Applied

Prismatic and
Reflective Optics

Dennis F. Vanderwerf

Bellingham, Washington USA

Library of Congress Cataloging-in-Publication Data

Vanderwerf, Dennis.
Applied prismatic and reflective optics / Dennis Vanderwerf.
p. cm. – (Press monograph ; 200)
Includes bibliographical references and index.
ISBN 978-0-8194-8332-4
1. Lenses–Design and construction–Mathematics. 2. Mirrors–Design and
construction–Mathematics. 3. Prisms–Design and construction–Mathematics. 4.
Fermat’s theorem. 5. Refraction. 6. Reflection (Optics) I. Title.
QC385.2.D47V36 2010
681 0.423–dc22
2010021193

Published by

SPIE
P.O. Box 10
Bellingham, Washington 98227-0010 USA
Phone: +1 360.676.3290
Fax: +1 360.647.1445
Email: Books@spie.org
Web: http://spie.org

Copyright
c 2010 Society of Photo-Optical Instrumentation Engineers
All rights reserved. No part of this publication may be reproduced or distributed in
any form or by any means without written permission of the publisher.
The content of this book reflects the work and thought of the author(s). Every
effort has been made to publish reliable and accurate information herein, but the
publisher is not responsible for the validity of the information or for any outcomes
resulting from reliance thereon.
Printed in the United States of America.
About the cover: The image on the cover shows linear Fresnel solar concentration
at work. The solar concentrator powers an air-conditioning system at South West
Gas Corporation’s facilities in Phoenix, Arizona. The system was designed by
HelioDynamics (photograph courtesy of Lee Langan).
Contents

Preface ............................................................................................................... xiii

Chapter 1 Introduction and Background ........................................................ 1

1.1 Snell’s Law of Refraction .............................................................................. 1
1.2 Optical Dielectric Materials ......................................................................... 2
1.3 Fresnel Reflection at a Dielectric Surface................................................ 4
1.4 External Reflection at an Optical Surface ................................................ 5
1.5 Internal Reflection at an Optical Surface ................................................. 6
1.6 Reflection Phase Shifts at a Planar Interface .......................................... 7
1.7 Antireflection and Reflection Coatings ..................................................... 9
1.8 Effective f /# of a Converging Light Beam ............................................. 9
1.9 Refraction and Translation of Skew Rays at Planar Surfaces ........... 10
1.10 Convergent Beam through a Tilted Plate ................................................. 13
1.11 Reflection and Translation of Skew Rays at Planar Surfaces ............ 17
1.12 Reflection Matrix ............................................................................................. 18
1.13 Orientation of Viewed Images through Prisms ...................................... 18
1.14 Intersection Coordinate Matrix ................................................................... 19
1.15 Three-Mirror Beam-Displacing Prism ...................................................... 21
1.16 Refraction Matrix ............................................................................................. 24
1.17 Four-Mirror Beam-Displacing Prism ........................................................ 25
1.18 90-deg Beam-Deviating Prism..................................................................... 28
References ....................................................................................................................... 32

Chapter 2 General Prisms and Reflectors ...................................................... 33

2.1 Equilateral Prism .............................................................................................. 33

2.2 Abbe Dispersing Prism .................................................................................. 35
2.3 Pellin–Broca Dispersing Prism ................................................................... 36
2.4 Penta Prism......................................................................................................... 38
2.5 Right-Angle Prism ........................................................................................... 39
2.6 Porro Prism......................................................................................................... 40
2.7 Dove Prism ......................................................................................................... 42
2.8 Brewster Laser-Dispersing Prism ............................................................... 44
2.9 Littrow Prism ..................................................................................................... 46

v
vi Contents

2.10 Schmidt Prism ................................................................................................... 49

2.11 Pechan Prism ..................................................................................................... 53
2.12 Schmidt–Pechan Prism................................................................................... 54
2.13 Cube-Corner Retroreflector .......................................................................... 56
References ....................................................................................................................... 60

Chapter 3 Polarization Properties of Prisms and Reflectors ......................... 61

3.1 Prisms Producing Polarized Light .............................................................. 61

3.1.1 Uniaxial double-refracting crystals ........................................... 61
3.1.2 Nicol polarizing prism ................................................................... 61
3.1.3 Glan–Foucault polarizing prism ................................................. 63
3.1.4 Glan–Thompson polarizing prism ............................................. 64
3.1.5 Glan–Taylor polarizing prism...................................................... 64
3.1.6 Beam-displacing polarizing prism ............................................. 65
3.1.7 Wollaston polarizing prism .......................................................... 66
3.1.8 Nomarski polarizing prism ........................................................... 67
3.1.9 Rochon polarizing prism ............................................................... 67
3.1.10 MacNeille polarizing beamsplitter cube .................................. 68
3.1.11 Birefringent multilayer reflective polarizing film ................. 70
3.1.12 Polarizing beamsplitter elements using birefringent
polarizing film ................................................................................... 71
3.1.13 Wire-grid polarizing beamsplitter .............................................. 72
3.1.14 Polarizing beamsplitter using frustrated total internal
reflection ............................................................................................. 73
3.1.15 Polarizing beamsplitter prism with common polariza-
tion output .......................................................................................... 74
3.2 Prisms Controlling the Polarization of Light .......................................... 75
3.2.1 Fresnel rhomb retarders ................................................................. 75
3.2.2 Total-internal-reflecting cube-corner retarders ...................... 78
3.2.3 Phase-coated total-internal-reflecting right-angle prism
retarders............................................................................................... 80
3.3 Polarization Preservation in Prisms and Reflectors .............................. 82
3.3.1 Polarization-preserving total-internal-reflecting prism ....... 82
3.3.2 Polarization-preserving two-piece reflective axicon ............ 87
3.3.3 Polarization-preserving total-internal-reflecting cube-
corner prism ....................................................................................... 89
3.3.4 Stokes parameters ............................................................................ 89
3.3.5 Depolarizing cube-corner prism ................................................. 90
3.4 Plane of Polarization Rotation Using Total-Internal-Reflecting
Prisms and Reflectors ..................................................................................... 92
3.4.1 90-deg polarization-rotating prism with coaxial beam
output ................................................................................................... 92
3.4.2 90-deg polarization-rotating prism with retroreflected
beam output ....................................................................................... 92
Contents vii

3.4.3 90-deg polarization-rotating prism with orthogonal

beam output ....................................................................................... 92
3.4.4 Double Fresnel rhomb polarization rotator with
collinear beam output ..................................................................... 93
3.4.5 Four-mirror 90-deg polarization rotator with collinear
beam output ....................................................................................... 94
References ....................................................................................................................... 95

Chapter 4 Specialized Prism Types ................................................................ 97

4.1 Dispersing Prism .............................................................................................. 97

4.1.1 Refracting direct-vision prism ..................................................... 97
4.1.2 Reflective dispersing prisms with collinear output .............. 98
4.1.3 Direct-vision prisms with wavelength tuning ........................ 99
4.1.4 Total-internal-reflecting dispersing prism ............................... 99
4.1.5 Multiprism negative dispersion ................................................... 101
4.2 Refracting Achromatic Compound Prism................................................ 101
4.3 Anamorphic Prisms for Beam Compression and Expansion ............ 103
4.3.1 Beam expander with orthogonal output ................................... 104
4.3.2 Beam compressor with coaxial output ..................................... 105
4.3.3 Beam expander with collinear output ....................................... 106
4.3.4 Wedge prism beam compressor/expander ............................... 107
4.3.5 Anamorphic prism pair with coaxial output ........................... 108
4.3.6 Multiprism dispersive compressors and expanders .............. 109
4.4 Achromatic Anamorphic Prism ................................................................... 111
4.4.1 Air-spaced prism pair with coaxial output .............................. 111
4.4.2 Compound prisms with orthogonal output.............................. 113
4.4.3 Refracting/total-internal-reflecting prism pair with or-
thogonal output ................................................................................. 113
4.5 A Misalignment-Tolerant Beam-Splitting Prism................................... 116
4.6 Axicon Prism ..................................................................................................... 116
4.7 A Variable Phase-Shifting Prism ................................................................ 116
References ....................................................................................................................... 119

Chapter 5 Prism and Mirror System Design, Analysis, and Fabrication ....... 121

5.1 Prism Design and Analysis ........................................................................... 121

5.1.1 Sectional element approach for prism design ........................ 122
5.1.2 Right-angle prism sections ........................................................... 124
5.1.3 Experiential design of multiple reflectors ............................... 124
5.1.4 Matrix methods for design and analysis .................................. 125
5.1.5 Evolutionary prism design using a genetic algorithm ......... 126
5.1.6 A three-mirror tabletop lectern projector................................. 127
5.1.7 Prism aberrations ............................................................................. 128
viii Contents

5.2 Prism Quality Specifications ........................................................................ 130

5.2.1 Surface quality and flatness specifications .............................. 130
5.2.2 Optical material properties ........................................................... 130
5.2.3 Specifying angular accuracies ..................................................... 131
5.2.4 Tolerancing a Dove prism ............................................................. 131
5.2.5 Techniques for prism angle measurement ............................... 131
5.3 Survey of Fabrication Methods ................................................................... 135
5.3.1 Ground and polished glass prism ............................................... 135
5.3.2 Fabrication of a Penta prism by measurement of the
angular deviation error ................................................................... 135
5.3.3 Molded, pressed, and fire-polished prisms.............................. 137
5.3.4 Fabrication of large prisms ........................................................... 137
5.4 Some prism-mounting methods .................................................................. 137
References ....................................................................................................................... 138

Chapter 6 A Selection of Prism Applications ................................................. 141

6.1 Laser Scanning .................................................................................................. 141

6.1.1 Reflective scanning prism ............................................................. 141
6.1.2 Refractive prism-beam scanning and steering ....................... 141
6.1.2.1 Single-wedge prism ....................................................... 141
6.1.2.2 Wedge prism pairs .......................................................... 143
6.1.2.3 LADAR guidance system using prism pairs.......... 145
6.1.2.4 Rotating square-plate linear scanner ........................ 146
6.2 Interferometry and Spectroscopy ................................................................ 149
6.2.1 Laser interferometer with prism polarization rotator .......... 149
6.2.2 Polarization interferometer using a Wollaston prism .......... 149
6.2.3 Multipass optical cell for laser interferometer ....................... 150
6.2.4 Nomarski polarized-light interferometer ................................. 151
6.2.5 Aplanatic prism spectrograph ...................................................... 152
6.3 Prismatic Optical Devices ............................................................................. 153
6.3.1 Prism switch for fiber-optic connections ................................. 153
6.3.2 Laser gyro readouts ......................................................................... 153
6.3.3 Reflecting wedge prism for optical reader .............................. 155
6.3.4 Total-internal-reflecting touch switch using a Dove prism 157
6.3.5 Inspection device for window surfaces .................................... 158
6.4 Viewing, Display, and Illumination Systems .......................................... 159
6.4.1 Direct-view system for a microdisplay..................................... 159
6.4.2 Binocular surgical loupe with flare reduction ........................ 160
6.4.3 Inversion prism for range finders................................................ 161
6.4.4 Prism transforming transmitted intensity profile .................. 161
References ....................................................................................................................... 163
Contents ix

Chapter 7 Projection Displays ........................................................................ 165

7.1 Color-Separating and Color-Combining Prisms .................................... 165

7.1.1 Three-channel Philips RGB separating prism........................ 165
7.1.2 Philips prisms in reflective LCD projection displays .......... 166
7.1.3 Crossed dichroic x-cube prisms for projection displays..... 168
7.1.4 Prisms for digital light processing projection ........................ 170
7.1.5 Other types of color-separating prisms for projectors ......... 173
7.2 Polarizing Beamsplitters for Projection Displays ................................. 175
7.2.1 MacNeille polarizing beamsplitters........................................... 175
7.2.2 Cartesian polarizing beamsplitters ............................................. 176
7.2.3 Wire-grid polarizing beamsplitters in projection displays . 177
7.3 Illuminators for Projection Displays.......................................................... 179
7.3.1 Hollow tunnel integrators.............................................................. 179
7.3.2 Solid light pipes................................................................................ 180
7.3.3 Effect of light-pipe cross section on uniformity .................... 180
7.3.4 Solid microprismatic light homogenizer.................................. 181
7.3.5 Tapered-tunnel illuminator for projection displays .............. 183
References ....................................................................................................................... 184

Chapter 8 Microprismatic Arrays .................................................................... 187

8.1 Roof Prism Linear Array ............................................................................... 187

8.2 Square Prismatic Hollow Light Guide ...................................................... 195
8.3 Circular Prismatic Hollow Light Guide.................................................... 197
8.4 Luminaire with Contoured Prismatic Extractor ..................................... 199
8.5 Elliptical Light Guide with Directional Output ..................................... 199
8.6 Prismatic Backlighting Devices .................................................................. 202
8.7 Brightness Enhancement for Liquid Crystal Displays ........................ 209
8.8 Polarizing Prismatic Sheet ............................................................................ 216
8.9 Prismatic Reflective Polarizer Film ........................................................... 217
8.10 LCD Backlights Producing Polarized Light ........................................... 217
8.11 Prismatic Array Beamsplitters and Combiners ...................................... 222
8.12 Polarization Converters Using Prismatic Arrays ................................... 226
8.13 Cube-Corner Arrays ........................................................................................ 228
8.14 Dove Prism Arrays .......................................................................................... 231
References ....................................................................................................................... 234
Some commercial nonsequential ray-tracing programs:.................................. 235

Chapter 9 Fresnel Lenses ............................................................................... 237

9.1 Basic Refractive Fresnel Lens Design ...................................................... 237

9.1.1 Design example: Fresnel lens collimator/searchlight .......... 240
9.2 High-Transmission Fresnel Lens Doublet ............................................... 242
9.3 Reflective Fresnel Lenses .............................................................................. 245
9.3.1 First-surface reflector design parameters ................................. 245
9.3.2 Second-surface reflector design parameters ........................... 247
x Contents

9.4 Refractive Planar Circular Fresnel Lens Solar Applications ............. 248
9.4.1 Multilens solar furnace .................................................................. 248
9.4.2 Multilens-array solar simulator ................................................... 248
9.5 Refractive Meniscus Fresnel Lenses.......................................................... 248
9.6 Reflective Planar Linear-Focus Solar Concentrators ........................... 250
9.6.1 Tilted linear-focus reflective solar concentrator .................... 250
9.6.2 Linear-focus concentrator using a linear Fresnel lens
and a crossed linear total-internal-reflecting array ............... 250
9.6.3 Planar reflective spot-focus concentrator using orthogo-
nal refractive and reflective linear Fresnel lenses ................. 253
9.7 Curved Linear Fresnel Lens Solar Concentrators ................................. 255
9.8 Flexible Fresnel Lens Solar Concentrators.............................................. 260
9.8.1 Sectional planar solar concentrators .......................................... 260
9.8.2 Inflatable curved solar concentrators ........................................ 260
9.9 Fresnel Lenses Using Total Internal Reflection ..................................... 261
9.9.1 Low-profile overhead projector................................................... 262
9.9.2 Curved catadioptric Fresnel lenses ............................................ 262
9.9.3 Photovoltaic solar concentrator using total internal
reflection ............................................................................................. 264
9.10 Fresnel Lenses for Rear-Projection Screens ........................................... 264
9.11 Fresnel Lens Manufacture ............................................................................. 265
9.12 Achromatic Fresnel Lenses........................................................................... 265
9.12.1 Combination of high- and low-dispersion materials ............ 267
9.12.2 Achromatic catadioptric Fresnel lenses ................................... 267
9.12.3 Dispersion-compensated achromatic Fresnel lens................ 271
9.12.4 Design example: achromatic dual-grooved Fresnel lens
for overhead projector .................................................................... 273
9.12.5 Achromatic zone plate using a Fresnel lens ........................... 274
9.13 Diffraction and Coherence Effects in Fresnel Lenses .......................... 276
9.13.1 Diffraction compensation in a Fresnel lens reflector ........... 276
9.13.2 Phase-optimized Fresnel lens ...................................................... 277
9.13.3 Phase-optimized Fresnel lens for use in an IR intrusion
detector ................................................................................................ 278
9.14 Design of a Fresnel Lens Illuminator Using Genetic Algorithms ... 278
References ....................................................................................................................... 281

Afterword ........................................................................................................... 285

Index ................................................................................................................... 287

Preface
This text deals primarily with the optics of refracting and reflecting planar surfaces
in the form of prismatic refracting and reflecting components, and the design,
analysis, and applications of these components. Optical prisms consist of multiple
planar surfaces, constructed to a specified geometry and formed from optical
glass or plastic. The surfaces may have thin-film coatings that contribute to their
functionality. Optical prismatic elements can be classified into two general types:
those that are used in imaging systems, such as binoculars or projectors, and
those used in nonimaging systems, such as spectrometers, illuminators, and solar
concentrators. In addition to well-known prism systems, new applications of prisms
are being introduced in the fields of electro-optics, metrology, prismatic films and
arrays, projection displays, and others.
Chapter 1 introduces and reviews the optical concepts that are useful for
the topics developed in the succeeding chapters. In Chapter 2, some better-
known prism types are discussed, along with the essential ray-trace equations that
define their specific properties. This includes both single and compound prisms,
along with cube-corner retroreflectors. Birefringent prisms and polarizing beam-
splitting prisms that produce polarized light are discussed in Chapter 3, including
prisms that affect the polarization state of light, such as polarization-preserving
prisms and prisms that rotate the plane of polarization. Prisms with collinear and
coaxial dispersion properties, achromatic multiprisms, and anamorphic designs
for beam expansion and compression are examined in Chapter 4. In Chapter 5,
several methods of prism design are reviewed, including some of the more recent
methodologies. This chapter also covers prism fabrication, tolerancing, choice
of optical material, and some mounting methods. Specific uses of prisms in
optical systems, such as scanning, beam steering, spectroscopy, interferometry,
light coupling and switching, and viewing and illumination are presented in
Chapter 6. Chapter 7 covers the use of prisms as dichroic color beamsplitters and
combiners, polarizers, and light-beam homogenizers and integrators in projection
displays. Microprism arrays are very useful for light guides, luminaires, brightness-
enhancement sheets, backlight displays, and sheet polarizers. These applications
are detailed in Chapter 8. Last, Chapter 9 covers Fresnel lens optics and the use
of both refractive and reflective lenses in illumination, solar concentration, and
direct-view displays. Several design methods for producing achromatic and phase-
corrected Fresnel lenses are also presented.
References and examples are drawn from specialized texts, journal articles,
conference proceedings, trade publications, and patent literature. I wish to
acknowledge the editorial assistance of Gwen Weerts of SPIE Press for her
suggestions and contributions during the composition of this book.
Dennis F. Vanderwerf
Austin, Texas

xi
Chapter 1
Introduction and Background
1.1 Snell’s Law of Refraction
One of the most important laws in the analysis and design of prisms, and optical
systems in general, is Snell’s law of refraction, named for Willebrord Snell. It
relates the angles of incidence and refraction at the boundary of two materials with
differing refractive index (sometimes called the index of refraction). The refractive
index n is defined as the ratio of the velocity of light in a vacuum c to the velocity
of light in the material vmat :
c
n= . (1.1)
vmat

Since the velocity of light is reduced when traveling through optical materials, n
is greater than unity. For the special case of air, which has a refractive index of
approximately 1.0003, we assume the refractive index of air to be unity for most
optical calculations.
Snell’s law can be derived geometrically or from Fermat’s principle, named for
Pierre de Fermat.1,2 It is usually stated in the following form:

n sin I = n0 sin I 0 , (1.2)

where n is the refractive index of the incident medium, and n0 is the refractive
index of the transmitting medium. I is the angle of incidence, measured relative
to the boundary surface normal, and I 0 is the angle of refraction at the boundary
surface of the second medium (see Fig. 1.1). Snell’s law is applicable to plane or
curved surfaces, and both rays lie in a common plane called the plane of incidence.
A related law for reflecting surfaces is the law of reflection. It can also be derived
geometrically or by using Fermat’s principle. It is stated in the following form:

I = I0, (1.3)

where I is the angle of incidence, and I 0 is the angle of reflection, as illustrated in

Fig. 1.2. Since both incident and reflected rays are in the same medium, refractive
index is not a factor in the directional change, and both rays lie in the common
plane of incidence.

1
2 Chapter 1

Figure 1.1 Snell’s law of refraction.

Figure 1.2 Law of reflection.

1.2 Optical Dielectric Materials

When applying Snell’s law, the refractive indices depend on the wavelength λ of
the incident light (other than air). Refractive indices of optical materials for various
wavelengths are usually obtained from data tables from the manufacturer, or in
optical material reference books, and are determined by careful measurement of
the indices for various spectral lines. The refractive index n is usually specified
with a subscript indicating the spectral line used for measuring the index—for
example, nC at λ = 656.3 nm, nd at λ = 587.3 nm, and nF at λ = 486.1 nm.
There are also various formulas known as dispersion equations that approximate
the continuous change in refractive index with wavelength. The dispersion of an
optical material is defined as the change in refractive index with wavelength and
is an important consideration for most prism designs. For visible light, the Abbe
number, or ν-number, gives a measure of the material dispersion and is defined as

nd − 1
ν= . (1.4)
nF − nC

By comparing the ν-number between various glasses, relative dispersions can be

compared, since a lower ν-number indicates a material of higher dispersion. This is
Introduction and Background 3

important in the choice of glass type for prism design, especially where dispersion
reduction (e.g., achromatic prisms) is a requirement. In Fig. 1.3, a ray of visible
light is refracted and dispersed at a planar air–glass surface.
There are many optically transparent dielectric optical materials, in the form
of glasses and plastics, suitable for use in optical element fabrication. Although
there are hundreds of types of optical glass available for use by a lens designer,
as listed in the Schott, Ohara, or Hoya glass catalogs, a more limited number is
usually employed for optical prisms. Table 1.1 gives some representative types
used in commercially available prisms. The most important factors to be considered
in choosing a material for prism construction are the availability, cost, intended
spectral range, stability, transmission quality, and ability to be accurately machined
and polished, or in some cases, molded.

Figure 1.3 Dispersion of refracted light for crown glass.

Table 1.1 Typical prism materials.

Description Code or nC nd nF ν-number
brand name λ = 656.3 nm λ = 587.3 nm λ = 486.1 nm
Crown glass BK7 1.5143 1.5168 1.5228 64.2
Optical crown glass B270 1.5202 1.5229 1.5291 58.5
Flint glass F2 1.6150 1.6200 1.6331 36.4
Flint glass F5 1.5988 1.6034 1.6146 38.0
Extradense flint glass SF10 1.7209 1.7283 1.7464 28.5
Acrylic plastic Lucite, 1.4892 1.4918 1.4978 57.5
(PPMA) Plexiglas
Polycarbonate Lexan 1.5799 1.5855 1.5994 30.0
Fused silica UV grade Herasil 1.4565 1.4586 1.4632 67.6
Suprasil
Fused quartz IR grade Infrasil 1.4565 1.4586 1.4632 67.6
Vitreosil
IR
4 Chapter 1

1.3 Fresnel Reflection at a Dielectric Surface

When light is refracted at a dielectric optical surface, a portion of the incident light
is reflected, so all of the light is not transmitted. The amount of reflected light is
derived from the Fresnel amplitude coefficients for reflection, named for Augustin
Jean Fresnel:
n0 cos I − n cos I 0
rp = , (1.5)
n cos I 0 + n0 cos I
or
tan(I − I 0 )
rp = , (1.6)
tan(I + I 0 )
n cos I − n0 cos I 0
rs = , (1.7)
n cos I + n0 cos I 0
or
−sin (I − I 0 )
rs = , (1.8)
sin (I + I 0 )

where r p is the reflection coefficient for p-polarized light (electric field vector
parallel to the plane of incidence), and r s is the reflection coefficient for s-polarized
light (electric field vector perpendicular to the plane of incidence). Figures 1.4(a)
and (b) show the angles of incidence, refraction, and reflection I, I 0 , I 00 ; the electric
field vectors Ek , E⊥ ; and the propagation vectors k, k0 , and k00 , which define the
direction of the incident, refracted, and reflected rays. The orthogonal magnetic
field vectors are shown as dotted, since the magnetic induction for dielectric
materials is negligible.
The Fresnel reflections R p and R s are given by

tan2 (I − I 0 )
R p = r2p = , (1.9)
tan2 (I + I 0 )
sin2 (I − I 0 )
R s = r2s = . (1.10)
sin2 (I + I 0 )

The surface transmissions are

T p = 1 − Rp, (1.11)
T s = 1 − Rs. (1.12)

For unpolarized light, the reflection R is given by the average of R p and R s :

1 tan2 (I − I 0 ) sin2 (I − I 0 )
" #
R= + . (1.13)
2 tan2 (I + I 0 ) sin2 (I + I 0 )
Introduction and Background 5

Figure 1.4 (a) Fresnel reflection vectors for s-polarized light (n0 > n). (b) Fresnel reflection
vectors for p-polarized light (n0 > n).

For normal incidence of light (I = I 0 = 0), the reflection for unpolarized light R is
given simply by

(n0 − n)2
R= . (1.14)
(n0 + n)2

The boundary transmission T for unpolarized light is then given by

T = 1 − R. (1.15)

1.4 External Reflection at an Optical Surface

The reflection parameters of light depend on whether the ray is incident on a
medium of higher or lower refractive index. We will first consider the case where
the medium is of a higher refractive index, or external reflection. Such an example
is an air–glass interface. Figure 1.5 shows the Fresnel reflection coefficients R s and
6 Chapter 1

Figure 1.5 External reflection at surface n0 > n.

R p at an air–glass interface as a function of the angle of incidence I for BK7 optical

glass. The average value for unpolarized light is also shown. At a specific angle of
incidence IBrew , known as Brewster’s angle for David Brewster, the p-polarized
light is completely refracted, with IBrew + I 0 = 90 deg. The reflected light is then
completely s-polarized, and from Snell’s law it follows that

n0
!
IBrew = arctan . (1.16)
n

For an air–BK7 glass interface, with n0 = 1.5229 and n = 1.0, IBrew = 56.71 deg.

1.5 Internal Reflection at an Optical Surface

In the case where the light ray is incident on a medium of lower refractive index,
we have internal reflection. Figure 1.6 illustrates the reflection coefficients for a
BK7 glass–air interface as the angle of incidence varies. It is apparent that there
is also an incidence angle for which R p = 0. Using Eq. (1.16), IBrew = 33.29 deg,
which is the complement of the Brewster angle for external reflection. There is also
an angle of incidence for which both R p and R s approach unity, such that the light
is completely reflected for all angles of incidence greater than this critical angle.
This is called total internal reflection (TIR) and can provide 100% reflectance at
Introduction and Background 7

Figure 1.6 Internal reflection at surface n > n0 .

all wavelengths of interest. No applied reflective coating can exactly achieve this
reflectance value.
The critical angle Icrit for total internal reflection is calculated from

n0
!
Icrit = arcsin . (1.17)
n

For a BK7–air interface, with n0 = 1.0 and n = 1.5229, Icrit = 41.04 deg. Both
Brewster reflection and total internal reflection are important considerations in
many prism designs.

1.6 Reflection Phase Shifts at a Planar Interface

When light is reflected from a planar surface, the component of the electric vector
E parallel to the plane of incidence Ek and the component perpendicular to the
plane of incidence E⊥ undergo a phase shift ∆ϕ. The phase shift for externally
reflected light varies as a function of the angle of incidence I for an interface
where n1 (incident medium) > n2 (transmitting medium). Assume that n1 = 1.0
and n2 = 1.51. The E⊥ phase shift ∆ϕ⊥ remains constant at 180 deg for all I values
between 0 and 90 deg. The E phase shift ∆ϕk = 0 deg, for 0 deg ≤ I < IBrew , then
changes abruptly to 180 deg for IBrew ≤ I ≤ 90 deg.
For internally reflected light (n1 > n2 ), ∆ϕ⊥ = 0 deg for 0 deg ≤ I ≤ Icrit , and
∆ϕ⊥ varies continuously from 0 to 180 deg for Icrit < I ≤ 90 deg. This continuous
8 Chapter 1

variation can be calculated3,4 from the following:

 p 
 − sin2 I − n∗2 
∆ϕ⊥ = 2 arctan   , (1.18)
cos I 

where n∗ ≡ n2 /n1 , and n∗ < 1.0, ∆ϕk = 180 deg for 0 deg ≤ I < IBrew , ∆ϕk = 0 deg
for IBrew ≤ I ≤ Icrit , and ∆ϕk varies continuously for Icrit ≤ I ≤ 90 deg (TIR
region). This continuous variation can be calculated from the following:
 p 
 − sin2 I − n∗2 
∆ϕk = 2 arctan   . (1.19)
n∗2 cos I 

The relative phase shift is defined as δ = ∆ϕ⊥ − ∆ϕk and is calculated from
 p 
 cos I sin2 I − n∗2 
δ = 2 arctan   . (1.20)
sin2 I 

The maximum value δmax is calculated from

(1 − n∗2 )
" #
δmax = 2 arctan , (1.21)
2n∗

δmax occurs at the incident angle I (δmax ), where

s
2n22
I(δmax ) = arctan . (1.22)
n21 − n22

For n1 = 1.51 and n2 = 1.0, Icrit = 41.47 deg, IBrew = 33.51 deg, δmax = 45.94 deg,
and I (δmax ) = 51.34 deg.
Table 1.2 shows some calculated values for selected parameters, where the
smallest I value is slightly larger than Icrit . There are two values of I for which
δ = 45.0 deg, because there will be two values of I for every value of δ.
In Fig. 1.7, the δ variation is plotted as I varies from normal to a grazing
incidence angle.

Table 1.2 Phase changes for total internal reflection, in degrees.

I 41.472 45 48.63 51.34 54.62 75 90
∆Φ⊥ −0.2676 −38.63 −56.21 −67.03 −78.81 −139.6 −180.0
∆Φk −0.6102 −77.26 −101.2 −112.9 −123.81 −161.7 −180.0
δ 0.3426 38.63 45 45.94 45 22.08 0
Introduction and Background 9

Figure 1.7 Phase shift for internal reflection.

1.7 Antireflection and Reflection Coatings

By the use of optical thin-film coatings, the reflectance of optical surfaces can
be significantly reduced. These antireflection (AR) interference coatings can be
a single-layer film or a multilayer film of various thicknesses and materials. The
reflectivity is wavelength dependent. By judicious choice of coating materials, it
is possible to reduce the reflectance to zero for a single wavelength using two
layers. By using more layers and different coating materials, the reflectance can be
lowered over an extended wavelength range.5 The reflectance of a common glass
surface can be reduced from about 4% to less than 0.5% over the visible spectrum
by a cost-effective multilayer coating.6
High-reflectance surfaces can be obtained from metallic coatings, such as
aluminum or silver. These metallic coatings can be further overcoated with
multilayer thin films to enhance the reflectance or make them more durable.
For example, over the visible spectrum, the reflectance of an aluminum coating
can be increased from about 92% to about 98% using four alternating layers
of silicon dioxide and titanium dioxide.7 It is also possible to produce a high-
reflectance (>99%) multilayer coating using alternating high- and low-index
dielectric coatings, with select wavelength band reflectance. Multilayer coatings
can be designed having other unique optical characteristics.

1.8 Effective f /# of a Converging Light Beam

It is often convenient to describe a converging beam of light in terms of an
effective f -number, designated by f /#. Consider a convergent beam of light with
ray angle θ0 , produced by collimated light incident on an aplanatic lens, as shown
10 Chapter 1

in Fig. 1.8(a). Then, the effective f /# of the lens is defined by

n
f /# = , (1.23)
2n0 sin θ0

where n is the refractive index on the object side of the lens, and n0 is the refractive
index on the image side.8 For an aplanatic lens in air, n = n0 = 1.0, n0 = n0 0 = 1.0,
and the minimum allowable f /# = f /0.5. Without considering the actual lens, we
can describe a convergent light beam with half-angle θ0 as having this effective
f /#. For example, if θ0 = 15 deg, the beam f /# = f /1.9; if θ0 = 30 deg, the beam
f /# = f /1.0; and if θ0 = 60 deg, the beam f /# = f /0.58.
Consider a convergent f /1.0 beam (θ = 30 deg) that is incident on the front
surface of a glass cube (n0 = n1 0 = 1.517) in air and is focused on the back surface
of the cube by movement of the cube away from the lens [see Fig. 1.8(b)]. A
resultant focal shift ∆S0 occurs. The internal ray angle θ0 in the cube is reduced
to 19.2 deg, yielding an effective beam f /# of f /0.12 at the focus, using a value
of n0 = 1.517. However, if the cube is moved toward the lens, such that the focus
falls outside the cube in air, then the effective beam f /# at the focus is returned to
f /1.0, with a resultant focal shift ∆S0 as shown in Fig. 1.8(c).

1.9 Refraction and Translation of Skew Rays at Planar Surfaces

Up to now, we have been considering rays that lie only in the meridional
(tangential) plane. Although this type of ray trace is useful, it is often necessary to
trace rays that are incident on an optical surface in an arbitrary plane of incidence.
These are called skew or oblique rays. Figure 1.9(a) shows the refraction of a
general skew ray. The law of refraction can be written in vector form as the
following cross-product:

n(K × k) = n0 (K0 × k). (1.24)

Here, n is the refractive index of the incident medium, n0 is the refractive index of
the refracting medium, K is the incident ray vector, K0 is the refracted ray vector,
and k is the vector normal to the refracting boundary.
The following derived equations can be used in a Cartesian coordinate system to
calculate the angles of refraction and intersection coordinates at the next occurring
surface:
n  n 
Ki 0 = Ki + k i cos I 0
− cos I , (1.25)
n0 n0

where
X
cos I = Ki ki , where (i = x, y, z), (1.26)
Introduction and Background 11

and
X
cos I 0 = K i 0 ki , where (i = x, y, z), (1.27)

where I is the angle of incidence, I 0 is the angle of refraction, ki is the direction

cosine of the normal for the planar boundary in the forward direction, Ki is the
direction cosine of the incident ray, and Ki 0 is the direction cosine of the refracted
ray. The summation is over the x, y, and z coordinates. The K x 0 , Ky 0 , and Kz 0
values then define the direction of the ray incident on the next surface. I 0 is usually
calculated directly from I using Snell’s law [see Fig. 1.9(b)].

Figure 1.8 (a) f /# of lens in air, (b) f /# in glass cube, and (c) f /# through glass cube.
12 Chapter 1

Figure 1.9 (a) Refraction of a skew ray, vector representation. (b) Refraction of a skew ray,
Cartesian coordinates.

The translation of the ray to the next planar surface is calculated from the
intersection of the ray with the next planar surface. Since prism design deals mainly
with planar surfaces, we do not have the added complication of calculating the ray
intersection at a curved surface, as for a lens. The normal form of the planar surface
is given by

xk x + yky + zkz = P, (1.28)

where P is the perpendicular distance from the plane to the origin.

The line representing the ray is described by the following set of equations:

x − xn−1 y − yn−1 z − zn−1

= = = d, (1.29)
Kx Ky Kz

where (xn−1 , yn−1 , zn−1 ) are the intersection coordinates from the previous surface,
and d is the ray distance between the two surfaces.
Introduction and Background 13

Figure 1.10 Translation of light ray between planar surfaces.

The intersection coordinates (x, y, z) are then calculated from the simultaneous
solution of Eq. (1.28) and Eq. (1.29), and these become the initial values for the
translation to the next surface (see Fig. 1.10).

1.10 Convergent Beam through a Tilted Plate

We now use Eq. (1.25) through Eq. (1.29) to trace an f /1.9 convergent beam of
light through a tilted glass plate (or slab or cube) of thickness I, with a nominal
refractive index n0 = 1.5168. The traces can be performed in the tangential plane,
an oblique plane, or the sagittal plane. In Fig. 1.11, a convergent f /1.9 beam
of light (maximum half-angle θ = 15 deg) originates from a circular reference
surface having a radius R0 . This could be the exit pupil of a positive lens that
focuses the beam to a point at a distance z f from the reference surface. Any ray
ρ originating from this reference plane from a point (x0 , y0 , z0 ) is specified by the
angle ϕ measured from the positive x axis and the radial distance r from the origin,
where ρ = r/ sin θ, x0 = r cos ϕ, y0 = r sin ϕ, and z0 = 0. The direction cosines of
these rays are then calculated from the following equations:

x0
K x1 = = −cos ϕ sin θ (1.30)
ρ
−y0
Ky1 = = −sin ϕ sin θ (1.31)
ρ
zf
Kz1 = . (1.32)
ρ
14 Chapter 1

Figure 1.11 Glass plate in a convergent beam.

If a glass plate of thickness T is inserted in the beam normal to the optical axis,
there is a longitudinal displacement of the focus from the original focus.
Now, the glass plate is positioned a distance d01 from the reference surface and
tilted around the y axis by angle ω, as in Fig. 1.12. The direction cosines of the
plate entrance surface 1 and the plate exit surface 2 are then

k x1 = k x2 = −sin ω, (1.33)
ky1 = ky2 = 0, (1.34)
kz1 = kz2 = cos ω. (1.35)

Figure 1.12 Astigmatism from a tilted plate in a convergent beam—tangential plane.

Introduction and Background 15

The angle of incidence I1 and angle of refraction I1 0 at plate surface 1 are then
calculated from Eq. (1.26) and Snell’s law:

I1 = arccos(K x1 k x1 + Ky1 ky1 + Kz1 kz1 ), (1.36)

 n  
I1 0 = arcsin 0 sin I1 . (1.37)
n

The direction cosines of the refracted ray at plate surface 1 are then calculated from
Eq. (1.25):
n
K x1 0 = 0
(K x1 − k x1 cos I1 ) + k x1 cos I1 0 , (1.38)
n
n
Ky1 0 = 0 (Ky1 − ky1 cos I1 ) + ky1 cos I1 0 , (1.39)
 nn 
Kz1 0 = 0 (Kz1 − kz1 cos I1 ) + kz1 cos I1 0 . (1.40)
n

These values are then used for the ray incident on the exit plate surface 2. The
intersection coordinates (x1 , y1 , z1 ) at the plate surface 1 are calculated using
Eq. (1.28) and Eq. (1.29), where P01 = d01 cos ω. Then,
! !
Ky1 Kz1
k x1 + ky1 + kz1
K x1 K x1
x1 = " ! # " ! #, (1.41)
Ky1 Kz1
P01 − ky1 y0 − x0 − kz1 z0 − x0
K x1 K x1
!
Ky1
y1 = y0 + (x1 − x0 ), (1.42)
K x1
!
Kz1
z1 = z0 + (x1 − x0 ). (1.43)
K x1

For surface 2 of the plate, P02 = P01 +T . Then, the direction cosines of the refracted
ray (K x2 0 , Ky2 0 , Kz2 0 ) and the intersection coordinates (x2 , y2 , z2 ) at surface 2 are
calculated by the reapplication of Eq. (1.38) through Eq. (1.43). The intersection
coordinates (x3 , y3 ) of the refracted ray at the original focal plane (z3 = z f ) can be
calculated from

K x2 0
!
x3 = x2 + (z f − z2 ), (1.44)
Kz2 0
Ky2 0
!
y3 = y2 + (z f − z2 ). (1.45)
Kz2 0

It is useful to calculate the focal-point coordinates (x3 , y3 , z3 ) in both the

tangential and sagittal planes for various ray angles in the convergent beam. For
the nonsymmetric tangential plane (ϕ = 0 deg and 180 deg, and y3 = 0), the z3
16 Chapter 1

coordinate is calculated from the intersection of the upper ray U with the lower ray
L. At this intersection, x3L = x3U and x3U = x3L . Then,

K x2 0U U K x2 0L L
! !
x2L
− +
x2U z − z
Kz2 0U 2 Kz2 0L 2
z3 = , (1.46)
K x2 0U K x2 0L
−
Kz2 0U Kz2 0L
K x2 0U
!
x3 = x2 +
U
(z3 − zU
2 ). (1.47)
Kz2 0U

For the symmetric sagittal plane (ϕ = 90 deg and 270 deg, and x3 = 0), the z3
focus coordinate is calculated at the position where the ray intersects the z axis, or
where y3 = 0. It follows that

Kz2 0
!
z3 = z2 − y2 . (1.48)
Ky2 0

We now apply the previous equations to trace through a 45-deg tilted BK7 glass
plate (or slab or cube) with a thickness of 5 mm. Plates of this tilt angle appear
in many optical applications. The beam half-angle of the boundary ray is 15 deg,
and the beam radius at the reference plane is 25 mm. The calculated focal point
of the undeviated beam is z f = 93.30 mm. With the plate inserted, the focal
point in the tangential plane for the boundary rays moves outward and downward
with coordinates x3 = −1.853, y3 = 0, and z3 = 96.40. In the sagittal plane, the
focal point moves outward along the z axis with coordinates x3 = 0, y3 = 0, and
z3 = 95.020. The paraxial foci are calculated from a very small cone (θ ≈ 0.5 deg)
centered on the optical axis, where zt3 (paraxial) in the tangential plane = 96.322,
and z3s (paraxial) in the sagittal plane = 94.974.
The resultant astigmatism is then calculated from the difference of these paraxial
foci:

Astigmatism = z3s (paraxial) − zt3 (paraxial) = −1.348. (1.49)

This value is significant and needs to be considered, especially for imaging systems
in which a tilted plate or slab is an optical component. The longitudinal spherical
aberration (LSA) is calculated from the following equation:

LSA = z3 (θ) − z3 (paraxial). (1.50)

Since the tilted plate is nonrotationally symmetric around the optical axis, the LSA
will vary with both the ray angle θ and the planar angle ϕ of the ray fan. Figure 1.13
plots the LSA as a function of θ in the tangential and sagittal planes. Color and
coma are also introduced by a tilted plate in a convergent light beam. There are
Introduction and Background 17

Figure 1.13 LSA as a function of θ in tangential and sagittal planes.

concise formulas available for the calculation of astigmatism, chromatic aberration,

spherical aberration, and coma for the tilted plate.9

1.11 Reflection and Translation of Skew Rays at Planar Surfaces

For a reflective surface, the law of reflection can be described by the following
vector equation:

K0 = K − 2k cos I, (1.51)

where K is the incident ray vector, K0 is the reflected ray vector, k is the upward
vector normal at the reflecting surface, and I is the angle of incidence [see
Fig. 1.14(a)].
For a Cartesian coordinate system, the following equations result:

Ki 0 = Ki − 2ρki , (1.52)

where
X
ρ= Ki ki = cos I, where (i = x, y, z). (1.53)

Here, Ki is the direction cosine of the incident ray, ki is the direction cosine of the
reflecting surface normal pointing into the mirror, and I is the angle of incidence.
Equation (1.53) is interpreted as follows: the arc cosine of ρ yields the angle of
incidence I [see Fig. 1.14(b)].
18 Chapter 1

Figure 1.14 (a) Reflection of a skew ray, vector representation. (b) Reflection of a skew
ray, Cartesian coordinates.

1.12 Reflection Matrix

Equations (1.51) and (1.52) can be combined and recast in matrix form to assist in
multiple sequential reflector computations. For a review of matrix and determinant
mathematics, see, for example, Kreyszig,10 or for optical applications, Kloos.11
The following matrix equation results:
 0
K x  K x 
 
 K 0  = R  K  , (1.54)
 y0   y 
Kz Kz

where R is the reflection matrix:

1 − 2k2x −2k x ky −2k x kz 

 
R =  −2k x ky 1 − 2ky2 −2ky kz  . (1.55)
 
−2k x kz −2ky kz 1 − 2kz2
 

This defines the direction of the reflected ray. For a series of multiple reflecting
surfaces, the reflection matrix can be multiplied to calculate the direction of the
final ray:

R = (Rn ) (Rn−1 ) (. . .) (R1 ), (1.56)

where the matrix multiplication is performed in the opposite direction in which

the light strikes the mirrors. In general, matrix multiplication is noncommutative,
unless both matrices are diagonal and of the same order.

1.13 Orientation of Viewed Images through Prisms

The terminology to describe the image orientation of the object viewed through
a mirror system or a reflecting prism can be confusing. We follow closely the
definitions of Malacara and Malacara12 for inversion (geometric reflection about
Introduction and Background 19

a horizontal axis) and reversion (geometric reflection about a vertical axis) to

describe the image orientation of an object as viewed through the mirror system or
prism, not being concerned for the present with how the orientation was produced.
If the viewed image can be read normally, then it is called a right-handed image or
readable image. Readable images can be rotated over any rotation angle and remain
readable. If the viewed image is unreadable, then the image has been transformed
from a right-handed to a left-handed coordinate system (sometimes called a change
in parity). An unreadable image will remain unreadable at any rotation angle. If the
unreadable image is erect, it is said to be reverted. A reverted image is produced
by the reflection from a vertical wall-mounted mirror. If the unreadable image is
upside down, the image is said to be inverted. An inverted image is produced, for
example, by the reflection from a horizontal body of water. By these definitions,
both inverted and reverted images are unreadable. The sequential operations of
inversion and reversion will produce a readable image that is rotated 180 deg.
By these definitions, a prism that produces a readable upside-down image (i.e.,
rotates the image by 180 deg) is not an inverting prism. (See also the use of the
terms inversion and reversion, as used by Levi.13 ) However, prisms that rotate an
image 180 deg, regardless of readability, are often called inverting prisms in the
literature. Figures 1.15(a) to (d) illustrate several of these cases, where the dotted
box represents a mirror system or reflecting prism of unspecified design. The exit
image represents how the object appears when viewed, and is not a “projected”
image. When the determinant of R, |R| = −1, an image viewed through the
prism will be unreadable. It does not, however, tell us anything about rotation or
displacement of the viewed image.
To determine the actual orientation of the viewed image, we consider how
the object coordinate system (x0 , y0 , z0 ) is transformed or rotated into a viewed
coordinate system (x0 , y0 , z0 ). This can be determined by use of the following matrix
multiplication, where RT is the transpose of the reflection matrix R.14
Then,
 0
 x  cos α x cos αy cos αz   x0 
  
y0  = RT cos β cos β cos β  y  , (1.57)
 0   x y z  0
 
z cos γ x cos γy cos γz z0

where RT represents a rotation of the viewed coordinate system relative to

the object coordinate system, (α x , β x , γ x ) are the direction angles of the rotated
coordinate system relative to the x0 axis, (αy , βy , γ x ) are the direction angles relative
to the y0 axis, and (αz , βz , γz ) are the direction angles relative to the z0 axis.

1.14 Intersection Coordinate Matrix

The intersection coordinates (xi , yi , zi ) at the next reflecting surface can be
calculated by the same procedure as for plane refracting surfaces, as in Eq. (1.28)
and Eq. (1.29), ensuring that the next reflective surface is in a position to intercept
the incident ray. Again, there are advantages to calculating the intersection
20 Chapter 1

(a) (b)

(c) (d)

Figure 1.15 (a) Viewed image is undeviated, erect, and unreadable. (b) Viewed image is
displaced, rotated 180 deg, and readable. (c) Viewed image is displaced, deviated 90 deg,
erect, and readable. (d) Viewed image is deviated by angles α and β, erect, and readable.

coordinates using matrices. This calculation can also be defined in matrix form
as follows:13,14

 1   1 

   
 x 
 n  = C  xn−1  ,
 
(1.58)
yn  yn−1 
 
zn zn−1

where C is the intersection-coordinate matrix:

1 0 0 0
 
 
PK /ρ 1 − k K /ρ −k K /ρ −k K /ρ 
C =  x(i−1) x x(i−1) y x(i−1) z x(i−1)  , (1.59)
 PKy(i−1) /ρ −k x Ky(i−1) /ρ 1 − ky Ky(i−1) /ρ −kz Ky(i−1) /ρ 
PKz(i−1) /ρ −k x Kz(i−1) /ρ −ky Kz(i−1) /ρ 1 − kz Kz(i−1) /ρ

where i = 1, 2, 3, . . . , P is the perpendicular distance from surface i to the origin,

and ρ is defined as in Eq. (1.53). Computation of the coordinate matrix C for a
surface requires knowledge of the preceding surface coordinates and the direction
cosines of the incident ray. For a series of multiple reflecting surfaces, the mirror
Introduction and Background 21

coordinate matrices can be multiplied to calculate the coordinates of the final ray
at a defined surface:

C = (Cn ) (Cn−1 ) (. . .) (C1 ). (1.60)

1.15 Three-Mirror Beam-Displacing Prism

In general, an odd number of reflecting surfaces produces an unreadable image,
and the image direction and rotation are determined by the mirrored surface
orientations. An even number of mirrored surfaces will produce a readable image.
Consider the three-mirror reflecting prism in Figs. 1.16(a) and (b). Reflection from
mirror surfaces M1 and M2 occurs by total internal reflection for most optical
glasses. Reflecting surface M3 must be coated with a reflective material. This is a
limited-aperture prism, with the square entrance and exit apertures Ap × Ap defined
as shown. θ = 60 deg, ϕ = 120 deg, ψ = 150 deg, input face length = 2Ap,
M1 length = 2Ap, M2 length = 3Ap, and M3 length = 2Ap/ sin α. If we consider
entrance rays normal to the planar entrance aperture, then the exit ray will emerge
normal to the exit aperture, and since there is no ray deviation due to refraction,
we treat this as a pure three-mirror system. Place the (x, y, z) coordinate system
origin along a central ray entering the center of the aperture. The distance of the
origin from the entrance facet is somewhat arbitrary, but we place it at the position
where the extension of the M3 facet crosses the z axis. The direction cosines of the
incident ray are

K x01 = 0 Ky01 = 0 Kz01 = +1, (1.61)

and for the reflecting surfaces, the direction cosines of each mirror normal are

k x1 = cos (30 deg) ky1 = 0 kz1 = cos (60 deg), (1.62a)

k x2 = 0 ky2 = 0 kz2 = 1, (1.62b)
k x3 = −cos (60 deg) ky3 = 0 kz3 = −cos (30 deg), (1.62c)

From Eq. (1.55), the reflection matrices for each reflecting surface are

 −0.50 0 −0.866

 
R1 =  0 1 0  ,
 (1.63a)
−0.866 0 0.50
1 0 0 
 
R2 = 0 1 0  , (1.63b)
0 0 −1
 0.750 −0.50 −0.433
 
R3 =  −0.50 0 −0.866 ,
 (1.63c)
−0.433 −0.866 0.250
22 Chapter 1

Figure 1.16 (a) A three-mirror reflecting prism; viewed image is displaced and inverted.
(b) A three-mirror reflecting prism, perspective view.
Introduction and Background 23

and from Eq. (1.56),

−1 0 0
 
R = (R3 )(R2 )(R1 ) =  0 1 0 . (1.64)
0 0 1

The determinant |R| = −1. Therefore, the image is unreadable.

Then,

 0
K x  K x0  0 0
     
 K 0  = R  K  = R 0 = 0 . (1.65)
 y0   y0     
Kz Kz0 1 1

This simply states that the exiting ray, the ray reflected from M3 , is parallel to the
entrance ray. However, it is not collinear, being displaced by a defined distance.
The transposed reflection matrix RT is

−1 0 0
 
RT =  0 1 0 , (1.66)
0 0 1

which in this case is identical to R, and the matrices are symmetric. For the x0
axis, α x = 180 deg, αy = 90 deg, and αz = 90 deg; for the y0 axis, β x = 90 deg,
βy = 0 deg, and βz = 90 deg; and for the z0 axis, γ x = 90 deg, γy = 90 deg, and
γz = 0 deg. The direction of the x0 axis is reversed and the directions of the y0 and
z0 axes remain unchanged, and therefore the viewed image is inverted.
The intersection coordinates at each surface can be calculated using Eq. (1.58).
The P values for the reflecting planes are as follows:

P1 = (Ap/2) sin (60 deg), (1.67a)

P2 = 2Ap sin (60 deg), (1.67b)
P3 = 1.5Ap cos (60 deg). (1.67c)

Assume for these calculations that Ap = 1.0. Using Eq. (1.53), ρ1 = 0.50 and
I1 = 60 deg, ρ2 = 0.50 and I2 = 60 deg, and ρ3 = 0.8660 and I3 = 30 deg.
Equation (1.59) then yields the coordinate matrices C1 , C2 , and C3 :

 1 0 0 0
 
 0 0 1 0

C1 =  , (1.68a)
 0 0 1 0
0.866 1.732 0 0
24 Chapter 1

 1 0 0 0 
 
 −3.0 0 1 1.732
C2 =   , (1.68b)
 0 0 1 0 
1.732 0 0 0
 1 0 0 0 
 
−0.750 0.50 1 −0.866
C3 =   . (1.68c)
 0 0 1 0 
−0.433 −0.289 0 0.50

Then,

 1   1 

   
 x 
 3  = C  x0  ,
 
(1.69)
y3  y0 
 
z3 z0

where

 1 0 0 0 
 
 −3.0 1 0 1.732

C = (C3 )(C2 )(C1 ) =  , (1.70)
 0 0 1 0 
0.866 0.577 0 0.50

yielding x3 = −3.0, y3 = 0, and z3 = 0.866. Thus, the exiting central ray through
prism surface 2 is displaced by a distance 3Ap from the incident central ray.

1.16 Refraction Matrix

As we defined a reflection matrix for a mirror, we can define a refraction matrix
for refraction at a boundary between materials having different refractive indices.
Using Eqs. (1.38), (1.39), and (1.40), the direction cosines of the refracted ray are
n  n 
Kx = 0 Kx + kx ρ − 0 ρ ,
0 0
(1.71a)
 nn    nn  
Ky = 0 Ky + ky ρ − 0 ρ ,
0 0
(1.71b)
nn    nn  
Kz = 0 Kz + kz ρ − 0 ρ ,
0 0
(1.71c)
n n

and
X
ρ = cos I = Ki ki , where i = (x, y, z), (1.72)
r  n  2
ρ0 = cos I 0 = 1− sin I . (1.73)
n0
Introduction and Background 25

Equations (1.71a), (1.71b), and (1.71c) can be recast in matrix form as

 
 1   1 
 
K 0  K 
x
 K 0  = R  K  ,
x
(1.74)
 
 y   y 
Kz 0 Kz

where the refraction matrix R is defined as

1 0 0 0 
 

k [ρ0 − (n/n0 )ρ] n/n0 0 0 

R =   x
. (1.75)
ky [ρ0 − (n/n0 )ρ] 0 n/n0 0 

kz [ρ0 − (n/n0 )ρ] 0 0 n/n0

For a general prism in which both refraction and reflection occur, we cannot
directly multiply refraction matrices R by reflection matrices R of the form in
Eq. (1.55), since they are of a different order. To obtain a system matrix S by matrix
multiplication, we introduce a modified fourth-order matrix R0 for the reflection
matrix, such that

 1 0 0 0
 
−2k ρ 1 0 0

R0 =  x
, (1.76)
−2ky ρ 0 1 0
−2kz ρ 0 0 1

where ρ is defined as in Eq. (1.72). We call R0 the fourth-order reflection matrix,

and it yields the direction cosines of the reflected ray as R. Then,
 
 1   1 
 
K 0 
x 0 K x 
 
 K 0  = R  K  . (1.77)
 
 y   y 
Kz 0 Kz

The use of the third-order matrix R is preferable for pure mirror systems,
because it requires only the direction cosines of the reflective surface normals. Both
R and R0 will yield the same direction cosines of the reflected ray. The coordinate
matrix C for refractive surfaces is obtained from Eq. (1.59), where ρ is defined as
in Eq. (1.72).

