0% found this document useful (0 votes)
27 views

An Efficient FDTD Algorithm For The Anal

This document discusses an efficient FDTD algorithm for analyzing microstrip patch antennas printed on anisotropic dielectric substrates. It proposes a 3D FDTD approach using D, E, and H fields to handle arbitrary anisotropic media. Numerical results demonstrate how the resonant frequencies of such an antenna are influenced by the anisotropy of the substrate.

Uploaded by

othmanem449
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
27 views

An Efficient FDTD Algorithm For The Anal

This document discusses an efficient FDTD algorithm for analyzing microstrip patch antennas printed on anisotropic dielectric substrates. It proposes a 3D FDTD approach using D, E, and H fields to handle arbitrary anisotropic media. Numerical results demonstrate how the resonant frequencies of such an antenna are influenced by the anisotropy of the substrate.

Uploaded by

othmanem449
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

1142 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 47, NO.

7, JULY 1999

An Efficient FDTD Algorithm for the Analysis of


Microstrip Patch Antennas Printed on a
General Anisotropic Dielectric Substrate

An Ping Zhao, Jaakko Juntunen, and Antti V. Räisänen

Abstract—In this paper, an efficient three-dimensional finite-difference


time-domain (FDTD) approach based on the D-, E-, and H-fields is
proposed to handle arbitrary anisotropic dielectric media; and, partic-
ularly, the way of imposing the electric-wall boundary condition on the
surface of perfect electric conductors is discussed in details. By combining
the proposed FDTD approach with the material-independent perfectly
matched-layer absorbers, the performance of a line-fed microstrip patch Fig. 1. The 3-D FDTD unit cell used for an arbitrary anisotropic dielectric
antenna deposited on general anisotropic dielectric substrate is investi- medium.
gated. The scattering parameters of the antenna as a function of the
rotation angle of the optical axis of the anisotropic substrate are presented 2) It is less accurate because the updating procedure for E was
for the first time. Numerical results demonstrate how the resonant done in terms of the spatial derivatives within two space steps
frequencies of the antenna are influenced by the anisotropy.
of H [10], [11].
Index Terms— FDTD, general anisotropic dielectric media, MIPML 3) It is hard to apply discontinuous boundary conditions between
absorber, patch antennas. two different dielectric media, as normal (with respect to the
interface of the media) electric-field components are defined on
the edges of the variants of Yee’s grid [11].
I. INTRODUCTION
4) It is almost impossible to impose perfect electric conductor
During recent years, great interests have been shown in using (PEC) conditions due to the same reason explained in 3).
microstrip lines deposited on anisotropic substrate since the substrate
In this paper, to overcome the above difficulties, an efficient
anisotropy could have important implications on the operation of
three-dimensional (3-D) FDTD algorithm based on the D-, E-, and
microstrip circuits [1], [2]. With the increasing complexity of ge-
H-fields is proposed to handle arbitrary anisotropic dielectric media.
ometry and material property, designing these circuits requires more
In particular, in the proposed approach, neighboring quantities within
and more dedicated and sophisticated computer-aided-design (CAD)
only one space step are used in the averaging approximation [see
tools to predict the characteristics or performance of the circuits,
(3)] in space (also spatial derivatives within two space steps are
as most of these structures do not lead themselves to analytical
not involved in the updating equations) and no normal electric
solutions. The finite-difference time-domain (FDTD) method, which
field components are defined at the edges of the FDTD unit cell
was introduced by Yee in 1966 [3], has been proven to be one of
(see Fig. 1). In addition, unlike the algorithms based on E and
the most powerful CAD tools for solving this class of problems
H, Berenger’s perfectly matched layer (PML) absorbing boundary
due to its high flexibility. However, in the past, the application
condition (ABC) [12] can be very simply and easily adopted (in
of the FDTD method has been mainly focused on the circuits
terms of material-independent PML (MIPML) absorbers [13]-[15])
deposited on the materials with isotropic [4]–[8] or diagonal tensor
in the present approach. Consequently, the approach proposed in this
permittivity [9]. It is obvious that, under certain circumstances,
paper is more flexible and accurate than the previous approaches.
many circuit materials exhibit more or less anisotropic behavior that
This paper is organized as follows. In Sections II and III, the
cannot be simply ignored in the circuit simulation and, thus, an
proposed 3-D FDTD algorithm based on the D-, E-, and H-fields
accurate theoretical characterization could be the key issue for critical
is described. The numerical implementation for the treatment of
designs and successful applications of integrated circuits based on
PEC’s is discussed in Section IV. To demonstrate the effectiveness
anisotropic materials. Therefore, extension of the FDTD method to
and usefulness of the proposed algorithm, the influence of the
cover arbitrary anisotropic materials is highly preferred.
anisotropy on the scattering parameters of line-fed microstrip patch
In the past, although FDTD approaches based on the E- and H-
antennas deposited on anisotropic dielectric substrates is investigated
fields were developed (in connection with Yee’s grid [10] and the
in Section V and, finally, a conclusion is presented in Section VI.
variants of Yee’s grid [11]) for arbitrary anisotropic dielectric media,
these approaches have the following drawbacks in real applications.
II. MAXWELL’S EQUATIONS IN 3-D ARBITRARY ANISOTROPIC MEDIA
1) It is difficult to calculate the field components located at a half-
space step [e.g., Ex (0:5; j; k)] inside the mesh truncation since Suppose that ["] is the relative permittivity tensor of an arbi-
neighboring quantities within two space steps are involved in trary anisotropic dielectric (loss-free) medium, thus, in this medium,
the updating equations (see [10, eq. (21)] and [11, eq. (10)]). Maxwell’s equations can be written as
@Dx @Hz @Hy
@t
=
@y
0 @z (1a)
Manuscript received January 21, 1999.
@Dy @Hx @Hz
A. P. Zhao is with the Electronics Laboratory, Nokia Research
@t
=
@z
0 @x (1b)
Center, Itamerenkatu 11-13, FIN-00180, Helsinki, Finland (e-mail: an-
@Dz @Hy @Hx
[email protected]).
J. Juntunen and A. V. Räisänen are with the Radio Laboratory, Department @t
=
@x
0 @y (1c)
of Electrical and Communications Engineering, Helsinki University of
@Hx @Ey @Ez
Technology, FIN-02150 Espoo, Finland. 0
@t
=
@z
0 @y (1d)
Publisher Item Identifier S 0018-9480(99)05300-4.

