Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Direct Calculation of Permeability and Permittivity For A Left-Handed Metamaterial

Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

APPLIED PHYSICS LETTERS VOLUME 77, NUMBER 14 2 OCTOBER 2000

Direct calculation of permeability and permittivity for a left-handed


metamaterial
D. R. Smith,a) D. C. Vier, N. Kroll, and S. Schultz
Department of Physics, University of California, San Diego, La Jolla, California 92093
共Received 16 June 2000; accepted for publication 14 August 2000兲
Recently, an electromagnetic metamaterial was fabricated and demonstrated to exhibit a
‘‘left-handed’’ 共LH兲 propagation band at microwave frequencies. A LH metamaterial is one
characterized by material constants—the permeability and permittivity—which are simultaneously
negative, a situation never observed in naturally occurring materials or composites. While the
presence of the propagation band was shown to be an inherent demonstration of left handedness,
actual numerical values for the material constants were not obtained. In the present work, using
appropriate averages to define the macroscopic fields, we extract quantitative values for the effective
permeability and permittivity from finite-difference simulations using three different approaches.
© 2000 American Institute of Physics. 关S0003-6951共00兲03440-9兴

In a paper published in 1968, Veselago introduced the utilizing a commercial electromagnetic-mode solver.5 The
concept of a left-handed 共LH兲 medium—a medium in which standard procedure is to discretize a unit cell of the structure
both the permittivity ⑀ and the permeability ␮ are simulta- 共Fig. 1, inset兲, and apply periodic boundary conditions, al-
neously negative.1 Veselago predicted that LH media would lowing a phase advance in one of the directions. In the case
have unique and potentially useful properties, as many basic where we are looking for an effective magnetic response
electromagnetic phenomena, such as the Doppler shift, from the rings, we focus on electromagnetic modes polarized
Snell’s law, and Cerenkov radiation would be ‘‘reversed.’’ such that the magnetic field is parallel to the ring axes. Fig-
The limitation in demonstrating a LH media, however, was ure 1 shows the resulting dispersion curve, which displays a
the lack of a material example possessing a magnetic re- frequency band gap. The effective refractive index n( ␻ ) is
sponse to electromagnetic radiation sufficient to result in a related to both the permeability and permittivity according to
negative permeability. We recall that the magnetic suscepti- n( ␻ )⫽ 冑⑀ eff(␻)␮eff(␻); thus, the band gap of Fig. 1, charac-
bility can be negative 共e.g., diamagnetism兲, but is typically terized by imaginary values of n( ␻ ), only indicates that one
very small, and therefore, insufficient to bring ␮ ⫽1⫹ ␹ or the other material parameter has fallen below zero.
negative. A means of demonstrating the negative permeability was
In 1999, Pendry et al. proposed an artificial system, put forward by Smith et al.,4 who combined the SRR me-
made only of nonmagnetic conducting materials that could dium with a second medium of interacting wires 共Fig. 2,
exhibit a large response to the magnetic field of electromag- inset兲. The wire medium alone can be characterized by an
netic radiation.2 By building a resonance into the structure, effective permittivity having the same form as a dilute, col-
very large positive and negative effective magnetic suscepti- lisionless plasma, or
bilities could be obtained. While the idea of creating mag-
netic materials from conductors is not new,3 the structures
that Pendry et al. suggested were particularly suited for this
purpose, having a very high Q factor and nearly exclusive
magnetic-field coupling. The generic frequency dependence
of the effective permeability that Pendry et al. calculated was

F ␻ 20
␮ eff共 ␻ 兲 ⫽1⫺ , 共1兲
␻ 2 ⫺ ␻ 20 ⫺i ␻ ⌫

where F, ⌫, and ␻ 0 are constants relating to the geometry


and materials composing the artificial medium. Note that,
due to the resonance, a region of negative permeability oc-
curs over a finite frequency band.
The effective magnetic medium used in the recent dem-
onstration of an artificial LH medium4 was a periodic array
of split ring resonators 共SRRs兲, which we also study here. FIG. 1. Dispersion curve for an array of split ring resonators 共SRRs兲. The
All of the simulation data presented here result from finite- strips composing the SRRs are 0.8 mm wide and 0.05 mm thick. The outer
difference calculations of SRRs with a square cross section, ring has an outer width of 6.6 mm, with a gap of 0.2 mm between outer and
inner rings. The width of the splits is 0.3 mm. The rings are in a unit cell
with dimensions 8⫻8⫻10 mm 共10 mm in the direction of the wire兲. The
a兲
Author to whom correspondence should be addressed; electronic mail: horizontal axis is the phase advance per unit cell, or kd, where k is the
drs@sdss.ucsd.edu wavenumber 共2␲ divided by the free-space wavelength兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 177.20.130.9
0003-6951/2000/77(14)/2246/3/$17.00 2246 © 2000 American Institute of Physics
On: Wed, 20 Aug 2014 15:19:36
Appl. Phys. Lett., Vol. 77, No. 14, 2 October 2000 Smith et al. 2247

