Professional Documents
Culture Documents
(Methods in Experimental Physics 26) Daniel Malacara (Eds.) - Physical Optics and Light Measurements-Academic Press (1989) PDF
(Methods in Experimental Physics 26) Daniel Malacara (Eds.) - Physical Optics and Light Measurements-Academic Press (1989) PDF
Experimental Physics
VOLUME 26
Founding Editors
L. MARTON
C. MARTON
Volume 26
Daniel Malacara
Centro de lnvestigaciones en Optice
Leon, Gto
Mexico
. .
LIST OF CONTRIBUTORS . . . . . . . . . . . . . . . . . ix
PREFACE. . . . . . . . . . . . . . . . . . . . . . . . . . xi
...
LIST OF VOLUMESI N TREATISE . . . . . . . . . . . . . . . xi11
1 . Interference
by D A N I E LMALACARA
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . 1
1.2. Two-Beam Interferometers . . . . . . . . . . . . . . 7
1.3. Multiple-Beam Interferometers . . . . . . . . . . . . . 25
1.4. Multiple-Pass Interferometers . . . . . . . . . . . . . 37
1.5. Applications of Interferometry . . . . . . . . . . . . . 38
References . . . . . . . . . . . . . . . . . . . . . 45
3. Optical Polarization
by FREDERIC R . STAUFFER
4 . Holography
by R. D. BAHUGUNA
A N D D . MALACARA
6. Detectors
by T. 0. POEHLER
ix
This Page Intentionally Left Blank
PREFACE
Two books covering the field of modern optics have been prepared in
this series “Methods of Experimental Physics”, separating the material into
two parts, one with the title “Geometrical and Instrumental Optics”, and
the other with the title “Physical Optics and Light Measurements”.
The purpose of these books is to help the scientist or engineer who is
not a specialist in optics to understand the main principles involved in
optical instrumentation and experimental optics.
Our main intent is to provide the reader with some of the interdisciplinary
understanding that is so essential in modern instrument design, develop-
ment, and manufacture. Coherent optical processing and holography are
also considered, since they play a very important role in contemporary
optical instrumentation. Radiometry, detectors, and charge coupled imaging
devices are also described in these volumes, because of their great practical
importance in modern optics. Basic and theoretical optics, like laser physics,
nonlinear optics and spectroscopy are not described, however, because they
are not normally considered relevant to optical instrumentation.
In this volume, “Physical Optics and Light Measurements”, Chapter One
describes the theory and applications of interference and interferometers.
Chapter Two studies diffraction, its basic theoretical fundamentals, and
some practical applications. Polarized light and its uses are considered in
Chapter Three. Holography and holographic methods are studied in detail
in Chapter Four. The photometric and radiometric principles are covered
in Chapter Five. Finally, Chapter Six considers detectors.
There might be some overlapping of topics covered in different chapters,
but this is desirable, since the points of view of different authors, treating
different subjects, may be quite instructive and useful for a better under-
standing of the material.
This book has been the result of the efforts of many people. Professor
H. W. Palmer started this project and spent many fruitful hours on it.
Unfortunately, he did not have the time to finish his editorial work due to
previous important commitments. I would like to express my great appreci-
ation of and thanks to Professor Palmer and all of the authors, without
whom this book could never have been finished. I also thank Dr. R. E.
Hopkins and many friends and colleagues for their help and encouragement.
Finally, I appreciate the great understanding of my family, mainly my wife
Isabel, for the many hours taken away from them during the preparation
of these books.
DANIEL MALACARA
Leon, Gto. Mexico.
This Page Intentionally Left Blank
METH0DS OF
EXPERIMENTAL PHYSICS
Editors-in-Chief
Robert Celotta and Judah Levine
Daniel Malacara
Centro de lnvestigaciones en Optica A.C.
Apdo. Postal 948
37000 Leon, Gto. Mexico.
1 .l.Introduction
screen
1.0--
.8-
(b)
6-
.4 -
.2-
FIG. 1. Young's experiment. (a) Experimental arrangement and (b) Diffraction pattern.
&l--
4 ----
screen
s*
(a)
SO" rce
dipole antennas and radio waves. The total width of the central maximum
is given by sin 0 = A l l .
The Fresnel biprism and Lloyd's mirror also produce interference fringes
by division of the wave front, as shown in Fig. 2. If two waves are to
interfere producing fringes with good contrast, the polarization states of
both waves must be the same. This condition is always satisfied in the
Fresnel biprism. However, in Lloyd's system, only one beam is reflected.
Therefore, the reflection coefficients and the phase shifts under reflection
must remain nearly constant over the range of incident angles used. This
is possible only near grazing incidence.
4 INTERFERENCE
A second reason for using grazing incidence in the Lloyd system is that
the spacing between fringes decreases rapidly as the separation between
the virtual sources S, and S2 is increased. The sources must be quite close
to each other to make fringes visible, and this is possible only near grazing
incidence. If the screen in the Lloyd system is placed near the edge of the
mirror, we can observe that a dark fringe appears at this edge. This happens
because there is a phase shift upon reflection with grazing incidence. Wolfe
and Eisen" have studied the coherence requirements in Lloyd's mirrors.
Division of Amplitude. This class of interference occurs when both
interfering beams are obtained by division of the amplitude of the original
wave front by means of a partially reflecting optical surface. Then, both
beams travel different paths, and interference occurs when they are recom-
bined. Typical examples are Newton rings and the Michelson interferometer
described in the next section.
observing eye
observed fringes
some of the light is reflected. Some of the light that goes through the first
lens is reflected from the upper flat surface of the second lens. The two
reflected beams interfere constructively when the phase difference is an
integral multiple of 27~.We can see that the optical path difference is twice
the surface separation. If one of the surfaces is spherical, each fringe is a
ring that represents the locus of points with equal surface separation. These
circular fringes, also called Newton rings, are a particular case of fringes
of equal thickness.
Fringes of Equal Inclination. I n this case, each fringe is the locus of
points in the field of view with the same angle of the incidence 8 at the
interferometer. As an example, let us consider two parallel reflecting surfaces
as in Fig. 4. As in the previous example, the two interfering beams are
produced by division of amplitude. Fringes of equal thickness cannot be
formed because the optical path difference (OPD) is the same for the entire
field. One way to change the phase difference and hence to observe fringes
is to introduce a range of angles of incidence by using an extended light
source.
Circular fringes are observed with an angular radius 8 given by
mA
cos 8 =-)
2d
where rn is an integer smaller than or equal to 2 d / A .
To directly observe these fringes, the lens and the screen may be replaced
by the unaided eye (focused at infinity). Then, the observation must be
made almost perpendicularly to the optical surfaces, and the eye must be
placed very close to them, because of the small diameter of the pupil of
the eye. If the observing distance d between the surfaces is so large that
the fringes cannot be resolved with the naked eye, a telescope must be used.
monochromatic source
of light
4
in contact
where m is an integer, d is the thickness of the gap between the two reflecting
surfaces, and n is the refractive index of the gap medium. This expression
assumes that the fringes are observed normally to the reflecting surfaces,
and hence the two following rules should be followed to obtain good
accuracy:
(a) If the surfaces are flat, a collimating lens is required in order to
observe the fringes from the focus ofthis lens. Another acceptable alternative
is to use a large extended source of light and then to observe the fringes
from a minimum distance of about five times the diameter of the flat surface
under test.
(b) If the surfaces are spherical, the observation point should effectively
be placed at the center of the curvature, as shown in Fig. 6.
Figure 7 shows some typical defects that may be observed with this
interferometer.
where d is the thickness of the gap, and n is its refractive index. If the field
is free of fringes, we infer that the product nd is constant over the gap. We
can conclude that d is a constant only if n is assumed to be constant. If
straight and parallel fringes with separation S are observed, the two faces
forming the gap are flat and with an angle 8 in radians between them, given
by
A
e=-.
2 nS
This makes this interferometer quite useful for measuring small wedges
in glass plates.
TWO- BEAM INTERFEROMETERS 9
observation
k point
I
surface (a 1
under t e s t
observat Ion
polnt
k
are circles with angular radii given by Eq. (1.2), where d is given by
d = -OPD,
2 ’
and OPD, is the optical path differences along the optical axis.
TWO-BEAM INTERFEROMETERS 11
light source t- c
The interfering beam coming from mirror MI traverses the beam splitter
three times, while the beam from mirror M 2 passes through the plane only
once. To equalize the thickness of glass that each beam traverses, a plate
P2 of the same thickness and orientation as P, is inserted into the beam to
M 2 . This extra plate is called a compensating plate, and the interferometer
containing one is said to be compensated.
When the interferometry is not compensated, only one beam goes through
the inclined beam splitter, and thus the optical path difference for a given
value of 8 is different in different planes. In that case, the Haidinger fringes
are not circular but elliptical in shape, as described by Guild.I3
light
source
4observed
pattern
F i c . 9. Michelson interferometer.
12 INTERFERENCE
The fringes have good contrast only if the OPD, is smaller than the
coherence length. Therefore, the fringes are observed with white light only
if the variation of the OPD within the spectral region being observed
becomes smaller than about A/4. Since the index of refraction is a function
of the wavelength, the former condition cannot be satisfied unless the total
amount of the glass traversed by both beams is the same, and then the
OPD, is adjusted to be smaller than a few wavelengths. Hence, white-light
fringes can be observed only in a compensated interferometer.
If the mirrors MI and M 2 are not perpendicular to their optical axes,
equal-thickness fringes will be observed instead of equal inclination fringes.
However, they are exactly equal-thickness fringes only if the observing eye
is very far from the interferometer, or a lens is used as in Fig. 9 to optically
place the eye at an infinite distance from the interferometer.
If the eye is close to the interferometer and if the mirrors are not
perpendicular to their optical axes, intermediate fringe shapes are observed.
These fringes are curved, with the convexity directed toward the narrower
part of the wedge formed by the two mirror images. The three kinds of
Michelson fringes are shown in Fig. 10. The extended light source permits
the fringes to be observed only at a well-defined plane near the virtual
images of the light source. These fringes are said to be “localized.”
r-i
I 1
I 1
sample 1 I
t
I 1
The Jamin interferometer is shown in Fig. 11. The light from the light
source is split at the surfaces of a thick glass plate. Then, the two beams
are recombined at a second identical glass plate. For some applications,
the advantage of this arrangement over the Michelson interferometer is that
the light passes through the sample only once. The plates have a large
thickness so that enough room is left for the sample. This interferometer
was used for refractometry, but it has now been replaced by other kinds of
interferometers.
The Mach-Zehnder interferometer, shown in Fig. 12, can be considered
as a Jamin interferometer in which the four reflecting surfaces have been
replaced by four plates. This instrument is generally used to detect small
changes or space variations in the refractive index of the samples.
Both, the Jamin and the Mach-Zehnder interferometers, are compensated
if correctly adjusted, so that white-light sources can be used if desired.
However, it must be pointed out that this adjustment is particularly difficult
in the case of the Mach-Zehnder.
The Twyman-Green interferometer14*" is one of the most useful
modifications of the Michelson interferometer, used mainly to test the
quality of optical components. The system is illuminated with a collimated
14 INTERFERENCE
,attern 2
9 -
p2 pattern I
2
+
3
\
Q
beam of light, as shown in Fig. 13. The fringes that are observed by placing
the eye at the focus of lens L are of equal-thickness.
The Twyman-Green interferometer is not generally compensated, and
large values of the OPD are frequently obtained, so that the light source
Y
F I G . 13. Twyman-Green interferometer.
TWO-BEAM INTERFEROMETERS 15
--(C)
If the plate faces are flat but not parallel and n is constant, straight and
parallel fringes appear.
Using the arrangements in Figs. 14(b) and 14(c), the quality of a lens or
photographic objective can be tested. It is assumed in this test that the
convex or concave mirrors are perfectly spherical. If the lenses are assumed
to be perfect, the quality of the spherical mirror is tested. Interference
patterns associated with various lens aberrations are shown in Fig. 15. Many
other types of optical elements, such as prisms and diffraction gratings, can
also be tested with this interferometer.
I ate ra I radial
front
sheared
wavef ront s
microscope
objective
gas LASER
rl
lens
under test
FIG. 17. Two lateral shearing interferometers.
wave fronts as shown in Fig. 17(a). The lens under test is placed between
the light source and the interferometer.
Murty” has proposed a simpler version of a lateral shearing interferometer
in which the large temporal coherence of laser light permits large values
for the OPD [see Fig. 17(b)].
Lateral shearing interferometers provide rapid and easy identification of
lens aberrations as shown in Fig. 18. Given an interferogram, the wave front
that produced it can be computed by using several methods, such as those
described by Saunders,” M a l a ~ a r a , ’and
~ Rimmer and W ~ a n t . * ~
Many radial shear interferometers have been designed, but the first one
was invented by Brown.25Basically, this is a Jamin interferometer that works
in convergent light and has a small meniscus lens in one of the beams. To
compensate the interferometer, a small parallel plate is placed in the other
beam [see Fig. 19(a)]. Another interesting radial shear interferometer, shown
in Fig. 19(b), was designed by M ~ r t y . ~ ~
Radial shear interferometers are sensitive to all types of aberrations, as
the Twyman-Green interferometers, but to a lesser degree.
TWO-REAM I N T E R F E R O M E T E R S 19
RADIAL L I SHEARED
WAVEFRON rs
BROWN'S INTERFEROMETER
WAVEFRON r
UNDER T E S r
I . ."& .^CC."
EYlSPHERlCAL
AM SPLITTER
MURTY'S INTERFEROMETER
FIG. 21. Koester's prism interferometers. (a) Gates interferometer and ( b ) Saunders
interferometer.
22 INTERFERENCE
call DF, and an undiffracted one (zero order), which we call DO. The
undiffracted beam DO covers only a small area near the center of the surface
under test, while the diffracted beam covers the whole surface. The first
scatter plate is imaged onto the second plate, which is a mirror image of
the first. The beam DO produces two beams upon passing through the second
plate, one again undiffracted (DOO) and one diffracted for the first time
(DOF). The beam D F also originates two beams, one undiffracted (DFO)
and one diffracted a second time (DFF). The beam DO0 is observed as a
small bright spot near the center of the mirror. The beam D F F is very faint
and unobserved. The beams DOF and DFO form a fringe pattern. The pattern
is sometimes interpreted as a radial shear interferogram with a very large
amount of radial shear; however, it is best to think of it as a Twyman-Green
pattern.
There i s a large variation in common-path interferometers, many of them
using birefringent materials, as described by FranGon’ and Mallick.”’
fringes
FIcj. 23. Rayleigh interferometer.
f l a t mirrors
s c r e e n or
d photographic
plate
--La"
'IrnI"
flat mlrrors
The light received from opposite sides a and b of the star disc has an
angular separation 8. The optical path difference from point a assumed to
be on the axis is represented by OPD,. Thus, the optical path difference
from point b, off-axis, is given by
OPD,, = OPD, - ed, (1.8)
where d is the separation between the two small mirrors receiving the light
from the star. The irradiances produced by the two sources at points a and
b can be obtained from Eq. (1.1) as
I, =21[1 +cos(koOPD,)],
and
Ih = 21[ 1 + COS( koOPDh)]. (1.9)
Since the two point sources are not coherent with respect to each other,
the total irradiance on the pattern is
1, = 1, +Ib
=41 + ~ I [ c okoOPD,)
s( + cos ko(OPD, - Od )]. (1.10)
The fringe visibility versus the slit separation is plotted in Fig. 25. The
angular diameter of the star may be calculated from the measurement of
the distance d, which produces the first minimum in the visibility. As an
example, the star Betelgeuse has an angular diameter of about 0.05 sec of
arc, which lets d equal about three meters for green light.
A modern version of the Michelson stellar interferometer is the irradiance
interferometer in which the light from each slit is not made to interfere.
Instead two irradiance detectors are used, and then the irradiance variations
in the two signals are electronically correlated. This instrument has a
t" t"
Sl s2
FIG. 26. Multiple reflection in a Fdbry-Perot interferometer.
26 INTERFERENCE
(b)
FIG. 27. Fabry-Perot configurations.
The two parallel surfaces may be in two plates as in Fig. 27(a), or in one
plate as in Fig. 27(b). Such plates are sometimes called etalons, from the
French word for "standard."
The transmitted irradiance 1 ( 6 ) as a function of the phase difference,
assuming the amplitude transmission and reflection coefficients in Fig. 26,
may be given by
(1.11)
where it has been assumed that rl and rz are equal, in other words, that
the phase shift upon reflection is either 0" or 180". Here, I is the incident
irradiance, the asterisk denotes the complex conjugate, and the phase
difference is given by
477
6 =- nd cos 0 2 , (1.12)
A
( 1 . 1 1 ) reduces to
T , T2
I ( 6 ) = I” (1.13)
6’
( 1 - R,)2+4R, sin2-
2
as shown by Ba~meister,~’where the average reflectance R, is defined as
(R~R~)~’~.
This expression can be generalized to include the case in which there are
phase shifts E , and E~ upon reflection as measured in the medium of the
spacer (shown by Baumeister and Jenkins3’) by writing Eq. (1.12) as
47T
6 = - nd cos O2 -$(
A
+ E ~ ) . (1.14)
(1.16)
As shown in Fig. 28, the finesse can be extremely large even for relatively
small values of R,.
(1.19)
Two rings with different wavelengths may have the same angular
diameters, because their value of m may also be different. However, this
ambiguity is eliminated by measuring two sets of rings with different values
of d.
Fringes of equal thickness may also be formed with a Fabry-Perot
interferometer by forming a small wedge with the two reflecting surfaces.
In this case, an extended source is not needed, and the light is normally
collimated by a lens. The instrument may now be called a multiple-reflection
Fizeau interferometer.
Fringes of equal chromatic order appear when a white light source is
used to illuminate the interferometer adjusted with parallel faces, and then
the outgoing light is seen through a spectrograph. Narrow bright bands are
seen in the spectrum. This effect is called a channeled spectrum, and the
fringes are called fringes of equal chromatic order or Edser- Butler fringes.
Tomkins and Fred39 have used these fringes to calibrate a spectrograph
by using the fact that they are located at positions A N = 2 n d / ( N - 4 / 2 ~ ) ,
MULTIPLE-BEAM INTERFEROMETERS 29
where N is an integer, and C#I is the total phase change on the two internal
reflections.
T ~ l a n s k y ~ has
~ . ~also
* ~used
' these fringes to measure the microtopography
of crystal surfaces and films.
/ \
We should also notice that the emerging rays are all coincident along the
incident ray and not the parallel and laterally separated rays as in the
Fabry-Perot interferometer. To keep the rays within the limits of the paraxial
optics, two small circular diaphragms are frequently used in front of the
mirrors.
The only adjustment that this interferometer needs is to fix the correct
distance between the mirrors with a tolerance of a few microns. Since the
mirrors are spherical, there is no need to adjust their parallelism.
The main use of this instrument is to examine the spectrum of lasers, as
described by Herriott4' and Fork et al.,44by oscillating one of the mirrors
a small distance (usually about h / 4 by means of an electromechanical
device).
2nd cos e
A= (1.22)
m
-glass
m
-c-i-elatl coating
dielectric spacer
-glass
ANGLE OF INCIDENCE
( 1.23)
where A, and A. are the wavelengths for an angle of incidence 4 and for
normal incidence, respectively. Figure 32 shows a plot of (Ao- & ) / A o versus
4. A consequence of this angle sensitivity is that the narrower the passband,
the more accurate the angle of incidence and collimation have to be.
Half-widths of interference filters can have a wide range of values, between
about 80 A and about 1 A. In order to reduce the passband width (or
half-width), interference filters are often made with more than one cavity,
that is, with several spacers with reflecting films made of dielectric multi-
layers instead of metal.
surface
surface 2
I
I
V/II
I coating Y ' ' H
I
+ @ N 2 I f d
glass
substrate fl @ ,& N,
,reflectance
uncoated substrate
FIG. 34. Change in the reflectance of a single antireflecting layer with the optical
thickness nd.
MULTIPLE-BEAM I N T E R F E R O M E T E R S 35
0 600 700nm
’0° wavelength
.3c
.2c
-
0
-"
-
.I 0
( I I I
4 1 so0 600 700nrn
wavelength
I reflectance
l M 3
These are interferometers in which at least one of the wave fronts traverses
the normal trajectories more than once. The instruments, which have been
described in detail by H a r i h a ~ - a nand
~ ~ by Langenbe~k,~’
have some advan-
tages, as we shall see in this section.
A double-pass interferometer is designed with the purpose of separating
the symmetrical and the antisymmetrical parts of the wave aberration and
to display them in separate interferograms. Figure 39 shows the double-pass
Twyman-Green interferometer described by Hariharan and Sen.” The two
beams going out from the interferometer are reflected back to it by a small
mirror M 3 . Then, the patterns are observed by means of the beam splitter
S 2 , where there are four wave fronts. By moving the light source slightly
off-axis, two different patterns are observed. One pattern represents the
symmetrical component and the other one the antisymmetrical component
of the wavefront.
Double-pass Fizeau interferometers have also been designed by Sen and
Pu~~tambekar’”’~ and by Puntambekar and Sen” in order to reduce the
coherence requirements of the light source.
The multiple-pass principle can also be used in order to increase the
sensitivity of an interferometer. Two examples are the multiple-pass
Twyman-Green and Fizeau4’ interferometers shown in Fig. 40.
38 INTERFERENCE
aux i lio r y be a rn
divider
test
surf ace
coated mirror
surface
1.5.3. Refractometry
Interferometry can also be used to measure very small changes in the
refractive index of a gas or to compare the refractive indices of two liquids.
As mentioned before, Michelson3' applied interference for the first time for
this purpose. The most commonly used instrument for this application is
the Rayleigh interferometer4 described in Section 1.2.7. A Jamin or a
Fabry-Perot interferometer can also be used to measure refractive
indices.
shown by hop kin^.^' Some lateral shearing interferometers made for this
purpose are described by Fran~on.'
I I
second step was to measure the wavelength of the laser line by inter-
ferometric comparison with the Krypton line. Next, the vacuum speed of
light was computed from the product of the wavelength and the frequency,
and this result, equal to c = 299,792,498 meters/sec. was defined to be exact.
In conclusion, the definition of the meter fixes the speed of light, and any
source of known frequency may become a reference standard for length.
An obvious application of these techniques is the calibration of scales in
many instruments, for example in the fabrication of diff raction gratings.73
Interferometry using amplitude-modulated beams can be used to measure
longer distances. It is like having a much longer wavelength, but the accuracy
of the measurements can be much greater than that obtained by other
methods. Instruments using this principle in order to measure long distances
of the order of several meters have been made by Dukes and Gordon.74 A
good description of these instruments can be found in a paper by Bruning.68
Even the distance from the earth to the moon has been measured with this
principle by H a m m ~ n d . ~ '
as well as l a ~ - g e ~angles
~ * ~ ' can also be measured very accurately
by interferometric procedures.
1.5.8. Microscopy
The principle of multiple-beam interference in thin films can be used
with great advantage to study the microtopography of small crystal or glass
surfaces. This subject has been fully described by Tolansky4' and by Krug
et al." One of the many applications of this method is the observation of
the polishing defects of a metal or glass surface.
MONOCROMATIC
COLIMATED
LIGHT SOURCE
OBSE E R OR
LIGH' ETECTOR
one moves clockwise, and the other one counterclockwise. The result is that
there is a fringe shift proportional in magnitude to the angular rotation w ,
as follows:
4Aw
AN=- ( 1.28)
CA '
1.5.10. Spectroscopy
Probably one of the most popular and oldest applications of inter-
ferometers is the measurement of the spectral components of light beams,
as was already mentioned in Section 1.3. The resolution of interferometric
methods is so good that the hyperfine structure46and the wavelength values3'
can be measured with high accuracy.
Another very important and relatively new spectroscopic method that we
want to describe here is called Michelson-Fourier spectroscopy. This
method has been treated by P. Connes'" and Vanasse and Strong." It works
on the principle that the temporal coherence or wave-train shape determines
the frequency spectrum and vice versa. A Twyman-Green interferometer
can be used to measure the wave-train shape, and then the spectrum can
be computed mathematically. Suppose that we want to determine the spec-
trum of the light source illuminating the interferometer in Fig. 13. This is
done by moving one of the flat mirrors along its optical axis to change the
OPD. A light detector then replaces the eye in order to permit the average
irradiance of the interference pattern to be recorded. The irradiance
measured by the detector is a function of the optical path difference.
Defining (1/2)Zl(k)dk as the average irradiance due to either of the two
beams on a narrow band dk centered at k, the irradiance d l at the detector
will vary sinusoidally with the OPD as follows:
d l = 11(k ) [1+ COS( kOPD)] dk. ( 1.29)
Since different wavelength light beams are mutually incoherent, the irradi-
ance I due to all colors acting simultaneously is
I= lom
I,(k)[l+cos(kOPD)] dk. ( 1.30)
If we now define
lom
I,= I,(k) dk, (1.31)
and
I(OPD)=
I,' I , ( k ) c o s ( k . O P D ) dk, (1.32)
gives
Z(0PD) C O S ( ~ OPD)
- d(0PD). ( 1.34)
where the value of OPD,,, is twice the maximum mirror displacement. The
larger this OPD,,, , the better the spectral resolving power.
There are two clear advantages of interference spectroscopy over conven-
tional dispersing spectroscopy. The first advantage is called Jucquinot and
Dufour” or etendue advantage. The spectroscopic field of view of an
interferometer is circular and much greater than the narrow and long field
given by the slit of a spectrometer. For instance, with a Fabry-Perot
interferometer, the spectrum of the whole image of the sun can be simul-
taneously studied, while with a spectroheliograph, this has to be done by
dividing the image of the sun into slits.
The second advantage is called Fellgett9’ or multiplex advantage. In a
spectroscope, different wavelengths are studied either at different times if
a photoelectric detector is used or with different photographic regions if a
photographic plate is used. In interference spectroscopy, on the other hand,
all spectral lines are measured simultaneously with a single photoelectric
detector. This is equivalent to coding each optical frequency by an electrical
cosine signal with different frequency, and then decoding numerically by
Fourier analysis. In this manner, the signal-to-noise ratio is much greater
than in conventional spectroscopy.
We could describe many other useful applications of interferometry, but
it would be impossible to mention all of them in a single chapter.
Acknowledgment
The author wishes to express his gratitude to Professor W. H. Steel for his many valuable
suggestions.
References
1. M . Born and E. Wolf, Principles ofoptics, Pergamon Press, New York, 1959.
2. A. H . Cook, Interference of Electromagnetic Waves, Clarendon Press, Oxford, 1971.
3. M . Fraqon, Optical Interferometry, Academic Press, New York, 1966.
4. C. Candler, Modern Interferometers, Hilger and Watts, London, 1951.
5. W. H. Steel, Inferferomerry, 2nd ed., Cambridge University Press, London, 1983.
46 INTERFERENCE
Daniel Malacara
Centro de lnvestigaciones en Optica. A.C
Apdo. Postal 948
37000 Leon, Gto. Mexico
2.1. Diffraction
Diffraction phenomena have been some of the most interesting and most
thoroughly studied manifestations of light in the history of physics. Many
good textbooks present this subject in detail like those by Born and Wolf,’
Meyer,2 Marechal and F r a n ~ o n ,and
~ Ra~leigh.~ The
, ~ earliest
known mention of this phenomenon appears in the works of Leonard0 da
Vinci (1452-1519).However, Grimaldi7 is the first one to seriously treat the
subject in his book. It is interesting to know that Newton* was aware of
Grimaldi’s observations.
Huygens’ was the first one to develop a wave theory (see Fig. l), although
he also seems to have been aware of Grimaldi’s observations, since he did
not attempt to explain the existence of diffraction bands. An attempt to
explain diffraction qualitatively was made by Young, 1 0 * ’ 1 - 1 2 who assumed
that the wave passes undisturbed through the aperture, except at the edges,
where the edge wave is formed. In this case, he thought that the interference
between the direct wave and the edge waves formed the diffraction pattern.
It was not until 1818 that FresnelI3 used the wave theory together with the
principle of interference to explain diffraction quantitatively. He assumed
mutual interference between all secondary waves, adding on the screen the
amplitudes of each Huygens wavelet, and taking into account any phase
differences.
The Huygens principle, modified in this way, is called the Huygens-
Fresnel principle; it leads to quantitatively good predictions of the light
distribution on the screen. This theory, nevertheless, is not perfect. One of
the problems is that there is no explanation as to why the wavelets do not
travel backwards to the source, and another is that the resultant phase on
the screen, computed by this theory, is 7r/2 less than the experimental
value.
49
M E T H O D S O F E X P E R I M E N T A L PHYSICS Copyright 0 1988 by Academic Press. Inc.
Vol. 26 All rights of reproduction in any form reserved
ISBN 0- 12-475971-8
50 DIFFRACTION A N D SCATTERING
and portion C, a spherical shell which intersects the diffracting screen and
which has its center at the observation point P.
The amplitude over the diffracting screen must be zero. Therefore the
integral over portion B must also be zero. The integration over the sphere
C also goes to zero when its radius goes to infinity.
If the angle between the incident and the diffracted rays is called 8, it
may be shown that:
cos( n, s) + cos( n, r ) = 1 + cos 8. (2.2)
Hence, Eq. (2.1) becomes
(2.3)
rs
which is the well-known Kirchhoff diffraction integral, where u is the
diffracting aperture.
We notice here a few differences between the results of this theory and
those of Fresnel. The first important difference is the presence of the complex
number i in front of the integral. This means that the phase of the observation
point P, obtained with the Kirchhoff theory, has a difference 7r/2 with
respect to the phase obtained with the Huygens-Fresnel theory. This also
means that the phase of the illumination at point P, using a very small
diffraction pinhole (only one Huygens wavelet), and the phase of the
illumination at the same point with no diffraction, differ by 7r/2.
52 DIFFRACTION A N D SCATTERING
Second, we have the presence of the factor (1 + cos 0)/2. Usually this is
called the inclination factor, first derived by Stokes. It implies that the
amplitude of each secondary wave decreases when the inclination angle
increases. This inclination factor appeared naturally in Kirchhoff’s theory
and is only postulated in the Huygens-Fresnel theory.
The third difference is the presence of the product l / r s in the integral in
Eq. (2.3) which would appear in the Huygens-Fresnel theory only when
the decrease of the amplitude, as l / r s , is considered. Experimental
verifications of Kirchhoff’s scalar theory of diffraction were made by Silver”
in 1962.
Even Kirchhoff’s theory is incomplete, not only because of the approxima-
tions made, but mainly because the vectorial transverse nature of light waves
is ignored.
The scalar theory of diffraction produces, in general, very accurate results,
but a further refinement is achieved if the vectorial nature of light20.2’.22.23324.2s
and the optical properties of the screen or stops are considered by means
of the electromagnetic theory.
separated into two waves, one coming from the aperture and another one
coming from its edge. In other words, they showed that a diffraction wave
appears as arising from the scattering at the edge of the aperture. Descrip-
tions of the Miyamoto-Wolf theory can be found in R ~ b i n o w i c z . ~ ~ . ~ ~
Even a corpuscular quantum approach to the diffraction problem in
gratings has been given, namely by Duane.40
Depending on the positions of the light source and the observation plane,
we may consider two types of diffraction. Fresnel diffraction occurs when
either the light source or the observation plane or both are at finite distances
from the diffracting aperture or object. Fraunhofer diffraction occurs when
both the light source and the observation plane or both are at finite distances
from the diffracting aperture. In the following sections, we will consider
some selected configurations for Fresnel diffraction. Figure 3 shows some
interesting Fresnel diffraction patterns.
c - c
a l b
\
\
\
\
\
Assuming that the distances a and b are very large compared to the
height S, the segment QR may be approximated by the sum of the sagittas
of the two arcs passing through Q and R, respectively.
