See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235410466

Negative Index Materials Using Simple Short Wire Pairs

Article in Physical Review B · January 2006

DOI: 10.1103/PhysRevB.73.041101

CITATIONS READS
450 525

5 authors, including:

Jiangfeng Zhou Gary Tuttle

University of South Florida Iowa State University
92 PUBLICATIONS 10,170 CITATIONS 77 PUBLICATIONS 3,943 CITATIONS

SEE PROFILE SEE PROFILE

Costas M. Soukoulis
Iowa State University
635 PUBLICATIONS 55,990 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jiangfeng Zhou on 10 November 2015.

The user has requested enhancement of the downloaded file.

 RAPID COMMUNICATIONS

PHYSICAL REVIEW B 73, 041101共R兲 共2006兲

Negative index materials using simple short wire pairs

Jiangfeng Zhou,1 Lei Zhang,1 Gary Tuttle,1 Thomas Koschny,2,3 and Costas M. Soukoulis2,3
1Department of Electrical and Computer Engineering and Microelectronics Research Center, Iowa State University,
Ames, Iowa 50011, USA
2
Ames Laboratory and Department of Physics and Astronomy, Iowa State University, Ames, Iowa 50011, USA
3Institute of Electronic Structure and Laser–FORTH, and Department of Materials Science and Technology,

University of Crete, Heraklion, Crete, Greece

共Received 27 October 2005; published 4 January 2006兲

Negative refraction is currently achieved by a combination of artificial “electric atoms” 共metallic wires with
negative electrical permittivity ␧兲 and artificial “magnetic atoms” 共split-ring resonators with negative magnetic
permeability ␮兲. Both ␧ and ␮ must be negative at the same frequency, which is not easy to achieve at higher
than THz frequencies. We introduce improved and simplified structures made of periodic arrays of pairs of
short metal wires and continuous wires that offer a potentially simpler approach to building negative index
materials. Using simulations and microwave experiments, we have investigated the negative index n properties
of short wire-pair structures. We have measured experimentally both the transmittance and the reflectance
properties and found unambiguously that n ⬍ 0. The same is true for ␧ and ␮. Our results show that short
wire-pair arrays can be used very effectively in producing materials with negative refractive indices.

DOI: 10.1103/PhysRevB.73.041101 PACS number共s兲: 41.20.Jb, 42.25.Bs, 42.70.Qs, 78.20.Ci

The first demonstration of a left-handed 共LH兲 material1,2 imaginary part has been observed yet at the THz region. One
by the UCSD group, in 2000, following the work by Pendry reason is that it is very difficult to measure with the existing
et al.,3,4 underscores the relevance of utilizing structured ma- topology of SRRs and continuous wires both the transmis-
terials to create electromagnetic response not available in sion, T, and reflection, R, along the direction parallel to the
naturally occurring materials. This LH material made use of plane of the SRRs. So there is a need for alternative, im-
an array of conducting, nonmagnetic elements to achieve a proved, and simplified designs that can be easily fabricated
negative effective permeability, ␮共␻兲, and an array of con- and experimentally characterized. Currently, there is much
ducting continuous wires to achieve a negative effective per- interest in pushing the frequency range for NIM behavior
mittivity, ␧共␻兲, the simultaneous combination of which had into the infrared and optical regions of the spectrum.
never before been observed in any previously known mate- Designing a material structure to have an effective nega-
rial. While a number of theoretical concerns were raised tive refractive index can be achieved by having simulta-
shortly after the reporting of these experiments, subsequent neously negative permeability and negative permittivity over
measurements5–8 on similar metamaterials have firmly estab- some range of frequencies. Nearly all negative index mate-
lished the foundation of negative refraction. LH materials rials that have been investigated up until this point have been
will display unique “reversed” electromagnetic properties, as based on the use of split-ring resonators, which have been
discussed by Veselago9 long before such materials existed. described in detail elsewhere.1,2,5,6 Conceptually, an SRR
All of these reversals stem from the fact that a LH material is starts with a loop of wire that provides an inductive response
characterized by a negative refractive index, a property that to an incident field. Cutting out a short section of the loop
does not exist in any known natural material, and represents leads to a capacitive gap. The inductance 共L兲 and capacitance
a new regime of physics. 共C兲 together form an LC resonator, with resonance frequency
In the past few years there has been ample proof for the ␻R = 1 / 冑LC. For an electromagnetic wave incident with its
existence of negative index materials 共NIMs兲 in the GHz wave vector parallel to the plane of the SRR and the mag-
frequency range.1–8 All NIM implementations to date have netic field perpendicular to the SRR, the response at the reso-
utilized the topology proposed by Pendry, consisting of split- nance frequency will be as if the SRR had a negative perme-
ring resonators 共SRRs兲 共rings with gaps, providing the nega- ability. To build a negative index material, an array of SRRs
tive ␮兲 and continuous wires 共providing the negative ␧兲. is combined with a grid of continuous metal wires, which is
Many groups were able to fabricate1,2,5–8 NIMs with an index known to exhibit a negative permittivity,3 so that the two
of refraction n = −1 with losses of less than 1 dB/ cm. Re- effects together lead to NIM behavior.
cently, different groups observed indirectly10–13 negative ␮ at Recent theoretical work14 has shown that using pairs of
the THz region. In most of the THz experiments10–12 only finite length wires would not only allow replacing the SRRs
one layer of SRRs was fabricated on a substrate and the as magnetic resonators but could also give simultaneously a
transmission, T, was measured only for propagation perpen- negative ␧ in the same frequency range, and therefore a
dicular to the plane of the SRRs, exploiting the coupling of negative n, without the need for additional continuous wires.
the electric field to the magnetic resonance of the SRR via The condition to obtain simultaneously negative ␮ and ␧ by
asymmetry.11 This way it is not possible to drive the mag- pairs of finite metallic wires is very restrictive. Recent
netic permeability negative. Also, no negative n with a small experiments15 have not shown evidence of negative n in the