FIG. 2. Dispersion curve 共dashed line兲 for the combined SRR and wire FIG. 3. Calculated permeability 共open and closed circles, triangles兲 and
mediums, using the same dimensions as in Fig. 1. The solid lines correspond permittivity 共diamonds兲 for the left-handed propagation band of the compos-
to the dispersion curves obtained without the wires 共as in Fig. 1兲, shown ite SRR-wire structure. Open circles: 具 B 典 / 具 H 典 method; closed circles: z,n
here to help identify the LH region. method; triangles: currents method. The permittivity was calculated by the
z,n method only.

␻ 2p d is the length of an edge of a cell, and i is one of the x, y, or


⑀ eff共 ␻ 兲 ⫽1⫺ 2 , 共2兲
␻ z components. The averages for 具E典 and 具H典 involve line
integrals along the edges of the unit cell, while the averages
where ␻ p is an effective plasma frequency relating to the of 具B典 and 具D典 involve surface integrals over the faces of the
details of the wire radius and spacing. The essential feature cell. By following this scheme, the details of the actual struc-
of Eq. 共2兲 is that below ␻ p , the effective permittivity is ture are removed into the field averages, from which we can
negative. When the wire and SRR mediums are combined, define the effective material parameters as
the overlap of the negative permittivity of the wire medium
with the frequency gap region of the SRR medium results in 具 B i典 具 D i典
a propagation band 共Fig. 2, dashed curve兲 where there was ␮ eff
i, j
⬅ , ⑀ eff
i, j
⬅ . 共4兲
具H j典 具E j典
previously attenuation for the SRR medium alone. The pres-
ence of a propagation band in the combined structure thus Having defined these field averages, we can now outline
implies that the SRR gap is due to a negative permeability, three computational methods to calculate the material param-
and furthermore, indicates that for this frequency region the eters. The most straightforward of these is to use the field
structure is left handed. averages of Eq. 共3兲 to compute the material parameters di-
While this is a valid demonstration of an effective LH rectly by the prescription shown in Eq. 共4兲. When there is
medium, it is necessary to obtain quantitative values for the appreciable phase advance across a unit cell, however, we
material constants, both to rule out other possible mecha- note that the averages are ‘‘incorrect’’ by a factor relating to
nisms for the propagation band, and to facilitate the design of the phase variation. To illustrate the problem, suppose we
practical structures. In designing improved microwave ab- have a plane wave propagating through vacuum ( ⑀ ⫽ ⑀ 0 , ␮
sorbing materials, for example, achieving the condition ⑀ eff ⫽ ␮ 0 ), and we discretize the space into cubes with length d.
⫽⫺⑀0 , ␮eff⫽⫺␮0 would be very desirable, as electromag- From Eqs. 共3兲 and 共4兲, we have, for example,
netic waves at any angle of incidence would be entirely
transmitted into such a material, with nearly zero-reflected 具 B 典 ⫽ ␮ 0 H 0 共 e ikd ⫺1 兲 /ikd⫽ ␮ 0 H 0 f 共 k 兲 , 共5兲
power.1
where H 0 is the applied magnetic field. Clearly, the ratio
Given that we have an effective medium, with only
具 B 典 / 具 H 典 , where 具H典 is taken at the zero phase advance side
vacuum and responding currents ( ⑀ ⫽ ⑀ 0 , ␮ ⫽ ␮ 0 ) it is clear
of the cell, does not equal ␮ 0 when kd is finite. Nevertheless,
that we must apply appropriate averaging methods to com-
we can still maintain a sensible definition of the quantities
pute effective material parameters. In fact, we may view the
averaged over a surface, by dividing out the factor f (k) ap-
process as a discretization of Maxwell’s equations, in which
pearing in Eq. 共5兲 from all flux averaged quantities.
we replace the details of each unit cell by averaged fields and
Alternatively, we can utilize the calculated surface cur-
material constants. We follow Pendry et al.2 in introducing
rent density in the SRR 共and wire兲 to determine the material
the following averaging scheme for the fields, and will dis-
response. In this second method, we make use of the defini-
cuss this further in a separate publication:
tions relating the magnetization and polarization to the mag-
netic flux and electric field, respectively, or M⫽ ␹ M H and
具 E i 典 ⫽d ⫺1 冕 E•dxi , 具 D i 典 ⫽d ⫺2 ⑀ 0 冕 E•dsi , P⫽ ␹ E E. The average magnetization and polarization can be
found by integration of the current densities as follows:
共3兲