S 2 S2 a + b
OpD=QR=-+-=-
2a 2b 2ab
s2, (2.4)
S=kOPD=
n( a +b ) S2.
(2.5)
abh
To simplify the mathematics, we consider this as a two-dimensional
problem. Defining a nondimensional variable u as:
(2.7)
If we divide the wave front into narrow parallel strips of width ds, we
may use the Huygens-Fresnel principle and add the contribution of all the
strips to the amplitude at point P. This sum is made by considering the
FRESNEL DIFFRACTION 55
phase differences between the different strips and by adding them vectorially
as shown in Fig. 5.
Each vector, representing the contribution of each slit, has a magnitude
directly proportional to the width ds and hence to dv. Each vector forms
an angle 6 with the x axis. The resultant amplitude at the point P is directly
proportional to the resultant vector R.
We can mathematically generate a curve, as in Fig. 5, that can be used
to find the amplitude at point P for any width and position of the diffracting
slit. Using Eq. (2.7), we find:
x= lo" ($)
cos dv, (2.10)
y= lo" ($)
sin dv. (2.11)
56 DIFFRACTION A N D SCAITERING
f q A2
Equations (2.10) and (2.1 1 ) are the Fresnel integrals, whose numerical values
are given in table form in many book^.^'.^^
The curve produced with these integrals is the Cornu spiral shown in
Fig. 6.
Strictly speaking, we should have used the inclination factor (1 + cos 8 ) / 2
here, since the amplitude of each secondary wavelet decreases when the
angle 8 increases. We assumed, however, that 8 remains very small
everywhere, and therefore the inclination factor is always very close to 1.
The Corm spiral allows us to find the amplitude of the illumination at
the point P on the axis. If the diffracting slit is centered on the optical axis
and has a width S, we first compute u by using Eq. (2.6). Then we choose
two symmetrical points A , to A, on the Cornu spiral, such that their
separation measured along the spiral be equal to u. The distance from A ,
and A * , measured along a straight line, represents the amplitude of the
illumination at P.
If the diffracting slit is decentered off the optical axis, we again use Eq.
(6.6) to compute the decentering and the slit width in terms of u. Then two
FRESNEL DIFFRACTION 57
points A; and A2 are chosen such that their positions as measured along
the curve corresponding to the slit width and to the decentering. Again, the
separation between these two points, measured along a straight line, rep-
resents the amplitude at P.
The whole diffraction pattern for a diffracting slit, with a width corre-
sponding to a certain value of u, is found by moving two points, A; and
A;, along the curve in such a way that the distance between these two points
along the curve remains constant and equal to u. The distance between the
two points in a straight line will change continuously to give the amplitude
of the whole diffraction pattern. The irradiance which is the square of the
amplitude is shown in Fig. 7(a). In order to relate the resultant irradiance
to the value of the unobstructed irradiance, the squares of amplitudes
obtained from Fig. 6 must be divided by 2.
If we have only straight edges, we may consider that we have a decentered
slit, with one edge on the optical axis and the other edge at infinity. In this
case, we have one fixed point 2, and a moving point A; on the Curnu
spiral. We then obtain the diffraction pattern in Fig. 7(b). If there is no
INTENSITY
I I
I r
GEOMETRICAL LIMITS
(A)
INTENSITY
G GEOMETRICAL EDGE
(B)
FIG.7. Diffraction light distribution in (a) a wide slit, (b) a straight edge.
58 DIFFRACTION A N D SCATTERING
K=
v(a +b ) (2.13)
abh .
The contribution of each zone is directly proportional to its area, and
this area in turn is directly proportional to its width ds. Taking this into
account, we can find a graph analogous to the Cornu spiral, which can be
easily shown to be given by:
[
x + y-- ;I*=[&l'~ (2.14)
This is a circle with its center on the y axis, tangent to the x axis, and with
radius A/2K. This means that, if the radius S of the diffracting circle is
increased continuously, the amplitude on the axis at the point P oscillates,
as shown in Fig. 8(a). This is physically unacceptable, because these oscilla-
tions in the amplitude do not occur when the diffracting aperture is very
large. The problem disappears when taking into account the inclination
factor (1 +cos 8)/2, because the radii of the circles will become smaller
when the aperture grows, increasing the angle 8. The curve is now a spiral,
as in Fig. 9. When the aperture becomes very large, the amplitude reaches
the value A/2K at the point P on the axis. The color at the center of the
diffraction pattern produced by a circular aperture and illuminated with
white light has been studied by Bergsten and H ~ b e r t y . ~ ~
F R E S N E L DIFFRACTION 59
AMPLITUDE AT
POINT P
-
A
2K
AMPLITUDE
AT POINT P
FIG.8. Amplitude variation on axis with a diffracting circle, (a) without obliquity factor,
( b ) with obliquity factor.
(2.15)
and
ab
s . =[ -
n'n a+b
h ( n - 1/2) (2.16)
Defining So= SIin a n d f = 2Si/h (for a zone plate with a transparent center
zone), we may show that
(2.17)
which is the equivalent of the thin-lens formula for the positions of the
object and the image, at a and b, respectively.
In spite of the close similarity between a thin lens and zone plate, there
are some important differences. The first difference for a lens is that the
focus is not unique, but there are an infinite number of them with focal
lengths fm, given by
2s;
fm =- m =integer, (2.18)
mh '
where rn is any positive or negative odd integer.
It should also be noticed that for every convergent focus there is a
corresponding diverging focus. In other words, a zone plate acts simul-
taneously as a convergent and a divergent lens. A very useful application
of zone plates is for alignment, as described by Van
A pinhole camera uses a small pinhole, instead of a lens, to form the
image. A consequence of Eq. (2.17) is that in order to optimize the resolution
and light-collecting power, the pinhole must have a radius S, and the
photographic-plate-to-pinholedistance should be f:
A Gabor plate is quite similar to a zone plate, with the difference being
that its transmittance does not change in steps but in a quasi-sinusoidal
manner. It is normally produced by the interference between a flat and a
spherical wave front.
62 D I F F R A C T I O N A N D SCATTERING
/- I/ I
-I+-POINT
COLLIMATOR I \‘I
DIFFRACTING
APERTURE
When the light source and the observing screen are both at infinite
distances from the diffracting aperture, we say that we are observing Fruun-
hofer difruction.
The light source may really be at infinity as in the case of a star, but
more frequently the light source is placed “optically at infinity” by means
of a collimating lens. In this case the light source is at the focus of a
convergent lens, as shown in Fig. 1 1 . The observing screen cannot be really
placed at infinity, but it is again done optically by placing it in the focal
plane of a convergent lens. The fact that the observing screen is at infinity
may be interpreted by saying that what we observe on the screen in Fraun-
hofer diffraction is the angular distribution of the light after the diffracting
aperture.
Figure 12 shows the slow transition from Fresnel to Fraunhofer diffrac-
tion, with a point source at infinity, by increasing the distance b of the
observing screen from the diffracting aperture. We say that we have reached
the far field or Fraunhofer diffraction region when b >> D‘/A, where D is
the diameter of the diffracting object.
The inclination factor is not important in the case of Fraunhofer diffrdc-
tion, and if we also take the case of normal incidence ( r = constant), the
F R A U N H O F E R DIFFRACTION A N D FOURIER T R A N S F O R M S 63
0
44
64 DIFFRACTION A N D SCATTERING
U(P)=i{ { AcdS,
S (2.19)
On the other hand, from Fourier theory, the function A(x, y ) can be given
by:
(2.24)
The functions A(x, y ) and U (k,, k,) are two-dimensional Fourier trans-
forms of each other.
Thus, except for unimportant constants, we can say that the angular
diffraction pattern is the Fourier transform of the amplitude function on
the diffracting aperture.
In a similar way, for diffraction apertures in one dimension, we may write
the one-dimensional Fourier transforms as
(2.25)
FRAUNHOFER DIFFRACTION A N D FOURIER TRANSFORMS 65
and
U ( k , )= A ['
J -0
elk\' dx, (2.27)
assuming that the amplitude A ( x ) remains constant inside the slit. Integrat-
ing, and using Eqs. ( 2 . 2 1 ) and ( 2 . 2 2 ) , we obtain:
U (k ) = U,
sin( k a sin 6)
k a sin 6 ' I (2.28)
A
sin 6 =-. (2.29)
2a
Therefore, the diffraction pattern becomes narrower when the slit width 2 a
increases.
For a rectangular aperture, with width 2 a and length 2b, we can use Eq.
( 2 . 2 3 ) to obtain
(B)
k o sin 6
FIG. 13. Fraunhofer diffraction pattern for single slit. ( a ) Amplitude distribution.
( h ) lrradiance distribution.
u(e)=2u0 J,(ka[
sin 0 )
ka sin 0 1’
where a is the radius of the aperture and J , ( x )is a first-order Bessel function
(2.31)
FRAUNHOFER DIFFRACTION A N D FOURIER TRANSFORMS 67
- AIRY DISC
-- ka sin 8
(6)
F l c . 14. Fraunhofer diffraction pattern for a circular aperture. ( a ) Amplitude aperture.
( b ) Irradiance distribution.
2.3.3.Babinet Principle
Two pupil functions, A l ( x , y ) and A 2 ( x , y ) are complementary to each
other if
A,(x,y)+A,(x,y)=constant. (2.34)
If the amplitude is constant over the whole x, y plane, then we say that
two diffracting apertures are complementary if the transparent parts in one
are the opaque parts in the other, and vice versa. For example, the com-
plementary aperture of a circular aperture is the whole x, y plane, excluding
only the opaque disc of the same size as the circular aperture.
Let us assume that we have two arbitrary complementary apertures, giving
diffraction pattern U , ( P ) and U , ( P ) . If U ( P ) is the amplitude of the
illumination produced without any diffraction apertures, the Babinet prin-
ciple’’ says that
(2.35)
Thus, except for the point at the origin, the Fraunhofer diffraction patterns
of complementary apertures are identical in shape but with amplitudes of
opposite sign.
(2.36)
70 D I F F R A C T I O N A N D SCATTERING
(2.40)
Irradiance is then given by
(2.41)
(2.42)
lim
S/2-mrr
[
sin'( y)
(I") (2.44)
I
I
I
I
I
I I I
I I I
A A
(2.47)
_A -
- mN. (2.50)
AA
This is the well-known expression for the resolving power of a grating.
If we define the effective width We, of a grating as
we,,= w cos 82 = Nd cos 8 2 (2.51)
and use Eq. (2.49) we find a n alternative expression showing that the
resolving power of a grating is limited only by its effective width and the
wavelength, as shown by Michelson,5s
(2.52)
DIFFRACTION GRATINGS 15
(2.53)
We have plotted this irradiance in Fig. 20(a) for a grating with five slits
of widths 2a = d/3.
A line with some order of interference does not appear when its direction
coincides with the direction of a zero of the second factor in Eq. (2.54).
The zeros of the second factor may be shown to occur at angles such that
2a(sin O1 -sin Oz) = nh, n = a n y integer excluding zero. (2.55)
The line positions are given by Eq. (2.38). Thus, the missing order numbers
are:
d
m = n--, (2.56)
2a
where n is any integer, excluding zero, such that m is also an integer. The
influence of the slit width is illustrated in Fig. 21, which shows the diffraction
patterns for two slits with different widths but with the same separation.
-6 -4 -2 I 2 4 6 m
0
(B)
FIG.20. lrradiance distribution in a grating with slits with finite width. ( a ) Finite-size grating.
( b ) Infinite-size grating.
There are some other types of spurious lines and ghosts that may
appear in a diffraction grating spectrum, but all of them are due to ruling
imperfections.
GROOVE
\
.
B L A Z E D GRATING I/
ECHELLE
GRATING
where h is the step height as shown in Fig. 23. The blaze angle is often
specified as the wavelength that fulfills this condition in the first order.
WoodG6pointed out that very efficient gratings for the infrared could be
obtained by using coarse-blazed gratings with this condition ( y = 0). Since
the light is reflected on the grooves' faces in a mirror like manner, he called
them echellettes, Harrison6736x considered that since resolving power depends
only on the total grating width, the echellette could be made with spacings
as large as about 100 grooves/mm. The typical blaze angle for these gratings,
named echelles by Harrison, is 63"26'. Echelles are very useful in order to
achieve high resolving powers in the infrared.
Gratings can also be concave and can be used in many configurations
that will not be described here. The reader is referred to the book by Sawyer69
for a more complete description.
The art of grating manufacture is a very complicated and interesting one.
The reader can consult the excellent papers by Harrison,67768S t r ~ n g , ~
and
"
Stroke7' for details.
In closing the subject of diffraction gratings, it should be mentioned that
there are also two- and three-dimensional gratings. Figure 24 shows the
diffraction patterns for two two-dimensional arrays of apertures.
FIG.24. Diffraction of a two-dimensional array of holes. (a) A regular array and (b) its
diffraction pattern. (c) An irregular array, and (d) its diffraction pattern.
Let us consider two very close stars, with their Airy discs overlapping in
the image plane. The light of the two stars is mutually incoherent, so that
there is no fixed phase relation between the two images. The resultant
combined image is then obtained by adding the intensities of the two images.
RESOLVING POWER O F OPTICAL INSTRUMENTS 81
-
RADIUS OF AIRY DISC
w=-
2T
(2.62)
L’
The frequency response curve of a lens described first by O’Nei1173 in
1954 is called the optical transferfunction (OTF) and is defined as the ratio
of the harmonic components in the image to those in the object. It has two
m
-wK
3
0
Lb
parts. The plot of the amplitude ratio versus spatial frequency is often called
the modulation transferfunction (MTF). The second part, the plot of the
phase ratio, is less frequently used.
The best way to measure the MTF is to form the image of an object that
contains all spatial frequencies in equal amounts. The plot of the amplitudes
of the spatial frequency components in the image thus formed would then
represent the frequency response of the lens.
An object that contains all spatial frequencies in equal amounts is a point
source, as may be proved without much difficulty. The image of this point
source is called point spreadfunction. The point spread function is the same
as the Fraunhofer diffraction pattern of the lens aperture only if the lens
is perfectly free of geometrical aberrations.
We shall assume a point source at infinity. The point spread function in
the focal plane of a lens with focal length F, with coordinate x F and y F
can be obtained from Eq. (2.23) if we first obtain the amplitude as follows:
A(XF, yF) =
II-: T ( ~y ), e ~ ( k / F ) ( r ~ f i + i ,)i ldx dy,
the wave front due to aberrations of the lens. The point spread function
S ( x F ,y F ) is then given by:
. [I -‘m
The variables x and y in these integrals do not remain after the integration
1
T * ( x ,y ) e - l ( k I ~ ) ( X X f i + Y v F ) d x d y . (2.65)
(2.66)
Once the point spread function is computed from this expression or
becomes known by some other means, the optical transfer function
F ( w , , w , , ) may be computed by using the Fourier transform of S ( x F ,y F )
86 DIFFRACTION A N D SCATTERING
as follows:
F(w,,w,) =
I:I s ( x F y, F )e - “ W \ X F + ’ > ” ,
obtaining thus:
\ \ INFOCUS-,
\ \
0 I I 1 I I I I dr
-.2 -
0 0.2 0.4 0.6 0.0
(i&P
FIG.30. Some optical transfer functions.
88 DIFFRACTION A N D SCATTERING
The appearance of the image is quite different if the two points are coherent.
This is, for example, the case if the object is illuminated by a single point
source or a laser.
Ernest Abbe74 developed a theory, usually called the Abbe theory of the
microscope, which allows the computation of the resolving power of a lens
when the object is illuminated with coherent light. His main assumption
was that the spatial frequency structure of the object to be imaged acts like
a diffraction grating when illuminated by a coherent light source, as shown
in Fig. 31.
If the object acts as a diffraction grating, the light will be diffracted into
many beams with different orders of diffraction. In this case the image may
be regarded as the interference pattern arising from the superposition of
all diffracted beams.
If the lens is so small that only the zero order of diffraction passes through
the lens, no interference takes place at the image plane, and therefore no
image is formed.
The object structure is first detected at the image plane when the lens is
large enough to let the two first orders of diffraction pass through the lens.
Under these conditions, the image is not an exact reproduction of the object,
because the orders of diffraction with magnitude greater than 1 are not
reaching the image plane. The image, in this case, has a sinusoidal profile
with a spatial frequency equal to that of the object (see Fig. 32).
When the lens is larger, so that the order of diffraction + 2 and -2 passes
through the lens, the image will more closely resemble the object, as shown
in Fig. 32.
T H E A B B E T H E O R Y O F T H E MICROSCOPE 89
The resolving power limit may be easily found by using the diffraction
grating equation, (2.38) for the first order rn and normal incidence:
d=- A (2.73)
sin 8’
(2.74)
Ar2
A6 2
of some dielectric material has been evaporated to produce the phase shift.
The illuminating system has a diaphragm with an annulus aperture and
a condenser. This illuminating system is designed so that all the nondiffrac-
ted light passes through the ring on the phase plate. The diffracted light
tends to pass out of the ring on the phase plate. Figure 35 shows the objected
observed in a normal and in a phase-contrast microscope.
(A) ( B)
FIG. 35. Pictures taken using a microscope. (a) Normal. (b) Phase contrast.
92 DIFFRACTION A N D SCATTERING
2.7. Scattering
Most of the light that our eyes receive comes not directly from the light
source but through scattering. Looking at almost any object, what we see
SCATTERING 93
is the light scattered from its surface. If we look at the blue sky or at clouds,
we also see scattered light. These examples give us an idea of the great
importance of scattering phenomena. Furthermore, the subject is very com-
plicated, and many good books have been written covering it, for example,
those by Van de Hulst” and Kerker.79
Scattering phenomena may be classified according to the size or shape
of the scattering object or to the wavelength. In general, scattering is an
anisotropic phenomenon, depending on the shape and orientation of the
94 DIFFRACTION A N D SCATTERING
Young's eriometer does not work for particles of varied sizes, because
the Fraunhofer diffraction pattern is formed from the irradiance sum of all
the individual patterns. Parrent and Thompson" and Silverman et aLX2
pointed out this problem and proposed to observe the pattern in the far
field (s >> n 2 / A ) of the individual particles but close enough to be in the
near field of the whole sample. Under those conditions, the diffraction
patterns of each particle will be in different places instead of all being
superimposed. It is interesting that the diffraction patterns are not Airy
patterns, because they may be interpreted as Fresnel patterns, or alterna-
tively, as Fraunhofer patterns with a coherent background.
If the object is very large compared with the wavelength, the mechanisms
producing scattering are diffraction on the edge of the object and refractions
and reflections inside it. These particles may be water drops, such as in fog,
rain, mist, and haze.
The rainbow is one example of scattering by water drops. This
phenomenon has been considered by poets as well as scientists. The scientific
description is often assumed to be an exercise in geometrical optics, but it
is much more complicated than that, as shown by Khare and Nussenzveigs3
and Van de H ~ l s t . ' ~
The best approximate theory to explain the rainbow was formulated by
Airy. An exact but complicated theory is obtained applying the general Mie
theory.
The glory is another beautiful atmospheric phenomenon due to scattering
by water drops, as described in the book by Mir~naert.'~ It is observed as
a rainbow surrounding the sun or moon, covered by a thin veil of clouds.
More often, persons standing on a high point, observe their shadows
projected onto low clouds, showing a colored ring around each of their
heads. The glory can be seen by each observer, with the ring only around
his or her own head.
acts as a driving force for the dipoles. With this force, the movement equation
would become
(2.79)
The choice of sign for the driving force is made by taking into account that
the electron feels the force in the direction opposite to E, because of its
negative charge -9. However, this equation is incomplete, because we know
that any accelerated charge radiates, and therefore the dipole charges will
emit an electromagnetic wave. Due to this radiation, the electrons experience
some kind of reaction force that opposes their movement. This reaction
force has been shown by Lorentz to be
F , = - . -2q2 d3y
(2.80)
3~x10’ dt3’
Thus, the complete movement equation is
- = - k y + - 2q2 --qEo
m -d- 2Y d3y
e-””. (2.81)
dt2 3cx lo7 dt3
The effect of the magnetic field of the driving wave is negligible compared
with that of the electric field and thus it is ignored. By proposing again a
solution of the type
Y = A e-‘”‘ (2.82)
SCATTERING 97
(2.84)
Therefore, we see that the dipole oscillates with a frequency equal to that
of the driving wave. In order to study the amplitude and phase of the dipole
oscillation, we shall consider three different cases. We assume that y w 3 / w i
is very small compared with ( w 2 - w ' ) .
(1) If w << w o , as is the case for most materials at frequencies in the visible
range, A is negative, and hence y is opposite in sign to E.
Thus, the phase of the dipole vibration is the same as that of the driving
wave, as shown in Fig. 39. We say that the electrons vibrate in phase with
the driving wave if they always move following the force they feel, that is,
opposite to the driving electric field.
(2) If w >> w o , as in the case of free electrons or most materials at X-ray
frequencies, A is positive. Thus, the dipole vibration is 180" out-of-phase
with respect to that of the driving wave, as shown in Fig. 39.
W<<Wo
w
0
a
I DIPOLE
I
1
FIG.40. Radiation pattern for a dipole.
q2w4 x 10-~
(S)= 8.rrcr2 - A: sin2 6, (2.85)
where 6 is the angle between the given direction and the dipole orientation.
The angular distribution of the dipole radiation is plotted in Fig. 40. The
complete radiation pattern would be obtained by revolving the pattern about
the dipole axis, thus obtaining a toroidal distribution.
The irradiance ( S ) is inversely proportional to the square of the distance
r from the dipole to the point under consideration. The symbol A represents
the maximum amplitude of oscillation of the dipole, as given in Eq. ( 2 . 3 8 ) ;
thus, if A,?,= A . A*,
q4E:x w 4 sin2 6
(S)= 8.rrm2cr2 . (2.86)
SCAITERING 99
(2.87)
The emitted wave has a phase difference kx with the electron vibration
only at distances from the dipole that are greater than the wavelength. At
very small distances from the dipole, the emitted wave has a phase shift of
180" with respect to the dipole vibration. The reason for this phase difference
at small distances is that at low frequencies, the charges in the dipole move
in a direction such that the electric field produced by the dipole tends to
cancel that of the driving wave.
4
r sin t? -1
I
I
I
SCATTERED
LIGHT
-
INCIDENT
LIGHT
where
(2.92)
(2.94)
FIG.42. Two pictures taken with (a) normal black-and-white film, (b) infrared film and filter.
102 DIFFRACTION A N D SCATTERING
Acknowledgment
The author wishes to thank Professor W. H. Steel for his very valuable comments.
References
1. M. Born and E. Wolf, Principles ofOptics, 5th ed., Pergamon Press, New York, 1975.
2. C. F. Meyer, The Diflraction of Light, X-Rays and Material Particles, 2nd ed., Edwards
Bros., Ann Arbor, Michigan, 1949.
3. R. W. Wood, Physical Optics, Macmillan, New York, 1936.
4. A. Marechal and M. Francon, Diflraction, Structure des Images, editions of the Revue
d’Oprique The‘orique et Instrumentale, Paris, 1960.
5. L. Rayleigh, Scientific Papers, Cambridge University Press, London, 1899.
6. L. Rayleigh, Phil. Mag. 41, 107 (1871); 41, 274 (1871); 41, 447 (1871).
7. F. M. Grimaldi, Physico-Mathesis de Lumine, Bologna, 1655.
8. 1. Newton, Opticks, London, 1704 (Reprint: Dover, New York, 1952).
9. C. Huygens, Traite‘ de la LumiPre, Leyden, 1960.
10. T. Young, Phil. Trans. Roy. Sor. 92, 12 and 387 (1802).
1 1. T. Young, A Course of Lectures on Natural Philosophy and Mechanical Arts, London, 1807.
REFERENCES 103
12. T. Young, Miscellaneous Works of the Late Thomas Young, Vol. 1 (George Peacock, ed.),
John Murray, London, 1855.
13. A. Fresnel, Oeuures ComplPtes, Vol. 1, pp. 89 and 129; vol. 2, pp. 261 and 479 (H. de
Senarmont, E. Verdet, and L. Fresnel, eds.), Imprimerie Imptriale, Paris, 1866.
14. G. Kirchhoff, Sitz. Ber. Kgl. Preuss. Akad. Wiss., 22 June 1882.
15. G. Kirchhoff, Sizt. Akad. Wiss. Berlin, 641 (1882). Reprinted in: Wied. Ann. Phys. 18, 663
(1883).
16. G. Kirchoff, Cesammelte Abhandlungen Nachtrag, Barth, Leipzig, 1891.
17. G. Kirchoff, Lecfures on Mathematical Optics, Part. 2 (Boltzmann, ed., Teubner, Leipzig,
1891). (In German.)
18. G. G. Stokes, Trans. Camb. Phil. Sac. 9, 1 (1849).
19. S. Silver, “Microwave Aperture Antenna and Diffraction Theory,” J. Opt. Sac. Am. 52,
131 (1962).
20. A. Sommerfeld, Optics, p. 325, Academic Press, New York, 1954.
21. J. D. Jackson, Classical Electrodvnamics, 2nd ed., p. 391, Wiley, New York, 1962.
22. F. Kottler, Ann. d. Physik 71, 457 (1923).
23. F. Kottler, Ann. d. Physik 72, 457 (1923).
24. B. B. Baker and E. T. Copson, The Mathematical Theory of Huygens Principle, 2nd ed.,
Clarendon Press, Oxford, 1950.
25. J. C. Maxwell, A Treatize on Electricity and Magnetism, Oxford, 1873.
26. H. Poincart, Mathematical Theory o f l i g h t , Part 2, Georges Carre, Paris, 1892. (In French.)
27. A. Sommerfeld, Nachr. Akad. Wiss, Gotringen, Math.-Phys., K 1 1, 338 (1894).
28. C. J . Bouwkamp, Rep. Progr. Phys. (London) 17, 35 (1954).
29. F. Kottler, “Diffraction at a Black Screen,” in Progress in Optics, Vol. 4, Chap. 7 and Vol.
6, Chap. 8 (E. Wolf, ed.), North-Holland Publ., Amsterdam, 1971.
30. F. Kottler, Ann. Phys. 70, 405 (1923a).
31. E. Maey, Wied. Ann. Phys. 49, 96 (1893).
32. A. Sommerfeld, Math. Ann. 47, 317 (1896).
33. A. Rubinowicz, Ann. Phys. 453, 257 (1917).
34. A. Rubinowicz, Ann. Phys. 473, 339 (1924).
35. K. Miyarnoto, A . General Theory of the Boundary Diflraction Wave and Related Investiga-
tions, Ph.D. Thesis, Inst. of Optics, The Univ. of Rochester, 1961.
36. K. Miyamoto and E. Wolf, “Generalization of the Maggy-Rubinowicz Theory of the
Boundary Diffraction Wave, Part 1,” J. Opt. Sac. Am. 52, 615 (1962a).
37. K. Miyamoto and E. Wolf, “Generalization of the Maggy-Rubinowicz Theory of the
Boundary Diffraction Wave, Part 11,” J. Opt. Soc. Am. 52, 626 (1962b).
38. A. Rubinowicz, Progress in Opfics, Vol. 4 Chap. 5 (E. Wolf, ed.), North-Holland Publ.
Amsterdam, 1965.
39. A. Rubinowicz, “Simple Derivation of the Miyamoto-Wolf Formula for the Vector Poten-
tial Associated with a Solution of the Helmholtz Equation,” J. Opt. Sac. Am. 52, 717 (1962).
40. W. Duane, frog. Nut. Acad. Sci. 9, 158 (1923).
41. E. Jahnke and F. Emde, Tables ofFunctions with Formulae and Curves, Leipzig and Berlin,
Teubner. (Reprint: Dover, New York, 1945, 4th ed., p. 35.)
42. T. Pearcey, Tables of Fresnel Integrals to Six Decimal Places, Cambridge University Press,
Cambridge, 1956.
43. R. Bergsten and S. Huberty, “White-light Fresnel Diffraction by a Circular Aperture,”
J. Opt. Sac. Am. 67, 643 (1977).
44. A. C. S. Van Heel, Advanced Optical Techniques, Chap. 13, Wiley, New York, 1967.
45. F. M. Schwerd, Die Beugungserscheinungen aus den Fundamentalgesetzen der Undulations-
theorie Analyfisch entwickelt und in Bildern dargesrellt, Schwan und Goetzsche Hofbuch-
handlung, Mannheim, 1835.
104 DIFFRACTION AND SCATTERING
81. G. B. Parrent, Jr. and B. J . Thompson, “On the Fraunhofer (Far Field) Diffraction Patterns
of Opaque and Transparent Objects with Coherent Background,” Oprica Acra 11, 183
(1964).
82. B. A. Silverman, B. J. Thompson, and J. H . Ward, J. Appl. Meteor. 3, 792 (1964).
83. V. Khare and H . M. Nussenzveig, Phys. Rev. Lert. 33, 976 (1974).
84. M. Minnaert, The Nature of Light and Color in rhe Open Air, Dover, New York, 1954.
This Page Intentionally Left Blank
3. OPTICAL POLARIZATION
Frederic R. Stauffer
Sacramento Peak Observatory
Sunspot, New Mexico 88349
3.1. Introduction
Polarization effects are a result of the wave interaction of light with the
physical world and have long been demonstrated with a double refractive
index crystal known as calcite. Light waves in two orthogonal planes are
laterally separated after passing through calcite, resulting in two polarized
images. Christian Huygens in the late seventeenth century experimented
with and successfully described calcite’s double images by using wave-front
constructions. But, at this time, light was thought to be a longitudinal wave
like sound, and polarized light remained undiscovered. In 1808, E. L. Malus
discovered that reflected light viewed with a properly oriented calcite crystal
passed a single image and extinguished the other image. Malus concluded
that the reflected light was polarized.
In 1815 Brewster showed the significance of the incidence angle on the
polarization of the reflected beam. Maximum polarization (lOOo/~ for dielec-
trics) occurs at an incident angle, now termed Brewster’s angle, depending
on the material’s determined optic parameters. In 1816, Fresnel and Arago
demonstrated with Brewster angle polarizers that two orthogonal polarized
beams did not interfere, which led to the important conclusion that light
is a transverse wave propagation. Fresnel proceeded to simply state the
polarization effects produced by light’s reflection and refraction at boun-
daries between materials further supporting light’s wave nature. Later in
the century, Maxwell’s and Faraday’s efforts produced Maxwell’s equations
describing the fundamental interrelationship between the basic electromag-
netic vectors. The necessary theories to understand polarization were
complete.’.’ The theories are based on observations. To understand
polarization, it is best to experiment with its different aspects. Computer
optical design programs have also made the polarization equations easier
to use.
107
METHODS OF EXPERIMENTAL PHYSICS Copyright 63 1988 hy Academic Press. Inc.
