

[62] F. Ramos-Mendieta and P. Halevi. Surface modes in a 2d array of square dielectric

cylinders. Solid State Commun., 100(5):311 – 314, 1996.

[63] B. Wang, W. Dai, A. Fang, L. Zhang, G. Tuttle, T. Koschny, and C. M. Soukoulis.

Surface waves in photonic crystal slabs. Phys. Rev B, 74(19):195104, 2006.

[64] H. Raether. Surface Plasmons on Smooth and Rough Surfaces and on Gratings. Springer,

1988.

[65] R. Moussa, B. Wang, G. Tuttle, Th. Koschny, and C. M. Soukoulis. Effect of beaming

and enhanced transmission in photonic crystals. Phys. Rev. B, 76(23):235417, 2007.

[66] E. Moreno, L. Martn-Moreno, and F.J. Garca-Vidal. Efficient coupling of light into and

out of a photonic crystal waveguide via surface modes. Photonics Nanostruct. Fundam.

Appl., 2(2):97 – 102, 2004.

[67] Steven K. Morrison and Yuri S. Kivshar. Engineering of directional emission from

photonic-crystal waveguides. Appl. Phys. Lett., 86(8):081110, 2005.

[68] A. I. Rahachou and I. V. Zozoulenko. Light propagation in finite and infinite photonic

crystals: The recursive green’s function technique. Phys. Rev. B, 72(15):155117, Oct

2005.

[69] A. Mekis and J. D. Joannopoulos. Tapered couplers for efficient interfacing between

dielectric and photonic crystal waveguides. J. Lightwave Technol., 19(6):861, 2001.

[70] D. Tang, L. Chen, and W. Ding. Efficient beaming from photonic crystal waveguides via

self-collimation effect. Appl. Phys. Lett., 89(13):131120, 2006.

[71] J. Bauer and S. John. Molding light flow from photonic band gap circuits to microstruc-

tured fibers. Appl. Phys. Lett., 90(26):261111, 2007.

 94

[72] K. Guven and E. Ozbay. Directivity enhancement and deflection of the beam emitted

from a photonic crystal waveguide via defect coupling. Opt. Express, 15(23):14973–14978,

2007.

[73] T. W. Ebbesen, H. J. Lezec, T. Thio H. F. Ghaemi1, and P. A. Wolff. Extraordinary

optical transmission through sub-wavelength hole arrays. Nature, 391:667–669, 1998.

[74] L. Martı́n-Moreno, F. J. Garcı́a-Vidal, H. J. Lezec, K. M. Pellerin, T. Thio, J. B. Pendry,

and T. W. Ebbesen. Theory of extraordinary optical transmission through subwavelength

hole arrays. Phys. Rev. Lett., 86(6):1114–1117, Feb 2001.

[75] L. Martı́n-Moreno, F. J. Garcı́a-Vidal, H. J. Lezec, A. Degiron, and T. W. Ebbesen.

Theory of highly directional emission from a single subwavelength aperture surrounded

by surface corrugations. Phys. Rev. Lett., 90(16):167401, Apr 2003.

[76] H. J. Lezec, A. Degiron, E. Devaux, R. A. Linke, L. Martin-Moreno, F. J. Garcia-

Vidal, and T. W. Ebbesen. Beaming Light from a Subwavelength Aperture. Science,

297(5582):820–822, 2002.

[77] R. Sambles. Photonics: More than transparent. Nature, 391:641–642, 1998.

[78] S. I. Bozhevolnyi, J. Erland, K. Leosson, P. M. W. Skovgaard, and J. M. Hvam. Waveg-

uiding in surface plasmon polariton band gap structures. Phys. Rev. Lett., 86(14):3008–

3011, Apr 2001.

[79] C. Genet and T. W. Ebbesen. Light in tiny holes. Nature, 445:39–46, 2007.

[80] J. P. Berenger. A perfectly matched layer for the absorption of electromagnetic waves.

J. Comput. Phys., 114(2):185 – 200, 1994.

[81] J.-P. Berenger. Perfectly matched layer for the fdtd solution of wave-structure interaction

problems. IEEE Trans. Antennas Propag., 44(1):110–117, Jan 1996.

 95

[82] W. Dai, B. Wang, Th. Koschny, and C. M. Soukoulis. Experimental verification of

quantized conductance for microwave frequencies in photonic crystal waveguides. Phys.

Rev. B, 78(7):073109, 2008.

[83] R. Landauer. Spatial variation of currents and fields due to localized scatterers in metallic

conduction. IBM, J. Res. Dev., 1(3):223, 1957.

