Synthesis of Artificial Magnetic Conductors by Using Multilayered Frequency Selective Surfaces

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

196

IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 1, 2002

Synthesis of Artificial Magnetic Conductors by Using


Multilayered Frequency Selective Surfaces
Agostino Monorchio, Member, IEEE, Giuliano Manara, Senior Member, IEEE, and Luigi Lanuzza

AbstractIn this letter, the combination of a multilayered dielectric structure, a capacitive frequency selective surface (FSS)
and a perfectly electric conducting (PEC) ground plane is proposed
for realizing high-impedance surfaces. These surfaces behave like
an artificial magnetic conductor (AMC) in a specific frequency
range; the inclusion of the dielectric layers allows us to enhance the
angular properties of the AMC as well as the frequency bandwidth
of the device. In order to obtain the proper values of the design parameters, a genetic algorithm (GA) is employed that makes use of
an electromagnetic solver based on the method of moments (MoM)
to evaluate the scattering properties of the structure. A key step in
the design procedure is the inclusion in the fitness function of the
electromagnetic response of the high-impedance surface with respect to the illumination angle. The synthesized structure shows
the desired frequency performance and reveals robust as concerns
the stability of the solution with respect to a wide interval of illumination angles.
Index TermsArtificial magnetic conductors, frequency
selective surfaces, genetic algorithms, metamaterials, photonic
bandgap.

I. INTRODUCTION

T is well known that a hypothetical perfect magnetic conductor might be very useful in a large variety of microwave
applications; for instance a magnetic ground plane can improve
the performance of printed dipole antennas, creating their
equiverse image currents. Recently, photonic bandgap (PBG)
structures have been widely investigated for their behavior as
artificial magnetic conductors (AMC) at the corresponding
stopband frequency, considerably reducing the tangential
magnetic field component. Either three-dimensional [1] or
two-dimensional [2] structures have been proposed to realize
this goal, both dielectric and metal-dielectric. In particular, a
two-dimensional structure can be realized by using a frequency
selective surface (FSS) screen backed by a perfectly electric
conducting (PEC) ground plane. This configuration is desirable
for its low-cost manufacturing and for the easy integration
in microwave devices. In this work, we propose to insert the
FSS screen in a multilayered dielectric arrangement in order
to obtain the desired reflecting properties of the AMC at wide
incidence angles as well as for wide frequency bands. However,
the design of such a complex structure is not an easy matter,
because it involves the optimization of many parameters, as
for instance, the FSS screen basic periodicity cell shape and
dimensions, as well as the dielectric layers thickness and

Manuscript received August 5, 2002; revised September 18, 2002.


The authors are with the Department of Information Engineering,
University of Pisa, I-56126 Pisa, Italy (e-mail: a.monorchio@iet.unipi.it;
g.manara@iet.unipi.it; l.lanuzza@iet.unipi.it).
Digital Object Identifier 10.1109/LAWP.2002.807956

electric properties. Genetic algorithms are global stochastic


search methods that are very suitable to this purpose [3]. They
are able to determine a global minimum (or a maximum) of a
multivariables function representing the problem to be solved,
by evolving proper populations of solutions. The chromosomes
associated with the individuals of these populations are a set of
genes, representing coded versions of the optimization parameters. As shown in the recent literature (see for instance [4][9]),
a genetic algorithm (GA) evolutionary strategy can be suitably
applied to design FSS filters. In this work, GA is employed
to determine the shape and the dimensions of the metallic
elements of the FSS screen, as well as the permittivity and the
thickness of each dielectric layer for realizing an AMC in the
pertinent frequency band. A key step in the design procedure
is the inclusion in the fitness function of the electromagnetic
response of the high-impedance plane with respect to the
incidence angle. The synthesized structure shows the desired
frequency behavior and reveals robust as concerns the stability
of the solution with respect to wide angles of illumination. An
example of a designed high-impedance surface is presented
and numerically tested to demonstrate the effectiveness of the
optimization procedure.
II. FORMULATION
In Fig. 1, we show the structure of the multilayered FSS
screen backed by a PEC ground plane that has been used as a
working example. Indeed, a single dielectric FSS screen is not
able to provide the proper reflecting behavior in wide frequency
bands and/or wide intervals of incidence angles. To this aim, the
thickness and the dielectric permittivity of each layer, together
with the shape and the dimensions of the unit periodicity cell,
can be used as additional parameters to be optimized in order
to get the desired frequency response [4]. Each parameter has
been codified in a binary string to form a chromosome, representing the whole structure, as shown in Fig. 2. To achieve a
realistic design, the permittivities of the dielectric layers were
selected from among a set of 16 commercially available products, together with some hypothetical (but still realistic) exploratory values. The maximum number of different dielectric
layers has been initially fixed to eight, being the layers placed
indifferently above or below the FSS screen. However, extended
numerical experiments showed that satisfactory results can be
found by using only two or three dielectric layers over the FSS
screen and one layer as substrate. The basic periodicity cell is
subdivided into elementary pixels coded as 1s or 0s depending
on whether they are covered by a printed metal element or not
(see Fig. 2); either symmetric or asymmetric shapes of the elementary cell have been considered [4], [5].

