Tutorial Texts Series

• Metrics for High-Quality Specular Surfaces, Lionel R. Baker, TT65

• Field Mathematics for Electromagnetics, Photonics, and Materials Science, Bernard Maxum, TT64
• High-Fidelity Medical Imaging Displays, Aldo Badano, Michael J. Flynn, and Jerzy Kanicki, TT63
• Diffractive Optics–Design, Fabrication, and Test, Donald C. O’Shea, Thomas J. Suleski, Alan D.
Kathman, and Dennis W. Prather, TT62
• Fourier-Transform Spectroscopy Instrumentation Engineering, Vidi Saptari, TT61
• The Power- and Energy-Handling Capability of Optical Materials, Components, and Systems, Roger M.
Wood, TT60
• Hands-on Morphological Image Processing, Edward R. Dougherty, Roberto A. Lotufo, TT59
• Integrated Optomechanical Analysis, Keith B. Doyle, Victor L. Genberg, Gregory J. Michels, Vol. TT58
• Thin-Film Design Modulated Thickness and Other Stopband Design Methods, Bruce Perilloux, Vol. TT57
• Optische Grundlagen für Infrarotsysteme, Max J. Riedl, Vol. TT56
• An Engineering Introduction to Biotechnology, J. Patrick Fitch, Vol. TT55
• Image Performance in CRT Displays, Kenneth Compton, Vol. TT54
• Introduction to Laser Diode-Pumped Solid State Lasers, Richard Scheps, Vol. TT53
• Modulation Transfer Function in Optical and Electro-Optical Systems, Glenn D. Boreman, Vol. TT52
• Uncooled Thermal Imaging Arrays, Systems, and Applications, Paul W. Kruse, Vol. TT51
• Fundamentals of Antennas, Christos G. Christodoulou and Parveen Wahid, Vol. TT50
• Basics of Spectroscopy, David W. Ball, Vol. TT49
• Optical Design Fundamentals for Infrared Systems, Second Edition, Max J. Riedl, Vol. TT48
• Resolution Enhancement Techniques in Optical Lithography, Alfred Kwok-Kit Wong, Vol. TT47
• Copper Interconnect Technology, Christoph Steinbrüchel and Barry L. Chin, Vol. TT46
• Optical Design for Visual Systems, Bruce H. Walker, Vol. TT45
• Fundamentals of Contamination Control, Alan C. Tribble, Vol. TT44
• Evolutionary Computation Principles and Practice for Signal Processing, David Fogel, Vol. TT43
• Infrared Optics and Zoom Lenses, Allen Mann, Vol. TT42
• Introduction to Adaptive Optics, Robert K. Tyson, Vol. TT41
• Fractal and Wavelet Image Compression Techniques, Stephen Welstead, Vol. TT40
• Analysis of Sampled Imaging Systems, R. H. Vollmerhausen and R. G. Driggers, Vol. TT39
• Tissue Optics Light Scattering Methods and Instruments for Medical Diagnosis, Valery Tuchin, Vol. TT38
• Fundamentos de Electro-Óptica para Ingenieros, Glenn D. Boreman, translated by Javier Alda, Vol. TT37
• Infrared Design Examples, William L. Wolfe, Vol. TT36
• Sensor and Data Fusion Concepts and Applications, Second Edition, L. A. Klein, Vol. TT35
• Practical Applications of Infrared Thermal Sensing and Imaging Equipment, Second Edition, Herbert
Kaplan, Vol. TT34
• Fundamentals of Machine Vision, Harley R. Myler, Vol. TT33
• Design and Mounting of Prisms and Small Mirrors in Optical Instruments, Paul R. Yoder, Jr., Vol. TT32
• Basic Electro-Optics for Electrical Engineers, Glenn D. Boreman, Vol. TT31
• Optical Engineering Fundamentals, Bruce H. Walker, Vol. TT30
• Introduction to Radiometry, William L. Wolfe, Vol. TT29
• Lithography Process Control, Harry J. Levinson, Vol. TT28
• An Introduction to Interpretation of Graphic Images, Sergey Ablameyko, Vol. TT27
• Thermal Infrared Characterization of Ground Targets and Backgrounds, P. Jacobs, Vol. TT26
• Introduction to Imaging Spectrometers, William L. Wolfe, Vol. TT25
• Introduction to Infrared System Design, William L. Wolfe, Vol. TT24
• Introduction to Computer-based Imaging Systems, D. Sinha, E. R. Dougherty, Vol. TT23
• Optical Communication Receiver Design, Stephen B. Alexander, Vol. TT22
• Mounting Lenses in Optical Instruments, Paul R. Yoder, Jr., Vol. TT21
Tutorial Texts in Optical Engineering
Volume TT65

Bellingham, Washington USA

Library of Congress Cataloging-in-Publication Data

Baker, L. R. (Lionel R.)

Metrics for high-quality specular surfaces / Lionel R. Baker.
p. cm. -- (Tutorial texts in optical engineering ; v. TT65)
Includes bibliographical references and index.
ISBN 0-8194-5576-8
1. Surfaces (Technology)--Measurement. 2. Optical measurements. 3. Optical instruments--
Quality control. I. Title. II. Series.

TA418.7.B35 2004
681'.428--dc22 2004021009

Published by

SPIE—The International Society for Optical Engineering

P.O. Box 10
Bellingham, Washington 98227-0010 USA
Phone: +1 360 676 3290
Fax: +1 360 647 1445
Email: [email protected]
Web: http://spie.org

Copyright © 2004 The 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.

 Introduction to the Series
Since its conception in 1989, the Tutorial Texts series has grown to more than 60
titles covering many diverse fields of science and engineering. When the series
was started, the goal of the series was to provide a way to make the material
presented in SPIE short courses available to those who could not attend, and to
provide a reference text for those who could. Many of the texts in this series are
generated from notes that were presented during these short courses. But as
stand-alone documents, short course notes do not generally serve the student or
reader well. Short course notes typically are developed on the assumption that
supporting material will be presented verbally to complement the notes, which
are generally written in summary form to highlight key technical topics and
therefore are not intended as stand-alone documents. Additionally, the figures,
tables, and other graphically formatted information accompanying the notes
require the further explanation given during the instructor’s lecture. Thus, by
adding the appropriate detail presented during the lecture, the course material can
be read and used independently in a tutorial fashion.

What separates the books in this series from other technical monographs and
textbooks is the way in which the material is presented. To keep in line with the
tutorial nature of the series, many of the topics presented in these texts are
followed by detailed examples that further explain the concepts presented. Many
pictures and illustrations are included with each text and, where appropriate,
tabular reference data are also included.

The topics within the series have grown from the initial areas of geometrical
optics, optical detectors, and image processing to include the emerging fields of
nanotechnology, biomedical optics, and micromachining. When a proposal for a
text is received, each proposal is evaluated to determine the relevance of the
proposed topic. This initial reviewing process has been very helpful to authors in
identifying, early in the writing process, the need for additional material or other
changes in approach that would serve to strengthen the text. Once a manuscript is
completed, it is peer reviewed to ensure that chapters communicate accurately the
essential ingredients of the processes and technologies under discussion.

It is my goal to maintain the style and quality of books in the series, and to
further expand the topic areas to include new emerging fields as they become of
interest to our reading audience.

Arthur R. Weeks, Jr.

University of Central Florida
Contents

Preface / xi

List of Abbreviations / xiii

List of Symbols / xiv

Chapter 1 Surface Metrics / 1

1.1 Introduction / 1
1.2 Why Measure Surfaces? / 4
1.2.1 System function / 4
1.2.2 Appearance / 5
1.2.3 Manufacturing efficiency / 6
1.2.4 Benefits / 6
1.3 Definition of Surface Metrics / 6
1.3.1 Surface metrics influencing quality / 8
1.3.2 Causes of defects in surface topography / 9
1.4 Chapter Conclusions / 10
References / 10

Chapter 2 Surface Form / 13

2.1 Introduction / 13
2.2 Optical Height Probes / 14
2.3 Optical Slope Probe / 16
2.4 Interferometers / 16
2.4.1 Twyman-Green interferometer / 17
2.4.2 Laser Fizeau interferometer / 18
2.4.3 Four-step measurement of phase / 19
2.4.4 Interlaboratory comparison of interferometers / 21
2.4.5 Interferometer error sources and calibration / 21
2.5 Form Tolerances / 23
2.5.1 Approximating a spherical surface / 25
2.5.2 Sagitta error / 25
2.5.3 Irregularity function / 25
2.5.4 Irregularity / 25
2.5.5 Approximating an aspheric surface / 25
2.5.6 Rotationally symmetric irregularity / 25
2.5.7 Total RMS deviation (RMSt ) / 26
2.5.8 RMS irregularity (RMSi ) / 26
2.5.9 RMS asymmetry (RMSa ) / 26
vii
viii Contents

2.5.10 Form indication on drawings / 26

2.6 Chapter Conclusions / 26
References / 27

Chapter 3 Surface Roughness / 29

3.1 Introduction / 29
3.2 Typical Optical Component Roughness Values / 31
3.3 Deterministic Methods / 32
3.3.1 Stylus method / 33
3.3.2 Profilometry metrics / 33
3.3.3 Microinterferometer / 34
3.4 Parametric Methods / 34
3.4.1 Surface point spread functions / 34
3.4.2 Total integrated scatter measurement / 36
3.5 Surface Roughness Indications in Drawings / 37
3.6 Chapter Conclusions / 37
References / 39

Chapter 4 Surface Waviness / 41

4.1 Introduction / 41
4.2 Fourier Analysis of Height Profile / 43
4.3 Spatial Frequency Zones / 43
4.4 Computation of Texture / 43
4.5 Chapter Conclusions / 45
References / 45

Chapter 5 Surface Imperfections / 47

5.1 Introduction / 47
5.2 Imperfections and Subconscious Thoughts / 49
5.3 Effect of Surface Imperfections / 49
5.4 Impact of Imperfections on Market Access / 50
5.5 Description of Imperfections / 51
5.5.1 Terminology / 51
5.5.2 Size of imperfections / 51
5.5.3 Substrates/materials / 51
5.5.4 Location / 51
5.5.5 Characteristics / 52
5.6 Influence of Imperfections on Quality / 52
5.6.1 Cosmetic influence / 52
5.6.2 Functional influence / 52
5.7 Causes of Imperfections / 53
5.8 Reduction of Damage / 53
5.9 Imperfection Measurement / 54
Contents ix

5.9.1 Why measure imperfections? / 54

5.9.2 Characterization and measurement of imperfections / 54
5.10 Comparison of Measurement Methods / 55
5.11 Imperfection Size Versus Visibility / 56
5.11.1 Surface step as an imperfection / 56
5.11.2 Step measurement by interferometry / 58
5.12 The Eye as a Sensor / 58
5.12.1 Benefits / 58
5.12.2 Disbenefits / 59
5.13 Disbenefits of Inspection / 59
5.14 National Standards for Scratch Assessment / 60
5.14.1 United States / 60
5.14.2 Germany / 61
5.14.3 France / 61
5.14.4 United Kingdom / 61
5.15 Level of Agreement Achieved Using National Standards / 61
5.16 Scratch Reference Standards / 63
5.17 Target Specification for Imperfection Measurement / 65
5.18 Need for Standards / 66
5.19 ISO TC 172 Optics and Optical Instruments / 66
5.20 Comparison of Two Methods Proposed by ISO in 1996 / 67
5.20.1 Method I / 67
5.20.2 Method II / 67
5.20.3 Comparison of Methods I and II / 67
5.21 Chapter Conclusions / 68
References / 68

Chapter 6 Measurement of Imperfections by Obscuration / 71

6.1 Introduction / 71
6.2 Optical Component Inspection / 71
6.3 Radiometric Obscuration by Imperfections / 71
6.4 Calibration Graticules / 73
6.5 LEW and SED Measurement Requirements / 74
6.6 LEW and SED Simple Viewing System / 74
6.7 Analogue Microscope Image Comparator (AMIC) / 75
6.7.1 Description / 75
6.7.2 Theory / 77
6.7.3 Method of operation / 78
6.8 Digital Microscope Image Comparator (DMIC) / 79
6.8.1 Description / 79
6.8.2 Results and discussion / 83
6.9 Chapter Conclusions / 84
References / 84
x Contents

Chapter 7 Surface Imperfection Quality Control / 85

7.1 Introduction / 85
7.2 Survey of Tolerances / 85
7.2.1 British Standard BS4301 (1991) / 85
7.2.2 American Standard MIL-O-13830A:1963 / 86
7.2.3 German Standard DIN 3140: Part 7, 1978 / 86
7.2.4 French Standard / 86
7.3 Acceptable Thresholds for Scratches and Roughness / 87
7.4 Inspection and Measurement Flow Diagram / 89
7.5 Chapter Conclusions / 90
References / 91

Chapter 8 Far-Field Nanoscopy / 93

8.1 Introduction / 93
8.2 Comparison between Subjective and Objective Measurements
of Imperfections / 93
8.3 Relative Contrast of Standard Scratches / 94
8.4 Measurement of Imperfections and Contamination in Assemblies / 96
8.5 Measurement of Imperfections in Coatings / 100
8.6 Use of MIC to Measure Surface Texture / 101
8.7 Use of MIC to Examine Phase Objects / 107
8.8 Use of MIC in AC Mode / 111
8.9 Use of MIC On-Machine / 114
8.10 Chapter Conclusions / 115
References / 116

Chapter 9 Strip Product Inspection / 117

9.1 Introduction / 117
9.2 Laser Beam Scanners / 118
9.3 Camera Inspection Systems / 122
9.4 Chapter Conclusions / 123
Acknowledgment / 124
References / 124

Appendix 1. Quality Metrics for Digital Cameras / 125

Appendix 2. Surface Cleaning / 135

Glossary / 139

Contacts and Further Reading / 145

Index / 147
Preface
This book provides a basic working knowledge of the definition, measurement, and
standardization of a number of different metrics used to characterize high-quality
specular surfaces. It should be of interest to optical component and systems design-
ers, quality assurance engineers, and designers of quality assurance instrumenta-
tion, as well as those with a need to set acceptance thresholds for surface form and
finish in accordance with ISO standards. The quality of finish of specular or mir-
rorlike surfaces is also of more general interest to engineers from the electronics
and precision mechanical industries. These include the automotive, defense, phar-
maceutical, and biotechnology sectors, as well as those concerned with the new
technologies of integrated optics and microfabrication. Products with a specular
finish also requiring surface quality control include, for example, glossy computer
printing paper, plastics laminates, and rolled strip. The quality of nonspecular sur-
faces that may be ground or painted and diffuse light are mostly excluded from this
study.
The design and working of traditional as well as some new techniques and
instrumentation for the inspection of specular surfaces and for the measurement
of quality metrics applied to high-quality surfaces are described. It is hoped that
an understanding of the provisions and methods of operation of recently available
international standards, including the setting of form and finish tolerances, will
provide insight into the changing needs of those concerned with the design and
manufacture of specular surfaces with characteristics defined either by function or
appearance.
A parametric approach to the characterization of the total topography of a sur-
face leads on to a comparison of methods for the measurement of form and finish
and to a better understanding, based on recent research, of the calibration and prac-
tical use of instruments for measuring the basic metrics of form, texture, and im-
perfections. New metrics for the objective measurement of imperfections, adopted
in a recently published ISO standard, are defined, and designs of analogue and
digital comparison microscopy systems for their measurement are described. Their
embodiment in optical component quality control procedures is also covered. It
is anticipated that some of the new tools described here may have application in
the characterization of the surface appearance of a wide range of products with a
specular finish.
The potential for use of relatively low cost comparison microscopes for mea-
suring other parameters such as contamination, microtopography, and surface tex-
ture is explored and supported by practical sections involving the measurement of
image luminosity by digital camera. Due to the nanometric sensitivity to surface-
height variations of these methods, the term far-field nanoscopy (FFN), as opposed
to near-field nanoscopy (NFN), has been used as a generic title involving the use
of an instrument called a nanoscope. A study of methods for measuring the spatial
image quality of digital cameras includes, in an appendix, a proposal for a new
xi
xii Preface

spatial image quality metric based on the measurement of optimum print width.
A simple method for assessing the contrast resolution of digital cameras is also
described. A review of surface-cleaning techniques needed for the realization of
these procedures is included in a second appendix. The book ends with a chapter
on the latest automated laser beam scanning techniques used for inspecting very
wide specular surfaces found typically in strip-product manufacture.
The content of this book is based on collaborative research and discussions,
extending over several years, with workers from industries around the world and
with colleagues serving on committees of the British Standards Institution (BSI)
and the International Organization for Standardization (ISO). Much of the origi-
nal research on the measurement of imperfections, by analogue comparison mi-
croscopy, reported here, was undertaken while I was working at Sira Ltd., although
new digital methods, also described, have been developed since. I would like to
thank all of my colleagues for their support and my wife, Dorothy, for her limitless
understanding and patience.
The principal objective in writing this book is to stimulate and motivate others
to carry forward research on metrics for characterizing the function and appear-
ance of specular surfaces that has occupied much of my own time and thoughts
in recent years. Much previous work, excellent publications, and a variety of in-
struments already exist in the field of surface metrology. The niche I have started
to address here, on behalf of the industrial user, occurs between traditional visual
methods, still applied in most surface-quality assessments, and techniques employ-
ing modern, slow, computer-aided, off-machine, and costly but precise instrumen-
tation requiring laboratory accreditation. The driver has been the repeated request
from industry for fast, low-cost, noncontacting methods with traceability in support
of subjective assessments.
It is hoped that describing the methods here will encourage their further vali-
dation by industry, and lead on to the drafting of future surface measurement stan-
dards. I have tried to apply a physical and practical approach, related to the needs
of industry, by minimizing the mathematical complexity of the subject. To enable
an extended study, references are provided together with a glossary defining new
and frequently used technical terms.

L.R.B

Orpington, Kent, UK
2004
List of Abbreviations
The following is a list of the important abbreviations used.
AMIC analogue microscope image comparator
BS British Standard
CCD charge-coupled detector
CR contrast resolution
DIN Deutsche Industrie-Norm
DMIC digital microscope image comparator
FFN far-field nanoscopy
GD geometrical depth
IR infrared
ISO International Organization for Standards
LCD liquid crystal display
LEW line-equivalent width
MIC microscope image comparator
MIL U.S. military standard
MTF modulation transfer function
NFN near-field nanoscopy
NIC nominal information capacity
NPL National Physical Laboratory
OPW optimum print width
OTF optical transfer function
PPI pixel packing index
PSD power spectral density
PSF point spread function
Ra average roughness
RMS root mean square
RMSa RMS asymmetry
RMSi RMS irregularity
RMSt total RMS deviation
Rq RMS roughness
RTH Rank Taylor Hobson
SD standard deviation
SED spot-equivalent diameter
SEM scanning electron microscope
SFB spatial frequency bandwidth
SLR single lens reflex
SPM scanning probe microscope
TIC true information capacity
TIS total integrated scatter
TV television
UV ultraviolet
VGI veiling glare index
xiii
List of Symbols
This list shows some of the symbols used and the variables they most frequently
represent. In some instances they are used to represent other variables but, if so,
their use is always described in the associated text.

I light intensity
φ phase of wavefront
V modulation of fringe pattern
λ wavelength of light
L sampling length
x abscissa
y ordinate
Rq root-mean-square roughness, nm
f cycles per unit length
θ angular setting of analyzer in MIC
f/No lens aperture ratio
Hz Hertz, cycles/second
µm micrometers
nm nanometers
mm millimeters
m meter
kV kilovolts
NA numerical aperture
LEW line-equivalent width, µm
SED spot-equivalent diameter, µm
DC direct current
AC alternating current

xiv
Chapter 1
Surface Metrics
The metrology of specular surfaces demands a continuing dialogue between the dual
processes of inspection and measurement.

1.1 Introduction
The above plea by the author is made to stress the essential difference between the
two processes of inspection and measurement. The following definitions are taken
from the Chambers Dictionary.
Specular: mirrorlike
Inspect: to look into
Measure: the ascertainment of extent by comparison with a standard
The need for a clear understanding of the meaning of terms, often loosely applied in
an industrial situation, arises unavoidably when drafting standards. A measurement
standard aims to improve communication between a supplier and a customer by
codifying measurement parameters typifying current good practice. It follows that
purely subjective assessments, although essential for inspection, should be sup-
ported ultimately by an objective measurement traceable to national standards. To
help further in understanding the subject, a glossary defining a selection of impor-
tant technical terms is provided.
It should be noted here that an international standard is only published after re-
ceiving substantial agreement by an international community of experts. All stan-
dards are subject to review after five years but can be revised at any time, should
the need arise through the advance of technology or due to the discovery of er-
rors.
Optical components, with which this book is mostly concerned, usually require
a degree of surface quality unsurpassed by most manufactured products. A tradi-
tional precision mechanical engineering workshop presented with a design requir-
ing a surface-shape accuracy of 10 nm and residual RMS roughness of 1 nm, even
accepting the vagueness of this specification, would probably be less than keen to
quote. Optical workshops, however, have a long tradition of working to this level
of drawing tolerance. The precise control of the passage of light through an optical
system requires the use of tolerances related to its wavelength. The tolerances on
mechanical components, however, where fit, lubrication retention, and wear rates
1
2 Chapter 1

are of primary concern, may be one or two orders of magnitude less demanding.
The materials used in the optical workshop, however, are usually brittle and so
the design of surface-generating machinery has progressed over the years along
different lines for the two disciplines.
Recent developments in photonics, however, are having the effect of bringing
together these hitherto largely separate disciplines. Machine bearings with greater
stiffness and operating precision, improved machine mounting, on-machine mea-
surement, robotic control, and improved surface-generating systems are now giv-
ing rise to new machines embodying selected features from mechanical and opti-
cal workshop technologies. Innovative techniques1,2 for the generation of precision
optical surfaces are reviewed from time to time.
In spite of advances in technology, reaching the high degree of perfection re-
quired by optical surfaces is costly and demands detailed knowledge of a variety
of processing technologies combined with considerable operator skill. Since the
latest techniques of topographical analysis indicate that no practical surface can
be perfect, residual errors must be quantifiable and toleranced in terms that can be
related to quality.
Here we shall review and explain current thinking on the process of inspec-
tion of high-quality surfaces and on the measurement and standardization of total
surface topography in terms of its constituent parameters. These metrics include
surface form, surface finish, texture, and imperfections. Surface form describes the
macroscopic or global shape of a surface. Surface finish includes the microscopic
texture of the surface and localized imperfections. Texture embodies roughness and
waviness. Imperfections include localized defects, such as digs and scratches.
These metrics are chosen for measurement because they can have influence on
the functional and cosmetic quality of the component. The extent of this influence
will depend on where the component is situated within a system and on the par-
ticular application. It is therefore important to have a clear understanding of the
influence it will have on quality, as perceived by the customer, when specifying
a metric tolerance, whether based on theory or practice. Moreover, these metrics
should be measurable by objective means to a stated uncertainty and traceable to
national standards. Purely subjective assessments of, for example, surface scratches
have given rise in the past to misunderstandings and so should be avoided if pos-
sible. The results of an international survey of a selection of constructional para-
meter tolerances for a variety of different applications are presented in Chapter 7
just as a general guideline to current practice based largely on subjective assess-
ments. To complete the picture, some consideration will be given to the influence of
these metrics on quality and also to the measurement of surface contamination. The
current status of automatic inspection technology over large areas of high-quality
surfaces is also presented.
A number of international standards relating to this subject have been published
recently or are in draft form. Their historical development, new features, and their
methods of operation are critically reviewed because they can provide insight into
the need for future developments.
Surface Metrics 3

Attempts to assess the quality of a specular surface have traditionally been car-
ried out subjectively by looking first at the image of a distant object reflected in
the surface and then at the surface itself. If the surface is flat, the image is undis-
torted, and if no surface damage can be seen, we may conclude that the surface is
of high quality. This inspection process, as we shall see, can achieve a high level
of sensitivity in some respects but lacks precision and accuracy. Such observations
can, however, be a necessary and valuable precursor to carrying out the subsequent
process of measurement involving comparison with standards.
A plane light wave reflected from a nominally flat clean surface will carry an
impression of residual height variations across the surface. We can expect that the
spread of light obtained when the reflected beam is brought into focus, called the
point spread function (PSF), will bear intensity information related quantitatively
to the shape of the surface. Due to the mechanism of propagation, the light intensity
variations arising from slowly varying form errors (low spatial frequencies) will be
found near the center of the PSF and more rapidly varying surface errors (high
spatial frequencies), due to poor finish, will be away from the center. If a scratch
that contains a wide range of spatial frequencies is present, there will be a spread
of light right across the PSF in the form of a line at right angles to the direction of
the scratch.
Although this simple optical technique of examining the PSF can reveal surface
errors of nanometer dimensions and is useful as a rapid tool for surface inspection,
it has not been widely adopted for measurement. The lack of phase information,
which tells us about the direction of travel of parts of the beam in the PSF, means
that, without some prior information3 about the character of the defect, we cannot
calculate the shape of the wavefront and hence deduce surface shape errors. Fortu-
nately, developments in optical technology have given rise to the computer-aided
interferometer4–6 that is capable, in principle, of measuring the shape of optical
wavefronts to very high accuracy. Recent developments in image position sensing
that enable the direction of travel of rays of light to be determined with great preci-
sion are creating renewed interest in electronic means for measuring7–9 wavefront
shapes first reported in 1965.
These optical techniques that combine the processes of inspection and mea-
surement have been supported by the development of surface profilers from the
fields of precision engineering and materials science. Sharp probes can now be
scanned, in a production environment, over a surface to reveal surface height vari-
ations of atomic dimensions.10 The employment of widely differing disciplines
from the separate fields of optics and mechanics to measure the same surface pro-
vides valuable information regarding the strengths and weaknesses of both. We
can, in this way, obtain a higher level of confidence regarding our uncertainty of
measurement.
A wide variety of optical technologies for engineering metrology have been de-
veloped in recent years including, for example, photogrammetry, holographic in-
terferometry, fringe projection, moiré interferometry, and speckle methods. These
methods appear not, so far, to have found significant application in the high-
precision optical field.
4 Chapter 1

The material presented here is based on research conducted over the last
30 years at a number of organizations around the world. Work continues in an
attempt to ensure that any standards proposed are supported by the best available
technology and accepted by both the manufacturers and users of optical-quality
surfaces.

1.2 Why Measure Surfaces?

Good reasons are required for measuring surfaces. The process is costly. We need
to select, buy, calibrate, and maintain instruments. They must be installed in clean
measurement areas. Operating staff need to be trained and given the opportunity
to keep abreast of new developments. Laboratory accreditation may be needed to
comply with quality assurance requirements. Then we have to consider the cost
of labor and materials in undertaking and reporting measurements, and in dealing
with the inevitable feedback from customers. Negotiation with design authorities
may be needed to modify constructional tolerances to achieve acceptable produc-
tion yields. All these costs influence profitability in a competitive manufacturing
environment, so we need to be aware of the benefits to be derived from measure-
ment to ensure that we do enjoy them.

1.2.1 System function

The quality of an image produced by an optical system is determined by the op-
tical design chosen, the quality of the optical materials, and the manufacturing
processes. We shall not be considering here the influence of the optical materials,
the coatings of optical surfaces, the cementing process, or the errors associated
with centering, edging, and mounting on the final image quality. Although these
topics may occasionally be touched on, our main concern here is the specification
and measurement of the optical surfaces to be generated.
Errors in the shape or form of a surface have a direct effect on the shape of
a wavefront passing through an optical system. As little as a quarter of a wave-
length of light deformation of a wavefront shape can transfer approximately 20%
of power out from the center of the PSF to its edges. This broadens the PSF and
gives rise to a reduction in the capability of the system to resolve extended images.
If a diffraction-limited image quality is required over an extended image plane us-
ing ultraviolet radiation, the nature of the design process requires the use of a large
number of surfaces, some of which will need to be made to tolerances perhaps two
orders of magnitude better than this figure.
A reduction in the limit of resolution of an image is accompanied by a loss
in contrast of structures that can still be resolved, although significant amounts of
wavefront deformation can be tolerated before the level of veiling glare extending
over the whole image plane can be detected. A much greater contribution to veiling
glare11 arises from poor-quality optical coatings and from the surface finish and the
coatings of the mechanical parts of the system.
Surface Metrics 5

Optical systems designed for metrology, such as aerial survey cameras, require
accurate geometrical correspondence between the object and image, as well as
high-quality images. Low image distortion requires small tolerances on surface
form as well as very accurate lens centration.
System function can also be degraded by the quality of the component surface
texture. Residual light scatter from imperfectly polished surfaces, although having
little effect on the PSF or the optical transfer function (OTF) of a system, could
impair low-contrast image detection. The surface texture of mirrors used in a laser
gyroscope is required to be of the highest standard because light scatter has a direct
influence on drift rate.
The remaining metric related to finish is surface imperfections. These can occur
as digs and scratches. If they appear on a graticule placed in an image plane, their
effect may be described as functional since they would add to the pattern present,
whereas if they exist elsewhere their effect may be considered as cosmetic. The
exception to this arises in the case of optical components used in a laser system.
Imperfections, even of nanometer dimensions, can, when exposed to high laser
power or energy pulses, trigger weaknesses in the material structure, causing sur-
face damage or even complete shattering of the component. Surface imperfections
on a component used with laser radiation can therefore result in a reduction in its
life, and so may be regarded as functional.
At the time of writing, material imperfections in the inner layer of toughened
glass panels used in cars, windows, walls, and roofs are causing considerable
concern.12 Unstable impurities in the form of nickel sulphide crystals can cause
shattering of panels without any warning. Means are required for detection, mea-
surement, and classification since bubbles of the same size, always present, cause
no problem.

