Professional Documents
Culture Documents
Full Chapter Optical Inspection of Microsystems 2Nd Edition Wolfgang Osten PDF
Full Chapter Optical Inspection of Microsystems 2Nd Edition Wolfgang Osten PDF
https://textbookfull.com/product/optical-inspection-of-
microsystems-first-edition-wolfgang-osteneditor/
https://textbookfull.com/product/springer-handbook-of-geographic-
information-2nd-edition-wolfgang-kresse/
https://textbookfull.com/product/elements-of-optical-networking-
basics-and-practice-of-glass-fiber-optical-data-
communication-2nd-edition-bruckner/
https://textbookfull.com/product/essentials-of-modern-optical-
fiber-communication-2nd-edition-reinhold-noe-auth/
Clinical psychology a modern health profession 2nd Ed
2nd Edition Wolfgang Linden
https://textbookfull.com/product/clinical-psychology-a-modern-
health-profession-2nd-ed-2nd-edition-wolfgang-linden/
https://textbookfull.com/product/introduction-to-optical-
microscopy-2nd-edition-jerome-mertz/
https://textbookfull.com/product/the-mexican-revolution-s-wake-
sarah-osten/
https://textbookfull.com/product/the-physics-of-thin-film-
optical-spectra-an-introduction-2nd-edition-olaf-stenzel-auth/
Optical Inspection of
Microsystems
Optical Inspection of
Microsystems
Second Edition
Edited by
Wolfgang Osten
Cover image: Courtesy of Michał Józwik, Faculty of Mechatronics, Warsaw University of Technology.
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to
publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materi-
als or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material
reproduced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained.
If any copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in
any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, micro-
filming, and recording, or in any information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copyright.
com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a
not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been granted a
photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifi-
cation and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
v
vi Contents
Chapter 13 Optoelectronic Holography for Testing Electronic Packaging and MEMS............. 379
Cosme Furlong
Index............................................................................................................................................... 563
Preface to Second Edition
2007 was a good year to publish a book dedicated to optical methods for the inspection of micro-
systems. The technological basis for the design and fabrication of microcomponents was meanwhile
mature, and many devices and applications were already implemented or under test. Consequently,
reliable procedures for quality assurance gained importance, and the search for efficient inspection
methods was on the agenda of many vendors. Full-field optical methods established an alternative
approach in comparison with conventional tactile and/or pointwise sampling methods, owing to their
noncontact, nondestructive, and areal working principle. Thus, this book was very well received by
the readers and translated to Chinese1 a couple of years after its first edition. However, 10 years later,
it was time to think about an update. As the technology progressed, naturally higher expectations
were placed on quality assurance and thus the inspection techniques used. For instance, challenges
such as checking the system’s internal parts after packaging came up. New requirements for the
calibration of measurement tools, new rules for the estimation of uncertainty in measurement, and
meaningful surface features had to be considered.
Consequently, the publisher and editor decided to prepare an updated and extended edition, tak-
ing into account the progress in the field. This second edition is now available. Three new chapters
were added, focusing on the following:
All these new chapters were written by leading experts in the particular fields. Nine of the original
chapters have been updated, while only five chapters have retained the content of the first edition.
Thus, the reader finds an almost completely updated and extended book. Publisher, authors, and
editor hope that the second edition will find an even well-interested audience and help to solve the
even more challenging problems in microsystems technology.
The editor is very grateful to all contributors and the publisher for the repeated fruitful coopera-
tion. Special thanks goes to Michelle Kazensky of Lumina Datamatics and Marc Gutierrez and
Kari Budyk at CRC Press.
Wolfgang Osten
Lilienthal, 2019
vii
Preface to First Edition
The miniaturization of complex devices such as sensors and actuators is one of the biggest challenges
in modern technology. Different manufacturing technologies—for instance, the so-called LIGA
technique and UV lithography—allow the realization of nonsilicon and silicon microparts with a
high aspect ratio and structural dimensions in the range from nanometers to millimeters. LIGA is an
acronym standing for the main steps of the process, i.e., deep x-ray lithography, electroforming, and
plastic molding. These three steps make it possible to mass-produce high-quality microcomponents
and microstructured parts, in particular from plastics, but also from ceramics and metals at low
cost. Techniques based on UV lithography or advanced silicon etching processes (ASE) allow for
direct integration of electronics with respect to the realization of advanced microelectromechanical
systems (MEMS) devices. Further technologies such as laser micromachining, electrochemical
milling (ECF), electrodischarge machining, and nanoimprint lithography (NIL) offer, meanwhile,
an economical, high-resolution alternative to UV, VUV, and next-generation optical lithography.
Increased production output, high system performance, and product reliability and lifetime are
important conditions for the trust in a new technology and deciding factors for its commercial success.
Consequently, high quality standards are a must for all manufacturers. However, with increasing
miniaturization, the importance of measurement and testing is rapidly growing, and therefore the
need in microsystems technology for suitable measurement and testing procedures is evident. Both
reliability and lifetime are strongly dependent on material properties and thermomechanical design.
