Professional Documents
Culture Documents
Bachelor of Engineering
2013
Abstract
Comment [B3]: Leave only a line of 12point blank space between consecutive
paragraphs. Ok
electronics has created an array of new and difficult problems and demands for the field of
Wafer dicing. During wafer dicing saw, broken silicon wafer may have a possibility to go out
of the process without any interruption. Because of this, it drives the researcher to come up
with the engineering long term solution to improve the wafer dicing saw process by introducing
a blade breakage detector for semi-automatic dicing saw machine to prevent possibilities of a
broken silicon wafer to come out during wafer dicing saw process.
Designing, development and construction of blade breakage detector prevents the potential
wafer dicing saw quality issues such as broken silicon wafers and chuck damage. It also helps
to minimise approximately US$96,000 per annual to the company from corrective maintenance
and resource management budget.
ii
Disclaimer
I hereby declare that this thesis is my own work and contains no material which has been
accepted for the award of any degree or diploma from any tertiary institution. To the best of
my knowledge and belief, this thesis contains no material previously written or published by
another person, except where due reference is made in the text.
Signed:
Akash Anil
22 July 2013
iii
Acknowledgements
I would like to express my gratitude to my principal supervisor Professor Kazem Abhary for
his overall guidance throughout my project. His encouragement and willingness to share his
wisdom have been essential during each stage of the research and writing of this thesis. I
would also like to thank my industry supervisor Mark Anthony Siazon for his unselfishness
to provide his insightful comments and suggestions during the whole project process. There is
no doubt this work could not have completed without the observations, comments and
discussions contributed by my two supervisors.
I also sincerely acknowledge the assistance provided by the following individuals:
Mr Mizan Jefree, our multimedia personnel, for his help in providing some of the
valuable photos on the company and the connector itself. His expertise has helped me
in providing a clearer picture to the reader in this whole project thesis.
My Brother, Ashish Anil, for his comments on the presentation of the project thesis.
This had helped me in providing a clearer and more consistent project presentation.
I would like to thank the whole design team members for contributing their valuable
suggestions and comments on the blade breakage detector design
iv
Table of Contents
Title page
Abstract
ii
Disclaimer
iii
Acknowledgments
iv
Table of Contents
List of Figures
vii
List of Tables
viii
Appendices
ix
2
2
4
7
7
7
8
9
9
9
9
9
11
12
14
14
15
17
19
20
21
22
23
24
24
25
26
27
28
29
29
29
30
31
35
35
35
35
36
36
37
37
38
38
38
39
40
41
43
44
44
44
44
45
46
46
48
48
50
50
51
51
52
52
52
53
53
53
54
56
57
58
59
60
61
61
62
vii
List of Figures
Chapter 1 - Project Background and Significance
Figure 1.1 Portion of a Diced Wafer after Cleaning
Figure 1.2 Typical Wafer Dicing Saw Blade Holder Assembly
Figure 1.3 Damage Chuck Table
Figure 1.4 Excessive Cut
Figure 1.5 Damaged Wafer
4
5
6
6
6
12
15
16
28
34
34
35
35
36
36
37
38
40
41
42
47
48
50
51
53
54
55
57
viii
List of Table
Chapter 1 - Project Background and Significance
Chapter 2 - Literature Review and Project Methodology
Table 2.1 ANSI Recommended Allowances and Tolerances
Table 2.2 Table of Preferred Sizes
Table 2.3 Gantt chart
13
15
31
44
56
58
60
61
62
ix
Appendices
Appendix 1
1.1 Introduction
This chapter will give a high level detail of the project "The Design and Development of a
Blade Breakage Detector for Semi-Automatic Dicing Saw Machine". It discusses why the
system capability is needed and its importance to the proponent. It shows the background of
the study, outlining the scenarios, aims and scope.
1.2 Company Background
GLOBALFOUNDRIES (Singapore) is the worlds first full-service semiconductor foundry
with a truly global manufacturing and technology footprint. Launched in March 2009 through
a partnership between Advanced Micro Devices (AMD) [NYSE: AMD] and the Advanced
Technology Investment Company (ATIC), GLOBALFOUNDRIES provides a unique
combination of advanced technology, manufacturing excellence and global operations. With
the integration of Chartered Semiconductor in January 2010, GLOBALFOUNDRIES
significantly expanded its capacity and ability to provide best-in-class foundry services from
mainstream to the leading edge.
GLOBALFONDRIES manufacturing operations are based in Singapore, Germany and it is
headquartered in Silicon Valley. A leading-edge fabrication is currently under construction in
Saratoga County, New York. These sites are supported by a global network of R&D, design
enablement, and customer support in Singapore, China, Taiwan, Japan, the United States,
Germany, and the United Kingdom.
The proponent of this thesis is currently employed in GLOBALFOUNDRIES Singapore as a
Senior Associate Engineer. Reporting under the Equipment Engineering Team, the
proponents major job function is providing sustained equipment health throughout the shift.
The proponent also provides engineering support, such as machine qualifications and
solutions to issues affecting the production floor within the confines of the process.
Wafer fabrication is a meticulous procedure composed of many sequential processes which
involves engineers metallurgists, chemists and physicists to produce a complete electrical or
photonic circuit. The term wafer fabrication generally refers to the process of fabricating
integrated electrical circuits/components on silicon wafers. Prior to the fabrication of wafer,
pure silicon ingots are used to produce raw silicon wafers, which then sliced and shaped into
thin wafers through a process wafering.
The crucial step in wafer fabrication is the designing of the electrical circuits by defining its
function, specifying the input-output signals and voltages. These circuit speculations are
entered into electrical circuit design software, such as simulation program with integrated
circuit emphasis (SPICE), and then imported into circuit layout programs. SPICE is a
powerful program used in integrated circuit and board level design to check the integrity of
the circuit design and to predict the behavior of the circuit. For wafer mask production it is an
essential layer to be defined. The circuits resolution increases rapidly with the step in design,
as the scale of the circuits at the start of the design process is already being measured in
fractions of micrometers. Each step thus increases circuit density for a given area.
The silicon wafers start out blank and pure. The circuits are built in layers in clean rooms.
First, photo sensitive resistance patterns are photo masked in micrometer detail onto the
wafers surface. The wafers then exposed to short wave ultraviolet light and the unexposed
areas thus etched away and cleaned. Hot chemical vapors are deposited on to the desired
zones and baked in high heat, which permeate the vapors into the desired zones. In some
cases, ions, such as O2+ or O+, are implanted in precise patterns and at a specific depth by
using radio frequency driven ion sources. These steps are often repeated many hundreds of
times, depending on the complexity of the desired circuit and its connections.
As the worlds largest chip manufacturer, GLOBALFOUNDRIES strives to make every facet
or semiconductor manufacturing state of the art from semiconductor process development and
manufacturing, through yield improvement to final test and optimization, and lastly
packaging. Employees in the technology and manufacturing group are part of a worldwide
network of manufacturing, assembly and test facilities that is committed to its goal of
producing world class microchips that is unsurpassed in quality and performance and the
companys capability to deliver the market needs for high quality microchips.