1.2.2 Appearance
The appearance of optical components has improved over the years. Before the
clarity of optical materials had reached its present high standard,13 small surface
imperfections and polish defects on a telescope lens would have been disguised by
particle suspensions and some striae. As these usually had relatively little influence
on the image seen, the manufacturers were ready to advise customers that lenses
were designed for looking through and not at.
It is perhaps unfortunate that the appearance of an optical component usually
bears no obvious relation to its performance. We cannot see the small deformations
of a surface that could ruin an image or judge the little influence that an easily seen
scratch will usually have on system use. The customer may, as a consequence,
be likely to regard a system having a visible scratch as lacking in quality. If the
manufacturer will pass a component with a scratch that can be seen, how will he
have judged the significance of other errors not visible to the eye?
Surface defects, such as residual roughness and imperfections, will always be
present to some degree and will be seen as an indication of component appearance
6 Chapter 1

quality, even though functional quality may be unimpaired. Some objective means
for measurement and agreed acceptance tolerances are therefore necessary.

1.2.3 Manufacturing efficiency

Manufacturing efficiency can always be improved, providing present performance
is known. In order to be aware of performance, we need to have knowledge of the
shape and dimensions of products produced so that they can be compared with
those appearing on specifications. If tolerances are exceeded, then product yields
will be reduced. In order to improve manufacturing efficiency we need to know
why this has happened. It is important, therefore, that the measurement process
provides some diagnostic information that can be acted upon to rectify any drop in
yield. Measurement is, therefore, essential for quality control, fault diagnoses, and
process optimization.

1.2.4 Benefits
We have seen that the information provided by measurement is necessary for ef-
ficient manufacture but, if wisely interpreted, might also lead to benefits resulting
from increased yields, cost reductions, and product improvements.
Still further benefits have been found to arise when a particular measurement
has been made on a number of different products by different organizations, and
the results are compared in a round-robin interlaboratory comparison. Several such
exercises have been carried out over the last 50 years involving optical measure-
ments, such as OTF, surface form errors, and laser beam profiles. The resulting
spread of results, usually much greater than expected, when analyzed have pro-
vided much valuable information on instrument design problems, measurement
procedures, and the need for calibrated reference components. This knowledge can
then be embodied in new measurement standards of benefit to all manufacturers
and users of optical systems.

1.3 Definition of Surface Metrics

Modern techniques for surface examination, such as the scanning electron micro-
scope (SEM) and the scanning probe microscope (SPM), are capable of revealing
structures of atomic dimensions. A lateral resolution of 1 nm and vertical resolu-
tion of 0.05 nm may be achieved with the SPM. It is to be expected, therefore, that
any surface tested with sufficient care will be found to be less than perfect. But
fortunately perfection is not required in real applications.
The whole amount of topographical information that could be recorded from
measurements on an optical surface of, say, 50 mm diameter is so great as to be
of little practical value without some form of data reduction. Figure 1.1 is an SEM
image of real scratches on a glass substrate. The great amount of topographical
information seen in the scratches and in the substrate extending even over such
small areas illustrates this point.
Surface Metrics 7

Figure 1.1 Magnified SEM images of real scratches on a glass substrate.

The simple geometrical laws of light propagation involving refraction and re-
flection form the basis of optical system design. In order to ensure that a given ray
from a point in an object arrives at the desired image point, all optical surfaces
encountered on the way must have a slope determined by the designer. The varia-
tion in value of this slope across the surface, when integrated, constitutes the form
or shape of the surface. Most optical surfaces are spherical and of known radius
that can vary from 1 mm to infinity, although increasingly surfaces of precisely
known nonspherical form are being produced. The usual method of surface gen-
eration based on grinding and lapping ensures that errors in form are macroscopic
laterally (greater than 1 mm extent) and microscopic vertically (less than 500 nm).
Errors in surface form can usually be attributed to the generating machine. Since
they can seriously degrade the performance of the optical component, for example,
in terms of limiting resolution, it is essential that their magnitude be kept within
limits defined on the optical drawing.
The final stage of surface generation usually involves polishing away the re-
mains of surface irregularities left by lapping. These may occur as random or peri-
odic height variations called, respectively, roughness and waviness, extending over
the whole surface where they are collectively termed the surface texture or as lo-
calized imperfections usually consisting of digs and scratches. These remaining
microscopic defects of texture and imperfections are referred to as the finish of the
8 Chapter 1

surface. The lateral spread of individual finish defects is usually less than 0.01 mm
and their depth is usually less than 100 nm.
Diffraction and light scatter govern the significance of finish. Although optical
system resolution may not be much affected by residual surface finish, an image
could suffer reduced contrast and the presence of imperfections will degrade the
cosmetic quality and therefore the value of the component. Imperfections cause
additional problems with laser optics and low-light-level imaging systems, where a
scratch can produce a disturbing line of light across an image plane. Imperfections
can also give rise to radiation absorption and high field concentrations that may
trigger surface damage in the presence of high-power/energy laser beams.
Unwanted light scatter can also arise from the imperfect deposition of thin film
coatings and from surface contamination occurring before, during, or after system
assembly. Unfortunately, most surface treatments after surface generation serve to
degrade the quality of the surface to some extent.
Since all of the parameters defined above influence in some way the quality of
the component, they should all be toleranced by the designer.

1.3.1 Surface metrics influencing quality

Figure 1.2 provides an exaggerated pictorial and underneath a schematic summary
of the parameters or metrics that together define total surface topography. These
metrics are chosen because of their relevance to the processes of design, surface
generation and measurement, or to the appearance of a component. They are de-
fined in the glossary.

Figure 1.2 Surface metrics influencing quality.

Surface Metrics 9

1.3.2 Causes of defects in surface topography

The design and use of instrumentation for measuring these important parameters
will be described, including reference to relevant ISO standards recently published
or still in draft form. But first we need to review the likely causes of defects, in
case they can be reduced or even eliminated.
Significant form errors are most likely to be due to problems with the surface-
form generator. They require immediate correction. Surface deformations of
100 nm or less may be due to the processing technology involving perhaps temper-
ature control and/or mounting/cementing stresses.
Surface roughness, often evident as microdefects distributed over a lapped sur-
face and seen with a magnifier, should be removed by further polishing. If, how-
ever, the surface has been generated by single-point diamond turning, the rough-
ness may be due to tool wear14,15 and the waviness due to tool vibration.
Surface imperfections in the form of digs and scratches can be due to a dry
polisher or to component handling and/or cleaning. Surface degradation can also
arise from a corrosive environment or, in the case of high energy/power systems,
to laser damage. Other defects, such as stain, bubbles, striae, and surface crazing,
are due to materials or coating problems.
It will be seen from the above that the study of our subject, which may be
described as the morphology of optical surfaces, requires consideration of many
other interrelated fields and disciplines. These include instrumentation, technology,
manufacturing processes, testing, specifications, training, and standards. Figure 1.3

Figure 1.3 Optical surface morphology.

10 Chapter 1

presents these relationships in an interactive pictorial form and illustrates the stages
involved before reaching the final point of acceptance or rejection of a compo-
nent.

1.4 Chapter Conclusions

We can summarize the measurement problems to be discussed as follows.
Modern surface analysis instruments, e.g., SEM and SPM, provide more deter-
ministic data than can be used conveniently and indicate that no surface is likely to
be perfect.
Metrics based on macroscopic (low spatial frequency, found by applying a low-
pass filter) and microscopic (high spatial frequency, found by applying a high-pass
filter) spreads in the lateral direction across a surface provide useful information
on manufacturing processes and on component performance. A midrange spatial
frequency or band-pass filter enables waviness to be measured.
The four metrics, comprising total topography, that need to be measured and
standardized include: form, roughness, waviness, and imperfections.

References
1. D. Golini, “Beating the grind,” OE Magazine, Vol. 1, No. 8, pp. 20–21 (2001).
2. J. Bilbro, “Optics in orbit,” OE Magazine, Vol. 1, No. 8, pp. 23–24 (2001).
3. T. Ha et al., “Size determination of microscratches on silicon oxide wafer sur-
face using scattered light,” Precision Eng., Vol. 27, pp. 265–272 (2003).
4. J. C. Wyant et al., “An optical profilometer for surface characterization of mag-
netic media,” ASLE Trans., Vol. 27, pp. 101–113 (1984).
5. B. Bhushan et al., “Measurement of surface topography of magnetic tapes by
Mirau interferometry,” Appl. Opt., Vol. 24, pp. 1489 (1985).
6. J. C. Wyant et al., “Development of a three-dimensional non-contact digital
profiler,” Trans. ASME J. Tribol., Vol. 108, No. 1, pp. 1–8 (1986).
7. L. R. Baker and J. N. Whyte “New instrument for assessing lens quality by
pupil scanning (spot diagram generation),” Japanese J. Appl. Phys., Vol. 4,
Supp. 1, 121–127 (1965).
8. L. R. Baker and T. L. Williams, “New electronic wavefront plotter,” Appl. Opt.,
Vol. 4, pp. 285–287 (1965).
9. ISO 15367-2, Test methods for determination of the shape of a laser beam
wavefront—Part 2: Hartmann-Shack sensors.
10. S. Jordan, “Scanning microscopy moves into production,” Photonics Spectra,
Vol. 38, March, pp. 73–74 (2004).
11. S. Martin, “Glare characteristics of lenses and optical instruments in the visible
region,” Optica Acta, Vol. 19, pp. 499–513 (1972).
12. D Cohen, “Flawed beauty,” New Scientist, Sept. 22, pp. 38–41 (2001).
13. D. Battistoni, “FT interferometry measures homogeneity,” Photonics Spectra,
Vol. 38, pp. 62–66 (2004).
Surface Metrics 11

14. L. R. Baker and J. K. Myler. “In-process measurement of surface texture,”

Proc. of SPIE, Vol. 802, pp. 150–156 (1987).
15. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scat-
tering, Optical Society of America, Washington, DC (1997).
Chapter 2
Surface Form
2.1 Introduction
An account of the early history of the manufacture, testing, and use of optical
components can be found in Twyman’s book. Of the four surface metrics worthy
of measurement, form, as we have seen, is probably the most important due to
its direct influence on system performance. If two easily abradable surfaces are
rubbed together in all directions, their area of contact increases to the point where
both surfaces assume a spherical form of the same radius. This basic process is
used in surface generation, but we still need means for determining the end point
depending on acceptance tolerances.
A flat contacting metal template can be made to the desired shape and placed
against a ground surface to assess the accuracy of form produced by viewing vari-
ations in the width of gap. Skill is needed to achieve a precision of 0.01 mm.
Polished surfaces require a much higher degree of measurement accuracy. This
can be obtained by using some form of interferometer. In its simplest form, a care-
fully made reference plate can be put into close contact with the surface under test.
When illuminated from above with a diffuse monochromatic source, as shown for
flat surfaces in Fig. 2.1, Newton’s fringes can be seen at the interface.
The advantage of this method is that the whole surface can be seen, so that
areal cover as opposed to line cover, using the template method, is available. High
sensitivity of 50 nm is typically achieved if the fringes are analyzed by eye, due to
the small unit of optical wavelength employed, but surface damage can arise if the
surfaces are brought into contact. Skill is needed in this case to avoid dust in the
interface zone.
A third technique for form measurement involves probe gauging. A proxim-
ity probe is scanned in a straight line over the surface to be measured and height
variations are recorded. The method, which is slow to achieve areal cover, is
more usually applied in coordinate measuring machines using a contacting probe.
It typically achieves an accuracy of 0.001 mm and requires some skill to ap-
ply.
13
14 Chapter 2

Figure 2.1 Newton interferometer.

2.2 Optical Height Probes

Since optical surfaces are often sensitive to damage, attempts have been made to
replace the contacting probe by a noncontact optical equivalent. Figure 2.2 illus-
trates the basic approach for measuring the form of a surface by use of an optical
probe.
Means must be found for measuring the changes in the distance H as the probe,
measuring this parameter, is traversed along a straight line across the surface in the
direction D but without contact. The slideway defining the reference line should
permit movement within the desired tolerance of measurement and avoid rotation
of the probe mount.
There are two basic approaches for solving this problem involving intersecting
rays or parallel waves creating interference. Figure 2.3 shows a schematic diagram
of the arrangement where a left and a right pencil of light are projected onto the
surface at an angle.
The surface is viewed from above and, when moved vertically, the separation
of the spots on the surface changes and becomes zero when the surface is level
with the point of intersection of the rays. Further movement will cause the spots to
diverge, causing a reversal of the L and R spot positions. In this way, the surface
can be accurately located at the point of intersection of the rays using, for example,

Figure 2.2 Optical height probe.

Surface Form 15

Figure 2.3 Intersecting ray probe.

Figure 2.4 Optical interference probe.

a red and a green ray for visual observation, or by the use of an electro-optical
position-sensing device. Provided the probe can be accurately moved horizontally
relative to the surface, the vertical movement of the probe needed to bring about
beam coincidence will be a measure of the surface form. Although at the point of
coincidence of the rays small changes in surface slope will be unimportant, the
method does depend on enough scatter to enable surface images to be formed and
measured with sufficient precision.
The second approach to the measurement of surface form relies on detecting
the interference pattern formed when a vertical pencil of light is reflected from
a reference and then a test surface, as illustrated in Fig. 2.4. Since reflection at
the optically denser medium will impose a half-wave phase advance, and if the
surfaces were in contact at some point, the two interfering beams would be in
antiphase and so the light intensity would be zero at that point. The beam reflected
at the less dense medium on return will have traversed twice the gap between the
surfaces and if this distance is half a wavelength, destructive interference will again
take place. When the pencil of light is scanned without tilt over the surfaces, null
signals will arise where the gap is an integral number of half wavelengths of the
radiation used. If the test surface is also flat, the locus of null signal bands will
be straight but, if not, the departure from straightness of these signals will provide
a measure of the form error of the test plate. As we shall see later, this form of
16 Chapter 2

dynamic probe scanning is replaced in practice by the use of imaging and a CCD
detector.

2.3 Optical Slope Probe

As an alternative to measuring surface height variations, a pencil of light can be
projected onto a polished surface at an angle and the variations in the angle of
reflection can be measured by an image position sensor as the beam, sampling a
small area, is scanned over the surface, as indicated in Fig. 2.5.
The variations in surface slope, recorded in this way, across the surface un-
der test can be integrated so as to yield surface form or, indeed, the shape of a
wavefront1,2 generated by an optical system. The use of an array of small lenses,
so as to perform static as opposed to dynamic scanning, over a surface or across
the pupil of an image-forming system in association with a CCD now enables ray
directions across an aperture to be determined.3 A precision of form measurement
approaching that achieved with an interferometer can be obtained provided sur-
face imperfections and contamination, causing errors in the detection system, are
avoided. A detailed analysis of these techniques and their associated errors has
been given.3

2.4 Interferometers
Surface form and its tolerance will be specified on the drawing of an optical com-
ponent. Information codifying the types of spherical and aspherical surface-form
deviation can be found in ISO 10110-5:1996 and will be described later. Toler-
ance magnitudes vary with the application and are typically in the region of 0.5 to
500 nm.
Methods for measuring surface form are either mechanical or optical. Mechan-
ical methods rely on contacting the surface with some form of lightly loaded stylus
or probe. This is scanned over the surface to precisely known coordinate positions,
and changes in its height from the surface are recorded.
Possible damage to the surface can be avoided, as we have seen, by the use
of noncontacting optical methods. These may involve the equivalent of the probe

Figure 2.5 Optical slope probe.

Surface Form 17

method to measure height variations, by triangulation ranging or slope determi-

nation followed by integration. However, since the availability of the laser, form
measurement is now usually carried out by computer-aided interferometry.4
This measurement is undertaken by allowing a wavefront reflected from the sur-
face under test to interfere coherently with a reference wavefront. By introducing
a slight tilt to one wavefront, a pattern of interference fringes is created. The de-
parture from straightness of these fringes embodies the form deviations required.
Since the intensity distribution of two-beam interference fringes is known to be
precisely sinusoidal in shape, it is possible to measure form deviations to 0.5 nm
under carefully controlled conditions.

2.4.1 Twyman-Green interferometer

International Technical Reports, presently under preparation, for measurement of
surface form are as follows:
“Interferometric measurement of optical wavefronts and surface form of optical
elements”
ISO 14999-1 Part 1: Definitions and fundamental relationships
ISO 14999-2 Part 2: Measurement and evaluation techniques
ISO 14999-3 Part 3: Calibration and validation of interferometric test equipment
and measurements
A further document (Part 4), also under preparation, deals with the evaluation of
tolerances.
These documents, currently in draft form, cover the full range of interferom-
eters likely to be met. The Twyman-Green and laser Fizeau interferometers, de-
scribed below, are most commonly used. Figure 2.6 is a schematic representation
of a Twyman-Green interferometer.
Light from a mercury discharge lamp illuminates a pinhole placed at the focus
of the collimator. The parallel beam produced is divided into a reference and test
beam at the beam splitter. The reference beam is returned by a plane mirror and, af-
ter passing again through the beam splitter, is brought into focus at the eye. The test
beam is also brought into focus at the same point after reflection from the surface
under test, and two passages through the compensating diagonal. This diagonal is
required to equalize the two optical path differences, thereby enabling the gener-
ation of high-contrast fringes created by interference between the two beams that
may include a number of different wavelengths arriving at the eye. The fringes may
be assessed directly by eye, photographed, or recorded by the use of an electronic
image sensor, such as a TV or digital camera.
Due to the separate paths taken by the beams before interference, the instru-
ment is sensitive to vibration and atmospheric turbulence. Both of these problems
have been largely overcome by use of the Fizeau configuration employing a one-
wavelength source such as a laser.
18 Chapter 2

Figure 2.6 Twyman-Green interferometer.

2.4.2 Laser Fizeau interferometer

A schematic representation of three laser configurations is given in Figs. 2.7(a)–(c).
Fringe formation arises in the cavity between the reference and test surfaces as
with Newton’s arrangement. The vertical lines ending with arrowheads are lenses.
In each case, an expanded laser beam illuminates the whole aperture of the cavity
and the interfering beams reflected by the reference and test surfaces are removed
for observation by a beam splitter.
In Fig. 2.7(a), the reference and test surfaces are nominally plane.
In Fig. 2.7(b), a well-corrected collimator is placed in the cavity so that a con-
vex surface, with its center of curvature placed at the focus of the collimator, can
be tested.
In Fig. 2.7(c), the high-quality in-cavity collimator, needed for testing convex
surfaces, is avoided by replacing it with a lower-quality component brought for-
ward out of the cavity. This generates a converging wavefront with its focus coin-
cident with the center of curvature of both the reference and test surfaces. In this
way, the beams intersecting the reference and test surfaces are substantially normal
to both in all three configurations.
Although a visual assessment of the interferogram can be made, measurement
of wavefront deviation is usually undertaken by computer analysis of the image
stored by an electronic image sensor, such as a TV camera. Figure 2.8 shows a
typical Fizeau instrument.
Surface Form 19

Figure 2.7 Laser Fizeau configurations.

2.4.3 Four-step measurement of phase

At a given point in the interferogram, a continuous variation of the optical path

length between the reference and test surfaces would result in a sinusoidal variation
in the intensity of light detected. Since the peaks and troughs of intensity will occur
20 Chapter 2

Figure 2.8 Vertical Zygo interferometer.

at half-wavelength intervals, due to the double-pass configuration, intermediate

levels of intensity can be related to a specific path or phase difference.
In order to simplify this analysis, it is usual to introduce accurately known
phase shifts of the reference beam as indicated below. The variation of light inten-
sity I in an interferogram with phase φ is given by

I1 = I0 (1 + V cos φ),

where I0 is the mean value and V is the fringe modulation. If we now introduce
reference beam shifts of π/2, π, and 3π/2, then

 
π
I2 = I0 1 + V cos φ + = I0 (1 − V sin φ),
2
I3 = I0 [1 + V cos(φ + π)] = I0 (1 − V cos φ),

 
3π
I4 = I0 1 + V cos φ + = I0 (1 + V sin φ).
2
From these equations we can deduce
I4 − I2
tan φ = .
I1 − I3
This means that we can calculate the phase difference between the reference and
test surfaces, or wavefront deformation, at any point in the interferogram to a pre-
cision of perhaps 0.5 nm from four intensity measurements. A larger number of
intensity measurements will reduce the uncertainty of measurement still further.
Surface Form 21

Application of the four-step phase stepping technique has been recently applied
to white light wavefront measuring systems.5

2.4.4 Interlaboratory comparison of interferometers

A level of precision of 0.5 mm may be hoped for from commercially available
interferometers. However, a recent study6 to determine the agreement achieved
between 11 different laboratories of peak-to-valley measurements of form error
using six components arrived at rather different conclusions.
The instruments included in the study were as follows:

• Zygo Mark ll
• Zygo Mark lV
• Zygo GPI
• WYKO 400
• WYKO 6000
• P-OE Interfire with WYKO phase shift and software
• Specac FOTI 100 with Phase Shift Technology ZMOD software
• Owner-build Fizeau with WYKO phase shift and software

The test components chosen and calibrated by the UK National Physical Labora-
tory (NPL) included

T1 – Nominal flat, 100 mm diameter

T2 – Concave sphere, nominal 10.6 mm radius of curvature, 15 mm diameter
T3 – Concave sphere, nominal 111 mm radius of curvature, 62 mm diameter
T4 – Convex sphere, nominal 111 mm radius of curvature, 62 mm diameter
T5 – Concave sphere, nominal 426 mm radius of curvature, 75 mm diameter
T6 – F/7 doublet collimator, 38 mm diameter

The agreement between instruments/operators was, at best, as follows:

• 0.1 wavelengths for P-V on flats

• 0.2 wavelengths for P-V on spheres
• 0.04 wavelengths for RMS form error
• 0.1–0.25% for radius of curvature

The spread of surface form errors of the six components is shown in Fig. 2.9.

2.4.5 Interferometer error sources and calibration

Sources of error, which may be random or systematic (for example, zero-scale
errors), may arise from test equipment, the particular setup used, or from the envi-
ronment.
22 Chapter 2

Figure 2.9 Spread of surface form errors found in six components.

Sources of uncertainty should be sought from test requirements, test procedure,

personal bias, instrument resolution/scale errors, values of standards/test pieces,
drift since last calibration, approximations, and random effects. These uncertain-
ties can be combined to include the standard deviation of individual uncertainties.
The square root of the sum of the squares of these values is then modified by a
confidence factor to give a final estimated value.
So-called absolute values of small wavefront errors can be obtained using a
Twyman-Green interferometer and a two-step process with reference to Fig. 2.10
as follows:
Step 1: Record W1 when WR is the interferometer error and P is the known surface
form of a flat inserted in the test arm. Then,
W1 = WR + 2P.
Step 2: Record W2 when the test optics, W0 and test sphere with known form S,
replace the flat. Then,
W2 = WR + 2W0 + 2S,
from which WR can be removed to give
1
W0 = [W2 − W1 − 2(S − P)].
2
For the most accurate measurements, consideration must be given to the interfer-
ometer calibration. Items to be considered include

• Coordinates of a point in the wavefront

• Phase steps used
Surface Form 23

Figure 2.10 Absolute values of small wavefront errors.

• Camera responsivity
• Performance of the software used

The levels of intensity and slope thresholds as well as the presence of spurious
images arising from interference, diffraction, speckle, or residual surface rough-
ness also need to be considered.
Instruments may have a “self-check” facility to ensure a desired performance
of piezoelectric displacement transducers. Overall performance may be checked
over a given dynamic range by using test pieces calibrated by a national standards
laboratory. Precise wavefront tilts can be made to create known fringe spacing. Ul-
timately, though, the highest confidence of uncertainty of measurement can best be
achieved by circulating test specimens among different laboratories, where some
of which should preferably be using alternative physical principles (for example,
ray intersection measurements).
Some practical tips for reducing measurement errors can be found in Ref. 7 and
some problems to be overcome in the testing of extremely high-quality material are
described in Ref. 8.

2.5 Form Tolerances

The components of form deviation are illustrated by reference to Fig. 2.11, and a
typical output from a computer-aided interferometer indicating the analysis of the
measured deviation into various metrics relating to design functions is shown in
Fig. 2.12.
24 Chapter 2

Figure 2.11 Diagram showing surface form tolerances.

Figure 2.12 Measured surface and its decomposition into error types.
Surface Form 25

It is helpful to analyze the measured form of a surface deviation [see

Fig. 2.12(a)] into a number of component parts that can be toleranced by the de-
signer and related to system performance. Thus, the surface-form deviation mea-
sured can be analyzed into the best-fit spherical surface and the residual irregular-
ity function. This can be further analyzed into a rotationally symmetric aspherical
component and a final residual determining the RMS asymmetry.
The ISO 10110-5:1996 document9 and an explanatory reference10 define the
following metrics, requiring tolerancing in terms of fringe spacings where one unit
is equal to one half the light wavelength. Computer-aided interferogram analysis11
provides quantitative and accurate data.

2.5.1 Approximating a spherical surface

This is the best-fit spherical surface, shown as in Fig. 2.12(b), for which the RMS
difference from the total surface deviation is a minimum.

2.5.2 Sagitta error

It can be seen from Fig. 2.11 that this error is the distance along the axis separating
the desired surface from the approximating surface.

2.5.3 Irregularity function

This is the theoretical surface (c) in Fig. 2.12, defined as the difference between
the total deviation function (a) and the approximating spherical surface (b).

2.5.4 Irregularity
This is the peak-to-valley difference between the irregularity function and the plane
that best approximates it. It is more useful as a measure of quality of manufacture
rather than image quality, as it lacks information on where in the aperture an error
occurs.

2.5.5 Approximating an aspheric surface

This is the rotationally symmetric surface (d) in Fig. 2.12 for which the RMS dif-
ference from the irregularity function is a minimum.

2.5.6 Rotationally symmetric irregularity

This is the peak-to-valley difference between the approximating spherical surface
and the plane that best approximates it.
26 Chapter 2

2.5.7 Total RMS deviation (RMSt )

This is the RMS difference between the test surface and the nominal theoretical
surface. This parameter has optical significance in terms of the amount of radiation
removed from the imaging rays.

2.5.8 RMS irregularity (RMSi )

This is the RMS value of the irregularity function (c) in Fig. 2.12. This parame-
ter has optical significance in terms of the amount of radiation removed from the
imaging rays.

2.5.9 RMS asymmetry (RMSa )

This is the RMS value of the difference between the irregularity function and the
approximating aspheric surface shown as (e) in Fig. 2.12. This parameter has op-
tical significance in terms of the amount of radiation removed from the imaging
rays.

2.5.10 Form indication on drawings

Three versions of indications are possible as follows:

3/A(B/C) or 3/A(B/C) RMSt,i, or a < D

or

3/--RMSt,i, or a < D,

where 3 is the code number for the surface form tolerance; A is either the maximum
sagitta error or (−), indicating the total radius of curvature tolerance as given in
the radius of curvature dimension and not specified here; B is either the maximum
permissible value of irregularity or (--), indicating no irregularity is given; C is
the permissible rotationally symmetric irregularity tolerance, if given; and D is the
maximum permissible value of the RMS quantity of the type specified by one of
the letters t, i, or a.
The ISO 10110-5:1996 standard also includes valuable annexes on digital inter-
ferogram analysis, visual interferogram interpretation, and the physical relevance
of RMS errors.

2.6 Chapter Conclusions

Despite recent advances in interferometer design, it is difficult to achieve a spread
of agreement between workshop instruments and operators of better than 50 nm
for flats and 100 nm for spherical surfaces. Claims for higher accuracy should
be treated with caution. The presence of scratches can lead to erroneous results.
Surface Form 27

Surfaces should be inspected prior to testing, and the sensitivity of programs to

imperfections needs to be measured. Operators need training in setup procedures,
particularly with regard to aperture setting and component mounting. Training is
also needed in the correct assessment and treatment of errors. An interferometer
visual channel for in situ surface inspection to check for alignment and surface
damage is considered essential.

