Ram 2
Ram 2
Ram 2
RD-R169 M
I
AN INTRODUCTION TO RRDAR RBSORDENT MATERIALS (
P G LEDERER FEB 86 RSRE-85916 DRIC-R-99572
)U)
ROYAL SIGNALS AND RADAR ESTRBLISHNENT NALVERN (ENGLAND)
UNCLASSIFIED F/G 01/9
Ill
II
3.6~
1A L
I"-"' IB~
IIII
11111.25
11'-.4 u111111.
-III
p.
,,,. "°?
. -. .
.-. . ." " ". : " ,., ' . , .,.,? . .
~.- - . .-,'- .-. . . . . **.' .. *',-
'
, .-.
',.**:....-:-
-. , -." . .
-~~~- - r~ -7 - - - -
UNLIMITED803z
Lfl
* 00
O(n
AN INTRODUCTION TO RADAR ABSORBENT
MATERIALS (RAM)
Author: P G Lederer
RSRE
UNLIMITED 8 6 7 15 103
ROYAL SIGNAtS ANID RADARESTABLISHMENT
Report No 85016
AUTHOR: P G Lederer
:-I.
SUMMARY
J . - -
Copyright , .
S83/16
CONTENTS
1. INTRODUCTION
2. ELECTROMAGNETIC FUNDAMENTALS
3. ABSORBER TYPES
5. MEASUREMENTS
6. ABSORBER DESIGN
7. REFERENCES
C. °°
- 7_ -W.
.4 1. INTRODUCTION
restrictions on size and weight are much more severe. The design of
absorbers, therefore, must be tailored to each application, and this
requires an understanding of the factors which can influence performance.
Most absorbers rely on the bulk electromagnetic properties of materials,
but, at microwave frequencies, the wavelength of the radiation is of the
same order a- the dimensions of absorbing structures (a few mm to a few cm)
* and, crc.i,,equently, geometrical factors also play an important role.
A practical electromagnetic absorber must satisfy two main requirements-
it must reduce the reflection of incident radiation by the specified degree
over the frequency band of interest and it must achieve this - and continue( tO
* achieve it throughout its life - in its operational environment.
2
e, PP%.
- E, El r- r~~ It (2.2)
The universal constants, E and pot are respectively the permittivity and
0
permeability of free space and are related by
0
cp
0
= 1/c2
3
I
,,1 (2.3)
E E exp(jwt-yx)
(2.4)
H H exp(jwt- yx)
-Y a+ jB = jw(c . * (2.5)
The real and imaginary parts of y define the way in which the amplitude
and phase respectively of the wave vary with distance of propagation, a
being called the attenuation factor and B the phase factor, given by
B 2r (2.6)
- .5
E
Z =-=(2.7)
4
Z
o~c
01-I
(C 0)*
is therefore
z ~. .4S
2z Rflctonan Rfrcto (2.8) ..
E.
1 +E r t-.- (a)
(2.9)
H. + H = H (b)
1 r t
However, since electric and magnetic fields are related by the intrinsic
impedances of the medium according to equation (2.7), it follows that
E. ,
E1 r E tCc
Z = Z--2(C) (2.9)
1 1 2
E
r E.' (2.10)
1
lit. .5
A1EDILLM I MIUMj a
M~l~u u ,.. .,
ZI
j w.
REFRACTED:( ) EL= Et
Er P.EL. .7
Z, .~'-
MlCOEIV",
Uj
.j(
I,,,I
W
6'
Z 2 + Z1 (.1
of the two media and hence on their dielectric and magnetic properties.