1098-0121/2006/73共4兲/041101共4兲/$23.00 041101-1 ©2006 The American Physical Society

 RAPID COMMUNICATIONS

ZHOU et al. PHYSICAL REVIEW B 73, 041101共R兲 共2006兲

1 1 1 co
fm = = = , 共1兲
2␲冑LC ␲ l冑␧r␧o␮o ␲l冑␧r

where co is the speed of light in vacuum. From 共1兲, the

magnetic-resonance frequency is inversely proportional to
the length of the wires in the pairs, but does not depend on
the wire widths or separation between the wires. Resonance
frequencies determined by detailed simulations of short wire-
pair structures with various combinations of parameters
共l , w , ts兲 show good agreement with the simple formula
above. However, our simulations have shown that it is diffi-
cult to obtain a negative n with only pairs of short wires.
This is because the electric resonance of the short wires is
usually well above the magnetic-resonance frequency, thus
preventing ␧ and ␮ from becoming simultaneously negative.
To realize NIM behavior, the short wire pairs must be com-
bined with a continuous wire grid that provides the extra
negative permittivity. In our structure, two additional con-
tinuous wires are placed on either side of the short wire pairs.
Repeating this basic structure periodically in the x, y, and z
directions would result in a NIM structure.
The short wire-pair arrangement has a distinct advantage
over conventional SRRs. Since all of the features of the wire-
pair NIM lie in parallel planes, conventional microfabrica-
tion techniques can be used to build the structures. In par-
FIG. 1. 共Color online兲 共a兲 Schematic representation of one unit ticular, the methods for making complex multilevel
cell of the wire-pair structure. 共b兲 Photograph of one side of a interconnects in integrated circuits could be applied directly
fabricated microwave-scale wire-pair sample. to making wire-pair NIMs.
To examine the potential usefulness of wire-pair struc-
tures as NIMs, we characterized the properties of the wire
short wire-pair cases that were studied. This is in contrast pair of Fig. 1共a兲 using simulations and microwave measure-
with the claims16 that one can get negative n at THz frequen- ments, and then used these results to determine the expected
cies. properties of NIMs built from the wire-pair building blocks.
Simulations of wire-pair structures were done with CST Mi-
In this Communication, we report our investigations of
crowave Studio 共Computer Simulation Technology GmbH,
wire-pair structures as alternatives to conventional SRR-
Darmstadt, Germany兲, which uses a finite-difference time-
based NIMs. The basic structure of a single unit cell of the
domain method to determine reflection and/or transmission
wire-pair NIM is shown in Fig. 1共a兲. In the wire-pair ar- properties of metallodielectric structures. In the simulations,
rangement, the conventional SRR is replaced with a pair of the dielectric properties of the metal patches were handled
short parallel wires and the continuous wires are preserved. with a frequency dependent Drude model. The detailed cal-
The short wire pair consists of a pair of metal patches sepa- culations were used to determine reflection and transmission
rated by a dielectric spacer of thickness ts. In essence, the coefficients from a single unit cell. Experimental transmis-
short wire pair is a “two-gap” SRR that has been flattened to sion and reflection data were obtained by building and mea-
result in the wire-pair arrangement.15 For an electromagnetic suring microwave-frequency versions of the wire-pair struc-
wave incident with wave vector and field polarization as tures. These were fabricated using Rogers 5880 printed
shown in Fig. 1共a兲, the short wire pair will exhibit both in- circuit board stock with a dielectric layer thickness of
ductive 共along the wires兲 and capacitive 共between the upper 254 ␮m and a listed relative dielectric constant of 2.53. The
and lower adjacent ends of the short wires兲 behavior and will circuit board was coated on both sides with 10-␮m-thick
possess magnetic resonance providing a negative permeabil- layers of copper. The copper was formed in the wire-pair
ity. The inductance L of a short wire pair is approximately patterns using conventional photolithography techniques. For
given, as inductance of parallel plates, by L = ␮o(l · ts) / w, the samples reported here 共both simulations and experi-
where l is the length of the short wires, w is the width, and ts ments兲, the widths of all metal lines were 1 mm. The length
is the separation distance between the short wires. The ca- of the short wire pairs was 7 mm, and the unit-cell size was
pacitance C of the short wire pairs can be written as the two 9.5⫻ 7 ⫻ 2.274 mm3. The total sample size was 7 ⫻ 10⫻ 1
plate capacitor formula for the upper and lower halves of the unit cells, resulting in approximately square samples. A pho-
short wire pair C = ␧r␧o共l · w兲 / 共4ts兲, where ␧o is permittivity tograph of one side of a complete sample is shown in Fig.
in vacuum and ␧r is the relative dielectric constant of the 1共b兲. With these patterned dimensions on the printed circuit
region between the wires. Then the frequency of the mag- board material, the resonances for NIM behavior were ex-
netic resonance is pected to occur near 13.7 GHz.

