ACOUSTIC METAMATERIAL DESIGN AND APPLICATIONS

BY

SHU ZHANG

DISSERTATION

Submitted in partial fulfillment of the requirements

for the degree of Doctor of Philosophy in Mechanical Engineering
in the Graduate College of the
University of Illinois at Urbana-Champaign, 2010

Urbana, Illinois

Doctoral Committee:

Assistant Professor Nicholas X. Fang, Chair and Director of Research

Professor Jianming Jin
Associate Professor Gustavo Gioia
Associate Professor Harley T. Johnson
 ABSTRACT

The explosion of interest in metamaterials is due to the dramatically increased manipulation

ability over light as well as sound waves. This material research was stimulated by the

opportunity to develop an artificial media with negative refractive index and the application in

superlens which allows super-resolution imaging. High-resolution acoustic imaging techniques

are the essential tools for nondestructive testing and medical screening. However, the spatial

resolution of the conventional acoustic imaging methods is restricted by the incident wavelength

of ultrasound. This is due to the quickly fading evanescent fields which carry the subwavelength

features of objects. By focusing the propagating wave and recovering the evanescent field, a flat

lens with negative-index can potentially overcome the diffraction limit. We present the first

experimental demonstration of focusing ultrasound waves through a flat acoustic metamaterial

lens composed of a planar network of subwavelength Helmholtz resonators. We observed a tight

focus of half-wavelength in width at 60.5 KHz by imaging a point source. This result is in

excellent agreement with the numerical simulation by transmission line model in which we

derived the effective mass density and compressibility. This metamaterial lens also displays

variable focal length at different frequencies. Our experiment shows the promise of designing

compact and light-weight ultrasound imaging elements.

ii
 Moreover, the concept of metamaterial extends far beyond negative refraction, rather giving

enormous choice of material parameters for different applications. One of the most interesting

examples these years is the invisible cloak. Such a device is proposed to render the hidden object

undetectable under the flow of light or sound, by guiding and controlling the wave path through

an engineered space surrounding the object. However, the cloak designed by transformation

optics usually calls for a highly anisotropic metamaterial, which make the experimental studies

remain challenging. We present here the first practical realization of a low-loss and broadband

acoustic cloak for underwater ultrasound. This metamaterial cloak is constructed with a network

of acoustic circuit elements, namely serial inductors and shunt capacitors. Our experiment clearly

shows that the acoustic cloak can effectively bend the ultrasound waves around the hidden

object, with reduced scattering and shadow. Due to the non-resonant nature of the building

elements, this low loss (~6dB/m) cylindrical cloak exhibits excellent invisibility over a broad

frequency range from 52 to 64 kHz in the measurements. The low visibility of the cloaked object

for underwater ultrasound shed a light on the fundamental understanding of manipulation,

storage and control of acoustic waves. Furthermore, our experimental study indicates that this

design approach should be scalable to different acoustic frequencies and offers the possibility for

a variety of devices based on coordinate transformation.

iii
 ACKNOWLEDGMENTS

I would like to thank my advisor, Dr. Nicholas Fang, for providing me the wonderful opportunity

to finish my PhD degree and work on these exciting projects. His unwavering support, invaluable

guidance and suggestions in exploration this research and presenting the thesis are greatly

appreciated.

At the same time, many thanks to all the committee members, Dr. Jianming Jin, Dr. Gustavo

Gioia and Dr. Harley T. Johnson, for their invaluable suggestions and help.

Special thank to Dr. Leilei Yin for his great help on experiment techniques. I am very

grateful to everybody in our research group: Pratik Chaturvedi, Jun Xu, Keng Hao Hsu,Kin

Hung Fung, Tarun Malik, Anil Kumar, Howon Lee and Hyungjin Ma. I truly learned the most

from them. They were the first people I turned to for help with a challenging problem.

Finally, I would like to thank my parents for their constant encouragement and belief in me

during this course. I am greatly indebted to Chunguang Xia for his special efforts in spending

countless hours helping me. His support made the last few years of hard work possible. I would

also like to extend my thanks to all my friends who kept me in good spirits during my stay here.

iv
 Table of Contents
1. INTRODUCTION.................................................................................................................... 1
1.1. Metamaterial..................................................................................................................... 1
1.2. Thesis Organization.......................................................................................................... 4
2 ACOUSTIC TRANSMISSION LINE ..................................................................................... 9
2.1 Introduction ...................................................................................................................... 9
2.2 Locally Resonant Sonic Materials.................................................................................. 10
2.3 Acoustic Circuits ............................................................................................................ 11
2.4 Reflection and Transmission .......................................................................................... 15
2.5 Absorption and Attenuation of Sound in Pipe ................................................................ 23
2.6 Acoustic Isotropic Metamaterial .................................................................................... 36
2.7 Anisotropic Acoustic Metamaterial ................................................................................ 42
3 ULTRASOUND FOCUSING USING ACOUSTICMETAMATERIAL NETWORK .......... 51
3.1 Introduction .................................................................................................................... 51
3.2 Negative Refractive Index Lens ..................................................................................... 53
3.3 Phononic Crystal ............................................................................................................ 55
3.4 Ultrasound Focusing by Acoustic Transmission Line Network ..................................... 59
4 BROADBAND ACOUSTIC CLOAK FOR ULTRASOUND WAVES ................................ 86
4.1 Introduction .................................................................................................................... 86
4.2 Optical Transformation .................................................................................................. 88
4.3 Acoustic Cloak ............................................................................................................... 95
4.4 Numerical Simulation of Acoustic Cloak Based on Transmission Line Model ............. 98
4.5 Irregular Transmission Line Network .......................................................................... 103
4.6 Experimental Study of Acoustic Cloak Based on Transmission Line Model .............. 110
5 SUMMARY AND FUTURE WORK .................................................................................. 130
5.1 Summary ...................................................................................................................... 130
5.2 Future Work .................................................................................................................. 131
APPENDIX A: LUMPED CIRCUIT MODEL .......................................................................... 133
APPENDIX B: FRESNEL LENS DESIGN BY ACOUSTIC TRANSMISSION LINE ........... 140
APPENDIX C: NEGATIVE INDEX LENS BASED ON METAL-INSULATOR-METAL (MIM)
WAVEGUIDES ........................................................................................................................... 148
APPENDIX D: SCATTERING FIELDS FROM THE CLOAK ................................................ 161
APPENDIX E: EXPERIMENTAL SETUP AND DATA ACQUISITION ................................. 174
APPENDIX F: CICUIT MODELING ........................................................................................ 179

v
 1 INTRODUCTION

1.1 Metamaterial

Over the past eight years, metamaterials have shown tremendous potential in many disciplines of

science and technology. The explosion of interest in metamaterials is due to the dramatically

increased manipulation ability over light as well as sound waves which are not available in

nature. The core concept of metamaterial is to replace the molecules with man-made structures,

viewed as “artificial atoms” on a scale much less than the relevant wavelength. In this way, the

metamaterial can be described using a small number of effective parameters. In late 1960s, the

concept of metamaterial was first proposed by Veselago for electromagnetic wave1. He predicted

that a medium with simultaneous negative permittivity and negative permeability were shown to

have a negative refractive index. But this negative index medium remained as an academic

curiosity for almost thirty years, until Pendry et al2,3 proposed the designs of artificial structured

materials which would have effectively negative permeability and permittivity. The negative

refractive index was first experimentally demonstrated at GHz frequency. 4,5

It is undoubtedly of interest whether we can design metamaterial for the wave in other

systems, for example, acoustic wave. The two waves are certainly different. Acoustic wave is

longitudinal wave; the parameters used to describe the wave are pressure and particle velocity. In

1
electromagnetism (EM), both electric and magnetic fields are transverse wave. However, the two

wave systems have the common physical concepts as wavevector, wave impedance, and power

flow. Moreover, in a two-dimensional (2D) case, when there is only one polarization mode, the

electromagnetic wave has scalar wave formulation. Therefore, the two sets of equations for

acoustic and electromagnetic waves in isotropic media are dual of each other by the replacement

as shown in Table 1.1 and this isomorphism holds for anisotropic medium as well. Table1.1

presents the analogy between acoustic and transverse magnetic field in 2D under harmonic

excitation. From this equivalence, the desirable effective density and compressibility need to be

established by structured material to realize exotic sound wave properties. The optical and

acoustic metamaterial share many similar implementation approaches as well.

The first acoustic metamaterial, also called as locally resonant sonic materials was

6
demonstrated with negative effective dynamic density. The effective parameters can be

ascribed to this material since the unit cell is sub-wavelength size at the resonance frequency.

Furthermore, by combining two types of resonant structural, acoustic metamaterial with

simultaneous negative bulk modulus and negative mass density was numerically demonstrated.7

Recently, Fang et al.8 proposed a new class of acoustic metamaterial which consists of a 1D

array of Helmholtz resonators which exhibits dynamic effective negative modulus in experiment.

2
Table 1.1 Analogy between acoustic and electromagnetic variables and material characteristics

Acoustics Electromagnetism (TMz) Analogy

∂P ∂E z
= −iωρ xu x = −iωμ y H y
∂x ∂x

∂P ∂Ez
= −iωρ y u y = iωμ x H x
∂y ∂y

∂u x ∂u y ∂H y ∂H x
+ = −iωβP − = −iωε z Ez
∂x ∂y ∂x ∂y

Acoustic pressure P Electric field Ez − Ez ↔ P

Particle velocity u x u y Magnetic field H x , H y H y ↔ −ux H x ↔ u y

Dynamic density ρ x ρ y Permeability μ x μ y ρx ↔ μ y ρ y ↔ μx

Dynamic compressibility β Permittivity ε z εz ↔ β

The concept of metamaterial extends far beyond negative refraction, rather giving enormous

choice of material parameters for different applications. One of the most interesting examples is

the invisible cloak by transformation optics. 9, 10. A simpler version of two-dimensional (2D)

11
cloak was implemented at microwave frequency. Later on, this new design paradigm is

extended theoretically to make an “inaudible cloak” for sound wave. 12,13,14,15 The sound wave is

directed to flow over a shielded object like water flowing around a rock. Because the waves

reform their original conformation after passing such a shielded object, the object becomes

3
invisible to the detector. The cloak for surface wave is also proposed in hertz frequency range. 16

Since the inception of the term metamaterials, acoustic metamaterials have being explored

theoretically but there has been little headway on the experimental front. The development of

acoustic metamaterial will yield new insight in material science and offer great opportunities for

several applications.

1.2 Thesis Organization

The central theme of this thesis is to design and characterize the acoustic metamaterial for

potential application in ultrasound imaging and sound controlling. This dissertation is organized

into four chapters. Besides the current chapter which intends to give a brief introduction of the

acoustic metamaterial and the motivations of this dissertation, the other three chapters organized

as following:

The second chapter describes the approach to build an acoustic metamaterial based on the

transmission line model. The basic concept and derivation of lumped acoustic circuit is

introduced. In the lumped circuit model, the aluminum is assumed as acoustically rigid,

considering the acoustic impedance of aluminum is around eleven times of that of water. A

more careful analysis including elasticity of the solid suggested that at low frequency the

majority of acoustic energy can be predominantly confined in the fluid, when such an excitation
4
originates from the liquid. 17 On the other hand, aluminum does participate in the wave

propagation, and may increase the loss.18 Moreover, the loss and limitation of current lumped

circuit model are discussed.

The third chapter deals with one of the most promising application of acoustic metamaterials,

obtaining a negative refractive index lens which can possibly overcome the diffraction limit. An

acoustic system is simulated by the analogous lumped circuit model in which the behavior of the

current resembles the motion of the fluid. Based on this lumped network, an acoustic negative

index lens is implemented by a two-dimensional (2D) array of subwavelength Helmholtz

resonators. The experimental studies are presented, demonstrating the focusing of ultrasound

waves through the negative index lens.

The fourth chapter is to construct an anisotropic cloak for sound waves at kHz range.

Relying on the flexibility of the transmission line approach, an acoustic metamaterial exhibiting

effective anisotropic density and bulk modulus is proposed to construct the cloak. An object can

be shielded inside the cloak and thus becomes invisible to the detector. Given the simulation

results, the sound-shielding capability is explored experimentally by measuring the scattered

pressure field.

5
References

1
Veselago,V. G.,“The electrodynamics of substances with simultaneously negative values of µ

and ε,” Sov. Phys. Usp., Vol. 10, No. 4,509(1968)

2
J. B. Pendry, A. J. Holden, D. J. Robbins, W. J. Stewart. “Magnetism from conductors and

enhanced nonlinear phenomena.” IEEE Trans. Microwave Theory Tech. 47, 2075–2084 (1999).

3
J. B. Pendry, A. J. Holden, W. J. Stewart and I. Youngs, “Extremely low frequency plasmons in

metallic mesostructures.” Phys. Rev. Lett. 76, 4773–4776 (1996).

4
Smith, D. R. et al. “Composite medium with simultaneously negative permeability and

permittivity”. Phys. Rev. Lett. 84, 4184–4187 (2003).

5
Shelby, R. A., Smith, D. R. & Schultz, S., “Experimental verification of a negative index of

refraction”, Science 292, 77–79 (2001).

6
Z. Liu, X. Zhang, Y. Mao, Y. Y. Zhu, Z. Yang, C. T. Chan, and P. Sheng, “Locally Resonant

6
Sonic Materials”, Science 289, 1734 (2000).

7
Y. Ding, Z. Liu, C. Qiu, and J. Shi, “Metamaterial with Simultaneously Negative Bulk Modulus

and Mass Density”, Phys. Rev. Lett. 99, 093904 (2007).

8
N. Fang, D. J. Xi, J. Y. Xu, M. Ambrati, W. Sprituravanich, C. Sun, and X. Zhang, “Ultrasonic

Metamaterials with Negative Modulus”, Nat. Mater. 5, 452 (2006).

9
J. B. Pendry, D. Schurig, D. R. Smith,” Controlling Electromagnetic Fields”, Science 312, 1780

(2006)

10
U. Leonhardt, “Optical Conformal Mapping”, Science 312, 1777 (2006).

11
D. Schurig et al, “Metamaterial Electromagnetic Cloak at Microwave Frequencies.”, Science

314, 977-980(2006)

12
Milton G W, Briane M and Willis J R, “On cloaking for elasticity and physical equations with

7
a transformation invariant form”, New J. Phys. 8, 248 (2006)

13
S. A. Cummer and D. Schurig, “One path to acoustic cloaking”, New J. Phys. 9, 45 (2007).

14
Torrent D and Sánchez-Dehesa J, “Anisotropic mass density by two dimensional acoustic

metamaterials”, New J. Phys. 10 023004 (2008)

15
Daniel Torrent and José Sánchez-Dehesa1, “Acoustic cloaking in two dimensions: a feasible

approach”, New J. Phys. 10 063015 (2008)

16
M. Farhat et al, “Broadband Cylindrical Acoustic Cloak for Linear Surface Waves in a Fluid”,

Phy.Rev.Lett 101, 134501 (2008)

17
C.R. Fuller and F. J. fahy, Journal of Sound and Vibration 81(4), 501-518, (1982)

18
H. LAMB, Manchester Literary and Philosophical Society-Memoirs and Proceedings, 42(9),

(1898)

8
 2 ACOUSTIC TRANSMISSION LINE

2.1 Introduction

Recently, there is a new research field that is known under the generic term of metamaterials.

Metamaterial refers to materials “beyond” conventional materials, which dramatically increased

our ability to challenge our physical perception and intuition. The exponential growth in the

number of publications in this area has shown exceptionally promising to provide fruitful new

theoretical concepts and potentials for valuable applications.

The physical properties of conventional materials are determined by the individual atoms

and molecules from which they are composed. There are typically billions of molecules

contained in one cubic wavelength of matter. The macroscopic wave fields, either

electromagnetic or acoustic wave, are averages over the fluctuating local fields at individual

atoms and molecules. Metamaterials extend this concept by replacing the molecules with

man-made structures, viewed as “artificial atoms” on a scale much less than the relevant

wavelength. In this way the metamaterial properties described using effective parameters are

engineered through structure rather than through chemical composition.1 The restriction that the

size and spacing of this structure be on a scale smaller than the wavelength distinguishes

metamaterials from photonic/phononic crystals. Photonic/phononic crystal is another different

class of artificial material with periodic structure on the same scale as the wavelength. Therefore

photonic/phononic crystals usually have a complex response to wave radiation that cannot be

9
simply described by effective parameters .However, the structural elements which make up a

metamaterial is not necessary periodic.

Metamaterial with negative refractive index and the application in superlens has initiated the

beginning of this material research. Early in late 1960s, metamaterial was first proposed by

Veselago for electromagnetic wave.2 He predicted that a medium with simultaneous negative

permittivity and negative permeability were shown to have a negative refractive index. The first

experimental demonstration of metamaterial with negative refractive index is reported at

3,4
microwave frequency. This metamaterial composed of a cubic lattice of artificial meta-atoms

with split ring resonators and metallic wires. However, metamaterial with negative index is not

the only possibility. Most recent developments explore new realms of anisotropic metamaterial

that can produce novel phenomena such as invisibility5,6,7 and hyperlens. 8,9
Moveover, it is of

great interest to extend the metamaterial concept to other classical waves, such as acoustic

wave.10,11,12 Since the analogy between light and sound waves, the electromagnetic and acoustic

metamaterials have been sharing much same design freedom while there has been less headway

on the experimental front of acoustic wave.

2.2 Locally Resonant Sonic Materials

Locally resonant sonic materials, which are a major step towards acoustic metamaterial, are

designed by including a resonant unit into the building block of phononic crystal. The key

difference between this sonic material and phononic crystal is that the individual unit cell is in

deep-subwavelength range compared with the resonant frequency, enabling effective properties
10
as mass density and bulk modulus to be assigned to this material. Although the static elastic

modulus and density need to be positive to maintain stable structure, these dynamic effective

acoustic properties are dispersive in nature and can turn negative at resonance. When the

resonance-induced scattered field prevails over the incident fields in background, the volume

change can be out of phase with applied dynamic pressure, implying negative bulk modulus. On

the other hand, the acceleration can be out of phase with the dynamic pressure gradient, showing

negative mass density effect. Liu 13 experimentally demonstrated the localized resonance

structure by coating heavy spheres with soft silicon rubber and encasing the coated spheres in

epoxy .Negative effective density was obtained due to a dipolar resonance at low sonic

frequency. Those anomalous phenomena resulted from strong coupling of the traveling elastic

wave in the host medium with the localized resonance rather than Bragg scattering. In the long

wavelength limit, the effective medium approach can be employed to offer a good estimation and

give an intuitive understanding of this complex system.14,15,16 It was demonstrated that an

acoustic metamaterial can possess simultaneous negative bulk modulus and mass density by

combining two types of structural units. While the monopolar resonances give rise to the

negative bulk modulus, the dipolar resonances lead to the negative mass density.17

2.3 Acoustic Circuits

A close analogy can be established between the propagation of sound in pipes or chambers and

electrical circuits. When the dimensions of the region in which the sound propagates are much

11
smaller than the wavelength, a lumped-parameter model is appropriate. The essential thing here

is that the phase is roughly constant throughout the system. (Appendix F)

2.3.1 Acoustic Impedance of a Pipe

Z0 Zl

x=0
Figure 2-1 A tube with rigid side wall terminated with acoustic impedance Zl

Assume a hollow cylindrical tube, open at one end and close another end with impedance Zl .

The origin of coordinates is chosen to be coinciding with the position of the open end of the tube.

We shall assume the diameter of the tube is sufficiently small so that the waves travel down the

tube with plane wave fronts. In order to make this true, the ratio of the wavelength of the sound

wave to the diameter of the tube must be greater than about 6. If an initial wave traveling in the

positive x direction pai, when the wave propagates at point x=l, a reflected wave traveling in the

negative x direction will in general be produced par, the corresponding particle velocity can be

written as
(2-1)
(2-2)

Where ,

The total pressure in the tube at any point is

12
 (2-3)
The total particle velocity is

= (2-4)

So the general expression for the acoustic impedance includes the reflected wave is

(2-5)

So we know the impedance at x=0, l as

(2-6)

(2-7)

Combine (2-6) and (2-7), we can express the impedance at the open end x=0 as a function of the

impedance ,18

(2-8)

2.3.2 Acoustic Inductance

Consider the water in a tube of length l and area S. Assume the tube is acoustically rigid and

open on both ends. Since all quantities are in phase when the dimension of the tube is much

smaller than the corresponding wavelength, it moves as a whole with displacement under the

action of an unbalanced force. The whole part moves without appreciable compression because

of the open ends.

Substitute (2-8) with ZAl=0,19

tan (2-9)

13
Since is much ssmaller thann wavelenggth, is a veryy small valuue, the tanggent can bee

replaced
d by the Tayylor series form,
f

t
tan (2-10))

Substituute (2-10) innto (2-9) yieelds

+… (2-11))

When , we can
c keep onnly the firstt term and nneglect the higer orderr terms witthin about 5

per centt error. So we

w can definne the acousstic inductance for an open
o end tubbe as

LA CA

(a) (b)
Figure 2-2 A pippe with (a) openo end annd (b) rigid end is anallogous to ann acoustic in
nductor and
d
acousticc capacitor respectivelyy

(2-12 )

As a ressult of the rradiation im

mpedance, thhe l in (2-122) should bee replaced byy an effectivve length of

′
the tubee added by a correctionn factor. 0.85 , where iss the radius of the tube.

2.3.3 Acoustic
A Caapacitance

If the tuube is rigidly closed at one end, suubstitute ∞ in (22-8) 19

14
 cot (2-13 )

For small value of , the cotangent can be replaced by the equivalent-series form

cot (2-14 )

Equation (2-13) becomes

(2-15 )

is valid within 5 percent for l up to series as a combination of an acoustic inductance and

capacitance. Furthermore, if the second term is small enough, we may neglect it, such that the

impedance of the cavity can be expressed as an acoustic capacitance.

(2-16 )

2.3.4 Helmholtz Resonator

A typical Helmholtz resonator as in Figure 2-3 can be presented as a series of inductance and

capacitance. The fluid inside the cavity is much easier to be compressed compared with that in

the neck part. Moreover, the pressure gradient along the open neck is much greater than that

inside the large cavity. Therefore the cavity displays capacitive property and leaves the smaller

neck as an acoustic inductor.

2.4 Reflection and Transmission

When an acoustic wave traveling in one medium encounters the boundary of a second medium,

reflected and transmitted waves are generated. For normal incidence, solids obey the same

equations developed for fluids, which is greatly simplified. The only modification needed is that

15
the speed of sound in the solid must be the bulk speed of sound, relying on both bulk and shear

module. The characteristic acoustic impedances and speeds of sound in two media and the angle

of incident wave determine the ratios of the pressure amplitudes and intensities of the reflected

and transmitted waves to those of the incident wave. For fluids, the characteristic acoustic

impedance is defined as .

Figure 2-3 Helmholtz resonator

2.4.1 Normal Incidence

Figure 2-4 Normal incidence

16
Let the boundary 0 be the boundary between two fluids with characteristic acoustic

impedance and . A plane wave traveling in the direction,

(2-17 )

when the incident wave strikes the boundary, generates a reflected wave and a transmitted wave

(2-18)

(2-19)

Where , , is the angular frequency and , are the speed of sound.

The particle velocities are

(2-20 )

(2-21)

(2-22)

, , (2-23)

The boundary conditions require the continuity of pressure and the normal component of the

particle velocities must be equal at both sides of the boundary. The first condition implies that

there is no net force on the boundary plane separating the media. The continuity of the normal

component of velocity requires that the media remain in contact.

So at 0

(2-24)

(2-25)

17
Substitute (2-17)-(2-23) into (2-24) and (2-25), we can obtain the reflection and transmission

coefficients.

(2-26)

(2-27)

(2-28)

(2-29)

Where ,

The acoustic intensity of a harmonic plane progressive wave is defined as . The intensity

reflection and transmission coefficients are calculated.

(2-30)

1 (2-31)

Figure 2-5 Oblique incidence

18
In the limit when , 1, 1, 2, 0 .The wave is reflected with

amplitude equal to the incident wave and no change in phase. The transmitted wave has pressure

amplitude twice that of the incident wave. The normal particle velocity of the reflected wave is

equal to but 180 out of phase with that of the incident wave. Therefore the total normal

particle velocity is zero at the boundary. The boundary with is termed rigid. In fact,

such total reflection caused standing wave pattern in medium 1 and the boundary is the node for

the particle velocity and antinode for the pressure. While there is no acoustic wave propagates in

medium 2 since the particle velocity is zero and the pressure is static force. Given one example,

the density and speed of sound of aluminum are 2700kg/m3 and 6420m/s. While the density and

speed of sound of water is 1000 kg/m3 and 1500m/s. So the acoustic impedance of aluminum is

around 12 times of that of water. Therefore when acoustic wave travels through water in an

aluminum tube, the boundary can be assumed as rigid.