In comparison to conventional technologies, the situation in microsystems technology is extremely
complicated. Modern microsystems (MEMS and MOEMS) and their components are characterized
by high-volume integration of a variety of materials and materials combinations. This variety is
needed to realize very different and variable functions such as sensor and actuator performance,
signal processing, etc. Still, it is well known that the materials´ behavior in combination with new
structural design cannot be easily predicted by theoretical simulations. A possible reason for wrong
predictions made by FEM calculations with respect to the operational behavior of microdevices
is, for instance, the lack of reliable materials data and boundary conditions in the microscale.
Therefore, measurement and testing procedures are confronted with a complex set of demands. In
general, the potential for the following is challenged:
Measurement and inspection techniques are required that are very fast, robust, and relatively low
cost compared to the products being investigated. The reason for this demand is obvious: proper-
ties determined on much larger specimens cannot be scaled down from bulk material without any
experimental verification. Further on, in microscale, materials’ behavior is noticeably affected by
production technology. Therefore, simple and robust methods to analyze the shape and deformation
of the microcomponents are needed. Together with the knowledge of the applied load and appropri-
ate physical models, these data can be used for the derivation of material parameters and various
ix
x Preface to First Edition
system properties. It is obvious that neither a single method nor a class of measurement techniques
can fulfill these requirements completely. Conventional tensile test techniques (e.g., strain gauges)
are unable to test specimens from submillimeter-sized regions because of their limited local resolu-
tion and partly unwanted tactile character. Other approaches, such as, for instance, microhardness
measurements, do not reveal directional variations.
However, full-field optical methods provide a promising alternative to the conventional methods.
The main advantages of these methods are their noncontact, nondestructive, and fieldwise working
principle; fast response potential; high sensitivity and accuracy (typical displacement resolution of a
few nanometers, strain values of 100 microstrain); high resolution of data points (e.g., 1000 × 1000
points for submillimeter field of view); advanced performance of the system, i.e., automatic analysis
of the results; and data preprocessing in order to meet requirements of the underlying numerical
or analytical model. Thus, this book offers a timely review of the research into applying optical
measurement techniques for microsystems inspection. The authors give a general survey of the
most important and challenging optical methods such as light scattering, scanning probe micros-
copy, confocal microscopy, fringe projection, grid and Moiré techniques, interference microscopy,
laser Doppler vibrometry, holography, speckle metrology, and spectroscopy. Moreover, modern
approaches for data acquisition and processing (for instance, digital image processing and correla-
tion) are presented.
The editor hopes that this book will significantly push the application of optical principles for
the investigation of microsystems. Thanks are due to all authors for their contributions, which give
a comprehensive overview of the state of the art in the fascinating and challenging field of optical
microsystems metrology. Finally, the editor is grateful for the cooperation shown by CRC Press
represented by Taisuke Soda, Preethi Cholmondeley, Gerry Jaffe, and Jessica Vakili.
Wolfgang Osten
Stuttgart, 2007
Editor
Wolfgang Osten earned an MSc/Diploma in physics at the Friedrich Schiller University Jena in
1979. From 1979 to 1984, he was a member of the Institute of Mechanics in Berlin, working in the
field of experimental stress analysis and optical metrology. In 1983, he earned a PhD at the Martin
Luther University Halle-Wittenberg for his thesis in the field of holographic interferometry. From
1984 to 1991, he was employed at the Central Institute for Cybernetics and Information Processing
ZKI in Berlin, making investigations in digital image processing and computer vision. Between
1988 and 1991, he headed the Institute for Digital Image Processing at the ZKI. From 1991 until
2002, he joined the Bremen Institute of Applied Beam Technology (BIAS) to establish and direct
the Department Optical 3D Metrology. From September 2002 until October 2018, he was a full pro-
fessor at the University of Stuttgart and director of the Institute for Applied Optics. From 2006 until
2010, he was the vice rector for research and technology transfer of the Stuttgart University, and
from 2015 until 2018, he was the vice chair of the university council. His research work is focused
on new concepts for industrial inspection and metrology by combining modern principles of optical
metrology, sensor technology, and image processing. Special attention is directed to the develop-
ment of resolution-enhanced technologies for the investigation of micro- and nanostructures.
