Sie sind auf Seite 1von 46

Microwave Optics

JOHN BROWN
Reader i.n Electrical Engineering, University College, London, England
Page
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
11. A Comparison of Light Waves and Microwaves.. . . . . 110
111. Microwave Diffraction Phenomena.. . . . . . 112
A. Relevance t o Antenna Design. . . . . . . . . . . . . . . . . . . 112
B. Diffraction Theory as Applied to Microwaves.. . . . . . . . . . . 116
C. Microwave Diffraction Measurements. . . . . . . . . . . . . . . . . 123
IV. Optical Instruments Adapted for Microwavc Use. . . . . . . . . . . . . 128
A. Fixed-Direction Antennas. . . . . . . . . . . . . . . . . . . . . . . . . . . 129
B. Supergain Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
C. Scanning Antennas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
D. Interferometers.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
E. Spectrometers.. . . . . . . . . . . . . . . . 143
F. Interaction of Elect . . . . . . . . . . 148
V. General Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

The fundamental similarity between light waves and microwaves


leads naturally t o the use of well-established optical methods for the
design of microwave inst,ruments. Until recently this has been most
marked in the development of microwave antennas, for which advanced
optical techniques have been used. The steady reduction in the wave-
lengths available from microwave oscillators has led t o the construction
of many optical-type instruments, and such instruments are likely t o
play an increasingly important role in microwave measurements.
The object of this paper is twofold: first, t o draw attention t o the
parallel between optics and microwaves, illustrating this by a general
account of antenna design, and second, t o summarize the present position
in the development of optical-type microwave instruments. No attempt
is made t o give a detailed account of microwave-antenna design, a
subject which is fully covered in existing textbooks.
I. INTRODUCTION
The discovery of radio waves was a direct consequence of Clerk
Maxwell’s theoretical demonstration ( I ) of the electromagnetic nature
of light, and the essential similarity between radio waves and light waves
has been evident since Hertz’s earliest experiments in 1888 ( 2 ) .
107
108 JOHN BROWN

The foundations of microwave optics were securely laid in the years


immediately following by a number of elegant demonstrations of the
optical-like properties of radio waves. A brief selection of the more
striking experiments will serve t o show the extent to which the subject
was examined. Lodge ( 3 ) verified the basic laws of reflection and refrac-
tion and succeeded in focusing radio waves with the aid of a lens, using
pitch as the refracting medium. Righi (4) demonstrated the polarized
nature of radio waves from the transmission characteristics of wire grids:
for one direction of the wires virtually complete transmission resulted,
but when the grid was turned through go”, the incident wave was re-
flected. The relation of polarization to the directions of the electric and
magnetic fields was elucidated by Trouton ( 5 ) as a result of a theoretical
study of reflection at a dielectric surface for oblique incidence. He showed
that the optical plane of polarization is the plane containing the direction
of the magnetic field and the direction of propagation of the wave.
Here we encounter the first difference between radio waves and
optical waves, a relatively minor one of nomenclature: in radio work,
polarization is firmly associated with field directions and it is therefore
more convenient t o refer t o the direction of polarization, taken as that
of the electric field, rather than t o a plane of polarization. The plane-
polarized wave of optics becomes a linearly polarized wave in radio
nomenclature, and the term “plane” is reserved to indicate a constancy
of field properties in planes perpendicuhr t o the direction of propagatioii.
Of the many other experiments on the optical properties of radio
naves, we may single out that of Bose (6) on the phenomenon of total
refection as an interesting example of the possibility of using radio
methods t o carry out experiments which would be impracticable with
light waves. The arrangement used by Bose is indicated in Fig. I , the
incident radio wave being subject t o total internal reflection a t the
hypotenuse of the first prism, A . When a second identical prism, R , was
positioned as in the figure, a signal decayed exponentially with increasing
values of X , the separation between the prisms. This demonstrated the
existence of a wave outside the surface a t which total internal reflection
occurred and showed one of its most important properties, an exponential
decay with increasing distance from the surface. Waves of this type are
now referred to as “surface waves” ( 7 , 8 ) and are being extensively
investigated in connection with many aspects of radio propagation and
transmission.
Present-day applications of microwaves are closely associated with
waveguides, i.e., hollow metal conductors within which waves may
travel with small attenuation. Once again we find early interest in this,
shown by Rayleigh’s (9) theoretical analysis in 1897. Propagation can
MICROWAVE OPTICS 109

occur in waveguides only if the free-space wavelength of the radiation is


less than a critical value governed by the dimensions of the guide cross
section being used. When the wavelength is longer than this critical
value, the field in the guide is exponentially attenuated in the direction
of the guide axis and is said to be “evanescent.” Evanescent fields are
essentially of the same nature as the surface waves which exist in the
region between the two prisms in Bose’s experiment.
Despite the discovery of radio waves as a result of their relation to
light waves and the considerable body of physical experiments on their
optical character, the development of radio as a means of communication

R E F L E C T E D WAVE

1 I, 1L

TO
INCIDENT
c RECEIVING
’HAVE
AERIAL

FIG.1. Jagadir Bose’s experiment to demonstrate the fields which occur outside a
dielectric when total internal reflection occurs. A and B are identical prisms made
from a material whose refractive index is sufficiently high to ensure total internal
reflection a t the surface PQ.

owed relatively little to optical ideas. We find an obvious reason for this
in the values of the wavelengths used in practical applications. The early
experimental work described above was carried out with spark-type
generators producing relatively short wavelengths around 10 cm. The
invention of the thermionic valve provided a much more convenient
method of generating electromagnetic waves, but the available wave-
lengths from valve oscillators were much longer. I n thinking of optical
ideas, we naturally tend where possible to use the simplest concepts of
geometrical optics, an essential condition for this being that the dimen-
sions of the structure considered should be much greater than the wave-
length used. Radio wavelengths less than 10 meters were not generally
available until 1940, and it is obvious that geometrical optics can have
only a limited field of application for wavelengths longer than this. In
110 JOHX BROWN

particular, even the largest aiitennas used can have dimensions of only
a few wavelengths, so that any attempt to analyze such antennas using
optical ideas must introduce the coniplications of diffraction theory. On
the other hand, the distance between a radio transmitter and a receiver
is invariably much greater than the wavelength, so that geometrical
optics is of considerable value in analyzing the propagation of radio
waves. Examples of this arise in explaining the reflection properties of
land and sea surfaces by the optical Fresnel coefficients (10) and in
describing the effect of the ionosphere on radio waves in terms of reflec-
tion and refraction (11). Another radio problem which led to extensive
use of optical methods is the diffraction of radio waves by the curved
surface of the earth (12). Such problems were, however, regarded very
much as the province of the physicist, and the interest of the radio engi-
neer in optical ideas was not aroused until the advent of the ultrahigh
frequencies used in radar. The description “inicrowaves ” applied to this
region of the frequency spectrum makes it evident that engineers were
barely conscious of the existence of even shorter wavelengths! Before
pursuing the implications of optical ideas in the design of microwave
equipment and in the measurement techniques used, we will find it
convenient t o summarize the points of similarity and difference between
light waves and microwaves.
11. A COMPARISON
OF LIGHTWAVESA N D M I C R O W ~ Y E S

Light waves and radio waves, both being electromagnetic, have


basically the same properties: in particular, they propagate in free space
with the same velocity, c, and satisfy the same fundamental relation
between frequency and wavelength. Such differences as exist arise
entirely from the different orders of magnitude of the wavelengths in the
two cases. The longer limit of wavelength of the microwave region is
usually taken as about 30 mi, corresponding t o a frequency of 1,000 Mcs
and is the wavelength corresponding t o the division between longer
wavelengths, for which the circuitry concepts of low-frequency analysis
are applicable, and shorter wavelengths, for which electromagnetic field
concepts are essential t o an understanding of the observed phenomena.
The lower limit of wavelength in the microwave region is less clearly
defined, but can be taken as the shortest wavelength which can be gen-
erated b y radio techniques: this a t present is about 1 mm. The wave-
lengths of microwaves are thus many times larger than those of light
waves, and this points t o the first major difference between the two:
namely, t h a t with microwaves diffraction effects must always be con-
sidered, whereas in many simple optical problems diffraction can be
ignored.
MICROWAVE OPTICS 111

The next differences we consider arise primarily as a consequence


of the methods of generation: the early microwave experiments referred
to in the previous section relied on the use of spark-type generators which
produced radiation over a wide range of frequencies. All modern work is
carried out with electronic oscillators which generate essentially mono-
chromatic radiation. In optics, on the other hand, only specially selected
sources are monochromatic, and most work is done with sources which
radiate a band of frequencies. Further, the microwave oscillator is
coherent in the sense that it produces a single continuous wave with a
high degree of phase stability, whereas optical sources are usually inco-
herent, since they consist basically of a large number of oscillators,
whose phases are independent. The main consequence of a coherent
source is that interference effects can occur. Accordingly, interference
always occurs in microwave systems, whereas in optics special techniques
have to be used to allow the observation of interference.
The final difference associated with the sources is that microwave
valves produce linearly polarized radiation, but optical sources produce
unpolarized radiation, i.e., an assembly of linearly polarized waves of
random directions of polarization. The microwave signal, being invariably
monochromatic, coherent, and linearly polarized, is much simpler than
the incoherent, unpolarized light signal extending over a range of fre-
quencies and from many points of view is the easier to handle both
experimentally and theoretically. On the other hand, the inevitable
presence of diffraction in microwaves off sets these advantages.
The differences discussed in the preceding paragraphs arise because
of the different methods of generation, but it is convenient at this stage
also to mention the methods of detection. Clark Jones ( I S ) has sum-
marized the properties of detectors for visible and infrared radiation and
has compared these with an antenna connected to a microwave receiver.
This comparison makes it evident that the minimum detectable energy
is much smaller for microwaves than for light or infrared waves, a result
to be expected from the much smaller quantum of energy in the micro-
wave region. Practical microwave receivers using crystal rectifiers as
frequency changers are only a few times less sensitive than the theoreti-
cally best receiver for which the limit is set by thermal noise entering the
antenna. The introduction of Maser-type devices ( l h ) , which operate
as a result of the stimulation of energy transitions in semiconductors or
paramagnetic materials is likely to lead to a further improvement in
sensitivity. For laboratory measurements at power levels well above the
detection threshold, a variety of microwave detectors is available:
crystals can be used directly as rectifiers and thermal detectors such as
bolometers, thermojunctions, and thermistors all find application. For
112 JOHN BROWN

wavelengths less than 4 mm, the Golay detector (15), a pneumatic


instrument developed for use in the infrared spectrum is likely to be as
sensitive as the best crystals available.