Vol. 26 All rights of reproduclion in any form reserved
I S B N 0-12-475071-8
108 OPTICAL POLARIZATION
VxE+B=O
VXH-D=J Maxwell’s equations (3.1)
V-D=p
V.B=O
D = EE
B=pH material equations
J=oE
Plane and spherical wave solutions exist for this differential equation. Plane
waves in Cartesian coordinates are useful in describing electromagnetic
effects and are used as the example. For dielectrics, the conductivity is zero,
ELECTROMAGNETIC DESCRIPTION O F LIGHT 109
1
=- for a vacuum
C
A =wavelength
(3.4)
S=O
A = arc tan( Ev/ E H )
where E~ and E~ are the dialectric constants in the horizontal and vertical
directions. Now, if the phase difference is changed to an odd multiple of
7r/2, equal intensity orthogonal linear polarizations add together, resulting
in a helical electromagnetic field distribution along the propagation direction
propagating at light's velocity. The resultant plane projection of the light
beam results in a circular spinning vector produced by the helical distribu-
tion moving through space.
E = ( E H E+~E , , E ~e J * )e-J(k r - w r )
S=~ / 2 , EH = Ev for circular polarization,
E, E E~V eJTI2)
=E ~ ( f e-J(k r - w r ) (3.5)
A right-handed helix, E-, produces a clockwise rotating plane projection
looking towards its source, and vice versa for a left-handed helix, E,. These
spinning projections define right- (R)and left- (L) handed circular polariz-
ELECTROMAGNETIC DESCRIPTION OF LIGHT 111
The oppositely spinning E+ and E- vectors add around their paths, resulting
in an ellipse whose major axis, a, is the sum, and whose minor axis, b, is
the difference of the intensities. The major axis is tilted at an angle A, which
is half of the phase difference between the vectors. Figure 1 shows various
combinations of L and R and phase.
a=E++E
b=E+-E
ellipticity = b / a
handedness: E- > E, right handed
E- = E, indeterminate
E- < E, left handed
azimuth = A = 6/2
L/E, E, E_ Result.
tan (y)i,
a =
LONG = 2A.
-45.0 +45.0’
R
FIG. 2. Polarization ellipticity and handedness at positions as mapped on the Poincare
sphere. The handedness is L for the upper and R for the lower hemisphere. The sense of
rotation for handedness is as in Fig. 1, with the polarized light source in the page.
WAVE PROPAGATION IN ISOTROPIC MEDIA 113
FIG.3. Sign convention for the Fresnel equation interpretation and derivation. r, is perpen-
dicular and r,, is parallel to the page. 0 is a vector out of, and x is a vector into, the page.
114 OPTICAL POLARIZATION
( p ~ / p ~ ~ ~ J ” 4i
’ s=i n( ~ sin 4,
L ’ E ‘ / ~ ~ E ~ ) ’ ’ =
~ n sin +i = n’sin 4t, Snell’s law
n = refractive index = c/ o
Epretlected - P
6n’ cos 4I - n cos 4t
- rp = ,
pincident
6
P
n’cos 4 , + n cos 4t
Eptransmitted - 2n cos 4,
- tp=
E p incident
6
P
n‘cos + , + n cos
lrs12+lt,12= 1, 1,
Irp12+lfp12=
R, = rsr?, R , = rprp*,
where R, and R , are the reflectance and are related to the energy flow by
Poynting’s equation. In most cases, p / p ’ - 1 , simplifying the equations.
The Fresnel equations are derived according to Fig. 3 and predict the
phase and amplitude for S and P polarizations. The incidence plane is the
page, and it is also the rp plane. The vector perpendicular to the page is r,,
and it is an x or 0 into or out of the page. The equations’ phase shifts are
interpreted in Fig. 3. For example, at normal incidence, rs and rp are
indistinguishable, and the minus sign difference is because the rp vector is
antiparallel to r, in the figure.
WAVE PROPAGATION I N ISOTROPIC M E D I A 115
The reflection coefficients can also be written in terms of Snell’s law and
incidence angle only.
n’
v=-
n
cos 4, - ( v’ - sin’ 4i)”’
r, =
cos 4, + ( Y’ -sin2 4 ~ ” ”
(3.11)
v’ cos bi- ( v 2 - sin’ 4 ~ ” ’
rp =
+
v’ cos 4i ( v’ - sin’ 4i)’/2*
The positive sign of the square root term was chosen because the replaced
cosine term is positive real valued. When the square root goes to zero and
then becomes imaginary, the negative root is chosen because it predicts an
exponential decrease in the refracted beam’s amplitude, which is the real
situation. Eliminating v from equation (3.11) for rp gives an equation for
rp in terms of r, and the incident angle only.’
(3.12)
There are two cases for v, IvI > 1 and Ivl< 1 . For IvI > 1, for example a
beam incident from air to glass, the value of the square root term is always
positive, real, and less than cos so that r, is negative. The negative sign
is interpreted as a r phase shift for 8 , . The rp goes from a positive value,
116 OPTICAL POLARIZATION
-
- I
-
4
- I
.om
- - 10 n ___________________
-
4
-
L -
-
..-.-.-._._, y.- , , ,
10 n
4
4
-
-
L
4
0.0
0.u v@ ‘pe 9o.a
Incidence angle, 9
FIG. 5. Absolute magnitude for the reflection coefficients r,,p and phase shift S,,,, from a
more to a less dense medium.
(3.16)
For total internal reflection, expressions for the phase changes a,, 6, are,
(3.17)
(3.18)
Also, [A( 6,- &)/A&] slowly varies about the extremum, and this fact can
be used to introduce a phase shift between P and S which is fairly insensitive
to the incident angle. A Fresnel rhomb is a total internal reflection device
to produce a phase shift between P and S. The incident beam totally
internally reflects for both r, and rpoff of the two sides inclined so that the
incidence angle gives a ?r/4 phase difference between P and S on each
reflection. This gives a total retardation between P and S of ~ / 2 A. linearly
polarized beam with equal P and S components incident on the Fresnel
118 OPTICAL POLARIZATION
0
0.d 9O.U
Incidence angle, (pl
AV>I
- - - - - - -. &<I
l5
.
3
rP (s=45.0' rP , q=75.0'
FIG. 7. Complex plane representation for r, in terms of constant phase (radial lines) and
amplitude (semi-circles), and r p as a function of r, and incidence angle.
Z=
( a +2p) * [(a' - 4 ) + 4 ( a +2)p]'12 9
2(1 - P )
(1 + z ) - ( z 2 - a z + 1)'12 (3.20)
W = '
( 1 + z)(z' - az + 1
a=4tan2cpi+2,
p = [ 'I+w
-"]'.
...'..,
[O
.'......'.
(4
FIG. 8. (a) Constant amplitude circles representing either IzI or IwI, (b) IzI mapped into w
for 4, = 30". (c) IwI mapped into z for +a= 30". and (d) IzI normal incidence measurement
overplotted with a I wI measurement for 4, = 30" mapped into z.
(3.22)
where K is the absorption index, n and K are real, and n' is the complex
refractive index. The absorption coefficient k = nK is most often used. It is
important to remember the distinction between k and K .
Snell's law provides a complex-value refracted sine term when light is
incident from a dielectric to a metal, i.e., n , sin 4i= nl2 sin 4,. This means
that for a refracted wave in a conductor, the complex wave number produces
an imaginary term which represents constant beam amplitude surfaces
parallel to the boundary and a real term depending on the incidence angle
which relates constant phase surfaces. Since the constant amplitude and
constant phase surfaces do not necessarily agree, the electromagnetic wave
is said to be inhomogeneous (see Ref. 2, p. 611).
In Fig. 9, Eq. (3.11) is used to plot the reflection coefficients for aluminum
and silver as a function of the incidence angle. The effect of the attenuation
term smoothes the extremum termed the second Brewster angle. This also
means that the rp term is not extinguished as for a dielectric and increases
*T - - - - - - \
\
4
Lo \
0
0.0‘ 90.0-
Incidence angle,
AAI
_------- AA9
FIG. 10. A as a function of incidence angle for aluminum and silver.
with K . Silver has a large absorption coefficient and consequently has only
a small variation in reflectance as a function of the incidence angle. However,
aluminum or rhodium have a lower absorption coefficient and produce
relatively large polarization effects compared to silver.
The difference between S and P phase shifts is plotted in Fig. 10. An
incidence angle with a 7r/2 phase shift in the plot reflects a linearly polarized
beam as an elliptically polarized beam whose axes are along rp and r,. If
rp = r,, circular polarization is produced. This is called the principal angle,
and it is a way to produce a known elliptical polarization.
I I
I
manner, because the incidence angle is not the same for the three mirrors.
The four-mirror arrangement in Fig. 12 has all the mirrors at Brewster's
angle, thus a higher polarizance, low output of R,, the image is not rotated,
and polarization varies across the acceptance angle as for a single mirror.
Transmission Brewster-angle polarizers or pile-of-plane parallel plates
transmit the P component according to the amount of absorption and
therefore, theoretically, 100% for no absorption. Realistically, though, there
exist scattering, surface films, and crystal imperfections. Also, the beam is
deviated perpendicular to its propagation direction, and multiple reflections
make T, difficult to extinguish. These two problems can be corrected by
proper arrangement of the elements and by making them wedge-shaped to
reduce multiple reflection effects. These polarizers are useful in the IR with
materials such as selenium, silver chloride, as shown for Saran Wrap," and
in the UV with calcium fluoride, lithium fluoride, and magnesium fluoride.
There are several advantages of transmission polarizers: an undeviated
beam, high extinction, and high energy use. They are shorter than reflection
type polarizers and are an easy way to produce polarized light.2'
Reflection circular polarizers have been demonstrated, particularly for
the UV. As was shown above for metals, there is a principal angle where
the phase shift between r, and r p is 7712, so that an incident linearly polarized
beam adjusted to give r, = rp produces a circularly polarized beam. In a
paper by McIlrath et al., the authors show results for an aluminum (Al)
mirror exposed to air (overcoated by an aluminum oxide (AI,O,) layer of
about 25 A ) which produces /rsl = Irpl over all incidence angles and thus
can be used as a phase shifter between r, and rp. At an incidence angle
equal to 52" at Lyman-a, the phase difference is ~ / 2 Their . experimental
arrangement was an A1203 linear polarizer at Brewster's angle which fed
the A1 phase shifter mirror, and then analyzed by an LiF Brewster-angle
analyzer in combination with a detector. They produced a circular polarizer
which transmitted 12% with E - / E + = 6 to 10 at A =1216A with a 5"
acceptance angle at c#+ = 52". Circular polarizers are considered later in
discussing thin films and crystal polarizers.22
THIN FILMS 125
The reflection coefficient, rs.p, can be rewritten with relations between the
transmission and reflection coefficients.
+
- r I 2 r2, e-J2@ (3.24)
'S.P - 1 + r12r23 e - i 2 ~ .
(3.25)
126 OPTICAL POLARIZATION
The reflectance thus varies periodically with thickness with extrema given
by d R / d d =O. This occurs when
sin(S2,+2P)=0, S2,+2P=rn7r, rn=0,*1,*2 , . . . ,
The phase term (S2,+2P) can be adjusted with /3 either by film thickness
. ~ ~ 14 is a plot of
or by the incidence angle for a specific ~ a v e l e n g t hFigure
r, and rp versus film thickness for two incidence angles showing the periodic
nature due to interference. The maximum and minimum values are deter-
mined by the reflection coefficients and the material. For a dielectric film
and substrate, the maxima and minima occur at phase thickness equal to
odd and even multiples of a quarter wave. Figure 15 is a plot of the reflection
coefficients versus the incidence angle for two film thicknesses. In Fig. 16
the incidence angle and the film thickness are held constant, making the S
and P reflections vary periodically with wavelength.
3.5.1.2. Linear Polarizers and Phase Retarders. Writing the ellip-
sometric equation r p / r s=tan Ic/ e J Afor the general case of an absorbing film
and absorbing substrate gives an equation nicely mapped in the complex
plane. The points where tan II, equals zero or infinity are possible designs
for a P or S extinguishing polarizer, while points tan Ic/ = 1 and A = *77/2
produce a circular polarizer. This suggests that the phase difference can be
controlled between r,* and r2, via the incident angle, refractive indices, and
film thickness to make P or S reflective polarizers, linear partial polarizers
(which differentially attenuate the P and S components without shifting
the phase between the components), and wave retarders.25-28
The P or S extinguishing polarizers with a single dielectric film on a
metal substrate require large incidence angles for the higher reflecting
THIN F I L M S 127
OA la
Film thickness, d
1.0
FIG. 14. rr and r,, for reflections from air to glass film to glass substrate showing the effect
on r, and r,, as a function of film thickness, n , = 1.0, n, = 2.0, and n, = 1.5.
1.0 r I
....
Incidence angle, (h
Ir, I
-.-.-.- Irpl
0.u 90.0-
Incidence angle,
F I G . 15. r\ and rp for reflections from air to glass film to glass substrate with a 2h and a
Oh film, showing the film's effect as a function of incidence angle.
128 OPTICAL POLARIZATION
-1
4
-
L-
p = 30'
0.0
5000 A 5500 A
Wavelength, h
1.0
t = 60'
FIci. 16. rs and rp for reflections from air to glass film to glass substrate showing the film's
effect as a function of wavelength about a central wavelength A o .
These are suitable for beams with high-energy densities. The time depen-
dence of the wave equation, wt, is assumed fixed, and the equations are
expressed as second-order differential equations in space for the magnetic
and electric wave propagation. The multilayer’s effect on the incident beam’s
magnetic and electric field is expressed as a 2 x 2 unimodular transformation
matrix. The new beam produced from the first layer is applied to the second
layer, producing a modified beam for the third layer, etc. The total thin-film
stack can thus be expressed as a single unimodular transformation matrix.
To see the effect of the dielectric multilayer on the S and P polarization,
first consider the mth thickness z, at normal incidence. The matrix transfor-
mation of the electric and magnetic fields is
The diagonal and cross-diagonal terms are pure real cosine and pure
imaginary sine terms with arguments which are the phase thickness,
2rn,z,
A,=-. (3.28)
A0
To apply this at oblique incidence angles, the effective phase thickness and
effective index (a, p m )are used to describe the uniform plane wave at an
oblique incidence as a nonuniform plane wave at normal incidence, via
6, = A, cos 4, and
where 4, is the refracted angle through the layer. The last two relations
were introduced in Fresnel’s equations earlier and are substituted into
normal incidence Fresnel equations to produce nonnormal incidence
equations.
Periodic multilayer dielectric thin films have a transformation matrix,
also known as a Herpin matrix, which is equivalent to a two-film combination
for a particular wavelength. A symmetric three-layer film is a special case
which simplifies to an equivalent single-film matrix for all wavelengths. The
single film is described by an equivalent index, N, and an equivalent phase
thickness, r.
(3.29)
130 OPTICAL POLARIZATION
FIG. 17. A P-polarizing MacNeille prism using multiple reflections at Brewster's angle &.
(3.31)
visible spectrum, the retardation variation is 0.4" for a coated rhomb and
5.0" for an uncoated version.
Imperfections in total internal reflection devices also are due to strain
birefringence which is inversely proportional to wavelength, and for applica-
tions in the ultraviolet it becomes critical, especially if the retarder is to be
achromatic. Also, it is reported that polishing the faces that are not used
reduces strain 10-15 mm inside the glass by a factor of two over that for a
fine-ground face. Attention should be given to mounting the retarder, and
silicone rubber sealant as an adhesive does not introduce any measurable
strain.
Another error source is the refractive index of the glass and thin film.
For the glass, the surface is slightly different in phase retardation, about
0.5" more than the theoretical values, because surface shaping and polishing
changes the refractive index. Also, generally for most materials, the refractive
index increases with a decrease in wavelength, except for polyvinyl alcohol.
Retardation error sources can be corrected by adjusting the reflected beam's
phase difference experimentally with film thickness. The film can increase
achromatism, because the increase in refractive index with decrease in
wavelength can be balanced by the thin film's increased phase thickness
due to p. King has made Fresnel rhombs with either 144A magnesium
fluoride films on each reflecting surface or a 268A film on one surface
leaving the other uncoated, which produce a retardation of 90" f 0.2" from
3440 A to 6680 A compared to 90"*2.5" for an uncoated rhomb.
In Fig. 18, various other retardation devices are shown. They have about
the same chromaticity but various other advantages and disadvantages with
regard to beam displacement, physical size, and angular acceptance. A
Fresnel rhomb displaces the beam; and, if not optimized by thin films, it
has a 9" variation in retardation with a *7" incidence angle. Achromatic
device one, ADl, is self-compensating in incidence angle, because the first
two reflections on the minus side of the optimum angle will have two
reflections on the positive side, therefore almost canceling the off -angle
effects. It is a very long prism, 175 mm long for a 10-mm beam, and it would
be hard to meet theoretical predictions due to strain birefringence. It has
a 0.7" retardance variation for a *7" incidence-angle variation. The Mooney
rhomb is made of two dense flint 60" prisms and is self-compensating in
incidence angle with a performance similar to AD1. Also, its physical size
is small. Achromatic device two, AD2, is physically small and does not
deviate or displace the beam, but its incidence angle sensitivity is terrible
with a 13.2" retardation variation for a *3" incidence-angle variation.''
Half-wave or full-wave rhombs have also been designed to deviate the light
beam 90" without changing the polarization. A simple design is the dove
prism.
WAVE PROPAGATION I N A N I S O T R O P I C M E D I A 133
FIG. 18. Total internal reflection devices. (a) Fresnel rhomb, ( b ) Mooney rhomb, (c)
achromatic device 1 ( A D l ) , and (d) achromatic device 2 (AD2).
(3.32)
equation relating the phase velocity to the wave normal direction and the
material constants.
Considering the functional form of the energy density, the dielectric
tensor is seen to be symmetric, and at most six independent values need to
be determined. A coordinate system, termed the principal axes, can be fixed
to the tensor’s diagonal. In this coordinate system, the material equations
simplify to the diagonal terms.
E,,+Ex, Eyy+Ey, EZZ+EZ,
i = x, y , x.
Assume the media is homogeneous, nonconductive, nonmagnetic, and
electrically anisotropic. A plane wave, Eq. (3.4) simplifies the derivatives
in Maxwell’s equations.
V x H =j w D = -jks x H
V x E = -jpwH = -jks x E
(3.33)
(3.34)
where s is the unit vector for propagation direction and k defines equal
phase planes A apart. D and H are perpendicular to s, but E is not. E is
perpendicular to the Poynting energy flow vector, S.
WAVE PROPAGATION I N ANISOTROPIC M E D I A 135
S
E
FIG. 19. The orientation of the electromagnetic vectors H, D, and E, the wave normal s,
and the energy propagation vector S in an anisotropic medium.
(3.35)
symmetry axis directed into the rhomb’s corner. A surface cut perpendicular
to the optic axis on the corner accurately determines the crystal’s principal
axes.’
Unpolarized light incident on a uniaxial crystal produces two linearly
polarized beams determined by the optic constants for the plane and the
optic axis. For all field orientations with respect to the optic axis, an electric
vector parallel to the plane exists and behaves as for an isotropic medium.
This is termed the ordinary beam, 0, and is described by a spherical velocity
surface. The wave perpendicular to the 0 beam depends on the incident
beam’s orientation and varies from the optic constants for the plane to the
optic constants perpendicular to the plane. This is the extraordinary beam,
E, and its phase velocity as a function of propagation direction is a
spheroidal surface.
The phase velocity surfaces for a uniaxial crystal in terms of the principal
axes are found by applying Fresnel’s equation of wave normals. The z axis
is made parallel to the optic axis, and the angle from the wave normal to
z is 9.
v, = v, = v,, v, = v,,
s: + s: = sin2 9, s t = cos2 8,
( V ; - vf)[(V ; - ~ f sin’
) 9 + ( V : - v;)cos2 e] = 0, (3.36)
v;,= vf,
v;*= vf sin2 e + v’,cos2 e.
Figure 20 is a plot of the two phase velocity surfaces and an arbitrary wave
normal, s, with its two phase velocities. When s is in the optic axis direction,
the phase velocities are equal, and when s is perpendicular, the difference
between phase velocities is maximum.
z O.A. z O.A.
FIG. 20. Cross sections of phase-velocity surfaces for a negative and a positive uniaxial
crystal predicted by the Fresnel equation for wave normals. The phase velocities are determined
by the wave normal’s orientation.
WAVE PROPAGATION IN ANISOTROPIC MEDIA 137
2 Z Y
vx > vy > vz
FIG. 21. Three cross sections of phase-velocity surfaces described by the Fresnel wave
normal equation for a biaxial crystal.
138 OPTICAL POLARIZATION
The literature has references with material constants given for the more
commonly used crystals and wavelength ranges.
3.6.2.Crystal Polarizers
3.6.2.1. Prism Beamsplitters. Beam-splitting polarizers, such as the
Rochon, SCnarmont, and Wollaston are biprisms which separate the 0 and
E beams by introducing a discontinuity in the refractive index for either
the E beam (Rochon and Sknarmont) or for both the E and 0 beams
(Wollaston). Figure 22 shows the E and 0 rays for a Rochon and a
Wollaston. The Rochon and SCnarmont are basically the same with their
first prisms’ optic axis perpendicular to the entrance face.
The Rochon gives a slightly larger E-beam deviation than the SCnarmont,
because the beam sees the maximum birefringence. These prisms can be
used backwards, but the E-beam deviation will be slightly less. Snell’s law
at the prism discontinuity, now from nE to no, reverses the incident beam’s
refracted angle from greater to smaller deviation. The beam deviation and
FIG.22. ( a ) Rochon and ( b ) Wollaston polarization beam splitters. The optic axis direction
is oriented parallel I or perpendicular 0 to the page.
WAVE PROPAGATION IN ANISOTROPIC MEDIA 139
(3.38)
A Wollaston polarizing beamsplitter has the optic axis of its first prism
parallel to the face to produce an E and an 0 beam which sees the second
prism's optic axis rotated 90" about the beam axis to produce a maximum
birefringence and deviation for both beams. Wollaston prisms can produce
angular separations up to 20" with a slight asymmetry, and triple-element
Wollastons produce larger angular separations. At large angles through
crystal polarizers, image distortion occurs due to nE varying in angle with
respect to the optic axis.
Since the birefringence of Rochon's first prism is not used, it can be
substituted with strain-free glass of either index, nE or no. Glass eliminates
image distortion due to the effects of the birefringent crystal and is easy to
manufacture cheaply with a good tolerance. The residual birefringence in
glass needs to be c o n ~ i d e r e d . ~ ~
3.6.2.2.Prism Polarizers. Polarizing prisms are birefringent crystals cut
into two pieces with an interface of an index which totally internally reflects
the higher-index beam, for calcite the ordinary beam.' A Glan-Thompson
prism is illustrated in Fig. 23. A Nicol prism was the first polarizing prism
design, and it is essentially a calcite rhomb cut along a diagonal and
cemented together. Glan-type prism polarizers arrange the optic axis to give
i"
i
i
i
I
i
i
,-I _._.-.- 0
- - - - - - -. E
FIG. 23. Clan-Thompson prism drawn with normal and extreme rays. The optic axis is
perpendicular to the page, 0.
140 OPTICAL POLARIZATION
(3.42)
The usable maximum (symmetric about the prism's normal) incident angle
or semifield angle is about 8" in the ultraviolet, decreasing for longer
wavelengths, and is limited by the transmitted E beam being totally inter-
nally reflected at the cut. Large prism apertures can be made, since most
of the crystal is used.
An air-spaced Glan-Thompson is termed a Clan- Foucault The optic
axis is parallel to the cut surface, so that the transmitted beam has its
electromagnetic vibration oriented parallel to the cut surface, which has
the higher Fresnel reflection coefficient, r,. If the optic axis is rotated 90"
about the beam axis, the reflection coefficient is now rp,which is significantly
reduced. The transmission is increased in the ultraviolet by 10% for this
design, which is called a Clan- Taylor. The optic axis orientation decreases
the useful birefringence which reduces the semifield angle. The cement n,
is unity, and the angular acceptance primarily depends on the cut angle
and the crystal's wavelength-dependent birefringence. The L / A ratio is
about equal to one.
To increase the semifield angle or to decrease the L / A ratio for a given
maximum incident angle, double prisms are constructed.' Examples are the
Ahrens prism (two Glan-Thompson prisms next to each other), the Grosse
prism (an air-spaced Ahrens), and the Marple-Hess prism (a double Glan-
Taylor back to back). Since the number of components increases in double
prisms, the manufacturing difficulty is increased, and extinction perform-
ance is about an order of magnitude worse than for single polarizing prisms.
The Ahrens and Grosse prisms have half the L / A ratio of their Glan
single-prism counterparts. For example, an Ahrens prism with an L / A of
1.8 has been made with a 26" semifield angle. The Grosse prism has an L / A
about equal to one-half and is useful for tight physical limitations. Marple-
Hess prisms have definite advantages due to the cut angles S and -S. 0-rays
below the axis will be totally reflected by the first surface, and 0-rays above
the axis will be totally reflected by the second surface. The axial 0-ray and
its total internal reflection due to n , ( A ) decreasing in A is the design
parameter for the ordinary beam. The prism's limiting field angle is deter-
mined by the E-ray's total reflection, and the limiting wavelength depends
on the axial 0-ray's transmission. These prisms have a lower transmission
in the ultraviolet because of their increased L / A ratio.
Glass-calcite Glan-type prism polarizers are possible, because the second
prism is used to transmit the E-ray, and its birefringence is not used. The
second crystal prism can be replaced with an index n, glass prism. Or, the
first prism can be replaced with an no index glass, and an n, cement used
to assemble the prism. The E-ray is totally reflected in this case. These
prisms are usually not as good as all-crystal polarizers, because the greater
L/ A ratio and optical cement enhance the resulting residual strain.s3
142 OPTICAL POLARIZATION
d = mh. (3.45)
cos 6 +-
( 3 = dlsin(26,), (3.46)
with the resultant polarization ellipse axes aligned 45" to the wave plate's
extraordinary and ordinary axes. The analyzer is then oriented 45" to the
ellipse axes. Rotating the wave plate about the optic axis direction has no
effect on the birefringence. The light-path length in the crystal and retarda-
tion will change, but the analyzer will transmit. Rotating the wave-plate
surface perpendicular to the optic axis changes the birefringence from a
maximum to a zero value for a 90" rotation, and so extinction occurs.62
The wave plate's retardation is related to the intensity of the elliptical
polarization's major and minor axes, as seen earlier. It is a matter of
measuring the maximum intensity of the above crossed polarizers and
wave-plate arrangement and rotating the analyzer 90" to measure the
minimum or minor axis intensity,
(3.47)
(3.50)
where C:k/ is the elastic tensor element, ekliis the piezoelectric tensor element,
and Tj is the strain tensor element. Conversely, an applied electric field in
a properly cut piezoelectric crystal can generate shear or longitudinal waves
which are used for transducers.
The change in refractive index can be found by considering how the
impermeability matrix changes for either effect. Under the assumption that
the index change is small to the material’s initial index, the index change is
nips
Ans = --
2 ’
(3.51)
can be either offset with a bias voltage or can be eliminated by using crossed
crystal pairs to cancel each other’s birefringence (this also reduces tem-
perature-birefringence effects). The voltage applied to produce half-wave
retardation is in the kilovolts range. The drive voltages for modulators
depend also on the path length through the crystal and on electrode
configuration. With many passes through a crystal or with cascaded crystal
pairs, the drive voltage is reduced.
High-frequency modulation is limited by capacitance, mechanical stress,
and heating; gigahertz frequencies can be achieved. Low-frequency modula-
tion on the order of a second or longer has the problem of current leaking
across the crystal. The charge needs to be replaced. This is accomplished
with a slightly conducting grease for contacting the electrodes to the crystal.69
Longitudinal Pockels cells have either transparent electrodes or ring
electrodes on the entrance and exit surfaces. The phase shift is
(3.52)
where I is the distance through the crystal, and E is the electric field. A
specification for longitudinal . modulators is the voltage for half-wave
retardation, VAI2, which is needed to produce maximum transmission
between crossed polarizers. Large apertures can be made.
Transverse Pockels cells have electrodes on the crystal’s sides, and the
light beam passes perpendicular to the applied field. Retardation increases
with an increased path length through the crystal and with a decreased
entrance aperture or electrode gap. Capacitance increases with increased
crystal length so that the frequency bandwidth decreases. The retardation is
(3.53)
(3.55)
where n is the refractive index, n28 is the ac Kerr coefficient, Sn, and Snll
are the refractive index changes perpendicular and parallel to the E field,
T is the relaxation time, t is the gating pulse length, I?( t ) is the time average
over one period, and P ( t ) is the power density. For t > T, the birefringence
, n,,E2( t ) .
~ n ,-,~ n = (3.56)
The sample beam’s retardation is
I
6 = 27rn,,P( t ) ~ (3.57)
(Acne,) ‘
The on/off extinction ratio is typically about 1000: 1, determined mostly
by the extinction ratio of the entrance and exit polarizer, and the aperture
is the gating beam diameter.
150 OPTICAL POLARIZATION
3.6.4.2. Acousto-optic Effect. There are several ways to use the elasto-
optic effect depending on incident beam diameter and frequency modula-
t i ~ nA. uniaxial
~ ~ static strain deforms the isotropic media’s spherical phase
velocity surface into a spheroidal phase velocity surface with an optic axis
oriented along the strain direction with a birefringence,
6 = cos(Rr),
where R is the angular frequency, and k is the wave number for sound. If
the sound wavelength is comparable to the beam diameter, the phase
modulation or birefringence will vary across the aperture, and the intensity
modulation contrast between crossed polarizers will not be as large. This
limits the maximum frequency for resonating elasto-optic modulators to
less than 1 MHz.
Standing-wave photo-elastic modulators can drive in a self-resonating
mode and can be made with zinc selenide, calcium flouride, KRS-5 infrared
glass, germanium, and fused quartz. There are two types of construction,
either the bar type73or an isometric-contour element.74 The bar type is the
first design which is made of either one or two transducers connected to
either side of the optical element to be modulated. The transducers and the
optical element all resonate in a standing-wave mode with nodes at the
joints. The elements can be cemented together with silicone cement to reduce
strain birefringence. Bar-type modulators have two error sources which are
eliminated by the isometric-contour-type modulator. The errors are, ( 1 ) a
WAVE PROPAGATION I N ANISOTROPIC M E D I A 151
(3.63)
where V, is the material’s sound velocity, and & is the light frequency.
For high sound frequencies or for a long interaction path with light, the
behavior changes from a thin phase grating operating in the Raman- Nath
regime to the thick phase grating behavior similar to X-ray diffraction called
the Bragg regime. The light is diffracted into two orders, because the sound
wave is a sinusoidally varying density function needing one frequency term
to describe it, while for X-ray diffraction in particle distributions, higher
spatial harmonics are needed to describe the particles’ positions producing
a range of diffracted angles.