xi
Contributors
Anand Asundi Cosme Furlong
School of Mechanical and Aerospace Center for Holographic Studies and Laser
Engineering MicroMechatronics
Nanyang Technological University Mechanical Engineering Department
Singapore Worcester Polytechnic Institute
Worcester, Massachusetts
Petra Aswendt
ViALUX GmbH Claudiu Giusca
Chemnitz, Germany Cranfield University
Cranfield, United Kingdom
Alain Bosseboeuf
Centre de Nanosciences et de Nanotechnologies Christophe Gorecki
UMR CNRS 9001 Département LOPMD, FEMTO-ST
Université Paris-Sud Université de Franche-Comté
Université Paris-Saclay Besançon, France
Orsay Cedex, France
Marc Gronle
Philippe Coste Institut für Technische Optik
Centre de Nanosciences et de Nanotechnologies Universität Stuttgart
UMR CNRS 9001 Stuttgart, Germany
Université Paris-Sud
Université Paris-Saclay Han Haitjema
Orsay Cedex, France Mitutoyo Research Center Europe
Leuven, Belgium the Netherlands
Patrick Delobelle
Department LMA, FEMTO-ST Roland Höfling
Université de Franche-Comté ViALUX GmbH
Besançon, France Chemnitz, Germany
xiii
xiv Contributors
CONTENTS
1.1 Introduction............................................................................................................................... 2
1.2 Classification of Tasks...............................................................................................................2
1.3 Image Processing and Computer Vision Components..............................................................4
1.3.1 Behavior of Light, Colors, and Filters........................................................................... 5
1.3.2 Illumination................................................................................................................... 8
1.3.3 Lens Systems............................................................................................................... 12
1.3.4 Sensors......................................................................................................................... 14
1.3.4.1 CCD Sensor.................................................................................................. 15
1.3.4.2 CMOS Sensors.............................................................................................. 17
1.3.4.3 Color Sensors and Color Cameras................................................................ 18
1.3.4.4 Camera Types and Interfaces........................................................................ 19
1.3.4.5 Frame Grabbers............................................................................................20
1.4 Processing and Analysis of Image Data.................................................................................. 21
1.4.1 Computer Vision Process............................................................................................ 21
1.4.1.1 Image Acquisition......................................................................................... 21
1.4.1.2 Image Processing.......................................................................................... 22
1.4.1.3 Image Analysis............................................................................................. 22
1.4.1.4 Measurements and Decision-Making........................................................... 22
1.4.1.5 Presentation of Results.................................................................................. 22
1.4.2 Image Data Preprocessing and Processing Methods................................................... 22
1.4.2.1 Histogram..................................................................................................... 23
1.4.2.2 Point Transformations................................................................................... 23
1.4.2.3 Spatial Filtering............................................................................................26
1.4.3 Image Data Analysis Methods..................................................................................... 30
1.4.3.1 Point Operations........................................................................................... 30
1.4.4 Solving Measurement and Testing Tasks....................................................................34
1.4.4.1 Finding a Test Object or Region of Interest..................................................34
1.4.4.2 Position Recognition..................................................................................... 37
1.4.4.3 Measuring Geometric Features.................................................................... 39
1.4.4.4 Presence Verification.................................................................................... 42
1.4.4.5 Defect and Fault Detection........................................................................... 43
1.5 Commercial and Noncommercial Image Processing and Computer Vision Software........... 45
1.6 Image-Processing Techniques for the Processing of Fringe Patterns
in Optical Metrology........................................................................................................ 47
1.7 Conclusion............................................................................................................................... 48
References......................................................................................................................................... 49
1
2 Optical Inspection of Microsystems
1.1 INTRODUCTION
Not only is there a requirement for testing the electrical and dynamic behavior of MEMS, but there
is also a considerable demand for methods for testing these systems during both the development
phase and the entire manufacturing phase. With the aid of these test methods, it is possible to assess
such static properties as dimension, shape, presence, orientation, and the surface characteristics of
microsystems or their components. Using an optical measurement and testing technique based on
image processing and computer vision, a wide range of procedures can be applied, which enable
such properties to be recorded rapidly and in a robust and noncontact way.
If measurement and testing means are not based on special optical procedures but rather on
illumination with normal light and imaging with normal and microscopic optics, their resolution
capabilities extend only to just below the micrometer range. This is due to the diffraction of light
and the dimensions of imaging sensor elements in a lateral direction. Such a degree of resolution
is inadequate, as it is unable to cover the entire scale of microsystem structure sizes, which ranges
from just a few nanometers (e.g., surface roughness) to a few millimeters (e.g., external contours). In
order to measure sizes in the nanometer range, special imaging measurement and testing means are
required. These include interferometers, spectrometers, near-field/scanning electron/atomic force
microscopes (AFMs), and specialized reconstruction and analysis processes such as fringe process-
ing and scanning techniques, which are described in the following chapters.
The main advantages of implementing optical testing equipment using simple light sources and
normal and microscopic optics are the speed with which images can be recorded and analyzed and
the fact that they can be easily integrated into the manufacturing process, thus making the error-
prone removal of components from and reintroduction into the clean environment for test purposes
superfluous. For this reason, despite their limited resolution capabilities, they are still ideally suit-
able for testing large piece numbers, that is, in the manufacturing process in the areas of assembly,
function, and integrity testing and packaging.
Furthermore, the algorithms developed for image processing and computer vision are not only
suitable for analyzing images recorded using a video camera but can also be applied to the fields of
signal analysis, data analysis and reconstruction, and many other areas.
This chapter deals not only with the technical aspects of illumination and image recording
techniques but also with image processing and computer vision processes relevant to the optical
measurement and testing technique and their implementation in typical measurement and testing
tasks. Finally, several software products that are available commercially for image processing and
computer vision will also be described. However, first, a classification of typical measurement and
testing tasks in the field of microsystem development and production is given.
constructed as MEMS and are utilized especially in the automotive industry—a mass market with
high safety requirements (Airbag, Electronic Stability Program, Bosch, Germany [ESP®]). However,
MEMS and MOEMS are being used more and more in many other fields such as in the medical
industry or in biological and chemical diagnostics. The “Lab-on-a-Chip” is c apable of carrying out
complete investigation processes for chemical, biological, or medical analyses. Microspectrometers,
just a few millimeters in size, enable extremely small and cheap analysis devices to be constructed,
which can be used in flight equipment for monitoring terrestrial biological and climatic processes and
also in a wide range of technical applications in the form of hand-held devices. Using micromotors,
microdrives, micropumps, and microcameras, instruments can be constructed in the medical field for
keyhole diagnosis and surgical interventions.