1536-1225/02$17.00 2002 IEEE

MONORCHIO et al.: SYNTHESIS OF ARTIFICIAL MAGNETIC CONDUCTORS

197

Fig. 1. Configuration used for realizing the high-impedance surface.

Fig. 2. (a) Design parameters and (b) chromosome structure.

To evaluate the frequency and angular electromagnetic behavior of the surface, an electromagnetic solver based on the
method of moments (MoM) has been used, which is particularly
suited for the analysis of doubly periodic multilayer screens [8].
Since the MoM formulation is nothing but conventional, it will
be not discussed further for the sake of brevity.
The evaluation criterion of the structure performance has
been chosen as the root mean square difference between
the actual and the desired electric field reflection coefficient
for both TE and TM modes; i.e.,
j for a
PMC (this condition has been imposed separately on both the
real and the imaginary parts). To improve angular stability, two
analyses have been performed, one in the frequency domain
and the other by varying the incidence angle at the central
frequency of the band. As far as the frequency behavior is
concerned, we calculate the fitness function as
e

(1)

is the total number of frequencies


in the desired
where
band. A similar quantity is computed for different incidence
angles

(2)

is the total number of incidence angles


considwhere
ered at the central frequency of the desired band. In order to
ensure symmetry, we have imposed equal dimensions along the
periodicity directions for the unit cell. We note that the separation of the reflection coefficient into its real and imaginary part
also in the presence
allows us to maximize the amplitude of
of losses in the dielectric materials or in the conductor.
To get the global fitness value of the structure, a simple average between TE and TM responses is performed (ensuring
good polarization performance), while a weighted average is

198

IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 1, 2002

(a)

(b)
Fig. 3. (a) GA synthesized basic periodicity cell. (b) Complete view of the
FSS screen. Dark areas correspond to printed metallic elements.

evaluated between the frequency and incidence angle fitness


data, i.e.,
(3)
. A single point crossover has been used,
with
with probability
. The specific GA adopted in
this work employs a standard proportionate selection also called
the weighted roulette wheel selection scheme [3]. Moreover,
to improve the algorithm speed of convergence, a new kind of
selection strategy has been introduced. It is well known that
best results are usually found if GA starts from a good initial population. In our case, we build the initial population by
using the best chromosomes of several trial populations; once
the evolution has started, the next generations are created by
inserting only new chromosomes showing fitness values lower
than the older ones. In this way, overlapping generations are involved, resulting in a hybrid scheme between generational and
steady-state GAs [3]. A faster convergence might result in a
loss of genetic information, prematurely removing overall unfit
chromosomes, which, however, might contain good genes. To
obviate this problem, a variable linear mutation probability has
been applied. The mutation operator is rarely present in the initial evolution stages, when crossover acts maximally to achieve
a fast improvement of the populations, while it begins to act

Fig. 4. (a) Frequency and (b) angular properties of the synthesized structure
shown in Fig. 3. Frequency analysis is performed at normal incidence. In (b), a
comparison is made between GA solution and two =4 dielectric slabs backed
by a PEC with permittivity " = 1 and " = 9, respectively (amplitude >
0:14 dB).

when the fitness value settles [10]. In our implementation, muincreases if two successive generations
tation probability
increases linshow the same fitness value. In particular,
early between a minimum value of 1% to a maximum value
of 20% with 0.3% steps. If an improvement is observed,
returns to its lower value.
III. NUMERICAL RESULTS
Some numerical results are shown in this section to demonstrate the effectiveness of the optimization procedure. The design specifications call for a positive unit reflection coefficient
for the electric field, implying in phase total reflection for an
incident plane wave. The frequency range has been fixed to
1.71 1.88 GHz, and a symmetric shape for the FSS unit cell has
been chosen. The FSS structure is composed by three dielectric
layers (two as superstrates and one substrate) as shown in Fig. 1,
whose dielectric properties were chosen among a database comprising 16 materials with ranging from 1.1 to 80; the tangent
loss of materials was included in the design. As concerns the
total value of the fitness function, a weight of 65% is assigned
to the angular analysis while 35% is assigned to the frequency