1.17 Four-Mirror Beam-Displacing Prism

To produce a readable image, we can replace the single inclined M3 reflector of
the three-mirror reflecting prism example with two reflective contiguous mirrors
M3 and M4 , having a 90-deg included angle [Fig. 1.17(a)]. This is often called a
roof mirror, which by itself produces an erect and readable image. In this prism,
26 Chapter 1

the roof vertex line is inclined at the same angle β = 30 deg as the original single
mirror. The reflections from M3 and M4 now occur by total internal reflection.
Then,
k x3 = −cos (45 deg) sin (30 deg)
ky3 = cos (45 deg)
kz3 = −cos (45 deg) cos (30 deg),

and
k x4 = −cos (45 deg) sin (30 deg)
ky4 = −cos (45 deg)
kz4 = −cos (45 deg) cos (30 deg),

where M3 and M4 could be reversed in computational sequence with no resultant

difference.
Then, the corresponding reflection matrices are

 0.750 −0.50 −0.433

 
R3 =  −0.50 −0.866 ,

0 (1.78a)

−0.433 −0.866 0.250
 0.750 0.50 −0.433
 
R4 =  0.50 0.866  .

0 (1.78b)

−0.433 0.866 0.250

R1 and R2 remain as defined in Eq. (1.63a) and Eq. (1.63b).

Then,
−1 0 0
 
R = (R4 )(R3 )(R2 )(R1 ) =  0 −1 0 .
 (1.79)
0 0 1

The determinant |R| = +1, indicating that the image is readable.

Then,
 0
K x  K x0 
 
 0 0
   
 K 0  = R  K  = R 0 = 0 .

(1.80)
 y   y0     
Kz 0 Kz0 1 1

The exiting ray remains parallel to the entrance ray.

The transposed reflection matrix RT is

−1 0 0
 
RT =  0 −1 0 . (1.81)
0 0 1
Introduction and Background 27

Thus, both the x0 axis and the y0 axis are reversed, and the z0 axis is unchanged.
The viewed image is therefore readable and rotated 180 deg. This prism, as in
Figs. 1.17(a) and (b), has a circular area of nonessential glass removed and is called
a Leman prism (sometimes called a Sprenger-Leman prism or Leman-Sprenger
prism). Reflection at M3 and M4 occurs by TIR, each having an angle of incidence
of about 52 deg, eliminating the need for any reflective coatings. The Leman prism
is used in several optical system applications, such as a monocular spotting scope.

Figure 1.17 (a) A four-mirror reflecting prism with roof prism face; viewed image is rotated
180 deg and readable. (b) A four-mirror reflecting prism with roof prism face, perspective
view.
28 Chapter 1

1.18 90-deg Beam-Deviating Prism

The four-surface single prism in Fig. 1.18(a) is useful for producing an erect and
readable viewed image with a deviation angle δ = 90 deg. The square input
aperture of side Ap is inclined at 45 deg to the z axis. Figure 1.18(b) shows the
design parameters for BK7 glass (nd = 1.5168). There are two refractions and two
reflections. Total internal reflection occurs at surface 2 (BC), while surface 3 (CD)
must be coated with a reflective material. Surface AD is nonworking. The object
plane 0 and the viewing plane 5 are in the positions shown, with the coordinate
system origin on the central ray at the object plane.
The sequential ray-tracing equations are

I1 = 45 deg, (1.82a)
!
sin I1
I1 0 = arcsin , (1.82b)
n
I2 = I1 − I1 0 + 45 deg = I2 0 , (1.82c)
I3 = I2 − 45 deg = I3 ,
0 0
(1.82d)
I4 = 45 deg − I3 0 = I1 0 , (1.82e)
I4 = arcsin(n sin I4 ) = I1 ,
0
(1.82f)
δ = 45 deg + I4 = 90 deg.
0
(1.82g)

The prism dimensions are determined by the input aperture Ap and the intersection
of the lower refracted ray AC at the far corner of the prism. The prism dimensions
are then calculated from

Ap
AB = , (1.83a)
sin (45 deg)
BC = AB tan (θ − I1 0 ), (1.83b)
AB
CD = , and (1.83c)
sin ψ
sin (ψ − I1 0 )
" #
AD = CD . (1.83d)
sin I1 0

Let Ap = 25 mm. Then, AB = 35.4 mm, BC = 67.1 mm, CD = 50.1 mm, and
AD = 31.8 mm.
The system matrix S is given by

S = (R4 ) (R3 0 ) (R2 0 ) (R1 ), (1.84)

Introduction and Background 29

(a)

(b)

Figure 1.18 (a) A 90-deg beam-deviating prism, perspective view. (b) A 90-deg beam-
deviating prism, design parameters.
30 Chapter 1

and

 1   1 

   
K 
 x45  = S K x01  ,
 
 Ky45   Ky01  (1.85)
   
Kz45 Kz01

where

 1 0 0 0
 
 1 1 0 0

S =  . (1.86)
 0 0 1 0
−1 0 0 1

Then, K x45 = 1, Ky45 = 0, Kz45 = 0, and the ray is deviated 90 deg in the +x
direction.
The fourth-order determinant |S| = +1, and therefore, the image is readable.
Alternatively, since the two planar refracting surfaces do not change the readability
of a viewed image, it could be directly inferred that the even number of mirrors
in this prism will produce a readable image. The ray intersection coordinates at
each surface are calculated using Eq. (1.58) and Eq. (1.59) for each surface, in the
order in which the ray hits the surfaces. At the viewing plane 5, the deviated ray is
displaced 2.4460 units in the z direction.
To calculate the ray intersection coordinates at each surface, we construct the
coordinate matrix Ci at each surface using Eq. (1.59). Then,

 1 0 0 1
 
12.50 0 0 0

C = (C5 )(C4 )(C3 )(C2 )(C1 ) =  . (1.87)
 0 0 1 0
61.15 −1 0 0

Then,

 1   1 

   
 x 
 5  = C  x0  .
 
y5  y0  (1.88)
   
z5 z0

If x0 = +12.5 mm, then z5 = 48.65 mm, and if x0 = −12.5 mm, then z5 =

73.65 mm, such that the exit-beam width equals the input aperture height, and
there is no anamorphic expansion or compression of the exit beam. Table 1.3 gives
the relevant ray-trace data.
A useful construction for many prism types is the tunnel diagram, shown for
this 90-deg deviating prism (Fig. 1.19). Here, the prism is unfolded about each
reflecting surface, which shows the direct path of a ray. The tunnel diagram shows
Introduction and Background 31

Table 1.3 90-deg deviating prism ray-trace data.

Surface Intersection ρ, I, I0 Exit-ray Surface Perpendicular
number coordinates direction normal surface
(mm) cosines direction distance to
cosines origin (mm)
0 Origin surface x0 = 0 ρ0 = 0 K x01 = 0 k x0 = 0 P0 = 0
y0 = 0 I0 = 90 deg Ky01 = 0 ky0 = 0
z0 = 0 I0 0 = 90 deg Kz01 = 1.0 kz0 = 1.0

1 Refractor x1 = 0 ρ1 = 0.7071 K x12 = −0.2959 k x1 = −0.7071 P1 = 8.84

y1 = 0 I1 = 45 deg Ky12 = 0 ky1 = 0
z1 = 12.50 I1 0 = 27.787 deg Kz12 = 0.9552 kz1 = 0.7071

2 Reflector x2 = −11.22 ρ2 = 0.4662 K x23 = −0.9552 k x2 = 0.7071 P2 = 26.5

y2 = 0 I2 = 62.213 deg Ky23 = 0 ky2 = 0
z2 = 48.72 I2 0 = 62.213 deg Kz23 = 0.2959 kz2 = 0.7071

3 Reflector x3 = −37.50 ρ3 = 0.9552 K x34 = 0.9552 k x3 = −1.0 P3 = 37.5

y3 = 0 I3 = 17.215 deg Ky34 = 0 ky3 = 0
z3 = 56.86 I3 0 = 17.215 deg Kz34 = 0.2959 kz3 = 0

4 Refractor x4 = −23.65 ρ4 = 0.8847 K x45 = 1.0 k x4 = 0.7071 P4 = 26.5

y4 = 0 I4 = 27.868 deg Ky45 = 0 ky4 = 0
z4 = 61.15 I4 0 = 45 deg Kz45 = 0 kz4 = 0.7071

5 Viewing plane x5 = 12.50 ρ5 = 0 — k x5 = 1.0 P5 = 12.5

y5 = 0 I5 = 0 deg ky5 = 0
z5 = 61.15 I5 0 = 0 deg kz5 = 0

Figure 1.19 Tunnel diagram for the 90-deg deviating prism.

32 Chapter 1

that the input and exit surfaces are optically parallel, and there will be no dispersion
when collimated light enters the prism as shown. However, the prism will exhibit
the aberrations of a tilted plate when used in convergent or divergent light beams.
Therefore, unless there are additional elements to correct these aberrations, this
particular prism is recommended for use only in collimated light.

References
1. E. Hecht, “The propagation of light,” Chapter 4 in Optics, 2nd ed., 87–92,
Addison-Wesley, Reading, MA (1987).
2. T. V. Higgins, “All rays lead to geometrical optics,” Laser Focus World 30(4),
89–97 (1994).
3. M. Born and E. Wolf, Principles of Optics, 4th ed., 49–50, Pergamon Press,
London (1970).
4. J. Lekner, Theory of Reflection: of Electromagnetic and Particle Waves,
194–195, Springer, New York (1987).
5. T. V. Higgins, “Reflections on surfaces, coatings, and thin films,” Laser Focus
World 30(9), 61–67 (1994).
6. CERAK, Inc., “Coatings: selecting thin-film materials,” in The Photonics
Design and Applications Handbook, 46th ed., Book 3, 88–98, Laurin
Publishing, Pittsfield, MA (2000).
7. C. K. Carniglia, “Mirrors: coating choices make a difference,” in The
Photonics Design and Applications Handbook, 46th ed., Book 3, 307–310,
Laurin Publishing, Pittsfield, MA (2000).
8. M. R. Hatch and D. E. Stoltzmann, “The f-stops here,” Optical Spectra, 88–91
(June, 1980).
9. W. J. Smith, Modern Optical Engineering, 2nd ed., 99, McGraw-Hill, New
York (1990).
10. E. Kreyszig, Advanced Engineering Mathematics, 5th ed., 289–299, John
Wiley, New York (1983).
11. G. Kloos, “Optical components,” Chapter 2 in Matrix Methods for Optical
Layout, SPIE Press, Bellingham, WA (2007) [doi:10.1117/3.737850].
12. D. Malacara and Z. Malacara, “Prisms,” Chapter 18 in Handbook of Lens
Design, Marcel Dekker, New York (1994).
13. L. Levi, “Plane surfaces, mirrors, and prisms,” Chapter 8 in Applied Optics,
Vol. 1, John Wiley, New York (1980).
14. R. E. Hopkins, “Mirror and prism systems,” Chapter 7 in Applied Optics and
Optical Engineering, Vol. 3, R. Kingslake, Ed., Academic Press, New York
(1968).
Chapter 2
General Prisms and Reflectors
2.1 Equilateral Prism
The equilateral prism is one of the most available and most widely known of the
prism types. It is normally used as a dispersing prism, separating white light into
its component visible colors. The prism has three planar surfaces at equal 60-deg
angles (Fig. 2.1) and is commercially available in optical glass or plastic, with
specified surface and angular accuracies.
A multispectral light ray entering surface 1 at angle of incidence I1 is refracted at
angle I1 0 , incident on surface 2 at angle I2 , and refracted at angle I2 0 . The refraction
angles vary for each wavelength λ in the light ray according to Snell’s law, and the
exiting beam is dispersed into the familiar visible spectrum. The resulting deviation
angle δ(λ) is the change in direction of the exit beam from the entrance beam,
where
!
sin I1
I1 = arcsin
0
, (2.1a)
nλ
I2 = 60 deg − I1 0 , (2.1b)
I2 = arcsin(nλ sin I2 ),
0
(2.1c)
δ(λ) = I1 + I2 − 60 deg.
0
(2.1d)

There is an allowable range of incident angles for which the rays refracted at
surface 1 hit surface 2 and are refracted into a dispersed beam. In particular, if

Figure 2.1 Equilateral dispersing prism.

33
34 Chapter 2

I2 exceeds the critical angle at surface 2, then the ray undergoes total internal
reflection (TIR). For BK7 glass with nd = 1.5168 (λ = 587.3 nm), the critical
value I2crit = arcsin(1/nd ) = 41.25 deg. The resultant minimum allowable value
of I1 = 29.19 deg is calculated using Eqs. (2.1a) to (2.1c). Rays that undergo
TIR at surface 2 will exit through surface 3 as a nondispersed ray. The maximum
allowable value of I1 is 90 deg, or close to grazing. Also, some of the internally
refracted rays might directly hit surface 3 and undergo TIR. These rays are
refracted by surface 2 and are not dispersed. These specific cases are well described
by Southall.1
If we consider only rays that are directly refracted by surface 2, then we find that
the deviation angle δ acquires a minimum value for a particular angle of incidence.
This occurs when the rays pass through the prism symmetrically—that is, I1 0 = I2 ,
or I1 = I2 0 . For this case, all internal rays are parallel to surface 3 of the prism and
therefore pass directly to surface 2, and I1 0 = I2 = 30 deg. For BK7 crown glass at
λ = 587.3 nm, then I1 = 49.32 deg and δmin = 38.65 deg. When the prism produces
minimum deviation, then the Fresnel surface-reflection losses for unpolarized light
are minimized, and the prism transmission is maximized (see Fig. 2.2).
Since the equilateral prism is available in crown glass, extradense flint glass, and
acrylic plastic, the question arises as to how much the dispersion is increased by
using a dispersive flint glass.
We define the angular dispersion α(δ) in the visible spectrum as the difference
in the deviations for the blue F (486.1 nm) and red C (656.3 nm) wavelengths, as
in Eq. (2.2):

α(δ) = δmin (486.1 nm) − δmin (656.3 nm). (2.2)

Table 2.1 shows the minimum deviation angles, the angle of incidence at
minimum deviation, and the angular dispersion at minimum deviation for three
different visible wavelengths using three different glass types. It is seen that the
angular dispersion is increased about fourfold by choosing an extradense flint glass
over a crown glass.

Figure 2.2 Equilateral dispersing prism at minimum deviation.

General Prisms and Reflectors 35

Table 2.1 Minimum deviation angles for different glass types.

Glass type λ (nm) δmin (deg) I1 (deg) α (deg)
BK7 crown glass, ν = 64.2 486.1 38.43 49.21 0.748
587.3 38.65 49.32
656.3 39.18 49.59

Acrylic plastic, ν = 57.5 486.1 36.25 48.12 0.741

587.3 36.47 48.24
656.3 36.99 48.50

SF10 Extradense flint glass, ν = 28.5 486.1 58.73 59.37 2.93

587.3 59.57 59.79
656.3 61.67 60.83

The equilateral dispersing prism can be used as a component in a prism

spectroscope.2 Measurement of the angle of minimum deviation δ(λ)min for specific
wavelengths can be used to accurately calculate the refractive index n(λ) of the
prism as a function of wavelength by use of the following equation:

sin[δ(λ)min + 60 deg]
( )

2
n(λ) = . (2.3)
sin(30 deg)

For this type of spectroscope, the angles of minimum deviation are always
noninteger, and the observer or detector must rotate at a different rate than the
prism rotation.

2.2 Abbe Dispersing Prism

Another type of dispersing prism is the Abbe prism, named for Ernst Karl Abbe.
This is one of several types of “Abbe” prisms named after this pioneer in the field
of optics. This 60/90/30-deg prism is illustrated in Fig. 2.3. A multispectral ray
enters surface AB at angle of incidence I1 , is refracted at angle I1 0 , undergoes TIR
at surface BC at angle of incidence I2 and angle of reflection I2 0 , and is refracted
at surface AC at angle of incidence I3 and angle of refraction I3 0 . The resulting
deviation angle δ(λ) is calculated from the following ray-trace equations:
!
sin I1
I1 = arcsin
0
, (2.4a)
nλ
I2 = 90 deg − I1 0 = I2 0 , (2.4b)
I3 = I2 − 30 deg, (2.4c)
I3 0 = arcsin(nλ sin I3 ), (2.4d)
δ(λ) = 60 deg + I1 − I3 0 . (2.4e)

If the ray passes through both refracting surfaces symmetrically, where I1 = I3 0 ,

then δ(λ)const.dev = 60 deg, and the internal ray is parallel to surface 3, where
36 Chapter 2

Figure 2.3 Abbe dispersing prism with 60-deg constant deviation.

δ(λ)const.dev is equal to the angle α between the refracting surfaces. Then, I1 0 =

30 deg, and the deviation angle can be held at the integral value 60 deg for all
wavelengths by proper rotation of the prism to change the angle of incidence on
surface 1. This type of prism is referred to as a wavelength-dependent 60-deg
constant deviation dispersing prism. This is not the minimum deviation angle.
Table 2.2 shows some relevant values for a prism constructed of BK7 glass.

Table 2.2 Sample parameters for Abbe constant deviation dispersing prism.

Glass type λ (nm) I1 (deg) I10 (deg) δ(λ)const.dev

BK7 crown glass, ν = 64.2 656.3 (nC = 1.5143) 49.21 30.0 60.0
587.3 (nd = 1.5168) 49.32 30.0 60.0
486.1 (nF = 1.5228) 49.59 30.0 60.0

2.3 Pellin–Broca Dispersing Prism

The Pellin–Broca prism, named for Phillippe Pellin and André Broca, is another
wavelength-dependent constant deviation dispersing prism, where δ(λ)const.dev =
90 deg. As in Fig. 2.4, it is a four-sided prism ABCD, where α = 75 deg, and
β = 90 deg. The CD face is inactive, typically with 60 deg ≤ γ ≤ 90 deg. In this
case, γ = 60 deg, yielding 135 deg for the fourth angle.
The basic ray-trace equations to calculate the deviation angle are
!
sin I1
I1 = arcsin
0
, (2.5a)
nλ
I2 = α − I1 0 = I2 0 , (2.5b)
I3 = I2 − 90 deg + α, (2.5c)
I3 0 = arcsin(nλ sin I3 ), (2.5d)
δλ = 90 deg − I1 + I3 0 . (2.5e)
General Prisms and Reflectors 37

Figure 2.4 Pellin–Broca dispersing prism with 90-deg constant deviation.

When I1 = I3 0 , then δλ = 90 deg, I1 0 = 30 deg, and I2 = I2 0 = 45 deg.

If the critical angle at surface BC is less than 45 deg, then TIR occurs, and no
reflective coating is required on this face. By varying the angle of incidence I1 , the
wavelength that produces the constant deviation angle δλ const.dev can be isolated.
Table 2.3 shows the values for three visible wavelengths for a Pellin–Broca prism
constructed of BK7 glass. Using Eq. (2.5), the angular dispersion α(λ) ≈ 0.75 deg
at δ(λ) = 90 deg. The I1 values are the same as for the similar Abbe prism described
previously. Moreover, if the prism is rotated around a point on face BC, two-thirds
the distance from B to C, then the input and output beams can be kept stationary.3
A variation of this standard Pellin–Broca prism is the Brewster’s-angle
Pellin–Broca prism (Fig. 2.5). The incident ray I1Brew , enters surface 1 at Brewster’s
angle for a specific refractive index of the prism. For example, if nd = 1.5168 (BK7
glass, λ = 587.3 nm), then I1Brew = 56.60 deg. For δλ = 90 deg, α = 78.40 deg.
This results in very high transmission of p-polarized light at this wavelength.
Brewster’s-angle Pellin–Broca prisms are available commercially in BK7, UV-
grade fused silica, and crystal quartz for high-power laser pulses. Since we can
achieve an exact 90-deg deviation angle only for a specific Brewster’s angle
at a specific wavelength, δλ = 90 deg for other wavelengths requires that the
incident angle I1 deviate slightly from Brewster’s angle. The angular dispersion
α(λ) ≈ 0.98 deg at δλ = 90 deg. Table 2.4 gives typical parameters for a Brewster’s-
angle Pellin–Broca dispersing prism.

Table 2.3 Sample parameters for Pellin–Broca dispersing prism.

Glass type λ (nm) I1 (deg) I10 (deg) δconst.dev

λ (deg)
BK7 crown glass, ν = 64.2 656.3 (nC = 1.5143) 49.21 30.0 90.0
587.3 (nd = 1.5168) 49.32 30.0 90.0
486.1 (nF = 1.5228) 49.59 30.0 90.0
38 Chapter 2

Figure 2.5 Brewster’s-angle Pellin–Broca prism.

Table 2.4 Sample parameters for a Brewster’s-angle Pellin–Broca dispersing prism.

Glass type λ (nm) I1 (deg) I10 (deg) δconst.dev

λ (deg)
BK7 crown glass, ν = 64.2 656.3 (nC = 1.5143) 56.46 33.40 90.0
587.3 (nd = 1.5168) 56.60 (Brewster’s angle) 33.39 90.0
486.1 (nF = 1.5228) 56.95 33.40 90.0

2.4 Penta Prism

The Penta prism is a solid prism having five sides with one included angle of 90 deg
and three included angles of 112.5 deg each, as shown in Fig. 2.6. One surface is not
used optically. Planar surfaces AB and CD are usually reflection coated to ensure
operation over a wider range of input angles at surface 1. Although the Penta prism
cannot discriminate wavelengths as the Pella–Broca prism can, it has the useful
property of producing a constant 90-deg deviation angle δ, independent of wave-
length and the angle of incidence at surface 1. The basic ray-trace equations are
!
sin I1
I1 = arcsin
0
, (2.6a)
n
I2 = I1 0 + 22.5 deg, (2.6b)
I3 = 45 deg − I2 0 , (2.6c)
I3 0 = I3 , (2.6d)
I4 = 22.5 deg − I3 0 , (2.6e)
I4 0 = arcsin(n sin I4 ), (2.6f)
δ = 90 deg + I1 − I4 0 . (2.6g)

Since I1 0 = I4 , the Penta prism can be used to maintain a 90-deg deviation

angle without the need for precise prism alignment perpendicular to the plane of
incidence. Table 2.5 shows some sample data illustrating this property.
General Prisms and Reflectors 39

Figure 2.6 Penta prism with 90-deg constant deviation.

Table 2.5 Sample parameters for a Penta prism.

Glass type λ (nm) I1 (deg) I10 = I4 (deg) δconst.dev
λ (deg)
BK7 crown glass, ν = 64.2 656.3 (nC = 1.5143) 0 0 90.0
+10 +6.585 90.0
−10 −6.585 90.0
±20 ±13.05 90.0
587.3 (nd = 1.5168) 0 0 90.0
+10 +6.574 90.0
−10 −6.574 90.0
±20 ±13.03 90.0
486.1 (nF = 1.5228) 0 0 90.0
+10 +6.548 90.0
−10 −6.548 90.0
±20 ±12.98 90.0

2.5 Right-Angle Prism

The right-angle prism in Fig. 2.7 consists of three surfaces, two refracting and
one reflecting, with included angles 45/90/45 deg. If the incident light is close to
normal to a side face, then reflection from the hypotenuse face can occur by TIR.
For a wider angular range of incident angles, the hypotenuse should be reflection
coated. The right-angle prism is nondispersive, and the deviation angle δ varies
with angle of incidence ±I1 , being simply defined as

δ = 90 deg − 2I1 . (2.7)

The right-angle prism can therefore be used as a variable achromatic beam deviator
by rotation of the prism. When used in an imaging application, the direct-viewed
40 Chapter 2

Figure 2.7 Right-angle reflecting prism.

image of a distance object is reverted. If the reflecting hypotenuse face is converted

to a 90-deg roof, then the prism is called an Amici roof prism (for Giovanni Amici).
Reflection is then accomplished by TIR, and the direct-viewed image is readable
(see Fig. 2.8).

2.6 Porro Prism

The Porro prism, named for Ignazio Porro, is a special type of right-angle prism
where the incident light enters the hypotenuse face at near-normal incidence.
The light is retroreflected from the right-angled faces by TIR and exits at the
hypotenuse face. A Porro prism system can be constructed from two identical

Figure 2.8 Amici roof prism.

General Prisms and Reflectors 41

air-spaced right-angle prisms, oriented orthogonally, as shown in Fig. 2.9(a). For

an entrance aperture of dimensions A × A, the direct-viewed image is displaced in
both the x and y directions by a distance A, being readable and rotated 180 deg.
Figure 2.9(b) illustrates a double Porro prism with shaped corners and beveled
edges to reduce the weight and increase compactness. Shaping and beveling, along
with tolerancing, is an integral part of prism design and fabrication. This shaped
and beveled form of the Porro prism is used in binoculars to erect the image formed
by the objective, where it is called a Type I Porro.
Another type of double Porro prism system is shown in Fig. 2.10, which is a
modification of the Type I system by Abbe. Here, it is shown as a construction
from three right-angle prisms. It is often called a Porro–Abbe prism, or simply
a Type II Porro. In this prism system, the direct-viewed image is displaced by a
distance A in the y direction only. The Type II Porro can be made slightly more
compact than the Type I Porro.

Figure 2.9 (a) Double Type I Porro prism system, original design. (b) Shaped and beveled
Type I Porro prism system.
42 Chapter 2

Figure 2.10 Double Type II Porro prism system, Abbe modification.

Figure 2.11 Brewster’s-angle Porro prism.4 Adapted with permission from the Optical
Society of America.

A single-element Brewster’s angle Porro prism can be constructed.4 The prism

can be considered to be a lower-half Porro prism (ADE), and an upper-half
Brewster’s angle wedge (ABCD), as in Fig. 2.11. The corner angles are α =
78.4 deg, β = 101.6 deg, and γ = ε = 90 deg for nd = 1.5168 (BK7 glass).
The prism is identical in construction to the Brewster’s-angle Pellin–Broca prism
of Fig. 2.5 but is used differently. Light enters face AB at Brewster’s angle, where
IBrew = 56.60 deg, and is retroreflected at surfaces AE and DE by TIR. It then exits
surface AB at Brewster’s angle. This prism can be used as a 1D retroreflector with
very low loss for p-polarized light, less than 10−5 loss between 0.4 and 1.1 µm.

2.7 Dove Prism

A Dove prism, named for Heinrich Wilhelm Dove, sometimes called a
Harting–Dove prism, is useful as an image rotator (Fig. 2.12). As shown, a readable
General Prisms and Reflectors 43

Figure 2.12 Dove prism design parameters.

erect object is viewed as inverted. Since it is often used in an optical system to erect
an upside-down image, the Dove prism is sometimes called a derotator. When used
in collimated light (the only recommended mode), there is no dispersion, as can be
easily seen from a tunnel diagram. Incident light rays at the top of the input face
are directed to the bottom of the exit face, and vice versa, with TIR occurring at
the base face. When the Dove prism is rotated around the optical axis by an angle
ω, the viewed image is rotated by an angle 2ω.
The design parameters are as follows:
A
ϕ = arcsin , (2.8a)
B
I = 90 deg − ϕ, (2.8b)
!
sin I
I = arcsin
0
, (2.8c)
n
θ = 90 deg − ϕ − I 0 , (2.8d)
!
1 1
L=A + , (2.8e)
tan θ tan ϕ

where A= aperture height, B= length of input face, ϕ= base angle of input face, I=
angle of incidence at input face, I 0 = angle of refraction at input face, n= refractive
index of the prism, θ= angle of internal rays with base face, and L= length of the
prism. Table 2.6 shows sample parameters for A = 1.0, varying B, with refractive
indices nd = 1.5168 (BK7) and nd = 1.4586 (UV fused quartz), with an aspect
ratio γ = L/A.
The Dove prism is not a very compact device, since the γ value is usually
between 4 and 6. Since the upper and lower incident rays have a longer internal
optical path than a central incident ray, the Dove prism can be used as a phase
shifter between adjacent input rays. Thus, the Dove prism can be used as a delay
line between separated narrow beams.
A roof Dove prism substitutes a 90-deg roof for the base face of the conventional
Dove prism. The reflection matrix R for the roof Dove prism is

1 0 0 
 
R = 0 −1 0  . (2.9)
1 0 −1
44 Chapter 2

Table 2.6 Sample design parameters for a Dove prism, where A = 1.0.
nd B Φ (deg) I (deg) Θ (deg) L=γ
1.5164 1.20 56.44 33.56 12.18 5.295
1.30 50.28 39.72 14.80 4.615
1.4142 45.0 45.0 17.21 4.228
1.50 41.81 48.19 18.76 4.063

1.4586 1.20 56.44 33.56 11.29 5.674

1.30 50.28 39.72 13.73 4.922
1.4142 45.0 45.0 16.0 4.487
1.50 41.81 48.19 17.46 4.298

The determinant of R is +1, and the viewed image is readable, but still rotated
180 deg.
A compound version of the single Dove prism is the double Dove prism
(Fig. 2.13). Two identical single dove prisms are positioned base to base. This
functions similarly to a single Dove prism, but the aperture is doubled, or the L/A
ratio is halved. The base of one prism can be reflectorized and the faces cemented
together, or a small air gap can be mechanically held for TIR. Each half of the
prism produces an inverted image, but if the full aperture is used as a direct-vision
prism, the viewed image is not continuous from top to bottom. Rotating double
Dove prisms are often used in scanning systems. Figures 2.14(a) to 2.14(c) show
perspective views of the Dove prism, the roof Dove prism, and the double Dove
prism.

2.8 Brewster Laser-Dispersing Prism

Figure 2.15 illustrates a laser-dispersing prism where the entrance and exit angles
are at a nominal design Brewster’s angle. For a prism material of UV-grade fused
silica (nd = 1.4586), the Brewster’s angle IBrew = 56.57 deg. This type of prism
is often used for p-polarized visible lasers where very low surface reflection loss
is desired. The prism apex angle α is chosen such that the refracted internal ray
is parallel to the prism base at the Brewster’s angle, where α is twice the value of

Figure 2.13 Double Dove prism design parameters.

General Prisms and Reflectors 45

Figure 2.14 (a) Dove prism producing an inverted image. (b) Roof Dove prism producing a
readable image rotated 180 deg. (c) Double Dove prism producing an inverted discontinuous
image.

Figure 2.15 Brewster laser-dispersing prism.

46 Chapter 2

each refracted internal angle I int . Since IBrew = 90 deg + I int , the prism is isosceles
with base angle β = γ ≈ IBrew .

2.9 Littrow Prism

The basic Littrow prism, named for Joseph Johann Littrow, is a dispersing
30/60/90-deg uncoated prism (Fig. 2.16). An incident ray I1 is refracted at the AB
face, undergoes TIR at the BC face, and is refracted outward at the AC face, pro-
ducing a wavelength-dependent deviation δ(λ). The basic ray-trace equations are
!
sin I
I1 = arcsin
0
, (2.10a)
n
I2 = 60 deg − I1 0 = I2 0 , (2.10b)
I3 = I2 0 − 30 deg, (2.10c)
I3 0 = arcsin(n sin I3 ), (2.10d)
δ(λ) = 90 deg − I1 + I3 0 , (2.10e)

where n is the refractive index at wavelength λ. For example, a BK7 prism with
I1 = 20 deg would yield δ(486.1 nm) = 96.471 deg, δ(587.3 nm) = 96.274 deg,
and δ(656.3 nm) = 96.192 deg.
If a high-reflectance coating is applied to the AC face, the prism functions as
a wavelength-independent (nondispersing) prism with a constant deviation angle
δλ const.dev = 60 deg. This prism is often called a 60-deg Bauernfeind prism [see
Figs. 2.17(a) and 2.17(b)]. The basic ray-trace equations are
!
sin I1
I1 = arcsin
0
, (2.11a)
n
I2 = 60 deg − I1 0 = I2 0 , (2.11b)

Figure 2.16 30/60/90-deg Littrow dispersing prism.

General Prisms and Reflectors 47

Figure 2.17 (a) 30/60/90-deg Littrow 60-deg deviation reflecting prism. (b) Perspective
view of 30/60/90-deg Littrow 60-deg deviation reflecting prism.

I3 = I2 0 − 30 deg = I3 0 , (2.11c)
I4 = 30 deg − I3 0 , (2.11d)
I4 0 = arcsin(n sin I4 ), (2.11e)
δλ const.dev = 60 deg + I4 0 − I1 . (2.11f)

Table 2.7 shows sample parameters for a Littrow 30/60/90-deg reflecting prism.
For a direct-view system, the image is erect, readable, and deviated 60 deg.
If the prism of Fig. 2.16 is split through the apex angle, and the face AC
opposite the hypotenuse AB is coated with a very high-reflectance (>99%)
multilayer dielectric coating, we obtain a Littrow laser-dispersion prism, as shown
in Fig. 2.18. If a ray is incident on face AB at the Brewster’s angle, and the vertex
angle α = 90 deg − I1Brew , then the ray is retroreflected. A p-polarized laser beam
will be transmitted with very low loss. For example, using ultraviolet-grade fused
silica (UVFS) glass (nd = 1.4586 at λ = 587.3 nm), I1Brew = 55.567 deg, and
α = 34.433 deg = I1Brew . For a multispectral incident laser beam, the reflected rays
48 Chapter 2

Table 2.7 Sample parameters for a Littrow 30/60/90-deg reflecting prism.

Glass type λ (nm) I1 (deg) I40 = I1 (deg) δconst.dev
λ (deg)
BK7 crown glass, ν = 64.2 656.3 (nC = 1.5143) 0 0 60.0
+20 +20 60.0
−20 −20 60.0
587.3 (nd = 1.5168) 0 0 60.0
+20 +20 60.0
−20 −20 60.0
486.1 (nF = 1.5228) 0 0 60.0
+20 +20 60.0
−20 −20 60.0

Figure 2.18 Brewster–Littrow laser-dispersing prism.

will be dispersed. The basic ray-trace equations are

I1 = I1Brew , (2.12a)
!
sin I1
I1 0 = arcsin , (2.12b)
n
I2 = I1 0 − I1 = I2 0 , (2.12c)
I3 = α − I2 0 , (2.12d)
I3 0 = arcsin(n sin I3 ), (2.12e)
δ(λ) = I1 − I3 0 , (2.12f)
α(δ) = δ(656.3 nm) − δ(486.1 nm). (2.12g)

For this prism, the calculated angular dispersion α(δ) ≈ 0.77 deg. This Littrow
dispersing prism can be used as a component in a tunable laser. By rotation of the
prism, the retroreflected wavelength can be selected, especially for gas lasers that
operate at discrete wavelengths.5
General Prisms and Reflectors 49

2.10 Schmidt Prism

The Schmidt prism, named for Bernhardt Woldemar Schmidt, is a 22.5/67.5/67.5-
deg reflecting roof prism as shown in Figs. 2.19(a) and 2.19(b). A ray incident
normal to the entrance face exits the exit face at a 45-deg deviation angle. The
deviation angle δ and the readability and rotation of a direct-view image can be
predicted from the reflection matrix, assuming that there are no refractions.
For the coordinate system shown, the direction normals for the surfaces are:

k x1 = −cos (67.5 deg), ky1 = 0, kz1 = cos (22.5 deg), (2.13a)

k x2 = cos (67.5 deg), ky2 = 0, kz2 = cos (22.5 deg), (2.13b)
k x3 = −cos (45 deg), ky3 = −cos (45 deg), kz3 = 0, (2.13c)
k x4 = −cos (45 deg), ky4 = cos (45 deg), kz4 = 0, (2.13d)
k x5 = cos (67.5 deg), ky5 = 0, kz5 = −cos (22.5 deg), (2.13e)
k x6 = cos (67.5 deg), ky6 = 0, kz6 = cos (22.5 deg). (2.13f)

Considering no refraction at surface 1 or 2, Eq. (1.55) gives

 0.7071 0 −0.7071

 
R2 =  0 1 0  ,
 (2.14a)
−0.7071 0 −0.7071
 0 −1 0
 
R3 = −1 0 0 , (2.14b)
0 0 1
0 1 0
 
R4 = 1 0 0 , (2.14c)
0 0 1
0.7071 0 0.7071 
 
R5 =  0 1 0  .
 (2.14d)
0.7071 0 −0.7071

The prism reflection matrix R is then

−1 0 0
 
R = (R5 )(R4 )(R3 )(R2 ) =  0 −1 0 . (2.15)
0 0 1

The direction cosines of the incident ray are K x1 = −cos (7.5 deg), Ky1 = 0 deg,
and Kz1 = cos(22.5 deg). Then, the direction cosines of the exit ray K x6 , Ky6 , Kz6
are

K x6  K x1 

   
 Ky6  = R  Ky1  . (2.16)
   
Kz6 Kz1
50 Chapter 2

Then, K x6 = 0.3827 = cos(67.5 deg), Ky6 = 0, Kz6 = 0.9239 = cos(22.5 deg),

and the beam deviation δ = 2 × 22.5 deg = 45 deg. The determinant |R| = +1,
indicating that the image is readable. For this symmetric matrix, the transpose
RT = R, and the rotated coordinates (x0 , y0 , z0 ) of the exit beam relative to (x, y, z)
are
 0
 x   x
 
y0  = RT y . (2.17)
 0   
z z

Then, x0 = −x, y0 = −y, z0 = z, and the direct-view image is rotated 180 deg. The
Schmidt prism finds use in eyepieces and viewing systems, providing a 45-deg
deviated and readable erect image of an upside-down object. Since a light beam is
split at the roof, the 90-deg vertex angle is held to an accuracy of 1 to 5 arcsec, or
down to 0.25 arcsec for very critical applications.
The question arises whether the roof surfaces can operate by TIR or whether a
reflective coating should be applied. Schmidt prisms are offered commercially with
and without this reflective coating. Referring to Fig. 2.19(c), the direction cosines
of the ray reflected from roof surface 3 to roof surface 4 are calculated from

K x34  K x23 

   
 K  = R  K  . (2.18)
 y34  3  y23 
Kz34 Kz23

Since K x23 = −cos (22.5 deg), Ky23 = 0, and Kz23 = −cos (67.5 deg), then
K x34 = 0, Ky34 = cos(22.5 deg), and Kz34 = −cos (67.5 deg). Then, using
Eq. (1.53), the angle of incidence at both surfaces 3 and 4 is 49.21 deg. For a
Schmidt prism of BK7 glass, the critical angle I2crit is 41.25 deg at λ = 587.3 nm.
For collimated light incident perpendicular to the entrance surface, uncoated roof
surfaces could function by TIR.
Figure 2.19(d) shows a convergent light beam incident on the Schmidt prism,
where the focus is outside the prism. The minimum f /# of this beam is determined
by any TIR failure at surface 2. In the tangential (x-z) plane shown, the maximum
I1 value is given by

I1 max arcsin n sin(45 deg − I2crit ) ,

 
(2.19)

where I1 max is the angle of incidence of the critical lower ray shown. This would
be the same as for a TIR right-angle prism, and its value is I1 max = 5.7 deg for
BK7 glass (nd = 1.5168). By the use of SF10 glass (nd = 1.7283), I1 max can be
dramatically increased to 16.8 deg, yielding an f /1.7 beam.
Using the x0 , y0 , z0 coordinate system (z0 axis perpendicular to entrance
surface 1), the direction cosines of the critical lower tangential ray A1 are given
General Prisms and Reflectors 51

Figure 2.19 (a) Schmidt prism, side view layout. (b) Schmidt prism, end view layout. (c)
Schmidt prism, perspective view. (d) Schmidt prism with convergent f /5 incident beam
(I1 = 5.7 deg for n = 1.5168).
52 Chapter 2

simply by

K x01 0 = −cos (90 deg − I1 max ) = −0.09932, (2.20a)

Ky01 0 = 0, (2.20b)
Kz01 0 = cos(I1 max ) = 0.9951. (2.20c)

For the upper tangential ray A2 :

K x01 0 = cos(90 deg − I1 max ) = 0.09932, (2.21a)

Ky01 0 = 0, (2.21b)
Kz01 0 = cos(I1 max ) = 0.9951. (2.21c)

For the sagittal ray B1 :

K x01 0 = 0, (2.22a)
Ky01 0 = −cos (90 deg − I1 max ) = −0.2890, (2.22b)
Kz01 0 = cos(I1 max ) = 0.9951, (2.22c)

For the sagittal ray B2 :

K x01 0 = 0 (2.23a)
Ky01 0 = cos(90 deg − I1 max ) = 0.2890, (2.23b)
Kz01 0 = cos(I1 max ) = 0.9951. (2.23c)

By referencing the tangential and sagittal rays to the original (x, y, z) coordinate
system, we can use the prism geometry defined by Eqs. (2.11a) through (2.11f).
The recalculated direction cosines for the tangential ray A1 are

K x01 = −cos (90 deg − 22.5 deg + I1 max ) = −0.2890, (2.24a)

Ky01 = 0, (2.24b)
Kz01 = cos(22.5 deg − I1 max ) = 0.9573. (2.24c)

The recalculated direction cosines for the tangential ray A2 are

K x01 0 = −cos (90 deg − 22.5 deg − I1 max ) = −0.4726, (2.25a)

Ky01 0 = 0, (2.25b)
Kz01 0 = −cos (22.5 deg + I1 max ) = 0.8813. (2.25c)

The direction angles (α x , βy , γz ) of the sagittal ray B1 are α x = 67.62 deg, βy =

84.3 deg, and γz = 23.18 deg. The resultant direction cosines are K x01 = −0.3808,
Ky01 = −0.09932, and Kz01 = 0.9193. The direction cosines for the sagittal ray B2
are correspondingly K x01 = −0.3808, Ky01 = 0.09932, and Kz01 = 0.9193. Using
General Prisms and Reflectors 53

the refraction matrix R from Eq. (1.75), and the reflection matrices in Eqs. (2.14a)
to (2.14d), the calculated internal angles of incidence for the boundary rays are
summarized in Table 2.8.
Using Eq. (1.23), the minimum f /# of this entrance beam in air is ≈ f /5 for
highest efficiency without encountering light leakage at any surface.

2.11 Pechan Prism

The Pechan prism is a two-element prism separated by an air space, with five
reflections. The basic structure is shown in Figs. 2.20(a) and (b). The lower input
half is a bisected Penta prism used as a 45-deg deviator (sometimes called a 45-
deg Bauernfeind prism), and the upper output half is similar to a Schmidt prism
with the roof section replaced by a planar-coated reflector surface. The air space
is mechanically held at ≈0.05 mm. The output rays are nearly coaxial with the
input rays, but with a considerable increase in optical path length. Neglecting any
refraction, the prism reflection matrix R is given by

−1 0 0
 
R = (R5 )(R4 )(R3 )(R2 )(R1 ) =  0 1 0 , (2.26)
0 0 1

where R1 , R2 , R3 , R4 , and R5 are calculated using Eq. (1.55). Here K x 0 = −K x ,

Ky 0 = Ky , and Kz 0 = Kz . A ray entering surface 1 normally (K x = 0, Ky = 0, Kz = 1)
will exit surface 3 with no angular deviation. Since |R| = −1, the image is
unreadable, and using Eq. (2.17), x0 = −x, y0 = y, and z0 = z. Therefore, a direct-
viewed object would be undeviated, unreadable, and rotated 180 deg. A Pechan
prism could be used as a compact image rotator like the Dove prism but can be
used in convergent or divergent light beams. The same f /# limitations apply as for
the Schmidt prism.
The Pechan prism has four transmitting surfaces and four reflecting surfaces.
Two of the surfaces perform double duty, functioning in both TIR and transmission
modes. To achieve high throughput, the coated reflector surfaces should have a very
high reflectance coating, such as the 64-layer dielectric OASIS coating produced
by Optricon, with a reflectance exceeding 99%. In addition, the external entrance
and exit surfaces should be antireflection coated. For the internal surfaces, adding
an antireflection coating may affect the TIR property, resulting in some light
leakage due to frustrated total internal reflection (FTIR).6,7

Table 2.8 Schmidt prism internal incident angles for f /5 convergent entrance beam
(n = 1.5168). Units are in degrees.
Plane I1max I2 I3 I4 I5 I6
Tangential A1 +5.70 41.25 50.64 50.64 41.25 3.75
Tangential A2 −5.70 48.76 47.96 47.96 48.76 3.75
Sagittal B1 +5.70 45.12 45.72 52.73 45.12 3.75
Sagittal B2 −5.70 45.12 52.73 45.72 45.12 3.75
54 Chapter 2

Figure 2.20 (a) Pechan prism design layout. (b) Pechan prism, perspective view.

2.12 Schmidt–Pechan Prism

The Schmidt–Pechan prism is a compound prism that has similar properties to the
roof Dove prism but is more compact and can be used in convergent or divergent
light beams. It is sometimes called a Pechan roof prism, or simply a roof prism in
the binocular trade. It consists of a Schmidt prism air spaced from the same section
of the Penta prism as used in the Pechan prism, yielding six reflective surfaces
[see Figs. 2.21(a) and (b)]. Neglecting refraction at surfaces 1 and 7, the reflection
matrix R is:
−1 0 0
 
R = (R6 ) . . . (R1 ) =  0 −1 0 . (2.27)
0 0 1

The determinant |R| = +1, indicating that the image is readable. Using
Eq. (2.17), the rotated coordinates (x0 , y0 , z0 ) of the exit beam, relative to (x, y, z) of
General Prisms and Reflectors 55

Figure 2.21 (a) Schmidt–Pechan prism design layout. (b) Schmidt–Pechan prism,
perspective view.

the entrance beam, are x0 = −x, y0 = −y, z0 = z, and the direct-view image is rotated
180 deg. In addition to antireflection coatings, a special phase-correction coating is
often applied to one or both of the roof surfaces. This phase-correction coating
was first introduced in Zeiss roof binoculars to compensate for a polarization
phase shift from reflections at the roof surfaces. Contrast and resolution of the
viewed image are claimed to be improved. A binocular viewing instrument with
a specified 9-layer phase-correction reflective coating on the roof surfaces of
a Schmidt–Pechan prism is described by Ito and Noguchi.8 Cojocaru has also
described phase-retarding thin films for totally reflecting prisms.9 When used
in quality binoculars, Schmidt–Pechan prisms are often constructed of BaK4
glass (nd = 1.5688, ν = 55.98) to provide a larger acceptance angle without TIR
leakage.
56 Chapter 2

2.13 Cube-Corner Retroreflector

The simplest form of a cube-corner reflector consists of three first-surface mirrors,
each aligned at right angles to the others. The reflection matrix for the hollow cube-
corner reflector of Fig. 2.22 is obtained from the reflective surface vectors k x , ky ,
kz , and

−1 0 0 1 0 0  1 0 0 

     
R1 =  0 1 0 , R2 = 0 −1 0  , R3 = 0 1 0  . (2.28)
0 0 1 0 0 −1 0 0 −1

Then,

−1 0 0 
 
Rcc = (R3 )(R2 )(R1 ) =  0 −1 0  .
 (2.29)
0 0 −1

If used as an imaging system, the determinant |Rcc | = −1, indicating an

unreadable image. Also, the image is inverted. Since K x 0 = −K x , Ky 0 = −Ky ,
and Kz 0 = −Kz 0 , the cube-corner functions as a constant deviation reflector
or retroreflector. Retroreflection occurs only when an incident ray hits all three
mirrors, and the cube-corner needs to be oriented correctly with respect to a
reference axis. There are several methods of achieving this, and one method is to
define the reference or optical axis as the line that trisects the cube base solid angle.
The direction angles of the reference axis with respect to the original coordinate
system are α = β = γ = 54.74 deg.
The aperture geometry of a cube-corner reflector is usually hexagonal or
triangular, as shown in Figs. 2.23(a) and 2.23(b). For the hexagonal case, three
full sides of the cube are exposed to incident light, and the leading corners define a
plane of incidence. For rays normal to this plane of incidence, every incident ray is
retroreflected, or the hexagonal aperture is the effective aperture, and the geometric
efficiency is 100%.

Figure 2.22 Cube-corner reflector geometry.

General Prisms and Reflectors 57

Figure 2.23 (a) Cube-corner reflector with hexagonal aperture. (b) Cube-corner reflector
with triangular aperture.

For a triangular aperture cube-corner, each side of the cube reflector is truncated
to form three 45-deg right triangles. Then, for rays entering the geometric aperture,
some incident rays will not be retroreflected, since part of the full cube is missing.
Eckhardt has defined an effective aperture that varies with the incident angle of
the light.10 For the triangle cube-corner, there are two separated triangular planes
defined by the forward single-side corners of the original cube, and the back
double-side corners of the truncated cube (Fig. 2.24). The effective aperture is
the hexagonal-shaped overlap region of these triangles, which ensures that all
rays entering this aperture will be retroreflected. Consider incident light along the
reference axis of the cube, the equilateral triangles having sides of length a, inside
the full cube hexagon having sides of length b, and the effective aperture hexagon
having sides of length c. Then,
√ 
 3  2
Area (triangle) =   a , (2.30a)
4
!
3 2
Area (cube hexagon) = b cot(30 deg), (2.30b)
2
!
3 2
Area (overlap hexagon) = c cot(30 deg), (2.30c)
2

where a2 = 2, b2 = [1 − cos(120 deg)], and c = a/3.

Let a ≡ 1.0. Then, the area (triangle) = 1.299, and the effective aperture area
(overlap hexagon) = 0.866. The retroreflection efficiency of the triangular cube-
corner for normal incidence light is

Area (triangle)
Efficiency = = 0.666. (2.31)
Area (overlap hexagon)
58 Chapter 2

Figure 2.24 Effective aperture of a triangular cube-corner reflector.

Thus, the efficiency of the triangular aperture cube-corner is ≈67% that of

the hexagonal-aperture cube-corner for normal incidence of light. The effective
aperture for off-axis rays can be calculated by projection of the overlap region onto
a plane normal to the reference axis. The dropoff in efficiency for other angles of
incidence is plotted in Fig. 2.25.
The hollow retroreflector can be considered as a solid-glass prism with a
refractive index n ≡ 1.0. The solid-glass cube-corner reflector prism is the type

Figure 2.25 Retroflection efficiency for hollow cube-corner reflectors.10 Adapted with
permission from the Optical Society of America.
General Prisms and Reflectors 59

that is usually available from commercial manufacturers. The entrance aperture is

usually circular, since the cube-corner is formed at the end of a short glass cylinder,
as shown in Fig. 2.26. These cube-corner prisms are commercially available with
a typical aperture diameter d in the 15–50-mm range, and a total length l in the
10–40 mm range, using BK7 glass. The cube-corners can reflect by TIR, or a
reflector coating can be applied to increase the acceptance angle and minimize
polarization effects. The corner angles are usually held to an accuracy of ±1
arcsec. Figure 2.27 shows the variation in retroreflective efficiency for hexagonal,
triangular, and circular apertures of solid cube-corner prisms for various angles of
incidence, having a refractive index n = 1.5.

Figure 2.26 Solid glass cube-corner reflector with circular aperture.

Figure 2.27 Retroreflection efficiency for solid glass cube-corner reflectors (n = 1.50).10
Adapted with permission from the Optical Society of America.
60 Chapter 2

References
1. J. P. C. Southall, Mirrors, Prisms and Lenses, 3rd ed., 113–132, Macmillan,
New York (1946).
2. R. Kingslake, “Dispersing prisms,” Chapter 1 in Applied Optics and Optical
Engineering, R. Kingslake, Ed., Vol. 5, 1–15, Academic Press, New York
(1969).
3. W. M. McClain, “How to mount a Pellin–Broca prism for laser work,” Appl.
Opt. 12(1), 153 (1973).
4. H. Moosmüller, “Brewster’s angle porro prism: a different use for a
Pellin–Broca prism,” Appl. Opt. 37(34), 8140–8142 (1998).
5. “Tunable operation,” in Introduction to Laser Technology, Section 10, Melles
Griot 2009 Technical Guide, p. 10.15 (2009).
6. H. Osterberg, “Coating of optical surfaces,” Section 21 in Military Standard-
ization Handbook—Optical Design, MIL-HDBK-141, 27–29, Defense Supply
Agency, Washington, DC (1962).
7. L. Li, “The design of optical thin film coatings with total and frustrated total
internal reflection,” Optics and Photonics News 14, 24–30 (2003).
8. T. Ito and M. Noguchi, “Viewing optical instrument having roof prism and a
roof prism,” U.S. Patent No. 6,304,395 (2001).
9. E. Cojocaru, “Simple relations for thin-film coated, phase retarding totally
reflecting prisms,” Appl. Opt. 33(14), 2878–2681 (1994).
10. H. D. Eckhardt, “Simple model of corner reflector phenomena,” Appl. Opt.
10(7), 1559–1566 (1971).
Chapter 3
Polarization Properties of Prisms
and Reflectors
3.1 Prisms Producing Polarized Light
3.1.1 Uniaxial double-refracting crystals
Certain types of crystals, such as calcite (Iceland spar or calcium carbonate) exhibit
the property of double refraction or birefringence, as first observed in calcite by
Erasmus Bartholinus in 1669. For the class of crystals called uniaxial, there is only
one direction where all light rays travel along the same path at a constant velocity.
This direction defines the optic axis or principal axis, and any plane that contains
the optic axis is called a principal plane (sometimes called a principal section).
The optic axis is not a specific line, but indicates a direction in the crystal where
there is no double refraction. For all rays not traveling along the optic axis, the
velocity is determined by a pair of refractive indices called the ordinary refractive
index no and the extraordinary refractive index ne , and the path of an incident ray
is split into two rays, the so-called o-rays and e-rays. Birefringence is specified by
the number (no − ne ). Moreover, these o-rays and e-rays are polarized and vibrate
in mutually perpendicular planes. Only rays traveling parallel to the optic axis will
not be split, and no is therefore assigned to this direction. One way to represent
this refractive index variation is by use of the indicatrix.1 Figure 3.1(a) shows a
positive uniaxial indicatrix in the shape of an oblate spheroid, where ne > no , and
Fig. 3.1(b) shows a negative uniaxial indicatrix in the shape of a prolate spheroid,
where no > ne . Both have circular symmetry in planes normal to the optic axis, and
when the indicatrix has a spherical shape, ne = no , and the crystal is isotropic.