This article is copyrighted ␮ ⫺1


具 H i 典 ⫽as 冕
d ⫺1 inB•dx
0indicated i, 具 BReuse
the article. ⫺2

i 典 ⫽dof AIP B•ds i . is subject to the terms
content
On: Wed, 20 Aug 2014 15:19:36
1
2V
冕 r⫻jdV, P⫽
1

jdV, Downloaded to共6兲
M⫽at: http://scitation.aip.org/termsconditions.
i␻V
IP: 177.20.130.9
2248 Appl. Phys. Lett., Vol. 77, No. 14, 2 October 2000 Smith et al.

structure consisting of the SRR medium and the wire me-


dium. Figure 3 shows data for the LH composite structure.4
The calculations confirm that over the propagation band
shown as the dashed curve in Fig. 2, both the permittivity
and permeability are less than zero. Both methods requiring
only field averages are in agreement, while the method using
currents shows a rapid divergence from the other curves,
increasing with increasing phase advance. Nevertheless, the
same qualitative behavior is indicated by all three methods.
We conclude by presenting in Fig. 4 the composite per-
meability for the SRR medium, which displays the resonant
form predicted by Eq. 共1兲. The curve was pieced together
from the permeability calculation of the SRR medium alone,
and the permeability calculation of the SRR with the wires.
The smoothness of the transition at ␮ eff⫽0, where the data
FIG. 4. Calculated permeability as a function of frequency for the SRR from the SRRs alone join the data from the SRRs and wires,
medium. is indicative that any interactions between the wires and the
SRRs are negligible.
where it is assumed that the currents are confined within each Having now demonstrated the capability to characterize
cell. the material constants of artificial media, we are in a position
A third method for determining ⑀ eff and ␮ eff involves the to efficiently design and optimize LH structures for specific
calculation of the wave impedance z from the averaged field applications.
quantities. This latter method is robust, and is similar to that This work was supported by DARPA through a grant
introduced by Contopanagos et al. for calculations on photo- from ONR 共Contract No. N00014-00-1-0632兲, and the DOE
nic band-gap materials, where the free-space wavelength is 共Contract Nos. DE-FG-03-093ER40793 and DE-FG-03-
typically on the order of, or even smaller than, the unit-cell 93ER40759兲.
dimensions.6 An average Poynting vector can be defined
from the averaged fields as

冉 冊 V. G. Veselago, Sov. Phys. Usp. 10, 509 共1968兲.


1
1 1 具E典2
S⫽ Re共 具 E典 ⫻ 具 H典 * 兲 ⫽ Re
2
k̂ , 共7兲 J. B. Pendry, A. J. Holden, D. J. Robbins, and W. J. Stewart, IEEE Trans.
2 2 z Microwave Theory Tech. 47, 2075 共1999兲.
3
I. S. Schelkunoff and H. T. Friis, in Antennas: Theory and Practice, edited
where z⫽ 冑␮ eff /⑀eff. By combining the determination of z by S. Sokolnikoff 共Wiley, New York, 1952兲, p. 584.
with the dispersion relation, which provides n⫽ 冑⑀ eff␮eff
4
D. R. Smith, W. Padilla, D. C. Vier, S. C. Nemat-Nasser, and S. Schultz,
Phys. Rev. Lett. 84, 4184 共2000兲.
共Figs. 1 and 2兲, we can extract values for ⑀ eff and ␮ eff . 5
T. Weiland, AIP Conf. Proc. 297, 291 共1994兲.
We applied all three methods to compute the material 6
H. F. Contopanagos, C. A. Kyriazidou, W. M. Merrill, and N. G. Alex-
parameters for the SRR medium alone, and for the composite opoulos, J. Opt. Soc. Am. A 16, 1682 共1999兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 177.20.130.9
On: Wed, 20 Aug 2014 15:19:36

You might also like