152 OPTICAL POLARIZATION
where S' and S are the transformed and incident polarization vectors. The
phase-modulated beam will always have one linear component unmodu-
lated. If the modulator is rotated 45", then the modulated exiting linear
polarization changes from horizontal to 45" linear.
s; = s,,
S ; = S, cos S - S, sin 6.
m
cos 6 =cos(&,,, sin wt)=Jo(6,,,)+2 C JZn(Smax) sin(2nwt).
fl=l
The time averages for the sine terms are zero, and the remaining term is
J,( S,,,) = 0 for S,,, = 2.405, the first zero of J, . This means zero polarization
can be produced in either horizontal linear and circular polarizations, or
in 45" linear and circular polarizations, and can be used to calibrate instru-
mental polarization to an accuracy of 0.01 to o . o o ~ ~ ~ . ~ ~
If an optic modulator were driven by a sawtooth function such that the
phase modulation varied linearly from 0 to 2rr, the time-averaged output
light would have only a linear polarized component. If a second modulator
is added with its optic axis 45" to the first modulator and driven at an
integral multiple of the first frequency nw, then the resulting polarization
is zero.
Another method to produce depolarized light makes use of the time-
average effects for rotating retardation plates.83
s= plate retardation
st,= s, total intensity polarized + unpolarized
- 2( 1 + cos 6)Sl
S' -I horizontal linear polarization
S ; = $(1 + cos 6)S2 +45" linear polarization
s; = s3 cos 6 circular polarization
A rotating achromatic quarter-wave plate will pass only unpolarized and
linearly but not circularly polarized light. A rotating achromatic half-wave
plate will pass only unpolarized and circularly but not linearly polarized
light, and the circular polarization's handedness is reversed. A combination
of two wave plates rotating in opposite directions produces an element
passing only unpolarized light.
chosen such that the net molecular polarization lines up parallel or perpen-
dicular to the crystal's long axis; this corresponds to positive or negative
-
birefringence, A n 0.2. For example, tunable birefringence can be produced
by treating the container's walls such that negative nematic crystals line up
parallel to the light beam in the off state. An electric field applied to the
walls will flip the liquid crystals perpendicular to the electric field and light
beam and produce a retardation proportional to the voltage up to a threshold
voltage. Or, the cell's interior wall can be treated to align the positive
nematic crystal structure perpendicular to the beam; this is done by rubbing
the glass to form parallel microgrooves and then treating the surface with
a high-energy surfactant. Then, a 90" twist is introduced between the entrance
and exit surface, rotating the beam 90". A voltage of 5-10 volts will com-
pletely disrupt the rotation and extinguish the beam between crossed
polarizers. Liquid crystals are low-current, low-voltage, high-contrast, thin-
display p o l a r i z e r ~ . ~ ~
anomalous effect. The resonance and Wood's anomalies are close together
for this type of surface. For a strongly modulated surface, the reactance
increases and the anomalies separate.
Blazed gratings in the Littrow configuration (the incident and. diffracted
beams have the same angle), with right triangle-shaped grooves and the
90" corner at the apex, can be described by the blaze angle only. First-order
diffraction is most strongly influenced by anomalies which decrease for
higher orders because A / d decreases. Higher-order anomalies and diffracted
angle deviations can be deduced in terms of the first-order Littrow configur-
atiom9*
First-order gratings with a blaze angle of less than 5" behave close to
scalar theory and show few anomalies. The P and S polarizations have a
peak in their efficiency at the blaze angle equal to 100%. For gratings blazed
between 5" to 18", Wood's anomalies at the Rayleigh wavelength for the
-1 and + 2 order at A/d = 2 / 3 , about 19.5", decrease in strength with the
blaze angle. The S grating efficiency is 100% at the blaze angle, and the P
efficiency peak is decreasing. For gratings blazed at 18" to 22", the anomalies
are suppressed because the Rayleigh pass-off anomaly occurs in the same
direction as the blaze angle. The S efficiency becomes more uniformly high
for diffraction angles greater than about 19". The P efficiency is a minimum
of about 80%. For blaze angles greater than 22", the S efficiency for
diffraction angles greater than 19" is high up to 60" and is 100% at the blaze
angle. The P efficiency peak is 100% at the Rayleigh wavelength A / d = 2 / 3 .
Gratings with an angular deviation between the incident and reflected beams
have lower P and S efficiencies, and the anomalies and peak efficiencies
shift in wavelength. Gratings with known P and S efficiencies can be used
as a polarizing element in an instrument.
Light sources can be polarized for a number of reasons. Lasers are usually
linearly polarized coherent emission sources. Incandescent and fluorescent
emissions are polarized because the emitted radiation is refracted at the
surface; and radiation refracted at a grazing angle to the surface is strongly
P polarized.93 Metal filaments, ribbons, and Nernst glowers have demon-
strated strong polarization effects for radiation from the highly angled
cylindrical edges. Also, deuterium lamps have been studied, which demon-
strate strong polarization effects for old lamps with a thin metal film
deposited on the glass envelope.94 New lamps are not as strongly polarized
across the light spot, but the polarization is still nonuniform away from the
central light spot, which is possibly due to polarized reflections from the
POLARIZATION DETERMINATION A N D MATHEMATICAL DESCRIPTION 159
3.9. Pola rizat ion Deter mi nation and Mat hemat i ca I Descri ption
E = E x e x+ EyeY=
(3 ,
(3.70)
160 OPTICAL POLARIZATION
E‘=M*E
M = ( kx
0
0
k,
), 1 Eb
kx
k, E,
Ex )
partial linear polarizer in x - y
6 phase retardation in x - y
reflection in x - y
Ex cos a + E, sin a
-sina cosa -Ex sin (Y + E, cos a
rotation of an angle a counterclockwise looking into the beam
1 0 cos2 a cos a sin a
(I Q u V)
(I M C S)
(So s, s
2 SJ
POLARIZATION DETERMINATION A N D MATHEMATICAL DESCRIPTION 161
The following relations hold between the Stokes' vectors and the time-
averaged electromagnetic vectors,
I:=(E:)+(E:)+I~,
2 Q 2 + U 2 + V2,
0 = ( E 3 -W;L
U = 2( E,E,,) cos S = Re( I?$,,),
The Stokes' vectors can also be written in terms of the light's polarization
azimuth a,ellipticity e, and the percent polarization P.
( Q 2 + U 2 + V2)'/2
P= 3
IT
1 1 0 0
-1 0 0
0 0 0
90" linear polarization
162 OPTICAL POLARIZATION
1 0 1 0
-1 0 0
left circular polarization
0 0
0
rotation of an angle cy counterclockwise looking into the beam.
The 4 x 4 transformation matrices can also be determined experimentally
to take out an instrument’s polarization. A fairly complete list of matrices
is given by Shurcliff,2’ and recent articles go into details for error analysis
with Jones’ and Mueller’s matrices for polarization measurement^.^'-^^'
References
61. S. Chu, R. Conti, P. Bucksbaum, and E. Commins, “Making Mica Retardation Plates:
A Simple Technique,” Appl. Opt. 18, 1138 (1979).
62. J. Strong, Procedures in Experimental Physics, p. 388, Prentice-Hall, New York, 1952.
63. W. C. Davis, “Simplified Wave Plate Measurements,” Appl. Opt. 20, 2879 (1981).
64. E. Hartfield and B. J. Thompson, “Optical Modulators,” in Handbook ofOptics, Chap.
17 (W. G. Driscoll and W. Voughan eds.), McGraw-Hill, New York, 1972.
65. E. K. Sittig, “Elastooptic Light Modulation and Deflection,” in Progress in Optics, Vol.
10, Chap. 6, p. 231 (E. Wolf, ed.), North-Holland Publ., Amsterdam, 1972.
66. C. Forno and 0. C. Jones, “Hexamine Electro-Optic Light Modulators,” J. Phys. E 7,
101 (1971).
67. B. P. Blumich and R. Germer, “Electro-Optic Shutters with PLZT Ceramics for Infrared
Applications,” J. Phys. E 12, 770 (1979).
68. E. R. Kocher and R. P. Novak, “Fast Electro-Optical Shutter with Programable Trans-
mission Control,” 1. Phys. E 13, 542 (1980).
69. T. G. Baur, D. E. Elmore, R. H. Lee, C. W. Querfeld, and S. R. Rogers, “Stokes 11. A
New Polarimeter for Solar Observation,” Solar Physics 70, 395 (1981).
70. J. Etchepare, G. Grillon, R. Muller, and A. Orszag, “Kinetics of Optical Kerr Effect
Induced by Picosecond Laser Pulses,” Opt. Comm. 34, 269 (1980).
71. M. A. Duguay, “The Ultrafast Optical Kerr Shutter,” in Progress in Optics, Vol. 14, Chap.
4, p. 163 (E. Wolf, ed.), North-Holland Publ., Amsterdam, 1976.
72. A. Korpel, Applied Optics and Optical Engineering, Vol. 6, p. 89 (R. Kingslake and B. J.
Thompson, eds.), Academic Press, New York, 1980.
73. J. C. Cheng, L. A. Nafie, S. D. Allen, and A. I. Braunstein, “Photoelastic Modulator for
the 0.55-13 pm Range,” Appl. Opt. 15, 1960 (1976).
74. J. C. Kemp, “Photoelastic-Modulator Polarimeters in Astronomy,” in Polarizers and
Applications (G. B. Trapani, ed.), Proc. Soc. Phot. Opt. Inst. Eng. 307, 80 (1981).
75. A. Korpel, Electro Optics-Principles and Applications (B. Thompson and J. B. DeVelis,
eds.), Proc. SOC.Phof. Opt. Inst. Eng. 38, 3 (1973).
76. G. B. Brandt, M. Gottlieb, and J. J. Conroy, “Improved Contrast for Acousto-Optical
Modulators Using Spatial Filtering,” J. Opt. Soc. A m . 67, 1269 (1977).
77. A. Korpel, “Acousto-Optics,” in Applied Oprics and Optical Engineering, Vol. 6, Chap. 4
(R. Kingslake and B. J. Thompson, eds.), Academic Press, New York, 1980.
78. 1. C. Chang, “Acousto-Optic Tunable Filters,” Opt. Eng. 20, 824 (1981). ,
79. J. B. Houston, Jr., M. Gottlieb, Shi-Kay Yao, I. C. Chang, J. Tracy, L. M. Smithline, and
G. J. Wolga, “The Potential for Acousto-Optics in Instrumentation: An Overview for the
1980s.” Opt. Eng. 20,712 (1981).
80. M. Gottlieb and G. W. Roland, “Infrared Acousto-Optic Materials: Applications,
Requirements and Crystal Development,” Opt. Eng. 19,901 (1980).
81. Shan-Ling Lu and A. P. Loeber, “Depolarization of White Light b y a Birefringent Crystal,”
J. Opt. SOC.A m . 65, 248 (1975).
82. A. P. Loeber, “Depolarization of White Light by a Birefringent Crystal. 11. The Lyot
Depolarizer,” J. Opt. SOC.A m . 72,650 (1982).
83. B. H . Billings, “Monochromatic Depolarizer,” J. Opt. SOC.A m . 41,966 (1951).
84. R. A. Saref, in Electro Optics-Principles and Applications (B. J. Thompson and J. B.
DeVelis, eds.), Proc. SOC. Phof. Opt. Inst. Eng. 38,23 (1973).
85. E. H. Land, “Some Aspects of the Development of Sheet Polarizers,” J. Opt. SOC.A m .
41, 957 (1951).
86. R. V. Jones and J. C. S. Richards, “The Polarization of Light by Narrow Slits,” Proc.
Roy. SOC.(London), A225, 122 (1954).
87. R. E. Slocum, “Evaporative Thin Metal Films as Polarizers,” in Polarizers and Applicafions
(G. B. Trapani ed.), Proc. SOC.Phot. Opt. Inst. Eng. 307,25 (1981).
166 OPTICAL POLARIZATION
R. D. Bahuguna*
Rose-Hulman Institute of Technology
5500 Wabash Avenue
Terre Haute, Indiana 47803
D. Malacara
Centro de lnvestigaciones en Optica. A X
Apartado Postal No. 948
37000 Leon, Gto. Mexico
4.1. Introduction
The invention of the laser has had a great impact on the science of optics.
In particular, the high coherence of the laser light has completely revitalized
those methods of optics which use interference as the basic phenomenon.
Among these is the method of wave-front reconstruction. Also called
holography, this method was invented by Denis Gabor in 1948'*2.3 to improve
the resolution in the image obtained with the electron microscope. Although
the initial goal did not prove successful, holography in itself became an
important subject of investigation. However, due to the lack of a coherent
light source, it remained in obscurity for more than a decade.
In the 1960s, with the availability of the laser, a major advance was made
at the University of Michigan Institute of Science and Technology. Leith
and Upatnieks in 1962495*6 applied the laser light to holography producing
excellent three-dimensional images which astonished all those who saw
them. In the same year, Denisyuk' made a remarkable contribution to
holography by combining it with the volume-recording process, invented
by Lippmann' in 1891, to make color photographs. Using a thick emulsion
to record the interference fringes, a method was produced for color hologra-
phy in which reconstruction could be carried out with a point source of
white light. His hologram is, in effect, a hologram and a narrow-band
spectral filter, combined into a single structure.
*Work was done while the author was at Centro de lnvestigaciones en Optica, A.C.,
Apartado Postal No. 948, 37000 Leon, Gto. Mexico.
167
METHODS OF t X P E R I M t N T A L PHYSICS Copyright 01988 by Academic Press. Inc.
Vol ?b All rights of reproduction in an) form reserred.
I S B N 0-12-475971-8
168 HOLOGRAPHY
later see the effect of thick emulsions. Finally, we discuss the nonlinear
effects and aberrations involved in the process.
t(x)=to+BE(x), (4.2)
where to is a constant background transmittance and B is a parameter
determined by the emulsion and the processing conditions. Since E ( x ) for
a given exposure time is directly proportional to Z(x), we may write using
eqns. (4.1) and (4.2):
t(x)= t,+B'(1012+lR12+OR*+O*R) (4.3)
where B'= B T ; t is the exposure time.
(C 1 (d)
FIG. 2. Method to determine the conjugate image: Join the reference point source R and
the object point 0, by a straight line. Draw a perpendicular R P to the plate H through R.
Measure the angle RPO = a say. Draw a line passing through P at an angle ( - a ) and let it
meet O , R at 0;. 0;is then the conjugate image. ( a ) , (b) and (c) are different possible
recording geometries; (d) is similar to ( a ) except that one is viewing the image in reflection.
Reflection images of 0, and 0:are seen as 0,and 0;respectively.
T H E O R Y OF HOLOGRAPHY 171
FIG. 3. Basic arrangement to reproduce the images of a hologram: (a) recording set up;
(b) reconstruction with a beam identical to the reference beam; (c) reconstruction with a beam
antiparallel to the reference beam.
172 HOLOGRAPHY
3 Emulsion
and
e = * ( n - 4/21.
We will consider each case separately as follows:
(a) When the illuminating beam is identical to the reference beam ( 0 =
4/2), the object wave is reconstructed, and the virtual image formed. No
conjugate image is observed.
(b) When the illuminating beam is in the opposite direction to the initial
reference wave 6 = - ( n - 4 / 2 ) , a real image is reconstructed. This real
image is pseudoscopic, and no virtual image is formed.
174 HOLOGRAPHY
(c) When the illuminating beam is in the direction of the object beam
( 0 = - 4 / 2 ) , only an imperfect real image, with aberrations, is formed. The
complete object cannot be seen, becasue the Bragg condition is not satisfied
for all waves from the object.
(d) When the illuminating beam is in an opposite direction to the object
wave ( 0 = T - 4 / 2 ) , a partial, imperfect virtual image, with aberrations, is
reconstructed. The complete image is not observed for the same reason
given before.
Summarizing, we can say that in thick holograms only one image is
reconstructed, at,the same position where the object was when the hologram
was formed. If the reconstructing beam is parallel to the reference beam,
the image is orthoscopic. However, if the reconstructing beam is antiparallel
to the reference beam, the image is pseudoscopic.
So far, we have discussed the transmission volume holograms. The reflec-
tion type can be made as shown in Fig. 5 . The object wave 0 and the
reference wave R come from opposite directions. The interference fringes
thus recorded are approximately planes parallel to the emulsion. The dis-
tance between the planes is h / 2 . If such a hologram is illuminated with a
beam identical to the reference beam, it will reconstruct the virtual image
of the object. The reconstruction is viewed in reflection rather than trans-
mission.
The major feature of the refection holograms is that a single-color wave
front may be reconstructed with white-light illumination. The Bragg planes
act as an interference filter and have a high reflectivity only for those
wavelengths that produced them. However, in practice, such a hologram
recorded at one wavelength may reconstruct in a slightly shorter wavelength.
Object wave 0
i-
' Reference wave R
For example, the recording in red color may reconstruct in green. This is
because of emulsion shrinkage during photographic processing. The Bragg
planes pull together so that the final spacing is smaller than the one at
recording, leading to reconstruction at a different wavelength. It is worth
pointing out that white-light illumination results in a loss of image resolution,
since the hologram acts as a spectral filter with a bandwidth of about 100 A,
which is too large for high-resolution imagery.
L S
\
d
-
.
$
-4
H
0
FIG.8. Recording of a Fraunhofer hologram with a free space configuration.
The following are the two geometries with which to record such a
hologram.
(1) This configuration is a free-space geometry (see Fig. 8). The subject
(a small transparency) is illuminated by a broad plane-parallel beam. The
recording plate is kept in the far field of the subject. The light scattered
from the subject interferes with the uniform background to form the
hologram.
This geometry was utilized by Thompson et aLZ5for measuring the size
and shape of dynamic aerosol particles. The twin-image problem in this
geometry is apparently removed.
(2) In this configuration (see Fig. 9), the subject is placed close to the
lens and is illuminated by a plane-parallel beam. The recording plate is
kept at the back focal plane of the lens, and the reference wave is incident
obliquely from the side.
L, H
I
I
1
p,
L
Q2
where Ar is the extent of the source, Z, its distance from the hologram,
and Z1the distance of a point on the object from the hologram.
Further, the equation describing the effect of the finite bandwidth Ah on
the resolution Au of the reconstructed image is given by the following
relation:
where Or is the angle of the reference beam with the normal to the hologram
plane (the subject and image are considered to be on-axis).
Equations (4.8) and (4.9) reveal that for small 2,,extension of the source
and finite bandwidth have little effect. Thus reconstructions from image-
plane holograms are sharp and achromatic even with white-light-extended
sources. However, for image points which are far outside the plane of the
hologram, the reconstruction shows color dispersion and blur.
In conclusion, image-plane holography is particularly suitable whenever
display of holograms, double-exposure interferometric holograms, or vibra-
tion-analysis holograms are desired in white light.
4.3.5.Color Holograms
Multicolor holography was first suggested by Leith and Upatnieks6 The
basic idea of color holography is to use three-color object and reference
beams to record a hologram on a single photographic plate. If the object
and the reference beams contain the three primary colors, the hologram
would, when illuminated with a beam identical to the reference beam,
reconstruct the object wave in full color.
There are basically two main techniques to record color holograms ( a )
holograms recorded as planar diffraction gratings and (b) holograms recor-
ded as volume diffraction gratings. The difference in response of the two
grating types is made evident by considering the grating equations:
d(sin i + sin 0) = A (planar grating), (4.10)
2d sin i =A (volume grating). (4.11)
The planar grating equation reveals that for a given A, diffracted light
would always be observed for any incidence. Thus the response is that
broad-band and special techniques are required to separate or eliminate
the spurious images which result from light of wavelength A l , diffracting
from the component hologram recorded with the wavelength A2.
The volume grating equation, on the other hand, states that the incident
and diffraction angles are equal, and thus only wavelengths satisfying Eq.
DIFFERENT TYPES O F HOLOGRAMS 181
(4.11) would be significantly diffracted. Thus the spurious images are sup-
pressed by the filtering action of the volume grating. Further, there is no
loss of resolution, contrast, or spatial frequency bandwidth.
We now describe the various techniques used to record color holograms.
4.3.5.1. Plane Hologram Techniques. There are several basic tech-
niques to make color plane holograms, as will be described next.
In the three-reference-beam method due to Leith and U p a t n i e k ~ ,the
~
object beam is a mixture of three primary colors. The reference beam is a
combination of three separate beams in the three different primary colors.
Each reference beam is inclined at a different angle to the recording plate.
In this way, the spurious or the cross-talk images resulting from diffraction
of light of wavelength A , and from the component hologram recorded at
wavelength A 2 are spatially separated.
This method, however, is not very practical because the field of view is
very much restricted. Furthermore, during reconstruction, the illuminating
beam must be accurately aligned so as to be identical to the reference beam.
The spatial multiplexing method is essentially due to Collier and
Penningt01-1.~'Here, we spatially multiplex, in a nonoverlapping manner,
the several holograms corresponding to the several colors used, i.e., the
various colored interference patterns are not allowed to overlap on the
recording medium. During reconstruction, the colored illuminating light is
incident in such a way that each component hologram is illuminated with
only that color light used to form it.
There are several experimental arrangements which could be used to
form such a hologram. One possible scheme is shown in Fig. 12. A mask
F consisting of strips of red, green, and blue filters is imaged onto the
hologram plane by a lens with the multiwavelength reference beam. The
image, which consists of discrete areas of single-wavelength light, acts as
the reference wave front. The light in each strip then interferes with the
light of the same color coming from the object, giving nonoverlapping
multiplicity of single-color holograms. The reconstruction step consists of
placing the hologram in register and illuminating through the same filter
mask.
Some resolution is sacrificed in this method, since each component
hologram will be smaller than the total recording. Also, this scheme results
in the production of some noise, because in a A ,-component hologram there
will also be light present of A2 and A 3 from the object, thereby decreasing
the diffraction efficiency of the hologram ( A , , A 2 , A 3 are the three primary
colors).
The coded-reference-beam method, also due to Collier and Pennington,30
involves coding of the reference beam. The amplitude and phase of the
reference wave are made to vary across the hologram plane in a significantly
different manner for each wavelength. During reconstruction the developed
plate has to be replaced precisely in the same position that it occupied
during the formation of the hologram.
A possible arrangement for recording such a hologram is shown in Fig.
13. The ground-glass screen D serves as the reference-beam coding plate.
Each colored component of the reference beam, when passing through the
diffuser, produces a complicated speckle pattern on the hologram plane.
The component speckle patterns are randomly different from each other
with respect to phase and intensity at a point, meaning that each colored
reference wave is uniquely coded. Collier and Pennington have noted that
when only light of wavelength 0.514 km was used, an excellent image was
obtained. However, when the wavelength was changed by only 0.0128 km
and the coding plate was illuminated in exactly the same way, only a uniform
noise was observed.
L2 0
front focal plane of the lens, a monochromatic image can be obtained. This
is because the image of the slit is formed on the same side of the plate as
the observer, thus allowing the observer to place his eyes directly at the
position of the slit image. But the size of the image becomes limited due
to vignetting by the lens aperture.
The pseudoscopic configuration is shown in Fig. 17(a). The geometry is
almost similar to the one used for orthoscopic configuration, with the only
difference that the reference wave is convergent instead of divergent. In the
reconstruction step [see Fig. 17(b)], the conjugate illumination with white
\
. \
'\
\
\
\
light is used to give real images of the slit through which the observer can
see the image of the object. Unfortunately, the real image that is obtained
positions the front of the object closest to the lens, farthest from the viewer.
Thus the viewer is presented with the familiar “inside-out” or reversed-depth
experience of a pseudoscopic image. However, there is an advantage that
the image is bright and sharp and is undistorted and unvignetted by the
lens aperture. Therefore, for applications where the depth perception is not
important, the pseudoscopic configuration would be more desirable.
The astigmatic recording scheme is yet another method for producing
rainbow holograms in one step and produces a good orthoscopic image
with large field of view in the horizontal direction. The recording is illustrated
in Fig. 18. The cylindrical lens L, images the object 0 in the vertical direction
near the hologram plate H. The slit S is also imaged but beyond the recording
plate. A divergent beam is used as the reference wave front. The image of
the object can be reconstructed with a white-light point source in the same
position as the reference source. Since in the vertical direction the hologram
is an image plane hologram, the image in this direction is sharp (see Section
4.3.4).
It is worth pointing out that in the one-step rainbow process using
conventional spherical lenses, the field of view is narrow in the horizontal
direction, which is a major drawback. This is primarily because of the
difflculty in getting a large aberration-free spherical lens. The astigmatic
recording has removed this obstacle, since long convex cylindrical lenses
are easily available.
The lensless method of Bahuguna and Santoyo4’ is particularly suitable
for transparencies. In this method, a master hologram of a diffuse slit is
recorded. By using conjugate illumination a wave converging to a slit is
reconstructed. This reconstructed wave is utilized to illuminate the object
transparency for creating a rainbow hologram. Since the master hologram
can be used for making a rainbow hologram of an unlimited number of
transparencies, this method is essentially a one step method. A variation
of the above method has been described by B a h ~ g u n a ~ ~ .
FIG. 19. Binary computed holograms to test aspherical surfaces (from Ref. 53).
S O M E APPLICATIONS OF HOLOGRAPHY 191
D D T
P H P H
(a) (b)
FIG. 21. Holographic interferogram of an aluminum block, tilted between two exposures.
194 HOLOGRAPHY
FIG. 22. Time-average interferograms of a 35-mm film can at its second resonance (from
Ref. 12).
SOME APPLICATIONS OF HOLOGRAPHY 195
A , A2 OBJECT
OBJECT
without any fringes. However, if the sandwiches are now formed by the
pairs of plates A l with B2 or by BI with A * , the reconstructed image of the
object will show the fringes.
The tilting of a hologram laterally moves the image of the object when
reconstructing the image of an object. However, if the sandwich pair is
tilted, the images of both holograms move almost together, producing only
a very small tilt of one image with respect to the other. This property of
sandwich holograms permits compensation or simulation of small rigid
movements of the body by means of small tiltings of the sandwich. In this
manner, the fringe pattern due to the object stresses is easily isolated from
any possible object rigid motion fringes.
where the upper sign is for the primary image and the lower sign is for the
conjugate image. Here m is the linear magnification of the hologram, is
the ratio of the two wavelengths, and Z,, Z,, and 2, are the z-coordinates
of the object-point, the reference-point source and the illuminating point
source respectively.
The above principles have been effectively utilized to form holographic
microscopes. For details of the theory, construction, and working, the reader
is referred to the literature, specially the papers by E l - S t ~ m El-Sum
,~~ and
Kirkpatri~kB , ~ a~ e ~ Van
, ~ ~Ligten and O ~ t e r b e r g and
, ~ ~ Thompson et al.25
s, s,
FIG. 24. Recording a reconstruction of an acoustic hologram
EXPERIMENTAL PROCEDURES I N HOLOGRAPHY 199
deviation from this condition and poor coherence of the source affect the
resolution in the image.
Normally, for a small object of a few centimeters in size, a 2-mW He-Ne
laser is quite adequate in terms of the required power. For larger objects,
one can use a higher-power He-Ne laser. If still higher power is needed,
one may want to use an argon laser which gives output in the range of a
few watts. To study transient phenomena, Brooks er have shown that
a pulsed laser is a must.
Because of the diffraction effects at the walls of the laser cavity, the
TEMoo mode (the usual mode of operation) may be modulated, giving a
speckled appearance to the expanded beam. This is especially the case when
the laser resonant cavity is formed with two large-radius mirrors. One can
get rid of this modulation by focusing the laser output through a microscopic
objective and putting a spatial filter (a fine pinhole) of diameter between
10 to 40 pm, depending on the magnification of the microscope objective
being used, at the focus of the beam. The output is a clean expanded beam.
Y
a
5 a;
1000
e m -
Moo-
WAVELENGTH OF LIGHT
F i c . 25. Spectral sensitivity curves for some Kodak holographic silver halide emulsions.
202 HOLOGRAPHY
WAVELENGTH OF LIGHT
FIG. 26. Spectral sensitivity curves for some Agfa-Gevaert holographic silver halide
emulsions.
Kodak
Agfd-Gevaert
EXPOSURE IN ERGS/cm'
FIG. 27. Curves of amplitude transmittance versus exposure, in a logarithmic scale, for
some holographic silver halide emulsions, using a wavelength of 6328 A.
204 H 0 LOG RAPH Y
reversed with respect to the case of the direct bleach method, the relief
however remaining at regions where development took place. Thus the relief
image tends to cancel the phase variations due to the silver halide, and
consequently the speckle noise is reduced.
Acknowledgments
The authors want to thank Prof. N. Abramson for his very valuable comments, Mr. Fernando
Mendoza for his help with the laboratory work, and Mr. Krishna Morales for the drawings
in this chapter.
References
1. D. Gabor, “A New Microscopic Principle,” Nature 161, 777 (1948).
2. D. Gabor, “Microscopy by Reconstructed Wavefronts,” !‘roc. Roy. Soc. A 197,454 (1949).
3. I). Gabor, “Microscopy by Reconstructed Wavefronts: 11,” Roc. Phys. Soc. B 64, 449
(1951).
4. E. N. Leith and J. Upatnieks, “Reconstructed Wavefronts and Communication Theory,”
’
J. Opt. Soc. Am. 52, 1123 (1962).
5. E. N. Leith and J. Upatnieks, “Wavefront Reconstruction with Continuous Tone Objects,”
J. Opt. Soc. Am. 53, 1377 (1963).
6. E. N. Leith and J. Upatnieks, “Wavefront Reconstruction with Diffused Illumination
and Three-dimensional Objects,” J. Opt. Soc. Am. 54, 1295 (1964).
7. Y. N. Denisyuk, “Photographic Reconstruction of the Optical Properties of an Object
in its Own Scattered Radiation Field,” Sou. Phys. Dokl. 7 , 543 (1962).
8. G. Lippmann, “La Photographie de Couleurs,” Compr. Rend. Acad. Sci. Paris 112, 2 7 4
(1891).
9. S. A. Benton, “Hologram Reconstruction with Extended Incoherent Sources,” J. Opt.
Soc. Am. 59, 1545A (1969).
REFERENCES 207
37. B. L. Sopori and W. S. C. Chang, “3-D Hologram Synthesis from 2-D Pictures,” Appl.
Opt. 12, 419 (1973).
38. N. D. Haig, “Three-Dimensional Holograms by Rotational Multiplexing of Two-
Dimensional Films,” Appl. Opt. 12, 419 (1973).
39. T. Kasahara et al., “Stereo-Radiography using Holographic Techniques,” Jap. J. Appl.
Phys. 8, 124 (1969).
40. G. Groh, and M. Koch, “3-D Display of X-ray Images by Means of Holography,” Appl.
Opt. 9, 775 (1970).
41, M. Grossman, P. Meyrucys, and J. Fontaine, “Three-Dimensional Holographic Synthesis
of X-ray Pictures”, in Holography in Medicine and Biology (G. Von Bally, ed.), Springer-
Verlag. Berlin, 1979.
42. H. Chen and F. T. S. Yu,“One-Step Rainbow Hologram,” Opfics Left. 2, 85 (1978).
43. H. Chen, “Color Blur of the Rainbow Hologram,” Appl. Opt. 17, 3290 (1978).
44. H. Chen, “Astigmatic One-Step Rainbow Hologram Process,” Appl. Opt. 18, 3728
(1979).
45. H. Chen, F. T. S. Yu, and A. M. Tai, “One-Step Rainbow Holography Old and New,”
in Recent Advances in Holography, Proc. SOC.Phot. Opt. Instr. Eng. 215, 162 (1980).