With the volume demand increasing in microchips particularly in Asia, this brings a challenge
to the company to have a continuous capacity to produce high quality and reliable product
based from the specifications communicated internally within GLOBALFOUNDRIES and to
its customers. Quality standard and improvement system must be implemented in able to do
this and ultimately improve its business processes to facilitate labour intensive task that grows
proportionally with the volume being produced by its factory. These improvements can be
done either through breakthrough systems or by utilizing the capabilities of Computer
Technology simply through automation.
performance parameters have been optimized for silicon. In order to cut hard and brittle
materials it is necessary to have different blades and equipments, which operates in different
parameters. It is a crucial step/key as this leads in successful high yield dicing process. Tool
sharpness is the primary importance in a cutting process and it is unavoidable to maintain the
sharpness of the tool throughout the cutting operation. When cutting hard material such as
sapphire, the cutting edges become dull quite rapidly. Because the dulled cutting edges cannot
be re-sharpened in the usual manner, it is desirable that they be pulled loose from the blade, or
else be fractured to expose new sharp cutting edges (David E. Bajune, 2000).
To study the application and robustness of various sensor types for detecting the broken
blade
The Proponent - This study will help the proponent to gain more knowledge, additional
skills and information that would help him in his future career. It would be an honour on the
part of the proponent when the system design is fully implemented eventually.
For Other Researchers - This study will help the other researchers to gain more knowledge,
additional skills and information that would help them to improve their manufacturing
processes.
This chapter mentioned that the main purpose of this project, "The Design and Development
of a Blade Breakage Detector for Semi-Automatic Dicing Saw Machine". Thus, it will include
analysis, design, development and evaluation of the mentioned capability in an effort to
improve wafer dicing saw process to the benefit of GLOBALFOUNDRIES and its customers.
8
2.1 Introduction
This portion of the research will present literatures and studies related to the concept of the
project being proposed. The purpose of such is to be able to give the reader a wider view and
better understanding of the topic being discussed and provide a background on how it affected
"The Design and Development of a Blade Breakage Detector for Semi-Automatic Dicing Saw
Machine.
2.2 Source of Literature
Different data gathering tools and instruments are used by the proponent to gather information
about the existing system. These tools are needed in creation of the proposed system of the
study.
Internet Research - The researcher consulted several websites during project development
and understanding the system capability.
Library Research - Reference material and books available in the library were consulted
to provide more backbone into the programming structure and add robustness into the
new system.
Questionnaires - Several questionnaire sheets were made to help in assessing the usability
and functionality of the existing and new system design based on the answers given by the
respondents. The Nielsens attribute of usability evaluation approach was also employed
to look at the overall health of the existing and new procedure. This was chosen because of
its proven effectiveness in assessing design usability.
10
surface finish. Careful attention to those aspects of engineering drawing can greatly
improve the cost and quality of a design.
2.3.2.1 Tolerances
Tolerances must be placed on the dimensions of a part to limit the permissible
variations in size because it is impossible to manufacture a part exactly to given
dimensions. A small tolerance results in greater ease of inter-changeability of parts,
but it also greatly adds to the cost of manufacture (Figure 2.1).
. This system
is being replaced by the unilateral tolerance, in which the deviation in one direction
from the basic dimension is given. For example,
or
In the case of bilateral tolerance, the dimension of the part would be permitted to vary
between 1.996 and 2.004 in for a total tolerance of 0.008 inch. If unilateral tolerance is
specified, the dimension could vary between 2.000 and 2.008, and again the total
tolerance is 0.008 inch. Unilateral tolerances have the advantages that they are easier
to check on drawings and that a tolerance can be changed with the minimal
disturbance to other dimensions. The American National Standards Institute (ANSI)
has established eight classes of fit that specify the amount of allowance and the
tolerance on the hole and a mating shaft, see Table 2.1.
13
The ANSI system considers that the hole size d is the basic dimension, because most
holes are produced by using standard-size drills and reamers. Therefore, the shaft can
be more easily produced to a non-standard dimension. Consider a basic hole size of
2.000 inch and a class 3 (medium) fit (Englewood Cliff, 1978).
Allowance: 0.0009 22 = 0.0014 inch
Tolerance: 0.0008 2 = 0.0010 inch
Hole:
Maximum dimension 2.001 inch
Minimum dimension 2.000 inch
Shaft:
Maximum dimension 2.000 - .00014 = 1.9986 inch
Minimum dimension 1.9986 - 0.001 = 1.9976 inch
Therefore, the maximum clearance between shaft and hole is
2.0010 - 1.9976 = 0.0034 inch
And the minimum clearance between shaft hole is
2.00 - 1.9986 = 0.0014 inch
2.3.2.2 Dimensions
The engineering drawing provides the manufacturing department with the information
necessary for producing the part. Therefore, it is important that the dimensions of the
14
part be clear and complete. The dimensions given should be sufficient in number to
make it unnecessary for shop personnel to perform involved calculations for setting up
the production equipment. On the other hand, too many dimensions can cause
problems by resulting in ambiguity and leaving the manufacturing department with a
choice. Another difficult situation arises when successive points on a drawing must be
dimensioned. Note that it is possible for production parts manufactured within
tolerance to not fit because of interference due to accumulation of tolerances.
However, if all dimensions start at a datum line, all parts made to within tolerance will
assemble properly and the difficulty with interference is eliminated.
2.3.2.3 Preferred Sizes
Standard components such as bolts, bearings and electric motors need to be made
according to a rational scheme by which the size (or weight, speed, power etc.) covers
the usual range of needs. A geometric rather that arithmetic progression of size is most
logical. Each size is larger than the preceding size by a fixed percentage. At the small
size end of the range there will be more items than at the large size end. From an
economic standpoint, the number of standardized sizes should be kept to the smallest
number that will provide for the desired range of applications.
Table 2.2 Table of Preferred Sizes
15
2. Unlock your imagination. You must rekindle the vivid imagination you had as a child.
One way to do so is to begin to question again. Ask why and what if, even at the
risk of displaying a bit of navet. Scholars of the creative process have developed
thought games that are designed to provide practice in unlocking your imagination and
sharpening your power of observation (E. Raudsepp, 1982).
3. Be persistent. We already have dispelled the myth that creativity occurs with a
lightning strike. On the contrary, it often requires hard work. Most problems will not
succumb to the first attack. They must be pursued with persistence. After all, Edison
tested over 6000 materials before he discovered the species of bamboo that acted as a
successful filament for the incandescent light bulb. It was also Edison who made the
famous comment, Invention is 95 percent perspiration and 5 percent inspiration.
4. Develop an open mind. Having an open mind means being receptive to ideas from any
and all sources. The solutions to problems are not the property of a particular
discipline, nor is there any rule that solutions can come only from persons with college
degrees. Ideally, problem solutions should not be concerned with company politics.
Because of the NIH factor (not invented here) many creative ideas are not picked up
and followed through.
5. Suspend your judgment. We have seen that creative ideas slowly, but nothing inhibits
the creative process more than critical judgment of an emerging idea. Engineers by
nature, tend toward critical attitudes, so special forbearance is required to avoid
judgment at an early age.