111. MICROWAVE
DIFFRACTION
PHEXOMENA
A . Relevance to Antenna Design
As was already seen optical ideas were not generally applied in radio
engineering until the development of oscillators capable of producing
wavelengths less than 30 cm. At longer wavelengths, the possible size
of antennas relative t o the wavelength is so small that little help can be
expected from optics in solving design problems. Further, a t wavelengths
of 1 m and greater, antennas (16) usually consist of some arrangement
of wires within which prescribed currents are caused t o flow by suitable
connections t o twin-wire or coaxial feeder lines. The radiated field origi-
nates from the current in the wires, and the analysis of radiation prop-
erties follows naturally from an extension of the methods used a t low
frequencies t o calculate the magnetic field established by currents. Early
microwave experiments showed t h a t waveguides act as more efficient
carriers of microwave energy than two-conductor transmission lines, in
that the attenuation per unit length is reduced. This immediately poses a
problem t o the antenna designer in that waveguides do not lend them-
selves conveniently t o the excitation of currents in wires. The use of
waveguides t o replace transmission lines makes i t virtually essential t o
replace the conductor-array type of antenna by some more easily realiza-
ble alternative. Now, a n open-ended waveguide radiates energy over a
range of directions covering more than a hemisphere and approximates
closely t o a n optical point source. This immediately suggests that optical
methods be used t o focus the radiation into a beam of any desired shape.
As an example, we shall consider one of the simplest requirements, the
design of a n antenna t o produce a pencil beam, such as is radiated by an
optical searchlight, and we shall use this t o contrast the problems arising
with those of optical design.
Starting from our open-ended waveguide as the equivalent of a n
optical point source, we can produce a pencil beam by any of the standard
methods of collimation, for example, by using a paraboloid reflector as
illustrated by Fig. 2 . Looking a t this from the standpoint of geometrical
optics, we see that the rays from the point source are reflected a t the
paraboloid surface and are collimated, provided the source is placed a t
the paraboloid focus. Ideally, we should then have a parallel-sided beam
of circular cross section whose diameter equals that of the rim of the
paraboloid. I n practice, this is far from the case because of the relative
MICROWAVE OPTICS 113

sizes of the paraboloid diameter and the wavelength used. For most
paraboloid antennas in current use, the diameter-to-wavelength ratio
lies in the range 10 to 100. The aperture of the paraboloid is thus com-
parable to an optical pinhole, and it is obvious that diffraction will be
the dominant factor in determining the nature of the radiated beam.
In operation, an antenna is required to establish a field a t a great
distance, and its behavior can conveniently be described in terms of the
radiation pattern, i.e., the way in which the field strength varies with
direction a t a fixed distance, and power gain, which gives a measure of
the absolute field strength in any particular direction, usually that of

t' N

. P

FIG. 2. The use of a paraboloid reflector to produce a collimated beam. I n the


figure the focus F is shown t o lie in the aperture plane, this being often done in prac-
tice. The reflecting surface is illuminated by a waveguide placed at F . The direction
of the distant point P is specified by the spherical polar angles, 6 and 4: P N is the
normal from P to the plane z = 0.

maximum radiation. If we think in terms of the image-forming properties


of an optical system, the radiation pattern is essentially the image formed
by the point source in a focal surface a t infinity. The radiation pattern
and the power gain can both be related to the electromagnetic field
distribution in the aperture of the aerial by the Huyghens-Kirchhoff
diffraction integral. A comprehensive discussion of this has been given
by Silver (17). The precise choice of surface to be regarded as the aerial
aperture is not critical and may, for example, be the reflecting surface
of the paraboloid in Fig. 2. On the other hand, the integrations involved
in applying the Huyghens-Kirchhoff method are usually simplest for a
plane aperture, a convenient choice in our example being the circular
area A defined by the rim of the paraboloid.
114 JOHN BROWN

Extensive calculations have been made t o determine the types of


radiation patterns which result from possible aperture distributions, and
in these the Kirchhoff integral can be simplified by restricting the cal-
culation of the fields t o points a t great distances from the antenna. This
simplification corresponds t o considering the Fraunhofer diffraction
region (It?), in which the field a t the observation point is calculated as if
the contributions from the different parts of the aperture travel equal
distances as far as changes in amplitude are concerned, but have phase
differences due t o small differential pathlength changes. We shall see
later t h a t while Fraunhofer diffraction theory gives a good guide to the
behavior of antennas, there are occasions concerned with antenna mens-
urement in which allowance must be made for the differential changes in
amplitude for the contributions. I n this case, we consider the near-field
or Fresnel diffraction region.
The application of Fraunhofer theory t o the calculation of aerial
radiation patterns leads t o a relatively simple result, involving a Fourier
transform relation between the functions representing the radiation
pattern and the aperture field distribution (19, 20). The radiation pattern
is expressed in terms of functions of the spherical polar angles 8 and 9
(Fig. 2) with the following results:

F ( x ~ , s=~ ~1) /IA +


~ ( z , y )exp [ + j k ( ~ l z ~2y)ldzdy (1)
where E”(Xl,S2) is the radiation pattern, as given by the electric field a t a
distant point (r,e,+); C1 is a constant which depends on the particular
value of r considered; E ( x , y ) is a complex function giving the amplitude
and phase of the electric field in the aperture; and X1,S z are functions
defined by
X1 = sin 8 cos 4 S z = sin 8 sin 4 (2)
Many important properties of radiation patterns can be deduced
from these equations, and, what is more important, the best choice of
aperture distributions t o achieve close approximations t o desired radia-
tion patterns can be determined. We find in our pencil-beam example
t h a t a n almost essential condition is that the aperture distribution should
have constant phase over the aperture A , i.e., t h a t A should coincide
with a wavefront. The qualification “almost essential ” arises because
of the possibility of supergain antennas, which will be discussed in
Sec. IV,B. Now, the geometrical optics design which led t o the choice
of a paraboloid as the reflecting surface to collimate the radiation from
the point source leads t o the condition that A is a wavefront, since the
wavefronts are the surfaces normal t o the collimated rays. We thus find
t h a t the use of diffraction theory t o relate the radiation pattern t o the
MICROWAVE OPTICS 115

aperture distribution confirms that geometrical optics does give an


acceptable design for an antenna to produce a pencil beam.
So far nothing has been said of the influence of changes in magnitude
only in the aperture distribution. Such changes have a relatively smaller
effect than do changes in phase and alter only the details of the pattern
rather than its fundamental character. For example, the patterns in
Fig. 3 show that “tapering” the magnitude, i.e., progressively reducing
the magnitude from the center of the aperture towards the edges, slightly
increases the angular width of the main lobe of the beam and considerably
depresses the side lobes relative to the main lobe.

-0 %I
FIG.3. Radiation patterns for a circular aperture of radius a: (a) uniform aperture
illumination, (b) aperture field proportional to (1 - +//az)), r being the distance from
the center, (c) aperture field proportional to (1 - r2/a2)2.

At this stage, we should consider the question of whether the fields


in the region between the point source and the radiating aperture do in
practice conform to those predicted by geometrical optics. Measurements
both of the aperture field and of the final radiation pattern confirm that
the agreement is adequate for design purposes provided that the aperture
dimensions are a t least 10 wavelengths. We are thus led to a combination
of geometrical optics and diffraction theory in designing microwave
antennas: the former provides the basis for designing the collimating
system, and the latter predictions of the radiation pattern which will be
obtained.
The above argument justifies the use of direct analogs of optical
instruments in microwaves, but also shows the need to consider diffrac-
tion effects in assessing the microwave performance. Other factors which
116 JOHN BROWN

influence antenna design will be discussed in See. IV. Before leaving this
introductory treatment, we may note one difference between the pencil-
beam antenna and its optical analog, the searchlight. This difference
arises because the microwave source is coherent, while the searchlight
source is not. Since the antenna aperture is analogous t o a pinhole, the
angular spread of the radiation pattern is determined by the aperture
dimensions, apart from the minor variations arising from magnitude
changes as illustrated by Fig. 3. The searchlight is illuminated by an
incoherent source, and the width of its beam is governed mainly b y the
source area, which causes the formation of a n extended image in the focal
surface a t infinity. This does not arise in the microwave case, since the
coherent radiation from the waveguide feed leads t o the same behavior
as if the source were a single point.
B. Diffraction Theory a s Applied to Xicrowaves
The importance of diffraction theory in microwave antennn design
has been demonstrated in the last section. Other aspects of microwave
work in which diffraction theory plays a n important part, t o be discussed
in this section, are
a . Scattering of a n incident wave by conducting and dielectric oOstacles.
The interest in this is two-fold: first, on the engineering side, scattering
by obstacles such as aircraft or ships plays an inherent part in radar
operation and must be estimated before the over-all performance of
any radar system can be assessed. Also, both in radar and microwave
communication links, the effect on the propagation of the radio waves of
obstacles such as trees, buildings, etc., and reflection by rough surfaces
such as the sea are relevant t o problems such as the siting of stations.
The second aspect of this topic is of interest t o physicists in that diffrac-
tion phenomena can often be more easily investigated by microwaves
than by light waves.
b. Problems arising in aerial measurements. This point has already
been mentioned and is of vital importance in that i t is often impossible
t o obtain a sufficiently large test site t o permit the examination of aerial
properties under far-field conditions. We are then forced t o examine
theoretically what measurement errors can arise from the need t o observe
the radiation pattern in what is essentially the near field of the aperture.
c. T h e investigation of a n y diffraction effects which lead to significant
differences in operation between microwave instruments and the optical
analogs from which they have been developed.
Our discussion centers mainly around two points. The first, which is
mainly concerned with a above, is the question of what diffraction theory
can tell us and the second, which applies equally t o b and c, is concerned
MICROWAVE OPTICS 117

with the near-field diffraction region and the problem of making adequate
allowance not only for the presence of an aperture through which the
field is radiated, but also of the aperture which acts as a receiver.
The theory of diffraction has been widely studied for many years,
and elaborate mathematical techniques have been used. Surveys of
various aspects of this theory have been given by Severin ( d l ) ,Silver (ad),
Zucker (dS), and Mentzner (24).An examination of the work done makes
it evident that only for simple shapes of obstacle such as the half-plane,
the sphere, cylinders of circular, elliptical, and paraboloidal cross sec-
tions, and the cone is it possible to achieve exact solutions. Further, these

FIG.4. Complementary gratings: the shaded regions represent a very thin conduct-
ing sheet. In applying Babinet's theorem, we must rotate the polarization by go", as
indicated by the electric field directions.

