Beruflich Dokumente
Kultur Dokumente
Absrruct-The scattering from a zero thickness plane having finite sheet tionalelements(surfacepatches)ratherthanthinwiresseems
resistance and perforated periodically with apertures is calculatedfor a priori to be preferable.
arbitrary plane wave illumination. The surface current density within the
Inthefollowingapproach, we employsmoothsubsectional
unit cell is approximated by a finite number of current elements having
rooftop spatial dependence. The transverse electric field is expressed in current elements to represent the current on the conductor, and
terms ofthe current, and the electric field boundary condition is satisfied satisfy the electric field boundary condition over each conductor
in an integral sense over the conductor, generating a finite dimension subsection. Since the conductor shape in the unit cell is defined
matrix equation whose solution is the current density. Since the conduc- through the location of subsectional current elements. any shape
tor shape is defined through the locations of subsectional current ele-
aperture can be accommodated, and since the field is tested over
ments, arbitrary shaped apertures can be handled. The reflection coef-
ficient and current distribution are calculated for square apertures in the conductor, the sheet maybe perfectly conducting or resistive.
both perfectly conducting and resistive sheets, andfor cross-shaped aper-
tures. Finite resistivity is shown to cause the magnitude of the transverse 11. REDUCTION OF THE SCAMERINGPROBLEM TO A
magnetic(TM)reflectioncoefficienttodecrease more rapidlyandits FIKITE DIMENSION MATRIX EQUATION IN
phasetodecrease less rapidly,as theangle of incidenceapproaches UNKXORY CURREKTS
glancing. Throughdetailedplotsofthecurrentdensity,thecurrent
crowding around the apertures is made clearly evident. Fig. 1 shows the structure to be considered. The apertures are
periodically spaced. having periodicities dl and d2 in the x and y
directions. respectively. They need not be rectangular. The con-
I. INTRODUCTION ductor, located in the plane z = 0. has zero thickness and sheet
resistance R , 2 0.
C OKDUCTINGplaneshavingperiodicaperturesareusedfor
shielding, as ground planes and reflector surfaces for antennas,
forgroundplanes in printedcircuitassembliesthatconnect
The incident wave has the form
EOe/(wt-k'.r) (1)
integratedcircuits,andforband-passradomes.Althoughmany
approaches have beenused totreatperiodicsurfaces,noneare where r is the position vector and for incidence from z > 0. the
well suitedforperiodicarraysofarbitraryshapedapertures in incident plane wave has the wave vector
perfectly conducting or resistive sheets. k' = k i x o + k;.yo - ki20
Reflectionfrom an arrayofrectangularaperturesin a per-
fectlyconductingsheetwasinvestigatedthroughavariational = ko (sin 8 cos @xo+ sin 6 sin @yo- cos 8zo). (2)
approach by Kieburtz and Ishimaru [ l ] . Chensolvedtheaper- and ko = 6 where o is the angular frequency, po and eo
tureproblem[2]as well as thecomplementaryproblem[3] are the permeability and permittivity of free space, and Q and 8
(reflection from a periodic array of conducting plates) by repre- are the polar and azimuthal angles of incidence, respectively.
sentingtheelectricfieldandcurrentdensity.rcspectively,by
modefunctionsorthogonal overselectedrectangularregions. A . Fourier SeriesRepresentation of the Fields
However,themodalrepresentations in [2]and[3]preclude
The total field is the sum of the incident field (denoted by a
the analysisofarbitraryshapedapertures or plates. or finite
superscript i) and the scattered field (denoted byasuperscript
resistance conductors.
s). Sincethestructure is periodic.theFloquetcondition[7]
LuebbersandMunk[4]computedthetransmissioncoeffi-
must be satisfied: the surface current density as well as all electric
cient for an array of thin slotsby approximating the field through
andmagneticfieldcomponentscanberepresentedbyperiodic
anexpansionwhichincludesonlytheelectricfieldcomponent
functionsof x and y multiplied by thesameFloquetfactor
transverse to the slot. Since the tangential field is not considered,
the approach cannot be extended to wide slots or apertures in
-
exp (-~kf p ) where kf is the transverse part of the wave vector
k', and p is the transverse part of the position vector r.
general. In [ 5 ] . Pelton and Munk calculate the scattering from a
Consistent with the Floquet condition, the scattered electric
periodic array of crossed dipoles, while in [ 6 ] ,Munk and Burrell
and magnetic fields,andthesurfacecurrentdensitycanbeex-
' considerthefieldsproducedby an infinitearrayofpiecewise
pressed as
linear current elements. However, it is not clear how such thin
wire elements could be used to represent a sheet which is solid
exceptforapertures.Anapproachemployingsmooth subsec- .