If, furthermore, medium 1 is air (which, to a very good approximation can
z2
Z2= (2.1 b)
- zz
2 (2.11b)."2
zo
2 P
2 1
C-1
-2
6]
'Ie
then the impedances of the two media are equal, (Z2 Z I) and, according
to equation (2.9), the reflection coefficient falls to zero - this condi-
tion is referred to as a perfect impedance match. If medium 1 is free
.
space, then the matching condition requires that p 2 E2' which implies
thickness beyond which lies medium 3, (and possibly media 4, 5 etc beyond
that) then additional reflections will occur at these subsequent inter-
faces which will contribute to the total reflected wave seen in medium 1
(Figure 2.2). The total reflected wave, therefore, depends, not only on 7
the nature of medium 2, but also on the materials comprising the subsequent
layers and the layer thicknesses. The incident wave sees an effective
impedance, ZL, at the first boundary, which is no longer simply Z2 but is
modified by some function of d 2 , Z3 , d3 , Z4 , d4 etc:
Z -Z
= L 1 (2.11c)
L +1
and the concept of impedance matching is still valid -but now there are
7
the subsequent layers, ZT (see Figure A1.1):
ZL z2 tanh y 2 d2 (2.13)
ZL Z2 coth y2 d2 (2.14)
is lossless.
Appendix A3.
87
aL) M'ETR L
IMEDJUM2
/AEOILM
MEDIULM I
3. ABSORBER TYPES
z -+z(3.1)T.;
Z
r r
= L
L 0
L o
and that the reflected power falls to zero if this impedance is matched
*= (3.2)
and that the interface between such a material and free space would
wave as it propagates through it. The power in the wave decays with e-
= V F w(a2+b2 sin(tan b
- 00 sn-4an - "
9
Thus, in order to maximise a we require c'. C" r and p"r all to be large,
and this is in addition to the matching requirement, equation (3.2). Not
surprisingly, materials meeting these requirements over wide frequency
bands do not exist, and it is invariably necessary to employ one of the
impedance matching strategies, described in the following sections.
* upon the dimensions of its surface texture, but offers a uniform high
absorption level over a wide frequency range above this (Figure 3.6a),
and is also found to work well over wide ranges of incidence angle. These
desirable features make this type the usual choice for lining of microwave
anechoic chambers. A range of sizes is commercially available for chambers
of different lower operating frequency limits - pyramid depths of the
* order of 6 feet being required for operation down to below 1 GHz.
10
,~
K.1 7t
] N:CJVENTVWMAV
INC
Fig 3. cross-sea
Ameac of
aprid-tpe
fuctionalve aborberasmotl
taeeadbIasepdZodigdsrbto
3.3 Tapered Loading
z r= z (a)
21 3
and (3.4)
Itanh 'y
2 d2 b
12
.
L
rWPEfJ wVE Zo zz X2
Fig 3.3 A matching layer interposed between the air and the absorbing
med ium
Fig 3.4 The tuned layer Fig 3.5 The Salisbury screen
absorber absorber
_Z... - _---- 41 -..- 1 -6
layer are combined in such a way that their adverse effects cancel, while
at the same time, the overall thickness is kept to an absolute minimum.
These features make this type of absorber very popular despite the limited
bandwidth which inevitably results from a tuned system. It is frequently
referred to as a quarter-wave absorber, although this is, in fact, a
misnomer as will become apparent. It consists, as shown in Figure 3.4, of
a thin layer of lossy material, of thickness, d2 meacZ2 n
propagation factor, ywith a conductive backing. The effective impedance
of this system is found, by equation (2.13), to be
ZL Z tanh (3.5)
L 2 hY 2d 2
ZL Z (3.6)
using a lossless layer with suitable values of pi' and E' where energy
* conservation considerations would prohibit this.r r
2
2 = a( 2 1/4 -1 a (.)
(a+b ) cos(itan -)
where
a E j'-"j
rr rr
b c
E10 CVW'
rr rr
14
Thus, the larger the electromagnetic parameters are, the thiniter can the
absorber be made for a specified frequency provided, of course, that theirP
values are also compatible with the requirements for impedance matching.
Tuned, or narrowband absorbers of this type are commercially available ~
and typically achieve a peak absorption in excess of 25 dB (see Appendix
AO), and a bandwidth over which 10 dB absorption is achieved (10% power
reflection) of 10-20% of the tuned frequency (Figure 3.6c). This perfor-
I. ance at a tuned frequency in the region of 10 GI-z (for which X =30 mm) V
0
* can typically be provided by a sheet of 1.5-2 mm thickness. .