0018–9480/99$10.00  1999 IEEE


IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 47, NO. 7, JULY 1999 1143

@Hy @Ez
0
@t
= @x 0 @E
@z
x
(1e) Therefore, by combining the updating procedures for H and D with
the above updating scheme for E, the FDTD approach based on
@H @E
0 z = x
@t @y
0 @E
@x
y
(1f) the D-, E-, and H-fields can be used to study arbitrary anisotropic
dielectric materials. In this approach, however, to correctly impose
where 0 is the permeability in vacuum. To easily handle arbitrary the electric-wall boundary condition on the surface of PEC’s, special
anisotropic dielectric media, the magnetic field, H, electric field, attention should be paid to the tangential electric displacement t D
E, and electric displacement, D, are involved together in (1). The (e.g., Dx and Dy ).
relationship between E and D is of the following form:

Ex =
("yy "zz 0 "yz "zy ) D + ("xz "zy 0 "xy "zz ) D IV. TREATMENT OF PEC’S
x y
3 3 On the surface of PEC’s, the tangential electric field t must be E
("xy "yz 0 "xz "yy )
+ 3 Dz (2a) zero, and the implementation of such a condition is straightforward
in the FDTD updating formulation [3]. However, from (1) one can
(" " 0 " " ) (" " 0 " " )
Ey = yz zx yx zz Dx + xx zz xz zx Dy see that, although in the proposed algorithm the PEC condition for
3 3 H
H (i.e., the normal magnetic field n = 0) can be automatically
("xz "yx 0 "xx "yz ) E
satisfied while t = 0 is applied, the PEC condition for E (i.e.,
+ Dz
(" " 0 " " )
3
(" " 0 " " )
(2b)
E t = 0) cannot be kept because E is actually updated from D.
Ez = yx zy yy zx Dx + xy zx xx zy Dy Therefore, in order to make sure that the PEC condition is correctly