[84] R. Landauer. Can a length of perfect conductor have a resistance? Phys. Lett. A,

85(2):91 – 93, 1981.

[85] E. N. Economou and C. M. Soukoulis. Static conductance and scaling theory of local-

ization in one dimension. Phys. Rev. Lett., 46(9):618–621, Mar 1981.

[86] R. Landauer. Spatial variation of currents and fields due to localized scatterers in metallic

conduction. IBM, J. Res. Dev., 32(3):306, 1988.

[87] A. D. Stone and A. Szafer. What is measured when you measure a resistance?the landauer

formula revisited. IBM, J. Res. Dev., 32(3):384, 1988.

[88] G. Timp, A. M. Chang, P. Mankiewich, R. Behringer, J. E. Cunningham, T. Y. Chang,

and R. E. Howard. Quantum transport in an electron-wave guide. Phys. Rev. Lett.,

59(6):732–735, Aug 1987.

[89] B. J. van Wees, H. van Houten, C. W. J. Beenakker, J. G. Williamson, L. P. Kouwen-

hoven, D. van der Marel, and C. T. Foxon. Quantized conductance of point contacts in

a two-dimensional electron gas. Phys. Rev. Lett., 60(9):848–850, Feb 1988.

[90] B. J. van Wees, L. P. Kouwenhoven, E. M. M. Willems, C. J. P. M. Harmans, J. E.

Mooij, H. van Houten, C. W. J. Beenakker, J. G. Williamson, and C. T. Foxon. Quantum

ballistic and adiabatic electron transport studied with quantum point contacts. Phys.

Rev. B, 43(15):12431–12453, May 1991.

[91] D. S. Fisher and P. A. Lee. Relation between conductivity and transmission matrix.

Phys. Rev. B, 23(12):6851–6854, Jun 1981.

 96

[92] M. Büttiker, Y. Imry, R. Landauer, and S. Pinhas. Generalized many-channel conduc-

tance formula with application to small rings. Phys. Rev. B, 31(10):6207–6215, May

1985.

[93] E. A. Montie, E. C. Cosman, G. W. ’t Hooft, M. B. van der Mark, and C. W. J.

Beenakker. Observation of the optical analogue of quantized conductance of a point

contact. Nature, 350:594–595, 1991.

[94] S. Albaladejo, J. J. Sáenz, M. Lester, L. S. Froufe-Pérez, and A. Garcı́a-Martı́n. Op-

tical conductance of waveguides built into finite photonic crystals. Appl. Phys. Lett.,

91(6):061107, 2007.

[95] L. C. Botten, A. A. Asatryan, N. A. Nicorovici, R. C. McPhedran, and C. Martijn

de Sterke. Generalisation of the transfer matrix formulation of the theory of electron

and photon conductance. Physica B: Condens. Matter, 394(2):320 – 324, 2007.

[96] I. V. Shadrivov, S. K. Morrison, and Y. S. Kivshar. Tunable split-ring resonators for

nonlinear negative-index metamaterials. Opt. Express, 14(20):9344–9349, 2006.

[97] D. A. Powell, I. V. Shadrivov, Y. S. Kivshar, and M. V. Gorkunov. Self-tuning mecha-

nisms of nonlinear split-ring resonators. Appl. Phys. Lett., 91(14):144107, 2007.

[98] GIL I., BONACHE J., GARCIA-GARCIA J., and MARTIN F. Tunable metamaterial

transmission lines based on varactor-loaded split-ring resonators. IEEE Trans. Microwave

Theory Tech., 54(11):2665–2674, 2007.

[99] D. A. Powell, I. V. Shadrivov, and Y. S. Kivshar. Multistability in nonlinear left-handed

transmission lines. Appl. Phys. Lett., 92(26):264104, 2008.

[100] E. Ozbay, K. Aydin, S. Butun, K. Kolodziejak, and D. Pawlak. Ferroelectric based

tuneable srr based metamaterial for microwave applications. Microwave Conference,

2007. European, pages 497–499, Oct. 2007.

 97

[101] A. Degiron, J. J. Mock, and D. R. Smith. Modulating and tuning the response of

metamaterials at the unit cell level. Opt. Express, 15(3):1115–1127, 2007.

[102] L. Kang, Q. Zhao, H. Zhao, and J. Zhou. Magnetically tunable negative permeabil-

ity metamaterial composed by split ring resonators and ferrite rods. Opt. Express,

16(12):8825–8834, 2008.