15
~0
/%
a --
00
0
Ln M C1
0 0
REFLCTIO
LOS (dB
Figd3.6 Somestypicalreenvt
aspecrafor the macnistypes of aM37
16 'w-1
increases, the tapered loading wideband absorber is approached. In fact
it is common practice to approximate a defined impedance taper (eg expo-%
nential) by means of a properly designed triple layer structure. The -
177
It may also be possible to effect bandwidth improvements without
unacceptable side effects by bringing into play entirely different concepts.
For instance a rough front surface texture will tend to scatter incident
radiation in all directions (if the dimensions of the surface texture are
significant compared to the wavelength). This does not, of course,
constitute absorption but it does reduce the specularly reflected signal.
It may also prove advantageous to gain some measure of control over the
effective impedance of a multilayer structure by judicious use of circuit
analogue techniques. A periodic plan3r array of conducting elements,
such as short narrow strips, crosses or discs present to an incident wave
a complex impedance which depends upon the spacing and dimensions of these
elements and the wavelength, and can be regarded as a circuit element in
an electrical network model of the absorber (see section 6). The simplest
example of this technique is, in fact, the Salisbury screen, for which the
lumped impedance associated with the resistive sheet is real and frequency
independent, but arrays of conductive elements permit frequency selective
surfaces to be realised. The main advantages of this technique are that . •
such frequency selective surfaces are easy to produce by metallic deposi-
tion through an appropriate mask and their size and weight are negligible.
18
4. MATERIALS FOR RAM
There are various mixing laws which relate the effective electrical
parameters of a composite material to those of its constituents. For
particles which can be regarded as spherical - which includes most lossy
fillers used in RAM - the dilution of permittivity is governed by a
logarithmic law (5)
where jEj is the magnitude of the complex permittivity and 6 is the phase
angle. Here, the subscripts c, f and b refer to the composite, the filler "e
and the binder respectively and v is the volume fraction of the filler in
the final mixture. A similar pair of equations holds for the complex
permeability. It is important to note that the electrical properties of
a composite material, and consequently its electromagnetic response when
deployed as an absorber, depend upon the volume composition and not, as
19
.. *%
20
Another area towards which current research efforts are directed is
the development of structu'ral RAM. The tuned layer is designed as a%
coating to be applied to a metallic structure, but an inherently absorbing
structural material would be able to displace the metal, thereby elimina-
ting the need for the coating and its associated maintenance. This places
the additional requirement of mechanical strength on the material. Where
only a lightweight panel is called for, capable of supporting its own %
weight or very light loads, a rigid plastic foam or a honeycomb structure,
loaded with a suitable lossy material, might suffice. On the other hand, 4
*1 the member may be asked to support loads for which metal parts are normally
used, and, for these situations, there is an interest in incorporating
lossy fillers in fibre-reinforced composites.
layers, these limitations are less severe. The addition of magnetic loss
confers three main advantages. Firstly, the additional loss mechanism
increases the attenuation constant, a, (equation (3.3)) compared to the
dielectric only material. Secondly, the factor by which the free space
wavelength is reduced in the material (equation (3.7)) is greater than it
otherwise would be, so that a given wavelength is supported by a physically
thinner layer. It is true that these objectives would also be served by a
21
non-magnetic dielectric with very large values of permittivity and
dielectric loss, such as a ferroelectric, but this would magnify the
impedance mismatch problem which is the substance of the third advantage
of a magnetic dielectric. The combined magnetic and dielectric properties
allow a much closer approach to the matching condition of equation (3.2),
thereby considerably alleviating this problem. Thus the concept of a
broadband absorber of reasonable thickness and without any transition
regions, such as was suggested in section 3.1, assumes the possibility
of a physical realisation. This is approached by solid sintered ferrite,
which is commercially available in the form of ceramic tiles, 5-10 mm
thick, that provide a reflectivity reduction of about 10 dB (10% power
reflection) from below 100 MHz to 1-2 GHz, with a peak at a frequency
between 200 and 800 MHz depending on composition and thickness. These
tiles do, in fact, retain some resonant character, which serves to enhance
their absorption over part of their operating bandwidth. Their high
frequency performance is degraded because the magnetic constants of the
ferrites decrease with increasing frequency in this region. Tuned layers
for frequencies in the range 2-40 GHz generally use a dispersed powdered .a
The two major classes of magnetic fillers are metallic powders and
the ferrites. The only metal powder in common use is carbonyl iron, which
has been for many years the most widely used magnetic filler. It is
produced by the pyrolysis of iron carbonyl and is readily available.
