Progress in Electromagnetics Research, Vol. 114, 51-66, 2011
Progress in Electromagnetics Research, Vol. 114, 51-66, 2011
Progress in Electromagnetics Research, Vol. 114, 51-66, 2011
1
(),
0
)
(
0
,
0
)
x
y
PEC
W
W
I
LIM
VACUUM
...
2a
region 0
region 1
x
u
y
u
Figure 1. The conguration of the analyzed structure.
2. SOLUTION DERIVATION
2.1. Problem Statement
The physical conguration of the examined device is shown in Fig. 1,
where the considered Cartesian coordinate system (x, y, z) is also
dened. An innite dipole of current I (in Amperes) is located
centrally into a slab of nite thickness 2W (region 1), lled with a
low-index metamaterial (LIM) of relative permittivity
1
(). This
dielectric volume is backed by a reecting metallic screen posed at
y = W, while the whole structure is located into vacuum (region 0).
This specic type of antenna is preferred because it has been found
to possess substantial directivity properties. It is also a state-of -the-
art structure which can be extensively used in the future and thus it is
worth to examine the eect of the additional equipment on this device.
There are several textbooks analyzing the conguration, the excitation
and the characteristics of the low-index metamaterial slab antenna. For
example in [31] the radiating properties of this structure are presented
and discussed. The developed eld is scattered by a cluster of U
identical perfectly conducting (PEC) cylinders of small radius a W,
placed at arbitrary positions (x, y) = (x
u
, y
u
) , u = 1, . . . , U, parallel
to z axis. Low index metamaterials are plasmonic substances working
close to their resonant frequency
P
. In particular, a simplied form
of
1
() is given by:
1
() =
2
2
P
2
, (1)
54 Valagiannopoulos
where is the operating circular frequency of a harmonically
dependent time e
+jt
. The resonant frequency of a plasmonic
substance can be estimated through various techniques. Numerous
studies have been performed in measuring the plasma frequency of
dielectric materials as chromium [32] or superconducting ones like in
Levitated Dipole Experiment [33]. Moreover it is rather common to
assume the parameter of the host substances as well-known [34]. All
the areas are magnetically inert, while, due to the symmetry, the
electric eld of each region is z-polarized. In the following analysis,
the symbols k
0
=
0
, k
1
= k
0
_
1
() are used for the operating
wavenumbers in free-space and LIM slab respectively. In addition, the
well-known radiation functions dened by: g
0/1
() =
_
2
k
2
0/1
, are
always evaluated with a nonnegative real part (and in case it zero, with
a positive imaginary part).
The scope of this work is to provide a semi-analytical treatment
to the dened boundary value problem which is in accordance with
results derived from commercial software simulations. Also, we aim at
estimating the benecial or disadvantageous eect of this random pin
lattice on the propagation features and the radiation characteristics of
the considered device. The novelty of the presented structure does not
lie in the sort of the radiator which is well-known, but in the scattering
cluster of pins which is located within the near eld. The positive or
the damaging inuence of a set of electrically small cylinders on the
operation of this antenna has not yet investigated neither analytically
nor numerically.
2.2. Absent Cluster
Let us reproduce the explicit formulas of some quantities referring to
the structure in the absence of this grid of thin cylindrical scatterers.
The Greens function of electric type, in our example, equals the axial
electric eld when the structure is excited by a lamentary electric
current with magnitude
j
0
(in Amperes). When the innite dipole
is located along the axis (x, y) = (X, Y ), the Greens function for any
obstacle to the vacuum region 0 (Y > W), is comprised by the singular
free-space term describing the eect of the dipole itself, added to the
smooth term expressing the inuence of the slab antenna. The singular
term is dened as follows:
G
n
(x, y, X, Y ) =
j
4
H
(2)
0
_
k
0
_
(x X)
2
+ (y Y )
2
_
=
1
4
_
+
e
g
0
()|yY |
g
0
()
e
j(xX)
d. (2)
Progress In Electromagnetics Research, Vol. 114, 2011 55
The notation H
(2)
0
corresponds to the 0th-order Hankel function of
second type. The smooth component is given by:
G
m
(x, y, X, Y ) =
_
+
C
G
()e
j(xX)g
0
()(y+Y )
d, (3)
where the integrand function C
G
() has been rigorously determined:
C
G
() =
e
2Wg
0
()
4g
0
()
g
0
() sinh(2Wg
1
()) g
1
() cosh(2Wg
1
())
g
0
() sinh(2Wg
1
()) +g
1
() cosh(2Wg
1
())
. (4)
The incident electric eld into region 0, where the cluster of pins is
positioned, owns the form:
E
0,inc
(x, y) =
_
+
C
E
()e
jxg
0
()y
d, (5)
where:
C
E
() =
e
Wg
0
()
2
j
0
I sinh(Wg
1
())
g
0
() sinh(2Wg
1
()) +g
1
() cosh(2Wg
1
())
. (6)
Note that for W 0 the electric eld vanishes, which makes sense
because the opposite image of the source neutralizes the eect of the
excitation dipole. Detailed derivations of the aforementioned formulas
are contained in [35].