3.1.2 Nicol polarizing prism

One of the first prism polarizers to utilize a birefringent crystal was developed by
William Nicol in 1828 and is known as the Nicol prism. The Nicol prism shown
in Fig. 3.2 is constructed from negative uniaxial calcite, where no = 1.6584 and
ne = 1.4864 for λ = 589.3 nm. Calcite is a widely used material because of
its clarity, stability, high spectral transmission range (200–5000 nm), and high
birefringence. Two triangular sections are optically coupled at the hypotenuse by
a thin coating of optically clear cement such as Canadian balsam (ncement ≈ 1.54),

61
62 Chapter 3

Figure 3.1 (a) Positive uniaxial indicatrix (ne > no ). (b) Negative uniaxial indicatrix (no > ne ).

Figure 3.2 Nicol prism polarizer made of calcite, no = 1.6584, ne = 1.4864.

with the optic axis direction as shown. An incident unpolarized ray is split at the
entrance surface, with both rays becoming linearly polarized. By controlling the
incident angle of the rays at the interface, the o-ray can undergo total internal
reflection (TIR), where I o crit = arcsin(ncement /no ) ≈ 68 deg. Since ncement > ne ,
the e-ray is always transmitted and exits the prism as linearly polarized light. This
separation of o-rays and e-rays by TIR is a useful technique that is used in other
types of polarizing prisms. Although the exit ray is parallel to the incident ray, there
is a slight lateral displacement (noncollinear), the angular field is limited, and the
interface cement will suffer damage at high power levels.
Polarization Properties of Prisms and Reflectors 63

3.1.3 Glan–Foucault polarizing prism

One of several Glan-type polarizing prisms is the Glan–Foucault prism, shown in
Fig. 3.3(a). The calcite prisms are air spaced at the interface, and each optic axis is
perpendicular to the plane of reflection. There is no separation of the ray paths in
the first prism section, but the o-ray moves slower in the first section and undergoes
retardance with respect to the e-ray. Again, TIR is used to separate the o-ray from
the e-ray, and s-polarized light is emitted from the exit face. The o-ray is usually
absorbed by blackening the side face. The field of view is determined by TIR failure
of the o-ray at the glass–air interface 2, or TIR of the e-ray at this interface. For
calcite, no = 1.6557 and ne = 1.4852 at λ = 630 nm. The corresponding critical
angles are I o crit = 37.16 deg and I e crit = 42.32 deg. As shown in Fig. 3.3(b), the
maximum angle of incidence I1 o for the o-ray at entrance surface 1 is estimated
by I1 o max = arcsin[no sin(38.5 deg − I o crit )] ≈ 2.4 deg. The maximum angle of
incidence I1 e for the e-ray is estimated by I1 e max = arcsin[no sin(I o crit −38.5 deg)] ≈
6.3 deg. This results in a narrow asymmetric field of view about the central axis
in the tangential plane. A nominal angular field is given as 6 deg at λ = 633 nm
by a commercial supplier of the Glan–Foucault prism, United Crystals Company.

Figure 3.3 (a) Glan–Foucault prism polarizer made of calcite, Icrit (no ) = 37.1 deg, Icrit (ne ) =
42.3 deg. (b) Asymmetric field of view of Glan–Foucault prism polarizer.
64 Chapter 3

Other specifications are a damage threshold of 30 W/cm2 continuous wave (CW)

or 300 W/cm2 pulsed laser radiation and transmittance of s-polarized light > 60%
at λ = 633 nm.

3.1.4 Glan–Thompson polarizing prism

The Glan–Thompson prism shown in Fig. 3.4 uses two cemented calcite prisms
with each optic axis perpendicular to the plane of reflection. Using TIR separation
at the glass–optical cement interface, s-polarized light is transmitted, while the
reflected p-polarized light is absorbed by a blackened side face. The transmission of
s-polarized light is > 90%, and the angular field is approximately doubled to about
12 deg compared to the Glan–Foucault prism, but the Glan–Thompson prism can
accept only up to 8 W/cm2 CW or 100 W/cm2 pulsed radiation due to the lower
damage threshold of the optical cement.

3.1.5 Glan–Taylor polarizing prism

The last of the Glan group to be described here is the Glan–Taylor prism, shown in
Fig. 3.5(a). Two air-spaced calcite prisms are oriented with both optic axes parallel
to the plane of reflection and parallel to the entrance and exit faces. Using TIR
separation, p-polarized light is transmitted, while the reflected s-polarized light is
either absorbed by a blackened side face or emitted through a clear exit window.
The transmission of p-polarized light is > 85%, and the angular field is about 6 deg.
It can accommodate the highest radiation level of the Glan group—up to 30 W/cm2
CW or 500 W/cm2 pulsed radiation. A modified form of the Glan–Taylor prism,
shown in Fig. 3.5(b), can produce orthogonal s-polarized and p-polarized output
beams. In addition, if the angle of incidence at the interface is close to Brewster’s
angle, there will be little reflection of p-polarized light. However, the intensity of
the s-polarized reflected beam will be much less than the transmitted p-polarized
beam.

Figure 3.4 Glan–Thompson prism polarizer made of calcite, Icrit (no ) = 37.1 deg, Icrit (ne ) =
42.3 deg.
Polarization Properties of Prisms and Reflectors 65

Figure 3.5 (a) Glan–Taylor prism polarizer made of calcite. (b) Glan–Taylor prism polarizer
having orthogonal outputs made of calcite.

3.1.6 Beam-displacing polarizing prism

Figure 3.6(a) shows a cleaved calcite rhomb where the optic axis is inclined in the
principal section. Each side has corner angles α = 78.08 deg and β = 101.92 deg.
The optic axis direction is determined by equally trisecting a β-β-β oblique corner
of the crystal. A ray enters at the edge of the principal section. The undeviated
s-polarized o-ray vibrates perpendicular to the principal section, and the deviated
p-polarized e-ray vibrates in the principal section, where both rays lie in the
principal section. The exiting p-polarized ray is displaced and parallel to the exiting
s-polarized ray. Typical exit ray separation is nominally 4 mm. This polarizing
prism has the advantage that both exit beams are completely polarized and of equal
intensity, although obviously the entrance beam diameter must be small. As in
Fig. 3.6(b), a calcite prism pair can produce a variable beam displacement between
the p-polarized and s-polarized rays.2 Here, one wedge prism is slid relative to
another wedge prism.
66 Chapter 3

Figure 3.6 (a) Beam-displacement prism polarizer. (b) Variable beam-displacement prism
polarizer.2

3.1.7 Wollaston polarizing prism

Another type of beam-splitting polarizing prism is the Wollaston prism, (for
William Hyde Wollaston), shown in Fig. 3.7. It usually consists of two calcite right-
angle prisms optically cemented together at the hypotenuse. The optic axis of each
section is orthogonal to that of the other section. An unpolarized ray traversing the
first prism section is not split, but the o-ray is retarded with respect to the e-ray.
The o-ray vibrates parallel to the optic axis and the e-ray perpendicular to the optic
axis. Upon entering the second section, the o-ray becomes the e-ray, and vice versa.
The o-ray, now slower, is bent toward the interface normal, and the e-ray is bent
away from the interface normal. Prisms with a deviation angle δ from about 5 to
45 deg between the exit beams can be obtained, depending on the right-angle prism
base angles. For very high-power applications, the prisms may not be cemented
together, resulting in a reduction of transmission.
Polarization Properties of Prisms and Reflectors 67

Figure 3.7 Wollaston prism polarizer (calcite).

3.1.8 Nomarski polarizing prism

The Nomarski prism, named for Georges Nomarski, is a modified Wollaston prism
(Fig. 3.8). The optic axis of the first right-angle calcite prism is skewed as shown,
while the optic axis of the second prism is oriented the same as for the Wollaston
prism. This angled optic axis causes the ordinary and extraordinary rays to intersect
outside the prism, forming an interference plane. The resulting phase shifts can
be detected by an analyzer. The exact distance of this interference plane from
the prism is determined by the angle of the skewed optic axis and is set by the
manufacturer. Normarski prisms are used in differential interference contrast (DIC)
microscopes.

3.1.9 Rochon polarizing prism

Related to the Wollaston polarizing prism, the Rochon prism (for Alexis Marie
Rochon) has the optic axis of the first calcite prism section in the direction of the
incident ray (Fig. 3.9), and there is no distinction between the o-ray and the e-ray
in this section. The split at the second section interface produces no deviation of
the s-polarized o-ray, while the p-polarized e-ray is deviated from the interface
normal. The first calcite section can be replaced by a more robust isotropic glass
section, choosing a glass with a refractive index and dispersion close to either of the

Figure 3.8 Nomarski prism polarizer (calcite).

68 Chapter 3

Figure 3.9 Rochon prism polarizer (calcite).

refractive index values of calcite.3 Glass-calcite Rochon prisms are commercially

available using FK5 glass (nd = 1.4875, νd = 70.41), which is close to ne and the
dispersion of calcite. Typical beam-deviation angles are 5, 10, and 15 deg.
Other double-refracting crystals used in commercially available polarizing
prisms are crystal quartz (circular and low birefringence), alpha-BBO (ne =
1.6021, no = 1.6776 at λ = 552 nm), YVO4 (ne = 2.2154, no = 1.9929
at λ = 630 nm), magnesium fluoride, and titanium dioxide. Polarizing prisms’
surfaces are usually antireflection coated.

3.1.10 MacNeille polarizing beamsplitter cube

Another method to produce polarized light is by the deposition of alternating high-
and low-index film layers, as described by MacNeille.4 Figure 3.10 illustrates
a seven-layer transparent thin-film stack having refractive indices n1 and n2 ,
deposited between transparent bulk material having a refractive index n3 , where
n1  n2 , and n1 > n3 . All internal rays within the layers hit the next layer at
Brewster’s angle. A fraction of the reflected light at each layer interface is therefore
completely s-polarized. If we choose n1 = 2.3 (zinc sulfide) and n2 = 1.38

Figure 3.10 Polarizing thin-film stack.

Polarization Properties of Prisms and Reflectors 69

(magnesium fluoride), then Brewster’s angles α and β are given by

!
n2
α = arctan = 31.0 deg, (3.1)
n1
!
n1
β = arctan = 59.0 deg, (3.2)
n2

where α + β = 90 deg and Snell’s law is satisfied at each layer interface.

A useful incident angle from the bulk material to the first layer is θ = 45 deg.
The required refractive index of the bulk material is then calculated from Snell’s
law:
" #
sin α
n3 = n1 ≈ 1.67. (3.3)
sin(45 deg)

Since θ between n3 and n2 (53.9 deg) is not Brewster’s angle, this incident ray is
not completely s-polarized on reflection.
To maximize the intensity of the reflected s-polarized ray at each layer, the layer
thickness is controlled such that the ray reflected from the next layer is in phase
with the incident ray. To achieve this, the physical thicknesses t1 and t2 of the layers
are controlled to be
λ
t1 = , (3.4)
4 (n1 2 + n2 2 )/n1 2
p

λ
t2 = p , (3.5)
4 (n1 2 + n2 2 )/n2 2

where λ is the wavelength of the incident light, nominally 550 nm. For these seven
layers, approximately 50% of the incident light is reflected as s-polarized, while
the other half is transmitted as p-polarized light.
Figure 3.11 shows a 50R/50T polarizing beamsplitter (PBS) cube, where the
deposited layers lie on the hypotenuse of a right-angle prism, and another right-
angle prism is coupled to the hypotenuse using a thin coating of optical cement
having a refractive index close to n3 . From Eq. (3.3), a suitable material for the
cube would be SF5 glass (nd = 1.673). Both the reflected s-polarized light and the
p-polarized transmitted light are at least 95% polarized over the visible spectrum,
and the beamsplitter is usable for 40 deg ≤ θ ≤ 50 deg, or ±5 deg from the ideal
incident angle at the interface. The extinction ratio is the ratio of the transmitted or
reflected primary polarization component to the opposite polarization component.
It is possible to increase the angular field of MacNeille PBS cubes by modifying the
beam-splitting coating, albeit with a reduced usable wavelength range.5 Modern
commercial PBS cubes of a modified MacNeille design can achieve an input f /#
down to ≈ f /2.5 and a transmission extinction ratio ≈ 1,000:1. The minimum
working f /# of a PBS cube for projection display applications, without noticeable
loss of contrast, has been stated to be ≈ f /3.3.6
70 Chapter 3

Figure 3.11 MacNeille 50R/50T PBS cube in convergent beam of half-angle θ.

3.1.11 Birefringent multilayer reflective polarizing film

A type of reflecting polarizer film has been developed by 3M that uses a multilayer
stack of biaxial birefringent polymer layers.7 It is designed to produce high
reflectance for light with its plane of polarization parallel to one axis, and high
transmission for light with its plane of polarization parallel to a second axis, both
over a wide range of incident angles. A biaxial birefringent material, where the
refractive indices differ along all three axes, can be produced by stretching the
multilayer stack in one direction (uniaxial stretching). Figure 3.12 shows a single
interface between layers for a biaxial birefringent film, and the associated refractive
indices. For light incident in the x-y stretch plane, n10 = n1x , n20 = n2x for
p-polarized light, and n10 = n1y , n20 = n2y for s-polarized light. For light incident in
the y-z nonstretched plane, n10 = n1y , n20 = n2y for p-polarized light, and n10 = n1x ,
n20 = n2x for s-polarized light. The x direction is the extinction direction, and the
y direction is the transmission direction. Typical values are n1x = 1.88, n1y = 1.64,
n1z = variable, n2x = 1.65, n2y = variable, and n2z = variable. For the large index
differential of 1.88 − 1.65 = 0.23 in the stretch direction, there is a high reflectance
of s-polarized light for a stack of hundreds of layers, and the angular transmission
of p-polarized light depends on the n1z /n2z index ratio.

Figure 3.12 Two-layer single interface biaxial birefringent film.

Polarization Properties of Prisms and Reflectors 71

An example of a reflective polarizing film consists of 601 alternating layers of

polyethylene (PEN) and a copolymer of 2,6 napthalene dicarboxylate, methyl ester,
dimethyl isophthalate, and dimethyl terephtalate with ethylene glycol (coPEN),
with controlled layer thicknesses. The multilayer film was heated and stretched
in the x direction, producing a film ≈ 0.5 mm thick. The transmission versus
wavelength of the polarizing sheet is plotted in Fig. 3.13, where the “a” curve
corresponds to p-polarized light at I = 0 deg, the “b” curve corresponds to
p-polarized light at I = 60 deg, and the “c” curve corresponds to the s-polarized
light extinction at I = 0 deg in the stretched x direction.
There is very high transmission of p-polarized light (80–100%) over a wide
angular range, and a very high extinction (reflectance) of s-polarized light in the
visible 400–700-nm range. Adding an antireflection coating will further increase
the transmission of p-polarized light.

3.1.12 Polarizing beamsplitter elements using birefringent polarizing

film
A PBS has been developed using a type of 3M birefringent polarizing film instead
of the MacNeille stack.8 The polarizing film can be encased in a cube split along
the diagonal [Fig. 3.14(a)], or between thin, tilted glass plates [Fig. 3.14(b)].
These types of PBS elements are characterized by fixed polarization x-y axes
of the polarizing film that do not vary with the angle of incidence, as for the
MacNeille prism. Unlike the MacNeille PBS cube, the polarization properties are

Figure 3.13 Transmission versus wavelength for birefringent reflective polarizing film.7
72 Chapter 3

Figure 3.14 (a) A birefringent film PBS cube. (b) A birefringent film PBS plate.

also independent of the refractive index of the cube material. Since the polarization
axes are invariant with respect to the PBS, this type is especially useful in
convergent or divergent light beams, and is known as a wide-angle Cartesian
PBS. The extinction ratio (ratio of transmitted or reflected primary polarization
component to the opposite polarization component) can exceed 10,000:1 for a
birefringent PBS cube, for both transmitted and reflected rays.
For either the cube or plate configuration, it is important to use a nonbirefringent
glass, such as PBH56 (n ≈ 1.85), to avoid visible polarization effects. The
multilayer reflective polarizing beamsplitter film typically has 892 layers at about
0.15-mm thickness and a refractive index n ≈ 1.55. The preferred orientation for
large cone angle, high-contrast optical systems is such that p-polarized light is
transmitted along the x axis, and s-polarized light is transmitted along the y axis.

3.1.13 Wire-grid polarizing beamsplitter

A wire-grid polarizer is a thin planar element consisting of spaced fine parallel
metallic lines (normally aluminum or gold wires) that are deposited on a clear
dielectric substrate, as shown in Fig. 3.15. The line width w is approximately the
same as the line spacing (period) d, and both must be small with respect to the
light wavelength λ. Incident unpolarized light is reflected by the metallic lines as
s-polarized light and transmitted through the dielectric substrate as p-polarized
Polarization Properties of Prisms and Reflectors 73

Figure 3.15 Wire-grid polarizer.

light, where d < λ/2. Wire-grid polarizers were first produced for use in the
microwave region, because a wider grid spacing is more easily fabricated. For use
as a visible-light PBS, the line spacing must have dimensions ≤ 100 nm. Planar
polarizers were first produced by Moxtek, Inc., as the ProFlux wire-grid PBS.9,10
Visible-light (420–700 nm) wire-grid planar polarizers are also commercially
available from suppliers such as Edmund Optics, with a clear aperture up to
44 mm2 .
A wire-grid PBS cube has been constructed by encasing a wire-grid polarizer
between two diagonal halves of a glass cube.11 The wire-grid polarizer substrate is
Corning 1737F glass, which is cemented to the BK7 glass cube with Norland 61
cement. The wires are oriented perpendicular to the triangular edges of the prism
halves, providing reflection of s-polarized light and transmission of p-polarized
light. This wire-grid polarizer has an incident angle ≈ 45 deg, which limits the
useful wavelength range to longer than mid-visible.

3.1.14 Polarizing beamsplitter using frustrated total internal reflection

A polarizing prismatic beamsplitter can be designed that uses a thin film with
incident angles greater than the critical angle.12 It is based on frustrated total
internal reflection (FTIR) and thin-film interference effects. The basic PBS is
illustrated in Fig. 3.16, where p-polarized light is reflected from the film interface,
and s-polarized light is transmitted. The thin-film structure consists of a stack of
high- and low-index layers. For the low-index layers, the internal angle of incidence
must be greater than the critical angle, and the reflection phase changes depend on
the refractive index ratios and the angle of incidence. If the thicknesses of the low-
index layers are thin, then the evanescent wave can tunnel to the next high-index
layer, FTIR occurs at the high-index/low-index boundaries, and p-polarized light
74 Chapter 3

Figure 3.16 Thin-film PBS using frustrated total internal reflection.12

is reflected as from a metal reflector. At the low-index/high-index interfaces, the

phase is varied such that for many thin layers, the transmitted light is s-polarized
and independent of wavelength and angle of incidence. There is a minimum angle
of incidence Imin , where Imin > Icrit , where the PBS will function. The theory of
this PBS is detailed in Ref. 12. This PBS is in contrast to the conventional type of
PBS, which transmits p-polarized light and reflects s-polarized light.
As a cited example, a 45-layer film is encased in glass prism halves, where n0 =
1.75. For the high-index layer, n1 = 2.35, and for the low-index layer, n2 = 1.38.
The layer thicknesses d are varied between 8.3–44.2 µm for the high-index layers,
and between 30.5–107.9 µm for the low-index layers. The minimum angle of
incidence at the film is calculated to be 62.50 deg, and the calculated critical angle
is 52.05 deg. The usable angular range at the film is 64–77 deg, corresponding
to ±11.4 deg at the prism entrance face. The extinction ratios are ≈ 30,000:1 for
reflected p-polarized light, and > 50,000:1 for s-polarized transmitted light.

3.1.15 Polarizing beamsplitter prism with common polarization output

It is sometimes desirable to produce a beam having a common polarization mode
and output direction. The modified PBS of Fig. 3.17(a) adds two right-angle
prisms, a rhomb prism, a half-wave retarder thin plate or birefringent polymer
film, and a 50R/50T nonpolarizing beamsplitter film.13 Collimated light entering
the PBS cube is split into transmitted p-polarized light and reflected s-polarized
light. The s-polarized light is retroreflected and displaced by prism 1. A half-
wave retarder film is embedded between prism 2 and prism 3, which converts
the s-polarized light to p-polarized light. A nonpolarizing beam-splitting film is
embedded between prism 1 and right-angle prism 2, and the reflected half is
integrated with the directly transmitted p-polarized light from the PBS cube. The
transmitted light is reflected by the rhomb prism 3 to exit collinear to the lower
beam with a common polarization mode. Figure 3.17(b) illustrates an alternative
Polarization Properties of Prisms and Reflectors 75

Figure 3.17 (a) Prism polarization converter with λ/4 retarder.13 (b) Prism polarization
converter with λ/2 retarder.13

arrangement with a quarter-wave retarder on the entire hypotenuse face of prism 1.

In both cases, the output aperture is doubled in the x direction, and the output
subbeams are not integrated. Devices of this type are often called polarization
converters.

3.2 Prisms Controlling the Polarization of Light

3.2.1 Fresnel rhomb retarders
It can be generally stated that every prism that has reflecting surfaces will affect
the polarization state of incident light in some manner. Also, most light that is
reflected from terrestrial objects viewed by the unaided eye or through a viewing
device such as binoculars is partially polarized. That is why polarized sunglasses
are useful in reducing horizontally reflected glare. When the polarization state of
76 Chapter 3

the input beam can be controlled and specified, certain prism types are useful
for changing the state of polarization in a predictable manner. A useful way to
specify the polarization of a light beam is the polarization ellipse in Fig. 3.18.
Since the most general state of polarization is elliptic, the polarization ellipse can
represent an arbitrary polarization state. The azimuth angle Ψ is defined as the
angle between the major semiaxis a and the x axis. The ellipticity angle X is defined
as X = arctan(b/a), where b is the semiminor axis, while the ellipticity is defined
as (b/a). Thus Ψ = 0 deg, X = 0 deg would indicate linearly polarized light along
the x axis, and Ψ = 90 deg, X = 45 deg would indicate circularly polarized light.
A well-known prism type is the Fresnel rhomb, named for Augustin Jean
Fresnel. Figure 3.19 shows a single Fresnel rhomb. Through two total internal
reflections, 45-deg linearly polarized input light is converted to circularly polarized
output, or it functions as a 90-deg or quarter-wave retarder. The 45-deg linearly
polarized input, with equal components along the x axis and y axis, allows the TIR
phase changes to be calculated from the equations in Sec. 1.6. From Eq. (1.20), a
45-deg phase shift occurs at incident angles of 47.87 deg and 55.22 deg for BK7
glass (nd = 1.5168) in air. The larger incident angle is normally chosen, as it

Figure 3.18 Polarization ellipse.

Figure 3.19 Fresnel rhomb quarter-wave retarder.

Polarization Properties of Prisms and Reflectors 77

provides less wavelength dependence of the retardance. A conventional BK7 glass

(nd = 1.5168) Fresnel rhomb has opposite acute vertex angles α = 55.22 deg,
a physical length L, and a beam displacement D. For a square aperture having
sides A ≡ 1.0 unit, L = A tan α[1 + sin(2α − 90 deg)] = 1.94 units, and
D = A [1 + sin(2α − 90 deg)] = 1.35 units.
Figure 3.20(a) shows how a half-wave retarder can be constructed from two
identical quarter-wave rhomb retarders. The two rhombs can be cemented together,
brought into optical contact, or air spaced. The output beam remains collinear
with the input beam, but the retarder length is doubled. Both single and double
Fresnel rhombs are available commercially with aperture diameters in the 10–20-
mm range, and at optimized wavelengths in fused quartz, FK5, and BK7 glass.
For quarter- and half-wave rhomb retarders in BK7 glass (I = 55.22 deg at λ =
587.56 mm), and using Eq. (1.20) at each reflecting surface, the total retardance is
obtained. These rhomb retarders are nondispersive prisms. However, they are often
called achromatic retarders because the change in retardance with wavelength is
much less than for crystal quartz plate retarders.14
A quarter-wave double Fresnel rhomb can be designed by joining two eighth-
wave Fresnel rhombs. It would be useful for producing circularly polarized
collinear output from a linearly polarized input. Figure 3.20(b) shows the design
parameters for BK7 glass. Each of the four internal angles of incidence I =
74.76 deg, yielding a relative phase shift δ = 22.5 deg at each reflection. If
A = 20 mm, α = I = 74.76 deg, then the total length 2L = 273 mm, with D = 0.
This is not a very compact retarder. Table 3.1 gives the retardance variation over
the visible spectrum for these rhomb retarders.

Figure 3.20 (a) Double Fresnel rhomb half-wave retarder. (b) Double Fresnel rhomb
quarter-wave retarder.
78 Chapter 3

Table 3.1 Retardance variation with wavelength for single and double Fresnel rhombs.
Wavelength λ (nm) λ/4 single Fresnel λ/2 double Fresnel λ/4 double Fresnel
rhomb (deg) rhomb (deg) rhomb (deg)
90-deg retardance 180-deg retardance 90-deg retardance
656.27 89.8 179.6 89.9
587.56 90.0 180.0 90.0
546.07 90.2 180.4 90.1
486.13 90.6 181.2 90.3
435.83 91.0 182.0 90.5
404.66 91.2 182.4 90.7

The single Fresnel rhomb can be further achromatized by coating one of the
reflecting surfaces with a 20-nm-thick layer of MgF2 and changing the acute
rhomb angles to α = 51.5 deg.15 This results in a maximum phase retardation of
exactly 90 deg, which further reduces variation of the retardance with the internal
angle of incidence. The resulting improvement in Fig. 3.21 provides about 0.4-deg
retardance variation over the wavelength range 334.1 nm to 546.1 nm.14

3.2.2 Total-internal-reflecting cube-corner retarders

A prismatic cube-corner reflector formed as a single prism is shown in Fig. 3.22,
where the face reflections occur by TIR. The internal reflections introduce phase
changes that alter the azimuth angle and ellipticity of incident polarized light, such
that incident linearly polarized light is both rotated and converted to elliptically
polarized light. The polarization changes that occur during reflection for a TIR
cube-corner can be shown in a polarization pupil map. For a BK7 glass cube-corner
in air, Fig. 3.23 shows a six-segment pattern of the various polarization states of
the output beam for a linearly polarized input beam.16

Figure 3.21 Phase retardance versus wavelength for uncoated and coated Fresnel
rhombs using BK7 glass. (Adapted from Ref. 14, with permission from the Optical Society
of America.)
Polarization Properties of Prisms and Reflectors 79

Figure 3.22 TIR cube-corner reflector.

Figure 3.23 Polarization pupil map for perfect TIR cube-corner reflector (BK7/air
interface). (Reproduced from Ref. 16, courtesy of ZEMAX.)

Player has analyzed the polarization properties of uncoated TIR cube-corner

reflectors for changes in retardance δ and azimuth angle Ψ.17 For this retroreflector,
with input light parallel to √the optical axis, the angle of incidence at each of the
three surfaces is I = arcsin 2/3 = 54.74 deg. Using Eq. (1.20), the relative phase
80 Chapter 3

shift δ for the cube-corner at a single face is then

p
δ = arctan 2 − 3/n2 , (3.6)

where n is the refractive index of the cube-corner substrate in air.

The retardance δcc for three reflections on the cube-corner is
! !
3δ δ 

 cos 3 cos
 
2 3 
δcc = 2 arccos  +  , (3.7)

 4 4 
 

and the azimuth angle Ψ is

√
! 
δ

 2 3 sin
 
2

Ψ = 0.5 arctan  !  . (3.8)
 
!
3δ δ 
+ sin

 sin 
2 2

Player has noted that if the refractive index n is selectively chosen—e.g., SF14
glass—then the cube-corner retardance δ ≈ 90 deg and operates as a quarter-wave
retroreflecting retarder.17 Table 3.2 gives the retardance and azimuth angles for
various cube-corner substrates. For SF14 glass, the variation in the cube-corner
retardance with wavelength is given in Table 3.3. Over the visible spectrum, there
is a low variation of δ and an even lower variation of Ψ.

3.2.3 Phase-coated total-internal-reflecting right-angle prism retarders

A single low-index-glass right-angle prism can function as a 90-deg retarder by

a single internal reflection by coating the reflecting face with a single layer of
dielectric film.18 The film layer produces additional retardance to the inherent
TIR retardance. Figure 3.24 illustrates the retarder prism with a refractive index

Table 3.2 Variation of retardance and azimuth angle of TIR cube-corner reflector based
on various glass types.
Glass type Single-face Cube-corner Azimuth
retardance δ (deg) retardance δ (deg) angle Ψ (deg)
Ideal (nd = 1.7675) 54.03 90.00 16.9
BK7 (nd = 1.5168) 45.29 76.35 16.32
SF14 (nd = 1.7618) 53.89 89.79 16.89
LASF9 (nd = 1.8503) 55.85 92.78 17.03
Polarization Properties of Prisms and Reflectors 81

Table 3.3 Wavelength variation of retardance and azimuth angle of TIR cube-corner
reflector using SK14 glass.
Wavelength λ (nm) Cube-corner retardance δ (deg) Azimuth angle Ψ (deg)
706.52 89.3 16.87
656.27 89.48 16.88
587.56 89.79 16.89
479.99 90.6 16.93
435.83 91.15 16.95
404.66 91.68 16.98

Figure 3.24 Right-angle TIR prism quarter-wave retarder. (Adapted from Ref. 18, with
permission from the Optical Society of America.)

nprism ≈ 1.51, and Fig. 3.25 plots the phase shift difference δ as a function of
normalized thickness d/λ for various film refractive indices at an angle of incidence
α = 45 deg. Retardance of 90 deg can be achieved for dual d/λ values, and for film
indices between 2.2 and 2.4, with small variation of retardance with wavelength.
Thus for d/λ ≈ 0.125, the required film thickness at λ = 546 nm would be
d ≈ 68 nm. This prism retarder can be further achromatized by additional thin
dielectric layers. For example, a three-layer coating can produce δ = 90.0 deg,
varying < 0.02 deg at α = 45 deg, over a ±15% bandwidth. For these multilayer
coatings, both the refractive indices and thicknesses are varied.
Three conjoined right-angle prisms, with coated reflecting faces, can produce
useful 90-deg or 180-deg achromatic retardation with all internal angles of
incidence α = 45 deg.19 There are four internal reflections producing collinear
input and output rays. Figure 3.26(a) illustrates a BK7 glass (n ≈ 1.52) prism
with any one of the four surfaces coated with a thin dielectric film of high-index
ZnS (n ≈ 2.39), overcoated by a thicker layer of MgF2 (n ≈ 1.38). A retardance
of 180 deg is produced with a standard deviation of 1.66 deg over the visible
wavelength range of 400–700 nm. Figure 3.26(b) shows a fused quartz (n ≈ 1.46)
82 Chapter 3

Figure 3.25 Right-angle TIR prism retarder—phase retardation versus normalized film
thickness. (Adapted from Ref. 18, with permission from the Optical Society of America.)

prism having all four surfaces coated, producing a retardance of 90 deg over the
visible range, with a standard deviation of 1.03 deg. The coatings also consist of a
thin film of ZnS, overcoated by a thicker coating of MgF2 . Table 3.4 summarizes
the design parameters for these right-angle prism retarders.

3.3 Polarization Preservation in Prisms and Reflectors

3.3.1 Polarization-preserving total-internal-reflecting prism
A useful application of prisms is to deviate or displace a polarized input beam
without significantly affecting the polarization of the output beam. Such prisms
are called polarization-preserving prisms. In the case of total-internal-reflecting
surfaces, which normally produce relative phase shifts for oblique polarized light,
these phase shifts are compensated by the prism geometry, input-beam polarization
direction, and film coatings on the facets.

Table 3.4 Design parameters for 90-deg and 180-deg right-angle prism retarders.19
Prism glass Thickness of Thickness of Average phase Standard Number of
first-layer second-layer shift (deg) deviation of coated
ZnS (nm) MgF2 (nm) retardance (deg) surfaces
BK7 15.63 67.53 179.84 1.66 1
Fused quartz 0.17 50.0 89.73 1.03 4
Polarization Properties of Prisms and Reflectors 83

Figure 3.26 (a) Coated right-angle prism 180-deg phase retarder using BK7 glass.
(b) Coated right-angle prism 90-deg phase retarder using BK7 glass. (Adapted from Ref. 19,
with permission from the Optical Society of America.)

Figure 3.27 illustrates an uncoated prism pair that displaces and deviates the
incident light by 180 deg and preserves the polarization.20 The input prism
produces three total internal reflections in the x-z vertical plane, and the output
right-angle prism produces two total internal reflections in the y-z horizontal plane.
The three reflections I1 = +75 deg, I2 = −60 deg, and I3 = +75 deg in the first
prism are balanced by the two reflections of I4 = +45 deg and I5 = +45 deg in the
second prism, where the sign is positive for counterclockwise beam reflection and
84 Chapter 3

Figure 3.27 Polarization-preserving prism pair. (Adapted from Ref. 20, with permission
from the Optical Society of America.)

negative for clockwise reflection. Then,

I1 + I2 + I3 = I4 + I5 = 90 deg . (3.9)

The phase shifts are equal in orthogonal planes, and the prism pair will then
transmit any mode of polarized light unchanged.
Figure 3.28 shows a polarization-preserving prism that uses four orthogonal
internal reflections and produces a parallel displaced output.21 As shown, the
electric field vector E of the input beam is oriented at 45 deg to the x axis and
has components E x and Ey . The beam enters the prism normally and undergoes
an orthogonal TIR reflection at each surface. The angle of incidence I at each

Figure 3.28 Polarization-preserving prism with coaxial output. (Adapted from Ref. 21, with
permission from the Optical Society of America.)
Polarization Properties of Prisms and Reflectors 85

TIR surface is 45 deg, where I > Icrit (Icrit ≈ 41.2 deg for BK7 glass). The
E x component undergoes p-polarization TIR at surface 1, s-polarization TIR at
surfaces 2 and 3, and p-polarization TIR at surface 4. The Ey component undergoes
s-polarization TIR at surface 1, p-polarization TIR at surfaces 2 and 3, and
s-polarization TIR at surface 4. The components are rotated as shown, and there is
a relative phase shift at each reflection that produces internal elliptical polarization
states. However, by the use of these four reflecting surfaces, the accumulated phase
shifts for both E x and Ey are equal, and there is no relative phase shift δ between
the input and exit beams. The beam exits the prism parallel to the input beam with
a displacement D, and with the same polarization state and orientation as the input
beam. This polarization-preserving prism is achromatic.
Another method to preserve the polarization of more general types of reflecting
prisms is by the coating of thin-film dielectric layers on the TIR surfaces.
Polarization is preserved by designing the retardance δ to be close to zero at each
TIR surface, where

δ = ∆Φ⊥ − ∆Φk ≈ 0, (3.10)

∆Φ⊥ and ∆Φk are defined in Eqs. (1.18) and (1.19), and δ is defined from
Eq. (1.20):
 p 
 cos I sin2 I − n0 /n1 
δ = 2 arctan   , (3.11)
sin2 I

where the glass index ng > n0 and I > Icrit . Cojocaru has computed the refractive
indices for one-, two-, and three-layer coatings, using quarter- and half-wave thick
coatings, to achieve polarization preservation on a glass substrate.22 For a single-
layer coating, the following are given by

λ0
n21 = n0 ng , where d1 = , (3.12)
4
 2 
 2n  λ0
n2g =  20  − n20 , where d1 = , (3.13)
sin Ig 2

where n1 is the coating index and n0 = 1.0 (air). The optical thickness d1 at design
wavelength λ0 , angle of incidence Ig , and physical coating thickness t1 is given by
q
d1 = t1 n21 − n2g sin2 Ig . (3.14)

From Eq. (3.14), the retardance for TIR at Ig = 45 deg and ng = 1.61 for an
uncoated surface is δ = 51.1 deg. The computed retardance for an ideal quarter-
wave coating (n1 = 1.27) on a glass substrate (ng = 1.61) as a function of λ/λ0
for Ig = 45 deg is plotted in Fig. 3.29(a). Figure 3.29(b) plots the retardance
dependence on Ig , for ng = 1.61, n1 = 1.27, and λ0 = 500 nm.
86 Chapter 3

Figure 3.29 (a) Wavelength dependence of retardance for single λ0 /4 layer (n0 = 1, ng =
1.61, n1 = 1.27). (b) Retardance variation versus angle of incidence for single λ0 /4 layer
(n0 = 1, ng = 1.61, n1 = 1.27). (Adapted from Ref. 22, with permission from the Optical Society
of America.)

In general, the retardance for a single-layer coated polarization-preserving TIR

prism depends on four factors:
• The prism glass index ng
• The layer index n1
• The optical thickness-to-wavelength ratio d1 /λ
• The angle of incidence Ig .
Polarization Properties of Prisms and Reflectors 87

For example, the dependence of the layer index n1 is illustrated in Fig. 3.30
for several values of the prism glass index ng . Wang et al. have analyzed these
factors theoretically and experimentally using a fused-quartz right-angle prism
with ng = 1.457, Ig = 45 deg, n1 = 1.23, and d1 /λ = 0.5.23 In Fig. 3.30, the
dependence of the retardance on the layer index n1 is illustrated for several values
of the prism glass index ng .

3.3.2 Polarization-preserving two-piece reflective axicon

It is also possible to design polarization-preserving first-surface metallic reflectors.
A two-piece biconical axicon reflector element is often used in laser applications
(Fig. 3.31). It is possible to preserve the polarization of light after two reflections
by the application of a single dielectric layer to the metallic axicon surfaces.24 The
dielectric material is the same for both reflecting surfaces, but the normalized film
thickness ζ (ratio of physical thickness to film-thickness period) has a value ζ1 on
one reflecting surface and ζ2 on the other for a given angle of incidence. To achieve
polarization-preservation of the system,
! !
Rp Rp
= f (I1 , ζ1 ) f (I2 , ζ2 ) ≡ 1, (3.15)
Rs 1 Rs 2

where R p and R s are the complex reflection coefficients for p-polarized and
s-polarized light at each surface, and I1 and I2 are the angles of incidence at each

Figure 3.30 Retardance dependence on layer index for several glass prism indices ng
(d1 /λ = 0.5, n1 = 1.27, Ig = 45 deg). (Adapted from Ref. 23, with permission from the Optical
Society of America.)
88 Chapter 3

Figure 3.31 Polarization-preserving two-piece reflecting axicon. (Adapted from Ref. 24,
with permission from the Optical Society of America.)

surface. Values of ζ1 and ζ2 pairs are calculated by an iterative process, restricting

solutions such that 0 ≤ ζ1,2 < 1. For example, a polarization-preserving axicon
was designed for a single-layer MgF2 coating (n = 1.38) on an aluminum substrate
(complex refractive index n = 1.212 − 6.924i), at the He-Ne laser wavelength
of 0.6828 µm. Figure 3.32 shows the computed ζ1,2 values versus the angles of
incidence, where I1 = I2 . For this system, polarization preservation is not possible
when I1,2 ≤ 34 deg. Representative results are summarized in Table 3.5, including
the overall reflectances for uncoated and coated (polarization-preserving) axicons.
As a bonus, the MgF2 coating protects the aluminum mirror coating and also
enhances the reflectance for p-polarized light at higher angles of incidence.

Figure 3.32 Normalized film thicknesses versus angle of incidence for polarization-
preserving axicon. (Adapted from Ref. 24, with permission from the Optical Society of
America.)
Polarization Properties of Prisms and Reflectors 89

Table 3.5 Representative results for MgF2 /Al polarization-preserving axicon.24

I1 , I2 (deg) ζ1 (µm) ζ2 (µm) Uncoated Uncoated Coated reflectance
p-reflectance s-reflectance
35 0.48173 0.51081 0.7906 0.8546 0.7048
45 0.43021 0.564371 0.7627 0.8733 0.7133
60 0.38793 0.611417 0.6896 0.9089 0.7352

3.3.3 Polarization-preserving total-internal-reflecting cube-corner prism

It is possible to preserve the polarization state of retroreflected light by a TIR cube-
corner prism with the use of phase-compensation coatings.25 Both azimuth angle
and ellipticity changes are considered for a TIR cube-corner of BK7 glass. In one
method, an interference stack on all three faces can reduce the normal TIR phase
difference from δ = 45.29 deg at each face to δ = 0 deg. Vertical linear s-polarized
light is incident on the cube-corner and exits the cube-corner as shown in Fig. 3.33.
Data for a four-layer stack is given in Table 3.6.
Layer 1 is the first layer on the BK7 substrate, and the optical thickness is a
quarter-wave layer with a physical thickness t.

3.3.4 Stokes parameters

Another way to describe the polarization state of a light beam is by the use
of the Stokes parameters, named for George Gabriel Stokes. The four Stokes
parameters, S 0 , S 1 , S 2 , and S 3 , are related to the polarization ellipse. One method
to describe the Stokes parameters is by a set of four sequential transmitting filters
that are irradiated by randomly polarized light. Then the Stokes parameters are

Figure 3.33 Polarization-preserving TIR cube-corner prism.25

Table 3.6 TIR cube-corner stack design producing δ = 0 deg at each face.25
Layer Material Refractive index n (λ = 633 nm) Optical thickness4nt (nm)
1 SiO2 1.46 813
2 TiO2 2.45 1066
3 SiO2 1.46 1090
4 TiO2 2.45 1702
90 Chapter 3

operationally defined by the following set of equations:26

S0 = 2H0 , (3.16a)
S1 = 2H1 − 2H0 , (3.16b)
S2 = 2H2 − 2H0 , (3.16c)
S3 = 2H3 − 2H0 , (3.16d)

where S 0 is the incident irradiance. H0 , H1 , H2 , and H3 are the measured

irradiances through each sequential filter. S 1 , S 2 , and S 3 represent different
polarization states such that:
• S 1 > 0 represents horizontal linearly polarized light.
• S 1 < 0 represents vertical linearly polarized light.
• S 1 = 0 represents elliptically polarized light where Ψ = ±45 deg, circularly
polarized light, or unpolarized light.
• S 2 > 0 represents linearly polarized light, where Ψ = +45 deg.
• S 2 < 0 represents linearly polarized light, where Ψ = −45 deg.
• S 2 = 0 represents linearly polarized light at other Ψ values.
• S 3 > 0 represents right handedness.
• S 3 < 0 represents left handedness.
• S 3 = 0 represents no specific handedness.
The Stokes parameters are usually normalized such that S 0 = 1 represents an
incident beam of unit intensity and the values of S 1 , S 2 , and S 3 are either 0 or 1.
Thus, the Stokes parameter set [S 0 S 1 S 2 S 3 ] = [1 0 1 0] describes the output
as linearly polarized light oriented at Ψ = +45 deg. Moreover, if two incoherent
beams are superimposed, each having specified Stokes parameter sets, then the
resultant polarization state is the sum of the components of each Stokes parameter
set. For unpolarized light, S 0 is a positive quantity, and S 1 = S 2 = S 3 = 0. For
completely polarized light, S 0 2 = S 1 2 + S 2 2 + S 3 2 , and for partially polarized light,
the degree of polarization V is
p
S 12 + S 22 + S 32
V= . (3.17)
S0

3.3.5 Depolarizing cube-corner prism

By control of the orientation and tilt of a solid-glass TIR cube-corner, it is possible
to produce unpolarized output.27 Figure 3.34 shows a solid-glass TIR prism with
a circular aperture (previously shown in perspective in Fig. 2.26). The cube-corner
is viewed normal to the circular aperture along the direction of the input light
beam. The back corners of the cube are A, B, and C, and the cube apex is D. The
forward cube-corners are E, F, and G. The back corners are projected on the planar
circular aperture (x-y plane) as A0 , B0 , C0 , and D0 , such that the back edges B-D,
A-D, and C-D are projected on the circular aperture as B0 -D0 , A0 -D0 , and C0 -D0 .
Polarization Properties of Prisms and Reflectors 91

Figure 3.34 TIR cube-corner prism orientation angle θ.27

The angle θ of the projected line B0 -D0 with the horizontal x axis orients the
cube-corner. A major diagonal of the cube connects the front corner E with the back
corner B.
For a BK7 solid-glass (n ≈ 1.52) cube-corner in air, the cube-corner is rotated
around the major diagonal E-B, forming an angle σ with the incoming light beam of
intensity S 0 , and is uniformly irradiated. The incoming linearly polarized light has
a polarization orientation angle Ψ. Figure 3.35 plots the calculated absolute values
of the spatially integrated nonnormalized Stokes parameters |S 1 00 |, |S 2 00 |, |S 3 00 |,
calculated from the six Stokes parameters of the incident and reflected light at each
of the three cube-corner surfaces. For details of the calculations, see Kalibjian.27
There are two triple null points for which |S 1 00 | = |S 2 00 | = |S 3 00 | ≈ 0. S 0 = 1.0. This
occurs, for example, at σ = 6.12 deg, θ = 30 deg or 150 deg, and Ψ = 11.6 deg
or 101.6 deg. The retroreflected light is therefore unpolarized, and the cube-corner
reflector functions as a depolarizer for linearly polarized input.

Figure 3.35 Stokes parameters S versus polarization orientation Ψ for TIR cube-corner
(σ = 6.117 deg; θ = 30 deg, 150 deg; n0 = 1.5113).27
92 Chapter 3

3.4 Plane of Polarization Rotation Using Total-Internal-Reflecting

Prisms and Reflectors
3.4.1 90-deg polarization-rotating prism with coaxial beam output
To rotate the plane of polarization of a linearly polarized input beam by 90 deg,
prisms are again constructed where the total internal reflections are restricted
to orthogonal planes. Figure 3.36 shows a prism with three orthogonal TIR
reflections, where the input linearly polarized light is oriented along the x axis,
such that E = E x .28 The internal beam remains linearly polarized and is rotated
during the internal reflections as shown. The output beam is displaced and parallel
to the input beam with the plane of polarization rotated 90 deg.

3.4.2 90-deg polarization-rotating prism with retroreflected beam output

In Fig. 3.37, a prism with three orthogonal TIR reflections produces a retroreflected
output beam with the polarization plane rotated 90 deg. The linearly polarized
input beam is oriented along the x axis.28 Figure 3.38 shows another type of
retroreflecting prism that rotates the plane of polarization 90 deg for a linearly
polarized beam. Here, the incident linearly polarized light can be oriented at
any angle in the x-y plane.29 The internal beam undergoes four orthogonal TIR
reflections, where two are s-polarized, and two are p-polarized. The net relative
phase shift is zero, the output is linearly polarized, and the prism is achromatic.

3.4.3 90-deg polarization-rotating prism with orthogonal beam output

Another prism system has six orthogonal TIR reflections for a linearly polarized
input beam (Fig. 3.39).29 Three internal reflections produce s-polarized phase
shifts, and three produce p-polarized phase shifts. The linearly polarized output
beam is perpendicular to the input beam, and the plane of polarization is rotated
90 deg relative to the input beam.

Figure 3.36 90-deg polarization-rotating TIR prism with linearly polarized input and coaxial
output. (Adapted from Ref. 28, with permission from the Optical Society of America.)
Polarization Properties of Prisms and Reflectors 93

Figure 3.37 90-deg polarization-rotating TIR prism with linearly polarized input and
retroreflected output. (Adapted from Ref. 28, with permission from the Optical Society of
America.)

Figure 3.38 90-deg polarization-rotating TIR prism with retroreflected output (α = 90 deg,
β = 45 deg).29

3.4.4 Double Fresnel rhomb polarization rotator with collinear beam

output
As shown in Sec. 3.2.1, when the double Fresnel rhomb is designed as a half-wave
retarder, the plane of polarization of a 45-deg-oriented linearly polarized input
beam is rotated 90 deg, and the output beam is collinear with the input beam. The
retardation of a BK7-glass double Fresnel rhomb varies continuously from about
91.0 deg (λ = 360 nm) to 89.6 deg (λ = 900 nm) and is considered achromatic in
retardance. (See Table 3.1 for some calculated retardance values.) Rotation of the
double Fresnel rhomb by angle ϕ about the optic axis rotates the linear polarization
of a light beam by 2ϕ.
94 Chapter 3

3.4.5 Four-mirror 90-deg polarization rotator with collinear beam output

Smith and Koch have analyzed the use of multiple first-surface metallic mirrors to
rotate the plane of polarization of a linearly polarized input beam and produce a
collinear output.30 Reflections are not restricted to orthogonal planes, but can be off
axis. By geometric phase analysis, they have shown that a minimum of four mirrors
is required to produce a 90-deg polarization rotation and produce collinear output.
Although there are a multitude of solutions, a sample configuration is shown in
Fig. 3.40.

Figure 3.39 90-deg polarization-rotating TIR prism with orthogonal output.29

Figure 3.40 Four-mirror 90-deg polarization rotator with nonorthogonal reflections and
collinear output.30
Polarization Properties of Prisms and Reflectors 95

References
1. E. E. Wahlstrom, Optical Crystallography, 3rd ed., 54–55, John Wiley, New
York (1948).
2. G. Brasen et al., “Polarizing beamsplitter,” U.S. Patent No. 7,230,763 (2007).
3. E. O. Ammann and G. A. Massey, “Less-expensive Rochon prisms,” NASA
Report Number MFS-20554, National Technology Transfer Center, Wheeling,
WV (1970).
4. S. M. MacNeille, “Beam splitter,” U.S. Patent No. 2,403,731 (1946).
5. J. Mouchart et al., “Modified MacNeille cube polarizer for a wide angular
field,” Appl. Opt. 28(10), 2847–2853 (1989).
6. A. E. Rosenbluth et al., “Contrast properties of liquid crystal light valves in
projection displays,” IBM J. Res. Develop. 42, 359–386 (1998).
7. J. M. Jonza et al., “Polarizing beam-splitting optical component,” U.S. Patent
No. 5,962,114 (1999).
8. C. L. Bruzzone et al., “Polarizing beam splitter,” U.S. Patent No. 6,721,096
(2004).
9. E. Gardner and D. Hansen, “An image quality wire-grid polarizing beam
splitter,” SID Symp. Dig. 34, 62–63 (2003).
10. R. T. Perkins et al., “Broadband wire grid polarizer for the visible spectrum,”
U.S. Patent No. 6,122,103 (2000).
11. T. Baur, “A new type of beam splitting polarizer cube,” Proc. SPIE 5158,
135–141 (2003) [doi:10.1117/12.510767].
12. L. Li and J. A. Dobrowolski, “High-performance thin-film polarizing beam
splitter operating at angles greater than the critical angle,” Appl. Opt. 39(16),
2754–2771 (2000). See also L. Li and J. A. Dobrowolski, “Thin film polarizing
device,” U.S. Patent No. 5,912,762 (1999).
13. J. J. Lee, “Polarizing prism for panel type liquid crystal display front projector
and optical prism using the polarizing prism,” U.S. Patent No. 5,717,472
(1998).
14. J. M. Bennett, “A critical evaluation of rhomb-type quarterwave retarders,”
Appl. Opt. 9(9), 2123–2129 (1974).
15. R. J. King, “Quarter-wave retardation systems based on the Fresnel rhomb
principle,” J. Sci. Instr. 43, 617–622 (1966).
16. M. Nicholson, “How to model corner cube retroflectors,” ZEMAX Application
Note, Zemax Development Corp, Bellevue, WA (2007).
17. M. A. Player, “Polarization properties of a cube-corner reflector,” J. Mod. Opt.
35(11), 1813–1820 (1988).
18. E. Spiller, “Totally reflecting thin-film phase retarders,” Appl. Opt. 33(20),
3544–3549 (1984).
96 Chapter 3

19. I. Fillinski and T. Skettrup, “Achromatic phase retarders constructed from

right-angle prisms: design,” Appl. Opt. 23(16), 2747–2751 (1984).
20. W. H. Steel, “Polarization-preserving retroreflectors,” Appl. Opt. 24(21),
3433–3434 (1992).
21. E. J. Galvez, “Achromatic polarization-preserving beam displacer,” Opt. Lett.
26, 971–973 (2001).
22. E. Cojocaru, “Polarization-preserving totally reflecting prisms,” Appl. Opt.
31(22), 4340–4342 (1992).
23. Z. P. Wang et al., “Polarization-preserving totally reflecting prisms with a
single medium layer,” Appl. Opt. 36, 2802–2807 (1997).
24. R. M. A. Azzam and M. Emdadur Rahman Khan, “Polarization-preserving
single-layer-coated beam displacers and axicons,” Appl. Opt. 21(18),
3314–3322 (1984).
25. L. H. Lee and J. J. Bockman, “Phase-compensated cube-corner in laser
interferometry,” U.S. Patent No. 7,165,850 (2007).
26. E. Hecht, “Polarization,” Chapter 8 in Optics, 2nd ed., 321–326, Addison-
Wesley, Reading, MA (1987).
27. R. Kalibjian, “Cube-corner depolarizer,” U.S. Patent No. 7,254,288 (2007).
28. N. I. Petrov, “Achromatic polarization rotator,” Appl. Opt. 45(25), 6340–6343
(2007).
29. W. A. Challener IV, “Achromatic polarization-rotating right-angle prism
system,” U.S. Patent No. 5,751,482, (1998).
30. L. L. Smith and P. M. Koch, “Use of four mirrors to rotate linear polarization
but preserve input–output collinearity,” J. Opt. Soc. Am. A 13, 2102–2105
(1996).
Chapter 4
Specialized Prism Types
4.1 Dispersing Prism
4.1.1 Refracting direct-vision prism
A direct-vision prism disperses the input light with no significant angular deviation
of the output. It is often called a direct-view spectroscope. The simplest type of
direct-vision prism is shown in Fig. 4.1, where the displaced and dispersed output
is coaxial with the input. This compound prism usually consists of crown- and
flint-glass prisms cemented at the interface surface. Let the crown glass have a
refractive index nd = 1.5159 and a ν-number = 70.0, and the flint glass have a
refractive index nd = 1.9626 and a ν-number = 19.7. If we choose the slope angle
α of the interface, then the vertex angle β of the first prism can be calculated from

(n2 − 1) sin α
tan β = q , (4.1)
n21 − n22 sin 2α − cos α

where n1 is the design refractive index of the crown glass, and n2 is the design
refractive index of the flint glass.1 To produce a dispersed output that is collinear
with the input, two of the above prisms are combined to form a double Amici prism
(Fig. 4.2). Using the same crown and flint glasses, if we choose α = 45.0 deg, then
β = 98.123 deg.