2.4.2 Oblique Incidence

Assume that the incident, reflected and transmitted waves make the respective angles , , .

(2-32)

(2-33)

(2-34)

(2-35)

(2-36)

19
 (2-37)

Continuity of pressure and normal component of particle velocity at 0 yields

(2-38)

(2-39)

Since (2-39) must be

true for all y, this means

sin sin (2-40)

(2-41)

Equation (2-41) is the statement of Snell’s law

So (2-39) can be further simplified as

(2-42)

(2-43)

So the reflection and transmission coefficients are

(2-44)

(2-45)

Where ,

Where the Snell’s law reveals

/
cos 1 sin 1 sin (2-46)

If and , define sin / ,

20
 /
cos sin 1 (2-47)

cos becomes pure imaginary. The transmitted pressure is

(2-48)
/
sin 1 (2-49)

The transmitted pressure field decays perpendicular to the boundary and propagates in the y

direction, parallel to the boundary. For incident angle greater than the critical angle, the incident

wave is totally reflected and in the steady state, no energy propagates away from the boundary

into the second medium. Even though the transmitted wave possessed energy, but it propagates

parallel to the boundary. As an example, the speed of sound in aluminum is greater than the one

in water. As a result, as a plan wave propagates in water inside an aluminum tube and the

incident angle in the solid/fluid interface is almost 90o, the wave is totally reflected from the

aluminum and confined inside the water.

2.4.3 Reflection from the Surface of Solid

Define the normal specific acoustic impedance as

(2-50)
·

Where is the unit vector perpendicular to the interface.

So the pressure reflection coefficient can be written as

(2-51)

21
Solids can support two types of elastic waves: longitudinal and shear. If the transverse

dimensions of an isotropic solid are much larger than the wavelength of the acoustic wave, the

appropriate phase speed for the longitudinal waves is

(2-52)

Where and are bulk and shear modulus of the solids and is the density.

For the case of normal incidence, the transmission and reflection coefficient between solid

and fluid is the same as those with two fluids interface. However, when plane wave obliquely

incident on the surface of a solid, the wave transmitted into the solids might be refracted in three

different cases. The wave may propagate along the surface of the solids. Another possibility is

that the wave can propagate in a manner similar to two-fluid interface. Moreover, the wave may

be converted into two waves, a longitude wave and a transverse wave.

For most solids, the normal specific acoustic impedance has two parts, resistance and

reactance, respectively. . The pressure reflection coefficient can be revised as

(2-53)

This means the reflected wave at the boundary may either lead or lag the incident wave by

certain angle. When 90 , approaches unity.

22
2.5 Absorption and Attenuation of Sound in Pipe

The previous sections are under the assumption that all losses of acoustic energy could be

neglected .There are two kinds of loss in acoustic wave .The first source is associated with the

boundary conditions and the other type of loss is intrinsic to the medium. The losses in the

20
medium can be further subdivided into three basic types: viscous losses, heat conduction

losses and losses associated with internal molecular processes. Viscous losses occur when there

is relative motion between adjacent portions of the medium. Heat conduction losses are caused

by the conduction of thermal energy from high temperature condensations to lower temperature

rarefactions. Losses resulted from molecular processes is by converting kinetic energy of the

molecular into stored potential energy ,rotational and vibration energies and energies of

association and dissociation between different ionic species and complexes in ionized solutions.

On the other hand, the loss due to the boundary is more significant when the volume of the fluid

is small in comparison with the area of the walls, as when the pipe is narrow. The acoustic

velocity amplitude increases from zero at the wall to the maximum value in the center of the

pipe. Therefore there exit dissipative forces due to the shearing viscosity of the fluid. In addition

to these viscous losses, heat conduction between fluid and the solid wall also causes energy loss.

Usually it was assumed that the condensations and rarefactions in fluid are adiabatic and

resulting in temperature change. However, for solid wall, the temperature is nearly constant, thus

causing the tendency for heat to be conducted from the fluid medium to the solid walls during

condensation and vise verse during rarefaction. The heat transfer increases the entropy of the

23
whole system and thus dissipates acoustic energy. In addition to those losses, there is direct

absorption of the acoustic energy from the fluid medium by the wall.

2.5.1 Intrinsic Absorption from Viscosity in Medium

Consider a nonlinear Navier-Stokes equation in the absence of external body forces 20

· · (2-54)

Where and are coefficient of shear and bulk viscosity respectively. is zero in

monatomic gases and finite in fluids. It measures the dissipation involving the conversions of

energy between molecular motion, internal molecular states and structural potential energy

states. counts when turbulence, laminar flow, vorticity occurs. In linear acoustics,

these are usually confined to small region near boundaries.

Assume linear acoustic wave

· (2-55)

Where condensation , is instantaneous density and is equilibrium density.

And the adiabatic condition yields

(2-56)
Substitute (2-56) (2-57) into (2-55) yields lossy acoustic wave equation

1 p (2-57)

Where is relaxation time and c is the thermodynamic speed of sound which is not necessarily

the phase speed.

Assume harmonic vibration , (2-58) can be further simplified into Helmholtz equation
p k p 0 (2-58)
24
 /
/ 1 (2-59)

Solve the absorption coefficient and phase velocity

/
(2-60)
√
/
√2 (2-61)

The solution to the wave equation for a plane wave traveling in the x direction is
ω α ω β
p Pe Pe e (2-62)
The typical value for is about 10 for fluids. So 1 is valid for very large frequency

range.

/2 (2-63)

1 (2-64)

The absorption coefficient is proportional to the square of frequency and the phase velocity

is function of frequency, so the propagation is dispersive.

The loss in fluids involved with bulk viscosity is caused by the structure relaxation. Water,

for example, is assumed to be a two-state liquid. The normal state has lower energy and in the

state of higher energy, the molecules have a more closely packed structure. In static state, at

equilibrium most of the molecules are in the first energy state. However with incident

compression wave, more molecules transfer from the first state with more open space to the more

closely packed second state. The time delays in this process and in the reversal resulted in a

relaxation dissipation of the acoustic energy. The structure relaxation can be taken into account

by adding nonvanishing absorption coefficient of bulk viscosity. Direct measurement indicates

25
that in water is around three times of . The absorption coefficient of water is measures as

25 ·

2.5.2 Viscous Absorption in Pipes at a Rigid Wall

Approach I

For moderately small pipe ( 10 ), a laminar motion exits throughout the cross section

of the pipe, the velocity increases from zero at the wall to the maximum in the center. The

velocity difference causes viscous forces between two adjacent layers of the fluid medium, given

by

(2-65)

Where S is the cross section area and u is particle velocity. is the coefficient of shear

viscosity. is a measure of the diffusion of momentum by molecular collisions between regions

of the fluid possessing different net velocities, so it is manifested in producing absorption even in

pure longitudinal motion. It is independent of frequency and depends only on temperature.

Because the temperature fluctuations in acoustic propagation are very small, can be assumed

to be a function of the equilibrium temperauter. These viscous processes delay the system to

reach equilibrium, during which the density and temperature of the fluid are changed by

expansion or compression. These resulted in the conversion of acoustic energy to random

thermal energy.

26
 u
dr
r

dx

Figure 2-6 Acoustic loss due to shear viscosity in pipes

Consider the fluid inside an annular ring of volume2 , the net force by pressure gradient

on such a ring is given by

·2 (2-66)

The net force on the annulus due to viscous force is

·2 (2-67)

By Newton’s second law

2 ·2 ·2 (2-68)

Hence the equation of motion is

(2-69)

Assume harmonic wave motion ,

So (2-70) can be simplified as

(2-70)

Assume K = (1-j) , we can write (2-71) as

(2-71)

27
The solution of (2-72) is
/
(2-72)

Where A is obtained by boundary condition that 0 at , so we have

/
1 (2-73)

The average velocity amplitude can be calculated

1 (2-74)

If the radius of the pipe

10 (2-75)

(2-76)

and hence

1 1 1 (2-77)

Substitute (2-78) into (2-75), we obtain

1 1 (2-78)

So the effect of viscosity is to introduce an addition reactance term

(2-79)

And the resistance term

(2-80)

From (2-79) (2-80), we can write the effective density as

1 (2-81)

So the effective velocity of wave propagation can be derived as

28
 1 (2-82)

Since

(2-83)

Where is the average particle displacement.

We can modify (2-83) in terms of

(2-84)

Solving this equation yields

(2-85)

Where and

So the damping increases as frequency increases. For water, the viscosity is equal to 0.01g/cm,

and density is 0.998g/cm3, 0.01 /

If the length of the pipe l is much smaller than wavelength, we can use (2-79) to derive the

acoustic impedance.

2
1

(2-86)

Compare with the ideal case, the viscous forces caused modification of the impedance of the

close end tube by an additional reactance term as well as an acoustic

resistance .

(2-87)

29
The above expressions are valid only for 10 , the ratio , will not exceed √2/

10 . So the effect of viscosity can be neglected in this case with error within 10 per cent.

Approach II

In a pipe with constant cross section, viscous loss can cause by shear at the boundary of the rigid

wall. Consider a nonlinear Navier-Stokes equation (2-55) in the absence of external body forces.

Assume lossless plane wave propagating along a rigid wall with boundary perpendicular to the

constant phase front, we can neglect · which is due to viscous loss in the bulk

of fluid and has been discussed in Sec 2.5.1.

· (2-88)

Assume acoustic plane wave exit in the positive z space, propagation along x direction. With

0 , the wave has particle velocity and pressure as function of x,t only. The presence of

rigid wall in the region 0 with its boundary at 0 introduces an additional wave

component as function of x,z,t. The total particle velocity near the boundary of the rigid

wall should be zero because of viscosity. Furthermore, should vanish at large z.

at 0 (2-89)
0 at ∞ (2-90)
Substitute into (2-90) and consider =0

(2-91)

0 (2-92)

(2-93)

30
 is the pressure associated with . If we neglect which we will justify in the end

(2-94)

The solution for this one-dimensional diffusion equation for frequency dependence
/
(2-95)
2 / (2-96)

is the viscous penetration depth or acoustic boundary layer thickness.

(2-97)
/
(2-98)

The viscosity near the boundary of the rigid wall caused an additional wave component which

/
decay along z direction , also the wave vector has z component equal to . From (2-93)

we can derive , so we can neglect in the above derivation when

The average particle velocity

1 2 (2-99)

The acoustic impedance of the fluid with the viscous boundary layer is

1
1 2
1

31
 ( (2-100)
Substitute

(2-101)

It was observed that we have obtained same impedance expression for a tube considering the

viscosity loss by two different approaches. The acoustic elements has a acoustic resistance

resulted from viscous loss in the rigid wall boundary and the effective density of the fluid

is increased as 1 , occurring as the incensement in the acoustic inductance.

The power dissipated because of the viscous boundary layer can be interpreted as .

Integrating over one period T and averaging over a wavelength gives the average power density

during one cycle of motion

(2-102)

And the total energy density of the propagation wave is

(2-103)

The quality factor from viscous losses at the wall is defined as

/2 (2-104)

Also , , so
/
(2-105)

2.5.3 Thermal Conduction

Calculation of the absorption resulting from thermal conduction between the fluid and the

isothermal walls of the pipe is quite straightforward. When the fluid is subjected to an acoustic

32
vibration, the temperature in compressed regions will be increased while the temperature in

rarefied regions will be decreased. The kinetic energy of translation in fluid is proportional to the

temperature. The molecules in region with higher temperature have greater kinetic energies that

diffuse into the surrounding cooler regions through intermolecular collisions. As energy leaves

the region, it is lost from the acoustic wave and converted into random thermal energy of

molecular motion. When there is plane wave propagation along x direction through a lossless

fluid in the pipe with equilibrium absolute temperature which can be found from the

equation of state and the adiabat to be 20

1 (2-106)
(2-107)

The additional temperature must maintain equilibrium temperature at the pipe wall

and go to zero for large distance z away from the wall. The behavior of the temperature in this

boundary layer region is described by the diffusion equation

(2-108)

Where is the specific heat at constant pressure / · e]. is the thermal conductivity

/ · . The above equation can be solved

1 1 (2-109)

2 / (2-110)

The change in thermal energy is related to the change in temperature, ∆ is the gain in

thermal energy of a unit volume of the fluid.

33
 ∆
(2-111)
∆

In the pipe of radius a, the wave is propagating along x direction, and the boundary layer

damped quickly along z direction. The loss in the boundary layer from thermal conduction is

∆
2 (2-112)
∆

The absorption coefficient is defined as

2 / (2-113)

The total acoustic energy is

(2-114)

So the absorption coefficient is calculated as

/
1 (2-115)

The skin depth for viscosity and thermal conduction are related by Prandtl number

(2-116)

So
/
(2-117)

Compared with (2-107)

We obtain the ratio of the two types of absorption coefficients

(2-118)

2.5.4 Total Loss in a Pipe

The total absorption coefficient for wall losses is

34
 = 1 (2-119)

The presence of the viscous boundary layer also modifies the phase speed of the acoustic wave.

As in (2-103), the viscosity not only resulted in a resistance term, but one reactance term. The

effect is equivalent to the fluid having effectively greater density 1 . For adiabatic

compression, the acoustic velocity depends inversely on the square root of the density. The

correction of the speed of sound for the viscous boundary layer

1 1 (2-120)
Similar conclusions can be reached for the thermal boundary layer. The density corrected
for temperature fluctuation is 1 , and the attendant correction to the phase
velocity is
1 1 (2-121)

So the total corrected phase velocity caused by the rigid wall

1 1 (2-122)

As is proportional to√ , and is related to , the speed of sound approaches the

value in free field asymptotically.

(2-123)

2 / (2-124)

Additional attenuation can be caused by the hear conduction at the walls of the pipe.21 The

study based on the assumption that the layer of fluid in contact with the walls can have neither

velocity nor change in temperature shows the effective kinematic coefficient of viscosity is

35
 1 √ (2-125)
√

2.6 Acoustic Isotropic Metamaterial

Because of the strong dispersion inherent to the resonant elements in the sonic material, the

effective material property was only obtained in a narrow frequency range. In addition, such

resonances led to undesired material absorption. In this thesis, a new approach was employed to

build a two-dimensional acoustic metamaterial based on transmission line model22.

2.6.1 Isotropic Distributed Transmission Line

In an inviscid medium, the two-dimensional (2D) time harmonic acoustic wave equations are
∂P
= − jωρuZ
∂z (2-126)
∂P
= − jωρux
∂x
∂u z ∂u x
+ = − jωβP
∂z ∂x (2-127)

Substitute (2-127) into (2-126) yields

∂ 2 P ∂P (2-128)
+ + k 2 P = 0 k = ±ω βρ
∂x 2 ∂z 2

Where P is pressure, u is particle velocity, β is compressibility, ρ is density and k is

propagation constant.

In electromagnetism, a dielectric medium can be modeled using distributed transmission

network.23 Similar analogy can be found between an acoustic system and a distributed network.

The basic unit cell in the transmission line is composed of distributed series impedances and

36
shunt admittances as shown in Figure 2-7. The 2D telegrapher’s equation for the distributed

structure can be expressed as

∂Vy
= −ZIz
∂z
∂Vy
= −ZI x (2-129)
∂x

∂I z ∂I x (2-130)
+ = −YVy
∂z ∂x

Substitute (2-130) into (2-129)

∂ 2Vy ∂ 2Vy
+ + k 2Vy = 0 k = ± − ZY (2-131)
∂x 2 ∂z 2

where Vy is the voltage , Ix , Iz are the currents ,Z is the impedance per unit length , Y is the

admittance per unit length and k is propagation constant.

Comparison of the above two sets of equations shows that the distributed transmission

network can be used to model acoustic medium properties by mapping voltage and current to the

pressure and particle velocity respectively. In this analogy, the impedance Z in the transmission

line describes the material density ρ and the admittance Y corresponds to the compressibility β .

ix + d i x
vy + dvy

(Z/2)

(Z/2)
iz iz + d iz

vy vy + dvy
(Z/2)
(Z/2)

Y
ix y
x

vy z

Figure 2-7 Unit cell for a 2D isotropic distributed transmission line

37
2.6.2 Isotropic Acoustic Metamaterial Network

The distributed acoustic transmission line mentioned in previous section can be physically

implemented by an acoustic lumped circuit network. An acoustic element can be predominantly

of either capacitance or inductance nature, depending on the relative compressibility of the fluid

inside the element. The building block in the acoustic transmission line for an effective medium

with positive refractive index (PI) is shown in Figure 2-8 (a). In this structure, the channels

connecting the cavity act as a series of acoustic inductors and the cavity as an acoustic

capacitor.24 On the other hand, the building block of the acoustic metamaterial with negative

index (NI) is shown in Figure 2-8 (b). The channels are analogous to a series of acoustic

capacitors while the through hole works as an acoustic inductor.25,26

Using the lumped circuit model, the propagation of acoustic wave through the PI network in

Figure 2-8 (e) is described as

∂P Pn+1,m − Pn−1,m jωLPU z

≈ =−
∂z dP dP
∂P Pn,m+1 − Pn,m−1 jω LPU x (2-132)
≈ =−
∂x dP dP
∂U x ∂U z U x n,m+1 − U x n,m U z n+1,m − U z n,m jω CP P
+ ≈ + =− (2-133)
∂x ∂z dP dP dP

Where P is pressure, U x U z are x and z component of volume velocity, L P and CP are acoustic

inductance and capacitance, d P is the periodicity and ω is angular frequency.

Compared the above equations with the microscopic sound wave equations (2-128) and (2-129),

we can derive the effective density and compressibility as

38
 LP S P l
ρ eff , P = = ρw P (2-134)
dP dP

CP VP
β eff , P = = (2-135)
S P d P ρ w cw2 S P d P

Where both effective density and compressibility are positive, ρ w and cw are the density and

speed of sound of water. The geometry parameter l P is the channel length, S P is the cross

section area of the channel and V P is volume of the cavity as shown in Figure 2-8(c).

The corresponding propagation constant is

lPVP
ω
ω LP CP S P cw2
kP = = (2-136 )
dP dP

The phase and group velocities are represented by

ω dP dP
vϕ , P = = =
kP LP C P l PV P
S P c w2

−1
⎛ ∂k ⎞ dP dP
vg , P =⎜ P ⎟ = = (2-137)
⎝ ∂ω ⎠ LP CP lPVP
S P cw2

Relative effective acoustic refractive index nP can be determined as

lPVP
cw cw LP C P SP
nP = = = (2-138)
vϕ , P dP dP

which is demonstrated to be positive.

39
 Similarly, the propagation of acoustic wave in the NI networks in Figure 2-8 (f) are

estimated as

∂P Pn +1, m − Pn −1, m Uz
≈ =−
∂z dN jω C N d N
∂P Pn ,m +1 − Pn , m −1 Ux (2-139)
≈ =−
∂x dN jω C N d N

∂U x ∂U z U x n ,m+1 − U x n ,m U z n+1,m − U z n ,m P (2-140)

+ = + =−
∂x ∂z dN dN jω LN d N

Where L N and CN are acoustic inductance and capacitance, d N is the periodicity.

The effective density and compressibility are derived as

SN S N ρ w c w2
ρ eff , N = − 2
= −
ω CN dN ω 2V N d N

1 S0
β eff , N = − 2
=− 2
(2-141)
ω LN d N S N ω ρ w lN d N S N
Both the effective density and compressibility are negative and function of frequency. The
geometry parameter l N is the length of the open hole, S0 is the cross section area of the open

hole and SN is the cross section area of the main channel as shown in Figure 2-8(d).The
propagation constant is defined as
kN = −
1
= −
1 (2-142)
ωdN LN C N lN V N
ωdN
S 0 c w2

We found the phase and group velocity are

ω lNVN
vϕ, N = = −ω2dN LN CN = −ω2dN
kN S0cw2

40
 (a) (b)

(c) (d)

(e) (f)

Figure 2-8 The unit cell of (a) PI and (b) NI medium. The two-dimensional geometry of the
unit cells in the (c) PI and (d) NI medium and corresponding lumped circuit in the (e) PI and (f)
NI medium.

−1
⎛ ∂k ⎞ l V
vg , N = ⎜ N ⎟ = ω2d N LN CN = ω2 dN N N2 (2-143)
⎝ ∂ω ⎠ S0cw
41
The refractive index is calculated as

cw cw cw2
nN = =− 2 =− (2-144)
vφ , N ω d N LN C N l N VN
ω 2d N
S0

The negative root is chosen in equation (2-146) to guarantee a positive group velocity. From

the above derivation, we found that an acoustic metamaterial with negative refractive index can

be modeled by a series of capacitors with a shunt inductor.

2.7 Anisotropic Acoustic Metamaterial

Recently, a new design paradigm called conformal mapping and coordinate transformation has

inspired a series of key explorations to manipulate, store and control the flow of energy, in form

of either sound, elastic waves or light radiation. In electromagnetism, because of the coordinate

invariance of Maxwell’s equations, the space for light can be bent in almost arbitrary ways by

providing a desired spatial distribution of electric permittivity ε and magnetic permeability μ.27,28

Similar design approach can be applied to acoustic waves by a engineered space with desired

distribution of effective density and compressibility.29,30,31 A set of novel optical/acoustic

32,33,34,35
devices were proposed based on transformation optics ; they usually call for complicated

medium with anisotropic and spatially varying material parameter tensor to accomplish the

desired functionality. Therefore, the 2D isotropic transmission line model is extended in this

section to build an anisotropic acoustic metamaterial which promise potential application for a

myriad of fascinating devices based on coordinate transformation.

42
2.7.1 Anisotropic Acoustic Distributed Transmission Line

In an inviscid medium, the two-dimensional (2D) time harmonic acoustic wave equations in a

cylindrical coordinate are

∂P (2-145)
= − jωρφ uφ
r∂φ
∂P
= − jωρ r u r
∂r
1∂ 1 ∂uφ (2-146)
(rur ) + = − jωβP
r ∂r r ∂φ

Where P is pressure, u is particle velocity, β is compressibility, ρ ρ is density along radial

and angular direction respectively.

Assume an anisotropic transmission line in a cylindrical coordinate. The basic unit cell is

composed of serial impedances and shunt admittances as shown in Figure 2-9. The 2D

telegrapher’s equation for the distributed structure can be expressed as

∂V (2-147)
= − Iφ Zφ
r ∂φ
∂V
= −I r Zr
∂r
1 ∂ 1 ∂ Iφ (2-148)
(r I r ) + = −VY
r ∂r r ∂φ

Where V is the voltage,I , I are the currents along r and direction respectively, Z , Z

are the impedance per-unit length along radial and angular direction and Y is the admittance

per-unit length.

Comparison of the above two sets of equations shows that the distributed transmission

network can be used to model an anisotropic acoustic medium by mapping voltage and current to

the pressure and particle velocity respectively. In this analogy, the impedance Z , Z in the

43
transmission line describes the material density ρ , ρ accordingly and the admittance Y

corresponds to the compressibility β . This analog implies that we can model the cylindrical

cloak with an inhomogeneous anisotropic transmission line by modifying the value of distributed

impedance and admittance accordingly.

Figure 2-9 Unit cell for a 2D anisotropic distributed transmission line

2.7.2 Anisotropic Acoustic Metamaterial Network

The anisotropic transmission line can be physically implemented by an acoustic lumped circuit

network which consists of 2D array of the building blocks as shown in Figure 2-10. In each unit

cell, the cavity with large volume in center works as an acoustic capacitor whereas the channels

connecting to four neighboring cells act as serial inductors. The analogous inductor-capacitor

(LC) circuit is shown in Figure 2-10 (b). This unit cell has the same topology as the one in PI

isotropic medium. However, the unit cells here are positioned periodically along diagonal

direction in a cylindrical coordinate. Moverover, the geometry of the unit cell is tuned

individually to build an anisotropic effective acoustic medium. As the unit cell is much smaller
44
than the operational wavelength, the LC based circuit network is seen as effective medium by the

acoustic waves. Based on the lumped circuit model, the propagation of the acoustic wave

through the unit cell in Figure 2-10 (a) can be written as,

∂P Pn +1, m − Pn −1, m jωLrU r

` ≈ =− (2-149)
∂r Δr Δr
∂P Pn, m +1 − Pn, m −1 jωLφUφ
≈ =−
r∂φ rΔφ rΔφ
1∂ 1 ∂U r U −r U U −U jω C P
( rUr ) + φ ≈ n+1 r n+1,m n r n−1,m + φ n,m+1 φ n,m−1 =− (2-150 )
r ∂r r ∂φ rnΔrn rnΔφm Δr

where U , U are r and φ component of volume velocity , L and L are acoustic

inductance, C is acoustic capacitance and ω is angular frequency.

