I.
$1.5 v!
1.8.3.?

(3.

 

 

 

 

:3: : . i
by...“ a".

.u.x.3§|(
A

. i (.3 1305:}.

 

 

 

 

ween;

20:0

 

LIBRARY
Michigan State
University

 

 

 

PLACE iN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11m muss-p.14

 

 

PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS
FOR MICROELECTRONIC DEVICES

By

Kiran K. Baikerikar

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Chemical Engineering

2000

ABSTRACT

PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS
FOR MICROELECTRONIC DEVICES

By

Kiran K. Baikerikar

Plastic encapsulated microelectronic devices consist of an integrated circuit die
that is physically attached to a leadframe, electrically interconnected to input-output
leads, and molded in a plastic that is in direct contact with the die, leadframe, and
interconnects. The plastic is often referred to as the molding compound or encapsulant,
and is used to protect the integrated circuit die (silicon chip) from adverse mechanical,
chemical, thermal, and electrical environments. Encapsulation of microelectronic devices
is typically accomplished using a transfer molding process. in which the molding
compound is cured by heat. Most transfer molding processes suffer from signiﬁcant
problems arising from the high operating temperatures and high pressures required to ﬁll
the mold. These aspects of the current process can lead to thermal stresses, incomplete
mold filling, and wire sweep (ﬂow-induced deformation of the wires that interconnect the
silicon chip to the leadframe during molding). As the size of the encapsulant and the
associated wires shrink to smaller and smaller dimensions, so does the operating window
for transfer molding; hence, it is difﬁcult to control wire sweep in the high pin count, ﬁne
pitch packages. These problems lead to productivity decreases due to increased
packaging-related rejects.

In this research, a new strategy for encapsulating microelectronic devices using

photopolymerizable liquid encapsulants (PLEs) has been investigated. These highly

ﬁlled PLEs consist of an epoxy novolac-based vinyl ester resin (~25 wt.%), fused silica
(70-74 wt.%), and small amounts of a photoinitiator, thermal initiator, and silane
coupling agent. For these encapsulants, the use of light rather than heat to initiate the
polymerization allows precise control over when the reaction starts, and therefore
completely decouples the mold ﬁlling and the cure. The relatively low viscosity of the
PLE allows low operating pressures and minimizes problems associated with wire sweep.
Because photoinitiation allows active centers to be formed rapidly at room temperature,
the in-mold cure time for the PLEs is no longer than the in-mold cure times of current
transfer molding compounds.

In this thesis, an extensive series of studies on the thermal and mechanical
properties, as well as the viscosity and adhesion of photopolymerizable liquid
encapsulants, are reported in order to demonstrate that a UV—curable formulation can,
indeed, have the necessary material properties required for microelectronic encapsulation.
Speciﬁcally, the viscosity, ﬂexural strength and modulus, coefﬁcient of thermal
expansion, glass transition temperature, thermal stress parameter, thermal conductivity,
thermal diffusivity, speciﬁc heat, and adhesive peel strength of the PLEs have been
characterized. In addition, the effects of the illumination time, postcure time, fused silica
loading, and the inclusion of a thermal initiator on the thermal and mechanical properties
of the ﬁnal cured encapsulants have been investigated. The results show that the material
properties of the PLEs are the same, if not better, than those exhibited by conventional
transfer molding compounds and demonstrate the potential of using PLEs for

encapsulating microelectronic devices.

ACKNOWLEDGMENTS

I would like to thank, ﬁrst and foremost, Dr. Alec Scranton for serving as my
graduate advisor throughout this project. I have really enjoyed working on this research
project and am grateful to Dr. Scranton for giving me the opportunity to work in the areas
of photopolymerization and microelectronic encapsulation. I would also like to thank the
members of my doctoral committee, Dr. Martin Hawley, Dr. Robert Ofoli, and Dr. Jack
Giacin for their guidance and helpful suggestions. I would like to thank Mike Rich for
training me in the polymer composites laboratory and for always taking the time to
answer any questions that I had about thermal and mechanical testing. In addition, I
would like to thank Dr. Richard Schalek for his help and guidance in using the
environmental scanning electron microscope.

I would like to thank the current and former graduate students in Dr. Scranton’s
research group whom I had the pleasure of working with over the years, including
Bernhard Drescher, Julie Jessop, Arvind Mathur, Vijaykumar Narayanan, Khanh
Nguyen, and Katy Padon. These ﬁ‘iends were always quick to help someone out and I
appreciate the help that they often gave me.

I would also like to thank my loving parents, Dr. Kamalakar and Vijaya
Baikerikar, for always encouraging me and giving me their undivided support. My
parents always sacriﬁced for me and instilled in me the power of hard work and
dedication. I would not be where I am today without their guidance, and for that, I am

eternally grateful.

iv

Finally, I would like to express my appreciation for the person that I am most
fortunate to share my life with, my wonderful wife, Naina. She has never shied away
from offering me encouragement and support when I most needed it. I am lucky to be

with her and am forever grateful for her compassion and unconditional love.

TABLE OF CONTENTS

 

 

 

 

LIST OF TABLES x
LIST OF FIGURES xii
1. INTRODUCTION 1
1.1. Overview and Importance of Microelectronic Encapsulation ............................ l

1.2. Scope and Aim of Research ............................................................................... 3

1.3. References .......................................................................................................... 5

2. BACKGROUND 6
2.1. Components of a Microelectronic Device .......................................................... 6

2.2. Components of the Encapsulation Material ....................................................... 8
2.2.1. Epoxy Resin ........................................................................................... 8

2.2.2. Curing Agent .......................................................................................... 9

2.2.3. Accelerator ............................................................................................. 9

2.2.4. Filler ..................................................................................................... 10

2.2.5. Coupling Agent .................................................................................... 11

2.2.6. Flame Retardant ................................................................................... 11

2.2.7. Mold Release Agent ............................................................................. 12

2.2.8. Coloring Agent ..................................................................................... 12

2.3. Material Properties used to Evaluate Encapsulation Materials ........................ 12
2.3.1. Viscosity ............................................................................................... 13

2.3.2. Flexural Modulus ................................................................................. 13

2.3.3. Coefﬁcient of Thermal Expansion ....................................................... 14

2.3.4. Glass Transition Temperature .............................................................. 14

2.3.5. Thermal Conductivity .......................................................................... 14

2.3.6. Adhesion ............................................................................................... 15

2.4. Semiconductor Manufacturing Process ............................................................ 15
2.4.1. Wafer Separation and Die Attach ......................................................... 16

2.4.2. Die Interconnection .............................................................................. 16

2.4.3. Transfer Molding .................................................................................. 17

2.4.4. Deﬂashing and Code Marking ............................................................. 17

2.4.5. Postcure ................................................................................................ 18

2.4.6. Trim-and-Form ..................................................................................... 18

2.4.7. Solder Dipping/Solder Plating ............................................................. 18

2.4.8. Bum-in ................................................................................................. 19

2.4.9. Testing .................................................................................................. 19

2.5. Encapsulation of Microelectronic Devices by Transfer Molding .................... 19

2.6. Disadvantages of Transfer Molding ................................................................. 21

2.7. Overview of Photopolymerizations .................................................................. 22

2.8. Photopolymerization of Thick Polymers and Composites ............................... 23

vi

2.9. Advantages of using Photopolymerizations to Encapsulate Microdevices ...... 24

 

 

 

 

2. 10. References ....................................................................................................... 26
RESEARCH OBJECTIVES 27
SELECTION OF MATERIALS FOR THE PHOTOPOLYMERIZABLE
LIQUID ENCAPSULANT 29
4.1. Resin Matrix ..................................................................................................... 29
4.2. Filler ................................................................................................................. 31
4.3. Photoinitiator .................................................................................................... 32
4.4. Thermal Initiator .............................................................................................. 34
4.5. Silane Coupling Agent ..................................................................................... 35
4.6. References ........................................................................................................ 42
VISCOSITY CHARACTERIZATION OF HIGHLY FILLED
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS 43
5.1. Introduction ...................................................................................................... 43
5.2. Background on the Viscosity of Concentrated Suspensions ............................ 46
5.3. Experimental .................................................................................................... 49
5.3.1. Materials ............................................................................................... 49
5.3.2. Sample Preparation .............................................................................. 50
5.4. Results and Discussion ..................................................................................... 52
5.4.1. Effect of Fused Silica Particle Size Distribution .................................. 52
5.4.2. Effect of Silane Coupling Agent .......................................................... 55
5.4.3. Effect of Higher Filler Loadings .......................................................... 55
5.5. Conclusions ...................................................................................................... 56
5.6. Nomenclature ................................................................................................... 59
5.7. References ........................................................................................................ 64
MECHANICAL AND THERMOMECHANICAL PROPERTIES OF
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS 66
6.1. Introduction ...................................................................................................... 66
6.2. Experimental .................................................................................................... 67
6.2.1. Preparation of Photopolymerizable Liquid Encapsulant ...................... 67
6.2.2. Flexural Modulus and Strength ............................................................ 68
6.2.3. Degree of Cure ..................................................................................... 69
6.2.4. Coefﬁcient of Thermal Expansion and Glass Transition Temperature 72
6.3. Results and Discussion ..................................................................................... 73
6.3.1. Effect of UV Illumination Time on F lexural Modulus and Strength... 73
6.3.2. Effect of Postcure Time on F lexural Modulus and Strength ................ 74
6.3.3. Effect of Thermal Initiator on Flexural Modulus and Strength ........... 76
6.3.4. Degree of Cure ..................................................................................... 77
6.3.5. Effect of Postcure Time on CTE below Glass Transition .................... 80
6.3.6. Effect of Thermal Initiator on CTE below Glass Transition ................ 81
6.3.7. Effect of Postcure Time and Thermal Initiator on CTE above Glass
Transition .............................................................................................. 81

vii

6.3.8. Effect of Postcure Time on Glass Transition Temperature .................. 82

 

 

 

6.3.9. Effect of Thermal Initiator on Glass Transition Temperature .............. 83
6.3.10. Thermal Stress Parameter .................................................................... 83
6.4. Conclusions ...................................................................................................... 85
6.5. References ........................................................................................................ 98
EVALUATION OF REDUCED IN-MOLD CURE TIMES FOR
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS 99
7.1. Introduction ...................................................................................................... 99
7.2. Experimental .................................................................................................. 100
7.2.1. Preparation of Photopolymerizable Liquid Encapsulant ..................... 100
7.2.2. F lexural Modulus and Strength ........................................................... 101
7.2.3. Coefﬁcient of Thermal Expansion and Glass Transition Temperature102
7.3. Results and Discussion ................................................................................... 103
7.3.1. Flexural Modulus and Strength ........................................................... 103
7.3.2. Effect of Reduced UV Illumination Times on F lexural Modulus
and Strength ........................................................................................ 103
7.3.3. Effect of Reduced UV Illumination Times on Coefﬁcient of Thermal
Expansion below Glass Transition ..................................................... 104
7.3.4. Effect of Reduced UV Illumination Times on Coefﬁcient of Thermal
Expansion above Glass Transition ...................................................... 106
7.3.5. Effect of Reduced UV Illumination Times on Glass Transition
Temperature ........................................................................................ 106
7.3.6. Effect of Reduced UV Illumination Times on Thermal Stress
Parameter ............................................................................................ 106
7.4. Conclusions .................................................................................................... 107
7.5. References ...................................................................................................... 112
THERMAL PROPERTIES OF PHOTOPOLYMERIZABLE LIQUID
ENCAPSULANT S 113
8.1. Introduction .................................................................................................... 113
8.2. Experimental .................................................................................................. 114
8.2.1. Preparation of Photopolymerizable Liquid Encapsulant ..................... 114
8.2.2. Specimen Preparation for Thermal Conductivity Testing ................... 115
8.2.3. Laser Flash Method ............................................................................. 116
8.3. Results and Discussion ................................................................................... 118
8.3.1. Thermal Conductivity ......................................................................... 118
8.3.2. Speciﬁc Heat ....................................................................................... 121
8.3.3. Thermal Diffusivity ............................................................................. 122
8.4. Conclusions .................................................................................................... 122
8.5. Nomenclature ................................................................................................. 124
8.6. References ...................................................................................................... 130
ADHESION BETWEEN PHOTOPOLYMERIZABLE LIQUID
ENCAPSULANTS AND ALLOY 194 131
9.1. Introduction .................................................................................................... 131

viii

10.

9.2. Experimental .................................................................................................. 133

 

9.2.1. Preparation of Photopolymerizable Liquid Encapsulant .................... 133
9.2.2. Solvent Degreasing Pretreatment of Alloy 194 .................................. 134
9.2.3. Treatment of Alloy 194 Strips with Silane Coupling Agents ............ 135
9.2.4. Treatment of Alloy 194 Strips with Azole Compounds ..................... 136
9.2.5. Specimen Preparation for Peel Test ................................................... 136
9.3. Results and Discussion ................................................................................... 137
9.3.1. Peel Strength of Surfaces Treated with Silane Coupling Agents ....... 137
9.3.2. Mechanism of Bond Failure between PLE and Surfaces Treated
with Silanes ....................................................................................... 139
9.3.3. Peel Strength of Surfaces Treated with Azole Compounds ............... 140
9.4. Conclusions .................................................................................................... 140
9.5. References ...................................................................................................... 150
RESEARCH SUMMARY AND RECOMMENDATIONS FOR
FUTURE WORK 151
10.1. Summary of Results ...................................................................................... 151
10.1.1. Viscosity ........................................................................................... 152
10.1.2. Mechanical and Thermomechanical Properties ............................... 153
10.1.3. Reduction of In-mold Cure Time ..................................................... 155
10.1.4. Thermal Conductivity, Speciﬁc Heat, and Thermal Diffusivity ...... 155
10.1.5. Adhesion ........................................................................................... 156
10.2. Recommendations for Future Work .............................................................. 157
10.2.1. Increased Light Intensities ............................................................... 158
10.2.2. Development of Photomolding Process to Encapsulate
Actual Microdevices ............................................................... 158
10.2.3. Accelerated Testing .......................................................................... 160
10.2.4. Glob-top Encapsulation .................................................................... 160
10.3. References .................................................................................................... 162

Table 2.1.

Table 5.1.

Table 5.2.

Table 6.1.

Table 6.2.

Table 6.3.

Table 6.4.

Table 6.5.

Table 6.6.

Table 6.7.

Table 6.8.

LIST OF TABLES

Chapter 2
Sequence of steps in the assembly of microelectronic devices ................. 16
Chapter 5

Volume fractions of individual fused silica distributions
in the binary mixtures ................................................................................ 60

Viscosity of PLEs with 70.0, 72.0, and 74.0 wt.% fused silica ﬁller ........ 60
Chapter 6

Effect of the inclusion of thermal initiator on ﬂexural strength for
samples that were not postcured. ................................................................. 95

Effect of the inclusion of thermal initiator on ﬂexural strength for
samples that were postcured for 6 hours. .................................................. 95

Effect of the inclusion of thermal initiator on ﬂexural modulus for
samples that were not postcured ................................................................ 95

Effect of the inclusion of thermal initiator on ﬂexural modulus for
samples that were postcured for 6 hours ................................................... 96

Heat released during DSC postbake and additional cure for
photopolymerized encapsulants subject to various cure cycles. ............... 96

Effect of thermal initiator on the CTE below glass transition of the
photopolymerized encapsulants ................................................................. 96

Effect of thermal initiator on the glass transition temperature of the
photopolymerized encapsulants ................................................................ 97

Effect of thermal initiator on thermal stress parameter of the
photopolymerized encapsulants ................................................................ 97

Chapter 7

Table 7.1. Effect of reduced UV illumination times on ﬂexural modulus for
both UV- initiated and dual initiated samples containing 70. 0, 72. 0,
and 74. 0 wt. % fused silica. Samples were photocured with UV light
of 200 mW/cm2 UVA intensity and then postcured for 6 hours at
170°C ....................................................................................................... 109

Table 7.2. Effect of reduced UV illumination times on ﬂexural strength for both
UV-initiated and dual initiated samples containing 70. 0, 72. 0, and
74. 0 wt. % fused silica. Samples were photocured with UV light of
200 mW/cm2 UVA intensity and then postcured for 6 hours at
170°C. ...................................................................................................... 109

Table 7.3. Effect of reduced UV illumination times on coefficient of thermal
expansion below glass transition for both UV-initiated and dual
initiated samples containing 70. 0, 72. 0, and 74. 0 wt. % fused silica.
Samples were photocured with UV light of 200 mW/cm2 UVA
intensity and then postcured for 6 hours at 170°C ................................... 110

Table 7.4. Effect of reduced UV illumination times on coefficient of thermal
expansion above glass transition for both UV- initiated and dual
initiated samples containing 70. 0, 72. O, and 74. 0 wt. % fused silica.
Samples were photocured with UV light of 200 mW/cm2 UVA
intensity and then postcured for 6 hours at 170°C ................................... 1 10

Table 7.5. Effect of reduced UV illumination times on glass transition
temperature for both UV-initiated and dual initiated samples
containing 70. 0, 72. O, and 74. 0 wt. % fused silica. Samples were
photocured with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C. ............................................................... 111

Table 7.6. Effect of reduced UV illumination times on thermal stress parameter
(0.) for both UV-initiated and dual initiated samples containing
70. 0, 72. 0, and 74. 0 wt. % fused silica. Samples were photocured
with UV light of 200 mW/cm2 UVA intensity and then postcured for

6 hours at 170°C. ..................................................................................... 111
Chapter 8
Table 8.1. Calculated volume fractions corresponding to 70.0, 72.0, and 74.0
wt.% fused silica loadings. ...................................................................... 120
Chapter 10

Table 10.1. Comparison of material properties between PLEs and molding
compounds .............................................................................................. 152

xi

Figure 2.1.
Figure 2.2.

Figure 4.1.
Figure 4.2.

Figure 4.3.

Figure 4.4.

Figure 4.5.

Figure 4.6.
Figure 4.7.

Figure 5.1.

Figure 5.2.

Figure 5.3.

LIST OF FIGURES

Chapter 2
Schematic of a microelectronic device ........................................................ 6
Transfer molding process for encapsulating microelectronic devices ...... 20
Chapter 4
General chemical structure of DERAKANE epoxy vinyl ester resins ...... 37

Cure time as a function of ﬁber loading for various DERAKANE
epoxy vinyl ester resins ............................................................................. 38

Effect of ﬁller loading on the coefﬁcient of thermal expansion for

three different DERAKANE resins ............................................................. 39
Chemical structure of bis(2,4,6-trimethylbenzoyl)phenyl phosphine
oxide .......................................................................................................... 40
Absorbance spectrum of bis(2,4,6-trimethylbenzoyl)phenyl
phosphine oxide. The concentration is 5.1 x 10'4 g/ml in acetone ........... 40
Chemical structure of benzoyl peroxide .................................................... 41
Chemical structure of 3-methacryloxypropyltrimethoxysilane ................. 41
Chapter 5

Relative viscosity as a function of volume fraction of PSD-A. The
balance of the ﬁller is either PSD-B or PSD-C. The viscosity of the
suspending medium, 770, is 0.0456 Paos (45.6 CF) ..................................... 61

Relative viscosity as a function of volume ﬁaction of PSD-C in the
ﬁller. The balance of the ﬁller is either PSD—A or PSD-B. The
curves correspond to values obtained from the Krieger-Dougherty
equation. The viscosity of the suspending medium, 770, is 0.0456
Pa.s (45.6 cP) ............................................................................................ 62

Relative viscosity as a function of weight percent of silane coupling
agent. The viscosity of the suspending medium, 00, is 0.0456 Pa-s
(45.6 CF) .................................................................................................... 63

xii

Chapter 6

Figure 6.1. Flexural modulus as a function of UV illtunination time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica ..................................... 87

Figure 6.2. Flexural strength as a function of UV illumination time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica ..................................... 88

Figure 6.3. Flexural modulus as a function of postcure time for samples
containing 70. O, 72. 0, and 74. 0 wt.% fused silica. Samples were
photocured for 2 minutes with UV light of 200 mW/cm2 UVA
intensity and then postcured at 170°C 1n a laboratory oven ...................... 89

Figure 6.4. Flexural strength as a function of postcure time for samples
containing 70. 0, 72. 0, and 74. 0 wt.% fused silica. Samples were
photocured for 2 minutes with UV light of 200 mW/cm2 UVA

intensity and then postcured at 170°C m a laboratory oven ...................... 90

Figure 6.5. Representative DSC reaction proﬁle at 110°C for the liquid
formulation containing 74.0 wt.% fused silica, 1.0 wt.% silane
coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on
resin weight), 1.2 wt.% benzoyl peroxide (based on resin weight),
and the balance epoxy novolac-based vinyl ester resin ............................. 91

Figure 6.6. CTE below glass transition as a function of postcure time for
samples containing 70. 0, 72. 0, and 74. 0 wt. % fused silica. Samples
were photocured for 2 minutes with UV light of 200 mW/cm2 UVA
intensity and then postcured at 170°C 1n a laboratory oven ...................... 92

Figure 6.7. CTE above glass transition as a function of postcure time for
samples containing 70. 0, 72. 0, and 74. 0 wt. % fused silica. Samples
were photocured for 2 minutes with UV light of 200 mW/cm2 UVA
intensity and then postcured at 170°C 1n a laboratory oven ...................... 93

Figure 6.8. Glass transition temperature as a function of postcure time for
samples containing 70. 0, 72. 0, and 74. 0 wt. % fused silica. Samples
were photocured for 2 minutes with UV light of 200 mW/cm2 UVA
intensity and then postcured at 170°C 1n a laboratory oven ...................... 94

Chapter 8

Figure 8.1. Thermal conductivity of PLEs at 25°C as a function of fused silica
loading ..................................................................................................... 125

Figure 8.2. Comparison of experimental thermal conductivity results to the
predicted values obtained using the model of Agari and Uno from 0
to 80.0 wt.% fused silica (0 to 65.6 vol.%) ............................................. 126

xiii

Figure 8.3.

Figure 8.4.

Figure 8.5.

Figure 9.1.
Figure 9.2.
Figure 9.3.

Figure 9.4.

Figure 9.5.

Figure 9.6.

Figure 9.7.

Figure 9.8.

Comparison of experimental thermal conductivity results to the
predicted values obtained using the model of Agari and Uno from
60.0 to 80.0 wt.% fused silica (41.7 to 65.6 vol.%) ................................ 127
Speciﬁc heat of PLEs at 25°C as a function of fused silica loading ....... 128

Thermal diffusivity of PLEs at 25°C as a function of fused silica

loading ..................................................................................................... 129
Chapter 9

Scanning electron micrograph of Alloy 194 strip before cleaning ......... 142

Scanning electron micrograph of Alloy 194 aﬁer cleaning .................... 143

Peel strength between photopolymerizable liquid encapsulants and
Alloy 194 strips treated with silane coupling agents ............................... 144

Scanning electron micrograph of peel test specimen showing
cohesive failure between PLE ﬁlled with 72 wt.% fused silica and
Alloy 194 strip without any silane. The particles are remnants of the
PLE adhered to the Alloy 194 surface after the 180 degree peel test ..... 145

Scanning electron micrograph of peel test specimen showing
cohesive failure between PLE ﬁlled with 72 wt.% fused silica and
Alloy 194 strip treated with MPMS. The particles are remnants of
the PLE adhered to the Alloy 194 surface after the 180 degree peel
test ........................................................................................................... 146

Scanning electron micrograph of peel test specimen showing
cohesive failure between PLE ﬁlled with 72 wt.% fused silica and
Alloy 194 strip treated with AAMS. The particles are remnants of
the PLE adhered to the Alloy 194 surface after the 180 degree peel
test ........................................................................................................... 147

Scanning electron micrograph of peel test specimen showing
adhesive failure between PLE ﬁlled with 74 wt.% fused silica and
Alloy 194 strip treated with BTCA. Note the lack of any signiﬁcant
amounts of the PLE remaining on the copper surface ............................ 148

Scanning electron micrograph of peel test specimen showing
adhesive failure between PLE ﬁlled with 74 wt.% ﬁrsed silica and
Alloy 194 strip treated with PBI. Note the lack of any signiﬁcant
amounts of the PLE remaining on the copper surface ............................ 149

xiv

Chapter 1
INTRODUCTION

1.1. OVERVIEW AND IMPORTANCE OF MICROELECTRONIC
ENCAPSULATION

The microelectronics, communication, and computer industries have undergone a
major shift from highly sophisticated and expensive mainframe systems to small-scale
systems, such as personal computers and communication devices. This trend arises from
the continuous push for the integration of many functions on a single chip. Very large
scale integration (V LSI) and ultra large scale integration (ULSI) ultimately are made
possible by the ability to fabricate devices with smaller and smaller feature sizes.