Figure 4.1 A direct-vision prism with coaxial output.

97
98 Chapter 4

Figure 4.2 The double Amici direct-vision prism with collinear output.

4.1.2 Reflective dispersing prisms with collinear output

There are several types of reflecting dispersing prisms of a single glass type that
produce collinear output (output beam on same axis as input beam). The Kessler
direct-vision prism is a classic design that uses two refractions and two total
internal reflections (Fig. 4.3). Figure 4.4 shows a Sherman-type prism that uses two
refractions and four reflections.2 Two reflections are from coated reflector surfaces,
and two undergo total internal reflection (TIR). This prism is compact and produces
a collinear output, but it requires a high-index glass to achieve the dispersion of a
60-deg equilateral refracting prism (see Table 2.1). Figure 4.5 illustrates another
compact dispersing prism that uses two refractions and two reflections from coated
surfaces.3

Figure 4.3 The Kessler direct-vision prism with collinear output.

Figure 4.4 The Sherman-type direct-vision prism with collinear output.

Specialized Prism Types 99

Figure 4.5 A direct-vision prism with collinear output using two refractions and two
reflections. (Adapted from Ref. 3 with permission from the Optical Society of America.)

4.1.3 Direct-vision prisms with wavelength tuning

All of the direct-vision prisms described produce collinear output at a single
wavelength. It is possible to arrange a pair of the prism types in Fig. 4.5 to vary
the wavelength of the collinear output. In Fig. 4.6, a pair of two reflection prisms
are coupled to rotate by the same angle in opposite directions about the indicated
ray positions.4 The ray path between the prisms is coaxial to the input and output
beams but is not collinear to these beams. Another type of rotational coupling for
a pair of identical dispersing prisms produces collinear output wavelength tuning
(Fig. 4.7).5 Each prism has two refractions and two internal reflections, with a
dispersed coaxial output (output beam parallel to input beam). The rotation axis
positions are as shown, and the rotations are in opposite directions.

4.1.4 Total-internal-reflecting dispersing prism

Where collinear or coaxial output is not a requirement, a single TIR prism can be
designed to disperse light, or to be nondispersive. The TIR prism shown in Fig. 4.8

Figure 4.6 A coupled dispersing prism pair with wavelength tuning and collinear output.
(Adapted from Ref. 4 with permission from Elsevier.)
100 Chapter 4

Figure 4.7 A coupled dispersing TIR prism pair with wavelength tuning and collinear
output.5

Figure 4.8 A single TIR dispersing prism.

has planar entrance and exit refracting surfaces with an intermediate TIR surface.
The sequential ray-tracing equations are as follows:

I1 = θ1 − α, (4.2a)
!
sin I1
I1 0 = arcsin , (4.2b)
n
I2 = 180 deg − I1 0 − α − β, (4.2c)
I2 0 = I2 , (4.2d)
I3 = β − I2 0 , (4.2e)
I3 0 = arcsin(n sin I3 ), (4.2f)
δ = θ1 + θ3 , (4.2g)

where θ1 is the incident ray angle, α is the refracting groove angle, β is the
reflecting groove angle, θ3 is the exit ray angle, and δ is the deviation angle. I2
must exceed the critical angle at the reflecting surface. For example, if θ1 = 60 deg,
α = 35 deg, β = 80 deg, and nd = 1.5168 (BK7 glass), then I2 = 48.82 deg,
I1 = 25 deg, θ3 = 51.74 deg, and δ = 111.74 deg.
Specialized Prism Types 101

The TIR prism can be designed to have normal positive dispersion, negative
dispersion, or no dispersion (e.g., achromatic). The conditions for each are as
follows:
• If (2β + α) > 180 deg, the prism has positive dispersion.
• If (2β + α) < 180 deg the prism has negative dispersion.
• If (2β + α) = 180 deg the prism is nondispersive.
Thus, a glass or plastic prism has positive dispersion if α = 40 deg and β = 75 deg,
negative dispersion if α = 20 deg and β = 75 deg, and no dispersion if α = 30 deg
and β = 75 deg. These relationships are valid for any normal refractive index and
angle of incidence I1 , provided that TIR occurs at surface 2. This prism type could
be used to compensate for other dispersions in an optical system.

4.1.5 Multiprism negative dispersion

Consider the pair of identical prisms oriented as shown in Fig. 4.9, where the
incident face of prism 1 is parallel to the exit surface of prism 2, and the exit
surface of prism 1 is parallel to the entrance surface of prism 2. The prisms are
configured such that all rays are incident at the Brewster’s angle, and all operate
at minimum deviation.6 The color components of the exiting rays are spatially
and temporally separated and are coaxial to the incident ray and to each other. This
two-prism system can generate negative group-velocity dispersion. By the addition
of another identical prism pair with controlled orientation and separations, as in
Fig. 4.10, an in-line (collinear) output is obtained. The output of this four-prism
arrangement also has negative group velocity dispersion. The amount of dispersion
can be adjusted, and the theory is developed by Fork et al.6

4.2 Refracting Achromatic Compound Prism

A well-known refracting achromatic prism is constructed from two bonded prisms
having different refractive indices and dispersions (Fig. 4.11). The first prism is
usually of low-dispersion glass, and the second prism is of high-dispersion glass.

Figure 4.9 A prism pair with coaxial output and negative dispersion.6
102 Chapter 4

Figure 4.10 A four-prism system with collinear output and negative dispersion.6

Figure 4.11 An achromatic prism combination of high- and low-dispersion glass.

The initial values of the vertex angles for each section, α1 and α2 , are calculated
from the paraxial approximation for a two-element achromatic prism:7

δν1
α1 = , (4.3)
(n1 − 1)(ν1 − ν2 )
δν2
α2 = , (4.4)
(n2 − 1)(ν2 − ν1 )

where δ is the deviation angle.

Specialized Prism Types 103

The sequential ray-tracing equations for an angle of incidence I1 are

!
sin I1
I1 = arcsin
0
, (4.5a)
n1
I2 = α1 − I1 0 , (4.5b)
" #
sin I2
I2 = arcsin
0
, (4.5c)
(n1 /n2 )
I3 = I2 0 − α2 , (4.5d)
I3 0 = arcsin(n2 sin I3 ), (4.5e)
δ = I1 + I3 0 − α1 + α2 , (4.5f)

where α1 and α2 are entered as positive numbers and are related by the following:

n2 0 + sin I2 0 − sin(δ − I1 + α1 )
tan α2 = , (4.6)
n2 0 + cos I2 0 − cos(δ − I1 + α1 )

where n2 0 is the refractive index value that produces the target δ value. For visible
light, the deviation angles are calculated from Eqs. (4.3a) to (4.3g) for three
wavelengths, nC (λ = 656.3 nm), nd (λ = 587.3 nm), and nF (λ = 486.1 nm).
A merit function MF(δ) can be defined as
q
MF(δ) = (δ − δc )2 + (δ − δd )2 + (δ − δ f )2 . (4.7)

If the glasses have been selected, the prism angles α1 and α2 could then be
varied to reduce MF(δ) to an acceptable value over the visible spectrum. Other
variables to consider are I1 in Eq. (4.1) and n2 0 in Eq. (4.6), where n2C ≤ n2 0 ≤
n2F . Design methods, including glass selection, for several compound achromatic
prism systems are described by Mercado.8 Figure 4.12(a) shows a two-element
achromatic prism with a deviation angle δ = 6 deg. Here, α1 = 42.0239 deg,
n1d = 1.52855, ν1 = 76.98, and α2 = 25.5872 deg; n2d = 1.65160 and ν2 = 58.40.
Figure 4.12(b) plots the change in the deviation angle over the visible spectrum.

4.3 Anamorphic Prisms for Beam Compression and Expansion

Anamorphic prism systems change the height or width of an incident beam along
one dimension by a specified amount. For example, an elliptically shaped beam
can be converted to a circular beam, or the aspect ratio of a beam can be modified.
In fact, most refracting prisms produce some magnification change between input
and output beams. The problem is to specify its magnitude and direction for the
intended application. The most useful output directions are collinear, coaxial, and
orthogonal (output beam perpendicular to input beam). These anamorphic prisms
can be of the dispersing type, or in some cases, produce an achromatic output.
104 Chapter 4

Figure 4.12 (a) An achromatic compound prism with a deviation angle δ = 6 deg.8 (b)
Deviation angle variance of an achromatic compound prism over the visible spectrum.8

4.3.1 Beam expander with orthogonal output

Figure 4.13 shows an anamorphic beam expander that produces an orthogonal
output and is based on the a Littrow-type prism.9 Vertex angle β is cut at Brewster’s
angle θBrew , and the angle of incidence I1 at surface 1 is also θBrew for high
transmission of p-polarized light. For n = 1.5 and θBrew = 56.3 deg, TIR occurs at
surface 2, and the dispersed output is orthogonal. The magnification (expansion) of
the beam could be calculated from the ray intersection coordinates using Eq. (1.58)
and Eq. (1.59). Alternatively, using the angles of incidence I and refraction I 0
at each surface of the prism, the total anamorphic prism magnification MAG
(or demagnification for beam compression) can be calculated from the following
Specialized Prism Types 105

Figure 4.13 An anamorphic beam expander with orthogonal output.9

product:
k
Y cos Ii 0
MAG = , (4.8)
i=1
cos Ii

where i is the refracting surface number and k is the number of refracting surfaces.
For this example, the magnification is

cos I1 0 cos I3 0 cos(33.69 deg) cos(0 deg)

MAG = × = × = 1.5,
cos I1 cos I3 cos(56.31 deg) cos(0 deg)

which is the same as the prism design refractive index n. The dispersed output is
only exactly orthogonal at the design wavelength, with a slight deviation for the
dispersed rays.

4.3.2 Beam compressor with coaxial output

Figure 4.14 illustrates a beam-compressing prism with a coaxial output that has
two refractions and two total internal reflections.10 The design parameters are the
vertex angle α, the prism refractive index n, and the beam-compression factor
MAG = A0 /A. For coaxial output, the design refractive index and vertex angle
are related by

cos α
n= . (4.9)
cos 3α

The sequential ray-tracing equations are as follows:

I1 0 = I1 = 0, (4.10a)
106 Chapter 4

Figure 4.14 A beam-compressing prism with coaxial output.10

I2 0 = I2 = 90 deg − α, (4.10b)
I3 0 = I3 = I2 0 − α, (4.10c)
I4 = I3 0 − α, (4.10d)
I4 0 = arcsin(n sin I4 ). (4.10e)

Using Eq. (4.7), if α = 17 deg, then n ≈ 1.52 and I4 0 = 90 deg − α. Then, the beam
compression is

cos I1 0 cos I4 0 1.0 0.292

MAG = × = × = 0.376.
cos I1 cos I4 1.0 0.777

4.3.3 Beam expander with collinear output

A single prism with two refractions and one total internal reflection (Fig. 4.15)
can produce an expanded collinear output with an anamorphic magnification of
2×.11 The intended use is to change the elliptically shaped collimated beam from
a laser diode to a circular beam. The prism material is BK7 glass with a published
refractive index n = 1.51119 at λ = 780 nm. The vertex angles are α = 27.912 deg
and β = 41.388 deg. Let the entrance surface 1 be tilted at an angle ϕ with respect

Figure 4.15 An anamorphic beam expander with collinear output at 2× magnification.11

Specialized Prism Types 107

to the horizontal optic axis. The sequential ray-trace equations are

I1 = 90 deg − ϕ, (4.11a)
!
sin I1
I1 0 = arcsin , (4.11b)
n
I2 = I1 0 + α = I2 0 , (4.11c)
I3 = I2 0 − β, (4.11d)
I3 0 = arcsin(n sin I3 ), (4.11e)
δ = I3 0 − 90 deg + α + β − ϕ. (4.11f)

From Eq. (4.8), the magnification MAG is

cos(38.65 deg) cos(40.0 deg)

MAG = × = 2.0.
cos(70.9 deg) cos(25.17 deg)

Since the deviation angle δ ≈ 0 deg, the output is coaxial. The prism length is then
adjusted so that the exit ray is on the same axis as the central entrance ray, resulting
in a collinear output.

4.3.4 Wedge prism beam compressor/expander

A wedge prism usually has a vertex angle α between 2 and 20 deg and is often used
for beam steering (see Sec. 6.1.2). If one surface of the prism is reflectorized, a thin
anamorphic beam expander can be created.12 The output beam is orthogonal to the
input beam. It is realized that any prism designed for anamorphic compression
can be used as an expander by reversing the direction of the input and vice versa.
Figure 4.16 shows a reflecting wedge prism anamorphic compressor that converts
an input beam with an aspect ratio of 2:1 to an output beam with an aspect ratio
of 4:3.13 The ray-trace equations are

I1 = ϕ, (4.12a)
!
sin I1
I1 0 = arcsin , (4.12b)
n
I2 = α + I1 0 = I2 0 , (4.12c)
I3 = I2 + I2 0 + I1 − I1 0 − ϕ, (4.12d)
I3 0 = arcsin(n sin I3 ), (4.12e)
δ = ϕ + I3 0 . (4.12f)

Here, α  (I3 − I1 0 )/2 and ϕ is the tilt angle of surface 1 from the vertical. To
account for the varying thickness of the prism, ϕ and α are adjusted until the
desired compression ratio is obtained. For a prism of B270 optical crown glass
(nd = 1.5229) with ϕ = 16.9 deg and α = 14.0 deg, an anamorphic compression
A0 /A = MAG ≈ 0.375 can be obtained.
108 Chapter 4

Figure 4.16 An anamorphic reflecting wedge prism compressor.13

4.3.5 Anamorphic prism pair with coaxial output

A useful method to produce a variable magnification is by using a pair of identical
air-spaced prisms “a” and “b”. By varying the tilt angles ϕa and ϕb , the anamorphic
magnification can vary from about 2× to 6×. These prisms are sold commercially
as separate components or as a mounted assembly by suppliers such as Melles
Griot or Edmund Optics. They are usually antireflection coated. Figure 4.17 shows
a pair of prisms with a vertex angle α = 29.43 deg and SF11 glass (n ≈ 1.765).

Figure 4.17 An air-spaced anamorphic prism pair beam expander.

Specialized Prism Types 109

The ray-trace equations are

I1 = α + ϕa , (4.13a)
!
sin I1
I1 0 = arcsin , (4.13b)
n
I2 = I1 − I1 0 − ϕa , (4.13c)
I2 0 = arcsin(n sin I2 ), (4.13d)
δa = I2 0 + ϕa , (4.13e)
I3 = δa + α − ϕb , (4.13f)
!
sin I3
I3 = arcsin
0
, (4.13g)
n
I4 = α − I3 0 , (4.13h)
I4 0 = arcsin(n sin I4 ), (4.13i)
δab = ϕb − I4 0 , (4.13j)

where ϕa and ϕb are positive for clockwise rotation and chosen such that I3 = I1
and δab ≈ 0 for coaxial output. The anamorphic magnification of the pair is
calculated from Eq. (4.8). The vertical separation between the input and output
beams is determined by the spacing between the prisms. Table 4.1 shows sample
data for magnifications of 2×, 3×, and 4×.
When operating at 3× magnification, the incident angle at each prism is closest
to the Brewster angle for SF11 glass, where IBrew = 60.47 deg for n = 1.765.

4.3.6 Multiprism dispersive compressors and expanders

The anamorphic prism pair beam expander of Sec. 4.3.5 is of the compensating pair
type, sometimes called the up-down configuration. The dispersion is reduced, but
not eliminated, when compared with the additive pair configuration in Fig. 4.18,
sometimes called the up-up configuration. Duarte and Piper have calculated the
cumulative dispersion for a series of anamorphic beam prism pairs in both
configurations.14 For use as beam expanders in pulsed dye lasers, the angle of
incidence at each surface is set equal, and the exit angles are set close to zero. The
prisms are identical and designed for BK7 glass (n ≈ 1.515), with a vertex angle
α = 41.5 deg and a glass dispersion dn/dλ = 0.71×10−4 nm at the laser wavelength
λ = 510 nm. Duarte and Piper calculated and compared the cumulative dispersions

Table 4.1 Sample data for an anamorphic prism pair, α = 29.43 deg, and SF11 glass.
ϕa 21.2 deg 30.7 deg 35.2 deg
ϕb 6.1 deg 0.0 deg −2.4 deg
I1 50.63 deg 60.13 deg 64.63 deg
I3 50.63 deg 60.13 deg 64.63 deg
δab 0.00117 deg 0.00260 deg 0.00273 deg
MAG 2.0 3.0 4.0
110 Chapter 4

Figure 4.18 A two-prism up-up configuration.

for single-, double-, and four-prism beam expanders at magnifications of 5.34× and
60×. In particular, the four-prism compensating pair arrangement at 5.35×, with the
angles of incidence close to Brewster’s angle, yielded a 4.8 factor reduction in the
cumulative dispersion compared to the additive pair arrangement.14
By control of the prism vertex angles and prism angles of incidence on
a separated prism pair in the up-down configuration, the positively directed
dispersion of the first prism can be compensated by the negatively directed
dispersion of the second prism, resulting in a near-achromatic prism pair. Trebino
has analyzed arrangements for multiple-prism beam expanders, considering
the impact on total transmission and magnification.15 A common four-prism
achromatic down-up-up-down configuration is shown in Fig. 4.19, consisting
of two achromatic pairs, with an achievable magnification MAG ≈ 40, and a
transmission > 50%.

Figure 4.19 A four-prism down-up-up-down configuration with MAG ≈ 40. (Adapted from
Ref. 15 with permission from the Optical Society of America.)
Specialized Prism Types 111

It was found that the total transmission for a given magnification can be
optimized by the up-up-up-down configuration of Fig. 4.20, where the negatively
directed dispersion of the fourth prism compensates for the additive positively
directed dispersion of the first three prisms.
It was determined that for an achromatic N-prism beam expander, when
MAG ≈> [2−1/(2N−1 −1)]N , that the transmission at this magnification is optimized
for an up-up . . . up-down configuration.
The number of prisms is not restricted to an even number. A three-prism up-up-
down configuration that is achromatic with optimal performance at MAG ≈ 20 is
shown in Fig. 4.21.

4.4 Achromatic Anamorphic Prism

The achromatic anamorphic multiprisms described in Sec. 4.3.6 are restrictive to
choice of prism vertex angles, angles of incidence, and the steering direction of the
beam output. However, there are several types of anamorphic prisms that can be
made achromatic by other methods.

4.4.1 Air-spaced prism pair with coaxial output

By making an air-spaced anamorphic prism pair of a high-dispersive and a low-
dispersive glass, this type of beam expander/compressor can be made achromatic.

Figure 4.20 A four-prism up-up-up-down configuration with MAG ≥ 10. (Adapted from
Ref. 15 with permission from the Optical Society of America.)
112 Chapter 4

Figure 4.21 A three-prism up-up-down configuration with MAG ≥ 5. (Adapted from Ref. 15
with permission from the Optical Society of America.)

The design principles are similar to the refractive achromatic compound prisms
described in Sec. 4.2. In Fig. 4.22(a), an achromatic anamorphic prism pair beam
compressor provides a coaxial beam output over a 20-nm range of the design
wavelength.16 It converts the elliptically shaped output of a diode laser light
source to a circularly shaped output beam, with a magnification MAG = 0.333,
to compensate for small variations in the laser diode wavelength. The pair is
composed of a first prism of KF9 glass (nd = 1.52346, νd = 51.54) and a second
prism of SF11 glass (nd = 1.78472, νd = 25.68). All prism surfaces are nonnormal
to the incident and exit beams to prevent any reflection returning to the light source.
The relevant design values at a laser-diode design wavelength λ = 800 nm are given
here:

αa = 35.927 deg
αb = 27.384 deg
ϕa = I1 = 1.0 deg
ϕb = −29.217 deg
I2 = 36.587 deg
I3 = 1.5 deg
I4 0 = 56.601 deg
δa = I2 0 − αa − ϕa = 27.717 deg
δab = 0 deg
MAG = 0.333.

Figure 4.22(b) plots the change in the deviation angle δab in the wavelength
range 500 nm ≤ λ ≤ 1000 nm.
Specialized Prism Types 113

Figure 4.22 (a) An air-spaced achromatic anamorphic prism pair compressor.16 (b)
Deviation angle versus wavelength for an achromatic prism pair.16

4.4.2 Compound prisms with orthogonal output

Another type of compound achromatic anamorphic beam expander shown in
Fig. 4.23(a) has an orthogonal output.17 The first glass is low-dispersion BK7
crown glass (n = 1.5112) and is optically bonded to the second high-dispersion
glass SF11 (n = 1.7660). Both indices are referenced to light at λ = 780 nm. The
reflecting surface requires a reflective coating, and the prism has an anamorphic
magnification MAG ≈ 1.75. Another achromatic compound beam expander shown
in Fig. 4.23(b) uses TIR at the reflecting surface and also provides coaxial output
with a magnification MAG ≈ 1.75.17

4.4.3 Refracting/total-internal-reflecting prism pair with orthogonal

output
By combining a refracting prism having positive dispersion with a catadioptric
prism or TIR prism (a prism using both refraction and reflection) having negative
dispersion (Sec. 4.1.4), an achromatic beam compressor with orthogonal output
can be constructed.13 Figure 4.24 shows the configuration where the material for
both prisms “a” and “b” is BK7 glass (nd = 1.5168, νd = 64.17). The sequential
114 Chapter 4

Figure 4.23 (a) A compound achromatic anamorphic beam expander with orthogonal
output.17 (b) A compound achromatic anamorphic beam expander with orthogonal output
using TIR.17

Figure 4.24 An air-spaced achromatic anamorphic prism pair beam compressor using a
common material.13
Specialized Prism Types 115

ray-tracing equations are

I1 = ϕ1 , (4.14a)
!
sin I1
I1 = arcsin
0
, (4.14b)
n
I2 = α1 + I1 0 , (4.14c)
I2 0 = arcsin(n sin I2 ), (4.14d)
δa = I2 0 − ϕ1 − α1 , (4.14e)
I3 = δa + 90 deg − α2 + ϕ2 , (4.14f)
!
sin I3
I3 = arcsin
0
, (4.14g)
n
I4 = 180 deg − α3 − I3 0 − α2 = I4 0 , (4.14h)
I5 = I4 0 − α3 , (4.14i)
I5 0 = arcsin(n sin I5 ), (4.14j)
δab = 90 deg − I5 0 + ϕ2 . (4.14k)

Some design values at the design wavelength λ = 587.6 nm are listed here:

α1 = 18.6 deg
α2 = 68.0 deg
α3 = 38.2 deg
ϕ1 = 30.0 deg = I1
ϕ2 = 8.02 deg
I2 = 37.85 deg
I3 = 49.96 deg
I4 = 43.48 deg
I5 = 5.28 deg
I5 0 = 8.02 deg
δa = 19.93 deg
δab = 90.0016 deg

and

δa (λ = 486.1 nm) − δa (λ = 656.3 nm) = 0.430 deg,

δab (λ = 486.1 nm) − δab (λ = 656.3 nm) = −0.000568 deg .

The anamorphic magnification MAG = 0.375 is calculated from Eq. (4.8). The
prism pair can be used to convert a 2:1 aspect ratio input beam to a 4:3 aspect ratio
output beam.
116 Chapter 4

4.5 A Misalignment-Tolerant Beam-Splitting Prism

Figure 4.25(a) illustrates a beam-splitting prism comprised of two bonded sections
with a beam-splitting coating on the interface.18 The external surfaces are
accurately parallel to the opposing faces, and the reflecting surfaces utilize TIR
or are coated with a high-reflectance film. The corner angles are α = 45 deg and
β = 135 deg, while the beam-splitting coating reflects 50% and transmits 50%.
When a light beam is incident normal to the input aperture, the beam is split and
the output is displaced and parallel to the input beam, or the output is coaxial. The
displacement is determined by the length of the prism for the application.
In Fig. 4.25(b), the prism is rotated in a clockwise direction by angle ϕ.
This could be uncontrolled misalignment or an intentional tilt to prevent light
being reflected back along the original path. For a rotated prism (clockwise or
counterclockwise), the output beams remain coaxial to the input beam. This
rotation-tolerant prism is optically equivalent to a tilted plate (Sec. 1.10). In
contrast, for a cube beam-splitting prism, the output beams do not remain
orthogonal if the prism is rotated, although both prism types can be used in
convergent or divergent beams.

4.6 Axicon Prism

An axicon prism has a conical-shaped surface. It is sometimes called an axicon lens
or rotationally symmetric prism. The most common configuration is a refracting
plano-convex axicon, as shown in Fig. 4.26. With collimated light incident on the
plano side, a linear focus is formed on the optic axis and a ring focus is formed
on an image plane. The angle θ, measured from the axicon surface to a plane
perpendicular to the optic axis, describes the surface. The prism can be modeled
using the single design parameter θ.19 Axicon prisms are available commercially
in BK7 glass and quartz.
A type of reflecting plano-concave axicon prism shown in Fig. 4.27 uses a TIR
conical surface.20 Collimated light entering the plano surface undergoes TIR at the
conical surface (θ = 45 deg) and forms an illuminated concentric ring 360 deg
around the sides of the axicon prism.

4.7 A Variable Phase-Shifting Prism

The phase of a coherent beam of light can be varied by means of a shifting prism.21
As shown in Fig. 4.28, an isosceles prism with vertex angle α and refractive index
n1 has light incident at an angle of minimum deviation Imin . The surround has a
refractive index n0 (normally air). For the prism in position 1, the optical path
distance (OPD) between the points A and B is OPD1 = n0 L + n1 l0 + n0 L. When the
prism is shifted by a distance Y to position 2, the OPD between the same points
is OPD2 = n0 l + n1 l0 + n0 l, where Y, L, l0 , and l are defined as shown. Then the
difference in the optical path distance ∆OPD12 is

∆OPD12 = 2(n1 l − n0 L). (4.15a)

Specialized Prism Types 117

Figure 4.25 (a) A beam-splitting compound prism with coaxial output.18 (b) A tilted beam-
splitting compound prism with undeviated coaxial output.18
118 Chapter 4

Figure 4.26 A refracting plano-convex axicon prism.

Figure 4.27 A reflecting plano-concave axicon prism.20

Figure 4.28 A prismatic phase shifter.21

Specialized Prism Types 119

When the prism is operated at a minimum deviation angle of incidence Imin , then
∆OPD12 is derived by Childers21 to be

2Y sin(α/2)   α 
∆OPD12 = n0 − n1 cos Imin − . (4.15b)
cos Imin 2

The phase of the exit beam is then precisely controlled by varying Y, without any
translation or deviation of the light beam.

References
1. J. P. C. Southall, Mirrors, Prisms and Lenses, 3rd ed., 493–499, Macmillan,
New York (1946).
2. B. Sherman, “Dispersion prism with no deviation,” U.S. Patent No. 3,057,248
(1962).
3. M. V. R. K. Murty and A. L. Narasimhan, “Some new direct vision dispersion
prism systems,” Appl. Opt. 9(4), 859–862 (1970).
4. M. V. R. K. Murty, “In-line dispersion prisms and methods of tuning different
wavelengths,” Opt. Laser Technol. 16, 255–257 (1984).
5. R. D. Tewari et al., “Modified in-line dispersion prism,” Opt. Eng. 31(6),
1340–1341 (1992) [doi:10.1117/12.57696].
6. R. L. Fork et al., “Negative dispersion using prism pairs,” Opt. Lett. 9(5),
150–152 (1984).
7. W. J. Smith, Modern Optical Engineering, 2nd ed., 90–91, McGraw-Hill, New
York (1990).
8. R. L. Mercado, “Color-corrected prism systems,” U.S. Patent No. 4,704,008
(1987).
9. A. B. Marchant, “Method and apparatus for anamorphically shaping and
deflecting electromagnetic waves,” U.S. Patent No. 4,759,616 (1988).
10. S. D. Fantone, “Optical system with anamorphic compression,” U.S. Patent
No. 4,627,690 (1986).
11. J. F. Forkner, “Anamorphic prism for beam shaping,” U.S. Patent
No. 4,623,225 (1986).
12. K. Yoshifusa and T. Yokota, “Beam converting apparatus with a parallel light
beam input and output from one prism plane,” U.S. Patent No. 5,007,713
(1991).
13. D. F. Vanderwerf, “Polarized illumination system for LCD projector,” U.S.
Patent No. 5,995,284 (1999).
14. F. J. Duarte and J. A. Piper, “Dispersion theory of multiple-prism beam
expander for pulsed dye lasers,” Opt. Commun. 43, 303–307 (1982).
120 Chapter 4

15. R. Trebino, “Achromatic N-prism beam expanders: optimal configurations,”

Appl. Opt. 24(8), 1130–1138 (1985).
16. F. C. Leuke, “Achromatic anamorphic prism pair,” U.S. Patent No. 5,596,456
(1997).
17. M. Sugiki, “Anamorphic prism,” U.S. Patent No. 4,750,819 (1988).
18. C. F. Buhrer, “Optical beam splitter prism,” U.S. Patent No. 4,671,613 (1987).
19. “Modeling axicons,” ZEMAX Application Note, Zemax Development Corp.,
Bellevue, WA (1999).
20. A. R. Henderson, “Prisms,” UK Patent Application No. 2,001,775A (1978).
21. B. A. Childers, “Prismatic phase shifter,” NASA Tech Brief LAR-14637,
NASA Langley Research Center, Hampton, VA (1999).
Chapter 5
Prism and Mirror System Design,
Analysis, and Fabrication
5.1 Prism Design and Analysis
More optical tools are available for the analysis of prism and mirror systems than
for the design of new systems. Some questions to be considered when approaching
the design of a prism for an optical system are:
• Is the prism to be used in an imaging optical system, or as an illumination control
element, or both?
• For imaging applications, what is the required orientation of the viewed or
projected image? Does the image shape need to be preserved—e.g., no linear
magnification?
• Can an existing design, preferably one available from a commercial supplier, be
used?
• Can the design be fabricated economically?
• Is the required prism a single or compound element?
• Are compound prisms air-spaced or bonded?
• Are surface coatings required—e.g., antireflection or beam-splitting?
• Does the prism need to be achromatic?
• Will the prism be used in convergent or divergent light?
• What aperture size is required?
• What is the wavelength range of interest?
• Are there size and weight limitations?
• What are the mounting tolerance requirements?
• What are the prism material clarity requirements, angular accuracy of faces, and
surface smoothness?
• What is the thermal environment?
There is an extensive library of prism designs in the literature that may provide
the requirement of your optical system. Many times, these designs can be modified
by a change of glass material or the addition of special coatings. Often two or more
existing prisms can be combined in series.

121
122 Chapter 5

5.1.1 Sectional element approach for prism design

Often, sections of simpler prisms can be combined to form another prism type
with new properties. Rothstein has described the combination of single isosceles
TIR prisms.1 In Fig. 5.1, a single isosceles prism has vertex angle α and base angles
β. It is tilted at angle ϕ and usually truncated to eliminate the nonworking lower
portion. The directions of the three working surfaces are specified by the angles
ψ with a horizontal optical axis, where ψ1 = 90 deg − ϕ, ψ2 = α/2 + ϕ, and
ψ3 = 90 deg − α − ϕ. For a prism with refractive index n with an air surround, the
sequential ray-tracing equations are

I1 = ϕ, (5.1a)
!
sin I1
I1 0 = arcsin , (5.1b)
n
I2 = 90 deg − ψ2 + I1 − I1 0 = I2 0 , (5.1c)
I3 = ψ2 + ψ3 − I2 0 , (5.1d)
I3 0 = arcsin(n sin I3 ). (5.1e)

The deviation angle δ = 90 deg − ψ3 + I3 0 . Table 5.1 gives several calculated

values for a BK7 glass prism in air (nd = 1.5168). Although there is no anamorphic
magnification, the prism is dispersing, and the deviation angle δ depends on the
value of ϕ.
Figure 5.2 shows another TIR deviator composed of two identical isosceles
prisms folded along a side to form a single prism with two TIR reflections. The
prism is tilted at angle ϕ. The directions of the four working surfaces are specified

Figure 5.1 A single isosceles prism with vertex angle α and base angles β, tilted at angle ϕ.
Prism and Mirror System Design, Analysis, and Fabrication 123

Table 5.1 Sample data for isosceles TIR prism, nd = 1.5168.

α (deg) β (deg) ϕ (deg) I2 (deg) δ (deg)
20.0 80.0 15.0 70.18 50.0
20.0 80.0 5.0 76.71 30.0
20.0 80.0 0.0 80.0 20.0
20.0 80.0 −10.0 86.57 0.0
30.0 75.0 15.0 65.18 60.0
30.0 75.0 5.0 71.71 40.0
30.0 75.0 0.0 75.0 30.0
30.0 75.0 −10.0 81.57 10.0
45.0 67.50 15.0 57.68 75.0
45.0 67.50 5.0 64.21 55.0
45.0 67.50 0.0 67.50 45.0
45.0 67.50 −10.0 74.07 25.0

Figure 5.2 A double TIR prism consisting of two isosceles prisms. (Adapted from Ref. 1
with permission from the Optical Society of America.)

by the angles ψ with the optic axis, where ψ1 = 90 deg − ϕ, ψ2 = 90 deg − β + ϕ,

ψ3 = 270 deg − 3β + ϕ, and ψ4 = β − ψ3 . The sequential ray-tracing equations are:

I1 = ϕ, (5.2a)
!
sin I1
I1 0 = arcsin , (5.2b)
n
I2 = 90 deg − ψ2 + I1 − I1 0 = I2 0 , (5.2c)
I3 = 180 deg + ψ2 − ψ3 − I2 0 = I3 0 , (5.2d)
I4 = ψ3 + ψ4 − I3 0 , (5.2e)
I4 0 = arcsin(n sin I4 ). (5.2f)
124 Chapter 5

The deviation angle δ = 90 deg − ψ4 + I4 0 . Several values of the parameters are

given in Table 5.2 for a BK7 glass prism.
The deviation angle is constant for any tilt angle, provided that TIR occurs at
both reflecting surfaces. The magnification remains constant at MAG = 1.0, and
the prism is achromatic. Reflecting isosceles prisms are used as sections of the
Pechan prism [see Fig. 2.20(a)].

5.1.2 Right-angle prism sections

Several well-known single-prism types contain basic right-angle prism sections
or modified right-angle prisms. For example, the wavelength-discriminating
Pellin–Broca prism in Sec. 2.3 is a fusion of three right-angle prisms, two
30/90/60-deg sections, and one 45/90/45-deg section (see Fig. 5.3). The reflecting
Littrow prisms in Sec. 2.9 use a single right-angle prism. It is noted that a reflecting
isosceles prism and a right-angle prism are sections of the Penta prism of Fig. 2.6.

5.1.3 Experiential design of multiple reflectors

Prism or multiple-mirror design by pure trial and error implies little previous
knowledge of the principles used and neglects the experience of the designer.

Table 5.2 Sample data for double isosceles TIR prism, nd = 1.5168.
β (deg) ϕ (deg) I2 (deg) I3 (deg) δ (deg)
67.50 15.0 57.68 77.32 90.0
67.50 5.0 64.21 70.79 90.0
67.50 0.0 67.50 67.50 90.0
67.50 −10.0 74.07 60.93 90.0
75.0 15.0 65.18 84.82 60.0
75.0 5.0 71.71 78.29 60.0
75.0 0.0 75.0 75.0 60.0
75.0 −10.0 81.57 68.43 60.0
78.750 15.0 68.93 88.57 45.0
78.750 5.0 75.46 82.04 45.0
78.750 0.0 78.75 78.75 45.0
78.750 −10.0 85.32 72.18 45.0

Figure 5.3 Pellin-Broca prism composed of three right-angle sections.

Prism and Mirror System Design, Analysis, and Fabrication 125

Experiential knowledge often leads to choices that produce a better and quicker
solution. Smith presents a methodology for designing mirror systems and
certain prism types that uses an iterative procedure combined with experiential
knowledge.2 Smith describes the layout of a four-mirror system that projects a
correctly oriented image to a rear-view screen positioned orthogonal to the object.
A solution is obtained using a minimal number of mirrors, and direction changes
of each reflected ray are restricted to the same plane or to an orthogonal plane (see
Fig. 5.4).

5.1.4 Matrix methods for design and analysis

Prism elements and mirrors of known geometry can be inserted into commercial
optical design programs. As components in multiple-element optical systems,
they can be positioned, sized, toleranced, and corrected for certain aberrations.
However, for the determination and control of prism image orientation, these ray-
tracing programs are not as useful. To address this, several specialized techniques
have been developed for the design and analysis of prisms and reflectors.
The design and analysis of mirror systems using fourth-order reflection matrices
has been described by Pegis and Rao.3 Procedural examples are given for
the design of a reversion prism with no deviation, a Wollaston prism, and a
Penta prism. A matrix technique for determining the general orientation of a
viewed image through a series of k planar reflectors was developed by Walles
and Hopkins.4 Here, a coordinate transformation matrix directly produces the
orientation if the reflection matrices are multiplied in the same order as the light
ray reflections occur.
A more general analysis of a series of planar reflective and refractive surfaces
has been described by Liao and Lin.5 The analysis considers skew rays, and
by the use of fourth-order homogeneous transformation matrices, the location

Figure 5.4 A four-mirror system with reflections in orthogonal planes.2

126 Chapter 5

and orientation of a local coordinate system is calculated for each surface. The
technique is applied to a Pechan prism and a glass cube-corner prism. Tsai and Lin
have addressed the actual design of a prism using a minimum number of reflecting
and refracting surfaces by defining a merit function Γ based on the required change
in image orientation.6 It is shown that the use of the following merit function:

−1 0 0
 
Γ =  0 −1 0 , (5.3)
0 0 −1

which specifies an image inversion along three axes, independent of prism

alignment, leads directly to a solid-glass cube-corner reflector prism.

5.1.5 Evolutionary prism design using a genetic algorithm

A prototype methodology has been developed that uses a genetic algorithm (GA)
for the initial conceptual design of a prism having undefined geometry. Bentley
and Wakefield have investigated GAs in the evolutionary design of optical prisms
without any prior knowledge of a specific prism type.7 Basically, a population
of possible solutions is evaluated according to a fitness criterion, and the fittest
offspring are reproduced to form an improved population. Designs that deviate
from this criterion are penalized. The binary-coded parameter to be optimized is a
genotype, and the decoded design modified by the GA is a phenotype.
In evolutionary prism design, the process often begins with a purely random
initial geometry. The light path through the prism is not specified. The direction
and size of the input light and the position and direction of the desired output
are specified (see Fig. 5.5). To simplify ray tracing, the light is assumed
monochromatic, and surface reflections (other than TIR or coated surfaces) are
neglected. The GA begins an iterative process of evaluation and produces new
populations of improved designs. Design fitness is evaluated by deviations of the
output light from the target direction and intersection at a specified plane. A basic
requirement for any evolved design is the setting of limitations on prism size.

Figure 5.5 Input and output target-ray vector directions for initial prism geometry.7
Prism and Mirror System Design, Analysis, and Fabrication 127

Designs that fall outside the set limits in size are penalized with poor fitness but are
not completely eliminated. The evolved prisms are composed of primitive sections,
and a design where a section becomes detached from the group is called fragmented
and is given a high penalty.
Certain types of prism designs present difficulties in reaching an optimized
solution due to the presence of easily reached local minima, called deceptive
attractors. For the design of a rhomboid prism, Fig. 5.6 shows three failed attempts,
and Fig. 5.7 shows two deceptive attractors for this problem. Evolutionary design
has also been applied to right-angle, roof, and rotating Dove prisms, and to a
Penta prism.7 In another approach, fixed right-angle evolved sections were further
evolved to produce nearly perfect Abbe and Porro prisms with four internal
reflections.
For the application of GAs in the design of a Fresnel lens illuminator, see
Sec. 9.14.

5.1.6 A three-mirror tabletop lectern projector

For many applications, the use of a multiple-reflecting glass prism is preferable to
the use of a series of mounted planar mirrors, mainly because of the stability of
the integral prism structure. However, for systems with larger optical beams and
sizes, the use of mirrors is the only choice. The lectern projector in Fig. 5.8 has
a horizontal glass stage on which a 285 × 285-mm overhead-type transparency is
projected to an integral rear dual-focal-length projection lens.8 The screen has a

Figure 5.6 Failed attempts at rhomboid prism design using a genetic algorithm. (Adapted
from Ref. 7 with permission from Wiley.)

Figure 5.7 Deceptive attractors for evolutionary design of rhomboid prism. (Adapted from
Ref. 7 with permission from Wiley.)
128 Chapter 5

Figure 5.8 A tabletop lectern rear-screen projector.8

size of about 450 × 450 mm, and the entire lectern is approximately 1.5 m wide by
1 m high and deep. A dual-focal-length projection lens projects at magnifications
of 1.6× or 2.4×. The ray path needs to be displaced and rotated by 90 deg. Since
the information on the transparency is projected from the reverted back side, three
folding mirrors are required, taking into account the 180-deg image rotation by
the projection lens. Figure 5.9 illustrates the layout of the projection lens L1 and
the mirrors M1 , M2 , and M3 . The mirrors can be sized by experimental or analytic
methods and are trapezoidal in shape.

5.1.7 Prism aberrations

Although prisms are composed of flat surfaces, they exhibit many of the
aberrations that are usually associated with lenses, but the aberrations are related
to plane symmetrical systems. Sasián has classified and calculated significant
aberrations for a refracting prism with a plane of symmetry and a vertex angle α.9
Figures 5.10(a) and 5.10(b) show the basic ray paths. Some of these aberrations
are:
• Constant lateral chromatic (uniform over field of view)
• Constant astigmatism (uniform over field of view)
• Image anamorphism (varies over aperture)
• Constant coma (uniform over field of view)
• Linear astigmatism (varies linearly over field of view)
• Field tilt (tilts the image plane)
• Quadratic distortions (keystone and curvature of imaged lines).
Prism and Mirror System Design, Analysis, and Fabrication 129

Figure 5.9 Optical path of rear-projection tabletop lectern.8

Figure 5.10 (a) Side view of isosceles refracting prism showing optical ray axis. (b) Top
view of isosceles refracting prism showing chief and marginal rays. (Adapted from Ref. 9
with permission from the Optical Society of America.)
130 Chapter 5

These aberrations can be eliminated for certain specific ray paths through the
prism. The mathematical description of these aberrations and their elimination is
described in Ref. 9.

5.2 Prism Quality Specifications

Commercial suppliers often classify prisms by grades such as student,

demonstration, standard, high-tolerance, precision, and calibration quality. It is
necessary to examine the printed specifications of the supplier to determine the
applicability for its intended use.

5.2.1 Surface quality and flatness specifications

Surface quality is usually specified by the scratch and dig standard. For a scratch
and dig of 60/40, the largest observable scratch width does not exceed 60
hundredths of a millimeter, and the maximum measured diameter of a dig, pit, or
bubble does not exceed 40 hundredths of a millimeter. Surface flatness (sometimes
called figure) is specified by the maximum wavelength change over the entire
surface at λ = 632.8 nm (HeNe laser) using a standard test plate where the space
between the test plate and the surface changes by λ/2 for every counted fringe. For
example, normal surface quality and flatness for a “standard” right-angle prism
could be 60/40 scratch and dig, and λ/2 for BK7 glass. A “precision” or “high-
tolerance” right-angle prism might have 20/10 scratch and dig surface quality and
λ/4 surface flatness for BK7 glass or λ/10 for fused silica.

5.2.2 Optical material properties

Some of the optical requirements for prism glass are spectral transmission, clarity,
color, bubbles, occlusions, stress, and striae. Stress in glass produces birefringence,
and stress birefringence is usually expressed in nm/cm for a given wavelength of
light. Stress can be significantly reduced by the process of annealing, where the
glass is heated to an annealing point and then slowly cooled to the strain point
temperature, and then dropped to room temperature at a determined rate. Two
benefits of annealing glass are (1) removal of internal strains that produce stress
birefringence effects, and (2) normalizing the glass so that the refractive index is
uniform throughout the material.10 The refractive index variation for glass can be
reduced to about ±1 × 10−4 to ±2 × 10−5 by fine annealing. Striae are local abrupt
changes in refractive index. To designate striae, glasses are often classified as A, B,
C, or D grade.11 Prisms that transmit high-energy beams or beam-splitting prisms
must contain few small bubbles or inclusions. Examples of glass descriptions
would be “BK7 Grade A fine annealed glass,” or “UV-grade synthetic fused silica.”
Some of the glass physical properties that can affect optical performance and
fabrication are hardness (Knoop test), grindability, viscosity, coefficient of linear
expansion, thermal conductivity, and chemical resistance.
Prism and Mirror System Design, Analysis, and Fabrication 131

5.2.3 Specifying angular accuracies

The required accuracy of the face angles for prisms depends on the type of prism
and may vary between faces of the prism. For an Amici roof prism, angle tolerance
between the refracting faces may be 30 arcsec, but the TIR roof angle tolerance
must be held to 90 deg ± 5 arcsec. The Penta prism is sometimes supplied with a
tighter tolerance < 10 arcsec for the 90-deg angle and a looser ±3-arcmin tolerance
for the other angles.

5.2.4 Tolerancing a Dove prism

The Dove prism is used as an optical component in several interferometric
applications. In particular, for a rotational shearing interferometer, a high-precision
rotating Dove prism is used to shear or rotate the wavefront. Since many
commercially available Dove prisms do not have the required accuracy for
this interferometer, Herrera and Strojnik have analyzed Dove prism tolerance
requirements for interferometric use.12 The prism geometric parameters are height
H, width W, length L, and base angle β. The glass is BK7 at λ = 633 nm, and for
an effective square aperture of 19.75 × 19.75 mm, the effect of glass homogeneity
was considered to be negligible. Also, polarization changes for a Dove prism were
considered to be negligible. Surface flatness over the 60% surface area analyzed
was assumed to be λ/10. For this analysis, tolerance errors were considered for
prism length (∆L), base angles (∆β1 , ∆β2 ), and pyramidal angles (θ1 , θ2 ) of the
entrance and exit faces [see Figs. 5.11(a) and (b)]. Base values H = W = 25.4 mm
and L = 107.5 mm were chosen to duplicate dimensions of a commercially
available prism.
Collimated light from a square object grid was traced through the prism, while
the vertices defining the edges were varied to simulate manufacturing errors. The
image of this object grid displayed the effect of these errors, where errors in β
and θ, but not L, generate a critical optical path difference (OPD). Maximum
OPD occurs when ∆β1 = −∆β2 or θ1 = θ2 . It was determined that to achieve
wavefront deviations < λ/10, the base angle tolerance must be ±0.37 arcsec, and
the pyramidal angle tolerance must be held at ±0.52 arcsec. A commercially
available Dove prism with an angular tolerance of ±2 arcmin would produce an
OPD about twice as large as the requirement for this interferometer.

5.2.5 Techniques for prism angle measurement

Once a prism is fabricated with target angle tolerances, the prism angles should
be measured to verify the required accuracy. Commercial instruments such as the
PrismMaster goniometer using an electronic autocollimator can measure prism
angles in reflection or transmission from 0 to 360 deg with a resolution up
to 0.036 arcsec.13 Various other techniques are available using angle gauges,
autocollimators, and interferometers.
Rao has developed alternative noninterferometric prism angle-measurement
techniques for specific prisms.14,15 A method for measuring the error α in the 90-
deg angle of a 45/90/45-deg prism is illustrated in Fig. 5.12. The diagonal face of
132 Chapter 5

Figure 5.11 (a) Dove prism showing base-angle errors ∆β1 and ∆β2 and length error
∆L. (b) Dove prism showing pyramidal angle error (θ1 + θ2 ). (Adapted from Ref. 12 with
permission from Elsevier.)

Figure 5.12 Technique for the measurement of error in the 90-deg angle of a right-angle
prism.14

the prism is first placed on a well-cleaned optical flat. An optical plate with its faces
parallel to less than 0.5 arcsec is temporarily attached in an approximate vertical
position to the polished face of the test prism. Collimated light from a HeNe laser
reflects from one of the parallel faces of the plate and is reflected to a vertical
rear screen, where its position A is recorded with a micrometer eyepiece. The test
Prism and Mirror System Design, Analysis, and Fabrication 133

prism is then rotated 180 deg, and the position B of the reflected spot on the screen
is measured. The error α in the right-angle is then calculated from

AB
α= , (5.4)
4OC

where AB is the distance between the spots, C is the midpoint position, and OC is
the distance from the plate to the screen.
Figure 5.13 shows a procedure for measurement of the error β of the 45-deg
angles. The diagonal face now rests on three steel balls, and a horizontal laser
beam is incident on the test prism face. The reflected spot position A is measured
on a horizontal rear screen. The prism is then rotated 180 deg, and the reflected
spot position B is measured. The error C is then calculated from

AB
β= , (5.5)
4OC

where a positive error in β for one of the 45-deg angles indicates an equal negative
error in the other 45-deg angle. For OC = 5 m, and a measurement accuracy of
0.01 mm for AB, the calculated accuracy of the prism angle measurements is about
1 arcsec.
Interferometric techniques for prism angle measurements can provide an
accuracy of up to 0.1 arcsec. Nunez and Sanchez have described interferometric
techniques for measuring prism vertex angles and pyramidal error.16 An aligned
Twyman–Green interferometer and test prism [Fig. 5.14(a)] produce fringe patterns

Figure 5.13 Technique for the measurement of error in the 45-deg angles of a right-angle
prism.14
134 Chapter 5

that indicate the absence of pyramidal error (A, B, C) and the presence of pyramidal
error (D, E, F), as shown in Fig. 5.14(b).
The pyramidal error ε p can be calculated from the following:

λ
εp = , (5.6)
4p(n − 1)

where

λ = wavelength of illumination,
p = fringe period,
n = prism refractive index.

Figure 5.14 (a) Interferometric technique for measurement of pyramidal error in a prism.
(b) Interferometric fringe patterns indicating absence and presence of pyramidal error.16
Prism and Mirror System Design, Analysis, and Fabrication 135

The percentage error dε p of the measured pyramidal angle is the derivative of

Eq. (5.6):
" # " # " #
dλ λd p λdn
dε p = − − . (5.7)
4p(n − 1) 4(n − 1)p2 4p(n − 1)2

Using stated values of n = 1.61996, dn = 0.001, λ = 632.99 nm, dλ = 2.6 × 10−6 ,

p = 2.98 mm, and d p = 0.01 mm, the percentage error in the pyramidal angle is
0.52%.15

5.3 Survey of Fabrication Methods

5.3.1 Ground and polished glass prism
The most accurate surface figure for flat surfaces is achieved by grinding with
subsequent block polishing. By this method, a surface flatness up to λ/20 can be
achieved for precision applications and interferometric use.

5.3.2 Fabrication of a Penta prism by measurement of the angular

deviation error
Most prisms are fabricated and then checked for the accuracy of the prism angles.
Proper application of the Penta prism requires a precise angular deviation δ =
90 deg. Chatterjee and Kumar have developed a technique for monitoring the
accuracy of the deviation angle of a Penta prism during the fabrication process
so that surface adjustments can be made.17 Figure 5.15 illustrates the ray path of
a Penta prism with a collimated laser beam (HeNe at 633 nm) incident normal to
the entrance surface. The reversed ray path of a Fresnel surface-reflection ray from
the exit surface is also shown, exaggerated for clarity. The angles adjacent to the
right-angle corner angle are given by 112 deg+α1 and 112 deg+α2 , the right-angle
corner angle is given by 90 deg − γ, and the angle between the reflecting faces is
given by 45 deg + β, where the angle errors α1 , α2 , γ, and β can have plus or minus
values. It is shown that

δ = 90 deg + (n − 1)γ − 2nβ, (5.8)

where n is the refractive index of the prism. The error ε in the deviation angle is

ε = δ − 90 deg = (n − 1)γ − 2nβ, (5.9)

and if n = 1.5, thenε = 0.5γ − 3β. This shows that for this index, an error in β has
six times the effect as an angle in γ. Using Snell’s law and assuming small angles,
the angle of refraction I1back 0 of the back-reflected ray at the entrance surface is

I1back 0 = nI1back = n(2γ − 4β), (5.10)

136 Chapter 5

Figure 5.15 Technique of monitoring the accuracy of pentaprism angles during the
fabrication process. (Adapted from Ref. 17 with permission from the Optical Society of
America.)

and
I1back 0
ε= − γ. (5.11)
2
From Eq. (5.10), I1back 0 is twice as sensitive to an error in β than an error in γ
for any refractive index, and using Eq. (5.11), the deviation angle error ε can be
calculated from measured values of I1back 0 and γ. During fabrication, I1back 0 and
γ are measured, and the resultant deviation ε is calculated from Eq. (5.11). The
general procedure is as follows:
1. From the ground prism blank, the entrance face AB and the exit face AE are
block polished to λ/10 surface flatness.
2. Face DC is set at noncritical angles 112.5 deg ± 1.0 arcmin. Although DC is a
nonworking surface of the Penta prism, it is polished for measurement viewing
of the prism right angle.
3. The right angle is finished to an angular accuracy of a few fractions of an arc
second by standard techniques. The sign of the angular error γ is determined
using interference fringe analysis of the split wavefront reflected from the right
angle (90 deg + γ) through face DC using a HeNe laser, a Fizeau interferometer,
and an adjustable reference plate R. This right angle becomes a reference angle
for the correction of the error in the 90-deg deviation angle δ.
4. Face BC is block polished to good figure and reflectorized. Using a Fizeau
interferometer at the face AB, the magnitude and sign of I1back 0 is measured
from the two-beam interference pattern formed by the back reflections of light
from the entrance face AB and the exit face AE as the face ED is polished.
Prism and Mirror System Design, Analysis, and Fabrication 137

5. The angular deviation angle error ε is calculated. If acceptable, face ED is

reflectorized. If not, the prism can be reworked and remeasured.
Measured values for a 25-mm-wide × 45.0-mm-long Penta prism were reported
at γ ≈ 0.25 arcsec and I1back 0 ≈ 2.5 arcsec, yielding a deviation angle error
ε ≈ ±1.25 + 0.25 arcsec.17

5.3.3 Molded, pressed, and fire-polished prisms

Molded and pressed-glass optical prisms are being produced that provide high-
angle accuracies and surface flatness. Special precision molding glasses (Schott
P-series) are available having a low transformation temperature.18 Fire polishing
of glass can melt the surface of certain pressed or molded prisms enough to produce
a smoother surface. Typical fire-polishing temperatures range from about 700 ◦ C
to 760 ◦ C, and the surface melt time is from 5 to 20 min.
Many mass-produced plastic prisms of smaller size are produced by injection
molding. To suppress flow marks on the optical surfaces during filling of the cavity,
the gate is usually positioned along an edge of the formed prism. Injection-molded
plastic Amici roof prisms are used as a component in range-finder cameras.19

5.3.4 Fabrication of large prisms

A very large isosceles prism has been fabricated for the High Efficiency and
Resolution Canterbury University Large Echelle Spectrograph (HERCULES).20
It was prepared from a 23-kg prism blank of Schott Grade A fine annealed BK7
glass. The vertex angle was 49.5 deg, with a height of 276 mm, a base of 268 mm,
and a length of 255 mm. A Twyman–Green interferometer was used to test glass
homogeneity and wavelength distortion. Surface figure was checked using a 200-
mm-diameter Zerodur reference flat.