References
1. L. R. Baker and J. N. Whyte, “New instrument for assessing lens quality by
pupil scanning (spot diagram generation),” Japanese J. Appl. Phys., Vol. 4,
Supp. 1, pp. 121–127 (1965).
2. L. R. Baker and T. L. Williams, “New electronic wavefront plotter,” Appl. Opt.,
Vol. 4, pp. 285–287 (1965).
3. ISO 15367-2, Laser beam wavefront shape—Part 2. Hartmann-Shack sensors.
4. J. C. Wyant et al., “Development of a three-dimensional non-contact digital
profiler,” Trans. ASME J. Tribol., Vol. 108, pp. 1–8 (1986).
5. R. J. Barron et al., “White light wavefront measuring system using geometric
phase stepping,” J. Mod. Opt., Vol. 51, pp. 1–11 (2004).
6. J. D. Briers, “Interferometric optical testing: an inter-laboratory comparison,”
J. Opt A: Pure Appl. Opt., Vol. 1, pp. 1–14 (1999).
7. J. Kumler and R. Malcom, “Practical tips for reducing measurement errors,”
Photonics Spectra, Dec., pp. 104–110 (2002).
8. B. Wang and M. Ward, “Measuring birefringence in calcium fluoride at
157 nm,” Photonics Spectra, Nov., pp. 62–68 (2002).
9. ISO 10110-5:1996, Optics and optical instruments—Preparation of drawings
for optical elements and systems—Part 5. Surface form tolerances.
10. M. Bray, “Ensuring quality: Standards point the way to performance,” The
Photonics Design and Applications Handbook, Laurin Publishing, Pittsfield,
MA, pp. 395–398 (2001).
11. D. Malacara et al., Interferogram Analysis for Optical Testing, Marcel Dekker,
New York (1998).
Chapter 3
Surface Roughness
3.1 Introduction
The prime purpose of surface metrology is to ensure the quality of manufacture
defined as fitness for the intended purpose of the component. Hence, there is
the need for tolerances relating to parameters that can be measured and used to
characterize the performance or appearance of a component. It is an advantage if
these metrics can also provide diagnostic information on the processes of manu-
facture.
Whereas deviations from the ideal form, or surface macrotopography, of a com-
ponent computed by the designer have a direct and predictable influence on the
performance of the system involved, the effect of residual finish, or surface mi-
crotopography, is not so obvious. Finish, not usually considered by the designer
as it is not amenable in the same way as form to computation, is dealt with by
defining a range of different metrics that will be discussed in the remaining chap-
ters.
Ideally, all optical surfaces should be finished to the state where further surface
treatment results in no improvement in surface quality. In practice, however, de-
fects of atomic dimensions can now be visualized and so thresholds of acceptance
of measurable parameters are required. As we have seen, it is convenient to ana-
lyze finish into metrics called texture, which extends over the whole surface, and
imperfections that are highly localized. Texture can involve a random component
of surface height variations called roughness, together with a periodic component
termed waviness. This chapter deals with the most common methods for the mea-
surement of surface roughness.
Roughness can be assessed subjectively by running a fingernail across the sur-
face or by visual observation. A sensitive nail tip can detect surface height vari-
ations down to 100 nm. Visual assessment achieves a sensitivity of, perhaps, an
order of magnitude better. There are basically two approaches when an objec-
tive measurement to roughness characterization is required. These are illustrated
schematically in Fig. 3.1. The original “deterministic” method, adopted from en-
gineering and shown in the lower half of the figure, employs a stylus dragged
along a line (x) on the surface. The vertical movements, recorded effectively as
29
30 Chapter 3

Figure 3.1 Methods for surface characterization.

changes in the separation of the reference and test surfaces (z) using a very sharp
stylus, are taken to provide an accurate measure of surface irregularities. Noncon-
tact optical probes, usually based in some way on the principle of triangulation
ranging, can do the same job. Either the stylus radius or the wavelength employed
with the optical probe limits the spatial resolution, or bandwidth, of the recorded
data.
The most recent interpretation of this principle, employed in the SPM, can
achieve a spatial resolution of 1 nm and a vertical resolution of 0.05 nm. The SPM
can function in several different regimes. Manufacturers of these extremely sen-
sitive instruments provide advice and information on the precautions to be taken
during their use and on the interpretation of output data. Damage to even fragile
surfaces can be minimized by the use of tapping mode imagery. The spring can-
tilever assembly supporting the probe tip is driven at its resonant frequency up to
1 MHz. During traverse across the surface, the tip movement is reduced due to loss
of energy at contact and this change in amplitude is a measure of surface features.
The second metric, shown in the upper half of the figure, involves the measure-
ment of some parameter that can be related to quality. Parametric methods can be
used for characterizing the function or appearance of a component. They depend
on measuring the way in which a beam of light incident on the surface is modi-
fied by the processes of diffraction, polarization, and absorption. Since, however,
the phase of the light is not usually recorded, it is impossible to solve the inverse
problem and calculate the surface height variations causing the scatter.
Surface Roughness 31

The incident beam can be made to vary in intensity, wavelength, polarization,

and spatial coherence. As indicated in Fig. 3.1, measurements of the beam after in-
teraction with the surface include specular reflectance, total integrated scatter, po-
lar distribution, speckle, and polarization. Useful information related to the surface
condition obtained in this way can include roughness, waviness, imperfections,
optical constants, and contamination.
The approach to texture measurement favored in ISO 10110-8 is based largely
on the technique of profilometry widely employed to characterize engineering sur-
faces. As with form measurement, contacting or noncontacting probes may be
used, but information gathered extends over only a small portion of the compo-
nent so as to avoid inclusion of form data.
Surface roughness specifications are applicable to matte and ground surfaces as
well as to specular (polished) surfaces. There is also provision in the above standard
to characterize roughness by counting the number of pits, left by polishing, along
a line of length 10 mm.
The metric most widely used to measure roughness is the RMS value ob-
tained from surface height variations with stated lower and upper limits of sam-
pling length. If a probe is employed, alternative tracks across the surface may give
different results if the surface is not uniform.
The most popular instrument used for optical surface topographical analysis is
the computer-aided microinterferometer. This has the advantage of being noncon-
tact and providing aerial coverage. Although RMS values are quoted to 0.01 nm,
the accuracy actually achieved may be open to doubt. Variations in the optical con-
stants across a surface if they occur, for example, from contamination or surface
crystals, because of the phase changes they induce, may give rise to significant
errors in the surface heights recorded.
The effect of residual roughness is to scatter image-forming rays and produce
an unwanted haze over the image plane. This reduces contrast and limits the dy-
namic range of image intensity. The cause of roughness is usually imperfect polish.
Scatter of light can also arise from optical coatings employed to change reflectance
values, or from contamination.

3.2 Typical Optical Component Roughness Values

Modern applications1 of optical systems employed in the laser and information
technology industries call for surfaces of the highest quality attainable. Figure 3.2
shows the microtopography of a selection of high-quality surfaces obtained by the
use of an atomic force microscope and their areas of use. Note the extremely small
area over which the measurements have been made. It is not unusual for RMS
values to be quoted in the literature without supporting information on the sampling
interval and extent of the scan.
32 Chapter 3

Figure 3.2 Optical component roughness values.

3.3 Deterministic Methods

Direct measurement of the microtopography of a surface can be undertaken by
the use of a contacting stylus or by noncontacting optical means, such as those
used for the measurement of surface form. Information at a molecular level can
Surface Roughness 33

be obtained by scanning probe microscopes involving the measurement of surface

current changes or microforce perturbations.

3.3.1 Stylus method

A schematic representation of a typical stylus probe developed first for the engi-
neering industry is shown in Fig. 3.3.
The lightly loaded stylus, linked to a ferrite core, experiences a vertical move-
ment as it is moved relative to the surface under test. This is detected by induction
in the coils, amplified, and recorded. An optical pickoff may also be used. The
modern stylus instrument is now well established and can achieve a high vertical
resolution of 0.1 nm over a range of 100 micrometers and a horizontal resolution
of 100 nm. There are, however, significant disadvantages including sensitivity to
vibration, line and not areal cover is provided in one scan, the method is slow in
operation, and damage of fragile surfaces can easily occur.

Figure 3.3 Schematic representation of a contacting stylus.

3.3.2 Profilometry metrics

The output data from a single trace of profile must be reduced to parametric forms
to be useful. Figure 3.4 shows a trace over a known sampling length with a selected
sampling interval.
The average roughness Ra is the average of the absolute values of the surface
height variations measured from the mean surface level. This parameter is normally
used for machined, as opposed to optical, surfaces. At the time of writing, it is the
most widely used metric relating to surface finish.
The RMS roughness Rq is the square root of the mean value of the squares of
the distances of the chosen points from the mean surface level.
It should be remembered that these statistical parameters have no unique value
as their magnitude depends on the sampling length, the sampling interval, and the
surface area selected for the measurement. In the absence of large surface devia-
tions, these metrics will have a similar value. Otherwise Rq will be larger than Ra .
34 Chapter 3

Figure 3.4 Profilometry metrics.

Some applications require roughness to be recorded over an extended area, giv-

ing rise to global roughness, whereas for others a localized measure may be a
sufficient indication of quality. Considerable care is required in undertaking this
measurement, as small variations in the sampling interval, for example, can cause
significant errors in the roughness value.

3.3.3 Microinterferometer
The principle of two-beam interference for measuring the difference between a ref-
erence and test wavefront employed for measuring surface form can also be applied
for measuring roughness, provided the surface can be viewed under magnification.
The most widely used technique is based on the Mirau microscope interferometer
shown in Fig. 3.5. A lateral resolution of 0.5 µm and vertical resolution of 0.02 nm
are possible values using computer processing of image data.
These figures for resolution and uncertainty of measurement achieved in prac-
tice have so far not been supported by interlaboratory comparisons. The measure-
ment is still deterministic as it provides profile information, but it has an advantage
over the probe approach since areal as opposed to line information is provided.

3.4 Parametric Methods

These methods provide information directly related in some way to the optical
performance of the system and therefore may be more useful for quality control.

3.4.1 Surface point spread functions

We know from experience that if we view a distant point source after reflection
from a flat mirror, the shape of the virtual image tells us something about the quality
of polish of the mirror surface. If the virtual image remains as a point source,
the mirror is well polished, but if a diffuse image is seen, then we conclude the
surface lacks polish. Using a collimator to bring the reflected beam to a focus and
measuring the distribution of intensity at the image, i.e., the PSF, we can undertake
Surface Roughness 35

Figure 3.5 Mirau microscope interferometer.

Figure 3.6 Surface point spread functions.

quantitative measurements of surface roughness. Figure 3.6 shows schematically

how the energy in the PSF is transferred from the center to the edge of the point
image as the roughness is increased.
Using visible light, RMS roughness values of 10 nm can be seen, but the image
becomes diffuse when this value increases to 100 nm. If the surface suffers from
waviness errors, the reflected image will be broken up into a number of diffraction
36 Chapter 3

images or spectra. Without phase information, the surface topography cannot be

calculated, but the method could be used in a comparator mode over a range of
roughness values of 100:1 and for detecting waviness.

3.4.2 Total integrated scatter measurement

As an alternative2,3 to measuring the drop in peak intensity of the surface-point
spread function, roughness can also be determined by measuring where that light
has been scattered.
If R is the specular reflectance of a surface where only radiation in the specular
direction is measured, R0 is the total reflectance where all the reflected radiation
is measured, δ is the RMS surface roughness, and λ is the wavelength of radiation
used, then, provided the height variations are less than a wavelength of radiation,
it can be shown from diffraction theory that

  
  
4πδ 2 4πδ 2
R = R0 exp − ∼ R0 1 − .
λ λ
 
(R0 − R) 4πδ 2
But the total integrated scatter (TIS) = = or
R0 λ
 
λ √
δ= π TIS.
4

If we take the base level of TIS as equal to 10−4 , then δ = λ/1257.

This means that measuring radiation scattered by a surface is a sensitive way
of determining surface roughness.
The apparatus needed to measure the metric TIS is shown in Fig. 3.7

Figure 3.7 TIS apparatus.

Surface Roughness 37

The beam from a helium-neon laser passes through a mechanical beam-splitter

chopper, providing a detector output from the incident beam (Di ) and, after enter-
ing the hemispherical collecting mirror, falls at normal incidence on the specimen
to be tested. The specular beam reappears from the hemisphere and falls on a re-
flected beam detector (Dr). The radiation scattered by the surface is collected by
the hemisphere and focused onto the scattered radiation detector (Ds). The detector
outputs provide R0 = Ds/Di and R = Dr/Di needed to calculate the TIS and the
RMS surface roughness. The chopper and filters are needed to modulate and select
the radiation from any ambient light.
The advantages of this method are that a single metric is obtained integrated
over the area illuminated and the technique is noncontacting. The disadvantages,
however, are that a very clean environment is needed, surface imperfections cause
errors, and the surface must not transmit or great care is needed to avoid collecting
spurious signals. Due to its parametric nature, only statistical rather than determin-
istic information is provided. The method is more likely to find application in a
research, rather than a production, environment.
Due to the importance of roughness on surface quality, which affects many
different industries, the subject has been studied at length in Refs. 4–10.

3.5 Surface Roughness Indications in Drawings

Surface roughness indications in drawings are described in ISO 10110-8. Figure 3.8
deals with a single method, based on profilometry, for matte or ground surfaces and
three alternative codes for specular surfaces.
The first code for indicating the quality of a polished surface involves counting
the number of microdefects, resulting from incomplete polish, along a straight line
of length 10 mm using a low-power microscope or stylus. It is assumed that the
better the polish, the fewer microdefects will remain. There is unlikely to be a high
correlation between this parameter and the optical properties of the surface.
The second code requires measurement of the RMS value, with sampling
lengths chosen between stated limits. This parameter, as we have seen, has a di-
rect influence on specular reflectance and TIS, but does not reveal waviness or
patterns across the surface that may be readily visible. Microdefects with their P
values may also be included in the code.
The third code addresses the problem of patterns and waviness by employing
the parameter of power spectral density (PSD). This requires the surface profile
to be broken down into its spatial frequency components as described in the next
chapter. Since each spatial frequency component diffracts radiation at a particular
angle, the method is appropriate for studying the angular distribution of radiation
scattered by a surface. Microdefects with their P values may also be required.

3.6 Chapter Conclusions

Roughness is due to random surface height variations. It is a part of texture and ex-
tends over the whole surface. It is usually caused by incomplete polishing away of
38 Chapter 3

Figure 3.8 Surface roughness indications in drawings.

pits left by the form generator. Measurement can be by deterministic (analytical)

or parametric (functional) methods. The metric most widely used by the optical
industry to measure roughness is the RMS (Rq ) value of surface height variations
along a line. Rq is not a unique descriptor. It depends on the sampling interval
and length, and also on the location of the measurement on the surface. Ra is the
metric most popular with engineering industries. Roughness scatters radiation and
Surface Roughness 39

reduces image contrast. Scatter from coatings and contamination add to this ef-
fect.

References
1. C. Langhorn and A. Howe, “Optical morphology: Just how smooth is that
surface?” Photonics Spectra, June, pp. 130–133 (1998).
2. H. E. Bennett, “Scattering characteristics of optical materials,” Opt. Eng.,
Vol. 17, pp. 480–488 (1978).
3. J. M. Bennett, “Recent developments in surface roughness characterization,”
Meas. Sci. Technol., Vol. 3, pp. 1119–1127 (1992).
4. E. L. Church et al., “Relationship between surface scattering and microphoto-
graphic features,” Opt. Eng., Vol. 18, pp. 125–136 (1979).
5. E. L. Church, “Fractal surface finish,” Appl. Opt., Vol. 27, pp. 1518–1526
(1988).
6. M. Bray, “Ensuring quality: Standards point the way to performance,” The
Photonics Design and Applications Handbook, Laurin Publishing, Pittsfield,
MA, pp. 395–398 (2001)
7. T. Vorburger and J. Fu, “In the rough,” OE Magazine, Vol. 2, No. 3, pp. 31–34
(2002).
8. ISO 10110-8:1997, Optics and optical instruments—Preparation of drawings
for optical elements and systems—Part 8. Surface texture.
9. ISO 4287:1997, Geometrical Product Specifications (GPS)—Surface texture:
Profile method-terms, definitions and surface texture parameters.
10. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scat-
tering, Optical Society of America, Washington, DC (1997).
Chapter 4
Surface Waviness
4.1 Introduction
Most surfaces when viewed from sufficiently far away will appear to have a uni-
form texture. On close and detailed examination using modern techniques, how-
ever, all surfaces will reveal some patterning, even if only at the atomic level.
Surface waviness is the periodic component of surface texture. It arises most
frequently from induced vibrations of a single-point surface generator, and when
surface profiles are Fourier analyzed it usually occupies a spatial frequency band
between those of surface form and roughness. Optical surfaces, such as metal mir-
rors or infrared (IR) optics, generated by single-point diamond turning, are often
influenced to some degree by waviness. The periodic nature of such a surface re-
sults in diffraction of a reflected or transmitted beam, giving rise to multiple images
rather than a single image when the beam is brought into a focus. A surface gener-
ated by the more common process involving area contact between the lap and the
work surface is unlikely to suffer in this way. Polishing by the use of a flexible lap
at speed can give rise to an orange-peel effect that can exhibit dominant spatial fre-
quencies when Fourier analyzed. Similar patterns are sometimes seen on painted
surfaces.
Very small periodic surface height variations of less than a nanometer can be
measured by a microscope interferometer even in the presence of residual rough-
ness. In practical terms, where waviness gives rise to spurious images varying in in-
tensity, measurement of these images can be carried out radiometrically and thresh-
olds set depending on the application. For machine diagnostic purposes, however,
as well as for setting acceptance thresholds, ISO 10110-8:19971,2 recommends
measurement of the PSD function as an alternative approach to setting thresholds
for surface texture including roughness and waviness. This metric is the square of
the Fourier transform of the measured surface profile along a line. The software
employed by a computer-aided microinterferometer will often compute this algo-
rithm. A typical surface has a maximum PSD at low spatial frequencies and an
exponential decrease as the spatial frequency increases. Waviness will give rise to
peaks at one or more spatial frequencies.
41
42 Chapter 4

In the one-dimensional case the PSD, suggested as an alternative measure of

texture in ISO 10110-8, can be expressed as
A 1 1
PSD = , for <f < ,
fB 1000 × D 1000 × C
where f is the spatial frequency of the surface height variations in µm−1 , B is
the power to which the spatial frequency is raised, C and D are the minimum and
maximum sampling lengths in millimeters, and A is a constant expressed in µm3−B
where usually 1 < B < 3. In this way, the texture threshold is given by specifying
the four values A, B, C, and D.

Figure 4.1 Effect of tool wear on surface quality.3

Surface Waviness 43

4.2 Fourier Analysis of Height Profile

An understanding of the practical significance of waviness can best be acquired by
reference to a practical example. Figure 4.1 shows four microscope images of a
surface generated by turning, using a single point diamond tool.3 Image (a) shows
the surface when the tool is still sharp. This is followed by images (b), (c), and (d)
as the tool gradually loses its edge. At (d) the extra stress involved is starting to set
up machine vibrations, creating a periodic variation of tool cutting depth. These
TV images were obtained using a vertical illuminator and low-power microscope
with a spatial frequency filter in the back focal plane of the objective lens. This
simple arrangement,3 suitable for use on-machine, provides an image of intensity
inversely proportional to the slope of the surface. The periodic nature of the surface
can be clearly seen from the line trace of image intensity displayed at the top of
each image. See also Sec. 8.6.
We have already seen in Fig. 3.6 how the waviness of a surface gives rise to
peaks of intensity at various diffraction angles. The closer the observing lens is to
the surface, the more diffracted light will be collected and any surface pattern will
be more likely to be seen. If the lens is too far away, waviness will not be evident.
It is important, therefore, to match the aperture of the collecting lens to the period
of the pattern to be assessed.
The preferred metric for measuring waviness is the PSD. It involves analyz-
ing the surface height distribution along a line across the surface as shown, for
example, in Fig. 4.2(a). The remaining curves (b), (c), (d), and (e) indicate the
amplitudes of increasing spatial frequency of Fourier components present in (a).
These components can be regarded as periodic gratings where the finer the
grating, the larger the angle at which diffraction images will be formed.

4.3 Spatial Frequency Zones

Figure 3.6 shows a PSF obtained by bringing a parallel beam reflected from the
surface into focus. A typical surface produces a PSF divided into three spatial fre-
quency zones, but overlap can of course occur.
Surface-form errors influence the central peak of the PSF and poor polish
(roughness) will scatter radiation away from the peak. Waviness usually directs
energy between these two extremes. This analysis is based on a rule-of-thumb ob-
servation that the order of magnitude of the spatial spreads of these parameters, as
a fraction of the aperture size, is generally as follows: form:waviness:roughness =
0.1:0.01:0.001.

4.4 Computation of Texture

As we have seen, texture includes roughness4 and waviness. The basic data requires
recording the variation in surface heights measured from a mean line, Y(x), for
44 Chapter 4

Figure 4.2 Fourier analysis of surface height profile.

incremental movements across the surface (x). If the sampling length is L, then the
RMS roughness Rq is given by

 L 1/2
1
Rq = 2
Y (x)dx .
L 0

This metric is the most widely used by the optical industry. It provides an estimate
of scattered radiation likely to reduce image contrast.
The surface patterning or waviness resulting in a change in the appearance of
the surface requires measurement of the PSD for each spatial frequency compo-
nent, f , measured in cycles per unit length according to the equation
 2
1  L 
PSD(f ) =  Y(x) exp(−i2πfx)dx .
L 0

This metric may relate to the image of a point source spreading into a repeated dif-
fraction pattern, thereby seriously degrading image quality through the formation
of multiple images. The combined effect of form errors, waviness, and roughness
on the image of a point source can be seen in Fig. 4.3. When undertaking measure-
ments, care should be taken to select areas free of surface imperfections so as to
avoid false data.
Surface Waviness 45

Figure 4.3 Spatial frequency zones.

4.5 Chapter Conclusions

Waviness is a part of texture and extends over the whole surface. It is usually caused
by vibration due to poor machine mounting and/or tool wear, and is most often
inspected and measured by a microinterferometer. Incident radiation is diffracted,
causing the formation of multiple images. Waviness is of particular concern in laser
and infrared optics.

References
1. ISO 10110-8:1997, Optics and optical instruments—Preparation of drawings
for optical elements and systems—Part 8. Surface texture.
2. M. Bray, “Ensuring quality: Standards point the way to performance,” The Pho-
tonics Design and Applications Handbook, Laurin Publishing, Pittsfield, MA,
pp. 395–398 (2001).
3. L. R. Baker and J. K. Myler, “In-process measurement of surface texture,” Proc.
of SPIE, Vol. 802, pp. 150–156 (1987).
4. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scatter-
ing, Optical Society of America, Washington, DC (1997).
Chapter 5
Surface Imperfections
5.1 Introduction
Imperfections are described in ISO 10110-7:19961 as “localized defects within the
effective aperture of an optical surface produced by improper treatment during or
after the fabrication process.” They degrade the quality of a surface finish. Since
modern methods of surface examination can resolve atomic dimensions, it can be
assumed that no surface is completely free of imperfections. But most surfaces gen-
erated are of acceptable quality and so there arises the need to define imperfection
tolerances related to application.
The need to specify these tolerances on optical drawings has been recognized
for a long time. Countries including the USA, the UK, and Germany developed
standard scratch and dig plates that provided reference imperfections with a wide
range of severities. These were used to identify the severity of a test imperfection—
by eye—by comparing its visibility with the reference values. After a period of
time it became clear that the operation of national standards was always going to
be variable, as the reference plates were difficult to manufacture consistently and
results depended on the observer’s technique and visual performance. Even worse,
correlation between different national standards was completely lacking.
The reason for this was traced to the fact that the microtopography of real
scratches is extremely complex; therefore, it is not practically possible to suggest a
statistical function that could be taken as a measure of severity. Moreover, since the
eye is so sensitive to imperfections on a high-quality substrate, it was impossible
to devise a deterministic approach likely to be appropriate for use by industry.
After much research, it was concluded that imperfection severity could only be
defined by creating a parameter related to practical use and which had traceability
to national standards. The dark-field visibility metric was subjective and lacked
traceability. Since optical imperfections are only of concern due to their ability
to interact with optical radiation, scattering methods were nevertheless obvious
candidates for such a metric.
Extensive tests involving the measurement of light patterns scattered by nu-
merous imperfections over wide angles led to the conclusion that the only com-
mon factor was that all imperfections remove some light from the incident beam
47
48 Chapter 5

by the process of scattering. It eventually became clear that, rather than recording
data based on polar scatter patterns, it would be better to integrate the total light
scattered and use this parameter as a measure of severity. Since this was equivalent
to the amount of light removed by scattering, this measurement could be greatly
simplified by finding the width of a totally absorbing artifact that removed the
same amount of light by the process of absorption. In this way, the severity of any
scratch could be quantified by measuring its line-equivalent width (LEW) and any
dig could be quantified by measuring its spot-equivalent diameter (SED).
These metrics can be measured in terms of the relative contrast of their im-
ages formed in parallel light by a lens of low numerical aperture compared with
images of absorbing artifacts of known size, imaged and viewed in the same way.
As will be described later, this operation can be carried out by the use of a micro-
scope image comparator that can also be used for measuring other features such
as contamination on a surface. These generic techniques, involving the use of spa-
tial frequency filtering and comparator microscopy are described here as far-field
nanoscopy (FFN). They form a platform from which a variety of different mea-
surements can be made.
The standard, ISO 10110-7:19961, specifies two methods for measuring im-
perfections. Method I requires measurement of the area of obscuration caused by
the imperfection, but gives no indication of how this can be achieved. Method II
depends on measuring the visibility of the imperfection under precisely controlled
conditions of illumination and viewing. The personal visibility discrimination of
the observer is established by prior assessment. Tests have established that the
width of imperfections greater than 10 µm can be measured with a measuring
microscope, but smaller imperfections are much more difficult to quantify due to
diffraction effects. Under these circumstances, radiometric rather than geometric
obscuration should be measured using the LEW or SED parameter.
A further major problem in the practical measurement of an imperfection is
identifying its location on the surface. The human operator is able to inspect a large
area and quickly identify an isolated imperfection worthy of measurement. As we
shall see in Chapter 9, the need to inspect high-quality strip products, such as float
glass, of width 1 m or more automatically on-line has resulted in the widespread
use of laser beam scanners for this purpose. These operate on the principle of mea-
suring scattered radiation, but in this case speed of operation is more important than
the ability to resolve the very small imperfections of concern to manufacturers of
optical components.
If speed of operation, sensitivity, and economy in use by industry are major
factors in optical component manufacture, it is doubtful whether a laser scanner
of this type could compete with visual inspection followed by measurement of
borderline cases. This chapter attempts to describe how the perception of the sig-
nificance of imperfections by the optical manufacturing industry, and by users of
optical systems, has developed over the years. The extremely difficult problem of
how to measure the severity of an imperfection by means accessible to even the
smallest workshop will then be addressed in the following chapter.
Surface Imperfections 49

5.2 Imperfections and Subconscious Thoughts

It has been known that some optical companies appear to declare their lack of
interest in this subject since, in their view, to be seen as being involved with imper-
fections would be tantamount to admitting their products were less than perfect!
Others accept that imperfections are always bad news, but just how bad they are
depends on the application and, indeed, on the attitude of the customer. Figure 5.1
illustrates the important part played by the quality assurance (QA) manager and
standards in settling arguments that can arise between the customer and the pro-
duction manager.

Figure 5.1 QA manager caught between the customer and the production man-
ager. (Reprinted with permission of G. Dunn.)

The production manager is naturally keen to maximize yields and from experi-
ence knows that imperfections, although they may be visible, rarely influence the
performance of a system. Most customers, at least of technical products, accept
that very small imperfections can arise but ask: Was the designer, who is responsi-
ble for setting a tolerance, sufficiently aware of the requirements of the particular
application?
The QA manager has the responsibility of ensuring fitness for purpose and
therefore should be able to resolve such problems, provided a standard exists em-
ploying test methods that have traceability to national standards of measurement.
Herein lies the problem. At the time this book was in the planning stage, no such
standard existed.

5.3 Effect of Surface Imperfections

Experience over the years has shown that the performance of optical systems, such
as cameras and binoculars that are used with an extended object field, is unlikely
50 Chapter 5

to be affected by the presence of digs and scratches that can be readily seen on
their surfaces. This observation is supported by the fact that scratched surfaces,
unacceptable from a cosmetic point of view, have little influence on the measured
modulation transfer function (MTF)2,3 of a system. The effect on veiling glare due
to poor polish and coating scatter will usually have much more of an influence than
scratches.
The effect of imperfections on system performance is much more significant
when the field of view is restricted to a point, as is the case with laser optics and
spectrographs. The increased range of image luminosity values generated magni-
fies the significance of radiation diffracted by digs and scratches to the extent that
a single scratch can project a line of radiation at right angles to its length in the
image plane.
The increasing use of very small lenses, as in microengineering, where the area
of the lens becomes comparable with the area of the imperfection, will result in
much closer tolerances on imperfections.
Consumer products, such as spectacles, although rapidly damaged by careless
cleaning, continue to function over a long period of time. The effect of imper-
fections, however, from a cosmetic point of view at the point of sale, cannot be
ignored.

5.4 Impact of Imperfections on Market Access

The changing structure of optical industries around the world combined with the
development of new market areas for optical quality surfaces is having an effect
on the significance being attached to imperfections of all types. Some products
simply will not function if imperfections of nanometer dimensions are present.
We have seen that measurement of other surface metrics, such as form, require
more consideration to be given to detecting the presence of imperfections that may
influence uncertainty of measurement. Access to an increasing number of market
areas will be harder if the problems of imperfections are not adequately addressed.
The cutting, welding, and surface conditioning of materials by high-power or
high-energy lasers require optical components that are free of even very small im-
perfections that could trigger serious damage to any of the optical components
used. New technologies, such as microfabrication, x-ray, and EUV lithography,
are critically dependent on surface quality. The rapidly expanding requirements of
informatics, nanoengineering, and automation depend on the use of high-quality
precision optical components.
Increasing productivity and cost reduction initiatives are encouraging more
companies to subcontract optical component manufacture. This process demands
critical attention to the tolerancing of all constructional parameters, including im-
perfections at the design stage as well as support from international standards.
Problems can even then arise if workers without the necessary training and ex-
perience assemble the delivered components. Current attempts to de-skill optical
production may not be compatible with the drive to improve surface quality.
Surface Imperfections 51

The development of new products and market areas is often accompanied by

the development of new methods of measuring constructional parameters. Before
these can be embodied in standards, they need to be critically assessed by experts to
ensure that they can provide the desired uncertainty of measurement. It is essential,
therefore, that industries keep in close touch with R&D groups developing the new
test procedures, and also with experts drafting the new standards that they will need
to ensure good future communication with their customers.

5.5 Description of Imperfections

5.5.1 Terminology
Imperfections are sometimes classified as area, edge, long, and round. A few terms
used for each class are given below.

• Area imperfections: stains, striae, water spots, mold marks, cement separa-
tions.
• Edge imperfections: chips, optical cement fractures.
• Long imperfections: scratches, sleeks, lint, chuck marks.
• Round imperfections: bubbles, digs, coating voids, dirt, stones.

5.5.2 Size of imperfections

Imperfections may be straight or curved. Their width can be considered to vary
from 0.1 mm down to 10 nm. Long imperfections may extend across the diameter
of the component or exist down to 500 nm. The depth of imperfections may be in
the range 0.1 nm to 0.1 mm.
Measurements have shown that the microtopography of imperfections is very
variable, with the result that such detailed measurements are of little practical value
in quality control.

5.5.3 Substrates/materials
Modern optical systems employ a wide variety of substrates that might suffer dam-
age including, for example, glass, metal, plastics, crystals, semiconductors, thin
films, and cements.