6. Set problem boundaries. We place great emphasis on proper problem definition as a
step toward problem solution. Establishing the boundaries of the problem is an
essential part of problem definition. Experience shows that this does not limit
creativity, but rather focuses it.
A creative experience often occurs when the individual is not expecting it and after period
when they have been thinking about something else. The secret to creativity is to fill the
mind and imagination with the context of the problem and then relax and think of
something else. As you read or play a game there is a release of mental energy which your
preconscious can use to work on the problem. Frequently there will be a creative Ah-ha
experience in which the preconscious will hand up into your conscious mind a picture of
what the solution might be. Since the preconscious has no vocabulary the communication
between the conscious and preconscious will be by pictures or symbols. This is why it is
important for engineers to be able to communicate effectively through three-dimensional
sketches.
19
2.3.4 Invention
An invention is something novel and useful. As such, we generally can consider it to be
the result to creative thought. A study of a large number of inventions (G. Kivenson,
1977) showed that inventions can be classified into seven categories:
1. The simple or multiple combinations. The most elementary form of invention is a
simple combination of two existing inventions to produce a new or improved result.
2. Labour-saving concept. This is a higher level of invention sophistication in which an
existing process or mechanism is changed in order to save effort, produce more with
the same effort, or dispense with a human operator.
3. Direct solution to a problem. This category of invention is more typical of what we
can consider to be engineering problem solving. The inventor is confronted with a
need and sets out deliberately to design a system that will satisfy the need.
4. Adaptation of an old principle to an old problem to achieve a new result. This is a
variation of category 3. The problem (need) has been in existence for some time, and
the principle of science or engineering that is key to its solution also has been known.
The creative step consists in bringing the proper scientific principle to bear on the
particular problem so as to achieve the useful result.
5. Application of a new principle to an old problem. A problem is rarely solved for all
time; instead, its solution is based on the then current limitations of knowledge. As
knowledge (new principles) becomes available, its application to old problems may
achieve startling results. As an example, the miniaturization of electronic and
computer components is creating a revolution in many areas of technology.
6. Application of a new principle to a new use. People who are broadly knowledgeable
about new scientific and engineering discoveries often are able to apply new principles
in completely different disciplinary areas or areas of technology.
7. Serendipity. The mythology of invention is full of stories about accidental discoveries
that led to great inventions. Lucky breaks do occur, but they are rare. Also, they hardly
ever happen to someone who is not already actively pursuing the solution of a
problem. Strokes of good fortune seem to be of two types. The first occurs when the
inventor is actively engaged in solving a problem but is stymied until a freak
occurrence or chance observation provides the needed answer. The second occurs
when an inventor suddenly gains a valuable insight or discovers a new principle that is
not related to the problem he is pursuing. He then applies the discovery to a new
problem and the result is highly successful.
20
supplied and returned via glass fiber optic cables. Glass fiber exhibits very good optical
qualities and typically carries high temperature ratings. Plastic fiber can be cut to length in
the field and can be flexible enough to accommodate various routing configurations.
Important parameters to consider when specifying fiber optic proximity sensors include
detecting range, position measurement window, minimum detectable object, and response
time. The detecting range is the range of sensor detection. For presence sensors, this goes up
to the maximum distance for which the signal is stable. For position measurement sensors,
this is the distance range over which the position versus output response is linear and stable.
The position measurement window is the width of linear region for the sensor. For example,
if the sensor could measure between 14 and 24 centimeters, this window would be 10
centimeters. The minimum detectable object is the smallest sized object detectable by the
sensor. The response time is the time from target object entering detection zone to the
production of the detection signal.
Choices for optical types for fiber optic proximity sensors include through beam, retroreflective, polarized retro-reflective, diffuse, divergent, convergent, fixed field background
suppression, and adjustable field background suppression. Through beam (or opposed
mode) sensors incorporate a transmitter and a receiver on opposite sides of the target and
evaluate absence or presence based on transmittance or blockage of the beam respectively.
Retro-reflective sensors incorporate the emitter and detector in one body with parallel
beams; a reflector opposite the target path is used to reflect the emitted beam back to
detector; presence is detected when reflected beam is interrupted by an opaque object.
Polarized retro-reflective sensors are variants of the retro-reflective in which the emitter and
receiver have polarizing filters 90 degrees apart, thereby making the reflector the only
recognizable source. With a diffuse sensor presence is detected when any portion of the
diffuse reflected signal bounces back from the detected object. Divergent beam sensors are
short-range diffuse-type sensors without any collimating lenses. Convergent, fixed focus, or
fixed distance optics (such as lenses) are used to focus the emitter beam at a fixed distance
from the sensor. Fixed-field sensors are designed to have a distance limit beyond which they
will not detect objects, no matter how reflective. Adjustable field sensors utilize a cut-off
distance beyond which a target will not be detected, even if it is more reflective than the
target. Some photoelectric sensors can be set for multiple different optical sensing modes.
Reflective properties of the target and environment are important considerations in the
choice and use of photoelectric sensors.
23
Common configuration features for photoelectric sensors include beam visibility, light-on or
dark-on modes, light and dark programmability, adjustable sensitivity, self teaching, and
laser source. The body style of the sensor can be threaded barrel, cylindrical, limit switch,
rectangular, slot, ring, and window or frame. The sensor may be self-contained and may
have a remote head (www.globalspec.com, 2005).
2.3.8 Light Emitting Diode
As stated by Luis G. Sison (2008), in the book: Introduction to Semiconductor Devices and
Circuits that a diode made of indirect semiconductors, such as silicon and germanium, the
excess minority carrier concentrations present in forward-bias undergo thermal
recombination. When the diode is made of direct semiconductors, such as gallium arsenide
and indium phosphate, the same excess carriers can undergo recombination that results in
photoemission. The result is a light emitting diode. The wavelength of the emitted light
depends on the band gap of the semiconductor. The wavelength varies from the visible
wavelengths (0.4 to 0.7 microns) to the infrared wavelengths (> 0.7 microns). Mixing
different semiconductor elements and introducing impurities leads to different band gaps.
A light emitting diode (LED) is a semiconductor device that emits incoherent, narrow
spectrum light when it is electrically biased in the forward direction. This effect is a form of
electroluminescence. Light emitting diodes are small extended sources with extra optics
added to the chip, which emit a complex intensity spatial distribution. The emitted light may
be in the infrared, visible, or near-ultraviolet region of the spectrum, depending on the
composition and condition of the semiconducting material used. Light emitting diodes have
a broad range of uses (www.newworldencyclopedia.org, 2006). Many are used for a variety
of signs and signals, such as traffic lights, motorcycle lights, light bars on emergency
vehicles, elevator push-button lights, status lights on electronic equipment, exit signs, and
other message displays. Different types of light emitting diodes are used for architectural
lighting and Christmas lights. Infrared light emitting diodes are used in remote controls for
televisions and video camera recorders. Some light emitting diodes are used for
phototherapy for acne.