Compared the above equations with the microscopic sound wave (2-149) and (2-150), we

find the effective density and compressibility of the system interpreted as

Lr Sr Lφ Sφ C
ρeff ,r = ρeff ,φ = βeff = (2-151 )
Δr rΔφ Sr Δr

According to the definition of the acoustic inductance and capacitance

lr lφ V
Lr = ρw Lφ = ρ w C= (2-152 )
Sr Sφ ρ w cw2

Substitute (2-152) into (2-151) yields,

lr lφ V
ρeff ,r = ρw ρeff ,φ = ρw βeff = (2-153 )
Δr rΔφ ρwcw2 Sr Δr

where ρ is the density of water and c is sound speed in water. The geometry parameters

and are the channels length and and are the cross section area of the channels along

r and φ direction respectively. V is the volume of the large cavity. Equation (2-153)

indicates the dependence of effective density and compressibility on the structure geometry

45
and filling medium which is water in the experiment. Therefore, this discrete network allows

for a practical implementation of an anisotropic effective acoustic medium with spatial

gradients by modifying the geometry and placement of the building blocks.

(a) (b)
Figure 2-10 (a) One unit cell for an anisotropic acoustic metamaterial and (b) the corresponding
lumped circuit element.

References
1
J. B. Pendry, "Metamaterials and the Control of Electromagnetic Fields," Proceedings of the

Ninth Rochester Conference on Coherence and Quantum Optics (2007).

2
Veselago,V. G.,“The electrodynamics of substances with simultaneously negative values of µ

and ε,” Sov. Phys. Usp., Vol. 10, No. 4,509 (1968).

3
Smith, D. R., et al., “Composite Medium with Simultaneously Negative Permeability and

Permittivity”, Phys. Rev. Lett. 84, 4184–4187, (2000).

46
4
Shelby, R. A., et al.,” Experimental VeriÞcation of a Negative Index of Refraction”, Science

292, 77–79, (2001).

5
J. B. Pendry, D. Schurig, D. R. Smith,” Controlling Electromagnetic Fields”, Science 312,

1780, (2006).

6
U. Leonhardt, “Optical Conformal Mapping”, Science 312, 1777 (2006).

7
D. Schurig et al, “Metamaterial Electromagnetic Cloak at Microwave Frequencies.”, Science

314, 977-980 (2006).

8
Jacob, Z., Alekseyev, L. V. & Narimanov, E. “Optical hyperlens: Far-field imaging beyond the

diffraction limit.” Opt. Express 14, 8247–8256 (2006).

9
Z. Liu, H. Lee, Y. Xiong, C. Sun, X. Zhang, “Far-Field Optical Hyperlens Magnifying

Sub-Diffraction-Limited Objects”, Science 315, 1686 (2007).

10
N. Fang, D. Xi, J. Xu et al., “Ultrasonic metamaterials with negative modulus”, Nature

Materials, 5,452-456 (2006).

11
S. Zhang, L.Yin, and N. Fang, “Focusing Ultrasound with an Acoustic Metamaterial

Network”, Phys. Rev. Lett. 102, 194301 (2009).

47
12
J. Li, L. Fok, X. Yin, G. Bartal and X. Zhang, “Experimental demonstration of an acoustic

magnifying hyperlens”, Nature Materials 8, 931 - 934 (2009).

13
Z. Liu, X. Zhang, Y. Mao, Y. Y. Zhu, Z. Yang, C. T. Chan, and P. Sheng, “Locally Resonant

Sonic Materials”, Science 289, 1734 (2000).

14
V. Fokin, M. Ambati, C. Sun, and Xiang Zhang, “Method for retrieving effective properties of

locally resonant acoustic metamaterials”, Phys. Rev. B 76, 144302 (2007).

15
Z. Liu, C. T. Chan, and P. Sheng, “Analytic model of phononic crystals with local

resonances”, Phys. Rev. B 71, 014103 (2005).

16
Y. Wu, Y. Lai, and Z. Q. Zhang, “Effective medium theory for elastic metamaterials in two

dimensions”, Phys. Rev. B 76, 205313 (2007).

17
Y. Ding, Z. Liu, C. Qiu, and J. Shi, “Metamaterial with Simultaneously Negative Bulk

Modulus and Mass Density”, Phys. Rev. Lett. 99, 093904 (2007).

18
L. E. Kinsler, Fundamentals of Acoustics, 1982, Wiley, New York.

19
L.L.Beranek, Acoustics, 1954, McGRAW-HILL,New York.

20
Lawrence E. Kinsler, Austin R. Frey, Alan B. Coppens, and James V. Sanders, Fundamentals

of Acoustics, 1999, Wiley, New York.

48
21
R.D.Fay, J.Acoust.Soc.Am., 12,62, (1940).

22
N. Fang, D. Xi, J. Xu et al., “Ultrasonic Metamaterials with Negative Modulus”, Nature

Materials, 5,452-456 (2006).

23
A. Grbic, G. V. Eleftheriades’, “Overcoming the diffraction limit with a planar left-handed

transmission-line lens”, Phys. Rev. Lett., 92, 117403 (2004).

24
L.L.Beranek, Acoustics, 1954, McGRAW-HILL,New York.

25
W. P. Mason, “A study of the regular combination of acoustic elements, with applications to

recurrent acoustic filters, tapered acoustic filters, and horns”, Bell Syst. Tech. J . 6, 258 (1927) .

26
G. W. Stewart, “Acoustic wave filters”, Phys.Rev.20, 528 (1922).
27
Pendry J B, Schurig D and Smith D R ,” Controlling Electromagnetic Fields “, Science

312 ,1780 (2006).

28
U. Leonhardt, “Optical Conformal mapping”, Science 312, 1777 (2006).

29
S. A. Cummer and D. Schurig, “One path to acoustic cloaking”, New J. Phys. 9, 45 (2007).

30
Cummer SA et al, “Scattering theory derivation of a 3D acoustic cloaking shell”, Phys Rev

Lett. 100, 024301(2008).

49
31
Milton G W, Briane M and Willis J R, “On cloaking for elasticity and physical equations with

a transformation invariant form”, New J. Phys. 8, 248 (2006).

32
A. V. Kildishev, V. M. Shalaev, “Engineering space for light via transformation Optics”, Opt.

Lett. 33, 43 (2008).

33
M. Rahm, S. A. Cummer, D. Schurig, J. B. Pendry, “Optical Design of Reflectionless Complex

Media by Finite Embedded Coordinate Transformations”, D. R. Smith, Phys. Rev. Lett. 100,

063903 (2008).

34
M. Rahm et al., “Transformation-optical design of adaptive beam bends and beam expanders”,

Opt. Express 16, 11555 (2008).

35
Pendry J .B .and Li Jensen, “An acoustic metafluid: realizing a broadband acoustic cloak”,

New J. Phys. 10 115032 (2008).

50
3 ULTRASOUND FOCUSING USING ACOUSTICMETAMATERIAL NETWORK

3.1 Introduction

High-resolution acoustic imaging techniques are the essential tools for nondestructive testing and

medical screening. However, the spatial resolution of the conventional acoustic imaging methods

is restricted by the incident wavelength of ultrasound. This is due to the quickly fading

evanescent fields which carry the subwavelength features of objects. To overcome this

1
diffraction limit, a remarkable perfect lens is proposed by John Pendry, which offers the

promise to build a device allowing super-resolution imaging of an object. This perfect lens is

based on focusing the propagating wave and recovering the evanescent field through a flat

negative-index slab. Since then research on metamaterials has been stimulated by the opportunity

to develop artificial media that refract waves in negative direction. Several different

metamaterials have been proposed and demonstrated to present negative index of refraction.

2,3,4,5,6

The successful demonstration of electromagnetic (EM) superlens7,8,9,10 has inspired the

search for the analogous acoustic negative-index lens. In fact, phononic crystals11,12,13,14,15 were

first investigated to develop negative-refractive devices for sound waves. Beam steering in

phononic crystals can be achieved by Bragg scattering, leading to enhanced diffraction in

negative direction. Ultrasound focusing from negative refraction by a three-dimensional

phononic crystal was first demonstrated experimentally by Yang et al.13 A focal spot around five

51
wavelengths in width was observed in the far field at 1.57 MHz. Recently, a finer resolution was

achieved by focusing the ultrasound field emitted by a subwavelength line source using a

two-dimensional (2D) phononic crystal slab.15

However, for lens design based on phononic crystals, the dependence of band structure on

the lattice periodicity usually requires the spatial modulation to be the same order of magnitude

as the acoustic wavelength, which would makes such structure impracticably large. Locally

resonant sonic materials16 made a major step towards the acoustic metamaterial development.

Since the lattice constant is much smaller than the relevant wavelength, effective medium

properties can be attributed to this sonic material at low frequency. With appropriate resonances

included into the building block, acoustic metamaterials with either negative effective mass

density or bulk modulus or both have been demonstrated.17,18,19,20 These anomalous phenomena

resulted from strong coupling of the traveling elastic wave in the host medium with the localized

resonance in the building block. However, to the best of our knowledge, there is no experimental

demonstration of focusing ultrasound waves in these negative index acoustic metamaterials.

In this section, we experimentally investigated the focusing of a point source from a

designed ultrasonic metamaterial consisting of a planar network of subwavelength Helmholtz

resonators. To facilitate the design, we adapted the 2D transmission line (TL) method which is

widely used in the development of negative index EM metamaterials.8,9,10 In this approach, the

acoustic system is converted to an analogous lumped circuit model in which the motion of the

fluid is equivalent to the behavior of the current in the circuit. Similar to permittivity and
52
permeability in the EM metamaterial, 8 the effective density and compressibility of the network

structure are found to be related to the capacitance and inductance in this lumped circuit. Earlier,

in the one-dimensional version of this ultrasonic metamaterial, the elastic modulus is found to be

negative at specific frequency range theoretically and experimentally.21

3.2 Negative Refractive Index Lens

The refractive index is a fundamental parameter describing the interaction between waves and

material. In late 1960s, Veselago22 first considered the theoretical possibility for a medium

having simultaneous negative permittivity and negative permeability. When ε (ω) < 0, μ(ω) < 0

at certain frequency, the refractive index is n = − εμ and the negative sign is taken to satisfy

causality. When light passes from a positive (n > 0) to a negative (n < 0) medium, Snell’s law

implies that the angle of refraction is negative, showing the refracted ray emerges on the same

side of the normal as the incident ray .

In 2000, Pendry1 proposed that a thin slab of metamaterial with negative refractive index

could make a ‘perfect lens’ without any loss of details. A conventional lens cannot focus light

onto an area smaller than a square wavelength due to the diffraction limit. The reason for this

limit is that the evanescent waves which carries the subwavelength details of the object is

exponentially decaying, leading to the loss of those fine features in the image. However, utilizing

negative refractive index material, a ‘perfect lens’ can focus propagation waves and also amplify

evanescent waves to generate subwavelength imaging.

53
 As shown in Figure 3-1, negative refraction allows a flat slab lens of negative index to focus

all the diverging light rays from an object into two images: one inside the slab and the other one

outside the slab. The evanescent waves have been enhanced across the lens and decay again after

emerging from the negative index lens. Therefore the amplitude at the two image planes reaches

their original level. At the same time, the propagating waves pass through the negative index lens

with a reversed phase front, leading to zero phase change at the image planes. By completely

recovering both propagating and evanescent waves in phase and amplitude, a perfect image is

obtained.

An approximation of the ‘perfect lens’ called as ‘super lens’ was built in optical frequency

range under near-field condition. In the near field, since the electric and magnetic components

are decoupled, the ‘super lens’ only needs negative dielectric permittivity for one polarization

light. The optical superlensing effect on the scale of 60-nm (λ/6) was observed by excitation of

surface plasmons 7 through the metal/dielectric layer structure. The sub-diffraction-limited image

was recorded by optical lithography at 365 nm wavelength.

In microwave frequency, subwavelength focusing was realized by different groups. One

example is to use a planar transmission-line structure. 10 The negative index lens is a planar slab

consisting of a grid of printed metallic strips over a ground plane, loaded with series capacitors

and shunt inductors. In the experiment, the loaded grid is sandwiched between two unloaded

printed grids that act as effective homogeneous media with a positive refractive index.

54
Figure 3-1 (a) A flat lens brings all the diverging rays from an object into two focused images

(b) the NI medium can enhance the evanescent waves across the lens, so the amplitude of the

evanescent waves are identical at the object and the image planes.

3.3 Phononic Crystal

Great interest towards negative index material has been induced in phononic crystal study since

Pendry proposed the ‘superlens’ idea in 2000. Phononic crystals are the acoustic or elastic

analogue of photonic crystals (PC) for electromagnetic wave. It composes of periodic structure

of different acoustic velocities and densities with lattice constants on the scale of the wavelength

of sound (or ultrasound). Initially, phononic crystals attracted a lot of interest because of the

exhibition of band gaps, over which wave propagation is forbidden due to Bragg scattering.

Novel acoustic waveguide, sound insulators and filters have been proposed and demonstrated.

55
23,24,25,26,27
The phononic crystas also have a rich variety of physical properties of interest to

fundamental and applied research. There are longitudinal and transverse bulk waves as well as

surface waves in phononic crystal, providing distinct advantages for studying propagation and

scattering properties of classical waves.

In the low frequency region below first band gap, the phononic crystal is transparent to

sound. The refractive acoustic devices have been proposed and demonstrated using phononic

crystal in this frequency region when the wavelength is moderately long compared to the lattice

constant. The Fabry-Perot-like interferometer and a converging lens made by periodic

distributions of rigid cylinders in air have been physically realized and tested for airborne

sound.28 The acoustic focusing and imaging effect were numerical studied through lenslike
29,30
phononic cystal device with a few wavelengths size.

The negative refraction in the periodic crystal is achieved by the band-folding effect with

locally positive material property ( μ,ε for EM wave and ρ , E for acoustic and elastic wave),

everywhere inside the crystal, which is fundamentally different from those in metamaterial with

effective negative material property. The direction of power flow is given by the gradient vector
r
∇krω(k ) and thus negative refraction can be obtained because of anisotropy of crystal

equifrequency contour or surface.

Negative refraction was first demonstrated in PC by employing the anisotropic shape of

equifrequency sufaces (3D) or contours (2D) to achieve antiparallelism of vectors S (Poynting

vector) and k (wave vector) ( S • k < 0 ).31 Furthermore, the negative refraction can also occur

without negative index or backward wave effect ( S • k > 0 ).32,33 Luo et al studied the possibility

to employ the convex equifrequency contuores located at the corners of the first Brillouin zone in

56
the first band to achieve negative refraction in a 2D PC.15 The superlensing effect was

demonstrated experimentally by Cubukcu et al with a 2D PC made of a square array of dielectric

rods in air. The refraction was achieved in the convex equifrequency contour in the first band.

The full width at half maximum of the focus was 0.21λ (where λ is the wavelength of the
34
incident wave).

Analogous phenomena are found in phononic crystal for acoustic wave. Outside the band

gap, wave propagation in phononic crystals is strongly influenced by crystalline anisotropy,

leading the group velocity, which represents the direction of energy transport by Bloch waves

inside the crystal, no longer parallel to the wave vector.35 In atomic crystals, analogous effects in

phonon focusing have been extensively studied,36 but experiments are limited to the long

wavelength regime where λ >> a .

Negative refraction of acoustic waves in two-dimensional (2D) phononic crystals37 has

been theoretically demonstrated in the first band resulted from euqifrequency contours along the

ΓΜ direction without employing the backward wave effect ( S • k > 0 ) in a way similar to what

is found in PC15. They numerically demonstrated the near-field imaging of a point source with

the full width at half maximum of 0.14 λ , which overcomes diffraction limit. Far-field imaging

by negative refraction through a 2D phononic crystal consisting of hexagonal arrays of steel

cylinders in air are realized. The left-handed behavior S • k < 0 is found in the second band.

The refractive index matches only with extremely high filling fraction, causing low

transmission.38 The ultrasound focusing phenomena result from negative refraction of a 3D

phononic crystal was first demonstrated experimentally in 2004.39 The negative refraction was

achieved with source incident obliquely with respect to the [111] direction. Compared with the

57
field without lens, a focal spot around five wavelengths was observed in the far field at 1.57

MHz as presented. Ke et al reports experimental observation of focusing and negative refraction

of acoustic wave in a 2D phononic crystal consisted of a triangular array of steel rods immersed

in water. The circular equifrequency contour in the second band guaranteed the negative

refraction for all incidence angles. Negative refraction was demonstrated by monitoring the

transverse displacement of the sound beam transmitted through a flat phononic crystal with

respect to the incident beam. The lateral resolution of the far-field image was found to be a

couple of wavelength.40 Recently, better resolution was observed experimentally with a similar
41
2D phononic crystal as in by imaging a subwavelength line source which was positioned about

half a wavelength away from the crystal surface. Prism-shaped phononic crystal was used to

verify the negative refraction in the second band where the antiparallel directions of wave vector

and group velocity were induced by the circular equifrequency contours.

In the above discussion, the negative refraction is induced due to the band-folding effect in

the phononic crystal. The strongly modified dispersions resulted from Bragg scattering. The

imaging conditions rely on appropriate intrinsic parameters of the chosen components in

phononic crystal, such as the contrast of mass density and sound speed, the volume filling

fraction, and the lattice structure. The dependence of band structure on the lattice periodicity

usually requires the spatial modulation must be of the same order of magnitude as the sonic

wavelength, which would make such structures impracticably large.

Another mechanism which can help to achieve negative refraction is local resonant structure.

Negative effective mass density or elastic modulus can occur at certain frequencies if an

appropriate resonance is included into the building block. Although the static elastic modulus and

density need to be positive to maintain stable structure, these dynamic effective acoustic
58
properties are dispersive in nature and turn negative at resonance. When the resonance-induced

scattered field prevails over the background incident field, the volume change can be out of

phase with applied dynamic pressure, implying negative bulk modulus effect. And the

acceleration is out of phase with the dynamic pressure gradient, showing negative mass density
42
effect. In 2000 Liu and colleagues43used localized resonance structure in a three-component

phononic crystal to realize band gaps at low frequency with a lattice constant two orders of

magnitude smaller than the relevant wavelength. Negative effective elastic modulus was realized

due to a dipolar resonance at low sonic frequencies. Those anomalous phenomena resulted from

strong coupling of the traveling elastic wave in the host medium with the localized resonance

rather than Bragg scattering. In the long wavelength limit, it is more appropriate to call this

“phononic crystal” as acoustic metamaterial since the lattice constant is much shorter than the

acoustic wavelength of interest. Therefore the effective medium approach can be employed to
44,45
offer a good estimation and give an intuitive understanding of this complex system.

It was demonstrated that an acoustic metamaterial can possess simultaneously negative bulk

modulus and mass density by combining two types of structural units. While the monopolar

resonances give rise to the negative bulk modulus, the dipolar resonances give rise to the

negative mass density.46

3.4 Ultrasound Focusing by Acoustic Transmission Line Network

For lens design based on phononic crystals, the dependence of band structure on the lattice

periodicity usually requires the spatial modulation to be the same order of magnitude as the

acoustic wavelength, which would makes such structure impracticably large. Locally resonant

59
sonic materials made a major step towards the acoustic metamaterial development. Since the

lattice constant is much smaller than the relevant wavelength, effective medium properties can be

attributed to this sonic material at low frequency. With appropriate resonances included into the

building block, acoustic metamaterials with either negative effective mass density or bulk

modulus or both have been demonstrated. These anomalous phenomena resulted from strong

coupling of the traveling elastic wave in the host medium with the localized resonance in the

building block. However, to the best of our knowledge, there is no experimental demonstration

of focusing ultrasound waves in these negative index acoustic metamaterials. In this thesis, an

acoustic lens with negative refractive index was implemented based on transmission line network

for ultrasound focusing.

3.4.1 Numerical Study of Acoustic Lens with Negative Refractive Index

To study the ultrasound focusing by a negative index lens, a two-dimensional circuit model is

simulated employing commercial software SPICE. The calculated lumped model is an

approximation of the distributed acoustic system. The acoustic metamaterial with negative

refractive index (NI) is composed by a two-dimensional (2D) 50× 50 periodic cascaded array of

the unit cell as in Figure 3-2 (a). In order to build a PI/NI interface, an acoustic metamaterial

with positive index (PI) is implemented by 50 × 50 circuit cells as shown in Figure 3-2 (b). In

the circuits, a very small resistance is connected to each inductor. The boundary of the

60
simulation model is grounded by a resistor with value equal to the characteristic impedance of

the transmission line to reduce the reflection from the boundary.

Figure 3-3 (a) and (c) illustrates the normalized pressure magnitude and phase distribution of

the sample at steady state when a continuous signal at 50 kHz is introduced at the center of the PI

part [node (25, 25)]. The x-y axes are labeled according to the cell number. The maximum field

magnitude was normalized to unity. The focal point is expected around node (75, 25) since the

relative index value equals –1 at 50 kHz. Focusing is evident by the increased transmission and

confinement of the fields near the focal plane [near node (75, 25)]. Therefore, a tight spot with

central magnitude about –12dB is presented Figure 3-3 (a). Moveover, concavity waterfronts are

observed in both NI and PI as evident in Figure 3-3 (c).

(a) (b)
Figure 3-2 Unit cells for acoustic metamaterial with (a) positive and (d) negative refractive
index based on transmission line model

61
(a) (b)

(c) (d)

(e) (f)

(g) (h)

62
Figure 3-3 Pseudo colormap of scaled pressure amplitude distribution due to a point source

illuminating a 2D transmission model of the (a) PI/NI and (b) PI/PI interface. Phase distribution

through (c) PI/NI and (d) PI/PI interface. Normalized pressure along the horizontal central line

through the (e) PI/NI and (f) PI/PI interface model. Lineplot of scaled pressure on three nodes

along the propagation direction in (g) PI/NI and (h) PI/PI interface.

The interface of PI/PI is also simulated as a comparison. The normalized pressure magnitude

and phase distribution of the model is presented in Figure 3-3 (b) and (c). Away from the source,

the cylindrical wavefront approaches plane wave fronts.

In Figure 3-3 (e) and (f), the pressure along the central line (y=25) in the two cases are

plotted respectively. In both plots, the source appeared as a peak with maximum value in node

(25, 25). In the NI medium, a pressure magnitude peak around –12dB appear at (75, 25) while

the pressure filed decays without no focusing in Figure 3-3 (f).

Figure 3-3 (g) and (h) shows the scaled pressure pulse as function of time at three different

nodes A (45, 25), B (50, 25) and C (55, 25). In Figure 3-3 (g), compared with the pressure pulse

at A, the one at B presents phase delay as result of longer distance from the source.

However, the pulse at C shows phase advance compared with the one at B even though C is

further away from the point source. Therefore, the phase velocity is negative in the NI medium.

However, in Figure 3-3(h), the comparison between three pulses clearly shows the advancement

of the wave front when there is only PI medium.

63
Figure 3-4 Normaalized presssure along the horizonntal central line througgh the PI/N
NI interfacee

model with
w source of differentt frequenciees.

To investigatee the influeence of relative refracctive indexx on focus property, a series off

simulation are impplemented at malized pressures alongg the centrall

a different frequenciess. The norm

line thrrough the soource are pllotted in Fiigure 3-4 with

w the sourrce inducedd in the cennter of PI att

node (2
25, 25).

At 50
5 kHz, the focusing point occurreed at node ((75, 25) in the
t center oof the NI ressulting from
m

the indeex match. As

A frequenccy decreasees to 40 kH
Hz, the posittion of focuus moves fuurther away
y

from thhe interface. On the othher hand, at higher freqquency 60 kHz

k and 70 kkHz, the focus appearss

nearer to
t the interfa
face.

Ownning to thee flat interfface, the veertical posiition of thee source haas no influeence in thee

focusing mechanissm. In order to demoonstrate this property,, the pressuure field iss calculated
d

illuminaated by twoo point sourrces. Two foocuses are observed

o gure 3-5 (b),
inn Figure 3-55 (a). In Fig

the line plots of thee pressure field

fi along horizontal
h d
direction preesent two peeaks in amp
plitude

64
 (a) (c)

(b) (d)

Figure 3-5 Pseudo colormap of scaled pressure (a) amplitude and (b) phase distribution due to

two point sources illuminating the 2D transmission line model of the PI/NI interface. Lineplot of

pressure along three different (c) lines (d) and columns.