As the feature sizes on a semiconductor chip are reduced, the probability of
contamination or damage (by scratching, handling, corrosion, etc.) is dramatically
increased, and consequently there is a drastic reduction in the yield of the process.'°3 For
this reason, silicon chips are encapsulated in a plastic or ceramic package to protect the
chip from adverse mechanical, chemical, electrical, and thermal environments. Another
consequence of the increasingly large integration of features on a chip is the use of
smaller and ﬁner wires to electrically connect the semiconductor chip to the leadframe.
During the encapsulation of microelectronic devices by the transfer molding process,
these extremely ﬁne wires are subjected to ﬂow stresses from the molding compound and
may be deformed or displaced (a phenomenon known as wire sweep). If the deformation
of a bonding wire during encapsulation is sufﬁciently large, it may cause a short circuit
(due to adjacent wires touching) or an open circuit (due to wires breaking).4 If this

happens, the device cannot be repaired and must be discarded. Thus, minimizing wire

sweep during encapsulation is of prime importance, and is becoming more critical due to
the use of an increasing number of wires in microelectronic packaging.

The process of chip encapsulation is particularly important since the chip has
already undergone numerous processing steps and is just a few steps away from being a
ﬁnished product (and thus has considerable value that is lost if the encapsulation step
fails). With the currently used transfer molding process, it is very difﬁcult to control the
problem of wire sweep in a high pin count, ﬁne pitch package. New trends in die design
and packaging pose an increasing challenge to the wire sweep problem?’3 Indeed, due to
the continuing advances in device capabilities and rapid changes in circuit board
assembly methods, the packaging and encapsulation steps are more important than ever
before, and may, for the ﬁrst time, impose limits on the design and performance of the
ﬁnal microelectronic device.5

Microelectronic devices are typically encapsulated in a protective therrnoset body
from which a number of leads extend to allow electrical contact and interconnection
between the encapsulated semiconductor device and a printed circuit board.
Encapsulation of microelectronic devices has traditionally been accomplished using a
transfer molding process. In this process, the thermoset molding compound (typically a
solid epoxy preform) is dielectrically preheated and then placed into the pot of the
molding tool. A transfer cylinder, or plunger, is used to push the molding compound into
the runner system and gates of the mold.’ The molding compound then ﬂows over the
chips, vvirebonds, and leadframes, encapsulating the entire assembly.

Most transfer molding processes suffer from signiﬁcant problems arising from

high operating temperatures (the molding compound is a solid at room temperature) and

high pressures required to ﬁll the mold (even in the melt state, the molding compound has
a high viscosity, and the viscosity increases further with reaction). These aspects of the
current process can lead to incomplete mold ﬁlling, thermal stresses (since the reaction
temperature is much higher than the ﬁnal use temperature), and wire sweep. Therefore,
current transfer molding processes are plagued by two important problems: i) wire sweep
due to the high initial melt viscosity, and ii) inadequate time to ﬁll the mold due to rapid
cure (concurrent with mold-ﬁlling) at the elevated operating temperatures. As the
package sizes and the associated wires shrink to smaller and smaller dimensions, so does
the operating window for transfer molding; hence, it is difﬁcult to control wire sweep in
the high pin count, ﬁne pitch packages. In turn, these problems lead to decreases in
productivity for the new ﬁne pitch microelectronic devices due to increased packaging-
related rejects.
1.2. SCOPE AND AIM OF RESEARCH

In this research, we are investigating a new strategy for microelectronic
encapsulation using photopolymerizable liquid encapsulants (PLEs). These highly ﬁlled
PLEs consist of an epoxy novolac-based vinyl ester resin (~25 wt.%), fused silica (70-74
wt.%), and small amounts of a photoinitiator, thermal initiator, and silane coupling agent.
For these encapsulants, the use of light rather than heat to initiate the polymerization
allows precise control over when the reaction starts, and therefore completely decouples
the mold ﬁlling and the cure. The relatively low viscosity of the PLE allows low
operating pressures and minimizes problems associated with wire sweep. Because

photoinitiation allows active centers to be formed rapidly at room temperature, the in-

mold cure time for the PLEs is no longer than the in-mold cure times of current transfer
molding compounds.

In recent years, a number of investigators (including our research group) have
demonstrated the feasibility of photopolymerizing thick polymers and compositesﬁ’7 In
addition, Hanneman et al. recently demonstrated the feasibility of using
photopolymerizations to produce molded acrylate parts.8 These authors molded unﬁlled
poly(methylmethacrylate) and found that they could produce high quality parts more than
ﬁve millimeters thick. Therefore, the potential exists to develop a photomolding process
to encapsulate microelectronic devices. Before this can be accomplished, however, a
suitable photopolymerizable formulation must be developed which possesses the
appropriate material properties that can effectively protect microelectronic devices. This
is the main objective of this research project.

In this thesis, an extensive series of studies on the thermal and mechanical
properties, as well as the viscosity and adhesion of photopolymerizable liquid
encapsulants, are reported in order to demonstrate that a UV-curable formulation can,
indeed, have the necessary material properties required for microelectronic encapsulation.
Speciﬁcally, we have characterized the viscosity, ﬂexural strength and modulus,
coefﬁcient of thermal expansion, glass transition temperature, thermal stress parameter,
thermal conductivity, thermal diffusivity, speciﬁc heat, and adhesive peel strength of
PLEs containing 70.0-74.0 wt.% fused silica. In addition, the effects of the illumination
time, postcure time, fused silica loading, and the inclusion of a thermal initiator on the

thermal and mechanical properties of the ﬁnal cured encapsulants have been investigated.

1.3. REFERENCES
. W.R. Runyan and KB. Bean, Semiconductor Integrated Circuit Packaging, Addison
Wesley, Reading, MA, (1990).

. F. Primeaux, “Molded Plastic Package with Wire Protection”, US Patent 5331205,
(1994)

. M.B. McShane, A.H. Woosley, F. Primeaux, “Method for Encapsulating
Semiconductor Devices with Package Bodies”, US Patent 5344600, (1994).

. S. Han and K.K. Wang, J Electron Packaging, 117, 178, (1995).

. L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, New York, NY, (1990).

. V. Narayanan and AB. Scranton, Trends in Polym Sci, 5, 415, (1997).

. V. Narayanan, K.K. Baikerikar, A.B. Scranton, RadTech ’98 North America U V/EB
Conference Proceedings, 31, (1998).

. T. Hanemann, R. Ruprecht, J .H. HauBelt, Polym Mat Sci Eng, 77, 418, (1997).

Chapter 2
BACKGROUND

2.1. COMPONENTS OF A MICROELECTRONIC DEVICE
The microelectronic device (also known as the plastic package) consists of: (i) a
silicon chip, (ii) a metal support or leadframe, (iii) metal wires that electrically
interconnect the chip circuits to the leadframe, (iv) an adhesive to physically attach the
chip to the leadframe, and (v) the plastic encapsulation material (molding compound) to

protect the chip and the wire interconnects, as shown in Figure 2.1:

Molding comoomd

  
  
 
 
  

Silicon chip

Gold wire

Gold/Silver plating

ead frame

Figure 2.1. Schematic of a microelectronic device.

The leadfrarne is the main supporting structure of the device to which every other
element is attached. Etched or stamped from a thin metal strip, the leadfrarne carries the
chip throughout the assembly process and is imbedded in plastic after molding. The
functional requirements of the leadfrarne as described by Robock and Nguyenl are: (i) a

holding ﬁxture that indexes with tool-transfer mechanisms as the package proceeds

through various assembly operations, (ii) a dam that prevents the encapsulation material
ﬁ'om rushing out between leads during the transfer molding operation, (iii) chip attach
substrate, (iv) support matrix for the encapsulation material, and (v) electrical and
thermal conductor from chip to board. Typically, the leadfrarne is made of Alloy 42
(58% iron, 42% nickel), Alloy 50 (50% iron, 50% nickel), or a copper alloy (Alloy 194,
Alloy 151), and is plated with gold and silver or palladium, either completely or in
selected areas over nickel or nickel/cobalt.2 Selecting the appropriate leadfrarne material
requires the evaluation of a number of issues. For example, the leadfrarne material plays
a signiﬁcant role in determining the degree of adhesion between the molding compound
and the leadframe, the therrnomechanical shrinkage stresses, the heat dissipation capacity
of the package, and the buckling of the leadfrarne after molding. All of these issues have
direct implications for packaging yield and reliability.3 Alloy 42 is one of the most
commonly used leadfrarne materials because of its low coefﬁcient of thermal expansion
(CTE), as well as its resistance to buckling under the compressive loads applied by the
molded body (Manzione, 96). The small mismatch in coefﬁcients of thermal expansion
between Alloy 42 (CTE = 4.5 x 10'6 °c") and the silicon chip (CTE = 2.6 x 10* °C°')
results in lower therrnomechanical shrinkage stresses. In addition, Alloy 42 exhibits a
higher degree of adhesion to molding compounds because it does not form a passivating
oxide ﬁlm as easily as the copper alloys.3 However, its main drawback is its low thermal
conductivity, which limits the amount of heat dissipated from the chip to the printed
circuit board.

Because of their higher coefﬁcients of thermal expansion, copper alloys (CTE =

16 x 10" °C") will have larger therrnomechanical shrinkage stresses. However, they do

have much higher thermal conductivities than other alloys and this is the principal reason
for their use. Heat conduction along the leadfrarne is an important heat dissipation
mechanism in microelectronic devices and the higher thermal conductivity of copper
alloys makes them an attractive choice for high-power devices.3

The silicon chip is usually mounted to the leadfrarne with what is known as a die
attach adhesive. Previously, solders were used to mechanically attach the chip to the
leadframe, but now these are being displaced by conductive polymer adhesives because
they are less expensive and their application is more easily automated.3 Presently,
solventless conductive epoxies ﬁlled with silver ﬂakes are the most widely used die
attach adhesive. To electrically connect the bonding pads of the die with the leads of the
frame, gold, aluminum, or even copper wires are used. For gold wires, the typical

diameters are 0.0025 to 0.0032 cm, while for aluminum, they are up to 0.005 cm.3

2.2. COMPONENTS OF THE ENCAPSULATION MATERIAL

Encapsulation materials are usually proprietary mixtures of at least eight
categories of raw materials, including the epoxy resin, curing agent (hardener),
accelerator (catalyst), ﬁller, coupling agent, ﬂame retardant, mold release agent, and
colorant."3
2.2.1. Epoxy Resin

The epoxy resin is commonly a cresol-novolac epoxy and is the base upon which
the encapsulation material is built, accounting for 25 to 30 wt.% of the formulation. As
Robock and Nguyenl explain, “It is the organic binder that, when cured, holds all the

constituents together and imparts the inherent characteristics of high T8: shrinkage on

cure, low moisture absorption, and chemical inertness to the plastic package”. The
excellent high temperature properties of epoxy novolacs are the primary reason they have
replaced the bisphenol epoxies, phenolics, and silicones as the major compound used for
integrated circuit packaging.
2.2.2. Curing Agent

Curing agents or hardeners such as phenols, amines, and acid anhydrides are used
to crosslink the epoxy resins. Depending on the type of resin and hardener in the
formulation, the concentration of hardener can range from as low as 1 part of hardener
per 100 parts of resin (1 phr) to 100 phr.I Currently, phenol novolac and cresol novolac
hardeners are the most widely used because of their excellent moldability, electrical
properties, and heat and humidity resistance.3
2.2.3. Accelerator

Accelerators act as catalysts to accelerate crosslinking of the epoxy when heat and
pressure are applied in the molding process, thus reducing in-mold cure times and
improving productivity."3 They also tend to affect the electrical properties and humidity
resistance of the cured molding compound. Accelerators that are usually used include
amines, imidazoles, organophosphines, urea derivatives, and Lewis bases and their
organic salts.‘1 The problem with accelerators is that their reactions are initiated at
comparatively low temperatures, thus deteriorating the moldability or pot life of the
molding compound.4 Therefore, accelerators that promote rapid curing of resins when
heated to a certain temperature, without promoting curing at lower temperatures, are
preferred. This is why organophosphine compounds and Lewis base salts are widely

used in epoxy molding compounds.

2.2.4. Filler

Epoxy resins, curing agents, and accelerators alone cannot be used for packaging
microelectronic devices, because these materials do not yield optimal material properties
to the ﬁnal encapsulant. For this reason, inert materials called ﬁllers are used. The
inclusion of ﬁllers in the encapsulation material offers a number of beneﬁts, most
notably, lower coefﬁcient of thermal expansion, increased thermal conductivity,
improved mechanical properties, and prevention of excessive resin bleed at the parting
line of the molding tool.3 In order to obtain these advantages, ﬁllers are added in large
amounts, anywhere from 70 to 80 wt.%. Fillers that have been used in epoxy molding
compounds include fused silica, crystalline silica, aluminum oxide, and aluminum nitride,
with ﬁised silica being the most commonly used. The widespread use of ﬁrsed silica is
due to its very low coefﬁcient of thermal expansion (CTE = 0.5-0.6 x 10'6 °C") and its

3 However, the

low abrasiveness, which provides a reasonable molding tool lifetime.
primary drawback of fused silica is its low thermal conductivity (1.5 W/m-K) and this is
the reason high thermal conductivity ﬁllers such as crystalline silica (3.0 W/moK),
aluminum oxide (42.0 W/moK), and aluminum nitride (150-220 W/m.K) are ﬁnding
more use.5 High thermal conductivity ﬁllers are required as a cost-effective means of
removing heat from plastic packages. Besides selecting the appropriate ﬁller, the proper
selection of the ﬁller particle size and distribution also plays a key role in determining the
encapsulation material’s properties. Typically, fused silica particles are irregularly
shaped ﬂakes that range in size from 10 to 25 microns, but the size distribution can be

wide with some particles reaching 125 microns.3 This can present a problem because the

large particles can become trapped in the gates (a gate admits the molding compound into

10

the cavity of the device to be molded) which can be as thin as 150 microns or caught
between the leads of packages where the interlead spacing is 100 microns.3 Usually, a
blend of small particles of different size distributions is used in order to maximize ﬁller
loadings, without a substantial increase in viscosity. This is one of the main reasons for
the use of spherical particles, which are used in conjunction with ﬂakes.
2.2.5. Coupling Agent

The purpose of coupling agents is to bind the polymer matrix and the inorganic
ﬁller in order to improve the physical properties and maximize the ﬁller-to-polymer ratio.
Coupling agents used for molding compounds include silanes, aluminum chelates,
titanates, and zircoaluminates, with silanes being the most commonly used coupling
agent. Silanes are organofunctional compounds that act as molecular bridges at the
interface between the polymer matrix and the ﬁller particle surfaces of the composites.
They are linear molecules which have a reactive group at one end capable of being
incorporated into the matrix of a curing polymer and a silicon-based group at the other
end which bonds to inorganic ﬁllers. Beneﬁts of silanes include reduced viscosity of the
molding compound, improved processability, better coating adhesion to substrates,
improved humidity and heat resistance, and improved mechanical and electrical
properties.
2.2.6. Flame Retardant

Epoxy materials are not inherently ﬂame retardant, so ﬂame retardancy is
required for epoxy molding compounds to insure safety. Product safety requirements for

plastic packages used in electronic equipment specify an Underwriter’s Lab ﬂame

ll

retardant ratio of 94V-O. This can be achieved only by the addition of ﬂame retardants
such as tetrabromobisphenol-A and antimony trioxide to the molding compound.l
2.2.7. Mold Release Agent

Because of the excellent adhesion of epoxies to all types of surfaces, mold release
agents are used to ease the release of the cured encapsulants from compounding and
molding equipment. The most commonly used mold release agents for epoxy molding
compounds include hydrocarbon waxes, silicones, metallic salts of organic acids, and
ﬂuorocarbons. Of these, hydrocarbon waxes, such as carnauba wax, are the most widely
used mold release agents for microelectronic encapsulation.2
2.2.8. Coloring Agent

Coloring agents are used primarily for aesthetic purposes, since the ﬁlled molding

2 They are also used to

compound without any coloring agent is a translucent yellow.
distinguish different device types in packages. Carbon black is the predominant coloring
agent for plastic packaging of microelectronic devices, even though it degrades the

moisture resistance and slightly increases the electrical conductivity of the molding

compound. For this reason, it is used in concentrations less than 0.5%.

2.3. MATERIAL PROPERTIES USED TO EVALUATE
ENCAPSULATION MATERIALS

An encapsulant is an electrically insulating, plastic material that protects an
electric or electronic component from adverse mechanical, chemical, electrical, and
thermal environments. Therefore, materials used for this application must deliver a
balanced combination of properties in the areas of viscosity, mechanical strength and

toughness, chemical resistance, electric and dielectric insulating performance, thermal

12

conductivity, adhesion, and thermal and moisture stability in the general temperature
range of —65 to 200°C.6 Some of the more important material properties used to evaluate
encapsulation materials are described in more detail below.
2.3.1. Viscosity

The microelectronic encapsulation process requires a low viscosity of the molding
compound so it can ﬂow into the mold with minimum deformation of the leadfrarne and
wirebonds.2 During the encapsulation of microelectronic devices by the transfer molding
process, these extremely ﬁne wires are subjected to ﬂow stresses from the molding
compound and may be deformed or displaced (a phenomenon known as wire sweep). If
the deformation of a bonding wire during encapsulation is sufﬁciently large, it may cause
a short circuit (due to adjacent wires touching) or an open circuit (due to wires
breaking).7 If this happens, the device cannot be repaired and must be discarded. Thus,
reducing the viscosity of the encapsulation material in order to reduce wire sweep is of
prime importance, and is becoming more critical due to the use of an increasing number
of wires in microelectronic packaging.
2.3.2. Flexural Modulus

An important mechanical property for microelectronic encapsulants is the ﬂexural
modulus, which characterizes the stiffness of a material under an applied load.8 The
optimal value of the ﬂexural modulus provides a balance between two conﬂicting
requirements. A high value of the ﬂexural modulus is desired to maximize the protection
of the chip from an applied load; however, lower values of the ﬂexural modulus lead to
lower thermal stresses. Recommended ﬂexural modulus values for microelectronic

encapsulants are around 1000 kgp/mmz.3

13

2.3.3. Coefﬁcient of Thermal Expansion

The linear coefﬁcient of thermal expansion (CTE) is deﬁned as the ratio of the
change in the length of the sample to the change in temperature per unit initial length.9
This parameter is very important in semiconductor encapsulation because a signiﬁcant
CTE mismatch between the encapsulant, leadframe, and the silicon chip can lead to the
build up of internal stresses in the semiconductor device, and could lead to cracking of
the chip or encapsulant. Current encapsulants exhibit on values (CTE below glass
transition) in the range of 16-25 um/m°C.
2.3.4. Glass Transition Temperature

The glass transition temperature is deﬁned as the temperature (or more precisely,
the temperature region) at which a vitreous material changes from a hard and relatively
brittle condition to a viscous or rubbery condition.2 Above this temperature region, the
mechanical, dielectric, and thermal properties change dramatically. Indeed, below the
glass transition temperature, the coefﬁcient of thermal expansion is low and nearly
constant, but above the glass transition temperature, the coefﬁcient of thermal expansion
is very high. As a result, encapsulants must have high glass transition temperatures.
Commercial encapsulants typically exhibit glass transition temperatures in the range of
140 to 180°C.
2.3.5. Thermal Conductivity

Since the microelectronic device generates a signiﬁcant amount of heat during
use, the thermal conductivity of an encapsulating compound plays an important role in
the thermal management of the device. Indeed, the thermal conductivity determines the

amount of heat that can be removed from the device and thus affects operating

14

temperatures of electronic components. As semiconductor devices increase in transistor
densities, the heat dissipation requirements also increase, resulting in the need for
packaging materials with enhanced thermal conductivity.
2.3.6. Adhesion

The adhesion of an epoxy molding compound to the leadfrarne is a critical
parameter for encapsulated microelectronic devices. The on-going trends of increasing
silicon chip size and thinner packages, as well as the push towards surface mount
operations, have increased the importance of proper adhesion between the molding

10'” Speciﬁcally, improved adhesion between the molding

compound and leadfrarne.
compound and the leadfrarne leads to enhanced reliability and resistance to package and
chip cracking. In addition, inadequate adhesion leads to popcorn problems in surface
mount devices, as well as delarnination and subsequent thermomechanical failure?“
Furthermore, a high degree of adhesion limits moisture-induced corrosion by preventing

penetration of water along the leadfrarne. '2

2.4. SEMICONDUCTOR MANUFACTURING PROCESS
In order to gain a better appreciation for plastic packaging and its role in the
microelectronics industry, it is imperative to review the steps in the assembly of

microelectronic devices. The sequence of steps is ﬁrst listed here in Table 2.1. and then

described bcrcw.3

15

Table 2.1. Sequence of steps in the assembly of microelectronic devices

 

 

 

 

 

 

 

 

1. Wafer Separation 7. Postcure

2. Die Attach 8. Trim-and-Form

3. Die Interconnection 9. Solder Dipping/Solder Plating
4. Molding 10. Bum-in

5. Deﬂashing 11. Testing

6. Code Marking

 

 

2.4.1. Wafer Separation and Die Attach

The ﬁrst step involves sawing the silicon wafers into individual chips using a
diamond-tipped saw blade. The sawed wafers are then placed on the paddle of the
leadﬁ'ame, which has already been applied with the die attach adhesive. The adhesive not
only physically attaches the chip to the leadframe, but also, because it is conductive,
allows electrical contacts to be made to the back of the chip. To cure the adhesive, the
leadframes are passed through an oven.
2.4.2. Die Interconnection

To electrically interconnect the silicon chip and the leadfrarne (die
interconnection), two methods are used: wire bonding and tape automated bonding
(TAB). In plastic encapsulated microdevices, wirebonding is the dominant
interconnection technology, and therefore it will receive primary attention.2 In wire
bonding, gold or aluminum wires are used to electrically interconnect the bonding pads of
the chip with the leads of the frame. This is done with automated machinery which ﬂame
softens the tip of the wire and then mashes it against the bonding pad of the chip. The
wire is then looped out and formed into a wedge bond on the leadfrarne ﬁnger.3 The
bond length and the loop shape are chosen so as to minimize wire displacement, with

bond lengths ranging from 0.15 to 0.30 cm and loop heights about 0.075 cm above the