There are other magnetic metal and alloy powders, such as cobalt and
nickel, but these have not been widely explored for use in RA1M. The 'e
other major class of magnetic filler, the ferrites, is considered in
the following section.
22
4.4 Ferrites
class, the spinel ferrites, have a cubic crystal structure and can be
represented by the general formula, MFe 2 04 , where M is usually a divalent
transition metal ion, a combination of two or more such ions or alterna-
tively a combination of mono- and tri-valent ions that maintains overall
electrical neutrality. MnFe0 Ni Zn 1 xFe 0 and Li Cr Fe 0 are
examples of spinel ferrites. There is another class, usually containing
large divalent metal ions such as Ba, Sr, Ca, or Pb, which crystallises
into a hexagonal structure and these, too, may contain combinations of
divalent or bi- and tetra-valent ions. Not only does the wide variety of
compositions lead to a wide variation in the electromagnetic material
constants of ferrites, but, even for a given composition, these properties
are also dependent upon the microstructure of the material and hence on
the precise details of its manufacture.
The real part of the permeability, which has a constant low frequency
value, p if, falls to unity over a frequency band extending over approxi-
mately two decades, half of this reduction being achieved at the mid-point
of this band, f The magnetic loss rises to a peak, also at the frequency,
fr but is small at frequencies above and below this band. This magnetic
dispersion behaviour is very similar to the ideal dielectric relaxation
described by Debye (6), and is the result of the progressive failure of
the magnetisation (or polarisation in Debye theory) to follow the oscil-
lations of the driving field as frequency is increased. Figure 4.1, however,
shows that typical ferrites also exhibit a slight rise in the real part of
*the permeability at the low frequency end of the dispersion region and a
corresponding dip at the high frequency end, which is indicative of a
magnetic resonance phenomenon in addition to the relaxation (la,7).
23
Mat erialI
Parameters
r r
*r r
1f
ifo
0L
ci 1 f f0 0 0
0
FRQEC
Fi .1
Tegeea fr f h letomgeicmtrilpraee
Fig
4.1 sTecr
genr forrftheeetoagei aeia aaee
.-. - vr~'g ~ *- - r -° - .,--
Moreover, the magnetic loss often displays a small secondary peak or, at
least, a broadening, on the high frequency side of the relaxation disper-
sion which can lead to a small upper frequency extension of the useful
operating band of absorbers in which they are incorporated. The real
and imaginary parts of the complex permeability are in fact related to
each other by a pair of transform relations known as the Kramers-Kronig
relations, which are a fundamental result of the Causality Principle
relating the magnetic response of a material to the field causing it (la,8).
For spinel ferrites, jIf may typically lie in the range 10 to 1000
and f between 5 and 500 MHz. However, these two parameters are found
r
not to be independent but are inversely related according to Snoek's
empirical formula
where the constant has a value of 5600 MHz. The occurrence of the
dispersion region for spinel ferrites below 1 GHz accounts for their
direct use, in the form of sintered tiles, as absorbers in this frequency
region. For hexagonal ferrites, p1f is usually found to be somewhat
lower (less than 30) and f somewhat higher (greater than I GHz) than the
spinels, but the constant in equation (4.2) is found to be several times
larger (typically 15-30 GHz). The microwave properties of some ferrites
have been collected by von Aulock (7).