2.3.9 Amplifier
Generally, an amplifier or simply amp is any device that changes, usually increases, the
amplitude of a signal. The relationship of the input to the output of an amplifier - usually
expressed as a function of the input frequency - is called the transfer function of the
24
amplifier, and the magnitude of the transfer function is termed the gain. In popular use, the
term usually describes an electronic amplifier, in which the input "signal" is usually voltage
or current. In audio applications, amplifiers operate loudspeakers used in public address
(PA) systems to make the human voice louder or play recorded music. Amplifiers may be
classified according to the input (source) they are designed to amplify (such as a guitar
amplifier, to perform with an electric guitar), the device they are intended to drive (such as a
headphone amplifier), the frequency range of the signals (audio, intermediate frequency,
radio frequency, and very high frequency amplifiers, for example), whether they invert the
signal (inverting amplifiers and non-inverting amplifiers), or the type of device used in the
amplification like valve or tube amplifiers, field effect transistors amplifiers, etc.
(www.wikipedia.org, 2004).
2.3.9.1 Operational Amplifiers
Operational amplifiers (op amps) are linear integrated circuits that can be used to
amplify signal frequencies that extend from 0 hertz (direct current) to well above 1
mega hertz. Operational amplifiers have two input terminals and one output terminal.
One of the most important qualities of an operational amplifier is that it amplifies only
the difference between its two input signals, while rejecting or severely attenuating
signals common to both inputs. This allows operational amplifiers to be used in systems
where a large amount of electrical noise is present. In this case, the desired signal is
amplified, while the noise common to both inputs is attenuated.
According to Mitchel E. Schultz (2008), operational amplifiers (op amps) are the most
commonly used type of linear integrated circuit (IC). By definition, an operational
amplifier is a high-gain, direct coupled, differential amplifier. An operational amplifier
referred to as the 741 has become an industry standard. This operational amplifier,
which is contained in an eight pin integrated circuit, is made by several manufacturers.
They are, however, all equivalent since the specifications are nearly identical from one
manufacturer to another.
2.3.9.2 Operational Amplifier Applications
Although the following is certainly not an exhaustive list, it does serve to illustrate the
range of operational amplifier applications:
1. Amplifiers: Operational amplifiers are used to amplify signals that range from direct
current (DC) through the higher radio frequencies (RF). The amplifier can be made
25
channel from the process; it is used to time-share the analog-to-digital converter among
the input channels. The alternative is to have a separate analog-to-digital converter for
each input channel, which would be costly for a large application with many input
channels. Since the process variables need only sampled periodically, using a
multiplexer provides a cost-effective alternative to dedicated analog-to-digital
converters for each channel.
4. Amplifier. Amplifiers are used to scale the incoming signal up or down to be compatible
converter is to convert the incoming analog signal into its digital counterpart.
2.3.11 Wafer Dicing
Wafer dicing is the process by which the individual chips or integrated circuits on a silicon
wafer are separated. A scribe line is an area between the die left empty of a circuitry, where
the dicing saw can pass through, thereby separating the dies (chips) from a wafer. Scribe
line is otherwise referred as street, kerfs, and saw-kerfs. The space should be enough to
allow separations of the die by cutting or breaking without causing any damage to the die.
The wafer saw will be able to cut in a straight line. The dicing process can be accomplished
by scribing and breaking, by mechanical sawing (normally with a machine called a Dicing
Saw) or by laser cutting. After the dicing process the individual silicon chips are
encapsulated in the packages.
27
According to Peter McCabe (2005), the first step in the dicing process is an evaluation of
wafer thickness, street width, and the material, which might be silicon, silicon on sapphire,
silicon germanium, or more exotic materials. The evaluation helps in choosing the best
blade. A standard silicon wafer (20 mils thick and 4 mils streets), for example, can be sawn
with a 1 mil thick blade having a 2 to 4 microns grit and a 30 mils exposure. After the
evaluation and blade choice, the wafer is mounted with the adhesive film onto a metal wafer
cutting ring. First, a disk of particle free paper is centred on the mounting chuck to protect
the wafers face. Next, the wafer is positioned face down on the paper. Then, the wafer tape
is stretched, sticky side down, over the back of the wafer and the cutting ring, and pressed
with a roller to smoothly and evenly distribute the tape. Excess tape is cut away from the
edge of the ring.
The ring mounted wafer is loaded on the saw-dicing chuck for programming. The X and Y
step centres are measured and stored. Following the theta angular alignment, a sample cut is
made near the edge of the wafer for verification. Pattern recognition features are selected
and stored if multiple wafers are to be cut. During dicing, the wafer is periodically checked
for alignment and cut quality. Following dicing, the wafer is moved to the cleaner to wash
away residual dust so that it cannot adhere to bond pads. For saws with fewer cooling water
jets, a surfactant may be used to bind with and help remove any remaining silicon dust.
choice for silicon wafers. Resin matrix or electroformed blades might be used for harder
or softer materials. Nickel blades are selected for a specific cutting task based on blade
width, grit size, and exposure. Blade width is determined by the width of the street to be
cut. Grit size describes the size of the diamond chips on the blade. Typical grit size for
silicon wafers ranges from 2 to 6 microns. On a wafer with a narrow street, a thin blade
with a smaller grit size allows a smoother cut. Exposure measures the height of the
exposed blade surface available to cut the material. For example, a 20 mils thick wafer
requires a blade with at least 25 mils exposure, because the blade must cut completely
through the wafer and some distance into the supporting tape (Peter McCabe, 2005).
The diamond blade is a cutting tool in which each exposed diamond particle comprises a
small cutting edge. Three basic types of dicing blades are available commercially:
1. Sintered Diamond Blade, in which diamond particles are fused into a soft metal such
as brass or copper, or incorporated by means of a powdered metallurgical process.
2. Plated Diamond Blade, in which diamond particles are held in a nickel bond
produced by an electroplating process.
3. Resinoid Diamond Blade, in which diamond particles are held in a resin bond to
create a homogeneous matrix.
Silicon wafer dicing is dominated by the plated diamond blade, which has proved most
successful for this application.
2.4 Project Methodology
This unit will guide the reader on how the study was designed and will explain the procedures
used in executing the research. It will also cover the setting which tools used on how the
research was conceived and completed.
2.4.1 Research Design
The method that is used in this study is the Borg and Galls Research and Development.
This method is both qualitative and quantitative. The study is "The Design and
Development of a Blade Breakage Detector for Semi-Automatic Dicing Saw Machine" that
can be used by manufacturing personnel. In this method, data is collected in order to test and
evaluate initial hypothesis concerning the study. The data will also be used to answer
preliminary questions and concerns regarding the study. The Research and Development
method will also be used to achieve the objectives of the study. Interviews and
questionnaires are given to respondents to understand their issues and concerns on the
29
existing method of wafer dicing saw process. The capability being proposed supports the
wafer dicing saw process without affecting quality. The proponent will be getting
information from Staff Engineers, Line Sustaining Engineers and Production Technicians in
able to find the most effective way of assessing and solving the problem.
2.4.2 Procedures
The proponent used detailed steps in system development which are outlined below:
1. Data Gathering and Conceptualization:
Ask users of the machine of possible risks involved on the current process.