5.4 Some prism-mounting methods

Mounting methods for optical components are generally classified as one of three
types: kinematic mount, semikinematic mount, or nonkinematic mount. Mounting
methods for prisms differ from that of mirrors and lenses, since prisms often have
irregular shape, may have inputs and outputs that are not coaxial, or may have
reflecting faces (TIR and coated) that require protection.
A mounted prism has six possible degrees of freedom: three translational along
perpendicular axes, and three rotational about these axes. When the prism is held
in position at six points, each point uniquely assigned to one degree of freedom,
the prism is independently constrained at each point and is a kinematic mount. The
contact points should be outside the clear aperture of the prism. The prism can
then be held in position by pressure at a single point. If the prism mount is defined
at more than six support points, it is overconstrained and is nonkinematic. A true
kinematic mount is superior in repositioning a prism that has been removed for
some purpose such as cleaning or recoating but may induce stress birefringence
due to excessive force at the support points.
138 Chapter 5

The semikinematic mount spreads each of the six contact points over a small
area. The area should be a raised pad lapped coplanar with respect to the prism
surface to minimize any point contact.21,22 This is the preferable type of mount
for most prism elements, since there is negligible stress induced in the prism by
contact force. The prism can be secured by a clamp at a single point.
Another type of nonkinematic mount consists of mounting the prism directly
to a planar substrate by thin adhesive layers, where the mounting face is not used
optically. This type of mount is resistant to shock and vibration, but the element is
not easily removed or repositioned.
Positional mountings are used when the prism needs to be accurately rotated
or repositioned, especially in laboratory applications. Various types of gimbaled
three-axis prism mounts are available commercially. The design of these mounts
can be mechanically complex. A rotary beamsplitter prism mount has been
designed with three axes of rotation, where each axis operates independently of
the other two and does not affect their positioning.23 A two-axis angular adjust
motorized mount for a right-angle prism having high sensitivity and stability has
been described for use in an autoboresight. The angular range is ±0.75 mrad with
a resolution of 10 µrad.24

References
1. J. Rothstein, “Isosceles total internal reflectors as optical elements,” Appl. Opt.
2(11), 1191–1194 (1963).
2. W. J. Smith, Modern Optical Engineering, 2nd ed., 113–115, McGraw-Hill,
New York (1990).
3. E. J. Pegis and M. M. Rao, “Analysis and design of plane-mirror systems,”
Appl. Opt. 2(12), 1271–1274 (1963).
4. S. Walles and R. E. Hopkins, “The orientation of the image formed by a series
of plane mirrors,” Appl. Opt. 3(12), 1447–1452 (1964).
5. T.-T. Liao and P. D. Lin, “Analysis of optical elements with flat boundary
surfaces,” Appl. Opt. 42(7), 1191–1202 (2003).
6. C.-Y. Tsai and P. D. Lin, “Prism design based on changes in image
orientation,” Appl. Opt. 45(17), 3951–3959 (2006).
7. P. J. Bentley and J. P. Wakefield, “Conceptual evolutionary design by a genetic
algorithm,” Eng. Design Automation 2(3), 119–131 (1997).
8. D. F. Vanderwerf, “Dual-magnification rear-projection lectern,” U.S. Patent
No. 4,561,740 (1985).
9. J. M. Sasián, “Aberrations from a prism and a grating,” Appl. Opt. 39(1), 34–39
(2000).
10. F. Twyman, Prism and Lens Making: A Textbook for Optical Glassworkers,
505, CRC Press, Boca Raton, FL (1988).
Prism and Mirror System Design, Analysis, and Fabrication 139

11. “Glass, Optical,” MIL-G-174B, Defense Supply Agency, Washington, DC

(1988).
12. E. G. Herrera and M. Strojnik, “Interferometric tolerance determination for a
Dove prism using exact ray trace,” Opt. Commun. 281, 897–905 (2008).
13. Trioptics GmbH, Wedel, Germany.
14. S. M. Rao, “Method for measurement of the angles of 90-, 45-, 45-
deg and 60-, 30-, 90- deg prisms,” Opt. Eng. 36(1), 197–200 (1997)
[doi:10.1117/1.601159].
15. S. M. Rao, “Methods for making prism with submultiple of half angles:
applications to the measurement of the angles of Pechan and Pellin–Broca
prisms,” Opt. Eng. 41(11), 2945–2950 (2002) [doi:10.1117/1.1512660].
16. A. Jaramillo-Nunez and C. Robledo-Sanchez, “Measuring the angles and
pyramidal error of high-precision prisms,” Opt. Eng. 36(10), 2868–2871
(1997) [doi:10.1117/1.601516].
17. S. Chatterjee and Y. P. Kumar, “Simple technique for the fabrication of a
penta prism with high-accuracy right-angle deviation,” Appl. Opt. 46(26),
6520–6525 (2007).
18. Schott Glass Catalog, available online at www.schott.com.
19. K. Tanaka and I. Kasai, “Roof prism,” U.S. Patent No. 5,946,147 (1999).
20. J.B. Hearnshaw et al., “HERCULES: a high-resolution spectrograph for small
to medium-sized telescopes,” in IAU 8th Asia-Pacific Regional Meeting, 289,
11–15 (2003).
21. P. R. Yoder Jr., “Attributes of the successful optic-to-mount interface,”
Chapter 2 in Design and Mounting of Prisms and Small Mirrors in Optical
Instruments, SPIE Press, Bellingham, WA (1998).
22. P. R. Yoder Jr., “Optomechanical design in five easy lessons,” SPIE’s
oemagazine, 29–32 (February 2004).
23. D. F. Arnone and F. S. Lueke, “Rotary beamsplitter prism mount,” U.S. Patent
No. 5,694,257 (1997).
24. E. J. Stolfi, “Motorized- axis-angular fine adjustment prism mount,” U.S.
Patent No. 4,722,592 (1988).
Chapter 6
A Selection of Prism
Applications
6.1 Laser Scanning
6.1.1 Reflective scanning prism
A rotating reflective prism in the shape of a polygon cylinder, or spinner, is a well-
known scanning technique. Figure 6.1 shows a rotating polygon where a fixed-
direction single laser beam partially illuminates a face. The reflected beam scans
continuously until a facet corner is encountered, where there is a discontinuity or
retrace interval between the end of the facet scan and the succeeding scan from
the next facet. The pyramidal faceted scanner in Fig. 6.2 reduces or eliminates the
retrace interval.1 A convergent beam illuminates at least two facets to produce
a continuous-scan interval between these facets having a circular focal trace
with constant angular velocity. Thus, two illuminated facets will produce one
continuous-scan interval, three illuminated facets will produce two continuous-
scan intervals, and illumination of an entire pyramid having n facets will produce
n − 1 continuous-scan intervals.

6.1.2 Refractive prism-beam scanning and steering

6.1.2.1 Single-wedge prism
A single-wedge prism with vertex angle α and deviation angle δ, when rotated
about an axis normal to one of the surfaces, produces a circular scan on a screen

Figure 6.1 A rotating reflective polygonal scanner.

141
142 Chapter 6

Figure 6.2 A rotating reflective pyramidal faceted scanner. (Adapted from Ref. 1 with
permission from Elsevier.)

Figure 6.3 Wedge prism oriented with input face normal to incident ray (“A” orientation).

or generates a cone with half-angle δ. The exact deviation angle depends on the
vertex angle, prism refractive index, and the direction of the prism with respect to
the incident beam. When the wedge prism is in “A” orientation as in Fig. 6.3, the
deviation angle δ can be calculated from

I1 = I1 0 = 0, (6.1a)
I2 = α, (6.1b)
δA = I2 0 − α = arcsin(n sin α) − α. (6.1c)
A Selection of Prism Applications 143

When the wedge prism is in “B” orientation as in Fig. 6.4, the deviation angle is
calculated from

I1 = α, (6.2a)
!
sin α
I1 = arcsin
0
, (6.2b)
n
I2 = I1 − I1 0 , (6.2c)
δB = I2 0 = arcsin(n sin I2 ), (6.2d)

For analyzing wedge prisms as scanners, the small angle or paraxial approximation
is often used. Then, Eq. (6.1c) reduces to δ = nα − α = (n − 1)α, and Eq. (6.2d)
reduces to δ = nI2 = n(α − α/n) = (n − 1)α. Thus, the paraxial values δparax are the
same for both prism orientations. Table 6.1 lists the exact and paraxial δ values for
a BK7 glass (nd = 1.5168) wedge prism using some sample vertex angles α.
6.1.2.2 Wedge prism pairs
By placing two wedge prisms with vertex angles α1 and α2 in series with a
collimated laser beam incident on prism 1, the deviated ray from prism 1 is incident
on prism 2. By rotating each independently about the optical beam axis, the final
beam deviation from prism 2 can be controlled (see Fig. 6.5). When the first prism
rotates at an angular velocity ±ω1 and the second prism rotates at an angular

Figure 6.4 Wedge prism oriented with input face inclined to incident ray (“B” orientation).

Table 6.1 Exact and paraxial ray deviations for wedge prisms at sample vertex angles.
α (deg) δA (deg) δB (deg) δparax (deg)
2 1.0344 1.0339 1.0336
6 3.123 3.108 3.101
10 5.271 5.201 5.168
14 7.527 7.326 7.235
18 9.951 9.497 9.302
144 Chapter 6

Figure 6.5 Operation of the double-wedge Risley prism.

velocity ±ω2 , a variety of scan patterns are produced. The prism pair is known
as a Risley prism.
Marshall has generated a series of Risley prism scan patterns using the following
procedure:2 A refractive index n = 1.50, and the paraxial form of the ray deviation
is used for each prism, such that δ1 = α1 /2 and δ2 = α2 /2, where δ1 and δ2
are vector quantities. Then the vector addition δ = δ1 + δ2 represents the total
deviation. The x and y components of δ are

δ x = δ1 cos ω1 t + δ2 cos(ω2 t − ϕ), (6.3a)

δy = δ1 sin ω1 t + δ2 sin(ω2 t − ϕ), (6.3b)

where ϕ is the relative orientation (e.g., phase angle) between the two prisms. Scan
patterns were then generated for specified values of the ratios (ω2 /ω1 ), (α2 /α1 ),
and ϕ, and plotting δ x against δy . Risley prisms can produce many types of scan
patterns, including those with loops and cusps. In terms of application, probably
the most useful scan patterns are circles, lines, ellipses, and spirals for scanning an
area. Figures 6.6(a) to 6.6(d) illustrate several scan patterns.
The Risley prism as described earlier has chromatic dispersion when used in
broadband light. In many applications, especially in the infrared, it is necessary
to minimize the angular dispersion over a wide spectral range. One method is to
use achromatic compound-wedge prisms (see Sec. 4.2). LiF/ZnS compound-wedge
Risley-type prisms have been designed to have minimum dispersion over the 2-
to 5-µm IR region, with a maximum steering angle δmax = 45 deg.3 There are
two possible orientations for these achromatic prism pairs. The “A” orientation
has nonparallel faces between prism 1 and prism 2, while the “B” orientation
has parallel faces between the two prisms. Both orientations produce a deviation
angle δmax . If a deviation angle is desired, prism 2 can be rotated 180 deg about
the reference axis. However, the rotated “A” orientation of Fig. 6.7(a) produces
a small angular deviation over the spectral range, or an on-axis blind spot, and
exact steering to δ = 0 deg is not possible. This is unacceptable for broadband
A Selection of Prism Applications 145

Figure 6.6 Several types of Risley scan patterns.2

IR countermeasure steering devices. The rotated “B” orientation of Fig. 6.7(b)

provides exact steering to δ = 0 deg, since both the interior faces and the incident
and exit faces are parallel, and this is the preferred orientation.3
6.1.2.3 LADAR guidance system using prism pairs
By the use of two-wedge prism pairs, where each prism is independently rotated
in one pair, and the orientation can be changed for either pair, a laser detection
and ranging system (LADAR) can be constructed.4 In Fig. 6.8(a), high-speed
(≈10,000 rpm) rotating prisms 1 and 2 are oriented as shown, and fixed-position
prisms 3 and 4 are oriented as shown. A collimated laser beam is incident on
prism 1. When the rotation velocity ω2 of prism 2 is slightly higher than the rotation
velocity ω1 of prism 1, a spiral scanning pattern is produced. An alternative prism
arrangement [Fig. 6.8(b)] has prism 2 oriented 180 deg with respect to prism 1,
while prisms 3 and 4 remain fixed in position. Another spiral scanning pattern of
smaller diameter is produced for ω2 > ω1 . In Fig. 6.8(c), the orientation of prism 4
can be changed to 180 deg by a rapid acceleration motor, producing a shift in the
center of the spiral scan pattern. Thus, both the size and position of the scanning
pattern can be changed.
When a missile is launched at a target aircraft, an initial large scan identifies the
target. The scan pattern is then shifted to sweep and track the aircraft, and the scan
pattern is reduced in size for more precise tracking.
146 Chapter 6

Figure 6.7 (a) Achromatic Risley prism “A” configuration with prism 1 rotated 180 deg,
showing an on-axis blind spot (δ > 0 deg). (b) Achromatic Risley prism “B” configuration
with prism 1 rotated 180 deg and δ = 0 deg.

6.1.2.4 Rotating square-plate linear scanner

A simple rotating refracting square plate of refractive index n can function as a
linear scanner. A varying displacement s is produced by the varying angle of
incidence I of a small-diameter collimated laser beam as the plate edges rotate
around its center. The required plate thickness in the direction of the rotation axis is
determined by mechanical stability during rotation. In addition, the plate thickness
must accommodate the diameter of the incident laser beam. In Fig. 6.9(a), a square
plate ABCD having sides of length d is rotated about its center. In this position, a
fixed laser beam passes through the center of AB and the axis of rotation. When
the plate rotates in a clockwise direction, the exit beam is displaced by a distance
s, where
 
cos I
s = d sin I 1 − p  .
 
(6.4)
n2 − sin2 I

For the position shown in Fig. 6.9(b), the scan direction is reversed when
the incident beam passes through corner B from face AB to face BC, and the
transmitted beam moves from one end of the scan line to the other. The cycle is
repeated as each face and corner passes through the incident beam. The maximum
A Selection of Prism Applications 147

(a)

(b)

(c)

Figure 6.8 (a) Scan pattern for rotating and fixed wedge-prism pairs. (b) Smaller-diameter
scan pattern for reoriented rotating prism and fixed-wedge-prism pairs. (c) Shifted scan
pattern for rotating and reoriented fixed-wedge-prism pairs.4

value of s occurs when I = 45 deg and ensures that the beam always exits the
opposite face. The scan is repeated as the next corner C intersects the incident
beam, as shown in Fig. 6.9(c). Then, the scan length SL = 2s and is given by
 
 
 1 − cos(45 deg) 
SL = 2d sin 45 deg  q  . (6.5)
 2 2 
n − sin (45 deg)
148 Chapter 6

Figure 6.9 (a) Reference position of a rotating square-plate scanner. (b) Rotating square-
plate scanner at bottom of scan, I = +45 deg. (c) Rotating square-plate scanner at top of
scan, I = −45 deg.
A Selection of Prism Applications 149

6.2 Interferometry and Spectroscopy

6.2.1 Laser interferometer with prism polarization rotator
A uniaxial birefringent prism of specified geometry can be used as a primary
component in a double-pass laser interferometer system. It is used in conjunction
with a phase conjugate mirror (PCM) to produce an output beam with 90-deg
polarization rotation.5 Figure 6.10 shows the basic optical arrangement and the
two prisms used. Prism 1 is made from calcite with the principal axis oriented as
shown, with angles α = 38.5 deg, β = 83.5 deg, and γ = 90 deg. Prism 2 is a
right-angle retroreflecting prism.
Linearly polarized light oriented at 45 deg enters face AD of prism 1 normally,
where it is separated into an o-ray and an e-ray. The calcite refractive index for
the o-ray is no = 1.6428 and ne = 1.4799 for the e-ray. The o-ray undergoes TIR
at point a, and the e-ray is refracted toward prism 2. The internally reflected ray at
face AB undergoes total internal retroreflection to point b on face AB and exits face
AD to the PCM, collinear with the entrance ray. The retroreflected ray from prism 2
is directed to the same point b, recombines with internally reflected rays, and exits
face AD to the PCM. The reflected ray from the PCM passes again through the
system, and the recombined rays at point a are reflected from the PBS as output.
The distance between prism 1 and prism 2 is adjusted in the double pass such that
the difference in path lengths between the o-rays and the e-rays produces a 90-deg
polarization rotation in the output.

6.2.2 Polarization interferometer using a Wollaston prism

In laser interferometry, it is often necessary to resolve ambiguities in fringe
counting by providing two outputs, one horizontally polarized and the other
vertically polarized, each output 90 deg out-of-phase with the other (phase
quadrature). Figure 6.11 illustrates a polarization interferometer that produces
two such interference pattern outputs.6 Linearly polarized light at 45 deg reflected
from a beam-splitting mirror and right-angle prism 1 is converted to a circularly
polarized beam by a quarter-wave 90-deg retarder. This beam combines with the
reference beam reflected from right-angle prism 2 at the beam-splitting mirror.
The resultant interference pattern has circularly and linearly polarized components.

Figure 6.10 A laser interferometer producing a 90-deg rotation of input polarized light.5
150 Chapter 6

Figure 6.11 Polarization interferometer using a Wollaston prism. (Adapted from Ref. 6 with
permission from Elsevier.)

A Wollaston prism separates the interference patterns of the vertical and horizontal
components, producing two interference patterns, one horizontally polarized and
the other vertically polarized, both of which are in phase quadrature.

6.2.3 Multipass optical cell for laser interferometer

A multipass optical cell increases the optical path length of a laser beam in a
compact physical space. It is often used as a laser interferometer component
to increase the difference between interfering beams. The cube-corner reflector
(CCR) can be used to advantage for this purpose. Figure 6.12(a) shows a pair of
offset TIR CCRs that produce a series of five double passes of an input laser beam
entering in the plane of the axes.7 One CCR has a flat reflectorized segment at
the vertex. The output beam exits collinear to the input beam, and the number of
reflections is limited by overlapping of beams at the flat reflectorized segment.
By the use of an additional TIR right-angle prism and two CCRs with offset
axes, the number of passes can be increased considerably.8 Here, the input beam
is not in the plane of the CCR axes. When the reflected beam leaves the CCR,
the right-angle prism returns the beam to the CCR for another series of passes.
A Selection of Prism Applications 151

A perspective view of an arrangement where there are 16 single passes of the light
beam is shown in Fig. 6.12(b). The output beam exits coaxial to the input beam.
The number of passes will eventually be limited, even with the use of additional
prisms, when the laser beam diameters completely fill the CCR apertures.

6.2.4 Nomarski polarized-light interferometer

A Normarski polarized interferometer is useful for obtaining interferograms of
fast light pulses with a short coherence length. A configuration for a Nomarski
polarized interferometer (Fig. 6.13) consists of two identical Wollaston prisms.9
Linearly polarized light at a 45-deg angle enters the first Wollaston prism and is
separated (or sheared) into o-rays and e-rays at the diagonal boundary. Lens L1

Figure 6.12 (a) A pair of cube-corner reflectors providing five passes. (b) A pair of cube-
corner reflectors and a right-angle prism providing 16 passes (only input, output, and right-
angle prism rays are shown). (Adapted from Ref. 8 with permission from the Optical Society
of America.)
152 Chapter 6

Figure 6.13 A Nomarski-polarized interferometer using two Wollaston prisms.9

transmits collimated light through the sample S, and it is converged into the second
Wollaston prism by lens L2 . By control of the diagonal angle α in the prisms, the
output light rays can be brought close to parallel. By adjustment of the distances
between the lenses and prisms, the output can be made convergent or divergent.
For calcite Wollaston prisms of thickness l1 = l2 = 8 mm, α = 79 deg, and
collimating lens focal lengths f1 = f2 = 50 mm, an output shear distance of ≈
3.5 mm is obtained. When the two prisms are in line, changes in the values of α, l1 ,
and l2 have negligible influence on the performance of the interferometer, where
the optical path difference between the two output beams can be brought to less
than the coherence length of a subnanosecond light pulse.

6.2.5 Aplanatic prism spectrograph

A double-pass aplanatic prism spectrograph is illustrated in Fig. 6.14.10 The basic
configuration consists of a dispersing prism with spherical concave and convex
surfaces, backed by a spherical concave mirror. A series of aplanatic (no spherical
aberration or coma) refractions at the prism faces produces a series of virtual and
real images at the aplanatic conjugates for these faces. The dispersion of the prism
material causes small changes in the location of these secondary images, resulting
in a spectrum at the focal plane close to perpendicular to the incident light. The
magnification is close to 1:1 with a throughput > 90% in the visible spectrum.

Figure 6.14 A double-pass aplanatic prism spectrograph.10

A Selection of Prism Applications 153

The focal plane is flat and aberration free. The detector placed at the focal plane
is usually a digital camera CCD matrix array. It remains a matter of definition
whether the dispersing element is considered to be a prism or a section of a convex-
concave lens. This prism spectrograph is offered commercially as the PARRIS
Imaging Prism Spectrometer and sold by LightForm, Inc.11

6.3 Prismatic Optical Devices

6.3.1 Prism switch for fiber-optic connections
A switching device can be constructed that uses an optical prism as a transfer
element between the inputs and outputs of a series of optical fibers without a
physical disconnect and reconnect.12 In Fig. 6.15(a), there are two parallel light-
beam channels, each between a set of input and output optical fibers. The beams
are formed by a molded lens element attached to the end of each fiber. Between one
set of fibers is a hexagonal prism element that does not affect the beam direction
when positioned as shown. When the prism is translated to the position shown
in Fig. 6.15(b), the input and output are switched between the two channels by
refraction and total internal reflection within the prism.
A top view of a group of eight fiber-optic channels symmetrically placed on
a circle around a central fiber-optic channel is shown in Fig. 6.16. The fiber-optic
lens elements are mounted in holes in two rigid plates. By means of stepper motors,
the prism can be rotated and translated to align any of the peripheral outputs to the
central input. Once calibrated, this switching can be performed remotely.

6.3.2 Laser gyro readouts

The ring-laser gyroscope (gyro) is used extensively in inertial guidance systems.
The basic operation is illustrated in Fig. 6.17. A linear laser with cavity modes
consisting of two waves traveling in opposite directions is mounted in a triangular
cavity configuration with three mirrors. The beam directions are designated as
clockwise (CW) or counterclockwise (CCW), and the resonance condition occurs
when the cavity length equals an integral number of wavelengths. The rotation of
the cavity produces a difference in the path length between the CW and CCW
beams. If these two beams are output through one of the mirrors, the resultant
interference pattern can provide information on the rotation rate.
The fringe patterns give the instantaneous phase difference between the CW and
CCW beams, with a prism readout device that combines the beams (Fig. 6.18).13
One of the gyro mirrors transmits a small percentage of the light (<0.1%) from
the cavity. A prism ABC of refractive index n has a vertex angle α close to, but
not exactly, 90 deg, such that α = 90 deg + α0 . The divergence δ of the beams
exiting face AB after two internal reflections is δ = 2nα0 . For example, when
δ = 15 arcsec and for a HeNe wavelength λ = 0.633 nm, there is a 3-mm fringe
spacing. By the use of a detector much smaller than this fringe spacing, the rotation
rate and direction can be determined by the rate that fringe intensity maxima move
past the detector.
154 Chapter 6

Figure 6.15 (a) A two-channel optical-fiber prism switch in a neutral position.12 (b) A two-
channel optical-fiber prism switch in a switched position.12

Figure 6.19 illustrates another type of prism-based readout.14 Two symmetric

prisms ABCD and AB0 C0 D are optically bonded along AD with a 50T/50R
(transmittance/reflectance) beamsplitter coating. This compound prism is optically
bonded to one of the partially transmitting mirrors of the laser gyro. Faces AB
and AB0 also have partially reflecting coatings. The CW and CCW laser beams
from the laser gyro intersect at a common point P1 on the laser gyro mirror. By
control of the prism geometry, the reflected CW and CCW beams from surfaces AB
and AB0 intersect at the common point P2 . The transmitted component of the CW
beam becomes collinear to the reflected component of the CCW beam, and both
are directed to detector D1 . The reflected component of the CW beam becomes
collinear to the transmitted component of the CCW beam, and both are directed
to detector D2 . Thus, each detector receives 50% CW + 50% CCW, forming an
A Selection of Prism Applications 155

Figure 6.16 An eight-channel optical-fiber prism switch showing a single switch position.12

Figure 6.17 Basic layout of a ring-laser gyroscope.

interference spot whose intensity is proportional to the phase difference between

the components. From the intensity variation at either detector, the rotation rate
can be determined. To determine the rotation direction, additional techniques are
required.14

6.3.3 Reflecting wedge prism for optical reader

A wedge prism can be used to modify the aspect ratio of a light beam in an optical
system (see Sec. 4.3.4). Figure 6.20 shows the components of an optical disk reader
that uses a semiconductor laser light source.15 Elliptical cross-section light emitted
from the semiconductor laser element (λ = 780 nm) is collimated by lens L1 and
156 Chapter 6

Figure 6.18 A type of optical prism readout device for a laser gyro. (Adapted from Ref. 13
with permission from Elsevier.)

Figure 6.19 A symmetric prism readout device for a laser gyro.14

enters the top surface of reflectorized wedge prism represented in cross section by
ABC. Surface AB is antireflection coated, and surface AC has a high-reflectance
coating. The wedge prism of refractive index n is inclined at angle ϕ such that
the elliptical cross-section input beam is expanded in one dimension to a circular
cross-section output beam. The anamorphic expansion is W2 /W1 ≈ 3:1 when the
prism is inclined at an angle ϕ = 27.4 deg, n = 1.765, and α = 12.4 deg, and the
output and input beams are orthogonal.
The circular profile beam is focused to the reading plane of an optical disk by
tracking and focusing lens L2 . The modulated light from the disk then passes back
through the system, where the beam cross section is compressed by the wedge
prism. The beam is focused by lens L1 and diffracted by a holographic element to
a signal detector.
A Selection of Prism Applications 157

Figure 6.20 An optical disk reader using a wedge prism in double-pass mode.15

6.3.4 Total-internal-reflecting touch switch using a Dove prism

The evanescent wave that penetrates a surface when TIR occurs can be the basis
of a touch switch when frustrated total internal reflection (FTIR) occurs. A touch
switch that controls the on-off state of an electrical device is based on the use of
FTIR in a Dove prism.16 A series of multiple TIR reflections occur within the
Dove prism, where the light beam enters and exits parallel to the base of the
prism [Fig. 6.21(a)]. A photodetector connected to an electronic control circuit
determines the on-off state of the electrical device. When the top surface of
the prism is touched by a finger, skin moisture and oils produce FTIR, and the
transmitted light intensity is reduced. Finger pressure, beam size, the number of
TIR points along the prism, and the sensitivity of the electronic control circuit
affect the on-off sensitivity.
The geometry for multipoint TIR is shown in Fig. 6.21(b). The minimum length
L of the TIR surface of the Dove prism can be calculated from15

2(d − y1 )
L= + 2N(d − y1 ) tan I + 2(n − 1)y1 tan I, (6.6)
tan ϕ

where

d = Dove prism thickness,

y1 = distance of incident beam from the bottom of the prism,
ϕ = Dove prism slope angle,
N = number of TIR points on the touch surface,
I = internal angle of incidence at the TIR point.
158 Chapter 6

Figure 6.21 (a) Optical touch switch showing total internal reflections in a Dove prism. (b)
Geometry for frustrated multipoint.16

The sensitivity and reliability of the touch switch can be increased for larger N
values. It can be shown from Eq. (6.6) that for a fixed value of L, N can be increased
by decreasing the thickness d of the prism, and also that for fixed L and d values,
N can be increased by reducing the value of ϕ.

6.3.5 Inspection device for window surfaces

For the purpose of visual inspection and detection of surface and subsurface defects
in a pane of window glass, light from a fiber-optic light guide is coupled into the
window pane using an attached light-guiding prism, as shown in Fig. 6.22.17 The

Figure 6.22 Light-coupling prism for window-defect inspection device.17

A Selection of Prism Applications 159

slope of the entrance face is ≈ 70 deg with respect to the window surface, and
the polished bottom face of the prism is coupled to the window surface using an
index-matching fluid. Light entering the glass plate undergoes TIR at both the top
and bottom surfaces of the window, and the prism length is controlled to avoid
reemergence of light reflected off the bottom window surface through the prism.
The presence of a window surface defect or interior defect changes the direction
of the light as it escapes, with the defect appearing as a bright spot on a darker
background. The suppression of stray light entering the window glass allows an
automated-defect-inspection technique to be implemented.

6.4 Viewing, Display, and Illumination Systems

6.4.1 Direct-view system for a microdisplay
Compact direct-view display systems are a necessary component for binocular
head-mounted displays. A system can be designed for a reflective imager as in
Fig. 6.23.18 Light is collected from an LED RGB array light source by a reflector.
The light then passes through a collimating Fresnel lens, rectangular aperture stop,
high-transmission diffusing plate, and prepolarizer that transmits s-polarized light.
The display panel is preferably of the reflective LCOS type. An image of the
display panel is formed at the exit pupil by the eyepiece optics. The beam-control
optics consists of two low-birefringence glass prism sections, a 30/90/60-deg
bottom section and a smaller 30/90/60-deg top section to provide the light path
as shown. Faces BC and CD of the bottom prism are coated with a high-reflectance
phase-optimized coating to reduce undesirable polarization phase shifts. Total
internal reflection from prism face AB also produces skew ray depolarization
and has a phase-optimized coating. A planar polarizing beamsplitter (PBS)—
preferably a wire-grid polarizer (WGP)—is placed along AD, and the two prism
sections are optically bonded along AD.

Figure 6.23 Direct-view microdisplay and illumination system.18

160 Chapter 6

S-polarized light from the Fresnel lens is reflected from prism faces CD and BC,
undergoes TIR at face AB, and is reflected to the LCOS (liquid crystal on silicon)
imager from the WGP. When the LCOS imager is in the dark state, there is no
change in polarization state of the reflected ray, and the reflected light from the
PBS does not reach the viewer. In the bright state, the reflected p-polarized light
is transmitted by the PBS and reaches the viewer, with intermediate polarization
states providing gray scale.

6.4.2 Binocular surgical loupe with flare reduction

A binocular surgical loupe may be used for surgery or other related applications.
It is worn as a headpiece with identical loupes to provide binocular vision over
an illuminated field. Figure 6.24 illustrates the components of a single loupe that
provides a well-illuminated area over the field of view, and by the use of circularly
polarized light, minimizes viewed flare from the beam-splitting optics.19
Light from a fiber-optic illuminator is collimated by lens system L1 and passes
through a one-piece circular polarizer CP1 to a 50T/50R beam-splitting cube. The
transmitted component is reflected from mirror M1 for overall illumination of the
object, while the reflected component is focused by the objective lens system L2
for directed illumination. The viewing optics consists of the objective lens L2 ,
the transmitting beam-splitting cube, a circular polarizer CP2 , a Schmidt–Pechan
prism, and an eyepiece L3 . The Schmidt–Pechan prism provides an erect image and
lengthens the optical path between L2 and L3 for increased image magnification.
The circular polarizers suppress flare that occurs at the beam-splitting surface but
does not affect the mostly depolarized viewed illumination reflected back from the
object.

Figure 6.24 Biological surgical loupe using beam-splitting cube and Schmidt–Pechan
prism.19
A Selection of Prism Applications 161

6.4.3 Inversion prism for range finders

For telescopic laser range finders, it is necessary to view the object, illuminate the
object with a laser light beam, and read the distance-measurement information. It
is useful to perform these three functions along a common axis, where the readout
display would be in the field of view of the eyepiece optics. In Fig. 6.25, an afocal
telescopic system provides three channels for these functions that lie close to a
common axis.20 Prism 1 is a right-angle prism that has a beam-splitting dichroic
coating along a section that is optically bonded to prism 2. A reflective coating
is applied to the uncoated section of prism 1, and prism 2 has a roof section for
correct orientation of the image. Both the bonded section and the roof section are
at about a 60-deg angle with respect to the optical axis.
In the illumination mode, light from the laser diode is reflected from three faces
of prism 1 and the beam-splitting coating and does not reach the eyepiece. In the
view mode, light from the objective lens is internally reflected twice by prism 1,
transmitted through the beamsplitter, internally reflected three times by prism 2,
and then exits along the optical axis to the eyepiece optics. In the readout mode,
light from the display is reflected by the folding mirror, imaged by the display lens,
reflected by the beamsplitter, internally reflected three times by prism 2, and then
exits along the optical axis to the eyepiece optics.

6.4.4 Prism transforming transmitted intensity profile

The two-section prism device in Fig. 6.26 is useful for smoothing the distribution
of an incident light beam having higher intensity at the center than at the edge. It
consists of trapezoidal prism ABCD optically bonded to right-angle prism CDE
with a 50T/50R beam-splitting coating at the interface.21 Consider a collimated
incident beam having a Gaussian distribution incident on face AE. The incident

Figure 6.25 Telescopic range-finder prism assembly with three light channels along the
optical axis.20
162 Chapter 6

Figure 6.26 Two-section prism assembly transforming intensity distribution of the input
beam.21

distribution profile is preserved at the exit surface BE by TIR at surface AB and

transmission through the beam-splitting coating. Another component undergoes
a symmetric inversion due to TIR at surface AB and reflection from the beam-
splitting coating. These half-intensity components combine at the exit surface BE
to form a more uniform output light beam. Figure 6.27(a) illustrates the incident
Gaussian-light-beam profile, and Fig. 6.27(b) shows how the original profile and
the inverted profile combine to form a uniform output.
This procedure can be extended to a three-section prism, as in Fig. 6.28.
The input is a collimated beam with a much sharper falloff. A rhomboid prism
ABCD is optically bonded to a trapezoidal prism CDEF, which is bonded to
a right-angle prism EFG. At each bonded interface, there is a beam-splitting
coating whose reflectance and transmittance produce a nearly equal intensity at
each output section. To accomplish this, the beam-splitting coating on interface
EF has a reflectance R = 0.5 and a transmittance T = 0.5, while the beam-
splitting coating on interface DC has a reflectance R = 0.333 and a transmittance
T = 0.667. Figure 6.29 shows the incident-light-beam distribution, the internal
profile inversions, and the more uniform composite output.

Figure 6.27 Input and output light-intensity profiles for two-section prism assembly.21
A Selection of Prism Applications 163

Figure 6.28 Three-section prism assembly transforming intensity distribution of the input
beam.21

Figure 6.29 Input and output light-intensity profiles for three-section prism assembly.21

References
1. L. Beiser, “Laser scanning systems,” in Laser Applications, M. Ross, Ed.,
Vol. 2, 71–86, Academic Press, New York (1974).
2. G. F. Marshall, “Risley prism scan patterns,” Proc. SPIE 3787, 74–86 (1999)
[doi:10.1117/12.351658].
3. B. D. Duncan et al., “Wide-angle achromatic prism beam steering for
infrared countermeasure applications,” Opt. Eng. 42, 1038–1047 (2003)
[doi:10.1117/1.1556393].
4. B. Sallee and J. K. Vinson, “Optical system for ladar guidance application,”
U.S. Patent No. 6,371,405 (2002).
164 Chapter 6

5. D. A. Rockwell, “Polarization rotator with frequency shifting phase

conjugate mirror and simplified interferometric output coupler,” U.S. Patent
No. 5,483,342 (1996).
6. J. C. Owens, “Laser applications in metrology and geodesy,” in Laser
Applications, M. Ross, Ed., Vol. 1, 84–86, Academic Press, New York (1971).
7. V. Ya. Barash et al., “A laser displacement interferometer,” USSR Invention
Brevet No. 1,275,205 (1984).
8. A. L. Vitushkin and L. F. Vitushkin, “Design of a multipass optical cell based
on the use of shifted corner cubes and right-angle prisms,” Appl. Opt. 37(1),
162–165 (1998).
9. H. Yu and S. Meng, “Wollaston prism design and working parameters in the
Nomarski polarized light interferometer,” Opt. Eng. 35(8), 2310–2312 (1996)
[doi:10.1117/1.600805].
10. D. W. Warren et al., “Compact prism spectrograph based on aplanatic
principles,” Opt. Eng. 36(4), 1174–1182 (1997) [doi:10.1117/1.601237].
11. LightForm, Inc., 601 Route 206, Hillsborough, NJ 08844.
12. C. F. Buhrer, “Multi-position optical fiber rotary switch,” U.S. Patent
No. 5,115,481 (1992).
13. F. Aronowitz, “The laser gyro,” in Laser Applications, M. Ross, Ed., Vol. 1,
139–141, Academic Press, New York (1971).
14. S. P. Callaghan, “Readout for a ring laser,” U.S. Patent No. 4,582,129 (1986).
15. K. Yoshifusa and T. Yokota, “Beam converting apparatus with a parallel light
beam input and output from one prism plane,” U.S. Patent No. 5,007,713
(1991).
16. S. Sumriddetchkajorn, “Optical touch switch based on total internal
reflection,” Opt. Eng. 42(3), 787–791 (2003) [doi:10.1117/1.1544457].
17. H. Weidner et al., “Improved illumination device for inspecting window
surfaces,” NASA Tech Brief KSC-12127, NASA John F. Kennedy Space
Center, FL (2000).
18. P. L. Gleckman and M. Schuck, “Compact near-eye illumination system,” U.S.
Patent No. 6,542,307 (2003).
19. B. Clark, “Optical system which allows coincident viewing, illuminating and
photographing,” U.S. Patent No. 5,078,469 (1992).
20. A. Perger, “Prism system for image inversion in a visual observation beam
path,” U.S. Patent No. 6,292,314 (2001).
21. K. Matsuaka et al., “Optical element having function of changing the cross-
sectional intensity distribution of a light beam,” U.S. Patent No. 4,641,920
(1987).
Chapter 7
Projection Displays
7.1 Color-Separating and Color-Combining Prisms
7.1.1 Three-channel Philips RGB separating prism
One of the first commercial applications of trichroic separation prisms was
developed by Philips Corporation for use in early color television cameras. Widely
known as the Philips prism, it consists of three prism sections and is used today in
three-channel charge-coupled device (CCD) video camcorders. Figure 7.1 shows
one configuration.1 Converging light from a camera objective lens is split into
red, green, and blue components that are directed to separate photocathode or
CCD receivers. The geometry must be such that the optical path lengths for each
color are equal and there are two reflections for each split ray to preserve image
orientation. Face 2 of prism A has a first dichroic layer that reflects green light
and transmits red and blue light. There is a small air-gap separation between face
2 and face 3 of prism B. Reflected green light from face 2 is directed to the green
sensor by TIR from entrance face 1. A second dichroic layer on face 4 of prism
B reflects the red light to face 3, where it is directed to the red sensor by TIR.
Prism C is optically bonded to prism B, and the transmitted blue light is directed
to the blue sensor. The varying angle of incidence on each dichroic layer causes

Figure 7.1 A Philips prism for three-color separation.

165
166 Chapter 7

the transmission and reflection curves to be shifted toward shorter wavelengths.

However, transmitted rays that are incident on the first dichroic layer at larger
angles of incidence have smaller angles of incidence on the second dichroic layer,
and vice versa. Hence, the wavelength shifts that occur in each dichroic compensate
each other. This application of the Philips prism uses a single-pass mode with
unpolarized light.

7.1.2 Philips prisms in reflective LCD projection displays

The Philips prism can be used as a color-separating and combining prism in
projection displays using reflective LCD imager s. A typical projection system is
shown in Fig. 7.2. The reflective light modulator is preferably of the liquid crystal
on silicon (LCOS) type, which has fabrication advantages and high-resolution
capability in a small-size imager. The pixels of LCOS spatial-light modulators
provide variable rotation of the polarization plane of incident polarized light, such
that by the use of an analyzer, transmission states can be continuously varied
from bright to dark. A PBS cube can be used as the polarizer/analyzer, and
the Philips prism provides the color separation and recombination. However, the
s-p polarization rotational splitting at the dichroic surfaces of the Philips prism
increases with increasing angle of incidence on the coatings, resulting in a loss of
image contrast (ratio of transmitted light in the bright state to the transmitted light
in the dark state) and color fidelity. This is often called skew-ray depolarization or
compound-angle depolarization.
One method to address this depolarization is by a modification the geometry
of the conventional Philips prism to decrease the angles of incidence on the
dichroic coatings while maintaining the TIR requirements. The need for TIR in a
conventional Philips prism limits the minimum angle of incidence on the dichroic
surfaces to about 25–30 deg. Figure 7.3 illustrates a modified three-element

Figure 7.2 Typical projection system using a Philips prism.

Projection Displays 167

Figure 7.3 A modified three-element dichroic prism with a matched noncubic PBS.
(Adapted from Ref. 2 with permission from the Optical Society of America.)

trichroic prism design with a matched noncubic PBS.2 For a prism refractive
index n ≈ 1.52 and an f /4 beam, the minimum angle of incidence on the
dichroic coatings is reduced to about 16 deg, resulting in improved optical system
performance. The four-element trichroic prism in Fig. 7.4 reduces the angles of
incidence on the dichroic surfaces.3
The angle of incidence on the first dichroic surface is I1 ≈ 11 deg, and the
angle of incidence at the second dichroic surface I2 ≈ 20 deg. This decreases the
phase change between the s-polarized and p-polarized transmission curves, and
performance is improved. Another method is to insert wavelength-compensation
plates between the input/output surfaces of a Philips prism and the reflective light
modulators.4 These plates are positioned normal to the system’s optical axis, and
the waveplate thickness can vary from near zero up to λ/2. Polarization rotation and

Figure 7.4 A four-element trichroic prism with reduced angles of incidence on the dichroic
surfaces.3
168 Chapter 7

ellipticity are introduced at both the dichroic surfaces and the TIR surfaces, and the
polarization characteristics of each channel are analyzed by means of polarization
pupil maps. The retardation of each waveplate is then calculated to reduce the
polarization orientation and ellipticity to a minimum in the “off” or “dark” state for
each channel. By the deposition of phase-control coatings on the dichroic surfaces
of the Philips prism, the depolarization of beams having finite numerical aperture
can be reduced.5,6 For a Philips prism used in double-pass projection mode, the
cross-polarization leakage can be reduced to ≈ 1 × 10−3 by these phase-control
coatings.

7.1.3 Crossed dichroic x-cube prisms for projection displays

An important type of three-channel color prism for projection displays is the
crossed dichroic prism or x-cube prism. It is more compact than an equivalent-
aperture Philips prism and consists of four 45/90/45-deg sections. Section A
has a blue reflecting (red/green transmitting) dichroic coating on face 1 and a
red reflecting (blue/green transmitting) coating on face 2. Section B has a red
reflecting (blue/green transmitting) dichroic coating on face 3, and section D has
a blue reflecting (red/green transmitting) coating on face 4. Section C is uncoated.
The sections are bonded together using a clear optical adhesive to form a cube.
Figure 7.5 shows the light paths when the x-cube prism functions as a color
integrator. It can be used in reverse to separate white light into red, blue, and green
components.
The typical use of an x-cube prism and three PBS cubes is a projection system
using LCOS light modulators, as shown in Fig. 7.6. Because of the 45-deg angle
of incidence at the dichroic surfaces, there are strong polarization shifts at these

Figure 7.5 Light paths for an x-cube prism used as a color integrator.
Projection Displays 169

Figure 7.6 Typical use of an x-cube prism and three PBS cubes for an LCOS projection
system.

surfaces that degrade projected image contrast and color fidelity. This effect can be
reduced by phase-shift coating s on the dichroic surfaces or by external waveplates.
Polarization-insensitive x-cube prisms are commercially available.
By the use of off-axis projection optics and the placement of the polarizers in
the light path behind the x-cube prism, depolarization effects in the x-cube prism
can be minimized (Fig. 7.7). Unpolarized light enters a beam-separation prism at
an angle such that it is directed to an x-cube prism by TIR, and there are negligible
depolarization effects from the dichroic coatings of the x cube. Linear polarizers
are positioned between the x-cube prism and each reflective LCD imager such
that a dark state is achieved in the “off” position. When the imagers are activated,
polarized light returns to the x-cube prism, where depolarization effects can occur.

Figure 7.7 An off-axis projection system with minimized x-cube depolarization effects.7
170 Chapter 7

The beam enters the beam-separation prism at an angle, where it is transmitted

to the projection lens. The image retains a high contrast and brightness since the
viewer perceives only the light intensity, not the polarization state. This off-axis
arrangement can also be used to minimize the depolarization effects in a projection
system using a Philips prism.

7.1.4 Prisms for digital light processing projection

Digital light processing (DLP) projection systems make use of the digital
micromirror device (DMD) from Texas Instruments. This device consists of a
high-resolution array of hinged micromirrors with a separation ≈ 1 µm. Each
micromirror can be electronically tilted about the mirror diagonal at an angle of
about ±10 to ±12 deg at a rate of over 5000 times per second. By the use of
a light-directing TIR prism, reflected light can be either directed to a projection
lens (“on”state) or to a heat sink or light dump (“off” state). The concept of DLP
projection for a single micromirror is shown in Fig. 7.8. Gray scale is achieved
by controlling the ratio of “on” time to “off” time for each micromirror. Light-
deflecting TIR prisms for DLP projection are commercially available.
For a three-chip color DLP projection system, color-splitting and combining
prisms are used. Both Philips prisms and x-cube prisms can be used. Figure 7.9
shows a section of a DLP projector using a Philips prism in double-pass mode as
a color splitter and combiner. Figure 7.10 illustrates a section of a DLP projector
where the x-cube prism is used as a color combiner, and the color splitting is done
in a separate module.8 Although DLP projectors do not require polarized light,
polarization shifts between the s-polarized and p-polarized components of light at
the 45-deg incidence angle on the dichroic coatings of an x cube can degrade the
brightness and color temperature of the projected image. Chen et al. have modified
the x-cube dichroic coatings to reduce these polarization wavelength shifts and
have reported a projected brightness increase of about 33% and an increase in the
color temperature of about 18% for a DLP projector.9

Figure 7.8 DLP projection scheme for a single micromirror.

Projection Displays 171

Figure 7.9 A section of a DLP projector using a Philips prism as a color splitter and
combiner.

Figure 7.10 Section of a DLP projector with an x-cube as a color combiner.8

The modified crossed–dichroic color-combining prism of Fig. 7.11 consists

of four optically bonded sections: an isosceles prism section, two quadrangular
sections, and a pentagonal prism section.10 The base angle β is such that 30 deg ≤
β ≤ 45 deg, and the acute angle ϕ between AC and BD is such that 60 deg ≤
ϕ ≤ 90 deg. Bonded faces AE and EC have a blue-reflecting dichroic coating,
and bonded faces BE and ED have a red-reflecting dichroic coating. The angle of
incidence of the light beams on the dichroic surfaces can be reduced to ≈ 30 deg,
reducing any depolarization effects.
There are several types of light-directing TIR prisms for DLP projection.
In Fig. 7.12, a DLP projector uses a three-section crossed TIR light-directing
prism. For a DMD that has three stable states—e.g., +10 deg, flat, −10 deg—the
172 Chapter 7

Figure 7.11 A modified crossed–dichroic color-combining prism.10

Figure 7.12 A DMD projector using a three-section crossed TIR light-directing prism.
(Adapted from Ref. 11 with permission from The Society for Information Display.)

output of two synchronized pulsed sources can be combined.11 A three-section

beam-separating TIR prism for DLP projection can provide a large separation
of the exiting on-state beam and the off- and flat-state beams (Fig. 7.13).12 This
produces minimal spillage of the off- and flat-state beams into the projection
path of the on beam, yielding a high-contrast projected image. Light-directing
TIR prisms are sold commercially, such as the LightGateTM from Unaxis or the
VikuitiTM TIR prism from 3M.
Projection Displays 173

Figure 7.13 A three-section TIR light-directing prism for high-contrast DLP projection.12

7.1.5 Other types of color-separating prisms for projectors

Figure 7.14 shows a projector with a color-separating and combining prism
consisting of three identical prism sections and usable with LCOS imagers.13 Face
AD of prism 1 has a dichroic coating that reflects red light and transmits blue and
green light. Face BD has a dichroic coating that reflects green light and transmits
blue light. The three prisms are optically bonded at the contact surfaces. The LCOS
imagers for red, blue, and green light are positioned as shown. The TIR surfaces AC
and AB have multilayer polarization phase shift–compensating coatings to improve
contrast. Polarized light from a PBS enters the prism through surface AC, and the
polarization-modulated output is directed to a projection lens.

Figure 7.14 Projector with a color-separating/combining prism consisting of three identical

prism sections.13
174 Chapter 7

Another color-separating/combining prism in Fig. 7.15 consists of four sections

and is usable with reflective LCD imagers.14 A blue reflecting dichroic coating
is applied to face AE of prism 1 and face CE of prism 3. A red reflecting dichroic
coating is applied to face BE of prism 2 and face DE of prism 1. The prism sections
are optically bonded at the contact surfaces. Entrance face AD is a TIR surface for
reflected blue and red rays. A blue imager is positioned opposite exit face CD, a
red imager opposite exit face AB, and a green imager opposite the exit face FG of
prism 4.
A three-channel color-separating/combining system for small reflective LCD
imagers (e.g., LCOS) can be designed as a compact module using QuadCubeTM
architecture.15 It consists of four PBS cubes and various spectrally selective
polarization filters or coatings (ColorSelect
R
filters).16 This quad architecture can
be configured in various arrangements to provide collinear or orthogonal output
to a projection lens. A configuration that provides collinear output is shown in
Fig. 7.16, where the components are separated for clarity.17 A randomly polarized

Figure 7.15 A four-section color-separating/combining prism for reflective LCD imagers.14

Figure 7.16 A quad-cube color-splitting/combining prism with polarization beam splitting.17

Projection Displays 175

white-light source provides s-polarized light through a plate polarizer, which is

then incident on a green polarization filter 1 in front of PBS 1.
For the green component, selective filter 1 rotates the plane of polarization of
green light 90 deg, while the red and blue components are unaffected. The p-
polarized green light is transmitted by the PBS 1 and passes through PBS 2 to
the green reflective LCD imager. For the bright state, the LCD imager rotates the
plane of polarization 90 deg, and the s-polarized green light is reflected from PBS
2 and PBS 4. A green polarization filter 4 at the exit face of PBS 4 rotates the plane
of polarization 90 deg, and the p-polarized light enters the projection lens.
For the red component, the incident s-polarized light is reflected in PBS 1 and
passes through a red polarization filter 2 positioned between PBS 1 and PBS 3,
where it is converted to p-polarized light and passes through PBS 3 to the red
reflective LCD imager. The reflected s-polarized light from the imager is reflected
in PBS 3 and is converted to p-polarized light by a red polarization filter 3
positioned between PBS 3 and PBS 4. It passes unaffected through PBS 4 and
green polarization filter 4 to the projection lens.
For the blue component, the incident s-polarized light is reflected in PBS 1,
passes through red filter B, and is reflected in PBS 3. It passes to the blue imager as
s-polarized light. The reflected p-polarized light from the imager passes unaffected
through PBS 3, red polarization filter 3, PBS 4, and green polarization filter 4 to
the projection lens. In practice, the polarization filters can be applied to the PBS
surfaces as optical coatings, and the PBS cubes can then be optically bonded to
form a single module.

7.2 Polarizing Beamsplitters for Projection Displays

7.2.1 MacNeille polarizing beamsplitters
All practical projection systems use convergent or divergent light beams,
and for compact LCOS systems, low- f /# beams are required. This places a
constraint on the performance of components, especially polarizing beamsplitters.
The conventional MacNeille PBS described in Sec. 3.1.10 produces skew-ray
depolarization when used in the fast (≤ f /2.5) illumination beams of most LCOS
projectors. The rotation of the s-polarization and p-polarization axes produces a
loss of contrast and uniformity in the projected image. One method to improve
this depolarization is by the insertion of a quarter-wave retarder between the PBS
and the reflective imager. Accurate alignment of the retarder is required, and this
method is applicable when the PBS functions both as a polarizer and an analyzer.
Another compensation method has been described that inserts an oblique-aligned
uniaxial material on a glass substrate (o-plate compensator) at an input surface
of a conventional PBS.18 The material can be a birefringent liquid crystal polymer
(LCP plate compensator). For an f /2.0 beam, it was calculated that < 0.1% of non-
s-polarized light was transmitted over a 200-nm bandwidth for the compensated
PBS, as compared to 0.8% for an uncompensated PBS.
A three-panel LCOS projector that uses polarization-compensated MacNeille-
type PBSs is described by Chen et al.19 For color management, retarded stack
176 Chapter 7

filters (RSFs) are used that orthogonally rotate the polarization over a specified
spectral range. To minimize filter leakage from the skew-ray depolarization of
the PBSs, they are designed to have reflection invariance, with ≈ 0.2% leakage
over its color band for an f /2.5 beam. In Fig. 7.17, a projector uses four PBSs, a
blue/yellow RSF (rotates polarization of transmitted first color, retains polarization
of transmitted second color), a green/magenta RSF, a magenta/green RSF, and three
LCOS imagers. There is also an optional clean-up polarizer, a half-wave plate at 0
deg between PBS 1 and PBS 2 to minimize effects of polarization mixing between
these PBSs, and a polarization-rotating quarter-wave plate at 45 deg between PBS
2 and PBS 4. It is stated that this architecture can achieve a contrast ratio exceeding
1000:1.