24
5. MEASUREMENTS
25
Microwav ..:
Source
r .0
A%
P ref network
P test analyser
--
B
HI
* 4*
ref
Microwav network
, /\-
_ m (5.1)
InP.
1
P
raa P.
a •,..
(5.2)
L = a
(5.3)
P
a
P
L- log-
a dB (5.4)
the desired signal are not readily predictable and vary with frequency.
26
The imperfect directional properties of the couplers degrade their ability
to isolate the return signal from the incident one. Non-ideal power
detector characteristics lead to measurement errors which depend on power".'
level and temperature. Fortunately, these problems can be overcome by
cross-polarisation effects.
guide or co-axial line. Many methods are described in the literature but,
because relatively few classes of materials are magnetic, most yield the
2--~
-_.,
_ , - _- . -' - _ -. -. , / " , .- - - - - " - . . " . ' " . • • - .... ...- o ' i
magnetic waves to propagate along its length and inhibits any tendency to
radiate outside its walls. These walls may comprise a boundary between
two suitable dielectric materials, as is the case in optical fibres, but
at microwave frequencies, and in the context of this chapter, they more
usually take the form of metallic walls - either a single conductor such
as circular or rectangular waveguide or a pair of conductors such as a
28
IW
a) b)
Medium 2
b) Zoc
medium 2
To 4 ~- To
Microwave Matched
Source _____________________________ Load
z -z
zL 1
FL + Z1
Z - Ztanh y 1 (5.5)
sc 2 nh 2
Z = Z2 coth y 2 1 (5.6)
measurements of the open and short circuit impedance, and it became the
standard method for many years. It will be evident that equations (5.5)
and (5.6) together constitute a pair of complex simultaneous equations
which can be solved for the two complex quantities Z and from which,
2 from2hi
29
in turn, the complex permittivity and permeability may be derived. By
multiplying and dividing (5.5) and (5.6) we find
and
tanh y 2 1 = (Zsc/Z )2 (5.8)
wall along which a probe may be moved in order to sense the electric field
as a function of position (12). The two properties of the pattern that
are of particular relevance are the ratio of the maximum to minimum
amplitudes, known as the Voltage Standing Wave Ratio (VSWR), S, and the
distance from the boundary of the first voltage minimum, x. The effective
impedance, ZL, seen at the boundary is then given by (1)
Z -j tanB x
L S lo L- 0
(5.9)
1 tan lx.
1j S
30
.... .-.
rIL
this method over a wide range of frequencies using the same apparatus,
but a large amount of adjustment of the slotted line and the open circuit
is necessary, making it a very time-consuming and painstaking exercise.
Another potential hazard, which is common to all transmission line methods, 14
is the error due to an imperfect fit of the sample into the guide. Any
air gaps between the sample and the guide walls can seriously affect the
reflection and transmission properties of that section of the guide and
The open circuit adjustment necessary in the open and short circuit
method can be avoided if the transmission coefficient can be accurately
measured. The filled section of line is inserted between two unfilled
sections with matched terminations to eliminate any unwanted reflections
(Figure 5.5). The reflection coefficient of this arrangement is shown
in Appendix A3 to be given by
-2y 2 d
r 1(1-e
r = (5.10)
2 -2y 2 d
1-r e
-Y 2 d 2
e (1-r)
T-
2 =2Y2 d
I-r e (5.11)
- Z°
pZ
1 -
2 + 00
-
-z , --
31
.. ,
32
-.
all frequencies from 0 to ). The response r(t), then takes the form of
a transient waveform with a gradual decay, and this can be analysed as a
similar infinite series of sine-waves each of which represents the 0.
response of the network to the corresponding input. In fact, the fre-
quency domain functions, V and R, are related to their time domain
counterparts, v and r, by the Fourier transform, F, thus (13):
and (5.14)
m
R(jw) = F(r(t)) = r(t) ej t dt
F(r(t)) (5.15)
The permittivity and permeability can then be found by the same means as
described in the previous section.