Although, in comparison with microelectronic components, microsystems possess a clear three-
dimensional structure (e.g., microspectrometers, electrical micromotors, and gears for drives made
by using microinjection molding techniques), classic MEMS structures, especially sensor and mir-
ror systems, are essentially two-dimensional in shape. This feature is the result of constructing
MEMS based on semiconductor materials and on the manufacturing processes used in conjunction
with such materials. Another conspicuous characteristic of MEMS-based systems is the common
use of hybrid constructions, where the drive and analysis electronics are located on a semiconductor
chip and the actual MEMS (e.g., the sensor element) is located on a separate chip.
Both these properties influence the tests realizable for MEMS, using image processing and com-
puter vision. These are essentially performed using incident light arrangement, where both image
recording and illumination take place at the same angle. The transmissive light arrangement (which
can be much better controlled), where the object to be tested is situated between the illumina-
tion source and the image recording system, can be used to advantage if MEMS are more three-
dimensional in shape. This is becoming more and more the case.
From the point of view of image processing and computer vision, the solutions listed below are
possible for the following examples of testing tasks:
• Test object or regions of interest: Test objects need to be located if the position of the
objects to be tested varies either in relation to each another or in relation to the reference
system of the observation area. This is often the case with hybridly constructed systems,
where individual components are located relatively inaccurately next to one another.
Regions of interest need to be located, for example, in cases where MEMS components
are processed or adjusted individually (e.g., when measuring the cross-section of laser-
trimmed resistances and capacitors) or if a measuring area needs to be determined using
a small measurement range for a high-resolution measuring or test system (multiscaled
measuring/testing).
• Position recognition: The position of components needs to be recognized if they are
not aligned for a test or if their position, when installed, shows degrees of freedom (e.g.,
resistance, diode, and gear). Another example is when tests need to be carried out on the
components themselves, such as the recognition of component coding and the measure-
ment of geometric features. In some cases, it is necessary to merge the best two or more
images taken from the same scene (e.g., with different sensors or from different positions)
in one image. This very important task in image processing is called “image registration.”
There are a lot of approaches and methods to realize this “best fit” of images in medical,
remote sensing, and industrial applications [1].
• Measuring geometric features: Investigating a component from a metrological point of
view shows whether production tolerances have been adhered to; this affects both the
mechanical and electrical behavior of the object. The location of geometric features using
contour or edge recognition and the calculations based on them to obtain straightness,
circularity, length, enclosed surfaces, angle, distance, diameter, etc, form the principles of
optical metrology.
4 Optical Inspection of Microsystems
• Presence verification: The monitoring of production and assembly processes often requires
a purely qualitative statement regarding certain features, without specific knowledge of
their geometrical characteristics. For example, in production processes, tool breakage can
be monitored by checking a work piece for the presence of bore holes, grooves, etc. With
assembly processes, the focus of interest is usually on the assembled result, that is, the
presence of all components requiring assembly.
• Fault detection: In contrast with presence verification, in the case of fault detection, fea-
tures are checked for deviations from the required standards. To detect faults, for example,
text recognition is used for reading identification markings on components, color checking
for verifying color-coded components, and texture analysis for investigating structured
surfaces for flaws (e.g., scratches).
In order to solve these examples of measuring and verification tasks based on two-dimensional
image data, image processing and computer vision have a whole range of proven processing and
interpretation methods available. Shading correction, the averaging of image series, spatial and
morphological filtering, edge detection, pattern and geometric feature matching, segmentation, con-
nectivity analysis, and the metrology of geometric features denote some of these methods. Other
techniques that are required, for example, in interferometry, spectroscopy, and holography, and that
also permit imaging metrology, are described in the following chapters.
implementation of suitable algorithms but rather much earlier on, that is, in the depiction of the
images. Particular attention must be given to the illumination of a scene. A type of illumination
well-adapted to the task in question produces images that can be analyzed using simple algorith-
mic methods. Correspondingly badly illuminated scenes may produce images from which even
the most refined algorithms are unable to extract the relevant features. It is also equally important
to take the dynamic behavior of imaging sensors into consideration. If illumination is too bright,
sensor elements may reach saturation range and produce overilluminated images, which could lead
to false conclusions. As far as metrological tasks are concerned, it is essential to understand the
imaging properties of lenses. For these reasons, the following sections are concerned with the most
important aspects of illumination, imaging sensors, camera technology, and the imaging properties
of lenses.
the refraction index n, the electromagnetic waves of the light are refracted at different degrees,
depending on their wavelength λ, and disintegrate into the colors of the rainbow. This is known as
dispersion.