exact solutions are often in a form unsuited for numerical calculation, a


striking example being that for the conducting sphere. Although an exact
solution for the diffracted field when a plane wave is incident on a con-
ducting sphere can be obtained in a series form by quite a simple analysis
( 25), the computation required t o give numerical results is extremely
difficult. The single most important parameter, the scattering cross
section, which provides a measure of the power in the diffracted field,
has indeed been computed only quite recently (26).
One general theoretical result of great interest has emerged, namely
an extension of Babinet's theorem for complementary gratings. Booker
(27) has shown that if the diffracted field is known when a linearly
polarized wave is incident on a grating such as that in Fig. 4a, then the
solution for the complementary grating, Fig. 4b, can be immediately
deduced provided that the direction of polarization of the incident wave
118 JOHN BROWN

is rotated through 90 deg. This result has been of immense help in relating
the properties of slot aerials t o those of conducting dipoles.
For diffraction by most obstacles of engineering importance, such as
aircraft, we can only hope for approximate solutions, and it is this aspect
of the theory which requires the most active inr-estigation. It should
perhaps be stressed that often no very great accuracy is needed in these
solutions: for example, if the scattering cross section of an aircraft is
known t o within lo%>,the range of detection by a givcn radar can be
predicted t o within less than 3 yc.Similarly, in applying diffraction theory
t o calculating antenna performance, we find it useless t o strive for great
accuracy because of constructional tolerances, stray reflections, etc.,
which modify the aperture distribution t o a n extent which can usually
only be determined by a final measurement (28, 29).
We may classify the approximate methods used a t present roughly
into three groups. The first relies on the fact that it is much easier t o
calculate electrostatic fields near a conducting obstacle than t o find the
field excited by a n incident high-frequency wave. For obstacles which
are small compared with the wavelength, the electrostatic field often
provides a n adequate approximation t o the diffraction problem. Bethe
(30) used a corresponding magnetostatic approximation with great effect
in estimating the behavior of small holes of the kind met with in wave-
guide problems. The electrostatic approximation can be used as the first
term of a n asymptotic series expansion for a diffraction field, in which
the terms involve increasing powers of the ratio (obstacle dimension/
wavelength). Solutions of this type were first explored by Luneberg and
are being actively investigated by Kline and his co-workers (31).
The second group of methods is based on geometrical optics and is in
principle the same as that outlined in the previous section to explain
the functioning of a paraboloid reflector. The application of this method
to scattering by a conducting obstacle involves firstly the calculation
of the current density on the obstacle surface and secondly the calcula-
tion of the scattered field from the current distribution. The surface
currents are estimated as if the incident field is reflected a t each point
of the surface in the same way as a plane wave is reflected by a plane
surface. Senior (32) has demonstrated that this method can give results
in close agreement with known exact solutions, provided unnecessary
approximations are avoided in the evaluation of the scattered field from
the current distribution. This method has two obvious weaknesses in
that the current distribution is assumed zero on those parts of the surface
lying in the geometrical shadow of the incident wave and in that the
effect of singularities such as edges and corners are ignored. Keller (33)
has elaborated the method t o overcome these weaknesses by the introduc-
MICROWAVE OPTICS 119

tion of “diffracted rays,” which are excited either a t singularities or


when the incident wave grazes the surface. Such rays enter the shadow
region in addition t o altering the field in the illuminated region. An
attractive feature of Keller’s method is its close relation t o a simple
physical picture of the mechanisms underlying diffraction phenomena.
Millar (34) has applied a similar technique t o diffraction by a circular
aperture in a conducting plane and has obtained good approximations
by supposing t h a t the incident wave is diffracted a t each point of the
aperture edge in the same way as a plane wave is diffracted in the classical
Sommerfeld half-plane problem. The appearance of these relatively
straightforward methods of solution during the last few years represents
a striking step forward and leads t o the hope that results of sufficient
accuracy for engineering use will be forthcoming in the future.
The last group of methods t o be mentioned is slightly different from
the others in that i t provides a means of improving the accuracy of
approximate solutions. It is the variational technique, widely used in all
branches of physics. The wave equation plus the boundary conditions
which must be satisfied by the electromagnetic field are used t o express a
desired quantity, such as the scattering cross section, in terms of an
unknown boundary function, e.g., the surface current density on a con-
ducting surface. This expression is then manipulated until it is stationary
with respect t o changes in the unknown boundary function. If now an
approximation for this function, accurate t o the first order of small
quantities, is inserted into the variational expression, the answers for
desired quantity will be accurate to the second order. The variational
method can of course be combined with the methods described in the
previous paragraphs by using as the approximation to the boundary
function the value obtained by either method. A comprehensive account
of the procedure for setting up variation solutions has been given by
Borgnis and Papas (55).
We now turn t o points b and c, which were raised a t the beginning
of this section. We can consider all the points of interest by examining
the problem of measuring a n aerial radiation pattern: in general, the
measurement involves a n arrangement of the type indicated in Fig. 5 .
The reciprocal properties of antennas are such that it is immaterial
whether the pattern is measured when the antenna under test is used as a
transmitter or a receiver, and in the present discussion we consider the
former. A very powerful technique for dealing with this kind of problem
is available in the plane-wave spectrum concept introduced by Booker
and Clemmow (36). The cardinal principle in this approach is that any
electromagnetic field can be expressed in terms of a n assembly of plane
waves, provided that we allow the term “plane wave” t o include the
120 JOHN BROWN

surface wave of'the kind arising in Bose's prism experiment. This assem-
bly can either involve a summation of discrete terms, as, for example,
when a plane wave is incident on a diffraction grating, or an integration
over a continuous range of possible directions of travel for the plane
waves. The latter possibility arises when we consider diffracting apertures.

a DIRECTION OF
MAXIMUM RADIATION

FIG.5 . Layout to measure the radiation pattern of an antenna 2'. T is used as it


transmitting antenna, and the output from the fixed receiving antenna R is observed
as a function of the angle 00.

Suppose that the electric field over a plaiie aperture taken as t h e


coordinate plane z = 0 has components

Then the Booker-Clemmow analysis shows that the field throughout the
whole of the region x >_ 0 is given by

(Ci - Slk)F(SI,S2)exp [--jk(Slx + S2y + C2)]dSldS2 (4)


c

[-SIS2i , S Z[)- j k ( S 1 ~ :+ S z y + Cz)]dSldS2


+ (1 - S 2 ) j - C S Z ~ ] F ( S ~ exp
C
(5)
where E(z,y,z) and H(x,y,z) are, respectively, the electric and magnetic
field-strength vectors; i, j, k are unit vectors in the directions of the x,
MICROWAVE OPTICS 121

y, and z axes, respectively,


c = (1 - 812 - S,2)3*

Yo = (eo/po)%("is called the wave admittance for a plane wave in free


space; and cot po are, respectively, the permittivity and permeability of
free space. MKS units are used throughout. The form of these equations
shows the spectrum idea very clearly, the direction of an individual plane
wave being specified by the variables S1 and S2 and its amplitude and
phase by the complex spectral function, F(SI,S2). Further, this spectral
function is related to the aperture field distribution in the plane z = 0
by the Fourier transform relation:

So far the equations are quite general and can be used for any kind
of diffracting aperture, e.g., the Sommerfeld half-plane. If we restrict
the discussion to apertures of finite size, such as we have in antenna work,
the integrations in Eqs. (4) and (5) can be evaluated by stationary phase
methods giving expressions for the fields at distances from the aperture
much greater than the Rayleigh range, defined as a2/X, where a is the
largest antenna dimension. These far-field approximations show that the
plane-wave spectrum is identical to the radiation pattern if S1 and S2 are
expressed in terms of the spherical polar angles by Eq. (2). In other
words, we are back a t the same result as that implied by Eq. (l),which
was originally derived from the Huyghens-Kirchhoff integral.
We may look upon our radiation pattern measurement as being the
determination of the plane-wave spectrum radiated by the antenna, and
this suggests a parallel t o the measurement of the frequency spectrum
of a time waveform. This is achieved by passing the signal through a
narrow-band filter and observing the amplitude and phase of the output
as the mid-band frequency of the filter is altered. We must therefore
look for a corresponding filtering process in the antenna measurement:
the variable analogous to frequency is direction, as specified by say the
angle eo in Fig. 5. Now, if the receiving antenna R is a t a very great
distance from the test antenna T , all the plane waves in the radiated
spectrum will interfere destructively except for those within a very small
range of directions centered on O0. The output from R is therefore a very
good approximation to the spectral function for this angle, and the
radiation pattern can be plotted by rotating the test aerial through the
desired range of Oo. This argument therefore shows that the far-field
radiation pattern can be measured accurately provided the aerial separa-
tion T R is sufficiently great, hardly a surprising result. We can pursue
122 J O H N BROWN

our filter analogy yet further, however, by including the effect on the
result of the plane-wave spectrum of the fixed receiving antenna R, and
lve arrive a t certain conclusions which have not until very recently been
evident t o workers in this field.
As already mentioned, the receiving properties of an antenna are
closely linked to those when it is used as a transmitter: in particular, the
output from a waveguide connected t o the antenna varies with the
direction of a n incoming signal of constant power according to a function
identical t o the radiation pattern. A directive receiving antenna is thus a
filter in the sense that i t discriminates against certain directions of travel
for incoming plane waves, and this leads us t o a second possibility, first
appreciated by Woonton et al. (57), for measuring radiation patterns.
A large fixed antenna R is used so t h a t its radiation pattern is so narrow
as t o be directly equivalent t o the narrow-band filter of the frequency
measuring problem. All the filtering required then rests in this antenna,
and i t is no longer necessary to have destructive interference in the
transmission region between T and R : the antenna separation 1’R can
then be as small as may be desired. Woonton and his co-workers have
demonstrated experimentally the validity of this method of measurement.
A situation of immense practical interest is that in which the antenna
site is insufficiently large for destructive interference t o give an adequate
filtering action. Can we then enhance the filtering action by using a
sufficiently large receiving antenna t o give the equivalent of a longer site:’
The answer t o this is complicated by the essentially different nature of
the filtering processes in the transmission region between the antennas
and in reception by the fixed antenna. Brown (58) has derived the follow-
ing general relation for the signal observed in the waveguide connected
to the aerial I? for the conditions shown in Fig. 5:

where D is a complex quantity giving the amplitude and phase of the


received signal; F I T and FIR are the plane-wave spectra for the test and
fixed antennas, respectively, each calculated for unit power radiated by
the appropriate aperture; and
S2‘ = -C sin Bo + 8 2 cos 80 c‘ = c cos 80 + SZsin O0 (9)
The integral in Eq. (8) has been evaluated approximately for a few
specific numerical examples from which i t is evident that the size of the
MICROWAVE OPTICS 12s

fixed antenna has a considerable influence on the errors in pattern


measurements made on too small sites, This means that the correction
procedures which are based on results derived from classical diffraction
theory can only be used if the fixed antenna is omnidirectional. Much
further work remains to be done on this topic, but it is now evident that
the Booker-Clemmow plane-wave spectrum concept provides a very
powerful technique for problems of this kind. Such problems arise not
only in antenna measurements but in the adaptation for use at micro-
waves of conventional optical instruments, as will be seen in Secs. IV,C
and IV,D.
C . Microwave Diflvaction Measurements
Measurements of diffraction patterns a t microwave frequencies are
of interest both to the engineer as an essential part of the study of many
propagation problems and to the physicist as a means of obtaining very
much more detailed information than can be done by optical methods.
I n principle, the technique is extremely simple: a small receiving antenna
is introduced into the field, and the signal, which is picked up, is amplified
to a level suitable for display on a meter or an oscilloscope or for record-
ing by any of the standard techniques. A mechanical drive for the receiv-
ing antenna mount can move the antenna through the field region quite
rapidly, so that detailed plots of the field can be obtained in a short time.
There are two major sources of difficulty: first, the pickup antenna
and its associated feeder must not produce any marked alteration of the
field, and second, the possibility of stray reflection from laboratory walls,
apparatus stands, etc., is nearly always a potential danger. The first of
these is overcome by using antennas which are very small compared with
the wavelength being used and by positioning the cable or waveguide
feeder so that its length is perpendicular to the direction of the electric
field, this greatly reducing the excitation of currents on the outside wall
of the feeder. I n general, the most satisfactory antenna is a small (i.e.,
much less than the wavelength) unipole or dipole, this giving a signal
proportional to the electric field. If for any reason, it is essential to exam-
ine the magnetic field, a loop antenna, again of dimensions much less
than the wavelength, must be used, but careful design is needed to avoid
coupling to the electric field as well. Stray reflections are eliminated by
careful siting of the apparatus to insure that a series of reflections from
different walls or obstacles is needed to return energy to the region being
investigated and by covering possible reflecting surfaces with absorbing
materials.
The general level of agreement between theoretical and experimental
results obtained by many workers shows that when the above precautions
124 JOHN BROWN