5.5.4 Location
An imperfection can be undesirable wherever it occurs in an optical system. Most
are on surfaces or in coatings, but some exist within the optical material or in a
cement layer. Imperfections arising from the process of assembly are particularly
difficult to quantify when they are buried within the system. Chips at the edge of a
component may give rise to strain extending into the operating aperture.
52 Chapter 5

5.5.5 Characteristics
Microscopic examination of natural imperfections has shown that some, particu-
larly the larger ones, reveal signs of brittle chip formation. Incident radiation will
be scattered and usually little will be directly transmitted. These are termed fully
developed imperfections.
Smaller imperfections, typically those with a width of less than 0.01 mm, can
assume a ductile form with smooth edges. Since these can transmit some radiation,
they are termed partially developed imperfections. In microscopic terms, they are
known as phase objects. Their image appearance depends critically on the method
of illumination and viewing, with the result that their width is difficult to measure
by conventional means.
Depending on the method of generation, surface damage may extend below
the surface and be concealed by subsequent polishing. Such imperfections cannot
be easily detected by physical means, but can be exposed by chemical etching. In
analogy with photographic images, it is suggested here that these might be termed
“latent” imperfections.

5.6 Influence of Imperfections on Quality

In the early days of lens manufacture, optical materials suffered from a number
of imperfections resulting from their method of manufacture, but images could
nevertheless be formed. Customers were advised not to worry because lenses were
made to be looked through and not at. Such advice has never really been heeded.
A visible imperfection, for example, on the front component of a camera lens, will
be perceived as an indication of the quality of the lens as a whole. As mentioned
below, there is an increasing number of applications where cosmetic quality is
still important and the function of some systems can be significantly degraded by
imperfections.

5.6.1 Cosmetic influence

As we have seen, if the imperfection is visible, for example on the outside surface
of a system, this will influence its perceived quality. Even very small imperfections
of nanometer dimensions may be visible to a degree depending on any coating
used and on whether it is viewed in transmitted or reflected light. This is because
visibility depends on these factors. Imperfection appearance depends on the degree
to which it generates a perturbation on the wavefront entering the eye, and this is
influenced by the direction of travel of light and whether reflections are involved.

5.6.2 Functional influence

If the imperfection is near an image plane, it might be regarded at part of the image
and so is considered to have functional significance. The tolerance may have to be
very small if the image is seen under high magnification.
Surface Imperfections 53

Although an imperfection will scatter radiation, it occupies a very small pro-

portion of the area of the lens aperture and so may contribute very little to veiling
glare measured at an image plane.
If, however, a point source is used, such as when measuring the glare spread
function3 , the greater dynamic range of measurement may reveal the spread of
radiation from a scratch extending at right angles to its length in the image plane.
Low-light-level optical systems may be sensitive to imperfections when there are
sources inside or close to a field of view.
Since the shape of the wavefront generated mainly influences the MTF of an
optical system, imperfections extending over a small fraction of the aperture will
have little effect on image quality.
The increasing use of laser scanners and high-power/energy lasers for materials
processing and for energy research is creating more interest in the tolerancing of
imperfections to ensure fitness for purpose. Imperfections on optical systems used
to focus lasers will distort the beam concentration, and the associated radiation
scatter from an imperfection will increase the drift rate of a laser gyroscope. Much
more significant will be component shattering, triggered by an imperfection result-
ing from the passage of a high power/energy laser beam. These are situations where
the severity of an imperfection has a direct influence on the component quality.

5.7 Causes of Imperfections

The cost of inspection and measurement of imperfections can be minimized and
yields increased if clean environments are provided and greater care is taken at
various stages of component manufacture. Physical damage is usually caused by
surface impact or abrasion. Optical damage can arise from exposure to high-
power/energy lasers, and chemical damage can result from environmental pollu-
tion, from mold, or from insects. Optical materials including substrates and coat-
ings can vary significantly in their sensitivity to damage.
Surface microdefects similar to digs may result from imperfect polishing or
from accidental particle or tool impact. Edge chips may be stress induced from
chuck jaws or impact. Scratches can arise from abrasion induced by contaminants
in polishing or cleaning media, or from careless handling at the stages of inspec-
tion, wrapping, or assembly. Sleeks may occur with dry polishing or cleaning.
Strongly adhering particles may occur from airborne dust, from paint, or from the
coating process.
Further causes to be considered include staining, striae, inhomogeneity, mater-
ial bubbles, and poor-quality optical cements.

5.8 Reduction of Damage

Damage may be reduced just by increasing awareness of possible causes, or in-
vestment may be required to provide clean-room conditions and to extend staff
training. The following factors and actions are among those to be given serious
54 Chapter 5

consideration once appropriate measures are in hand to quantify the incidence and
severity of surface imperfections.

• Assess and, if necessary, improve environmental dust control.

• Quantify the occurrence of imperfections and take actions to reduce han-
dling.
• Improve design of lens carriers
• Improve wrapping materials.
• Take greater care in mounting.
• Review and improve the cleaning process.
• Review and improve coating technology.
• Consider use or avoidance of protective coatings.
• Consider the possibility of using harder and/or more stable materials.
• Review the type and frequency of training/education of operators.
• Seek the opinions and ideas of operators.
• Renegotiate acceptance thresholds.

5.9 Imperfection Measurement

Since the measurement of any parameter costs money, it is reasonable to consider
the possible benefits. There is much anecdotal evidence that the tolerance levels of-
ten set in practice by designers are closer than is required by most applications. This
is done to provide a margin of safety since customer complaints can be costly to
resolve. Acceptable methods of measurement enable justifiable levels to be agreed
upon and prices to be related to quality.

5.9.1 Why measure imperfections?

Measurement of imperfections will enable better control of manufacturing,
whether quality is to be based on cosmetic or functional performance. Better con-
trol should lead to higher yields and increased profits.
When objective test methods with traceability to national standards are avail-
able, the way will be clear to prepare international standards. The use of these
standards should improve communication between manufacturer and customer and
enable research to be carried out to establish realistic tolerances for all applications,
whether quality is based on component appearance or function.

5.9.2 Characterization and measurement of imperfections

Before an imperfection can be measured, we must decide what parameters influ-
ence and characterize its visibility or functional significance. Surface roughness,
for example, can be characterized by the RMS metric. Measurement of the micro-
topography of an imperfection can also be carried out using a stylus, by SPM, or
Surface Imperfections 55

by replication and SEM. High spatial and depth resolution can be achieved in this
way, but with the enormous amount of data produced it is extremely difficult to
define and compute a metric that could be related to severity.
The application of microinterferometry to imperfection visualization also pro-
duces attractive displays of topography, but again it is difficult to quantify their
optical significance. Practical problems of these deterministic methods include
the high cost of the test equipment and the time required to measure a single
scratch.
One way of reducing the data is just to measure the width and length of a
scratch and calculate the area of the surface it obscures. Provided the width of the
scratch is greater than 10 µm, an ordinary microscope can be used for this purpose.
The uncertainty of measurement becomes too great when smaller imperfections
that behave as phase rather than amplitude objects, giving rise to edge fringes that
confuse measurement. This form of area metric, although the basis of DIN and
ISO standards, has no optical significance because there is no account taken of
depth.
A number of workers have tried to use the distribution of intensity in the far-
field diffraction pattern of an imperfection illuminated by a laser beam to measure
its severity. To be effective, it is necessary to illuminate both the imperfection and
the substrate, but since the area of the imperfection is likely to be much less than
the area of the substrate illuminated, the contribution to the diffraction pattern of a
small imperfection will be less than that of the substrate.
By far, the most common method of characterizing imperfections is to compare
their visibility under dark-field conditions with a set of reference imperfections.
We shall see that this approach, although favored for inspection and to locate the
position of an imperfection, leaves much to be desired in terms of measurement
uncertainty.
The inverse of this method uses the comparison of the bright-field contrast of
an imperfection with calibrated artifacts. Although apparently less sensitive, this
method has the advantage over all of the previous approaches in being objective and
having traceability to national dimensional standards. A combination of these two
approaches, to provide sensitive inspection of all components followed by mea-
surement of the usually much smaller number of borderline cases, is thought to be
the best solution.

5.10 Comparison of Measurement Methods

The difficulty of measuring imperfections and setting meaningful tolerances is per-
haps evident from the variety of methods and different standards adopted by differ-
ent countries. International trade in the supply of optical components and systems
has been impeded by the lack of an objective method of tolerancing and measur-
ing imperfections. The geometrical complexity of real imperfections renders the
theoretical analysis of radiation/imperfection interaction extremely difficult.
56 Chapter 5

5.11 Imperfection Size Versus Visibility

It is tempting to assume that the greater the width and depth of an imperfection,
the greater will be its visibility. The fact that this is unlikely to be the case can be
seen from studying the visibility of the simplest of all imperfections—the step.

5.11.1 Surface step as an imperfection

Due to the complexity of the analysis involved in computing the visibility of real
scratches, it is worthwhile to study the relatively simple case of an isolated step as
was carried by Hopkins.4
In Fig. 5.2, a step on a reflecting substrate is illuminated from a distant, effec-
tively coherent source. The shape of the wavefront reflected from the surface is
indicated. The distribution of intensity across the image of the step formed by the
lens is shown on the right for step depths on the substrate of λ/20, λ/8, and λ/4,
where λ is the wavelength of the radiation used. The image of the step is seen as
a dark line against a bright background, and its contrast increases with step height
until all the radiation is removed when its height is only λ/4. If C is the percentage
contrast of the central image and δ is the step height on the wavefront resulting
from interaction with the surface, then, providing δ is small compared with λ, it
was shown that C = 140(πδ/λ)2 .
Interestingly, this expression does not contain the numerical aperture of the
imaging lens and only depends on the wavefront step height and the wavelength of
the radiation used. Assuming the eye can detect a line image of contrast of 2%, a
step height of only 0.04λ can be seen, although in the paper it is claimed that much
smaller steps on crystals have been observed. The use of modern image contrast en-
hancement techniques able to detect a change in intensity of 10−6 would, however,
enable step heights of 0.1 nm to be detected using radiation of wavelength 500 nm.
These considerations5 suggest that even the human eye falls into the category of
far-field nanoscopy.

Figure 5.2 Distribution of intensity in the image of a step.

Surface Imperfections 57

This analysis is also useful in explaining the dependence of scratch visibility

on how the component is illuminated and viewed. It follows that, just by measuring
the contrast of their image, very small perturbations of a surface might be assessed.
In a practical situation, of course, a wavefront step has to be created by a step on a
surface, but the effect it has on the wavefront depends greatly on how the surface
is examined, as is illustrated in Fig. 5.3.
Figure 5.3 illustrates the large variations in image contrast C to be expected
when a step of height D = λ/20 on a substrate of refractive index N = 1.5 is
viewed in transmission, in reflection from the air side, and then in reflection from
the medium side.
Taking the visibility discrimination of the eye to be 2%, a step of height 0.07λ
could be seen when viewed in transmission through a medium of refractive index
1.5. This would decrease to 0.02λ when viewed in reflection from the air side and
decreased further to 0.01λ when viewed from the medium side.
This analysis helps to explain why the eye can detect extremely small surface
imperfections (even of zero width, as with a step!) whose profile would not be
easy to measure, and why it is essential to specify the viewing conditions when
inspecting a surface. Due to the quadratic relationship between depth and contrast,
very small changes in depth will be magnified in terms of contrast changes. This
observation suggests that an optical rather than geometrical approach should be
used to quantify imperfections.

Figure 5.3 Variation of step image contrast with viewing conditions.

58 Chapter 5

5.11.2 Step measurement by interferometry

The height of a step on a surface can be readily measured by a microinterferometer.
Figure 5.4 shows a plane wavefront incident vertically on a surface with a step of
height of λ/8. The reflected wave will have traveled twice this value on its return to
interfere with the reference wave. If, now, one beam is tilted, straight-line fringes
will be seen with a discontinuity at the step equal to a fraction of the fringe spacing,
where one complete fringe represents a height change of λ/2.
Modern computer-aided interferometers are sometimes claimed to be capable
of a height resolution of λ/1000. As discussed in Chapter 2, some caution is, how-
ever, in order because small changes in test conditions and variations of the optical
constants across the surface, if they occur, may give rise to errors of unknown
magnitude.

Figure 5.4 Step measurement by interferometry.

5.12 The Eye as a Sensor

The process of inspection employing the eye in association with controlled hand
movements has, over many years, proved to be the preferred way for the routine
quality control of surface quality. In spite of advances in machine vision, it has been
difficult to surpass the use of the eye as a sensor for this particular application.

5.12.1 Benefits
The dynamic range of sensitivity to light of the human eye, from detection to the
onset of pain, must be at a record level for any sensor at 1014 . Thanks to diffraction
effects, this performance enables changes in surface heights of a few nanometers
Surface Imperfections 59

to be detected over a wide range of levels of illumination. Moreover, due to the

fast response time of the eye, small imperfections scattered over a high-quality
substrate can be readily seen.
Rapid manipulation of the component enables inspection of both flat and
curved surfaces. Once detected, the eye is also able to classify the imperfection
and, in most cases, complete the process of quality control. Finally, it has to be ad-
mitted that complete replacement of subjective controls by automation is unlikely
to be economic for some time to come.

5.12.2 Disbenefits
As one would predict, there are some significant disbenefits resulting from using
the eye for the control of surface quality. Careful selection of individuals must be
followed by a period of training, during which the imperfection detection level of
an individual will rise to a level that could change from day to day and even during
the day due to fatigue. Detection levels are likely to gradually improve over the
long term.
The process does of course involve contact with the component including, on
occasion, the use of a brush to distinguish between dust and digs. Such manual
inspection could give rise to further damage and decrease yields.
Since the process is subjective by nature, it is difficult to monitor and avoid
drifts in tolerance levels. The capital investment involved will, in any case, include
special lighting and most likely a laminar-flow cabinet.

5.13 Disbenefits of Inspection

The degree of inspection needed and its associated cost will vary with the product
being manufactured and the method of manufacture. A low-cost product would
not justify detailed inspection and measurement, whereas a product expensive to
manufacture and sensitive to imperfections would need to be supported by the best
available means for quality control.
Assuming a well-organized manufacturing process, probably the most likely
cause of scratches will arise from the process of cleaning. Some forms of contam-
ination due to molecular contact with a surface can only be removed by physical
contact involving rubbing or polishing. This carries with it a high risk of surface
damage. Ultrasonic cleaning also involves dissipation of energy at the surface that
may cause damage to relatively sensitive surfaces. See also Sec. 8.11.
Small imperfections on a curved surface may only be detected by careful ma-
nipulation of a component through several degrees of freedom so as to pick up the
necessary low-angle scattered radiation. Handling in this way and by jig loading,
transportation, and wrapping of components all play their part in increasing the
risk of surface damage.
Since even highly skilled inspectors require a finite time for inspection, this
process must add to costs that could be avoided if only damage-free manufacturing
could be achieved. Due to the subjective nature of inspection, involving the use
60 Chapter 5

of reference pieces and supported by careful training, experience has shown that
there is still plenty of scope for lack of agreement between inspectors when assess-
ing the surface quality of some components. The experienced inspector, moreover,
may over time acquire a more critical approach leading to a trend toward lower
yields. Quality assurance procedures call for objective records of performance pa-
rameters supported by traceability to national standards that are not available in
the absence of objective methods of measurement. Unexpected delays in produc-
tion can also arise from the lack of agreed acceptance imperfection thresholds and
the occurrence of complaints regarding quality that could be costly to resolve.
It may be concluded that even with the best available manufacturing processes,
the costs associated with inspection have to be accepted as a necessary precursor
to the measurement by objective means of borderline cases.

5.14 National Standards for Scratch Assessment

Periodic attention has been devoted to ideas for improving the inspection process
over the last few decades. The national standards committees6–11 of the U.S., Ger-
many, France, and the UK have each produced their own standards. The main
characteristics of these are summarized in Table 5.1 below.

5.14.1 United States

The standard, MIL-0-13830A:1963—Optical Components for Fire Control Instru-
ments; General Specification Governing the Manufacture, Assembly, and Inspec-
tion of, was drafted for military systems including sights, telescopes, and range
finders. It has been widely specified by governments over the last 40 years, but
is currently undergoing revision. Imperfection tolerances are stated in relation to
scratch visibility reference plates. Separate glass substrates each have a single fine
scratch produced by a diamond tool or by etching. Five levels of scratch severity are

Table 5.1 National standards for scratch assessment.

Country Type of Range of severity Principle of
reference measurement
scratch
U.S. Engraved MIL: 10, 20, 40, Dark-field visibility
MIL-0-13830A:1963 lines 60, 80
Germany Chromium 2.5, 4.0, 6.3, 10 µm Area measurement
DIN 3140: Part: 1978 lines width
France Chromium 7 µm width Threshold visibility
AFNOR lines
UK Line width Any chosen Line-equivalent
BS 4301:1991 standard width
Surface Imperfections 61

used and designated: S-10, 20, 40, 60, and 80. The numbers are intended to indi-
cate visibility8 and not width. Traceability is based upon visual comparison with a
set of master standards held by the U.S. Army. The standard recommends that opti-
cal components shall be inspected by “approved optical methods and equipment in
accordance with applicable detail specifications.” A simple dark-field illumination
system is described for this purpose.

5.14.2 Germany
The standard DIN 3140: Part 7—1978 Dimensional and Tolerance Data for Op-
tical Components: Surface Defects involves measuring the surface area of every
imperfection above a certain value. The severity or grade of an imperfection is
equal to the edge length of a square of equal area. Since no account is taken of
depth, all imperfections of equal area are judged equal although they may vary in
visibility due to differences in depth. No recommendations are given on methods
of measurement, but current practice appears to favor visual comparison of area
with opaque lines of known size on a comparator plate or by the use of a traveling
microscope.

5.14.3 France
The early standard AFNOR NFS 10-006:1979 Experimental; Optical Elements:
Surface Imperfections has been replaced by a more objective method now covered
in ISO 10110-7:1996, to be described9 in Sec. 5.20.

5.14.4 United Kingdom

The early standard BS 4301:1982 Preparation of Drawings for Optical Elements
and Systems embodied both functional and cosmetic imperfections. The former
required the measurement of width and length of imperfections and the latter re-
quired a comparison of visibility with standards made as grooves in a glass sub-
strate of the same width but varying in depth. This was eventually replaced by BS
4301:1991, which introduced the concept of line-equivalent widths12,13 to replace
the need to make reference grooves, and then by BS ISO 10110-7:1996, described
in Sec. 5.20.

5.15 Level of Agreement Achieved Using National Standards

The pressing need to reduce optical component manufacturing costs in recent years
led companies to subcontract, as a cheaper alternative to in-house production. This
process required more attention to constructional tolerances and a greater need
for objective measurement with traceability. Manufacturers suspected there was
little equivalence between various imperfection tolerances arising from these var-
ious standards and that this was giving rise to overstrict tolerancing and increased
62 Chapter 5

costs. Accordingly, Sira (formerly the British Scientific Instrument Research As-
sociation) was requested to quantify the level of agreement to be expected and, if
possible, propose an objective method of measurement that might form the basis
of an international standard.
A total of 17 experienced inspectors, taken from commercial and UK Ministry
of Defence (MOD) establishments active in the manufacture and/or use of quality
optical surfaces, carried out the comparison trials. Plane transparent samples were
used to facilitate data analysis although it was realized that different element di-
mensions could influence the results. This was taken into consideration by noting
inspectors’ comments when drawing conclusions.
Sets of comparator standards employed by the various standards were obtained.
About this time, an experimental microscope image comparator12 for measuring
imperfections was offered by Sira to be included in the comparison trials.
Experimental detail of this assessment of current scratch standards can be found
in Ref. 7. A typical result of the many tests carried out is shown in Fig. 5.5.

Figure 5.5 Levels of agreement using French, German, UK, and U.S. standards.
Surface Imperfections 63

The maximum percentage of agreement obtained between the inspectors is

plotted for various samples bearing scratches, employing methods described in the
national standards.
Using BS 4301:1982 (top left), around 60% agreement was recorded for sam-
ples A, C, and H, whereas using the French method (top right), over 90% agreement
was found using samples B, C, E, F, and G. The German approach (bottom left)
gave less than 50% agreement on all samples due to subjective variations in area
assessment. The only objective method employing the microscope image compara-
tor (bottom right) should, of course, have recorded 100% agreement, as variations
between individuals are taken out. The drop in samples G and H was attributed to
a wrong control setting.
Immediately after completing the trials, the inspectors were asked for their
opinion on various criteria relating to each type of inspection station in terms of
its ability to satisfy general requirements. The UK and U.S. inspection stations, in
conjunction with visibility standards, were considered to provide a quick and sim-
ple facility for assessing cosmetic quality. Both large and small components could
be examined relatively easily, and it was considered a major advantage that a view
of the total sample surface was available. The major drawback was that the test of
visibility was unacceptably subjective.
The French station was well received, being easy and fast to use in a produc-
tion environment. Inspectors, however, disliked the setup procedure and thought
the system would prove tiring when in continual use. Use of a comparison plate to
determine surface area, as required in the German approach, was not well received.
No confidence was expressed in this technique for a production environment, al-
though some thought views might change with more practice. Measurement of
surface area with a traveling microscope was regarded with more enthusiasm but
deemed too time consuming to be of use in a production environment.
Most inspectors found the microscope image comparator very easy to use once
it was set up. The main objection was the small field of view. The hope was ex-
pressed that the field of view could be increased, in which case most requirements
would be satisfied.
The study concluded that visual inspection using some form of dark-field view-
ing in conjunction with some reference pieces would satisfy most needs, provided
it was backed up by an objective method of measurement with traceability, such as
some form of image comparator. This would have the great advantage of reducing
the need for more skilled inspectors.

5.16 Scratch Reference Standards

Since measurement requires comparison with a standard, a great deal of effort has
been expended over the years in trying to create reference artifacts of varying sever-
ity to simulate naturally occurring scratches.
A desirable appearance specification for these artifacts should include indepen-
dence of the following:
64 Chapter 5

Wavelength
Magnification
Direction of viewing
Coherence of illuminating wave
Direction of illuminating wave

Needless to say, little success has so far been achieved in those listed above, but
the following technologies have been tried:

Ruling
Success in ruling grooves of a precisely known form depends on very accurate
control of the tool and a nonbrittle substrate. Glass is therefore unsuitable, accept
for very small grooves when plastic deformation may occur. Softer materials such
as gold or copper have been used with some success.

Etching
Chemical etching was used as the basis of early BS reference standards, but the
process was difficult to control and therefore too expensive. Ion-beam etching suf-
fers from similar problems.

Electron-beam machining
The generation of any scratch topography was thought possible by the multiple use
of extremely fine lines as described in Refs. 14–16, but again the process is difficult
to control and costly to implement.

Thin-film deposition
This is the alternative to etching lines, but as with all these simulated techniques,
appearance will be governed more by geometry and refractive index than by scatter,
which may dominate over real scratches; and cost of preparation will be high.

Embedded fibers
This approach has the advantage that, by careful choice of the refractive index
of the fiber and embedding medium, a controllable wavefront perturbation can be
achieved. The effect of chromatic and temperature variations combined with the
cost of measurement and fabrication, however, has led to the idea being abandoned.

Injection molding
This is probably the favored method of producing low-cost scratch artifacts. A mas-
ter surface can be ruled in a soft metal and specimens reproduced cheaply by pre-
cise injection molding. Since the profile stands proud of the plastics, surface dam-
age can easily occur by cleaning; and since scratch appearance depends critically
on profile, some form of calibration is necessary, which means this is really only a
secondary standard.
Surface Imperfections 65

5.17 Target Specification for Imperfection Measurement

The first requirement for the design of any measuring instrument is agreement
on the specification of performance with the customers, and what it is that we
are measuring. Reliability combined with economic purchase and low-cost staff
training, together with easy system integration and operation, is taken as read. Now
we need to consider sensitivity, dynamic range, and uncertainty of measurement.

• Depth (1000 nm > depth > 0.1 nm)

A practical range of depth of most digs and scratches is 1000 nm to 0.1 nm.
The latter value is thought to be of interest to studies on laser damage, also
noting that the unaided eye can detect surface height changes of only a few
nanometers.
• Width (0.1 mm > width > 50 nm)
Imperfections of width greater than 0.1 mm should be rare and are likely to
cause rejection on grounds of appearance, whereas values less than 50 nm,
at least with present knowledge, are considered less important.
• Length (component diameter > length > 0.1 mm)
A scratch could extend across the diameter of a component and be unlikely
to be less than 0.1 mm in length. Also bear in mind that it can vary in width
along its length.
• Surface radius of curvature (infinity > radius > 1 mm)
Imperfections on flat surfaces such as windows and filters have always been
of concern, but increasingly small high-power lenses used in microscopes,
eye implants, and information technologies, employing highly curved sur-
faces, have been found to be functionally degraded by quite small imperfec-
tions.
• Surface coatings
Since a thin-film coating on a surface with imperfections can change the way
in which radiation is scattered, the instrument must be able to examine any
surface whether coated or uncoated.
• Optical radiation used
The degree with which radiation interacts with an imperfection will depend
greatly on wavelength. The wavelength and spectral band chosen should re-
late to the application and sensitivity of measurement required.
• Transmission or reflection
Optical components may be transmitting or reflecting to any degree—all
must be capable of measurement. The examination of assembled systems
where any damage caused by the processes of assembly can only be assessed
is also desirable.
• Component size
Since the diameter of components can be upwards of 1 mm, the method of
measurement should be capable of quantifying imperfections of very large
mirrors as well as microlenses.
66 Chapter 5

• Speed of measurement
Since a skilled inspector can examine the surface of a component of 100 mm
diameter in about 10 seconds, the instrument should perform at least as well
as this.
• Sensitivity of measurement
This will depend on the application, but previous experience suggests that it
should be possible to measure down to a MIL 10 scratch.
• Uncertainty of measurement
The metric related to imperfection severity should be capable of measure-
ment by objective means with traceability to national standards. The toler-
ance level and uncertainty of its measurement will depend on the component
application. A target for uncertainty of measurement might be taken as 0.2
LEW.

It is no wonder that the whole of optical production so far has managed without
an instrument operating to this specification, and it may be some time before such
an instrument is available. Nevertheless, considerable progress has been made in
recent years and much of the remainder of this book will be devoted to describing
what has been done and achieved, and how.

5.18 Need for Standards

Standard methods for measuring imperfections are needed to support trading by
improving communications between companies and by removing barriers to trade.
Efficiency of manufacture can be improved by helping the diagnosis of manufactur-
ing faults and by less dependence on highly skilled inspectors. Existing imperfec-
tion tolerances, set at a high level to reduce costs associated with returns, can now
be relaxed as a result of an ability to control quality with confidence and thereby
achieve a consequential saving due to increased yields. The result is an increase in
competitiveness and more satisfied customers.

5.19 ISO TC 172 Optics and Optical Instruments

Progress in optical and information technology after World War II and in particular
the invention of the laser in 1960 has led to an ever-increasing range of new optical
and photonic products and market areas. In 1978, the International Organization for
Standards (ISO) decided to create a new technical committee, TC/172, to support
the industries. The various subcommittees (SCs) involved are listed below:

SC 1 Fundamental standards
SC 2 Materials for optical processing
SC 3 Optical materials and components
SC 4 Telescopes
SC 5 Microscopes
SC 6 Geodetic instruments
Surface Imperfections 67

SC 7 Ophthalmic endoscopic, metrological instruments, and test methods

SC 8 Ophthalmic optics
SC 9 Electro-optical systems

The Working Groups 1 and 2 of SC 1 are principally concerned with the subjects
discussed here. A comprehensive list of terms used in relation to optical materials
is available in ISO 9802.17

5.20 Comparison of Two Methods Proposed by ISO in 1996

The standard ISO 10110-7: Surface Imperfection Tolerances specifies the presenta-
tion of design requirements for optical elements and systems in technical drawings
used for manufacturing and inspection. Indications and rules are given for the level
of acceptability of surface imperfections. Allowable surface imperfections may be
specified either by measurement of their area of obscuration (Method I) or by mea-
suring their visibility (Method II).

5.20.1 Method I
Surface imperfections are characterized in the form N × A, where N is the number
of allowable surface imperfections of maximum permitted size and A is the grade
number, which is equal to the square root of the surface area of the maximum al-
lowed imperfection, expressed in millimeters. Separate indications are given for
coating imperfections, scratches longer than 2 mm, and edge chips. No recommen-
dations are given for the method of measurement of imperfections.

5.20.2 Method II
This method requires the use of inspection stations in which the method of illumi-
nation and viewing of the element under test is carefully controlled and the visual
sensitivity of the observer is calibrated. Separate stations are needed for elements
viewed in transmission and reflection.
The sample is diffusely illuminated in an integrating sphere by a controlled
light source that can be set at a number of fixed levels, and viewed against a
background of continuously variable luminance. The observer is first calibrated
by viewing a calibrated reference defect. This is seen under a specified illumina-
tion and then the background is adjusted until the defect is just not seen. If now the
reference defect is replaced by the element under test and any imperfections are
visible, the component will be rejected.

5.20.3 Comparison of Methods I and II

These two methods are compared in the Table 5.2 below.
The main problem with Method I is the poor agreement found between ob-
servers when measuring small imperfections. Method II has the advantage of a
68 Chapter 5

Table 5.2 Comparison of Methods I and II.

Area Method I Visibility Method II
Every imperfection is Whole surface is seen
measured
No optical significance Has optical value
Inaccurate for small Component size limitation
imperfections
Time consuming Quick test
No obvious link to application Insensitive to small
imperfections
Appears to be objective Subjective test
Operator fatigue Operator fatigue

go/no-go mode of operation with full-field observation, but the disadvantages of

subjectivity and insensitivity to small imperfections. It is not obvious how thresh-
olds for the two approaches can be related. As it will be described later, a survey of
industry has shown that modern optical systems manufacture requires component
features to be measured to ensure adequate quality.