7.2.2 Cartesian polarizing beamsplitters

The wide-angle Cartesian polarizing beamsplitter was discussed in Sec. 3.1.12.
The Cartesian PBS can be used to advantage in projection systems using reflecting
LCD imagers, where the input beam f /# can be reduced to ≤ f /2.5.20 Figure 7.18
illustrates a three-channel LCOS projector that uses an x-cube color-combining
prism and three Cartesian polarization beam-splitting cubes. The x cube is of the
type shown in Fig. 7.5.
The multilayer polarizing beamsplitter (MLPB) film has a lower refractive index
than the surrounding glass cube and functions as a 45-deg tilted low-index plate
immersed in a higher-index medium. Astigmatism is generated when the PBS
is used in a divergent or convergent beam. This astigmatism can be reduced by
modifying the design of the PBS.21 One method is shown in Fig. 7.19(a). An
MLPB/adhesive layer having a low refractive index n1 and a glass layer having
a high refractive index n2 are sandwiched between the two halves of a glass cube
having an intermediate refracting index n0 such that n2 > n0 > n1 . The thickness d1

Figure 7.17 A three-channel LCOS projection system using compensated MacNeille

polarizing beamsplitters.19
Projection Displays 177

Figure 7.18 A three-channel LCOS projection system using three Cartesian polarizing
beamsplitters.20

of the MLPB film and the thickness d2 of the high-index layer are adjusted such that
the astigmatism caused by the high-index layer reduces the astigmatism introduced
caused by the MRPB/adhesive layer. Typical values are n0 ≈ 1.85 (PMH55 glass),
n1 ≈ 1.56, n2 ≈ 1.92 (PBH71 glass), d1 = 225 µm, and d2 = 3.8 mm.
Another method of astigmatism reduction uses a transparent wedge layer
between the 225-µm-thick MLPB film and one half of the glass cube [Fig. 7.19(b)].
The wedge material is not critical and can be an optical adhesive. The wedge
reduces the astigmatism caused by the MLPB film. For a PBS cube with a height
of h = 35 mm, the wedge angle α is such that 0.15 deg ≤ α ≤ 0.25 deg, and the
wedge thickness w = 129 µm.
An optical engine module has been developed consisting of three astigmatism-
compensated Cartesian polarizing beamsplitters and an x-cube color prism.22
When used in a prototype projector operating at f /2.0, contrast has been measured
at > 1500:1. This module is sold commercially as the 3M VikuitiTM LCOS Optical
Core.

7.2.3 Wire-grid polarizing beamsplitters in projection displays

The design principles of a wire-grid polarizer (WGP) for use in the visible
region were discussed in Sec. 3.1.13. Planar wire-grid polarizers have found
widespread use and are usually used at 0 or 45-deg incidence in reflective LCD
projection systems. Wire-grid polarizers do not exhibit the depolarization effects
of a conventional MacNeille PBS at f /#s approaching f /2.0. However, the wire-
grid polarizer is more wavelength dependent, and astigmatism is introduced when
178 Chapter 7

Figure 7.19 (a) Astigmatism correction in a Cartesian PBS using a high-index glass plate.
(b) Astigmatism correction in a Cartesian PBS using an optical wedge.21

used in the tilted mode.23 When used with an LCOS projection system at 45 deg
in an f /2.0 beam, polarization rotation generated in the WGP is compensated
by an opposite rotation in the WGP in the reflected beam, and no quarter-
wave plate is required for improved contrast.24 Yu and Kwok have measured the
optical properties of the WGP for projection displays.25 They have measured light
efficiency of ≈ 80%, and for optimum contrast have recommended an optimal tilt
angle of 35 deg instead of the conventional 45 deg.
A projection architecture using color-selective green/magenta and red/blue
polarization filters and four wire-grid polarizing beamsplitters (WGPBSs) is shown
in Fig. 7.20.26 Wire-grid polarizers are components in the UltreX-3 light engine
that is used in a rear-projection system.27 A WGP at 0-deg incidence can also be
used as a prepolarizer in other projectors using PBS cubes, such as the Cartesian
PBS system of Fig. 7.18.
Projection Displays 179

Figure 7.20 A three-channel projection architecture using wire-grid polarizers.26

7.3 Illuminators for Projection Displays

7.3.1 Hollow tunnel integrators
The light pipe or tunnel integrator is often used in projection displays to provide
uniform illumination from a light source to the imaging optics. One form is a
hollow pipe having a constant square or rectangular cross-sectional area with a
reflective coating on the wall interiors. Light from a nonuniform light source (or
a focused image of the source) positioned at the input aperture enters the light
pipe. Multiple kaleidoscopic reflections within the pipe produce an array of images
of varying uniformity, as shown in Fig. 7.21.28,29 When viewed through the exit
aperture, the superposition of these images becomes the effective light source,
producing a homogenized output of improved uniformity at the exit aperture.
Hollow light pipes are capable of transporting light of high heat flux but must
be assembled from four reflectorized sections. Figure 7.22 shows a construction
method for a hollow light pipe of quadrilateral cross section.30 The internal surfaces

Figure 7.21 Reflected-light-source images from a hollow light beam homogenizer.

180 Chapter 7

Figure 7.22 Construction of a hollow beam homogenizer with a quadrangular aperture.30

of the side and top pieces can be coated with reflecting glass or polished metal, and
the pieces can be held together by shrink-wrap tubing.
Several factors that determine the required length of a light pipe for a given
aperture or aspect ratio are (1) cross-sectional shape of the light pipe, (2) f /# of
the incidence beam, (3) centering accuracy of the light source, (4) light-source
distribution pattern, and (5) light-source size. Cassarly has modeled several light
pipe configurations using the LightTools R
nonsequential ray-tracing program,
taking into account these factors among others.31 In general, longer light pipes
and lower f /# input beams increase the number of superposed images and provide
better uniformity.

7.3.2 Solid light pipes

Light pipes constructed from solid material have some advantages in manufacture
and mechanical stability. From an optical standpoint, TIR at the interior surface
provides high and nondegradable reflection. However, high–heat-flux light sources
may affect the thermal stability, the wall may require some protection to maintain
TIR, and the end faces may require an antireflection coating.
A rectangular solid light pipe has been analyzed by Chang et al.32 Scrambling
within the light pipe produces an output to a liquid-crystal light valve with the
approximate cross-sectional shape of the light pipe. For a 4:3 aspect ratio light-
pipe cross section, an optical efficiency of 57% and light uniformity of 70% was
calculated.

7.3.3 Effect of light-pipe cross section on uniformity

The cross section of a light pipe can have a number of shapes. It has been found
that when the reflected images of the entrance aperture completely fill the reflection
plane, good uniformity can be achieved. The reflected cross-sectional images for
the light pipe of Fig. 7.21 fulfill this condition. Hexagonal and equilateral triangle
shapes can also be close-packed to fill the reflection plane and will produce good
uniformity. However, round and pentagon-shaped cross sections do not completely
fill the reflection plane and will produce poor uniformity, which has been verified
by ray tracing.31 In particular, the circular-cross-section light pipe produces a
centered hot spot.
Projection Displays 181

7.3.4 Solid microprismatic light homogenizer

Figure 7.23 illustrates a solid cylindric light-beam homogenizer that operates in
collimated light.33 A series of circular microprismatic grooves in the shape of
isosceles triangles are formed on the input and output surfaces having a diameter
D. The grooves are isosceles triangles with a base angle β. An incident collimated
beam is refracted at internal deviation angles ±δ, which are totally internally
reflected at the circular walls. The cylinder length L is adjusted such that rays
incident on the perimeter of the entrance surface are refracted to the center of the
exit surface, and rays incident near the center of the entrance surface are refracted
to the perimeter of the exit surface.
The angle of incidence I1 = β at the entrance surface, and δ = I1 −I1 0 = β−I1 0 . To
maximize δ and prevent a refracted ray from hitting an adjacent facet, the following
condition is required:

90 deg + I1 0


β= . (7.1)
2

Equation (7.1) is solved iteratively for β, and the length of the homogenizer
is calculated from L = D/(2 tan δ). The rays reflected off of the walls are
collimated at the exit aperture. The homogenizer is achromatic, since dispersion
generated at the input surface is canceled by opposite dispersion at the output
surface. Figure 7.24(a) models a distribution of an input beam, and Fig. 7.24(b)
indicates the homogenized output. For BK7 glass material (nd = 1.5168),
calculated parameters are β = 62.98 deg, δ = 27.01 deg, D = 25 mm, and L =
24.52 mm.

Figure 7.23 A solid cylinder light-beam homogenizer with microprismatic grooves on input
and output apertures.33
182 Chapter 7

Figure 7.24 (a) Modeled light distribution of input beam for microprismatic light-
beam homogenizer. (b) Light distribution of output beam for microprismatic light-beam
homogenizer.
Projection Displays 183

7.3.5 Tapered-tunnel illuminator for projection displays

If the input aperture of a hollow light pipe is illuminated with a divergent or
convergent circular light beam having a maximum half-angle θ, then the optical
throughput, or étendue E, is defined by the geometric quantity:

πA
E = πA sin2 θ = [mm2 -steradian], (7.2)
4 ( f /# )2

where A = area of aperture (mm2 ), and the f /# = 1/(2 sin θ) = 1/(2NA),

where NA is the numerical aperture. Thus, for an f /2.0 beam, θ = 14.5 deg and
NA = 0.25, and for an f /1.0 beam, θ = 30.0 deg and NA = 0.50.
A hollow illuminator for a projection system has been described where the
reflecting walls have the shape of a compound parabolic reflector (CPR) with
defined input and output aperture dimensions.34 The maximum value of the
divergence half-angle θo at the output aperture can be related to the maximum
input divergence half-angle θi , where θi  θo , and the étendue can be preserved
from the input aperture to the output aperture with maximum collection efficiency
by the following condition:

Ai sin2 θi = Ao sin2 θo . (7.3)

In Fig. 7.25, the maximum angle output ray angle θo at the upper-left edge
originates at the bottom-right edge of the opposite face.

Figure 7.25 Hollow illuminator having the shape of a compound parabolic concentrator.34
184 Chapter 7

For rectangular apertures, the condition for étendue preservation can be further
refined by the following:

ai bi sin θi a sin θi b = ao bo sin θo a sin θo b , (7.4)

where

ai , ao = heights of input and output rectangular apertures,

bi , bo = widths of input and output rectangular apertures,
θi a , θo a = input and output half-angles along rectangle heights,
θi b , θo b = input and output half-angles along rectangle widths.

A CPR illuminator for square apertures is shown in Fig. 7.26. For an f /2.0 output
beam, θo a = θo b ≈ ±15 deg.

Figure 7.26 A CPR illuminator for a projection display with a square output aperture.34

References
1. L. P. G. Verdijk and E. Tienkamp, “Color-separating prism arrangement
of which some surfaces adjoin dichroic layers,” U.S. Patent No. 4,009,941
(1977).
2. H.-S. Kwok et al., “Trichroic prism assembly for separating and recombining
colors in a compact projection system,” Appl. Opt. 39(1), 168–172 (2000).
3. J. W. Bowron, “Four prism color management system for projection systems,”
U.S. Patent No. 6,644,813 (2003).
4. B. Bryars, “Systems, methods and apparatus for improving the contrast ratio in
reflective imaging systems utilizing color splitters,” U.S. Patent No. 5,986,815
(1999).
5. M. R. Greenberg and B. J. Bryars, “Skew ray compensated color separation
prism for projection display applications,” SID Digest 31(1), 88–91 (2000).
Projection Displays 185

6. A. E. Rosenbluth et al., “Design of phase-controlled coatings to correct

skew–ray depolarization in LCOS projectors,” Displays 23(3), 121–138
(2002).
7. B. J. Bryars, “Off-axis projection display system,” U.S. Patent No. 6,398,364
(2002).
8. J. Huang, “Digital light processing projector,” U.S. Patent No. 7,144,116
(2006).
9. H.-C. Chen et al., “Improving the illumination efficiency and color temperature
for a projection system by depositing thin-film coatings on an x-cube prism,”
Opt. Eng. 45(11), 113801 (2006) [doi:10.1117/1.2393154].
10. F.-C. Ho and J.-J. Huang, “Optical prism assembly,” U.S. Patent No. 7,224,531
(2007).
11. D. Maes, “Dual TIR prism, a way to boost the performance of a DLPTM
projector,” Barco, Society of Information Display MEC Spring Meeting, Jena,
Germany (March, 2008).
12. S. M. Penn, “Prism for high contrast projection,” U.S. Patent No. 6,959,990
(2005).
13. J. Huang, “Color separation prism assembly compensated for contrast
enhancement and implemented as reflective imager,” U.S. Patent No.
6,704,144 (2004).
14. A. L. Huang, “Color-separating prism utilizing reflective filters and total
internal reflection,” U.S. Patent No. 6,517,209 (2003).
15. M. G. Robinson et al., “High contrast color splitting architecture using color
polarization filters,” SID Symp. Digest 31, 92–95 (2000).
16. G. Sharp et al., “LCOS projection color management using retarder stack
technology,” Displays 23, 139–144 (2002).
17. T. Suzuki, “Color-separating and recombining optical system,” U.S. Patent No.
6,984,041 (2006).
18. M. G. Robinson et al., “Wide field of view compensation technique for cube
polarizing beam splitters,” SID Symp. Digest 34, 874–877 (2003).
19. J. Chen et al., “High contrast MacNeille PBS based LCOS projection systems,”
Proc. SPIE 5740, 78–91 (2005).
20. C. L. Bruzzone et al., “Reflective LCD projection system using wide-angle
Cartesian polarizing beam splitter,” U.S. Patent No. 6,486,997 (2002).
21. D. J. Aastuen et al., “Projection system having low astigmatism,” U.S. Patent
No. 6,786,604 (2004).
22. C. L. Bruzzone et al., “High-performance LCOS optical engine using
Cartesian polarizer technology,” SID Symp. Digest 34, 126–129 (2003).
186 Chapter 7

23. S. Arnold et al., “An improved polarizing beamsplitter LCOS projection

display based on wire-grid polarizer,” SID Symp. Digest 32, 1282–1285
(2001).
24. E. Gardner and D. Hansen, “An image quality wire-grid polarizing beam
splitter,” SID Symp. Digest 34, 62–63 (2003).
25. X.-J. Yu and H.-S. Kwok, “Application of wire-grid polarizers to projection
displays,” Appl. Opt. 42(31), 6335–6341 (2003).
26. A. J. S. M. De Vaan and S. C. McClain, “Projection device with wire grid
polarizers,” U.S. Patent No. 6,873,469 (2005).
27. C. Pentico et al., “Ultra high contrast color management system for projection
displays,” SID Symp. Digest 34, 130–133 (2003).
28. M. M. Chen et al., “The use of a kaleidoscope to obtain uniform flux over a
large area in a solar or arc imaging furnace,” Appl. Opt. 2(3), 265–271 (1963).
29. L. J. Krolak and D. J. Parker, “The optical tunnel—a versatile electrooptical
tool,” J. SMPTE 72, 177–180 (1963).
30. B. Wagner, “Method of producing a hollow mixing rod, and a mixing rod,”
U.S. Patent No. 6,625,380 (2003).
31. W. J. Cassarly, “Design of efficient illumination systems,” Optical Research
Associates, SPIE Short Course Notes, SPIE 44th Annual Meeting, Denver, CO
(1999).
32. C.-M. Chang et al., “A uniform rectangular illuminating optical system for
liquid crystal light valve projectors,” SID Euro Display ’96, Birmingham,
England, 258–260 (1996).
33. D. F. Vanderwerf and A. J. Herbert, “Prismatic light beam homogenizer for
projection displays,” U.S. Patent No. 6,024,452 (2000).
34. A. L. Duwaer and J. F. Goldenberg, “Light valve projection system with
improved illumination,” U.S. Patent No. 5,146,248 (1992).
Chapter 8
Microprismatic Arrays
A microprismatic array is a series of prismatic elements that are positioned next
to each other in a 2D pattern or, in some cases, a 3D pattern. These prismatic
elements can be identical or vary in geometry and have smaller dimensions than
conventional single prisms. They are often replicated in optical plastic and can be
a rigid sheet or a flexible film, with the sheet area much greater than the thickness.
These prismatic sheets or films can subsequently be shaped into other forms, such
as a rectangle or cylinder, for specialty applications.

8.1 Roof Prism Linear Array

The linear array in Fig. 8.1 consists of a flat sheet of repeating isosceles roof prisms,
having a vertex angle ϕ. The prismatic array has a refractive index n, surrounded
by air with refractive index n0 . For an individual prism, an incident light-ray vector
with direction cosines K x1 , Ky1 , Kz1 is incident and refracted at planar surface 1.
The refraction matrix R1 is derived from Eq. (1.75), where the surface normals are
k x1 = 0, ky1 = 0, and kz1 = −1, referenced to the coordinate system shown, where

1 0 0 0 
 

0 n0 /n 0 0 
 
R1 =  , (8.1)
 0 0 n0 /n 0 
kz1 [ρ1 0 − (n0 /n)ρ1 ] 0 0 n0 /n

where

ρ1 = K x1 k x1 + Ky1 ky1 + Kz1 kz1 = cos I,

and
p
ρ1 0 = [1 − (n0 /n) sin I]2 = cos I 0 .

The direction cosines of the surface normals for surface 2 are k x2 = 0, ky2 =
cos[(180 deg − ϕ)/2], kz2 = −cos[(180 deg − ϕ)/2], and for surface 3 are k x3 = 0,
ky3 = −cos[(180 deg − ϕ)/2], kz3 = −cos[(180 deg − ϕ)/2].
A perspective view of an individual prism of an array, with apex angle ϕ,
and several possible paths of an incident ray are shown in Fig. 8.2. Ray A0

187
188 Chapter 8

Figure 8.1 Section of a microprismatic array.

Figure 8.2 Possible light-ray paths in a microprismatic array.

enters planar surface 1, normally at a 0-deg incident angle, undergoes TIR at

surfaces 2 and 3, and is retroreflected through surface 1. Ray A enters planar
surface 1 at an oblique angle, undergoes TIR at surfaces 2 and 3, and is refracted
upward at surface 1. Ray B is refracted at surface 1, undergoes TIR at surface 2,
is refracted at surface 3, and is transmitted downward. Ray C is refracted at
surface 1, directly refracted outward at surface 3, and transmitted downward. Both
B and C ray paths are possible up to near-grazing incidence angle I1 ≈ 90 deg.
Alternatively, rays incident from opposite directions could be reflected at surface 3
and refracted outward through surface 2, or directly refracted outward through
surface 2. Therefore, to calculate the spread of possible exit angles for an incident
light beam, both the refractive and reflective matrices are required for surfaces 2
and 3.
Microprismatic Arrays 189

From Eq. (1.76), the fourth-order reflection matrices R2 0 and R3 0 are:

 1 0 0 0
 
 0 1 0 0

R2 0 =  , (8.2)
−2ky2 ρ2 0 1 0
−2kz2 ρ2 0 0 1
 1 0 0 0
 
 0 1 0 0

R3 0 =  , (8.3)
−2ky3 ρ3 0 1 0
−2kz3 ρ3 0 0 1

and the refraction matrices R2 , R3 , and R4 are

1 0 0 0 
 

0 n/n0 0 0 
 
R2 =   , (8.4)
ky3 [ρ2 0 − (n/n0 )ρ2 ] 0 n/n0 0 
0
kz3 [ρ2 − (n/n0 )ρ2 ] 0 0 n/n0
1 0 0 0 
 

0 n/n0 0 0 
 
R3 =   , (8.5)
ky3 [ρ3 0 − (n/n0 )ρ3 ] 0 n/n0 0 
0
kz3 [ρ3 − (n/n0 )ρ3 ] 0 0 n/n0
1 0 0 0 
 

0 n/n0 0 0 
 
R4 =  
 , (8.6)
 0 0 n/n0 0 
kz4 [ρ4 0 − (n/n0 )ρ4 ] 0 0 n/n0

The appropriate system matrices S are given by

Reflected rays A, A0 : SA = R4 R3 0 R2 0 R1 , (8.7a)

Transmitted ray B: SB = R3 R2 0 R1 , (8.7b)
Transmitted ray C: SC = R3 R1 . (8.7c)

The direction cosines of the exiting rays, K x 0 , Ky 0 , and Kz 0 , can then be

calculated from the direction cosines of the incident rays, K x , Ky , Kz , by a matrix
multiplication similar to Eq. (1.65), such that
 
 1   1 
 
K 0  K 
 x 
 Ky 0  = S  Ky  .
 x (8.8)
 0   
Kz Kz
190 Chapter 8

If the vertex angle ϕ exceeds a certain value, then incident normal rays can
be directly refracted through either surface 2 or surface 3. This happens when
the angle of incidence I at surfaces 2 and 3 is less than the critical angle Icrit ,
as in Fig. 8.2. This occurs when ϕ > 180 deg − 2Icrit , e.g., ϕ ≥ 96 deg for
n = 1.5. When either ρ2 > cos I2crit or ρ3 > cos I3crit , there is TIR failure, and
light is refracted through the array. The range of incident angles I1 can then be
calculated that define the boundary regions in which reflection or transmission
occur. Figure 8.3 illustrates a hemispheric model1,2 in which an oblique ray
incident at point (x1 , y1 , z1 ) of planar surface 1 passes through the hemispheric
surface having a radius R0 . The intersection coordinates at the hemispheric surface
(x0 , y0 , z0 ) are given by x0 = R0 sin I1 sin ω, y0 = R0 sin I1 cos ω, and z0 = R0 cos I1 ,
where R0 , I1 , and ω are defined as shown.
If x1 = y1 = z1 = 0, then the direction cosines for an incident oblique ray are
calculated from

K x = (x1 − x0 )/R0 = −sin I1 sin ω, (8.9a)

Ky = (y1 − y0 )/R0 = −sin I1 cos ω, (8.9b)
Kz = (z1 − z0 )/R0 = −cos I1 . (8.9c)

If n = 1.5, then ρcrit = 1/n = 0.6667, or Icrit = 41.81 deg.

The most frequent uses for a roof prism array have ϕ = 90 deg. The normal
direction cosines for the reflecting facets 2 and 3 are given by k x2 = 0, ky2 =
cos(45 deg), kz2 = −cos(45 deg), and k x3 = 0, ky3 = −cos(45 deg), kz3 =
−cos(45 deg). For the case where a ray undergoes TIR at both facets, the direction
cosines of the exit ray at planar surface 4 (same as 1), are calculated directly from
Eq. (8.8), using SA .

Figure 8.3 Transmission and reflection regions of microprismatic array using a model
hemisphere.1
Microprismatic Arrays 191

Table 8.1 gives the resultant exit ray directions for various incident angles in
the reflection region. TIR occurs where I2 and I3 both exceed, or are close to, the
critical angle. Retroreflection occurs when I1 ≤ 4.7 deg and ω = 0 deg. Specular-
type reflection occurs for any value of I1 when ω = 90 deg, I1 = I4 , and the entrance
and exit angles are in the same plane. Skew reflections occur at other I1 and ω
values, where I1 = I4 , but the exit plane is rotated relative to the incident plane.
The resultant line of retroreflection and the boundaries between the reflection and
transmission regions are as shown in Fig. 8.4. The reflection/transmission boundary
is determined by the I1 and ω values for which I3 = Icrit . Rays with a high exit
angle, as in ray path B of Fig. 8.2, may not clear the adjacent groove in the array
and can be refracted out through the input surface. To fully evaluate these types of
rays in prismatic structures, the technique of nonsequential ray tracing is useful.
Several commercial nonsequential ray-tracing optical programs are listed at the
end of the references. For light that is directly transmitted by refraction, as in ray
path C, the range of output angles I5 , measured relative to the z axis, is reduced
from the range of input angles over a large portion of the transmission region.
This is referred to as normalized transmission, where the output angles are brought
closer to the array normal. This occurs when I1 ≥ 14.8 deg for various values of
ω. Table 8.2 gives some representative input and exit angles in the transmission
region. The nonnormalized/normalized transmission boundary is determined for
the I1 and ω values for which I5 ≈ I1 .

Figure 8.4 Square linear prismatic light guide with light source centered on input aperture.
(Adapted from Refs. 3 and 4 with permission from the author and the Optical Society of
America.)
 Table 8.1 90-deg roof-prism array in a reflection region (n = 1.5, Icrit = 41.81 deg, θmax = 27.6 deg).
192

Ray path I1 Entrance ω Kx I2 (deg) I3 (deg) Kx 0 I5 Exit xo

angle (deg) θ (deg) Ky Ky 0 angle (deg) yo
Kz Kz 0 zo
(R0 = 100 mm)
A0 0 0 00 45 45 00 0 00
90 00 00 00
Retroreflection −1 0 10 100 0
A0 4 78 0 00 48 18 41 82 00 4 78 00
90 −0 0833 reflection/transmission 0 0833 8 333
Retroreflection −0 9965 boundary 0 9965 98 65
A 6 98 45 −0 0859 48 37 41 82 −0 0859 6 98 8 593
85 1 −0 0859 reflection/transmission 0 0859 8 593
Skew reflection −0 9926 boundary 0 9926 98 26
A 89 55 −0 1267 48 59 41 82 −0 1267 89 12 67
82 7 −0 0887 reflection/transmission 0 0887 8 874
Skew reflection −0 9880 boundary 0 9881 98 80
A 13 5 65 −0 2116 49 31 41 84 −0 2116 13 5 21 16
77 8 −0 0987 reflection/transmission 0 0987 8 866
Skew reflection −0 9724 boundary 0 9724 97 24
A 45 68 6 −0 6584 59 87 41 82 −0 6584 45 65 84
48 8 −0 2580 reflection/transmission 0 2580 25 80
Skew reflection −0 7071 boundary 0 7071 70 71
A 63 8 65 −0 8132 67 49 41 81 −0 8132 63 8 81 32
35 6 −0 3792 reflection/transmission 0 3792 37 92
Skew reflection −0 4415 boundary 0 4415 44 15
A 88 9 62 45 −0 8866 72 0 41 83 −0 8866 88 9 88 66
27 6 −0 4625 reflection/transmission 0 4625 46 25
Skew reflection at θmax −0 0018 boundary 0 0018 0 1745
A 45 90 −0 7071 51 42 51 42 −0 7071 45 70 71
45 00 00 00
Specular-type reflection −0 7071 0 7071 70 71
Chapter 8
 Table 8.2 90-deg roof-prism array in a transmission region (n = 1.5, Icrit = 41.81 deg, θmax = 27.6 deg).
Microprismatic Arrays

Ray path I1 ω Kx I2 (deg) I3 (deg) Kx 0 I5 xo

Entrance θ (deg) Ky Ky 0 Exit angle yo
angle (deg) Kz Kz 0 from z-axis (deg) zo
(R0 = 100 mm)
B 10 0 00 51 64 −38 36 00 00
−66 44
90 −0 1736 TIR transmission −0 9167 17 36
transmission blockage
−0 9848 0 3997 98 48
C 10 0 00 misses 38 35 00 23 55 00
90 −0 1736 0 3996 nonnormalized 17 36
−0 9848 −0 9167 transmission 98 48
C 14 8 0 00 misses 35 19 00 14 8 00
90 −0 2554 0 2559 normalized transmission 25 54
−0 9668 −0 9668 boundary 96 68
C 21 8 45 −0 2626 misses 36 0 −0 2626 21 8 26 26
74 8 −0 2626 0 2621 normalized transmission 26 26
−0 9285 −0 9286 boundary 92 85
B 45 0 00 73 13 −16 88 00 00
−71 11
90 −0 7071 TIR −0 9444 70 71
transmission blockage
−0 7071 −0 3287 70 71
(continued on next page)
193
 194

Table 8.2 (continued)

Ray path I1 ω Kx I2 (deg) I3 (deg) Kx 0 I5 xo
Entrance θ (deg) Ky Ky 0 Exit angle yo
angle (deg) Kz Kz 0 from z-axis (deg) zo
(R0 = 100 mm)
C 45 0 00 misses 16 88 00 00
19 20
90 −0 7071 −0 3287 70 71
normalized transmission
−0 7071 −0 9444 70 71
C 45 64 2 −0 6366 misses 39 37 −0 6366 63 66
45
50 5 −0 3078 0 3082 30 78
normalized boundary
−0 7071 −0 7069 70 71
B 85 0 00 86 62 −3 38 00 00
−50 08
90 −0 9962 TIR 0 6641 99 62
transmission blockage
−0 0872 0 7476 8 716
C 85 0 00 misses 3 38 00 00
38 92
90 −0 9962 −0 6417 99 62
normalized transmission
−0 0872 −0 7669 8 716
B 88 9 grazing 62 4 −0 8862 72 02 −41 80 −0 8862 −70 14 88 62
27 6 −0 4633 TIR reflection/transmission −0 3427 46 33
−0 0018 boundary 0 3121 0 1745
C 88 9 grazing 62 4 −0 8862 misses 41 80 −0 8862 88 62
70 14
27 6 −0 4633 reflection/transmission 0 3150 46 33
normalized transmission
−0 0017 boundary −0 3397 0 1745
Chapter 8
Microprismatic Arrays 195

The (x0 , y0 , z0 ) values on the hemisphere defining these boundary regions are
calculated using Eqs. (8.9a) to (8.9c) and a reasonable value of R0 . Figure 8.3
shows the resulting regions on the model hemisphere. Here θmax indicates the
maximum ray angle θ, relative to the groove direction (x axis), for which a ray
will be reflected by double TIR for any angle of incidence I1 from 0 to 90 deg. In
general, θ = arccos(sin I sin ω). However, there are angles of incidence < 90 deg
for which θ > θmax . Tables 8.1 and 8.2 give representative data for the reflection
and transmission regions. For an extended array, both the reflection region and the
transmission region are useful for several types of optical illumination devices.
The prismatic array can be a rigid sheet (about 4–6 mm thick), where the prismatic
triangular grooves are precisely replicated in optical plastic with a typical period
width of about 3 mm.
In another type of array, a flexible film (about 0.5 mm thick) can be produced
where the prismatic triangular grooves have a typical period width of about
0.25 mm. The optical plastic is usually acrylic, with nd = 1.4918, or polycarbonate,
with nd = 1.5855. Polycarbonate is more impact resistant, but acrylic has a lower
dispersion. The flexible prismatic film is produced and sold by the 3M Company
as 3M optical lighting film (OLF) in several configurations.

8.2 Square Prismatic Hollow Light Guide

One of the first applications of a 90-deg prismatic array sheet in the reflection
region was for a prismatic hollow light guide (also called a light conduit or light
pipe). For the configuration in Fig. 8.4, four identical rigid sections of the array are
arranged to form a square hollow structure, with the grooves facing outward and
running along the guide.3 The sections can be molded from acrylic plastic, typically
1 to 10 mm thick. A light source with a controlled beam spread is positioned at one
end, and the light rays are transported down the guide by total internal reflection
from the prismatic surfaces. The maximum beam half-angle θmax , as calculated by
Whitehead et al.3,4 is
s
1 − n2 sin2 (22.5 deg)
θmax = arccos . (8.10)
1 − sin2 (22.5 deg)

For n = 1.5, θmax = 27.6 deg, corresponding to an f /1.08 beam.

Light guides of this type are normally used to efficiently transport light from
the input end of the guide to the exit aperture of the guide, or to provide extended
illumination through an area along the length of the guide. In the first case, high
reflectance must be maintained with minimum light loss through the walls. In the
second case, the light must be efficiently transmitted through the guide wall over
the desired illumination area. The transport efficiency of a light guide depends on
several factors, such as
• Imperfections of the array surfaces,
• Dielectric absorption loss in the array material,
196 Chapter 8

• Fidelity of the prism geometry—e.g., peak sharpness and surface flatness,

• Dust and contamination on the optical surfaces.
Considering these factors, calculation of the expected light throughput of a light
guide is an approximation at best. In general, for maximum transport of light, the
base length of the microprisms should be much greater than the distance of the
microprism base to the inner planar surface, and the cross-sectional area of the
four walls should be much less than the light-guide cross section. Also, a practical
light intensity for transport through the light guide is on the order of 10 W/cm2 .
In Fig. 8.5, a light transporter with an input beam at one end of the guide produces
output at the other end. The light guide can also be encased in a reflective sheath,
such that any escaping light can be redirected back through the wall to increase the
transport efficiency.
The second case provides more practical applications for illumination devices.
Normal light leakage through the walls causes the light guide to glow. To increase
the transmitted light over a specified area for use as an illuminator, an extractor is
used, where the extractor is any element or modification that allows light to escape
from the guide that would normally be reflectively contained inside. It can be as
simple as a tapered cutout or an array of patterned holes along a single wall. The
problem then becomes how to maximize the extraction of light along a selected
linear area on the guide, with close to uniform intensity along the illuminated
length. One method is to place diffusing screens on the wall interior. This scatters

Figure 8.5 A high-throughput prismatic light guide.

Microprismatic Arrays 197

the light transmitted to the prismatic wall, and light is then transmitted. Several
other methods to extract light are rounding the peaks of the prismatic grooves,
curving the prism facets, modifying the prism vertex angle, or increasing the
divergence half-angle θ of the light in the guide, such that θ > θmax . The divergence
angle θ can be increased by adjusting the light-source module angular output,
adding interior diffusion screens perpendicular to the guide axis, or placing a
convex mirror at the end of the guide.5 Figure 8.6 illustrates a typical configuration
for a square light guide luminaire, with a controlled input light module at one end,
a tilted reflector at the other end, planar diffuse reflectors surrounding three walls
to redirect transmitted light back into the guide, and a diffusing plate along the
bottom wall to even out the extracted light.

8.3 Circular Prismatic Hollow Light Guide

Circular prismatic hollow light guides can also be constructed using flexible 3M
optical lighting film, usually formed from acrylic or polycarbonate material. These
light guides (sometimes called light tubes) have certain advantages over the square
prismatic light guide. For the square light guide of Fig. 8.4, rays from a centered
point source are limited in reflection by the angular divergence θmax . For example,
an oblique light ray hitting the wall at I1 = 45 deg, θ = 55 deg, ω = 54.2 deg
(n = 1.5), at the position shown, will be transmitted through the prismatic wall
by TIR failure. For the circular prismatic light guide in Fig. 8.7, all rays from
a centered point source produce an angle of incidence I1 that lies in a plane
with the surface normal. All reflected rays then remain in the reflection region
and undergo specular-type reflection, where ω = 90 deg and cos θ = sin I1 .
Since the angle of incidence can vary from I1 ≈ 0 deg at the input end of the
guide to I1 ≈ 90 deg at the end of an infinitely long tube, θmax can approach 90
deg at the input, with little light loss through the wall. The angles of incidence

Figure 8.6 A square prismatic light guide luminaire.

198 Chapter 8

Figure 8.7 A circular prismatic light guide with sample light-ray paths.6

remain constant for subsequent reflections, significantly increasing the transport

efficiency of the guide. However, for a high-divergence-angle incident beam, the
number of interior reflections increases over the length of the tube, producing more
attenuation through the dielectric material. In practice, the input light consists of a
finite-size source with an auxiliary reflector, e.g., parabolic or elliptical, smooth
or segmented. To ensure that the majority of rays are internally reflected, the
divergence half-angle is often controlled to be about 30 deg, even for a circular
light guide.
For a flexible prismatic film that is curled into a cylinder, there is a minimum
practical diameter Dmin that can be formed. For a film having a thickness
T measured from the smooth side to the prism vertex, this diameter can be
approximated as Dmin ≈ CT, where C is a constant related to the modulus of
elasticity. For example, a 0.4-mm-thick acrylic microprismatic film having about
3 prisms per millimeter can be formed into a cylinder with a minimum diameter of
about 80 mm. This film can be self-supporting up to a diameter of about 450 mm.
Film curvature also affects the reflected angle from total internal reflection at the
prisms. The change in the reflection angle, compared to a planar array, is estimated
to be about ±1.3 deg.6
If maximum light transport is desired, the prismatic grooves must have a precise
geometry, and the dielectric must be of the highest clarity to minimize internal
scattering. The transported light flux depends on the type of light source and any
auxiliary reflector and on the material of the guide. Figure 8.8 plots the amount
of light transported down tubes of varying length-to-diameter ratios (sometimes
called aspect ratio or tube length in diameter units) for a polycarbonate tube.7 The
maximum recommended aspect ratio is ≈ 60. Data points are calculated from a
measured transport factor TF = 23.2 diameter/dB, using a tungsten source with a
divergence half-angle ≈ 30 deg.2 The light loss in decibels = 10 log (Φ/Φ0 ), where
Φ0 is the initial beam flux, and Φ is the flux at a specified distance down the tube.
For example, for a tube with a 15-cm diameter, the transported flux will be ≈ 80%
Microprismatic Arrays 199

Figure 8.8 Transported light intensity in circular microprismatic light guide.7

(−1-dB loss) for a tube ≈ 3.5 m long. When used for maximum light transport, the
curved microprismatic film is often directly encased in a diffusely reflecting rigid
cylinder. This preserves the cylindrical shape of the film and directs any escaping
light from the guide back into the tube, where it can be further transported. Once
the light is transported from one remote location to another, the light is often used
for illumination. Having the light source in a remote location allows control of
lamp changing and maintenance, heat buildup, and spectral properties of the light.

8.4 Luminaire with Contoured Prismatic Extractor

The light guide of Fig. 8.9 is used as a luminaire with a contoured prismatic
extractor surface and an internal linear diffuse reflecting element.8 Ideally, the
light source emits light with ray angles 0 < θ < 50 deg relative to the guide x
axis. Figure 8.10 shows a cross section of the light guide, where the rays shown
are the projections of the rays on the x-y plane. It has been determined through
measurement and computation analysis that light is transmitted most efficiently
through the guide wall when projection of the light ray on a plane perpendicular to
the axis of the guide forms an angle γ ≈ 60 deg with the normal of the inner surface
of the extractor surface. Table 8.3 gives typical rays that yield γ ≈ 60 deg, where
I, ω, and θ are as previously defined, and ϕ is the slope angle of the contoured
smooth inner surface. Here, IP is the projected angle of I in the y-z plane, given by
IP = arctan(tan I cos ω) and ϕ = γ − IP .

8.5 Elliptical Light Guide with Directional Output

Consider the elliptical reflective cylinder with major and minor axes a and b, as in
Fig. 8.11. The two foci are directed along the length of the cylinder, and the inside
of the cylinder is either coated with a highly reflective coating or is formed from
200 Chapter 8

Figure 8.9 A microprismatic light guide luminaire with a contoured extractor.8

Figure 8.10 Cross section of contoured light guide luminaire.8

Table 8.3 Some optimum light guide wall-transmission ray angles (γ ≈ 60 deg). Units are
in degrees.
I ω θ IP ϕ
60.0 85.0 30.4 8.58 51.4
65.0 85.0 25.5 10.6 49.4
60.0 75.0 33.2 24.1 35.9
70.0 80.0 22.3 25.5 34.5
60.0 60.0 41.4 40.9 18.1
75.0 75.0 21.1 44.0 16.0
85.0 85.0 7.1 44.9 15.1
Microprismatic Arrays 201

Figure 8.11 Focal properties of an elliptical reflective cylinder.

a highly reflective film such as 3M Silverlux. The focus coordinates are given by
±x f , where
√
xf = a2 − b2 . (8.11)

For an oblique ray originating at either focus and intersecting the reflective ellipse
at (x1 , y1 , z1 ), the slope ϕ of the surface is
 
 
 
bx1
ϕ = arctan  r  .
 
(8.12)
2

 a2 1 − x 1


a2

Then, the direction cosines of the surface normal are

k x1 = sin ϕ, (8.13a)
ky1 = cos ϕ, (8.13b)
kz1 = 1. (8.13c)

The distance D f 1 from the focal point to the surface intersection is

q
Df1 = (x1 − x f )2 + y1 2 + z1 2 , (8.14)

and the direction cosines of the incident ray are

(x1 − x f )
Kx f 1 = , (8.15a)
Df1
y1
Ky f 1 = , (8.15b)
Df1
z1
Kz f 1 = . (8.15c)
Df1
202 Chapter 8

Then, the direction cosines of the reflected ray are

K x12 = K x f 1 − 2ρk x1 , (8.16a)

Ky12 = Ky f 1 − 2ρky1 , (8.16b)
Kz12 = Kz f 1 − 2ρkz1 , (8.16c)

where ρ = K x f 1 k x1 + Ky f 1 ky1 + Kz f 1 kz1 .

The ray angle θ is given by θ = arccos Kz12 , and the intersection coordinates
(x2 , z2 ) of the reflected ray with the y-z plane (y2 = 0) are
!
K x12
x2 = x1 − y1 , (8.17a)
Kz12
!
Kz12
z2 = z1 − (x2 − x1 ). (8.17b)
K x12

These calculations show that x2 = −x f , or that any skew reflected ray originating
at x f will always intersect the second focal line at a distance z2 from the origin.
Several cases are shown in Table 8.4, where the linear dimensions are in arbitrary
units.
Table 8.4 Ray intersections for an elliptic reflective cylinder.
a b xf x1 , y1 , z1 θ (deg) x2 , y2 , z2
4 3 2.65 2, 2.60, 10 14.99 −2.65, 0, 28.88
4 3 2.65 3, 1.98, 5 21.96 −2.65, 0, 18.84
10 5 8.66 4, 4.58, 10 33.17 −8.66, 0, 30.60
10 5 8.66 8, 3, 5 31.56 −8.66, 0, 32.55
4 3 −2.65 2, 2.60, 10 28.03 2.65, 0, 15.02
4 3 −2.65 3, 1.98, 5 50.12 2.65, 0, 6.68
10 5 −8.66 4, 4.58, 10 53.40 8.66, 0, 14.85
10 5 −8.66 8, 3, 5 73.54 8.66, 0, 5.91

This result can be used for the design of a light-guide luminaire with directional
output.9 In Fig. 8.12, a half-elliptic reflective cylinder has linear diffusely reflecting
extractors running along the focal lines. Light scattered from either extractor is
reflected from the elliptical wall toward the other extractor, where it is again
scattered. Scattered or reflected light rays that hit the linear prismatic array exit
aperture are normalized on transmission. γ is the angle between normal to the
prismatic array (n = 1.5) and the projected angle of the incident ray with a plane
perpendicular to the axis of the luminaire.

8.6 Prismatic Backlighting Devices

Backlighting devices often use linear prismatic arrays to extract or transmit light
over a larger rectangular area in a low-profile enclosure. They have applications for
back-illuminated advertising displays and liquid crystal display (LCD) illuminators
for computer screens and television. In Fig. 8.13(a), a low-profile light fixture
Microprismatic Arrays 203

Figure 8.12 A linear elliptic luminaire with linear directional output from a microprismatic
extractor.9

provides directional or normalized output using a linear 90-deg microprism planar

array.1 A 90-deg linear microprismatic array element is positioned above a series
of tubular tungsten filament or fluorescent lamps, with the grooves parallel to the
lamp axis. An optional clear or diffusely transmissive element can be positioned
above the prism array. A planar diffuse or specular reflector is positioned below
the lamps. As shown in the cross-sectional view of Fig. 8.13(b), forward ray C is
directly transmitted by the prismatic array (I1 ≥ 15 deg for n ≈ 1.5) and brought
closer to the array normal. Forward ray A is retroreflected (I1 ≤ 15 deg for n ≈ 1.5)
and then reflected back to the prismatic array, where it is normally transmitted.
Back rays B and D undergo planar reflection and are then transmitted normally
through the prismatic array. A small percentage of rays, such as ray E, are not
normalized. An essentially directional output is thus achieved.
Another type of backlighting device uses a reflective linear 90-deg microprism
planar array.10 In a cross section of this microprism array [Fig. 8.14(a)], an incident
ray having a ray angle θ1 is refracted at facet 1, total internally reflected at facet 2,
reflected at the planar mirror surface 3, and refracted at facet 4 of the adjacent
prism with exit angle θ4 0 = θ1 . The ratio of element thickness t to prism height
h is t/h ≈ 3. The ray-deviation angle is δ = 90 deg. Figure 8.14(b) shows the
same prism array rotated by angle ϕ, at the same entrance-ray angle θ1 , with the
following ray-trace equations:

I1 = 45 deg + θ1 − ϕ, (8.18a)
!
sin I1
I1 = arcsin
0
, (8.18b)
n
I2 = 90 deg − I1 0 = I2 0 , (8.18c)
I3 = I2 0 − 45 deg = I3 0 (8.18d)
I4 = 45 deg − I3 0 , (8.18e)
I4 0 = arcsin(n sin I4 ), (8.18f)
θ4 = 45 deg − ϕ − I4 0 , (8.18g)
δ = 90 deg − θ1 + θ4 . (8.18h)
204 Chapter 8

Figure 8.13 (a) A low-profile linear prismatic illuminator with normalized output.1 (b) Cross
section of a low-profile illuminator.

The deviation angle δ for the array remains fixed at 90 deg, and θ1 = θ4 for all
θ1 and ϕ values. This also holds for any value of n that provides TIR at facet 2.
This reflective linear microprismatic array is therefore a wavelength-independent
constant 90-deg deviation element over a defined acceptance angle. The backlight
fixture in Fig. 8.15 uses a horizontal planar prismatic array with linear collimated
light sources at both ends. A transmissive diffusing plate is placed at the exit
window. Figure 8.16 shows a curved array with a single linear collimated light
source. The curvature improves the light uniformity at the exit window, and the
curvature profile is not critical. The backlight of Fig. 8.17 has an additional
reflective prismatic film that spreads the light from a collimated point source to
an extended beam directed to the curved reflective linear prismatic array.
Microprismatic Arrays 205

Figure 8.14 (a) Linear reflecting microprismatic array with a 90-deg deviation angle. (b)
Tilted linear reflecting microprismatic array with a 90-deg deviation angle.

Figure 8.15 Dual-lamp backlight with flat reflective microprismatic array.10

206 Chapter 8

Figure 8.16 Single-lamp backlight with curved reflective microprismatic array.10

Figure 8.17 Single-lamp backlight with dual-reflective microprismatic arrays.10

The backlight fixture in Fig. 8.18 uses a curved transmitting linear microprism
extractor array to provide directional output over a large area.11 The prisms face
downward and utilize TIR to bend the light; this type of array is sometimes called
a turning film. The linear light source is substantially collimated and can be tilted
to produce an incident ray angle θ at the lower prismatic surface. The isosceles
microprisms have a vertex angle α and base angles β, and ϕ is the slope of the
curved array at any position along the curve. Figures 8.19 and 8.20 illustrate typical
ray paths, and the basic ray-trace equations are

I1 = θ + ϕ − α/2, (8.19a)
!
sin I1
I1 = arcsin
0
, (8.19b)
n
I2 = α + I1 0 , (8.19c)
I2 0 = I2 , (8.19d)
I3 = β − I2 0 , (8.19e)
I3 0 = arcsin(n sin I3 ), (8.19f)
δ = 90 deg − I3 0 + ϕ + θ. (8.19g)
Microprismatic Arrays 207

Figure 8.18 Low-profile backlight using curved refractive microprismatic array.11

Figure 8.19 Section detail of curved linear microprismatic extractor for backlight.

For example, if θ = 10 deg, ϕ = 0 deg, and α = 70.61 deg, the output is

essentially normal to the prismatic array, for nd = 1.5855 (polycarbonate), with
I2crit = 38.10 deg. The microprismatic array is dispersive.
The high-aspect light box of Fig. 8.18 has a length L, width W, and height
H. A semicircular curve is drawn from the upper edge of the box near the light
source to the lower edge of the box. The required radius of curvature R is given
by R = (W 2 + H 2 )/2H, where the center of curvature is located on the x axis. If
L = W = 300 mm, and H = 60 mm (5:1 box aspect ratio), then R = 780 mm.
A transmissive extractor film consisting of isosceles microprisms is bent along
this curve such that all collimated light rays with θ = 0 deg from the source
are intercepted. For light to emerge close to the normal of the exit window, the
vertex angle of each microprism must be varied as a function of the curve slope
angle ϕ = arcsin(Y/R), such that δ = 90 deg, or I3 0 = −ϕ. The exit window can
208 Chapter 8

Figure 8.20 Halfway point of curved linear microprismatic extractor (Y = L/2).

Table 8.5 Sample design parameters for a curved transmissive extractor. Units are in
degrees.
Y ϕ α β I1 I2 , I2 0 I3 0
0 0 75.06 52.47 37.53 52.46 0.0082
W/2 11.09 64.03 57.99 20.93 51.01 −11.10
W 22.62 52.08 63.96 3.42 48.92 −22.62

be a clear plate, a diffuse transmissive plate, printed graphics, or a transmissive

electronic display. Some sample design parameters are shown in Table 8.5.
Another type of low-profile backlighting device is shown in Fig. 8.21.12 A
solid light guide of optical plastic has a series of linear microprismatic grooves
molded into the lower surface. A linear light source is placed at one end, and
a planar reflector at the other. A highly reflective diffuse scattering surface is
positioned adjacent to the lower grooved surface. A diffuse transmitting element
is positioned adjacent to the guide upper surface. Light rays that undergo TIR at
the leading prism facet are directed to the upper surface where they are diffusely
transmitted. Rays that undergo TIR at the upper surface and are transmitted by the
forward prism facet are scattered by the diffuse reflector and reenter the guide to
be transmitted at the upper planar surfaces. Rays that undergo TIR at the trailing
prism facet can be further reflected by the upper surface of the end mirror until they
can be extracted by the prismatic array.
The backlighting device in Fig. 8.22 uses a solid rectangular light guide of
optical plastic with trapezoidal input prisms, miniature (10- to 12-mm diameter)
fluorescent lamps, and involute-shaped reflectors at each end.13 A series of
trapezoidal microprisms that function as TIR extractors are placed on the top
planar surface of the light guide. The input prisms and extracting microprisms are
optically cemented to the light guide or molded as an integral part of the guide.
The backlighting device in Fig. 8.23 has a tubular light source in a highly
reflective housing, with a horizontal prismatic array providing directional input into
a hollow rectangular light guide.14 The vertical sides of the guide have height H and
depth D, with an internal reflective surface chosen to be specularly reflective when
Microprismatic Arrays 209

Figure 8.21 A solid light guide luminaire with linear microprismatic array.12

Figure 8.22 Solid light guide backlight with trapezoidal prismatic array extractor.13

the aspect ratio H/D > 10 and diffusely reflective when H/D < 10. The inside
of the top side of the box has a specular, highly reflective surface. The rear face
of the light guide has a highly reflective (>85%) surface having a narrow scatter
dispersion angle between 5 and 15 deg. A suitable reflective material is radiant
light film embossed VM2000, available from 3M. The forward surface is a vertical
prismatic array that extracts the light from the light guide. A protective clear or
diffusely transmitting view plate is positioned over this prismatic array.

8.7 Brightness Enhancement for Liquid Crystal Displays

The design of backlights for transmissive direct-view liquid crystal displays has
been a major area of design research. One of the components in many of these
backlights is a microprismatic array that enhances the viewing brightness in an
angular range close to the display normal. Such a backlighting assembly uses a
serpentine-shaped tubular fluorescent light source, a 90-deg linear prismatic film,
and a back reflector to redirect stray radiation in the direction of the LCD.15
210 Chapter 8

Figure 8.23 Light box backlight with vertical and horizontal linear microprismatic arrays.14

The concept of brightness enhancement derives from the normalized prism array
as shown in Fig. 8.13(b), where a means is provided to efficiently recycle light
that is reflected downward from the microprismatic array. For a high-reflectance
(>98%) back reflector, any direct or reflected light is redirected to the prismatic
array, usually at a different angle. The transmitted light then increases the forward
brightness. Figure 8.24 shows a direct-view LCD display backlight using a series
of cold-cathode fluorescent tubes (CCFTs) with high-reflectance-scattering lower
and side reflectors. A transmissive diffusing plate, a brightness-enhancement 90-
deg microprismatic array film, and a transmissive LCD are positioned above the
light guide. The linear microprisms run in the same direction as the fluorescent
tubes. Upward light is scattered by the diffusing plate, and a portion is transmitted
by the microprismatic array. About 50% of the light hitting the array is reflected
downward. The downward- or side-directed light is scattered and redirected to
the array. Light transmitted by the array is normalized, resulting in brightness
enhancement close to the normal axis of the display.
These prismatic arrays are sold commercially in the form of a flexible film by
3M as VikuitiTM brightness-enhancement film (BEF). For handling and durability,
the approximately 25-µm-thick UV-cured acrylic prism array is laminated to a 250-
µm polyester film material, PET. In one type of BEF, designated BEFII 90/50, there
is a fixed 90-deg prism apex angle and a fixed peak-to-peak separation (often called
pitch) of 50 µm. The 90-deg BEF yields the maximum gain, typically about 1.6, for
a view direction normal to the array, and about 1.2 at a view angle of ±30 deg, with
a fairly sharp cutoff at wider view angles. Increasing the prism angle increases the
view angle, but reduces the gain, while decreasing the prism angle reduces both
the view angle and the gain. Embossing a matte surface on the planar side of the
Microprismatic Arrays 211

Figure 8.24 LCD backlight with a brightness-enhancement linear microprismatic array.