are observed, accurate measurements can be made. A large number of


such results have been published and it will suffice here t o give one ex-
ample (39):in Fig. 6, the field diffracted by a half-plane is shown, as
measured by several workers, and compared with Sommerfeld’s theory.
The general order of agreement indicates that microwave measurements
are adequate as a check on theoretical results.
Besides the straightforward method outlined above, several elegant
techniques designed t o eliminate possible sources of trouble have been
developed. One particularly attractive method is due t o Cullen and
Parr (40) and is shown in outline in Fig. 7. The pickup antenna is replaced
by a small dipole mounted on a nylon thread which is rotated a t a con-
venient speed. This dipole scatters a small portion of the incident energy
when it is parallel t o the electric field, but none when i t is perpendicular
A
25c
20
I
- 1.5
I0
ID
0.5
0 1
0 1 2 3 4 5
X/X
FIG.6. Comparison of experimental and calculated values of electric field strength
in the plane of a semi-infinite conducting screen: z is the distance from the edge of the
screen. The points show experimental values and the solid line gives values calculated
from a n empirical expression due to Andrews (reproduced from ref. 3.9).

t o this field. The scattered radiation is therefore modulated a t a frequency


determined by the rate a t which the dipole spins and can be detected by
the same antenna which is used to launch the field being examined. The
modulation on the scattered signal from the dipole serves t o distinguish
it from any unmodulated reflections from the room walls, etc. The receiv-
ing system is so designed that the amplitude, phase, and direction of
polarization of the field a t the dipole position can all be determined. The
facility for measuring phase is particularly valuable, since this is some-
thing which cannot readily be done a t optical wavelengths.
The spinning dipole equipment has been used by Cullen and Matthews
(41) t o examine the field near the vicinity of a lens focus, and their results
provide a noteworthy demonstration of the way in which microwave
methods can be used t o elucidate optical problems. Figure 8 shows the
contours of constant amplitude and constant phase around the focus of a
lens of aperture 120 cm square, illuminated by a waveguide feed placed
MICROWAVE OPTICS 125

FIG.7. Arrangement of spinning dipole t o measure the field radiated by a horn.


The dipole is carried by a nylon thread which is rotated by the two magslips (repro-
duced from ref. 40).

t
g
- 40-
0
9a
50-
W
-I D I R E C T I O ~ ~OF
a PROPAGATION
In
V
60-
In
W
a
W
70-
z I I I I I I
2
I-
L 60 80 100 120 140 160 180

AXIAL SCALE READING, CM.

FIG. 8. Contours of equal intensity (solid lines) and equal phase (dotted lines)
near a lens focus (reproduced from ref. 4 1 ) .

a t a n equal distance on the other side of the lens. The major point of
interest is the demonstration of the phase change of 180 deg undergone
in passing through the focus. This is shown by the curvature of the phase
contours, implying t h a t the wavelength on the lens axis exceeds the free-
space wavelength. A direct measurement of the anomalous phase change
126 JOHN BROWN

near the focus, which is, of course, a smooth change and not a discon-
tinuous jump, is provided by moving the dipole through the path shown
in Fig. 9. The dipole is moved from point A by a distance equivalent t o n
wavelengths as indicated by the phase-sensitive properties of the appa-
ratus, until point E on the other side of the focus is reached. The path BC
is a portion of a constant phase curve, C being any convenient point in
the region of the first side lobe. The dipole is then moved parallel to the
DIRECTION OF

t
PROPAGATION
L I N E OF
CONSTANT PHASE

LINE OF ‘
CONSTANT PHASE

FIG. 9. Diagram used to illustrate the “anomalous phase change” on passing


through a focus (reproduced from ref. 41).

axis along CPD, the distance CD again being made equivalent to n indi-
cated wavelengths and finally brought back to the axis, point E , D E being
another constant phase line. E’ is found t o lie one wavelength further
from the focus than A , so that the phase change along EB exceeds t h a t
along DC by 360 deg. From the symmetry of the path, it follows that the
fields a t points 0 and P in the focal plane have a phase difference of
180 deg.
Hey et al. (4.2)have used n method for measuring the scattering cross
section of a n object, which has points of similarity to the spinning dipole
method. Again, the signal which carries the wanted information is sub-
jected t o a frequency shift so that it can be clearly distinguished from
stray reflections, etc. I n this method, the frequency shift is caused by
moving the obstacle a t a steady rate along the axis of the aerial used as a
MICROWAVE OPTICS 127

common transmitter receiver (see Fig. 10). The Doppler effect gives the
frequency shift, the rate being chosen to give a value of 10 cps.
Measurements have been made on spheres and disks, the former being
shown in Fig. 11. Excellent agreement has been obtained between the
OSCILLATOR

pSPHERE MIXER

TRANSMI ATTENUATOR
-RECEI
10 c/s
AMPLIFIER

METER

w
FIG. 10. Arrangement of apparatus used to measure scattering cross sections
(reproduced from ref. 42).
d/A

log,, ( d / A 1

FIG. 11. The equivalent echoing area of spheres of diameter d (reproduced from
ref. 4 2 ) .

experimental values and the theoretical results, showing that the equip-
ment can be relied on to give accurate scattering cross sections when
more complicated obstacles are used.
The experimental methods described above are invaluable in extend-
ing the range of numerical results to those problems for which theoretical
128 JOHN BROWN

solutions are not likely t o be forthcoming. Further, they provide the


possibility of checking the assumptions made by theoreticians in their
quest for simpler methods of solving diffraction problems. Further
progress in this is likely t o stem from a joint effort by theoreticians and
experimenters.
IV. OPTICALINSTRUMENTS USE
ADAPTEDFOR MICROWAVE
I n principle any type of optical instrument can be reproduced a t
microwave frequencies, but so far, apart from a widespread application
of reflectors and lenses as antennas, there have been no obvious uses for
more than a few special instruments. Interferometers and spectrometers
have been developed for such requirements as measuring the velocity of
electromagnetic radiation, and the electrical properties of materials and
typical examples will be described in Secs. IV,D and IV,E.
Either reflectors or lenses can be used as antennas, and most of the
common optical systems now have their microwave counterparts. The
relation of geometrical optics and diffraction theory t o aerial design has
been outlined in Sec. I I I , A , and we shall now examine some further
differences between optical and microwave design which result from the
differing requirements of the two fields. The design of' a n optical system
is largely concerned with reducing the various defects which arise in
image formation because of factors such as spherical aberration, off-axis
aberrations, chromatic aberration, etc., all of which must be considered
in relation t o the depth of focus and angle of view required. A microwave
antenna is usually designed to have a specified radiation pattern a t a
very large distance, so that only the imaging of an object at infinity
need be considered. Depth of focus is therefore not required.
The angle of view required from an antenna depends very much on
the application. An antenna used in a repeater link need accept a signal
from one fixed direction only, so that it may be designed for purely
on-axis operation. At the other extreme, a n antenna for a search radar
should be capable of accepting a signal from any direction. A simple way
t o do this is to rotate the complete antenna so that i t points in each
direction successively, but this becomes mechanically difficult for the
large antennas which are needed t o give narrow beams. Much attention
has been given to designing scanning antennas in which the direction of
the radiation pattern maximum can be altered over a specified range by
moving only a small part of the antenna, such as the feed in Fig. 2. The
problems involved in designing such antennas are essentially the same
as those in producing wide-angle optical systems. I n this paper, we will
look at these two extreme cases with a view t o seeing how optical ideas
are used and what additional complications arise.
MICROWAVE OPTICS 129

A . Fixed-Direction Antennas
From an optical standpoint, the design of a fixed-direction antenna
is a trivial problem. There is noarestriction on the shape of the surfaces
which can be used for reflectors or lenses, since the manufacturing
techniques available are such that aspheric surfaces can be made as
easily as spheric. Spherical aberration is thus eliminated by using either
a paraboloid reflector or any lens which gives perfect collimation for a
feed placed a t its focus. Off-axis aberrations are immaterial, since the
feed remains on-axis. The remaining optical aberration, chromatic,
determines the frequency range for which the antenna may be used, but in
practice this range is often restricted by other than optical considerations.
We shall now examine the requirements which must be met by a
microwave antenna but which are not usually serious in optical design,
the three most important being ( a ) power gain, ( b ) side-lobe level, and
(c) the reflection of transmitted energy back from the antenna into the
primary feed.
The first is a measure of the directivity of the antenna and is defined
for a transmitting antenna as the power flux at a point lying in the direc-
tion of maximum radiation to the power flux at the same point if the
same total power were radiated uniformly in all directions. The power
gain thus indicates how much the signal a t a given point is amplified by
using the directive antenna in place of an omnidirectional one. The
larger the antenna gain, the smaller need be the transmitter power to
achieve a specified signal level a t some distant point: every effort is thus
made to make the power gain as large as possible. The gain increases
proportionally t o aperture area, since increasing the aperture dimensions
decreases the angular cross section of the radiated beam, thus concen-
trating the available power more completely in the wanted direction. A
practical limit to antenna size is always fixed by considerations of space,
weight, and cost, and the next consideration is to insure that any antenna
has a radiation pattern consistent with what may be expected from
diffraction theory. This theory enables us t o calculate the power gain for
an aperture with a known field distribution, and a convenient standard
of reference is provided by the constant-phase constant-amplitude dis-
tribution for which
Power gain = 4rA/X2 (10)
A being the aperture area. Practical antennas have gains ranging up to
80% of the value predicted by the above equation, although as will be
seen in the next, section, there is no fundamental reason why this value
should not be exceeded. A possible source of reduction in power gain is
130 JOHN BROWN

FIG.12. Lens-corrected horn under assembly: the refracting material is an artificial


dielectric embedded in expanded polystyrene. The lens is assembled from slabs in
which conducting elements are positioned to give the desired lens profile (American
Telephone and Telegraph Company).