65
 (a) (b)

Figure 3-6 Pseudo colormap of scaled pressure (a) amplitude and (b) phase distribution due to

a beam sources illuminating the 2D transmission line model of the PI/NI interface

around column 75 corresponding to two focusing spots. In Figure 3-5 (d), the line plot shows the

peak positions of the two sources and focuses along vertical direction. The open circles are the

pressure at the PI/NI interface. The focusing points are observed in the same horizontal position

as the corresponding sources due to the flat interface.

The imaging of a beam source by the negative index lens is also studied. The beam source is

introduced by applying pressure field at ten nodes vertically in phase in the center of the PI part.

In Figure 3-6 (a), the wave is convergent in the center of the NI part. In Figure 3-6 (b), phase

distribution in the NI presents similar pattern as the one in the PI part.

66
3.4.2 Experimental Demonstration of Acoustic Lens with Negative Refractive Index

3.4.2.1 Experimental Setup and Data Acquisition

Figure 3-7 shows the experimental setup to study the focusing phenomena of the acoustic

metamaterial. To prepare the sample, we machined a 2D array of periodically connected

subwavelength Helmholtz resonators in an aluminum plate and the resonators are filled with

water. As shown in previous work,47,48,49 a main transmission channel with recurrent side

branches, which are closed at the outer end, is analogous to a circuit of a series of inductors with

shunt capacitors. On the other hand, when the side tubes inserted in the main channel is open on

the outer end, the acoustic system can be described by a lumped network of a series of capacitors

with shunt inductors. The left and right half parts in the sample are 2D periodic versions of those

two different types of topology respectively. One unit cell from each half part is enlarged and

shown in the two insets respectively.

The left half part is composed of a 2D array (40 by 40) of larger cavities connected with

main channels. The volume of the cavity is around ten times of that of one section of the

channels. Consequently, when an incident acoustic wave is applied onto the fluid in the channels,

the pressure gradient through the channels is much greater than that inside the cavity. Hence, it is

as if the fluid in the cavity were at rest relative to those in the channels.48 So when the plug of

fluid in the channels oscillates as a unit, there are adiabatic compressions and rarefactions of the

fluid inside the larger cavity. Such an acoustic system is analogous to an inductor–capacitor

67
circuit as shown in the inset with the channels acting as a series of inductors ( LP ) and the cavity

providing the stiffness element as capacitors (CP ) . The periodicity (3.175mm) of the sample is

one-eighth of the wavelength at around 60 KHz frequency range. Table 3-1shows the geometry

of these two different unit cells. Given these values, the lumped circuit model is a valid

50
approximation for the distributed acoustic system with only 10% error. Following the

8,9,10
approach of EM circuit analysis , the effective density and compressibility of this network

LP SP C
can be expressed in the form as ρeff ,P = , βeff , P = P ,where d P is the periodicity, S P
dP SP d P

is the cross section area of the channels. Both effective density and compressibility are positive.

cw LP CP
Effective relative acoustic refractive index nP can be determined by nP = , where
dP

cw is speed of sound in water. We call this half part as effective positive index medium.

The right half part of the sample is the dual configuration of the left half part, in which there

is an array (40 by 40) of orifices connected with channels. The volume of one section of the main

channel is designed as around ten times of that of the orifice. Since the fluid in the orifice is not

confined, it experiences negligible compression while the fluid in the channels experience less

48
movement in average compared with that in the orifice. Consequently, when the fluid in the

orifice oscillates as a unit, there are adiabatic compressions and rarefactions of the fluid inside

the main channels. Such an acoustic system is described as a lumped network with a series of

capacitors (C N ) for the main channel part and a shunt inductor ( LN ) due to the orifice. The

periodicity is the same as that in the left part, so the effective mass density and compressibility
68
 SN
can be similarly estimated as ρeff , N = − 2
ω CN d N

1
βeff , N = − 2
, where d N is periodicity and S N is the cross section area of connecting
ω LN d N SN

cw cw
channels. Both parameters are negative. The refractive index n N = =− 2 is
vφ ω d N LN C N

negative. So this network structure acts as a medium exhibiting negative index (NI) of refraction.

The two half parts are designed with effective indices of equal and opposite value and matched

impedance ρ eff / β eff at the design frequency 60.5 kHz.

For experimental confirmation of ultrasound focusing in this acoustic metamaterial, we

measured the pressure field through this PI/NI interface. The ultrasound waves were launched

from a horn shaped transducer with a tip of 3 mm diameter in size. The tip is inserted into a hole

drilled through the center of the PI part [(column, row) = (20, 20)] to illuminate the sample. A

waveform generator (Tektronix AFG 310) is used to drive the transducer. The source generated a

burst of sine waves with a width of 5 periods. To map the pressure field, a hydrophone was

mounted on two orthogonal linear translation stages. By stepping the miniature hydrophone

(RESON TC4038-1) to the positions above those through holes in the NI part and recording the

pressure amplitude at every step, we acquired the spatiotemporal field distribution of the

ultrasound wave focusing pattern. The pressure field is afterwards amplified by Stanford research

systems model SR650 and captured using a digital

69
Figure 3-7 Schematic showing the experimental setup. The sample with PI/NI interface is

composed of an array of different designed Helmholtz resonators machined from an aluminum

plate. A burst of monotonic signal with a width of five periods was used to drive the transducer

as an underwater sound source in the center of the half part with positive index. One needle-sized
70
hydrophone detected the ultrasonic signals from the array of open holes in the negative index

half part. Unit cells of each half part and the corresponding inductor–capacitor circuit analogy

are shown in the insets. For the PI part in the top inset, the channels surrounding the cavity act as

a series of acoustic inductors L P and the cavity is a shunt capacitor C P . The building block of

the acoustic metamaterial in the bottom inset corresponds to a circuit with a series of acoustic

capacitors C N due to the channels and an acoustic inductor LN due to the open hole in the

middle

Table 3-1 Geometry parameters of the sample

PI network NI network

Periodicity (mm) 3.175 3.175

Cross-sectional area of the main channels (mm2) 1.23 4

Diameter of the through holes (mm) N/A 1

Length of the through holes (mm) N/A 0.96

Cavity Volume (mm3) 6.89 N/A

oscilloscope (Agilent DSO6104A) and then downloaded to a computer for post processing and

analysis. The field pattern in the NI part at frequency of interest was obtained by Fourier

transformation of the acquired pulse at each grid point, allowing the wave amplitude of each

frequency component to be plotted as a function of position in the NI part. The limitations on the

transducer working spectrum prevent us from probing the acoustic wave field below 60 kHz.
71
3.4.2.2 Results and Discussion

Figure 3-8 (a) shows the pressure field map in the NI part at 60.5 KHz with the PI/NI interface

along x=0. The pressure amplitude is normalized to unity. A tight spot is observed in experiment

as is evident from the plot. The pressure cross the focal plane along y direction is plotted in

Figure 3-8 (c). The full width at half maximum (FWHM) was found to be 12.2mm,

corresponding to a resolution of 0.5 wavelength in water.

For numerical verification, lumped circuit simulation of this acoustic network was

performed by using commercial circuit simulator SPICE. Lumped circuit simulations were

performed by using commercial circuit simulator SPICE to study the ultrasound focusing in this

acoustic metamaterial network. The acoustic inductance and capacitance in the circuit are

determined from the geometry of the sample. To match the experimental condition, the circuit

model has the same unit cell number as the sample. A pulse of sine wave with a width of five

periods is launched on the center of the PI part [(column, row) = (20, 20)] as in the experiment.

A map of the field amplitude at each frequency of interest was obtained by Fourier

transformation (FT) pulses at each nodes and reading the FT amplitudes. Comparison of Figure

3-8 (a) and (b) shows that the field plots found through simulation is in remarkable agreement

with the experimental results. In Figure 3-8(c), the measured data in blue line is shifted to left by

3.175 mm for comparison purpose. The comparison demonstrates a very good match in the focus

width between the measurement and the numerical simulation. In Figure 3-9, the measured

72
Figure 3-8 Pseudo colormap of the normalized pressure field distribution at 60.5 KHz. (a)

Measured and (b) simulated field map of the acoustic NI metamaterial and (c) Line plot of

pressure field cross the focal plane parallel to interface.

73
 (a) (b)

(c) (d)

(ee)

Figure 3-9 Pseud

do colormaap of measu
ured pressurre amplitude distributioon at (a) 61
1.5 KHz (b))

74
62.5 KHz (c) 63.5 KHz (d) 64.5 KHz (e) 65.5 KHz

pressure fields at different frequencies are presented. The pressure amplitude is normalized to

unity. Tight spots are observed in these pressure maps. We also plotted the full width at half

maximum (FWHM) at different frequencies in Figure 3-10 (a). The optimal focus imaging is

observed at 60.5 kHz from both experimental and numerical results. The FWHM increases with

increasing frequency. The focal point moves closer to the PI/NI interface (at x=0) at higher

frequency.

(a) (b)

Figure 3-10 Measured and calculated FWHM of the focus as a function of frequency. Blue solid

line is calculated using acoustic circuit model and red circles represent experimental data. (b)

Measured and calculated focal length as a function of frequency. The analytical curve (blue solid

line) is based on ray propagation and the green dashed line is numerical results from acoustic

circuit model. Experimentally obtained data are shown by red circles.

75
Figure 3-11 The absolute value of effective refractive index as function of frequency. The

relative refractive index is -1 at 60.5 kHz.

The focal length defined as the distance between focus and PI/NI interface is plotted in

Figure 3-10 (b) as a function of frequency. Ray acoustics is utilized to estimate the focal length

as shown in the blue solid line. The magnitude of the refractive index in the acoustic

metamaterial decreases from 1.19 to 0.85 as the frequency increases from 56 to 66 KHz. And this

analysis predicts that the negative refractive index approaches -1 relative to the PI part at 60.5

KHz. The decrease of the index magnitude over this frequency range causes the focal length

decreasing from 79.27 to 37.6mm. The lumped circuit simulation gives the dashed green curve

while the red stars show the measurement data. The three curves present a good match in trend.

However, around 10 mm shift is observed.

In order to achieve high-quality focus imaging, the ratio of refractive index should be -1 at

the PI/NI interface. Only when the index is matched, based on ray acoustics, the angle of

refraction equals the angle of incidence for each ray such that all rays can be brought to the same
76
focal spot in the NI part. However, due to the loss and variation of the inductance and

capacitance from their design values resulted from machining tolerance; the refractive index is

not exactly matched in the measurement. Therefore, the distance between the focus and the

interface varies for different incident angle as result of aberration 8. Due to this index mismatch,

we observed that the focal spot elongated along x direction while remain narrow along the

direction parallel to the interface in the experiment. And the focus is in a position closer to the

interface than the source. The best focusing resolution is observed at 60.5 KHz. We expect that

the ratio of refractive index might approach -1 at lower frequency. However verification of this is

beyond the operation frequency range of our transducer in the experiment. The slight material

loss in the measurement also significantly degrades the focusing resolution as discussed in

51,52
several papers. It was noted that single PI/NI interface does not allow the enough growth of

8
evanescent fields to achieve sub diffraction focusing while sandwich structure (two PI/NI

interfaces) offers better chance to overcome the diffraction limit.10

3.4.3 Conclusion

In summary, the emission of a point source at kilohertz frequency was brought to a focus through

the PI/NI interface because of the negative refraction in this ultrasonic metamaterial, which is

expected to be a step toward a novel acoustic imaging lens. The resolution of 0.5 wavelength was

recorded by focusing the acoustic field of a point source. This is not sub diffraction imaging, but

among the best achievable passive acoustic imaging elements. The unit cell of the acoustic
77
network is only one eighth of the operating wavelength, making the lens in a compact size.

Compared with conventional lenses, the flat thin slab lens takes advantages in that there is no

need to manufacture the shapes of spherical curvatures and the focus position is insensitive to the

offset of source along the axis. Also this negative index lens offers tunable focal length at

different frequencies. More generally, this design approach may lead to novel strategies of

acoustic cloak for camouflage under sonar.

References

1
J. B. Pendry, “Negative refraction makes a perfect lens”, Phys. Rev. Lett. 85, 3966 (2000).

2
D. R. Smith et al, “Composite medium with simultaneously negative permeability and

permittivity”, Phys. Rev. Lett, 84, 4184 (2000).

3
R. A. Shelby, D. R. Smith and S. Schultz, “Experimental verification of a negative index of

refraction”, Science 292, 77 (2001).

4
S. Foteinopoulou, E. N. Economou, and C. M. Soukoulis, “Refraction in media with a negative

refractive index”, Phys. Rev. Lett, 90, 107402 (2003).

5
A. A. Houck, J. B. Brock and I. L. Chuang, “Experimental observations of a left-handed

material that obeys snell's law”, Phys. Rev. Lett. 90, 137401 (2003).
78
6
H. J. Lezec, J. A. Dionne, H. A. Atwater, “Negative refraction at visible frequencies”, Science

20,430 (2007).

7
N. Fang, H. Lee, C. Sun, and X. Zhang, “Sub-diffraction-limited optical imaging with a silver

superlens”, Science 308, 534 (2005).

8
G. V. Eleftheriades, A. K. Iyer, P. C. Kremer, “Planar negative refractive index media using

periodically L-C loaded transmission lines”, IEEE Trans. Microwave Theory and Tech. 50, 2702

(2002).

9
A. K. Iyer, P. C. Kremer, G. V. Eleftheriades, “Experimental and theoretical verification of

focusing in a large, periodically loaded transmission line negative refractive index metamaterial”,

Opt. Express 11, 696 (2003).

10
A. Grbic, G. V. Eleftheriades’, “Overcoming the diffraction limit with a planar left-handed

transmission-line lens”, Phys. Rev. Lett. 92, 117403 (2004).

11
C.Y. Qiu, X. D. Zhang and Z. Y. Liu, “Far-field imaging of acoustic waves by a

two-dimensional sonic crystal”, Phys. Rev. B 71, 054302 (2005).

79
12
X. D. Zhang and Z. Y. Liu, “Negative refraction of acoustic waves in two-dimensional

phononic crystals”, Appl. Phys. Lett. 85, 341 (2004).

13
S. X. Yang et al., “Focusing of sound in 3D phononic crystal”, Phys. Rev. Lett. 93,

024301(2004).

14
M. Z. Ke et al., “Negative-refraction imaging with two-dimensional phononic crystal”, Phys.

Rev. B 72, 064306 (2005).

15
A. Sukhovich, L. Jing and J. H. Page, “Negative refraction and focusing of ultrasound in

two-dimensional phononic crystals”, Phys. Rev. B 77, 014301 (2008).

16
Z. Liu, X. Zhang, Y. Mao et al., “Locally resonant sonic materials”, Science 289, 1734

(2000).

17
J. Li and C. T. Chan, “Double-negative acoustic metamaterial”， Phys. Rev. E 70, 055602

(2004).

18
Z. Y. Liu, C. T. Chan, and P. Sheng, “Analytic model of phononic crystals with local

resonances”, Phys. Rev. B 71, 014103 (2005).

80
19
Y. Q. Ding, Z.Y. Liu ,C. Y. Qiu and J. Shi , “Metamaterial with simultaneously negative bulk

modulus and mass density”, Phys. Rev. Lett. 99, 093904 (2007).

20
J. Li, Z. Liu, and C. Qiu, “Negative refraction imaging of acoustic waves by a

two-dimensional three-component phononic crystal”, Phys. Rev. B 73, 054302 (2006).

21
N. Fang, D. J. Xi, J. Y. Xu, M. Ambrati, W. Sprituravanich, C. Sun, and X. Zhang, “Ultrasonic

metamaterials with negative modulus”, Nat. Mater. 5, 452 (2006).

22
Veselago,V. G.,“The electrodynamics of substances with simultaneously negative values of µ

and ε,” Sov. Phys. Usp., Vol. 10, No. 4,509 (1968).

23
E. N. Economou and M. M. Sigalas, “Classical wave propagation in periodic structures:

Cermet versus network topology”, Phys. Rev. B 48, 13434 (1993).

24
M. S. Kushwaha, P. Halevi, L. Dobrzynski, and B. Djafari-Rouhani, “Acoustic band structure

of periodic elastic composites”, Phys. Rev. Lett. 71, 2022 (1993).

25
M. Kafesaki and E. N. Economou,” Multiple-scattering theory for three-dimensional periodic
81
acoustic composites”, Phys. Rev. B 60, 11993 (1999).

26
J. O. Vasseur, P. A. Deymier, B. Chenni, B. Djafari-Rouhani, L. Dobrzynski, and D. Prevost,

“Experimental and Theoretical Evidence for the Existence of Absolute Acoustic Band Gaps in

Two-Dimensional Solid Phononic Crystals”, Phys. Rev. Lett. 86, 3012 (2001).

27
Torres, M.; Montero de Espinoza, F.R.; Aragon, J.L. “Ultrasonic Wedges for Elastic Wave

Bending and Splitting without Requiring a Full Band Gap”, Phys. Rev. Lett. 86, 4282 (2001).

28
F. Cervera, L. Sanchis, J. V. Sanchez-Perez, R. Martinez-Sala, C. Rubio, and F. Meseguer,

“Refractive Acoustic Devices for Airborne Sound”, Phys. Rev. Lett. 88, 023902 (2002).

29
N. Garcia, M. Nieto-Vesperinas, E. V. Ponizovskaya, and M. Torres, “Theory for tailoring

sonic devices: Diffraction dominates over refraction”, Phys. Rev. E 67, 046606 (2003).

30
Bikash C. Gupta and Zhen Ye, “Theoretical analysis of the focusing of acoustic waves by

two-dimensional sonic crystals”, Phys. Rev. E 67, 036603 (2003).

31
M.Notonmi, “Theory of light propagation in strongly modulated phontonic crystals:

Refraction like behavior in the vicinity of the photonic band gap”, Phys.Rev.B 62 10696 (2000).

82
32
Luo, S. G. Johnson, J. D. Joannopoulos, and J. B. Pendry, “All-angle negative refraction

without negative effective media”, Phys. Rev. B 65, 201104 (2002).

33
E. Cubukcu, K. Aydin, E. Ozbay, S. Foteinopoulou, and C. M. Soukoulis, “Negative

refraction by photonic crystals” , Nature(London) 423, 604 (2003).

34
E. Cubukcu, et al “Subwavelength resolution in a two-dimensional photonic-crystal-based

superlens”, Phys.Rev.Lett.91, 207401(2003).

35
J. H. Page et al. “Phononic crystals”, phys. stat. sol. (b) 241, No. 15, 3454–3462 (2004).

36
J. P. Wolfe, Imaging Phonons: Acoustic Wave Propagation in Solids (Cambridge University

Press, Cambridge, (1998).

37
X. D. Zhang and Z. Y. Liu, “Negative refraction of acoustic waves in two-dimensional

phononic crystals”, Appl. Phys. Lett. 85, 341 (2004).

38
C.Y. Qiu ,X. D. Zhang and Z. Y. Liu. “Far-field imaging of acoustic waves by a

two-dimensional sonic crystal”, Phys. Rev. B 71, 054302 (2005).

83
39
S. Yang, J. H. Page, Z. Liu, M. L. Cowan, C. T. Chan, and P. Sheng, “Focusing of Sound in a

3D Phononic Crystal”,Phys. Rev. Lett. 93, 024301 (2004).

40
M. Ke, Z. Liu, C. Qiu, W. Wang, J. Shi, W. Wen, and P. Sheng, “ Negative refraction

imaging with two-dimensional phononic crystals”, Phys. Rev. B 72, 064306 (2005).

41
A Sukhovich, L Jing, J.H. Page, “Negative refraction and focusing of ultrasound in

two-dimensional phononic crystals”, Phys. Rev. B, 77, 014301 (2008).

42
V. Fokin, M. Ambati, C. Sun, and Xiang Zhang, “Method for retrieving effective properties of

locally resonant acoustic metamaterials”, Phys. Rev. B 76, 144302 (2007).

43
Z. Liu, X. Zhang, Y. Mao, Y. Y. Zhu, Z. Yang, C. T. Chan, and P. Sheng, “Locally Resonant

Sonic Materials”, Science 289, 1734 (2000).

44
Z. Liu, C. T. Chan, and P. Sheng, “Analytic model of phononic crystals with local resonances”,

Phys. Rev. B 71, 014103 (2005).

45
Y. Wu, Y. Lai, and Z. Q. Zhang, “Effective medium theory for elastic metamaterials in two

84
dimensions”, Phys. Rev. B 76, 205313 (2007).

46
Y. Ding, Z. Liu, C. Qiu, and J. Shi, “Metamaterial with Simultaneously Negative Bulk

Modulus and Mass Density”, Phys. Rev. Lett. 99, 093904 (2007).

47
L. E. Kinsler, Fundamentals of Acoustics, 1982, Wiley, New York

48
G. W. Stewart, “Acoustic wave filters”, Phys.Rev.20,528 (1922).

49
W. P. Mason, “A study of the regular combination of acoustic elements, with applications to

recurrent acoustic filters, tapered acoustic filters, and horns”, Bell Syst. Tech. J . 6, 258 (1927).

50
L.L.Beranek, Acoustics,1954,McGRAW-HILL,New York

51
D. R. Smith et al, “Limitations on subdiffraction imaging with a negative refractive index

slab”, Appl. Phys. Lett. 82 ,1506 (2003).

52
N. Fang and X. Zhang, “Imaging properties of a metamaterial superlens”, Appl. Phys. Lett. 82,

161 (2003).

85
 4 BROADBAND ACOUSTIC CLOAK FOR ULTRASOUND WAVES

4.1 Introduction

Recently, a new design paradigm called conformal mapping and coordinate transformation has

inspired a series of key explorations to manipulate, store and control the flow of energy, in form

of either sound, elastic waves or light radiation. In electromagnetism, because of the coordinate

invariance of Maxwell’s equations, the space for light can be bent in almost arbitrary ways by

1,2
providing a desired spatial distribution of electric permittivity ε and magnetic permeability μ.

A set of novel optical devices were proposed based on transformation optics3,4,5; they usually call

for complicated medium with anisotropic and spatially varying permittivity and permeability

tensor to accomplish the desired functionality. Recent advances in synthetic structured

metamaterial 6 , 7 whose properties are determined by its subwavelength structure, offers the

potential to physically implement these complicated media. By modifying the shape and

arrangement of these subwavelength constituent elements, anisotropy and spatial variation can be

achieved in the artificial metamaterials.

Among the most exciting examples is perhaps an electromagnetic cloak that can render the

objects invisible. The first experimental demonstration of such a cloak was reported in

microwave using structured metamaterial composed of metallic resonant rings8. However, the

invisibility effect was only obtained in a narrow frequency range because of the strong dispersion

inherent to the resonant elements used to build the cloak. In addition, such resonances led to

undesired material absorption in the cloak. To mitigate these constrains, several different

86
 9,10,11,12,13
schemes of cloaking utilizing non-resonant structure were proposed .One example is a

so-called carpet cloak which compresses the cloaked space into a thin sheet14,15. However, the

waves travel faster through the carpet cloak than through the outer space, such a faster-than-light

speed thereby sets a fundamental restriction for broadband application in ambient air. Therefore,

the experiments of these carpet cloaks were so far conducted in a dielectric medium with higher

index.

In contrast, cloaking of other classical waves such as acoustic waves do not suffer from

such limitation for electromagnetic cloaks. 16 , 17 , 18 However, in general the elastodynamic

equations do not have this invariance symmetry as proven by Milton et al19. Fortunately, acoustic

waves in fluids follow such form invariance and several theoretical schemes of transformation

20,21,22
have been proposed. Theoretical analysis of an acoustic cloak23was reported based on the

equivalence between transverse electric electromagnetic waves and acoustic waves in a

two-dimensional (2D) geometry. Yet, this 2D acoustic cloak requires anisotropic mass density

which is not common in naturally-occurring materials. Consequently the experimental studies of

acoustic cloak have been hampered by the difficulty in creating suitable materials and so far

remain challenging.