(
3
"xx "yy 0 "xy "yx )
3 E
imposed for t , the values of the tangential electric displacement t D
+ Dz on the surface of the PEC have to be calculated or justified. In what
3 (2c)
follows, it will be shown that different calculation procedures for t D
where 3 = "0 ("xx "yy "zz + "xy "yz "zx + "xz "yx "zy 0 "xz "yy "zx 0 should be adopted while the PEC is located at different positions.
"xy "yx "zz 0 "xx "yz "zy ) and "0 is the permittivity in vacuum. If the PEC is located inside the mesh domain (e.g., the metal
microstrip line), e.g., at k = k0 , then Dx and Dy on this PEC can
be calculated, respectively, from (3a) and (3b) by imposing Exn = 0
III. EXTENSION TO ARBITRARY ANISOTROPIC DIELECTRIC CASES
Fig. 1 shows a 3-D unit cell used for the proposed FDTD approach, or Eyn = 0 as follows:
where the locations of the components of E and H are the same as
those in [3] and the positions of the components of D are located at Dxn (i +0:5; j; k0 )
the same positions (in the x-, y -, and z -directions, respectively) of ("xz "zy 0 "xy "zz ) Dn(i; j 0 0:5; k )+ Dn(i; j +0:5; k )
= 0 4(
the components of E; and on the edge of the 3-D unit cell, only the "yy "zz 0 "yz "zy ) y 0 y 0

tangential (with respect to the surface of the edge) electric field and + Dy (i +1; j 0 0:5; k0 )
n
electric displacement field components are involved. The FDTD unit
cell shown in Fig. 1 is applicable to arbitrary anisotropic dielectric
+ Dyn(i +1; j +0:5; k0 )
materials; and the updating procedures used for D and H are similar
to those used for E and H in the isotropic cases [3].
0 4((""xyyy""yzzz00""xzyz""yyzy)) Dzn(i; j; k0 0 0:5) + Dzn (i; j; k0 +0:5)
Since in the FDTD mesh the components of E and D are position + Dzn(i +1; j; k0 0 0:5)
+ Dzn(i +1; j; k0 +0:5)
dependent, (2) must be rearranged while the proposed FDTD method
is applied because this equation is valid only when the concerned field (4a)
components are defined at the same position. By using an averaging Dyn (i; j +0:5; k0 )
= 04((""yz ""zx 00""yx ""zz )) Dxn(i 0 0:5; j; k0 )
approximation procedure in space on D, (2a) and (2b), for example,
can be written, respectively, in the FDTD form as xx zz xz zx
Exn (i + 0:5; j; k) + Dxn(i 0 0:5; j +1; k0 )
= ("yy "zz 0 "yz "zy ) Dxn(i +0:5; j; k)+ ("xz "zy 0 "xy "zz )
3 43
+ Dxn(i +0:5; j; k0 )
+ Dxn(i +0:5; j +1; k0 )
1 Dyn (i; j 0 0:5; k)+ Dyn(i; j +0:5; k) + Dyn (i +1; j 0 0:5; k)
+ Dyn(i +1; j +0:5; k) 0 4((""xzxx""yxzz00""xxxz""yzzx)) Dzn (i; j; k0 0 0:5) + Dzn (i; j; k0 +0:5)
0 "xz "yy ) Dn(i; j; k 0 0:5)+Dn(i; j; k +0:5) + Dzn(i; j +1; k0 0 0:5)
+ ("xy "yz43 z z
+ Dzn(i; j +1; k0 +0:5) : (4b)
+ Dzn(i +1; j; k 0 0:5)
+ Dzn(i +1; j; k +0:5) (3a)
D
Equations (4a) and (4b) calculate the values of t on the surface of
the PEC located inside the mesh domain. Imposing this implicit PEC
n
Ey (i; j +0:5; k) D E
condition to t makes sure that the condition t = 0 is correctly
0 "yx "zz ) Dn(i 0 0:5; j; k)+ Dn(i 0 0:5; j +1; k)
= ("yz "zx 43
applied; and failing to do this may cause instability.
x x However, if the PEC is located exactly on the mesh truncation (i.e.,
n
+ Dx (i +0:5; j; k) the ground metal plane), e.g., at k = 0, then (4a) and (4b) are not
valid since in these equations some points [e.g., Dyn (i; j; k0 0 0:5)
+ Dxn(i +0:5; j +1; k) in (4a)] are outside the mesh domain and, thus, not available. To be