Conceptualize an idea that can resolve the problem which poses possible risks of
safety injury brought about by a scattered laser and prevent it.
2. System Design:
Studying related topics about barriers and automation to come up with a design that
will prevent a laser from going out.
3. Machine Pilot:
Gather information about the materials to be used for the automated device.
Do researches, consult manuals and or catalogue for the materials and parts to be used,
and take note of each component that is deemed applicable for the device.
4. System Installation:
5. Functionality Testing:
Confirm validity and effectiveness by doing survey with the tool users.
Item Activity
1 BackgroundResearch
2 ProjectProposalandObjective
3 DataGathering
4 Analysis
5 ConceptualDesignDevelopment
6 ReviewandSelection
7 DetailedElectricalSystemModelling
8 DetailedElectronicsSystemModelling
9 DetailedMechanicalSystemModelling
10 DesignCompilationandSoftwareIntegration
11 DesignReviewEvaluation
12 DesignEnhancement
13 PrototypeDesign
14 TestingandImprovement
15 ProjectRelease
2013
M ay
Jun
Jul
A ug
Sep
O ct
N ov
D ec
Jan
Feb
M ar
A pr
M ay
Jun
Jul
A ug
Sep
O ct
N ov
D ec
2012
31
32
33
3.1 Introduction
This chapter aims to present the analysis of the current system and operation of the SemiAutomatic Dicing Saw Machine.
3.2 Type of Materials
3.2.1 Semiconductor Wafer
Silicon and GaAs are the two most common materials used for semiconductor wafers.
Wafers are typically available in sizes up to 12" in diameter, with 6" and 8" most popular.
Typical thicknesses are between 100 to 650 microns.
The second medium that is used to hold the tape is what called a film frame. This is a
metal frame that is widely used in some applications. Mainly to accommodate some
machines and systems that only accept this type of medium to hold the wafer on the table
or work area. It is basically a round metal frame that is roughly .050 thick with notches
at the bottom. As opposed to the ring, the film frame is only composed of one piece of
metal and the tape just adheres to the inferior side of the frame.
3.3.3 Mounting
Tape is most common and frequently used mounting methods for majority of dicing
applications. Tapes are preferred my majority of dicing applications, because of their ease
of use, ease of handling, and fact that it can also be used later in die mounting process.
Large variety of tapes are available today. Low strength, medium strength, and high
strength tapes Mounting is done manually or utilizing wafer mounting systems. Tapes are
available with different adhesives with what is called tackiness characteristics. Adhesion
characteristics of the most common tapes are 215 to 315 per 25 mm. Tape thickness varies
depending on application and material being diced. For example for cutting silicon wafers
tape thickness of .010" (0.25mm) is typically used, where for thick alumna substrates tape
thickness of .080 (2.03mm) is implemented.
in contact with the mounting tape before and after dicing will determine whether to use
medium or high strength tapes.
Wafer or material thickness is another consideration when selecting the right mounting
adhesive. Very thin dies work well with waxes and medium strength tapes. Whereas,
thicker wafers work better with medium to high strength tapes. Cutting too deep into the
mounting media will cause the blade to load. Minimize the depth in which the blade
touches the mounting media. It is recommended to cut no more than twenty five microns
into the media. Cutting too deep into the media will cause premature blade failure. The
blade can also agglomerate with the cutting media, resulting in chipping.
3.3.4 Wax or Glue Mounting to Media
Waxing or gluing material to solid media is usually most reliable and preferred form of
mounting. Wax mounting is typically used for mounting very thin and brittle materials.
There is a large variety of waxes that can be used for this purpose. These are typically
available in lumps, moulded bricks, flakes, chips and powders. Wax provides best
adhesion of material to supporting substrate. Allowing deep cutting into substrate. This
frequently eliminates the lip effect and cracks at the bottom of material. Also, wax
expands as it fills in the gaps of a non-flat substrate. Providing best adhesion of material or
dies, equates in superior cut quality and consistency.
Some disadvantages of wax mounting is that is a relatively messy process. Specially
taking off the material and cleaning dies afterwards. Waxes have various characteristics
that make the best suited for specific applications than others. This depends on their
melting point, flash point, specific gravity, structure, hardness, brittleness, flexible and
elastic characteristics.
Wax or glue mounting offers the following advantages:
No lip effect capability to cut much deeper into the base material. This typically
results in better cut quality, specially on back side of material.
Substrates not perfectly flat can be mounted. The wax or glue compensates by filling
in the gaps
Different hardness base media can be used to control blade wear and simultaneously
dress the blade. Common base media used is glass or ceramic.
39
Whether it is a ring or a film frame that is used to process a wafer, the mounting is
absolutely crucial. It is paramount that there are no air bubbles in between the tape and the
wafer because this will cause many problems during the dicing process. These problems
could amount to blade or material breakage and some pieces of the material falling off the
tape. Since a vacuum is used during dicing one can see how the material can break once it
is activated if there is a lump between the tape and the wafer. If the blade passes through
the part with the air bubble and the piece comes off, it could break the blade or it could
scratch the rest of the wafer. In order to avoid air bubbles and to ensure a proper
mounting, the technique requires that the tape is placed close to the wafer without being in
contact with it at first. Using the digital pulp of the finger the tape is then carefully
pressed onto the wafer making contact and adhering to it. Starting at one extreme of the
wafer and slowly working across it until the mounting is complete. Now that the wafer is
mounted as shown in Figure 3.9, be it on a ring or a film frame while using the right tape,
it is ready to be diced.
diced, it needs to be mounted using a ring or film frame and dicing tape. Then the wafer can
go on the chuck and the vacuum is activated to prevent it from moving during the violent
process. The blade is set to cut just slightly into the tape without going all the way through.
This is done so that the wafer is diced completely while the tape remains intact. Among other
hardware, the dicing machine is composed of optics that let you see the pattern on the wafer,
as well as software that allows for programming according to the specific dicing need.
During the silicon wafer dicing process, the silicon wafer is divided into single units or dice.
A rotating abrasive disc (blade) performs the dicing, while a spindle at high speed of 30,000
to 60,000 rpm (linear speeds of 83 to 175 m/sec), rotates the blade. The blade is made of
abrasive grit and diamonds that are embedded in an electroplated nickel matrix, the binder.
During the separation of dice, the blade crushes the substrate material or wafer and removes
the created debris simultaneously. Material removal occurs along dedicated dicing lines or
streets between the active areas of the dice. When the dicing blade moves along these streets,
a groove is formed in the substrate material. The groove width or kerfs is proportional to the
thickness of the blade.
The throughput of the dicing process is measured by the number of wafers diced per hour.
The speed at which the blade advances along the substrate. The feed rate actually determines
the throughput the higher the feed rate, the higher the throughput. To achieve maximum
dicing yield and productivity, it is important to operate the dicing system at the highest
throughput possible within the process quality limits. The optimal selection of the process
parameters discussed previously enables this objective. In other words, the optimal process is
the one that employs the highest possible feed rate without decreasing cut accuracy, creating
intolerable chipping, breaking the blade or shortening the blade`s life.