5.21 Chapter Conclusions

We have seen that damage to a high-quality surface can arise at any time during
surface generation, system assembly, at testing, or in use. The imperfections arising
can influence the appearance and/or function of most optical components. Their
impact on cost and quality is of concern to an extremely wide range of industries
and consumers throughout the world.
Until recently, imperfection tolerances, although specified at the component
design stage, were assessed almost entirely by subjective methods and occasion-
ally led to controversy. The recent development of objective methods of measure-
ment of imperfections with traceability to national standards, to be described later,
should enable the development of tolerances at the design stage set at levels aimed
at reducing component costs and, therefore, increasing profitability.

References
1. ISO 10110-7:1996, Optics and optical instruments—Preparation of drawings
for optical elements and systems. Part 7. Surface imperfection tolerances.
2. T. L. Williams, The Optical Transfer Function of Imaging Systems, Institute of
Physics Publishing, Bristol, UK (1999)
3. S. Martin, “Glare characteristics of lenses and optical instruments in the visible
region,” Optica Acta, Vol. 19, pp. 499–513 (1972).
4. H. H. Hopkins, “Phase structures seen in the ordinary microscope,” Revue
d’Optique, pp. 142–152 (1952).
Surface Imperfections 69

5. L. R. Baker, “Optical surface assessment: Parametric characterisation of im-

perfections,” Proc. of SPIE, Vol. 3739, pp. 480–487 (1999).
6. L. R. Baker and J. Singh, “Comparison of visibility of standard scratches,”
Proc. of SPIE, Vol. 525, pp. 64–69 (1985).
7. A. J. Cormier, “Assessment of current scratch standards,” Proc of SPIE,
Vol. 805, pp. 152–159 (1987).
8. M. Young, “Objective measurement and characterization of scratch standards,”
Proc. of SPIE, Vol. 362, pp. 86–92 (1982).
9. A. Huard, “Visibility method to classify microscopic surface defects for both
reflection and transmission systems,” Proc. of SPIE, Vol. 525, pp. 64–69
(1985).
10. J. A. Slater, OSTAG 821 Scratch Assessment Report 16/85, UKMOD, Royal
Arsenal, Woolwich, London (1985).
11. J. A. Slater, “Scratches: At what price quality,” Proc. of SPIE, Vol. 654, pp. 68–
## (1986).
12. L. R. Baker, “Microscope image comparator,” Opt. Acta, Vol. 31, pp. 611–614
(1984).
13. L. R. Baker, “Inspection of surface flaws by comparison microscopy,” Appl.
Opt., Vol. 27, No. 22, pp. 4620–4625 (1988).
14. E. G. Johnson, Jr., “Simulating the scratch standards for optical surfaces: The-
ory,” Appl. Opt., Vol. 22, p. 4056 (1983).
15. M. Young and E. G. Johnson, Jr., “Tunable scratch standards,” Proc. of SPIE,
Vol. 525, pp. 70–77 (1985).
16. M. Young, “Scratch-and-dig standard revisited,” Appl. Opt., Vol. 25, No. 12,
pp. 1922–1929 (1986).
17. ISO 9802:1996, Raw optical glass—Vocabulary.
Chapter 6
Measurement of Imperfections
by Obscuration
6.1 Introduction
Since an imperfection can occur anywhere over the surface of a component of
any size or even within an assembled system, measurement cannot take place until
it has been found and located in the measurement area. Inspectors often prefer
to retain inspection methods for locating imperfections that they themselves have
developed. This is inevitable due to the subjective nature of visual inspection. This
chapter describes the currently favored methods for the objective measurement of
imperfections.

6.2 Optical Component Inspection

Many different proposals have been made for the rapid inspection of optical el-
ements for imperfections. A prime consideration is comfort for the observer, but
with sufficiently high levels of illumination to permit detection of the smallest im-
perfections of interest. Clean laminar airflow over the specimen will minimize dust
deposits. Dark-field viewing, with the aid of a magnifier if necessary, with the eye
positioned to collect radiation scattered over low angles, is generally regarded as
the preferred arrangement. A typical inspection station1 is shown in Fig. 6.1.
The component under inspection, shown here, is being viewed in transmitted
light by magnifier. A component to be inspected in reflected light is placed in the
bottom corner of the box and tilted so that the specular light just misses the eye.
Due to the directional effect of light scattered by a scratch, the component may
need to be rotated in its plane so that the eye can receive as much scattered light as
possible.

6.3 Radiometric Obscuration by Imperfections

We have suggested that a useful metric to quantify the severity of an imperfection
in optical terms is to determine the extent to which it obscures incident radiation.
Application of the method of substitution, where the radiation removed from an
71
72 Chapter 6

Figure 6.1 A typical inspection station.

incident beam by the imperfection is equated to that removed from the same beam
by an opaque line or spot of known dimensions, reduces the number of instrument
parameters needing accurate calibration. Figure 6.2 illustrates schematically the
difference in widths expected between partially and fully developed imperfections
having the same image contrast.
The ray diagram at level A shows a component illuminated by a small distant
light source. The low-aperture lens used to form a bright-field image of the imper-
fection collects some of the radiation scattered by the imperfection in the forward
direction, together with the background radiation. The choice of a very low aper-
ture lens helps remove any fine structure present in the image of the scratch, which
appears as a dark line against a bright background. The more visible the scratch,
due to a greater level of scattered radiation, the higher will be the contrast of the
scratch image.
Image contrast is measured as (Imax − Imin )/(Imax + Imin ), where Imax is the
intensity of radiation surrounding the image, and Imin is the minimum intensity at
the image of the scratch.
At level B, a MIL:60 scratch of width 0.050 mm gave the same image contrast
of 0.40 as an opaque line, or fully developed scratch, of width 0.004 mm. This
scratch therefore has a severity of 4 LEW. Similarly, at levels C and D a MIL:40
scratch has a severity of 1.6 LEW and a MIL:10 scratch has a severity of 0.25 LEW.
A simple audience demonstration of this principle of measurement, suitable for
use on an overhead projector, can be made by securing, side by side and 10 mm
apart on a transparent plastic plate, a number of wires 20 mm long and of increas-
Measurement of Imperfections by Obscuration 73

Figure 6.2 Partially and fully developed imperfections can have the same contrast.

ing diameter. A suitable series of diameters, based on the ISO recommendations,

is 16, 25, 40, 63, 100, 160, 250, 400, and 630 µm. A transparent cursor bearing
two parallel wires, of 63 µm diameter and 3 mm apart, can be traversed in paral-
lel guides so as to straddle each of the other wires in turn. When projected on a
screen, the audience will see the sequence of wire images to be of gradually in-
creasing contrast. As the cursor is moved along to approach each wire in turn, it
will be clearly seen when the cursor stops symmetrically over the wire of the same
diameter. Only then will the three images be of equal contrast, even to those in
the audience with no previous experience in inspecting for imperfections. The wire
images should be carefully focused before moving the cursor across.

6.4 Calibration Graticules

The substitution method requires the use of calibrated graticules. A reference
graticule, to be used in transmitted radiation, should carry nonreflecting opaque
lines and dots on a nonreflecting glass substrate. A reference graticule, to be used
in reflected radiation, should carry transmitting lines and dots on a reflecting sub-
strate, apart from those areas occupied by lines or dots, which should be nonre-
flecting.
74 Chapter 6

An example of a reference graticule1 that could be calibrated by a national

standards body is shown in Fig. 6.3. Present methods of measuring line widths are
thought to achieve an uncertainty of 10 nm.

Figure 6.3 BS 4301 (1991) graticule, line widths, and spot diameters in micrometers.

6.5 LEW and SED Measurement Requirements

The simple measurement system shown in Fig. 6.2A operates by removing the test
sample and replacing it with a calibrated graticule as indicated in Fig. 6.3. In the
case of a scratch, the width of line is found that gives the same contrast as the image
of the scratch. In a similar way, the diameter of a graticule spot is found that gives
the same contrast as the image of a dig under test. Imperfections with dimensions
greater than 10 µm are assessed in terms of their diameter.
The following conditions should be observed in order to maintain an acceptable
uncertainty of measurement:

• The test area and the reference graticule should be illuminated and imaged
under the same conditions.
• Parallel but not coherent illumination should be used so as to avoid speckle
in the image plane.
• A low lens numerical aperture, such as 0.01, should be used and it should be
big enough to provide sufficient signal for measurement, but small enough
to prevent imaging of fine structure present in the imperfection.
• A TV or digital camera is preferred to visual observation of an image, which
could be influenced by fatigue.

6.6 LEW and SED Simple Viewing System

As has been indicated, probably the majority of optical components manufactured
only need inspection rather than the measurement of any imperfections. Some com-
panies may nevertheless feel the need for a low-cost version of an instrument to
measure imperfections, although this may be rarely used.
Simple viewing systems, enabling a direct visual comparison of the contrast of
an imperfection with the contrast of the images of lines on a reference graticule,
are shown schematically in Fig. 6.4.
Measurement of Imperfections by Obscuration 75

Figure 6.4 Simple viewing systems to measure LEW and SED values.

Operators of these systems may soon feel the need for additional magnification
and try the use of a magnifier or, better still, a low-power microscope to view the
imperfection.

6.7 Analogue Microscope Image Comparator (AMIC)

The microscope image comparator2 (MIC) was designed to aim for the target spec-
ification set down in Sec. 5.17. Although visual observation of images through an
eyepiece was possible, a TV display that could be seen by a number of operators
was considered desirable. As the technology improved, a digital still camera3 be-
came available as a low-cost alternative for a TV camera.

6.7.1 Description
A schematic arrangement of a typical analogue microscope image comparator is
shown in Fig. 6.5.
Light from a tungsten lamp S is focused by the condenser lens L1 onto the
pinhole P. After passage through the polarizer Z1 , the light from P, which is at
the focus of L2 , passes as a parallel beam into the polarizer beam splitter B. The
beam passing straight through B is transmitted by the quarter-wave plate Q1 before
falling at normal incidence onto the plate R, which carries a transparent reference
slit in a reflecting substrate. The light reflected by R can now be imaged, after a
76 Chapter 6

Figure 6.5 Schematic diagram of a microscope image comparator.

return passage through Q1 and B by lens L3 , onto the TV camera after transmission
by the analyzer Z2 and by the spatial frequency filter F placed in the back focal
plane L3 . This spatial frequency filter can provide attenuation or phase shifting, or
a combination of both, but is usually just a pinhole of diameter sufficiently small
to remove the fine structure from the imperfection image by reducing the effective
aperture of L3 . The parallel beam from L2 , reflected by B into the test channel,
is transmitted by the quarter-wave plate Q2 and illuminates the test specimen T at
normal incidence. As in the reference channel, the light reflected by T can now be
imaged after a return passage through Q2 and B by L3 , also onto the TV camera.
The plates Q1 and Q2 are rotated in turn in a setup operation to maximize the
intensity of the beams from T and R on the TV camera. As the light falling on T
and R is circularly polarized and at normal incidence, the amount of light reflected
is found to be independent of imperfection orientation.
If the imperfection is on a transmitting substrate such as a lens T
instead of on
a flat mirror T, the light would need to be returned by the use of a retroreflecting
screen R
and quarter-wave plate Q
2 now placed between the lens and this screen.
This alternative configuration is shown at the bottom right-hand corner of the fig-
ure.
A TV line profiler can be used, showing the distribution of intensity of a scratch
image as an alternative to viewing the image contrast. As will be shown below, the
tangent squared of the analyzer angle, at the setting of equal image contrast from
both channels, is the ratio of peak radiation removed by the test and reference
artifacts/imperfections.
Measurement of Imperfections by Obscuration 77

The spot-equivalent diameter of a dig can be found by calibrating the analyzer

angle with circular artifacts of known diameter.
An imperfection measured in reflection and then in transmission may not have
the same contrast value. The mode of imperfection illumination must always be
stated.
In transmitting mode, a calibration graticule will have opaque lines/spots on
a transmitting substrate; and in reflecting mode, the plate will carry transmitting
slits/spots on a reflecting substrate.
A power-balancing lens with antireflection coatings is needed above T when it
has power and is measured in reflection.

6.7.2 Theory
If A and B are the peak intensities of a light beam polarized in each of two perpen-
dicular directions, then, by the law of Malus, the intensity at the TV camera from
the combined channels is given by

A(1 − t) cos2 θ + B sin2 θ

at the test imperfection,

A cos2 θ + B(1 − r) sin2 θ

at the reference artifact, where θ is the angular setting of Z2 , and t and r are the
fractional drops in intensity due to the presence of the test imperfection and the
reference artifact. If θ is now rotated to equalize the contrast of the features in the
image plane, then

A(1 − t) cos2 θ + B sin2 θ = A cos2 θ + B(1 − r) sin2 θ.

This can be reduced to

At = Br tan2 θ.

If A and B are first set equal by an appropriate rotation of the polarizer Z1 , then
t
= tan2 θ.
r
It follows that the ratio of the contrasts of two imperfections/artifacts under com-
parison t1 /t2 is equal to the ratio of the tangent squared of the two analyzer set-
tings θ1 and θ2 required to equalize the contrast of the test imperfection with the
reference artifact built into the instrument, or

t1 tan2 θ1
= .
t2 tan2 θ2
78 Chapter 6

6.7.3 Method of operation

To bring about the conditions described above, Z2 is rotated continuously, but,
since the specular reflectances of T and R are most likely different, the background
intensity on the TV will be modulated. This modulation is reduced to zero by care-
ful rotation of Z1 . The angular setting θ1 of Z2 can now be set to the value where
the contrast of the image of the imperfection on T is equal to that of the reference
artifact on R. A second value θ2 could, of course, be obtained using a calibrated
line in order to relate any imperfection to a line of known width. In this way, the
angular setting of the analyzer can be calibrated in relation to lines of known width
of a set of reference imperfection specimens.
It is common for imperfections to have microstructure. The effect of this in
the image can be minimized by the use of an imaging lens with a small aperture
able to provide a smooth line-type shadow, the maximum contrast of which can
be measured. An advantage of using a small-aperture lens is the extended depth of
focus available. This can be further increased by using a telecentric lens with its
front focus at F and its back focus at the image sensor. Although the instrument
is not “resolving” the imperfections, or for that matter the lines on the calibration
graticule, it is responding to the amount of light they remove in terms of reduced
contrast at the image plane.
The instrument, when used in reflection, can be calibrated with a series of trans-
parent slits of known width cut into a reflecting substrate. The edges should be as
steep as possible in a film typically 100 nm thick. The total light scattered by a
particular imperfection is in this way equated to the light transmitted by a slit of
“equivalent” width cut into a reflecting substrate.
When the imperfection is on a transmitting substrate, as with the lens T
instead
of on a flat mirror T as shown in Fig. 5.10, the light needs to be returned by the use
of a retroreflecting screen R
and quarter-wave plate Q
2 now placed between the
lens and this screen. Calibration of the comparator operating in transmission can
be carried out using black opaque chromium lines of known width on a transparent
nonreflecting substrate such as glass. It should be noted that the severity of an
imperfection seen in reflection would, in general, be greater than when the same
imperfection is quantified in transmission as explained in Sec. 5.11. A schematic
representation of the control of the contrast values of the image of a line of width
0.01 mm, in the center of the field, and two parallel reference lines is shown in
Fig. 6.6.
In the top diagram, the two parallel lines in the reference channel have a con-
trast lower than that of the test line image between them. The analyzer dial reads
36 deg. The reference channel lines in the bottom diagram have a higher contrast
when the dial is moved round to 46 deg. Searching for a balance of contrasts for
all three of the line images, as in the middle diagram, results in a dial setting of
41 deg. Assuming the reference lines have unit contrast, the value for the test line
is tan2 41 = 0.76.
A laboratory version of the microscope image comparator being used to cali-
brate a set of MIL-Standards imperfections is shown in Fig. 6.7.
Measurement of Imperfections by Obscuration 79

Figure 6.6 Equalizing line image contrasts to quantify imperfection.

Due to the compact nature of the instrument, it can also be mounted on a trans-
parent inverted saucer made from a plastic material suitable for sliding across the
surface of a large optical component. In this way, imperfections likely to cause
concern can be quantified on large mirrors or windows. Figure 6.8 illustrates an
imperfection comparator used in the way.

6.8 Digital Microscope Image Comparator (DMIC)

The general availability of relatively low-cost digital cameras and associated soft-
ware, such as Photoshop®, has led to the development of a simpler alternative
version3 of the AMIC.

6.8.1 Description
This instrument uses a digital camera to record the image of an imperfection or
calibration line, the contrasts of which are later measured by downloading into
a computer. Figure 6.9 shows a schematic representation of a typical arrange-
ment.
80 Chapter 6

Figure 6.7 Sira microscope image comparator used with a line-profile analyzer.

Figure 6.8 Microscope image comparator in use on a large mirror.

Measurement of Imperfections by Obscuration 81

Figure 6.9 Schematic representation of a microscope image comparator employ-

ing a digital camera.

The instrument is based on a standard microscope with vertical illuminator 1–3

and spatial frequency filter 8, placed in the back focal plane of the microscope ob-
jective 7. A tungsten-halogen lamp and condenser illuminates the pinhole source 1
that is at the focus of the collimator 2. The parallel beam produced is reflected
downwards by the polarizing beam splitter 3 to illuminate the test specimen 4. The
beam transmitted by 4 is returned by the retroreflector 5 and two passages through
the quarter-wave plate 6.
This plate is needed to allow the returned beam to be transmitted by 3 and
imaged by 7, having passed through 8 and the high-aperture field lens 9. This is
needed to converge the field rays down to be accepted by the small-aperture lens
in the camera 10.
If 4 is a reflecting specimen, 5 is removed as it is no longer required and 6
is placed between 3 and 4. Since even a quarter wavelength of defocus resulting
from an object movement of ∼0.3 mm can cause a change in value of the image
contrast of ∼20%, care is required in selecting the best focus. If available, a visual
image channel employing an eyepiece can be used for selecting the best “peak”
82 Chapter 6

focus or, alternatively, the digital camera can be plugged into a TV. The LCD of
the camera may not have sufficient resolution to perform this task with the neces-
sary precision. With this instrument, an imperfection image appears as a dark line
or dark spot, if a dig is present, seen against a bright background. Having focused
and operated the camera shutter, the next task is to download the stored image into
a PC supplied with image processing software. (The procedure to follow is based
on Photoshop®.) The image magnification on the screen should be adjusted to pro-
vide at least 10 pixels across the image to be measured. To minimize the effect
of contrast variations along the length of scratches, a selection window of 1 × 50
pixels is preferred. Image luminosity values are found from selecting Image in the
Menu Bar and then the digital display is found in Histogram. The selected win-
dow is scanned manually across the image to determine the minimum luminosity
value Imin . The maximum value Imax is obtained by displacing the selected window
to one side of the image. The contrast C percent is then calculated as

Imax − Imin
100.
Imax + Imin

Although a luminosity value may be quoted to five significant figures, because

of residual image clutter from variations in shape along a scratch, errors in focal
setting, residual polish defects, and dust on components, the uncertainty of mea-
surement of a high-contrast feature is probably no better than ±5%. The effects of
variation in contrast along the length of a scratch and aliasing, if present, should be
dealt with by quoting the maximum contrast measured.
If the scratch is curved or a dig is being considered, pixels worthy of selection
and measurement can be identified using the Magic Wand facility with the image
luminosity tolerance set at, say, 10. The maximum and minimum luminosity val-
ues to be recorded from the background and at the imperfection should be those
displaying peak population levels in the selected Histogram display.
The field lens is needed in this configuration because in most low-cost digi-
tal cameras the lens cannot be removed. The camera used here was an Olympus
Camedia C-860L with 1.3 M pixels. The disadvantage of using a field lens is that
it can introduce field curvature and image distortion that can only be avoided by
taking measurements at the center if the field. At the time of writing, a low-cost
digital camera with removable lens could not be found.
In order to convert the measured contrast of an imperfection image to LEW
or SED units, the instrument requires calibration using opaque lines and dots of
known dimensions. An appropriate range of line widths is 1 to 16 µm. A graticule
bearing lines and spots according to the logarithmic series in ISO 10110-7:1996,
shown in Fig. 6.3, is commercially available. The instrument calibration indicated
in Fig. 6.10 shows a plot of image contrast against line width using log/log scales
for the system described in Fig. 6.5. The vertical bars represent an estimate of
the uncertainty of measurement of image contrast due principally to uncertainty in
selecting the best image focus.
Measurement of Imperfections by Obscuration 83

Figure 6.10 Image contrast/line width using log/log scales.

6.8.2 Results and discussion

In order to explore its range of operation and sensitivity, and also to illustrate its
application with reference to existing and widely used reference standards, the in-
strument was used to measure the severity in LEW units of two samples of plastic
MIL scratch and dig paddle available commercially. Table 6.1 indicates the level of
agreement obtained when two samples are measured in transmission and reflection.
The contrast of the MIL 10 scratches in transmission were judged to be less
than 1% and so their LEW values are not recorded. The approximate geometrical
widths of the MIL scratches in micrometers are also shown here for comparison.
An imperfection to be measured was located on the stage of the instrument, po-
sitioned in the center of the field, and carefully focused before operation of the
shutter. After downloading into the PC, the image contrast was calculated and
recorded. The logarithm of this number was then marked on the ordinate of the
calibration curve and the LEW magnitude of the imperfection determined as the
antilog value of its abscissa value. The dotted lines shown arise from the average
values of the two samples of MIL 10, 20, 40, and 60 reference scratches, shown in
Table 6.1, measured in transmission (T) and reflection (R). The downward pointing
arrows indicate the logarithm of their LEW values.
It can be seen that the LEW values of imperfections in reflection are approach-
ing the contrast levels of fully developed scratches. The same scratches, seen in
transmitted light, are of lower contrast and are partially developed. Imperfections
greater than 10 µm in width are usually fully developed and can be measured
in terms of their geometrical width. The possibility might exist, however, that a
smooth imperfection, because of its particular depth, could perhaps be described
as “over developed” if, due to interference effects, it appears to remove more radi-
ation than a fully developed scratch of the same width.
84 Chapter 6

Table 6.1 Line-equivalent widths of two samples of MIL scratch and dig paddle mea-
sured in transmission and reflection.
MIL
MIL width Sample 1 (LEW) Sample 2 (LEW)
grade (µm) Transmission Reflection Transmission Reflection
10 2.5 – 1.7 – 1.2
20 4.0 0.5 3.2 1.1 3.6
40 6.3 2.4 5.4 2.0 6.2
60 10 5.2 10.0 5.8 10.0
80 16 10.0 13.2 12.6 11.7

The availability of this simple low-cost objective method for measuring imper-
fections that can be calibrated, using opaque lines and spots of known size, should
enable workers to undertake tests to determine acceptable imperfection tolerances
for different applications. The method should also find use in better understanding
the mechanism by which even small imperfections can trigger laser damage.
Recent experience in selecting a digital camera for this type of measurement
has demonstrated the need for simple low-cost methods for classifying the relative
performance of different models. Appendix 1 defines new metrics for this purpose
and describes how they can be applied.

6.9 Chapter Conclusions

The quality control of optical components for imperfections should involve visual
inspection to identify those with borderline severity for subsequent measurement.
This requires instrumentation with sensitivity better than the eye, and measure-
ments traceable to standards. A microscope image comparator employing a TV
display can be used to carry out real-time measurements of imperfection severity
in terms of the LEW or SED metrics. A low-cost alternative employing a digital
camera and computer software can perform in a similar way. This technology has
now been embodied in an ISO standard.4

References
1. L. R. Baker, “Standard for surface damage,” Opt. Eng., Vol. 31, pp. 1685–1689
(1992).
2. L. R. Baker, “Microscope image comparator,” Opt. Acta, Vol. 31, pp. 611–614
(1984).
3. L. R. Baker, “Surface damage metrology: Precision at low cost,” Proc. of SPIE,
Vol. 4779, pp. 41–51 (2002).
4. ISO 14997:2003, Optics and optical instruments: Test methods for surface im-
perfections of optical components.
Chapter 7
Surface Imperfection Quality
Control
7.1 Introduction
Although it is now generally accepted that all surfaces may be to some extent
less than perfect, there is still little objective information available on the relation
between the severity of a particular imperfection and its impact on the quality of
the system in which it occurs. It is to be hoped, now that objective methods of
measuring imperfections have been developed, that the necessary research will take
place to determine imperfection tolerances appropriate to particular applications.
Meanwhile, the data given below provide some insight into the current practice of
component design. This chapter is concerned with the quality control of optical
components.

7.2 Survey of Tolerances

A survey1 of 30 organizations across the world, almost equally divided between
manufacturers of optical systems and research-and-development groups, was un-
dertaken in 1993 in an attempt to determine the mean levels of imperfection and
roughness tolerances in use at that time. A measure of the importance attached to
this subject can be judged from the fact that 88% of those responding indicated
they would be in favor of seeing the data published. A few organizations (12%)
expressed concern that, due to the complexity of the subject and its contentious
history, some readers might mistakenly regard the table containing the data as a
standard rather than simply a guideline to quality. Readers were invited to report
to the author the results of studies to update the values quoted, but apart from one
observation that there could be no equivalence between different approaches, ad-
ditional data has not so far been forthcoming.

7.2.1 British Standard BS4301 (1991)2

Tolerance bands for scratches of four levels of severity are given in Table 7.1. The
severity parameter, or metric, is the contrast or LEW of the scratch viewed un-
85
86 Chapter 7

Table 7.1 Tolerance bands for four levels of scratch severity.

Reference letter of LEW greater than LEW not greater than
maximum LEW (µm) (µm)
A – 1.0
B 1.0 2.5
C 2.5 6.3
D 6.3 16.0

der bright-field conditions when compared with a series of opaque lines of known
width. Both are viewed in transmitted light. Scratches on mirrors viewed in reflec-
tion are compared with transparent slits of a known width on a reflecting substrate.

7.2.2 American Standard MIL-O-13830A:1963

Table 7.2 shows the approximate relations3 between MIL values, their contrast
under bright-field conditions using the MIC, and their LEW values.

Table 7.2 Approximate relations of MIL values, contrast,

and LEW.
MIL 10 20 40 60
Visibility 0.05 0.10 0.20 0.40
LEW 0.25 0.63 1.60 4.00

7.2.3 German Standard DIN 3140: Part 7, 1978

Scratches are quantified in this standard in terms of their area in comparison with
a graticule having lines of various widths and lengths. No guidelines are provided
on how width should be measured. In practice, a microscope could be used when
a scratch scatters most of the incident light. A microscope image comparator is
used for partially developed imperfections. The German standard is equivalent to
the British standard when the obscuration is measured radiometrically

7.2.4 French standard

The bright image of a scratch, seen under dark-field conditions, can be reduced
in visibility to zero by directing a portion of the light into the measurement field.
The amount of light needed to render an imperfection just visible is a measure
of its severity. Different experimental arrangements4 are needed for quantifying
imperfections in transmission and reflection.
Five levels of imperfection visibility for surfaces viewed in transmission, and a
similar number for those viewed in reflection, are defined. These are designated T1
for the highest quality down to T5 for transmission optics and R1 to R5 for mir-
rors. Although this has yet to be established, correspondence with MIL 10 for sur-
faces of the highest quality may perhaps be assumed, in which case the relations in
Table 7.3 could apply.
Surface Imperfection Quality Control 87

Table 7.3 Five levels of imperfection visibility for surfaces viewed in transmission
and reflection.
Grade T1 or R1 T2 or R2 T3 or R3 T4 or R4 T5 or R5
MIL 10 20 40 60 80

It should be repeated here that the inevitability of the uncertainty of subjective

assessments of imperfections means that close correspondence between different
national standards, applied to the severity of a particular imperfection, should not
be expected. This fact again underlines the need for objective methods of measur-
ing imperfections with traceability to national standards.

7.3 Acceptable Thresholds for Scratches and Roughness

The availability of means for measurement of LEW and SED enables optical de-
signers to quote imperfection tolerances in objective terms. Table 7.4 is an attempt
to provide approximately equivalent values based on the data provided by the above
survey. The precise levels entered on a particular drawing, to be agreed upon with

Table 7.4 Imperfection thresholds and approximate equivalent values based on dif-
ferent national standards, (a) To apply to UV systems, move one level down from vis-
ible optics, (b) The lateral cutoff length must be stated. For visible optics, use 15 µm
when concerned with scattering > 2.4 deg and 800 µm for scattering > 1 mrad.
Description (a) British American Gennan French Surface (b)
c: cosmetic grade or MIL value width µm grade roughness
f : functional LEW RMS nm
1. Camera lens
1.1 Front c B 40 1.60 T3 2
1.2 Inside c D 80 10.00 T5 2
1.3 Near focus f A 20 0.63 T2 2
2. Projection lens
2.1 Front c C 60 4.00 T4 2
2.2 Inside c D 80 10.00 T5 2
3. Process lens
3.1 Front c C 60 4.00 T4 2
3.2 Inside c D 80 10.00 T5 2
4. Copier lens
4.1 Front c C 60 4.00 T4 2
4.2 Inside c C 60 4.00 T4 2
5. A-focal systems
5.1. Objective c C 60 4.00 T4 2
5.2. Prism c B 40 1.60 T3 2
6. Microscope <A, 0.25 10 0.25 T1 1
OG f
88 Chapter 7

Table 7.4 (Continued).