[103] D. Wang, L. Ran, H. Chen, M. Mu, J. A. Kong, and B. Wu. Active left-handed material

collaborated with microwave varactors. Appl. Phys. Lett., 91(16):164101, 2007.

[104] J. Carbonell, V. E. Boria, and D. Lippens. Nonlinear effects in split ring resonators loaded

with heterostructure barrier varactors. Microwave Opt. Technology Lett., 50(2):474–479,

2007.

[105] K. Aydin and E. Ozbay. Capacitor-loaded split ring resonators as tunable metamaterial

components. J. Appl. Phys., 101(2):024911, 2007.

[106] L. D. Landau and E. M. Lifshitz. Course of Theoretical Physics: Mechanics. Butterworth-

Heinemann; 3rd edition, 1982.

[107] I. V. Lindell, A. H. Sihvola, S. A. Tretyakov, and A. J. Viitanen. Electromagnetic Waves

in Chiral and Bi-isotropic Media. Artech House, 1994.

[108] J. Applequist. Optical activity: Biots bequest. American Scientist, 75(1):58–68, 1987.

[109] L. D. Barron. Molecular Light Scattering and Optical Activity. Cambridge, 1982.

[110] A. Serdyukov, I. Semchenko, S. Tretyakov, and A. Sihvola. Electromagnetics of Bi-

anisotropic Materials: Theory and Applications. Gordon and Breach Science, 2001.

[111] S. Tretyakov, I. Nefedov, A. Sihvola, S. Maslovski, and C. Simovski. Waves and energy

in chiral nihility. J. Electromagn. Waves Appl., 17(5):695–706, 2003.

[112] D. R. Smith, S. Schultz, P. Markoš, and C. M. Soukoulis. Determination of effective per-

mittivity and permeability of metamaterials from reflection and transmission coefficients.

Phys. Rev. B, 65(19):195104, Apr 2002.

 98

[113] D. R. Smith, D. C. Vier, T. Koschny, and C. M. Soukoulis. Electromagnetic parameter

retrieval from inhomogeneous metamaterials. Phys. Rev. E, 71(3):036617, 2005.

[114] T. Koschny, P. Markoš, E. N. Economou, D. R. Smith, D. C. Vier, and C. M. Soukoulis.

Impact of inherent periodic structure on effective medium description of left-handed and

related metamaterials. Phys. Rev. B, 71(24):245105, Jun 2005.

[115] D. Kwon, D. H. Werner, A. V. Kildishev, and V. M. Shalaev. Material parameter

retrieval procedure for general bi-isotropic metamaterials and its application to optical

chiral negative-index metamaterial design. Opt. Express, 16(16):11822–11829, 2008.

[116] M C K Wiltshire, J B Pendry, and J V Hajnal. Chiral swiss rolls show a negative

refractive index. J. Phys.: Condens. Matter, 21(29):292201 (5pp), 2009.

[117] C. Monzon and D. W. Forester. Negative refraction and focusing of circularly polarized

waves in optically active media. Phys. Rev. Lett., 95(12):123904, 2005.

[118] Semchenko, I. V., Khakhomov, S. A., and Tretyakov, S. A. Chiral metamaterial with

unit negative refraction index. Eur. Phys. J. Appl. Phys., 46:20607, 2008.

[119] Q. Cheng and T. J. Cui. Negative refractions in uniaxially anisotropic chiral media. Phy.

Rev. B, 73(11):113104, 2006.

[120] C. Zhang and T. J. Cui. Negative reflections of electromagnetic waves in a strong chiral

medium. Appl. Phys. Lett., 91(19):194101, 2007.

[121] Y. Tamayama, T. Nakanishi, K. Sugiyama, and M. Kitano. An invisible medium for

circularly polarized electromagnetic waves. Opt. Express, 16(25):20869–20875, 2008.

[122] C. Qiu, H. Yao, L. Li, S. Zouhdi, and T. Yeo. Backward waves in magnetoelectri-

cally chiral media: Propagation, impedance, and negative refraction. Phys. Rev. B,

75(15):155120, 2007.

[123] C. Qiu, H. Yao, L. Li, S. Zouhdi, and T. Yeo. Routes to left-handed materials by

magnetoelectric couplings. Phys. Rev. B, 75(24):245214, 2007.

 99

[124] A. Papakostas, A. Potts, D. M. Bagnall, S. L. Prosvirnin, H. J. Coles, and N. I. Zheludev.

Optical manifestations of planar chirality. Phys. Rev. Lett., 90(10):107404, 2003.