041101-2
 RAPID COMMUNICATIONS

NEGATIVE INDEX MATERIALS USING SIMPLE SHORT … PHYSICAL REVIEW B 73, 041101共R兲 共2006兲

FIG. 3. 共Color online兲 Extracted electromagnetic properties of a

FIG. 2. 共Color online兲 Simulated 共red thin line兲 and measured periodic array of wire-pair unit cells, using the simulated 共red thin
共blue thick-dotted line兲 response to electromagnetic radiation inci- line兲 and measured 共blue thick-dotted line兲 data of Fig. 2. Real 共a兲
dent on the wire-pair structures: 共a兲 transmission, 共b兲 reflection. and imaginary 共b兲 parts of the refractive index. Real 共c兲 and imagi-
nary 共d兲 parts of the permeability. Real 共e兲 and imaginary 共f兲 part of
the permittivity. The negative-index behavior can be seen clearly
Transmission and reflection properties of a single-layer near 14 GHz in 共a兲.
structure were measured over the frequency range of 12–16
GHz using a network analyzer 共HP 8510兲 and a pair of stan-
dard gain horn antennas serving as a source and receiver, as simulations. There are three potential causes for the shift: 共1兲
shown in the insets of Fig. 2. In the transmission measure- The external resonances in the measurements may be mask-
ments, the microwaves were incident normal to the sample ing the true peak 共dip兲 in the measured data. 共2兲 The actual
surface. This is a tremendous simplification relative to the dielectric constant of the circuit board material may be
conventional SRRs and wires where the incident electromag- slightly lower than the value used in the simulations. 共3兲 The
netic waves have to propagate parallel to the sample surface. wires on the front and back of the dielectric layer may be
With the conventional orientation of the SRRs, it is almost misaligned slightly, which would have the same effect as
impossible to do these types of measurements at the THz shortening the wires. The dip in the reflection at 15.7 GHz is
region, since only single-layer samples are usually due to the fact that the sample impedance z = 1 at that fre-
fabricated.10,11 Transmission measurements were calibrated quency, and so no reflection is possible.
to the transmission between the horns with the sample re- Using the transmission and reflection results from a single
moved. The reflection measurements were done by placing layer, we can extract the effective refractive index that would
the source and receiving horns on the same side of the result if a periodic multilayer sample were built using the
sample and bouncing the microwave signal off the sample. single-layer structure as a building block. The details of the
The source and receiver horns were each inclined with an numerical retrieval procedure have been described
angle of about 7.5° with respect to normal on the sample elsewhere.17–19 In performing the retrieval, we have assumed
surface. The reflection measurement was calibrated using a a z-direction size of the unit cell of 2.274 mm. This interlayer
sample-sized sheet of copper as a reflecting mirror. In both spacing is an adjustable parameter in the retrieval process.
measurements, the electric field of the incident wave was Smaller spacing would lead to stronger negative index fea-
polarized parallel to the long dimension of the wires. 共For tures in the periodic structure, as long as the distance be-
perpendicular polarization, the transmission was nearly tween the short wires is small compared to the length of the
100%, independent of frequency in the resonance region, and unit cell, but may also result in more complicated fabrication
reflection was essentially zero.兲 procedures in building a multilayer structure. In choosing
The calculated and measured transmission spectra are 2.274 mm as the separation distance for the numerical ex-
shown in Fig. 2共a兲. Figure 2共b兲 shows the calculated and traction, we attempted to strike a balance between good
measured reflection spectra. There is good qualitative agree- negative index properties and having a separation distance
ment between simulations and measurements. The measured that is in line with the other feature sizes of the structure. The
spectrum does show resonance peaks and valleys due to re- extracted permittivity, permeability, and refractive index are
flections between the receiving horn and the sample. Also, shown in the various parts of Fig. 3. The plots show that the
there is a distinct frequency difference between critical real part of the permittivity is negative over most of the
points in the two sets of curves. The frequencies of the mea- measured range.20 The real part of the permeability is nega-
sured transmission resonance peak and the corresponding re- tive over a resonance band near 13.8 GHz for the simulation
flection dip near 14.0 GHz are about 2% higher than in the and at about 14.0 GHz for the measurement. Notice also that