7. Eye lens c C 60 4.00 T4 2
8. Field lens f <A, 0.25 10 0.25 T1 2
9. Graticule f <A, 0.25 <10 0.25 T1 2
10. Grating f <A, 0.25 <10 0.25 T1 1
11. Mirror
11.1 Laser f <A, 0.25 <10 0.25 R1 1
11.2 Other c A 20 0.63 R2 2
11.3 In cavity f 0.10 5 0.10 <R1 0.5
12. IR Optics 3–5/8–12 3–5/8–12 3–5/8–12 3–5/8–12 3–5/8–12
µm µm µm µm µm
12.1 Front c B-C 40–60 1.60–4.00 R3 –R4 3–5
12.2 Inside f C-D 60–80 4.00–10.00 R4 –R5 3–5
12.3 Near focus f B-C 40–60 1.60–4.00 R4 3–5
13. Laser lenses
13.1 Low power c A 20 0.63 T2 2
13.2 High power f <A, 0.25 <10 0.25 T1 1
14. Parallel plates
14.1 Beam C 60 4.00 T4 2
splitters c
14.2 Filters f C 60 4.00 T4 2
14.3 Cells f B 40 1.60 T3 2
14.4 Holographic f A 20 0.63 T2 2
14.5 HUD f A 20 0.63 T2 2
14.6 Window c B 40 1.60 T3 2
14.7 Polarizer f A 20 0.63 T2 2
15. Substrates
15.1 Photon <A, 0.25 10 0.25 T1 2
detector f
15.2 Wafer f 0.10 5 0.10 <R1 0.5
16. Fibre optics f <A, 0.25 10 0.25 R1 1
17. Spectacle lens
17.1 Glass c A 20 0.63 T2 2
17.2 Plastics c A 20 0.63 T2 2
17.3 Implant f B 40 1.60 T3 2
18. Condensers c D 80 10.00 T5 5
19. Magnifiers c C 60 4.00 T4 2
20. Protective D 80 10.00 T5 2
glasses c
Surface Imperfection Quality Control 89

the customer, may be higher or lower depending on component size and location
of the imperfection within the total system.
The table excludes information on the lengths or number of imperfections, be-
cause disputes relating to these parameters are less likely. Digs are also excluded
for the same reason. The RMS surface roughness tolerance is included because this
parameter is of increasing interest.
Table 7.4 lists, in the left-hand column, a variety of different optical systems
and components employing optical surfaces needing to be toleranced in terms of
their sensitivity to imperfections and roughness. The four central columns provide
suggested surface imperfection thresholds according to standards employed in the
United Kingdom, United States, Germany, and France. Reference should be made
to the standards published in these countries when applying this table, always bear-
ing in mind the need to obtain agreement with the customer on matters relating to
quality.
The right-hand column lists RMS roughness values5,6 providing an indica-
tion of the quality of polish likely to be necessary for the different applications
listed. For components used in the infrared optics appearing in item 12, two thresh-
olds are given relating separately to the 3–5 and the 8–12 µm bands of the spec-
trum.

7.4 Inspection and Measurement Flow Diagram

The procedure used for the quality control of components regarding surface imper-
fections will vary from firm to firm. The flow diagram2 shown in Fig. 7.1 provides
an illustration of a possible arrangement.
The diagram indicates that all production, whether to be examined in trans-
mitted (T) or reflected (R) light at level 1, enters an inspection station at level 2
configured according to whether the surface is to be examined in transmitted (T) or
reflected (R) light. Components without imperfections are sent to the accept box
at level 8, and those unacceptable according to some subjective criteria, such as
that employed in Method II of ISO 10110-7 (Ref. 4) agreed upon with the cus-
tomer, are sent to the reject box at the same level in the diagram. Components with
imperfections of doubtful severity pass to level 3. The locations of imperfections
with respect to edge markers are noted and the components are classified accord-
ing to the drawing indications listed in ISO 10110-7 (Ref. 4) as general, coatings,
long scratches, and edge chips, and then passed down for measurement, in either
transmission (T) or reflection (R) mode, in level 4.
If an imperfection has its smaller dimension greater than 10 µm, its grade can
usually be measured by comparison with lines on a graticule. Smaller imperfec-
tions require the use of a microscope image comparator, operating in either trans-
mission (T) or reflection (R) mode, to determine grade levels. Imperfections below
a threshold are ignored, as noted at level 5, unless the drawing specifies otherwise,
and concentrations as indicated lead to rejection as noted at level 6. A final deci-
sion whether to accept a component with imperfections of doubtful severity takes
place at level 8.
90 Chapter 7

Figure 7.1 Imperfection quality control.

7.5 Chapter Conclusions

Objective methods of measuring imperfections are now available for use by manu-
facturers of optical components when justified by cost implications. Conventional
manufacturing processes can accommodate the procedures involved. Experience in
using the new metrics of LEW and SED should lead to a better understanding of
Surface Imperfection Quality Control 91

the influence of imperfection severity on the cost and quality of optical components
and systems.

References
1. L. R. Baker, “Thresholds for surface imperfections,” Opt. Eng., Vol. 33,
pp. 2800–2802 (1994).
2. L. R. Baker, “Standard for surface damage,” Opt. Eng., Vol. 31, pp. 1685–1689
(1992).
3. L. R. Baker and J. Salerno, “In pursuit of perfection,” Photonics Spectra, Nov.,
pp. 151–156 (1992).
4. ISO 10110-7:1996, Optics and optical instruments: Preparation of drawings for
optical elements and systems. Part 7. Surface imperfection tolerances.
5. ISO 10110-8:1997, Optics and optical instruments: Preparation of drawings for
optical elements and systems. Part 8. Surface texture.
6. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scatter-
ing, Optical Society of America, Washington, DC (1997).
Chapter 8
Far-Field Nanoscopy
8.1 Introduction
Although the microscope image comparator was designed for quantifying the
severity of surface imperfections, its ability to measure luminosity values of any
image and compare values with images of similar objects suggests there may be
other uses for this instrument. This generic technology has for convenience been
termed far-field nanoscopy (FFN). Previous use1 of the term “nanoscopy” appears
to have been restricted to optical near-field probe measurements and could there-
fore be termed near-field nanoscopy (NFN).
This chapter illustrates the potential of FFN for replacing subjective assess-
ments of surface quality by objective assessments of surface quality. It is aimed
at filling the gap between visual assessment and that using expensive, slow pro-
filometers. It indicates the potential of the technique for enabling the measurement
of texture, surface height variations, imperfections, and contamination. Structures
of nanometer dimensions, even when buried within a system already assembled,
can be measured.
Since FFN often employs a microscope, albeit of low aperture, a wide range
of complex, adaptive, and intelligent spatial frequency filtering techniques, such
as dark-ground, partially coherent, and phase contrast, can be employed to com-
pare surfaces with similar topographical structures for the purpose, for example,
of quality control of surface finish. A wide range of specular surfaces can be mea-
sured.

8.2 Comparison between Subjective and Objective

Measurements of Imperfections
Although over the years there has been much criticism of subjective methods for
quantifying the severity of surface imperfections, published accounts of the degree
of agreement possible have proved hard to find. Figure 8.1 illustrates the spread of
results that can be expected based on tests2,3 carried out in 1985.
Comparison trials were carried out by eight experienced inspectors from the
UK MOD. The tests employed 18 samples carrying typical imperfections. Each
sample was inspected and graded according to U.S. standard MIL-0-13830A. The
93
94 Chapter 8

Figure 8.1 Comparison between subjective and objective assessments of imperfec-

tion severity.

heavy line shows there was full agreement on the severity of imperfections on
samples No. 2 and 5, but there was an unacceptable level of agreement on more
than 80% of all of the samples.
The horizontal line at the 100% level records the level of agreement expected
and achieved by the use of the microscope image comparator employing objective
means of measurement. It cannot be assumed from these results, however, that any
subjective assessment obtained with a particular imperfection will always agree
with the measured value. This is expected due to the lack of precise control of the
method of illumination and viewing of the sample, and variations of the observer’s
visual acuity.

8.3 Relative Contrast of Standard Scratches

A number of national standards organizations, and some companies, have designed
their own comparator standards for assessing surface imperfections. The severity
of a particular imperfection is determined by visual comparison under defined il-
Far-Field Nanoscopy 95

lumination and viewing conditions with a reference graticule. In this way, optical
components can be classified into different quality grades that are related to the
requirements of individual systems. The standard or reference scratches are fabri-
cated using a variety of processes including ruling, etching, thin film deposition,
and replication. Each approach has its own advantages and disadvantages in terms
of range of visibility, cost, and change of visibility under different viewing condi-
tions. A description of various approaches is given in Sec. 5.14 and also in Ref. 4.
The relative visibilities of a variety of different standard scratches available at
the time have been measured repeatedly using the microscope image comparator
described in Sec. 6.7. The results are presented graphically in Fig. 8.2.

Figure 8.2 Relative contrast of standard scratches.

The curve is a plot of the angle θ of Z2 , of each of the DIN standard lines
producing an image contrast equal to that of R, against the logarithm of their width
for lines of width varying from 2.5–40 µm. The short vertical lines on the curve
indicate the limit of precision with which a setting of the instrument dial could be
made. The dotted part of the curve is an extrapolation based on the assumption that
the visibility of the very narrow lines would decrease linearly with width down to
a width of 0.00025 mm.
A variety of other reference imperfections available at the time, including
British standard scratches, the Kodak dig/scratch paddle, Rank Taylor Hobson stan-
dard scratches, and MIL standard scratches were also measured and the data added
to the graph. The vertical locations of these various standards on the graph have
been separated for convenience of presentation. The angle setting θ for each im-
perfection was measured and its position along the abscissa set to intersect the DIN
curve already plotted. The value of the abscissa can then be taken as the logarithm
of the LEW of that particular imperfection.
96 Chapter 8

It can be seen from the figure that the MIL 60, RTH 2, Kodak 60, and DIN 0025
have approximately the same LEW value of 2.5 µm and therefore all produce an
image of contrast relative to the instrument internal standard of approximately 0.3.
These results show that any in-house scratch reference standards can be calibrated
in terms of their LEW.

8.4 Measurement of Imperfections and Contamination

in Assemblies
The inspection and measurement of imperfections on a single optical component
are difficult enough using conventional dark-field techniques, but the task is prac-
tically impossible when the component is within an assembled system. Although
imperfections and contamination may be visible when a system is held up to the
light, the task of measurement through other optical components with power is
problematic.
As we have seen, the microscope image comparator employs a low-aperture
imaging system that may be compared optically with the eye when examining a
system for imperfections in transmitted light. In order to compensate for the ef-
fect of additional lens power between the imaging lens and the imperfection being
measured, the calibration lines need to experience the same power conditions.
Figure 8.3 shows a MIC configuration suitable for examining assembled optical
systems for imperfections and contamination.

Figure 8.3 MIC for use with assemblies.

Far-Field Nanoscopy 97

Light from a tungsten lamp S is focused by the condenser lens L1 onto the
pinhole P1 . After passage through the polarizer Z1 , the light from P1 that is at the
focus of L2 passes as a parallel beam into the polarizing beamsplitter B. The beam
passing straight through B is brought to a focus by a reference compound lens,
shown here as a triplet L3 , on to the center of curvature of a retroreflective mir-
ror M1 after passing through the quarter-wave plate Q1 . Upon reflection at M1 and
retransmission through Q1 and L3 , the beam from this reference channel can now
be reflected by B and brought to a focus by the projection lens L4 , after passage
through the analyzer Z2 onto the plane of a spatial frequency filter F. The TV cam-
era is placed further back to receive an image of whichever of the surfaces of L3
is selected for study on the TV as a result of movement of L3 , Q1 , and M1 rigidly
linked together, in the direction shown by the arrows.
The parallel beam from L2 , reflected by B, is, in this test channel, brought to
a focus by L5 through the quarter-wave plate Q2 onto the center of curvature of
the retroreflective mirror M2 . As in the reference channel, the returned beam is
brought into focus on F placed at the back focal plane of L4 and, by the combined
movement of L5 , Q2 , and M2 , each of the surfaces of the test lens L5 can be im-
aged in turn onto the TV. In this way, corresponding surfaces of L3 and L5 can be
brought simultaneously into focus for the purpose of comparing the contrast of an
imperfection with a reference line.
Rotation of Z2 by angle θ has the effect of exposing radiation from the two
channels alternatively to the TV in accordance with the Law of Malus—namely
cos2 θ from one channel and sin2 θ from the other that is polarized at right angles
to it. If the intensity at every point in the background field at the TV is the same
for both channels, the sum of these two signals will be unity and there will be
no residual modulation at the corresponding point on the TV monitor as Z2 is
rotated. If there is modulation of this background field, it can be reduced to zero
by rotation of Z1 . In this way, the background light intensity will be constant and
the imperfection image intensity will be modulated by an amount proportional to
the amount of light it has removed from the beam. In the same way, the image of
a speck of dust will be modulated with amplitude proportional to the amount of
light it has removed from the beam. The system can be calibrated using absorbing
artifacts of known size deposited on the surfaces of L3 . They can take the form of
lines and spots as previously discussed. Their intensity will also be modulated but
in antiphase with an imperfection image.
The reference lens L3 should be of the same design as the test lens L5 to provide
parfocality and automatic compensation for scatter losses, ghosts, and intersurface
reflections. The relative magnitude of any imperfection or dust particle compared
with the radiometric obscuration of the reference graticule can be found by record-
ing the angular rotation θ of Z2 needed to equalize the contrast of the image of the
imperfection and the reference graticule seen side-by-side on the TV monitor. If
the amounts of light removed by the imperfection and graticule are I and G, then it
has been shown in Sec. 6.7.2 that I/G = tan2 θ, assuming the backgrounds of the
two channels have been equalized by an appropriate rotation of Z1 . It is possible in
98 Chapter 8

this way to measure a particle or imperfection severity on L5 in comparison with

the reference lens/graticule artifacts.
The class of image seen on the TV monitor depends on the design and size of
the spatial frequency filter F. A bright-field image is produced when F takes the
form of a pinhole, as with the conventional MIC, when the image is dark against
a bright field. Its diameter is chosen so that the largest imperfection of interest just
removes all the light from the image.
A dark field is obtained when P1 has an opaque stop at its center, providing
annular oblique illumination, of size such that its image at F is just larger than the
diameter of F. In this way, only light scattered by the imperfection will be collected
by F and projected onto the image plane, where the imperfection or dust particle
will be seen as a bright line or spot.
Some forms of contamination, such as stains or those influencing the nature
of surface coatings, absorb and/or scatter little light but do cause a phase per-
turbation on the transmitted wavefront that could influence the performance of a
high-quality optical system. Such phase effects can be visualized by the use of a
complex filter at F, as in the phase contrast microscope.5 Here the image of the
pinhole P1 is formed on an absorbing disk just larger than its image, but which also
shifts the phase of the direct light by a quarter of the mean wavelength of radiation
used. This light then expands, overlaps, and interferes with the diffracted light of
less intensity, which passes around this obstruction, creating an image on the TV
where the intensity is proportional to the height of the phase perturbation caused
by the contamination. Only small-size discontinuous perturbations, as may arise

Figure 8.4 Surface damage and contamination in lens f/1.9, 32 mm focal length.
Far-Field Nanoscopy 99

from contamination, are seen. Once again, defects can be quantified by the use
of this two-beam imaging null comparator, assuming reference defects of known
characteristics are available for use in the reference channel. Sensitivity to changes
in surface height profiles of ∼0.1 nm has been claimed6 for phase contrast tech-
niques. The filter F can vary in shape, becoming circular, annular, or cruciform to
match the source shape so as to optimize sensitivity and reduce the effects of dif-
fraction. Some experimentation is required, depending on the type of imperfection
and contamination.
A four-element compound lens was examined with the system described above,
operating in bright-field mode, with results shown in Fig. 8.4. Each element was
brought into focus, in turn, on the TV. The first element has scratches and dust, the
second a great deal more dust, the third a filament of lint, and the fourth shows dust
and cleaning sleeks.
The effects of varying the design of F on the performance of the MIC can be
seen in Fig. 8.5, which shows the first element of the above lens seen with bright-
field, dark-field, and phase-contrast viewing.

Figure 8.5 Use of different F configurations.

100 Chapter 8

In Fig. 8.5(a), the two reference graticule lines have been introduced by ro-
tating Z2 until their contrast matches that of the more clearly seen scratch. The
tangent squared of the setting of Z2 is then the ratio of the amount of light removed
by that particular scratch to that removed by the graticule lines. The dark-field
view in Fig. 8.5(b) shows the scratch as a bright line against a dark background.
Although sensitivity to imperfections has increased, signal levels are dependent on
light intensity as is the case when operating in the inspection mode. In the phase
contrast view of Fig. 8.5(c), however, residual fine background structure due to a
combination of contamination, thin film defects, and surface roughness becomes
visible. Although the field of view of this system was limited to 3 mm, a beam
expander might be used to examine larger areas.

8.5 Measurement of Imperfections in Coatings

Coating imperfections can be functional or purely cosmetic. In both cases, nor-
mal practice in the optical industry is to use visual inspection to compare indi-
vidual imperfections with standard scratches as described in various national and
other standards. The test is subjective and often leads to dispute. Functional im-
perfections are sometimes treated more quantitatively than those rated as cosmetic.
Techniques include measurement of scattering, absorption, or reflectance, ellip-
sometry, photo-acoustic spectroscopy, electron microscopy, Auger spectroscopy,
bright- and dark-field microscopy, Nomarski microscopy, FECO fringes, interfer-
ometry, and knife-edge testing.7 Much of the work has been prompted by the need
to correlate coating imperfections and laser damage threshold,8 and several differ-
ent techniques have been developed.9,10 Very few of these techniques are suitable
for the on-line inspection of coating imperfections, especially when they are merely
cosmetic. The MIC is offered as a quick, accurate, low-cost technique to replace
existing subjective, qualitative visual assessment.
A variety of samples have been examined11 with the MIC, operating with a
field of view of 1.5 mm. Figure 8.6 shows four witness samples taken from routine
coating runs.
The traces below the photographs represent the intensities across the image
at the region of the bright cursor line—the two vertical dark lines are images of
the graticule lines in the reference channel. The top left image presents a substrate
scratch as a dark line and the orange-peel trace across the field indicates evidence of
low-angle scatter in this sample. A bright line in the center of the top right image
is caused by a sleek in a multilayer antireflection coating. The bright dots on a
darker background in the bottom left image result from pinholes in an antireflection
film. The final image shows the presence of material spits of 0.05 mm diameter in
a dichroic filter as dark spots absorbing radiation. The practical significance on
appearance or function of the measured severity of these various imperfections
needs to be established by further studies.
Far-Field Nanoscopy 101

Figure 8.6 Coating imperfections.

8.6 Use of MIC to Measure Surface Texture

Although MIC operates with a small numerical aperture and so provides low spa-
tial resolution, it is nevertheless capable of high resolution in the vertical direction.
The recent development and application in industry of new techniques for surface
generation, such as single-point diamond turning, laser beam ablation, ion beam
etching, and high-speed flexible lap polishing, is resulting in the availability of
components with unusual surface morphologies. Whereas conventional polishing
creates a surface with random height variations, new finishing techniques are often
characterized by an element of periodicity or patterning in the surfaces they pro-
duce. This fact, combined with the more widespread use of objective means such as
the optical transfer function and glare spread function to assess optical system per-
formance, is leading to a requirement for setting objective thresholds of acceptance
relating to all aspects of the texture of these high-technology surfaces.
102 Chapter 8

As we have seen in Chapters 2–4, the total topography of a surface includes the
macroscopic shape or form, usually measured by an interferometer, and the mi-
croscopic surface irregularities, known as finish, left by the final finishing process.
The localized elements of finish, involving digs and scratches, can be measured
with the MIC, but there also appears to be potential for using the same technique
for quantifying texture distributed over the whole surface as roughness and wavi-
ness. If this proves to be the case, we should then have the capability to assess all
aspects of finish with one instrument.
Although all surfaces have texture, the task of measurement and control only
becomes significant when the variability of the processes involved exceeds certain
limits. The degree of polish of an optical component has a direct effect on image
quality and also on component cost. If polished for too short a time, the lens cost
may be low but so also will be the quality of image produced due to light scatter.
If given a longer polishing time, the cost will increase, but image quality may not
increase in proportion. Ideally, one would like to terminate polishing at the instant
surface roughness is just tolerable. More polishing not only increases costs but can
also lead to the probability of surface degradation arising from accidental damage
by the polishing process itself.
In some engineering applications a finite level of texture is needed, as with
bearings, to ensure oil retention, but not so much as to lead to an excessive wear rate
and friction. In others, texture has a direct influence on the aesthetic appearance of
a surface. These considerations lead to the need for low-cost means for the on-line
measurement of surface texture, providing quality control based on agreed texture
tolerance levels.
The potential for using MIC for this purpose has been explored with results
shown in Figs. 8.7–8.10.
In the first example,12 MIC was used to examine the effect of diamond
tool wear on surface quality. Four aluminum-alloy samples were machined on a
Pneumo Precision diamond-turning lathe. Table 8.1 records the Ra values of spec-
imens a, b, c, and d obtained with successive cuts.
The roughness levels, as measured with a stylus, increase as the tool wears.
Surface waviness is concealed by the roughness present but appears in the MIC
displays shown in Fig. 8.7. Surface amplitude variations down to 1 nm can evi-
dently be detected. The periodicity of the surface waviness suggests it is due to
machine and tool vibration rather than the tool feed dynamics.
The power of the parametric method for measuring surface finish through the
use of the MIC, with its potential for automation, has been further demonstrated
with displays shown in Figs. 8.8–8.10.
Display A of Fig. 8.8 shows a scratchlike defect invisible to all but the most
experienced inspectors of optical components, while C appears, by eye, as a per-
fectly acceptable mirror. The white line across the image is the track of a video
line-intensity profile analyzer, the output of which is shown in the dark space at
the top of each image. Display D presents an image intensity profile related to the
waviness of the turned surface.
Far-Field Nanoscopy 103

Figure 8.7 Effect of tool wear on surface texture.12

Table 8.1 Ra values of diamond-turned samples.

Sample Ra (µm) Measurement conditions
a 0.006 0.5 mm sample length
b 0.006 normal to turning marks
c 0.007
d 0.010

In order to assess the possible value of the technique for inspecting high-quality
engineering surfaces,13 measurements of relative specular reflectance were carried
out on a specimen of a Rubert Microsurf Gauge. Figure 8.9 shows the appearance
of four surfaces and the analogue traces obtained with the line-profile analyzer.
The angular settings of the polarizer Z1 , required to produce no screen flicker
when the analyzer Z2 was rotated continuously, are plotted in Fig. 8.10 against
the stylus-measured Ra values taken at right angles to the lay of the surface. This
process depends on the relation between roughness and specular reflectance dis-
cussed in Sec. 3.4.2.
104 Chapter 8

Figure 8.8 MIC displays of finish showing scratches, grating lines and turning marks.

No attempt has been made to relate these measured values to those that might
be expected from theory, but there is a level of sensitivity and dynamic range,
which suggests that this objective comparative approach might have application
in routine noncontacting areal assessment of finish as an alternative to assessment
just by eye. We have seen that when a localized surface height approaches a quarter
of a wavelength, its image luminosity becomes nonlinear and so this method will
be restricted to surface heights significantly less than a wavelength. A simple test
involves reflecting a laser beam, using a laser pointer, from the surface in question.
Surface imaging can be used, providing the diffraction spread shows a clear peak
in the specular direction. If there is no peak, the surface is too rough to be measured
in this way.
A tilt of a small element of a surface, imaged in reflection by spatial frequency
filtering, results in a reduction of intensity in its image proportional to the tilt ex-
Far-Field Nanoscopy 105

Figure 8.9 MIC displays of 4 levels of surface texture.

Figure 8.10 Relation between polarizer setting and Ra value.

106 Chapter 8

perienced. In this way, the direct image of a slightly rough surface creates a distri-
bution of intensity in its image proportional to surface slope. Integration of these
gradients would give surface heights but, assuming surfaces with a periodic charac-
teristic, the step of integration may be unnecessary for comparative measurements.
Using the digital version of MIC described in Sec. 6.8, images of three Rubert
Gauge (RG) specimens with nominal Ra values of 12.5 nm, 25 nm, and 50 nm were
transferred to Photoshop® and several windows of 200 × 10 pixels were selected,
in turn, across the lay of the pattern chosen for measurement. Figure 8.11 shows
these images.
Using Image and Histogram from the Menu bar, the average standard devi-
ations (SD) of the selected image luminosities were recorded, together with the
mean luminosity values for each image. The standard deviations were divided by
the mean values and the results normalized to the 12.5 nm specimen. To within
the limits of sampling error, the surface roughness values for the remaining spec-
imens, as indicated in Fig. 8.11, agreed quite well with the nominal values. This
result suggests that there could be potential for the use of a digital camera, as de-
scribed in Sec. 6.8, for the comparative noncontact areal measurement of surface
texture over the range of 1–50 nm, using visible light. Measurement to a useful
uncertainty should be possible, providing calibrated reference pieces of approxi-
mately the same character are available.
Further confirmation of the potential of this approach for measuring a wider
range of specular surfaces was sought by tests on aluminum foil, a cover glass
used in projection slides, and the surface of glossy computer printer paper. The
digital camera version of the MIC was first calibrated by plotting the SD values of
three reference surfaces with known roughness values. The straight line shown in
Fig. 8.12 is a best fit for the three reference specimens.
The roughness values for each of the three test specimens mentioned above
(shown in Fig. 8.13), together with three reference specimens, were obtained from
their measured SD values. These were determined from the histogram display us-
ing a selected window of 100 × 100 pixels and the calibration graph. The SD values
were normalized with respect to mean luminosity values.

Figure 8.11 Digital images of three Rubert Gauge specimens.

Far-Field Nanoscopy 107

Figure 8.12 Image SD calibration.

Figure 8.13 Measurements of surface roughness.

The circular images shown at the bottom of the figure cover a field of 0.9 mm,
selected from a total field of 6.4 mm. One significant advantage of using the
nanoscope rather than an interferometer for this purpose, apart from cost, is that
specimens do not need to be particularly flat, providing the reflected image is sig-
nificantly smaller than the size of the spatial frequency filter. Information in the
image depends mostly on attenuation of the higher spatial frequencies.

8.7 Use of MIC to Examine Phase Objects

A homogeneous surface with small height variations, when viewed under vertical
illumination by a conventional microscope, provides an image of substantially uni-
form distribution of intensity. We have seen, however, that surface slopes can be
108 Chapter 8

imaged with the aid of spatial frequency filtering. In order to create an image where
the intensity distribution is directly related to surface height variations, a complex
spatial frequency filter is required. This partially absorbs the low spatial frequen-
cies, but also provides a phase shift of a quarter of a wavelength retardation of the
background radiation with respect to the wavefront carrying information about the
object structure. Should a pixel element in the surface also be retarded due to a
local depression, that area will appear brighter than its surroundings due to a re-
duction in the relative phase difference. If, however, the pixel advances in phase,
then the relative phase difference increases and so it will appear darker in the im-
age. It is thus possible, with phase-contrast viewing, to distinguish between hills
and valleys with reference to the surrounding surface and to provide a measure of
surface-height variations.
This type of complex spatial frequency filtering, as first discovered by
Zernike,14 has been employed in the transmitting phase-contrast microscope for
many years. Much work has been published on the theory of spatial frequency fil-
tering to visualize phase structures. In the experiments described below, the MIC
was used as a two-channel phase-imaging photometer.15 A schematic representa-
tion of this instrument is shown in Fig. 8.14.
Light from a tungsten lamp S is focused by the condenser lens L1 through the
polarizer Z1 onto an annular aperture P1 . The light from P1 , which is at the focus
of L2 , passes as a parallel beam into the polarizing beam splitter B. The beam

Figure 8.14 MIC as an imaging photometer.

Far-Field Nanoscopy 109

passing through B is transmitted by the quarter-wave plate Q1 and the graticule G

before falling on to the retroreflective screen R.
The light returned by R illuminates G so that, after its return passage
through Q1 and B, it can be imaged by L3 after transmission by the complex
frequency filter P2 placed in the back focal plane of L3 , and the analyzer Z2 onto
the TV camera. G can take the form of a narrow opaque line on a transmitting sub-
strate the contrast of which, when imaged on the TV, can be continuously changed
by rotating Z2 . This image relates to the reference channel of the photometer.
The parallel beam from L2 , reflected by B into the test channel, is transmitted
by the quarter-wave plate Q2 and illuminates the test specimen T at normal inci-
dence. As in the reference channel, the light reflected by T can also be imaged
after a return passage through Q2 and B by L3 onto the TV camera. The plates Q1
and Q2 are rotated in turn, in a setup operation, to maximize the intensity of the
beams on the TV. The filter P2 is typically a phase-retarding ring, onto which an
image of the annular source at P1 is imaged as in the conventional phase-contrast
microscope. Since R produces scattering of light over a few degrees, the reference
channel is little affected by P2 , and so operates as a normal microscope. The TV
camera adds the incoherent images from both channels.
If the light transmissions of the test and reference channels are KT and KR ,
and the proportional change in intensity g due to G in the reference channel, and
the value d resulting from a height perturbation in the test channel, then the peak
intensities at the two images, which can for convenience be arranged to be in close
proximity on the TV monitor, will be gKR and dKT .
It follows from the law of Malus, since the beams are polarized at right angles
and for the angular setting θ of Z2 , that the intensities of the two images, both of
which are summed with the background of the other channel, will be

KT (1 − d) cos2 θ + KR sin2 θ

at the image of the test feature, and

KT cos2 θ + KR (1 − g) sin2 θ

at the image of G. If Z2 is rotated to equalize the intensities of the reference and

test features, then

KT (1 − d) cos2 θ + KR sin2 θ = KT cos2 θ + KR (1 − g) sin2 θ,

which reduces to

KT d = KR g tan2 θ.