+ ("xx "zz 0 "xz "zx ) n (" " 0 " " )


Dy (i; j +0:5; k) + xz yx xx yz
D
able to calculate the t field on this PEC (i.e., at k = 0), a linear
3 43 approximation procedure [16], which is similar to the one successfully
1 Dzn(i; j; k 0 0:5)+ Dzn(i; j; k +0:5) + Dzn(i; j +1; k 0 0:5) used for the analysis of magnetized ferrites [17], is employed to
calculate the Dx , Dy , and Dz fields first on the PEC as follows:
+ Dzn(i; j +1; k +0:5) (3b) Dxn (i + 0:5; j; 0) = 2Dxn(i + 0:5; j; 1) 0 Dxn (i + 0:5; j; 2) (5a)
1144 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 47, NO. 7, JULY 1999

( + 0:5; 0) = 2Dyn(i; j + 0:5; 1) 0 Dyn(i; j + 0:5; 2)


Dyn i; j (5b)
Dzn (i; j; 0) =
1 3Dzn(i; j; 0:5) 0 Dzn(i; j; 1:5) :
2 (5c)

By combining (3a) with (5b) and (5c), and using the averaging
approximation in space on D, as well as imposing Exn =0
, we have
n( +0 5 0)
Dx i : ; j;

= 4(( ) 2 ( 0 5 1) ( 0 5 2)
0 ""xzyy""zyzz00""xyyz""zzzy Dyn i; j 0 : ; 0 Dyn i; j 0 : ;
) Fig. 2. Line-fed microstrip patch antenna printed on an anisotropic substrate
+ 2 ( +0 5 1) ( +0 5 2)
Dyn i; j : ; 0 Dyn i; j : ; (where the relative permittivities of the uniaxial anisotropic dielectric substrate
are "1 = 2:31 and "2 = 2:19).
+2 ( +1 0 5 1)
Dyn i ; j 0 : ;
( +1 0 5 2)
0 Dyn i ; j 0 : ; "yy= "1 (7b)
+2 ( +1 +0 5 1)
Dyn i ; j : ;
"zz= "1 sin2  + "2 cos2  (7c)
n ( +1 +0 5 2)
0 Dy i ; j : ; "xz = "zx = ("1 0 "2 )sin  cos  (7d)

0 4((""xyyy""yzzz00""xzyz""yyzy)) 3Dzn(i; j; 0:5) 0 Dzn(i; j; 1:5) "xy = "yx = "yz = "zy = 0 (7e)

where  is the angle between the optical axis and the x-direction;
+3Dzn(i +1; j; 0:5) 0 Dzn(i +1; j; 1:5) : whereas if the optical axis is placed in the yz -plane, then the relative
(6a) permittivity tensor is given by

Similarly, by combining (3b) with (5a) and (5c), as well as imposing "xx = "1 (8a)
Eyn =0 , we have "yy = "1 cos2  + "2 sin2  (8b)
Dyn (i; j +0:5; 0) "zz = "1 sin2  + "2 cos2  (8c)
("yz "zx 0 "yx "zz )
= 0 4("xx"zz 0 "xz "zx ) 2Dxn(i 0 0:5; j; 1) 0 Dxn(i 0 0:5; j; 2) " yz = "zy = (" 1 0 "2 )sin  cos  (8d)
"xy = "yx = "xz = "zx = 0 (8e)
+2Dxn(i +0:5; j; 1) 0 Dxn(i +0:5; j; 2)
+2Dxn(i 0 0:5; j +1; 1) where  is the angle between the optical axis and y -direction.