041101-3
 RAPID COMMUNICATIONS

ZHOU et al. PHYSICAL REVIEW B 73, 041101共R兲 共2006兲

the product of Im共␧兲 and Im共␮兲 is negative. This is a well These results show clearly the viability of using short
known and real effect, which comes from the periodic effects wire pairs to build negative index materials, if combined
of the retrieval procedure.11 It is not relevant to the discus- with additional continuous wires. It is likely that modifica-
sion of our results. The extracted real part of the refractive tions of the basic structure studied here may improve or alter
index is negative21 over a narrow band at 13.8 GHz for the the NIM properties. Also, wire-pair arrangements with sig-
simulations and 14.0 GHz in the experiments, dipping as low nificantly different geometries may lead to negative index
materials. The relative ease of fabricating wire-pair struc-
as −2 using measured data, and to less than −3 from the
tures may hasten the development of NIMs working at opti-
simulation. The ratio of the imaginary part of n to the real
cal wavelengths.
part of n is 41 , which means that we have left-handed propa-
gation with ␧, ␮, and n negative. Our preliminary numerical We gratefully acknowledge the support of Ames Labora-
results show that if our structure scales down by a factor of tory, which is operated by Iowa State University under Con-
200, it will give a negative index of refraction at THz fre- tract No. W-7405-Eng-82, EU FET project DALHM, and
quencies, with both ␧ and ␮ negative. DARPA 共Contract No. HR0011-05-C-0068兲.

1 D.R. Smith, W. J. Padilla, D. C. Vier, S. C. Nemat-Nasser, and S. chev, V. P. Drachev, and A. V. Kildishev, Opt. Lett. 30, 3356
Schultz, Phys. Rev. Lett. 84, 4184 共2000兲. 共2005兲.
2 R. A. Shelby, D. R. Smith, and S. Schultz, Science 292, 77 17 D. R. Smith, S. Schultz, P. Markos, and C. M. Soukoulis, Phys.

共2001兲. Rev. B 65, 195104 共2002兲.

3 J. B. Pendry, A. J. Holden, D. J. Robbins, and W. J. Stewart, 18 D. R. Smith, D. C. Vier, Th. Koschny, and C. M. Soukoulis, Phys.