2.3. Present Cluster
It is common knowledge that the basic analytic tool for treating
scattering problems is the so-called scattering integral [36]. This
pivotal formula, written for the scattered eld inside vacuum region
0, is particularized in our case to give:
E
0,scat
= j
0
U
u=1
_
S(u)
u
[G
n
+G
m
] dl, (7)
where S(u) is the metallic surface of the uth PEC pin and
u
the
unknown (supposedly constant) axial, z-polarized current (in A/m)
own upon it. Due to the small radius of the pin, we are going to
impose the boundary conditions for vanishing eld around the metallic
rods, only on U specic points: the centers of the circular bounds. That
yields to:
E
0,inc
(x
v
, y
v
) +E
0,scat
(x
v
, y
v
) = 0, (8)
for v = 1, . . . , U. Let us compute the scattering eld at these discrete
positions (x, y) = (x
v
, y
v
). Due to the electrically small cross section
56 Valagiannopoulos
of the metallic rods, the smooth integrands in (7), exhibit negligible
variation around them and thus, the corresponding line integrals can
be approximately evaluated as follows:
_
S(u)
u
G
m
(x
v
, y
v
, X, Y )dl
=
u
M
vu
=
u
2aG
m
(x
v
, y
v
, x
u
, y
u
). (9)
As far as the singular components are concerned, the integrals are
analytically evaluated via a standard procedure [37]:
_
S(u)
u
G
n
(x
v
, y
v
, X, Y )dl
=
u
N
vu
=
u
a
2j
_
H
(2)
0
(k
0
a) v = u
J
0
(k
0
a)H
(2)
0
_
k
0
_
(x
v
x
u
)
2
+ (y
v
y
u
)
2
_
v = u.
(10)
In this sense, a UU linear system with respect to the unknown vector
= [
u
], is formulated:
[M+N] =
1
j
0
e
inc
, (11)
where M = [M
vu
] and N = [N
vu
] for v, u = 1, . . . , U. Obviously, the
constant vector contains the samples of the incident eld at the centers
of the pins, namely: e
inc
= [E
0,inc
(x
v
, y
v
)]. Once the unknown currents
are determined, the scattering eld can be approximately computed
by (7), namely:
E
0,scat
(x, y) = 2aj
0
u=1
u
[G
n
(x, y, x
u
, y
u
) +G
m
(x, y, x
u
, y
u
)] .
(12)
If one introduces the equivalent cylindrical coordinate system
(, , z) and the corresponding notation (
u
,
u
), u = 1, . . . , U for the
pins axes, the asymptotic relations for the developed eld in the far
region { +, (0, )}, could be derived. Implementation of
stationary phase approximation [38] yields to:
e
0,inc
() k
0
C
E
(k
0
cos ) sin, (13)
e
0,scat
() 2aj
0
u=1
u
_
k
0
C
G
(k
0
cos )e
jk
0
u
cos(+
u
)
sin
j
4
e
jk
0
u
cos(
u
)
_
, (14)
for +. The -dependent factor
_
2
k
0
e
jk
0
+j
4
is omitted and
that is why the small letter e is used in dening the eld quantities
instead of the total ones which are denoted by E.