In this thesis, we overcome the above challenges in acoustic cloak design by introducing an

acoustic transmission line approach. By taking the analogy between lumped acoustic elements

and electronic circuit elements, this transmission line approach enabled a new class of acoustic

metamaterials24and ultrasound focusing through a metamaterial network. 25

As a demonstration,

we designed a 2D cylindrical cloak in order to hide an object in the center. This acoustic
87
cylindrical cloak is implemented by a 2D array of sub-wavelength cavities and connecting

channels with spatially tailored geometry. The acoustic wave propagation through this discrete

network can be described by a set of telegrapher’s equations. This approach enables the

realization of acoustic metamaterials with simple structure, ease of manufacturing and scaling,

offering the potential to achieve a variety of acoustic devices based on transformation. Moreover,

the acoustic cloak is expected to be low-loss and broadband with the use of non-resonant

constituent elements.

4.2 Optical Transformation

4.2.1 Rules of Covariant and Contravariant Transformation

In mathematics and theoretical physics, covariance and contravariance refer to how coordinates

change under a transformation. A covariant transformation is a rule that describes how certain

physical entities change under a change of coordinate system and the inverse transformation is

called as contravariant transformation. The general formulation of covariance and contravariance

guarantees that any vector should be invariant under a coordinate transformation.

If e1, e2, e3 are contravariant basis vectors of R3 (not necessarily orthogonal nor of unit norm)

then the covariant basis vectors of their reciprocal system are:

(4-1)
Note that ei and ei are mutually orthonormal

(4-2)

the contravariant coordinates of any vector v can be obtained by

88
 (4-3))

Likewisse, the covaariant compoonents of v can be obtaained

(4-4))

So v can
n be expressed as,

o
or (4-5))

Then, (4-6))

Thus th
he vector cann be easily converted between
b covvariant to coontravariantt basis

and (4-7))

If the contravariannt basis vecttors are orthhonormal thhen they aree equivalentt to the cov
variant basiss

vectors

4.2.2 Invariancce in Electrromagnetism

Assumee the coorddinate transformation as from a Cartesiann coodinates , , to

o a generall

system , ,

, , , , , , , , (4-8))

The len
ngth of a linee element iss given by

+2 +2 2
+2 (4-9))

Where

89
Define three units vectors , , , to point along the generalized , , axes .The length

of a vector along the axis direction can be interpreted as , where

Write the three component of the E field in new general coordinate system

· , · , · (4-10)

Figure 4-1 Small element in the general coordinate system , , .

Figure 4-2 Integration path for .

By stokes’ theorem, the line integration of E field can be written as,

· (4-11)

This equation can be further simplified

· (4-12)

Where , ,

The relation of the contravariant components and covariant components of H is

90
 · · ·
· · · (4-13)
· · ·

Or ∑ (4-14 )

From definition

· , · , · (4-15)

(4-16)

Use Faraday's law, the right side of (4-12) can be modified

· ∑ · (4-17)

Define ̂ Λ | · |

· ∑ (4-18)

Substitute (4-12)

∑ µ (4-19)

Similarly

∑ (4-20)

Where ̂ Λ | · |

Note that when the general coordinate system , , is orthogonal

| · | (4-21)

91
So the Maxwell’s equations are of the same form in any coordinate system but the values of

permittivity and permeability will change. This implies that, mathematically, the description of

the permittivity and permeability has the same effect on the electromagnetic fields as a

coordinate transformation.

4.2.3 Transformation Media and Invisible Cloak

It is known that if light travels in a gradient index medium, where the refractive index changes as

a function of position, the light ray will follow a curved path determined by the nature of the

gradient. Therefore by controlling the permittivity and permeability distribution in space, the

propagation path of electromagnetic wave can be controlled. The question is how to obtain the

precise description of the medium property as function of position to achieve the desired

redirection of optical rays.

Low index

High index

Figure 4-3 Wave propagation in a gradient medium.

The coordinate invariance of Maxwell’s equations we discussed in previous section enables one

possible design strategy to provide the prescription of the gradient medium to control and

manipulate wave propagation path. It was demonstrated that the arbitrary distortion of wave

fields can be recorded as a coordinate transformation. On the other hand, the desired

92
transformation can be equivalently achieved by providing the transformation materials in the

engineered space. Transformation media is defined as such that implement the same coordinate

transformations effect in Maxwell's equations. The design approach is quite straightforward.

Assume there are two different coordinate systems and two different spaces. In the first space,

which is named as virtual space, the Maxwell equations are usually written in Cartesian

coordinates in free space with a simple configuration of permittivity and permeability. Then we

imagine the Cartesian coordinate lines were distorted and the deformed coordinates could be

described by a coordinate transformation. The Cartesian grid in the virtual space will deform

under the transformation and this deformed grid shows ray trajectories in the physical space.

Next rewrite Maxwell’s equations using the new coordinate system in the new space, called as

physical space. In previous section, we presented that the Maxwell’s equation are of the same

form in any coordinate system but the values of permittivity and permeability will change under

the transformation. By providing the straightforward prescription for a material, the fields in the

physical space which take up the distorted configuration can be obtained. If we can make the

material, we can accomplish the same result in terms of controlling the trajectories of the rays of

light. This has spawned a new field called transformation optics, in pursuit of harness exotic

optical device. For example, the waves can be focused as desired or steered to avoid scattering

objects and flow around them like a fluid, recombined to their original trajectories.

93
Figure 4-4 (A) A field line in free space with the background Cartesian coordinate grid shown.

(B) The distorted field line with the background coordinates distorted in the same fashion. 1

Usually, the transformation will lead to complex material properties such as anisotropy and

inhomogeneous that doesn’t exist in nature. However, thanks for the development of

metamaterial, an extraordinary range of electromagnetic properties can be engineered in

artificially structured metamaterials, with a flexibility unmatched by any conventional material.

An extraordinary range of electromagnetic properties can be engineered in artificially structured

metamaterials, with flexibility unmatched by any conventional material. Taking full advantage of

this enormous flexibility has led to a variety of application based on transformation optics, such

as perfect lens,26 and magnification.27 Among all these applications, the invisible cloak might

be one of the most interesting one.

Invisibility cloaks work by steering light around a region of space and recombine on the

other side to the original trajectory, making any object inside that region and the cloak invisible.

In effect, an invisibility cloak creates the illusion the light had passed through the empty volume

of space. In order to do so, in two dimensional cases, the idea is to make a coordinate

94
transformation that takes a point in space and expands it into a cylinder the interior of which is

invisible to an observer on the outside. So no radiation can get into the concealed volume, nor

can any radiation get out. In two-dimensional (2D) a simple transformation that can realize this

goal is to compress all fields in the region 0 into the annual region in a

cylindrical coordinate

/ (4-22)

(4-23)

(4-24)

By coordinate transformation, we can obtain the prescription for the cloak in as

(4-25)

(4-26)

(4-27)

By using material with above prescription, the TM field will exclude all fields from the

central region in the engineered space and such make the object together with the cloak invisible.

4.3 Acoustic cloak

It is undoubtedly of great interest to extend the design approach based on transformation-based

solutions to waves in other systems, for example acoustic waves. The electromagnetic and other

wave system shares the universal concepts of wave vector, wave impedance, and power flow.

95
And the manner in which permittivity and permeability control these in electromagnetics is

closely analogous to that by material properties in other wave system. However, the special

symmetry in Maxwell equations is an important element of the equivalence between coordinate

transformations and material properties .The field equations for other waves may not usually

process this property. It was demonstrated that in general the equation of motion in elastic media

does not have form invariance. However, in 2D case, the acoustic waves in fluid and Maxwell

equations with one single mode have identical form under certain variable exchange.

In 2D cylindrical coordinate with z invariance, the time harmonic acoustic equation for fluid

with anisotropic but diagonal in the cylindrical coordinate are interpreted as

∂P (4-28)
= − jωρφ uφ
r∂φ

∂P (4-29)
= − jωρ r ur
∂r
1 ∂ 1 ∂uφ (4-30)
(rur ) + = − jωβP
r ∂r r ∂φ

Where is scalar pressure, , are particle velocity, , are density along and

direction respectively, is the compressibility. There is an only longitude wave in this

anisotropic medium.

In the same z invariant cylindrical coordinate, mode Maxwell’s equations for transverse

magnetic (TM) mode are

∂Ez (4-31)
= iωμ r H r
r∂φ

∂Ez (4-32)
= −iωμφ H φ
∂r
1 ∂ 1 ∂H r (4-33)
(rH φ ) − = −iωε z E
r ∂r r ∂φ
96
 The medium has anisotropic permittivity and permeability, but the tensor only has diagonal

component.Observing the above two sets of equations are equivalent under the following

variable exchange

E z → − p H φ → v r H r → −vφ μ r → ρ φ μφ → ρ r ε z → β (4-34)

In terms of boundary condition, in electromagnism, the normal component and

tangential component and continuous, while at a fluid interface, the normal component of

particle velocity and pressure are continuous. Thus the boundary conditions are preserved

as well under the variable exchange. This implies that the 2D acoustic equations have the same

form invariance under coordinate transformation as Maxwell’s equations; consequently, we can

apply the design approach of transformation optics approach in 2D acoustic scheme.

The invisible cloak can thus be realized acoustically if we can implement a gradient medium

with anisotropic density

(4-35)

(4-36)

(4-37)

Where , are the density and compressibility of the background medium. and are

the radius of the inner and outer boundary of the 2D cloak.

97
4.4 Numerical Simulation of Acoustic Cloak Based on Transmission Line Model

In order to build the acoustic cloak as described by (4-35)-(4-37), acoustic transmission line

approach is employed by cascading the unit cell in Figure 4-5 (d) in a lattice configuration which

is diagonal in a cylindrical basis. The distributed acoustic system corresponding to the circuit

unit cell can be implemented by one large cavity in the center with four channels connecting to

the four neighboring blocks, which will be discussed in more details in the following section.

In such a two-dimensional network, the cloak is approximated by sixteen homogeneous

concentric cylinders. The effective anisotropic density and compressibility in these rings was set

as a 16-step piecewise constant value as an approximation of the continuous medium. From the

first to the fourth layers, the unit cell size along radial direction is λ/7, λ/8, λ/9 and λ/9 (λ is the

wavelength in water at 60 kHz) respectively. Afterwards, the cylinder layers are evenly spaced

with the distance equal to λ/10 along the radial direction. On the other hand, the first cylinder

near the inner lining of the cloak is divided to 32 units around the circumference. To keep the

size of the constituent element smaller than λ/10 along circumferential direction, starting from

the second layer, the number of cells is doubled to 64, and further increased to 128 from the sixth

layer. Software SPICE is used to simulate the pressure field through the lumped acoustic circuit.

The calculation domain is presented in Figure 4-5(a). A point source is incident on an object

(0.67λin diameter) surrounded by a cloaking shell of thickness 1.6λ. The cloaked object is put in

center of water with circular boundary.

98
 (a) (b)

(c) (d)

Figure 4-5 (a) Computational domain (b) mesh model (c) one quarter of the mesh model (d)

circuit unit at each mesh node.

Figure 4-6 (a) and (b) show the scaled pressure filed amplitude for the cases with and

without cloak respectively. The x and y coordinate is normalized by the wavelength in water.

Comparison shows the wave is guided around the object in the shell and merged behind the

object with obvious reduction in back scattering. Nearly zero amplitude filed is observed inside

the cloak as expected. Continuous medium simulations of 2D cylindrical cloak were performed

by COMSOL Multiphysics finite element based solver. Figure 4-6 (c) and (d) are simulation

results for when there is cloak and when there is only object respectively.

99
 (a) (c)

(b) (d)

Figure 4-6 Pseudo colormaap of scaled

d pressure filed
f distrib
bution due tto a point source
s with
h

cloak by
y (a) lumpeed circuit an
nd (b) FEM
M simulationn. Pressure filed
f distribbution witho
out cloak by
y

(c) lum
mped circuit and (d) FE
EM simulattion. The black circle marks the outer boun
ndary of thee

cylindriical cloak.

100
 (a) (b)

(c)

Figure 4-7 Pseudo colormap of phase filed distribution due to a point source (a) in a free field

and (b) with a scattering object (c) with a cloaked object. The black circle marks the outer

boundary of the cylindrical cloak.

101
 Comparison with FEM model demonstrates that the discretized cloak based on lumped

circuit model can force the incident wave energy to bend around the central object. This allows

for a practical realization of the acoustic cloak. In continuous model, there is less distortion is

observed compared with lumped circuit simulation. One reason is reflection from the boundary

in circuit model. In FEM, a perfect match layer is used to achieve zero reflection from the

boundary. On the other hand, the mesh is much finer in FEM model. Another reason is additional

absorption loss which degrades cloaking performance is added in the circuit model as a small

resistor connected to each inductor.

Figure 4-7(a) presents the phase distribution of the free field when there is no object or cloak.

When an object is put in the center of the field, obvious scattering is observed .However, there is

object is put inside the cloak, it is observed that the field behind the cloaked object is exactly

same with the one when there is only free field.

We numerically demonstrated a 2D acoustic shell cloaking by lumped acoustic transmission

line design. The artificial anisotropic acoustic medium is built by an array of designed Helmholtz

resonators. The simulation results proved that the acoustic wave can be stretched and guided

through the cloak without penetrating into the object.

102
4.5 Irregular Transmission Line Network

Figure 4-8 One unit cell for a nonorthogonal transmission line

In order to conform the transmission line network to more general geometry and boundaries, the

regular Cartesian lattice is further extended to irregular nonorthogonal structured grids. Further,

it is often desirable to have a refined lattice in localized regions. With uniform lattice, this results

in global refinement of the mesh density, leading to unnecessary increase in the computional

effort and memory resources. This can be solved by nonuniform grids.

Consider a locally curvilinear coordinate system, the linear acoustic wave equations still

holds

du
ρ = −∇ p
dt
r 1 ∂p
∇ • (ρ0u ) = − 2
c0 ∂ t (4-38)

Use cell of arbitrary parallelepipeds shown in black lines to generate a lumped circuit

network. The pressure is collocated at the grid cell vertices black nodes while the particle

velocity is defined along edges of cells.

For the first equation which is linear Euler’s equation, use the finite difference expression for

103
∫ρ −
•=
udS ∫
ic2 p

the space derivatives and assume harmonic acoustic wave

pi, j − pi +1, j = −iωρi, j ui +1/ 2, j gi +1/ 2, j

(4-39)

pi, j −1 − pi, j = −iωρi, j ui, j −1/ 2 gi, j −1/ 2

(4-40)

pi, j − pi, j +1 = −iωρi, j ui, j +1/ 2 gi, j +1/ 2

(4-41)

pi −1, j − pi, j = −iωρ i, j ui −1/ 2, j gi −1/ 2, j

(4-42)

Where , is the edge length shown as black lines.

The second continuity equation can be written in integral form use Gauss’s theorem

(4-43)

This equation presents that the net flow with which mass flows into the volume through its

surface must equals the rate with which the mass within the volume increases. And the adiabat

tells .

Define a dual cell associated with each vertices where the pressure defines; which is

indicated in dotted blue lines. Consider a small parallelepiped volume

, , ∆ (4-44)

Where , is the area of the dual cell inside the dotted blue lines. ∆ is the thickness of

the volume with z direction perpendicular to the plane. The continuity equation can be written in

following discrete form

104
 ui +1/ 2, j ⋅ ρi , j ⋅ li +1/ 2, j ⋅ cos φi +1/ 2, j ⋅ Δz − ui , j −1/ 2 ⋅ ρi , j ⋅ li , j −1/ 2 ⋅ cos φi , j −1/ 2 ⋅ Δz
ui , j +1/ 2 ⋅ ρi , j ⋅ li , j +1/ 2 ⋅ cos φi , j +1/ 2 ⋅ Δz − ui −1/ 2, j ⋅ ρi , j ⋅ li −1/ 2, j ⋅ cos φi −1/ 2, j ⋅ Δz
ω
= −i pi , j ⋅ Si , j ⋅ Δz = −iωρi , j βi , j pi , j ⋅ Si , j ⋅ Δz
c2 (4-45)

We want to use a lumped circuit network to simulate the above acoustic equations. The

circuit consists of serial of impedance and shunt admittance. The telegraph’s equations for this

circuit network are

−Ii+1/2, j +Ii, j−1/2 +Ii−1/2, j −Ii, j+1/2 =Vi, j ⋅Yi, j

(4-46)

Vi, j − Vi +1, j = I i +1/ 2, j ⋅ Ri +1/ 2, j

(4-47)

Vi , j − Vi , j +1 = I i , j +1/ 2 ⋅ Ri , j +1/ 2
(4-48)

Vi −1, j − Vi , j = I i −1/ 2, j ⋅ Ri −1/ 2, j

(4-49)

Vi, j −1 − Vi , j = I i , j −1/ 2 ⋅ Ri , j −1/ 2

(4-50)

The two sets of equations are exact duality of each other under the variable exchange

, , , (4-51)

, ,
, (4-52)
, ,

, , (4-53)

, · , · , , (4-54)

This circuit model can be extended to simulate electromagnetic wave with single

polarization, for example TM mode. Collocate the component at grid cell vertices while

105
define the component tangent to the edge of the cells. Apply the same discretization approach

to integral form of Faraday’s equation and Ampere’s law

r r ∂ r r
∫ H ⋅ dl =
∂t ∫
D ⋅ dS (4-55)

r r ∂ r r
∫ E ⋅ dl = − ∂t ∫ B ⋅ dS (4-56)

H i +1 / 2, j ⋅ li +1 / 2, j + H i , j −1 / 2 ⋅ li , j −1 / 2 − H i −1 / 2, j ⋅ li −1 / 2, j − H i , j +1 / 2 ⋅ li , j +1 / 2 = iωεEi , j ⋅ S i , j (4-57)

Ei, j Δz − Ei, j +1Δz = iωμ H i, j +1/ 2 ⋅ g i, j +1/ 2 ⋅ Δz ⋅ cosφi, j +1/ 2 (4-58)

Ei , j Δz − Ei +1, j Δz = −iωμ H i +1/ 2, j ⋅ g i +1/ 2, j ⋅ Δz ⋅ cos φi +1/ 2, j (4-59)

E i −1, j Δ z − E i , j Δ z = −iωμ H i −1 / 2 , j ⋅ g i −1 / 2 , j ⋅ Δ z ⋅ cos φi −1 / 2 , j (4-60)

Ei, j −1Δz − Ei, j Δz = iωμHi, j −1/ 2 ⋅ gi, j −1/ 2 ⋅ Δz ⋅ cosφi, j −1/ 2

(4-61)

The equation set is also equivalent with the circuit equations under the variable change

Ei , j ~ Vi , j iωεSi , j ~ Yi , j (4-62)

iωμ g i +1/ 2, j cos φi +1/ 2, j (4-63)

H i , j −1/ 2 ⋅ li , j −1/ 2 ~ I i , j −1/ 2 Ri +1/ 2, j =
li +1/ 2, j

iωμ g i −1/ 2, j cos φi −1/ 2, j

H i , j +1/ 2 ⋅ li , j +1/ 2 ~ I i , j +1/ 2 Ri −1/ 2, j =
li −1/ 2, j (4-64)

H i +1/ 2, j ⋅ li +1/ 2, j ~ − I i +1/ 2, j iωμ g i , j −1/ 2 cos φi , j −1/ 2

Ri , j −1/ 2 = (4-65)
li , j −1/ 2
iωμ g i , j +1/ 2 cos φi , j +1/ 2
H i −1/ 2, j ⋅ li −1/ 2, j ~ − I i −1/ 2, j Ri , j +1/ 2 =
li , j +1/ 2 (4-66)

106
 The nonorthogonal circuit network is employed to simulate a cylindrical acoustic cloak in a

background medium with square boundary, which is water here. At each node, there are four

serial inductors and one shunt capacitors. The mesh of the background water is divided into two

parts. One part closer to the cloak is discretized by nonorthogonal mesh. On the other hand, the

background is meshed by uniform square mesh away from the cloak. The cloak is meshed by

nodes along the diagonal direction of a cylindrical coordinate. Similar to the background medium,

the object is composed of nonuniform mesh closer to the cloak and square mesh in the center. In

Figure 4-9 (c), the red dots marked the nodes of the dual cells. The circuit cells shown in Figure

4-9 (d) and (e) are the employed for cloak and the square mesh respectively.

A plane acoustic wave filed impinges from the left side boundary of calculating domain. The

radius of object is one wavelength and the acoustic impedance of object is set as seven times of

the water. The thickness of cloak is 1.6 wavelengths. The pressure field at steadystate is shown in

Figure 4-10 (a). Nearly zero amplitude is observed inside the cloak. The phase distribution is

shown in Figure 4-10 (b). Inside the cloak shell, the wave is bent around the object as merged

behind the object. The wave is recombined behind the cloaked object to original plane wave

front. To keep the circuit stable, small number of resistor is added into the circuit in serial with

the inductors. This resulted in strong acoustic intensity attenuation as observed in Figure 4-10 (a).

The distortion of the phase front near the four corners of the cloak is mainly caused by the coarse

nonorthogonal mesh surrounding the cloak. Due to the limit of the software, we cannot further

refine this part of mesh.

107
 (a)

(b) (c)

(d) (e)

Figure 4-9 (a) Computational domain (b) mesh model (c) and center part of the mesh model (d)

one circuit unit at each mesh node for cylindrical and (e) rectangular mesh.

108
 (a) (b)

(c) (d)

Figure 4-10 Pseudo colormap of (a) a scaled pressure field amplitude and (b) phase distribution

in the vicinity of the cloaked object (c) a scaled pressure field amplitude and (d) phase

distribution in a free water space.

Figure 4-10 (c) and Figure 4-10 (d) show the pressure field amplitude and phase

distributions at steady state of a free water space with the same mesh nodes as in Figure 4-10(a).

Same field attenuation is observed in Figure 4-10 (c) while the much less distortion in phase
109
front is presented in Figure 4-10 (d) compared with Figure 4-10 (b).

4.6 Experimental Study of Acoustic Cloak Based on Transmission Line Model

4.6.1 Experimental Setup and Data Acquisition

In our design, the 2D acoustic metamaterial cloak is designed to squeeze the cylindrical region

0<r<R2 into an annular region R1<r’<R2 where r and r’ are the radial coordinate in the original

and transformed system respectively. The acoustic waves are thus excluded from the extended

volume and smoothly bent inside the cloak, with no perturbation of exterior field. We choose the

inner and outer radius of the cloak as R1=13.5mm, R2=54.1mm.Because of the form invariance

of the 2D acoustic wave equation, this given warping of space can be achieved by providing the

desired distribution of the mass density and compressibility in the annular region interpreted as

.To facilitate the experimental realization of the

cloak, we used the simplified functional form of the material parameters as 0.5

0.5 2 as plotted in Figure 4-12 (b) which can keep the same

wave behavior inside the cloak due to the scalar wave nature of the 2D acoustic wave in fluid .

The acoustic metamaterial cloak with the above material specification is physically synthesized

by a planar network of acoustic circuits machined in an aluminum plate as shown in Figure 4-12

(a). These building blocks are cascaded in a lattice configuration which is diagonal in a

cylindrical basis. In such a topology, the cloak is approximated by sixteen homogeneous

concentric cylinders. From the first to the fourth layers, the unit cell size along radial direction is

λ/7, λ/8, λ/9 and λ/9 (λ is the wavelength in water at 60 kHz) respectively. Afterwards, the

110
cylinder layers are evenly spaced with the distance equal to λ/10 along the radial direction. On

the other hand, the first cylinder near the inner lining of the cloak is divided to 32 units around

the circumference. To keep the size of the constituent element smaller than λ/10 along

circumferential direction, starting from the second layer, the number of cells is doubled to 64,

and further increased to 128 from the sixth layer. The geometry of the sample is listed in table 1.

For all the unit cells, the depth and width of the channels , , , are constant value of

0.5mm. The inner radius of the cloak is 13.5mm. We noted that with the same design approach,

cloak with larger inner radius is feasible, but with much larger number of unit cells needed to

build the cloak.

(a) (b)

Figure 4-11 Schematic diagram of the experimental setup. A burst of monotonic signal with a

width of twenty periods was used to drive the transducer as an underwater point source in the

water tank. One needle-sized hydrophone detected the ultrasonic signals in the immediate

environment of (a) the object and (b) the cloaked object.

