You are on page 1of 7

162 QUADERNI DELLA SOCIET ITALIANA DI ELETTROMAGNETISMO, VOL. 1, N.

2 LUGLIO 2005

A network theory for fss-based ebg surfaces
M. Nannetti, M. Caiazzo, A. Cucini, S. Maci
1
Abstract This paper presents a novel method for the efficient derivation of an
equivalent network representation of electromagnetic bandgap (EBG)
structures composed by lossless frequency selective surfaces (FSS) printed on
stratified dielectric media. The formulation presented yields a two-port
dominant mode equivalent matrix representation of the FSS, and is applicable
in the range of frequencies where a single propagating Floquet mode occurs.
Otherwise, it can be generalized to the case of an arbitrary number of
accessible modes. The elements of the diagonalized FSS matrix respect the
conditions of driving point LC impedance functions, thus, they can be
approximated in terms of poles and zeros, using a simple analytical expression.
The final result is a compact form of the dispersion equation, whose solutions
identify the modes of the structure.

I. INTRODUCTION
In recent years, a great interest has been devoted to the study of the dispersion
properties of planar structures realized by frequency selective surfaces (FSS) on
grounded dielectric slab. This kind of structures are used to realize artificial
magnetic conductors (AMC) [1], electromagnetic band-gap (EBG) surfaces, or
surfaces which exhibit soft and/or hard equivalent boundary conditions [2].
As it is well known, an approximate model of an FSS may be given in terms of a
quasi-static LC impedance network placed in a TE or TM transmission line. In
absence of losses, the FSS impedance is purely reactive. Patch-type FSS are
described by a series LC network. At frequencies below the resonant frequency, the
FSS is capacitive. This capacitance may resonate in parallel with the inductance
provided by the section of short-circuited transmission line, thus, providing an AMC
surface. In a certain frequency band, the same structure can provide an inhibition to
the surface wave (SW) propagation, acting as an EBG surface. Similar
considerations apply to aperture-type FSS which can be described by a parallel LC
circuit. This simple circuital representation contains the essential physics to
qualitatively justify the basic aspects of the AMC properties. However, the lack of
description of important aspects such as the wavenumber dependence of the

1 Dipartimento di Ingegneria dellInformazione, Universit di Siena, Via Roma 56, 53100 Siena,
e-mail: nannetti@unisi.it, caiazzo@ieee.org, cucini@dii.unisi.it, macis@dii.unisi.it
ID 0001-02-2005 2005 SIEM

M. NANNETTI ET AL.: A NETWORK THEORY FOR FSS-BASED EBG SURFACES 163
equivalent circuit or the coupling between TE and TM polarizations, imposes a more
rigorous generalization of the model.
Recently, a method has been introduced for the efficient synthesis of the FSS
admittance matrix for the study of dispersion properties [3]. This method is based on
the application of the Fosters reactance theorem, which implies that FSS admittance
functions of frequency satisfy the pole-zero analytical properties of the driving point
LC admittance functions. The identification of the poles and zeros of the FSS
equivalent admittance allows to reconstruct the surface response over a large
frequency band. In this paper, the method is generalized in order to deal with a wide
class of FSS-based artificial surfaces. A new derivation of the FSS admittance
matrix is proposed and some general properties of this kind of structures are pointed
out.
II. METHOD OF MOMENT ANALYSIS AND FSS ADMITTANCE
MATRIX
The formulation will be derived for the case of a patch-type FSS-based surface,
shown in Fig. 1(a), being implied that the approach can be applied similarly to
aperture-type FSS. Due to the periodicity of the problem, the analysis is reduced to
that of a single periodic cell, with phase-shift boundary conditions. By applying the
equivalence theorem, an electric current distribution is assumed on the surface of the
metallic patches, radiating with the Greens function (GF) of the grounded slab. By
imposing the boundary conditions on the surface of the metallic patches, the electric
field integral equation (EFIE) is derived

(1) 0 ) ( = +
imp s
E J E

where E
imp
is the tangential impressed electric field, and E
s
(J) is the tangential
electric field, scattered by the unknown electric currents. By using a Galerkin
spectral MoM approach, the EFIE is converted into a matrix equation

V I Z MOM = (2)







164 QUADERNI DELLA SOCIET ITALIANA DI ELETTROMAGNETISMO, VOL. 1, N. 2 LUGLIO 2005

Figure 1 FSS-based surface (a) and equivalent 2-port network
representation for the dominant mode (b).