46. S. A. Benton, H. S. Mingace, Jr., a n d W. R. Walter, “One-Step White-Light Transmission
Holography,” in Recent Advances in Holography, Proc. SOC. Phot. Opt. Instr. Eng. 215,
156 (1980).
47. R. D. Bahuguna and F. M. Santoyo, “Simple Rainbow Holographic Technique for Two
Dimensional Transparencies,” Optics Lett. 9 381 ( 1984).
48. R. D. Bahuguna, “Simple Lensless One-step Rainbow Holographic Technique for 2-D
Transparencies, Applied Opt. 24, 61 1 (1985)
49. R. L. Fusek and L. Huff, “Use of a Holographic Lens for Producing Holographic
Stereograms,” in Recent Advances in Holography, Proc. SOC. Phot. Opr. Instr. Eng. 215,
32 (1980).
50. B. R. Brown a n d A. W. Lohmann, “Complex Spatial Filtering with Binary Mask,” Appl.
O p f 5, 967 (1966).
51. W. H. Lee, “Computer Generated Holograms: Techniques a n d Applications,” in Progress
in Optics (E. Wolf, ed.), Vol. 16, North Holland, Amsterdam, 1978.
52. A. W. Lohmann and D. P. Paris, “Binary Fraunhofer Holograms, Generated by Com-
puter,’’ Appl. Opt. 6, 1739 (1967).
53. J. Wyant, “Holographic and Moire Techniques,” in Optical Shop Tesfing (D. Malacara,
ed.), Wiley, New York, 1978.
54. J. S. Loomis, “Computer Generated Holography a n d Optical Testing,” in Recent Advances
in Holography, Proc. Soc. Phot. Opt. Instr. Eng. 215, 59 (1980).
55. L. B. Lesem, P. M. Hirsch, and J. A. Jordan Jr., “The Kinoform: A New Wavefront
Reconstruction Device,” IBM J. Res. Develop. 13, 150 (1969).
56. F. Broder and D. Malacara, “Holographic Interferometer to Test Optical Surfaces,” Appl.
O p f . 14, 2280 (1975).
57. J. C. Fouere a n d D. Malacara, “Holographic Radial Shear Interferometer,” Appl. Opt.
13, 2035 (1974).
58. D. Malacara and S. Mallick, “Holographic Lateral Shear Interferometer,” Appl. Opt. 15,
2695 (1976).
59. E. B. Aleksandrov and A. M. Bonch-Bruevich, “Investigation of Surface Strains by the
Hologram Technique,” Sou. Phys.-Tech. Phys. 12, 258 (1967).
60. A. E. Ennos, “Measurements of In-Plane Surface Strain by Hologram Interferometry,”
J. Sci. Instrum., Ser. 11, 1, 731 (1968).
REFERENCES 209
61. C. Froehly, J . Monnerer, J. Pasteur, and J. Ch. Vienot, “Etude des Faibles DCplacements
d’Objects Opaques et de la Distorsion Optique dans les Lasers a Solide par InterfCrometrie
Holographique,” Optica Acta 16, 343 (1969).
62. T. Tsuruta, N. Shiotake, and Y. Itoh, “Formation and Localization of Holographically
Produced Interference Fringes,” Optica Acta 16, 723 (1969).
63. J . E. Sollid, “Holographic Interferometry Applied to Meassurements of Small Static
Displacements of Diffusely Reflecting Surfaces,” Appl. Opt. 8, 1587 (1969).
64. N. E. Molin and K. A. Stetson, “Fringe Localization in Hologram Interferometry of
Mutually Independent and Dependent Rotations Around Orthogonal, Non-Intersecting
Axes,” Optik 33, 399 (1971).
65. N. L. Hecht, J. E. Minardi, D. Lewis, and R. L. Fusek, “Quantitative Theory for Predicting
Fringe Pattern Formation in Holographic Interferometry,” Appl. Opt. 12, 2665 (1973).
66. N. Abramson, “The Holo-Diagram: A Practical Device for Making and Evaluating
Hologams,” Appl. Opt. 8, 1235 (1969).
67. N. Abramson, “The Holo-Diagram. 11: A Practical Device for Information Retrieval in
Hologram Interferometry,” Appl. Opt. 9, 97 (1970).
68. N. Abramson, “The Holo-Diagram. 111: A Practical Device for Predicting Fringe Patterns
in Hologram Interferometry,” Appl. Opt. 9, 231 1 (1970).
69. N. Abramson, “The Holo-Diagram. IV: A Practical Device for Simulating Fringe Patterns
in Hologram Interferometry,” Appl. Opt. 10, 2155 (1971).
70. N. Abramson, “The Holo-Diagram. V: A Device for Practical Interpreting Hologram
Interference Fringes,” Appl. Opt. 11, 1143 (1972).
71. R. L. Powell and K. A. Stetson, “lnterferometric Vibration Analysis of Wavefront
Reconstruction,” J. Opt. SOC.Am. 55, 1593 (1965).
72. N. Abramson, “Sandwich Hologram Interferometry: A New Dimension in Holographic
Comparison,” Appl. Opt. 13, 2019 (1974).
73. N. Abramson, “Sandwich Hologram Interferometry: 2. Some Practical Calculations,”
Appl. Opt. 14, 981 (1975).
74. N. Abramson, “Sandwich Hologram Interferometry: 3. Contouring,” Appl. Opt. 15, 200
(1976).
75. N. Abramson, “Sandwich Hologram Interferometry: 4. Holographic Studies of Two
Milling Machines,’’ Appl. Opt. 16, 2521 (1978).
76. H. M. A. El-Sum, Reconstructed Waoefront Microscopy, Ph.D. Thesis, Stanford University,
November 1952.
77. H. M. A. El-Sum and P. Kirkpatrick, “Microscopy by Reconstructed Wavefronts,” Phys.
Rev. 85, 763 (1952).
78. A. V. Baez, “A study in Diffraction Microscopy with Special References to X-Rays,’’ 1.
Opt. SOC.Am. 42, 756 (1952).
79. R. F. Van Ligten and H. Osterberg, “Holographic Microscopy,” Nature 211, 282 (1966).
80. E. N. Leith and J. Upatnieks, “Holographic Imagery Through Diffusing Media,” J. Opt.
Sac. Am. 56, 523 (1966).
81. J. W. Goodman, “Wavefront Reconstruction Imaging Through Random Media,” Appl.
Phys. Lett. 8, 31 1 (1966).
82. A. Vander Lugt, F. B. Rotz, and A. Klooster, Jr., “Character Reading by Optical Spatial
Filtering,’’ in Optical and Electro-Optical Information Processing (J. T. Tipett el a/, eds.),
p. 125, MIT Press, Cambridge, Mass., 1965.
83. D. H. Close, “Holographic Optical Elements,” Opt. Eng. 14, 408 (1975).
84. W. C. Sweatt, “Describing Holographic Optical Elements as Lenses,” J. Opt. Sac. Am.
67, 803 (1977).
210 HOLOGRAPHY
85. J. Upatnieks, A. Vander Lugt, and E. N. Leith, “Correction of Lens Aberrations by Means
of Holograms,” Appl. Opt. 5, 589 (1966).
86. R. V. Mueller and N. K. Sheridon, “Sound Holograms and Optical Reconstruction,”
Appl. Phys. Lett. 11, 328 (1966).
87. J. D. Young and J. E. Wolf, “A New Technique for Acoustic Holography,” Appl. Phys.
Lett. 11, 294 (1967).
88. K. Preston, Jr. and J. L. Krenzer, “Ultrasonic Imaging Using a Synthetic Holographic
Technique,” Appl. Phys. Lett. 10, 150 (1967).
89. P. S. Green, ed., Acoustical Holography, Plenum Press, New York, 1974.
90. B. P. Hildebrand and B. B. Brenden, A n Introduction to Acoustical Holography, Plenum
Press, New York, 1974.
91. 8. B. Byron and H. D. Collins, “Acoustical Holography with Scanned Hologram Systems”
in Holographic Nondestructive Testing (R. K. Erfd, ed.), p. 405 Academic Press, New
York, 1974.
92. D. Malacara, “Optical and Electronic Processing of Medical Images,” in Progress in
Optics, Vol. 22, Chap. 1 (E. Wolf, ed.), North-Holland Publ., Amsterdam, 1985.
93. E. N. Leith and J. Upatnieks, “Holography with Achromatic Fringe Systems,” J. Opt.
SOC.A m . 57, 975 (1967).
94. R. E. Brooks, L. 0. Heflinger, and R. F. Wuerker, “Pulsed Laser Holograms,” IEEE J.
Quant. Electr. QE-2, 275 (1966).
95. H. M. Smith, ed., Holographic Recording Materials, Springer-Verlag. Berlin, 1977.
96. J. W. Gladden and R. D. Leighty, “Recording Media,” in Handbook ofoptical Holography
(H. J. Caulfield, ed.), Academic Press, New York, 1979.
97. R. A. Bartolini, H. A. Weakliem, and B. F. Williams, “Review and Analysis of Optical
Recording Media,” Opt. Eng. 15, 99 (1976).
98. M. Chang, “Dichromated Gelatin of Improved Optical Quality,” Appl. Opt. 10, 2551
(1971).
99. R. K. Curran and T. A. Shankoff, “The Mechanism of Hologram Formation in Dichro-
mated Gelatin,” Appl. Opt. 9, 1651 (1970).
100. R. A. Bartolini, “Characteristics of Relief Phase Holograms Recorded in Photoresists,”
Appl. Opt. 13, 129 (1974).
101. R. A. Bartolini, “Improved Development for Holograms Recorded in Photoresists,” Appl.
Opt. 11, 1275 (1972).
102. R. A. Bartolini, “Photoresists,” in Holographic Recording Materials, ( H . M. Smith, ed.),
Springer-Verlag, Berlin, 1977.
103. S. L. Norman and M. P. Singh, “Spectral Sensitivity and Linearity of Shipley AZ-1350
J Photoresist,” Appl. Opt. 14, 818 (1975).
104. J. P. Kirk, “Hologram in Photochromic Glass,” Appl. Opt. 5, 1684 (1966).
105. R. J. Araujo, “Recent Advances in Display Media,” in N A S A Symposium Proceedings,
p. 63, Cambridge, Mass., 1967.
106. D. R. Bosomworth and H. J. Gerritzen, “Thick Holograms in Photochromic Materials,”
Appl. Opt. 7, 95 (1968).
107. A. L. Mikaeliane, A. P. Axenchikov, V. 1. Brobinev, E. H. Gulaniane, and V. V. Shatun,
“Holograms on Photochromic Films,” IEEE J. Quant. Elect. QE-4, 757 (1968).
108. A. S. Mackin, “Holographic Recording on Electron Beam Colored Sodium Chloride
Crystals,” Appl. Opt. 9, 1658 (1970).
109. D. L. Ross, “Photochromic Indigoids. 111: A Photochromic Element Based on the cis-trans
Photoisomerization of a Thioindigo Dye,” Appl. Opt. 10, 571 (1971).
110. J. A. Jenney, “Holographic Recording with Photopolymers,” J. Opt. SOC.A m . 60, 1155
(1970).
REFERENCES 21 1
William L. Wolfe
Optical Sciences Center
University of Arizona
Tucson, Arizona 85721
5.1. Introduction
Table I presents a list of all the symbols used in this chapter along with
their definitions and units associated with them. The units are given not
just to define the dimensionality, but they are intended to help clarify the
meaning. SI units are generally used, although frequently used units like
centimeters and micrometers are indicated rather than meters where it is
appropriate. The symbols are listed alphabetically, but the ensuing dis-
cussion treats them topically.
Perhaps the simplest of the variables to be discussed is the independent
spectral variable. Most of us think in terms of the optical wavelength A of
the radiation to be measured. It is also useful to think in terms of the
frequency which can be expressed in several ways. The wave number i j is
just the reciprocal of the wavelength. The frequency v is the speed of light
times that ( v = c/A), and the radian frequency w is 27rv. Similarly, k, the
radian wavenumber is 27rl or 2 7 r l A . The nondimensional frequency x is
c2/AT (where c2 is the second radiation constant defined below, and T is
the temperature), a convenient independent variable for many calculations.
The spectral distribution of blackbody radiation can be given in terms of
any of these variables. In order to treat this somewhat generally, the symbol
y is used to represent any of these spectral variables.
Radiation transfer is ultimately expressed in terms of a quantity like
power, the time rate of energy. An equivalent concept is the time rate of
the transfer of a number of photons or of the amount of power which evokes
a visual response. Each of these is a flux and is denoted by the symbol P.
To distinguish between them, the subscripts u, q, and u respectively, are
recommended by the International Illumination Society.’ I have not used
a subscript when the symbol can refer to any of these or when it refers
specifically to energy transfer. I believe, no confusion will result. The rest
of the symbols which are used to describe radiative transfer are L, M , E,
I, and S. These are, respectively, the radiance or flux per unit normal area
and solid angle; the irradiance or incidance or incident flux per unit area;
S Y M B O L S , UNITS, A N D NOMENCLATURE 215
the emittance or exitance (emitted flux per unit area); the intensity or flux
emitted per solid angle-generally from a subresolution source; and the
radiance per unit path length in a medium which scatters or is a volume
emitter. This is also called the volume radiance. In several sections I have
dealt with irradiance and electric field strength in the same equation and
have used U for the electric field. These equations have also involved
m U U * for which I have used 9.I would prefer using I and E for these,
and J and H for intensity and irradiance, but that would be contrary to
the recommendations of the International Commission on Illumination
(CIE). Several writers have proposed nomenclature schemes to make these
definitions logical, consistent, and easy. Table I1 is a summary of these.
The fluometry scheme of Jones has as its main feature the unifying concept
that these flux terms are really all geometric concepts. He therefore used
sterance, incidance, exitance, intensity, and sterisent as generic terms to be
modified by “radiant,” “photon,” and ‘‘luminous.’’ Nicodemus,’ largely in
an effort to overcome the ambiguity of “intensity” (it is variously used to
mean E, M , I, and L by different scientists), has coined the geometric terms
of areance, and pointance for E or M and I, respectively. Worthing,” years
ago proposed that “-ancy” be used to mean “per unit area.” He went further
and proposed that the transmission, absorption, and reflection properties
be specified with “-ance” endings when a sample is specified, and “-ivity”
endings when a substance is specified.” This, of course, makes emittance
an ambiguous term if one uses the “conventional” terms. The nomenclature
system has not yet been proposed. The concept of path radiance may be
somewhat foreign to some readers. In a scattering medium or volume
radiator the radiance is not constant. It can increase with path length,
because each element of path contributes more radiation to the total. It
thus has units of W cm-* sr-’ cm-I, where the final cm-’ is the (reciprocal
of the) unit of path length. There is no recommended symbol for this. I
have chosen S for this quantity because it agrees with “sterisent,” and many
other appropriate letters have been preempted. I have also replaced with
P (to stand for phlux) for ease of typing. Finally, to indicate any of these
radiometric quantities, I have used the symbol R.
The atomic constants are listed in a conventional way: cI and c2 are the
first and second radiation constants, equal to 27rc2h and hc/ k g , respectively.
The speed of light in vacuo is c; Planck’s constant, h ; the Stefan-Boltzmann
constant a ; that for photon distributions, aq;and the Boltzmann constant
k g (because k is the radian wave number).
Some of the common symbols are x, y, and z for general distances. These
do not get confused with the spectral variables. Angles are denoted by 6
and 4 ; solid angles by R; areas by A; time by 1 ; temperature by T ; integers
by m, P, 4.
216 PHOTOMETRY A N D R A D I O M E T R Y
a, b, c constants -
A area mZ
B effective bandwidth Hz
BRDF Bidirectional Reflectance Distribution Function sr '
b base of a prism m
C constant -
C speed of light in a vacuum m s-'
CI first radiation constant w cm2 s-I
c2 second radiation constant cm K
D diameter m
D* specific detectivity cm Hz'/'W I
W width m
X nondimensional frequency or coordinate length - or m
Y general spectral variable or coordinate length various or m
I coordinate length m
SYMBOLS, UNITS, A N D NOMENCLATURE 217
TABLEI. (coni.)
a absorptivity or absorptance
Y absorption coefficient
E emissivity
1) quantum efficiency
1)" luminous efficiency
8 angle
A wavelength
v frequency
v wavenumber
P reflectance or reflectivity
U Stefan Boltzmann constant
u(A) slit function
T transmittance or transmissivity
@ angle
R solid angle
w radian frequency
Subscripts
a, b indicating different spatial points
a atmosphere
C collector or optical element
d detector
e emitted
h hemispherical
I incident
q photon
S source
U energy
V visible
1.2.i first, second ith of a group
W infinite (number of reflection)
Superscripts
BB blackbody
h hemispherical
m maximum, molecular
0 optics
P projected
time rate
- weighted average
TABLE11. Radiometric Nomenclature
~~ ~ ~
Field
Quantity Jones Nicodemus Worthing Theorists Meteorologists
Mqk = (f) nk
(5.3)
These are purely geometric relationships and apply to other spectral distribu-
tions. One important and very useful characteristic of the radiation is that
the geometrical characteristics are independent of the spectral distributions.
There are basically four spectral distributions: power vs frequency, power
vs wavelength, photons vs frequency, and photons vs wavelength.
(5.5)
(5.7)
(5.9)
The usual expression for the distribution of power vs wavelength can then
be found from
c2 hc hu
x=-=---
AT A k R T - kBT'
and
c 1.43883
AmT = L = - - - cm K =2898 [ p m K]. (511)
Xm Xrn
This is the common expression for the Wien displacement law." The
constants for the other forms are given in Table 111.
The distribution can also be expressed usefully in terms of the dimension-
less variable x by applying the technique described above:
Function
Dependent Independent
Variable Variable m Xmax Rmax
Any of the distributions will be of this form but with a different exponent
on the relative dimensionless frequency u (=x/x,).
The radiance can change either as a result of a change in temperature or
a change in the spectral variable.
Here it can be seen that the factor xex(e' - 1)-' is the factor for spectral
changes. Each variation individually can be viewed as an exponential
relation, as pointed out in part by
R,. = T P , (5.14)
where
p = xe'(e'- I)-'
and
R, = y 4 ,
where
* + xe '( e* - 1
q = rn )-I.
5.3.3.Generalized Integrals
The flux and change in flux over a finite spectral band are found by
integrating the appropriate monochromatic function over the spectral band.
The functions can be the radiance, photon sterance, or their respective
changes with temperature:
(5.15)
(5.16)
(5.18)
rn-4ee-m"[(rn~)3+3(rnx)2+6rnx+6],
e~mx[(rn~)2+2m~+2],
=-(-)
2c2h T 1 rn~4e-"A[(rn~)4+4(rn~)3+12(rnx)2
T c2 m;l
+24rnx+24],
C rn-' e-mx[(mx)3+3(rnx)2+6rnx+6].
5.3.4.Infinite Integrals
The value of the integral of a blackbody function is of course independent
of the spectral variable, and the simplest variable to use is the non-
dimensional frequency x. Then the total radiance is
(5.19)
SIMPLE RADIATIVE T R A N S F E R 223
2kiT' T'
=2.4041-- - 1.5202 x 10" -
c2h3 ?l
u T'
--.
- (5.20)
?l
Both of these are solved by resourse to zeta functions. This latter expression
can also be written as
L (5.21)
' = &
2.15 k B T '
This would indicate that the average value of the energy of a photon is
2.15 k B T .
This section deals with the radiative transfer in nonscattering media. The
primary purpose is to describe the characteristics of the radiometric variables
rather than the physical processes which give rise to absorption and emission.
The first part of the description then applies to transfer in nonabsorbing,
nonscattering, nonemitting media, and the second to absorbing and emitting
but nonscattering media.
224 PHOTOMETRY A N D R A D I O M E T R Y
8.9
0.8
0.7
8.6
8.5
0.4
0.3
8.2
0.1
0 1
0 1 2 3 4 5
NORNALIZED NONDIMENSIONAL FREQUENCY
FIG. 1. Normalized blackbody energy distribution as a function of normalized non-
dimensional frequency of the radiation.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
NORMALIZED NONDIMENSIONQL FREQUENCY
FIG. 2. Normalized blackbody photon distribution as a function of normalized non-
dimensional frequency of the radiation.
S I M P L E RADIATIVE TRANSFER 225
1 -
0.9 -
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2
0.1
0
NORFlALIZED NONDIMENSIONAL FREQUENCY
FIG.3. Normalized blackbody energy contrast distribution as a function of normalized
nondimensional frequency o f the radiation.
0.9
0.8
0.7
0.6
e. 5
0.4
8.3
0.2
0.1
0
NORNALIZED NONDIIIEHSIONAL FREQUENCY
FIG. 4. Normalized blackbody photon contrast distribution as a function o f normalized
nondimensional frequency o f the radiation.
226 P H o r o M E r u Y A N D RADIOMETRY
8.9
9.8
0.7
0.6
0.5
0.4
0.3
8.2
0.1
0.9
0.8
0.7
8.6
0.5
0.4
0.3
8.2
8.1
8
0 1 2 3 4 5 6 7 8 9 18
NORMALIZED tiONDIMENSIONAL FREQUENCY
FIG. 6 . Normalized blackbody photon integral as a function of normalized nondimensional
frequency of the radiation.
SIMPLE RADIATIVE T R A N S F E R 227
0.9
0.8
0.7
8.6
6.5
0.4
0.3
1:: / * , . . . . .
0 ~
1 -
0.9 .
0.8 -
8.7 -
0.6 -
0.5 -
0.4 -
0.3 -
:::t/
0
B 1 2 3 4 5
. . . . . . .
6 7 8 9 10
,
The expression on the right was written four different ways to illustrate the
different interpretations which have been applied to it. The first can be
viewed as a variation of the definition of radiance: since radiance is power
per projected area and solid angle, then power is radiance times projected
area and solid angle. The second form shows that the solid angle can be
written as a projected area divided by a distance squared. This form gives
a physical and symmetrical expression for the transfer: the product of two
projected areas divided by the square of the distance between them. The
third form just indicates that in this chapter (and other places in the
literature) the projected area is indicated with a subscript. The final form
uses the “p” to indicate a “projected solid angle.” This is a nonphysical
quantity, which when multiplied by an area and a radiance gives the flux
transfer.
The expression can be used to obtain the relation between radiance and
flux density for Lambertian bodies and the very useful configuration factor
that is found in extensive tables; see Sparrow and Cess2’ for more on this.
The relationship between Land M (regardless of whether photons, power,
or any other quantity is being transferred) is M = TL. One can imagine that
a differential area of a radiator which has the same radiance in all directions
FIG. 9. The geometry of radiative transfer. Radiation travels from a differential element
dA, of surface A , along R , , to dA, of A,. The angles between R , , and the surface normals
are 0, and 0,.
S I M P L E RADIATIVE T R A N S F E R 229
is at the center of a sphere. The flux per unit area d A , of this source radiated
into a hemisphere is given by
L d A , cos 8, dA2 cos e2
M=-= (5.23)
dA I x’dA,
The radiance is independent of angle, and O2 is zero, so that
M =L lo2=cos 8 sin e de d 4
= TL. (5.24)
It should be noted that the roles of emitter and receiver can be reversed
either by use of Helmholtz’s reciprocity theorem or by direct integration.
This relationship can be used in connection with the flux transfer theorem
to obtain a configuration factor (also called angle factor and view factor).
On a differential basis, one has
L I 2dA, cos 8, dA, cos O2
d R I 2= (5.25)
X-
The configuration factor K I 2is the fraction of radiation emitted by the first
surface that is received by the second. Thus,
(5.27)
Extended sources, of course, are not the only type of radiator one
encounters in practice. The second major class is generally called “point
sources,” although in theory there is no such.thing (at least one that radiates
appreciably). What is usually meant is a subresolution source, one which
is small compared to the image of the detector on the source. For such a
source, radiance is not a good descriptor, but the integral of radiance over
the area of source is appropriate. Such a quantity is usually called intensity
and labeled I by radiometrists.
It can be defined in two ways:
(5.29)
The radiance is a function of both the direction and the location on the
surface, whereas intensity is a function only of direction.
Before considering transfer in media that absorb but have inappreciable
emission, it is useful to consider two brief calculations involving flux on
the detector of a radiometer. The first calculation involves an extended
source, the second a subresolution one.
Assume that a radiometer views an extended source of area A , at a
distance x with radiance L. The radiometer has a collector area A , and
focal length f, detector area A , optical speed or Flnumber F, and trans-
mission r. Then, if all solid angles can be approximated by areas divided
by distances, the flux is given by
(5.31)
(5.32)
The total flux o n the detector must be obtained from one of the last two
expressions. The last expression on the right is in terms of the collector
area and the field of view of the detector. To keep a constant flux on the
detector, these two quantities must be varied inversely. Usually one wants
as small a detector field as one can get, but this must be accompanied by
a concomitant increase in collector area (even without diffraction consider-
ations). The penultimate expression shows that the flux concentration on
S I M P L E RADIATIVE T R A N S F E R 23 1
FIG. 10. Geometry of radiative transfer in a volume radiator, shown as a side view. An
annular volume element radiates a distance x/cosO to the surface.
S
L=-. (5.35)
Y
The basic assumptions in this derivation are that the exponential law of
absorption is valid and that the volume emission is the same in every
direction (an isotropic radiator). Integration over the hemispherical projec-
ted solid angle yields
7TS
M=-. (5.36)
Y
(5.37)
This gives the volume radiance in terms of the absorption coefficient y and
refractive index of the medium and the “standard” surface radiance. The
external radiation will be in a solid angle that is larger by n 2 than the
internal angle and is reduced by the surface reflection losses. Further,
radiation cannot exit at angles which exceed the critical angle.
If the partially transparent medium is of a finite rather than semi-infinite
thickness denoted by X , then the integration yields a radiance given by
L = n2~BB(1- e - ~ . ~ / ~ ~ s @ ) . (5.38)
As Gardon points out, the radiance from the semi-infinite plate follows
Lambert’s law, whereas that for the finite plate does not.
Several interesting characteristics of radiation in partially transparent
media can be developed from this beginning: the expression for radiation
from plane parallel plates, for directional and hemispherical emissivities,
Bouguer’s law for finite radiators, and the technique of inversion applied
to the atmosphere and to radiance solids like glass.
Integration of the radiance over the hemispherical projected solid angle
yields the following:
M = n 2 ~ n B [ 2 - e - y ‘ ( 1- y r ) - ( y r ) * ~ i ( - y t ) I , (5.39)
where
and
where
ei = angle of incidence
ee= angle of emission
234 PHOTOMETRY A N D RADIOMETRY
(5.45)
(5.47)
One can apply the same techniques of integration, but for the sake of
variety it will be done in terms of flux density rather than radiance. In a
small solid angle dR, the flux density is (1/47r)SdxdR in a small element
dx. This quantity multiplied by the transmission is integrated to the surface
--SdR
- (1-T). (5.48)
47rY
This is the radiant exitance just inside the surface; outside, it is
(5.49)
If the body is opaque, then T = 0, and Kirchhoff's law provides the following
result:
1dR MHH
M =- (1 - p ) =- ( 1 - p ) dR. (5.50)
4ffY 7r
From this, one can write that S / y = 47rM BB. Therefore, the nonreflected
volume radiation gives rise to one component of radiant exitance:
M I = ( 1 - p ) ( 1 - T ) M B B r - 'd R . (5.51)
That which is reflected from the top and the bottom and emerges again is
M2= P ~ T ~ M , . (5.52)
After one more trip, one has
M 3 = p27'M2 = p 4 r 4 M I . (5.53)
The sum of the infinite geometric series is
00
The emissivity is the same as the absorptivity, and Kirchhoff's law applies
to semitransparent substances.
There is one difference between this development and McMahon's: the
radiant exitance is derived from a consideration of radiance flowing into a
projected hemisphere. This changes the relationship between volume radi-
ance and radiant exitance by a factor of 2.
236 PHOTOMETRY A N D RADIOMETRY
IhoL"" dT.
the absorption region of a gas whose mixing ratio is known. Two atmospheric
gases meet this requirement: C O , and 0 2 .Kaplan3" proposed using the
667 cm-' (15 pm) band of C 0 2 for this purpose, and experiments have been
performed based on that ~uggestion.'~
Once the temperature profile has been established, other spectral bands
can be used for determining the profiles of other gases. The radiative transfer
equation for monochromatic radiation can be written in terms of the radiance
at a given frequency F and zenith angle 8 as'
dL(F, e)={-L(F, e ) + LBB[V,T(z)])yrn(V,z) p( z) secedz, (5.58)
where
L(F,O)= E(F)LBB(F, T ) r ( F ) -
I,' d(F' )'
LBB(F' T , d(1og P)
The radiance at the top of the atmosphere L(F,O) is found from the
d(log P ) . (5.59)
Lo=&rL-
I L
dr
d(logP)
d(1og P ) .
in the expression for each of the L( V ) until a best fit is found to the measured
values of Lo( F). Various techniques have been used to obtain convergence."
238 PHOTOMETRY A N D RADIOMETRY
Inversion
R I d iorondr
Temperature (OK)
FIG. 1 1 . Temperature profile derived from IRIS data by inversion of the radiative transfer
equation; data were recorded near Brownsville, Texas, 22 April 1969, at 1937 GMT. Data
from a radiosonde ascent at 1800 GMT are shown for comparison.
RADIOMETRIC TEMPERATURE MEASUREMENTS 239
V= lorn
~ ( A ) E ( A ) T ( A ) c , A - ' ( ~-" I)-' dA. (5.62)
T=(&)"4. (5.63)
band, constant emissivity but finite spectral band, etc. For constant
emissivity and infinite spectral band, the expression is
V=9 ~ a T
T=(&)"' (5.64)
One is usually in possession of the values for 9 and V but must assume
that E is 1. Then the relative error is
1 - &-1/4
The error is different for different values of E and is negative for positive
values of E. This means that the error is such that a higher temperature is
inferred than is really the case.
The next complication is that 9 ( A ) = T ( A ) = & ( A ) = 1, but that the band-
width is finite. Then the relative error is
1-
IA, c,A-'(e" - l)-' dA
(5.65)
aT4
A further complication is that % ( A ) and T are constant. One can usually
measure 9(A ) to sufficient accuracy. Then
V= J A.4
~ "l)-' dA.
~ ( A ) C , A - ~ (- (5.66)
(5.67)
(5.72)
(5.74)
then
(5.75)
In the region for which the Planck function must be used, the temperature
cannot be found explicitly. The voltage ratio is
(5.76)
242 PHOTOMETRY A N D RADIOMETRY
Generally one chooses the two spectral bands such that = E ? , but only a
knowledge of the ratio is necessary. The measurements are usually made
through an atmosphere of transmission T, so that an ET product in each
band should be used instead of just E . Then, since xI= c2/A I T = x2 = c2/A2T,
there is only one value of T which satisfies the set of conditions.
The errors resulting from an inaccurate assessment of the ratio of the ET
product in the two bands can be made by assuming the ratio is 1 and by
calculating the T values for other values of the ratio. The error can be
evaluated by calculating the value of T, using the proper ET ratio, then
calculating T for an ET ratio of 1. The relative difference between the two
temperatures is the error. As usual, the expression can be formulated in
terms of the dimensionless variable x. The solution, however, is a function
of both x1 and x2, not the ratio or the sum.