16

plane of the device.3 Longer wire spans (due to placing a small chip on a large leadfrarne
paddle) tend to be avoided because of the greater chance of wire deformation during
molding.
2.4.3. Transfer Molding

After the die attach and interconnection steps, the chip, wirebond, leadfrarne
assembly is ready for transfer molding, the most popular encapsulation method for
essentially all plastic packages in integrated circuit technology.2 Equipment for transfer
molding includes a transfer molding press with the appropriate molding tool, a dielectric
preheater, a cure oven, and frame and cassette loaders. In this process, the cassettes of
leadframes are loaded into the mold and then the mold is closed and clamping pressure is
applied. The epoxy molding compound, in the form of a solid preform, is dielectrically
preheated (to 90-95°C) and then placed into the pot of the molding tool. A transfer
cylinder, or plunger, is used to apply the transfer pressure and force the molding
compound through the runners and gates of the mold and into the cavities. The molding
compound then ﬂows over the chips, wirebonds, and leadframes, encapsulating the
semiconductor device. After 2 to 3 minutes at approximately 175°C, the epoxy molding
compound is cured to a sufﬁcient degree of conversion (typically 60 to 90%) in the mold.
The mold is then opened and the molded devices are ejected and ready for the next
processing steps.
2.4.4. Deﬂashing and Code Marking

During transfer molding, excess epoxy molding compound can inadvertently ﬂow
through the mold parting line and onto the leads of the device. This excess material

(known as ﬂash) can cause problems in the downstream operations of lead trimming,

17

forming, and solder dipping and/or plating, if not taken care of.2 Therefore, this material
is removed by a process known as deﬂashing, in which a mixture of pressurized air and
an abrasive are used to mechanically remove the material from the surface of the
leadfrarne. After deﬂashing, information is printed on the top surface of the package in
an operation known as code marking. This marking is typically done using laser writing
or polymer-based inks and includes information such as the device code, the
manufacturer, and the date of manufacture for product identiﬁcation and traceability.
2.4.5. Postcure

The plastic packages are now postcured and placed in large batches in ovens for 4
to 8 hours at 170-180°C in order obtain conversions near 100%, and thus fully develop
the mechanical and thermal properties of the cured encapsulant. Postcure is necessary
because the molding compound does not achieve complete * conversion in the 2 to 3
minute cycle time (conversions typically range from 60-90%) and it is inefficient to wait
for full conversion in the mold.
2.4.6. Trim-and-Form

In the lead trimming and forming operations, the peripheral leadfrarne metal that
joins all the leads is cut and the leads are bent (fornred) to deﬁne speciﬁc lead shapes.
Using a punch press in which a trim-and-fonn tool is mounted, the tool sequentially steps
through the cutting and forming of the lead ﬁngers.
2.4.7. Solder Dipping/Solder Plating

Since most integrated circuit devices are ultimately soldered to a circuit board,

solder dipping and plating are used to ease solder attachment. Solder dipping simply

18

involves dipping the leads into molten solder, while solder plating uses plating chemicals.
Typically, a thin coat of solder less than 0.0005 inch is applied.3
2.4.8. Burn-in

Bum-in is a process step where the devices are screened for premature failure at
high temperature and high electrical loading.3 Burn-in is accomplished by mounting the
devices on special circuit boards that apply the electrical loading, and these boards are
then plugged into large oven chambers that maintain a constant high temperature. A
commonly used burn-in cycle consists of a voltage loading 1-2 volts above the nominal 5
volt operating range typical for most devices at a temperature of 120°C for 9-24 hours.3
2.4.9. Testing

Testing is a ﬁnal means of evaluating devices, as well as, a means for process
analysis and improvement. A full function test on 100% of the packaged devices is
standard in the microelectronics industry. Techniques used for testing include x-ray
analysis, de-encapsulation, microscopy, scanning laser acoustic microscopy, differential

scanning calorimetry, and sectioning.3

2.5. ENCAPSULATION OF MICROELECTRONIC DEVICES BY
TRANSFER MOLDING

As was mentioned earlier, transfer molding is the predominant method for
encapsulating integrated circuits.8 Therefore, further details about this process will be
described here. Transfer molding is deﬁned as a method for molding thermosetting
materials in which the material is subjected to heat and pressure and then forced into a
closed mold cavity by this same pressure and held there under additional heat and

pressure until cure is complete.8 Transfer molding equipment for microelectronic devices

19

consists of a hydraulic press equipped with platens, one of which contains a chamber
known as the pot, in which the molding compound is placed and liqueﬁed by a
combination of pressure and heat. A plunger transfers the molten material at pressures
typically around 1000 psi into the cavities of the mold via a series of channels (known as
runners), as shown in Figure 2.2. Transfer times are usually between 15 and 20 seconds,
although it is not uncommon for some compounds and molds to go outside these limits.8
The molding compound then ﬂows over the chip-wirebond—leadframe assembly (located
in the cavities), encapsulating the microelectronic device. At the end of the transfer
stroke, the two mold halves remain clamped for a short time to allow the resin to react
further. Mold temperatures are normally 170-175°C. Once the molding compound has
reached an ejectabie hardness (typically after two to three minutes), the encapsulated
devices are removed from the mold and postcured to fully develop the material properties

of the encapsulant.

   
 

Plunger

  
  

Epoxy molding

wirebonded parts

Figure 2.2. Transfer molding process for encapsulating microelectronic devices.13

20

2.6. DISADVANTAGES OF TRANSFER MOLDING

Although transfer molding is the predominant means of encapsulating silicon
chips, it does suffer from a number of problems. Three problems commonly encountered
during transfer molding are: i) incomplete mold ﬁlling; ii) leadfrarne movement during
cavity ﬁlling; and iii) wire sweep, which is displacement or deformation of the wires that
provide electrical connection between the chip and the leadfrarne. Each of these molding
problems is directly related to the processing conditions and the rheological
characteristics of the molding compound used for encapsulation.l4 Of signiﬁcant interest
is wire sweep, which is the largest and predominant cause of defects in the encapsulation
of microelectronic devices.l

Most of the limitations of transfer molding in high volume production arise from
the fact that the resin system begins to react before and as it enters the mold due to high
operating temperatures. As a result, the resin becomes very viscous, necessitating the use
for high operating pressures during transfer molding. All of these factors contribute to
the wire sweep problem. As the package sizes and the associated wires shrink to smaller
and smaller dimensions, so does the operating window for transfer molding; hence, it is
difficult to control wire sweep in the high pin count, ﬁne pitch packages. In turn, these
problems lead to productivity decreases due to increased packaging-related rejects.
Moreover, the capital costs associated with transfer molding are quite high.

Photopolymerizable liquid encapsulants (PLEs) for microelectronic devices may
offer important advantages over traditional transfer molding compounds (which are solid

at room temperature), including alleviation of the three problems mentioned above. In

21

order to fully appreciate the beneﬁts that a photopolymerizable encapsulant can offer, it is

important to ﬁrst review the area of photopolymerizations.

2.7. OVERVIEW OF PHOTOPOLYMERIZATIONS

As the name implies, photopolymerizations are polymerization reactions that are
initiated by light. Photoinitiation is achieved by the addition of a photoinitiator (and
perhaps a photosensitizer) which produces active centers upon absorption of light of the
appropriate wavelength. Once the active centers are produced, the polymerization
proceeds by propagation of the active center with successive monomer units, just like in
traditional thermal polymerizations.

Photopolymerizations offer a number of advantages that make them attractive for
a variety of applications. For example, photopolymerizable formulations are typically
solvent-free, thereby minimizing volatile organic emissions. Moreover, since active
centers may be produced rapidly and efﬁciently using photochemical processes,
photopolymerizations provide high production rates, and are very energy efﬁcient
compared to thermal systems, in which the entire reaction system is raised to an elevated
temperature. Finally, photopolymerizations provide a great deal of control over the
initiation reaction: spatial control, since the light may be directed to locations of interest
in the system; and temporal control, since the light may be readily shuttered on or off.
These advantages have led to the widespread use of photopolymerizations in commercial
applications such as ﬁlms, inks, and coatings for a variety of substrates (i.e. paper, metal,
plastic, and wood). Although photopolymerizations have not been used to replace the

transfer molding process used for encapsulating microelectronic devices,

22

photopolymerizations have still found extensive use in the electronics manufacturing
industry; speciﬁcally, as UV curable adhesives (for applications in strain-relief, wire and
parts tacking, coil terminating, tamper-prooﬁng, structural bonding, and surface mount
component attachment), as well as, UV curable coatings for potting, glob-topping, and

conformal coatings.ls

2.8. PHOTOPOLYMERIZATION OF THICK POLYMERS AND COMPOSITES

Photopolymerization of thick polymers and ﬁlled composites (such as
encapsulants for microelectronic devices) is more challenging than photopolymerization
of thin-ﬁlm systems due to the exponential reduction in light intensity through the sample
resulting from absorption and scattering. Therefore, for thick polymers and ﬁlled
composites, proper selection of the illumination wavelength, initiating system (type of
initiator, photosensitizer, concentration, etc.), resin, ﬁber or ﬁller, and composite
thickness is imperative to ensure that the samples cure throughout (including the portions
of the sample furthest from the light). Speciﬁcally, the initiating wavelength must be
carefully selected to ensure that the resin and ﬁller are transparent to this wavelength, so
that only the photoinitiator signiﬁcantly absorbs light. This is accomplished by choosing
a photoinitiator that has good absorbance in a spectral region different from where the
other components of the system (resin, ﬁller, etc.) absorb. In addition, the photoinitiator
must “photobleach” at the initiating wavelength; that is, the initiating fragments should
not absorb light at the initiating wavelength. Thus, the photoinitiator becomes
transparent upon production of active centers, ensuring efficient penetration of light into

the sample.

23

2.9. ADVANTAGES OF USING PHOTOPOLYMERIZATIONS TO
ENCAPSULATE MICRODEVICES

As was mentioned earlier, photopolymerizations and a photopolymerizable liquid
encapsulant offer a number of advantages that could alleviate many of the problems
encountered in the conventional transfer molding process used to encapsulate
microelectronic devices. In particular, the temporal control of photopolymerizations
would allow the mold to be completely ﬁlled with a low viscosity resin before any
reaction takes place, and a fast therrnoset reaction could be initiated after the mold is
completely ﬁlled. This decoupling of mold ﬁlling and reaction would permit complete
control of the time allowed for ﬂow of the resin formulation and would reduce waste and
expensive clean-up by preventing premature reaction (gelation in the transfer lines). In
addition, since there would be no need for high pressures or external heating of the mold,
the photomolding process would be very energy efﬁcient. Most importantly, the
decoupling of mold ﬁlling and reaction would reduce the wire sweep problem.

Aside from the beneﬁts of using photopolymerizations, there are a number of
advantages to using a liquid formulation, as well. Speciﬁcally, liquid encapsulation
dispensing equipment is less expensive than molding presses, especially when in-line
automated factories are targeted. In fact, a cost analysis by Robinson et al. '6 comparing
the cost of molding and liquid encapsulation shows that, “liquid encapsulation is
signiﬁcantly lower in cost than molding and has the potential to displace mold
compounds in several applications”. Furthermore, the authors write that using liquid
encapsulants, “will allow package design changes without costly retooling. Minor

package and dispense changes are quickly implemented and immediately veriﬁable.

24

Shorter cure times, reduced cycle time, and low initial capital expense requirements will

also enable higher manufacturing productivity”. '6

25

2.10. REFERENCES

. P.V. Robock and LT. Nguyen, Microelectronics Packaging Handbook, eds. R.R.
Tummala and El. Rymaszewski, Van Nostrand Reinhold, New York, NY, 523-672,
(1989)

. M.G. Pecht, Plastic-Encapsulated Microelectronics, John Wiley & Sons, Inc., New
York, NY, (1995).

. L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, (1990).

. N. Kinjo, M. Ogata, K. Nishi, A. Kaneda, Adv Polym Sci, 88, 5, (1989).
. P. Procter and J. Solc, IEEE T Compon Hybr, 14, 4, 708, (1991).

. M. Shinohara, Electronic Materials Handbook, ed. M.L. Minges, ASM International,
Materials Park, OH, 802, (1989).

. S. Han and K.K. Wang, J Electron Packaging, 117, 178, (1995).
. M.T. Goosey, Plastics for Electronics, Elsevier, New York, NY (1985).

. R.R. Tummala and EJ. Rymaszewski, editors. Microelectronics Packaging
Handbook, Van Nostrand Reinhold: New York, NY, (1989).

10. Y. Nakamura, M. Yarnaguchi, A. Tanaka, M. Okubo, J Appl Polym Sci, 49, 331,

(1993).

11. S. Kim, IEEE T Compon Hybr, 14, 4, 809, (1991).

12. E. Suhir and T.M. Sullivan, IEEE Trans. CHMT, 911, (1992).

13. L.T. Nguyen, IEEE 43” Electronic Components and Technology Conference, 375,

Orlando, FL, June 1-4, (1993).

14. LL. Blyler, Jr., H.E. Bair, P. Hubbauer, S. Matsuoka, D.S. Pearson, G.W. Poelzing,

R.C. Progelhof, and W.G. Thierfelder, Polym. Eng. Sci, 26, 1399 (1986).

15. P. Swanson, Assembly Automation, 17, 23, (1997).

16. D.L. Robinson, M. Papageorge, and C. Naito, Proceedings of the International

Electronic Packaging Conference, 2, 924 (1993).

26

Chapter 3

RESEARCH OBJECTIVES

As Chapter 2 demonstrated, photopolymerizations offer tremendous potential for
the development of improved, high-speed, low-cost encapsulation methods for
microelectronic devices. This research project will address several issues that need to be
resolved in order to establish photopolymerizable liquid encapsulants (PLEs) as a feasible
alternative to traditional transfer molding compounds. Therefore, a general goal of this
project is to make an impact in the microelectronics encapsulation arena by
circumventing many of the problems and limitations associated with the current transfer
molding technology. Meeting the following general goals will allow us to demonstrate
the feasibility of PLEs for microelectronic encapsulants: i) Develop a highly ﬁlled
photopolymerizable liquid encapsulant that has the material properties required by the
microelectronics industry for traditional transfer molding compounds (i.e. ﬂexural
modulus and strength, coefﬁcient of thermal expansion, glass transition temperature,
thermal conductivity, adhesion, etc.); ii) Develop a photopolymerizable formulation with
a much lower viscosity than other transfer molding compounds in order to minimize
problems such as wire sweep; iii) Demonstrate that a highly ﬁlled photopolymerizable
liquid encapsulant can exhibit similar, if not faster, in-mold cure times than conventional
thermally cured transfer molding compounds.

To meet these general goals, the speciﬁc objectives of this research project are as

follows:

27

1. Identify appropriate materials (resins, ﬁllers, initiators, etc.) for a photopolymerizable
liquid encapsulant;

2. Characterize the viscosity of highly ﬁlled suspensions and determine the effect of
blending two different ﬁller particle size distributions on the viscosity;

3. Determine the effect of silane coupling agents on the viscosity of a highly ﬁlled
suspension;

4. Determine the effect of UV illumination time, postcure time, ﬁller loading, and
initiation scheme on the following material properties: ﬂexural modulus and strength,
coefﬁcient of thermal expansion, glass transition temperature, and thermal stress
parameter.

5. Determine the thermal conductivity of photopolymerizable liquid encapsulants.

6. Characterize the degree of adhesion between the photopolymerizable liquid
encapsulant and leadfrarne materials.

By meeting these research objectives, this research project will establish
photopolymerizable liquid encapsulants as a feasible and viable alternative for

microelectronic encapsulation.

28

Chapter 4

SELECTION OF MATERIALS FOR THE
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANT

4.1 RESIN MATRIX

Preliminary studies focused on determining the appropriate resin matrix for the
photopolymerizable liquid encapsulant. The resins used for this evaluation included
epoxy vinyl ester resins from the DERAKANE product line available from Dow
Chemical (DERAKANE 411-C-50, 441-400, 470-45, 510A, and 8084). Figure 4.1
shows the general structure of the DERAKANE resins. The primary properties that were
used to assess the resins were cure time, coefﬁcient of thermal expansion, and viscosity.

Cure time studies were performed on the full complement of DERAKANE resins
ﬁlled with powdered E-glass ﬁbers and the photoinitiator, benzoin ethyl ether (BEE).
These samples were photopolymerized in polyethylene vials using near-UV light with an
intensity of 200 mW/cm2 and a BBB concentration of 0.2 wt.% (based on resin weight).
For each resin, the time required for complete cure for 6.5 mm thick samples was
determined as a function of ﬁber loading, as shown in Figure 4.2. Clearly, shorter cure
times are desirable in order to meet the high-speed requirements of mass production.
Current molding compounds cure in 2 to 3 minutes at a temperature of 175°C, and
require a postcure of 4 to 8 hours at this elevated temperature. In addition, the typical
thickness of an encapsulant is one to three millimeters thick. In our photopolymerization
systems, the cure is performed with no external heating, and the cure time decreases

markedly as the sample thickness is reduced. Therefore, a cure time of a few minutes for

29

a 6.5 mm thick sample would easily meet the cure rate requirements for microelectronic
encapsulants. Figure 4.2 clearly illustrates that DERAKANE resins, 411, 441, and 470-
45, cure in the shortest amount of time at high ﬁber loadings and, therefore, the 510A and
8084 resins were eliminated from further study.

After the elimination of the 510A and 8084 resins, the coefﬁcients of thermal
expansion of the remaining resins were measured. The coefﬁcient of thermal expansion
(CTE) is a very important parameter in microelectronic encapsulation because a
signiﬁcant CTE mismatch between the encapsulant, leadframe, and the silicon chip can
lead to cracking of the chip or encapsulant. Since the CTE for the silicon chip is between
3 and 4 um/m°C, and for the leadfrarne between 12-16 tun/m°C, encapsulants with low
CTE values between 16 and 25 um/m°C are desired. Figure 4.3 contains experimental
results for the coefficient of thermal expansion of photopolymerized samples with 0 and
50 wt.% powdered E-glass ﬁbers for the three different resins. These samples were cured
for 3.5 minutes using 0.2 wt.% BEE and an intensity of 200 mW/cmz. As Figure 4.3
illustrates, only samples prepared with DERAKANE 470-45 and 50 wt.% ﬁber are within
the recommended 16 to 25 um/m°C range. This result, combined with the fact that
DERAKANE 470-45 had the lowest viscosity of the three remaining resins (viscosity at
25°C is 45.6 cP, as opposed to 100 cP for DERAKANE 411 and 400 cP for DERAKANE
441-400) led to the selection of DERAKANE 470-45 as the base resin for the
photopolymerizable liquid encapsulant.

In addition, acrylates were also evaluated to see their potential as a base resin
(since acrylates are predominantly used in photopolymerization systems). In particular,

1,3 butylene glycol diacrylate was selected because of its low viscosity (9 cP at 25°C)

30

and relatively high glass transition temperature (101°C). However, this resin performed
quite poorly when photocured; all samples created from this resin displayed numerous
cracks and were unusually brittle. Therefore, this resin was discontinued from further

study.

4.2. FILLER

The primary roles of the ﬁller are to reduce the coefﬁcient of thermal expansion,
improve the mechanical properties, and increase the thermal conductivity of the cured
encapsulant. Also, the addition of ﬁllers reduces encapsulant shrinkage by replacing
much of the resin with an inert compound that does not participate in the crosslinking
process. Fused silica is the most commonly used ﬁller in transfer molding compounds
since it possesses the optimum combination of properties. In particular, it has a low
coefficient of thermal expansion (0.5-0.6 x 10'6 °C") and it has a low enough
abrasiveness to provide a reasonable molding tool lifetime, as opposed to aluminum
nitride.I In addition, fused silica is transparent to UV light, and is therefore appropriate
for UV curing.2 Based upon these considerations, the ﬁller selected for the PLE was
crushed (angular), untreated fused silica from the commercial MIN-SIL product line
offered by MINCO, Inc. Three different fused silica products, each with its own
continuous particle size distribution, were used for the PLE. These ﬁllers were chosen
speciﬁcally for their particle sizes and associated distributions. The broadest distribution
of particles (which also contained the largest particles), MIN -SIL 40 (henceforth referred
to as particle size distribution A, or PSD-A) had a volume average particle size = 32.42

pm, and a median particle size = 22.11 pm. The second distribution of particles, MIN-

31

SIL 20, (henceforth referred to as PSD-B) had a volume average particle size = 19.37 pm
and a median particle size = 13.17 pm. The narrowest distribution of particles (which
contained the largest fraction of small particles), MIN-SIL 550 (henceforth referred to as
PSD-C) had a volume average particle size = 10.37 pm and a median particle size = 6.03
pm. For microelectronic encapsulants, fused silica particles larger than ~100 11m should
be avoided to prevent particles from becoming trapped in the gates (a gate admits the
molding compound into the cavity of the device to be molded and can be as narrow as
150 microns but are typically 200-250 microns) or caught between the leads of high
pincount packages (the interlead spacing can be as little as 100 microns).1 None of the
fused silica products that we have chosen to use has fused silica particles greater than 150
microns. Ideally, we would also want to use spherical fused silica particles, because
spherical fused silica is much more effective in decreasing the viscosity than crushed
fused silica.3 In addition, spherical silica leads to increased ﬁller loadings. However, a
spherical fused silica supplier in the United States was not found before these studies

were performed.

4.3. PHOTOINITIATOR
Since most monomers, oligomers, or prepolymers do not produce initiating
species with a sufﬁcient quantum yield upon light exposure, it is necessary to use low
molecular weight organic molecules called photoinitiators to start the polymerization.4 A
photoinitiator is the chemical species which produces active centers (free radicals or
cations) upon absorption of photons of the appropriate wavelength. Generally, free

radical photoinitiators fall into two categories: cleavable and non-cleavable

32

photoinitiators. Cleavable photoinitiators directly produce free radicals upon absorption
of photons and are further classiﬁed according to the mechanism by which they undergo
photolysis as tit-cleavable, B-cleavable, and y-cleavable. Non-cleavable photoinitiators
require a co-initiator such as an amine (electron transfer agent) or an alcohol (a hydrogen
donor) in order to produce free radicals.

Most of the commonly used free radical photoinitiators are tit-cleavable initiators
and these include a number of compounds, such as benzoyl phosphine oxides, benzoin
ethers, dialkoxyacetophenones, hydroxy alkyl ketones, amino ketones, morpholino

5 These (rt-cleavable free radical photoinitiators

ketones, and benzoyl oxime esters.
undergo a fairly simple sequence of photochemical transitions. Speciﬁcally, they are
excited to the singlet state, undergo efficient intersystem crossing to the triplet state, and
then undergo photolysis to form free radicals. The free radicals then initiate
polymerization in the presence of monomer, oligomer, or prepolymer.