Using the notation introduced by Orta et al. [4], the MoM matrix can be written
in compact form as

Q Z Q Z GF
H
MOM = (3)

where )] ( ), ( [ diag
p
TE
GF p
TM
GF
GF Z Z Z k k = is a diagonal matrix of order 2P
(with P denoting the number of Floquet waves (FW)), whose elements are the
spectral GF impedances
1
1
/
1
/
0
/
)] cot( ) ( ) ( [ ) (

= h k jY Y Z
z
E TM E TM E TM
GF
k k k
,
sampled at the FW spectral points k
p
. In the previous expression,
zm m
TM
m
k Y / ec =
and
0
/ e
zm
TE
m
k Y = are the modal transmission line TM and TE characteristic
admittances of the free space (m=0) or dielectric (m=1) region, with k
z0
= (k
2
k
x
2

k
y
2
)
1/2
and k
z1
= (c
r1
k
2
k
x
2
k
y
2
)
1/2
. In (3),
N n P p
TE
n p
TM
n p
Q Q Q
, 1 ; , 1 , ,
] , [
= =
= is a
2PN matrix (with N being the number of basis/test functions), with elements
p p n
TM
n p
Q o ) (
,
= k F and
p p n
TE
n p
Q o ) (
,
= k F , where denotes the Fourier
transform of the n-th basis function, and
) (k F
n
| | /
p p p
k k = o ,
p p
z o o = are the
spectral basis associated to the TM and TE component, respectively. Finally, the
superscript
H
denotes the transpose complex-conjugate matrix.
The derivation of the FSS admittance matrix will be drawn in the hypothesis that
at the most only the dominant FW is propagating in free space. Otherwise, the
formulation can be easily extended to an arbitrary number of accessible modes [5],
as shown in [6]. The generalization to accessible modes is of interest when dealing
with higher-order propagating modes or with near-field interaction with a proximity-
located antenna. With the single-mode hypothesis in mind, the MoM known vector
M. NANNETTI ET AL.: A NETWORK THEORY FOR FSS-BASED EBG SURFACES 165
V can be related to the impressed dominant FW voltage vector
imp
FW V be the
relationship
H
imp
FW V q V = , where
H
q is a N2 matrix obtained by the first two
columns of
H
Q . Analogously, FW I qI = . Thus, a relationship is obtained between
the dominant FW impressed voltage and total current, as


1
( )
H
imp imp
FW MOM FW FW FW I q Z q V Y V

= = (4)

By noting that the total dominant FW voltage is given by the summation of the
voltage associated to the impressed and scattered field,
imp s imp
GF FW FW FW FW FW V V V V Z I = + = , the FSS admittance matrix is obtained,
through Eq. (4), as

GF FW GF FW FSS Y Y Y Y Y
1
] [

= (5)

The previous expression allows to give a description of the FSS-based surface in
terms of an equivalent transmission line network, as in Fig. 1(b), where the FSS is
represented by a 2-port network, where each port is placed in parallel to the TM or
TE transmission line associated to the grounded dielectric slab.
III. ANALITYICAL APPROXIMATION OF THE FSS MATRIX
The 22 FSS admittance matrix is usually a full matrix. Under the hypothesis of
lossless FSS, the matrix elements are purely imaginary, thus implying that FSS Y can
be diagonalized by a rotation matrix

) ( ] , diag[ ) (
) 2 ( ) 1 (
o o = R Y Y R Y
FSS FSS
FSS (6)

where


(


=
o o
o o
o
cos sin
sin cos
) ( R (7)


166 QUADERNI DELLA SOCIET ITALIANA DI ELETTROMAGNETISMO, VOL. 1, N. 2 LUGLIO 2005
The rotation angle o and the eigenvalues , are dependent on the
frequency e and as well on the wavenumbers k
) 1 (
FSS
Y
) 2 (
FSS
Y
x
, k
y
. It must be noted that when the
direction of wave propagation is along any plane of symmetry of the FSS, TM and
TE modes are uncoupled and consequently o=0.
In absence of losses, it can be demonstrated that the eigenvalues ,
respect the Fosters reactance theorem. The main implication is that the functions
) 1 (
FSS
Y
) 2 (
FSS
Y
) , , (
) (
y x
i
FSS
k k Y e possess the same analytical properties of a passive driving-point
LC-function of frequency:
i) Poles and zeros lie on the real e axis, and are simple and alternate.
ii) A zero (patch-type FSS) or a pole (aperture-type FSS) is at e = 0.
iii) Poles and zeros are symmetrically displaced with respect to the origin.
On the basis of the previous points, the eigenvalues are approximated as
| | ( ) | | ( )
| | ( ) | | ( )
2
) (
2
2
) (
1
2
) (
2
) ( ) (
0
) (
) , ( / 1 ) , ( / 1
) , ( / 1 ) , ( / 1 ) , (
) , , (
2 1
y x
i
p y x
i
p
y x
i
z y x
i
z y x
i
y x
i
FSS
k k k k
k k k k k k C j
k k Y
e e e e
e e e e e
e