If KT and KR , which are proportional to the background intensities, are set equal
by rotation of Z1 , then
d
= tan 2θ.
g
110 Chapter 8

To bring about this condition, Z2 is rotated continuously; but since the reflectances
of R and T will in general be different, the background intensity on the TV will be
modulated. This modulation can be reduced to zero by rotation of Z1 . The angular
setting of Z2 can now be set to the value where the peak intensity in the image of
the feature on T is equal in magnitude to that of the image of G. If features on T
of known height are available, then the scale of Z2 can be calibrated directly in
feature height. It should be noted that if the feature on T is a valley it will produce
a signal brighter than the surroundings, but if it is a hill it will produce a darker
signal in phase with the signal from the absorbing line G.
Since g is a fixed constant of the instrument, we would expect the height of
a perturbation to be directly proportional to tan2 θ. The spatial resolution of the
instrument, which depends on the numerical aperture of L3 , was 0.003 mm in the
experiments described below. The depth resolution depends on the level of pho-
ton noise coupled with observation time. Using a TV Newvicon camera and line
profile analyzer to view a single line scan from the video signal, the limit of depth
sensitivity was about 1 nm. The lowest spatial frequency depends on the angular
subtense of the width of the annulus of the filter P2 . Taking a width of 0.5 mm, this
will be 20 times less than the cutoff frequency.
Since the intensity of light at a point in two overlapping and interfering beams
varies sinusoidally with path difference, the range of linearity is restricted to the
region where the size of the phase angle is approximately equal to the angle itself.
To within 4%, this corresponds to ±30 deg and so, for the mean wavelength of
observation of 750 nm, this provides a linear range of ±60 nm.
In order to check the linearity achieved in practice, values of θ that gave rise
to TV images of G and the test feature of equal intensity were obtained using a
number of artifacts of known dimensions and optical path differences. The artifacts
used were glass disks with simulated scratches formed as rectangular troughs of
width 0.017 mm, length 1 mm, and varying depths. Table 8.2 relates their code to
the geometrical depth (GD) of the trough in a glass plate of refractive index 1.517
when viewed in reflected or transmitted light, whether the trough was empty or
filled with water retained by a cover glass, and the magnitude of the optical path
difference perturbation induced on the wavefront.

Table 8.2 Wavefront perturbations obtained using five artifacts.

Code Geometrical Medium Viewing In Wavefront
depth GB perturbation
(nm) (nm)
AT 70 air transmission 36.2
BWT 110 water transmission 20.2
XT 23 air transmission 11.9
XWT 23 water transmission 4.2
XR 23 air reflection 46
Far-Field Nanoscopy 111

Figure 8.15 MIC as imaging photometer.

When T is viewed in transmission, Q2 is placed between T and a flat mir-

ror placed below T. This mirror, not shown in Fig. 8.14, returns the light back
through Q2 and T, and into B as before.
Figure 8.15 shows a plot of tan2 θ against the optical-path difference feature
height. It demonstrates a linear characteristic to within expected limits of accuracy.
Isolated features in a surface, such as hills and valleys, can in this way be dis-
tinguished and measured by a simple two-channel phase-contrast microscope, cal-
ibrated by reference structures of known height over a range of ±60 nm.

8.8 Use of MIC in AC Mode

So far, the MIC has been operated in open-loop mode employing a digital camera,
or in null mode where the loop is closed by operator intervention to equalize the
contrast of two images seen side-by-side. It has been found possible, however, to
continuously oscillate the plane of polarization of the illuminating beam and then
rotate the plane of polarization of the analyzer to drive the AC output of a point
receiver in the final image plane to a null value. The analyzer rotation required to
reach a null value is then a precise measure of image intensity at the selected point.
Figure 8.16 shows a schematic representation of a typical MIC operating in AC
mode.
It can be seen here that a Faraday rotator F1 placed between L1 and P2 is fed
by an oscillator at 80 Hz that is also connected to a phase-sensitive rectifier as a
reference signal. This oscillation of the plane of polarization creates an AC output
from the photomultiplier PM, mounted behind a small scanning aperture A in an
112 Chapter 8

Figure 8.16 MIC operating in AC mode.

aluminum film just behind the analyzer Z2 . This test signal is also fed into the
phase-sensitive rectifier. The DC output is fed into another Faraday rotator F2
placed between P3 and the lens L4 , which also collects the light reflected back
from A. This light is reflected and focused by L5 into the TV so as to provide an
image of the area on T under study. P3 is at the front focus of L4 and A is placed at
the back focus of L4 . In this way, the lens L2 with P3 at its back focus and T at its
front focus, combined with L4 , constitute a telecentric system. This arrangement
has the advantage that it gives rise to a large depth of focus at A.
The image projected onto A is scanned by the use of a motorized stage sup-
porting T with its output fed as an x displacement on the recorder, the y deflection
of which is obtained from the current flowing through F2 needed to bring about a
null input to the phase-sensitive rectifier.
An advantage of using polarized light in this type of instrument is that by the
use of the magneto-optic or electro-optic effects, extremely small angular rota-
tions of the plane of polarization, such as ±0.05 arc-seconds, have been found
possible.16 Liquid-crystal modulators have also been used for this purpose. They
have the advantage that low operating voltages can be used.
Experiments17 were carried out to determine the sensitivity of MIC in AC
mode, in comparison with conventional stylus probes, for measuring microstruc-
tures. Figure 8.17 at the top shows a probe trace across a trough etched into a glass
surface 23 nm deep and 0.015 mm wide.
The spatial resolution of a contacting probe depends on the tip radius, which
is typically 1 µm, and its height resolution, which is of the order of 1 nm. These
Far-Field Nanoscopy 113

Figure 8.17 MIC phase-contrast image of test defect.

performances can be extended downward by a factor of 10 to suit special needs,

but with the disadvantage of increased sensitivity to environmental change.
The light intensity levels within and to either side of the phase-contrast display
of this test defect, measured in reflection, are shown at the bottom of Fig. 8.17.
The vertical sensitivity of the optical noncontact approach is similar to that of the
probe. It should be noted, however, that phase contrast images are liable to produce
edge diffraction effects, dependent on the shape of the complex filter used, which
do not relate accurately to surface profile. Edge errors may also be present with a
contacting probe due to its finite radius.
The three displays in Fig. 8.18 illustrate the sensitivity of the MIC working in
AC mode. Figure 8.18(a) shows the distribution of intensity in the phase contrast
image of a rectangular trough of width 0.017 mm and depth 23 nm.
The width of the trace of 0.015 mm is determined by the small aperture of the
imaging lens. Measurements of the ratio of signal-to-noise suggest that the method
is capable of detecting changes in height of a few tenths of a nanometer. A higher
lateral resolution could be obtained by increasing the numerical aperture of L3 .
Figure 8.18(b) shows the distribution of intensity in the image of a transparent
slit of width 0.001 mm. The peak magnitude of the signal is related to the slit
width and the amount of light transmitted. It would appear from the amplitude of
the noise measurements that a change in line width of as little as 10 nm could be
detected, even though the depth of focus of the lens used was about 1000 times
114 Chapter 8

Figure 8.18 Displays showing MIC sensitivity in AC mode.

greater. It is thought that edges might be located to a few nanometers using the
comparator principle.
Figure 8.18(c) shows the distribution of intensity across the phase-contrast im-
age of a reflecting grating where the period is 0.025 mm and the depth is 9 nm.
Measurement of the ratio of signal-to-noise level here suggested a sensitivity to
surface height changes of approximately 0.1 nm.

8.9 Use of MIC On-Machine

The inspection and measurement of the surface quality of large optical components
(diameters greater than 0.1 m) to high accuracy are very difficult using existing
methods. Inspection by eye is difficult enough, but measurement by contacting
stylus or noncontacting optical probe usually requires the test piece to be mounted
on a small, motorized platform or instrument stage rather than on-machine.
Far-Field Nanoscopy 115

The increasing need for low-scatter, large optical surfaces in astronomy and
high-power laser transmission has suggested that there might be an opportunity
to use an on-machine version of the MIC for this purpose. The advantages of this
approach18 include ability, with one instrument, to measure roughness,19 waviness,
imperfections, and contamination, as well as specular reflectance. Additional fac-
tors in favor of using MIC on-machine include its low cost, small size, and low
weight, combined with an ability to both inspect and measure parameters of inter-
est in relation to performance thresholds.
Figure 8.19 shows a version of the MIC, employing a real-time TV monitor and
manual control that could be guided over a surface of extent limited only by the
length of its cables and access to its controls. Remote adjustment of the analyzer
and polarizer could be added if necessary.

Figure 8.19 MIC on-machine.

8.10 Chapter Conclusions

The technique of far-field nanoscopy, as evidenced in the microscope image com-
parator, provides an objective technique for measuring a variety of surface metrics
and microimage luminosity using the comparator or substitution method. A wide
range of applications includes measurement of contamination and noncontact areal
microtopography. The use of homodyne detection permits nanometer sensitivity
comparable with conventional probe sensors. Modern sensors, such as those em-
ployed in the digital camera, simplify the technology and reduce costs. Care is
116 Chapter 8

needed in the choice of camera because published specifications, as indicated in

Appendix 1, may lack information on true image information content with regard
to both spatial and contrast resolution.

References
1. J. P. Fillard, Near Field Optics and Nanoscopy, World Scientific, Singapore
(1996).
2. J. A. Slater, OSTAG 821 Scratch Assessment Report 16/85, UK MOD, Royal
Arsenal, Woolwich, London (1985).
3. L. R. Baker, “Subjective versus objective methods for surface inspection,”
Proc. of SPIE, Vol. 652, pp. 106–112 (1986).
4. L. R. Baker and J. Singh, “Comparison of visibility of standard scratches,”
Proc. of SPIE, Vol. 525, pp. 64–69 (1985).
5. J. Dyson, Interferometry as a Measuring Tool, Machinery Publishing, London,
pp. 170–173 (1970).
6. D. Malacara (ed.), Optical Shop Testing, John Wiley, New York, p. 270 (1978).
7. J. M. Bennett, “Recent developments in roughness characterization,” Meas.
Sci. Technol., Vol. 3, pp. 1119–1127 (1992).
8. R. P. Freese and K. J. Teegarden, “Laser induced damage in optical materials,”
NBS Spec. Publ., Vol. 568, pp. 313–332 (1979).
9. T. N. Sonderman, J. Stokowski, and D. Walker, “Laser induced damage in
optical materials,” NBS Spec. Publ., Vol. 620, pp. 159–161 (1980).
10. R. M. Wood, Laser-induced Damage of Optical Materials, IOP Publishing,
London (2003).
11. L. R. Baker and J. D. Briers, “The measurement and specification of coating
defects,” J. Phys. D:Appl. Phys., Vol. 21, pp. 88–91 (1988).
12. L. R. Baker and J. K. Myler, “In-process measurement of surface texture,”
Proc. of SPIE, Vol. 802, pp. 150–156 (1987).
13. L. R. Baker, “Automated measurement of surface texture,” Proc. of SPIE,
Vol. 730, pp. 91–96 (1987).
14. F. Zernike, “Phasen kontrastnerfarhren bei der mikroskopischen Beobach-
tung,” Physik Z, Vol. 36 (22/23), pp. 848–851 (1935).
15. L. R. Baker, “Surface metrology by phase contrast,” Proc. of SPIE, Vol. 1266,
pp. 175–183 (1990).
16. R. J. King and K. W. Raine, “Polarimetry applied to alignment and angle mea-
surement,” Opt. Eng., Vol. 20, pp. 39–43 (1981).
17. L. R. Baker, “Polarization micro-metrology,” Proc. of SPIE, Vol. 1166,
pp. 188–197 (1989).
18. L. R. Baker, “Areal measurement of topography,” Surf. Topography, Vol. 1,
pp. 207–213 (1988).
19. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scat-
tering, Optical Society of America, Washington, DC (1997).
Chapter 9
Strip Product Inspection
9.1 Introduction
The most economical way to inspect optical components for imperfections has
been found to be by using the eye to collect the low-angle light they scatter. At-
tempts to automate this process have so far not succeeded due to considerations of
sensitivity and cost. The use of the eye, however, becomes impractical when the
surface to be inspected is large and moving rapidly, as is the case with continuous-
strip product inspection.
Optical quality, including appearance, is of great importance to manufacturers
of most strip products. Automotive glass manufacturers, particularly those making
glass for laminated windscreens, where driver viewing at steep angles of incidence
occurs, find this configuration magnifies the effect of very small surface irregu-
larities. Moreover, the need to detect a particle of diameter 0.1 mm across a strip
of float glass 4 m in width at a rate of 1000 particles per second, should they oc-
cur, presents an extremely demanding problem. Although CCD cameras have been
tried, the most satisfactory solution found so far, with capability of scanning up to
three times faster, involves the use of laser beam scanning.1–3
The range of imperfections likely to occur with float glass includes seeds, bub-
bles, stones, tin oxide, and tin, while contaminants include, for example, glass dust,
flakes, chips, roll marks, and insects. Thresholding and classification based on mea-
surement and experience is vital to ensure maximum product yield. Contaminants,
which are removable, may arise with an incidence of 1000 times that of imperfec-
tions, which are not removable.
Since the task of automatic inspection is likely to be expensive, it is important
to understand how it can be justified. Customers for the product will wish to pay
a price related to a quality grade adequate for their application. Accurate quality
grading can only occur when objective quality measurement is possible. Under
these circumstances, overspecification can be avoided and productivity increased.
Objective measurement, supported where possible by traceability to national stan-
dards and involving ISO 9000, should lead to greater customer satisfaction. Finally,
the application of modern technology, avoiding the need for the human inspector,
should enable faster and more reliable quality grading. These factors and others are
illustrated in Fig. 9.1.
117
118 Chapter 9

Figure 9.1 Reasons for automatic inspection.

9.2 Laser Beam Scanners

A view of a typical float-glass inspection installation is shown in Fig. 9.2. This
shows a scanned laser being projected down at an angle to cover the width of the
glass strip.
Whereas a metal strip product, such as tinplate or aluminum foil, would employ
light reflected from its surface, float glass needs a second transmission channel.
Figure 9.3 shows the fan of a narrow beam of laser radiation, projected onto the
surface under inspection, being both transmitted and reflected. These beams are
returned to their source unit by the use of strip retroreflectors. The reflected beam

Figure 9.2 Typical float-glass inspection installation.

Strip Product Inspection 119

Figure 9.3 Inspection in reflection and transmission.

detects coating and other surface defects and the transmitted beam records defects
in the body of the product.
The types of imperfection that can be detected using transmitted light are illus-
trated in Fig. 9.4. A speck of tin within the body of the glass will reduce the inten-
sity of the laser spot falling upon it, whereas a small surface depression will cause
the beam to be spread and/or refracted. Digs and scratches will cause the beam to
be diffracted. These various types of imperfection will bring about changes in the
intensity, shape, and position of the returned laser beam when brought to a focus.
A schematic representation illustrating how these returned beams could be clas-
sified with a retroreflective scanning system operating in transmission is shown in
Fig. 9.5.
After cleaning up by passage through a pinhole the outgoing laser beam is re-
flected and transmitted through a small aperture, the backside of which acts as a
mirror, before falling on one face of the rotating 12-sided reflecting polygon. The
beam is projected downwards to the material to be inspected and, as the polygon is
rotated at high speed, a small laser spot is scanned across the width of the strip of
material. After transmission, in the case of glass or plastics strip, the beam falls on a
retroreflective screen and is returned along its original direction, but spread slightly
by the diffusing action of the retroreflector. The returned beam carries with it in-
formation about absorbing, refracting, and scattering imperfections with which it
may have interacted. This returned, and now extended, beam illuminates two faces
of the polygon and, in so doing, creates a bright-field and a dark-field detection
channel. The bright-field image of the laser diffraction pattern caused by an imper-
fection on or in the material collects its radiation from the original polygon face
after a further reflection before being focused onto a photon detector. The dark-
field channel collects light at the same time from an adjacent face of the polygon
and forms a similar diffraction image, but this time on a small opaque disk just in
front of a second photon detector. Figure 9.6 shows, in one possible arrangement,
how the shapes and intensities of their laser diffraction patterns can interact with
an opaque stop placed in front of a photon detector, which could separate different
imperfections.
120 Chapter 9

Figure 9.4 Imperfections detected in transmitted light.

The bright-field row shows photon detectors with circular apertures and the
shapes of a laser spot, first when no imperfection is present, followed by cases
when an absorbing defect, a scratch, a surface distortion, and a coating streak, are
present.
The dark-field row below shows detectors with an opaque circular stop at their
centers. Figure 9.7 indicates how positive and negative pulses from the average
video signal, arising across the width of the strip, can be compared with tolerance
levels to trigger quality levels specified by a customer.
As the strip moves forward at a known rate under the scanning laser beam
signal, pulses from the two detectors in the bright-field and dark-field channels are
detected, thresholded, and classified according to their characteristics. A hard copy
Strip Product Inspection 121

Figure 9.5 Retroreflective scanner system.

Figure 9.6 Diffraction patterns for different types of imperfection.

of an image of the scanned surface showing defects classified into types could be
provided.
Additional information could, of course, be obtained by collecting radiation
reflected from the strip by placing a second retroreflector above the surface as
shown in Fig. 9.3. This configuration is essential when inspecting opaque products
and is very successful at detecting small particles of tin on the surface of float glass.
122 Chapter 9

Figure 9.7 Video signal pulses are related to imperfection tolerances.

A major factor, when considering the justification for investing in automatic in-
spection, is its power, by the use of intelligent software, to rapidly separate images
of imperfections from those arising from contaminants. This is facilitated by the
identification of an appropriate combination of optical and electronic signatures.
A scanning system, to be successful, requires access to a wide range of diffrac-
tion, image, and imperfection shape analyzers combined with flexible algorithms
for multichannel electronic signal analysis. These powerful facilities are brought
to bear at the outset in the collection of on-line data from a potential host installa-
tion. In this way, the characteristics of the scanner can be matched most effectively
to a particular strip-product installation. Modern laser beam scanners, employing
intelligent sensor configurations, are capable of achieving less than one “false hit”
per 1000 m2 of float glass.

9.3 Camera Inspection Systems

Laser-beam scanners have several advantages, including high sensitivity, reliable
defect classification, and insensitivity to vibration and ambient light levels. Their
biggest disadvantage is cost due mainly to the high-speed polygon scanner. The
alternative possibility involves the use of a CCD digital camera. This technology is
advancing rapidly in terms of resolution, cost, and availability of intelligent soft-
ware. Figure 9.8 shows a schematic representation of a camera viewing a strip
product illuminated in reflection by a strip lamp mounted above.
Strip Product Inspection 123

Figure 9.8 Camera inspection system.

Not only is a camera system likely to be less expensive to purchase, it will also
be less costly to install. The method of strip illumination, however, requires careful
consideration in order to achieve adequate intensity of illumination and uniformity
across large widths of product, and at the same time the avoidance of disturbing
ambient light. The dynamics of scanning, moreover, mean that a single pixel may
detect more than its nominal area of strip image.
If we take a typical camera pixel size as 0.01 mm and an imperfection resolu-
tion at the surface to be inspected as 0.3 mm, the optical magnification required
will be ×30. It follows that to inspect a product of width 1 m, the diode array in
the image plane will be of length 33 mm and require 3300 pixels. Greater widths
of strip could be inspected by stacking the diodes or by using multiple cameras.
Improvements in solid state light-emitting diodes could speed up the process of
inspection.
Although present technology indicates laser scanners as the preferred technol-
ogy, camera systems are likely to be considered increasingly for some applications
in the future. A modern laser scanner can effectively achieve 30,000 pixels in one
line or 109 pixels in a square area. Even so, using a 100 MHz bandwidth of de-
tection and 100 intensity levels, 100 seconds would be needed to cover 109 pixels.
Using this system, detection of a MIL:10 grade of imperfection on a lens of diam-
eter 50 mm would take approximately 200 seconds, which is probably five times
longer than a skilled inspector would take.

9.4 Chapter Conclusions

The speed of movement and the high quality required of continuously manufac-
tured strip product is leading to a greater requirement for automatic inspection sys-
tems. These need to distinguish between genuine imperfections and contamination
that can be removed. A combination of optical and electronic signal processing,
together with intelligent software, can be embodied in a high-speed laser scan-
124 Chapter 9

ning system to reach this objective. It is, however, unlikely that this system will
be scaled down in size for routine lens inspection due to considerations of cost.
Due to recent progress in the development of digital cameras, these are likely to be
favored for specifications less demanding in terms of resolution, speed, and defect
classification.

Acknowledgment
Thanks are due to J. F. Claridge of Image Automation Ltd. for help with the drafting
of this chapter.

References
1. L. R. Baker, “Some scanning techniques used in automatic inspection,” (the
engineering uses of coherent optics), Proc. of Conference University of Strath-
clyde, 1975, Cambridge University Press, Cambridge, pp. 713–729 (1975).
2. G. M. Clarke and J. Bedford, “The application of the minicomputer to high
speed on-line optical inspection systems,” British Electro-optics, L. R. Baker
(ed.), Taylor & Francis, New York, pp. 65–72 (1977).
3. J. Claridge, “Retaining yields with better glass,” Int. Glass Review, No. 3
(1999).
Appendix 1
Quality Metrics
for Digital Cameras
A1.1 Image-forming Quality of Digital Cameras
The performance of digital cameras is improving at a considerable rate and so it is
important to understand the factors that are likely to limit their use in the fields of
surface inspection and measurement. We describe here test methods for comparing
the relative quality of digital cameras using only a computer.
The choice of digital camera for a particular purpose requiring specified infor-
mation content, includes consideration of the specification of the imaging lens, the
number of pixels required in the image format, and the size of each pixel. This,
together with the lens, determines the resolving power achievable. Unfortunately,
the manufacturers usually only provide the nominal number of pixels, and even this
information may include pixels providing color and edge smoothing rather than
resolution. Fortunately, it is not usually necessary to measure the complete MTF of
the camera system to obtain an approximate measure of the limit of resolution. This
is because the image processing electronics provide a measure of equalization of
the MTF over the spatial frequencies resolved. This ensures good edge sharpness
in images and we only need to determine the limiting spatial frequency or spatial
frequency bandwidth (SFB) in cycles/mm. The inverse of this number will be the
effective pixel or pixel cluster size in cycles/mm. In resolution terms a cluster of
two pixels will be required to detect one cycle.
The width and height of the image format can be measured by recording an
image of known size that fills the field at a known range. If the width of the image
format is Wi and the width of the object being recorded is Wo , at range Ra , then
using a lens of focal length F yields the relationship

Wi Wo Wo F
= or Wi = .
F Ra Ra

The format height can be found as a proportion of the width.

The division of picture width by pixel cluster size gives the effective number of
pixels contributing to resolution achieved. This is then multiplied by the equivalent
125
126 Appendix 1

number obtained for the height calculation to give a measure of true information
content (TIC). One metric for comparing different cameras, in terms of the effi-
ciency of utilizing pixels to create spatial information content, is the pixel packing
index (PPI), where PPI = TIC/NIC (nominal information capacity) quoted as a
percentage. A second and probably more useful metric is the optimum print width
(OPW) defined as the width of print, viewed at 500 mm, where the SFB of the eye
is matched to that of the camera. Other metrics concerned with contrast rendition
include the veiling glare index (VGI) and the contrast resolution (CR).

A1.2 Digital Camera Test Objects

Two tests used to compare cameras include measurement of the optical transfer
function (OTF)1,2 and the VGI3–5 . The OTF embodies the modulation transfer
function (MTF) and the phase transfer function (PTF). In practice, only the MTF
is usually measured because the PTF, which gives rise to image movement at high
spatial frequencies, is less significant.
Measurement of the MTF involves measuring the contrast of a range of grating-
like test patterns having an intensity distribution following a cos2 law. These
fringe patterns can be generated by computer using a MSWord computer program.
Figure A1.1 shows an example of what can be done.

Figure A1.1 Test patterns for measuring the MTF.

Quality Metrics 127

Figure A1.2 Sector star test pattern.

The pattern can be created from vertical contacting lines of luminosity follow-
ing the above law. The patterns can be printed out and used as hard copy to test
a camera, but a much better way is to display the image on a computer screen as
a self-luminous test pattern. This test will probably reveal that the low spatial fre-
quencies (cycles per millimeter) are all of high contrast, approaching unity before
their contrast falls quickly to zero or the image becomes distorted due to interac-
tion between the test pattern and the geometry of the sensor pixels. The value of
spatial frequency, just as image distortion (aliasing) begins to occur, defines the
SFB. Since the determination of this metric only requires a small range of spatial
frequencies, a somewhat simpler test pattern, called a sector star, can be used as
shown in Fig. A1.2. The master copy of this image was made using pen and ink.
As before, this image can be most easily recorded when displayed on a com-
puter laptop screen as a self-luminous test pattern. A viewing range should be
chosen so that the structure halfway along the radius of the pattern is at the point
where aliasing is just starting or the contrast drops to zero. A convenient range is
1000 times the focal length of the camera lens. The recorded image is then down-
loaded into the laptop used to generate the test pattern. The size of the test-pattern
image should be increased to fill the screen, thus providing a magnification of 1000.
If d (mm) is the diameter of a circle on the enlarged image defined by the start of
aliasing or the beginning of the gray unresolved area, using a star test object with
36 sectors, the period of a single sector at this point will be πd/36 or this divided
by the magnification M in the image plane of the camera. The SFB at this radial
setting will, therefore, be 36M/πd or 11.459M/d c/mm. Beyond this point, to-
ward the gray area in the center, there may be evidence of an image, but due to its
distortion it will not add to the quality of the image.
Since the SFB of the eye at a range of 500 mm is 5 c/mm, the OPW will be

MD
11.459 = 2.292MD/d,
5d
128 Appendix 1

where D (mm) is the width of the camera image format and MD is its value in the
print plane.
A short-cut route6,7 to the same conclusion recognizes that the number of limit-
ing cycles will be the same in the camera image as the print, assuming no cropping.
It follows, therefore, that

5(OPW)
D(SFB) = 5(OPW) or SFB =
D

and

D(SFB) 36D 2.3D

OPW = = = ,
5 5πd d

where D is the width of the full field and D/d is a ratio this time determined using
the same scale on the computer in Photoshop® or equivalent software using the
Zoom facility. This approach avoids the need to know the magnification.
If required, measurement of the SFB can be carried out with the test pattern
placed at various points in the field and at different contrast levels. More informa-
tion on the measurement of the OTF/MTF can be found in ISO 9335:1995.
The measurement of the SFB enables determination of the PPI and OPW,
which in turn provides a relative measure of different cameras in terms of their ca-
pacity for recording spatial information. The only facilities needed to measure all
these parameters are the sector star pattern and a tape measure, and even this is not
required if Photoshop or an equivalent software is available for measuring image
widths. Information recording is, however, also influenced by non-image-forming
light scattered by component surface finish, contamination, and reflections from
lens mounts. Light scattered in this way is assessed by measurement of the VGI4 . It
can have a serious effect on image quality, resulting in a loss of low-contrast detail.
The test object for measuring the VGI metric consists of a black disk in a white
diffusing surface. This is most easily made from a cardboard box in the shape of a
cube of side 400 mm with a circular hole of diameter 30 mm cut in the center of
one side, that side painted white. This hole will reflect no light, provided the inside
of the box is painted matte black. Figure A1.3 shows an example of such a test
object.
A more analytical approach to quantifying camera contrast resolution requires
the use of a test chart such as that shown in Fig. A1.4. This was made in MSWord
using adjacent vertical lines of luminosity increasing 10 units at a time. The lower
half of the figure shows the distribution of pixels at the various contrast levels after
transferring the image to Photoshop® and using the histogram display.
The various contrast levels are shown to be evenly distributed and fully re-
solved over the full contrast range. Recording an image of the chart and displaying
the same information in histogram can be used to compare the performance of var-
ious cameras. Further analysis can be undertaken using Channels in Photoshop®
to study the changes in contrast resolution for the red, green, and blue parts of the
Quality Metrics 129

Figure A1.3 Test object for measuring VGI.

Figure A1.4 Contrast resolution test chart.

spectrum. The number of luminosity cells displayed in histogram could be used as

a metric related to CR.

A1.3 Digital Camera Experiments and Test Results

Measurements of the metrics SFB, PPI, OPW, CR, and VGI described above were
carried out using two digital cameras with nominal specifications listed below:
Fujifilm FinePix 6800Z, 3MP, 14.9 mm sensor, 8.3/24.9 mm focal length, f/2.8
Olympus Camedia C-860-L, 1.3MP, 9.4 mm sensor, 5.5 mm focal length, f/2.8
130 Appendix 1

Figure A1.5 Sector star image.

A1.4 SFB Measurements

The sector star test object (shown in Fig. A1.2) was displayed on a laptop screen,
and its image was recorded by each camera, in turn, at a range of 1000 times the
focal length used. When filling the screen with this recorded image at a magnifi-
cation of 1000, the SFB was obtained, typically, as indicated in Fig. A1.5. We can
use SFB = 5(OPW)/D.

A1.5 PPI Measurements

The PPI metric has been suggested as a means of comparing the extent to which
different cameras utilize pixels to record spatial information. The NIC is the num-
ber of pixels as given in the camera manual, whereas the TIC is the number of
pixel clusters in the field format. This value is obtained from the inverse of the
SFB squared divided into the area of the image format. The PPI is TIC/NIC quoted
as a percentage. The pixels lost from the spatial image are used to provide color
and other image enhancing effects.

A1.6 OPW Measurements

Although the pixel cluster size determines the image spatial resolution, the number
present will determine the OPW that can be produced to match the resolution or
SFB of the eye, namely 5 c/mm, when the print is viewed at a distance of 500 mm.
The observed values of the PPI suggests that the OPW values are likely to be
significantly less than may be expected from the megapixel values quoted. We can
use
2.3D
OPW = .
d

A1.7 VGI and CR Measurements

Although veiling glare provides no structural image and it can be removed elec-
tronically, its presence will mask fine low-contrast detail. Thus, for example, a
low-contrast image of a flying bird may be completely replaced by blue sky. Using
Photoshop to display an image of the test object shown in Fig. A1.3, a small win-
dow of, say, 2 × 2 pixels is selected at the center of the black hole image and its im-
age luminosity recorded using Histogram from Image in the menu bar. The selected
Quality Metrics 131

window is then moved to a location in the white surrounding area and the new lumi-
nosity value recorded. The first reading divided by the second, multiplied by 100,
is a measure of the VGI. Measurements of VGI, as described above, were carried
out on the same two cameras. The value obtained will depend on the range and
size of window selected, but it is a useful parameter for comparing cameras. More
information on the measurement of veiling glare can be found in ISO 9358:1994.4
A new high-quality SLR film camera might have a VGI of 1.5%, and any value
above 5% would be regarded as lacking in this regard. It may be difficult, however,
to achieve values of this magnitude for the small lenses typically used in digital
cameras, due to the inevitable residual scattering from contamination and the metal
and glass surfaces placed close to the image detector.