The proper selection of a photoinitiator for UV-curing thick and ﬁlled systems
depends upon a number of factors, such as the emission spectrum of the light source,
absorption spectrum of the photoinitiator, absorbance by the monomer and other
additives, absorbance and scattering properties of the ﬁller, and the photobleaching
exhibited by the photoinitiator. Speciﬁcally, there are a number of criteria that must be
satisﬁed. First, the photoinitiator must clearly have signiﬁcant absorbance in the region
where the light source emits. Second, the photoinitiator must exhibit good absorbance in
a spectral region different from where the other constituents of the formulation (resin,

ﬁller, additives, etc.) absorb. Finally, the photoinitiator must undergo signiﬁcant

photobleaching upon illumination. Photobleaching occurs when the products of

33

photolysis do not absorb in the same region as the original initiator molecule. The net
result of photobleaching is an increase in transmittance of the system as a function of
illumination time, which, in turn, allows for deeper penetration of the incident light.
Based on these considerations, the photoinitiator selected for the
photopolymerizable liquid encapsulant was bis(2,4,6-trimethylbenzoyl)phenyl phosphine
oxide (IRGACURE 819, Ciba), whose structure is shown in Figure 4.4. This
photoinitiator is in the family of bisacylphosphine oxide photoinitiators and is denoted as
BAPO. These photoinitiators are very versatile and have found widespread use in
pigmented and ﬁlled systems for the following reasons. First, bisacylphosphine oxides
have absorption spectra with maxima between 350 and 380 nm and extending well into
the visible, as shown by the absorption spectrum of bis(2,4,6-trimethylbenzoyl)phenyl
phosphine oxide in Figure 4.5. As a result, these photoinitiators are especially suited for
use with light in the near-UV-vis part of the spectrum, where many pigments and ﬁllers
are sufﬁciently transparent.6 Second, bisacylphosphine oxide photoinitiators exhibit
signiﬁcant photobleaching since photolysis results in cleavage of the carbon phosphorus
bond, thereby destroying the chromophore which absorbs in the near UV-vis region (350
to 410 nm).6 Thus, the optical density of the system containing this photoinitiator
decreases in the near UV-vis range during the curing process, and light can effectively

penetrate deep into the sample (which is imperative to cure thick samples).
4.4. THERMAL INITIATOR

Curing techniques using both light and heat to initiate polymerization have been

well documented.6 When a radical polymerization is activated by both a photoinitiator

34

and a thermal initiator, the process is referred to as “dual cure” or “dual initiating”. In
this initiating method, the photoinitiator is used to give a fast cure of the surface, which is
easily accessible for the incident light. The heat evolved from the exothermic
photopolymerization then elevates the temperature, resulting in the production of
additional active centers by the thermal initiator. The primary advantage of this dual
initiating method is that the degree of cure is greatly enhanced, since the thermal initiator
is used to create active sites in areas where light may not be able to sufﬁciently penetrate
the system. As a result, this method is especially appropriate for increasing the degree of
cure in systems where the sample is opaque, such as in a composite with high ﬁller
loading.

In order to effectively utilize a thermal initiator in a photopolymerizable
formulation, the thermal initiator should not absorb the initiating light and should be
susceptible to rapid dissociation at moderate temperatures (80 to 100°C). Based on these
considerations, benzoyl peroxide (Aldrich) was selected as the thermal initiator for the
photopolymerizable liquid encapsulant (peroxides are well known as thermal initiators

for radical polymerization). The chemical structure of benzoyl peroxide is shown in

Figure 4.6.

4.5. SILANE COUPLING AGENT
Silane coupling agents are organofunctional compounds that act as molecular
bridges at the interface between the polymer matrix and the ﬁller particles of the
composite. They are linear molecules which have a reactive group at one end capable of

being incorporated into the matrix of a curing polymer and a silicon-based group at the

35

other end which bonds to inorganic ﬁllers. Silane coupling agents are included in the
formulation because they offer several beneﬁts, including improved processability,
reduced particle settling, better adhesion to substrates such as the leadframe, and
improved mechanical and electrical properties.

The silane coupling agent chosen for the photopolymerizable liquid encapsulant
was 3-methacryloxypropyltrimethoxysilane (Z-6030, Dow Corning), whose structure is
shown in Figure 4.7. This particular silane was chosen because of its chemical
compatibility with the epoxy vinyl ester resin. It is particularly effective in improving the
properties of free radical-cured plastics, such as vinyl ester resins, and it has a very low

viscosity (2.5 cP at 25°C).

36

 

 

CH3—C—CH3

 

 

 

H2

I

.
|
“I
l
l
|
I

—OH

CH2
0
o=c
cl—CH3

CH2

Figure 4.1. General chemical structure of DERAKANE epoxy vinyl ester resins.

37

6.00 --

 

5.00 -
X

T
Q
Q

.5
o
o
l
j
u:
ﬂ
9
.>
~~

  
 
 

Time (min.)

 

 

 

 

2.00 a
411
1.00 --
0.00 i i i i i i
0% 10% 20% 30% 40% 50% 60%

Wt.% Fiber

Figure 4.2. Cure time as a function of ﬁber loading for various DERAKANE epoxy vinyl
ester resins.

38

 

 

I No fiber
55 ‘ 1:150 wt.% fiber

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ere (pm/TnoC)
r»
O

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

470-45 411
DERAKANE resin

Figure 4.3. Effect of ﬁller loading on the coefﬁcient of thermal expansion for three
different DERAKANE resins.

39

Figure 4.4. Chemical structure of bis(2,4,6-trimethy1benzoyl)phenyl phosphine oxide.

 

Absorbance
O
on

 

 

..
:l
o 1 1 1 1 1

I T I U I I I T V I T T T V T j' ' T T ' T I V I I I

320 340 360 380 400 420 440
Wavelength (nm)

 

 

Figure 4.5. Absorbance spectrum of bis(2,4,6-trimethylbenzoyl)phenyl phosphine oxide.
The concentration is 5.1 x 10'4 g/ml in acetone.

40

Figure 4.6. Chemical structure of benzoyl peroxide.

ii fH3
(CH3O)3SiCHZCHZCHZOC—C=CH2

Figure 4.7. Chemical structure of 3-methacryonypropyltrimethoxysilane.

41

4.6. REFERENCES

. L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, (1990).

. H. Bayer and B. Lehner, ACS Symp Ser, 417, 412, (1990).

. M.G. Pecht, Plastic-Encapsulated Microelectronics, John Wiley & Sons, Inc., New
York, NY, (1995).

. J .P. Fouassier, Photoinitiation, Photopolymerization, and Photocuring, Hanser
Publishers, New York, NY, (1995).

. J .P. Fouassier and J.F. Rabek, editors, Radiation Curing in Polymer Science and
Technology, Volume 2, Elsevier Applied Science, New York, NY, (1993).

. K. Dietliker, Radiation Curing in Polymer Science and Technology, Volume 2,

editors, J .P. Fouassier and J .F. Rabek, Elsevier Applied Science, New York, NY,
155-237, (1993).

42

Chapter 5

VISCOSITY CHARACTERIZATION OF HIGHLY FILLED
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS

5.1. INTRODUCTION

A microelectronic device consists of a silicon chip that is mounted on a metal
leadframe, and gold or aluminum wires that electrically connect the chip circuits to the
leadfrarne. Microelectronic devices are typically encapsulated in a therrnoset molding
compound, which serves to protect the silicon chip from both physical and chemical
damage. The three most commonly used techniques for encapsulation are transfer
molding, potting, and conformal coating, with transfer molding being the predominant
method to encapsulate integrated circuits.l In this process, the thermoset molding
compound (typically a solid epoxy preform) is dielectrically preheated and then placed
into the pot of the molding tool. Transfer molding is accomplished by lowering a plunger
that forces the molding compound into the runners and gates of the mold. The molding
compound then ﬂows over the chips, wirebonds, and leadﬁames, encapsulating the
microelectronic device. Although transfer molding is the predominant means to
encapsulate integrated circuits, it does suffer from a few problems. Three problems
commonly encountered during transfer molding are: i) incomplete mold ﬁlling; ii)
leadfrarne movement during cavity ﬁlling; and iii) wire sweep, which is displacement or
deformation of the wires that provide electrical interconnection between the integrated
circuit and the leadfrarne. Each of these molding problems is directly related to the
processing conditions and the rheological characteristics of the molding compound used

for encapsulation?’3

43

Photopolymerizable liquid encapsulants (PLEs) for microelectronic devices may
offer important advantages over traditional transfer molding compounds (which are solid
at room temperature), including alleviation of the three problems mentioned above.
These liquid encapsulants are formulated using low-viscosity epoxy novolac-based vinyl
ester resins that cure rapidly upon exposure to UV light. The use of light, rather than
heat, to initiate the polymerization allows the reaction to be triggered only after mold
ﬁlling is complete, and the resin remains in the low-viscosity, uncured state throughout
the mold ﬁlling process. Typically, liquid encapsulants are used when the part
conﬁguration cannot be easily transfer molded, or when the capital costs of transfer
molding are too high. In the past, liquid systems have been limited to low-end consumer
electronic products where cost and size were the primary issues and environmental
reliability was of secondary importance. However, recent research on liquid encapsulants
has demonstrated that liquid encapsulants can meet or exceed the performance of
traditional transfer molding compounds.‘1 In addition, liquid encapsulants offer added
beneﬁts such as allowing package design changes without costly retooling, reduced cycle
time, and low initial capital expense requirements.

Tire microelectronic encapsulation process requires a low viscosity of the molding
compound so it can ﬂow into the mold with minimum deformation of the leadfrarne and
wirebonds.s Monodisperse suspensions are not favored for this application (or any other
highly ﬁlled suspension for that matter) because of the resulting high viscosity of the
system.‘ It is generally known that blending particles of various sizes in the proper
proportions can signiﬁcantly reduce the viscosity of highly ﬁlled systems at a ﬁxed

loading level.7 This is clearly demonstrated by the work of authors such as Farriss,

44

Hoffman9, Rodriguez et al. '0, and Zaman and Moudgil6. In these polydisperse systems,
the spaces between larger particles may be occupied by smaller particles, and the
combination of the small particles and the suspending medium can behave essentially as
a ﬂuid to the larger particles.8 In addition, bimodal and trimodal particle size
distributions are used in highly ﬁlled systems to achieve maximum ﬁller loadings
(thereby improving composite material properties, i.e., decreasing the coefﬁcient of
thermal expansion and increasing the thermal conductivity for the case of thennoset
encapsulants). For example, the maximum packing fraction of a spherical monomodal
particle distribution is 64 vol.% due to the inherent spaces between particles. In contrast,
a bimodal distribution allows a packing fraction of 83 vol.% to be theoretically achieved,
and a properly designed trimodal distribution allows a packing fraction of 90 vol.% to be
theoretically achieved.ll

Silane coupling agents are organofunctional compounds that act as molecular
bridges at the interface between the polymer matrix and the ﬁller particles of the
composite. They are linear molecules which have a reactive group at one end capable of
being incorporated into the matrix of a curing polymer and a silicon-based group at the
other end which bonds to inorganic ﬁllers. Silane coupling agents are included in the
formulation because they offer several beneﬁts, including improved processability,
reduced particle settling, better adhesion to substrates such as the leadframe, and

improved mechanical and electrical properties. In addition, the coupling agents can

3

either increase or decrease the viscosity of the formulation. For this reason, it is

important to characterize the effect of silane coupling agents on the viscosity.

45

In our research, we are investigating novel, highly-ﬁlled PLEs formulated using
epoxy novolac-based vinyl esters that cure in a relatively short amount of time to yield a
composite exhibiting appropriate values for the ﬂexural modulus and strength, coefﬁcient
of thermal expansion, glass transition temperature, and chemical resistance.”‘13 We have
found that these PLEs exhibit appropriate cure characteristics and material properties, and
are therefore quite promising for microelectronic encapsulation. Because the viscosity of
the encapsulation material is such an important parameter, the viscosity of the
photopolymerizable liquid encapsulants has been characterized in this chapter. We have
identiﬁed a blend of two particle size distributions of fused silica that provides a reduced
viscosity at high ﬁller loadings, and we have characterized the effect of silane coupling

agents on the photopolymerizable liquid encapsulants.

5.2. BACKGROUND ON THE VISCOSITY OF CONCENTRATED
SUSPENSIONS

The viscosity of concentrated suspensions has been the subject of extensive
research due to the widespread use of concentrated suspensions in a host of applications,
including paints, foods, cements, ﬁlled polymers, cosmetics, etc. For efﬁcient processing
and handling of these highly ﬁlled solid-liquid suspensions, it is important to understand
how their rheology depends upon a variety of factors. For this reason, several authors
have developed models to predict the effect of the solids loading, particle shape, particle
size distribution, and the viscosity of the suspending medium on the viscosity of the
suspension (for reviews, see Zaman and Moudgi16, Barnes”, Han”, and Probstein et
al. 16). Most descriptions of concentrated suspensions are extensions of Einstein’s early

17,18

work on dilute suspensions of rigid, uniform-sized spheres. Einstein’s equation

46

relates the viscosity of the suspension, 77, to the viscosity of the suspending medium, 770,

and the volume fraction of the spheres, 45, as shown in Equation 5.1.

n = n.(1+2.5¢) (5.1)

However, this equation is not applicable to concentrated suspensions because it neglects
particle-particle interactions, and does not account for particle size.14
A number of authors extended Einstein’s equation to suspensions of higher

'9'22 using the maximum packing fraction, ¢,,., to account for the particle-

concentrations
particle interactions.'4 For example, the Krieger-Dougherty equation22 is often used to

describe highly ﬁlled suspensions and is given in Equation 5.2:

n=n.(1-f-)*""- (5.2)

where [77] is deﬁned as the intrinsic viscosity. The maximum packing fraction is the
volume fraction of particles that makes ﬂow impossible, i.e. the viscosity goes to inﬁnity.
Because of the sensitivity of the maximum packing fraction to the particle size
distribution and particle shape, the Krieger-Dougherty equation effectively takes these
variables into account when predicting the viscosity of a concentrated suspension. The
Krieger-Dougherty equation can be extended to a binary mixture of particles, as shown in

Equation 5.3.”

A +2.14... ¢2 4cm...
77 77.,( T...) ( _¢..2) ( )

In Equation 5.3, the subscript 1 denotes particles of one size, while the subscript 2
corresponds to a second particle size. Each of these subscripts can refer to a particle size

distribution if the maximum packing fraction is calculated accordingly.

47

For the purposes of modeling the experimental viscosity data, Equation 5.3 was
used. The maximum packing fractions ¢,,./ and 43:12 were calculated directly from our
particle size distributions using the equations developed by Ouchiyama and Tanaka23 and

reported in the paper by Poslinski et al.24 (Equations 5.4 through 5.7 shown below):

D3
2 ,.,

 

 

 

_ 1 (5.4)
2(1). ~ 0.1V. 71(1). +0.53 —(D. ~ D.)’]f.
where
2
2(1),.- 0.5211 -—8—i—1f.
_ i a _ (D, +Da)
ﬂ—1+13(8¢M DD. ZIDf-(DnDJ’lf. (5.5)
and
D, = inf, (5.6)
where
_V._
D,’
f. = L (5.7)
of

In these equations, 41:, is the maximum packing fraction of spheres of uniform size and a
value of 0.619 was used, as suggested by the authors. In addition, D, is the diameter of
the i-th component, f, is the number fraction of the i-th component, v, is the volume
fraction of the i-th component, D, is the average diameter of the different sizes of the
particulates, and (D, ~ D0) is deﬁned as:

(D, ~ Da)=0 for D, 3D,,
and

(D, ~D,,)=D, —D,, for D, > Do,

48

Once the maximum packing fractions were determined using these equations, the
viscosity data was ﬁt to Equation 5.3 by nonlinear regression using the two intrinsic

viscosities as the ﬁtting parameters.

5.3. EXPERIMENTAL

5.3.1. Materials

The photopolymerizable liquid encapsulant consisted of a base resin,
photoinitiator, silane coupling agent, and ﬁller. The base resin was an epoxy novolac-
based vinyl ester resin commercially available from Dow Chemical (DERAKANE 470-
45). This resin was chosen primarily because of its low viscosity (770 = 0.0456 Pa-s or
45.6 CF at 30°C). In addition, it exhibits an appropriate polymerization rate, and good
material properties upon cure. The photoinitiator was bis(2,4,6-trimethylbenzoyl)phenyl
phosphine oxide (IRGACURE 819, Ciba). This photoinitiator, which we will denote as
BAPO, has been shown to be especially appropriate for curing relatively thick polymers
and composites due to its efficient photobleaching.” 26 The silane coupling agent was 3-
methacryloxypropyltrimethoxysilane, (Z-6030, Dow Coming) and the ﬁller was fused
silica (MINCO, Inc.).

A The primary roles of the ﬁller are to reduce the coefﬁcient of thermal expansion
and enhance the thermal conductivity of the cured encapsulant. Fused silica is the most
commonly used ﬁller in transfer molding compounds since it possesses the optimum
combination of properties. In particular, it has a low coefﬁcient of thermal expansion
(05-06 x 10'6 °C") and it has low enough abrasiveness to provide a reasonable molding
tool lifetime, as opposed to aluminum nitride.3 Based upon these considerations, the

ﬁllers selected for this study were crushed (angular), untreated fused silica from the

49

commercial MIN-SIL product line offered by MINCO, Inc. Three different fused silica
products, each with its own continuous particle size distribution, were used in this study.
The broadest distribution of particles (which also contained the largest particles), MIN-
SIL 40 (henceforth referred to as particle size distribution A, or PSD-A) had a volume
average particle size = 32.42 pm, and a median particle size = 22.11 pm. The second
distribution of particles, MIN-SIL 20, (henceforth referred to as PSD-B) had a volume
average particle size = 19.37 pm and a median particle size = 13.17 pm. The narrowest
distribution of particles (which contained the largest fraction of small particles), MIN-SIL
550 (henceforth referred to as PSD-C) had a volume average particle size = 10.37 um and
a median particle size = 6.03 um. For microelectronic encapsulants, fused silica particles
larger than ~100 um should be avoided to prevent particles from becoming trapped in the
gates (which can be as narrow as 150 microns and are typically 200-250 microns) or
caught between the leads of high pincount packages (the interlead spacing can be as little
as 100 microns).3
5.3.2. Sample Preparation

To determine the effect of blending two different fused silica particle size
distributions on the viscosity of the PLE, the DERAKANE 470-45 resin was mixed with
70.0 wt.% fused silica, 0.20 wt.% BAPO photoinitiator (based on resin weight), and 1.43
wt.% of the silane coupling agent (based on ﬁller weight). The epoxy novolac-based
vinyl ester was blended with 0.20 wt.% BAPO photoinitiator (based on resin weight)
until the photoinitiator completely dissolved in the resin. Next, the silane coupling agent
was added dropwise to the desired 1.43 wt.% (based on the ﬁller weight) and the resin

mixture was again stirred on a magnetic stir plate. Fused silica (with the appropriate

50

particle size distribution) in the amount of 70.0 wt.% was added to the system, which was
thoroughly mixed by using a vortex mixer until the silica was completely dispersed and a
homogeneous mixture was achieved. Lastly, the PLE was degassed in a vacuum oven.

In this initial study, all samples contained 70.0 wt.% (0.526 volume fraction) of
fused silica. The fused silica particle size distribution was varied by preparing binary
mixtures of the individual distributions, PSD-A, PSD-B, and PSD-C, as shown in Table
5.1.

A second set of studies was performed to investigate the effect of the silane
coupling agent on PLE viscosity. To prepare these samples, the epoxy novolac-based
vinyl ester resin was mixed with 70.0 wt.% fused silica (using the optimal particle size
distribution determined from the ﬁrst set of experiments), 0.2 wt.% BAPO photoinitiator
(based on resin weight), and 0 to 1.43 wt.% of the silane coupling agent (based on ﬁller
weight). In the ﬁnal set of studies, the effect of higher ﬁller loadings on the PLE
viscosity was investigated. Speciﬁcally, the optimal blend determined from the ﬁrst set
of studies was investigated in systems containing fused silica loadings of 72.0 and 74.0
wt.%. It is important to investigate this apparently small increase in the ﬁller loading
because the encapsulant containing 74.0 wt.% ﬁller exhibits a lower coefﬁcient of
thermal expansion and an enhanced thermal conductivity compared to an encapsulant
containing 70.0 wt.% ﬁller.

Viscosities were measured using a Brookﬁeld DV I+ Viscometer equipped with
the Small Sample Adapter (SC4-34 spindle and 13R chamber). For this particular spindle
and chamber setup, the required sample volume was 10 ml. The temperature of the

sample was kept at 30°C using a therrnostated circulating water bath connected to the

51

water jacket on the sample chamber. Viscosities were recorded 30 minutes after the
spindle started turning to ensure that the system had equilibrated at the desired
temperature. The spindle speed was set to 2 RPM (for this particular spindle and
chamber geometry, the shear rate is 0.56 sec'l at 2 RPM). Each reported viscosity is the

average of three independent measurements.

5.4. RESULTS AND DISCUSSION

5.4.1. Effect of Fused Silica Particle Size Distribution

Figure 5.1 illustrates the effect of the fused silica particle size distribution on the
viscosity of the resulting PLE. All samples contained a total volume fraction of fused
silica of 0.526 (70.0 wt.%), 1.43 wt.% silane coupling agent (based upon ﬁller weight),
0.2 wt.% BAPO (based upon resin weight), and the balance epoxy vinyl ester resin. Each
bar in the bar chart represents the average of three measurements while the error bars
indicate one standard deviation above and below the mean. Figure 5.1 contains data for
samples containing between 0.263 and 0.526 volume fraction of PSD-A, and the
remainder (up to a total ﬁller loading of 0.526 volume fraction or 70.0 wt.%) either PSD-
B or PSD-C. Figure 5.1 clearly illustrates that, for a given ﬁller loading (0.526 volume
fraction total), the system viscosity can vary signiﬁcantly for different particle size
distributions. For example, the system containing a 50:50 mixture of PSD-A and PSD-B
(0.263 volume fraction of each) has a viscosity that is more than four times greater than
the system containing a 50:50 mixture of PSD-A and PSD-C. Moreover, Figure 5.1
shows that the viscosity of the encapsulant formulation actually increases as the amount
of PSD-B is increased, but decreases considerably as the amount of PSD-C is increased.

Therefore, it is readily apparent that the PSD-A/PSD-C combination provides much

52

better viscosity reduction than the PSD-A/PSD-B combination. The system containing
0.526 volume fraction (70.0 wt.%) of a 50:50 mixture of PSD-A and PSD-C exhibits a
relative viscosity of ~100 (a viscosity less than 5.0 Paos). This viscosity is more than
three times lower than that exhibited by a system containing 0.526 volume fraction of
PSD-A (with no second ﬁller component) and, therefore, the blending of these two
particle size distributions in order to reduce the viscosity at a given ﬁller loading was
successful. For microelectronic encapsulants, it is undesirable to reduce the viscosity
further than that exhibited by the 50:50 mixture of PSD-A and PSD-C since molding
problems such as resin bleed and ﬂash may occur.

Figure 5.2 contains data for samples containing between 0 and 0.263 volume
fraction of PSD-C, and the remainder (up to a total ﬁller loading of 0.526 volume fraction
or 70.0 wt.%) either PSD-A or PSD-B. All samples contained a total volume ﬁaction of
fused silica of 0.526 (70.0 wt.%), 1.43 wt.% silane coupling agent (based upon ﬁller
weight), 0.2 wt.% BAPO (based upon resin weight), and the balance epoxy vinyl ester
resin. Each value represents the average of three measurements while the error bars
indicate one standard deviation above and below the mean. The viscosities of the
samples containing 0.526 or 0.474 volume fraction PSD-B were out of the range of the
instrument (implying that the viscosity was above 30.0 Paos) for the spindle and shear
rate under investigation. Figure 5.2 demonstrates that the PSD-C/PSD-A system has a
much lower viscosity than the formulation containing PSD-C and PSD—B. In addition the
data in Figure 5.2 clearly illustrates what is known as the “Farris Effect” - a large
reduction in the viscosity seen near a fraction of 0.6 of large particles.'4 Our data is also

in agreement with that of Zaman and Moudgil6 who found that the lowest level of

53

viscosities is obtained if the volume ratio of small and large particles lies in the range of
50/50 or 45/55.