The analysis for finite spectral bands proceeds in almost the same way
as for the monochromatic analysis. The radiation received in each of the
spectral bands is a combination of the measured spot and any background
in the field of view. Now, however, for the radiances, one must write integral
expressions. Thus, for a spot of area A
V=
A [ E T ( A ) A - ~ ( ~ "l)-'
- dA + ( 1 - A )
I,, & ~ ( h ) A - ~ ( el ') - ' dA
A f'
Ah2
E T ( A ) A - ~ ( ~-" 1)-' dA +(1 - A )
I,,,' E T ( A ) A - ~ ( ~-' I ) - ' dA '
(5.77)
where
V = voltage ratio
Ah I = first spectral band
Ah2 = second spectral band
XI = C 2 / A / TI
XI = CI/A/TZ
-
[ ~ ( h ) A - ~ ( e " - l )dA
-'
&T = ( &T),R = (5.78)
[ K 4 ( e "- l ) - ' dh '
f
V
--
0
U
.-
> 0 3 -
2
0 2 -
01 -
0 3
a 05 I
Background temperature 0 s percentape of cargef temperalure Mortmum
error occurs when Dockground I S 7 5 % of torpet
1. The technique of using the ratio of radiation in two spectral bands can
be used to measure temperature differences over the surface.33
To emphasize the main point and to avoid getting lost in the details, we
shall use the calculations based on very narrow bands and shall ignore the
contributions of the background. Then the voltage ratio from the two spectral
measurements at any scanned point is given by (where responsivities have
been cancelled)
(5.79)
where
c2
x, =-
A~ T’
x --. c2
2 - ~ Z ~
RADIOMETRIC TEMPERATURE M E A S U R E M E N T S 245
The difference in this ratio when the system looks at two different spatial
points a and b can be written
(5.80)
etc.
In this difference expression, A l and A2 are both known. If the emissivity
ratio E ~ / Eis ~the same for both spatial positions, then one has
(5.81)
The temperature difference can be found uniquely, provided that the ratio
of emissivities is spatially invariant. Dereniak has described this technique
in more detail.34
It seems important to emphasize the difference between two measurement
situations. The ratio of radiances can be used to measure temperature if
the body is gray-or at least if the emissivity is the same in two different
spectral bands. However, the ratio of radiances can be used to measure the
difference in temperature of two spatially separated points if the ratio of
emissivity in the two spectral bands is spatially invariant-even if the
emissivities are different in the two spectral bands.
How does one choose the spectral bands? They should be narrow enough
that the theory applies, broad enough to get satisfactory signal-to-noise
ratio, far enough apart to reduce the errors, and close enough together that
the emissivities are highly correlated.
246 PHOTOMETRY A N D R A D I O M E T R Y
Devices used to measure the amount of radiant flux are generally called
radiometers, but there are radiometers of various sorts and kinds, and even
with different names. This section describes some of these different types
and their salient characteristics. Devices called radiometers by most people
measure the amount of energetic flux in a relatively wide spectral band and
with a flat response. Photometers measure the flux in the visible spectral
region with a response that matches that of the eye. Pyrometers are devices
which measure the temperature of relatively hot bodies by virtue of some
radiation property of the body. Pyrheliometers measure the radiation from
the sun. Some measure incident flux, and others measure net flux. Radio-
meters which have a very narrow spectral bandpass are generally called
spectroradiometers, although some are also inconsistently called spec-
trophotometers. In addition to these classifications, which are generally
spectral in nature, radiometers can be classified as to whether they measure
the spatial and/or temporal distribution of radiation. Spatial radiometers
can be line scanners or can scan in two dimensions, creating the information
necessary for constructing an image. Whether an image is created or not,
these are classified as imaging radiometers. Another distinction which can
be made is whether the instrument measures the average or dc value of the
flux or whether it measures the high-frequency fluctuations. The latter is
often called a temporal or fluctuation radiometer. The most recent types of
radiometers are those which detect coherent radiation, or at least radiation
which is in some way coherent with a specified source, and those which
might be called power and energy meters, which are spectrally uniform and
relatively slow.
D*Pd dA
SNR=
I,, m .
(5.83)
where A,, is the area of the entrance pupil. Then the noise equivalent
irradiance or noise equivalent flux density (NEFD) is given by
E m
NEFD =---. (5.85)
sN R - D* T O ~ O
Values of D* for the thermal detectors used in most instruments range from
lo8 to109 cm Hzl” W-’., th ey are typically 0.1 cm on a side; and 7,) is about
50%. Therefore, the NEFD is about 5 x lo-“’ W cm-* for a 1-Hz bandwidth
and 1-cm2 entrance pupil. The time constants are about 1 ms. A typical
radiometer of this type is shown in Fig. 14. The operation is as follows.75
Light from a distant source is focused onto the detector through the chopper
mirror, filter, and field stop. The detector views either the field of view when
the open part of the butterfly-shaped chopping mirror is in position or the
light from the blackbody which is brought to the detector in the same cone
as the main radiation by the conformal mirror. The position of the chopper
is monitored by a small detector which views the light from the lamp. In
this instrument, the “back” side of the chopper is used. A sighting scope,
a light stop, and rudimentary baffle are also shown.
OPTICAL DIAGRAM FIELD
PRIMARY STOP
CAVl T Y
BLACKBODY
PREC I S ION APERTURE , n
CONFORMAL MIRROR
FILTER
FIELD STOP
DETECTOR
SECONDARY
MIRROR
PELTIER COOLER cFCCE+
OR H E A T E R
CHOPPER MOTOR
REFLECTIVE
CHOPPER BLADE
SYNCHRONOUS
PIC K O F F
LAMP RETICLE
VIEWING
OPT1 CS LLIGHT
STOP
FIG. 15. Optical train of the balloon-experiment: 1 m primary ( I ) ; wobbling secondary (2);
field stop (3); off-axis mirrors (4,6); lamellar grating (5); bolometer (7); instrument box (8).
250 PHOTOMETRY A N D RADIOMETRY
is Ad/rOAc. Since NEFD values are typically lo-' and detector areas are
about 0.01 cm, it is easy to see how NEFD values of W cm-* are
obtainable with reasonable instruments. Better (lower) detector values,
larger entrance pupils, and smaller detectors can move these values down
another decade or two with great care. Then the limitations are radiation
fluctuations, emissions from the optics, and inhomogeneities in the atmos-
phere. Figure 15 shows one of these astronomical radiometers.
SNR = l O I 3 (5.86)
Pd dA = Pd AA = roPAAA. (5.87)
The power on the entrance pupil can be found in terms of the target radiance
L,, even though this is a subresolution target. In the spectral band AA, it is
A
P = L,A, 7 = L,A,R. (5.90)
d
Here L,A, represents the watts per steradian radiated on the average from
the entire target, and L? is the solid angle that the entrance pupil subtends
at the target. The range is approximately 33 x lo6 m, so that for the square
meter entrance pupil, the power is
P = L,A, x 10-15. (5.91)
Thus the signal-to-noise ratio is
(5.92)
If the optical transmission is 50%, the detector area is (50 pm)', and the
bandwidth is lo6 Hz,then a target intensity of lo3 W sr-' gives a signal-to-
noise ratio of 1. The system must be designed so that photon noise is low
enough for this D* of l O I 3 to be attainable. The problem now becomes one
of scanning and numbers of detectors. The smaller the instantaneous field
of view, the better the chance of seeing the difference between the target
and background. The faster the scan rate, the more looks obtained but the
larger the bandwidth. The smaller the detector, the better the signal-to-noise
ratio, and (other things being equal) the smaller the instantaneous field and
the bigger the bandwidth required. Suffice it to say that these radiometer
designs use many detectors and various ingenious optical and scanning
systems. They can also take on a variety of forms, because there are so
many parameters that can be varied.
The midcourse detection problem is a different one. Here we can assume
that there is no thruster power; the only signal is that from the ballistic
projectile. Its radiant intensity is then calculable if certain assumptions are
made. The radiant intensity in the forward hemisphere is &ALBB(A, T). Its
emissivity is between 0 and 1. Thus an estimate of 0.5 is reasonable. A better
estimate reiuires knowledge of the material, its state of aging, exposure,
etc. For estimation purposes 0.5 is pretty good. The projected areas of most
missiles and satellites range from 10m2 to 1OOOm' but are mostly about
100 m2. They sit on the launch pad, where they come into equilibrium with
their surroundings at about 300" K. So a midcourse object has the intensity
of a 300" K blackbody with an emissivity-area product of (cross section)
50 m2. It peaks at 9.6 pm, where its spectral radiant intensity is approximately
350 W sr-' pm-'. The SNR is again given by the same equation. The power
252 PHOTOMETRY A N D RADIOMETRY
(5.93)
where
T, = optics transmission (0.5)
EALT-’ =target radiant intensity (350 W sr-’ km-’)
A,/ R 2 = solid angle of sensor at the target
The noise equivalent range can be calculated by assuming Ad = 50 km,
B = lo6 Hz,and by setting S N R equal to 1. Then the noise equivalent range
is
NER = 6 x 105AA12.
The units are those of A;/’. If A;’’ is 10cm for instance, then the noise
equivalent range for these conditions is 60 km. We can see that this is a
more difficult problem that should be solved by the use of a larger aperture
and more detectors.
The radiometric problems of reentry are trivial. The distances are hun-
dreds of kilometers, and the bodies have become very hot from aerodynamic
friction. Clearly the calculations above show that detection is easy. The
problems relate to search rates and discrimination. The multicolor tech-
niques described above may be appropriate.
(5.94)
where
SNR =
I T ~ ~ ~ L
dhrD~AB/4f2B”’,
D * (5.95)
(5.100)
where
g = 2 or 4 for a photovoltaic or photoco,nductive detector
7 = detector quantum efficiency
The bandwidth can be calculated on the basis that there are about 500 lines
in a TV frame and the frame rate is 30 sec-’. Thus there are 7 x 10 picture
elements scanned every second. If only a single detector is used, it must
have a time constant less than 1 ps, and the NEAL will be about
lo-* W cm-’ sr-’. Such a radiance change corresponds to a change of about
10°K or 10% in emissivity, neither of which is very interesting for most
applications. Most such instruments therefore use 100 or more detectors to
reduce the bandwidth (or signal average in some other way). Since the
bandwidth can be reduced in proportion to the number of detectors used,
the NEAL is proportional to the square root of the number,
An important question to be asked about such scanning radiometers is
how, and how accurately, can they be calibrated. The calibration schemes
vary, based on the intended use. Almost all are ac-coupled instruments; in
temporal frequency space, there is a low-frequency cuton determined gen-
erally by an RC pi network in series with an amplifier. This means that the
average or dc value is lost. It must be returned as a pedestal level on which
small variations ride, and this combined signal is recorded on a cathode
ray tube (CRT), fi1m;magnetic tape, or other appropriate medium. If there
is little concern about the true radiometric level, then averaging the signal
level over some appropriate length of time (e.g., one line or one frame)
provides an excellent reference. If the scene does not change much, then
the small signals are displayed well and d o not get lost in one end of the
dynamic range. If, on the other hand, one is really interested in measuring
radiometrically, then on-board calibration is required. Probably the simplest
R A D l O M ETRIC I N S T R U M E N T S 255
way to d o this is with a small source that periodically floods the detector
(array). If this source is not collimated, it will not illuminate the detector
in the same way as the distant objects, but it can be calibrated. If the source
does not shine through all the optical elements, then its calibration gradually
becomes invalid as the elements deteriorate. Obviously this source should
be stable or measurable. The most accurate way seems to be to use a small
heated cavity at the focus of a well-protected collimating system. The
resultant beam shines at’the detector system through all the optical elements
from just outside the active field of view. Ideally, two or more such sources
are used, each at a different temperature and with their temperatures
monitored. Diode emitters are a reasonable second choice. They are not as
good, because they do not match the spectral input and because the stability
of infrared diodes (8-14 pm) has not yet been established. In the visible,
a stable silicon diode could be used, but any spectrally selective deterioration
of optical elements would alter the calibration of the instrument.
There are several ways to calibrate these instruments in the laboratory
and in the field. The usual specifications of performance are the NETD
and the MRT-the noise-equivalent temperature difference and the
minimum resolvable temperature. The first of these is defined as the
difference in temperature between two blackbodies which both fill the field
of view of the radiometric scanner that gives a peak-to-peak signal difference
equal to the rms noise. The second is the minimum temperature difference
that can be resolved by an observer at a specified spatial frequency. To
make the first measurement, one needs two large-area blackbodies (almost
a contradiction in terms) that can be carefully controlled in temperature.
The usual sources are large blackened tanks of water with stirrers and
thermometers of some type. They are set so that they are different in
temperature by some 10°K or 20°K and so that a large SNR is obtained.
The temperature difference is reduced until the signal and noise are about
equal. A curve of ATISNR is plotted. Each point is a measure of the
NETD-almost. Two types of errors occur: one because the sources are
blackbodies, and one because they are not. Let us see how this comes about.
The difference in signal-to-noise ratio is given by
The terms that would make the proportionality an equality are all geometric
and temporal bandwidth factors. The ~ ( h factor
) is the product of the
spectral transmission of the optics and the atmosphere. The scanning radio-
meter views first one body and then the other. It is assumed that they are
close together so that there is no difference in atmospheric transmission,
256 PHOTOMETRY A N D R A D I O M E T R Y
T ( A ) D * ( A AL(A,
) T ) dA. (5.103)
Only the radiance changes. Let us consider only that term. The radiance
projected toward the sensor is the sum of the radiation emitted from the
source and the background radiation reflected from the source to the sensor.
AL(A, T ) = A [ & ( A ) L R B ( AT ,) + p b E ( A , T ) ] , (5.104)
ALL - AE E+ -L1-aadTLT + A ( y )
- _E+ - -xex
A AT E
+-Apb+-.
PbAE
- (5.105)
E ex-1 T L L
The first term is typically about 1%, a change in emissivity from about 0.98
to 0.97 of these near-blackbody sources. The second term is about 5 AT/ T
or 0.017 AT The next two terms represent background contributions. In
fact, each would really be the sum of all the variegated components of the
background. The first one is the contribution of the variation in reflectivity
with a constant background. The second one represents the contributions
of different background terms if the reflection geometry with the two samples
is slightly different. These effects have been considered in more detail.I2
The analytical results there show that errors of several percent are easily
attained. One message is clear: the radiometric differences of such test
objects should be measured radiometrically.
When really large patterns for field work are to be used, the use of water
baths and other actively heated structures gets cumbersome. Bartell” has
described a test pattern design that can be used for NETD and MRTD
measurements. As Lloyd38shows, an MRTD test involves several important
psychophysical factors. These include the temporal and spatial integration
properties of the eye and its tendency to be a matched filter. Accordingly,
the test pattern must have groups of four bars with a 7 to 1 aspect ratio. If
one can accept that the effects of the psychophysical factors can be calcu-
lated, then simpler means, including edge gradients and line spread func-
tions, are a~ailable.~’
5.6.5. Spectroradiometers
These devices differ from standard radiometers principally in that they
use some technique for selecting narrow portion of the spectrum, and only
radiation in that band is measured. The radiometric calculations of the
RADIOMETRIC INSTRUMENTS 257
There are several direct methods for filtering, and there are some that
might be called exotic or indirect. Certainly the use of a prism, grating, and
filter must be considered direct and straightforward-even if they are not
always simple.
FIG. 16. Rays on a thin film illustrating the reflection from the first surface plus two from
the second.
for instance, there is a phase shift variation with angle of 50% between
central and marginal rays. This is equivalent to moving the central
wavelength from 2 pm to 1 pm or from 10 pm to 5 pm. When the filter is
used in a collimated beam, it can be tipped to change the passband position;
such a procedure entails somewhat complicated polarization effects which
must be calibrated. An interference filter in general has what might be called
side lobes or higher-order passbands. For the simplest case of obtaining
good transmission at Ao, the total reflected radiation should destructively
interfere. The first ray has n o phase shift. The once-reflected ray undergoes
a phase shift given by 2knd cos 0 + rr. The additive rr comes from the phase
shift which arises from reflection when light goes from a dense to a rare
material as it exits from the film. The third ray which undergoes two
reflections is just twice this, etc., which is shown schematically in Fig. 16.
For destructive interference, each of these rays should have a phase shift
with the wave reflected from the front surface that is exactly rr, or an odd
multiple of T. The design equation then is
mA
ndcos @=--, even m. (5.106)
4
Although more complicated designs exist, they generally have the charac-
teristics that a variety of combinations of m and A satisfy the conditions
for maximum t r a n s m i s ~ i o n For
. ~ ~ example, a filter designed for one-quarter
wavelength thickness at 10 pm will have a peak at 10 pm, 5 pm, 2.5 pm,
etc. The radiation in these higher-frequency passband regions is usually
absorbed by the substrate. Caution should be exercised by the investigator
when using such a filter in that this absorption can be highly temperature-
dependent-the substrate is an absorption filter and has the properties
described above.
Radiometers often use these filters in a fixed position or in a filter wheel,
a set of different filters which can be introduced into an appropriate part
RADlOM ETRIC INSTRUMENTS 259
of the optical train one by one. The device is usually circular (a wheel) but
can take on other forms, linear or cylindrical for example.
One very useful device for defining a passband may be thought of as a
continuously varying interference filter; its most popular form is the cir-
cularly variable interference filter or CVF.4' Its center wavelength (or
frequency) is determined by position on the filter, which consists of a circular
substrate on which are evaporated interference layers whose thicknesses
vary with circumferential position around the circle. The calibrations, cau-
tions, and limitations which apply to interference filters apply also to CVFs.
In addition, positional calibration must be considered. The bandpass is an
average over the area of the filter that is viewed, and the center wavelength
is a function of the position of that area-which is usually a slit. Half of a
CVF typically covers one octave of the spectrum, from A,, to 2Ao. Thus, one
generally has a bandpass of about A,/ 180".The RP is 180 per degree of the
CVF that is viewed.
The concepts of throughput and the optical invariant are required for
further useful discussions and comparisons of these instruments. The optical
invariant when applied to pupil and image planes is just that ny sin 6 is a
constant, where n is the refractive index of the medium, y is the height of
the image or pupil, and 6 is the angle the next image or pupil subtends at
the one being considered, as shown in Fig. 17. The throughput is the square
of this optical invariant.
Suppose that an optical system with area A . and focal length f focuses
radiation onto a CVF over a width x and length y (as defined by a mask
or slit in front of the CVF). Then the throughput is A o x y / f 2 .If the slit
extends from r / 2 to r (where r is the radius of the CVF) and is x wide at
the circumference of the CVF, then the slit area is ry/4. The resolving power
is 47r/x (using the average width of the slit). Therefore, one increases the
resolving power at the expense of throughput, which is directly related to
flux. There is a linear and inverse relationship between these two. The loss
in throughput resulting from reduction of slit width can be compensated
for by making the slit longer, thereby requiring the radius of the CVF to
be larger.
FIG.17. The Laprange invariant for pupils. The product of n , y , sin 0, is equal to n z y , sin 0 2 .
In this example, n , = n z .
260 PHOTOMETRY A N D RADIOMETRY
-g=f (&.b dn
(5.108)
The throughput Th can be calculated as the area of the beam on the prism
times the area of the slit divided by the square of the focal length:
h dx
T h = ( h w c o s 0)-. (5.109)
f’
If the above expression is used for the linear dispersion dx/dA, then
dA
Th = RP h2 cos 8 -= the throughput. (5.110)
f
RADlOM ETRlC INSTRUMENTS 26 1
,
FIG. 18. Prism geometry.
m = d cos 0 (s)
d0 - dD - m
dh dh d cos 0
d cos d D
dh =
m
A
RP=---=d(sin 4 + s i n 8)
(5.1 12)
dh d cos 0 d D
E=E,+E,+2ERbcos
:[ -ndcos8
1.
The optical path difference nd cos 8 is just the difference in the length of
(5.1 14)
the two arms, assuming that a compensator plate has been inserted in the
horizontal arm. The length of the horizontal arm then is
l=tu(t)+Io.
The output flux density for a monochromatic input is given by
(5.115)
E ( t ) =2
I
EA( t ) 4-EA COS[ ( t u ( t ) -4- 1.3)- (5.1 16)
from the side lobes. The resolving power can be written as nd cos B / A ; it
is the optical path difference measured in wavelengths.
The throughput of an interferometer like the Michelson is the area of the
mirrors times the area of the detector divided by the square of the focal
length of the final beam. If a T w y m a n - G ~ e e nversion
~~ is used, it is easy
to see that this is (2.44A/D)*(.rrD2/4)=A2.
The use of an interferometer like the Michelson for radiometric purposes
requires some special considerations. The most important of these are that
there normally is no dc reference available, and that the source intensity
must remain constant during the time of the scan. The lack of a d c reference,
of course, means that the spectrum can be obtained, but only on a relative
basis. A fairly obvious solution to this is to measure the total radiation with
an auxiliary radiometer and use it to set the radiation scale. Variations of
the source output intensity generate much more subtle results. It can be
seen that the spectrum is obtained by a cosine transform:
co
5.7. Measurements
Radiometric and photometric measurements are not difficult in principle.
They are very hard to accomplish with a precision or accuracy better than
about 1YO. Most field measurements which are made with portable equip-
ment and distant (in both space and time) from a primary calibration have
errors of more like 20%. The measurement of the spectral radiant intensity
of rocket plumes has been considered satisfactory when the error was less
than an order of magnitude! These errors can usually be attributed to the
fact that the measurement of the unknown was different from the calibration.
Accordingly, rule 1 of radiometric measurement is to calibrate the radio-
meter in a configuration that is as similar to the measurement situation as
possible.
The responsivity of a radiometer depends upon very many factors. It can
be a function of wavelength position, angle, polarization, flux level, time,
humidity, temperature, and many other things. Phase of the moon and the
name of the operator have also been proposed as factors influencing the
measurement. This great variability leads to rule 2, which is to think of
everything.
264 PHOTOMETRY A N D RADIOMETRY
These two rules are very general. The rest of this section is intended to
make them a little more concrete by discussions of examples. The discussion
starts with standards, the primordial element of every radiometric
measurement.
5.7.1. Standards
Any radiometric measurement must be considered in the light of the
calibration against known standards. It can be no better than the standard.
The National Bureau of Standards (NBS) maintains and provides radio-
metric standards in the form of 1000 W DXW-type tungsten-halogen lamps.
They are described in detail elsewhere.I2 The lamps are given a primary
calibration and their irradiance is fit to the following curve:
E ( A , t ) = A - 5 e(o,-4.7+~f+bflA) (5.117)
The values of the constants ah, 6, and c are found from least squares fitting
for different values of the wavelength A and time t. Four lamps are used
and the proper 3a error found from among these, as shown in Table IV.45
Table IV summarizes the uncertainties assigned by NBS to these standards
at different wavelengths.* Item l ( a ) is the error in primary calibrations;
l ( b ) represents long-term reproducibility rather than error with respect to
SI units. Item 2(a) represents the transfer to an irradiance scale, while 2(b)
is the error in reproducibility of the four standard lamps. Item 2(c) represents
the errors involved in the fitting procedure described above. Item 3 represents
the error in transferring the value of the four standard lamps to the working
standard being calibrated.
The total uncertainty with respect to SI units then is found from the rss
of l(a), 2(b), 2(c), and 3. Reproducibility is found from l(b), 2(a), 2(b),
2(c), and 3. Note that these are all three standard deviation errors.
It should be observed that the errors increase rapidly in going toward
250nm largely because the flux falls off so rapidly. It is enlightening to
calculate the ratio of the total 2800°K blackbody to the amount of radiation
at each wavelength (the approximate temperature of these sources). These
sources are limited in their spectral extent at the short wavelength end by
lack of flux and at the longer wavelength end by the absorption of their
glass or quartz. These are the certified standards available from NBS.
Instructions for their use come with them and have been published else-
where.'*
A very encouraging development in radiometric standards is that of the
electrically calibrated radiometer ( ECR).46 The receiver is blackened and
1. NBS Spectral Radiance Scale 250 350 450 555 654.6 800 1300 1600
nm nm nm nm nm nm nm nm
(a) Absolute error (with respect to SI units) 1.66 1.20 0.93 0.65 0.65 0.46 0.28 0.27
(b) NBS long term reproducibility 1.06 0.76 0.61 0.30 0.26 0.22 0.12 0.18
3
rn
2. NBS Spectral lrradiance Scale
(a) Systematic errors 0.16 0.16 0.10 0.10 0.08 0.08 0.07 0.07
R
c
(b) Random errors (estimated reproducibility of the mean 0.72 0.55 0.53 0.56 0.54 0.51 0.80 0.80
of standards I, - I,, 3 - u) I
(c) Model error 1.35 0.80 0.73 0.73 0.78 0.76 0.77 0.82
4
v1
3. Transfer Calibration of a Test Lamp
(a) Random errors (3u precision) 1.35 0.65 0.50 0.48 0.41 0.51 0.39 0.36
V= E,,(h)%!(h)dh. (5.118)
-m
If the overall responsivity takes into account the slit function as well as the
268 PHOTOMETRY A N D RADIOMETRY
V=% I--oo
a-
E,,(A)r(A)cT(A-Ao) dA. (5.119)
If the slit can be described accurately enough as a delta function, the integral
unfolds immediately:
v= BEA(AO)r(AO) = %(AO)EA(AO). (5.120)
The spectral response at A. can be found by the use of any sufficiently
well-known standard source. In practice, of course, there are no delta
functions. Slit functions peak and have long tails that are relatively wide.
The long tails which reach longer wavelengths at values of to lo-' are
the source of error in measurement-at least until they are corrected. The
interesting features are the means by which such corrections can be made-
and the limitations in making them. The slit function can be measured by
the use of a device like the ECR and a source (like a laser) with a line
width smaller than the spectral slit width. The spectral slit function can be
taken as known at least to the resolution of the best laser in the spectral
region of interest. But even though the function is known, the integral is
not unfolded simply. The output voltage is given by
V=% I-m
a-
E , ( A ) r ( h ) a ( A-Ao) dh.
V=% I m
-m
E , ( A ) a ( A - A o ) dh. (5.121)
vi= I_,
00
The more individual measurements that can be made, the more determined
EA(A)will become. The problem is one of estimating E A ( A )from a set of
samples with known sampling shape. One can argue that a(A- A i ) can
change as hi changes. This is inconsequential as long as o ( A - A i ) is known.
MEASUREMENTS 269
Another way to view this procedure is to recognize that the right side of
the equation is a convolution. Therefore, the Fourier transform of the voltage
is equal to the product of the transforms of the slit function and irradiance.
The procedure for obtaining the irradiance is then straightforward but
involves much calculation. It might be added in passing that the use of this
approach to measuring spectra leads to the possibility that the smallest
possible slit width is not the best. An interesting tradeoff between slit width,
signal-to-noise, and processing procedures arises. Too wide a bandpass
leads to processing errors which are too large; too small a slit width leads
to a signal-to-noise ratio which is too low.
The authors of the NBS article then provide a figure which shows the
true spectrum and the inferred one. The major portion of the error is at the
short wavelengths where there is very little radiation and where the longer-
wavelength scattered radiation predominates.
(5.123)
(5.124)
Iorn MA
M
Iom
&h =
--
- (5.126)
aT4'
M :B dA
low--.
M'-lA "
-
M
(5.127)
Eqh= [" Mz: dA u q 7-4
Jo .
270 PHOTOMETRY A N D RADIOMETRY
are important for samples which are not close to Lambertian. These
emissivities are defined in terms of radiances. The spectral directional
emissivity then is
(5.129)
(5.130)
use the reflected radiance as the source for the next transfer. The BRDF is
just the right function for this.
Usually the spectral nature of the measurements is determined by the
source and the detector. Only one system (Gier-Dunkle) uses a different
method. The spectral region is determined by such difficulties as ambient
fluxes, gradients, etc.
The main methods for the measurement of normal specular reflectivity
are those of Strong” and Bennett and Koehler.’l These are illustrated in
Figs. 19 and 20. In the Strong method, light shines into the sample area
that has no sample. The beam reflects once from the folding mirror. Then
the mirror is rotated to its sample position and the sample is inserted. Now
the beam reflects once off the folding mirror and twice off the sample. The
geometry of the apparatus prevents true normal-incidence measurements.
They certainly are directional. Strong’s apparatus operated between 0.3 pm
and 10 pm, with the spectral region limited primarily by the integrating
sphere used for putting uniform flux on the detector. The inaccuracy is
about 0.1%. The diagram of the Bennett-Koehler apparatus shows how it
works. The beam with and without the sample travels along the following
FIG. 19. Sketch of the Strong-type reflectometer used in the visible. The sample reflectance
R is obtained from the square-root of the ratio of two oscillograph galvanometer deflections
G , and G, which correspond to SAMPLE-IN and SAMPLE-OUT.
274 PHOTOMETRY AND RADIOMETRY
sets of mirrors: MI to M 6 , and MI3 to Mzl.These are the input and output
groups, respectively. When the sample is in, as shown, the light follows the
path M 6 - M , - S - M 8 - S - M 9 - M 1 3 .There are two reflections off the sample
as with the Strong reflectometer. When the sample is out, the path is
M6-M,-Ml1-M9-Ml3.The reflection of the mirror M I , replaces that of
two sample reflections and M 8 . The ratio is the square of the sample
reflectance, except for the ratio of M , to MI . To correct for this, the sample
as well as mirrors M6 and MI, are rotated 180". Then the light follows the
dashed lines with the sample in, i.e., M13-Ml,-S-M,l-S-M12- MI3.With
the sample out, the path is Ml3-MIO-M8-MI2-Ml3. This ratio is then
S-MII-S to M 8 . Thus the ratio from the first two measurements would be
S 2 M 8 / M , , ,and that from the second would be S 2 M l , / M 8 The . ratio of
these ratios gives the fourth power of the sample reflectance. The sample
labeled T is a transparent one, the transmission of which can be measured
with a sample-in sample-out technique. Although this technique does require
the extra measurements for its ultimate accuracy (about 0.001), it does not
require the use of an integrating sphere as does the Strong system (or it
would be very sensitive to small misalignments of the sample).
and the requirements for satisfactory signal-to-noise ratio dictate that these
quantities are sometimes larger than one would really desire.
The geometry of the measurement is shown in Fig. 21. The angles of
incidence are Oi and +i; those of reflection are Or and 4,. The sample is
placed with its normal along the z axis and is assumed to be flat (on the
average). Two geometries have been used in measurement^.^^'^^ The first
emulates the diagram just discussed; the second places the beam and
detector in a horizontal plane and tilts and rotates the sample. The first
geometry requires no coordinate transformation to describe the defining
geometry, but it is not capable of measuring with detectors that are cooled
in dewar flasks with liquid refrigerants.