BEF also gives a wider view angle, but reduces gain. Adding another sheet of BEF
near and orthogonal to the first sheet enhances the brightness in crossed viewing
directions.
The BEF prismatic structure may be modified to yield a softer cutoff, as
described by O’Neill and Cobb.16 One type of structure is shown in Fig. 8.25.
Here the 90-deg vertex angle is maintained, but the valley angles are varied such
that alternate valley angles have values of 70 deg and 110 deg. This results in
an alternating groove pitch. Figure 8.26 plots measured values of luminance for
various view angles, for a typical backlight suitable for liquid crystal displays.
Curves are shown for no BEF, 90/50 BEF, and a soft-cutoff BEF.
When two identical periodic structures are positioned close to each other, an
undesirable moiré pattern can be observed if the structures are slightly displaced.
This can occur in BEF, for example, from a reflection of the prismatic surface from
the planar surface of the BEF, or from other planar surfaces if elements are closely
stacked. It is possible to minimize this effect by varying the peak pitch P between
groups of microprisms or making the structured surface nonperiodic.17 Figure 8.27
shows one type of peak pitch variation where each group of prisms maintains a
constant peak pitch, and the fixed peak pitch P of any group can have a value

Figure 8.25 Groove structure of brightness-enhancement film with soft cutoff.16

212 Chapter 8

Figure 8.26 Measured luminance of backlights for various view angles.16

Figure 8.27 Groove structure of brightness-enhancement film for moiré pattern

reduction.17

between 20 and 60 µm. Moreover, the number of peaks per group varies between 2
and 20. A typical configuration would include groups with peaks spaced at 50, 40,
30, and 20 µm.
If all of the peaks of the BEF lie on a common plane and are in contact
with another planar surface, as in a closely stacked array, the area near the
peaks may be optically coupled to the planar surface, and the prism TIR is
frustrated in this region. This is sometimes called wet out and causes a noticeable
mottling and streaking in the display. Campbell et al. describe a method of
manufacturing a BEF that has a continuously varying peak height along the
length of a groove.18 This is accomplished by mounting the cutting tool on the
diamond turning machine to a fast tool servo actuator, with white-noise input
Microprismatic Arrays 213

Figure 8.28 Brightness-enhancement film with microprisms of varying height.18

filtered by a bandpass filter transmitting 4 to 5.6 kHz. Figure 8.28 illustrates a BEF
structure having this geometry where the valleys also vary, and the peak height
is preferably between 4% and 8% of the average peak height measured from the
planar side. The average period of the variations is preferably between 5 and 16
times the nominal structure height. A 90-deg peak angle can be maintained to
provide maximum brightness gain. In addition to reducing the optical coupling,
the varying-height prisms mask small cosmetic defects introduced during the
manufacturing process. A brightness-enhancement film with a random prismatic
structure is available commercially from 3M, designated as VikuitiTM BEF III-
10T. A typical LCD TV may contain from 4 to 40 CCFLs, depending on the
screen size. When BEF is used in this application, it is normally oriented with the
grooves in the vertical direction. For thermal stability at larger-screen TV sizes,
the BEF thickness is increased over that used on smaller screens, such as computer
monitors.
There are other types of brightness-enhancement arrays that use different
geometries or have different applications. Figure 8.29 illustrates a brightness-
enhancement film for a display that is front illuminated at a small angle θ from the
display plane, often called sidelighting.19 The film has spaced microprisms with
base angles α and β and a prism height h, with a distance d between prism peaks.
For a pixilated display—e.g., LCD—there should be at least two prisms per pixel,
and no blockage by the adjacent prism. Light is reflected by TIR at a prism face,
is diffusely scattered by a planar diffuse reflector beneath the pixel, and emerges
from the film at angles θ0 from 0 deg to about ±30 deg. Recommended ranges for
the design parameters are

0 deg ≤ θ ≤ 5 deg, (8.20a)

85 deg ≤ α ≤ 90 deg, (8.20b)
40 deg ≤ β ≤ 50 deg, (8.20c)
214 Chapter 8

5 µm ≤ h ≤ 20 µm, (8.20d)
10h ≤ d ≤ 29h, (8.20e)

where α ≈ 90 deg − θ.
Another type of brightness-enhancement film for a backlit display uses a linear
array of prismatic structures in the x-z plane with curved faces in the y-z plane,
forming a lenticular array (Fig. 8.30).20 This array is optically coupled to a solid
light guide such that the curved microprisms refract in two directions and provide
normalized light output over a 2D range.

Figure 8.29 Brightness enhancement of front-illuminated LCD.19

Figure 8.30 Brightness-enhancement film using curved prismatic facets.20

Microprismatic Arrays 215

Another brightness-enhancement film consists of a linear array of solid light-

extracting structures having curved sides approximating a parabolic shape.21
These microstructures are based on the compound parabolic concentrator (CPC)
developed for efficient collection of radiant energy. Figure 8.31(a) illustrates the
operation of an individual light extractor. Two opposite parabolic-shaped surface
sections 1 and 2 are canted with the focal points F1 and F2 , defining the input
aperture boundaries. A specified fan of rays through the aperture edge F1 , and
at ray angles ϕ relative to the z axis, undergoes TIR at the parabolic surface and
refraction at the planar output aperture. This defines the maximum beam angle θmax
that is incident on the output aperture, where typically 10 deg ≤ θmax ≤ 30 deg.
If a ray is traced from the input aperture edge F1 to the opposite edge of the
output aperture, the minimum value of a ray angle ϕmin that will be reflected by
the parabolic surface is determined. Smaller values of ϕ will be directly refracted
at the output aperture at an angle of incidence ≤ θmax . This film array is optically
coupled to a solid light guide, preferably of the same optical material [Fig. 8.31(b)].
A linear lamp and curved reflector inputs light into one end of the guide, with a

Figure 8.31 (a) Detail of compound parabolic extractor element. (b) Solid light guide
backlight with compound parabolic linear array extractor.21
216 Chapter 8

planar reflector at the other end. Multiple reflections within the light guide result
in light entering the entrance apertures, with TIR at either parabolic-shaped side
and refraction at the output aperture. Typically, the pitch p is between 10 µm and
200 µm, the height h is between 10 and 100 µm, and the ratio of output aperture
to input aperture is between 1.5 and 10.0. Factors to consider in the design of this
structure are the preservation of TIR at the parabolic surfaces, the desired value of
θ, and ease of manufacture.

8.8 Polarizing Prismatic Sheet

It is possible to produce partially polarized light for collimated light over an
extended area by the use of two prismatic arrays 1 and 2, as shown in Fig. 8.32.22
The prismatic arrays have vertex angles α1 and α2 and are separated by an air gap.
Unpolarized light rays entering the planar surface are incident on both prismatic
structures at Brewster’s angle, causing some of the s-polarized light to be reflected
to the side. The prismatic sheets have refractive indices n1 and n2 , and n0 = 1.0 for
the air surround. Then,
!
n0
I1brew = arctan , (8.21a)
n1
α1 = 180 deg − 2I1Brew , (8.21b)
!
n1
I2brew = arctan , (8.21c)
n0

Figure 8.32 Production of partially polarized light by prismatic sheets.22

Microprismatic Arrays 217

" ! #
n1
I2 = arcsin
0
sin I2brew , (8.21d)
n0
α2 = 180 deg − 2I2 0 . (8.21e)

If n1 = n2 = 1.53, then I1Brew = 33.17 deg, I2Brew = 56.83 deg, α1 = α2 =

113.7 deg, I2 0 = I1Brew , and I1 0 = I2Brew . Since some of the s-polarized light is
removed at the two prismatic surfaces, the light transmitted through the upper
planar surface is partially p-polarized, and in the same direction as the incident
light.

8.9 Prismatic Reflective Polarizer Film

By combining two 90-deg prismatic arrays and a repeating series of high and
low refractive index film pairs (often called a MacNeille pair) deposited on one
of the prismatic surfaces, a retroreflective polarizing film can be constructed.23
Figures 8.33(a) and (b) illustrate this construction. The composite film stack
is captured between two prismatic arrays by optically cementing one array to
the stack. For broadband visible light, a suitable high-index material is titanium
dioxide (nH = 2.2 to 2.5), and a suitable low-index material is silicon dioxide
(nL = 1.45). In practice, there may be three optical stacks, each containing six
MacNeille pairs, the first stack having a quarter-wave thickness centered at 400 nm,
the second centered at 550 nm, and the third centered at 700 nm. A suitable
prismatic-array substrate is polycarbonate (nd = 1.586), with a commercial optical
cement having an index nA = 1.56. Typical reflective polarizer film thickness
is 130 µm for smaller displays to 550 µm for large displays, such as LCD TV.
Figure 8.34 gives typical performance of the polarizing film. The transmitted p-
polarized light is ≈ 80%, and the reflected s-polarized light is ≈ 96% over the
visible spectrum. This reflective polarizer film is sold by 3M as VikuitiTM dual-
brightness-enhancement film (DBEF).

8.10 LCD Backlights Producing Polarized Light

An LCD usually consists of a matrix of liquid crystal material captured between
absorptive polarizing sheets. If the incident light can be prepolarized in the
polarization direction of these absorptive sheets, the transmission of the LCD can
be significantly increased, resulting in increased viewing brightness. A reflective
polarizer sheet can be utilized in a backlight to prepolarize almost all of the
incident light entering an LCD display. Figure 8.35 shows an LCD backlight
that uses a bank of fluorescent lamps. Unpolarized light (designated as s + p)
emitted through the diffuse transmitting window strikes the reflective polarizer
where the p-polarized light is transmitted and the s-polarized light is reflected. The
p-component passes through the oriented lower LCD polarizer with no significant
absorption. The reflected s-polarized radiation is depolarized by the lower
diffuser/reflector, and this unpolarized light again strikes the reflective polarizer.
Unpolarized light that strikes the lower diffuser/reflector directly from the lamps is
218 Chapter 8

Figure 8.33 (a) Polarizing thin-film stack between 90-deg microprismatic sheets. (b)
Retroreflective polarizing film. (Adapted from Ref. 22 with permission from The Society for
Information Display.)
Microprismatic Arrays 219

Figure 8.34 Typical performance of retroreflective polarizing film.23

Figure 8.35 Backlight-producing polarized-light output using reflective polarizing film.

also directed toward the reflective polarizer. This continuous polarization recycling
and conversion to p-polarization yields a significant brightness increase in the
viewed display, where the on-axis brightness is approximately doubled.
Another type of polarized backlight for LCD panels is shown in Fig. 8.36.24 A
linear light source is coupled to a series of solid light pipe slabs that are optically
isolated from each other. TIR within each section tends to collimate the light
entering a solid light guide, where the light is extracted upward by TIR at an array
220 Chapter 8

Figure 8.36 Polarized backlight for LCD panel using solid light guide with TIR microprism
array.24

of linear prisms on the lower surface. A reflecting polarizing sheet is placed at

the top surface of this light guide. The reflective polarizer sheet is oriented with
the microprism direction at a 45-deg angle to the microprism direction of the lower
TIR array. If the apex angle of the TIR array ≈ 90 deg, then the reflected s-polarized
light undergoes two 45-deg phase changes during TIR at the two prism faces,
and the upward-reflected light is circularly polarized as it reenters the reflective
polarizer sheet. The p-polarized light is again transmitted, and the s-polarized
light is reflected back toward the lower TIR prismatic array. Thus, the light is
continuously recycled such that the p-polarized light entering the lower polarizer
of the LCD panel is strengthened. The lower polarizer of the LCD panel has its
polarization axis oriented parallel to that of the reflective polarizer sheet.
Several types of polarized backlights utilize microstructured anisotropic layers.
Figures 8.37(a) and (b) illustrate a solid light guide with a machined array of
linear prismatic grooves on the top surface.25 The light guide is an acrylate, with
nguide = 1.48. The grooves are filled with a liquid crystal birefringent polymer
having an ordinary refractive index no = 1.51 and an extraordinary refractive index
ne = 1.65, with directions as indicated. Unpolarized light striking prism surface 2
at angles ≥ I e crit (≈64 deg) undergoes TIR and is extracted from the backlight as
s-polarized light. The transmitted p-polarized light is essentially undeviated. The
prism apex angle α = 90 deg, and prism height = 10 µm, with a groove spacing of
100 µm. The depth of the light guide is ≈ 2 mm. For this geometry, the s-polarized
light is not normalized but has a maximum intensity about 30 deg from the normal.
Another type of polarized backlight uses a similar solid light guide but has a
different prism extractor geometry and uses a commercially available birefringent
film.26 The film is a uniaxial-aligned liquid crystal polymer (LCP) from Dejima
Microprismatic Arrays 221

Figure 8.37 (a) Polarized backlight using birefringent polymer microprism array. (b) Detail
of groove structure for polarizing extractor linear microprism array. (Adapted from Ref. 25
with permission from The Society for Information Display.)

Optical Films. The microprisms can be formed onto the top surface of the light
guide by photoreplication from a master mold (sometimes called the 2P technique).
The LCP film is then pressed into the microprism array by a lamination process
as a relief structure, where the uniaxial alignment of the film is maintained.
Figure 8.38 illustrates the near-normal TIR extraction of s-polarized light by the
relief structure. The s-polarized light is estimated to be about 100 times more than
the p-polarized light over a ±10-deg cone from the top surface normal, while most
of the p-polarized light can be recycled within the solid guide using side and end
reflectors.
It is possible to produce a light guide emitting p-polarized light by the device in
Fig. 8.39(a).27 The backlight consists of a prismatic array of linear trapezoids that
have MacNeille-type polarization-separating thin-film stacks applied to the bases
and optically coupled to a solid light guide by an optical adhesive. The optimum
angle of incidence I at the film stack is about 64 deg at λ = 550 nm. Then, the
222 Chapter 8

Figure 8.38 Detail of relief structure of polarizing extractor linear microprism array.
(Adapted from Ref. 26 with permission from The Society for Information Display.)

base angle β of the trapezoid is (180 deg − I)/2 = 58 deg for the p-polarized
light to be reflected (TIR) from the side of the trapezoid and exit normal to the top
surface. In addition, the reflected s-polarized light from the stacks that strikes the
base and end of the light guide can be converted to p-polarized light by placing a
quarter-wave retarder over reflectors on the base and end sections. This converted
p-polarized light can then exit through the linear trapezoid array. Figure 8.39(b)
plots the transmitted p-polarized and s-polarized light as a function of the angle of
incidence I. In practice, the backlight can be efficiently used to produce emitted
p-polarized light about ±10 deg from the exit surface normal.

8.11 Prismatic Array Beamsplitters and Combiners

There are a number of applications where a linear prismatic array can be used
as a beamsplitter or a beam combiner for collimated light beams. Figure 8.40(a)
shows a prismatic array with vertex angle α used as a refractive beamsplitter
with deviation angles ±θ and 100% throughput. For certain input angles and
substrate materials, it is possible to design a refractive beam combiner with high
throughput. As shown in Fig. 8.40(b), for α = 52.1 deg, θ = ±45 deg, and
n ≈ 1.59 (polycarbonate), all input rays exit normal to the array. In the catadioptric
beamsplitter of Fig. 8.41(a), both refraction and total internal reflection deviate the
incident beam. The relationship between the angle θ, the prism vertex angle α, and
the material refractive index n is given by
( " !#)
α 3α
θ = 90 deg − + arcsin n cos . (8.22)
2 2

Some of the incident light is not deviated by ±θ but is misdirected within the
prismatic array.
Figure 8.41(b) illustrates a catadioptric beam combiner. Table 8.6 gives the
fraction of beam fill for various values of α and θ, for n = 1.492 (PPMA).
Microprismatic Arrays 223

Figure 8.39 (a) Polarized backlight using thin-film-stack linear microprism array. (b)
Transmission curves for polarized light through thin-film-stack linear microprism array.27
224 Chapter 8

Figure 8.40 (a) Refractive beam-splitting prismatic array. (b) Refractive beam-combining
prismatic array.

Table 8.6 Fraction of beam fill for a catadioptric beam combiner prismatic array.
α (deg) θ (deg) Beam-fill fraction
50.0 42.3 1.55
58.0 56.5 1.11
58.0 58.3 1.05
60.0 60.0 1.0
61 61.7 0.947
62.0 63.5 0.895
70.0 77.7 0.446
Microprismatic Arrays 225

(a)

(b)

(c)

(d)

Figure 8.41 (a) 70-deg catadioptric beam-splitting prismatic array. (b) 50-deg catadioptric
beam-combining prismatic array. (c) 60-deg catadioptric beam-splitting prismatic array. (d)
60-deg catadioptric beam-combining prismatic array.
226 Chapter 8

When the beam-fill fraction is greater than 1.0, some of the incident light
overspills the reflecting facet and does not exit normal to the array. When the
beam-fill fraction is less than 1.0, the reflected facets are underfilled, and although
all rays exit normal to the array, the exit beams from adjacent microprisms are
spatially separated. For the special case where α = 60 deg and θ = 60 deg,
the reflecting facets are exactly filled, and there is negligible refraction, as with
the 60-deg reflective beamsplitter shown in Fig. 8.41(c). For the 60-deg reflective
beam combiner shown in Fig. 8.41(d), the spatial integration is continuous, with
no separation between the individual transmitted microbeams.
Figure 8.42 shows an overhead projector that combines the light output from two
lamps using a 60-deg reflective beam combiner.28 Fresnel lenses 1 and 2 collimate
the light from each lamp and are oriented such that the entrance angles to the
reflective beam combiner is ±60 deg. The spatially integrated beams are focused to
the entrance pupil of the projection lens and directed to the screen by the folding
mirror. The projected screen illumination is approximately doubled from that of a
single lamp of equivalent wattage.

8.12 Polarization Converters Using Prismatic Arrays

A polarization converter is a device that separates an unpolarized light beam
into s-polarized and p-polarized beams, converts one of the beams to a common
polarization, and spatially integrates the beams into a common direction with
high throughput. Figure 8.43 illustrates. A polarization converter for illuminating

Figure 8.42 Overhead projector with dual lamps using 60-deg catadioptric beam-
combining linear microprismatic array.28
Microprismatic Arrays 227

Figure 8.43 Polarization-converting backlight using linear microprismatic arrays.29

an LCD panel uses beam-splitting and beam-combining linear prismatic arrays,

reflective polarizing film, a quarter-wave retarder film, and planar mirrors.29
Collimated light is incident on an entrance refracting beamsplitter with vertex angle
α ≈ 105 deg and refractive index n ≈ 1.58. The light is split into two beams at
beam angles θ ≈ ±24 deg. The bottom and end of the guide are planar reflectors. A
quarter-wave retarder is placed over the end reflector, and incident s-polarized light
becomes reflected p-polarized light. The reflective polarizing film is placed beneath
a top catadioptric beam combiner having a vertex angle α ≈ 63 deg and refractive
index n ≈ 1.58. The reflective polarizer transmits p-polarized light and reflects
s-polarized light, and the beam combiner normalizes and spatially integrates the
transmitted light through the top exit window. The dimensions of the box formed
by these components must be such that only s-polarized light from the reflecting
polarizer strikes the retarder. Also, all light rays must pass through the exit window,
with no light being returned to the entrance window. This condition is satisfied
when the height H and length L of the light guide satisfy the following:

tan θ = 2H/L, (8.23)

where the aspect ratio AR of the light guide is H/L. For θ ≈ 24 deg and H = 25
mm, the required length is L = 112 mm, with AR = 4.5.
Figure 8.44 illustrates a large-area planar polarization converter that utilizes
linear prismatic arrays, reflective polarizing film, and quarter-wave retarder film.30
Collimated unpolarized light enters the lower element at an entrance angle θ =
45 deg. The lower linear prismatic array consists of two linear 90-deg vertex angle
array structures on opposite sides of a clear substrate. The microprisms of the
lower surface have one face clear, while the opposite face has a reflective coating.
Incident light enters through the clear faces and passes through a quarter-wave
retarder sheet (e.g., Nitto-Denko–type NRF-QF03A) with principal axis direction
228 Chapter 8

Figure 8.44 Planar polarization converter using microprismatic arrays.30

as shown and strikes the reflecting polarizer sheet (e.g., VikuitiTM DBEF). The
transmitted p-polarized light passes directly to a refractive beam combiner with
vertex angle α = 52.1 deg, n = 1.58, and is emitted perpendicular to the planar
converter. The reflected s-polarized light passes through the retarder where it is
converted to elliptically polarized light. This elliptically polarized light is reflected
in the opposite direction by the reflective facets of the lower surface. After passing
through the retarder, the elliptically polarized light is converted to p-polarized light
and spatially integrated with the directly transmitted p-polarized light. The groove
pitches of the upper and lower element facets are typically 100 to 500 µm, while
the groove pitch of the TIR microprisms is typically 10 to 50 µm. With all elements
in contact, typical thickness of this polarization converter is between 6 and 8 mm.

8.13 Cube-Corner Arrays

Instead of a large-aperture single cube-corner, it is often desirable to form a larger
2D array of micro cube-corners, as shown in Fig. 8.45. In this configuration, each
cube-corner has an equilateral-triangle entrance aperture, with a typical cube depth
H ≈ 120 µm and typical triangular aperture altitude h ≈ 255 µm. These entrance
windows are hexagonal close-packed. Once a master die is produced, cube-corner
sheeting can be reproduced by microreplication. Figures 8.46(a) and (b) show the
geometry of an individual cube and the cross section of a line separating adjacent
cube-corners.31 The master die can be formed by machining three sets of linear
V-grooves, each set rotated by a 60-deg angle. This can be accomplished by
mounting a flexible machinable sheet on a drum-type diamond-turning machine,
with the sheet precisely rotated for each set of grooves. The included tool angle
β to produce the triangular cube-corner configuration can be calculated from the
Microprismatic Arrays 229

Figure 8.45 Cube-corner array die section.31

following relationships:
! !
β β L
sin 90 deg − = cos = , (8.24)
2 2 h
√
where the√ length of a cube edge L = h 6/3. Then the ideal tool angle β =
2 arccos( 6/3) = 70.5288 deg = 70 deg, 31 min, 43.7 sec. If the cube-corner
array is then replicated in an optical plastic sheet, the sheet can retroreflect by TIR
or by coating the back prism surfaces with a highly reflecting thin film. Types of
fabrication errors that can affect the divergence of the retroreflected beam are errors
in the 90-deg dihedral cube angles, mounting errors of the master substrate during
the 60-deg rotations, included tool angle error, and tool mounting error.
In some applications, such as traffic signs and warning displays, it is desirable
to further increase the viewing angle of reflected light over a conventional cube-
corner reflective array. This can be accomplished by a modification of the groove
structure of the master die.32 Figure 8.47(a) illustrates three sets of parallel V-
grooves, where two of the sets (A, B) intersect at an angle of 70 deg, and a third
set (C) intersects at an angle of 55 deg. The included tool angle for the (A, B)
set is βA,B = 60 deg, 36 min, and for the (C) set is βC = 88 deg, 51 min. This
produces “matched pairs” of cube-corner elements where the optical axis is tilted
with respect to the trisection of the internal base angle [see Fig. 8.47(b)]. When
the cube-corner array is replicated in an optical plastic of index n, the optimum tilt
angle is approximated by:

ϕ = 54.736 deg − arcsin(1/n). (8.25)

230 Chapter 8

Figure 8.46 (a) Geometry of single micro cube-corner. (b) Cross section of adjacent micro
cube-corners and required cutting tool angle.

For n = 1.49, ϕ = 12.6 deg, and for most optical plastics, ϕ varies between 12
and 13 deg. The resultant reflected angular half-brightness angle (50% falloff from
maximum) was found to be ≈ 40 deg in the x-axis plane and ≈ 35 deg in the y-axis
plane, for an array replicated in optical acrylic, with intermediate angles in other
perpendicular planes.
An alternative geometry for a cube-corner array uses hexagonal entrance
apertures instead of the triangular shapes described previously. This has the
efficiency advantage of using three full sides of the cube, instead of the truncated
sides of the triangle cube-corner (see Sec. 2.4). The percentage of the reflected
light can approach 100% of the incident light. Figure 8.48 shows a section of a
hexagonal cube-corner retroreflecting array. However, the manufacture of a master
die for a hexagonal cube-corner array presents challenges. One method described
by Brinksmeier et al. is a pin-building method where the top of each hexagonally
shaped pin has a precision-machined cube-corner.33 Pin alignment restricts the
Microprismatic Arrays 231

Figure 8.47 (a) Top section of wide-angle retroreflective sheet.32 (b) Perspective view of
tilted-axis micro cube-corner.

microcubes to a minimum size of about 500 µm. Another method is diamond

microchiseling (DMC), also described by Brinksmeier et al. Using this technique,
a cube microstructure of about 100 µm can be achieved.

8.14 Dove Prism Arrays

A type of 2D prismatic array that uses Dove prisms can be constructed to provide
a larger entrance aperture D relative to the length L. The single-roof Dove prism
shown in Fig. 2.14(b) provides a directly viewed readable image rotated 180 deg.
However, these prisms cannot be close stacked in an array, resulting in gaps and
loss of efficiency over the full aperture. To provide a close-stacked array, a double-
roof Dove prism having a square cross section can be constructed, as shown in
Fig. 8.49.34 The upper roof is a nonworking optical surface whose only purpose
is to achieve a continuous-array aperture with no voids. A thin air separation
must be maintained between the individual prisms, or the working roof surfaces
must be reflectorized. The array shown in Fig. 8.50(a) consists of 88 double-roof
Dove prisms with an effective input aperture of radius D, each prism having a
232 Chapter 8

Figure 8.48 Hexagonal cube-corner reflective array. Leading corners are circled
on sample microcube. (Adapted from Ref. 33 with kind permission of Springer
Science+Business Media.)

Figure 8.49 Double-roof Dove prism with square cross section.34

side of length b, with a height d. Let D = 10 cm and d = 1 cm, and assume a

length/aperture ratio γ = 6 for each individual Dove prism. Then d = 1.414b, and
the length L of each prism is γ × d ≈ 8.5 cm. Thus, the aspect ratio of the array
γarray ≈ 8.5/10 = 0.85.
Figure 8.50(b) shows a side cross section of the array. For imaging purposes, a
continuous integrated image is viewed for an object at a far distance and could be
used for large-field scanning and tracking. These arrays can also be used in pairs
to function as a variable beam deflector in transmission.
Microprismatic Arrays 233

Figure 8.50 (a) Double-roof Dove prism array for circular entrance beam.34 (b) Cross
section of double-roof Dove prism array.34
234 Chapter 8

References
1. J. F. Dreyer, “Light fixture providing normalized output,” U.S. Patent
No. 4,791,540 (1988).
2. S. G. Saxe, “Prismatic film light guides: performance and recent
developments,” Solar Energy Mat. 19, 95–109 (1989).
3. L. A. Whitehead, “Prism light guide having surfaces which are in octature,”
U.S. Patent 4,260,220 (1981).
4. L. A. Whitehead et al., “New efficient light guide for interior illumination,”
Appl. Opt. 21(18), 2755–2757 (1982).
5. L. A. Whitehead, “Prism light guide luminaire,” U.S. Patent No. 4,615,579
(1986); U.S. Patent No. 4,750,798 (1988).
6. S. Cobb, “Totally internally reflecting light conduit,” U.S. Patent
No. 4,805,984 (1989).
7. C. Sease, “Light piping: a new lighting system for museum cases,” J. Am. Inst.
Conserv. 32(3), 279–290 (1993).
8. S. G. Saxe, “Light pipe having optimized cross-section,” U.S. Patent
No. 5,309,544 (1994).
9. L. A. Whitehead, “Prism light guide luminaire with efficient directional
output,” U.S. Patent No. 5,339,382 (1994).
10. K. A. Aho et al., “Back-lit display,” U.S. Patent No. 4,874,228 (1988).
11. R. A. Miller et al., “High aspect lighting element,” U.S. Patent No. 5,190,370
(1993).
12. R. E. DuNah et al., “Flat, thin, uniform thickness large area light source,” U.S.
Patent No. 5,420,761 (1995).
13. J. Kuper, “Light directing optical structure,” U.S. Patent No. 5,761,355 (1998).
14. J. C. Wright and M. C. Lea, “Light-guide lights suitable for use in illuminated
displays,” U.S. Patent No. 7,164,836 (2007).
15. A. Abileah et al., “Lighting assembly for a backlit electronic display including
an integral image splitting and collimating means,” U.S. Patent No. 5,161,041
(1992).
16. M. B. O’Neill and S. Cobb, “Brightness enhancement film with soft cutoff,”
U.S. Patent No. 5,917,664 (1999).
17. S. Cobb et al., “Variable pitch structured optical film,” U.S. Patent
No. 5,919,551 (1999).
18. A. B. Campbell et al., “Optical film,” U.S. Patent No. 6,354,709 (2002).
19. K. E. Epstein and R. P. Wentz, “Front-lit liquid crystal display having
brightness enhancement film with microridges which directs light through the
display to a reflector,” U.S. Patent No. 5,608,550 (1998).
Microprismatic Arrays 235

20. S. C. Tang, “Brightness enhancement film,” U.S. Patent No. 6,277,471 (2001).
21. J. Lee and D. Kessler, “Brightness enhancement film using a linear array of
light concentrators,” U.S. Patent No. 7,160,017 (2007).
22. M. Suzuki, “Reflective polarizer sheet on the backlighting unit,” SID SID
Symp. Digest, pp. 813–816 (1997).
23. M. F. Weber, “Retroreflecting sheet polarizer,” U.S. Patent No. 5,559,634
(1996).
24. C.-Y. Tai et al., “Backlighting assembly utilizing microprisms and especially
suitable for use with a liquid crystal display,” U.S. Patent No. 5,390,276
(1995).
25. S. M. P. Blom et al., “Towards a polarized light-emitting backlight micro-
structured anisotropic layers,” J. SID 10(3), 209–213 (2002).
26. H. J. Cornelissen et al., “Polarized light LCD backlight based on liquid
crystalline polymer film: a new method of manufacture,” SID Symposium
Digest 35, 1178–1181 (2004).
27. S.-M. Huang et al., “Polarizing light guide plate unit and backlight unit
and display device employing the same,” European Patent Application
No. 1850156A1 (2006).
28. D. F. Vanderwerf, “Multiple lamp illumination system for projection displays,”
Proc. SPIE 2650, 54–62 (1996); “Projector with multiple light source,” U.S.
Patent No. 5,504,544 (1994) [doi:10.1117/12.237018].
29. R. J. Saccomanno, “Hollow cavity light guide for the distribution of collimated
light to a liquid crystal display,” U.S. Patent No. 6,443,585 (2002).
30. D. F. Vanderwerf, “Planar polarizer for LCD projectors,” U.S. Patent
No. 5,940,149 (1998).
31. R. F. Stamm, “Retroreflective surface,” U.S. Patent No. 3,712,706 (1973).
32. T. L. Hoopman, “Cube-corner retroreflective articles having wide angularity
in multiple viewing planes,” U.S. Patent No. 4,588,258 (1986).
33. E. Brinksmeier et al., “Manufacturing of molds for replication of micro cube-
corner retroreflectors,” Prod. Eng. Res. Devel. 2, 33–38 (2008).
34. T. Lian and M.-W. Chang, “New types of reflecting prisms and reflecting prism
assembly,” Opt. Eng. 35(12), 3427–3431 (1996) [doi:10.1117/1.601103].

Some commercial nonsequential ray-tracing programs:

• ASAP, from Breault Research Organization, Inc., Tucson, AZ
• ZEMAX, from Zemax Development Corp., Bellevue, WA
• TracePro, from Lambda Research Corp., Littleton, MA
• CODE V, from Optical Research Associates, Pasadena, CA
• LightTools, from Optical Research Associates, Pasadena, CA
Chapter 9
Fresnel Lenses

9.1 Basic Refractive Fresnel Lens Design

The refractive Fresnel lens consists of a series of adjacent microprisms designed to

provide a varying deviation angle over the lens area. These angles are programmed
to focus or collimate incident light. The most common Fresnel lens consists of a
series of concentric grooves replicated in optical plastic, with grooves on one side
and a planar surface on the other, and having positive power. It is often referred to as
a positive aspheric Fresnel lens. Figure 9.1 shows how a continuous aspheric lens
can be collapsed to a Fresnel lens surface, eliminating much of the bulk material.
The profile of the continuous aspheric surface can be described by the standard

Figure 9.1 Relationship between a continuous aspheric surface and a Fresnel lens
surface.

237
238 Chapter 9

equation of an aspheric surface, axially symmetric about the z axis:

cx2
z= + a1 x4 + a2 x6 + a3 x8 + a4 x10 , (9.1)
1+ 1 − (k +
p
1)c2 x2

where z and x are the coordinates of the surface, c is the vertex curvature, k is the
conic constant, and a1 , a2 , a3 , and a4 are the aspheric coefficients.
The basic geometry for ray tracing through a positive aspheric Fresnel lens is
shown in Fig. 9.2(a), and the refractive geometry at a single echelon is shown in
Fig. 9.2(b).1 Here, θ1 and θ2 are the incident and exit-ray angles, θ2 0 is the angle of
incidence at the exit surface, t is the lens thickness, and n is the refractive index of
the lens material. From this geometry and the application of Snell’s law, the groove
angle α can be calculated from the equation

sin θ1 + n sin θ2 0
tan α = . (9.2)
n cos θ2 0 − cos θ1

Consider a single-element Fresnel lens with collimated light incident normal to

the planar side. Here, δ = θ2 = I2 0 − α, as shown in Fig. 9.3(a). Figure 9.3(b)
graphs the deviation angle δ as a function of the microprism groove angle α.
The maximum deviation angle δmax is attained when TIR occurs at the grooved
surface—e.g., I2 = αmax = arcsin(1/n) ≈ 42.12 deg for an acrylic plastic lens
with n = 1.491, and δmax = 90 deg − αmax ≈ 47.9 deg. For larger values of α, the
lens is no longer refractive. When collimated light is incident on the grooved side
[Fig. 9.4(a)], where I2 = α−I1 0 and δ = θ2 = I2 0 , the deviation angle δ as a function
of the microprism groove angle is as shown in Fig. 9.4(b). Since δ is now limited
by TIR at the planar surface, up to 90-deg deviation is possible at α ≈ 84.0 deg for
an acrylic plastic lens.
The groove angle α0 of a Fresnel lens can also be calculated from the derivative
of Eq. (9.1), where c and k now refer to the grooved surface:

dz
tan α0 = , (9.3a)
dx
or
cx
tan α0 = p + 4a1 x3 + 6a2 x5 + 8a3 x7 + 10a4 x9 . (9.3b)
1 − (k + 1)c2 x2

This is a concise way of specifying the groove angles of a Fresnel lens for any
groove frequency, normally between two and eight grooves per millimeter. From
a table of calculated groove angles from Eq. (9.2), the angles can be least squares
fitted to Eq. (9.3b) to calculate the values of the variables c, k, a1 , a2 , a3 , and
a4 . Since Eq. (9.3b) is not linear in the variables, a general technique is used that
Fresnel Lenses 239

minimizes the sum of the squares of the residuals by successive corrections to these
variables.2 The residuals are defined as (tan α0 − tan α), and initial estimates of the
variables are required.

Figure 9.2 (a) Basic geometry for ray tracing through a positive aspheric Fresnel lens. (b)
Refractive geometry at a single echelon.
240 Chapter 9

Figure 9.3 (a) Fresnel lens with collimated light incident on the plano side. (b) Deviation
angle versus groove angle of a Fresnel lens with collimated light incident on the plano side.

9.1.1 Design example: Fresnel lens collimator/searchlight

Consider a circular refractive Fresnel lens collimator, as shown in Fig. 9.5. The lens
diameter Ap = 300 mm, the distance of the light source from the planar entrance
surface is f = 250 mm, the material is acrylic plastic, and the lens thickness
t = 3 mm. We use initial estimates of c = 1/[(n − 1) f ] = 0.0081 mm−1 , k = −1
(parabolic), and a1 = a2 = a3 = a4 = 0. A sample of equally spaced rays are
traced as shown. Each groove angle α is calculated from Eq. (9.2), and all groove
Fresnel Lenses 241

Figure 9.4 (a) Fresnel lens with collimated light incident on the grooved side. (b) Deviation
angle versus groove angle of a Fresnel lens with collimated light incident on the grooved
side.
242 Chapter 9

Figure 9.5 A circular refractive Fresnel lens collimator/searchlight.

angles are then least squares fitted to Eq. (9.3b), yielding α0 . The calculated α and
α0 values are equal to five significant figures, and the groove profile parameters are
summarized as follows:

c = 0.0080816 mm−1 ,
k = −0.91915,
a1 = 6.1322 × 10−10 ,
a2 = −4.970 × 10−14 ,
a3 = 1.3967 × 10−19 ,
a4 = −1.0211 × 10−24 ,
Sum of squares (SS) of residuals = 6.5733 × 10−12 .

The results are summarized in Table 9.1.

9.2 High-Transmission Fresnel Lens Doublet

In many applications of Fresnel lenses, the surface reflection losses in the outer
region of a single lens are excessive, and a dual-element Fresnel lens needs to
be used. This is the case, for example, in the condensor lens near the stage of
a conventional overhead projector. Consider a 300-mm-aperture lens with design
Fresnel Lenses 243

Table 9.1 Least-squares aspheric fit of groove angles of a Fresnel lens collimator.
x1 Planar surface (mm) x2 Grooved surface (mm) α (deg) α0 (deg) Residual tan(α0 ) − tan(α)
0.4960 0.50 0.23152 0.23152 2.6873 × 10−9
10.416 10.50 4.8519 4.8519 −1.570 × 10−7
20.337 20.50 9.4183 9.4183 2.2310 × 10−7
30.258 30.5 13.883 13.883 3.8844 × 10−7
40.181 40.5 18.204 18.204 2.4518 × 10−7
50.105 50.5 22.348 22.348 −5.6978 × 10−7
60.030 60.5 26.291 26.291 −4.6390 × 10−7
69.958 70.5 30.016 30.016 6.8274 × 10−7
79.888 80.5 33.516 33.516 −3.6294 × 10−7
89.820 90.5 36.789 36.789 −2.80 × 10−7
99.754 100.5 39.838 39.838 8.5820 × 10−7
109.69 110.5 42.672 42.672 −2.3251 × 10−7
119.63 120.5 45.30 45.30 6.9366 × 10−7
129.57 130.5 47.734 47.734 −1.5539 × 10−6
139.52 140.5 49.986 49.986 1.1441 × 10−6
149.47 150.5 52.0697 52.0697 −2.7640 × 10−7

conjugates f = 250 mm (light source position) and f 0 = 400 mm (focus position).

Figure 9.6 shows a section of the lens where θ1 > θ4 , and the angle of refraction
I2 0 at each groove of the first element is set equal to the angle of incidence I3 at
each groove of the second element. One can consider that the “air prism” between
the prism facets is operating at minimum deviation. This results in a divergence of
the rays exiting the first element. Let ϕ ≡ I2 0 = I3 , and ϕ0 ≡ I2 = I3 0 . From the
figure:

α2 − ϕ = ϕ − α3 , (9.4a)
α2 = I2 + I1 0 , (9.4b)
α3 = I3 0 + I4 0 , (9.4c)

where

(I1 0 + I4 )
ϕ = ϕ0 + . (9.4d)
2

Using Snell’s law to relate ϕ and ϕ0 , it follows that

(I1 + I4 ) 
 " 0 # 

n sin


 


2

 

ϕ = arctan  .
 
(9.4e)
 
0
+
" # 

 (I 1 I4 ) 

 n cos − 1


 

2
 

244 Chapter 9

Figure 9.6 Ray-trace parameters for a high-transmission Fresnel lens doublet.

Then, the groove angles can be calculated from

!
sin ϕ
α2 = arcsin + I1 0 , (9.4f)
n
!
sin ϕ
α3 = arcsin + I4 . (9.4g)
n

Table 9.2 gives sample data for this Fresnel lens doublet. The groove width
= 0.5 mm, and Y2 and Y3 are set near the midpoint of each groove. The lens
material is acrylic plastic (n = 1.492). The transmission T i at each surface is
calculated using Eqs. (1.13) and (1.15), and the total transmission T = T 1 T 2 T 3 T 4
for each ray traced. Fresnel lenses are usually mass produced through replication,
and antireflection coatings are rarely used.

Groove Profile for α2 : Groove Profile for α3 :

c = 0.0070568 mm−1 c = 0.0064347 mm−1
k = −1.0632 k = −0.61104
a1 = 8.8277 × 10−9 a1 = −1.2115 × 10−8
a2 = −3.0594 × 10−14 a2 = −5.8711 × 10−14
a3 = −3.4088 × 10−20 a3 = −6.6858 × 10−19
a4 = 2.5417 × 10−24 a4 = −9.9916 × 10−24
SS residuals = 1.7063 × 10−10 SS residuals = 5.8336 × 10−10
Fresnel Lenses 245

Table 9.2 Representative data for a Fresnel lens doublet.

x2 , x3 (mm) x1 (mm) x4 (mm) α2 (deg) α3 (deg) T = T1 T2 T3 T4 Transmission
0.250 0.2480 0.2487 0.1011 0.0869 0.8530
11.75 11.65 11.69 4.742 4.075 0.8530
23.25 23.06 23.13 9.334 8.022 0.8529
34.75 34.47 34.57 13.833 11.889 0.8529
46.25 45.88 46.02 18.198 15.642 0.8527
57.75 57.29 57.46 22.396 19.254 0.8524
69.25 68.70 68.91 26.403 22.706 0.8518
80.75 80.12 80.35 30.201 25.983 0.8508
92.25 91.53 91.79 33.780 29.078 0.8494
103.75 102.95 103.24 37.136 31.991 0.8475
115.25 114.36 114.68 40.270 34.724 0.8449
126.75 125.78 126.13 43.187 37.281 0.8417
138.25 137.20 137.57 45.896 39.673 0.8378
149.75 148.62 149.02 48.407 41.907 0.8330

In addition to surface-reflection losses, there are also geometric effects caused

by the risers of each element. One effect occurs when light is directly blocked by
the risers, resulting in reduced throughput, as illustrated in Fig. 9.7(a). Another
effect is a dark banding in the output light, as shown in Fig. 9.7(b). Where the
riser angle β and groove width w can be controlled, it is possible to eliminate
both of these effects, as shown for the Fresnel lens doublet in Fig. 9.7(c). To
accomplish this, the pitch and riser angles of the elements are controlled such that
light leaving the peaks and valleys of the first element hit the corresponding peaks
and valleys of the second element. A defined separation of the elements needs to
be maintained. The design of such a lens has been described by Dudnikov et al.3
However, there are maximum β values that allow release from the mold during
replication, and in the case of overhead projector condensers, there are minimum
allowable β values to avoid stage glare from spurious reflections off of the
risers.4

9.3 Reflective Fresnel Lenses

Reflective Fresnel lenses are useful components for several applications, including
illuminators, solar energy concentrators, and compact overhead projectors, where
both the light source and projection lens are positioned over the stage.

9.3.1 First-surface reflector design parameters

Figure 9.8 shows the basic geometry for a first-surface reflective Fresnel lens, with
axial design conjugates f and f 0 , where f 0 > f . The groove angles α1 can be
246 Chapter 9

Figure 9.7 (a) Reduced throughput of a Fresnel lens due to direct riser blockage. (b)
Dark banding in output of a Fresnel lens. (c) High continuous throughput of a Fresnel lens
doublet.3
Fresnel Lenses 247

Figure 9.8 Basic geometry of a first-surface reflecting Fresnel lens.

calculated from the following set of equations:

!
y
θ = arctan , (9.5a)
f
!
y
θ = arctan 0 ,
0
(9.5b)
f
(θ + θ0 )
α1 = . (9.5c)
2

9.3.2 Second-surface reflector design parameters

A detail of the refraction and reflection at a groove of a second-surface Fresnel
lens reflector is shown in Fig. 9.9. The groove angles α2 can be calculated from the
following set of equations, where I1 = θ, I3 0 = θ0 , and the material has a refractive
index n:
!
sin I1
I1 = arcsin
0
, (9.6a)
n
sin I3 0
!
I3 = arcsin , (9.6b)
n
(I1 0 + I3 )
α2 = . (9.6c)
2
248 Chapter 9

Figure 9.9 Groove detail for a second-surface reflecting Fresnel lens.

9.4 Refractive Planar Circular Fresnel Lens Solar Applications

There are few uses of planar circular Fresnel lenses for solar-energy-concentration
applications. This is mainly due to size limitations in fabrication, structural stability
of large lenses when mounted in a near-horizontal position, the need for precise
solar tracking, and off-axis optical aberrations that affect the achievable focusing
accuracy.

9.4.1 Multilens solar furnace

One application uses a spatially separated array of lenses to construct a solar
furnace.5 Figure 9.10 depicts a section of the solar furnace. The lenses are mounted
on a support structure, precisely oriented in the direction of the sun, such that each
lens is focused at the spherical tip of a heat-conducting rod. These rods have high
heat conduction and feed into boiler tanks though which a vaporizing fluid flows.
The heated fluid can be utilized to generate steam and drive a turbine.

9.4.2 Multilens-array solar simulator

Another application of planar circular Fresnel lenses is the construction of a
low-cost, large-area solar simulator.6 Solar simulators attempt to reproduce solar
radiation with regard to intensity, spectrum, and uniformity over a defined area, for
controlled testing of materials and structures. The simulator consists of 143 lenses
arranged in a close-packed hexagonal array, each lens having a tungsten-halogen
lamp positioned one focal length (≈300 mm) from the lens, producing a uniform
collimated beam over an area 1.2 m2 (see Fig. 9.11).

9.5 Refractive Meniscus Fresnel Lenses

A refractive Fresnel lens having a meniscus shape solves some of the problems
of the flat Fresnel lens. Of particular interest is the fact that the meniscus Fresnel
lens can be made aplanatic, free from spherical aberration and coma. The theory is
covered in several papers by Delano and Kleinhans.7–9 Erismann has described the
design of a polyethylene Fresnel lens with a spherical shape for use as an infrared
motion sensor, where collimated incident light is focused to a point.10 However,
Fresnel Lenses 249

Figure 9.10 A multilens solar furnace.5

Figure 9.11 Fresnel lens array for a large-area solar simulator.6

250 Chapter 9

its use in large-diameter solar concentrators is uncommon, for some of the same
reasons as for the flat Fresnel lens.

9.6 Reflective Planar Linear-Focus Solar Concentrators

Many of the applications of Fresnel lenses as solar concentrators involve the use
of reflective linear Fresnel lens es producing a linear focus. When used as solar
concentrators, the first- or second-surface linear reflectors, shown in Figs 9.8 and
9.9, and operating at ≤ f /1.0, produce significant blockage of light by the risers in
the lens outer region.
9.6.1 Tilted linear-focus reflective solar concentrator
In Fig. 9.12(a), a tilted flat first-surface linear Fresnel reflector produces no riser
blockage.10 The design constraint that eliminates riser blockage is

ϕ ≥ α ≥ 0, (9.7a)

or

−ϕ ≤ α0 ≤ 0, (9.7b)

where ϕ is the minimum tilt angle of the reflector, α is the facet angle with respect
to the x-y coordinate system, α0 is the facet angle with respect to a rotated x0 -y0
coordinate system, and α0 = α − ϕ. For the V-configuration of Fig. 9.12(b), with
focal length f 0 , half-acceptance angle γ, and minimum tilt angle ϕmin , the aperture
Ap is calculated from
!
sin γ 0
Ap = 2 f . (9.8)
cos ϕ

For the second-surface configuration of Fig. 9.13, there is no riser blockage when

ϕ ≥ α > 0, (9.9a)
−ϕ ≤ α0 < 0, (9.9b)

where the refraction at the planar surface requires that α > 0.

9.6.2 Linear-focus concentrator using a linear Fresnel lens and a
crossed linear total-internal-reflecting array
Another type of planar reflecting linear-focus solar concentrator [Fig. 9.14(a)] has a
linear Fresnel structure replicated on one side of a plastic sheet and an orthogonal
90-deg TIR prism array replicated on the other side. No reflective coatings are
required. Figure 9.14(b) shows a detailed section with the direction vectors defining
the prism facets. Rays are traced through one quadrant of the concentrator using
Eqs. (1.25), (1.26), and (1.27) for the refractions and Eqs. (1.52) and (1.53) for the
Fresnel Lenses 251

Figure 9.12 (a) A tilted first-surface reflective linear-focus solar concentrator.11

(b) V-configuration for tilted reflective linear-focus solar concentrator.11

Figure 9.13 A tilted second-surface reflective linear-focus solar concentrator.

252 Chapter 9

Figure 9.14 (a) A linear-focus linear Fresnel lens/crossed TIR linear array solar
concentrator. (b) Groove detail of linear-focus reflecting solar concentrator.
Fresnel Lenses 253

reflections, where
kx1 = −sin α ky1 =0 kz1 = −cos α
kx2 =0 ky2 = cos 45 deg kz2 = −cos 45 deg
kx3 =0 ky3 = −cos 45 deg kz3 = −cos 45 deg
kx4 = sin α ky4 =0 kz4 = cos α
There is a varying displacement of the incident and exit rays for this system. The
intersection coordinates at each surface can be calculated from Eq. (1.59), where
!
K x(i−1)
xi = xi−1 + (Pi − k xi x − kyi y − kzi z), (9.10a)
ρi
!
Ky(i−1)
yi = yi−1 + (Pi − k xi x − kyi y − kzi z), (9.10b)
ρi
!
Kz(i−1)
zi = zi−1 + (Pi − k xi x − kyi y − kzi z), (9.10c)
ρi
where i = (1, 2, 3, 4).
Consider a reflective concentrator of size 500 mm2 with a centered line focus
f 0 = 500 mm from the panel. The panel thickness t is 2 mm, the groove width w
at each surface is 0.5 mm, and the material is acrylic √ plastic. The P values are as
follows: P1 = P4 = (w/2) sin α, and P2 = P3 = w( 2/2) + (t − w) sin 45 deg,
relative to a local coordinate system (x0 , y0 , z0 ) at the top surface, and centered
with each refracting groove and the corresponding reflecting groove vertex. The
direction cosines and coordinates are calculated for a series of exit rays for an
arbitrary y value, and the groove angles α are iteratively adjusted until x5 ≈ 0 at
the image plane, where
!
K x4 0
x5 = x4 + f . (9.11)
Kz4
For this linear concentrator, the important displacement is (x1 − x4 ) between the
collimated entrance ray and the focused exit ray, orthogonal to the focal line. This
varies from about 0.0027 mm near the panel center to about 0.58 mm at the edge
of the panel. Figure 9.15 plots the calculated groove angle α as a function of the
distance x0 from the panel center.

9.6.3 Planar reflective spot-focus concentrator using orthogonal

refractive and reflective linear Fresnel lenses
If a linear refracting Fresnel lens and a linear reflecting Fresnel lens are
crossed orthogonally, 2D convergence of normally incident solar radiation can be
achieved.11 For the panel shown in Fig. 9.16, the incident surface is the planar side
of the refracting element. All rays incident along any two orthogonal lines on the
panel can be directed to a common focal point, while incident rays outside these
directions will deviate from this focus. This lack of radial symmetry results in an
extended spot focus.
254 Chapter 9

Figure 9.15 Groove angle versus distance from center for a linear-focus reflecting solar
concentrator.

Figure 9.16 A spot-focus reflective solar concentrator using crossed linear Fresnel lens
elements.11
Fresnel Lenses 255

For a 305-mm2 panel with a focal length f 0 = 305 mm and n = 1.491,

Fig. 9.17(a) illustrates a histogram of the distribution of intersection points within
annular rings at the focal plane, when all incident rays along the x axis and the y
axis are focused to a point (axial design). Figure 9.17(b) shows the distribution of
intersection points when rays incident along the diagonals are focused to a point
(diagonal design). In each case, the 95% energy containment radius R95 is shown.
This type of planar concentrator can be supported on the back side and enlarged
by a mosaic of square or rectangular elements. It requires two-directional tracking
and could be useful for circular targets of extended area.

9.7 Curved Linear Fresnel Lens Solar Concentrators

Refractive linear Fresnel lenses formed along a circular or an aspheric curve
are very useful as solar concentrators (see Fig. 9.18). They can be fabricated in
large sizes and have better structural stability than flat lenses. There are various
methodologies for their design. One method starts at the edge and designs toward
the center. Each microprism is designed for minimum deviation of the incident
ray, making sure that each succeeding prism does not block any exit rays from
the one that preceded it.12 As shown in Fig. 9.19, α is the prism angle, and the
(x, z, ϕ) coordinates specify the position and slope of the curved surface. To avoid
interference from the riser of the next groove, set β < β0 , where β0 = I2 0 +90 deg−α.
The basic ray-trace equations for collimated incident light are as follows, where δ
is the required deviation angle for focusing:

I1 = ϕ, (9.12a)
sin(δ − I1 ) + sin I1
α = arctan p , (9.12b)
n2 − sin2 I1 − cos(δ − I1 )
!
sin I1
I1 = arcsin
0
, (9.12c)
n
I2 = α − I1 0 , (9.12d)
I2 0 = arcsin(n sin I2 ), (9.12e)
δ = I2 0 − α + I1 . (9.12f)

For minimum deviation of the refracted ray, each microprism is designed such that
I1 = I2 0 . Then,

αmin = 2I1 0 , (9.13a)

δmin = 2(I1 − I1 0 ). (9.13b)

This minimizes surface reflection losses, thereby maximizing the lens

transmission, and reduces the effect of prism slope errors. The overall transmission
can exceed 90% with f /# < f /1.0. For any finite value of the lens refractive index,
I2 will eventually reach the critical angle as ϕ increases. Therefore, it is not possible
to design a concentrator that operates over a full hemisphere.
256 Chapter 9

Figure 9.17 (a) Histogram of ray intersection points at the focal plane for a spot-focus
concentrator (axial design). (b) Histogram of ray intersection points at the focal plane for a
spot-focus concentrator (diagonal design).
Fresnel Lenses 257

Figure 9.18 A curved linear refractive Fresnel lens.

Figure 9.19 Ray-trace detail for a curved linear Fresnel lens.

Another design method considers the collimation angle ±θ0 of solar radiation,
where θ0 = 0.27 deg = 4.7 mrad. The maximum theoretical limit of the geometric
concentration ratio C(θ0 ) ≤ 1/ sin θ0 , where C(θ0 ) is the ratio of the lens input
aperture A to the target width a.13 Thus, the ideal maximum value is C(θ0 ) ≈ 200.
For the curved linear Fresnel lens of Fig. 9.20, the edge-ray principle is applied
258 Chapter 9

Figure 9.20 Edge-ray principle for a curved linear solar concentrator. (Adapted from
Ref. 14 with permission from the Optical Society of America.)

to each microprism. By adjustment of (x, y, ϕ, α), solar rays of ray angle +θ0 are
focused at the right edge of the target of length a, while those of ray angle −θ0
are focused to the left edge of the target. The lens is designed outward from the
center.14 With precise 2D tracking, all solar rays will then fall within the target. In
practice, chromatic dispersion reduces the concentration ratio to about 40.
Kritchman has described a “color-corrected” curved Fresnel lens that doubles
the concentration ratio to about 80.15 These design methods produce a noncircular-
shaped cross-sectional profile for the lens, where normally α > ϕ. Another design
method for a curved linear Fresnel lens solar concentrator has been described
where the smooth and grooved surfaces are formed with constant radius of
curvature.16
Using the edge-ray principle, the 2θ0 solar collimation acceptance angle can
be enlarged to a general acceptance angle 2θ. This relaxes the precise tracking
requirements but reduces the concentration ratio.17 The limiting values of θ are
usually determined by the occurrence of TIR at the prism facet. Figure 9.21 shows
how the maximum acceptance angle θmax in the x-z plane can be determined when
I2 = α−I1 0 = I2crit and θmax = ϕ−I1max . For example, when α = 60 deg, ϕ = 45 deg,
and n = 1.492, then θmax = 17.68 deg. Similarly, θmax for an incident ray in the
transverse y-z plane can be calculated using Eqs. (1.25) to (1.27). The smallest
value of θmax in either plane near the edge of the lens determines a conservative
value for the acceptance angle.
From a practical standpoint, it is useful to reduce the tracking requirement
from full 2D movement to movement in a single polar direction. It has been
found that diurnal rotation of angle ω about the linear axis of the concentrator
minimizes shortening of the focal point position from the lens center. See Fig. 9.22,
Fresnel Lenses 259

Figure 9.21 Ray-trace detail for a curved linear Fresnel lens with acceptance angle 2θ.
(Adapted from Ref. 17 with permission from the Optical Society of America.)