The data in Figures 5.1 and 5.2 clearly illustrate that PLEs containing 0.526
volume fraction (70.0 wt.%) of a 50:50 mixture of PSD-A and PSD-C exhibit the lowest
viscosity; therefore, this mixture was selected for further experiments on (i) the effect of
the silane coupling agent and (ii) higher fused silica loadings. Our data is in general
agreement with the results of previous investigators who have observed a decrease in the
viscosity of concentrated suspensions as the ratio of the large particles to the small

d.8’ 9' 27’ 28 Based upon the volume average particle sizes, the

particles was increase
particle size ratio of PSD-A to PSD-C is 3.13, the particle size ratio of PSD-B to PSD-C
is 1.87, and the particle size ratio of PSD-A to PSD-B is 1.67. From Figures 5.1 and 5.2
it can be seen that blends of PSD-A and PSD-C have the lowest viscosity, blends of PSD-
B and PSD-C have a higher viscosity, and blends of PSD-A and PSD-B have the highest
viscosity.

The curves drawn in Figure 5.2 correspond to values obtained by ﬁtting the
experimental data to the Krieger-Dougherty equation (Equation 5.3). The nonlinear
regression analysis was performed using the commercial software package Origin 5.0
(Microcal Software, Inc.) after the maximum packing fractions of PSD-A, PSD-B, and
PSD-C were determined using Equations 5.4-5.7. For each calculation, the two intrinsic
viscosities were used as the ﬁtting parameters. The calculated maximum packing
fractions were 0.836 for PSD-A, 0.802 for PSD-B, and 0.754 for PSD-C. As expected,

the maximrun packing fraction increases as the breadth of the particle size distribution is

increased. For the PSD-A/PSD-C system, the best ﬁt intrinsic viscosities were 7.2 for

54

PSD-A, and 8.4 for PSD-C. For the PSD-B/PSD-C system, the best ﬁt intrinsic
viscosities were 8.9 for PSD-B, and 7.0 for PSD-C. These values for the intrinsic
viscosity are quite reasonable. Figure 5.2 illustrates that the experimental data is well
correlated by the Krieger-Dougherty equation.
5.4.2. Effect of Silane Coupling Agent

A series of experiments were performed to investigate the effect of the silane
coupling agent on the viscosity of the highly ﬁlled liquid encapsulants. The experimental
results are shown in Figure 5.3. All samples contained 0.526 volume fraction (70.0
wt.%) fused silica ﬁller (50% PSD-A, 50 % PSD-C) and 0.2 wt.% BAPO (based on resin
weight) and the balance epoxy vinyl ester resin. Each bar in this bar chart represents the
average value of three viscosity measurements and the error bars indicate one standard
deviation above and below the mean. The data show that as the amount of the silane
coupling agent was increased, the viscosity of the system decreased, with the sample
containing 1.43 wt.% silane exhibiting the lowest viscosity. As was mentioned earlier in
the introduction, it has been reported that coupling agents can either increase or decrease
the viscosity of the formulation.3 The data in Figure 5.3 illustrate that for our system, the
coupling agent does not increase the viscosity of the PLE formulations, but decreases it
slightly. Silane concentrations above 1.43 wt.% may decrease the viscosity further, but
they were not evaluated because they lead to an undesirable increase in the coefﬁcient of
thermal expansion of the cured encapsulant.29
5.4.3. Effect of Higher Filler Loadings

Further experiments were performed to investigate the effect of higher ﬁller

loadings on the viscosities of the photopolymerizable liquid encapsulants. Speciﬁcally,

55

fused silica loadings of 72.0 and 74.0 wt.% were studied. It is important to investigate
this apparently small increase in the ﬁller loading because the encapsulant containing
74.0 wt.% ﬁller exhibits a lower coefﬁcient of thermal expansion and an enhanced
thermal conductivity compared to an encapsulant containing 70.0 wt.% ﬁller.
Unfortunately, as the ﬁller loading is increased from 70.0 wt.% to 74.0 wt.%, the increase
in viscosity can be quite dramatic. For these studies, a 50:50 mixture of PSD-A and
PSD-C was used, and the PLEs contained 1.0 wt.% of the silane coupling agent (based
upon ﬁller weight). Tire experimental results are shown in Table 5.2. As expected, the
viscosity was found to increase as the ﬁller loading was increased from 70.0 to 72.0
wt.%, and although the viscosity increase appears dramatic, the PLE containing 72.0
wt.% silica was still quite easy to handle and process, and could still be appropriate for
encapsulating microelectronic devices. The viscosity of the PLE containing 74.0 wt.%
silica was too high to be measured with the speciﬁed spindle and shear rate (implying

that the viscosity was higher than 30.0 Pa-s).

5.5. CONCLUSIONS

In this chapter, we have characterized the viscosity of highly ﬁlled
photopolymerizable liquid encapsulants to investigate the blending of two different
particle size distributions to tailor the viscosity for a given ﬁller loading. For
encapsulating microelectronic devices, it is desired to achieve a low enough viscosity to
allow the molding compound to easily ﬂow over the chip, wirebond, and leadfrarne
assembly, thereby minimizing problems associated with incomplete mold ﬁlling,
leadfrarne displacement, and wire sweep. We have characterized the viscosity of PLEs

containing 70.0, 72.0, and 74.0 wt.% silica, and found that a blend of particle size

56

distributions with a particle size ratio of 3.13 (PSD-A and PSD-C) resulted in the best
viscosity reduction. For example, in experiments performed for a constant ﬁller loading
of 0.526 volume fraction (70.0 wt.%), the particle size distribution had a marked effect on
the viscosity of the PLE. A 50:50 mixture of PSD-A and PSD-C resulted in a PLE
viscosity three times lower than that obtained using only the PSD-A ﬁller, and four times
lower than a PLE prepared using a 50:50 mixture of PSD-A and PSD-B. In addition, the
50:50 mixture of PSD-A and PSD-C resulted in a PLE viscosity 40% lower that obtained
using a 50:50 mixture of PSD-B and PSD-C. The experimental results for the effect of
the particle size distribution on the PLE viscosity were well correlated by the Krieger-
Dougherty equation to account for the blending of two particle size distributions.
Experiments performed to investigate the effect of the silane coupling agent on the
viscosity of the PLE revealed that the viscosity decreased slightly with increasing
concentration of the coupling agent.

As the fused silica loading was increased from 70.0 to 74.0 wt.%, the PLE
viscosity increased markedly. This apparently small increase in the fused silica loading
can be important for microelectronic encapsulants because it can lead to a lower
coefficient of thermal expansion and an enhanced thermal conductivity. As the amount
of fused silica was increased from 70.0 to 72.0 wt.%, the viscosity increased by a factor
of 2.3, and as the fused silica loading was increased from 70.0 to 74.0 wt.%, the viscosity
increased more than ﬁve-fold. The current epoxy molding compounds are solid at room
temperature, therefore even the PLE containing 74.0 wt.% fused silica has a much lower

viscosity at ambient temperature than the traditional molding compounds. Thus, the

57

resulting PLEs exhibit low viscosities at ambient temperature while maintaining desirable

material properties for microelectronic applications.

58

5.6. NOMENCLATURE

D, = average diameter of the different sizes of the particulates
D, = diameter of i-th component

f, = number fraction of i-th component

PLE = photopolymerizable liquid encapsulant

PSD = particle size distribution

v, = volume fraction of i-th component

,8 = parameter in Equations 4 and 5

r] = viscosity of suspension

770 = viscosity of suspending medium

I]/ 770 = relative viscosity

[7)] = intrinsic viscosity

¢= volume ﬁaction

41,; = maximum packing fraction of ﬁrst particle size distribution
¢,,.2 = maximum packing fraction of second particle size distribution

¢° = maximum packing fraction of spheres of uniform size = 0.619

59

Table 5.1. Volume fractions of individual fused silica distributions
in the binary mixtures.

 

flipsoqx ¢PSD-B ‘I’PSD-C
100% PSD-A 0.526 0 -
90% PSD-A, 10% PSD-B 0.474 0.052 -
80% PSD-A, 20% PSD-B 0.421 0.105 -
70% PSD-A, 30% PSD-B 0.368 0.158 -
60% PSD-A, 40% PSD-B 0.316 0.210 -
50% PSD-A, 50% PSD-B 0.263 0.263 -
100% PSD-A 0.526 - 0
90% PSD-A, 10% PSD—C 0.474 - 0.052
80% PSD-A, 20% PSD-C 0.421 - 0.105
70% PSD-A, 30% PSD-C 0.368 - 0.158
60% PSD-A, 40% PSD-C 0.316 - 0.210
50% PSD-A, 50% PSD-C 0.263 - 0.263
100% PSD-B - 0.526 0
90% PSD-B, 10% PSD-C - 0.474 0.052
80% PSD-B, 20% PSD-C - 0.421 0.105
70% PSD-B, 30% PSD-C - 0.368 0.158
60% PSD-B, 40% PSD-C - 0.316 0.210
50% PSD-B, 50% PSD-C - 0.263 0.263

 

Table 5.2. Viscosity of PLEs with 70.0, 72.0, and 74.0 wt.% fused silica ﬁller.

 

Wt.% Fused Silica Viscosity (Pa-s) Standard Deviation (Pa-s)
70.0 5.92 0.114
72.0 13.9 0.995
74.0 Out of range (>300) -

 

60

 

600

 

El PSD-A & PSD-C

  

 

i
l
_.|

I PSD-A 81 PSD-B

 

 

 

 

§

 

 

l

N
8
l

l

Relative Viscosity 11an
s
l

    

  

3
l

 

 

 

 

 

 

 

 

 

 

0.526 0.474 0.421 0.368 0.316 0.263
Volume Fraction of PSD-A

Figure 5.1. Relative viscosity as a function of volume fraction of PSD-A. The balance
of the ﬁller is either PSD-B or PSD-C. The viscosity of the suspending medium, 770, is
0.0456 Pa-s (45.6 cP).

61

 

500

 

 

 

 

 

o PSD-A 8 PSD-C Experimental Data
o PSD-B & PSD-C Experimental Data
Fe ‘°° 1 +Pso-A a. PSD-C Calculated Values
:- EE +PSD—B a PSD-C Calculated Values
'5 300 -
§
5
g 200 -
a
3
G
a:
100 'l a
o T T T T I
0 0.05 0.1 0.15 0.2 0.25 0.3

Volume Fraction of PSD-C

Figure 5.2. Relative viscosity as a function of volume fraction of PSD-C in the ﬁller.
The balance of the ﬁller is either PSD-A or PSD-B. The curves correspond to values
obtained from the Krieger-Dougherty equation. The viscosity of the suspending medium,

”a, is 0.0456 Pa-s (45.6 cP).

62

250

 

N
8

 

uh
I

 

Relative Viscosity "In.
'2‘

J

 

 

 

0 0.25 0.5 0.75 1 1.43
Wt.% Silane Coupling Agent (based on filler weight)

Figure 5.3. Relative viscosity as a function of weight percent of silane coupling agent.
The viscosity of the suspending medium, 770, is 0.0456 Pa-s (45.6 CF).

63

10.
11.
12.

13.

14.

15.

5.7. REFERENCES

M.T. Goosey, Plastics for Electronics, Elsevier, New York, NY (1985).

L.L. Blyler, Jr., H.E. Bait, P. Hubbauer, S. Matsuoka, D.S. Pearson, G.W. Poelzing,
R.C. Progelhof, and W.G. Thierfelder, Polym. Eng. Sci, 26, 1399 (1986).

L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, (1990).

D.L. Robinson, M. Papageorge, and C. Naito, Proceedings of the International
Electronic Packaging Conference, 2, 924 (1993).

. M.G. Pecht, Plastic-Encapsulated Microelectronics, John Wiley & Sons, Inc., New

York, NY, (1995).

A.A. Zaman and BM. Moudgil, J. Rheol, 42, 21 (1998).

D]. Lee, Journal of Paint Technology, 42, 579 (1970).

R.J. Farris, Trans. Soc. Rheol, 12, 281 (1968).

R.L. Hofﬁnan, J. Rheol, 36, 947 (1992).

BE. Rodriguez, E.W. Kaler, and MS. Wolfe, Langmuir, 8, 2382 (1992).
M.A. Petti, 7'” International SAMPE Electronics Conference, 526 (1994).

K.K. Baikerikar and AB. Scranton, “Photopolymerizable Liquid Encapsulants for
Microelectronic Devices,” submitted to Polymer.

K.K. Baikerikar and AB. Scranton, RadTech ’98 North America U V/EB Conference
Proceedings, 712 (l 998).

HA. Barnes, J .F . Hutton, and K. Walters, An Introduction to Rheology, Elsevier
Amsterdam (1989).

CD. Han, Multiphase F low in Polymer Processing, Academic Press (1981).

16. RF Probstein, M.Z. Sengun, and T-C. Tseng, J. Rheol, 38, 811 (1994).

17.
18.

A. Einstein, Ann. Physik, 19, 289 (1906).
A. Einstein, Ann. Physik, 34, 591 (1911).

19. M.J. Mooney, J. Colloid. Sci, 6, 162 (1951).

20. R. Roscoe, Br. J. Appl. Phys, 3, 267 (1952).

64

21. T. Kitano, T. Kataoka, and T. Shirota, Rheol. Acta, 20, 207 (1981).
22. I.M. Krieger and T.J. Dougherty, Trans. Soc. Rheol, 3, 137 (1959).
23. N. Ouchiyama and T. Tanaka, Ind. Eng. Chem. Fund, 19, 338 (1980).

24. A.J. Poslinski, M.E. Ryan, R.K. Gupta, S.G. Seshadri, and F.J. Frechette, J. Rheol,
32, 751 (1988).

25. V. Narayanan, K.K. Baikerikar, and AB. Scranton, RadTech ’98 North America
U V/EB Conference Proceedings, 31 (1998).

26. V. Narayanan and AB. Scranton, Trends Polym. Sci, 5, 415 (1997).
27. K. Sweeney and R. Geckler, J. Appl. Phys, 25, 1135 (1954).
28. J .S. Chong, E.B. Christiansen, and AD. Baer, J. Appl. Polym. Sci, 15, 2007 (1971).

29. W.G. Kim and J.H. Ryu, J. Appl. Polym. Sci, 65, 1972 (1997).

65

Chapter 6

MECHANICAL AND THERMOMECHANICAL PROPERTIES OF
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS

6.1. INTRODUCTION

Microelectronic devices are complex and delicate devices that require effective
encapsulation and packaging to ensure their long-term reliability. The encapsulant must
exhibit excellent physical, electrical, and mechanical properties for the hostile and
extreme temperature cycling requirements. Indeed, the mechanical and therrno-
mechanical behavior of the encapsulant plays a key role in the reliability of the
microelectronic device.

The encapsulant is typically characterized by a set of mechanical and thermal
properties that determine the suitability of the molding compound for a given application
and process. Speciﬁc tests are performed on encapsulants to ascertain design parameters,
guarantee reproducible moldability, assure lot uniformity, and analyze package
construction.

In this chapter, an extensive series of studies on the mechanical and thermal
properties, as well as the degree of cure of photopolymerizable liquid encapsulants, is
described. We have characterized the ﬂexural modulus, ﬂexural strength, coefficient of
thermal expansion, glass transition temperature, and thermal stress parameter of PLEs
containing 70.0-74.0 wt.% fused silica. In addition, the effects of the illumination time,
postcure time, fused silica loading, and the inclusion of a thermal initiator on the thermal

and mechanical properties of the ﬁnal cured encapsulants have been investigated.

66

6.2. EXPERIMENTAL
6.2.1. Preparation of Photopolymerizable Liquid Encapsulant

The photopolymerizable liquid encapsulant is comprised of a base resin, fused
silica ﬁller, photoinitiator, silane coupling agent, and in some cases a thermal initiator.
The base resin used in these studies was an epoxy novolac-based vinyl ester resin
(DERAKANE 470-45, Dow Chemical) that was chosen for its low initial viscosity
0.0456 Pa-s (45.6 cP) at 25 °C, as well as, its appropriate thermal and mechanical
properties upon cure. The photoinitiator used was bis(2,4,6-trimethylbenzoyl)
phenylphosphine oxide (IRGACURE 819, Ciba). This photoinitiator, which we will
henceforth denote as BAPO, has been shown to be especially appropriate for curing
relatively thick polymers and composites due to its efficient photobleaching."2 In some
systems, the thermal initiator, benzoyl peroxide (Aldrich), was used in order to
investigate the effects of a dual photo/thermal initiation scheme on the resulting material
properties of the encapsulant. The ﬁllers used were crushed (angular), untreated fused
silica obtained from MINCO, Inc. A silane coupling agent, 3-
methacryloxypropyltrimethoxysilane (Z-6030, Dow Corning), was used to provide a
stable bond between the base resin and the fused silica ﬁller, and to improve the
processability of the liquid encapsulant formulation.

Fused silica, which is the most common ﬁller for commercial microelectronic
encapsulants, was selected for these experiments because it possesses the optimum
combination of properties. Speciﬁcally, it exhibits a low coefficient of thermal expansion
(~0.6 x 10'6 °C") and a low enough abrasiveness to provide a relatively long molding tool

lifetime, especially when compared to aluminum nitride.3 In addition, fused silica is

67

transparent to UV light, and is therefore appropriate for UV curing.4 To minimize the
viscosity of the PLE, fused silica products with two different particle size distributions
were used: (i) MIN-SIL 40 (volume average particle size = 32.42 pm, median particle
size = 22.11 pm) and (ii) MIN-SIL 550 (volume average particle size = 10.37 pm,
median particle size = 6.026 run). In Chapter 5, it was shown that PLEs comprised of a
50:50 wt.% blend of MIN-SIL 40 and MIN-SIL 550 provided the best rheological
properties and processability. Therefore, this blend has been used for all of the PLEs in
this investigation.

The PLEs were prepared by adding 0.2 wt.% BAPO photoinitiator (based on resin
weight) to the epoxy novolac-based vinyl ester resin and stirring at room temperature on
a magnetic stir plate until the photoinitiator completely dissolved (~3 minutes). In some
cases the thermal initiator, benzoyl peroxide, was also added in the amount of 1.2 wt.%
(based on resin weight) and, again the system was stirred until dissolution was complete
(~8 minutes). The silane coupling agent was added dropwise to the desired 1 wt.%
(based on ﬁller weight) and the resin mixture was stirred on the magnetic stir plate.
Fused silica was added in the amount of 70.0, 72.0, or 74.0 wt.% and the system mixed
by using a vortex mixer until the ﬁller was completely dispersed and a homogeneous
mixture was achieved. The PLE was then degassed in a vacuum oven. Once these steps
were completed, the PLE was ready to use for preparing samples required for the material
property characterization studies.

6.2.2. Flexural Modulus and Strength
Specimens for ﬂexural testing were prepared by photopolymerizing the liquid

encapsulant in rectangular silicone molds with unﬁltered UV light (200 mW/cm2 UVA

68

intensity) from a 3000 W arcless mercury vapor lamp (Fusion UV Systems, model
F450T). Light intensities were measured (over the 320-390 nm range) using a
UVICURE Plus high energy UV integrating radiometer. The specimens were 76.2 mm
long, 12.7 mm wide, and 3.2 mm deep. To investigate the effect of the illumination time,
the samples were illuminated for 60.0, 90.0, 120.0, 150.0, and 180.0 seconds. To
determine the effect of thermal postcure (after UV illumination) on the resulting thermal
and mechanical properties, some specimens were postcured at 170°C in a laboratory oven
for 2, 4, 6, and 8 hours.

The ﬂexural modulus and strength of the photocured samples were determined
using a United SFM-20 instrument in accordance with the ASTM D 790 method. The
ﬂexural properties were measured using the three point ﬂexural test with a span length of
50.8 mm (2 inches), a span-to-depth ratio of 16:1, a 454 kg (1000 lb.) load cell, and a
downdrive rate of 1.7 mm/min (0.07 in/min). F lexural modulus and strength values were
calculated using DATUM 97 software (United Testing Systems).

6.2.3. Degree of Cure

The degree of cure was determined using differential scanning calorimetry (DSC).
In these experiments, cure was monitored by measuring the heat released by the sample
as the exothermic reaction proceeded. The ultimate conversion achieved during a given
set of cure conditions was determined by measuring the additional cure attained at an
elevated temperature (180°C) after the completion of the speciﬁed cure cycle. Cure
conditions that were investigated include: (A) photopolymerization for two minutes
without the addition of a thermal initiator, (B) photopolymerization for two minutes

without the addition of a thermal initiator followed by postcure for 6 hours at 170°C, (C)

69

photopolymerization for two minutes with the addition of a thermal initiator, and (D)
photopolymerization for two minutes with the addition of a thermal initiator followed by
postcure for 6 hours at 170°C. In each of these experiments the photopolymerization was
initiated at room temperature using the 3000 Watt Fusion Systems lamp without external
heating or cooling. The temperature of 180°C was selected because it is higher than any
temperature that the samples experienced during the cure cycles (and higher than the
glass transition temperature of the samples) but lower than the degradation temperature of
the cured PLEs.

Samples for the DSC experiments were prepared in the following manner. First,
rectangular bars of the same dimensions as described above for the ﬂexural tests were
prepared according to the four different cure conditions. Four rectangular bars were
prepared for each cure condition (all samples contained 74.0 wt.% fused silica). After the
cure cycle was complete, a piece was cut from each rectangular bar and ground into
smaller pieces with a mortar and pestle to prepare a sample for the DSC studies. The
mass of each DSC sample was in the range of 18.0 to 25.0 mg. The samples were
weighed both before and after the DSC cure experiment and no change in mass was
observed for these solid samples.

The DSC experiments were performed using a Perkin Elmer DSC 7 Differential
Scanning Calorimeter interfaced with a computer via the TAC 7/DX Thermal Analysis
Instrument Controller. The Perkin Elmer Pyris Software package for Windows was used
for the temperature control programs. The experimental procedure was as follows. A
solid DSC sample weighing between 18.0 and 25.0 mg was placed in an aluminum DSC

pan, which, in turn, was placed in the sample holder of the DSC instrument. An empty

70

aluminum DSC pan was used in the reference holder. After the samples were placed in
the DSC pans, the sample cells were rapidly heated from 25°C to 180°C by the DSC
instrument. Once a temperature of 180°C was attained (in approximately 30 seconds) the
instrument began to collect data for heat ﬂow from the sample as a function of time. The
heat ﬂow was measured while the sample was maintained isotherrnally at 180°C for 150
minutes. It was generally observed that the heat ﬂux from the sample ceased (indicating
that the reaction was ﬁnished) in less than 20 minutes, therefore the duration of 150
minutes for the experiment was ample time to ensure that the reaction went to
completion. For the analysis of the data, the values of the heat ﬂux ﬁ'om the sample were
normalized by the mass of the sample.