losses in the antenna, e.g., absorption of power if a lossy lens material is


used or if an imperfect conductor is used for a reflector surface.
An antenna radiation pattern, like any diffraction pattern, has a
major lobe and an infinite number of minor lobes, or side lobes. Side lobes
are objectionable in that they represent the radiation of energy in un-
wanted directions if the antenna is connected to a transmitter or that they
permit the reception of undesired signals if the antenna is connected to a
receiver. The size of these side lobes relative to the main beam must be
MICROWAVE OPTICS 131

restricted in either case. A considerable degree of control over the side-


lobe level is given by adjusting the amplitude distribution across the
radiating aperture, the effect being calculable through the Fourier rela-
tion between aperture distribution and radiation pattern. The amplitude
distributions which give the smallest side lobes also give small values of
power gain, so that a compromise has to be effected between power gain
and side-lobe levels.
The reciprocity theorem for antennas tells us that statements about
the properties of antennas apply equally well whether the antenna is used
for transmission or for reception. The third point above is exceptional,
since it need be considered seriously only when the antenna is used for
transmission. Reflections from the antenna back into the feed cause
some of the radiated power to return to the oscillator, and this may
result in unstable operation. The level of reflected energy which is toler-
able depends on the oscillator tube characteristics and must not be
exceeded over the frequency range for which the antenna is to function.
This usually limits the working frequency more severely than do the
optical considerations of radiation pattern properties. In addition to this
effect on a transmitter, reflections are also objectionable in that they
represent a loss of power gain.
The choice of antenna for the type of application discussed here is
usually the paraboloid reflector, which is simple, is easy to make, and
has a power gain of 65 to 70% of the maximum practicable. An alterna-
tive which is sometimes used is the shielded lens, or lens-corrected horn
(Fig. 12), consisting of a correcting lens mounted in the aperture of a
pyramidal horn.
B . Xupergain Antennas
The gain of an antenna of aperture A has the value given by Eq. (10)
if the aperture distribution has constant amplitude and phase. This value
is closely related to the size of the Airy disk when a beam of light passes
through a pupil, and a t first sight it would appear that it represents a
maximum for the gain in the same way that the size of the Airy disk
places a limit on optical resolving power (43).On the other hand, almost
any low-frequency radio antenna has an effective power gain, although
the ratio A/X2 is extremely small, so that there is an obvious inconsist-
ency. This has been examined by many authors (44-47), and it is now
well established that the gain predicted by Eq. (10) is not a theoretical
limit but that there are major practical difficulties in exceeding it. It can
be shown that Eq. (10) represents the largest gain if the phase in the
aperture plane is kept constant. Any increase over this value requires a
distribution in which both phase and amplitude vary across the aperture,
132 JOHN BROWN

a result which has also been recognized by Marechal in his work on


improving optical resolving power. When there are no restrictions on
either the amplitude or phase of the amplitude distribution, there is no
theoretical limit to the gain which can be obtained from an aperture
of finite size.
Let us now consider what practical problems lie in the way of achieving
super-gain antennas, i.e., antennas with gains exceeding the value given
by Eq. (10): an indication of the answer is provided by low-frequency
aerials, which present an effective impedance to the oscillator of a largely
reactive nature. For satisfactory working, this reactance, which is usually
capacitive, must be tuned out by an inductance, and then an appreciable
fraction of the transmitter power is dissipated as ohmic losses in this
matching coil. The advantages of a high gain from a relatively small
aperture are thus largely off-set by the very low radiating efficiency. A
further obvious difficulty is that the cancellation of the antenna reactance
can only be effective a t one frequency, so that the over-all performance
of the system will be very frequency-sensitive.
How does this reactive effect arise in an antenna of the aperture type
discussed in previous sections? For an answer, we turn to the expression
for the radiation pattern or plane wave spectrum as given in Eqs. (3)
+
to (7). Provided that S12 S22 does not exceed unity, the quantity C
is real and the component plane wave in Eq. (4) is of the normal type.
+
When ( X I 2 SZ2)does exceed unity, however, C becomes imaginary and
the plane wave is attenuated in the z direction, being therefore evanescent.
Once again we find evanescent or surface waves playing an important
role in microwaves, and it is the fact that such waves are not evident in
optical problems which has led to the erroneous acceptance of the Airy
disk relation as a fundamental limitation on optical resolving power.
Any attempt to design an antenna with supergain requires an aperture
distribution which will strongly excite evanescent waves. Such waves do
not carry power away from the aperture as do ordinary plane waves, but
store energy during one half-cycle and return this energy during the
next. This is precisely what happens in a reactive circuit element, so that
the evanescent fields near the vicinity of the aperture play exactly the
same role as does the reactive part of the impedance of the low-frequency
antenna. It is obvious that this reactive field gives rise to the same
practical difficulties of increased ohmic losses and frequency-sensitive
operation.
Any appreciable increase in gain over the value is most unlikely to be
achieved in practice, but there are two possibilities which are likely to
receive considerable attention in the future. The first is the realization
of a moderate increase in gain, and the second is the design of super-
MICROWAVE OPTICS 133

directive antennas, which will exploit the reduction in antenna beam


width obtainable by using complicated aperture distributions, accepting
as an essential penalty greatly increased ohmic losses. The major prac-
tical difficulty in the second case is to achieve the required aperture dis-
tribution to a sufficient degree of accuracy.
C . Scanning Antennas
Scanning antennas are much more difficult to design than the fixed-
direction antennas discussed in Sec. IV,A because of the need to reduce
the off-axis aberrations. An obvious starting point is to use the designs
for wide-angle optical systems, such as the spherical mirror with a
Schmidt correcting plate (48),the Mangin mirror (@), lenses designed
according t o the Abbe sine principle (50), etc. The microwave antenna
designer has, moreover, many advantages over the optical designer in
the variety of manufacturing methods available and in being often able
to simplify the requirements to be met. These advantages include
a. The use of aspheric surfaces and the possibility of figuring these
surfaces to a high degree of accuracy. Tolerances of X/16 on surface shape
can usually be achieved in constructing microwave antennas.
b. The availability of materials with a wide range of refractive
indices. Kock’s (51) suggestion that arrays of conducting elements should
behave for microwaves in the same way that refracting materials do for
light waves has led to extensive work on what are usually referred to as
‘C
artificial dielectrics” ( 5 2 ) . Not only can such materials be designed to
have any specified refractive index, but they offer the possibility of lenses
in which the refractive index is a continuously varying function of
position.
c. The design of systems t o collimate in only one plane. This arises
because the requirements for scanning antennas are seldom the same in
the two principal planes: for example, it is often desired to sweep a
beam through a wide range of angles in the horizontal plane with little
or no movement in the vertical. A powerful method of design in such
cases is t o use a ‘‘crossed cylindrical” system in which separate collimat-
ing elements are used to focus the beam in the horizontal and vertical
planes. This is illustrated in Fig. 13, where the formation of the hori-
zontal beam is effected by some suitable antenna A , confined between two
parallel conducting plates, which radiates through the long thin slot S.
The radiation from X can then be collimated in the vertical plane by an
element such as the parabolic cylinder P of Fig. 13. In this example, A
can be designed t o scan the beam in the horizontal plane.
The freedom of design given by the above points is being exploited
t o produce antennas whose optical counterparts would be virtually
134 JOHN BROWN

impossible to make. The availability of a continuous range of refractive


index gives an extra degree of freedom to conventional optical design,
and this has been used by Ruze (53) to give scanners whose performance
is superior to those designed according to the Abbe sine condition.
Alternatively, completely new lens systems can be developed in which
the refractive index varies continuously throughout the lens volume.
The attractive possibilities resulting from this have long been obvious

FIG.13. Crossed-cylinder collimating system.

to optical workers, but practical difficulties have prevented their realiea-


tion. Luneberg (54) showed that if the refractive index, n ( r ) , depends
only on the radial distance r from the lens center, according t o the
formula
n(r) = [2 - ( r / ~ ) ~ ] $ ~ (11)
where a is the lens radius, then a collimated beam will result if a point
source is placed anywhere on the lens surface. This gives a perfect scan-
ning antenna, since the beam direction can be made t o vary through all
possible values by moving only the waveguide feed. Luneberg lenses have
been constructed by using a set of concentric shells of suitably chosen
refractive index (55) and have shown satisfactory performance. The
construction is difficult, however, and some simpler method is needed
before such lenses are widely used.
An even more striking departure from conventional optics is given
by the “bootlace” lens suggested by Jones et al. (56).This consists of two
arrays of antenna elements, such as dipoles, interconnected by lengths
of transmission line. Signals received by the first array from a suitable
primary source are subject to phase delays determined by the line lengths
MICROWAVE OPTICS 135

before being radiated from the second array. A two-dimensional or


cylindrical version of this lens is shown in Fig. 14. The most remarkable
feature of this lens is that it has four degrees of freedom: an ordinary
optical lens has two degrees of freedom, corresponding to the two sur-
faces, while microwave lenses of the kind discussed by Ruse have a
third degree of freedom resulting from the possibility of a continuous
variation of refractive index, in the direction normal t o the lens axis.
The bootlace lens preserves these three degrees of freedom, the positioning
I
!
I9 :
!

CONNECTING
p@ CABLES

FOCUS
-f
-
LAYER OF LAYER OF
RECEl V l NG TRANSMITTING
DIPOLES

I I
I I
I I

FIG.14. General arrangement of the “bootlace” lens.

of the elements of the two arrays corresponding to the choice of the two
surfaces and the variation in length of the connecting lines to the varia-
tion of refractive index. The fourth degree of freedom is the position of an
output element such as Q with respect t o its input element P .
Calculations of the behavior of such lenses are extremely simple
because of the “constraint” imposed on the “ray paths” through the
interior of the lens. These “ray paths” correspond to the transmission
of energy along the transmission lines and are completely determined by
the positioning of the radiating elements and are thus effectively fixed in
direction. An equivalent method of obtaining a fourth degree of freedom
would be to allow an axial variation of refractive index, but this would
lead to much more complicated design equations because of the lack of
the simplification resulting from the constraint on the ray paths. The
full possibilities of this form of lens construction remain to be explored,
but it is clear from the work carried out by Gent (57) that it offers great
scope for new forms of lens antennas. Further, the technique can be
136 JOHN BROWN

readily applied a t much longer wavelengths, and now that large aerials
are required for both radio astronomy and scatter propagation, there is a
prospect of lens antennas being used in what is normally regarded as a
pure radio portion of the frequency spectrum.
So far the designs considered can in principle be applied t o spherical
systems, in which the lens collimates simultaneously in both planes.
Other techniques are available when collimation in one plane only is
needed, as for antenna A in Fig. 13. Much work has been carried out on
“configuration focusing’’ (58, 59), in which a plane wave is excited within
a pair of parallel plates which are then bent t o produce any desired change
in the shape of the wavefronts. This gives a convenient equivalent to a
two-dimensional lens using a continuous variation of refractive index,
and Rinehart (60) has shown the general relation between the two types.
D. Interferometers
Three optical interferometers have so far been reproduced a t micro-
wave frequencies, those attributed t o Michelson, Fabry-Perot, and
Boltzmann. The principles of operation of each instrument are exactly
the same for microwaves as for optics, but in the former case diffraction
effects occur and must be considered if accurate results are to be ob-
tained. A further optical interferometer, the Jamin, is very closely
related t o a waveguide bridge circuit, widely used to measure atten-
uation and phase shift.
The essential parts of the microwave Michelson interferometer,
constructed by Culshaw (61) t o operate a t a wavelength of 1.25 em, are
shown in Figs. 15 and 16. The transmitting and receiving antennas are
each pyramidal horns fitted with polystyrene lenses, of aperture 6 in.
square. The radiation pattern has a width of 10 deg between its first
zeros. Although half-silvered mirrors can be made to operate successfully
a t microwaves, a more convenient beam divider is formed by two poly-
styrene sheets, each a quarter-wavelength thick, a t a separation which
can be adjusted t o give 50% transmission.
T o test the effect of diffraction on the performance of the Michelson
interferometer, Culshaw measured the wavelength of the radiation by
finding the distance between successive zero outputs from the receiver
when one of the reflecting plates is moved. This is essentially the optical
technique of counting fringes. The measurements were repeated for a
number of positions of the reflecting plates and horns with respect to the
beam divider, and the results are given in Table I. From our general dis-
cussion of diffraction in previous sections, we would expect accurate
results t o be obtained only if the reflecting plates and horns were placed
a t distances from the beam divider exceeding the Rayleigh range,
MICROWAVE OPTICS 137
a2/X (a = length of a side of the horn aperture, X = free-space wave-

,
length). For Culshaw’s interferometer, the Rayleigh range is 1.85 meters,
and inspection of Table I suggests that only for the largest horn and
reflector distances is there any hope of being free from diffraction errors.
The wavelength as calculated from the measured frequency, and the
VARIABLE
AIR S W C E

FREQUENCYfiT&%iECTED,(\,4 RADIATOR

STABILIZED
DSCl LL ATOR

RECEIVING
HORN.

m,

I 1 a
-
RECEIVER

t t
MIRROR m,

FIG. 15. Schematic diagram for a millimeter-wave Michelson interferometer


(reproduced from ref. 61).