If light refracted into the colors of the rainbow is merged again using a lens, white light results.
This observation leads to the phenomenon of mixing colors. If lights of different colors are mixed, the
colors are added together and form white. If, however, colored substances (e.g., pigments) illuminated
by white light are mixed, the colors are subtracted from one another and form black (Figure 1.5).
From this, the following questions arise: Why objects appear colored if they are illuminated by
white light? What are the effects of colored lights in conjunction with colored objects?
If an object appearing red to the human eye is illuminated by white light, the electromagnetic
waves of the spectrum corresponding to the color red are reflected and reach the eye. The light
from all other wavelengths is adsorbed by the object’s pigments and is transformed into warmth.
Naturally, the same also applies for objects made up of several colors, as shown in Figure 1.6.
As monochrome cameras are often applied in image processing, colored light can be used advan-
tageously to highlight colored objects. If, as shown in Figure 1.7, a red-and-green-colored object is
illuminated by green light, essentially only the green-colored areas of the object reflect the light,
leading to pale gray values in an image taken by a monochrome camera.
Today, light-emitting diodes (LEDs) are often used as colored light source in image
p rocessing if the area to be illuminated is not too large. Because of their specific color types,
LEDs cover the entire frequency range of visible light and the adjacent areas of near-IR and
UV, induce minimal loss of warmth, and can be switched on and off very quickly. However,
if intensive light sources are required, sources of white light such as halogen lamps are imple-
mented and combined with optical filters. The filters are generally made of colored glasses,
which selectively allow light of a certain wavelength to pass through and either adsorb or reflect
all other light wavelengths.
If a scene is illuminated with colored light but recorded with a monochrome camera, it often
makes sense in practice to place the filter directly in front of the camera lens rather than to filter
the light from the source of white light. In this way, any stray light from the environment that does
not possess the wavelength of the transmitted filter light is also filtered out and is therefore unable
to reach the camera’s sensors (see Figure 1.8).
In cases where the colors of an object are irrelevant, near-IR light sources are often used in
conjunction with a filter that transmits only this light. As a result, light conditions where the object
is illuminated are almost completely independent of visible light. If the IR light source is also
monochromatic, that is, only light from a narrow range of the spectrum is transmitted—or if the
filter is constructed as a narrow band pass (see figure 1.9)—the color distortions of the lens cause
fewer chromatic errors in the images recorded, which can be used advantageously, especially where
metrological tasks are concerned.
As far as the transmission of light is concerned, the behavior of optical filters is charac-
terized using spectral response plots. To do this, in general, the relative transmission of the
filter material is plotted as a function of the wavelength of the electromagnetic spectrum
(see Figure 1.9).
In the same way, it is also possible to characterize the spectral fractions of irradiated light from
a light source and the spectral sensitivity of light-sensitive sensors.
FIGURE 1.9 Spectral response plot of: (a) edge filters, (b) band pass filters, (c) light sources, and (d) CCD
sensor.
1.3.2 Illumination
The type of illumination plays a crucial role in image processing. By using an illumination adapted
to the special task in question, the features of a test object requiring measurement or testing can
often be highlighted better. Thus, the processing of images is simplified drastically in many cases.
Illumination can essentially be classified into two types in accordance with the arrangement of
the light source, test object, and image-recording system. If the test object is situated between the
light source and the image-recording system, this is known as a transmissive light arrangement.
Correspondingly, if the light source and the image-recording system are situated on the same side
as the test object, this is known as an incident light arrangement.
The simplest realization of a transmissive light arrangement is represented by a light panel
(see Figure 1.10), in which the light source (usually consisting of several lamps, tube lamps, or LEDs)
(a) (b)
FIGURE 1.10 Transmissive light arrangement: (a) principle of a light panel and (b) silhouette of an object.
Image Processing and Computer Vision for MEMS Testing 9
is positioned behind a ground-glass screen, which scatters light diffusely. This arrangement is
especially used to determine and measure the contours of flat, nontransparent test objects, because
the camera sees the shadowed image of the object. In cases where objects are translucent or trans-
parent, this arrangement enables internal structures to be recognized. However, the diffusely scat-
tered light from a ground-glass screen is disadvantageous if the dimensions of test objects are
particularly large in the direction of the optical axis of the camera. This is because surfaces that
are almost parallel to the axis could also be illuminated, thus falsifying the true silhouette of the test
object and the resulting measurement data obtained.
This disadvantage of diffusely scattered light can be reduced if an illumination system composed
of a concave lens and a spotlight source (see Figure 1.11)—a so-called collimator—is used instead
of ground-glass illumination. By placing the spotlight source in the focal point of the lens, the
light emerges as almost-parallel light rays. The LEDs (without lens optics) are especially useful as
light sources for this, because the light-emitting semiconductors are so small that they are almost
punctiform.
Despite the advantage of parallel light rays, illuminating collimators have the disadvantage of
producing inhomogeneous illumination. In accordance with the law regarding the decrease in light
intensity of a spotlight source, the light intensity of light rays passing through the lens decreases
from the optical axis toward the periphery, known as shading or vignetting. This effect, which is
disadvantageous as far as image processing is concerned, can be avoided if a homogeneously illu-
minated diffuse ground-glass light source, placed at the image level of a lens, is projected, instead
of using a spotlight source in the focal point of a collimator. If a telecentric lens (see Section 1.3.3)
is used instead of a normal lens to project the ground-glass light source, homogeneous illumination
with parallel light rays results.