In both cases, the azimuthal angle of incidence is chosen to be zero. This
is a free choice for the origin of the coordinate system. The incident flux
is almost always a laser beam which has been expanded and clipped by an
appropriate pinhole. It is then aimed at the sample as a slowly convergent
beam that is focused at the detector position. The beam could be collimated,
but this would eliminate one of the two methods of calibration described
below. The function of the beam is to illuminate a portion of the sample
with a uniform irradiance. The sample then scatters light to a detector-
receiver assembly-a form of radiometer. The sizes of these can be deter-
mined to some extent by writing the definition of BRDF in its limiting form:
Ai A n d
BRDF= lim -, (5.131)
L\A,-O
an-0
ai
-
AAi
where AAi is the illuminated area and Andis the solid angle of the detector
as subtended at the sample (=Ad/d2). The flux on the detector is an
integration over the sample area and the detector area. It is said that the
BRDF is measured at angle Or, but it is really an average from O,-AO, to
O,+AO, and over all points on the sample. Reasonable values are d =
1 m, = 1 mm. Then the averaging range is 1 mr. For most purposes, this
angular resolution is finer than necessary. The illuminated area should be
large enough to adequately represent the sample unless several measure-
ments are to be made of several portions of the same sample. Rough samples
typically have roughness zones as large as 500 pm. Thus a sample illumina-
tion of 5 mm is about right to characterize these. An F/100 beam will
encompass a 1-cm sample and focus at a detector 1 m away from the sample.
But the beam may not be uniform.
For specular reflection methods, a ratio technique may be used for
calibration. Somewhat similar approaches have been tried for these direc-
276 PHOTOMETRY A N D R A D I O M E T R Y
FIG. 21. BRDF angles. The incidence angle is Bi. The angle of specular reflection is Bo.
The arbitrary polar angle of reflection is 8,. The azimuthal angle of reflection is br.
tional methods. The two main calibration techniques are what we have
called the “reference method” and the “no-sample method”. We can gain
an understanding of them from an analysis of the detector output voltage.
It is found to be
%LAdCOS 8dA, COS 8,
v=%!Pd= (5.132)
d2
The reflected radiance L is by definition equal to the sample irradiance
times its BRDF. Thus,
v, = % E , P h A d COS8dA, COS os
(5.133)
d2
The detector is always arranged to be perpendicular to the radiometer line
of sight, so cos8d is 1 . If there is no sample, then all of the flux at the sample
area gets to the detector, and
VNs= %!E,Ai.
If we take as the sample area the illuminated portion only, then A, = Ai.
Therefore,
(5.134)
The BRDF is given by the product of the voltage ratio and the reciprocal
of the projected solid angle of the detector.
MEASUREMENTS 277
(5.135)
where VK is the voltage from the detector when the known sample with
BRDF pK is in place. This is simple in concept and calculation but requires
a reference whose BRDF is known for all conditions and comparison
measurements made at all positions. The obvious question then is: “How
did you measure the reference?” The procedure which seems to devolve
from this is to use a Lambertian reference. It has a BRDF which is Phr-’,
where Ph is the hemispherical reflectivity. Then
(5.136)
(5.137)
where V(0) is the voltage resulting from the reference illuminated normally.
This approach has the advantage that the instrument parameters are not
involved; however, a separate measurement must be made of Ph.
Measurements of specular, hemispherical, and directional transmittance
follow those of reflection with an obvious change in geometry.
FIG. 22. Schematic diagram of an NBS laser calorimeter. Explanation in the text.
the current, voltage, and time. Nitrogen is flowed through the cavity to
reduce scattering, and temperature sensors are used to record any stray
absorption in the baffle structure. They claim 13% single-measurement
uncertainty for power levels of about 100 kw in the wavelength range from
1 to 11 pm (13 kW). The unit is 1.2 m high, 1.5 m long and 0.65 m wide and
weighs about 400 kg.
Although they do not discuss the sources of uncertainty, one can attribute
them in general to differences between the measurement and the calibration.
It is unlikely that the calibration source is 100 kW; it is probably considerably
less. The errors then can arise from different amounts of conduction and
convection of heat away from the cavity. Although the baffle structure is
monitored, some errors will arise from spillover onto it. The relative contri-
bution of heating from the pump will be different, and no cavity is a 100°/~
absorber.
Calorimeters exist for measuring laser powers from milliwatts to hun-
dreds of kilowatts. The uncertainties are lowest, of course, for the middle
of the range where the calibration can be very similar to the measurement
and where there is sufficient energy that it causes an appreciable temperature
rise. Of course, for fluxes of less than 100 kW, the instruments use smaller
reservoirs, sometimes just the heating of an appropriate detector mass. At
the lower end of the scale, direct detection by pyroelectric detectors is very
competitive with these calorimetric techniques. When the measurements are
made in the visible, silicon detectors can be used.
MEASUREMENTS 279
V( t ) =
I I'%%
% cos 2 ~ ~ , t P ( ~ 0 dvo
) dt.
The optical frequencies are so much higher than the electrical frequencies
(5.138)
that very many optical cycles are integrated, and P ( v o ) may therefore be
considered a constant, Po. We are also assuming that the only temporal
variations in the power are in the modulator-which is assumed to be
cosinusoidal. .
V( t ) =
I,%Po cos 27~v,tdt. (5.139)
V( t ) =
I, %Po cos 27rv,t rect( v,t - 4 ) dt. (5.140)
The phase of the electronic switch is set relative to the zero phase of the
chopper. If it is zero, then the output V ( t ) is just
V( t ) =
I, %P0lcos 2 ~ v , t ldt. (5.141)
Nonzero phase shifts will provide different values for V( t ) . The final output
voltage is obtained by integrating over the time constant 7 which is set by
a low-pass filter. It is worth considering all of this more generally and in
the frequency domain.
280 PHOTOMETRY A N D RADIOMETRY
,
@ = L dA cos 0, dA2 cos 0,/ d '. (5.142)
In this case, the angles 0, and O2 are equal (0, = O2 = 0). Further, the
center-to-center distance is d = 2R cos 0. Thus the flux on dA2from dA, is
MEASUREMENTS 28 1
given by
@ = L dA, dA2COS’ 8/4R2 COS’ 8
= L dA, dA2/4R2. (5.143)
Since the points were chosen arbitrarily, this shows that dA, will irradiate
all elements dA, uniformly. Now if Ed is the irradiance on element dA,
caused by a source shining through the entrance port, then for a Lambertian
surface the total diffuse irradiation is
Ed,-=pEd+p’Ed+p’Ed+. * .
When the direct irradiation is added to this, the total becomes
Ed
E = ~ ~ ( i + ~ + .)=-
~ * + . (5.144)
(1 -PI’
The resultant light is diffused and is an inverse function of the absorptivity
of the wall. The hemispherical reflectivity p has a maximum value of
-
T-’ 0.3. Thus the ratio of total irradiance to direct irradiance is at most
about 1.4.
A more detailed story described by Goebel and others is given be lo^.^',^^
Imagine a sphere divided up into n spherical caps, each of which rep-
resents an entrance port, exit port, or some other discontinuity in the wall
of the sphere. It can be seen from Fig. 2 that the fractional area of any of
these caps is given by
2Rh
f = -4.nR2-
- - - 2R-
h
2R
=”
R - J R ’ - ~ ~1 -
2
Jq] = f ( l -C0S2 0)
= sin’ 8. (5.145)
The radiation entering is labeled O0and reflects from area a, an amount
p0@, uniformly in all directions. I f f ; = ai/A,, where a, is the ith special
282 PHOTOMETRY A N D R A D I O M E T R Y
area and A, is the spherical area, then the average wall reflectivity or Pw is
n
This is just the area-weighted reflectivity. The unabsorbed flux after the
second reflection is pwpoQo.This is uniform and is distributed over the wall
and the special areas so that the flux after the third reflection is just p ~ p o @ o .
The total unabsorbed flux after an infinite number of reflections then is
po - po/ ( 1 - p w ) . The flux on aperture a, then is given by
f;P O Q O
-@) + c 4 ’
(5.147)
= [ 1- P w ( 1
and the ratio of the flux out of ai to the input flux Q, is the cavity efficiency
FIG.24. Sphere efficiency as a function of the fractional hole area for different numbers of
holes.
MEASUREMENTS 283
8.6 -
0.5 -
0. 4 -
8.3 -
0.2 -
0.1 -
FIG. 25. Sphere efficiency as a function of the fractional hole area for different wall
reflectivities.
-@)] .
-I
& =hw[
1-Pw( 1 (5.149)
In this form, we can see that the simple model above is modified by J p ,
in the numerator. This represents the amount of flux getting out the ith
hole. It is also modified by 1 -ZJ; to indicate that some of the wall surface
is not available for reflection. Curves of sphere efficiency for different area
ratios and coatings are shown in Figs. 24 and 25. The message is clear.
Keep the reflectance high; make the ports as large as the uniformity consider-
ations allow. Another message is apparent: there is a strong dependence
on the value of p w . Any spectral variations in the coating are magnified in
the sphere. The relative error in efficiency is
(5.150)
284 PHOTOMETRY A N D RADIOMETRY
The visible region has a very special significance in the entire electromag-
netic spectrum. Only visible light can be detected in a useful way by humans,
and the process of seeing is important to work, leisure, recreation, and
everyday living. Accordingly, a special branch of radiometry, photometry,
deals with flux levels of radiation as they generate a response in the eye.
A typical problem is the calculation of the proper light source to illuminate
a work surface for a particular function. Psychologists have measured the
appropriate light levels for performing many different tasks. Light sources
with different spectral distributions but the same total radiation elicit
different amounts of response from the eye according to how the spectrum
of the source matches that of the eye. This can be expressed in terms of
response of the eye at its maximum K,the spectral distribution of the eye’s
response V ( A ) ,and the spectral distribution of power:
P,= K
jom V ( A ) P ( A )dA [lm]. (5.151)
PV
K jOm V ( A ) P ( A )dA
0v=-= (5.152)
pe j9mP(A)dA
P,= qv
jOWdA
P(A) = vvP. (5.153)
All the geometric relations which apply to flux in general apply to luminous
flux in particular. The main differences and special features rest in the
PHOTOMETRY: RADIOMETRY O F VISIBLE LIGHT 285
handling of the spectral normalization and the many units and names for
them which have grown like topsy with this area of application.
V= [a
3
-W
%(A)Pe(A)dA. (5.154)
V = BP, [m
-m
r(A)p(A) d. (5.155)
j-,
m
-a
V(A)@(A) dh
(5.157)
V = j p m s l ( A ) P ( A )dh. (5.158)
vv= Im
-m
K V ( A ) P ( A )dh Im -m
P ( h ) dh
=K I m
-m
V ( h ) p ( A )dh I*-m
p ( h ) dh. (5.159)
Thus for a calibration source with spectral distribution of flux given by Pc,
one has for the luminous responsivity
slv= V I P c =
9
I r ( A ) P c ( A )dh
(5.160)
K
I V ( h ) P c ( A dh
)
(5.161)
= P, r ( h ) p , ( A ) dA.
I V ( A ) p c ( h )dA
- ps I
V=%,K
I ( A ) p c ( A 1 dh
r ( A ) p , ( h ) dh.
v = slvPv
I V(A)pc(A)dh J r ( h ) p s ( A ) dh
(5.162)
REFERENCES 287
This result deserves discussion. The output is not always the luminous flux
on the receiver times the luminous response of that receiver. It is true if
the integral factor is equal to 1. For that to be true, either r ( A ) must equal
V(A), or p c ( A ) must equal p , ( A ) . If one calibrates a n eye with p c a n d then
uses the eye, such is the case. If one calibrates a TV tube with a tungsten
lamp a n d then uses the same kind of tungsten lamp t o illuminate the televised
scene, it is also true. It is not generally true, n o matter how logical it sounds.
This should be borne in mind in the photometric calibration of TV a n d
other kinds of tubes, which d o not usually have reponses that match the eye.
References
1. S. Chandresekhar, Radiatiue Transfer, Dover, New York, 1960.
2. G. Bauer, Measurements of Opfical Radiation, Focal Press, London, 1965.
3. R. A. Smith, F. E. Jones, and R. P. Chasmar, The Defection and Measurement oflnfrared
Radiation, Clarendon Press, Oxford, 1968.
4. M. Planck, “Zur Theorie des Gesetzes der Energieverteilung im Normalspectrum,” Ver-
handlungen der Deutschen Physikalischen Gesellschafr 2, 237 ( 1901).
5. W. Herschel, “Experiments on the Refrangibility of the Invisible Rays of the Sun,” Phil.
Trans. Roy. SOC.90, 255 (1800).
6. 1. W. Ritter, as described in Optical Anecdotes, p. 55, by D. J. Lovell, SPIE, Bellingham,
MA, 1981.
7. U S A Standard: Nomenclature and Dejinitions for Illuminating Engineering RP- 16. Unfted
States of America Standards Institute.
8. R. C. Jones, “Terminology in Photometry and Radiometry,” J. Opt. SOC.Am. 53,1314 (1963).
9. F. E. Nicodemus, as reported by 1. Spiro in Opt. Eng. 13, G183, (1974), and C. L. Wyatt,
Radiometric Calibration: Theory and Methods, Academic Press, New York, 1978.
10. A. G. Worthing and D. Halliday, Heat, Wiley, New York, 1948.
11. A. G. Worthing, in Temperature: Its Measurement and Control in Science and Industry,
Reinhold, New York, 1941.
12. W. L. Wolfe and G. J. Zissis, The InJrared Handbook, Office of Naval Research, Washington,
D. C., 1978.
13. J. Strong, Concepts of Classical Optics, Freeman, San Francisco, 1958.
14. M. G. Dreyfus, “Spectral Variation of Blackbody Radiation,” Appl. O p t 2, 1113 (1963).
15. R. D. Hudson, Infrared System Engineering, Wiley, New York, 1969.
16. M. R. Holter, S. Nudelman, G. H. Suits, W. L. Wolfe, and G. J. Zissis, Fundamentals of
InJrared Technology, Macmillan, New York, 1962.
288 PHOTOMETRY AND RADIOMETRY
17. P. Kruse, McLaughlin, and McQuistan, Elements of Infrared Technology, Wiley, New York,
1962.
18. M. A. Bramson, Infrared: A Handbook for Applications, Plenum, New York, 1968.
19. R. H. Kingston, Detection of Optical and Infrared Radiation, Springer-Verlag, Berlin, 1978.
20. J. N. Howard, Book review of Seven Place Tables of the Planck Function for the Visible
Spectrum, by D. Hahn, J. Metzdorf, B. Schleef and J. Seich, Academic Press, 1964, in
Appl. Opt. 4, 808 (1965); other volumes are mentioned.
21. E. M. Sparrow and R. D. Cess, Radiation Heat Transfer, Brooke-Cole, Belmont, Calif. 1966.
22. F. E. Nicodemus, “Radiance,” Am. 1. Phys. 31, 368 (1963).
23. J. Rainwater, “Generalization of the Abbe Sine Law in Geometric Optics,” Am. J. Phys.
32, 626 (1964).
24. R. Gardon, “The Emissivity of Transparent Materials,” J. Am. Ceram. SOC.39,278 (1956).
25. H. 0. McMahon, “Thermal Radiation from Partially Transparent Reflecting Bodies,”
1. Opt. SOC.Am. 40,376 (1950).
26. M. Czerny and L. Genzel, “Uber die Eindringtiefe raumlich diffuser Strahlung in Glas,”
Glastech. Ber. 25, 134 (1952).
27. D. Q. Wark and H. E. Fleming, Monthly Weather Reo. 94, 351 (1966).
28. H a n d and Conrath, Science 165, 1258 (1969).
29. J. C. Gille, “Inversion of Radiometric Measurements,” Bull. Am. Mer. SOC.49, 903 (1968).
30. L. D. Kaplan, “Inference of Atmospheric Structure from Remote Radiation Measure-
ments,” J. Opt. SOC.Am. 49, 1004 (1959).
31. S. Twomey, J. Geo. Res. 66, 2153 (1961); 1. Assoc. Comp. Mach. 10, 97 (1963); J. Frank
Inst. 279, 95 (1965); Monthly Weather Reo. 91, 659 (1963).
32. W. L. Wolfe and H. P. Stahl, “Some Calculational Results Using Multicolor Radiation
Inversion,” IR Phys. 20, 293 (1980).
33. W. L. Wolfe and E. L. Dereniak, in Imagingfor Medicine, Vol. I Chap. 13 (S. Nudelman
and D. D. Patton, eds.), Plenum, New York, 1980.
34. E. L. Dereniak, Ratio Temperature Thermography, Ph. D. Thesis, University of Arizona,
Tucson, 1976.
35. Electro-Optical Industries Brochure.
36. R. Hofmann, S. Drapitz, and K. W. Michel, “Lamellar Grating Fourier Spectrometer for
a Balloon-borne Telescope,” IR Phys. 17, 451 (1977).
37. F. 0. Bartell and W. L. Wolfe, “Cavity Radiators: an Ecumenical Theory,” Appl. Opt. 15,
84 (1976).
38. J. M. Lloyd, Thermal Imaging Systems, Plenum, New York, 1975.
39. J. D. Gaskill, Linear Systems, Fourier Transforms and Optics, Wiley, New York, 1978.
40. W. G. Driscoll and W. Vaughan, Handbook of Optics, McGraw-Hill, New York, 1978.
41. R. A. Sawyer, Practical Spectroscopy, Dover, New York, 1963.
42. R. J. Bell, Introductory Fourier Transform Spectroscopy, Academic Press, New York, 1972.
43. A. C. Candler, Modern Interferometers, Hilger and Watts, London, 1951.
44. R. D. Saunders and J. B. Shumaker, Optical Radiation Measurements: The 1973 Scale of
Spectral Irradiance, U. S . Government Printing Office, 1977.
45. J. Geist and W. R.Blevin, “Chopper Stabilized Null Radiometer Based upon an Electrically
Calibrated Polyelectric Detector,” Appl. Opt. 12, 2532 (1973); W. M. Doyle, B. McIntosh,
and J. Geist, Roc. SOC.Phot. Inst. Eng. 62, 166 (1975); and R. J. Phelan and A. R. Cook,
“Electrically Calibrated Pyroelectric Optical Radiation Detector,” Appl. Opt. 12, 2494
(1973).
46. J. Geist, E. F. Zalewski, and A. ‘R. Schafer, “Spectral Response Self-Calibration and
Interpolation of Silicon Photodiodes,” Appl. Opt. 19, 3795 (1980).
47. National Bureau of Standards, Optical Radiation News, 20 April (1977) and 24 April (1978).
REFERENCES 289
T. 0. Poehler
Applied Physics Laboratory
Johns Hopkins Road
Laurel, Maryland 20707
6.1. Introduction
D * = Ll VN ( A )‘I2 =- VS/vN(AfA)’”(cm(Hz)’~’/W)
(6.3)
H P
where Af is the electrical bandwidth in Hz.A detector whose performance
is specified at a particular wavelength A from a monochromatic source is
designated as D,*(A,f;Af).
The performance of low-noise detectors may also be limited by radiative
noise arriving at the detector from the background environment. When
operating under these conditions, a detector is said to be background-limited
in performance or in the BLIP mode. Since this measure of the detectivity
is dependent upon the half angle 6 from which the detector can see
background radiation, a normalization including this dependence on field
of view is used and is designated D**, where
D** = D* sin 6, (6.4)
where the half angle of the field of view is 6.
0.75 mm
4 b -
I Radiation
Bismuth Silver contacts
502
Silicon
infrared radiation in the 1- to 20-pm range where the “black” film is highly
absorbing, but are characterized by low-power-handling capability and long
response times. Thin-film thermocouples fabricated by vacuum evaporation5
of the components often exhibit improved properties in comparison to bulk
devices. These detectors, made by precision photolithographic techniques
with small dimensions and low mass, can yield lower time constants and
impedances better matched to electronic circuitry (Fig. 1). Evaporated arrays
of thermocouples composed of antimony and bismuth are used in commer-
cial thermopiles.
While thermopiles are satisfactory for many spectroscopic applications,
their low-power-handling capability and long time constant make them
inappropriate for many laser measurements. In particular, measurements
of CW laser power in the multiwatt range or pulsed lasers with pulse energies
of several joules require calorimeters capable of withstanding these inputs
without damage. These requirements have stimulated the construction of a
multitude of calorimeters for high-laser-power measurements. Typically,
these instruments use a disc6 or cone’ coated with an absorbing layer to
absorb the laser radiation, while the temperature of the absorber can be
monitored with a thermocouple to obtain an estimate of the energy
deposited. While absorbing discs are common, radiation trapping can be
increased by directing the radiation into a cone which is coated with
absorbing material. Multiple reflections of the incident beam will cause
nearly complete absorption and will also tend to increase the damage
threshold of the detector. A diagram of a typical absorbing cone calorimeter
typical of those used for power and energy measurements is illustrated in
Fig. 2. Commercial calorimeters are available which are capable of measur-
ing power of up to several kilowatts and short-time-duration pulses of
greater than 1000joules, using these principles. With the absorbing coatings
generally used on these calorimeters, the spectral range which can be
T H E R M A L DETECTORS 295
Black surface
I
Thermal
sensing
element
FIG. 2. Absorbing cone calorimeter, used for power and energy measurements.
6.3.2. Bolometer
The heating effect of absorbed radiation can be used to cause a change
in the resistance of a metal or semiconductor from its normal resistance at
room temperature in devices which are classified as bolometer detectors.
This change in resistance can be measured when the resistive bolometer
element is incorporated in a bridge circuit in combination with a matching
bolometer element not exposed to radiation to compensate for temperature
changes in the environment. Initially, bolometers were constructed of evap-
orated metal films, but these devices have been noisy and less sensitive than
thermocouples. Thermistor bolometers composed of semiconductor alloys
have been used in modern infrared work.899
A number of specialized bolometer detectors have been applied to
measurements of incident laser pulses at intensities much greater than those
of interest in other applications. Early work on laser-power measurements
utilized a tangled mass of enamel-insulated wire contained in an enclosure."
These "rats-nest" calorimeters effectively absorbed all the incident optical
energy after a sufficient number of reflections from the wire, and the change
in resistance of the wire was used directly as a measure of incident power.
The system could be calibrated by passing a known current pulse through
the conductor. The upper limit on the optical energy before the thin insulat-
ing coating on the wires was damaged was typically several joules per square
centimeter. Several thin-film bolometers have also been used as room-
temperature detectors of high-power-pulsed 10.6-km radiation.'' The low
296 DETECTORS
mass of these devices permits response times of less than 1 psec., and at
least one device has been reported capable of withstanding 100-5 input
energy.I2
Two important quantities necessary in characterizing the performance of
a sensitive bolometer are the specific heat capacity C and the normalized
temperature coefficient of resistance a, where
1 dR
a=--
R dT'
Here R is the bolometer resistance (ohms) and T is the temperature of the
element (OK). The responsivity of R of an element with a bias voltage V
applied is
R=aV/G, (6.6)
where G is the thermal conductance to a heat sink or bath. A high-
responsivity bolometer thus requires a large temperature coefficient of
resistance, as is observed in most semiconductors at cryogenic temperatures.
Hence, extremely sensitive bolometers can be obtained by using materials
such as carbon,I3 g e r r n a n i ~ m , or
' ~ ~ i l i c o n at
' ~ temperatures below 20°K.
Operation at low temperature also improves the response time T,
75-
C
G'
At low temperatures, the specific heat capacity is, of course, proportional
to T 3so that the speed of response can be made faster according to (T/ Td)3,
far below the Debye temperature, Td.Since both the responsivity and time
constant of a bolometer depend inversely on the thermal conductance,
bolometer performance can be tailored to a particular application by the
degree of thermal contact with the bath. This parameter is usually varied
by altering either the length or the composition of the lead wires to the
element. Carbon bolometers with R L- 1 x lo4V/W at 2°K have been con-
structed, while Ge and Si bolometers with R = 1 x lo5V/ W and response
times of 10 psec have been demonstrated. Since the specific heat capacity
of silicon at low temperatures is eight times smaller than that of germanium,
silicon has a faster response and is used in commercial bolometers. Since
these are thermal detectors, the spectral response is uniform throughout
much of the infrared spectrum and usually limited by the transmission of
windows or radiation shields used to admit radiation to the vacuum-enclosed
bolometer element. Doped-germanium bolometer elements have been used
effectively by Lowl4 for infrared astronomical measurements. Metal
bolometers cooled to near their superconducting transition temperature
where very large temperature coefficients of resistance occur have also been
T H E R M A L DETECTORS 297
suggested. Devices of this type using thin tin films are sensitive infrared
detectors,16 but require a high degree of temperature stability compared to
cooled semiconductor bolometers.
A typical cooled Si semiconductor bolometer arrangement is shown in
Fig. 3.” The bolometer is mounted in a vacuum rather than immersed in
the cryogenic fluid bath to allow the sample temperature to change in
response to incident radiation. The bolometer is connected to wires in good
contact with the temperature bath. Room-temperature radiation is prevented
from heating the bolometer by cooled filters, usually quartz and high-density
black polyethylene. The bolometer is often mounted at the focal point of
a spherical or conical cavity to enhance the absorption of radiation. A
simple dc bias circuit is often used in combination with a low-noise pream-
plifier and a signal-averaging system if the signal is repetitive. The maximum
bias current is limited by dc ohmic heating which will raise the equilibrium
detector temperature, or by excess noise generation in the bolometer as the
bias field is increased. This excess noise occurs at relatively low applied
electric fields preliminary to breakdown effects associated with impact
ionization phenomena, which typically occur at field strengths of a few
volts per centimeter in Ge and Si.
6.3.3.Golay Cell
The Golay cell is a thermal detector with a nearly uniform energy response
from the ultraviolet to the microwave region that has been widely used in
298 DETECTORS
Incident
radiation
Window
FIG. 4. Golay cell and optical system to measure small displacements of the membrane.
THERMAL DETECTORS 299
,Electrodes,
Rv = 7 (wpAR/ G ) (1 + w 2 7 ; ) 1 +O ~ T ~ ) - ” * , (6.8)
where 7 is the emissivity, R is the equivalent circuit resistance, and G is
the thermal conductance, while T~ and T are the electrical and thermal time
constants of the system, respectively. Since the T € and T are on the order
of approximately 1 sec in most cases, the high-frequency limit of this
equation is usually applicable, yielding
(6.9)
where c’ is the volume specific heat and E is the dielectric constant of the
material.
Alternately, if the detector is a current source operating into a low-input
impedance current amplifier, it will have a current responsivity R,
THERMAL DETECTORS 301
R.=-77P (6.10)
' c'h'
where h is the detector thickness. The operating regime is usually in the
high-frequency region where the current responsivity is independent of
frequency, while the voltage responsivity decreases at 6 db/octave. If the
detector is used as a voltage source, the ratio of the pyroelectric coefficient
to the dielectric constant is the important parameter in achieving a high
responsivity, while as a current source, the pyroelectric coefficient alone is
the significant factor.
A good pyroelectric material is typified by a large pyroelectric coefficient,
a small dielectric constant, a low thermal conductivity and heat capacity,
and a high absorption coefficient in the spectral region of interest.22Table
I shows properties of some of the pyroelectric materials of greatest interest.
In addition to a number of inorganic insulating crystals, such as strontium
barrium niobate (SBN)23 with pyroelectric coefficients as high as 1 . 2 ~
lo-' C c K 2 K-', useful pyroelectric effects are observed in polymers such
as polyvinylidene fluoride (PVF2).24*25 From Eqs. (6.8) and (6.9), it is
apparent that the proper choice of a pyroelectric detector material depends
on the required frequency response and mode of operation. In low frequency
applications, SBN with its large pyroelectric coefficient can yield
responsivities as high as 2.5 x lo3V/ W, but its high-frequency performance
Operating
Temperature R" D* T
Detector ( O K ) NEP(Wj (V/W) (cm Hz'"/W) (secj
~
Thermocouple 3 00 lo-"'
2.5~ 1 3x10' 3x
BoI ome t e r 3 00 I x 10' 2 x IOU 1x lo-'
Carbon Bolometer L Ix 10-l' 2 x 104 4.5 x io10 1x
Ge Bolometer 2 5 x lo-'' 1 x 10' 8 x 10"1.ox lo-'
Si Bolometer 2 5 x 10-I) 1 x 10' 2x
TGS Pyroelectric 3 00 X
I 1x10~ 1 x loy 1x lo-)
SBN Pyroelectric 3 00 5x10" 1x lo-'
Golay Cell 3 00 5x10-I' 2x10' 1.6~10~ 1x
6.3.6.Summary
The performance of a number of different thermal detectors is summarized
by the characteristics shown in Table 11. The room-temperature thermal
detectors, such as thermocouples and bolometers, are relatively slow and
exhibit moderately good detection sensitivity as measured by NEP and D*.
The cooled bolometers are significantly more sensitive, have better response
times, but are vastly more difficut to use. The pyroelectric detectors exhibit
a sensitivity similar to the other uncooled thermal detectors and are becom-
ing the standard room-temperature sensor for energy and power measure-
ments.
6.4.1. Photomultiplier
One of the most important direct photon effects used to detect radiation
is the photoemissive effect, also known sometimes as the external photo-
effect. Photoemissive devices are particularly well suited to detection of
low-intensity radiation and offer major advantages in detection of fast
low-level signals. In a photoemissive detector, the action' of incident radi-
ation is to induce emission of an electron from the surface of a photocathode
into space, where it is subsequently collected by an anode. There are a
number of devices in which the photoemissive effect is applied for sensing
in a simple photomultiplier or for imaging radiation in image intensifiers
or image tubes, primarily in the visible spectrum. The most commonly
employed device is the photomultiplier, where the photoemitted electrons
impinge upon a sequence of electrodes called dynodes, which act as secon-
dary electron Each electron which is incident on a dynode
induces emission of one or more additional electrons, so that an
amplification or gain mechanism exists for the production of large numbers
of electrons.
The spectral properties of the photomultiplier are primarily determined
by the photocathode. Photocathodes are generally thought to be of two
types, conventional photocathode^^^ and so-called negative electron affinity
cathodes.33734The two types of cathodes differ from each other in the
magnitude of the photoelectric work function associated with the energy
required to remove an electron from its equilibrium thermal level inside of
a solid to the energy level above the potential barrier at the solid surface
which allows the electron to escape from the solid. The minimum energy
which a photon must possess to induce photoemission equals the work
function, 4, of the metal [Fig. 6(a)]. Within the category of conventional
photocathodes, both metals and semiconductors are used as electrode
materials. The energy band diagrams associated with the photoemissive
properties in these types of materials are illustrated in Fig. 6(a) and (b),
respectively. In the conventional semiconductor electrode, the minimum
energy required for a photon to induce photoemission is still that quantity
which will raise the electron to a level higher than the potential barrier at
the surface of the material [Fig. 6(b)]. Semiconductor photocathodes with
a positive electron affinity generally require less energy than metallic photo-
cathodes and hence have spectral responses which extend to somewhat
longer wavelengths than conventional metallic photocathodes. In either
case, these photocathodes can respond to radiation of wavelengths extend-
PHOTON DETECTORS 305
Metal Vacuum
(a)
Semiconductor Vacuum
Photoexcitation-! \
Conduction
band
Fermi level
Valence band
FIG. 6. Energy band diagrams associated with the photoemissive properties of photo-
cathodes. ( a ) For metals, and (b) for semiconductors.
ing only from the visible spectrum into the near infrared. These material
limitations, which have been encountered over most of the history of the
use of photomultiplier tubes, have permitted only very marginal response
at wavelengths as long as 1 p. In the high-energy ultraviolet part of the
spectrum, the tubes often do not respond to wavelengths shorter than 0.2 p
as a result of the transmission spectra of typical window materials used in
their construction. There have been some extensions of the response to near
0.1 p by using rather delicate materials, such as lithium fluoride, and even
further extension of their response into the X-ray spectrum by using very
thin wafers and material such as gallium arsenide. An alternative procedure
that is used to produce some ultraviolet response in photomultiplier tubes
is to coat the entrance window of the tube with an ultraviolet-sensitive
material which is capable of fluorescing in a region where the tube does
respond sensitively. Materials such as sodium silicate have been used to
produce relatively short wavelength response in conventional tubes by using
this approach.