3.5 Dicing Process Flow
During normal operation with the current system, the mounted silicon wafer in mounting tape is
placed in the chuck table by the operator. Then will turn-on the vacuum to hold the mounted
silicon wafer. And then will perform alignment of x and y kerfs before starting the machine on
an operational mode. During wafer dicing saw process, spindle will calculate first the surface of
the chuck table before cutting the silicon wafer. During the operation, broken silicon wafer may
have the possibility to go out of the process without any interruption.
Back side becomes a yield issue when micro-cracks exceed a certain length, which may
increase the sensitivity of the devices to thermal cycling and lower their reliability. Dice
intended for flip chip packages are even more sensitive. When packaging these dice, the
back side of the die is exposed to the molding material. If the back side edges of the die
are cracked and chipped, the molding may be imperfect, including air bubbles near the
chipping. During the packaging process, these bubbles may cause mold cracking, which
can reduce yield.
Back-side chipping is more pronounced in polyimide-coated wafers where there are heavy
metal layers in the streets, or where heavy back-grinding has produced high tensile
residual stress at the bottom of the wafer. The wafer is a given in most assembly lines;
therefore, the tools available to the process engineer for controlling back side chipping are
limited to blade selection and process parameter optimization.
3.7 Optimization
Optimizing a dicing process entails achieving the highest possible throughput with minimal
chipping on the die edges. To reach maximum throughput, it is necessary to use the highest
possible feed rate. The feed rate is always dependent on the spindle speed and the blade type.
Combining these parameters and ensuring optimum throughput is the challenge.
Table 3.1 Factors affecting cut quality and blade torque
Blade
Coolant
Machine
Work Piece
Rotation speed
Vibrations
Wear
Flow
Adjustment
Feed rate
Spindle, rpm
Wafer thickness
Surface condition
It has been shown that all process parameters correlate with blade torque as shown in Table
3.1. The appearance of chipping, especially back side chipping, also correlates with blade
torque. For each set of process parameters, there is a torque value limit. Dicing quality
deteriorates and back side chipping appears above the limit. By measuring the torque online,
the other process parameters can be set in such a way that the torque limit will not be
exceeded and maximum feed rate will be obtained without chip formation.
Many experimental variables have to be considered during process optimization. Testing each
variable separately is tedious and consumes many wafers. The Design of Experiment method
44
is used to reduce the number of tests needed and to provide the combined effect of several
parameters. The Design of Experiment is a statistical method for evaluation of multi-variable
processes. All experimental variables are arranged in a matrix and tested in at least two
settings.
The outcome of the tests is measured and recorded. Analysis of variance is done on the
measured test results. The Design of Experiment tests and analysis reveal the major factors
that affect the response (e.g., back side chipping), as well as the effect of each factor when it
increases or decreases.
3.8 Monitoring
The common way to continuously monitor the dicing process is online, visually inspecting cut
position and chipping on the wafer`s top side. However, to monitor the formation of back side
chips, the diced wafer must be inspected off the machine. There also is a way to monitor back
side online using torque measurements.
The torque applied by the blade when cutting the substrate reflects changes in the different
factors that affect the process. Because these changes indicate variation in the process, the
torque also reflects conditions that could lead to formation of back side chipping. When the
torque limits are determined, the torque measurements become an efficient tool for online
monitoring of back side chipping.
On-line monitoring of blade torque informs the operator of any deviation. It does not require
extra inspection time and alerts in real time when there is danger of yield loss. Subsequent
offline inspection should be a complementary tool, used to calibrate the online monitor or to
verify the causes of deviations it detects.
3.9 Concluding Remarks
This chapter mentioned that the current system which is dicing saw process does not add
value to the finished device, therefore, the quality of the dicing process is measured by the
yield loss it may cause. Potential dicing yield loss factors include cut misplacement,
cleanliness and chipping.
45
46
4.1 Introduction
This chapter will discuss the evaluation and consideration of alternative solution for the SemiAutomatic Dicing Saw Machine.
4.2 Variables in Selecting Dicing Blade
The following variables shown in Figure 4.1, should be considered in selecting the right
dicing blade for the application:
production and quality control. At the same time as dicing blades become more inconsistent,
dicing saw operators must monitor the process more closely and make necessary parameter
changes. These parameter changes may extend dicing blade life back to twenty five thousand
cuts. However, the extra time spend adjusting could have been better utilized monitoring
other machines or operations.
Choosing the right blade for any particular application is crucial. These essential precision
tools come in a variety of sizes, types, as well as in different abrasive or grit sizes. They also
come in hubbed and hubless forms. The different types of blades are: resin bond for dicing
glass, quartz, sapphire and such materials. Nickel bond for dicing silicon, gallium arsenide,
germanium and indium phosphate. Metal sintered blades are good for dicing plastics, QFN
packages, PCB and FR4 type materials. They all utilize diamond as an abrasive and the
diamond (in most cases synthetic diamond) is what does the actual cutting. The key is in
choosing the right size diamonds for the particular application at hand. When dealing with
hubbed blades the exposure is to be considered to ensure that there is enough blade to cover
the material and sufficient room for wear. When dealing with hubless blades the flange needs
to be taken into account. The flange is a metal hub that holds the blade in place and sets it in
the spindle. In choosing the right hubless blade one needs to keep in mind the inside diameter
and outside diameter of the flange to determine the right size blade that is needed. The
essential specifications for a blade are size (Inside Diameter and Outside Diameter), type
(nickel, resin, sintered, hubbed or hubless), thickness, grit size, and exposure. Some
manufacturers use different diamond concentrations and that is to be acknowledged as well.
In general, the optimization process should start with blade selection and process parameter
optimization should follow. The blade dimensions are limited by wafer demands, required
kerfs width determines blade thickness. The thickness of the wafer determines the blade
height. During blade selection, two factors should be considered: blade life and cut quality.
Blade life is an important factor in the cost of ownership of a dicing system. Blade life
depends on blade bonding material properties. Blades with softer bond material typically
provide better cut quality, especially regarding control of back side chipping. These blades
also wear much faster than hard bonded blades. A trade-off is required between blade wear
and cut quality. The chip size should be acceptable and blade life sufficient. Use of Design of
Experiment can determine this required compromise without numerous trial and error tests.
4.2.1 Bond Hardness
Ability of the bond matrix to hold diamonds. As the hardness of the bond is increased, its
diamond retention capabilities increase as well. However, the trade off is slower cutting
speed. Life of the diamond blade is usually increased with hardness of its bond matrix.
Bonds are designated on their scale of hardness from soft, medium and hard. There are
dozens of variations and classification schemes based on bond degree of hardness or
softness. Using diamond dicing blades with optimum bond hardness for your application
is important to successful dicing operation. Bond matrix that is too soft for the material
being cut will release diamond particles faster than needed, resulting in faster wear and
shorter diamond blade life. On other hand bond matrix that is too hard will result in much
slower cutting speeds and require constant dressing to expose the next diamond layer. As
rule of thumb, harder materials such as sapphire and alumina generally require a softer
bond. Whereas softer and more brittle materials require a harder bond.