A1.8 Camera Test Results

The above tests were carried out on the two cameras listed above. The results are
given in Table A1.1.
The most notable result is that the 6-MP setting on the Fujifilm camera appears
to provide the same SFB as its 3-MP setting, thus confirming that the additional
pixels are electronic rather than real. Both cameras generate veiling glare to an
extent greater than might be expected from a film camera. A conventional 35-
mm zoom camera using 100 ASA film gave OPW = 400 mm at the wide-angle
setting. Even a cheap plastic panorama film camera gave OPW = 140 mm. The
digital cameras did, however, provide images with a pleasing edge sharpness, as
one might expect from MTF measurements.
Contrast resolution tests shown in Fig. A1.6 for the Fuji camera and Fig. A1.7
for the Olympus model expose the presence of noise and lack of contrast resolution
at both the low and high luminosity levels. While image correction is possible at
high levels, when significant signal is available, information is likely to be lost at
low levels due to noise and stray light. These results suggest the need for further
improvements in the noise levels of image sensors. An assessment of CR is difficult
due to the high noise level, but on a comparison basis it appears that the Fujifilm
camera resolves perhaps 80% of the cells provided by the test chart, whereas the
Olympus model only resolves approximately 50%. This is clearly an area where
more work is required.
Further measurements on nine different digital cameras having pixels in the
range of 1–4 MP indicated that their SFB values varied relatively little, and possibly
by amounts that might be accounted for by variations in lens quality rather than

Table A1.1 Camera test results.

Camera MP SFB PPI OPW VGI
setting c/mm % mm %
Olympus 1.3 86 11 90 8
Fujifilm 6 114 8 170 10
132 Appendix 1

Figure A1.6 Contrast resolution using the Fujifilm FinePix camera.

Figure A1.7 Contrast resolution using the Olympus Camedia.

Quality Metrics 133

pixel size. A need to double the width of a print using a camera with 1 MP would
appear to require a camera with four times this number of pixels.

References
1. L. R. Baker (ed.), Selected Papers on Optical Transfer Function: Measurement,
SPIE Press, Bellingham, WA (1992).
2. T. L. Williams, The Optical Transfer Function of Imaging Systems, IOP Pub-
lishing, London (1999).
3. S. Martin, “Survey of glare measurements in optical instruments,” Proc. of
SPIE, Vol. 274, pp. 288–308 (1981).
4. ISO 9358:1994, Optics and optical instruments: Veiling glare of image forming
systems: Definitions and methods of measurement.
5. S. Martin, “Glare characteristics of lenses and optical instruments in the visible
region,” Opt. Acta., Vol. 19, pp. 499–513 (1972).
6. T. L. Williams, Private communication (2004).
7. L. Baker, “Testing resolution,” OE Magazine, Vol. 4, No. 5, p. 40 (2004).
Appendix 2
Surface Cleaning
A2.1 Introduction
It has often been said, particularly by those responsible for cleaning, that a surface
is only clean at the moment of its generation. The free surface energy created by
the process of material polishing, it is thought, will maximize the rate at which
dust will be attracted and firmly adhere to the surface. Thereafter we can only
speak of degrees of cleanliness or amounts of contamination. Even after the most
thorough cleaning process has been initiated we still have the problem of minimiz-
ing recontamination to consider. Due to its negative effect on system performance,
contamination control from particles and films, and ways of cleaning surfaces have
been subjects of continuous study.1–4 Components and systems known to be par-
ticularly affected by contamination include semiconducting wafers, micro-optics,
high-power lasers, x-ray systems, low-light level imaging, and space optics.
A scientific study of this subject does, of course, require means for inspec-
tion and measurement. As we have seen in Sec. 6.2, component inspection is best
carried out viewing low-angle scattered light. Particulate contamination counting,
employing some form of laser beam scanning, has been extensively employed by
the microcircuit industry because contamination has a direct effect on chip yield.
In Sec. 8.4 we have seen that the MIC can be used to quantify contamination in
assembled optical systems. Any form of contamination that influences the propa-
gation of radiation, such as particles or films, can be studied quantitatively by this
method. The preferred method of quantifying and setting tolerances for contami-
nation, for example, by area, particle count, thickness, or weight, will depend on
the system application.
A large variety of techniques have been used, sometimes in combination, for
cleaning optical surfaces. These include

Clean air jet

Brushing
Aerosol spray
Drag wiping
Vapor degreasing
135
136 Appendix 2

Soap and water

Ultrasonic cleaning
Strip cleaning
Glow discharges
Sputter etching
Solvent spray under pressure

The choice of cleaning method will depend on the environment in which the com-
ponent will operate. An ultraclean service environment will require the best clean-
ing method available.

A2.2 Scrubbing
This form of mechanical cleaning using a cloth or tissue paper and solvent with,
perhaps, a polishing compound such as cerium oxide is the most common. It is
needed to remove the most adherent particles, but is most likely to cause sur-
face damage. This can be minimized by the use of a fluid, such as acetone or
trichloroethylene, to lubricate the cloth. Thorough washing is needed to remove
small particles created by the process. Even bearing in mind the risks, this me-
chanical form of cleaning is regarded as the most effective method.

A2.3 Immersion Cleaning

This is often the final cleaning measure because it is least likely to contaminate,
providing the chosen solvent itself is clean. Ultrasonic agitation may be added to
remove particularly adherent particles not carried off by simple immersion, but the
technique is less successful at removing very small particles that may be attracted
to the component on removal from the bath. Water is not the best choice of solvent
because drying spots are difficult to avoid.

A2.4 Vapor Degreasing

With this method, the component is placed above a boiling organic solvent, such
as perchloroethylene, where it is drenched in condensate, thereby flushing contam-
inates back into the boiling fluid. The result is that grease films should be removed,
but small strongly adherent particles may not be.

A2.5 Viscous Drag Cleaning

This approach is nearest to the most efficient method of scrubbing without the risk
of surface damage. Small particles may be blown away from a surface without
damaging it, given sufficient fluid pressure is applied parallel to the surface. This
can be maximized by using a fluid of maximum density projected as a high-speed
jet across the surface.
Surface Cleaning 137

Figure A2.1 Sequence cleaning of an optical surface.1

The results of a sequence of cleaning methods and their effect on removing

contamination can be seen in Fig. A2.1.
The contaminated surface was first cleaned ultrasonically to remove 24% of
particles, followed by vapor degreasing to remove 11%, followed by high-pressure
solvent flushing, which removed 99.98% of the contaminants.

A2.6 Ion Bombardment Cleaning

This method, often using argon ions, is widely employed for the final cleaning of
glass prior to vacuum coating. The ions are accelerated by the application of a
voltage in the range 1–2 kV and, when reaching the surface, cause local heating
to the extent that particles and films are released. Careful control of the current is
needed to avoid surface stress and cathode sputtering.

A2.7 Strip Cleaning

Adhesive films are available to clean and protect surfaces from damage. They are
applied by dipping into or brushing a solution over the surface. When dry, the film
138 Appendix 2

can be stripped off, carrying most particles with it, but most likely leaving some
residual organic contamination still needing removal.
The benefits of using clean optics are so great that improvements in these meth-
ods of cleaning efficiently and cheaply will continue to be studied. The application
of recent work in quantifying contamination should hasten this process. At this
time, however, high-pressure solvent spraying and surface flushing give the best
results.

References
1. I. F. Stowers and H. G. Patton, “Cleaning optical surfaces,” Proc. of SPIE,
Vol 140, pp. 16–31 (1978).
2. P. T. Ma, M. C. Fong, and A. L. Lee, “Surface particle obscuration and BRDF
predictions,” Proc. of SPIE, Vol. 1165, pp. 381–387 (1989).
3. J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scatter-
ing, Optical Society of America, Washington, DC (1997).
4. P. T. C. Chen, “Optical system contamination: effects, measurements and con-
trol VIII,” Proc. of SPIE, Vol. 5526 (2004) (in press).
Glossary
Approximating spherical surface
Spherical surface for which the root-mean-square (RMS) difference to the total
surface deviation function is a minimum

Approximating aspheric surface

The rotationally symmetric surface for which the rms difference from the irregu-
larity function is a minimum

Bright-field imperfection contrast

Ratio of the difference between the background maximum and minimum luminosi-
ties across an imperfection image to the sum of these values

Combined standard uncertainty

Square root of the sum of the squares of standard uncertainties

Contrast resolution (CR)

Percentage ratio of contrast levels resolved to the number present

Expanded uncertainty
Result of multiplying the combined standard uncertainty by a coverage factor k—
when k = 2, the confidence level of 95% is reached

Far-field nanoscopy (FFN)

Use of comparator microscopy and spatial frequency filtering in nanometrics

Fully developed imperfection

Imperfection that scatters all radiation incident upon it

Imperfections
Localized defects, such as digs and scratches, within the effective aperture of an
optical surface produce by improper treatment during or after fabrication

Inspect
To look into

Irregularity
Peak-to-valley difference between the irregularity function and the plane that best
approximates to it
139
140 Glossary

Irregularity function
Theoretical surface defined by the difference between the total surface deviation
function and the approximating spherical surface

Latent imperfections
Imperfections below the surface only revealed by chemical etching

LEW unit
The severity of a scratch that removes the same fraction of radiation as an opaque
line of width 1 µm when both are measured under the same codified conditions of
illumination and imaging

Line-equivalent width (LEW)

Width of a fully developed scratch or absorbing line which obscures the same
amount of radiation as a partially developed scratch

Measure
The ascertainment of extent by comparison with a standard

Metric
Procedure involving measured physical parameters providing a measure of surface
quality

Modulation
(Imax − Imin )object /(Imax − Imin )image , where I is the luminance of an object or the
luminosity of an image—spatial flux variations are sinusoidal

Modulation transfer factor

Ratio of image modulation to object modulation at a particular spatial frequency

Modulation transfer function (MTF)

Variation of the modulation transfer factor with spatial frequency

Nanometrics
Measurement of a parameter involving nanometer dimensions

Nanometrology
Measurement of dimensions over the range 0.1–100 nm

Nanoscope
Instrument used in far-field nanoscopy
Glossary 141

Near-field nanoscopy (NFN)

Use of scanning probe in nanometrology

Nominal information capacity (NIC)

Number of pixels in the image sensor

Obscuration comparison
Process of measuring the severity of an imperfection by comparing its peak contrast
under bright-field conditions with that of an obscuring artifact of known size

Optimum print width (OPW)

Width of print viewed at 500 mm where the SFB of the eye and camera are matched

Parameter
Physical variable relating to the topography or condition of a surface

Partially developed imperfection

Imperfection that transmits as well as scatters radiation incident upon it

Pixel packing index (PPI)

Percentage ratio of effective pixel clusters to the number of pixels

Point spread function (PSF)

The power distribution in the image of a point source

Power spectral density (PSD)

Frequency spectrum of the surface topography measured in inverse length units

RMS asymmetry (RMSa )

Root-mean-square value of the difference between the irregularity function and the
approximating aspheric surface

RMS irregularity (RMSi )

Root-mean-square value of the irregularity function

Rotationally symmetric irregularity

Peak-to-valley difference between the approximating spherical surface and the
plane that best approximates to it

Saggita error
Peak-to-valley difference between the approximating spherical surface and a plane
142 Glossary

SED unit
The severity of dig that removes the same fraction of radiation as an opaque circular
spot of diameter 1 µm when both are measured under the same codified conditions
of illumination and imaging

Spatial frequency bandwidth (SFB)

Smallest value of spatial frequency at the image sensor where the contrast drops to
zero or aliasing starts

Specular
Mirrorlike

Spot-equivalent diameter (SED)

Diameter of a fully developed dig or absorbing spot that obscures the same amount
of radiation as a partially developed dig

Standard deviation (SD)

Square root of the average value of the squares of the deviations from the mean of
the original readings

Standard uncertainty
Standard deviation of individual uncertainties

Surface finish
Topography of a surface including texture and imperfections, but excluding form

Surface form
The nominal theoretical surface specified by the design, excluding finish

Surface form deviation

Distance between the optical surface under test and the nominal theoretical surface,
measured perpendicular to the theoretical surface, which is nominally parallel to
the surface under test, but excluding finish

Surface imperfections
Localized defects, such as digs and scratches, within the effective aperture of an
optical surface produced by improper treatment during or after fabrication

Surface metrics
Parameters that can be measured related to the function or appearance of a surface

Surface roughness
Statistically random fine topographical structure extending over the surface
Glossary 143

Surface texture
The topography of a surface including roughness and waviness but excluding form
and imperfections

Surface waviness
Periodic pattern of height variations extending over the surface

Total RMS deviation (RMSt )

Root-mean-square difference between the optical surface under test and the desired
theoretical surface, without subtraction of any surface form deviation types

Total surface deviation function

Theoretical surface defined by the difference between the actual surface and the
desired theoretical surface

True information capacity (TIC)

Number of effective pixel clusters based on a measure of resolution

Veiling glare index (VGI)

Percentage ratio of minimum image luminosity in the image of a black image area
to a white area nearby

Visual contrast threshold

Smallest ratio of the brightness of an object to its background that can be seen by
a particular observer
Contacts and Further Reading
The following lists are based on information known to the author at the time of
writing. They are not intended to be comprehensive, but as entry points for further
investigation.

Contacts
Instruments for Surface Metrology

Manufacturer Type
Advanced Metrology Systems Diamond stylus
Chapman Instruments Inc. Differential interference
Digital Instruments STM, AFM
Jeol STM, SEM
Moller-Wedel Fizeau interferometer
Planer Industrial Diamond stylus
Rank Taylor Hobson Diamond stylus
Rodenstock Probe
Wyko Corp., Veeco Instruments Mirau interferometer
Zygo Corp. Fizeau interferometer

Research organizations concerned with surface metrology

• Centro de Investigaciones en Optica, A.C., Mexico
• Center for Optics Manufacturing, University of Rochester, U.S.A.
• Danish Institute of Fundamental Metrology, Denmark
• École Centrale de Lyon, France
• École Nationale Supérieure de Physique de Marseille, France
• Institute of Optical Research, RTT, Sweden
• Institut d’Optique Théorique et Appliquée, France
• National Institute of Standards and Technology, U.S.A.
• National Physical Laboratory, UK
• Naval Weapons Center, China Lake, U.S.A.
• Philips Centre for Manufacturing Technology, The Netherlands
• Physikalisch-Technische Bundesanstalt, Germany
• Sira Ltd., UK
• University of Arizona Optical Sciences Center, U.S.A.
• University of Birmingham, UK
• Universität Stuttgart, Institut für Technische Optik, Germany
• University of Warwick, UK
145
146 Contacts and Further Reading

Further Reading
D. Whitehouse, Surfaces and Their Measurement, Kogan Page, London (2003).
L. Blunt and X. Jiang, Advanced Techniques for Assessment Surface Topography,
Kogan Page, London (2003).

K. Stout, Development of Methods for the Characterisation of Roughness in Three

Dimensions, Kogan Page, London (2003).

K. Stout and L. Blunt, Three Dimensional Surface Topography, Kogan Page, Lon-
don (2003).

F. Twyman, Prism and Lens Making, Second Edition, Hilger & Watts Ltd., London
(1952).

D. Malacara (ed.), Optical Shop Testing, Wiley, New York (1978).

K. Creath, “Phase-measurement interferometry techniques,” Progress in Optics

XXVI, E. Wolf (ed.), Elsevier Science, Oxford (1988).

D. J. Whitehouse, Handbook of Surface Metrology, Avalon Technology, Cleveland,

OH (1994).

J. M. Bennett and L. Mattsson, Introduction to Surface Roughness and Scattering,

Optical Society of America, Washington, DC (1997).
Index
abrasion, 53 coating scatter, 50
absorbing artifacts, 97 combined standard uncertainty, 139
absorption, 30 comparator standards, 94
acceptance thresholds, 54 comparison microscopy, ix, 48
aerial, 31 comparison trials, 62
aesthetic appearance, 102 complex spatial frequency filter, 108
AFNOR NFS 10-006:1979, 61 components, 53, 60, 65, 68
aliasing, 127 computer-aided interferometry, 17
amplitude objects, 55 computer-aided microinterferometer, 31, 41
analogue microscope image comparator contamination, ix, 31, 48, 53, 59, 93, 96, 117,
(AMIC), 75 123
antireflection film, 100 contrast resolution, x, 126, 139
appearance, 5, 54 cosmetic imperfections, 2, 52, 54
approximating a spherical surface, 25, 139 cosmetic quality, 8, 52
approximating an aspheric surface, 25, 139
area, 55 damage, 50–53, 59, 84
area Method I, 68 dark-field, 55, 61, 63, 71, 98
area of obscuration, 67 dark-field detection, 119
aspherical surface-form, 16 depth of imperfection, 56, 61, 65
assembled systems, 65 deterministic data, 10
assemblies, 51, 53, 96 deterministic methods, 29, 32, 37, 47, 55
atmospheric turbulence, 17 diagnostic information, 6, 29
automatic inspection, 2, 117 diffraction, 8, 23, 30
automation, 50, 59, 117 digital camera, ix, 17, 79, 84, 125, 129
average roughness Ra , 33 digital interferogram analysis, 26
digital microscope image comparator (DMIC),
79
benefits of measurement, 54
digs, 5, 7, 53, 59, 119
bright field, 99
DIN 3140: Part 7 1978, 61
bright-field contrast, 55, 139
drift, 22
bright-field image, 72, 119
ductile form, 52
British standard scratches, 95
dust, 53–54, 59
brittle chip formation, 52
British standards (BS), 64 electro-optic, 112
BS 4301:1982, 61, 63 electron-beam machining, 64
BS 4301:1991, 61 embedded fibers, 64
BS ISO 10110-7:1996, 61 energy research, 53
bubbles, 53 environments, 53
error sources, 21
calibrated reference defect, 67 etching, 64
calibration, 21 expanded uncertainty, 139
calibration graticule, 77 eye, 52, 56, 58–59, 65
CCD cameras, 117
CCD detector, 16 far-field nanoscopy (FFN), ix, 48, 56, 93, 139
cement layer, 51 filters, 65
channels, 128 fingernail test, 29
characterization of imperfections, 54 finish, ix, 8
chemical damage, 53 fitness for purpose, 53
chemical etching, 52, 64 Fizeau interferometer, 17
classify imperfections, 59 float glass, 117
clean-room, 53 form, ix, 50
cleaning, 53–54, 59 four-step measurement of phase, 19
closer tolerances, 50 Fourier analysis, 43
coating, 51–54, 100 France, 60–61
147
148 Index

French standard, 86 ISO 10110-7: surface imperfection tolerances,

Fujifilm, 129, 131 47–48, 61, 67
fully developed imperfections, 52, 73, 83, 139 ISO TC 172, 66
functional significance, 2, 52, 54
jig loading, 59
German Standard DIN 3140: Part 7, 1978, 86 Kodak dig/scratch paddle, 95
Germany, 60, 63
glare spread function, 53, 101 laminar airflow, 71
grade, 61 laminar-flow cabinet, 59
graticule, 5 laser, 48, 50, 84
ground surfaces, 31 laser beam ablation, 101
laser beam scanning, x, 48, 50, 53, 117, 135
handling, 54, 59 laser Fizeau interferometer, 18
high spatial frequency, 10 laser gyroscope, 53
high-power/energy laser beams, 8, 53 laser optics, 50
high-pressure solvent spraying, 137 latent imperfections, 52, 140
high-speed flexible lap polishing, 101 length of imperfection, 37, 55, 61, 65
histogram, 106 lens carriers, 54
lens centration, 5
light scatter, 8
image comparator, 63 line-equivalent width (LEW), 48, 61, 74, 87,
image contrast, 57, 72 140
image distortion, 5 liquid-crystal modulators, 112
image format, 128 localized imperfections, 7
image plane, 52 location of imperfection, 51
immersion cleaning, 136 low spatial frequency, 10
impact, 53 low-light-level imaging systems, 8
imperfection, ix, 2, 5, 47, 49–57, 59–61, low-light-level optical systems, 53
64–68, 93, 139 luminosity values, 93
imperfection appearance, 52
imperfection thresholds, 60 machine vision, 58
imperfection tolerances, 66, 68 macroscopic, 10
imperfections, typical, 52 macrotopography, 29
in-house scratch reference standards, 96 magneto-optic, 112
indications of acceptability, 67 manufacturing efficiency, 6
influence of imperfections, 53 market access, 50
informatics, 50 materials processing, 53
information technologies, 65 matte, 31
inhomogeneity, 53 measurement, ix, 2, 53–55, 59–61, 63–66, 68,
injection molding, 64 140
metal strip product, 118
inspection, ix, 2, 48, 53, 55, 58–60, 71, 139
Method I, 48, 67
inspection stations, 67
Method II, 48, 67
inspector, 59–60, 62–63, 117
method of substitution, 71
instrument resolution, 21 metric, 140
intelligent sensor, 122 microcircuit industry, 135
intelligent software, 122 microdefects, 37
intensity profile analyzer, 102 microengineering, 50
interferometer, 13, 16, 58 microinterferometery, 34, 55, 58
interlaboratory comparison, 21 microlenses, 65
International Organization for Standards microscope, 55, 61, 65
(ISO), 66–67 microscope image comparator (MIC), 48,
international standards, 2, 54 62–63, 75, 93–94
ion beam etching, 101 microscope interferometer, 41
ion bombardment cleaning, 137 microtopography, ix, 29, 32, 47, 51, 54
irregularity, 139 midrange spatial frequency, 10
irregularity function, 25, 140 MIL 10 scratch, 66
ISO 10110-5:1996, 16 MIL standard scratches, 95
Index 149

MIL-0-13830A:1963, 60, 86, 94 Photoshop® , 79, 106, 128

MIL:10, 72 physical relevance of RMS errors, 26
MIL:40, 72 pixel cluster, 125, 130
MIL:60, 72 pixel packing index (PPI), 126, 128, 130, 141
Ministry of Defence (MOD), 62 point spread function (PSF), 3, 34–35, 141
modulation, 140 polar distribution, 31
modulation transfer function (MTF), 50, 53, polar scatter patterns, 48
126, 140 polarization, 30–31
mounting, 54 polishing, 50, 53, 59
polygon, 119
nanoengineering, 50 position sensor, 16
nanometer, 52 power spectral density (PSD), 37, 141
nanometer dimensions, 50, 140 power spectral density function (PSD), 41
nanometrics, 140 PPI, See pixel packing index
nanometrology, 140 probe gauging, 13
nanoscope, ix, 93, 140 probes, 3
national standards, 60, 66, 68 process optimization, 6
near-field nanoscopy (NFN), ix, 93, 141 profitability, 54, 68
Newton’s fringes, 13 protective coatings, 54
nickel sulphide crystals, 5
nominal information capacity (NIC), 141 quality control, 51, 58
numerical aperture, 56 quality metrics, ix
objective, 60–61, 63, 66
radiometric obscuration, 71
objective methods, 68
radius of curvature, 65
obscuration, 71, 141
random effects, 21
Olympus Camedia, 129
Rank Taylor Hobson standard scratches, 95
on-machine, 43, 114–115
real scratches, 6
opaque products, 121
reference components, 6
operator training, 54
reference graticule, 74
optical and electronic signatures, 122
optical cements, 53 reference plate, 13
optical coatings, 31 reflecting mode, 77
optical components, 13, 68 reflection, 52, 56, 77
optical constants, 31, 58 refractive index, 57
optical material, 51–53 relative contrast, 48
optical probe, 14, 16 responsivity, 23
optical production, 50 retroreflectors, 118
optical system resolution, 8 RMS, 25, 31, 54
optical transfer function (OTF), 5, 101, 126 RMS asymmetry (RMSa ), v, 26, 141
optical workshop, 2 RMS irregularity (RMSi ), v, 26, 141
optimum print width (OPW), x, 126–128, 130, RMS roughness Rq , 33, 36
141 rotationally symmetric irregularity, 25, 141
overspecification, 117 roughness, 2, 5, 7, 29, 54
round-robin interlaboratory comparison, 6
paint particles, 53 rubbing, 59
performance parameters, 54, 60 Rubert Gauge specimens, 106
parametric, ix, 141 Rubert Microsurf Gauge, 103
partially developed imperfections, 52, 73, 83,
86, 141 sagitta error, 25, 141
particulate contamination, 53, 135 sampling, 37
perchloroethylene, 136 sampling interval, 33
personal bias, 22 sampling length, 33
phase contrast, 98, 100, 113 SC 1 fundamental standards, 66
phase objects, 52 SC 2 materials for optical processing, 66
phase steps, 22 SC 3 optical materials and components, 66
phase transfer function (PTF), 126 SC 4 telescopes, 66
phase-contrast microscope, 108 SC 5 microscopes, 66
150 Index

SC 6 geodetic instruments, 66 surface height variations, 93

SC 7 ophthalmic endoscopic, metrological surface imperfections, 5, 67, 93, 142
instruments, and test methods, 67 surface metrics, 142
SC 8 ophthalmic optics, 67 surface morphologies, 101
SC 9 electro-optical systems, 67 surface patterning, 41, 44
scanner, 48 surface point spread functions, 34
scanning electron microscope (SEM), 6, 55 surface roughness, 35, 142
scanning probe microscope (SPM), 6, 30, 54 surface slope, 16
scratch, 2, 5, 7, 53, 55, 59, 62, 65, 119 surface step, 56
scratch and dig plates, 47 surface texture, 7, 101, 143
scratch visibility reference plates, 60 surface waviness, 41, 143
scrubbing, 136 surface-cleaning, x
sector star pattern, 127–128 system assembly, 68
sector star test pattern, 127
self-luminous test pattern, 127 TC/172, 66
sensitivity of measurement, 66 technologies, 65
SFB, See spatial frequency bandwidth template metal, 13
shattering, 53 terminology, 51
single-point diamond turning, 41, 101 test pieces, 22
size of imperfections, 51 testing, 32, 68
sleeks, 53 texture, ix, 2, 31, 93
solid state light-emitting diodes, 123 thin film coatings, 8
spatial frequencies, 3 thin film, 65
spatial frequency bandwidth (SFB), 127 thin film deposition, 64
spatial frequency filtering, 48 tolerance, 52, 54–55, 59–61
spatial frequency zones, 43 tolerance level, 54, 66
spatial image quality, ix tolerancing, 53, 55
tool wear, 102
specimens, 23
topography, 55
speckle, 23, 31
total integrated scatter (TIS), 31, 36
spectral band, 65
total integrated scatter measurement, 36
specular reflectance, 31, 37
total light scattered, 48
specular surfaces, 37
total RMS deviation (RMSt ), v, 26, 143
speed of measurement, 66 total surface topography, ix, 2, 143
spot-equivalent diameter (SED), 48, 74, 87, traceability, 54–55, 60–61, 63, 66, 68, 117
142 training/education, 53, 54, 59–60
staining, 53 transfer function, 126
standard deviation, 142 transmission or reflection, 52, 57, 65
standard uncertainly, 142 transmitting mode, 77
standards, 54–55, 60–63, 66 transportation, 59
step, 58 traveling microscope, 63
stress, 53 triangulation ranging, 17
striae, 53 true information capacity (TIC), 125, 143
strip cleaning, 137 TV recording, 17
strip illumination, 123 TV line profiler, 76
strip products, 48, 117 Twyman-Green interferometer, 17
stylus, 30, 33, 54
stylus method, 33 U.S., 60, 63
subjectivity, 59, 67–68 UK, 60–61, 63
substrates, 51 ultrasonic agitation, 136
surface area, 61 ultrasonic cleaning, 59
surface coatings, 65 uncertainty of measurement, 2, 50–51, 55, 66
surface contamination, 2, 8
surface damage, 5 vacuum coating, 136
surface deviation, 23 vapor degreasing, 136
surface finish, 2, 4, 142 veiling glare, 4, 50, 53
surface form, 2, 5, 142 veiling glare index (VGI), 126, 128, 143
surface generation, 7, 68 vibration, 17
Index 151

viewing conditions, 57 wavefront, 4, 52–53, 56, 58, 64

viscous drag cleaning, 136 wavelength, 65
visibility, 52, 54–57, 60–61 waviness, 2, 7, 10
visibility Method II, 68 width, 52, 55–56, 61, 65
visibility standards, 63
windows, 65
visual contrast threshold, 143
visual comparison, 61 wrapping, 53–54, 59
visual inspection, 48, 63
visual interferogram interpretation, 26 yields, 4, 49, 53–54, 59–60
About the Author

Dr. Lionel Baker graduated in physics from Imperial

College, London, in 1953, and received his Ph.D. there
in 1955 working on the diffraction theory of image for-
mation. After a three-year Civil Service Commission Re-
search Fellowship at the Royal Aircraft Establishment,
Farnborough, he joined Sira Ltd. in 1958.
He was a founding member of the Sira Board of Di-
rectors and served from 1972 until his retirement as Te-
chnical Director in 1990. He is named in numerous patent
applications, including instruments for measuring the optical transfer function, sur-
face imperfections, and stress in structures subjected to dynamic loading. He is the
author of 100 technical publications and lectures widely in the field of instrument
technology. He is chairman of the CPW/172 BSI Committee on optics and optical
instruments and has been active in drafting several ISO standards.
Dr. Baker is a Fellow of the Institute of Physics, SPIE, and OSA. He was the Ex-
ecutive Editor of the Journal of Modern Optics (formerly Optica Acta) for 17 years,
and more recently served as an Associate Professor at Brunel University. His hob-
bies include painting and drawing, cooking, wine making, and carpentry.
 SBN 978 0 8194 5576 5
9 0 0 0 0

P.O. Box 10
Bellingham, WA 98227-0010

ISBN-10: 0819455768
9 780819 455765
ISBN-13: 9780819455765
SPIE Vol. No.: TT65