=
(8)

where . The final output is an analytical form of the
FSS admittance for any frequency and wavenumber. The parameters

) (
2
) (
2
) (
1
) (
1
i
z
i
p
i
z
i
p
e e e e < < <
) (
,
i
j z
e ,
) (
,
i
j p
e ,
and have a weak dependence on the wavenumbers (k
) (
0
i
C
x
, k
y
), that can be easily
interpolated from data relevant to few values. Analogous considerations hold for the
rotation angle o. This result allows to deal with a simple form of the resonance
equation

0 ] det[ = + FSS GF Y Y (9)

as derived from the equivalent network in Fig 1b. By straightforward algebraic
manipulations the dispersion equation can be explicitly rewritten as

(10) 0 ) )( ( sin ) )( (
) 2 ( ) 1 ( 2 ) 2 ( ) 1 (
= + + +
TE
GF
TM
GF FSS FSS
TE
GF FSS
TM
GF FSS
Y Y Y Y Y Y Y Y o

The solutions of equation (10) identify, within the slow-wave region, the
wavenumbers of the SWs supported by the artificial surface.
M. NANNETTI ET AL.: A NETWORK THEORY FOR FSS-BASED EBG SURFACES 167
The numerical example presented here refers to a cross-dipole FSS-based surface
(dimensions in millimeters are given in the inset of Fig. 2). Figure 2 shows the
dispersion curves of the poles and zeros of the FSS admittance matrix eigenvalues,
along the contour of the irreducible Brillouin zone. It can be seen that the
dependence on the wavenumber is very weak. Figure 3 shows the Brillouin
dispersion diagram. We stress that the paths IX and IM (corresponding to
symmetry planes of the structure) are associated to pure TE or TM SW modes, while
along the path XM hybrid SW modes are present with possible TM or TE
predominance (quasi-TM or quasi-TE modes). Moreover, a mode turns from TM to
TE along the XM path. An EBG is obtained in the frequency range (11 11.7) GHz.




Figure 2 Dispersion curves of the poles (p) and zeros (z) of the FSS
admittance matrix eigenvalues, along the contour of the irreducible
Brillouin zone




168 QUADERNI DELLA SOCIET ITALIANA DI ELETTROMAGNETISMO, VOL. 1, N. 2 LUGLIO 2005


Figure 3 Dispersion diagram of the cross-dipole FSS-based
artificial surface.

REFERENCES
[1] F.-R. Yang et al., A novel TEM waveguide using uniplanar compact
photonic-bandgap (UC-PBG) structure, IEEE Trans. Microwave Theory
Tech., Vol. 47, no. 11, pp. 2092-2098, Nov.1999.
[2] P.-S. Kildal, Artificially soft and hard surfaces in electromagnetics, IEEE
Trans. Antennas Propagat., Vol. 38, no. 10, pp. 1537-1544, Oct. 1990.
[3] S. Maci, M. Caiazzo, A. Cucini, and M. Casaletti, A pole-zero matching
method for EBG surfaces composed of a dipole FSS printed on a grounded
dielectric slab, IEEE Trans. Antennas Propagat., Vol. 53, no. 1, pp. 70-81,
Jan. 2005.
[4] J. C. Vardaxoglou, Frequency selective surfaces, Research Studies Press
Ltd., Taunton, England, 1997, pp. 221-238.
[5] T. E. Rozzi, Network analysis of strongly coupled transverse apertures in
waveguide, Int. J. Circuit Theory Appl., Vol. 1, pp. 161-178, 1973.
[6] S. Maci and A. Cucini, FSS-based complex surfaces in Electromagnetic
Metamaterials: Physics and Engineering Aspects, editors: N. Engheta and
R. W. Ziolkowski, Wiley InterScience, in press.