4.2.2 Diamond Grit Size (Mesh Size)
Diamond grit size (mesh size) plays a major role in determining the surface finish quality,
smoothness, level of chipping and material microstructure damage wish to obtain. Finer
mesh size diamonds such as 20 microns are much smaller in size than coarser diamond
particles. And will give a very smooth surface finish with minimal amount of chipping on
edges. These mesh sizes are usually used for fine cutting of a full range of materials such
as: LiNbO3, YVO4, GaAs and optical materials. Courser diamond particles are much
larger in diameter and are frequently used fast cutting or material removal on more harder
materials such as silicon carbide, zirconia, Al2O3 and other advanced ceramics and high
metallic content materials which do not require a very fine surface finish.
49
The diamond grit size (mesh size) in a dicing blade also directly relates to the number of
crystals per carat and the free cutting capability of the dicing blade. The smaller the mesh
size, the larger the diamond crystals, while larger mesh size means smaller diamond. A
30/40 mesh blocky diamond has about 660 crystals per carat, while a 40/50 mesh diamond
will have 1,700 crystals per carat. Specifying the proper mesh size is the job of the dicing
blade manufacturer. Producing the right number of cutting points can maximize dicing
blade life and minimize the machine power requirements.
4.2.3 Diamond Concentration
The proportion and distribution of diamond abrasive particles, also known as
concentration, has an effect on overall cutting performance and price of precision diamond
blades. Diamond concentration, commonly referred to as CON, is a measure of the
amount of diamond contained in a diamond section of drill based upon volume. Diamond
concentration is usually defined as: Concentration 100 = 4.4 ct/cm layer volume (mesh
size + bond). Based on this definition a concentration of 100 means that the diamond
proportion is 25% by volume of diamond layer, assuming at diamond density is 3.52
g/cm3 and 1 ct = 0.2 g. Nominal diamond concentration in precision diamond blades
range from 0.5 ct/cm3 to 6 ct/cm3. This means diamond concentrations are available from
8 to 135. Selecting the right diamond concentration can be critical in optimizing dicing
operation. Diamond concentration will play a major role in determining the life and
cutting speed of the dicing blade.
4.2.3.1 High Diamond Concentration
Higher diamond concentration causes the blade to act harder and wear slower as
shown in Figure 4.3. Hence, higher diamond concentration is recommended and
usually used for cutting softer and more abrasive types of materials. However, the
trade off is significantly slower cutting speed. Since there is higher density of
diamond particles in bond matrix, there is not enough space for the debris and fine
powder created during dicing process to escape. Blade overloading is often a result
of selecting improper diamond concentration.
4.2.3.2 Low Diamond Concentration
Blade is plated on very accurate aluminium hub, it is stiffer. Hence, producing better
cut quality. This can be seen in the kerfs having straight cross section and longitudinal
cut through the street being straight and parallel to the street.
Fewer vibration on the blade during cutting, due to increase blade stiffness. Many
operators note the cut or kerfs is cleaner and there is less chipping.
Hubbed type dicing blades are pre-dressed, hence requiring very simple and shore
dressing procedure.
All the operators will utilize and experiment with different flange types best suited for
their application
High cooling flange are available, allowing water directly onto the blade. This
provides better blade and substrate cooling, resulting in better cut quality and longer
blade life.
sintered with specific metal alloys to achieve the best cutting performance possible on any
large variety of material. The metal bond surrounding the diamonds must wear away to
continuously keep re-exposing the diamonds for the diamond tool to continue cutting.
but more than electroplated (nickel bond) diamond blades, finish and minimum amount of
chipping.
Low cost
Disadvantages:
Edge geometry is lost faster than with other types of dicing blades
Some disadvantages of resin bond dicing blades is resin bond must remain very fragile in
order to expose new diamonds. Resin bond is the softest of the all the dicing bond
matrixes. For this reason, strong and high quality diamonds cannot be used in a resin
bond. High quality diamonds are harder than a resin bond matrix and would soon
disintegrate the bond that keeps them in place. This brings about the need for frequent
blade dressing, causing the cut to lose its roundness or form. Another disadvantage of
54
resin bond is its high wear rate, lack of stiffness and thickness limitation. Resin bond can
cut hard and brittle materials fast, but will provide much shorter life.
4.4.3 Electroplated (Nickel Bond) Dicing Blade
Electroplated (nickel bond) dicing blades shown in Figure 4.5, have a high diamond
concentration and give a freer, faster cutting action with minimum heat generation.
Diamonds have higher protrusion ratio, staying on the surface of the cut allowing for fast
material removal. Electroplated (nickel bond) dicing blades last less than metal bond, resin
bond, hybrid bond blades and are the least expensive diamond blades available. Silicon
wafer dicing is usually done with the plated diamond blade (hubbed or hub less) which has
proven most effective for this application. The kerfs are typically in the 1 to 3 mils range
using a nominal spindle speed of 30,000 RPM with feed rates as high as 8 inch per
seconds.
Advantages:
Disadvantages:
Nickel Bond
(microns)
--3-6
--------3-6
3-6
1-6
----17 - 30
3 - 10
3-6
--30 - 50
30 - 50
17 - 30
3-6
17 - 30
Metal Bond
(microns)
--2-6
--17
--------------10 - 17
------30 - 50
30 - 50
--9
---
Resin Bond
(microns)
53
9
45
45
53
30
------45 - 63
53
53
----88 - 105
----45
-----
It is important for the diamonds to be well protruded and exposed in order to easily penetrate
into the material being cut. Dicing blades that are poorly dressing will tend to push the
material, creating high loads, high cutting temperature or heat and poor cut quality. This can
also cause blade breakage.
Resin bond dicing blades have a soft binder and in most cases will require minimum dressing,
if any at all. Resin bond dicing blade are typically dressed in the material being cut. This is
why they are frequently called 'self dressing blades'.
Nickel bond and metal bond blades have much harder binder and much more aggressive
dressing is needed in order to achieve the above mentioned goals. Nickel bond dicing blades
require dressing. Hubbed nickel bond dicing blades are partly pre-dressed when they are
manufactured. And typically do not require any dressing. Dressing will remove nickel built up
or diamond protruding from the blade surface are sheared off.
Since nickel bond dicing blades are typically used on silicon. It is recommended to conduct
the dressing cycle also in silicon. Dressing programs usually involve various steps of table
speeds, starting slow and increasing until production speed is about to be reached. Great care
has to be taken that the dressing depth is deeper than the required production cutting depth.
This will insure that only dressed blade sections are used for production cutting.
To ensure high quality process results, new blades have to be dressed before starting
production. This step is required to expose the cutting diamonds in the blade surface and
condition the blade for continuous work without dramatic changes in cut quality. Dressing
consists of cutting a certain length of material, starting at low feed rate and increasing the rate
until the target is reached.
The duration of the dressing process is usually based on post dicing inspection results.
Because the mechanism of this process is not fully understood, it tends to be a lengthy
procedure that affects productivity. Chipping may result if dressing time is too short. In many
cases, extra dressing time is used to maintain a margin of safety.