0 Dxn(i 0 0:5; j +1; 2)


For both the anisotropic cases, the space steps used in the compu-
tation are 1x = 0:3891 mm, 1y = 0:4 mm, and 1z = 0:1985 mm;
+2Dxn(i +0:5; j +1; 1) and the total mesh dimensions are 58 2 80 2 22 in the x-, y -, and z -
directions, respectively. The rectangular patch is thus 321x 2 401y ,
0 Dx (i +0:5; j +1; 2)
n
the microstrip feeding port is 61x 2 271y , and the substrate is
("xz "yx 0 "xx "yz )
0 4("xx"zz 0 "xz "zx ) 3Dz (i; j; 0:5) 0 Dz (i; j; 1:5)
n n modeled as 41z . The time step is 1t = 0:4 ps. The half-width of
a Gaussian pulse is T = 15 ps, and the time delay t0 is set to be
3 :1T . On the open boundaries, the MIPML absorber [15] is used. The
+3Dz (i; j +1; 0:5) 0 Dz (i; j +1; 1:5) : interfaces
n n
of the MIPML media are placed five cells from the edge of
(6b) the patch antenna and ten cells above the surface of the antenna. The
D
Equations (6a) and (6b) calculate the values of t on the surface of performance of the MIPML with different parameters were tested,
and it was found that the reflection caused by the MIPML was less
the PEC located on the mesh truncation. Again, imposing this implicit
D
PEC condition to t ensures that the condition t E =0
is applied. than 65 dB when the following parameters were used:
However, it is recommended that t E =0
should also be applied on 1) number of layers N =8in all directions;
04 ;
D
the PEC even it has been implicitly imposed through t . Finally, it 2) normal theoretical reflection coefficient R (0) = 10
needs to be mentioned how the above implicit PEC condition can be 3) power of the conductivity distribution m .=4
applied when the PEC is located inside MIPML regions. If the unsplit- In addition, the advanced source excitation technique [18] (with
field formulations [15] are adopted for the MIPML absorber, then (4) uniform distribution of Dz under the microstrip line) is adopted and
and (6) can be used directly. However, if the split-field formulations the input plane is located 21
y inside the MIPML. The simulation is
[13], [14] are used, then the forms of (4) and (6) should be changed. performed for 8000 time steps and the reference plane for the port is
For example, either Dxy or Dxz should be updated, instead of Dx , set at 101 y away from the edge of the patch.
on the surface of the PEC since in this case Dx is split to Dxy and D
To demonstrate that t is not zero on the surface of PEC’s, the
Dxz [i.e., Dx =Dxy Dxz ].+ contour plots of Dx and Dy recorded (at 800 time steps) on the plane
where the microstrip and patch antenna lie for the case  = (i.e., = 4
V. NUMERICAL RESULTS the optical axis of the substrate is located in the xz -plane) are shown
in Fig. 3(a) and (b), respectively. It can be seen from Fig. 3 that Dy
Using the technique developed in this paper, a line-fed microstrip
is zero on the surface of the PEC (in our case, the microstrip line and
patch antenna printed on a general uniaxial anisotropic dielectric
patch are the PEC), but Dx is not. This is due to the fact that, in this
substrate (as shown in Fig. 2) is studied. The following two different
special case, "xy is zero and "xz is not. Although the (maximum)
anisotropic cases for the anisotropic substrate will especially be
considered: the optical axis lies either in the xz - or yz -planes. In
D
value of t on the surface of the PEC is much smaller (about 2.0%)
than its value near the edge of the PEC, ignoring it in the calculation
particular, when the optical axis is located in the xz -plane, the relative
will cause instability and fault numerical results. Fig. 4 shows the
permittivity tensor is given by
waveforms of the normalized incident and reflected vertical electric
"xx = "1 cos2  + "2 sin2  (7a) fields (recorded at the reference plane and just underneath the strip)
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 47, NO. 7, JULY 1999 1145

Fig. 5. S -parameter of the microstrip patch antenna as a function of  , where


(a) the optical axis lies in the xz -plane and the values of the angle  are zero,
=4, and =2, respectively.