111
In our experiment, the water-filled network structure behaves as an effective anisotropic medium

for incoming underwater ultrasound. In each unit cell as shown in Figure 4-12 (b), the cavity

with large volume in center works as an acoustic capacitor whereas the channels connecting it to

28,29,30
the four neighboring cavities act as serial inductors . As each unit cell is only around

one-tenth of the wavelength at operating frequency of 60 kHz, the network can be characterized

as an effective medium with correspondence between effective density and serial inductance

(L ,L ) and between compressibility and shunt capacitance (C). The relation can be interpreted
Lφ Sφ
in forms of ρeff ,r = Lr Sr ρeff ,φ = and βeff = C .This analogy effectively describes the
Δr , rΔφ Sr Δr

realization of the metamaterial cloak with the spatially varying parameter profile by tailoring the

geometry of the corresponding building blocks as listed in Figure 4-12 (c) . Please note that the

effective mass density ρ eff , r is mainly a measure of response through water flow in the narrow

channels to the pressure gradients along the radial direction. Since the volume of water in such

narrow channels is very small with respect to the whole unit cell, the effective mass density

along the radial direction for the building block appears to be smaller than the density of bulk

water, which is rather a constant. A similar argument can be applied for the effective density

ρeff ,φ along the angular direction. Therefore the anisotropy of mass density is readily introduced

in our metamaterial cloak. On the other hand, the effective medium displays an effective

dynamic compressibility larger than that of filling water, because it depends on the compliance

of the large cavity in between the connecting channels. The combination of the above low

effective density and large compressibility give rise to a large sound speed (about 5.5 times of

the speed of sound in water) along angular direction at the inner boundary of the cloak, without
112
relying on resonance. Such a large speed is critical for rerouting the paths of underwater sound

around the cloaked object without significant scattering. Such non-resonant scheme is not yet

achieved in the optical cloaking experiments.

(a)

Layer (mm)

1 2.05 0.10 3.00

3 1.37 0.22 2.29
5 1.24 0.41 2.06
7 1.24 0.30 2.06
9 1.24 0.41 2.06
11 1.24 0.52 2.06
13 1.24 0.63 2.06
15 1.24 0.74 2.06

(b) (c)

Figure 4-12 A 2D acoustic cloak for underwater ultrasound waves. (a) The configuration of the

acoustic cylindrical cloak synthesized by an acoustic transmission line, namely serial inductors
113
and shunt capacitors. The inset is the expanded view of the network. The cavities with large

volume work as shunt capacitors and those cavities are connected by narrow channels that act as

the serial inductors. (b) One building block of the acoustic circuit, each unit cell consists of one

large cavity in the center with channels connecting to the four neighboring blocks. The reduced

cloaking parameters are used in the design. The density , compressibility have constant

values and increases as radius changes from R1=13.5mm to the R2=54.1mm. (c) The

geometry parameters of the building blocks in the layers with odd number are presented in the

table. The depth and width , and , of the channels along radial and angular

directions have constant values of 0.5mm.

To demonstrate the shielding phenomena, the sample in Figure 4-13 is placed in a water tank

to measure the pressure fields in the immediate environment of the cloaked object to compare

with those without cloak. The tank edge is filled with absorbing rubber to reduce reflection.

Because of the high impedance contrast between water and air as well as between water and

glass, the system provides a 2D waveguide to confine the ultrasound wave propagation. Figure

4-14 presents the pressure field distribution at 60 kHz without any object in the water tank. The

side of the cloak machined with the network structure is placed against the bottom of the tank.

The cloak has a thickness of 3mm with the depth of the cavities smaller than 1.36 mm. The water

inside the cloak is connected to the surrounding water which is 1.5 mm deep through the

channels along the radial direction around the outer boundary of the cloak.

114
 Figure 4-13 2D acoustic cloak based on acoustic circuit network

The ultrasound signal from a spherical shape transducer is launched to the water as a point

source. A waveform generator (Tektronix AFG 310) is used to drive the transducer. The source

generated a burst of sine waves with a width of 20 periods. The pressure field around the cloak

sample in the water is detected by a miniature hydrophone (RESON TC4038-1), amplified by

Stanford research systems model SR650 and captured using a digital oscilloscope (Agilent

DSO6104A) and then downloaded to a computer for post processing and analysis. The

hydrophone is attached to a motorized translation stage. The control program of a customized

LabVIEW scans across the data acquisition region by moving the hydrophones in a small

increment 3mme to record the spatiotemporal distribution of the pressure field. The snapshot of

the field pattern can be plotted as a function of position. To verify the broad operational

bandwidth of the acoustic cloak, the transducer is excited over a discrete set of frequencies to

illuminate the sample. The transducer operating spectrum limits us to test the frequency range
115
from 52 kHz to 64 kHz.

Figure 4-14 Measured acoustic wave field distribution with a point source at 60 kHz in the water

tank without any objects

4.6.2 Results and Discussion

For experimental confirmation of the cloaking performance, we placed an object in a water tank

and compared the wavefronts of propagating ultrasound in our measurement, with or without the

presence of our cloak as shown in Figure 4-11. The object is a steel cylinder with size equal to

the inner radius of the cloak. The side of the cloak machined with the metamaterial network was

placed against the bottom of the tank in order to seal water inside. The ultrasound waves were

launched from a spherical shaped transducer as a point source with distance of 165mm (about 6.5

wavelengths) away from the center of the cloak. To map the pressure field, a hydrophone was

mounted on a horizontal linear translation stages to scan in x-y directions. By stepping the

hydrophone in small increments of 3mm and recording the acoustic pulse signal from the water
116
 (d)
(a)

(b) (e)

(c) (f)

117
Figure 4-15 Measured pressure field mappings of the bare steel cylinder and the cloaked steel

cylinder illuminated with a point ultrasound source. The cloak lies in the center of the water tank

and surrounds the steel cylinder. The scattering field patterns of the bare steel cylinder at (a) 60

kHz (b) 52 kHz and (c) 64 kHz. The pseudo colormaps in the immediate environment of the

cloaked steel cylinder at (d) 60 kHz (e) 52 kHz and (f) 64 kHz.

at every step, we acquired the 2D spatial field distribution of the ultrasound wave scattering

pattern.

It is evident in our experiment that the presence of steel cylinder alone in the water tank

produces considerable scattering and shadowing at 60 kHz as shown in Figure 4-15 (a). By

surrounding the steel cylinder with the metamaterial cloak in Figure 4-15 (d), however, the wave

trajectory was restored behind the cloak with diminutive distortion in the cylindrical wavefronts,

making the cloak and the hidden cylinder invisible under the hydrophone. Very small attenuation

of the transmitted fields is observed on the exit side of the cloak, demonstrating the low-loss

nature of the metamaterial cloak based on transmission line model.

To demonstrate the broadband nature of our designed cloak, the acoustic wave field

distributions at 52 kHz and 64 kHz are presented in Figure 4-15 (b) (c) (e) (f) for both cases with

and without cloak. The field maps from these measurements present similar cloaking behavior

with those at 60 kHz. This is not surprising since our metamaterial cloak is constructed by

non-resonant elements. Theoretically, the cloak is expected to operate over a wide frequency

range of 40 to 80 kHz. In fact, at frequency below 40 kHz, the scattering from the object with
118
radius of 13.5mm become negligible. At the same time, the effectiveness of cloak at high

frequency is restricted by two factors. The first is the breakdown of the effective medium

approximation at 120 kHz when the unit cell is comparable to one quarter of wavelength. By

using smaller size of unit cells, this limit can be lifted to higher frequency. The other limit is the

cutoff frequency at 80 kHz due to the low-pass topology of the circuit network. By modifying

the geometry of the building block, this cutoff frequency can be potentially extended. However,

in the current experiment we can only verify the cloaking behavior from 52 to 64 kHz in the

experiment due to the limited operating frequency range of the transducer. Figure 4-16 presented

the acoustic wave field maps at different frequency to demonstrate the cloaking functionality.

The scattering shadow caused by the steel cylinder as shown in Figure 4-16 (a-d) is reduces by

the surrounding cloak in Figure 4-16(e-h) at frequency 54, 56, 58 and 62 kHz respectively.

To further quantify the reduction of scattering and shadowing of the cloaked object, we

conducted a set of measurements over different frequencies. The peak values of pressure along

the wavefronts behind the cloak were obtained using a MATLAB program to process the

experimental data. To facilitate the comparison of the cloaking performance, we defined the
P , P ,
averaged visibility of an object as ∑ , where ,P , and P ,
P , P ,

are the maximum and minimum peak values along the wavefront numbered by . This can be

compared to the traditional measurement of so-called scattering cross-sectional area, but

performed for the convenience of the near field measurement and limited field of view in our

experiment setup. Figure 4-17 (a) shows one example of the measured peak pressure at 60 kHz

119
along one wavefront on the exit side of the object for both cases with and without cloak. This

wavefront is near the boundary of the cloak between 100 and 170 . In

Figure 4-17 (b), the averaged visibility of the cloaked object over all the wavefronts on the exit

side is compared with the one with only bare cylinder. The comparison clearly indicates that the

cloak preserves good shielding effectiveness over a broad frequency range even with impedance

mismatch at the outer interface of the cloak. We can read the visibility of 0.62 for the bare steel

cylinder, whereas the visibility of the cloaked steel cylinder is reduced to 0.32 at 60 kHz,

showing significant reduction in scattering and shadowing. The reduced visibility of the cloaked

object is further confirmed by 2D numerical simulation using commercial finite-element method

software COMSOL. In the finite element model, the cloak is approximated by sixteen layers of

homogeneous cylinders with reduced cloaking parameters. In order to match the measured

visibility at 60 kHz, a small attenuation of 5.8dB/m is empirically determined and added in the

simulated cloak model. The good agreement between the numerical and experimental results

indicates the effectiveness of our low-loss metamaterial over a wide range of frequency.

As a control experiment, we measured the field patter scattered by a steel cylinder with the

same dimension as the cloak but with no network structure at 60 kHz. The result is shown in

Figure 4-18, strong scattering and large shadow area are observed behind the cylinder. The

visibility of this cylinder is 1.38, which is much larger than 0.32 for the cloaked object whereas

the hidden object in center has visibility of 0.56. This demonstrates that the steel cylinder without

the designed structure actually causes more significant scattering than the smaller cylinder which

120
is hidden inside the cloak. Only with the machined building blocks, the cylindrical cloak can

guide the acoustic wave to reform the wavefronts on the exit side.

(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 4-16 Measured acoustic wave field mappings of two cases: with bare steel cylinder at (a)

54 kHz (b) 56 kHz (c) 58 kHz and (d) 62 kHz and when the steel cylinder is surrounded by the

metamaterial cloak at (e) 54 kHz (f) 56 kHz (g) 58 kHz and (h) 62 kHz.

121
 (a) (b)

Figure 4-17 Frequency dependence of the averaged visibility of the steel cylinder with and

without the acoustic cloak. (a) The measured peak values of the pressure field along the

wavefront lies between y=100 mm and y=170 mm with and without cloak at 60 kHz (b) Plot of

the averaged visibility. The blue and red solid lines are calculated averaged visibility for the

cases with and without cloak respectively. The 2D simulation performed using

finite-element-based solver COMSOL Multiphysics. The experimental results measured with and

without cloak are marked by the magenta circles and green triangular respectively.

To further demonstrate the effectiveness of our acoustic anisotropic metamaterial based on

lumped circuit network, we simulated the acoustic wave propagation through the acoustic circuit

network to compare with the results using finite element method (FEM) software COMSOL. In

the FEM model, the cloak is approximated by 16 layers of homogeneous cylinders with the

reduced cloaking parameters. In the circuit model, the cloak, hidden cylinder and the background

122
water are all discretized and simulated by the circuit network. The number of unit cells for the

cloak is set as same as the sample in the experiment. Comparison of Figure 4-19 (a) and (b)

indicates a good match between those two numerical approaches. More distortion is observed in

the circuit model in Figure 4-19 (b). This is mainly due to the difference in the boundary

conditions. In COMSOL model, perfect matched layer is utilized to have zero reflection from the

boundary, while in the circuit model the matching resistors are used to terminate the cells in the

boundary.

Figure 4-18 Measured acoustic wave field distribution in the vicinity of a steel cylinder with the

same geometry as the cloak

In Figure 4-20, the scattering from the cloaked object and the bare object to different directions is

simulated using COMSOL at frequency 60 kHz. The plot indicates that the backscattering from

the cloaked object is strongly reduced and also the forward scattering is considerably lower than

that of the bare object. Due to the limited size of the water tank, we cannot measure the field

between the source and the sample to study the backscattering. Refer to Appendix D for
123
numerical study using FEM.

(a) (b)

Figure 4-19 2D simulations for the acoustic cloak with reduced material parameters with black

lines indicating the inner and outer boundary of the cylindrical cloak. (a) The FEM simulation by

software COMSOL. The cloak is approximated by 16-step piecewise homogeneous cylinders. (b)

Lumped circuit model simulation of the cylindrical cloak performed by software SPICE.

Figure 4-20 Simulated scattered pressure field to different directions from the cloaked object

(red line) and the bare object (blue line) at frequencies 60 kHz. The plots are normalized to the

maximum value of the scattered pressure field of the bare object. The plane wave that illuminates

the cloaked/bare object travels in the direction .

124
4.6.3 Conclusion

In conclusion, we have demonstrated a 2D acoustic cloak that can significantly reduce the

visibility of the hidden object from underwater acoustic waves. This underwater acoustic cloak

can be readily implemented by a network of anisotropic acoustic transmission line. Such a new

class of acoustic metamaterial is built from non-resonant elements and offers a wide range of

effective density and compressibility that are not available by the constituent materials over a

broad frequency range. Moreover, this transmission line approach may have potential

applications for a myriad of fascinating devices beyond cloaking based on coordinate

transformation.

References

1
Pendry J B, Schurig D and Smith D R, “Controlling Electromagnetic Fields”, Science 312,

1780 (2006).

2
U. Leonhardt, “Optical Conformal mapping”, Science 312, 1777 (2006).

3
A. V. Kildishev, V. M. Shalaev, “Engineering space for light via transformation Optics”, Opt.

Lett. 33, 43 (2008).

4
M. Rahm, S. A. Cummer, D. Schurig, J. B. Pendry, “Optical Design of Reflectionless Complex

Media by Finite Embedded Coordinate Transformations”, D. R. Smith, Phys. Rev. Lett. 100,

063903 (2008).

125
5
M. Rahm et al., “Transformation-optical design of adaptive beam bends and beam expanders”,

Opt. Express 16, 11555 (2008).

6
Smith, D. R., Padilla, W. J., Vier, D. C., Nemat-Nasser, S. C. & Schultz, S. “Composite

medium with simultaneously negative permeability and Permittivity”. Phys. Rev. Lett. 84, 4184

(2000).

7
Smith, D. R., Pendry, J. B. & Wiltshire, M. C. K. “Metamaterials and negative refractive

index”. Science 305, 788_792 (2004).

8
Schurig D et al, “Metamaterial Electromagnetic Cloak at Microwave Frequencies” , Science

314, 977 ( 2006 ).

9
Cai, W., Chettiar, U. K., Kildishev, A. V. & Shalaev, V. M. , “Optical cloaking with

metamaterials”, Nature Photon. 1, 224–227 (2007)

10
Alu, A. & Engheta, N., “ Multifrequency optical invisibility cloak with layered plasmonic

shells”, Phys. Rev. Lett. 100, 113901 (2008).

11
Leonhardt, U. & Tyc, T. Broadband invisibility by non-euclidean cloaking. Science 323,

126
110–112 (2009).

12
Li Jensen and Pendry J. B., “Hiding under the Carpet: A New Strategy for Cloaking”, Phys

Rev Lett.101, 203901 (2008).

13
Alitalo, P., et al., Appl. Phys. Lett. 94, 014103 (2009).

14
R. Liu, et al.,” Broadband Ground-Plane Cloak”, Science 323, 366 (2009).

15
J. Valentine et al., “An optical cloak made of dielectrics”, Nature Material 2461, (2009).

16
Cummer SA et al, “Scattering theory derivation of a 3D acoustic cloaking shell”, Phys Rev

Lett. 100, 024301(2008).

17
S.Zhang et al , “Cloaking of matter waves”, Phys.Rev. Lett 100, 123002 (2008).

18
M. Farhat, S. Enoch, S. Guenneau, and A. B. Movchan, “Broadband Cylindrical Acoustic

Cloak for Linear Surface Waves in a Fluid”, Phys.Rev. Lett. 101, 134501 (2008).

19
Milton G W, Briane M and Willis J R, “On cloaking for elasticity and physical equations with

a transformation invariant form”, New J. Phys. 8, 248 (2006).

127
20
Pendry J .B .and Li Jensen, “An acoustic metafluid: realizing a broadband acoustic cloak”,

New J. Phys. 10 115032 (2008).

21
Chen H and Chan C T, “Acoustic cloaking in three dimensions using acoustic metamaterials”,

Appl. Phys. Lett. 91, 183518 (2007).

22
Cheng Y, Yang F, Xu J Y and Liu X J, Appl. Phys. Lett. 92, 151913 (2008).

23
S. A. Cummer and D. Schurig, “One path to acoustic cloaking”, New J. Phys. 9, 45 (2007).

24
N. Fang, D. Xi, J. Xu et al., “Ultrasonic Metamaterials with Negative Modulus”, Nature

Materials, 5,452-456 (2006).

25
S. Zhang, L.Yin, and N. Fang, “Focusing Ultrasound with an Acoustic Metamaterial

Network”, Phys. Rev. Lett. 102, 194301 (2009).

26
U. Leonhardt and T. G. Philbin, “General relativity in electrical engineering,” New J. Phys. 8,

247 (2006).

27
D. Schurig, J. B. Pendry, and D. R. Smith, “Transformation-designed optical elements,” Opt.

Express 15, 14772- 14782 (2007).

128
28
G. W. Stewart, “Acoustic wave filters”, Phys.Rev.20,528 (1922).

29
L.L.Beranek, Acoustics, 1954, McGRAW-HILL,New York.

30
L. E. Kinsler, Fundamentals of Acoustics, 1982, Wiley,New York.

129
 5 SUMMARY AND FUTURE WORK

5.1 Summary

A novel approach was designed in this thesis to build an acoustic metamterial for different

applications. Since the inception of the term metamaterials, acoustic metamaterials have being

explored theoretically but there has been little headway on the experimental front. In this work,

we demonstrated that a distributed acoustic system can be effectively described by the analogous

lumped circuit model in which the behavior of the current resembles the motion of the fluid.

Based on this analogy, an acoustic metamterial can be implemented by a two-dimensional

transmission line network to realize negative refractive index and further inhomogeneous

anisotropic density and compressibility.

In one application, a PI/NI interface is constructed by a two-dimensional array of acoustic

circuit network. The ultrasound waves from a point source were brought to a focus point through

the NI metamaterial lens. The unit cell of the acoustic network is only one eighth of the operating

wavelength, making the lens in a compact size. Compared with conventional lenses, the flat thin

slab lens takes advantages in that there is no need to manufacture the shapes of spherical

curvatures and the focus position is insensitive to the offset of source along the axis. Also this

negative index lens offers tunable focal length at different frequencies.

Another more interesting metamaterial device, acoustic cloak, is studied both numerically
130
and experimentally in this thesis. This underwater acoustic cloak is constructed by a network of

anisotropic acoustic transmission line. As a result of the non-resonant nature of the building

blocks, this acoustic cloak can significantly reduce the visibility of the hidden object over a wide

range of frequency. Moreover, this transmission line approach may have potential applications

for a myriad of fascinating devices beyond cloaking based on coordinate transformation.

5.2 Future work

Amazing progress has been made in metamaterial over the past few years. This new concept

provides great opportunities in the design and development of exotic functional materials and

devices. Many applications in optical and acoustic wave fields have been proposed and

demonstrated. In ultrasound, the realization of high-resolution acoustic imaging is a key for

nondestructive testing and medical screening. The negative index metamaterial provides an

unprecedented opportunity for high-resolution ultrasound imaging. The loss in current circuit

model is one of the key limitations from beating the diffraction limit. To decrease the size of the

building blocks, which is essential to achieve sub-wavelength resolution, new fabrication method

could be employed.

The new design paradigm of transformation optics offers enormous way to control the wave

propagation by providing a desired spatial distribution of material parameters. Beyond cloaking,

a myriad of fascinating devices can be investigated in the future, such as power concentrator,

131
wave bends and beam expanders. Also, there is much more ground to investigate for the future

such as nonlinear, optic-acoustic device.

The current circuit model is the first-order linear approximation of the corresponding

distributed acoustic system. With the increase of the acoustic energy, higher-order and nonlinear

effect should be taken into account to make the analogy more accurate.

132
 APPENDIX A: LUMPED CIRCUIT MODEL

Following the development of electrical filters, G.W. Stewart studied showed the combination of

tubes and resonators could be devised in a way very similar to electrical circuit filter. In the

model, the acoustic elements are much smaller than the wavelength such that the wave motion in

these subwavelength elements need not be considered.

In 1927, W.P.Mason gave a more rigorous study of the regular combination of acoustic elements

with considering the wave propagation in the elements. Also viscosity and heat conduction

dissipation are taken into account.

Model I:

Mason’s full-wave model

S2 S2

l l

1 2 3 4 S1

L 2L L

Figure A1 Acoustic transmission line

Consider a main conduction tube with equally spaced side branches, assuming the width of the

junction is much smaller than the wavelength and hence the pressure is constant in the side

133
=
Z
s ρ
ci
tan
0

branch and the main branch over the portion in the immediate contact with the side branch. Also,

the sum of the volume displacement at a junction of tubes is zero.

We can derive the equations ⎡ Z ⎤

coshΓ − 0 sinhΓ⎥
⎡ p4 ⎤ ⎢ S1 ⎡ p1 ⎤
⎢U ⎥ = ⎢ S ⎥⎢ ⎥
⎣ 4 ⎦ ⎢− 1 sinhΓ coshΓ ⎥ ⎣U1⎦
⎢⎣ Z0 ⎥⎦

Z L 1S 2
Where cosh Γ = (cosh 2α 1L + sinh 2α 1 L ) Z L1S2
2 Z s S1 1+ tanh α1L
2Z s S1
Z 0 = Z L1
Z S
1 + L1 2 coth α1L
2Z s S1

If the side branch is closed at one end

While if the side branch is open at one end

Stewart’s lumped circuit model

Consider an acoustic wave transmission line consisting of equal acoustic impedances in series,

divided into sections by acoustic impedances in what might be termed shunt branches.

P1 P2 (P3) P4 Z1 / 2
Z1 / 2 Z1 / 2
U1 U3 U4

Z2 Z2

Figure A2 Stewart’s lumped circuit model

We can derive the transfer matrix for such periodic lumped model as

⎡ Z Z12 ⎤
1+ 1 − ( Z1 +
⎡ p4 ⎤ ⎢ 2 Z 2 4 Z 2 ⎥⎥ ⎡ p1 ⎤
)
⎢U ⎥ = ⎢ ⎢ ⎥
⎣ 4⎦ ⎢ − 1 Z
1+ 1 ⎥ ⎣U 1 ⎦
⎢⎣ Z 2 2Z 2 ⎦ ⎥

134
 Z1 = iωL2 = iωρ0 2 L / S1 Z2 = 1/ iωC2 = ρ0c2 / iω2LS2
If the side branch is closed at one end

Z 1 = 1 / iω C1 = ρ 0 c 2 / iω 2 LS 1 Z 2 = i ω L 2 = i ωρ 0 l / S 2
While if the side branch is open at one end

There is a problem with the above equations when the side branch is open at one end and the

main tube is assumed as acoustic capacitor. Assume one section of the main tube has volume V,

so C = V 2 . When we write out the transfer matrix, we divided the section into half, so each half
ρc0
C V . On the other hand, the half capacitances appears in series, and the total capacitance
=
2 2 ρ c 02

should be C
=
V . So there are different answers as to the total capacitance.
4 4 ρc02

Model II

Mason’s full-wave model

S2 S2

l l

1 2 3 4 5 S1

2L 2L

Figure A3 Acoustic transmission line

In this model, we did not divide the main tube into half within one section, so similarly we can

derive

⎡ p4 ⎤ ⎡ A11 A12 ⎤ ⎡ p1 ⎤
⎢U ⎥ = ⎢ A A ⎥ ⎢U ⎥
⎣ 4 ⎦ ⎣ 21 22 ⎦ ⎣ 1⎦
Z L 1S 2
A11 = cosh 2α 1 L cosh 2α 1 L + sinh 2α 1 L sinh 2α 1 L + cosh 2α 1 L sinh 2α 1 L
Z s S1

135
 2
Z L1 Z ⎛ Z ⎞ S2
A12 = − sinh 2α 1 L cosh 2α 1 L − L 1 cosh 2α 1 L sinh 2α 1 L − ⎜⎜ L 1 ⎟⎟ sinh 2α 1 L sinh 2α 1 L
S1 S1 ⎝ S1 ⎠ Z s

S1 S S
A21 = − sinh 2α 1 L cosh 2α 1 L − 2 cosh 2α 1 L cosh 2α 1 L − 1 cosh 2α 1 L sinh 2α 1 L
Z L1 Zs Z L1

A 22 = cosh 2 α 1 L cosh 2 α 1 L +
Z L 1S 2
sinh 2 α 1 L cosh 2 α 1 L + sinh 2 α 1 L sinh 2 α 1 L Z L1 = ρ 0 c0 α1 = ik
Z s S1

If the side branch is closed at one end Z s = iρ 0 c0 tan kl

While if the side branch is open at one end Z s = −iρ 0 c0 cot kl

Stewart’s lumped circuit model

P1 P2(P3) P4
Z1 Z1
U1 U2 U3 U4

Z2

Figure A4 Stewart’s lumped circuit model

⎡ Z Z 12 ⎤
1+ 1 − (2 Z 1 +
⎡ p4 ⎤ ⎢ Z 2
)⎥
Z 2 ⎥ ⎡ p1 ⎤
⎢U ⎥ = ⎢ 1 ⎢ ⎥
⎣ 4⎦ ⎢ − Z
1+ 1 ⎥ ⎣U 1 ⎦
⎢ Z ⎥
⎣ 2 Z2 ⎦

Z1 = iωL2 = iωρ0 2 L / S1 Z 2 = 1 / i ω C 2 = ρ 0 c 2 / i ω lS 2
If the side branch is closed at one end
1 iρ 2 L ω / S 1
Z 1 = i ω L1 // = i ω L1 /(1 − ω 2 L1C 1 ) =
While if the side branch is open at one end iω C 1 1 − 4 L2 ω 2 / c 2

Z 2 = iωL2 = iωρ 0l / S 2

The problem with this model is P1 and P4 are not in symmetrical positions.