In recent years, many advances have been made in increasing the sensitiv-
ity in the long-wavelength portion of the spectrum and extending that
response to the wavelengths as long as 1.5 p.33-36 Prior to this work, photo-
306
Semiconductor Vacuum
Conduction band
Photoexcitation +
Fermi leve
Valence band
emissive detectors were not available for this portion of the spectrum. The
extension of spectral response of photomultiplier tubes into the infrared
has resulted from the discovery of the negative electronic affinity photo-
cathode. It was discovered that by overcoating the surface of certain p-type
semiconductors with evaporated layers of low-work-function materials, the
resulting structure has a negative electronic affinity. This is illustrated in
Fig. 7. Here, in contrast to the conventional photocathodes previously
described, the incident photon must have an energy which only equals or
exceeds the energy gap, E,, of a semiconductor to allow photoemission.
Hence, by judicious choice of the semiconductor and associated coating,
it is feasible to construct photocathodes where the spectral response does
extend well into the near infrared. Even here, however, the quantum
efficiency does diminish at increasing wavelengths, so that there is yet a
relatively severe limitation on use of photomultiplier systems at wavelengths
exceeding 1.2 p.
The conventional photomultiplier tubes are usually based on a photo-
emissive structure where a thin evaporated layer composed of a compound
including some alkali metal (with the predominate example being that of
cesium), metallic elements from Group V of the periodic table, such as
antimony and other species, are combined in a ternary or similar alloy.
Examples of the spectral response of representative conventional photo-
cathode?’ are shown in Fig. 8. These show some of the conventional
materials designated by the typical nomenclature referring to the photo-
cathodes as S1, S17, and S20. For example, the S1, which is silver (Ag)
on cesium oxide (CsO),has some response to wavelengths as long as 1.1 p.
The conventional photocathode with the most sensitivity in the red portion
of the spectrum is the so-called S 2 0 , composed of cesium-sodium-
potassium-antimony [ (Cs)Na2KSb].This system is able to respond, at least
weakly, to almost 0.8 p. Some extended response versions of the S 2 0
PHOTON DETECTORS 307
1 1 1 1 I 1 I
0.4 0.6 0.8 1
Wavelength (rm)
I
I 1
I I
I
1
Dynodes
II I
I
I
Focusing Photocathode
electrode (semitransparent)
Because of the high current amplification and the noise, the photomulti-
plier is one of the most sensitive instruments for detection of visible and
near infrared radiation. The photomultipliers have been used to detect
optical power levels as low as lo-'' W. As a detector of an optical signal
modulated at some low frequency, the limiting factor in sensitivity is the
so-called dark current of the photomultiplier. The random fluctuations
observed in a photomultiplier output are due to either cathode or dynode
shot noise, Johnson and 1/ f noise. The cathode shot noise is composed of
a fluctuating current emitted by the photocathode due to incident signal
power and another component observed in the absence of radiation, which
is the so-called dark current arising from random thermal excitation of
electrons. All electrons in the cathode which are thermally excited to energies
greater than the work function or electron affinity are emitted, and each
emitted electron is then effectively replaced by another electron from the
cathode. The minimum detectable optical power is determined by the
magnitude of this thermally generated dark current and typically leads to
a minimum detectable power on the order of W for simple video
detection. An obvious means of reducing the thermal dark current is to cool
the photo~athode.~'*~" If the temperature of a conventional photoemitter is
lowered to at least -2O"C, the thermal dark current becomes negligible. For
systems with lower thermal activation energy, such as the negative electron
affinity emitters, correspondingly lower temperatures are required. Cooling
of this type has traditionally been achieved by using coolants such as dry
ice to reduce the temperatures to a point where thermal carrier generation
is reduced to negligible levels. More recently, thermoelectric cooling has
been found to be more convenient and able to achieve temperatures which
are precisely those required for optimum cooling, without reducing the tube
temperature to a point where factors such as moisture and condensation
become troublesome.
hu 2 Ed
\G Donor level
hv > Eg
ductor may take place from states near the top of the valence band across
the energy gap into states near the bottom of the conduction band, producing
excess electron-hole pairs, or alternately from impurity levels creating either
excess electrons or holes (Fig. 12). The former are designated intrinsic
photoconductors, while the latter, in which the photoexcitation is from
impurity levels, are termed extrinsic. Extrinsic detectors may incorporate
either donor or acceptor impurities, leading to sensors where the radiatively
induced carriers are either electrons or holes, respectively. Extremely long
wavelength detection may be obtained by electron intraband excitations in
materials such as InSb, yielding response at wavelengths in the 100-km to
1-mm range.
Intrinsic photoconductors respond to photons of sufficient energy to cause
an interband transition on being absorbed, thus creating excess electron-
hole pairs in the semiconductor. The semiconductor will absorb only
photons with energy greater than the energy gap, E B ,so that the maximum
wavelength A which can be detected is
(6.11)
where h is the Planck’s constant and c is the speed of light. The photodetector
can respond to all radiation with wavelengths shorter than A,,,, with
maximum sensitivity for photon energies slightly greater than the gap energy.
The photoresponse of several common infrared photoconductors is shown
in Fig. 13. The variation in spectral sensitivity for photon energies greater
PHOTON DETECTORS 313
than the gap arises from the strong absorption of the radiation which
produces excitation only near the surface. The location of the maximum in
the spectral response is dependent on the thickness of the crystal and the
carrier recombination parameters.
The spectral regions in which intrinsic semiconductor photoconductors
are capable of responding depend on the energy band gaps in elemental
and compound semiconductors which have been developed. As shown in
Table 111, the range from the visible spectrum to approximately 8 pm is
well covered by a number of common semiconductors. While the points
shown are for a fixed temperature, the band gaps of most semiconductors
vary strongly with temperature, so that the maximum sensitivity for a given
material will change if the detector is cooled. For example, when InSb is
cooled from 300°K to 77"K, the energy gap increases from 0.18 eV to 0.23 eV
causing a reduction in A,, from 7 pm to 5.5 pm. Materials such as PbS
314 DETECTORS
Si 1.10 1.12
Ge 0.68 1.82
lnAs 0.33 3.75
InSb 0.18 6.88
PbS 0.4 3.1
PbSe 0.25 4.95
PbTe 0.31 4.0
GaAs 1.4 0.88
CdSe 1.74 0.71
CdS 2.4 0.52
Hg,-,Cd,Te( x = 0.2) 0.09 14
Pb,-,Sn .Te(x = 0.2) 0.083 15
I I I I
1.6 -
1.4 -
1.2 -
-
m
-
-
-
-
-0.4
0 0.2 0.4 0.6 0.8 1.0
HgTe CdTe
FIG. 14. Dependence of the resulting band gap on the fractional composition of the
compounds in Hg,-,Cd,Te.
Pump out
PO rt
Electrical leads
(6.12)
6.3.3.Extrinsic Photoconductors
The requirement for quantum detectors beyond wavelengths of approxi-
mately 10 pm frequently makes it necessary to use extrinsic semiconductors.
Impurity or defect states in a semiconductor create discrete energy levels
in the band gap of a semiconductor with relatively low activation energies.
Carriers can be photoexcited from these levels near the conduction or
valence band edges, creating a decrease in the detector’s electrical resistance
which can then be measured, as was previously shown in Fig. 12. Impurity
levels near the conduction band edge are called donor levels, since photoex-
citation of carriers will cause extra electrons to be added to the number
already in the band at thermal equilibrium. Levels adjacent to the valence
band edge are called acceptor levels, since excitation of carriers into these
levels will leave excess holes in the band. The nature of the impurity level
depends on the valence of the impurity relative to the host semiconductor.
Since these are states with very low activation energies, detectors based on
photoexcitation from these levels must be cooled to prevent thermal excita-
tion from masking the photoeffect. The technique is limited to semiconduc-
tors of a high degree of purity and crystal perfection where the level of
incidental impurities and defects is small.
In materials which d o meet these requirements, the ionization energy of
isolated impurities can be approximately treated by a description similar
to an isolated hydrogen atom. The ionization energy in a semiconductor
with a dielectric constant K and an effective carrier mass m* is given by
m*
E = 13.6 -
m K 2 eV9 (6.13)
Conduction band
- cu Egap = 0.75 eV
Au
0 2
-
0.33
Zn
- Ga
0.02
025 0.04
-
0.03
Valence band
Solay cell
\ Ge:Au
I I 1 1 1 1 l 1 I I I I I 1 Ill I I 1 1 1 1 1 1
1.o 10 100 1000
Wavelength ( r m )
Vs=(:)V", (6.14)
with (T being the total detector conductivity. For an absorbed power per
unit volume P,/Ax, the signal photoconductivity induced by the radiation
320 DETECTORS
Photo
voltage
Bias
SUPPlY
Load resistor
Radiation
-/ 1 1 1
/-
/
/ Semiconductor /
contacts
FIG. 18. Measurement of the photosignal generated when radiation is incident on a photo-
conductive detector.
is
(6.15)
(6.17)
where Af is the noise bandwidth and vT is the total noise voltage. Reducing
the detector temperature will reduce the thermal noise component, so that
the noise will be dominated by generation-recombination noise. The total
effect of cooling is a significant increase in detectivity, as was shown for
several intrinsic photoconductors in Fig. 13.
In addition to specification of responsivity and detectivity, the perform-
ance of either extrinsic or intrinsic photoconductors under steady-state
incident radiation can be described in terms of the short-circuit photocurrent
or the open-circuit photovoltage modes. The short-circuit photocurrent Z,,
is typically given by Z,, = TqNg, where 7 is the quantum efficiency, q is the
electronic charge, N is the number of photons absorbed in the sample per
unit time, and g is the gain of the photodetector. If the photoconductive
gain and photon-absorption rate are expressed in terms of fundamental
materials parameters, the short-circuit photocurrent can then be written as
(6.18)
(6.19)
Here the detector resistance was expressed in terms of its majority carrier
concentration n, mobility p, and respective length, width, and thickness
1, w, and x. As in the case of the short-circuit current, the open-circuit voltage
is then dependent on the incident power, the wavelength, and bias and is
inversely proportional to the majority-carrier concentration in the device.
These expressions for short-circuit current and open-circuit voltage, are
low-frequency or steady-state descriptions, and do not express information
about the response time of the device which is crucial in many applications.
Since a photoconductor is, in fact, a majority-carrier device, the frequency
response of the open-circuit voltage V, is then given by the expression
Incident
radiation
I 1
Contact
N-region P-region
Conduction band
Valence band
FIG. 19. (a) A simple p-n junction photodiode. (b) Associated energy-band diagram.
324 DETECTORS
Current
Saturation
current
Dark \
current Ill Voltage
Photocurrent
Photocurrent (short circuit)
Bias -
IT-$ Radiation
Photodiode
+~
source+ '6
I+ RL I Load
resistor Photovoltage
FIG. 20. (a) Observation of photovoltage and photocurrent in a photovoltaic device. (b)
Circuit to observe the photovoltage.
charge region between the p and n regions. This charge separation creates
an open-circuit voltage across the junction and leads to a voltage which
can be detected that is proportional to the absorbed photon density. The
direct observation of this photovoltage is the simplest and most straightfor-
ward mode of operation of a photovoltaic detector [Fig. 20(a)]. Junction
detectors are often operated as well under some form of reverse bias voltage
which leads to observation of a photocurrent [Fig. 20(a)] rather than a
photovoltage and are said to be operated in a photoconductive mode, despite
the fact that it is not equivalent to the photoconductive mode previously
described.
The short-circuit current associated with the photovoltaic detector is
identical in form to that for a photoconductive detector, with the exception
PHOTON DETECTORS 325
that the gain is unity, that is, the short-circuit current is given by
nqPA
I,=-, (6.20)
hc
The open-circuit voltage, which is the commonly observed photovoltaic
signal, is obtained as the product of the short-circuit current with the
dynamic resistance of the junction in the absence of bias. Hence the
open-circuit photovoltage V,, for a photovoltaic detector is approximately
(6.21)
Here I,,, is the saturation current of the junction in the absence of a photo
signal, k is Boltzmann’s constant, and T is the absolute temperature. Just
as in the case of the photoconductive detector, the photovoltaic signal
depends on the incident power and wavelength, but it does not depend on
the majority carrier lifetime. The photovoltaic effect, in fact, is essentially
independent of the majority-carrier lifetime, since in creating an electron-
hole pair, the determining factor is the minority-carrier lifetime, which is
significantly shorter. Since the frequency response is determined by the
product of the modulation frequency of the radiation w and the minority-
carrier lifetime 7, the response is essentially independent of frequency for
values of W T < 1. The point at which W T becomes unity is at a significantly
higher frequency for a photovoltaic device; therefore, photovoltaic detectors
generally display a higher frequency response or a faster time response than
photoconductive detectors of exactly the same material.
The most widely used photovoltaic detectors for the visible spectrum rely
on the common elemental semiconductors Ge and Si with cutoff wavelengths
of 1.5 pm and 1.1 pm, r e s p e c t i ~ e l yFor
. ~ ~applications at longer wavelengths,
photovoltaic structures have been developed using 111-V compounds or the
alloy systems Pb, -,Sn,Te and Hg, -,Cd,Te. For optical communications
systems, detectors utilizing Ge”,52 or 111-V alloys, such as
InxGal-xAsyP1-y,53-55 have been emphasized with peak response in the
1.1- 1.7-pm range. For longer-wavelength infrared extending to wavelengths
as long as 14 pm, the Hg,-,Cd,Te and Pb,-,Sn,Te alloy systems have been
vigorously de~eloped.’~-’~
While the discussion of photovoltaic detection has centered on the
classical p-n junction, there are a number of structures which have been
employed for photovoltaic detection. The most direct extension of the simple
p-n junction is the p-i-n photodiode which incorporates an intrinsic region
between the p and n portions of the traditional junction. If either the p or
n side of the junction is made very thin compared to the optical absorption
length in the medium, the incident optical radiation will readily penetrate
into the intrinsic region. In the intrinsic region, the absorption will produce
a large number of electron-hole pairs. As these electron-hole pairs are
produced in a region of high electric field, the carriers will rapidly be drifted
out of that region into the adjacent p and n sides of the junction. This rapid
carrier separation leads to an improved frequency response and efficiency
in the p-i-n diode as compared to that of a pn diode of a comparable
material.
In addition to being used to enhance detection, p-i-n configurations have
also been employed in so-called avalanche photodiodes.""."' An avalanche
photodiode is a structure with an internal gain achieved through the so-called
avalanche breakdown effect, which occurs in the high field regions of a p-n
or p-i-n junction under high reverse bias. The carriers generated by photo-
excitation are accelerated to velocities which are sufficiently high so that
upon collision with atoms in the lattice, additional free electrons are gener-
ated by the impact ionization process. These free electrons are also acceler-
ated and undergo more collisions and hence will also lead to the generation
of more free electrons. The so-called avalanche effect, which occurs within
a high field region in a reverse bias p-n or p-i-n junction, can lead to a
significant internal gain."2 Hence, a larger photosignal is produced in a
material with the same incident radiation power than would be produced
under conventional diode operation. While the photogenerated current for
a p-i-n diode is as previously described by Eq. (6.20), for conventional
photovoltaic devices the value of current, I,", in an avalanche diode is
nqPA
I,, = -M , (6.22)
hc
where M is the avalanche gain. The gain depends on device design and
bias conditions and may be as high as 100. The avalanche process is a
statistical one, where M fluctuates for each chain of events. Hence the
amplification process in these devices is subject to an additional noise-
generation process.
In many semiconductors, particularly those of significance for infrared
detection, production of both n- and p-type materials of high quality is not
always possible within the current state of the art. A method for forming
photovoltaic structures in materials where p-n junctions cannot be formed
is through creation of a Schottky barrier p h ~ t o d i o d e . " ' .A
~ ~Schottky barrier
is a junction formed at the interface of a metal semiconductor structure. In
a manner analogous to that described for the p-n junction, a potential
barrier is created at the metal-semiconductor interface, which will effectively
separate electron-hole pairs generated by incident optical radiation. A
diagram of a typical Schottky barrier junction is shown in Fig. 21. In most
PHOTON DETECTORS 321
EF -
Metal
Semiconductor
6.5. Noise
noise, the noise voltage and current also depend on the resistance value of
the detector. The mean square Johnson noise power, which then appears
even in the absence of electrical bias as a fluctuating current or voltage, is
given by P, = kTB, where k is the Boltzmann's constant, T is the absolute
temperature of the detector, and B is the measurement of bandwidth. The
open-circuit noise voltage V, is then given by
V, = (4kTRB)'/2, (6.23)
where R is now the resistance of the detector. In most solid state photodetec-
tors, the Johnson noise is present at all frequencies, but it is not the dominant
noise, except at high frequencies. At lower frequencies, several other sources
of noise dominate. This is illustrated in Figure 22, which illustrates an ideal
noise spectrum for a typical solid-state photodetector in the absence of
radiation. At low frequencies, another noise source, called l/f noise, domi-
nates, while at intermediate frequencies generation-recombination noise
dominates. The specific ranges over which each of these noise sources is
predominant varies according to the details of the composition and structure
of the detector, but typically below 1 kHz is the range for l/f; which is the
principal noise source. Between 1 kHz and as high as 1 MHz, generation-
recombination noise would be the dominant noise, while above that
frequency, the Johnson noise would be the predominant noise source.
At the lowest frequencies, the source of noise found in most solid-state
detectors is l/f noise characterized by a spectrum where the noise power
I I I I 1 1 I I 1 I
- -
-
Generation-recombination noise
-
Johnson
-
noise
I I 1 I I I I I I I
2 4 6 8 10
I =(%)([*]””
+ 1 w2r2 (6.25)
where p is the total number of free holes in the sample, r is the free-carrier
lifetime, B is the measurement bandwidth, and n is the total number of
electrons in the material. It is clear that when the frequency becomes large
with respect to the free-carrier lifetime, the noise power would diminish,
which will limit the upper frequency limit over which generation and
recombination noise is an important factor.
OPTICAL WINDOW MATERlAL 33 1
References
1. R. C. Jones, “Noise in Radiation Detectors,” Proc. Inst. Rad. Eng. 47, 1495 (1959).
2. R. A. Smith, F. E. Jones, and R. P. Chasmar, The Defection and Measurement of Infrared
Radiation, Oxford Univ. Press, London, 1968.
3. E. S. Barr, “The Infrared Pioneers, JI; Macedonio Melloni,” Infrared Phys. 2, 67 (1962).
4. A. H. Wilson, The Theory of Metals, Cambridge Univ. Press, London, 1954. Also H. P. R.
Frederikse, V. A. Johnson, and W. W. Scanlon, in Methods of Experimental Physics, Vol.
6B, Chap. 7 (L. Marton, ed.), Academic Press, New York, 1959.
5. G. W. Day, 0. L. Caddy, and R. J. Iversen, “Detection of Fast Infrared Laser Pulses with
Thin Film Thermocouples,” Appl. PHys. Lett 13, 289 (1968).
6. D. A. Jennings and E. D. West, “A Laser Power Meter for Large Beams,” Rev. Sci. Insfr.
41 565 (1970).
7. J. W. Stearn, “A Radiometer for Continuous Wave Laser Radiation,” J. Sci. Instr. 44, 218
(1967).
8. R. De Waard and E. M. Wormser, “Description and Properties of Various Thermal
Detectors,” Proc. Insf. Rad. Eng. 47, 1508 (1959).
9. G. Koren, Y. Yacoby, H. Lotem, M. Kosower, and G. Greenwald, “Thin Film Calorimeter
for Low-Energy Laser Pulse Measurements,” Appl. Phys. Left. 23, 73 (1973).
10. A. J. Schmidt and R. C. Greenhow, “A Practical Laser Output Meter,” J. Phys. E 2, 438
(1969).
11. B. Contreras and 0. L. Caddy, “Nanosecond Response Time Room-Temperature Infrared
Detection with Thin Film Bolometers,” Appl. Phys. Left. 17, 450 (1970).
12. J. Stricker and J. Rom, “A Fast Response Thin Film Wattmeter for Measurements of CO,
Laser Power,” Rev. Sci. Instr. 43, 1168 (1972).
13. W. S. Boyle and K. F. Rodgers, “Performance Characteristics of a New Low-Temperature
Bolometer,” J. Opr. SOC.Am. 49, 66 (1959).
14. F. J. Low, “Low Temperature Germanium Bolometer,” J. Opt. SOC.Am. 51, 1300 (1961).
15. R. Bachmann, H. C. Kirsch, and T. H. Geballe, “Low Temperature Silicon Termometer
and Bolometer,” Rev. Sci. Insrr. 41, 549 (1970).
REFERENCES 333
45. T. C. Harman and 1. Melngailis, Applied Solid State Science, p. 1, 4th ed. (R. Wolfe, ed.),
Academic Press, New York, 1974. Also: I. Melngailis, J. Luminescence, 7, 505 (1973).
46. 1. Melngailis and T. C. Harman, in Semiconductors and Semimetals, p. 11 1, 5th ed. (R. K.
Willardson and A. C. Beer, eds.), Academic Press, New York, 1970.
47. R. B. Emmons, “The Frequency Response of Extrinsic Photoconductors with Reduced
Background,” Infrared Phys. 10, 63 (1970).
48. P. R. Bratt, in Semiconductors and Semimetals, Vol. 12, p. 39 (R. K. Willardson and A. C.
Beer, eds.), Academic Press, New York, 1977.
49. H. Levinstein and J. Mudar, “Infrared Detectors in Remote Sensing,’’ Proc. IEEE 63, 6
(1975).
50. R. G. Smith, “Photodetectors for Fiber Transmission Systems,” Proc. IEEE 68,1247 (1980).
51. J. Conradi, “Planar Germanium Photodiodes,” Appl. Opt. 14, 1948 (1945).
52. H. Melchior and W. T. Lynch, “Signal and Noise Response of the High Speed Germanium
Avalanche Photodiodes,” IEEE Trans. Electron Devices, ED-13, 829 (1966).
53. M. Feng, J. D. Oberstar, T. H. Windhorn, L. W. Cook, G. E. Stillman, and G. E. Streetman,
“Be-Implanted 1.3 pm InGaAsP Avalanche Photodetectors,” Appl. Phys. Lett. 34, 591
(1979).
54. T. P. Lee, C. A. Burrus, Jr., and A. G. Dentai, “lnGaAsP/InP Photodiodes: Microplasma-
Limited Avalanche Multiplication at 1-1.3 k m Wavelength,” IEEE J. Quantum Electron,
QE-15, 30 (1979).
55. M. A. Washington, R. E. Nahory, M. A. Pollack, and E. D. Beebe, “High Efficiency
In,~.,Ga,As,.P,_,/lnP Photodetectors with Selective Wavelength Response Between 0.9
and 1.7 km,” Appl. Phys. Lett. 33, 854 (1978).
56. W. H. Rolls and D. V. Eddolls, “High Detectivity Pb,Sn,-,Te Photovoltaic Diodes,”
Infrared Phys. 13, 143 (1973).
57. C. C. Wang and S. R. Hampton, “Lead Teluride-Lead Tin Teluride Heterojunction Diode
Array,” Solid State Electron. 18, 121 (1975).
58. A. M. Andrews, J. T. Longo, J. E. Clarke, and E. R. Gertner, “Backside-Illuminated
Pb, - ,Sn,Te Heterojunction Photodiode,” Appl. Phys. Lett. 26, 439 (1975).
59. D. Eger, M. Oron, A. Zussman, and Z. Zemel, “The Spectral Response of PbTe/ Pb, -,Sn,Te
Heterostructure Diodes at Low Temperatures,” Infrared Phys. 23, 69 (1983).
60. G. E. Stillman and C. M. Wolfe, in Semiconductors and Semimetals, Vol. 12 (R. K.
Willardson and A. C. Beer, eds.), Academic Press, New York, 1977.
61. P. P. Webb, R. J. Mclntyre, and J. Conradi, “Properties of Avalanche Photodiodes,” R C A
Reu. 35, 234 (1974).
62. A. G. Chynoweth, in Semiconductors and Semimetals, Vol. 4 (R. K. Willardson and A. C.
Beer, eds.), Academic Press, New York, 1983.
63. S. M. Sze, Physics ofSemiconductor Deoices, Wiley, New York, 1983.
64. E. M. Logothetis, H. Holloway, A. J. Varga, and E. Wilkes, “Infrared Detection by Schottky
Barriers in Epitaxial PbTe,” Appl. Phys. Left. 19, 318 (1971).
65. A. F. Gibson, M. F. Kimmitt, and A. C. Walker, “Photon Drag in Germanium,” Appl.
Phys. Lett 17, 75 (1970).
66. A. F. Gibson, C. A. Rosito, C. A. Raffo, and M. F. Kimmit, “An Optical Bridge for the
Assessment of Modern Locked C 0 2 Laser,” J. Phys. D 5, 1800 (1972).
67. P. J. Bishop and A. F. Gibson, “Absorption Coefficient of Germanium at 10.6 pm,” Appl.
Opt. 12, 2549 (1973).
68. W. B. Davenport and W. L. Root, A n Introduction to the Theory of Random Signals and
Noise, McGraw-Hill, New York, 1958.
A Brewster angle, 116, 121, 122
Abbe theory of the microscope, 86 brightness temperature, 241
aberration correction by holography, 197 Burch interferometer, 20
aberrations, 83
in holography, 172
Achromatic
devices 1 and 2, 133 C
wave plates, 144 channeled spectrum, 28
acoustical holography, 197 character recognition, 197
acousto-optic effect, 150 chromatic dispersion in gratings, 72
Airy disc, 67, 80 circular
angle measurements, 41 aperture, diffracting, 58, 66
angular resolution of telescopes, 81 polarization, 110
anisotropic media, 133 polarizers, 124
antireflection coatings, 33 coded reference beam for color holograms,
astronomical radiometers, 249 182
atmospheric sounding techniques, 236 coherence length, 6, 41
color holograms, 180
common-path interferometers, 19
compensated interferometer, 11
B compensating
plate, 11
Babinet
retardation plate, 143
compensator, 143
compensators, 143
principle, 68
complex plane representation of
band-pass filters, 33
polarization, 118
beam expansion in holography, 200
composite holography, 184
beam splitters, 33
computer hologram, 189
Bennett-Koehler reflectometer, 274
conjugate image in holography, 170
Bessel function, first order, 66
Cornu spiral, 56
binary hologram, 189
Cotton-Mouton effect, 153
birefringence
crystal
in crystals, 133
polarizers, 138
induced, 146
retardation plates, 142
blackbody
types, 135
curves, 223
cubic crystals, 135
radiation, 219
blaze angle in gratings, 78
blazed gratings, 78, 158
bleaching of holograms, 204
0
bolometers, 295
Bragg depolarizers, 153
effect in holograms, 172 detection, synchronous, 279
regime, 151 detector polarization, 158
335
336 INDE X
M
MacNeille polarizing prisms, 130
0
Mach-Zehnder interferometer, 12, 17
magneto-optic effect, 153 off-axis holography, 168
matched filtering. 197 one-step method for rainbow holograms.
materials for optical windows, 331 186
Maxwell equations, 108, 134 optic axis of a crystal, 135
measurement optical
of directional reflectance, 274 activity, 138
of emissivity, 269 elements, holographic, 197
of laser power, 277 polarization, 107
of reflectivity, 272 transfer function, 82, 84
measuring wave plates, 145 window material, 331
mechanical stability in holography, 200 ordinary beam, 136
metal grid polarizers, 157 orthoscopic recording in holography, 186
metal mirrors protection, 33
metals, wave propagation for, 120
meter definition, 41
Michelson P
fringes, 12 P-polarization, 114
interferometer, 4, 9 Parseval’s theorem, 69
stellar interferometer, 22 partial coherence, 89
Michelson- Morley experiment, 38 partially transparent materials, 231
microscopy, 42 path radiance, 236
microtopography. 29 phase
Miyamoto- Wolf theory, 53 contrast microscope, 90
modulation transfer function, 85 difference in gratings, 72
Mooney rhomb, 133 holograms, 190
Mueller’s matrices, 159 modulation, 146
multilayer films, 35, 128 plate, 91
multiple reflection interferometer, 30 retarders, 126, 130
multiple reflections in a thin film, 126 shift upon reflection, 115
multiple-beam interferometer, 25 velocities in a crystal, 136
multiple-pass photocathodes, spectral response of, 307
Fizeau interferometer, 37 photoconductive
Twyman-Green interferometer, 37 detectors, 3 1 1
interferometers, 37 response, 3 19
multiple-reflection photoconductors, extrinsic, 317
Fizeau interferometer, 28 photographic emulsions for holography, 201
interferometers, 25 photometers, 246
lNDEX 339
photometric pyroelectric
and radiometric detector, 299
nomenclature, 214 materials, 301
symbols, 214 pyroheliometers, 246
units, 214
measurements, 263
standards, 264 Q
units, 286 quantum detector, 303
photometry, 213 quarter-wave stack, 35
photomultiplier, 304, 309
photon detectors, 303
photon-drag detector, 327 R
photoresists for holography, 204 radial shear interferometer, 18
photovoltaic devices, 322 radiance temperature, 241
pinhole camera, 61 radiation temperature, 239
plane hologram techniques, 181 radiative transfer, 223
plane-holograms, 169 in a vacuum, 228
Pockels electro-optic effect, 147 in transparent media, 231
PoincarC sphere, 112, 142 radiometer,
point spread function, 85 interferometric, 261
polarization prism disperser, 260, 261
circular, 110 radiometers, 246
determination, 159 astronomical, 249
effects, 128 filter, 257
in metals, 120 imaging, 252
in thin films, 125 scanning, 252
elliptical, 113 space defense, 250
linear, 110 standard, 246
mathematical description, 159 radiometric
optical, 107 instruments, 246
scramblers, 153 measurements, 263
polarizer standards, 264
made of four mirrors, 124 temperature measurements, 239
made of three mirrors, 123 radiometry, 213
polarizers, of visible light, 284
grating, 157 rainbow
linear, 126, 130 cylindrical hologram, 189
metal grid, 157 holograms, 184
slit, 157 Raman-Nath regime, 151
polarizing prisms, MacNeille, 130 ratio temperature, 241
principal angle, 122 difference, 243
prism Rayleigh
beam splitters, 138 interferometer, 22
disperser radiometer, 260 scattering, 99
polarizers, 139 real-time holographic interferometry, 191
processing of holograms, 204 recording of holograms, 204
pseudoscopic configuration in holography, rectangular aperture, 65
187 reference beam in holography, 170
pupil function, 68 reflectance, directional measurement of. 274
340 INDEX
V wave-front topography, 40
white-light compensation, 6
visible light, radiometry of, 284
Wien displacement law, 220
volume hologram, 172
Wollaston prism, 138
techniques for color holograms, 183
Wood anomalies, 157
volume-reflection hologram, 174
W
wave
equations, 108 Y
plates Young's experiment, 1
achromatic, 144
measuring, 145
propagation
for metals, 120
in anisotropic media, 133 2
in isotropic media, 113 zone plate, Fresnel, 59
This Page Intentionally Left Blank