FIG.16. General arrangement of Michelson interferometer (reproduced from ref. 6 2 ) .

velocity of electromagnetic waves appropriate to the atmospheric coiidi-


tions during the interferometer experiment is approximately 1 part in
3 X lo4 less than the final value in Table I. In interferometer applica-
tions such as the measurement of dielectric properties, this is quite an
adequate accuracy.
The accuracy of the interferometer can be improved by using greater
separations between the reflecting plates, the horns, and the beam
138 JOHN BROWN

TABLEI

Measured
Distances from beam divider, cm velocity of
- Measured electromagnetic
Transmitter Receiver Mirror Mirror wavelength, waves,
Position horn horn MI i142 cm 108 m/sec

1 150 150 175 175 1.2452 2.9972


2 100 100 125 125 1.2455 2.9980
3 50 50 75 75 1.2456 2.9982
4 25 25 50 50 1.2459 2.9989
5 310 150 600 40 1.2451 2,9970

Calculated velocity of electromagnetic waves under experimental conditions


= 2.9969 X 108m/sec.

divider, but this requires a large space for the equipment. This has been
done b y Froome (fib),using separations of up t o 21.5 m, and he has de-
duced the velocity of electromagnetic waves to a n accuracy of kO.7 km/
sec. I n this work, allowance is made for the effect of diffraction on the
measured wavelength. There is a need for further theoretical analysis t o

COMPOSITE
REFLECTORS

FIG. 17. Schematic diagram of millimeter-wave Fabry-Perot interferometer


(reproduced from ref. 63).

predict the differences between the wavelength as measured by a n


interferometer and the true free-space wavelength when separations
smaller than those considered by Froome are used. It is understood that,
work of this kind is in progress a t the Propagation Research Laboratory,
Boulder, Colorado, using a method based on the Booker-Clemmow
plane-wave spectrum concept.
Culshaw (63) has also developed a microwave version of the Fabry-
Perot interferometer, in which the beam reflectors consist of eight
quarter-wavelength polystyrene sheets, spaced at equal separations of
quarter free-space wavelength (Fig. 17). These reflectors give amplitude
reflection coefficients of 0.9977: increasing the number of sheets would
give reflection coefficients even closer t o unity, but the over-all perform-
ance of the instrument would not be improved because of increased losses
MICROWAVE OPTICS 139
by attenuation in the dielectric. The Fabry-Perot has two advantages
over the Michelson:
a. The response of the receiver as a function of reflector separation
is a series of sharply peaked curves (Fig. 18a), compared with the sinus-
oidal variation of receiver output with reflector position for the Michelson
(Fig. 18b). The very sharp response facilitates the accurate determination
of the separation between adjacent peaks. A similar difference in response
arises with waveguide techniques, the curves of Figs. 18a and b corre-
sponding to a cavity resonator output and a standing wave indicator
output, respectively. This suggests that the sharpness of the Fabry-Perot
response be expressed as an equivalent Q factor, and Culshaw has shown
that for his instrument the value is 60,000.

RECEIVER
RESPONSE

.ECTOR

FIG. 18. Typical receiver responses for (a) Fabry-Perot interferometer and (b)
Michelson interferometer (reproduced from ref. 63).

0. The Fabry-Perot is less sensitive to dif’fraction effects. This can


be most easily seen with the help of the plane-wave spectrum idea. The
transmitting antenna radiates a spectrum of plane waves covering a
range of directions around the axis of the instrument, and in the resonant
position only the wave in the axial direction has the correct phase relation
between the components arising from multiple reflections. The plane
waves in directions off-axis are thus “filtered out’) in the same kind of
way as those traveling very long distances between a transmitter and a
receiver (Sec. 111,B). This conclusion is strictly true only for a mono-
chromatic beam, and a detailed theoretical analysis is needed to deter-
mine the limitations imposed by frequency variations in the radiated
signal.
As already mentioned, interferometers can be used for dielectric
measurements, although normally it is preferable to use the much simpler
waveguide or cavity resonator methods. Difficulties with these latter
methods can arise when the wavelength becomes very short, i.e., less
than 5 mm, because of the small dimensions of waveguides, or when
artificial dielectrics with lattice structures which cannot be fitted into
normal resonator shapes are being measured. I n either case, the inter-
140 J O H S BROWN

ferometer provides a satisfactory answer, the Fabry-Perot being prefer-


able as the more accurate for both relative permittivity and loss tangent
measurements.
The two instruments so far described are completely analogous t o
optical instruments in that the radiation is of a free-space nature in all
the essential parts of the instruments. We may, however, substitute
waveguides for all or some of the possible paths, and this leads to a wide
range of variants of the basic types. For example, Froome used waveguide
for one arm of his Michelson interferometer (Fig. 19), the optical beam

RECEIVER
CORRECTED

6 5.21 5 m
+ MOVABLE
D REFLECTOR

-3.srn-

b- ADJUSTABLE
SHORT
VARIABLE ATTENUATORCIRCUIT

FIG. 19. Modified arrangement of Michelson interferometer as used by Froome


(reproduced from ref. 62).

~,
:?, : ;TA
VARIABLE
, PHASE
,?lYER,,

UNKNOWN ELEMENT

FIG.20. Mirrowave impedance bridge, similar in design to the Jamin interferometer.

divider being replaced by a hybrid waveguide junction (magic T). The


transmitted power divides between path A B , a section of rectangular
waveguide terminated by a short circuit and path CD,which lies in free
space. The two signals reflected back along these paths interfere in arm
EF of the hybrid, to which a receiver is connected, and fringes are
observed if the electrical length of either of the paths A B and CD is
altered. A useful feature of this arrangement is that a variable attenuator
can be included in the waveguide arm, this permitting the equalization
of the amplitudes of the two waves entering the receiving arm of the
hybrid. Sharp minima can thus be observed.
The microwave bridge of Fig. 20 provides an example of an inter-
ferometer for which all paths are enclosed in waveguide: the correspond-
MICROWAVE OPTICS 141

ing optical interferometer is that due to Jamin. The variable attenuator


and phase shifter in one arm of the bridge are adjusted to make the
receiver output zero : from the settings on these variables, the attenuation
and electrical length of the unknown element can be deduced.

FIG.21. Boltzmann interferometer.

The Boltzmann interferometer shown in Fig. 21 has an application


in the microwave field unlike any of the others which have been described.
The receiver output is observed as a function of the distance s between
the two reflecting surfaces. An analysis of the performance of this instru-
ment shows that the receiver output is given by (64)

where = 2s cos e/c (13)


and G ( w ) is the power frequency spectrum of the signal incident on the
reflectors.
If the radiation source is sufficiently far from the reflecting surfaces,
diffraction can be neglected and the incident wave can be regarded as a
single plane wave. The angle e in Eq. (13) can then be regarded as a
constant, so that the receiver output as a function of s gives the function
S ( r ) .We see from Eq. (12) that G ( w ) is the Fourier cosine transform of
S ( T ) ,and hence the measured S(r) can be used to calculate G(w). The use
of the Boltzmann interferometer in this way enables us to determine the
frequency spectrum from a wide-band source. Potok (65) has carried out
measurements on the radiation from sparked-type oscillators, and typical
interferograms are shown in Fig. 22.
The simple account of the Boltzmann interferometer given above
ignores the complications introduced by diffraction. So far, no detailed
examination of this has been made, but it is evident from Eq. (13) that
when a finite range of directions must be considered in the plane-wave
spectrum, the variable will not be directly proportional to s. The measure-
142 JOHN BROWN

ments then depend not only on the frequency spectrum but on the plane-
wave spectrum. We have here a parallel to the situation in the Fabry-
Perot interferometer, whose fldvantages can only be exploited t o the full
if the input signal is monochromatic.
Optical interferometers are often used for measuring lengths, the
principal limitation being on the maximum separation which can be
permitted between the partially reflecting surfaces of the Fabry-Perot,
or in the maximum difference in length between the two paths in the
Michelson interferometer. The factor which causes this limitation is the
line width of the light source, since this determines the longest time for
which the radiation can be regarded as coherent. The corresponding factor
in the microwave case is the frequency stability of the oscillator. Active
interest in the use of microwave interferometers for length measurements

(a) (b) (C)


FIG.22. Typical interferograms obtained by using the Boltmann interferometer:
(a) interferogram obtained from sparked source; (b) as (a) with filter in receiver input
tuned to the natural frequency of the source; (c) as (b) with filter tuned to twice
natural frequency. (Reproduced from ref. 65.)

is being shown by standards laboratories, and it appears probable that


their use will lead t o an increase in the maximum length which can be
measured by interferometric methods.
Interferometers of a rather different type from those described above
are now widely used in radio astronomy. The key problem here is the
provision of sufficiently narrow beams t o give adequate resolution for the
accurate measurement of the positions of the discrete astronomical radio
sources. When conventional antennas are used, very large structures are
required, such as the 250-ft diameter paraboloid a t Jodrell Bank (Eng-
land). An alternative solution which has proved of great value in many
observations is t o use two or more antennas spaced apart by a sufficient
distance: when the signals from these antennas are combined, interference
results in the formation of a multilobed beam, each lobe having a beam
width comparable t o that of a single antenna which extends over the
whole distance between the elements of the interferometer. Various ar-
rangements of the elements have been used for particular types of obser-
vation, and details are available in the textbooks on radio astronomy.
MICROWAVE OPTICS 143