R n 1
,Ei
J, = (ixxo + j , , y o ) e - i k f . p (7)
0, elsewhere.
where J , and J , are periodic functions ofx a n d y .
We now approximate the periodic part of the surface current Theparameters 7, and ib aresuitablychosenintervalsthat
densityasasuperposition of currentelementshavingrooftop define the sizes of the subsections.
RUBIN AND BERTONI: PERIODICALLY PERFORATED P L A i E 83 1
Fig. 2 shows rooftop functions used to represent the surface We nowcanrewrite (6) using (1 2) as
currep density, The functions overlap so that the current den-
sity J , is continuous piecewise-linear in the x direction and step-
wise-constant in the y direction. Similarly: the current density j,,
is continuous piecewise-linearin the y directionAandstepwise-
constant in the x direction. The subsections for J, have dimen-
. r ~ T~ in the x and y directions,respectively,while
~ i o n s ~ 2 and
the JY subsectionshavedimensions r, and 2rb. Thesubsec-
tionsoverlapsuchthateachpatchofarea T,Tb showninthe
figurehasat least onex-directedrooftopandatleastone y - and
directedrooftopoverlayingit.Withthisapproximationfor J,.
the charge density in each patch resulting from the divergence of
J, is a linear function times the Floquet factor. The net tangential
current flowing in a strip near the edge is finite.While the normal
component of the current in the rooftop approximation goes to
zerolinearlywithdistancefromtheedgeratherthanwiththe
square root. by taking T= and T b small enough the current can be
reasonably well approximated. Consistency with the edge condi-
tions for the current is thus assured [ 101. C. Application of the Boundary Condition
T o describe J,, it is necessary toincluderooftopfunctions Equation(1 5) expressesthetransversecomponentsofthe
whose rectangles of definition are centered on the boundary of scattered electric field in terms of the current coefficientsI,, and
the unit cell and cross from one cell into the neighboring cells. l y e .In order to find these unknown coefficients and therefore
Because the currents at opposite edges of the unit cell are related obtaintheapproximatesurfacecurrentdensity.theelectric
via theFloquetcondition,unknowncurrentcoefficients Zxa. field boundary condition E, = R,J, is applied.
I,, need only be defined at half of the points on the boundary of There are P + Q unknown current coefficients. The boundary
the unit cell. In Fig. 2, dots have been omitted from the redun- condition must therefore be tested at P + Q locations so that ( 1 5)
dant points on the boundary. For the subsections shown in Fig. +
gives a set of P Q equations in that many unknowns. As will
2, thcre are P = 30 coefficients I x a 7 and an equal number Q un- be discussed later: satisfying or testing the electric field at points
known coefficients Z y a . leads to numerical difficulties in solving the equations generated.
Other basis functionssuchaslinearelementsorpulsefunc- We therefore satisfy the boundary condition at z = 0 in an inte-
tions are not suitable to represent relatively solid sheets. Linear gralsense.This is done by integrating the x components of E,
elementssuchasthinwireshavefiniteradiiandit is not clear and J, over P line segments of length r, parallel to x and centered
what radii the wiresshouldhavenorhowfinea wire mesh is at each of the points (x,. y e ) . Two such segments alongx are in-
needed to represent a solid surface. Pulse functions (one-dimen- dicated by heavy lines in Fig.2. Similarly. the J) components of
sional)havebeenused to represent the residual current on ob- E, and J, are integrated over Q line segments of length ‘Tb parallel
staclesinparallel-planewaveguides [ 113 . However, a basis em- t o y and centered at each of the points (xp,,. yP+,),as indicated
ploying two-dimensional pulse functions would lead to fictitious inFig. 2 for one segment. Mathematically, the boundary condi-
line charges wherever the normal component of the current jumps tions take the form
discontinuously, and the electric field resulting from this approxi-
mation would differ markedly from the actual field.