33 -
%AfNP;.jIG GA'rc
Ap
Vi 4
RESPONSE , 4, ,
*AMPLITUDE
AL TIM
WINDOW
c)
34
V.P
In this case, known as the single reflection case, the time window of the
sampling gate can be restricted so that it includes only this decay, as
shown in Figure 5.6b, and the rear face can then be ignored. The network ,
then being investigated is the front boundary of the sample only and the N
observed response is the step response of the front boundary reflection A
coefficient given by equation (5.12). In the alternative approach, the
sample length is short so that the response decay is far from complete
before the arrival of the next reflection. It is now necessary to open
out the time window to include the whole response, which is that resulting
from the total reflection coefficient of a finite sample and includes all
the re-reflected part-waves as described in Appendix A3.
35
mtrasis better in the sigl reflection method (16)...-
The fortunes of the time domain and frequency domain approaches have
changed over the years with the development of microwave technology. The
original advantage of time domain techniques was that a wide frequency
range could be covered in a single rapid measurement, so that results
could be obtained much faster than in the frequency domain, where separate
measurements must be made at a series of discrete frequencies. This was
particularly attractive when automatic network analysers were expensive
and slotted lines were the standard technique. However, the Fourier
transformation of the time domain data into the frequency domain adds some
delay in the presentation of results and degrades their accuracy due to -
the uncertainties in time origin for the two integrations, especially at
the higher frequencies. Reduction of these timing errors can extend the
high frequency limit of time domain measurements to about 15 GHz, but this
adds considerably to the equipment cost which becomes comparable to that
of some of the automatic network analysers (ANAs) available nowadays. A
less sophisticated time domain system, however, might prove to be cost
effective at frequencies below a few GHz, and might complement an ANA
operating between 2 and 18 GHz. For the highest precision in ANAs, the
* recently introduced Hewlett Packard 8510 can be used over the entire
frequency range from 50 MHz to 26 C!4z, but at double the capital cost.
Unfortunately, there is no cheap way of making the wide-band material
parameter measurements necessary for a systematic approach to RAM design.
36 _
6. ABSORBER DESIGN
v, A 1 1v2 + A 12 (-i2)61
A2 1 v2 + A2 2 (-i 2
37
-% --. -. - -. -.-' .6-1 W- -M
VV. I I . - W- L- - I 7%
-,
Z1 1 Z2 2 Z3 Y3
Z 1Y1 Z2 Y2 Z3 Y3
Fig 6.1 An arbitrary multilayer absorber and its transmission line analogue
iv i A '2
Fig 6.2 A 4-terminal network showing the sign convention used for the input
and output voltages and currents
1 2 3 i-.
7-4
10
V T A1 v2 v3 A24
i |
!
Fig 6.4 A 4-terminal network with a load connected to its output terminals
and a section of transmission line connected to its input
Y - ".
V. ,
AI (6.2)
IiIi LU
where
A A1
=
A 1 (6.3)
A2 1 A 2 2J
[
If two such networks, A and A 2 , are connected in cascade as shown
[]
in Figure 6.3, equation (6.2) may be applied independently to each. This
gives
V1
iI
=
=A
,[] v2
i2
ad
and
[I:
v3
i3
=
=A
A
2
v4
i4
"""
v v and i3 = -.
= (A x A2 ) (6.4)
The same procedure may be repeated for three or more networks in cascade.
38
T- TV.
A AxA xAx......A
A (6.5)
1 2 3 n
-2
The effective input impedance of this arrangement, looking into the input
terminals of the network, can similarly be written as
V
Z.
in
=
i
1(6.7)
Expanding this using equation (6.1) and substituting equation (6.6), the
input impedance can be expressed as
A Z +A
zi 11 t 12 (6.8)
21 t 2
z. -z
in o
39
no backing at all, which can be represented as a short circuit (Zt = 0)
or a matched load (Zt =t Z0 ) respectively.