Progress In Electromagnetics Research, Vol. 114, 2011 57
3. NUMERICAL RESULTS
Prior to proceeding to the numerical simulation and commenting on
the produced graphs, the value range of the input parameters should
be claried. The plasma frequency of the low-index metamaterial
is kept xed throughout the examples:
P
= 20 Grad/sec and the
operational frequency is selected close to
P
. The thickness of the slab
2W is chosen on the order of a centimeter, while the radius a of the
pins possesses much lower values. As far as the output parameters are
concerned, two are the quantities of interest: (I) The direction along
which the maximum power is radiated by the antenna, denoted by
max
(
max
= 90
(as
does the antenna itself, in the absence of the scatterers). This means
that the scattering cluster does not aects signicantly
max
when it
is posed far from the vertical direction. On the other hand, there is a
rapid increase in
max
close to
= 60
m
a
x
(
d
e
g
)
/
P
=0.75
/
P
=0.80
/
P
=0.85
/
P
=0.90
(b)
0.8 0.9 1 1.1 1.2
0.5
1
1.5
2
2.5
3
operational frequency /
P
p
o
w
e
r
r
a
t
i
o
R
=5
o
=25
o
=45
o
=65
o
=85
o
(c)
Figure 3. (a) The examined case of multiple pins uniformly
distributed along a straight sloping line, (b) the maximum power
angle as function of the lines slope for various frequencies, (c) the
radiated power ratio as function of the operational frequency for several
slopes. Plot parameters:
P
= 20 Grad/sec, a = 5 mm, W = 30 mm,
I = 1 A, U = 12.
the two lines of the pins are normal each other. With this combination
of input quantities, the positioning of the metallic rods reinforces the
radiated power. In other words, it diminishes that portion of power
being reected back to the source due to the material discontinuity at
y = W. In this sense, the lattice of pins could play a matching role,
maximizing the transmitted power from the source to the outer space.
When the angular extent is large (small ), the performance of the
device gets poor; in particular, for <
P
the ratio R is negligible.
On the other hand, when >
P
, the measured R is found close to
60 Valagiannopoulos
x
y
...
U/2
U/2
...
(a)
0.8 0.9 1 1.1 1.2
0
1
2
3
4
5
6
7
operational frequency /
P
p
o
w
e
r
r
a
t
i
o
R
=15
o
=30
o
=45
o
=60
o
=75
o
(b)
5 10 15 20 25 30
0
1
2
3
4
5
number of cylinders U
p
o
w
e
r
r
a
t
i
o
R
/
P
=0.75
/
P
=0.80
/
P
=0.85
/
P
=0.90
(c)
Figure 4. (a) The examined case of two symmetric grids of cylinders
forming a radiation funnel, (b) the radiated power ratio as function of
the operational frequency for several angular apertures of the funnel
(a = 2 mm, U = 24), (c) the radiated power ratio as function of the
number of pins for various frequencies (a = 5 mm, = 60
). Plot
parameters:
P
= 20 Grad/sec, W = 30 mm, I = 1 A.
one, regardless of the chosen angle .
In Fig. 4(c), the quantity R is represented as function of the
number of pins for several oscillation frequencies with xed angle
= 60
/2
...
U-1
/2
(a)
1 2 3 4 5
90
95
100
105
110
115
horizontal position /W
m
a
x
i
m
u
m
a
n
g
l
e
m
a
x
/
P
=0.75
/
P
=0.80
/
P
=0.85
/
P
=0.90
(b)
2 4 6 8 10 12 14
80
85
90
95
100
105
110
115
horizontal position /W
m
a
x
i
m
u
m
a
n
g
l
e
m
a
x
/
P
=0.75
/
P
=0.80
/
P
=0.85
/
P
=0.90
(c)
Figure 5. (a) The examined case of a vertical distribution of cylinders
with one regulating pin, (b) the maximum power angle as function of
the horizontal distance between the pin and the cluster for various
frequencies ( = 5 W), (c) the maximum power angle as function of
the vertical position of the pin for various frequencies ( = ). Plot
parameters:
P
= 20 Grad/sec, a = 5 mm, W = 30 mm, I = 1 A,
= 60
, U = 15.
62 Valagiannopoulos
(x, y) = (
2
, ). The positions of the cylinders are given by:
x
u
=
2
, y
u
= W + 3au, (17)
for u = 1, . . . , (U 1). In Fig. 5(b), the direction of maximum
power is depicted as function of the horizontal position for various
frequencies, with xed . When the horizontal distance between
the regulating cylinder and the vertical cluster gets increased, the
maximum radiation angle is decreasing, tending to 90
. This is an
anticipated result, as the farther from the vertical axis the scatterers
are located, the less powerful is the eld with which the metallic
volumes are interacting. The behavior of the four curves are very
similar but the deviation of
max
from 90