E . Spectrometers
A microwave spectrometer is a straightforward adaptation of the
optical instrument, antennas being used to provide a collimated beam
and to act as a receiver. Several microwave spectrometers have been
constructed to operate under free-space conditions (66-68), there being
few essential differences between them except as to detailed dimensions
and wavelength of operation. A modified form has been developed in

FIG.23. General view of microwave parallel-plate spectrometer. I n operation the


top plate rests on the absorbing wedges which are positioned around the circumference.
A polystyrene prism is shown in position for a refractive index measurement (repro-
duced from ref. 69).

which the spectrometer is completely enclosed within parallel plates (69),


as shown in Fig. 23. The advantages of this form of construction are
a. The waves within the spectrometer region are cylindrical, whereas
in a free-space instrument, the waves are spherical. Field intensity
decreases in the first case as l/d+and in the second as l/r: the parallel-
plate arrangement gives the more efficient use of the available power.
b. The operating region can be completely screened by absorbing
spacing wedges, some of which can be seen in Fig. 23. This gives a freedom
from stray reflections and other interfering signals such as can be achieved
with a free-space spectrometer only if it is placed in an anechoic chamber.
c . The mechanical form of the structure makes it simpler to provide
rigid antenna mountings, and this leads to an increase in the angular
accuracy which can be achieved in measurements.
144 JOHN BROWN

The disadvantages of the parallel-plate system are that only linearly


polarized waves can be used and t h a t the direction of incidence of a plane
wave on a sample must always be parallel t o the plane of the plates. The
latter restriction is not particularly serious nor apparently is the first,
since in principle, results for any arbitrary direction of polarization can
be deduced from those for the two linear polarizations. The difficulty is
that the parallel plate spectrometer has so far been designed t o operate
only with the electric field perpendicular to the plate surfaces, the wave
then being a transverse electromagnetic one with phase velocity equal t o
the free-space value. If the other polarization, i.e., the electric field parallel
to the plates, is used, then the wave is of a waveguide type with a com-
ponent of magnetic field in the direction of propagation and a phase
velocity dependent on the plate spacing. The spectrometer could be made
to operate for this polarization, and allowance for the phase velocity could
be made in interpreting measurements. However, a very severe tolerance
would have to be placed on the plate spacing t o insure constancy of phase
velocity, whereas with the first polarization (electric field perpendicular
t o the plates) the requirement on plate spacing is only that there should
be no abrupt discontinuities or variations of sufficient rapidity to cause
unwanted reflections.
The above discussion of parallel-plate systems is quite general and
can be applied t o any form of microwave instrument based on optical
principles. For example, the Fabry-Perot interferometer could easily be
made in a parallel plate form, and i t is probable that widespread use of
this construction will arise in future work.
The uses of microwave spectrometers include refractive index deter-
minations by measuring the deviation of a beam passing through a prism
of the material to be measured, investigations of the reflection and
transmission properties of diffraction gratings, and measurements of the
diffraction patterns of cylindrical obstacles. The spectrometer is too
elaborate for measuring the refractive indices of solid homogeneous
materials but provides a simple method of obtaining results for materials,
such as certain artificial dielectrics, in which the refractive index may be
dependent on the direction of propagation with respect t o the principal
axes of the material. Such results can be obtained either with a prism or
by measuring the transmission characteristics of a parallel-sided sltth as a
function of angles of incidence. The relation between transmission-line
theory and the properties of plane waves (70) leads to the condition that
a plane wave incident on a loss-less slab of thickness t a t angle of incident i
will be completely transmitted, if
MICROWAVE OPTICS 145

where X is the free-space wavelength, N is an integer, and n is the refrac-


tive index for a wave traveling through the material in a direction corre-
sponding to the angle of incidence i.
Culshaw (66) has demonstrated the accuracy of this method for
materials of low loss, and Seeley (71) has extended it to permit a complete
determination of the properties of artificial dielectrics with any value of
loss tangent.
As with any microwave instrument, diffraction effects play an impor-
tant part in determining the accuracy with which spectrometer measure-
ments can be made. We shall consider the spectrometer as used t o meas-
ure refractive index by the prism method to illustrate the complications
which may arise and to indicate how they may be overcome. The wave
radiated by the transmitting antenna is not a parallel-sided beam as in
the optical case but has a spectrum of plane waves extending over a
range of angles, governed by the antenna aperture length a. The greater
this length, the smaller is the effective angular width of the spectrum,
but a maximum length is fixed by the need to avoid blocking too large
an arc of the spectrometer circumference. Further, the larger a, the
greater will be the Rayleigh distance and the more difficult it will be to
avoid the difficulties arising from working in the Fresnel diffraction
region.
The dimensions of the instrument shown in Fig. 23, were selected as a
compromise between the above effects and are
Spectrometer diameter d = 1 meter = 80X
Antenna aperture a = 12.7 cm = 10X
Operating wavelength X = 1.25 cm
Rayleigh range az/X = 125 cm = lOOX
The Rayleigh range is therefore a little larger than the diameter. A series
of electrical tests confirmed that satisfactory results could be achieved
with these dimensions. The spread of the radiation gives a receiver
response curve of just over 10-deg width at the half-power points. The
effective direction of an incoming wave is taken as the center of gravity
of the receiver response curve, and tests have shown that this can be
determined with an absolute accuracy of f0.25 deg and that changes in
direction can be measured to within k 0.1 deg.
The next possibility in the spectrometer is that the receiver may
accept not only the wanted signal but also an unwanted signal originating
in reflections a t the antenna carriages, or diffraction a t a prism corner, or
any similar source. When this happens, the receiver is under the influence
of two plane waves, but these can be separated because of a differential
phase change between the waves when the receiving antenna is moved.
146 JOHN BROWN

Here we have an example of the usefulness of the coherent nature of


microwave radiation, since it insures that the two incoming waves
establish an interference pattern from which information can be extracted.
The first point is that the interference causes a n oscillatory variation on
the curve of receiver output plotted against receiver angular position:
this means that any unwanted signal is made apparent. When the desired
signal is much stronger than the other, these oscillations can be smoothed
out t o give the response curve appropriate to the desired signal alone.

FIG. 24. Occurrence of interference ripples in the receiver response curves. The
wanted signal is assumed to travel in the direction OR, from the center of the spec-
trometer and the unwanted signal in the direction QR1. XX and X’X’ are equiphase
contours for the wanted signal, and Y Y and Y’Y’ are equiphase contours for the
unwanted signal (reproduced from ref. 69).

We can then interpret the measurements as if the unwanted signal were


absent.
The oscillations on the response curve can, however, be used t o obtain
information about the unwanted signal. This is most easily explained
with the help of Figs. 24 and 25. The receiving antenna is first moved
along the line XX in Fig. 24, and the power output plotted giving a curve
such as A in Fig. 25. The maxima of this curve occur when the electrical
center of the antenna is a t points R, and R z , for which the two plane
waves are in phase. The distance between these points is easily seen to be
d = X csc 4 (15)
The instrument is such th at 1-deg angular movement is exactly equal to
1-cm traverse of the receiving antenna, so that from curve A in Fig. 25,
d can be obtained being 3.65 cm in the example shown. There remains an
MICROWAVE OPTICS 147

ambiguity as to the sign of the angle 4, and this is resolved by taking a


new response curve when the antenna is pushed into the spectrometer
by one wavelength and is moved along X’X’, giving curve B in Fig. 25.
The maxima are displaced by the distance s, where
s = tan (+/2) (16)

and this gives the extra information needed to resolve the ambiguity.
This example has been considered in some detail to show that although
unwanted signals can easily arise in microwave instruments, they not
only give a clear indication of their presence but also produce interference
patterns from which their direction of origin can be traced. Further, from
the magnitude of the oscillation on the desired response, the relative
amplitudes of the two signals can easily be calculated.

ANGLE OF INCIDENCE

FIG.25. Enlarged portions of receiver response curves. The relative power output
is plotted against the angular position of the receiver.

We conclude this discussion of spectrometers with a brief note on


nomenclature. The name ‘(spectrometer ” is associated with spectros-
copy, and we might therefore expect microwave spectrometers to be
used in microwave spectroscopy. The instruments which are used in this
subject are in fact based on purely waveguide techniques, and since the
subject is a very specialized one, it will not be discussed here. Excellent
accounts are already available (72). Two questions do arise, however:
a. Are we justified in retaining the name “spectrometer” for the
instruments described in this section?
b . Will such instruments be used in spectroscopy a t the very short
wavelengths lying between the infrared region and the present micro-
wave region ?
An obvious justification for the retention of the name for the micro-
wave instrument is its essential identity in layout to the optical instru-
ment, It deserves the name for yet another reason, viz., that it will find
increasing application in measuring the diffraction patterns of cylindrical
148 JOHN BROWN

obstacles and from diffraction gratings. I n making such measurements,


we are essentially observing the angular spectra: once again we see, just
as we did in the discussion on interferometers, the closeness of the link
between the problems of diffraction and those of waveform analysis.
The answer to the second question is less certain. The measurements
required by microwave spectroscopists are those of the attenuation and
refractive index of materials over a range of frequencies, and while
these can be obtained by using a spectrometer, the Michelson or Fabry-
Perot interferometer is likely t o prove more convenient.
F . Interaction of Electromagnetic Waves and Materials
Many interesting and useful effects arise when light waves pass
through a material which is simultaneously subjected to a steady electric
or magnetic field. We may, for example, mention Faraday rotation, and
the Cotton-Mouton, Kerr, Stark, and Zeeman effects, all of which are
discussed in standard textbooks on optics. I n principle, such effects
must also occur for microwave radiation, but in general the magnitudes
of the effects are proportional t o some power of frequency so that they
are much smaller for microwaves than for light waves. Materials do
occur, however, in which these effects can be observed, and many impor-
tant developments have occurred. A detailed discussion would require
much more space than is available here, but it seems appropriate to
indicate the present situation, since i t illustrates the essential unity of
the subjects of optics and microwaves.
We start with what a t the moment are technically the most important
in microwave work, the Faraday rotation and Cotton-Mouton effect.
These have given rise t o a whole new class of waveguide components in
which the transmission properties are nonreciprocal ( 7 3 ) . A closely
related phenomenon is t h a t of ferromagnetic resonance (74), in which
the attenuation through a material shows a very strong peak as a dc
applied magnetic field is altered. The successful application of these
effects relies largely on the availability of suitable materials, and the class
of magnetic materials known as “ferrites ” proves the mostj useful because
of the unique combination of magnetic and insulating properties. De-
tailed discussions of recent advances in this subject will be found in
(75) and (76).
Many of the effects considered here can also be observed in semi-
conductors, and i t is probable that considerable progress in using such
effects to gain a fuller understanding of semiconductor processes and in
producing useful waveguide components will be made in the next few
years. So far, Rau and Caspari (77) have shown t h a t measurements of
Faraday rotation in a semiconductor such as germanium can yield infor-
MICROWAVE OPTICS 149