DSC experiments were also performed on the uncured liquid formulation
containing 74 wt.% ﬁrsed silica. These studies allowed the entire reaction exotherm
corresponding to complete cure to be established. Four independent DSC runs were
performed on the liquid formulation containing only the photoinitiator and four additional
runs were performed on the liquid formulation containing both the photoinitiator and
thermal initiator. In each of these experiments, a drop of the liquid formulation with a
mass between 18.0 and 25.0 mg was placed in an aluminum DSC pan and an empty
aluminum DSC pan was used in the reference holder. The heat ﬂow was measured while
the samples were maintained isotherrnally at 110°C for 250 minutes. A lower
temperature was used for these studies to minimize evaporation of the styrene in the
resins. The samples were weighed both before and after the experiment to determine
how much evaporation took place. The PLE samples with both photoinitiator and

thermal initiator lost an average of 1.13% of their total weight, while the PLE samples

71

with only the photoinitiator lost an average of 8.62% of their total weight. As will be
shown in Section 6.3.4, the liquid system without the thermal initiator did not polymerize
at 110°C since the photoinitiator is stable at this temperature. This explains why more
evaporation of the monomer was observed for this case. For the system containing both
the photoinitiator and the thermal initiator, the ﬁnal weight of the sample was used to
normalize the heat ﬂow data in the reaction exotherm proﬁle.
6.2.4. Coefﬁcient of Thermal Expansion and Glass Transition Temperature
Specimens for coefficient of thermal expansion (CTE) testing were cut from the
samples that had been used for ﬂexural testing so that the thermal stress parameter of the
photocured encapsulants could be estimated. The dimensions of the samples were 11.0
mm length x 8.0 mm width x 3.2 mm thickness. All samples were cut using a diamond
wafering saw. The CTE below glass transition (on) and CTE above glass transition ((12)
were both measured using a DuPont 943 therrnomechanical analyzer (TMA) interfaced
with a DuPont model 9900 thermal analyzer controller. The samples were heated ﬁ'om
room temperature (23°C) to 234°C at a constant rate of 3°C/min. The change in the
sample thickness during heating was recorded in the personal computer and or; was
obtained from the inclined line connecting two points on the TMA curve, 50°C and 70°C.
To determine 012, the inclined line connecting 200°C and 230°C was used. The glass
transition temperature was taken to be the temperature at the intersection of these two

lines.

72

6.3. RESULTS AND DISCUSSION

6.3.1. Effect of UV Illumination Time on F lexural Modulus and Strength

An important mechanical property for microelectronic encapsulants is the ﬂexural
modulus, which characterizes the stiffness of a material under an applied load.5 The
optimal value of the ﬂexural modulus provides a balance between two conﬂicting
requirements. A high value of the ﬂexural modulus is desired to maximize the protection
of the chip from an applied load; however, lower values of the ﬂexural modulus lead to
lower thermal stresses. Thermal stresses associated with high modulus encapsulants can
lead to failure of the microelectronic device. In particular, this stress has been reported to
cause interfacial cracking and breakage of the wirebond leads, as well as passivation
layer cracking and aluminum pattern deformationf”7 To reduce the thermal shrinkage
stresses, while maintaining appropriate protection against mechanical shock, vibration,
etc., a ﬂexural modulus near 1000 kgp’mmz (9800 MPa) is desired.3

Experimental results for the effect of ﬁller loading and UV illumination time on
the ﬂexural modulus and strength of the photocured encapsulants ﬁlled with 70.0 to 74.0
wt.% fused silica are shown in Figures 6.1 and 6.2, respectively. All samples contained
1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on resin
weight), and the balance epoxy novolac-based vinyl ester resin. Samples were
photocured with UV light of 200 mW/cm2 UVA intensity. Each bar in these bar charts
represents the average of 5 samples and the error bars indicate one standard deviation
above and below the mean. As expected, the ﬂexural modulus and ﬂexural strength
increase with fused silica content. In addition, the ﬂexural properties do not change

signiﬁcantly as the illumination time is increased beyond 120 seconds. Therefore, 120

73

seconds was chosen for the in-mold illumination time for subsequent experiments. Note
that the values of the ﬂexural modulus are near the desired value of 1000 kgp’mmz. The
results shown in Figure 6.1 suggest that the PLEs could allow the in—mold cure time to be
reduced signiﬁcantly below 120 seconds (which is the typical in-mold cure time for
transfer molding compounds). For example, the mechanical properties observed after 90
seconds are essentially equivalent to those after 120 seconds, and an ejectabie hardness is
certainly achieved even earlier. This result is signiﬁcant because the in-mold cure time
consumes approximately 70% of the overall cycle time and an in-mold cure time
reduction of even 15 seconds can translate to a productivity increase of about 10%.3

The data in Figure 6.2 illustrate that the ﬂexural strength of the samples as they
are taken from the mold, immediately after illumination, range from 3 to 4.5 kgp’mm2 (~
30 to 45 MPa). Current microelectronic encapsulants exhibit a ﬂexural strength of at
least 9 kgp’mmz, therefore the strength of the PLEs immediately after illumination is
below the range of desired values. To address this point, we investigated the effect of a
postcure step (after photocuring the samples for 120 seconds) on the ﬂexural modulus
and the ﬂexural strength.
6.3.2. Effect of Postcure Time on F lexural Modulus and Strength

After the several minute cycle in the molding tool, almost all molding compounds
are postcured for 4 to 8 hours at 170 to 175°C, in order to fully develop the material
properties of the encapsulant. Postcuring is a key step in microelectronic encapsulation
because encapsulation materials do not reach 100% chemical conversion during the
transfer molding process. Typically, the reaction is continued in the mold until the

encapsulant reaches a suitable ej ectable hardness and then the encapsulant is postcured.

74

The effects of postcure time and ﬁller loading on the flexural modulus and strength of the
photocured encapsulants are shown in Figures 6.3 and 6.4, respectively. Each bar in
these bar charts represents the average of 5 samples and the error bars indicate one
standard deviation above and below the mean. The results shown in Figures 6.3 and 6.4
illustrate that postcure has little effect on the ﬂexural modulus, but has a signiﬁcant
impact on the ﬂexural strength of the specimens. For example, a comparison of the
ﬂexural modulus data for no postcure to the data corresponding to 8 hours of postcure,
demonstrates that the values are not signiﬁcantly different (for all three fused silica
loadings, the mean ﬂexural modulus and the associated standard deviations are nearly the
same). In contrast, the ﬂexural strength increases consistently as the postcure time is
increased from zero to six hours. For example, the samples ﬁlled with 74.0 wt.% fused
silica exhibit an increase in the ﬂexural strength from a value of 4 kgg/mm2 with no
postcure to more than 8 kgfl/mm2 after a postcure time of 6 hours. We attribute the
increase in ﬂexural strength associated with the postcure step to additional thermal cure
that occurs at the elevated temperatures, and possibly annealing of the polymer. Other
authors have reported this result, as well. For example, Cook and Johannson reported
signiﬁcant increases in tensile strength, ﬂexural strength, and fracture toughness for

8 A sample that has vitriﬁed at room

composites postcured at 100°C for 24 hours.
temperature can experience additional cure at elevated temperatures due to the enhanced
molecular mobility that allows unreacted double bonds to encounter active centers. The

enhancement in cure associated with the postcure step was characterized using

differential scanning calorimetry, as will be explained in Section 6.3.4.

75

6.3.3. Effect of Thermal Initiator on Flexural Modulus and Strength

To investigate the effects of a dual photo/thermal initiation scheme, we included
the thermal initiator, benzoyl peroxide, in addition to the photoinitiator BAPO in some
PLEs. Tables 6.1 and 6.2 both illustrate the effect of the inclusion of the thermal initiator
on the ﬂexural strength of the resulting composites. The data in Table 6.1 corresponds to
samples that were .UV cured for 120 seconds under a light intensity of 200 mW/cmz,
while the data in Table 6.2 corresponds to samples that were UV cured for 120 seconds
under a light intensity of 200 mW/cm2 and then postcured for 6 hours at 170°C after UV
illumination. In addition to the fused silica at the indicated loading (70.0, 72.0, or 74.0
wt.%), all samples contained 1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2
wt.% BAPO (based on resin weight), and the balance epoxy novolac-based vinyl ester
resin. Samples with thermal initiator also contained 1.2 wt.% benzoyl peroxide (based on
resin weight). The reported ﬂexural strength values are the average of ﬁve samples, and
the indicated range corresponds to one standard deviation.

The data in Tables 6.1 and 6.2 illustrate that the inclusion of the thermal initiator
has a signiﬁcant effect on the ﬂexural strength of the cured encapsulants (for each ﬁller
loading, the ﬂexural strength increased by approximately 2 kgp’mm2 upon the addition of
the thermal initiator). As illustrated by the calorimetric studies described in the next
section, the increase in the ﬂexural strength can be attributed to an enhancement in the
degree of cure associated with the addition of the thermal initiator. Moreover, a 6 hour
postcure step at 170°C also enhances the ﬂexural strength of the encapsulants containing

both the photoinitiator and thermal initiator. The ﬂexural strength of the samples

76

containing the thermal initiator and 72.0 or 74.0 wt.% fused silica (and postcured for 6
hours) now fall in the range of typical ﬂexural strength values for encapsulants.

The effect of the thermal initiator on the ﬂexural modulus of the cured
encapsulants is illustrated by the data in Tables 6.3 and 6.4. The data in Table 6.3
corresponds to samples that were UV cured for 120 seconds under a light intensity of 200
mW/cmz, while the data in Table 6.4 corresponds to samples that were UV cured for 120
seconds under a light intensity of 200 mW/cm2 and then postcured for 6 hours at 170°C
after UV illumination. In addition to the fused silica at the indicated loadings (70.0, 72.0,
or 74.0 wt.%), all samples contained 1.0 wt.% silane coupling agent (based on ﬁller
weight), 0.2 wt.% BAPO (based on resin weight), and the balance epoxy novolac-based
vinyl ester resin. Samples with thermal initiator also contained 1.2 wt.% benzoyl
peroxide (based on resin weight). The reported ﬂexural modulus values are the average
of ﬁve samples, and the indicated range corresponds to one standard deviation. The data
in Tables 6.3 and 6.4 illustrate that the values of the ﬂexural modulus for the samples
containing both the photoinitiator and thermal initiator are essentially equivalent to those
for the samples containing only a photoinitiator. In addition, the values of the ﬂexural

2 value recommended by Manzione for

modulus are close to the 1000 kgp’mm
microelectronic encapsulants.3
6.3.4. Degree of Cure

The degree of cure was determined using differential scanning calorimetry by
measuring the heat released from the sample as the exothermic reaction proceeded. The

ultimate conversion achieved during a given set of cure conditions was determined by

measuring the additional cure attained at an elevated temperature after the completion of

77

the speciﬁed cure cycle. These experiments were performed to investigate the effect of
postcure and the inclusion of the thermal initiator on the additional cure. To ﬁnd the
conversion attained during this additional cure, the total heat released from the solid
samples (that had gone through a given cure cycle) was divided by the total heat released
from the liquid formulation.

A representative reaction DSC proﬁle for the uncured liquid formulation
containing both the photoinitiator and thermal initiator is shown in Figure 6.5. This
ﬁgure contains a plot of the heat ﬂow from the sample as a ﬁrnction of time and the
reaction proﬁle exhibits a shape that is characteristic of free radical polymerizations.9
Since the polymerization reaction is exothermic, the area under this curve is proportional
to the degree of cure. Although the data was collected for 150 minutes, the cure was
generally complete (the proﬁle becomes horizontal) in ~15 minutes, as illustrated in
Figure 6.5. The baseline for integrating the area under the curve was established by
horizontal extrapolation of the ﬁnal steady-state value. The integration was performed
using the commercial data analysis software package Origin 5.0 (Microcal Software,
Inc.).

Cure conditions that were investigated in this study include: (A)
photopolymerization for two rrrinutes without the addition of a thermal initiator, (B)
photopolymerization for two minutes without the addition of a thermal initiator followed
by postcure for 6 hours at 170°C, (C) photopolymerization for two minutes with the
addition of a thermal initiator, and (D) photopolymerization for two minutes with the
addition of a thermal initiator followed by postcure for 6 hours at 170°C. These samples

will henceforth be referred to as A, B, C, and D, respectively. After the given cure cycle,

78

each photopolymerized (solid) encapsulant was subject to a DSC postbake at 180°C for
150 minutes. Based upon the heat released during this DSC postbake experiment, the
degree of cure was determined indirectly by measuring the additional cure that could be
attained at this elevated temperature. For a given solid sample, the additional degree of
cure was determined by dividing the area under the corresponding DSC reaction proﬁle
by the area corresponding to complete cure. The latter value was obtained from the
reaction exotherm exhibited by the previously unreacted liquid formulation (Sample E).
The experimental data for samples containing 74 wt.% silica is shown in Table 6.5. The
values in this table represent the average of four samples, and the indicated range
corresponds to one standard deviation.

The data in Table 6.5 help to explain the effect of the cure cycle on the ﬂexural
strength that was identiﬁed in the previous section. Recall that ﬂexural strength was
enhanced by postcure and by the presence of the thermal initiator. The data in Table 6.5
illustrate that either of these factors leads to a signiﬁcant enhancement in the conversion
attained during the cure cycle (an enhancement of ~8%). For example, the effect of
postcure is illustrated by comparing the data for Sample A to that for Sample B, while the
effect of the inclusion of the thermal initiator is illustrated by comparing the data for
Sample A to that for Sample C. Therefore, the enhanced ﬂexural strength of these
samples may arise from an increased degree of cure. The data for Sample D further
illustrate that the highest degree of cure is obtained if a thermal initiator is included and
the sample is subject to postcure. This is the sample that also exhibited the highest

ﬂexural strength; however, if Sample D is compared to Sample B or Sample C, it is

79

unclear why the relatively small increase in conversion would lead to the relatively large
increase in the ﬂexural strength.
6.3.5. Effect of Postcure Time on CTE below Glass Transition

The linear coefﬁcient of thermal expansion (CTE) is deﬁned as the ratio of the
change in the length of the sample to the change in temperature per unit initial length.'0
This parameter is very important in semiconductor encapsulation because a signiﬁcant
CTE mismatch between the encapsulant, leadframe, and the silicon chip can lead to the
build up of internal stresses in the semiconductor device, and could lead to cracking of
the chip or encapsulant. Current encapsulants exhibit or. values (CTE below glass
transition) in the range of 16-25 um/m°C, while copper leadframes possess a value of 16-
17 um/m°C and silicon exhibits a value of 3-4 ilrrl/m°C.ll The service temperature for
microelectronic encapsulants is typically between 75 and 85°C, which is well below the
glass transition temperature of the encapsulant.3 For this reason, the value of the CTE

above the glass transition temperature, or;, is not as important as the CTE below the glass

transition temperature, on.

For the PLEs, the experimental results for the coefficient of thermal expansion
below glass transition are shown in Figure 6.6. Figure 6.6 illustrates the effect of the
postcure time on the observed CTE for samples ﬁlled with 70.0 to 74.0 wt.% fused silica.
All samples contained 1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.%
BAPO (based on resin weight), and the balance epoxy novolac-based vinyl ester resin.
Each bar in the following bar charts represents the average of 5 samples and the error
bars indicate one standard deviation above and below the mean. The data illustrate that

for all three ﬁller loadings, a signiﬁcant reduction in or; is observed after just 2 hours of

80

postcure. This reduced value of the CTE may be attributed to the enhanced cure (~8%
additional cure as determined using calorimetry) associated with the postcure step. Most
of this additional cure occurs within the ﬁrst two hours of postcure. After a two hour
postcure, all of the photocured samples exhibit or; values well within the range required
for current encapsulants. As expected, or. decreased as the fused silica loading was
increased from 70.0 to 74.0 wt.%. The 74.0 wt.% silica system looks especially
promising for microelectronic encapsulants because of its low or. value (~20 um/m°C).
6.3.6. Effect of Thermal Initiator on CTE below Glass Transition

The effect of the inclusion of the thermal initiator on the value of on is illustrated
in Table 6.6. All samples contained 70.0, 72.0, or 74.0 wt.% fused silica, 1.0 wt.% silane
coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on resin weight), and the
balance epoxy novolac-based vinyl ester resin. Samples with thermal initiator also
contained 1.2 wt.% benzoyl peroxide (based on resin weight). All samples were
photocured for 2 minutes with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C in a laboratory oven. CTE values are the average of ﬁve
samples, and the indicated range corresponds to one standard deviation. The data
illustrate that the thermal initiator does not have any appreciable effect on the value of on.
For all three fused silica loadings, the mean or. values and the associated standard

deviations are nearly the same.

6.3.7. Effect of Postcure Time and Thermal Initiator on CTE above Glass
Transition

The CTE above the glass transition temperature, on, was also measured as a
function of ﬁller loading and postcure time for systems containing only the photoinitiator,

as well as, for systems containing both the photoinitiator and thermal initiators. Figure

81

6.7 illustrates the effect of the postcure time on the CTE above glass transition for
samples ﬁlled with 70.0 to 74.0 wt.% fused silica and containing only the photoinitiator.
All samples contained 1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.%
BAPO (based on resin weight), and the balance epoxy novolac-based vinyl ester resin.
Figure 6.7 illustrates that the CTE above glass transition decreases with an increase in
fused silica content and is essentially independent of the postcure time. In addition, all of
the values ranged between 55 and 65 um/m°C.

The values for the samples postcured for six hours and containing 70.0, 72.0, and
74.0 wt.% fused silica (and no thermal initiator) were 66.74 i 3.45, 61.78 i 1.70, and
56.68 :1: 2.91 urn/m°C, respectively. The values for the samples postcured for six hours
and containing 70.0, 72.0, and 74.0 wt.% fused silica (and both photoinitiator and thermal
initiator) were 69.44 i 1.30, 63.46 i 2.37, and 59.24 :I: 2.39 uni/m°C, respectively. These
values were essentially independent of the presence of the thermal initiator.
6.3.8. Effect of Postcure Time on Glass Transition Temperature

To provide effective protection of the microelectronic device, an encapsulant
must exist in the glassy state at the use temperature (~80°C). Commercial encapsulants
typically exhibit glass transition temperatures in the range of 140 to 180°C. Figure 6.8
shows experimental data for the glass transition temperature of the photopolymerized
encapsulants as a function of the postcure time for samples containing 70.0, 72.0, and
74.0 wt.% fused silica. As expected, the glass transition temperature generally increases
with increasing postcure time, and reaches a plateau in ~6 hours, consistent with the

previous results for the ﬂexural strength and the CTE. After 2 hours of postcure, all of

82

the samples exhibited glass transition values between 145 and 155°C, within the range
required for microelectronic encapsulants.
6.3.9. Effect of Thermal Initiator on Glass Transition Temperature

The effect of the thermal initiator on the glass transition temperature is illustrated
in Table 6.7. All samples contained 70.0, 72.0, or 74.0 wt.% fused silica, 1.0 wt.% silane
coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on resin weight), and the
balance epoxy novolac-based vinyl ester resin. Samples with thermal initiator also
contained 1.2 wt.% benzoyl peroxide (based on resin weight). All samples were
photocured for 2 minutes with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C in a laboratory oven. Glass transition temperatures are the
average of ﬁve samples, and the indicated range corresponds to one standard deviation.
The data illustrate that the samples exhibit glass transition temperatures appropriate for
microelectronic encapsulants, and that the presence of the thermal initiator has little

effect on the glass transition temperature.

6.3.10. Thermal Stress Parameter

As was mentioned previously, lowering the stress parameter leads to improved
device reliability. The thermal stress parameter, a", is typically calculated using
Equation 6.1.3

o‘ = (exp, — aaEpgrrg — 7“,) (6.1)
Here 51,,g represents the coefficient of thermal expansion below glass transition of the
encapsulation material, a,- is the coefﬁcient of thermal expansion of the silicon device or
metal leadframe, Epg is the modulus of elasticity (tensile or ﬂexural) of the encapsulant,

and T8 is the glass transition temperature of the encapsulant (°C). In order to evaluate the

83

thermal stress parameter for the photopolymerized encapsulants of this study, the
experimental values for a,,,,, Epg (ﬂexural) and T8 were substituted into this equation
along with literature values for a, and T1. Speciﬁcally, we used the CTE of a copper
metal leadfrarne (16.3 um/m°C) for a,- and -65°C for T1, since this is the usual starting
temperature for temperature cycling tests as speciﬁed by Mil. Std. 38510 Group D.3

Data for the thermal stress parameter of photopolymerized encapsulants with and
without the thermal initiator are shown in Table 6.8. All samples contained 70.0, 72.0, or
74.0 wt.% fused silica, 1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.%
BAPO (based on resin weight), and the balance epoxy novolac-based vinyl ester resin.
Samples with thermal initiator also contained 1.2 wt.% benzoyl peroxide (based on resin
weight). All samples were photocured for 2 minutes with UV light of 200 lecm2 UVA
intensity and then postcured for 6 hours at 170°C in a laboratory oven. Thermal stress
parameter values are the average of ﬁve samples, and the indicated range corresponds to
one standard deviation. First of all, the data illustrate that the presence of the thermal
initiator has little, if any, effect on the thermal stress parameter. More importantly, Table
6.8 illustrates that the photopolymerized encapsulants containing 74.0 wt.% fused silica
exhibit a thermal stress parameter between 1.0 and 1.2 kgg/mmz. These values compare
quite favorably to those reported for current encapsulants. For example, using the
literature values for apg, Epg (ﬂexural) and T8, reported by Nakamura et al.12 and Lin et
al.'3 along with the previously mentioned values for a,- and T I, (16.3 rim/m°C and -65°C,
respectively) current epoxy encapsulants exhibit a thermal stress parameter of ~1.3

kgp’mmz.

84

6.4. CONCLUSIONS
In this chapter, we have characterized the degree of cure, ﬂexural modulus,
ﬂexural strength, coefﬁcient of thermal expansion, glass transition temperature, and
thermal stress parameter of novel photopolymerizable liquid encapsulants comprised of
an epoxy novolac-based vinyl ester resin (~25 wt.%), fused silica (70-74 wt.%), and
small amounts of a photoinitiator, silane coupling agent, and in some cases, a thermal
initiator. The effects of the fused silica loading, UV illumination time, and postcure time
on these thermal and mechanical properties were also investigated. The results indicate
that a photocurable encapsulant loaded with 74.0 wt.% fused silica is very promising for
this application since the material properties (ﬂexural strength, ﬂexural modulus,
coefﬁcient of thermal expansion, glass transition temperature, and thermal stress
parameter) meet or exceed those of current commercial encapsulants, while the viscosity
and in-mold cure time are reduced. These formulations are liquid at room temperature
(current formulations are solid) and cure to an ejectabie hardness in less than two minutes
for an initiating light intensity of 200 mW/cmz. The photocured samples exhibit
appropriate values for the above mentioned material properties. Speciﬁc observations are
as follows:
1. The ﬂexural modulus and ﬂexural strength increased with UV light illumination time
until reaching a plateau value in ~90 to 120 seconds. This result indicates that the in-
mold cure time of the PLEs could be as low 75 or 90 seconds at a moderate light

intensity of 200 mW/cmz.

85

. The ﬂexural strength and glass transition temperature increased with increasing
postcure time at 170°C before reaching a plateau in ~4 to 6 hours. Current
encapsulants are typically postcured 4 to 8 hours at 170 or 175°C.

. The coefﬁcient of thermal expansion below the glass transition decreased with
increasing postcure time at 170°C before reaching a plateau in ~4 to 6 hours.

. Inclusion of a thermal initiator, benzoyl peroxide, in the photopolymerizable liquid
encapsulant signiﬁcantly increased the ﬂexural strength of the samples, but had little
effect on the other material properties;

. The ultimate degree of cure attained in a given cure cycle was signiﬁcantly enhanced
by the inclusion of a thermal initiator (in addition to the photoinitiator) and by

postcure at an elevated temperature.

86

 

1,400
E170 wt% silica
B72 wt% silica
1:200 ' I74 we silica

1,000 '

 

Flexural Modulus (kg , Imm’)

 

 

 

 

 

90

 

 

 

 

 

120 150 180

UV Illumination Time (seconds)

Figure 6.1. Flexural modulus as a function of UV illumination time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica.