There is a good method for controlling dressing time that can increase productivity. Blade
torque measurements provide an online method for determining the optimal dressing
procedure. If the measured torque values follow a predetermined reference pattern during the
dressing procedure, the dressing is being done properly and the point of completion can be
determined. Dicing systems offering this capability can significantly increase production time.
4.6 Dressing Process
Dressing procedures are established by end user and based on experience and the application
concerned. There is not one common method that applies to all applications. The application
below is a good starting point.
Table 4.2 Dressing Parametric
Number of Cuts
Depth
Table Speed
5 - 10 cuts
50 microns
50 mm/s
5 - 10 cuts
PD + 100 microns
10 mm/s
5 - 10 cuts
PD + 100 microns
20 mm/s
5 - 10 cuts
PD + 100 microns
30 mm/s
Where:
PD = production cutting depth
When cutting silicon most nickel bond hubbed dicing blades do not require dressing.
However on gallium wafers, dressing will be beneficial and increase blade life and improve
cut quality. It is specially recommended to dress blades used on gallium wafers.
Sometimes a blade that requires dressing is mistaken for a defective blade (blade that cannot
be used). This is not necessarily the case. The most frequent cause of this affect is a diamond
particle or few diamond particles excessively protrudes from the nickel bond matrix, causing
excessive chipping. By dressing the blade, this problem will be eliminated.
58
Blade thickness
Blade bond type such as resin bond, nickel bond or metal bond
59
Blade Type
Nickel Bond
Resin Bond
Nickel Bond
Nickel Bond
Carbide Blades
Resin Bond
Resin Bond
Nickel Bond
Resin Bond
Resin Bond
4.8 Coolant
The amount of coolant and its supply is of little importance when cutting silicon with nickel
bonded dicing. Most dicing saw used deionised water for this application. If the wafers are
sensitive to electrostatic charges, the required conductivity is set using CO2. For this
application single nozzle may be sufficient. However, when cutting ultra hard and brittle
materials, the coolant flow direction and position will play a major role in determining blade
life and performance.
Dual nozzle configuration is best suited to supply adequate amount of coolant at point of
contact between blade and material. Coolant should be pointed at blade point of contact and
leading edge of blade. The dual nozzle system provides two separate streams of coolant into
cutting zone at angle of cutting edge of the blade. Each stream will cool one side of the blade.
The nozzles should be as close as possible to the point where the blade enters the substrate.
Coolant flow should blast with high velocity into the kerf to improve removing debris. Wide
kerfs premature blade water is a result of improper coolant amount or flow applied at cutting
zone. By correctly position the coolant stream in front of the blade, the coolant will flow on
top of the material. In addition, coolant will be drawn into the kerf.
When cutting silicon, the direction of cutting is not an ultimate importance. Often most wafers
are cut by moving the table back and forth. However, when cutting hard and brittle materials
this quiet different. The blade should enter into the material cutting down. As the blade
penetrates the surface of the material, chips removed by diamond particles become smaller the
deeper the blade penetrates. This cutting direction is most optimal to provide best cut quality.
60
When cutting in the opposite direction the blade enters the substrate almost tangential. The
chips removed are being small and start but their size increases to maximize where the blade
exits the face of the substrate. Much chipping at the surface is the result. Cutting in the up
direction creates more edge chipping than cutting down. The spindle torque is noticed to be
the large cutting up than when cutting down.
4.9 Process Stability
Steady blade torque values indicate a stable process. Once the torque values for a specific
process have been established, the other setup parameters should be adjusted to keep these
values consistent. Inconsistency of torque values is an indication of deviation in the process
because of blade vibration. Blade vibration ends in blade breakage or chipping and cracking
damage to the wafer.
System operation at steady torque requires stabilization of the feed rate, spindle speed and
coolant flow. The coolant applies drag force on the blade, which makes a significant
contribution to the torque. To keep the coolant torque effect steady, the coolant flow is
controlled, ensuring steady flow of coolant and no changes in drag force.
When the dicing machines have steady coolant flow and all other parameters are under
control, a steady torque is maintained. Any deviations from steady torque, if recorded, are due
to uncontrolled factors. These include changes in coolant flow because of nozzle clogging,
changes in nozzle adjustment, blade-to-blade variation, blade condition and operator errors.
Table 4.4 Matching Basic Material Properties to Blade Properties
Hardness
Density
Abrasiveness
Diamond
Size
Concentration
Bond
Hardness
Hard
Fine
Low
Soft
Soft
Coarse
High
Hard
High
Fine
Low
Soft
Low
Coarse
High
Hard
High
Coarse
High
Hard
Low
Fine
Low
Soft
61
Operating Condition
Machine
Operating Speed
Feed Rate
Horsepower
Coolant Volume
Cutting Depth
Old
New
High
Low
Fast
Slow
High
Low
High
Low
Shallow
Deep
Blade Action
Softer
Harder
Harder
Softer
Softer
Harder
Softer
Harder
Harder
Softer
Softer
Harder
62
References:
Alger, J. and Hays, C. (1964) Creative Synthesis in Design, New York: Prentice Hall
Amplifier, http://en.wikipedia.org/wiki/Amplifier
Automation, http://www.encyclopedia.com/doc/1O48-automation.html
Bajune, D. (2000) Substrate Diving Method, US6152803 A
Cliff, E. (1978) Design of Machine Elements, 5th edition, New Jersey: Prentice-Hall
Dearborn, M. (1987) Tool and Manufacturing Engineers Handbook, Volume 4, Society of
Manufacturing Engineers
Fiber Optic Proximity Sensors,
http://sensorstransducers.globalspec.com/learnmore/sensors_transducers_detectors/proximity_
presence_sensing/fiberoptic_proximity_sensors
Fowler, K. and Schmalzel, J. (2004) IEEE Instrumentation and Measurement Magazine
French, T., Svensen, C., Helsen, J. and Urbanik, B. (1974) Mechanical Drawing, 8th edition,
New York: McGraw Hill
Groover, M. (2001) Automation, production systems, and computer-integrated manufacturing,
2nd edition, New York: Prentice Hall
Kivenson, G. (1977) The Art and Science of Inventing, New York: Van Nostrand Reinhold
Company
Light Emitting Diode, http://www.newworldencyclopedia.org/entry/LED
Nyce, D. (2004) Linear Position Sensors: Theory and application, New York: Wiley
Raudsepp, E. (1976) Creative Growth Games, New York: Jove Publications
Raudsepp, E. (1982) Creative Growth Games, 2nd edition, New York: Jove Publications
Sawing Silicon, http://www.chipsupply.com/resources/pdf/sawingSilicon.pdf
Schey, J. (2000) Introduction to manufacturing processes, 3rd edition, New York: McGraw Hill
Schultz, M. (2008) Grobs basic electronics, 10th edition, New York: McGraw Hill
Terrell, D. (2004) Op Amps: Design, application and troubleshooting, 2nd edition, Boston:
Elsevier
Comment [B31]: Attach Project Diary
And Industry Supervisors Report.
63