[cosci
jZ sn
A= (6.10)
Lj/Z sina cos j-
where
a jyd (6.11)
A 01 (6.12) '
where
40
y (6.13)
R
41
Material
Interface.
Plane of
Incidence )
EE"
Fig 6.6 A pair of waves obliquely incident on a material interface showing the
field components resolved parallel to the interface. In a), the
electric field is polarised perpendicular to the plane of incidence,
while in b) it is parallel
.. °*,%°
e ** * * * *
Dj P.I E jJ
1 1i 1i+1 i+1
E: .'j
where e is the angle of incidence at the front face of the entire multi-
layer structure.
42
4J.
For homogeneous layers, the materials Will be chosen on the basis of the
criteria of chapter 4 and their electromagnetic parameters can be measured
by the techniques of chapter 5. If frequency selective surfaces are to
be incorporated, their equivalent lumped admittance must be determined
as a function of frequency from the appropriate equivalent circuit,
taking into account the properties of the medium in which they are
embedded. Finally, individual properties may be varied at will in order
* to refine the performance of the design and to assess the effect of such
* variations that may occur as a result of manufacturing tolerances.
43
.1
7. REFERENCES %
1973.
(4) US Patent 4 116 906 (1978); Japanese Patents 54 041 495 (1979),
54 110 497 (1979), 54 127 000 (1979), 54 152 402 (1978).
," ' " '..'.i
44 1
----
-1 W- 7- -T - - - - - - - - - - - - - - - - - - --
(1983).
" (18) I Anderson, Bell. Syst. Tech. J. 54 1725-31 (1975); lEE Colloq. on
"Multiband Techniques for Reflector Antennas", Oct 28, 1983;
•*.,p.
-."V
5.'
- 45
!% dr- . V . - . ~ r w~r' r r r c . V. Ir ..,. *
-2x
E 2 (x) = Ei2 (O) e (Al.)
2
E r2 3 E (0) e (Al.2)
r2(X) i2
Z -z E (0)
T 2 _ r2
r2
-- E(0)(Al.3)
T 2 i2
Thus the net electric field at an arbitrary point x in layer 2 is given by the
x -.
Y2 X +Y 2
E2 = Ei2(0)(e + r2 3 e ) (Al.4)
46
"I 7.4
" Z1' y1 y2
Z2 ' Y
42
- X
zL zT
* ..
." x =-d 2 x x=O
=0 - ..
2
a'b
a.'
4% as
,..
E(0) -Y x
H Wx-Z Ce -r2 e )(A 1.5)
2 2 2
Y x " .
e +r e
Z(x) E (x) Z 23 -(A1.6)
2(x) 2 -Y
2 x Y
2 x
e -r 23 e
y2 d2 -y2 d2
Z Z) e +CZTZ2 e
L 2 y
2 d -
ZT+Z2 e ZT- Z2 e
(A1 .7)
zT +Z2 tan
anh y 2
2z~
L
+ Z an
T a y22
2
47
-. 7~ ~0T-W ~ ~~
TV j -tW
Iv..
= j (A2.1)
r r 0 0
y = a + jS (A2.3)
Equating the real and imaginary parts of (A2.2) and (A2.4) leads to
221/4 -1
a= -wvo- -(a2+b) sin(itan (-b/a)) (A2.4)
0 0
and
a2+b2)1// 4 -
= r 0 •0 (a2+b2) cos(Itan (-b/a)) (A2.5)
Since
8l = T- , (A2.6)
48
- - - *. F- N VT - 'C I- , . - 'W..
In free space, where E' 1 and E" = = 0, the free space wave-
r r r r
length, Ao9 of a wave of angular frequency, w, is found to be
27r
A (A2.7)
0 0
(This also follows from the relationship vFP = 1/c where c is the velocity
0 0
of light.)
2r (A2.8)
m m
Thus the wavelength of a wave in a lossy medium is reduced below the free
space wavelength by the factor given by .