MONORCHIO et al.: SYNTHESIS OF ARTIFICIAL MAGNETIC CONDUCTORS

behavior
. In Fig. 3 we show a sample of the results
obtained.
cm. The
The dimensions of the unit cell are
dielectric layers permittivity and thickness values are
,
cm,
, thickn
cm, for an
thickn
cm.; for
equivalent single superstrate of total thickness
, thickn
cm. In
the substrate we found
Fig. 4 we show the frequency and angular electromagnetic performance of the structure. The relative bandwidth (defined as the
frequency range in which the phase of the reflected field varies
) is equal to 11.4%. The structure also shows a
in the range
good angular stability, with very low phase values up to an incidence angle of about 88 , at the central frequency[Fig. (4b)]. To
better appreciate the angular performance, two additional curves
are reported in Fig. (4b). They are relevant to the responses of a
dielectric slabs
couple of simpler structures consisting of a
(designed at the central frequency of the band) backed by a PEC
and
. Wider intervals in
ground plane with
terms of the incidence angle with low phase response can be
obtained by inserting additional dielectric layers. Finally, it is
worth pointing out that introducing the new selection strategy,
the GA is able to reach convergence in 100150 generations
on the average against the previous standard GA implementation, for which 800 to 1000 generations were needed to reach
the same fitness function value [5]. In terms of computing time,
a single run of the optimization procedure takes approximately
6 hours on a 1.54 GHz Athlon XP 1800 processor.

199

REFERENCES
[1] D. Sievenpiper, L. Zhang, R. F. Jimenez Broas, N. G. Alexopolous, and
E. Yablonovitch, High impedance electromagnetic surfaces with a forbidden frequency band, IEEE Trans. Microwave Theory Tech., vol. 47,
pp. 20592074, Nov. 1999.
[2] F. R. Yang, K. P. Ma, Y. Qian, and T. Itoh, A novel TEM waveguide
using uniplanar photonic-bandgap (UC-PGB) structure, IEEE Trans.
Microwave Theory Tech., vol. 47, pp. 20922098, Nov. 1999.
[3] J. M. Johnson and Y. Rahmat-Samii, Genetic algorithms in engineering
electromagnetics, IEEE Trans. Antennas Propagat., vol. 39, pp. 721,
Aug. 1997.
[4] G. Manara, A. Monorchio, and R. Mittra, Frequency selective surface
design based on genetic algorithm, Electron. Lett., vol. 35, no. 17, pp.
14001401, Aug. 19, 1999.
[5] A. Monorchio, R. Mittra, and G. Manara, Design of dual band
frequency selective surfaces using genetic algorithms, in Proc. 16th
Annu. Rev. Progress in Applied Computational Electromagnetics
(ACES), Monterey, CA, Mar. 2025, 2000, pp. 5662.
[6] E. Michielssen, J.-M. Sajer, S. Ranjithan, and R. Mittra, Design of
lightweight, broadband microwave absorbers using genetic algorithms,
IEEE Trans. Microwave Theory Tech., vol. 41, pp. 10241031, July/Aug.
1997.
[7] E. A. Parker, A. D. Chuprin, J. C. Batchelor, and S. B. Savia, GA optimization of crossed dipole FSS array geometry, Electronics Lett., vol.
37, no. 16, pp. 996997, Aug. 2001.
[8] S. Chakravarty and R. Mittra, Application of the micro-genetic algorithm to the design of spatial filters with frequency-selective surfaces
embedded in dielectric media, IEEE Trans. Electromagn. Compat., vol.
44, pp. 338346, May 2002.
[9] C. Zuffada, T. Cwik, and C. Ditchman, Synthesis of novel all-dielectric
grating filters using genetic algorithms, IEEE Trans. Antennas Propagat., vol. 46, pp. 657663, May 1998.
[10] J. A. Vasconcelos, J. A. Ramrez, R. H. C. Takahashi, and R. R.
Saldanha, Improvements in genetic algorithms, IEEE Trans. Electromagn. Compat., vol. 37, pp. 34143417, Sept. 2001.

You might also like