mation on the effective mass of the current carriers. On the engineering


side, Gunn and Hogarth (78) have developed a microwave attenuator
using a slab of germanium, the attenuation being controlled by a magnetic
field. I n optics, the Stark and Zeeman effects are most widely used by
spectroscopists and this is also true in microwaves.
When light waves are reflected or absorbed by a material, a mechanical
pressure is exerted: this is usually referred to as radiation pressure.
Carrara and Lombardini (79) demonstrated that a similar pressure was
exerted by microwave radiation, and this has been used by Cullen and
his associates (80-82) to develop a method of absolute power measure-
ment. The instruments so far constructed are capable of measuring
powers of 5mW and greater.
17. GENERALDISCUSSION
The very rapid development of microwave optics has been largely
due to the availability of information from light optics, but has now
reached a stage where further progress requires new lines of attack.
Many of these are also relevant to light optics, and both subjects will
profit from cooperative research efforts. Obvious examples of topics for
such efforts are the use of aspherical systems and the development of
supergain antennas and optical systems of higher resolving power. The
advantages to be gained by carrying out optical-type experiments a t
microwave frequencies was illustrated in Sec. I by Bose’s experiment and
are being exploited by workers in optometry. Microwave analogs t o the
optical system of the eye have been made and are likely t o lead t o useful
results.
We have seen the all-pervading influence of diffraction on microwave
optics, and it is clear that much work remains to be done in this field.
Again, cooperation, this time between theoreticians searching for approxi-
mation techniques and experimenters, is most likely to be successful.
The idea of the plane-wave spectrum is a powerful tool in handling the
particular type of diffraction problem of most interest to microwave
engineers, and there is considerable scope for further applications of
this idea.
In this paper, little has been said of the purely engineering problems
of producing optical-type microwave instruments, but this is of para-
mount importance, particularly for antennas when fairly large numbers
may have to be manufactured. An indication of future developments
here is given by the bootlace lens, which, although based on purely optical
ideas, breaks away from optical methods of construction. This may
well lead to the extension of the range of optical ideas t o even lower
frequencies.
150 JOHN BROWN

One of the most exciting and rapidly-changing parts of the subject


is the interaction between electromagnetic waves and materials, and
here we may confidently expect many new developments.
REFERENCES
1. Maxwell, J. C., Phil. Trans. Roy. SOC.London 166, 459 (1865).
2. Hertz, H., Ann. Phys. u. Chem. 34, 610 (1888);Phil. Mag. [5] 27, 289 (1889).
3. Lodge, 0. J., and Howard, J. L., Phil. Mag. [5] 28, 48 (1889).
4. Righi, A., Mem. accad. sci. Bologna 4, 487 (1894). See Carrara, N., Nuovo cimento
191 9, Suppl., 251 (1952); and IVhittaker, E. T., “History of the Theories of the
Aether and Electricity,” Rev. ed., p. 327. Nelson, Edinburgh, 1951.
5 . Trouton, F.T., Nature 39, 391 (1889).
6. Bose, J., cited in Somerfeld, A., “Optics,” p. 32. Academic Press, New York, 1954.
7. Barlow, H. M., and Cullen, ,4.L., Proc. Znst. Elec. Engrs. (London) 100, Pt. 111,
329 (1953).
8. Zucker, F. J., Nuovo cimento [9] 9, Suppl., 450 (1952).
9. Rayleigh, J. W. S., Phil. Mag. [5] 43, 125 (1897).
10. McPetrie, J. S., and Stickland, A. C., J . Inst. Elec. Engrs. (London)87, 135 (1940).
11. Ladner, A. W., and Stoner, C. R., “Short Wave Wireless Communication,”
5th ed., Chapter V. Chapman & Hall, London, 1950.
12. Bremmer, H., “Terrestial Radio Waves.’’ Elsevier, Sew York, 1949.
13. Jones, R. C., Advances in Electronics 6, 1 (1953).
1 4 . Wittke, J. P., Proc. I.R.E. 46, 291 (1957).
15. Golay, M. J. E., Rev. Sci. Znstr. 18, 347 (1947).
16. Aharoni, J., “Antennae.” Oxford Univ. Press, London and New York, 1946.
17. Silver, S., “Microwave Antenna Theory and Design,” Chapters 4-6. McGraw-
Hill, New York, 1949.
18. Sommerfeld, A., “Optics,” Chapter V. Academic Press, h’ew York, 1954.
19. Ramsey, J. F.,Marconi Rev. 9, 139 (1946); 10, 17, 41, 81, 157 (1947).
20. Spencer, R. C., “Astronomical Optics” (2. Kopal, ed.), p. 130. North Holland
Pub. Co., Amsterdam, 1956.
21. Severin, H., Nuovo cimento [9] 9, Suppl., 381 (1952).
22. Silver, S., Nuovo cimento [9] 9, Suppl., 401 (1952).
23. Zucker, F. J., Nuovo cimento [9] 9, Suppl., 442 (1952).
94. Mentzner, J. R., “Scattering and Diffraction of Radio Waves.” Pergamon Press,
London, 1955.
26. Stratton, J. A., “Electromagnetic Theory,” p. 563. McGraw-Hill, New York.
26. Hey, J. S., Stewart, G. S., Pinson, J. T., and Prince, P. E. V., Proc. Phys. Sor.
(London) B69, 1038 (1956).
27. Booker, H. G., J . Znst. Elec. Engrs. (London) 93, Pt. IIIA, 620 (1946).
28. Ruze, J., Nuovo cimento [9] 9, Suppl., 364 (1952).
29. Brown, J., Proc. Inst. Elec. Engrs. (London) 97, Pt. 111, 419 (1950).
30. Bethe, H. A., Phys. Rev. 66, 163 (1944).
31. Kline, M., Communs. Pure and Appl. Math. 4, 225 (1951).
3d. Senior, T. B. A., private communication (1957).
33. Keller, J. B., J. Appl. Phys. 28,426 (1957); Keller, J. B., Lewis, R. M., and Seckler,
B. D., ibid. 28, 570 (1957).
34. Millar, R. F., Proc. Inst. Elec. Engrs. (London) 103, Pt. C , 177 (1956); 104,
Pt. C, 87 (1957).
MICROWAVE OPTICS 151

36. Borgnis, F. E., and Papas, C. H., “ Randwertprobleme der Mikrowellenphysik,”


Springer, Berlin, 1955.
S6. Booker, H. G., and Clemmow, P. C., Proc. Znst. Elec. Engrs. (London) 97, Pt. 111,
11 (1950).
S7. Woonton, G. A., Borts, R. B., and Carruthers, J. A., J. A p p l . Phys. 21,428 (1950).
58. Brown, J., Znst. Elec. Engrs. (London), Monographs 286, 301 (1958).
39. Lewis, L. R., J. A p p l . Phys. 27, 837 (1956).
40. Cullen, A. L., and Parr, J. C., Proc. Znst. Elec. Engrs. (London) 102, Pt. B, 836
(1955).
41. Matthews, P. A., and Cullen, A. L., Proc. Inst. Elec. Engrs. (London) 103, Pt. C,
449 (1956).
48. Hey, J. S., Stewart, G. S., Pinson, J. T., and Prince, P. E. V., Proc. Phys. SOC.
(London) B69, 1038 (1956).
4s. Jenkins, F. A., and White, H. E., “Fundamentals of Optics,” 2nd ed., Chapter 15.
McGraw-Hill, New York, 1950.
44. Schelkunoff, S. A., Bell System Tech. J . 22, 80 (1943).
46. Chu, L. J., J. A p p l . Phys. 19, 1163 (1948).
46. Woodward, P. M., and Lawson, J. D., J . Znst. Elec. Engrs. (London) 96, Pt. 111,
363 (1948).
47. Toraldo di Francia, D., N u o v o cimento [9] 9, S u p p l . , 426 (1952).
48. Chait, H. N., Electronics 26, 128 (1953).
49. Cornbleet, S., unpublished work (1957).
50. Friedlander, F. G., J. Znst. Elec. Engrs. (London) 93, Pt. IIIA, 658 (1946).
51. Kock, kt-. E., Bell System Tech. J . 27, 58 (1948).
52. Brown, J., and Jackson, W., Proc. Znst. Elec. Engrs. (London) 102, Pt. B, 11 (1955).
65. Ruse, J., Proc. I.R.E. 38, 53 (1950).
64. Luneberg, R. K., “Mathematical Theory of Optics.” Brown University, Provi-
dence, R. I., 1944.
65. Kelleher, K. S., and Goatley, C., Electronics 28, 142 (1955).
56. Jones, S. S. D., Gent, H., and Browne, A. A. L., British Patent Application.
57. Gent, H., unpublished work (1957).
68. Myers, S. B., J . A p p l . Phys. 18, 221 (1947).
59. Brown, J., “ Microwave Lenses.” Methuen, London, 1953.
60. Rinehart, R. F., J . A p p l . Phys. 19, 860 (1948).
61. Culshaw, W., Proc. Phys. SOC.(London) B63, 939 (1950).
62. Froome, K. D., Proc. Roy. SOC.A213, 123 (1952); ibid. A223, 195 (1954).
6s. Culshaw, W., Proc. Phys. SOC.(London) B66, 597 (1953).
64. Farrands, J. L., and Brown, J., Wireless Engineer 31, 81 (1954).
65. Potok, M. H. N., Proc. Znst. Elec. Engrs. (London) 103, Pt. B, 781 (1956).
66. Culshaw, W., Proc. Znst. Elec. Engrs. (London) 100, Pt. IIA, 5 (1953).
07. Brady, J. J., Pearson, M. D., and Peoples, S. R., J. A p p l . Phys. 23, 964 (1952).
68. Ruse, J., and Young, M., J. A p p l . Phys. 22, 277 (1951).
09. Sollom, P. H., and Brown, J., Proc. Znst. Elec. Engrs. (London) 103, Pt. B, 419
(1956).
70. Booker, H. G., J. Inst. Elec. Engrs. (London) 94, Pt. 111, 171 (1947).
71. Seeley, J. S., Proc. Znst. Elec. Engrs. (London) 106, Pt. C, 18 (1958).
72. Ingram, D. J. E., “Spectroscopy at Radio & Microwave Frequencies.” Butter-
worths, London, 1955.
73. Hogan, C. L., Proc. Z.R.E. 44, 1345 (1956).
152 JOHN BROWN

74. Van Vleck, J. H., Proc. Z.R.E. 44, 1248 (1956).


75. Convention on Ferrites. PTOC. Inst. Elect. Engrs. (London)104,Pt. B, Suppl. &7
(1957).
7 6 . Ferrites issue. Proc. I.R.E. 44, 10, 1233-1468. (October, 1956).
77. Rau, R. R., and Caspari, M. E., Phys. Rev. 100,632 (1955).
78. Gunn, J. B., and Hogarth, C. A., J . A p p l . Phys. 26, 353 (1955).
79. Carrara, N., and Lombardini, P., Nature 143, 171 (1949).
80. Cullen, A. L., PTOC.Inst. Elec. Engrs. (London)99, Pt. IV, 100 and 112 (1952).
81. Cullen, A. L., and Stephenson, I. M., PTOC. Inst. Elec. Engrs. (London)99, Pt.IV,
294 (1952).
82. Cullen, A. L., and French, 13. A , , Proc. Inst. Elec. Engrs. (London)104, Pt. C,
456 (1957).

Das könnte Ihnen auch gefallen