(b) Fig. 6. S -parameter of the microstrip patch antenna as a function of , where


the optical axis lies in the yz -plane and the values of the angle  are zero,
D
Fig. 3. Contour plots of the field components of t recorded (at 800 time =6, =3, and =2, respectively.
steps for the case  = =4) on the plane where microstrip line and patch
lie. (a) Dx . (b) Dy .
for the case  = =4. As shown in Fig. 4, the stable state of the
structure is already reached at 8000 iterations. Once the waveforms
of the incident and reflected vertical electric fields on the reference
plane are obtained, the incident and reflected voltages between the
metal strip and ground plane can be calculated and, thus, the S -
parameter can be calculated [4], [5]. The scattering coefficient results
as a function of  are shown in Fig. 5. From Fig. 5, one can observe
that the operating resonant frequencies are decreased while the angle
 increases from zero to =2; such a reduction is more significant for
the higher operating resonance. In particular, while the values of  are
chosen to be zero, =4, and =2, the operating resonant frequencies
at the lowest operating resonance (i.e., near 7.5 GHz) are 7.47, 7.37,
and 7.28 GHz, respectively, whereas at the highest resonance (i.e.,
around 18 GHz), they are 18.03, 17.81, and 17.58 GHz, respectively.
Finally, we calculate the S -parameter of the antenna for the case
where the optical axis lies in the yz -plane. In this case, Dx on the
surface of the PEC is not zero, but Dy is zero since the optical
axis lies in the yz -plane. Fig. 6 demonstrates the S -parameter as a
function of . It can be seen from Fig. 6 that, for the case where the
optical axis lies in the yz -plane, the influence of the anisotropy on
Fig. 4. Waveforms of the normalized incident and reflected vertical electric
fields (Ez ) just underneath the metal strip of the microstrip patch antenna the S -parameter, which is similar to the case where the optical axis
for the case  = =4. is located in the xz -plane, is again observed.
1146 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 47, NO. 7, JULY 1999