As many MEMS or their components are based on silicon wafers, which are—according to their
nature—essentially two-dimensional in shape, the main type of illumination used in image process-
ing to test such elements is that of incident light arrangement. In contrast with transmissive light
arrangement, by using this type of illumination, both the test features of the object and all other
areas are equally illuminated. This results in the fact that features in the images being recorded are
often more difficult to differentiate in the subsequent image processing. From this point of view, a
type of illumination that is adapted to the task is of particular importance.
With incident light arrangements, an essential difference is made between dark field arrange-
ment and bright field arrangement, because they highlight very different aspects of a test object.
As far as the angle of observation is concerned, with dark field arrangement (see Figure 1.12),
the test objects are illuminated from an almost-perpendicular angle. In an ideal situation, as a result,
no or only a small amount of light falls onto the object surfaces, which are parallel to the angle of
(a) (b)
FIGURE 1.11 Transmissive light arrangement: (a) collimator and (b) projection of a homogeneous light
source.
10 Optical Inspection of Microsystems
(a)
(b)
FIGURE 1.12 Principles of dark-field illumination: (a) single sided and (b) circular light.
incidence and which are thus perpendicular to the angle of observation. These fields are seen by
the observer as being dark areas. In contrast, all other nontransparent object details are highlighted,
provided they are not positioned in the shadows of other object details. With nontransparent test
objects, dark-field illumination is therefore advantageous if raised sections of an object have to be
highlighted. Another example of using dark-field illumination to advantage is for detecting particles
on smooth, even surfaces. Transparent test objects such as glass bodies and plastics can also be
examined for inclusions or edge defects, as these features stand out well against a dark background,
especially when the latter is matt black in color. Generally, spot and linear, diffuse and directed
light sources are used for dark-field illumination. In many cases, low-angle glass fiber or LED ring
lights reduce the problem of shadow formation.
In contrast with dark field arrangement, in the case of bright field arrangement test, objects are
illuminated almost from the angle of observation. In simple cases, the light source for bright field
arrangement is realized using either one or several spotlight sources, high-angle glass fiber lights,
or LED ring lights placed around the camera lens. Using this illumination arrangement, a good dif-
ferentiation of features can be obtained where object surfaces are diffusely dispersive, provided the
colors of the features to be tested contrast well against their surroundings, such as in the case of dark
print markings and barcodes on pale or colored markings.
The bright field arrangement described earlier becomes problematic if object surfaces are shiny
or reflective. This is because the light sources may be projected in the recording camera system in
the form of reflections, thus impeding or preventing reliable feature analysis. Homogeneous illu-
mination can be achieved in these cases if all visible sides of an object are illuminated equally by
a diffuse light source. The light dome (half sphere) shown in Figure 1.13 is capable of doing this.
Light sources fixed at the edge of the light dome illuminate its internal surface, which is matt and
coated white.
In the case of flat, shiny surfaces, reflex-free bright-field illumination can be achieved if the
light is coupled coaxially to the angle of observation. As shown in Figure 1.14a, diffuse light from
a homogeneously illuminated surface—such as that used in transmissive light arrangements—is
deflected by 90° toward the angle of observation by using a semitransparent mirror. Owing to the
Image Processing and Computer Vision for MEMS Testing 11
(a)
(b)
FIGURE 1.13 Bright-field illumination: (a) spotlight source and (b) light dome.
(a)
(b)
FIGURE 1.14 Bright-field illumination: (a) coaxially diffuse and (b) coaxially directed.
homogeneity of the light source and the evenness of the object surface, object areas possessing
an equal degree of reflection are depicted in the camera image with the same level of brightness,
making this a reliable method for imaging and analyzing features on such surfaces.
A special form of coaxial bright-field illumination is utilized for telecentric lenses
(see Section 1.3.3). As shown in Figure 1.14b, light from a spotlight source is coupled to the beam
path of the lens by using a semitransparent mirror. Ideally, the light emitted from the lens is made up
purely of parallel light rays, which are reflected toward the telecentric lens from object surfaces only
12 Optical Inspection of Microsystems
if those surfaces are perpendicular to the optical axis. The properties of telecentric lenses prevent
the light of all other surfaces from reaching the camera, with the result that these appear as dark
areas in the recorded image. This feature can be used to make the edges of flat, three-dimensional
structures visible.
Besides the types of illumination described here, other forms of illumination can also be imple-
mented in image processing. These include types of structured illumination such as fringe projec-
tion and laser scanning for recording and measuring objects in three dimensions (see Chapter 6).
1 1 1
= + (1.1)
f i o
As can be seen in Figure 1.15a, the geometric construction of the image Im results from the object O
as follows: a line is drawn from each object point through the center of the lens; a second line paral-
lel to the optical axis is bisected with the principal plane of the lens; and starting at this intersection,
a third line is drawn through the focal point of the lens. The intersection of the first and third line
then gives the corresponding image point.