IEEE Trans. Microwave Theory Tech. 47, 2075 共1999兲; J. B. Rev. E 71, 036617 共2005兲.
19 Th. Koschny, P. Markos, E. N. Economou, D. R. Smith, D. C.
Pendry, A. J. Holden, W. J. Stewart, and I. Youngs, Phys. Rev.
Lett. 76, 4773 共1996兲. Vier, and C. M. Soukoulis, Phys. Rev. B 71, 245105 共2005兲.
4
J. B. Pendry, Phys. Rev. Lett. 85, 3966 共2000兲. 20
We have also done experiments and simulations for samples with
5 C. G. Parazzoli, R. B. Greegor, K. Li, B. E. C. Koltenbah, and M.
only pairs of short wires. For this case both the experiments and
Tanielian, Phys. Rev. Lett. 90, 107401 共2003兲; K. Li, S. J. the simulations have shown that indeed ␮ is negative, however ␧
McLean, R. B. Greegor, C. G. Parazzoli, and M. H. Tanielian, is not negative in this region and therefore n is also not negative.
Appl. Phys. Lett. 82, 2535 共2003兲. The extra continuous wires are needed to drive ␧ to become
6 M. Bayindir, K. Aydin, E. Ozbay, P. Markos, and C. M. Soukou-
negative, without changing the magnetic response shown in Fig.
lis, Appl. Phys. Lett. 81, 120 共2002兲; K. Aydin, K. Guven, M. 3共c兲.
21 In lossy materials it is possible to have the real part n be negative,
Kafesaki, L. Zhang, C. M. Soukoulis, and E. Ozbay, Opt. Lett.
29, 2623 共2004兲. without having the real parts of ␧ and ␮ simultaneously nega-
7
E. Cubukcu, K. Aydin, E. Ozbay, S. Foteinopoulou, and C. M. tive. This is the case of the recent work of S. Zhang, W. J. Fan,
Soukoulis, Nature 共London兲 423, 604 共2003兲; Phys. Rev. Lett. N. C. Panoiu, K. J. Malloy, R. M. Osgood, and S. R. J. Brueck.
91, 207401 共2003兲. 关Phys. Rev. Lett. 95,137404 共2005兲兴. This can happen if the
8 P. V. Parimi, W. T. T. Lu, P. Vodo, and S. Sridhar, Nature
imaginary parts of ␧ and ␮ are sufficiently large, because in a
共London兲 426, 404 共2003兲. lossy material n = n⬘ + in⬙, and we also have that n = ␧z and z
9 V. G. Veselago, Sov. Phys. Usp. 10, 509 共1968兲.
= 冑␮ / ␧. After some algebra we obtain that n⬘ = ␧⬘z⬘ − ␧⬙z⬙ and
10
T. J. Yen, W. J. Padilla, N. Fang, D. C. Vier, D. R. Smith, J. B. z = 冑␮⬘␧⬘ + ␮⬙␧⬙ / ␧2 + i共␮⬙␧⬘ − ␮⬘␧⬙ / ␧2兲, so it is possible to have
Pendry, D. N. Basov, and X. Zhang, Science 303, 1494 共2004兲. n⬘ ⬍ 0, provided that ␧⬙z⬙ ⬎ ␧⬘z⬘. In this scenario, which occurs
11
S. Linden, C. Enkrich, M. Wegener, J. F. Zhou, T. Koschny, and at the low-frequency side of the n⬘ ⬍ 0 region in Fig. 3共a兲, how-
C. M. Soukoulis, Science 306, 1351 共2004兲. ever, the imaginary parts lead to dominant losses such that we
12
S. Zhang, W. J. Fa, B. K. Minhas, A. Frauenglass, K. J. Malloy, have a transmission gap with some negative phase shift rather
and S. R. J. Brueck, Phys. Rev. Lett. 94, 037402 共2005兲. than LH transmission 共with some losses兲. This type of negative
13 N. Katsarakis, G. Konstantinidis, A. Kostopoulos, R. S. Penciu, T.
n should not be considered LH behavior. In our experiments,
F. Gundogdu, M. Kafesaki, E. N. Economou, T. Koschny, and although we have considerable imaginary parts, the behavior is
C. M. Soukoulis, Opt. Lett. 30, 1348 共2005兲. still dominated by the negative real part of n at the high-
14
V. A. Podlovsk, A. K. Sarychev, and V. M. Shalaev, J. Nonlinear frequency side where we find the LH behavior. As one can see
Opt. Phys. Mater. 11, 65 共2002兲; Opt. Express, 11, 735 共2003兲. from the experimental data of Figs. 3共a兲 and 3共b兲, we obtain
15 G. Dolling, C. Enkrich, M. Wegener, J. F. Zhou, C. M. Soukoulis,
n⬘ / n⬙ = 3.5 at n⬘ = −1, which ratio improves to ⬃15 for n⬘ =
and S. Linden, Opt. Lett., 30, 3198 共2005兲. −0.76. The simulation data gives at least n⬘ / n⬙ ⬇ 3.0 for n⬘ =
16
V. M. Shalaev, W. Cai, U. K. Chettiar, H. K. Yuan, A. K. Sary- −1.7.

041101-4

View publication stats