[125] A. V. Rogacheva, V. A. Fedotov, A. S. Schwanecke, and N. I. Zheludev. Giant gyrotropy

due to electromagnetic-field coupling in a bilayered chiral structure. Phys. Rev. Lett.,

97(17):177401, 2006.

[126] B. Bai, Y. Svirko, J. Turunen, and T. Vallius. Optical activity in planar chiral metama-

terials: Theoretical study. Phys. Rev. A, 76(2):023811, 2007.

[127] M. Kuwata-Gonokami, N. Saito, Y. Ino, M. Kauranen, K. Jefimovs, T. Vallius, J. Tu-

runen, and Y. Svirko. Giant optical activity in quasi-two-dimensional planar nanostruc-

tures. Phys. Rev. Lett., 95(22):227401, 2005.

[128] D. Kwon, P. L. Werner, and D. H. Werner. Optical planar chiral metamaterial designs for

strong circular dichroism and polarization rotation. Opt. Express, 16(16):11802–11807,

2008.

[129] J. Zhou, L. Zhang, G. Tuttle, T. Koschny, and C. M. Soukoulis. Negative index materials

using simple short wire pairs. Phys. Rev. B, 73(4):041101, 2006.

[130] B. Wang, J. Zhou, T. Koschny, and C. M. Soukoulis. Nonplanar chiral metamaterials

with negative index. Appl. Phys. Lett., 94(15):151112, 2009.

[131] J. D. Baena, L. Jelinek, and R. Marqués. Towards a systematic design of isotropic

bulk magnetic metamaterials using the cubic point groups of symmetry. Phys. Rev B,

76(24):245115, 2007.

[132] R. Marqués, L. Jelinek, and F. Mesa. Negative refraction from balanced quasi-planar

chiral inclusions. Microwave Opt. Technology Lett., 49(10):2606–2609, 2007.

[133] J. Zhou, T. Koschny, and C. M. Soukoulis. An efficient way to reduce losses of left-handed

metamaterials. Opt. Express, 16(15):11147–11152, 2008.

 100

[134] M. Wegener, J. L. Garcı́a-Pomar, C. M. Soukoulis, N. Meinzer, M. Ruther, and S. Linden.

Toy model for plasmonic metamaterial resonances coupled to two-level system gain. Opt.

Express, 16(24):19785–19798, 2008.

[135] N. I. Landy, S. Sajuyigbe, J. J. Mock, D. R. Smith, and W. J. Padilla. Perfect metama-

terial absorber. Phys. Rev. Lett., 100(20):207402, 2008.

[136] H. Tao, N. I. Landy, C. M. Bingham, X. Zhang, R. D. Averitt, and W. J. Padilla. A

metamaterial absorber for the terahertz regime: design, fabrication and characterization.

Opt. Express, 16(10):7181–7188, 2008.

[137] H. Tao, C. M. Bingham, A. C. Strikwerda, D. Pilon, D. Shrekenhamer, N. I. Landy,

K. Fan, X. Zhang, W. J. Padilla, and R. D. Averitt. Highly flexible wide angle of

incidence terahertz metamaterial absorber: Design, fabrication, and characterization.

Phys. Rev. B, 78(24):241103, 2008.

[138] Y. Avitzour, Y. A. Urzhumov, and G. Shvets. Wide-angle infrared absorber based on a

negative-index plasmonic metamaterial. Phys. Rev. B, 79(4):045131, 2009.

[139] M. Diem, T. Koschny, and C. M. Soukoulis. Wide-angle perfect absorber/thermal emitter

in the terahertz regime. Phys. Rev. B, 79(3):033101, 2009.

[140] B. Wang, T. Koschny, and C. M. Soukoulis. Wide-angle and polarization-independent

chiral metamaterial absorber. Phys. Rev. B, 80(3):033108, 2009.

[141] D. L. Jaggard and N. Engheta. Chiralsorb as an invisible medium. Electronics Lett.,

25:173, 1989.

[142] J. Monzon. Comment on “chiralsorb as an invisible medium”. Electronics Lett., 25:1060,

1989.

[143] C. F. Bohren, R. Luebbers, H. S. Langdon, and F. Hunsberger. Microwave absorbing

chiral composites: Is chirality essential or accidental? Appl. Opt., 31(30):6403, 1992.

 101

[144] J. H. Cloete, M. Bingle, and D. B. Davidson. The role of chirality and resonance in

synthetic microwave absorbers. Int. J. Electron. Commun., 55(4):233, 2001.