The complex amplitude of the modal surface current density
can be expressed in terms of the coefficients Zxa andIY, via the
Fourier series theorem,giving
P
(16aj
Jxnm = IxaRxnrnenm(Xa.Ya)
a=1
where at z = 0: X = ~ 0
p + p ,= 1.2, Q (16bj
.2nn .2nm
I--I I-Y
e n i 1 2 ( x , y=) e d l e d 2 . where JSx and Jsy are thex andy components of the surface cur-
In theforegoingequations, R,,, and Rynin aretheFourier rent density J,, and T , and T b are as previously defined.
series coefficients of R,(x. J)) and R J x , y ) . and from (91, (10). We now substitute J,, and Jsy. as given by (7) and (8). and
and (11) are E: and E$ from (15), into (16). Interchanging the order of sum-
mationandintegrationin(16),andcarryingouttheindicated
integrations, we obtain the set of equations
Q Equation
(1 in of7) form
is the amatrix
dimension
equa-
finite
cP
a=1
JxaZyxap + 2 Iya(Zvyap - R s F y a p )
a=1
(17b)
tion thatrelatesthecurrentcoefficients
dent electric field. When solved,
Ixa and Iya totheinci-
(17) yields the surface current
-
- -Eye, 0 = 1 , 2 , .... Q. density
existing
on conductor
the when
illuminated
plane
a by
Here, .Exo and E arethe X and y componentsoftheincident wave.
yo
electric field quantlty Eo, and The Fourier series in (1 8a)-(18d) are absolutely convergent.
This can be seen by substituting the expressions for the coeffi-
( :)
Sa k x n -
kg--:, cients R,,, and Rynj7z from (14) into(18a)-(18d). For 12,
)
770
z x x a p = -- mterms
infinite
thecontained
in
+the
magnitudes
03 the
of
2k0 n , m Sa(k;: kznm decrease
as
series
1
* %+a.xpcp. (1 Sf)
In (1 8), sa(t;)again is the sampling function, the asterisk indicates
complex conjugation, 6ti,u is the Kronecker delta defined as
1, u=u
L,IJ =
0, elsewhere, where rTE and rTM are the reflection coefficients relating the
TE and TM scatteredmagneticfields,and H;(TM): tothe
TE and TM incident fields, Hf(TE) and H f ( T M ) :x1 and x 2 are the
cross-polarization coefficients. related through reciprocity by
and
The unit vector along k; is cos @xo + sin @yo,while the unit
RUBIN AND BERTONI: PERIODICALLY PERFORATED PLANE 833
- - Chen
0 P - Solution of eq. (17)
I
3.9 4.4
T-
r----- 1
13.9
- -
22
T
I
31
i
5.1 4.9
7- I
I9 I
-
4 X
19.0 10.3 I
I
- -
41 46
1%
14.3 13.6
+ 5.8 T
I I_ I
I 113 I
I
-I
7.7 7.6
--L
10.6
-
6.1
7.0
10.3
-6.7
10.5
-II %1
I
I
-
121
-
71 67
I-
----_-
t7 i7 I
I.
L----- _I
Fig. 5. Current distribution (mA/cm) over a quarter section of the unit
cell when a normally incident plane wave illuminates a perfectly con-
ductingplaneperforatedwithsquareapertures (E, = 1V/cm, k o =
-
111
I
I
-
102
-
87
&
80
I
0.001 c m - 1 , N = M = 16). Dimensions in centimeters. L_ ___------- _I
Fig. 6. Currentdistribution(pA/cm) over a quarter section of the unit
cell when a normally incident plane wave illuminates a resistive plane
x axis.Theperiodicities d l and d2 are both 0.30 wavelength. perforated with square apertures (R, = IO4 a/., Ex = lV/cm, k o =
The current distribution existing when an isolated but otherwise 0.001 cm-1, A'= 11.1 = 16). Dimensions in centimeters.
identicalplate is illuminated by thesameincidentfield is cal-
culated in [9] . (In that reference, although rooftop functions are
used to represent the surface current density, the plotscorrespond
t o linearly interpolated current densities.) The' current densities
in thetwo casesareverysimilar. Fortheperiodiccase,the
distribution of IJSx I is somewhat flatter while IJ, I is smaller.
Thisbehavior is notunexpected, sinceas theperiodicity goes
from infinity (the isolated plate) to 0.15 wavelength (a contigu-
ous perfectgroundplane), IJ, I shouldapproachaconstant
distribution of 2 A/cm while lJSv I should approach.zero.