-- =
(A2.9) ' "
2 2
- (a +b ) /4cos(tan-1 (-b/a)) (.)
-4 .o.
I I (A2.10)
; =
o r r
:..-
4 -9°
,, , 49
j[:
APPENDIX A3. DIELECTRIC SLAB SANDWICHED BETWEEN TWO MEDIA
r 1=(A3.1)
z+z1
reflected and partly transmitted, but this time the boundary reflection
coefficient is
z-z
3
The reflected portion then propagates back again to the front interface,
suffering further attenuation and phase shift, where another partial reflection
and partial transmission occurs. At this point the direction of propagation is
from medium 2 to medium 1, so the reflection coefficient is -r 1*This process
continues until the internally reflected wave has diminished to insignificant
levels and it is evident that the total reflected and transmitted waves each
comprise a series of partial beams. The reflected wave, Er therefore, is '
given by the sum of the reflected partial beams
50
MEDIUMA I PMEDIV4 2 MiEDIUM 3
Y33
* rj
/ e, d
/r
/w
+.... ) (A3.3)
Summing this and dividing by the incident wave, E, gives the reflection
coefficient, F,
J2
Two specific cases can be considered immediately. If medium 3 is the same as
medium 1, then r2 -r I and
r -e-2d)
= 2 -2yd (A3.5)
1-r e
(A3.6)
1-r e-2yd
terms of the intrinsic impedance of the slab, Z, and the hyperbolic tangent of
(yd), which is obtained using equations (2.9) and (2.12).
(1r)(1-r -y
-22d (A3.7)
1 +r r e
12
51
7-7'
which, for the case when medium 3 and medium 1 are the same, reduces to
2 -yd
(1-r ) e
T-1- r2 e-Zyd
52
77-
P1
A 10 log 0 dB (A4.1)
10P2
P1 0 1O 1 1
1000 200~~10 1000
2
*Thus, if the power output of a network is 10% of the power level at the input,
the signal has incurred a power loss of, or been attenuated by, 10 dB - or
alternatively, the network can be considered to have a power gain of -10 dB.
In the case of microwave absorbers, an absorbent coating on a metal plate is
said to have a reflection loss of 10 dB (sometimes written as -10 dB to enphasise
that the return signal is reduced) if the reflected power is 10% of that reflected
* by the same plate without the coating.
Vi
A 20 lg1 1 dB(A4.2)
2
of the incident to reflected electric field, bu. the reflection loss, Ldo of a
microwave absorber is expressed in terms of a power ratio. Therefore
L =20 1 d (A4.3)
d lo 10 r
53
Notice that the phase information inherent in r, which is a complex quniy
is lost when its magnitude only is taken in calculating reflection loss.
54p
DOCUMENT CONTRQN. SH[EET
(As far as possible this sheet should contain only unclassified information. If it Is necessary to enter
classified information, the box concerned must be marked to indicate the classification eg (R) (C)or (S)
5a. Sponsoring Agency's 6a. Sponsoring Agency (Contract Authority) Name and Location
Code (ifknown)
7. Title
An Introduction to Radar Absorbent Materials (RAM)
7b. Presented at (for conference napers) Title, place and date of conference
8. Author 1 Surname, initials 9(a) Author 2 9(b) Authors 3,4... 10. Date pp. ref.
Lederer, P G ,P.T
11. Co tract Number 12. Period 13. Project 14. Other Reference
UNLIMITED
ibbiract
The purpose of this Introduction is to present, in a straightforward way,
the electromagnetic principles of Radar Absorbent Materials (RAM) for the
benefit of the non-electromagnetic-specialist who finds himself involved in this '
field. The fundamental theory of electromagnetic wave propagation in media and
at the interfaces between different media is reviewed and the various approaches
to absorber design are described in the light of this. The types of materials
required and the techniques for measuring both their electromagnetic prperties
,, and the performance of the finished absorber are outlined. Finally a means
of designing absorbers from a knowledge of the properties of its constituent
materials is described.
S80/48
I.