87

 

 

S"
or

D70 wt°/e silica
'72 wt% silica
'74 wt% silica

 

 

 

0|
_1_

 

 

.3“
or

 

&

 

5"
or

  
 

 

Flexural Strength (kg , Immz)
M
n in 00

 

 

 

 

 

 

 

60 90 120 150 180
UV Illumination Time (seconds)

Figure 6.2. Flexural strength as a function of UV illumination time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica.

88

 

 

I370 Wt.% silica
'72 wt.% silica
'74 wt.% silica

 

1.
i

 

 

    

 

Flexural Modulus (kg ,Imm’

 

 

 

 

 

 

 

Postcure Time (hours)

Figure 6.3. Flexural modulus as a function of postcure time for samples containing 70.0,
72.0, and 74.0 wt.% fused silica. Samples were photocured for 2 minutes with UV light
of 200 mW/cm2 UVA intensity and then postcured at 170°C in a laboratory oven.

89

 

 

C170 M% silica
9 "T '72 WW. silica
'74 “71% silica

 

    

 

 

 

 

 

 

Flexural Strength (kg ,Immz)

 

 

 

 

 

 

0 2 4 6 8

Postcure Time (hours)

Figure 6.4. Flexural strength as a function of postcure time for samples containing 70.0,
72.0, and 74.0 wt.% fused silica. Samples were photocured for 2 minutes with UV light
of 200 mW/cm2 UVA intensity and then postcured at 170°C in a laboratory oven.

a.
in

b

u
in

a
1

 

Heat Flow (VII/g resin)
to t:

i;

 

‘
l

o
in

 

 

o ' T I T I I I I T

012 3 4 5 c 7 a 9101112131415
11me (minutes)

Figure 6.5. Representative DSC reaction proﬁle at 110°C for the liquid formulation

containing 74.0 wt.% fused silica, 1.0 wt.% silane coupling agent (based on ﬁller

weight), 0.2 wt.% BAPO (based on resin weight), 1.2 wt.% benzoyl peroxide (based on
resin weight), and the balance epoxy novolac-based vinyl ester resin.

91

 

 

42
" I 070 wt% silica
4° I72 multic-

 

 

 

 

 

 

 

36 '—: ..
34 d_.
32 ._
30 “r
28 ‘—
26 "—
24 “‘

 

 

 

 

a1 (”mlm°C)

 

 

 

 

 

 

 

 

 

18

Postcure Time (hours)

Figure 6.6. CTE below glass transition as a function of postcure time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica. Samples were photocured for 2 minutes
with UV light of 200 mW/cm2 UVA intensity and then postcured at 170°C in a laboratory

oven 0

92

 

80..

 

 

1370 wet silica
572 was silica
I 74 wt% silica

 

 

 

 

70

40..

a, (umIm°C)

30i

 

 

 

 

 

 

 

 

 

Postcure Time (hours)

Figure 6.7. CTE above glass transition as a function of postcure time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica. Samples were photocured for 2 minutes
with UV light of 200 mW/cm2 UVA intensity and then postcured at 170°C in a laboratory

oven.

93

 

160

 

l370 wt% silica
E72 wI°/e silica
150 ~ '74 wt°/e silica

 

 

 

'87

130‘

'5 1.4".» . ' 53.
:1HJLA..L 11.10:- ;“f«->. ‘ -.;x‘-4" :4

5‘. ‘ .L.........;u_"" .1 .x

120 ‘

110 ‘

Glass Transition Temperature ( °C)

 

 

 

 

 

 

 

 

Postcure Time (hours)

Figure 6.8. Glass transition temperature as a function of postcure time for samples
containing 70.0, 72.0, and 74.0 wt.% fused silica. Samples were photocured for 2
minutes with UV light of 200 mW/cm2 UVA intensity and then postcured at 170°C in a
laboratory oven.

Table 6.1. Effect of the inclusion of thermal initiator on ﬂexural strength for samples that
were not postcured.

 

 

Flexural Strength Flexural Strength
Wt.% Fused (kgp’mmz) (kgp’mmz)
Silica Without With
Thermal Initiator Thermal Initiator
70.0 3.70 i 0.52 5.25 i 0.38
72.0 3.92 i 0.23 5.50 i 0.65
74.0 4.24 i 0.38 5.71 i 0.36

 

Table 6.2. Effect of the inclusion of thermal initiator on ﬂexural strength for samples that
were postcured for 6 hours.

 

 

F lexural Strength F lexural Strength
Wt.% Fused (kgp’mmz) (kgr/mmz)
Silica Without With
Thermal Initiator Thermal Initiator
70.0 6.71 :l: 0.69 8.83 i 1.68
72.0 7.59 i 0.24 9.28 :t 0.95
74.0 8.12 :t 0.69 9.92 d: 1.44

 

Table 6.3. Effect of the inclusion of thermal initiator on ﬂexural modulus for samples
that were not postcured.

 

F lexural Modulus

 

Flexural Modulus
Wt.% Fused (kgglmmz) (kgp’rrnnz)
Silica Without With
Thermal Initiator Thermal Initiator
70.0 995.28 :1: 130.20 939.12 :l: 111.47
72.0 1076.50 i 59.27 1051.48 i 89.64
74.0 1166.24 :1: 55.73 1121.40 i 88.30

 

95

Table 6.4. Effect of the inclusion of thermal initiator on ﬂexural modulus for samples
that were postcured for 6 hours.

 

 

Flexural Modulus F lexural Modulus
Wt.% Fused (kgp/mmz) (kgp’mmz)
Silica Without With
Thermal Initiator Thermal Initiator
70.0 1059.24 i 111.18 973.40 i 89.13
72.0 1160.11 1- 62.23 1139.05 1 123.83
74.0 1240.38 i 96.62 1206.61 i 116.55

 

Table 6.5. Heat released during DSC postbake and additional cure for photopolymerized
encapsulants subject to various cure cycles.

 

Sample A Sample B Sample C Sample D Sample E

 

Heat
Released 2°36 i 2.80 a: 1.22 2.41 s 0.58 1.02 :t 0.24 32337 i
. 0.97 11.91
(J/g resrn)
Additional 8.14 0.86 0.74 0.31 -
Cure (%)

 

Table 6.6. Effect of thermal initiator on the CTE below glass transition of the
photopolymerized encapsulants.

 

at (Hm/WC) at (Hm/WC)

 

Wt.% Fused Without With
Silica Thermal Thermal
Initiator Initiator
70.0 22.26 i 0.63 22.70 :1: 1.10
72.0 22.62 :i: 0.64 20.92 i 0.54
74.0 20.50 :l: 0.57 20.12 :t 1.41

 

96

Table 6.7. Effect of thermal initiator on the glass transition temperature of the
photopolymerized encapsulants.

 

Glass Transition Temperature Glass Transition Temperature

 

Wt.% Fused (°C) (°C)
Silica Without With
Thermal Initiator Thermal Initiator
70.0 148:2 150:2
72.0 150:4 151:1
74.0 152:3 150:1

 

Table 6.8. Effect of thermal initiator on thermal stress parameter of the
photopolymerized encapsulants.

 

 

0.0 0""
. (kn/mini) (kgrmmzl
W‘s/{’liigsed Without With
Thermal Thermal
Initiator Initiator
70.0 1.34 :l: 0.20 1.35 : 0.32
72.0 1.58 : 0.20 1.14 : 0.23
74.0 1.13:0.19 1.02:0.47

 

97

9.

10.

ll.

12.

6.5. REFERENCES

. V. Narayanan and AB. Scranton, Trends in Polym Sci, 5, 415, (1997).

V. Narayanan, K.K. Baikerikar, A.B. Scranton, RadTech ’98 North America U V/EB
Conference Proceedings, 3 1, (1998).

L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, New York, (1990).

H. Bayer and B. Lehner, ACS Symp Ser, 417, 412, (1990).

M.T. Goosey, Plastics for Electronics, Elsevier, New York, NY, (1985).

D.K. Shin and J.J. Lee, Advances in Electronic Packaging (ASME), 1, 253, (1997).
TH. Ho and CS. Wang, J Appl Polym Sci, 50, 477, (1993).

W.D. Cook and M. Johannson, J Biomed Mater Res, 21, 979, (1987).

AB. Scranton, C.N. Bowman, J. Klier, N.A. Peppas, Polymer, 33 (8), 1683, (1992).

RR. Tummala and E.J. Rymaszewski, editors. Microelectronics Packaging
Handbook, Van Nostrand Reinhold: New York, NY, (1989).

N. Kinjo, M. Ogata, K. Nishi, A. Kaneda, Adv Polym Sci, 88, 5, (1989).

Y. Nakamura, M. Yarnaguchi, A. Tanaka, M. Okubo, J Appl Polym Sci, 49, 331,
(1993).

13. LL. Lin, T.H. Ho, C.S. Wang, Polymer, 38, 1997, (1997).

98

Chapter 7

EVALUATION OF REDUCED IN-MOLD CURE TIMES FOR
PHOTOPOLYMERIZABLE LIQUID ENCAPSULANTS

7.1. INTRODUCTION

In conventional transfer molding processes used to encapsulate microelectronic
devices, the in-mold cure consumes the majority of the cycle time. For this reason,
reducing the in-mold cure times of molding compounds is the most efficient way to
reduce cycle times and increase productivity. For example, the in-mold cure time
consumes approximately 70% of the overall cycle time and an in-mold cure time
reduction of even 15 seconds can translate to a productivity increase of about 10%.l
With conventional systems, the cure time required before the parts can be ejected from
the mold varies from 2 to 3 minutes. However, future developments will require in-mold
cure times to be in the range of 60-90 seconds to achieve an overall cycle time of 2
minutes.I

Photopolymerizable liquid encapsulants (PLEs) are promising for reducing the in-
mold cure time compared to conventional thermal systems. As was shown in Chapter 6,
PLE samples containing a thermal initiator that were photocured for 120 seconds and
then postcured for 6 hours have the necessary material properties required by the
microelectronics industry. These results demonstrate that an in-mold cure time of 2
minutes is achievable using PLEs. However, since future productivity requirements
necessitate even shorter in-mold cure times, the objective of this chapter is to determine if

the UV illumination time (which corresponds to the in-mold cure time) for the PLEs can

99

be reduced even further (while still retaining the 6 hour postcure step at 170°C).
Speciﬁcally, in this chapter we will examine the material properties (ﬂexural strength and
modulus, coefﬁcients below and above glass transition, glass transition temperature, and
thermal stress parameter) of PLEs cured with UV illumination times of 60 and 90
seconds. In addition, the effects of both UV initiation (using only a photoinitiator) and

dual initiation (using both a photo- and thermal initiator) will be examined.

7.2. EXPERIMENTAL

7.2.1. Preparation of Photopolymerizable Liquid Encapsulant

The photopolymerizable liquid encapsulant is composed of a base resin,
photoinitiator, thermal initiator, fused silica ﬁller, and silane coupling agent. The base
resin used in these studies was DERAKANE 470-45 (Dow Chemical), an epoxy novolac-
based vinyl ester resin that was chosen primarily for its low initial viscosity (0.0456 Paos
or 45.6 CF at 30°C). The photoinitiator used was bis(2,4,6-trimethylbenzoyl)
phenylphosphine oxide (IRGACURE 819, Ciba) and was selected because of its effective
photobleaching in the wavelength range of 360-450 nm. The thermal initiator, benzoyl
peroxide (Aldrich), was also used in order to determine the effects of a dual initiation
scheme on the resulting material properties. The ﬁllers used were all crushed (angular),
untreated fused silica obtained from MINCO, Inc. Fused silica was selected because it is
the most commonly used ﬁller in transfer molding compounds and it possesses the
optimum combination of properties. Fused silica products with two different particle size
distributions were used: (i) MIN-SIL 40 (median particle size = 22.11 pm) and (ii) MIN-

SIL 550 (median particle size = 6.03 pm). All compositions used a 50:50 wt.% blend of

100

MIN-SIL 40 to MIN-SIL 550 for a particular ﬁller loading, as this was found to result in
a formulation with a low initial viscosity and improved processability. In addition, a
silane coupling agent, 3-methacryloxypropyl-trimethoxysilane (Z-6030, Dow Corning),
was used to provide a stable bond between the base resin and the fused silica ﬁller, as
well as, to improve the processability of the liquid encapsulant formulation.

Photocurable formulations were prepared by adding 0.2 wt.% (based on resin
weight) of IRGACURE 819 photoinitiator to DERAKANE 470-45 resin and stirring at
room temperature on a magnetic stir plate until the photoinitiator completely dissolved.
The silane coupling agent was then added dropwise to the desired 1 wt.% (based on ﬁller
weight) and the formulation was again mixed. Fused silica in the amount of 70.0, 72.0,
or 74.0 wt.% was added to the formulation and mixed by using a vortex mixer until the
ﬁller was completely dispersed and a homogeneous mixture was achieved. The
formulation was then degassed in a vacuum oven. Once these steps were completed, the
photocurable formulation was ready to use for preparing samples required for the
material property characterization studies.

To determine the effects of a dual initiation scheme, the thermal initiator, benzoyl
peroxide, was also added to the liquid formulation. In this case, benzoyl peroxide in the
amount of 1.2 wt.% (based on resin weight) was added (after the photoinitiator had
completely dissolved) and stirred until complete dissolution.

7.2.2. Flexural Modulus and Strength

Specimens for ﬂexural testing were prepared by photopolymerizing the liquid

encapsulant formulation in rectangular-shaped, silicone molds with unﬁltered UV light of

200 mW/cm2 UVA intensity from a 3000 W arcless mercury vapor lamp (Fusion UV

101

Systems, model F450T). Light intensities were measured using a UVICURE Plus high
energy UV integrating radiometer (over the 320-390 nm range). Specimens were 76.2
mm length x 12.7 mm width x 3.2 mm depth and were photocured for various times (60,
90, and 120 seconds). To fully develop the material properties, specimens were
postcured at 170°C in a laboratory oven for 6 hours after UV curing.

The ﬂexural strength and modulus of the photocured samples were determined
using a United SF M-20 instrument in accordance with the ASTM D 790 method. The
ﬂexural properties were measured using the three point ﬂexural test with a span length of
50.8 mm (2 inches), a 454 kg (1000 lb.) load cell, and a downdrive rate of 1.9 mm/min
(0.07 in/min). F lexural strength and modulus values were calculated using DATUM 97
software (United Testing Systems).

7.2.3. Coefﬁcient of Thermal Expansion and Glass Transition Temperature

Specimens for coefficient of thermal expansion (CTE) testing were cut from the
samples used for ﬂexural testing so that the thermal stress parameter of the photocured
encapsulants could be estimated. Dimensions of the cut-away samples were
approximately 11 mm length x 8 mm width x 3.2 mm thickness. All samples were cut
using a diamond wafering saw. The CTE below the glass transition (on) and CTE above
glass transition ((12) were both measured using a DuPont 943 therrnomechanical analyzer
(TMA) interfaced with a DuPont model 9900 thermal analyzer controller. The
photopolymerized samples were heated from room temperature (23°C) to 234°C at a
constant rate of 3°C/min. The change in the sample thickness during heating was

recorded in the personal computer and or; was obtained from the inclined line connecting

two points on the TMA curve, 50°C and 70°C. To determine a2, the inclined line

102

connecting 200°C and 230°C was used. The glass transition temperature was taken to be

the temperature at the intersection of these two lines.

7.3. RESULTS AND DISCUSSION

7.3.1. Flexural Modulus and Strength

Encapsulants must have excellent mechanical properties in order to withstand
mechanical shock, vibration, and handling during assembly. One of the more important
mechanical properties is the ﬂexural modulus, which characterizes the stiffness of a
material under an applied load (it is the slope of a plot of stress versus strain in the elastic
portion of the curve).2 By reducing the ﬂexural modulus, the thermal stress (also known
as the stress parameter) is also reduced, and recent research indicates that this leads to
improved device reliability. In particular, this stress causes interfacial cracking and
breakage of wirebond leading, as well as, passivation layer cracking and aluminum
pattern deforrnation.3'4 To reduce the thermal shrinkage stresses, a ﬂexural modulus near
or below 1000 kgp’mm2 is recommended.l

7.3.2. Effect of Reduced UV Illumination Times on Flexural Modulus and Strength

Table 7.1 shows the effect of reduced UV illumination times on ﬂexural modulus
for both UV-initiated and dual initiated samples containing 70.0, 72.0, and 74.0 wt.%
fused silica. Each entry in Table 7.1 represents the average of 5 samples and the range
corresponds to one standard deviation above and below the mean. All samples contained
1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on resin
weight), 1.2 wt.% benzoyl peroxide (based on resin weight), and the balance epoxy
novolac-based vinyl ester resin. Samples were photocured with UV light of 200 mW/cm2

UVA intensity and then postcured for 6 hours at 170°C in a laboratory oven. From Table

103

7.1, it can be seen that a reduction in UV illumination time followed by postcure has no
deleterious effect on the ﬂexural modulus of the PLEs. All of the PLEs (including those
cured for only 60 seconds) had sufﬁcient green strength to maintain their shape and
integrity when ejected from the mold. In addition, after postcure, all samples exhibited

2. For example, ﬂexural

ﬂexural modulus values near the desired value of 1000 kgr/mm
modulus values for samples ﬁlled with 70.0 wt.% fused silica ranged from 940 to 1060
kgp/mmz, those with 72.0 wt.% fused silica had modulus values between 1020 and 1160
kgg’mmz, and samples ﬁlled with 74.0 wt.% ﬁrsed silica had values between 1200 and
1260 kgg’mmz. In addition, in agreement with the results from Chapter 6, the inclusion of
the thermal initiator has no effect on the ﬂexural modulus.

Table 7.2 illustrates the effect of reduced UV illumination times on ﬂexural
strength for both UV-initiated and dual initiated samples containing 70.0, 72.0, and 74.0
wt.% fused silica. From Table 7.2, two ﬁndings are evident. First, in agreement with our
results ﬁ'om Chapter 6, the inclusion of the thermal initiator increases the ﬂexural
strength of the PLEs, typically by 1 to 2 kgp’mmz. The ﬂexural strength values for
samples with the thermal initiator, again, meet the requirements for molding compounds
(ﬂexural strength values of at least 9 kgp’mmz). Second, the ﬂexural strengths for UV
illumination times of 60 seconds and 90 seconds are essentially equivalent to the values
of 120 seconds. This is further evidence that the in-mold cure time of the PLEs can be

successfully reduced to 60 seconds.

7.3.3. Effect of Reduced UV Illumination Times on Coefﬁcient of Thermal
Expansion below Glass Transition

The linear coefﬁcient of thermal expansion (CTE) is deﬁned as the ratio of the

change in the length of the sample to the change in temperature per unit initial length.5

104

This parameter is very important in semiconductor encapsulation because a signiﬁcant
CTE mismatch between the encapsulant, leadframe, and the silicon chip can lead to the
build up of internal stresses in the semiconductor device, and could lead to cracking of
the chip or encapsulant. Current encapsulants exhibit or. values (CTE below glass
transition) in the range of 15-30 um/m°C, while copper leadframes possess a value of 16-
17 um/m°C and silicon exhibits a value of 3-4 ltrrl/m°C.6 The service temperature for
microelectronic encapsulants is typically between 75 and 85°C, which is well below the
glass transition temperature of the encapsulant.l For this reason, the value of the CTE

above the glass transition temperature, a2, is not as important as the CTE below the glass

transition temperature, 011.

Table 7.3 shows the effect of reduced UV illumination times on the coefﬁcient of
thermal expansion below glass transition for samples that were photocured with UV light
of 200 mW/cm2 UVA intensity and then postcured for 6 hours at 170°C. Again, each
entry in Table 7.3 represents the average of 5 samples and the range corresponds to one
standard deviation above and below the mean. All samples contained 1.0 wt.% silane
coupling agent (based on ﬁller weight), 0.2 wt.% BAPO (based on resin weight), 1.2
wt.% benzoyl peroxide (based on resin weight), and the balance epoxy novolac-based
vinyl ester resin. Samples were photocured with UV light of 200 mW/cm2 UVA
intensity and then postcured for 6 hours at 170°C in a laboratory oven. From Table 7.3,
again, there is no signiﬁcant difference in or; between specimens photopolymerized for
60 seconds (and then postcured for 6 hours) and specimens photopolymerized for 90 and

120 seconds (and then postcured for 6 hours). This is to be expected, since the reduction

105

in the coefﬁcient of thermal expansion occurs primarily in the postcure step. In addition,
the addition of the thermal initiator has little effect on on.

7.3.4. Effect of Reduced UV Illumination Times on Coefficient of Thermal
Expansion above Glass Transition

Table 7.4 shows the effect of reduced UV illumination times on the coefﬁcient of
thermal expansion above glass transition for samples that were photocured with UV light
of 200 mW/cm2 UVA intensity and then postcured for 6 hours at 170°C. The results are
similar to the results for the coefﬁcient of thermal expansion below glass transition.
Table 7.4 illustrates that the CTE above glass transition decreases with an increase in
fused silica content and is essentially independent of both the illumination time and the

inclusion of a thermal initiator.

7.3.5. Effect of Reduced UV Illumination Times on Glass Transition Temperature

Table 7.5 shows the effect of reduced UV illumination times on the glass
transition temperature (T!) for samples that were photocured with UV light of 200
mW/cm2 UVA intensity and then postcured for 6 hours at 170°C. The T8 values range
between 152 and 155°C for samples that were photocured for 60 seconds and postcured
for 6 hours, while the observed Tg values fell between 148 and 152°C for samples that
were photocured for 90 and 120 seconds (and then postcured for 6 hours). Therefore, as
in the previous chapter, the glass transition temperature for all of the PLE samples is

observed to be around 150°C.

7.3.6. Effect of Reduced UV Illumination Times on Thermal Stress Parameter

Table 7.6 shows the effect of reduced UV illumination times on the thermal stress

parameter (0.) for samples that were photocured with UV light of 200 mW/cm2 UVA

106

intensity and then postcured for 6 hours at 170°C. The thermal stress parameter, which is
calculated using Equation 6.1, depends upon the ﬂexural modulus, the coefﬁcient of
thermal expansion below glass transition, and the glass transition temperature. The
relatively high degree of uncertainty in the values of 0" shown in Table 7.6 arises from
the combined uncertainty of the three material properties used in the calculation. As the
value of 0' increases, the likelihood of thermally induced stresses also increases.
Therefore low values of o. are desired. Conventional microelectronic encapsulants
typically exhibit thermal stress parameters between 0.5 and 1.5 kgp’mmz. The data in
Table 7.6 illustrate that the PLEs in this study generally exhibit a. values ranging from
0.6 to 1.3 kgg’mmz, and are therefore within the range of values exhibited by current

encapsulants.

7.4. CONCLUSIONS

Current trends in microelectronic encapsulation will require in-mold cure times
for molding materials to be reduced from 120 seconds to 60-90 seconds. With current
thermally cured systems this reduced cure time is difficult to achieve. In this chapter, we
have shown that photopolymerizable liquid encapsulants (PLEs) based upon epoxy
novolac-based vinyl ester resins ﬁlled with fused silica can meet this requirement.
Photopolymerizable liquid encapsulants that were photocured for 60 and 90 seconds had
appropriate ejectabie hardness, and upon postcure, exhibited material properties (ﬂexural
strength and modulus, coefficient of thermal expansion, glass transition temperature, and

thermal stress parameter) that were equivalent or better than those of conventional

107

encapsulants. It was also shown that the addition of a thermal initiator does not

negatively affect any of the material properties, and leads to improved ﬂexural strength.