I
direction.and k , = 0.01cm-' (X, 2 6 m). Furthermore, the
periodicities are dl = d2 = 1 .O cm, R, = 0, the dimensions of 16.0 11.1 to., I
the aperture are as shown. and N = 11.1 = 14. As expected, the
current increases where the aperture restricts the flow and spreads
I
5.9
5.6 d 5.4
d 4
out, going around corners. elsewhere. The surface charge density
is small, as indicated by the near zero divergence of the current. I
The reflection coefficient for the structure of Fig. 8 is plotted f11.9 10.6 10.3 I
versus k o inFig. 9. Forsmall ko theincident wave is almost I I
totallyreflected,butat ko = 4.15cm-'aresonance is ob-
served (total transmission). However, the wavelength at resonance
I 1% 6.6
d
6.0
---L -
5.8
as the angle of incidence approaches glancing. For normal inci- [ 121 D. A. Hill and J. R. Wait, “Electromagnetic scattering of an arbitrary
dence, only a small number of subsections is needed to compute plane wave by a wire mesh screen with bonded junctions,” Can. J .
Phys.. vol. 54, no. 4. pp. 353-361, 1976.
the reflection coefficient, while a greater number is required as
theparameter ko/(90 - 0) increases. A cross-shapedaperture
wasinvestigatedandexhibited,asexpected,asharpresonance
for normal incidence.
REFERENCES
R . B.KieburtzandA.Ishimaru,“Aperturefields ofan array of
rectangular apertures.” IEEE Trans. Antennas Propngar., vol. AP-9.
pp. 506-514, Nov. 1961.
C. C. Chen,“Transmissionthroughaconducting screen perforated
periodically with apertures.” IEEE Trans. Microwave Theor)- Tech..
vol. MTT-18, pp. 627-632. Sept. 1970.
C. C. Chen. “Scattering by a two-dimensional periodic array of con-
ductingplates.“ IEEE Trans. Antennas Propagar.. vol. AP-18,pp.
660-665,Sept.1970.
R. J. Luebbers and B. A. Munk, ”Cross polarization losses in periodic
arrays of loaded slots,” IEEE Trans. Anlennas Propagar.,vol.-AP-23.
pp. 159-164, Mar.1975.
E. L. Pelton and B. A . Munk,“Scattering from periodicarrays of
crossed dipoles,” IEEE Trans. Antennas Propagat., vol. A€’-27, pp. Henry L. Bertoni (M’67-SM’79) was born in
323-330. May 1979. Chicago, IL, on November 15, 1938. He received
R. A. Munk and G . A. Burrell, “Plane-wave expansion for arrays of theB.S.degree in electricalengineering from
arbitrarilyorientedpiecewiselinearelements and its application in NorthwesternUniversity,Evanston, IL in 1960,
determining the impedance of a single linear antenna in a lossy half- the M.S. degree in electrical engineering in 1962
space.” IEEE Trans. Antentzns Propagat., vol. ”-27. pp. 331-343. and the Ph.D.degree in electrophysics in 1967,
May 1979. bothfrom the PolytechnicInstitute of Brooklyn
R . E. Collin. FieldTheor! of GuidedWaves. New York: McGraw- (now of New York), Brooklyn, NY.
Hill, 1960, Ch. 9. pp. 368-371. In 1966 he joined the faculty of the Polytechnic
h’. Marcuvitz, WaveguideHandbook. M.I.T. Rad. Lab. Ser., no. Institute of New York, Brooklyn. and now holds
IO. New York: McGraw-Hill,1951,Ch. 1. the rank of Professor in the Department of Electri-
A. W. Glissonand D. R.Wilton,“Simpleandefficient numerical calEngineeringandComputerScience. He was
methods for problems of electromagneticradiation and scattering from elected Speaker of the Faculty for 1981-1982. Since 1967 he has served as a
surfaces,“ IEEETrans.AntennasPropagat., vol. AP-25,pp. 593- consultant in the areas of radar, ultrasonics and radio propagation. His re-
607,Sept.1980. search has dealt with variousaspects of wave propagation andscattering.
Born and Wolf, Principles of Oprics. Oxford: Pergamon. 1959. Ch. Theseinclude the geometricaltheory of diffraction,electromagnetic and
11.5. optical waves at multilayered and periodic structures. magnetoelastic waves,
S. Wu and Y. L. Chow, ”An application of the moment method to and ultrasonics as applied to both signal processing and nondestructive evalu-
waveguidescatteringproblems,” IEEETrans.MicrowaveTheory ation. He is a member of the Acoustical Society of America,the International
Tprh.. vol. M’M-’IO. on. 744-749. Nov. 1972. Union of Radio Science. and Sicma Xi.