Model III

136
Mason’s full-wave model

S2 S2

l l

1 2 3 4 5 S1

2L 2L

Figure A5 Acoustic transmission line

⎡ p3 ⎤ ⎡ A11 A12 ⎤ ⎡ p1 ⎤
⎢U ⎥ = ⎢ A A ⎥ ⎢U ⎥
⎣ 3 ⎦ ⎣ 21 22 ⎦ ⎣ 1⎦
⎡ Z ⎤
cosh 2α 1 L − L1 sinh 2α 1 L
⎡ p3 ⎤ ⎢ S1 ⎥⎡ p ⎤
⎢U ⎥ = ⎢ S ⎥⎢ 1 ⎥
⎣ 3 ⎦ ⎢− 1 sinh 2α 1 L − 2 cosh 2α 1 L cosh 2α 1 L + L1 2 sinh 2α 1 L⎥ ⎣U 1 ⎦
S Z S
⎣⎢ Z L1 Zs ZsS1 ⎦⎥
A11 = cosh 2α1L
Z L1
A12 = − sinh 2α 1L
S1

S1 S
A21 = − sinh 2α 1 L − 2 cosh 2α 1 L
Z L1 Zs

Z L 1S 2
A22 = cosh 2α 1 L + sinh 2α 1 L
Z s S1

α1 = ik Z L1 = ρ 0 c0

If the side branch is closed at one end Z s = iρ 0 c0 tan kl

While if the side branch is open at one end Z s = −iρ 0 c0 cot kl

Stewart’s lumped circuit model

137
 P1 P2 (P3)
Z1 Z1
U1 U2 U3

Z2

Figure A6 Stewart’s lumped circuit model

⎡ 1 − Z1 ⎤
⎡ p3 ⎤ ⎢ ⎡p ⎤
⎢U ⎥ = ⎢− 1 Z ⎥ 1
⎣ 3⎦ 1 + 1 ⎥ ⎢⎣U 1 ⎥⎦
⎣ Z2 Z2 ⎦

If the side branch is closed at one end Z1 = iωL2 = iωρ0 2 L / S1

1 iρ 2 L ω / S 1
While if the side branch is open at one end Z 1 = iω L1 // = iω L1 / 1 − ω 2 L1C 1 =
iω C1 1 − 4 L2 ω 2 / c 2
Z 2 = iωL2 = iωρ 0l / S 2

Derivation of lumped element from Mason’s full-wave model

⎡ Z ⎤
cosh 2α 1L − L1 sinh 2α 1L
⎡ p3 ⎤ ⎢ S1 ⎥ ⎡ p1 ⎤
⎢U ⎥ = ⎢ S S Z S
⎥⎢ ⎥
⎣ 3 ⎦ ⎢− 1
sinh 2α 1L − cosh 2α 1L cosh 2α 1L +
2 L 1 2
sinh 2α 1L ⎥ ⎣U 1 ⎦
⎢⎣ Z L1 Zs ZsS1 ⎥⎦

⎡ 1 − Z1 ⎤
⎡ p3 ⎤ ⎢ ⎡p ⎤
⎢U ⎥ = ⎢ − 1 Z1 ⎥ ⎢ 1 ⎥
⎣ 3⎦ 1+ ⎥ ⎣U 1 ⎦
⎣ Z2 Z2 ⎦

Compare the above two transfer matrix, to make the lumped model a good approximation for the

full-wave analysis. We will need the following four conditions.

cosh 2α1L ≈ 1 (1)

−
Z L1
sinh 2α1L = − Z1 (2)
S1

S1 S 1 (3)
− sinh 2α1L − 2 cosh 2α1L = −
Z L1 Zs Z2

Z L1S 2 Z (4)
cosh 2α 1 L + sinh 2α 1 L = 1 + 1
Z s S1 Z2

Substitute (2) (3) into (4)

138
 cosh 2α1L +
Z L1S2 Z S
sinh 2α1L = 1+ (sinh 2α1L)2 + L1 2 sinh 2α1L cosh 2α1L
(5)
Zs S1 Zs S1

If 2α1L is very small, in other words, when the periodicity is subwavelength size
2 4 6
and use cosh x = 1 + x + x + x ... sinh x = x +
x3 x5 x7
+ + ...
2! 4! 6! 3! 5! 7!

so cosh 2α 1 L = 1 sinh 2α 1L = 2α 1L → 0

we can rewrite (1)-(5) equations as

cosh 2α 1 L ≈ 1 (6)

Z1 = iωρ 0
2L (7)
S1

1 Zs
(In parallel) (8)
Z2 =
iω
2 LS1 S2 cosh 2α1L
ρ0c02

cosh 2α 1 L +
Z L1S 2 Z
sinh 2α 1 L = 1 + 1 (9)
Z s S1 Z2

Z L1S2 Z S
cosh 2α1L + sinh 2α1L = 1+ (sinh 2α1L)2 + L1 2 sinh 2α1L cosh 2α1L (10)
Zs S1 Zs S1
Z L1S2 Z S
⇒ cosh 2α1L + sinh 2α1L = cosh 2α1L + L1 2 sinh 2α1L
Zs S1 Zs S1
Z L1S2 Z S
⇒ 1+ sinh 2α1L = 1+ L1 2 sinh 2α1L
Zs S1 Zs S1

This means when 2α1L is very small, the equation (4) is automatically satisfied. The above

equation suggest the lumped circuit model based on the Mason’s full wave solution as such: the

series impedance is always an inductor; the shunt admittance is the capacitance with the volume

of the one section of the main tube in parallel with the admittance of the side branch. This

derivation is valid for both cases when the end of the side branch is open or closed.

Error analysis:
λ
cosh x ≈ 1 when lattice = , error is within 5%
19.8
λ
sinh x ≈ x when lattice = , error is within 5%
11.6

139
APPENDIX B: FRESNEL LENS DESIGN BY ACOUSTIC TRANSMISSION LINE

ABSTRACT

We proposed an approach to construct a 2D Fresnel lenses by acoustic network. This lens is

composed of an array of Helmholtz resonators. The resonance at individual resonators results in

effective focusing even the plate has subwavelength thickness. The FEM simulation results

presented the ultrasonic wave propagation through the lenses together with the resulting

diffraction pattern.

INTRODUCTION

The Fresnel lens was first proposed as viable alternatives to conventional spherical lenses for

optical focusing and has been widely studied1-7. Planar focusing lens for acoustic wave was

designed according to the Fresnel zone plate theory 8-9.Since the lens has simple configuration; it

can be fabricated with easy and inexpensive technique. However, the efficiency of Fresnel lenses

reported so far is low.

In this section, we numerically investigated the planar acoustic Fresnel lens composed of a

1D array of Helmholtz resonators using FE method. The Helmholtz resonators are filled with

different medium, dividing plate into 29 Fresnel zones. The acoustic wave fields at different

frequencies are compared from the simulation results.

140
FRESNEL LENSES DESIGN BY ACOUSTIC NETWORK

The spherical wavefront from a point source S as shown in Fig.B1 (a) can be divided into series

annular regions called Fresnel zones10. When there is no Fresnel zone plate, the amplitude of

wave field at observation point P from each zone alternates between positive and negative values.

This means that the contributions from adjacent zones are out-of-phase and tend to nullify each

other. If we remove either all the even or all the odd zones, we will observe a tremendous

increase in irradiance at P. Fresnel zone plate is designed to pass only the odd (even) zones and

obstructs the even (odd) zones. Fig.B1 (b) shows the configuration of planar 1D Fresnel lens we

designed. The lens is composed of an array of Helmholtz resonators. The resonators which

resonate at 50 KHz in pass even zone are filled with water. The cells in odd zone are filled with

air to induce large impedance mismatch, resulting large reflection to obstruct vibration. In Fig.B1

(a), the edge of the mth zone ( Rm from center) is marked by point Am , distance from the plate to

source ρ0 and distance from the plate to the image r0 . Since the wave travels through mth zone

must be out of phase with the center zone by m λ / 2 , so

( ρ m + rm ) − ( ρ 0 + r0 ) = mλ / 2 (1)

Where ρ m = ( Rm2 + ρ 02 )1/ 2 and rm = ( Rm2 + r02 )1/ 2 expand both equations, since Rm is

comparatively small, we get

Rm2 Rm2
ρ m = ρ0 + rm = r0 + (2)
2ρ0 2 ρ0

141
Substitute (2) into (1), we obtain

1 1 mλ (3)
+ =
ρ0 r0 Rm2

rm
Fresnel zone plate

r0 Rm Am

ρ0 ρm

(a)

0.4mm

4.5mm
Aluminum ρ1=2700 kg/m3 v1=6260
2mm
Water ρ2=1000 kg/m3 v2=1483
Air ρ3=1.21 kg/m3 v3=344 1mm

(b) (c)

Figure B1 (a) Schematic diagram of zone plate focusing (b) Configuration of acoustic Fresnel

lens (c) Designed unit Helmholtz resonators

ρ 0 and r0 are interrelated by the same equation as for lens 1

+
1 1
= , where focal length f
ρ0 r0 f

is defined as f =
Rm2 when ρ 0 → ∞ the source of light is put so far away from the plate that the
mλ

incoming wave can be regarded as a plane wave .

Rm2 = mr0 λ (4)

142
 F

Cylindrical wave

Acoustic
Fresnel lens

Input plane wave

Figure B2 Computational model

RESULTS AND DISCUSSION

Figure B2 shows the simulation model. FEM (finite-element method) is employed to

study the focusing of the designed Fresnel plate lens. A collimated acoustic wave is incident on

the plate, which is put inside water medium. The focus point F is expected on the other side of

the plate. When there is no focusing plate, the pressure field in F is the sum of each half

wavelength zone

P = P1 + P2 + K + P29

and each term is approximately equal. Since each term has π phase difference, the pressure at F

would be P , whereas with the zone plate in place, P ≈ 7P1 .The intensity of acoustic field is
1

increased by a factor of196 .Since the resonance effect in pass zone, we expect the proposed plate

yield a very high efficiency.

The focus is assumed at r0 = 100 mm We calculate the position of each zone by equation

Rm2 = mr0 λ , setting λ = 30 mm (wavelength at 50 KHz in water). The dependence of focus position

on frequency is studied. A series simulations have been studied at 40, 50, 60 and 70 KHz. Figure

3 (a)-(d) presented the pressure field distribution above the plate , which is 550mm width along

143
horizontal axis. The y axis is the distance from the plate. We observe the focuses above the

plate along the central axis. Plane wave pattern is also presented resulting from the penetration of

acoustic wave through this thin plate. The focus position is found at y = 75,115,150,175mm, x = 0mm

at 40, 50, 60 and 70 KHz respectively. The pressure intensity cross the focus as function of x is

plotted in Fig 3 (f) - (i). A conventional Fresnel zone plate is also studied as comparison. There is

an array of through holes in the aluminum plate, constituting the pass zones. The acoustic wave

field through the zone plate put inside water is shown in Fig 3 (e). Since the thickness of the

metal plate is only 0.23 λ same as the Fresnel lens based on acoustic network, the plane wave

penetration through the plate is dominate. The amplitude of the focus is comparable with the side

lobe as shown in Fig 3 (j). More than 50dB pressure level difference is found between pass and

obstruct zones in Fresenl zone plate based on acoustic network in Fig3 (k) at 50 KHz.
2
The focal length of the lens changes with source frequency r0 = Rm = ηf . The design set
m λ0

r0 = 100 mm and λ = 30 mm , so η = 3e3m 2 .Figure 4 compared the analytical prediction of focus

length with numerical results. As shown in the figure, focal length increases with frequency. As

frequency increases, there are more numerical errors from the FEM model. For the same mesh

model, the accuracy will decrease with increasing frequency. As frequency was shifted away

from resonance frequency 50 KHz for the pass zone, the efficiency of the lens decayed.

CONCLUSION

In this paper, planar Fresnel lens is built by an array of designed Helmholtz resonators. The

Helmholtz resonators are filled with different medium, forming pass and obstruct zones. The

144
 (a) (ff)

(bb) (g))

(c) (h))

(d) (i)

(ee) (j)

(k)

Figure B3 A cross section of pressuree field distrribution thrrough the F

Fresnel len
ns based on
n

acousticc network at
a (a) 40 KH
Hz (b) 50 KHz
K (c) 60 KHz
K (d) 70 KHz and thhrough a metal Fresnell
145
lens at (e) 50 KHz. The line plot of pressure level cross focus at (e) y = 75 mm at 40 KHz (f)

y = 115 mm at 50 KHz (g) y = 150 mm at 60 KHz (h) y = 175 mm at 70 KHz for Fresnel network lens and

(j) y = 115 mm at 50 KHz for the metal Fresnel lens (k) the sound pressure level in Fresnel network

lens composed of an array of resonators at 50 kHz

simulation results demonstrated that a plane pressure wave is focused through the planar

structure. Compared with conventional Fresnel lens with the same thickness, the focus effect is

more efficient through those based on acoustic network design .The focal length of the lens can

be tuned at different frequency as well.

Figure B4 Focal length as a function of frequency

References

1. G. Hatakoshi and S. Tanaka, “Grating lenses for integrated optics”, Opt. Lett. 2, 142-144

(1978).

2. T. Suhara, K. Kobayashi, H. Nishihara, and J. Koyama, “Graded-index Fresnel lenses for

integrated optics”, Applied Optics 21, 1966-1971(1982).

146
3. K. Rastani, A. Marrakchi, S. F. Habiby, W. M. Hubbard, H. Gilchrist, and R. E. Nahory,

“Binary phase Fresnel lenses for generation of two-dimensional beam arrays”, Appl. Opt. 30,

1347-1354 (1991).

4. J. S. Patel and K. Rastani, “Electrically controlled polarization-independent liquid-crystal

Fresnel lens arrays”，Opt. Lett. 16, 532-534(1991).

5. H. Ren, Y.H. Fan, and S.T. Wu., “Tunable Fresnel lens using nanoscale polymer-dispersed

liquid crystals”, Appl. Phys. Lett. 83, 1515-1517(2003).

6. Y.H. Fan, H. Ren, and S.T. Wu, “Electrically switchable Fresnel lens using a

polymer-separated composite film”, Opt. Express 13, 4141-4147(2005).

7. L.C. Lin, H.C. Jau, T.H. Lin, and A.Y.Fuh,“Highly efficient and polarization-independent

Fresnel lens based on dye-doped liquid crystal”, Opt. Express 15, 2900-2906(2007)

8. K Yamada, T Sugiyama and H Shimuzi, “Theoretical Considerations on Acoustic Field

Formed by Fresnel Zone Type Focusing Radiator,” Jpn J Appl Phys 26, 180-182(1987).

9. Hadimioglu, B. et al., “High-efficiency Fresnel acoustic lenses”,Ultrasonics Symposium

1,579 – 582(1993).

10. Eugene Hecht , “Optics”, Addison Wesley 2002

11. Shin Chuen Chan et al., “Finite element analysis of multilevel acoustic Fresnel lenses”,

Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions 43, 670 – 677(1996).

147
 APPENDIX C: NEGATIVE INDEX LENS BASED ON METAL-INSULATOR-METAL

(MIM) WAVEGUIDES

In this appendix, we calculated the dispersion curves of several different waveguide designs.

Negative refraction at visible frequencies is proved experimentally in designed

metal-insulator-metal (MIM) waveguides recently1. From our simulation, negative refractive

index is found in Ag-Si3N4-Ag planar waveguides with 50nm dielectric core. As the dielectric

core thickness is increased to 200nm, a matched mode is found to across the negative slope

branch of the waveguide with 50nm core. In waveguide Ag-Si3N4-PMMA-Ag with 50nm Si3N4

and 150nm PMMA, there are also two mode branch which cross the negative slope of the

Ag-Si3N4-Ag planar waveguides with 50nm dielectric core. Electromagnetic field distributions

of two cascaded waveguides are simulated by 3D FDTD method. The center positioned

waveguide with 50nm Si3N4 core is made of crescent shape to investigate the focus effect of the

MIM waveguides. Several samples are fabricated by depositing 100nm silver with curved shape

on 50nm Si3N4 membrane.

MIM simulation

We investigate the dispersion curve property of several planar waveguides. Transfer matrix

approach was used to simulate the transmission behavior of a multilayer stack of films (Figure 1).

The approach is based on matrix formulation of field amplitudes in terms of Fresnel coefficients.

A recurrence formula relating field amplitudes in successive layers is expressed as:

148
 Figure C1 Transfer matrix approach for an optical system with multilayer stack of films.

⎛ Em+ −1 ⎞ 1 ⎛ exp(−iγ m ) rm exp(−iγ m ) ⎞⎛ Em+ ⎞

⎜ ⎟ ⎜ ⎟⎜ ⎟
⎜ E − ⎟ = t ⎜ r exp(iγ ) exp(iγ ) ⎟⎜ E − ⎟ (1)
⎝ m −1 ⎠ m ⎝ m m m ⎠⎝ m⎠

where Em+ Em− denote the field amplitude of the forward and backward propagating wave in the

mth layer respectively (Note that there is no backward propagating wave in

the nth layer, i.e En− = 0 ). t m , rm are Fresnel transmission and reflection coefficients at the

interface between (m-1)th and mth layer, respectively. γ m = d m−1k m−1 is the phase factor,

d m and k m are thickness and wave vector of mth layer respectively.

From the recurrence relation, an expression for transmission (T) and reflection (R) can be

computed as:

⎛ E0+ ⎞ ⎛ a b ⎞⎛ En+ ⎞
⎜ ⎟=⎜ ⎟⎜ ⎟ (2)
⎜ E − ⎟ ⎜ c d ⎟⎜ 0 ⎟
⎝ 0⎠ ⎝ ⎠⎝ ⎠

En+ 1
T= = (3)
E0+ a

E0− c
R= = (4)
E0+ a

⎛a b ⎞ 1 ⎛ exp(−iγ m ) rm exp(−iγ m ) ⎞
where ⎜⎜ ⎟⎟ = (C1)(C2 )...(Cn ); and (Cm ) = ⎜⎜ ⎟
⎝ c d ⎠ t m ⎝ rm exp(iγ m ) exp(iγ m ) ⎟⎠

149
 Ag
Ag 100nm

Si3N4 200nm Si3N4 200nm

z Ag 1000nm
100nm
Ag 100nm

x (a) (
(b)

(cc) (d)

(ee) (ff)

(g)

Figure C2 (a) An Ag-Si3N4-A

Ag with 50nnm dielectrric core andd (b) an Ag--Si3N4-Ag with
w 200nm
m
2
dielectrric core .Coomputed isotherm conntour of traansmission T p of (cc) the waveeguide with
h

50nm dielectric
d coore and (d) of the waveeguide withh 200nm dieelectric coree. Hy field distribution
n

along z axis of (e) the waveguuide with 500nm dielectrric core andd (f) of the w
waveguide with
w 200nm
m

dielectrric core. (g) Dispersionn curve of Ag-Si

A 3N4-Ag
g waveguidee with 50nm
m and 200nm
m dielectricc

core.

150
 The calculated transmission of two Ag-Si3N4-Ag waveguides with 50nm and 200nm

dielectric core is shown in Fig.C2 (c) and (d). The transmission of P-polarized light is

investigated. Several bright lines are observed in Fig.C2(c) and (d), which corresponds to

propagation mode of the waveguides. The Hy field distribution normal to the layers are shown in

Fig. C2 (e) and 2(f) for 50nm and 200 nm dielectric core respectively. Two discontinuity points

are observed on both curves in the positions of the two boundaries of Si3N4 with Silver. In

Fig.C2(g), dispersion curves of two Ag-Si3N4-Ag waveguides are presented. From frequency

700THz to 700THz, the blue dispersion curve of Ag-Si3N4-Ag waveguides with 50nm dielectric

core exhibits a negative slope. There is one matched mode from the cross of the dispersion curve

of the two waveguides at 750 KHz and k x / k0 = 1.5 .

100
50nm
200nm

100nm
z
(a)
x
200nm 1200nm

2000n
Current source

y
Si3N4 Ag PML

(b)

Figure C3 (a) cross section and (b) corresponding plane view of the cascaded waveguides

151
 (b)

(a) (c)

Figure C4 Ez Field distribution in the cascaded waveguides on (a) xy-plane at z=233nm;

(b)xz-plane at y=1000nm (c) yz-plane at x=600nm

Fig.C3 shows the cross section and top view of the waveguides. An Ag-Si3N4-Ag

waveguides with 50nm dielectric core is positioned between two Ag-Si3N4-Ag waveguides with

200nm dielectric core. The center waveguide was designed with curved boundary shape on left

side.

Three-dimensional FDTD method is employed to calculate the wave propagation through the

cascaded waveguides. The Ez Field distributions in the waveguides in fig 3 are presented in

figure 4. A one-fifth wavelength long current source J z at λ = 400nm is applied in the top

boundary of the dielectric and Ag. In Fig.C4 (a), the Ez Field distribution on xy-plane at

z=233nm is shown. The Ez field is propagating through the waveguide with quite a lot loss

around 20dB after through the center 50nm dielectric core. Fig.C4 (b) shows Ez Field on the

xz-plane at y=1000nm. The dark red point shows the source position. The field decays in Si3N4
152
along th A Ez decaays verticallly in Silverr layer. The Ez field ov
he propagattion path. Also ver yz-planee

at x=60 own as in Fig.C4 (c).. The crosss section is cut througgh the part with 50nm
00nm is sho m

dielectrric core, wh d from 250nm to 300nm

hich is clearlly observed m in the figu
ure. The fielld decays in
n

the botttom and top

p 100nm sillver part. Arround 95dB
B Ez field iss presented outside thee waveguidee

from 0 to 150 nm where

w theree is air. It su
uggested thaat 100nm sillver claddinng might not be enough
h

to ensurre completee opacity.

Figure C5 Dispeersion curv

ve of Ag-S
Si3N4-Ag with
w 50nm dielectric core (bluee dots) and
d

Ag-Si3N4-PMMA--Ag (red dotts) with 50n

nm Si3N4 an
nd 150nm PM
MMA

Insstead of 2000nm Si3N4, two layers of 50 nm Si

S 3N4 and 15
50nm PMM
MA are used
d as core forr

the posiitive wavegguide part. This

T design offers an eaasy way forr fabricationn and flexib
bility for thee

thickness of the dielectric part

rt. .Dispersio
on curves oof Ag-Si3N4-PMMA-Ag with 50nm
m Si3N4 and
d

150nm PMMA is shown in fiigure5 in reed dots. Thee blue dots presented tthe dispersion curve off

Ag-Si3N4-Ag with 50nm dieleectric core .There are two cross points
p in figgure showin
ng there aree

two posssible matchhed modes in

i these two
o waveguidees from 700
0THz-800TH
Hz.

Fig
g.C6 showss the crosss section and
a top viiew of thee waveguid
des. An Ag-Si
A 3N4-Ag
g

wavegu
uides with 50nm dieelectric corre is posittioned betw
ween two Ag-Si3N4-PMMA-Ag
- g

153
waveguides with 50nm Si3N4 and 150nm PMMA as core. The center waveguide was designed

with curved boundary shape on left side.