As shown in Figure 1.15b, it can be seen from the lens equation that both the position and the size
of the image Im alter if the object is moved along the optical axis. In the case of lenses, in order to
depict the object in focus, the displacement of the image Im needs to be corrected. This is achieved
by moving the lens along the optical axis, so that the focal plane (where the imaging sensors of the
camera are situated) is always in the same place. Owing to the limitation of the depth of field, it can
be deduced directly that the objects that are particularly large in the direction of the optical axis and
that are close to the lens cannot be depicted completely in focus. To be more precise, this results in
part of the image to be in focus and the surrounding areas to be slightly out of focus. This charac-
teristic may considerably impair metrological image analysis or even render it impossible.
The cause of this blurred imaging can be explained by the following observation (see Figure 1.16a).
Each point of the object O emits light homogeneously in all free directions in the form of spherical
waves. Part of the light reaches the lens, and this then depicts it as an image point in the zone of
sharp focus. Circles of confusion are formed outside the zone of sharpness instead of image points,
thus resulting in a blurred image.
(a) (b)
FIGURE 1.15 Imaging behavior of a thin lens: (a) geometric construction of the image of an object, and
(b) if an object moves toward a lens, the image moves away from the lens and gets enlarged.
Image Processing and Computer Vision for MEMS Testing 13
(a) (b)
FIGURE 1.16 (a) Formation of image points from object points and (b) telecentric imaging.
As the geometric construction shows, the light rays running parallel to the optical axis possess
two characteristic features. Independent of the distance between the object and the lens, they always
run parallel, and when they are projected, they always pass through the focal point F of the lens.
This results in the concept of generating a sharp image, independent of distance, by utilizing a
screen with a small aperture at the height of the focal point (focal pupil) in the beam bath of the lens
(see Figure 1.16b). Using this measure, a so-called object-sided telecentric lens is created.
As illustrated in Figure 1.17, telecentric lenses possess the following properties:
1. The field of view is equal to the size of the lens, because light rays running parallel to the
optical axis are not imaged outside the lens area.
2. The imaging of an object O (see Figure 1.17a) always results in (within limits) a sharp
image Im, independent of the distance between the object and the lens. This characteris-
tic is limited by the finite dimensions of the focal pupil, because almost no light passes
through an infinitely small point.
3. For all distances from the lens, the object is imaged with the same magnification as far as
the focal plane remains fixed.
4. Owing to the focal pupil, telecentric imaging is of low light intensity.
5. The displacement of the focal plane (see Figure 1.17b) along the optical axis alters the size
of an image but not its sharpness.
(a)
(b)
(c)
FIGURE 1.17 Behavior of one- and double-sided telecentric lenses: (a) within limits, object movement
always result in a sharp image; (b) within limits, movement of the focal plane changes size of image but
not sharpness; and (c) in case of double-sided telecentric lenses, sharpness and size of image (within limits) do
not alter if focal plane moves along the optical axis.
14 Optical Inspection of Microsystems
This characteristic predestines telecentric lenses to being used for tasks where precise metrological
image analysis is required.
The transition from an object- to a double-sided telecentric lens is achieved by introducing a
second lens behind the focal pupil (see Figure 1.17c) in such a way that the focal points of the two
lenses coincide. As a result, the focal rays are imaged into rays running parallel to the optical axis,
and subsequently, the alteration in image size, which takes place when the focal plane is moved
along the optical axis, no longer occurs. Double-sided telecentric lenses can be used to a dvantage
if imaging sensors are equipped with microlenses (one lens per pixel) to give a better light yield.
In this case, light entering obliquely may lead to vignetting, thus resulting in inhomogeneous
grayscale value distributions in the images recorded.
1.3.4 Sensors
Imaging sensors built into today’s cameras are based on semiconductor materials. The formation of
“electronic” images by using such sensors utilizes the principle of the internal photo effect. With
this principle, light quanta (photons) possessing a minimum material-specific energy release atomic
electrons from their bonded state (see Figure 1.18a). However, in the process, the freed electrons
remain in the material.
The external photo effect describes the process where electrons are emitted from a material.
Photomultipliers (not discussed here) are based on this effect, for example.
The light-sensitive elements of imaging sensors based on semiconductor materials are con-
structed as photodiodes and use the depletion-layer photo effect. The functioning principle of such
diodes is explained briefly below.
A silicone substrate is p-doped with boron atoms; owing to the three free valence electrons
of boron, there are insufficient electrons to bond with the four valance electrons of silicone. The
substrate is then coated with a thin layer of silicone n-doped with fluoride. As fluoride atoms pos-
sess five free valence electrons—that is, one electron more than necessary in order to bond with
silicone—some of the free surplus electrons fill in the missing electrons (holes) in the p-substrate
of the junction zone (see Figure 1.18b). Through this recombination, the p-substrate in the junction
zone becomes negatively charged and the n-doped silicon layer positively charged, thus preventing
additional electrons from migrating from the n-doped silicone to the p-doped silicone. The junction
zone is thus transformed into a depletion layer.