Figure 9.22 Diurnal rotational tracking for curved linear Fresnel solar concentrator.
(Adapted from Ref. 18 with permission from the Optical Society of America.)

where rotation of the entire connected unit occurs about the y axis. Kritchman has
analyzed the performance of a polar-tracking curved linear Fresnel lens for two
seasonal changes—namely, θy-z = 0 deg (equinox), and θy-z = 23.4 deg (solstice).18
A curved linear Fresnel lens concentrator with minimal tracking requirements
has been described by Leutz et al.19 The resulting geometric concentration ratio is
260 Chapter 9

on the order of 1.5 to 2.0 and is usable for photovoltaic applications. The edge-ray
principle is applied such that light at two orthogonal acceptance angles is directed
to a linear absorber. The prisms are designed at an angle of minimum deviation.
The concentrator is oriented with the linear axis in the east-west direction with a
seasonal tilt. Acceptance half-angles are approximately ±45 deg in the east-west
direction and ±30 deg in the orthogonal direction.

9.8 Flexible Fresnel Lens Solar Concentrators

9.8.1 Sectional planar solar concentrators
In Fig. 9.23, a sectional refractive lens solar concentrator uses thin, flexible, linear,
refracting Fresnel lenses supported by a frame.20 The lenses have the smooth
surface toward the sun and the grooved surface toward the absorbing target, which
may be a linear heat exchanger or a linear array of photovoltaic cells. The lenses
are held under light tension and are made from a material such as 0.4-mm-thick
polymethylmethacrylate (PMMA). Sections A and A0 are inclined at ϕA = 45 deg,
and sections B and B0 are inclined at ϕB = 13.75 deg. The refracting grooves are
designed for minimum deviation according to Eqs. (9.13a) and (9.13b). The lenses
are allowed to flex and bow under wind, gravity, or other environmental factors.
Due to the increased groove-angle tolerance of minimum deviation, it is estimated
that reasonable focus will be maintained when the lens bows up to ±5 deg, where
the bow angle is measured from the tangent point of the curve at the edge of each
section.

9.8.2 Inflatable curved solar concentrators

For some solar applications in space, a rigid curved Fresnel lens can be replaced
by an inflatable flexible Fresnel lens.21 In Fig. 9.24, the inflated solar concentrator

Figure 9.23 A sectional flexible Fresnel lens solar concentrator.20

Fresnel Lenses 261

Figure 9.24 An inflatable linear Fresnel lens solar concentrator.21

has the shape of a cylinder. The cylinder consists of two curved flexible reflecting
sections, such as aluminized polyester film, and a transparent Fresnel lens section.
The transparent forward section consists of a flexible polymer with a series of linear
Fresnel grooves formed on its interior surface and a smooth outer surface. The
reflective flat side sections are the same material as the curved reflecting section.
These sections are joined at the contiguous edges to a rigid back section that is a
strong thermal conductor. The entire concentrator assembly is then inflated with
a low-pressure gas, such as hydrogen, helium, or nitrogen, through a valve on the
back section. When the Fresnel lens section is oriented toward the sun, a linear
focus is formed on the back section. When the focal length f is greater than the
centerline axis of the cylinder, the Fresnel lens microprisms can be designed for
minimum deviations as per Eqs. (9.13a) and (9.13b), resulting in a lens that is
more tolerant to microprism slope errors. The linear receiver would have a 5- to
10-cm width for a photovoltaic array, and the back plate would conduct and radiate
excess heat into space. For the deflated concentrator, the flexible sections could be
folded against the rigid back plate and then deployed by inflation in space.

9.9 Fresnel Lenses Using Total Internal Reflection

As the entrance angles become larger, a conventional Fresnel lens may no longer
function as a refracting lens [see Figs. 9.3(b) and 9.4(b)]. In this region, a new
class of Fresnel lens can be used that focuses by a combination of refraction and
total internal reflection. These are often called catadioptric Fresnel lenses or TIR
Fresnel lenses. They are based on the TIR prism design described in Sec. 4.1.2, but
262 Chapter 9

the preferred prism material is a plastic such as acrylic. Figure 9.25 depicts several
grooves of a single-element catadioptric Fresnel lens.

9.9.1 Low-profile overhead projector

A catadioptric Fresnel lens can be used with a conventional refracting Fresnel
lens to form a two-element condensing lens in a low-profile overhead projector.22
Figure 9.26 illustrates the optical layout, which uses an off-axis section of the
catadioptric Fresnel lens. This allows the lamp to be positioned outside the clear
stage area for cooling purposes and avoids any transition area between refracting
and catadioptric regions of the lower lens. Because the condenser lens is close
to the image stage in an overhead projector, it would be difficult to match the
transmission at the boundary of these regions, and a brightness difference would
be visible on the projected stage area. The catadioptric Fresnel lens is designed to
provide light with a smaller ray angle to the centered refractive upper Fresnel lens.
The stage aperture is normally about 285 mm2 , and the height of the projector base
is between 75 mm and 100 mm.

9.9.2 Curved catadioptric Fresnel lenses

Much of the earlier work on curved catadioptric Fresnel lenses was for use in
automobile taillights (Fig. 9.27).23 Where a catadioptric Fresnel lens is used as
a compact illuminator or searchlight, it is possible to use a centered single-element
Fresnel lens with both dioptric and catadioptric regions. Modern plastic replication
techniques have generated new interest in this type of lens. Figure 9.28 shows a
plastic curved TIR circular diverging Fresnel lens having a height-to-diameter ratio

Figure 9.25 Groove detail of a single-element catadioptric Fresnel lens.

Fresnel Lenses 263

Figure 9.26 Low-profile overhead projector using an off-axis catadioptric Fresnel lens.22

Figure 9.27 A curved catadioptric Fresnel lens used as an automobile taillight.23

264 Chapter 9

Figure 9.28 A plastic high-acceptance-angle curved catadioptric Fresnel lens.24

on the order of 0.25 or less.24 The lens is capable of collecting light over almost a
full hemisphere, with a 90-deg deviation angle near the edge of the lens. The output
light can also be designed to provide convergent or collimated light. This type of
lens can be provided with curved facets and could also be used as a light collector.25

9.9.3 Photovoltaic solar concentrator using total internal reflection

An integral three-zone circular solar concentrator lens for a single photovoltaic cell
has been designed. The lens is composed of three adjacent zones or regions.26 The
central region is a continuous-refracting conic surface, described by Eq. (9.1), with
a1 = a2 = a3 = a4 = 0, with an exit-ray-angle range between 0 and 15 deg. The
intermediate zone is a refractive Fresnel lens with an exit-ray-angle range from
about 15 to 30 deg. The outer zone is a catadioptric or TIR Fresnel lens with an
exit-ray-angle range from about 30 to 70 deg. The square entrance aperture has a
side length of 120 mm, and a 5.5 mm2 photovoltaic cell is placed 61.56 mm below
the top center of the lens, yielding a geometric concentration ratio ≈ 476×.
Each of the regions is designed by the edge-ray principle to focus a solar ray
with acceptance half-angle of θ0 = ±0.84 deg to each edge of the photovoltaic
cell. (See Sec. 9.7 on the use of the edge-ray principle for a solar concentrator.)
The incident surfaces of the center- and intermediate-refracting regions lie on a
common horizontal plane, but the incident surface for each groove of the outer
TIR lens region is angled. The lens has a staircase shape in this region. The optical
losses for this solar concentrator are ≈ 8% from surface reflection losses, ≈ 6%
from misdirected rays, and ≈ 2% from blocked rays. For an acrylic (PMMA) lens,
the chromatic dispersion was determined at λ = 300, 600, and 1200 nm, and was
estimated to contribute to about 2% additional loss in efficiency. By control of the
facet angles, an overall optical throughput efficiency of ≈ 81% can be achieved.

9.10 Fresnel Lenses for Rear-Projection Screens

Rear-projection displays are of two general types: those with the projection lens
axis centered with the projection screen normal, and those where the projection
lens is off axis with respect to the projection screen normal. The centered
system of Fig. 9.29(a) uses a circular Fresnel lens in conjunction with a bulk-
diffusing or matte surface screen. The use of a collimating Fresnel lens produces
Fresnel Lenses 265

a more uniform distribution of light over the screen-viewing area, as shown in

Fig. 9.29(b).27 In some cases, the separate Fresnel lens and diffusing screen can
be replaced by a single element with the Fresnel grooves on the input side and a
surface relief microstructure diffuser on the output side.28 This reduces cost and
suppresses ghosting due to spurious surface reflections.
Much of the recent research is in off-axis rear projection, driven by the goal
of designing thinner rear-projection television systems. For this application, off-
axis catadioptric Fresnel lenses are extremely useful. Figure 9.30 shows a basic
arrangement for an off-axis system with the grooves facing the light source. It
is realized that off-axis projection produces a distorted image on the screen that
cannot be corrected by the Fresnel lens. The device that produces the projected
image—e.g., a DMD or LCOS imager—must provide a predistorted image to
compensate for the off-axis projection. Another property of the catadioptric Fresnel
lens for rear projection is depicted in Fig. 9.31(a), where the order of inputted rays
is reversed at the diffuser. This can cause some loss of resolution in the projected
image. Figure 9.31(b) shows how the order can be preserved by an outward curving
of the TIR facets.29
Some Fresnel lenses for off-axis projection use a catadioptric region for the
high entrance angles and a contiguous refractive region for the lower ray-entrance
angles. Figure 9.32 depicts an arrangement where the input surface is planar and the
grooves face the diffusing sheet.30 A light transmission match is attempted at the
boundary between these regions. Figure 9.33 illustrates a rear-projection Fresnel
lens having grooves on both sides, with catadioptric and refracting regions, and a
transition region.31

9.11 Fresnel Lens Manufacture

Fresnel lenses are usually replicated in plastic from a master that is produced on
a lathe using a precision diamond tool (diamond-turning machine). Compression
molding produces the highest quality in terms of groove sharpness and fidelity.
This is usually the technique used for individual circular lenses, but it is size
limited and requires cycle times of several minutes. A newer high-precision
molding (HPM) process reduces the cycle time to seconds, while maintaining
the quality of compression-molded lenses.28 Injection molding and casting
are alternative techniques. For linear Fresnel lenses, embossing techniques
can produce continuous elements of good quality. Photoreplication is another
manufacturing technique for certain types of Fresnel lenses. Standard surface-
tolerance parameters used for glass prisms are not always relevant for plastic
Fresnel lenses. Since manufacturing methods are under continuous improvement
and modification, practitioners in Fresnel lens manufacture and design should be
consulted on these topics.

9.12 Achromatic Fresnel Lenses

Achromatic Fresnel lenses are useful in several applications, and there are several
methods to design and produce achromatic Fresnel lenses.
266 Chapter 9

Figure 9.29 (a) A rear-projection screen using a centered refractive Fresnel lens. (b)
Transmitted luminance distribution for a centered Fresnel lens rear-projection screen.
(Adpated from Ref. 27 with permission by The Society for Information Display.)
Fresnel Lenses 267

Figure 9.30 A rear-projection screen with off-axis projection and a catadioptric Fresnel
lens.

9.12.1 Combination of high- and low-dispersion materials

One method, used in a virtual image display, employs a sealed acrylic Fresnel
lens doublet with the grooves facing inward, with a high-dispersion liquid filling
the space between.32 The intervening liquid lens is negative in power. Typically,
the lower-dispersion acrylic lens material has a ν-number ≈ 57.4, and the higher
dispersion liquid has a ν-number ≈ 32.5.

9.12.2 Achromatic catadioptric Fresnel lenses

Using the principles for a single dispersing prism described in Sec. 4.1.2, it is
possible to design a catadioptric Fresnel lens that has positive, negative, or no
dispersion. The sample groove shown in Fig. 9.34 produces no dispersion and
is therefore achromatic. An isosceles triangle is formed by the refracting groove
angle α, TIR groove angle β, and apex angle γ, where β = γ. The lens will then be
achromatic for any incident ray angle θ1 , where

α = θ1 − θ3 , (9.14a)
(180 deg − α)
β= . (9.14b)
2

Single Fresnel lenses of this type would be most useful as off-axis sections with
high incidence angles θ1 , and the geometric throughput losses may require some
relaxation of perfect achromaticity.33
By constructing a doublet consisting of two catadioptric Fresnel lenses, it is
possible to provide an achromatic lens that is usable at lower incident ray angles.34
268 Chapter 9

(a)

(b)

Figure 9.31 (a) Order reversal of inputted rays for a catadioptric Fresnel lens. (b)
Preservation of ray order using curved TIR facets.29
Fresnel Lenses 269

Figure 9.32 Rear-projection screen with Fresnel lens grooves facing the diffuser, with
refractive and catadioptric regions.30

Figure 9.33 Rear-projection screen with a Fresnel lens having grooves on both sides, and
a refractive/catadioptric transition region.31

Figure 9.35 shows a typical groove pair, where

α3 = θ4 − θ1 , (9.15a)
(180 deg − α3 )
β2 = , (9.15b)
2
α4 = θ4 − θ6 , (9.15c)
(180 deg − α4 )
β5 = . (9.15d)
2
270 Chapter 9

Figure 9.34 Catadioptric Fresnel lens producing no dispersion.

Figure 9.35 Achromatic catadioptric Fresnel lens doublet useable at low-incident ray
angles.33

It is also possible to combine a catadioptric Fresnel lens element having

negative dispersion with a positive refractive Fresnel lens element (having positive
dispersion), as shown in Fig. 9.36. Both elements are the same material, such
as acrylic plastic (nd = 1.492). By proper choice of the groove angles, the net
dispersion can be brought close to zero. Both elements are acrylic plastic, and
representative groove angles α3 groove pair are: α1 = 25.2 deg, β2 = 71.9 deg,
Fresnel Lenses 271

Figure 9.36 Achromatic Fresnel lens doublet using catadioptric and refractive elements.34

α3 = 12.2 deg, and θ1 = 63 deg, θ2 = 20 deg, θ3 = 26.5 deg. Also,

α1 < (θ1 − θ2 ), (9.16a)

β2 < (180 deg − α1 ), (9.16b)
θ2 < θ3 . (9.16c)

Alternatively, a catadioptric Fresnel lens having positive dispersion can be

combined with a negative-refracting Fresnel lens to provide an achromatic
doublet.35

9.12.3 Dispersion-compensated achromatic Fresnel lens

A positive-refracting Fresnel lens can be made achromatic by placing a diffractive
structure on the Fresnel lens grooves, since the dispersion of a grating is of
the opposite sign to that of a refractive lens.36 Figure 9.37 illustrates a Fresnel
lens singlet with a diffractive structure on the individual refracting grooves. The
diffractive grooves are referenced to the base groove, which would normally be
machined prior to the machining of the diffractive grooves. The base groove angle
β and the blazed diffractive groove angle β0 are referenced to a common line, where
β0 is the angle that gives the same exit-ray angle at the blaze wavelength λB . The
blaze angle α is referenced to the base groove angle, where

α = β0 − β, (9.17)

and the diffractive groove period Λ is measured along the base groove length.
272 Chapter 9

Figure 9.37 Dispersion-compensated Fresnel lens with diffractive structure.36

Figure 9.38 Detail of diffraction parameters on a refracting groove.

The refraction, diffraction, and deviation angle at the surface of a diffracting

groove is detailed in Fig. 9.38. The chromatic dispersion of the diffractive structure
is of opposite sign to that of a purely refracting groove, and can be specified
between two wavelengths, λ1 and λ2 , by the first-order grating equation:

λ
n0 sin θ0 dif = n sin θdif + , (9.18)
Λ

where

n0 = refractive index of the exit medium at wavelength λ,

θ0 dif = angle of diffraction, referenced to the base groove (grating plane),
n = refractive index of the incidence medium at wavelength λ,
θdif = angle of incidence, referenced to the base groove (grating plane).
Fresnel Lenses 273

When the exit medium is air, the angle of diffraction θ0 dif at wavelength λ can be
calculated from
!
λ
θ dif = arcsin n sin θdif +
0
, (9.19)
Λ
where Λ = diffractive groove period.
The grating is blazed when Snell’s law is applied to the grating facet:

n sin θ = n0 sin θ0 , (9.20)

where

θ = θdif + α = angle of incidence, referenced to grating facet,

θ0 = θ0 dif + α = angle of incidence, referenced to grating facet.

The blaze wavelength λB is calculated from Eq. (9.18):

λB = Λ sin(θ0 − α) − n sin(θ − α) .
 
(9.21)

The diffractive chromatic angular dispersion ψ at the surface is defined by

ψ = ϕ0 (λ1 ) − ϕ0 (λ2 ), (9.22)

where

ϕ0 (λ1 ) = exit angle for light of wavelength λ1 ,

ϕ0 (λ2 ) = exit angle for light of wavelength λ2 .

The first-order diffraction efficiency η of the blazed grating at wavelength λ is

" !#
λB
η = sinc π2
−1 . (9.23)
λ

By adjustment of Λ, it is possible to calculate a value of ψ that offsets the refractive

dispersion in a Fresnel lens system, resulting in a significant reduction in the
chromatic dispersion of the lens.

9.12.4 Design example: achromatic dual-grooved Fresnel lens for

overhead projector
For some overhead-projector applications, it is possible to use a single-element
Fresnel lens with grooves on both sides instead of the conventional Fresnel lens
doublet.37,38 See Fig. 9.39, where f = 189.7 mm, f 0 = 434.0 mm, and the
lens diameter A = 400 mm. The lens material is acrylic plastic (nd = 1.492 at
274 Chapter 9

Figure 9.39 A dual-grooved Fresnel lens for use in an overhead projector.

λ = 0.5876 µm). The red and blue wavelengths chosen for the color correction
were λ1 = 0.656 µm (red C line), and λ2 = 0.486 µm (blue F line). The refractive
groove width W is fixed at W ≈ 0.5 mm.
As shown in Fig. 9.40, the diffractive structure is placed on the exit surface of
the lens, and the blazed diffraction grating equations are applied to this surface.
Initial grating period and blaze angles are supplied to the last groove, and by
successive iterations, the values are recalculated for each adjacent inward groove
until the angular chromatic dispersion ψ is reduced to less than 0.001 deg over the
entire lens. Simultaneously, the difference between the refracting design exit angle
(yellow light) and the diffraction exit angle is held to less than 0.006 deg over the
entire lens.
Figure 9.41(a) displays the calculated diffractive groove period Λ over the entire
lens, from 0 ≤ x ≤ 200 mm. Figure 9.41(b) plots the variation in the region where
diffraction becomes a significant factor in color correction, chosen as Λ ≤ 25 µm.
Figure 9.41(c) plots the corresponding blaze angle α variation, and Fig. 9.41(d)
shows the calculated angular dispersion for the refractive lens and the primary
and secondary colors for the dispersion-compensated lens. Refractive color may
dominate near the lens center, where the diffraction effect is small, but in any case,
the dispersion is small in this region. The diffraction efficiency η is calculated to be
≈ 0.94 over the entire lens, averaged between the three wavelengths λ, λ1 , and λ2 .

9.12.5 Achromatic zone plate using a Fresnel lens

A diffractive Fresnel zone plate can be combined with a refractive Fresnel lens
to provide a color-corrected Fresnel zone plate. Here, it is desired to correct the
Fresnel Lenses 275

Figure 9.40 Dual-grooved Fresnel lens with diffractive structure on an exit surface.

(a) (b)

(c) (d)

Figure 9.41 (a) Diffractive groove period Λ over the entire lens. (b) Diffractive groove
period Λ in a region where Λ ≤ 25 µm. (c) Blaze angle α variation in region where Λ ≤ 25 µm.
(d) Uncorrected refractive color and dispersion-compensated color over the entire lens.
276 Chapter 9

inherent chromatic aberration of the zone plate by use of the oppositely dispersing
Fresnel lens. The design of such an achromatic zone plate for use in the UV and
x-ray regions has been described and is shown in Fig. 9.42.39 The material is silicon
(refractive index n ≈ 2.37). The power of the Fresnel lens is much less than that of
the zone plate, retaining the high resolution of the zone plate. For high throughput,
the thickness of the Fresnel lens can be reduced, and any introduced phase errors
are cancelled by choice of the zone positions on the plate. The silicon zone plate
and silicon Fresnel lens are fabricated on a single silicon substrate.

9.13 Diffraction and Coherence Effects in Fresnel Lenses

9.13.1 Diffraction compensation in a Fresnel lens reflector
Where the light is incoherent and the microprism size is on the order of 0.5 to
0.1 mm, diffraction effects in Fresnel lenses are not usually of concern. However,
when the groove pitch approaches the spatial coherence length of the light,
diffraction effects are apparent. Distinct diffraction orders appear near the focal
point of the lens or reflector. An analysis has been performed on a first-surface
Fresnel reflector that minimizes these diffraction orders by randomizing the groove
width of the reflector.40 Both coherent laser light and partially coherent sunlight
were considered. Figure 9.43 represents a completely randomized reflecting groove
pattern for a 1.25-cm-aperture Fresnel reflector with a focal length of 20 cm, where
the groove geometry is specified in terms of the groove height and the groove
width. The groove width was randomly varied between 50 and 100 µm over the
lens aperture, with a mean width of 78 µm and a standard deviation of 9.217 µm.
Compared with a 78-µm fixed-groove-width reflector of the same aperture and
focal length, the systematic diffraction orders at the focus were substantially
reduced.

Figure 9.42 Achromatic zone plate with Fresnel lens dispersion compensation.39
Fresnel Lenses 277

Figure 9.43 Representation of a Fresnel lens reflector having random groove widths.40

9.13.2 Phase-optimized Fresnel lens

Figure 9.44 illustrates the refracted wavefronts through several adjacent grooves
of a Fresnel lens where collimated light is incident on the grooved surface. If the
heights of the grooves are controlled such that the refracted wavefronts align with
the refracted wavefronts of the adjacent grooves, then the emerging wavefronts
will be in phase across the lens, and all rays will superpose coherently at the focus
point for a given wavelength λ.41 This is called a phase-optimized Fresnel lens, or
“tuned” Fresnel. Here, I is the angle of incidence, I 0 is the angle of refraction, α is
the groove angle, h is the groove height, and α = I 0 . From the geometry and Snell’s
law, it is seen that λ0 = λ/ cos(I − I 0 ), and the wavefronts of adjacent grooves have
a phase difference ∆ϕ, where
!
1 1
∆ϕ = 2πh 0 − . (9.24)
λ λ
For the wavefronts of adjacent grooves to be in phase, the groove height h is
adjusted such that
!
λ
h=k , where (k = 1, 2, 3, . . .). (9.25)
n cos(I − I 0 ) − 1
For conventional Fresnel lenses at a visible wavelength, the groove depths need
be modified by only a small percentage. In fact, the height change is so small
278 Chapter 9

Figure 9.44 Phase-compensated Fresnel lens. (Adapted from Ref. 41 with permission
from the Optical Society of America.)

compared to the normal vertical groove height h ≈ w tan α that Vanucci describes
a laser interferometric process to monitor the cutting depth of the diamond tool
during the master die fabrication.41

9.13.3 Phase-optimized Fresnel lens for use in an IR intrusion detector

Figure 9.45 illustrates another method for providing a phase-optimized Fresnel
lens.42 Collimated light is incident on the plano surface, and the rays are focused
to a detector for use as an IR intrusion device. A series of zones is composed of
refracting facets, each having an aspheric curvature. For each zone, the distance d
between the facet and the focal point is such that d = (i + k)λ, where i is an integer,
k is a fraction between 0 and 1, and λ is the design wavelength. The value of i is
constant within a zone and varies between zones, while the value of k is constant
over the entire lens. Each facet is adjusted to be in phase with the preceding facet by
adjustment of the facet curvature and thickness. Shifts between zones occur when
the facet thickness approaches a target value, where the I value for the next zone
is adjusted by an integral value. A preferred lens material is polyethylene, having
a refractive index n ≈ 1.51, with transmittance in the 0.7–1.4-µ m range.

9.14 Design of a Fresnel Lens Illuminator Using Genetic

Algorithms
In Fig. 9.46, a single Fresnel lens is used to illuminate a reading area from an
array of LED light sources.43 A symmetric array of five LED light sources is fixed
Fresnel Lenses 279

Figure 9.45 Phase-optimized Fresnel lens for intrusion detector.42

Figure 9.46 Multiple-LED illumination system using a Fresnel lens.43

in position 107.25 mm above a Fresnel lens, with the grooves facing the light
sources and having a diameter of 200 mm and a fixed groove width of 0.5 mm.
A circular reading area is placed 56.5 mm below the Fresnel lens and is divided
into Nr equal-area rings for measurement of the illumination uniformity. Conical
reflectors surrounding each LED light source and between the LED array and the
Fresnel lens ensure that all emitted rays are directed to the Fresnel lens.
280 Chapter 9

Figure 9.47 (a) The optimized integer groove angles of the evolved Fresnel lens with 1000
rays emitted from each LED, compared with the groove angles of the conventional Fresnel
lens. (b) A cross section of the optimized Fresnel lens.43

The object is to calculate a set of groove angles for the Fresnel lens that
maximizes the illumination on the reading surface with an acceptable degree of
uniformity. To accomplish this, genetic algorithms (GAs) are used as a search and
optimization technique. The goal is to maximize a performance index J, where

J = I − Ip, (9.26a)

and

I = 5N − 2(5N − Rt ), (9.26b)

where I is the effective number of rays hitting the reading surface, I p is a penalty
index for less uniform distribution of light rays over the reading surface, N is the
number of uniformly distributed rays emitted by each LED, and Rt is the total
number of rays incident on the reading surface. The GAs search for a set of n
parameters that maximize J. The primary design parameters are the groove angles
of the Fresnel lens, with the initial population set being the groove angles of a
conventional single-focus Fresnel lens. A parameter represented by a set of m
binary digits is called a gene, and the n genes representing the n parameters are
Fresnel Lenses 281

formed as a binary string, called a chromosome. These chromosomes evolve by

an iterative process into generations. The chromosomes are evaluated by a fitness
criterion, and the mutated chromosomes that are fitter pass their traits to the next
generation. By this process, the best chromosome is produced, which represents
the optimal parameter value.
For this analysis, the number of groove angles Ng = 20 and Nr = 4. Rays
are traced using the TraceProTM optical design program.44 It was found that an
optimized Fresnel lens could be obtained that improved the illumination and
uniformity at the reading surface. Figure 9.47(a) shows the optimized integer
groove angles of the evolved Fresnel lens with 1000 rays emitted from each LED,
compared with the groove angles of the conventional Fresnel lens. Figure 9.47(b)
illustrates a cross section of the optimized Fresnel lens.

References
1. D. F. Vanderwerf, “Approximating the Fresnel lens,” Electro-Optical Systems
Design 14, 47–51 (1982).
2. J. B. Scarborough, Numerical Mathematical Analysis, 6th ed., 545–551, John
Hopkins Press, Baltimore, MD, (1966).
3. Y. A. Dudnikov et al., “The design of a large-diameter Fresnel condenser from
Fresnel lenses,” Sov. J. Opt. Technol. 42(5), 451–454 (1975).
4. D. F. Vanderwerf, “Ghost-image analysis of a Fresnel lens doublet,” Proc.
SPIE 1331, 143–157 (1990) [doi:10.1117/12.22674].
5. R. F. Bard, “Solar furnace,” U.S. Patent No. 3,985,118 (1976).
6. K. Yass and H. B. Curtis, “Low-cost air-mass 2 solar simulator,” NASA-TM-
X-3059, NASA John H. Glenn Research Center, Cleveland, OH (1974).
7. E. Delano, “Primary aberrations of meniscus Fresnel lenses,” J. Opt. Soc. Am.
66(12), 1317–1320 (1976).
8. E. Delano, “Primary aberration contributions for curved Fresnel lenses,” J.
Opt. Soc. Am. 68(10), 1306–1309 (1978).
9. W. A. Kleinhans, “Aberrations of curved zone plates and Fresnel lenses,” Appl.
Opt. 16(6), 1701–1704 (1977).
10. F. Erismann, “Design of a plastic aspheric Fresnel lens with a spherical shape,”
Opt. Eng. 36(4), 988–992 (1997) [doi:10.1117/1.601292].
11. D. F. Vanderwerf et al., “Linear refractor/reflector solar concentrators,” Proc.
SPIE 161, 23–28 (1978).
12. M. J. O’Neill, “Solar concentrator and energy collection system,” U.S. Patent
No. 4,069,812 (1978).
13. W. T. Welford and R. Winston, The Optics of Nonimaging Concentrators,
Academic Press, New York (1978).
282 Chapter 9

14. E. M. Kritchman et al., “Highly concentrating Fresnel lenses,” Appl. Opt.

18(15), 2688–2695 (1979).
15. E. M. Kritchman, “Color-corrected Fresnel lens for solar concentration,” Opt.
Lett. 5, 35–37 (1980).
16. S. Shanmugam, “Design of a linear Fresnel lens system for solar
photovoltaic electrical power source,” Proc. SPIE 4572, 556–564 (2001)
[doi:10.1117/12.444226].
17. E. Lorenzo and A. Luque, “Fresnel lens analysis for solar energy applications,”
Appl. Opt. 20(17), 2941–2945 (1981).
18. E. M. Kritchman, “Linear Fresnel lens with polar tracking,” Appl. Opt. 20(7),
1234–1239 (1981).
19. R. Leutz et al., “Design of a nonimaging Fresnel lens for solar concentrators,”
Solar Energy 65(6), 379–388 (1999).
20. R. H. Appeldorn, “Refracting solar energy concentrator and thin flexible
Fresnel lens,” U.S. Patent No. 4,848,319 (1989).
21. M. J. O’Neill and A. J. McDanal, “Inflatable Fresnel lenses as concentrators
for solar power,” NASA Tech Brief, Lewis Research Center (1999). See also
M. J. O’Neill, “Inflatable Fresnel lens solar concentrator for space power,”
U.S. Patent No. 6,111,190 (2000).
22. J. C. Nelson and D. F. Vanderwerf, “Catadioptric Fresnel lens,” U.S. Patent
No. 5,446,594 (1995).
23. R. N. Falge, “Lens,” U.S. Patent No. 2,023,804 (1935).
24. W. A. Parkyn and D. G. Pelka, “Compact non-imaging lens with
totally internally reflecting facets,” Proc. SPIE 1528, 70–81 (1991)
[doi:10.1117/12.49131].
25. W. A. Parkyn et al., “Faceted totally internally reflecting lens with individual
curved faces on facets,” U.S. Patent No. 5,404,869 (1995).
26. C. P. Liu et al., “Optical design of a new combo solar concentrator,” Proc.
SPIE 7423, 74230X (2009) [doi:10.1117/12.832320].
27. A. Davis et al., “Fresnel lenses in rear projection displays,” SID Digest 32,
934–937 (2001).
28. M. F. Foley, “Technical advances in microstructured plastic optics for display
applications,” SID Digest 30, 1106–1109 (1999).
29. Y. Huang, “Total internal reflection Fresnel lens and devices,” U.S. Patent
No. 7,230,758 (2007).
30. F. R. Engstrom, “Total internal reflection Fresnel lens and optical system using
the same,” U.S. Patent No. 7,350,925 (2008).
31. M. D. Peterson and J. A. Gohman, “Fresnel lens for use with rear projection
display,” U.S. Patent No. 6,804,055 (2004).
Fresnel Lenses 283

32. A. Cox, “Application of Fresnel lenses to virtual image display,” Proc. SPIE
162, 130–137 (1978).
33. D. F. Vanderwerf, “Achromatic catadioptric Fresnel lenses,” Proc. SPIE 2000,
174–183 (1993) [doi:10.1117/12.163633].
34. D. F. Vanderwerf, “Catadioptric Fresnel lens,” U.S. Patent No. 5,446,594
(1995).
35. D. F. Vanderwerf, “Overhead projector with catadioptric Fresnel lens,” U.S.
Patent No. 5,317,349 (1994).
36. K. C. Johnson, “Dispersion-compensated Fresnel lens,” U.S. Patent
No. 5,161,057 (1992).
37. D. F. Vanderwerf, “Dual grooved Fresnel lens for overhead projection,” U.S.
Patent No. 4,900,129 (1990).
38. S. K. Eckhardt, “Dual grooved Fresnel lens for overhead projection,” U.S.
Patent No. 5,803,568 (1998).
39. W. Yun and Y. Wang, “Achromatic Fresnel optics for ultraviolet and x-ray
radiation,” U.S. Patent No. 6,917,472 (2005).
40. D. A. Gregory and G. Peng, “Random facet Fresnel lenses and mirrors,” Opt.
Eng. 40(5), 713–719 (2001).
41. G. Vannucci, “A ‘tuned’ Fresnel lens,” Appl. Opt. 25(16), 2831–2834 (1986).
42. I. K. Pasco, “Fresnel lens,” U.S. Patent No. 5,151,826 (1992).
43. W.-G. Chen and C.-M. Uang, “Better reading light system with light-
emitting diodes using optimized Fresnel lens,” Opt. Eng. 45(6), 063001 (2006)
[doi:10.1117/1.2210472].
44. TracePro, Lambda Research Corporation, Littleton, MA.
Afterword
From the early ground-and-polished prism facets of glass lighthouse Fresnel lenses
to modern microreplicated flexible Fresnel lenses for solar concentrators, new
applications of prismatic optical components for light control are constantly being
developed. In addition, innovations utilizing single and compound prismatic and
reflective optical components are emerging in the fields of metrology, polarization
control, projection systems, and illumination and display lighting, among others.
Several significant applications which may indicate future trends are:
• In the field of electronic projection, handheld LCOS-, DMD-, and LCD-
based “cell phone” projectors (picoprojectors) require compact optical systems.
Large-screen projection displays are being developed using colored laser-
diode sources. Complex monolithic optics (CMO) optical engines are being
investigated for digital cinema projectors.
• For screen illumination of large flat-panel LED televisions, there are LED light-
guiding optics and new types of color-combining prisms for flat-panel OLED
(organic light-emitting diode) displays.
• In the field of direct-view microdisplays, LED backlighting optics for handheld
devices such as cell phones, digital cameras, and tablet-type personal computers
is an area of continuous development. There are new designs for LCOS-based
near-to-eye (NTE) miniature displays.
• Ranging from observation of the very large to the very small, there are giant
segmented mirror telescopes (GSMTs) and new high-contrast stereo prism
microscopes.
• Beam-shaping prisms are being designed for high-power lasers.
• New-generation Fresnel lens–based solar concentrating photovoltaic systems
(CPVs) are being designed and sold by several manufacturers.
• In the field of machine vision and image processing, multispectral prism-based
smart cameras are being developed.
• Microstructured metamaterial prisms exhibiting negative refractive index are
being fabricated, and applications are being proposed.
The techniques and examples presented in this book were intended to provide
a good background to analyze, evaluate, and understand these types of optical
applications, among others. It is further hoped that the material in this book may
have inspired readers to create novel and useful devices utilizing prismatic and
reflective optical components.

285
Index

2D tracking, 258 angular

45-deg Bauernfeind prism, 53 accuracies, 131
60-deg Bauernfeind prism, 46 dispersion, 34
90-deg annealing, 130
beam-deviating prism, 28 antireflection coatings, 9
polarization-rotating prism, 92 aplanatic
total-internal-reflecting prism array, prism spectrograph, 152
250 refractions, 152
aspect ratio, 115, 198
A astigmatism, 16, 177
Abbe reduction, 177
number, 2 autocollimator, 131
prism, 35 automated defect inspection, 159
Abbe, Ernst Karl, 35 axicon prism, 116
accumulated phase shift, 85 azimuth angle, 80, 89
achromatic
compound beam expander, 113 B
Fresnel lens, 265 backlighting, 206, 208
N-prism beam expander, 111 device, 202
retarder, 77 Bartholinus, Erasmus, 61
afocal telescopic system, 161 base-angle tolerance, 131
air-spaced prism pair, 111 Bauernfeind prism
Amici prism 45-deg, 53
double, 97 60-deg, 46
roof, 40, 131, 137 beam
Amici, Giovanni, 40 combiner, 222
anamorphic catadioptric, 222, 227
beam expander, 107 compression factor, 105
expansion, 30 compressor, 105
prism expander, 104, 106
magnification, 104 achromatic compound, 113
pair, 108 anamorphic, 107
system, 103 N-prism, 111

287
Index 288

beam, (continued) chromosome, 281

fill fraction, 226 clean-up polarizer, 176
steering, 107 close-packed hexagonal array, 248
beam-deviating prism, 90-deg, 28 coaxial output, 99, 105
beamsplitter, 222 cold-cathode fluorescent tube, 210
beam-splitting collinear output, 98, 106
coating, 162 wavelength tuning, 99
film, nonpolarizing, 74 color-corrected Fresnel zone plate,
prism, 116 274
biaxial birefringent polymer layers, 70 coma, 16
binocular complex reflection coefficient, 87
head-mounted display, 159 compound parabolic
surgical loupe, 160 concentrator, 215
birefringence, 61 reflector, 183
birefringent illuminator, 184
film, 220 compound-wedge Risley-type prism,
polarizing film, 71 144
blaze compression molding, 265
angle, 271 concentration ratio, 258
wavelength, 273 constant deviation
blind spot, on-axis, 144 angle, 37
block polishing, 135 dispersing prism, 36
Brewster’s angle, 6, 44, 64, 68, 104, critical angle, 7
216 cube-corner
Pellin–Broca prism, 37 reflector
wedge, 42 hollow, 56
Brewster, David, 6 solid-glass, 58
brightness enhancement, 210 retardance, 80
film, 210, 213 cumulative dispersion, 109
dual, 217 curved
Broca, André, 36 catadioptric Fresnel lens, 262
linear Fresnel lens, 255, 257
C
calcite, 61 D
Cartesian polarizing beamsplitter, 176 deceptive attractors, 127
wide-angle, 72 degree
catadioptric of freedom, 137
beam combiner, 222, 227 of polarization, 90
Fresnel lens, 261, 267, 270, 271 depolarization effect, 169
prism, 113 deviation angle, 33, 237, 238
change in parity, 19 diamond-turning machine, 228
charge-coupled device matrix array, dichroic layer, 165
153 dielectric optical materials, 3
chromatic dispersion, 264, 272 differential interference contrast, 67
289 Index

diffraction E
compensation, 276 echelle spectrograph, 137
efficiency, 273 edge-ray principle, 257, 258, 260, 264
diffractive effective
groove period, 271 aperture, 56
structure, 271 f /#, 9
digital electric field vectors, 4
elliptical light guide, 199
light processing, 170
ellipticity, 89, 168
projector, 170
entrance aperture
micromirror device, 170 hexagonal, 230
direction equilateral
angles, 19, 52 prism, 33
cosine, 11, 187 triangle-entrance aperture, 228
directional output, 202, 203 e-rays, 61
direct-view étendue, 183
display, 159 preservation, 184
system, 47 evolutionary prism design, 126
direct-vision prism, 97 exit pupil, 159
dispersing prism, 33, 152 experiential design, 124
dispersion equations, 2 external reflection, 5
divergence angle, 197 extractor, 196, 202
double extraordinary refractive index, 61
eyepiece, 159–161
Amici prism, 97
Dove prism, 44 F
Fresnel rhomb, 77 f /#, 9
Fresnel rhomb polarization rotator, fabrication
93 error, 229
isosceles total-internal-reflecting methods, 135
prism, 124 of a Penta prism, 135
refraction, 61 Fermat’s principle, 1
double-pass field of view, 160
laser interferometer, 149 figure, 130
projection, 168 film
birefringent, 220
Dove prism, 42, 157
polarizing, 71
double, 44
brightness-enhancement, 210, 213
roof, 43 dual-brightness-enhancement, 217
rotating, 131 multilayer polarizing beamsplitter,
Dove, Heinrich Wilhelm, 42 176
dual-brightness-enhancement film, multilayer thin, 9
217 nonpolarizing beam-splitting, 74
dual-element Fresnel lens, 242 reflecting polarizer, 70
dual-grooved Fresnel lens, 273 turning, 206
Index 290

fire polishing, 137 G

Fizeau interferometer, 136
flare reduction, 160
flatness, 130
flexible linear refracting Fresnel lens,
260
fluorescent tube, cold-cathode, 210
four-mirror
90-deg polarization rotator, 94
beam-displacing prism, 25
fourth-order reflection matrix, 25,
125, 189
Fresnel
lens
achromatic, 265
catadioptric, 261, 267, 270, 271
collimator, 240
curved catadioptric, 262
curved linear, 255, 257
doublet, 242, 244, 245
dual-element, 242
dual-grooved, 273
flexible linear refracting, 260
inflatable flexible, 260
meniscus, 248
phase-optimized, 277, 278
planar circular, 248
positive aspheric, 237
reflective, 245
reflective linear, 250
reflector, 247
refractive, 237, 264
total-internal-reflecting, 261, 264
reflection, 4
reflector, linear, 250
rhomb, 76
double, 77
quarter-wave double, 77
rhomb polarization rotator
double, 93
zone plate, color-corrected, 274
Fresnel, Augustin Jean, 4, 76
frustrated total internal reflection, 53,
73, 157
Gaussian-light-beam profile, 162
gene, 280
genetic algorithm, 126, 280
geometric
concentration ratio, 257, 259
efficiency, 56
Glan–Foucault prism, 63
Glan–Taylor prism, 64
Glan–Thompson prism, 64
glass cube-corner prism, 126
glass plate, tilted, 13
goniometer, 131
gyroscope, ring-laser, 153
H
half-wave
plate, 176
rhomb retarder, 77
Harting–Dove prism, 42
hemispheric model, 190
hexagonal
aperture, 58
array, close-packed, 248
entrance aperture, 230
high-reflectance surfaces, 9
histogram, 255
hollow
cube-corner reflector, 56
light pipe, 179
homogeneous transformation matrix,
125
I
image contrast, 166, 169
index-matching fluid, 159
indicatrix, 61
inflatable flexible Fresnel lens, 260
injection molding, 137, 265
interference, thin-film, 73
interferometer
Fizeau, 136
laser, 150
Nomarski polarized, 151
polarization, 149
291 Index

internal reflection, 6 liquid crystal polymer plate

intersection coordinate, 13 compensator, 175
matrix, 19 Littrow
inversion, 18 30/60/90-deg reflecting prism, 47
inverting prism, 19 laser-dispersion prism, 47
irradiance, 90 prism, 46
isosceles Littrow, Joseph Johann, 46
roof prism, 187 Littrow-type prism, 104
total-internal-reflecting prism, 122 longitudinal spherical aberration, 16
isotropic, 61 loupe, binocular surgical, 160
K M
Kessler direct-vision prism, 98 MacNeille
kinematic mount, 137 pair, 217
Knoop test, 130 polarizing beamsplitter, 175
polarizing beamsplitter cube, 69
L
stack, 71
ladar guidance, 145
master die, 228, 229
laser
material properties, 130
diode, 106
interferometer, 150 matrix
double-pass, 149 array, charge-coupled device, 153
laser-dispersing prism, 44 fourth-order reflection, 25, 125, 189
law of reflection, 17 homogenous transformation, 125
LCD projection displays, 166 intersection coordinate, 19
least-squares fit, 243 methods for design, 125
LED light source, 278 system, 28
left handedness, 90 transposed reflection, 23
Leman prism, 27 meniscus Fresnel lens, 248
Leman–Sprenger prism, 27 meridional plane, 10
light merit function, 103
pipe, 179, 195 micrometer eyepiece, 132
transporter, 196 micromirror, 170
tube, 197 microprismatic
light-directing total-internal-reflecting array, 187
prism, 171 light homogenizer, 181
light-guide luminaire, 202 microreplication, 228
light-guiding prism, 158 microscope, 67
linear microstructured anisotropic layer, 220
Fresnel reflector, 250 minimum deviation, 34, 101, 243,
scanner, 146 255, 260
linear-focus solar concentrator, 250 angle, 119
liquid crystal on silicon mirror, roof, 25
imager, 160, 173 moiré pattern, 211
spatial-light modulator, 166 mounting tolerance requirements, 121
Index 292

multilayer P
polarizing beamsplitter film, 176 paraxial approximation, 143
thin films, 9 Pechan
multipass optical cell, 150 prism, 53, 124, 126
multiprism dispersive compressors, roof prism, 54
109 Pellin, Phillippe, 36
Pellin–Broca prism, 36, 124
N Penta prism, 38, 124, 125, 131, 135
negative phase
dispersion, 101 conjugate mirror, 149
group velocity dispersion, 101 difference, 155
uniaxial calcite, 61 quadrature, 149
Nicol prism, 61
shifter, 43
Nicol, William, 61
phase-coated total-internal-reflecting
Nomarski
retarders, 80
polarized interferometer, 151
phase-compensation coating, 89
prism, 67
phase-correction coating, 55
Nomarski, Georges, 67
phase-optimized
nonbirefringent glass, 72
coating, 159
noncubic polarizing beamsplitter, 167
Fresnel lens, 277, 278
nonkinematic mount, 137
phase-shift
nonpolarizing beam-splitting film, 74
coating, 169
nonsequential ray tracing, 180, 191
compensating coatings, 173
normalized
phase-shifting prism, 116
output, 203
Philips prism, 165, 166
transmission, 191
photoreplication, 221, 265
numerical aperture, 168, 183
Pierre de Fermat, 1
O planar
OASIS coating, 53 circular Fresnel lens, 248
oblique rays, 10 polarization converter, 227
occlusions, 130 plane of incidence, 1
off-axis rear projection, 265 polar tracking, 259
on-axis blind spot, 144 polarization
optical beamsplitters, wire-grid, 178
cell, multipass, 150 converter, 75, 226
disk reader, 155 ellipse, 76
lighting film, 195 interferometer, 149
path distance, 116 pupil map, 78
o-rays, 61 recycling, 219
ordinary refractive index, 61 rotating prism, 90-deg, 92
orientation of viewed images, 18 rotation, 167
orthogonal output, 104 rotator, four-mirror 90-deg, 94
overhead projector, 226, 242, 245, wavelength shift, 170
262, 273 polarization-preserving prism, 82
293 Index

polarized backlight, 219 dispersing, 33, 152

polarizer, wire-grid, 72, 177 constant deviation, 36
polarizing beamsplitter, 159 double isosceles
Cartesian, 176 total-internal-reflecting, 124
cube, 68 Dove, 42, 157
wire-grid, 73 double, 44
film, multilayer, 176 roof, 43
MacNeille, 175 rotating, 127, 131
noncubic, 167 equilateral, 33
wide-angle Cartesian, 72 four-mirror beam-displacing, 25
wire-grid, 178 Glan–Foucault, 63
Porro prism, 40 Glan–Taylor, 64
Type I, 41 Glan–Thompson, 64
Type II, 41 glass cube-corner, 126
Porro, Ignazio, 40 Harting–Dove, 42
Porro–Abbe prism, 41 isosceles
positional mountings, 138 roof, 187
positive total-internal-reflecting, 122
aspheric Fresnel lens, 237 Kessler direct-vision, 98
dispersion, 101 laser-dispersing, 44
p-polarized light, 4, 64 Leman, 27
prepolarizer, 159, 178 Leman–Sprenger, 27
pressed-glass optical prism, 137 light-directing
primitive sections, 127 total-internal-reflecting, 171
principal Littrow, 46
axis, 61 30/60/90-deg reflecting, 47
plane, 61 laser-dispersion, 47
section, 65 reflecting, 124
prism Littrow-type, 104
45-deg Bauernfeind, 53 magnification, anamorphic, 104
60-deg Bauernfeind, 46 Nicol, 61
90-deg beam-deviating, 28 Nomarski, 67
90-deg polarization-rotating, 92 pair
Amici air-spaced, 111
double, 97 anamorphic, 108
roof, 40, 131, 137 refracting/total-internal-
array, 90-deg reflecting,
total-internal-reflecting, 250 113
axicon, 116 Pechan, 53, 124, 126
beam-splitting, 116 roof, 54
Brewster’s-angle Pellin–Broca, 37 Pellin–Broca, 36, 124
catadioptric, 113 Penta, 38, 125, 131, 135
design, evolutionary, 126 Philips, 165
direct-vision, 97 polarization-preserving, 82
Index 294

prism, (continued) Q
Porro, 40 QuadCubeTM architecture, 174
Type I, 41 quarter-wave
Type II, 41 double Fresnel rhomb, 77
Porro–Abbe, 41 retarder, 175, 227
pressed-glass optical, 137 rhomb retarder, 77
quality, 130
reflective dispersing, 98 R
reflector,solid-glass cube-corner, 58 randomly polarized light, 89
reversion, 125 ray tracing, nonsequential, 191
rhomboid, 127 readable image, 19
rear-projection displays, 264
right-angle, 39, 161
reflecting
Risley, 144
Littrow prism, 124
Risley-type
polarizer
compound-wedge, 144
film, 70, 217
roof, 54
sheet, 220, 228
rotationally symmetric, 116
reflection
Schmidt, 49
and translation of skew rays, 17
Schmidt–Pechan, 54
coatings, 9
sections, right-angle, 124
coefficient, complex, 87
spectograph, aplanatic, 152 phase shifts, 7
spectroscope, 35 reflective
Sprenger–Leman, 27 axicon, two-piece, 87
switch, 153 dispersing prism, 98
system, anamorphic, 103 Fresnel lens, 245
three-mirror beam-displacing, 21 LCD imager, 166, 174
total-internal-reflecting, 113 linear Fresnel lens, 250
double isosceles, 124 reflector, hollow cube-corner, 56
light directing, 171 refracting/total-internal-reflecting
trichroic separation, 165 prism pair, 113
wedge, 107, 155 refraction
anamorphic compressor, 107 and translation of skew rays, 10
as scanners, 143 matrix, 24, 187
Wollaston, 66, 125, 150, 151 refractive
x-cube, 168 Fresnel lens, 237, 264
prismatic index, 1
hollow light guide, 195, 197 extraordinary, 61
sheets, 187 ordinary, 61
prism-based readout, 154 relative phase shift, 8, 85
pyramidal retarded stack filter, 176
angle tolerance, 131 retarder, 75
error, 133 achromatic, 77
faceted scanner, 141 quarter-wave, 175, 227
295 Index

retrace interval, 141 semiconductor laser light source, 155

retroreflection, 191 semikinematic mount, 137
efficiency, 57 Sherman-type prism, 98
reversion, 19 sidelighting, 213
prism, 125 skew rays, 10
rhomb retarder reflection and translation of, 17
half-wave, 77 refraction and translation of, 10
quarter-wave, 77 skew-ray depolarization, 166, 175
rhomboid prism, 127 Snell, Willebrord, 1
right handedness, 90 Snell’s law, 1
right-angle prism, 39, 161 solar
sections, 124 collimation acceptance angle, 258
right-handed image, 19 concentrator, 250
ring-laser gyroscope, 153 linear-focus, 250
Risley prism, 144 furnace, 248
scan patterns, 144 simulator, 248
Risley-type prism solar-energy concentration, 248
compound-wedge, 144 solid light pipe, 180
Risley-type prism, compound-wedge, solid-glass cube-corner reflector, 58
144 spatial coherence length, 276
Rochon, Alexis Marie, 67 spectograph, aplanatic prism, 152
roof spherical aberration, longitudinal, 16
Dove prism, 43 spinner, 141
mirror, 25 s-polarized light, 4, 64
prism, 54 spot-focus concentrator, 253
Amici, 40, 131, 137 Sprenger–Leman prism, 27
array, 190 square plate, rotating refracting, 146
array, 90-deg, 192 Stokes parameters, 89–91
isosceles, 187 Stokes, George Gabriel, 89
rotating stress birefringence, 130
Dove prism, 127, 131 striae, 130
refracting square plate, 146 surface quality, 130
rotationally symmetric prism, 116 system matrix, 28, 189

S T
sagittal tabletop lectern projector, 127
plane, 16 tangential
ray, 52 plane, 15
Schmidt prism, 49 ray, 50
Schmidt, Bernhardt Woldemar, 49 telescopic laser range finders, 161
Schmidt–Pechan prism, 54, 160 test plate, 130
scratch and dig standard, 130 thermal environment, 121
sectional element, 122 thin-film interference, 73
Index 296

three-mirror beam-displacing prism, two-piece reflective axicon, 87

21 Twyman-Green interferometer, 133,
tilted glass plate, 13 137
total internal reflection, 6 Type I Porro prism, 41
frustrated, 53, 157 Type II Porro prism, 41
phase changes, 76
total-internal-reflecting U
deviator, 122 uniaxial stretching, 70
extractor, 208
Fresnel lens, 261, 264 V
prism, 113 variable achromatic beam deviator, 39
array, 90-deg, 250 virtual image display, 267
double isosceles, 124
isosceles, 122 W
light-directing, 171 wavelength-compensation plates, 167
retarders, phase-coated, 80 wedge prism, 107, 155
touch switch, 157 anamorphic compressor, 107
transmissive extractor, 207 as scanners, 143
transport wet out, 212
efficiency, 198 wide-angle Cartesian polarizing
factor, 198 beamsplitter, 72
transposed reflection matrix, 23 wire-grid
trapezoidal polarizer, 72, 159, 177
microprism, 208 polarizing beamsplitter, 178
prism, 161 cube, 73
triangular aperture cube-corner, 57 Wollaston prism, 66, 125, 150, 151
trichroic separation prism, 165 Wollaston, William Hyde, 66
tunnel
diagram, 30, 43 X
integrator, 179 x-cube prism, 168
turning film, 206 ν-number, 2
 Dennis F. Vanderwerf has been involved in the fields of optics
and optical engineering for over 35 years. He has held technical
positions at the Roswell Park Cancer Institute, Buffalo, New
York, the NASA John H. Glenn Research Center, Cleveland,
Ohio, and the 3M Company in St. Paul, Minnesota and Austin,
Texas. He has worked in the areas of crystallography, solar
radiometry, flow visualization optics, optical solar concentrator
design, lens and projection systems optical design, new product
development, quality assurance, and intellectual property management. He has
received the NASA Apollo Achievement Award and the 3M Corporate Circle of
Technical Excellence Award. He holds a BS in physics from Canisius College,
Buffalo, New York, an MS in physics from Ohio State University, Columbus,
Ohio, and an MBA from the University of St. Thomas, St. Paul, Minnesota.
He has numerous optical journal, trade magazine, and conference proceedings
publications, and is a named or sole inventor on 29 U.S. patents in the fields of
optics and optical design. His current interests lie in scientific writing, novel optical
technology applications, and science and math education. Dennis F. Vanderwerf
resides in Austin, Texas.