100nm
50nm
100nm 150nm

z 100nm

x
(a)
200nm 60 0nm 1200nm

2000nm

PML
y

PMMA Si3N4 Ag

(b)

Figure C6 (a) cross section and (b) corresponding plane view of the cascaded waveguides

The Ez Field distributions in the waveguides are presented in Fig.C7. A one-fifth wavelength

long current source J z at λ = 400nm is applied in the top boundary of the dielectric and Ag. In

Fig.C7 (a), the Ez Field distribution on xy-plane at z=280nm is shown. Compared with fig 4(a),

less boundary feature is shown on the left side of the Ag-Si3N4-Ag to Ag-Si3N4-PMMA-Ag in

fig7 (a). Fig.C7 (b) shows Ez Field on the xz-plane at y=1000nm. More leakage of the field in air

154
from x=300nm to x=800nm is presented. The Ez field over yz-plane at x=600nm is shown as in

Fig.C7 (c). Around 95dB Ez field is presented below the waveguide in air. It proved again that

100nm silver cladding might not be enough to ensure complete opacity.

(b)

(a) (c)

Figure C7 Ez Field distribution in the cascaded waveguides on (a) xy-plane at z=280nm;

(b)xz-plane at y=1000nm (c) yz-plane at x=600nm

3D FDTD model

(a) (b)

Current
Figure C8 (a) Geometry for the 3D-FDTD model (b) top view of the 3D model

155
 Three-dimensional FDTD model is employed to simulate the electromagnetic field propagating

through negative index lens. The computational domain is 220 × 220 × 100 grid points, in the x, y, and

z direction, respectively. Figure 1(a) shows the model, a negative index (NI) lens is placed in the

middle of the model. We used Drude model to set ε and μ as following expression:
ω 2p ω 2p
ε = ε 0 (1 − ) μ = μ0 (1 − )
ω 2
ω2

For the PML (perfect matched layer), we used 15 layers with σ max = 10 . The analytical expression

n
for σ is given by σ n = σ max ( )2
L

where n is either x, y, or z and L is the PML’s thickness.

A dipole current source was located at (38, 110, [38 62]). The input signal is given by

where Tp = 1/ f0 is the period of one single cycle with f 0 = 30GHz in the simulation. The

three-derivative smooth window functions are given by g on (t ) = 10 xon

3
− 15 xon
4
+ 6 xon
5

goff (t ) = 1− [10 xoff

3
− 15 xoff
4
+ 6 xoff
5
] with xon = 1− (mTp − t ) / mTp , xoff = [1− (m + n)] / mTp

Fig. C (9-11) showed the Ez field distribution of the model for three different index lens

with n=-1,-3 and -0.23 respectively, In Fig. C10 (a), two focus points are observed in the xy

plane because of index match between the lens and free space. Collimation feature is observed in

fig3 (a) for lens n=-0.23 while divergence of the field is shown in Fig.C11 (a). Fig.C9 (b),

Fig.C10 (b) and Fig.C11 (b) presented the field along the boundary of the lens in yz-plane. From

all the three figures, propagation modes are observed parallel to y direction. The depth of the lens
156
 (a) (a) (a)

(b) (b) (b)

(c) (c) (c)

(d) (d) (d)

(e) (e) (e)

Figure C9 Ez Field Figure C10 Ez Field Figure C11 Ez Field
distribution for n=1 in the distribution for n=-0.23 in distribution for n=-3 in the
(a) xy-plane at z=50; (b) the (a) xy-plane at z=50; (a) xy-plane at z=50; (b)
yz-plane at x=68 (b) yz-plane at x=68 yz-plane at x=68
(c)x=110 ;(d)x=150 ;(c)xz- (c)x=110 ;(d)x=150 ;(c)xz- (c)x=110 ;(d)x=150 ;(c)xz-
plane at y=115 plane at y=115 plane at y=115

157
along z direction is two wavelengths for f 0 = 30GHz in vacuum. In Fig. C11 (b), strong

interference is observed. The interference caused standing wave mode since depth of lens along z

is six wavelengths.

The fields along the other boundary of the lens in yz-plane are shown in Fig.C9 (c), Fig.C10

(c) and Fig.C11(c). The focus is observed for index matched lens n=-1 in Fig.C9(c). Propagation

mode are observed in Fig.C9 (c) and Fig.C11 (c) while field is restricted to center area resulted

from collimation in Fig.C10 (c).The fields in the free space in yz-plane after through the lens are

shown in Fig.C9 (d), Fig.C10(d) and Fig.C11(d). The field is more divergent in fig 4(d)

compared with the one in Fig.C9 (d). In Fig.C10 (d), no propagation parallel to y direction is

presented.The fields in xz-plane are shown in Fig.C (9-11)(e). Two focuses are observed in Fig.

C10 (e).Plane wave phase front is observed in Fig.C10 (e) and convergent wave is presented in

Fig.C10 (e).

Simulation Study of the waveguide MDM with effective negative index

Figure C12 Calculated dispersion curves for an Ag-Si3N4-Ag waveguides with 50nm thickness

dielectric core

158
 Three-dimensional FDTD method is employed to calculate the dispersion curve for an

Ag-Si3N4-Ag waveguides with 50nm dielectric core. A point source excited the electromagnetic

field inside the waveguide. The wavevector is calculated by spatial Fourier transfer. The

dispersion curve exhibits a negative slope over 410-448nm and hence negative index behavior is

expected in this frequency domain. In range 455-465nm, the curve also shows negative slope. As

the electrical field distribution is Bessel function, the period continuous changing along the

propagation direction. The finite size of calculation domain causes reflection from the boundary

even PML (perfect matched layer) is used.

(a)

(b)

Figure C13 (a) cross section and (b) top view of the simulated waveguides

Wave propagation through a cascaded waveguides was simulated. Fig.C13 shows the cross

section and top view of the waveguides. An Ag-Si3N4-Ag waveguides with 200nm dielectric core

is connected with an Ag-Si3N4-Ag waveguides with 50nm dielectric core. The waveguide with

200 nm Si3N4 is a 170nm wide slot shape. The interface of two waveguides makes an angle

α = 79.3 with horizontal direction.

159
 The simulation results are presented in Fig.C14.A current source Jz at 428nm is induced

parallel to y direction at z=275nm. This wavelength falls in the negative-index regions for the

50nm dielectric core waveguide according to dispersion curve. Ez field at z=275nm is shown in

fig3. We observed negative refraction with angle ϕ2 = 12.50 when the beam propagates through

the interface. The incident angle ϕ1 is 10.7 0 . The effective index for 50nm Si3N4 waveguide is -4

at 428nm from calculation. Hence the effective index for 200nm core waveguide is around 4.66

by Snell’s law.

ϕ2

ϕ1

Figure C14 Ez Field distribution in the cascaded waveguides on xy-plane at z=275nm

References

1. H.J.Lezec, J.A.Dionne, H.A.Atwater, Negative refraction at visible frequencies, Science,

316,430(2007)

160
 APPENDIX D: SCATTERING FIELDS FROM THE CLOAK

Scattering harmonic propagation

The scattering from the designed cloak is studied using FEM in this appendix. The “Scattered

harmonic propagation” mode in COMSOL is selected for the following simulation. The

calculation domain is presented in Figure D2. A plane wave is incident on a hidden object (1.0λ

in diameter) surrounded by a cloaking shell of thickness 1.0λ. A perfect matched layer (PML) is

put around the cloaked object to reduce reflection.

Figure D1 (a) Picture of model navigator in COMSOL (b) the plot of the incident wave field

Figure D2 Computational domain

161
 (a) (c)

(b) (d)
Figure D3 Pseudo colormap of electrical field distribution due to a plane incident wave by
FEM simulation. Total electrical fields (a) with cloak and (b) without cloak. Scattering field (c)
with a cloak and (d) without cloak.

Fig. D3 (a) and (c) show the total electrical fields for the cases with and without cloak

respectively. Nearly zero amplitude filed is observed inside the cloak as expected except the

singular point in center. Much less distortion is observed when the hidden object is surrounded by

the cloak. The scattered fields are presented in Fig. D3 (b) and (d). Comparison shows the wave is

guided around the object in the shell and merged behind the object with obvious reduction in back

scattering. However, inside the cloak, there is strong scattering intensity as a result of the

162
singularity in the center from the coordinate transformation.

(a) (b)

(c) (d)

(e) (f)

(g) (h)

163
Figure D4 Pseudo colormap of Electrical field distribution due to a plane incident wave by FEM

simulation. The scattering fields (a) inside the cloak with the hidden object (b) from the bare

object when there is no cloak. The line plots of the total electrical field as function of angle when

there is (c) cloak (d) no cloak. The far fields (e) with cloak (f) with no cloak. The scattering

fields (g) with cloak (h) with no cloak.

The scattering fields inside the cloak with the hidden object and from the bare object when

there is no cloak are presented in Fig. D4 (a) and (b). The line plots of the total electrical field as

function of angle for cases when there is cloak and no cloak are shown in Fig. D4 (c) and (d).

The line is positioned along the boundary of the computation domain as marked in red in Fig. D2.

Fig. D4 (e) and (f) are the far fields for the cases with cloak and without cloak. The scattering

fields from the cloaked object and bare object are shown in Fig. 4D (g) and (h).

The line plots are compared in Fig.5D. As indicated in Fig.5D (a), the amplitudes of the near

field from both cases are comparable but rather very different pattern. However, the amplitude of

the far field when there is a cloak is much smaller than the case with only bare object. Compared

the scattering electrical field, the scattering is significantly reduced when the hidden object is put

inside the cloak.

164
ε=
r μ
=
r εφ= =
−μ
φr εz= z=μ
⎜⎝R
2 ⎟⎠R
2
1 −

(a) (b) (c)

Figure D5 The line plots of the (a) total electrical field (b) the far fields and (c) the scattering

fields as functions of angle.

Stratton-Chu formula

The far electromagnetic field can be calculated from the near field using the Stratton-Chu

formula .Because the far field is calculated in free

space, the magnetic field at the far-field point is given by .The Poynting vector gives the

power flow of the far field: .Thus the far-field radiation pattern is given by

. Because the Stratton-Chu formula only defines the angular distribution of the

electromagnetic field and does not depend on the distance from the near field, you can evaluate

the far field at the some boundaries you pick.

TE wave propagation of perfect cloak（Perfect Cloak_March08_2010.mph）

The full-wave simulation is carried out using TE mode in software COMSOL as presented in Fig.

D6. The cloak is descirbed by full specification of the material property.

165
 Figure D6 Picture of model navigator in COMSOL

The calculated Electric field Ez is shown in Fig. D7 (a). As shown, the electric field is guided

inside the cloak and reform to its original trajectory. The normal of the electrical field is zero

inside the cloak as indicated in Fig. D7 (b) except the singularity point in center. The scattering is

negligible outside the cloak according to Fig. D7(c).

(a) (b)

(c)

(d)

Figure D7 TE-wave simulation of perfect cloak (a) Electric field Ez (b) Normal of Electric field

Ez (c) Scattered electric field scEz (d) Material specification of cloak

166
ε=
μ
z =
⎜⎝−
z2
R2

TE wave propagation of simplified cloak（Simplified Cloak_March08_2010.mph）

The full-wave simulation is carried out using TE mode in software COMSOL as presented in Fig.

D6. The cloak is descirbed by reduced specification of the material property.

Figure D8 Picture of model navigator in COMSOL

(a) (b)

(c)

(d)

Figure D9 TE-wave simulation of reduced cloak (a) Electric field Ez (b) Normal of Electric field

Ez (c) Scattered electric field scEz (d) Material specification of cloak

167
There is more distortion in the calculated Electric field Ez in Fig. D9 (a) as a result of the

reduced parameter compared in Fig. D7(a). However, the electric field is bent inside the cloak

around the hidden object. The normal of the electrical field is nonzero at the interface between

the cloak and hidden object as indicated in Fig. D9 (b) .The forward scattering is observed in Fig.

D9 (c). On the other hand, there is much less back forward scattering.

The electric field Ez is calculated under a incident point source out as presented in Fig. D10

(a). However, the normal of the electrical field is the same in Fig. D10 (c) as the total field in Fig.

D10(a). The software COMSOL failed to give the correct results for scattering field under point

excitation. The scattered electric displacement is plotted in Fig. D10 (d). Other simulation

approach should be found to test these results.

Figure D10 TE-wave simulation of reduced cloak under a point source (a) Electric field Ez (b)

Normal of Electric field Ez (c) Scattered electric field scEz (d) Scattered electric displacement

168
Isc, loakcidenbjh loakc≈
I()x jh
b
o
en
id Irespac()−
fx frespac x

Isc, ylinderc( cy≈

I)x (
inderlx freI−
) )(
pacesx

Experiment measurement

The measurement data at 60kHz is processed to obtain the scattering pressure field. At each

measurement point , we acquired a pressure pulse as function of time. To obtain the intensity at

each gird point, we define the intensity as . The intensity

distributions are shown in Fig.D12 (a-d) by process the measurement data. Due to the point-like

transducer, there is stronger field observed closer to the source.

Figure D11 Measured pressure amplitude at a grid point

Strong scattering presented in Fig.D12 (a) when there is a bare object in the center of the water

tank. However, with a cloak, the intensity is more homogeneous on the exit side of the cloak. Fig.

D12(c) is the intensity distribution there is no object in the water tank. The intensity of the

measured pressure field of a cylinder with same dimension with the cloak but no acoustic circuit

machined is shown in Fig. D12 (d). Large shadowing area is observed resulted from this

cylinder.

The intensity of the scattering field is calculated as

169
 (a) (b)

(c) (d)

(e) (f) (g)

(h) (i) (j)

Figure D12 Measured intensity of the pressure field of (a) bare object (b) cloak (c) free space (d)

large cylinder. The intensity of the scattering field of (e bare object (f) cloak and (g) large

cylinder. (h)(i)(j) are the zoom in of the square area in (e)(f)(g) respectively.

170
 (a) (b)

(c) (d)

(e) (f) (g)

(h) (i) (j)

Figure D13 Measured intensity of the pressure field of (a) bare object (b) cloak (c) free space (d)

large cylinder. The intensity of the scattering field of (e bare object (f) cloak and (g) large

cylinder. (h)(i)(j) are the zoom in of the square area in (e)(f)(g) respectively.

171
The intensity of the scattering field of bare object, cloak and large cylinder is shown in Fig.D12

(e-g). Fig.D12 (h-j) are the zoom in of the square region marked by dashed lines in Fig.D12 (e-g)

respectively. The patterns of the backward scattering are very similar for three cases. However,

the backward scattering from the bare object is concentrated in the center area while there is

higher intensity in the corner scattered from the cloak. When the cylinder is not machined with

the acoustic circuit network, there is much larger shadowing area observed in Fig.D12 (g)

In another data processing, the intensity is defined asI ( x) ≈ ∫ P( x, t ) dt = ∑ Pn ( x) The same

n

measurement data at 60 kHz is processed to obtain the scattering pressure field. The intensity

distributions are shown in Fig.D13 (a-d) by process the measurement data. Due to the point-like

transducer, there is stronger field observed closer to the source.

Large shadowing area is presented in Fig.D13 (a) when there is a bare object in the center of the

water tank. However, with a cloak, the intensity is more homogeneous on the exit side of the

cloak. Fig. D12(c) is the intensity distribution there is no object in the water tank. The intensity

of the measured pressure field of a cylinder with same dimension with the cloak but no acoustic

circuit machined is shown in Fig. D12(d), shadowing much larger shadowing area compare with

the cloak in Fig. D12(b).

The intensity of the scattering field is calculated as

I sc , cloak ( x ) ≈ I cloak ( x ) − I freespace ( x )

I sc , hiddenobj ( x ) ≈ I hiddenobj ( x ) − I freespace ( x )
I sc , cylinder ( x ) ≈ I cylinder ( x ) − I freespace ( x )

The intensity of the scattering field of bare object, cloak and large cylinder is shown in Fig.D13
172
(e-g). Fig.D13 (h-j) are the zoom in of the square region marked by dashed lines in Fig.D13 (e-g)

respectively. A strong scattering in beam shape is observed from the bare object while scattering

shows stronger in the corner from the cloak. When the cylinder is not machined with the acoustic

circuit network, there is much larger shadowing area observed in Fig.D13 (g)

173
 APPENDIX E: EXPERIMENTAL SETUP AND DATA ACQUISITION

Negative index lens

To study the focusing phenomena of the acoustic metamaterial in experiment, we machined a 2D

array of periodically connected subwavelength Helmholtz resonators in an aluminum plate and

the resonators are filled with water. The left half part is composed of a 2D array (40 by 40) of

larger cavities connected with main channels. The volume of the cavity is around ten times of

that of one section of the channels. The periodicity (3.175mm) of the sample is one-eighth of the

wavelength at around 60 kHz frequency range. The right half part of the sample is the dual

configuration of the left half part, in which there is an array (40 by 40) of orifices connected with

channels. The volume of one section of the main channel is designed as around ten times of that

of the orifice.

For experimental confirmation of ultrasound focusing in this acoustic metamaterial, we

measured the pressure field through this PI/NI interface. The ultrasound waves were launched

from a horn shaped transducer with a tip of 3mm diameter in size. The tip is inserted into a hole

drilled through the center of the PI part ((column, row) = (20, 20)) to illuminate the sample. A

waveform generator (Tektronix AFG 310) is used to drive the transducer. The source generated a

174
 Figure E1 Geometry parameter of the negative index lens

Figure E2 Experimental setup for acoustic negative index lens

burst of sine waves with a width of 5 periods. To map the pressure field, a hydrophone was

mounted on two orthogonal linear translation stages. By stepping the miniature hydrophone

(RESON TC4038-1) to the positions above those through holes in the NI part and recording the

pressure amplitude at every step, we acquired the spatiotemporal field distribution of the

ultrasound wave focusing pattern. The pressure field is afterwards amplified by Stanford research

systems model SR650 and captured using a digital oscilloscope (Agilent DSO6104A) and then
175
downlooaded to a computer
c foor post proccessing andd analysis. The
T field paattern in thee NI part att

frequenncy of intereest was obttained by Foourier transsformation of

o the acquuired pulse at
a each gridd

point, allowing
a thee wave amp
plitude of eaach frequenncy componnent to be pplotted as a function off

positionn in the NII part. The limitations on the traansducer woorking specctrum preveent us from
m

probingg the acousttic wave fieeld below 60

6 KHz. Onne Labview
w program iis wrote to control thee

experim
mental scannning.

Function geneerator I
generate a perio
odic squre
pulse at 1Hz
1

The hydrophone is moved

m to
Function generator II
one position to measure the
triggered by th
he squre pulse,
pressure field, the acquired
generate 5 peeriods of sine
pressure pulse is sent to
wave to drive the
t transducer
Oscilloscope e

The dataa (pressure field at

At the same time, the
one position as function of
Oscilloscope is triggered by
time) is downloaded to
the squre pulse to record the
computeer from Oscilloscope
signal from amplifier
as text file

F
Figure E3 Sequence
S off tasks in a circular
c flow
w controlledd by Labvieew in the exxperiment

Figure E44 FFT of thee measurement data

176
Acoustic cloaking

To demonstrate the shielding phenomena, the sample of cloaking is placed in a water tank to

measure the pressure fields in the immediate environment of the cloaked object to compare

with those without cloak. The tank edge is filled with absorbing rubber to reduce reflection.

Because of the high impedance contrast between water and air as well as between water and

glass, the system provides a 2D waveguide to confine the ultrasound wave propagation. The

side of the cloak machined with the network structure is placed against the bottom of the tank.

The cloak has a thickness of 3mm with the depth of the cavities smaller than 1.36 mm. The

water inside the cloak is connected to the surrounding water which is 1.5 mm deep through the

channels along the radial direction around the outer boundary of the cloak.

The ultrasound signal from a spherical shape transducer is launched to the water as a point

source. A waveform generator (Tektronix AFG 310) is used to drive the transducer. The source

generated a burst of sine waves with a width of 20 periods. The pressure field around the cloak

sample in the water is detected by a miniature hydrophone (RESON TC4038-1), amplified by

Stanford research systems model SR650 and captured using a digital oscilloscope (Agilent

DSO6104A) and then downloaded to a computer for post processing and analysis. The

hydrophone is attached to a motorized translation stage. The control program of a customized

LabVIEW scans across the data acquisition region by moving the hydrophones in a small

177
increment 3mme to record the spatiotemporal distribution of the pressure field. The snapshot

of the field pattern can be plotted as a function of position. To verify the broad operational

bandwidth of the acoustic cloak, the transducer is excited over a discrete set of frequencies to

illuminate the sample. The transducer operating spectrum limits us to test the frequency range

from 52 kHz to 64 kHz. Similar sequence of tasks is controlled by a Labview program in the

experiment.

Figure E5 Experimental setup for acoustic cloaking device

178
 APPENDIX F: CICUIT MODELING

Circuit modeling vs. full-wave simulation

To develop an understanding of the difference between a full wave simulation of a distributed

system and the correspondingly lumped circuit modeling, I conduct a simple analysis to quantify

the comparison. In this analysis, I calculated the phase difference between two neighboring units

in a periodic structure.

Figure F1 A distributed acoustic system of the corresponding circuit model with (a) a high-pass

topology (b) and a low-pass topology

In the first example, a main transmission channel with recurrent side branches, which are

open at the outer end as show in Fig. F1(a), is simulated by FEM. This system is analogous to a
179
circuit of a series of capacitors with shunt inductors. On the other hand, an array of large cavities

connected by small necks can be described by a lumped network of a series of inductors with

shunt capacitors as presented in Fig.F1 (b).

To help the comparison, I chose those points in the center of the channel to calculate the

phase lag. And in the circuit model, the phase can be more easily read from the two connecting

nodes.

(a) (b)

Figure F2 Phase difference as a function of the ratio between wavelength and the unit cell size

for (a) high-pass and (b) low-pass topology

Fig.F2 presents the phase difference as a function of the ratio between wavelength and the

unit cell size. As shown in Fig. F2 (a-b), when the unit cell is very small compared with

wavelength; the two models gave the value of phase advance between two units. However, as

frequency increases, the unit cell appears larger to the incident acoustic waves. As a result, the

errors increase at shorter wavelength. Because of the high-pass topology, the calculation can not

180
be carried at wavelength longer than 16 times of the size of the unit cell. On the other hand, the

wave cannot be transmitted at higher frequency in low-pass topology with wavelength longer

than two times of the size of the unit cell.

This analysis demonstrates that the equivalent circuit modeling can provide the desired

information about wave propagation through the distributed acoustic system when the unit cell is

smaller than one-eighth of wavelength.

Circuit modeling of negative index lens

NI PI
PI

Figur F3 A flat lens brings all the diverging rays from an object into two focused images

To study the ultrasound focusing by a negative index lens, a two-dimensional circuit model is

simulated employing commercial software SPICE. The calculated lumped model is an

approximation of the distributed acoustic system. The acoustic metamaterial with negative

refractive index (NI) is composed by a two-dimensional (2D) 30 × 30 periodic cascaded array

of the unit cell as in Error! Reference source not found. (a). In order to build a PI/NI interface,

an acoustic metamaterial with positive index (PI) is implemented by 30 × 30 circuit cells as

181
shown in Error! Reference source not found. (b). In the simulation, the negative index lens is

sandwiched between two positive index medium. In the circuits, a very small resistance is

connected to each inductor. The boundary of the simulation model is grounded by a resistor with

value equal to the characteristic impedance of the transmission line to reduce the reflection from

the boundary.

(a) (b) (c)

Figure F4 Pseudo colormap of scaled pressure (a) amplitude and (b)phase distribution and (a)a

snapshot due to a point source illuminating a 2D transmission model of the PI/NI/PI interface.

Fig.F4 (a) and (b) illustrates the normalized pressure magnitude and phase distribution at steady

state when a continuous signal at 60 kHz is introduced at the center of the PI part. The x-y axes

are labeled according to the cell number. The maximum field magnitude was normalized to unity.

The focal point is expected around node (45, 15) and (75, 15) since the relative index value

equals –1 at 60 kHz. In Fig.F4 (a) and (b), focusing is evident by the increased transmission and

confinement of the fields near the focal plane (near node (45, 15) and (75, 15)). Moveover,

182
concavity waterfronts are observed near the two focuses as evident in Fig.F4(b). Fig.F4(c) is a

snapshot of the pressure field at steady state.

183