VI. CONCLUSION analysis of planar microstrip circuits,” IEEE Trans. Microwave Theory
Tech., vol. 38, pp. 849–857, July 1990.
In this paper, an efficient 3-D FDTD approach based on the [6] R. Mittra and P. H. Harms, “A new finite-difference time-domain
D-, E-, and H-fields that can be applied to arbitrary anisotropic (FDTD) algorithm for efficient field computation in resonator narrow-
dielectric materials is developed, and the numerical implementation band structures,” IEEE Microwave Guided Wave Lett., vol. 3, pp.
of the approach is presented in great details. With the help of the 316–318, Sept. 1993.
proposed approach, the influence of the anisotropy on the scattering [7] J. Fang, “Absorbing boundary conditions applied to model wave prop-
agation in microwave integrated-circuits,” IEEE Trans. Microwave The-
parameter of the line-fed microstrip patch antenna printed on uniaxial ory Tech., vol. 42, pp. 1506–1513, Aug. 1994.
anisotropic dielectric substrates is investigated for the first time. It [8] P. C. Cherry and M. F. Iskander, “FDTD analysis of high frequency elec-
is found that the operating resonant frequencies of the antenna are tronic interconnection effects,” IEEE Trans. Microwave Theory Tech.,
decreased while the angle (either  or ) increases. The information vol. 43, pp. 2445–2451, Oct. 1995.
[9] M. A. Megahed and S. M. El-Ghazaly, “Analysis of anisotropic
provided in this paper might be useful for designing a new type of
high temperature super-conductor planar structures on sapphire
microstrip patch antennas, which could be important for the purpose anisotropic substrate,” IEEE Trans. Microwave Theory Tech., vol. 43,
of circuit integration, deposited on the substrate consisting of arbitrary pp. 1989–1992, Aug. 1995.
anisotropic dielectric materials. Furthermore, if the magnetic flux [10] J. Schneider and S. Hudson, “The finite-difference time-domain method
density, B, is introduced into the model, then the present FDTD applied to anisotropic material,” IEEE Trans. Antennas Propagat., vol.
41, pp. 994–999, July 1993.
algorithm can be easily extended to the substrate with arbitrary tensor [11] S. G. Garcia, T. M. Hung-Bao, R. G. Martin, and B. G. Olmedo,
permeability. “On the application of finite methods in time domain to anisotropic
Although the proposed approach requires more computer memory dielectric waveguides,” IEEE Trans. Microwave Theory Tech., vol. 44,
due to the introduction of D and/or B, it is more flexible and accurate pp. 2195–2206, Dec. 1996.
than the previous approaches because no normal electric-field com- [12] J. P. Berenger, “A perfectly matched layer for the absorption of
electromagnetic waves,” J. Comput. Phys., vol. 114, pp. 185–200, 1994.
ponents are defined at the edges of the unit cell, spatial derivatives [13] A. P. Zhao, J. Juntunen, and A. V. Räisänen, “Material independent PML
within two space steps are avoided in the updating equations, and, absorbers for arbitrary anisotropic dielectric media,” Electron. Lett., vol.
most importantly, it can be connected straightforwardly with the 33, pp. 1535–1536, 1997.
MIPML absorber that is devoted to absorbing waves propagating in [14] A. P. Zhao, “Generalized material-independent PML absorbers used
for the FDTD simulations of electromagnetic waves in 3-D arbitrary
arbitrary complex (including anisotropic [14], dispersive, lossy [19],
anisotropic dielectric and magnetic media,” IEEE Trans. Microwave
and nonlinear [20]) materials. Theory Tech., vol. 46, pp. 1511–1513, Oct. 1998.
[15] , “Unsplit-field formulations for generalized material independent
REFERENCES PML absorbers,” in IEEE MTT-S Microwave Symp. Dig., vol. 2, 1998,
pp. 469–472.
[1] J. L. Tsalamengas, N. K. Uzunoglu, and N. G. Alexopoulos, “Propaga- [16] A. P. Zhao, J. Juntunen, and A. V. Räisänen, “A generalized compact
tion characteristics of a microstrip line printed on a general anisotropic 2-D FDTD model for the analysis of guided modes of anisotropic
substrate,” IEEE Trans. Microwave Theory Tech., vol. MTT-33, pp. waveguides with arbitrary tensor permittivity,” Microwave Opt. Technol.
941–945, Oct. 1985. Lett., vol. 18, pp. 17–23, 1998.
[2] Y. Chen and B. Beker, “Dispersion characteristic of open and shielded [17] M. Okoniewski and E. Okoniewska, “FDTD analysis of magnetized
microstrip lines under a combined principal axes rotation of electri- ferrites: A more efficient algorithm,” IEEE Microwave Guided Wave
cally and magnetically anisotropic substrates,” IEEE Trans. Microwave Lett., vol. 4, pp. 169–171, June 1994.
Theory Tech., vol. 41, pp. 673–679, Apr. 1993. [18] A. P. Zhao and A. V. Räisänen, “Application of a simple and efficient
[3] K. S. Yee, “Numerical solution of initial boundary value problems in- source excitation technique to the FDTD analysis of waveguide and
volving Maxwell’s equations in isotropic media,” IEEE Trans. Antennas microstrip circuits,” IEEE Trans. Microwave Theory Tech., vol. 44, pp.
Propagat., vol. 14, pp. 302–307, May 1966. 1535–1539, Sept. 1996.
[4] X. Zhang and K. K. Mei, “Time-domain finite difference approach to [19] A. Renko and A. P. Zhao, “Extension of the material-independent PML
the calculation of the frequency-dependent characteristics of microstrip absorbers to arbitrary lossy anisotropic dielectric media,” to be published
discontinuities,” IEEE Trans. Microwave Theory Tech., vol. 36, pp. in Microwave Opt. Technol. Lett., vol. 20, pp. 245–249, 1999.
1775–1787, Dec. 1988. [20] A. P. Zhao, “Application of the material-independent PML absorbers
[5] D. M. Sheen, S. M. Ali, M. D. Abouzahra, and J. A. Kong, “Application to the FDTD analysis of electromagnetic waves in nonlinear media,”
of the three-dimensional finite-difference time-domain method to the Microwave Opt. Technol. Lett., vol. 17, pp. 164–168, 1998.

You might also like