(a) (b)
FIGURE 1.18 Illustration of the principle of (a) the photo effect and (b) of the photodiode.
Another random document with
no related content on Scribd:
Fig. 18.—Modern.
Fig. 19.—Typical Grain-Handling Plant; 50 Tons per hour.
The writer was directly interested in the erection and installing of one
of the first plants installed in this country for the elevation and
conveying of coal, and a description of the various details may give a
good idea of a complete plant, handling coal on a commercial scale.
The conditions to be complied with are as follows: 20 tons of
“slack” per hour, to be raised 90 ft. above canal level or 80 ft. above
road level.
The coal is brought alongside the power-house by canal barges of
25 tons capacity, or by tipping steam waggons from the railway
sidings, a distance of one mile away. In both cases the coal is
required to be elevated into overhead bunkers of 600 tons capacity
placed vertically over the boilers.
Fig. 23.—Pneumatic Unloading of Coal at Messrs. Boots, Ltd.
(Nottingham).
The main discharger valves are driven by worm gearing, the latter
having one right-hand thread and one left-hand thread, so that the
end thrust on the worms is neutralized. Ball bearings are provided
and the small motor which drives all three valves is coupled up with
an electrical device designed by the author. This device ensures that
if anything happens to the top discharge valves—so that the 3 h.p.
motor driving them cuts out, owing to an overload or other cause—
then the main motor also is cut out by the opening of its circuit
breaker. This prevents any “flooding” of the pipes and dischargers.
It may be mentioned that the valves are so designed that a portion
of the weight of each valve is carried by the vacuum, so that the
vertical wearing lift on the valves when at work is very slight.
The intake pipes for the coal are 5 ins. diameter, and they are
provided with heavy cast iron bends, having extra thick metal on the
outside radius to allow for the wearing effect of coal passing at the
rate of 20 tons per hour.
The pipe into the barge is provided with a flexible steel pipe at the
suction nozzle end, for convenience of handling. India-rubber piping
has been tried, but the extra cost does not justify its continued use.
The nozzle is made as light as possible for convenience of
handling, and is fitted with a special “free air” inlet for the regulation
of the amount of air necessary to blend with the coal.
Ash Handling. In addition to unloading coal, the above plant is
capable of dealing with hot ashes which are first crushed in a
portable clinker breaker, electrically driven, which runs under all the
ash hoppers of the boilers. The ash when crushed gravitates into
funnel-topped tee-pieces, inserted in the main ash-conveying pipe,
whence it is immediately sucked up into an overhead ash hopper to
await the convenience of the waggons which dispose of it on the
“tips.”
Flue Cleaning. A 3 in. suction pipe has been run round the boiler-
house in such positions that flexible hose can be attached for flue
cleaning purposes. In this case the cleaners simply use an enlarged
nozzle such as is supplied with a domestic equipment and the dust is
removed from the flues, economizer soot chambers, etc., into the
ash hopper without trouble or dust.
The success of this plant is best indicated by the fact that, at the
moment of writing, a duplicate plant is being erected. Owing to the
growth of the business, and its demand for power and steam, the
original plant has to be worked continuously on coal, so that the ash
and flue dust problem has become acute again.
Portable Floating Plant. A third plant ordered by the same firm is
of considerable interest. This is intended to be mounted in a barge
so as to be portable. Owing to lack of space in close proximity to the
power-house, considerable difficulty is found in keeping adequate
stocks of coal on the site except the 600 tons in the overhead
bunkers. In order to secure continuity of working, it is essential that
as much fuel as possible be stored, and for this purpose a coal pile
has been made about half a mile away from the works, adjoining the
canal. Ashes can be disposed of on certain fields a few miles outside
the city in swamps and pools caused by subsidences, due to colliery
workings.
The portable plant is therefore arranged to operate as follows: the
barge is self-propelled by a 30 h.p. paraffin engine which can be
coupled by clutches to either the propeller or a Roots blower, the
latter being the exhauster for the portable suction plant.
The barge is loaded with ashes for disposal, and then proceeds
under its own power to the site where they are to be dumped. The
clutch is operated disconnecting the propeller and operating the
blower. The suction side of the blower is coupled up with the pipe
line in the boat and the barge feeds the plant by means of the
flexible hose: the discharge pipe is raised over the towing path so as
not to interfere with passing traffic, and the ashes are blown out into
the swamps previously mentioned. It will readily be recognized how
simple this unloading becomes compared with trying to dig out the
ashes with either a spade or a fork.
The empty barge then returns to the coal pile and takes up a load
of coal in a similar manner, then proceeding to the power-house
under its own power and being unloaded by the original fixed
pneumatic installation in the ordinary way.
The coal arriving by road is tipped into a concrete hopper
excavated below the ground level, and so designed with sloping
sides that it is self feeding into a suction pipe connected to the
bottom of the hopper. The same procedure occurs except that in this
case the coal enters the main discharger at the top (E, Fig. 24).
It is interesting to note that the very fine dust collected from the air
filter is eagerly sought after by the foundry trade, and what would at
first appear to be a waste product impossible to burn, is actually a
valuable by-product of the plant.
CHAPTER VI
THE INDUCTION CONVEYOR