108

Table 7.1. Effect of reduced UV illumination times on ﬂexural modulus for both UV-
initiated and dual initiated samples containing 70.0, 72.0, and 74.0 wt.% fused silica.
Samples were photocured with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

F lexural Modulus F lexural Moglulus
UV Illumination o . . (kgp/mm ) (kgf/mm )
Time (seconds) Wt' /° Fused Sllrca Without With
Thermal Initiator Thermal Initiator
60 70.0 987.97 : 49.80 1059.77 : 131.59
60 72.0 1163.80 : 72.36 1118.11 : 81.59
60 74.0 1209.34 : 98.95 1247.32 : 66.39
90 70.0 945.18 : 44.39 966.90 : 59.53
90 72.0 1023.84 : 111.34 1119.90 : 157.61
90 74.0 1254.49 : 137.12 1234.05 : 108.82
120 70.0 1059.25 : 111.18 973.40 : 89.13
120 72.0 1160.11 : 62.23 1139.05 : 123.83
120 74.0 1240.38 : 96.62 1206.61 : 116.55

 

 

 

 

 

Table 7.2. Effect of reduced UV illumination times on ﬂexural strength for both UV-
initiated and dual initiated samples containing 70.0, 72.0, and 74.0 wt.% fused silica.
Samples were photocured with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

F lexural Strgngth F lexural Strgngth
UV Illumination o . . (kgp’mm ) (kgp’mm )
Time (seconds) Wt' /° Fused Srlrca Without With
Thermal Initiator Thermal Initiator
60 70.0 7.18 : 1.61 8.72 : 0.97
60 72.0 8.37 : 1.38 9.41 : 0.95
60 74.0 8.87 : 1.30 9.80 : 1.55
90 70.0 7.88 : 0.71 9.07 : 0.73
90 72.0 8.41 : 0.83 8.75 : 1.00
90 74.0 8.29:1.13 11.11 :1.09
120 70.0 6.71 : 0.69 8.83 : 1.68
120 72.0 7.59 : 0.24 9.28 : 0.95
120 74.0 8.12 : 0.69 9.92 : 1.44

 

 

 

 

109

 

Table 7.3. Effect of reduced UV illumination times on coefﬁcient of thermal expansion
below glass transition for both UV-initiated and dual initiated samples containing 70.0,
72.0, and 74.0 wt.% fused silica. Samples were photocured with UV light of 200
mW/cm2 UVA intensity and then postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

. . or; m/m°C ou( m°C
[ii/silligéﬂiii‘s’i Wt.% Fused Silica iii/lithom ) “With )
Thermal Initiator Thermal Initiator
60 70.0 22.10 : 1.57 20.84 : 1.35
60 72.0 19.80 : 0.87 20.20 : 1.43
60 74.0 20.02 : 1.27 19.36 : 1.23
90 70.0 22.04 : 0.71 22.30 : 1.26
90 72.0 19.40 : 1.21 20.20 : 0.51
90 74.0 18.58 : 0.45 19.44 : 1.20
120 70.0 22.26 : 0.63 22.70 : 1.10
120 72.0 22.62 : 0.64 20.92 : 0.54
120 74.0 20.50 : 0.57 20.12 : 1.41

 

 

 

 

 

Table 7.4. Effect of reduced UV illumination times on coefﬁcient of thermal expansion
above glass transition for both UV-initiated and dual initiated samples containing 70.0,
72.0, and 74.0 wt.% fused silica. Samples were photocured with UV light of 200
mW/cm2 UVA intensity and then postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

. . or; m°C) a2 m°C)
ﬁznlélggigz; Wt.% Fused Silica \(Nuinth/out. (“With .
Thermal Imtrator Thermal Imtrator

60 70.0 70.70 : 2.18 65.26 : 1.75
60 72.0 66.46 : 3.22 64.18 :t 1.35
60 74.0 62.74 :t 1.68 63.14 : 1.56
90 70.0 71.56 : 4.85 70.56 : 3.52
90 72.0 66.46 : 1.29 63.98 : 1.83
90 74.0 60.30 :l: 1.83 58.96 :i: 2.70
120 70.0 66.74 : 3.45 69.44 : 1.30
120 72.0 61.78 : 1.70 63.46 :1: 2.37
120 74.0 56.68 : 2.91 59.24 :l: 2.39

 

 

 

 

110

 

Table 7.5. Effect of reduced UV illumination times on glass transition temperature for
both UV-initiated and dual initiated samples containing 70.0, 72.0, and 74.0 wt.% fused
silica. Samples were photocured with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

. . T °C T °C
Iii/1111;123:2112? Wt.% Fused Silica Withoui 00th.).
Thermal Initiator Thermal Inrtrator
60 70.0 156:2 155 :2
60 72.0 153:1 154:2
60 74.0 153 : 2 153 :1
90 70.0 152 : 1 149 : 2
90 72.0 152:2 150:2
90 74.0 150:1 150:2
120 70.0 148:2 150:2
120 72.0 150:4 151:1
120 74.0 152:3 150:1

 

 

 

 

 

Table 7.6. Effect of reduced UV illumination times on thermal stress parameter (0“) for
both UV-initiated and dual initiated samples containing 70.0, 72.0, and 74.0 wt.% fused
silica. Samples were photocured with UV light of 200 mW/cm2 UVA intensity and then
postcured for 6 hours at 170°C.

 

 

 

 

 

 

 

 

 

 

 

 

 

. . o. f/mm2 a. k p’mrnf
[ii/1111;121:2382? Wt.% Fused Silica “(hihout ) (With. . )
Thermal Inrtrator Thermal Inrtrator
60 70.0 1.27 : 0.37 1.06 : 0.38
60 72.0 0.88 : 0.21 0.95 : 0.36
60 74.0 0.99 : 0.39 0.84 : 0.36
90 70.0 1.18 : 0.18 1.24 : 0.24
90 72.0 0.68 :l: 0.23 0.95 : 0.23
90 74.0 0.62 : 0.17 0.84 : 0.34
120 70.0 1.34 : 0.20 1.35 : 0.32
120 72.0 1.58 :l: 0.20 1.14 : 0.23
120 74.0 1.13:0.19 1.05:0.38

 

 

 

 

111

 

9

7.5. REFERENCES

. L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand

Reinhold, New York, (1990).

M.T. Goosey, Plastics for Electronics, Elsevier, New York, NY, (1985).
D.K. Shin and J .J . Lee, Advances in Electronic Packaging (ASME), 1, 253, (1997).
TH. Ho and CS. Wang, J Appl Polym Sci, 50, 477, (1993).

RR. Tummala and E.J. Rymaszewski, editors. Microelectronics Packaging
Handbook, Van Nostrand Reinhold: New York, NY, (1989).

N. Kinjo, M. Ogata, K. Nishi, A. Kaneda, Adv Polym Sci, 88, 5, (1989).

112

Chapter 8

THERMAL PROPERTIES OF PHOTOPOLYMERIZABLE
LIQUID ENCAPSULANTS

8.1. INTRODUCTION

Thermal management plays a vital role in the packaging of high performance
electronic devices. Speciﬁcally, it is important that the silicon chip does not overheat
during operation. Since the device generates a signiﬁcant amount of heat during use, the
thermal conductivity of an encapsulating compound plays an important role in the
thermal management of the device. Indeed, the thermal conductivity determines the
amount of heat that can be removed from the device and thus affects operating
temperatures of electronic components. As semiconductor devices increase in transistor
densities, the heat dissipation requirements also increase, resulting in the need for
packaging materials with enhanced thermal conductivity.

In order to attain high thermal conductivities, a high weight fraction of ﬁller must
be used in the molding compound. Typically, large chip, small package devices are
transfer molded using a polymer composite consisting of 20-30 wt.% (30-45 vol.%)
epoxy resin and 70-80 wt.% (55-70 vol.%) ﬁrsed silica ﬁller. Although the polymer
portion of the composite provides excellent electrical insulation and environmental
protection, it does not offer adequate thermal conductivity or dimensional stability. As a
result, ceramic ﬁllers such as fused and crystalline silica, aluminum nitride, beryllium
oxide, and silicon carbide are used to provide improvements in thermal conductivity and
dimensional stability. Fused silica is the most widely used ﬁller in epoxy molding

compounds for microelectronic encapsulation because the resulting encapsulants exhibit

113

excellent performance in most key areas, including dimensional stability (a low
coefﬁcient of thermal expansion), high moisture resistance, and superior stress and
electrical characteristics.1 Fillers such as crystalline silica, aluminum nitride, boron
nitride, and silicon carbide do lead to encapsulants that exhibit higher thermal
conductivities. However, due to its overall performance advantages, fused silica remains
the predominant ﬁller in epoxy molding compounds and was, therefore, the ﬁller used in
this study.

The goal of these experiments is simply to determine the thermal conductivity of
the photopolymerizable liquid encapsulants as a function of fused silica loading in order
to conﬁrm that the thermal conductivity values are in the range necessary for
microelectronic encapsulants. Furthermore, the experimental results were compared with

the model of Agari and Uno2 for predicting the thermal conductivity of a composite.

8.2. EXPERIMENTAL

8.2.1. Preparation of Photopolymerizable Liquid Encapsulant

The photopolymerizable liquid encapsulant (PLE) used for the thermal
conductivity experiments is comprised of a base resin, fused silica ﬁller, photoinitiator,
thermal initiator, and silane coupling agent. The base resin was an epoxy novolac-based
vinyl ester resin (DERAKANE 470-45, Dow Chemical) that was chosen for its low initial
viscosity (0.0456 Pats or 45.6 CF at 30°C), as well as, its appropriate thermal and
mechanical properties upon cure. The photoinitiator used was bis(2,4,6-
trimethylbenzoyl) phenylphosphine oxide (IRGACURE 819, Ciba). This photoinitiator,
which will henceforth be denoted as BAPO, has been shown to be especially appropriate

for curing relatively thick polymers and composites due to its efficient photobleaching.

114

The thermal initiator used was benzoyl peroxide (Aldrich). The ﬁllers used were crushed
(angular), untreated fused silica (MIN-SIL 40 and MIN-SIL 550) obtained from MINCO,
Inc. A silane coupling agent, 3-methacryloxypropyltrimethoxysilane (Z-6030, Dow
Corning), was used to provide a stable bond between the base resin and the fused silica
ﬁller, and to improve the processability of the liquid encapsulant formulation.

The PLEs were prepared by adding 0.2 wt.% BAPO photoinitiator (based on resin
weight) to the epoxy novolac-based vinyl ester resin and stirring at room temperature on
a magnetic stir plate until the photoinitiator completely dissolved. The thermal initiator,
benzoyl peroxide, was also added in the amount of 1.2 wt.% (based on resin weight) and,
again the system was stirred until dissolution was complete. The silane coupling agent
was then added dropwise to the desired 1 wt.% (based on ﬁller weight) and the resin
mixture was stirred on the magnetic stir plate. The fused silica (a 50:50 wt.% blend of
MIN-SIL 40 and MIN-SIL 550) was added in the amount of 70.0, 72.0, or 74.0 wt.% and
the system was mixed by using a vortex mixer until the ﬁller was completely dispersed
and a homogeneous mixture was achieved. The PLE was then degassed in a vacuum
oven to remove air bubbles. Once these steps were completed, the PLE was ready to use
for preparing specimens required for the thermal conductivity studies.

8.2.2. Specimen Preparation for Thermal Conductivity Testing

Specimens were prepared by photopolymerizing the liquid encapsulant in
rectangular silicone molds with unﬁltered UV light (200 mW/cm2 UVA intensity) for 90
seconds from a 3000 W arcless mercury vapor lamp (Fusion UV Systems, model F450T).
Light intensities were measured (over the 320-390 nm range) using a UVICURE Plus

high energy UV integrating radiometer. Following UV illumination, the specimens were

115

postcured in a laboratory oven for 6 hours at 170°C. The ﬁnal dimensions of the cured
specimens were 76.2 mm length, 25.4 mm width, and 3.2 m depth. The cured
specimens were then machined into disks with a diameter of 12.7 mm and a thickness of
1.0 mm for thermal conductivity testing.

8.2.3. Laser Flash Method

Thermal conductivity measurements are difficult and time-consuming, and
require relatively large specimens. However, thermal conductivity may be calculated
from measurements of thermal diffusivity, speciﬁc heat, and bulk density using Equation
8.1.

k = apCp (8.1)
Here, k is the thermal conductivity, 0: is the thermal diffusivity, p is the bulk density, and
Cp is the speciﬁc heat.

The laser ﬂash method (ASTM E1461-92) for determining thermal conductivity
has been applied to a wide range of materials, including composites, plastics, ceramics,
metals, glasses, crystals, and foams. One advantage of this technique is that it requires
only a small amount of material. In this technique, the measurement of the thermal
diffusivity of a material is carried out by rapidly heating one side of a ﬂat disk sample
and measuring the temperature as a function of time (the temperature rise curve) on the
opposite side. Based upon this temperature proﬁle, the through-plane diffusivity is
measured, and the thermal conductivity is calculated using the known values of the
speciﬁc heat and bulk density.

All measurements to determine the thermal conductivity by the laser ﬂash method

were performed by Holometrix, Inc. (Bedford, MA) using the Holometrix Therrnaﬂash

116

2200 Laser Flash system and in accordance with ASTM E1461-92. In this system, the
sample disk made from the PLE was aligned between a neodymium glass laser (1.06 pm
wavelength) and an indium antimonide (InSb) IR detector in a tantalum tube furnace. In
order to prevent penetration of the laser beam into the sample, the PLE samples were

coated with a 0.1 pm gold ﬁlm. In addition, because the PLE samples do not have a high

value of emissivity or absorptivity, the samples were also coated with a 5 pm graphite
ﬁlm before testing. The graphite ﬁlm increases the energy absorbed on the laser side and
increases the temperature signal on the rear face of the sample. A Type C thermocouple
in contact with the PLE sample was used to control the temperature of the sample and its
surroundings between 20 and 2000°C. Once the sample was stabilized at 25°C, the laser
was ﬁred several times over a span of a few minutes and the necessary data was recorded
for each laser shot. The laser beam energy strikes and is absorbed by the front surface of
the sample, causing a heat pulse to travel through the thickness of the sample. The
resulting sample temperature rise is fairly small (ranging from about 0.5 to 2°C) and is
kept in the optimal range by adjustable ﬁlters between the laser and the furnace. A lens
focuses the back surface image of the sample onto the IR detector and the temperature
rise signal vs. time is ampliﬁed and recorded with a high speed A/D converter. The
thermal diffusivity is determined from a numerical analysis of the 1R detector output. In
addition, the speciﬁc heat can be measured with the laser ﬂash method by comparing the
temperature rise of a reference sample of known speciﬁc heat tested under the same
conditions. The instrument is fully automated to control all systems and record, analyze,

and report the thermal difﬁrsivity, speciﬁc heat, and the calculated thermal conductivity.

117

8.3. RESULTS AND DISCUSSION

8.3.1. Thermal Conductivity

Experimental data showing the thermal conductivity of the PLEs as a function of
fused silica loading are shown in Figure 8.1. All samples contained 70.0 to 74.0 wt.%
fused silica, 1.0 wt.% silane coupling agent (based on ﬁller weight), 0.2 wt.% BAPO
(based on resin weight), 1.2 wt.% benzoyl peroxide (based on resin weight), and the
balance epoxy novolac-based vinyl ester resin. Samples were photocured for 90 seconds
with UV light of 200 mW/cm2 UVA intensity and then postcured for 6 hours at 170°C.
Each bar in the bar chart represents the average of 3 samples and the error bars indicate
one standard deviation above and below the mean. As expected, the thermal conductivity
of the PLEs increases as the fused silica content is increased, and the values range
between 0.56 and 0.62 W/m-K. Clearly, the thermal conductivity of a molding
compound is largely dependent on the ﬁller used and, therefore, the thermal conductivity
values of the PLEs ﬁlled with fused silica (k = 1.5 W/m-K) will not be as high as molding
compounds loaded with higher thermal conductivity ﬁllers, such as silicon carbide (k =
85 W/moK), aluminum nitride (k = 150-220 W/moK), and boron nitride (k = 250-300
W/moK). However, the thermal conductivity values of the PLEs are all within the 0.5 to
0.7 W/moK range of epoxy molding compounds ﬁlled with fused silica. As a result, the
PLEs are more suitable for applications with low stress requirements, rather than devices
with high heat outputs.

In order to predict the thermal conductivity of encapsulants at higher ﬁller
loadings, theoretical and empirical models are often used. Although other additives are

typically present in encapsulants, such as initiators and coupling agents, the concentration

118

added is rather small when compared to that of the ﬁllers. Thus, from a thermal transport
point of view, the encapsulant behaves like a two-phase composite.3 Although there are
several second order models used to predict the thermal conductivity of a two-phase
composite (Agari and Unoz, Hatta and Taya“, Hashin and Shtrikmans, and Nielsen"), the
model developed by Agari and Uno describes the thermal conductivity of encapsulants
quite well.7'8 Indeed, the model is especially suited for composites with high ﬁller
loadings (30 to 60 vol.%) and irregularly shaped ﬁller particles, such as fused silica. The
model is described by Equation 8.2,
log k, = (11C2 log(kf) + (1 — ¢) log(C,k,,,) (8.2)

and can be rearranged to give Equation 8.3,

k, = kf’ [(C,k,,)"“’] (8.3)
where kc, kf, and km are the thermal conductivities of the composite, ﬁller, and resin
matrix, respectively, and 45 is the volume fraction of the ﬁller. In addition, C1 is a
constant used to measure the effect of ﬁller particles on the secondary structure of the
resin matrix and C2 is a constant used to measure the ease with which the ﬁller particles
begin to form conductive chains.2 For the purpose of modeling the experimental thermal
conductivity data, reported literature values for fused silica (k; = 1.5 W/moK) and for
epoxy novolac-based vinyl ester resin (km = 0.182 W/m-K) were used. The fused silica
volume fractions, (11, were calculated using Equation 8.4,

W

¢= (8.4)
W+(1-W)&

m

 

119

where W is the weight fraction of fused silica, p,- is the density of the ﬁller (,0; = 2.20

g/cm3 for fused silica), and pm is the density of the resin matrix (p,,, = 1.048 g/cm3 for
epoxy novolac-based vinyl ester resin). Table 8.1 shows the calculated volume fractions

corresponding to the 70.0, 72.0, and 74.0 wt.% fused silica loadings.

Table 8.1. Calculated volume fractions corresponding to 70.0, 72.0, and 74.0 wt.% fused
silica loadings.

 

 

 

 

Weight Fraction Fused Silica W Volume Fraction Fused Silica (ii
0.70 0.526
0.72 0.551
0.74 0.576

 

 

 

 

Once the volume fractions were determined, the experimental thermal conductivity data
was ﬁt to Equation 8.3 by nonlinear regression using C; and C2 as the ﬁtting parameters.
The nonlinear regression analysis was performed using the commercial software package
Origin 5.0 (Microcal Software, Inc.).

Figure 8.2 compares the experimental thermal conductivity results to the
predicted values obtained using the model of Agari and Uno. The curve drawn in Figure
8.2 corresponds to values obtained by ﬁtting the experimental data to Equation 8.3, from
0 to 80.0 wt.% fused silica (0 to 65.6 vol.%). The best ﬁt values for C1 and C2 were
0.997 and 1.064, respectively. These values are quite reasonable since C. values reported
by Agari and Uno for various composite systems are approximately 1 and C2 values
range from 1 to 1.5.9 Since C1 is nearly 1, the secondary structure of the epoxy novolac-
based vinyl ester resin is largely unaffected by the fused silica particles. Figure 8.2

demonstrates that the experimental thermal conductivity data is well correlated by the

120

model of Agari and Uno. The model can be used to effectively predict the thermal
conductivity of the PLEs at both lower and higher ﬁller loadings.

Figure 8.3 provides a “close-up” comparison of the experimental data to the
model of Agari and Uno between 60.0 and 80.0 wt.% fused silica (41.7 and 65.6 vol.%).
This range of fused silica loadings was chosen because most encapsulants are ﬁlled with
60.0 to 80.0 wt.% fused silica. Looking at Figure 8.3, the model predicts that the thermal
conductivity of the PLEs will be 0.44 W/moK at 60.0 wt.% fused silica and 0.74 W/m.K
at 80.0 wt.% fused silica. Therefore, PLEs ﬁlled with 60.0 wt.% ﬁrsed silica would not
have an appropriate value (since the values do not fall within the 0.5 to 0.7 W/moK range
of epoxy molding compounds ﬁlled with fused silica). Figure 8.3 shows that the PLEs
must be ﬁlled with at least 66.0 wt.% fused silica in order have thermal conductivity
values within this range.

8.3.2. Speciﬁc Heat

Speciﬁc heat is a measure of the quantity of heat a material must absorb to raise
its temperature a speciﬁed increment.lo Figure 8.4 shows the effect of ﬁrsed silica
loading on the speciﬁc heat of the PLE samples. Each bar in the bar chart represents the
average of 3 samples and the error bars indicate one standard deviation above and below
the mean. The speciﬁc heat of the unﬁlled epoxy novolac-based vinyl ester resins is
approximately 1.2 J/g-K as reported by the manufacturer. As shown in Figure 8.4, the
speciﬁc heat for the PLEs ranges between 0.83 and 0.87 J/g-K. In agreement with the
results of Bujard et al. ' ', the speciﬁc heat decreases when the weight content of the fused

silica increases. In addition, the values of the speciﬁc heat for the PLEs ﬁlled with 70.0

121

wt.% fused silica are very close to the values reported by Bujard et al. for epoxy molding
compounds ﬁlled with 70.0 wt.% fused silica.
8.3.3. Thermal Diffusivity

The thermal diffusivity indicates how rapidly heat can diffuse through the
molding compound.ll The higher the thermal diffusivity of a substance, the higher the
rate of temperature propagation. That is, materials with high thermal diffusivities will
respond quickly to changes in their thermal environment, while materials with low
thermal diffusivities will respond more slowly, taking longer to reach a new equilibritun
condition.l2 Figure 8.5 shows the effect of fused silica loading on the thermal diffusivity
of the PLEs. Each bar in the bar chart represents the average of 3 samples and the error
bars indicate one standard deviation above and below the mean. The results show that
the thermal diffusivity increases as the fused silica content increases, and that all of the

values range between 3.7 x 10'3 and 4.2 x 10'3 cmz/s.

8.4. CONCLUSIONS

In this chapter, the thermal conductivity, speciﬁc heat, and thermal diffusivity of
photopolymerizable liquid encapsulants (PLEs) were measured as a function of fused
silica loading by the laser ﬂash method. Photopolymerizable liquid encapsulants loaded
with 70.0, 72.0, and 74.0 wt.% fused silica exhibited thermal conductivities between 0.56
and 0.62 W/moK, which is within the 0.5 to 0.7 W/moK range exhibited by most fused
silica loaded molding compounds. The thermal conductivity values of the PLEs were
approximately three times that of the unﬁlled epoxy novolac-based vinyl ester resin (k =

0.182 W/mcK).

122

The experimental thermal conductivity data was ﬁt to the second order model of
Agari and Uno in order to predict the thermal conductivity of the PLEs at ﬁller loadings
lower than 70.0 wt.% and ﬁller loadings higher than 74.0 wt.%. Indeed, the experimental
thermal conductivity data was well correlated by the model of Agari and Uno. This
model can be used to accurately predict the thermal conductivity of the PLEs over a wide
range of ﬁller loadings.

The speciﬁc heat for the PLEs ranged between 0.83 and 0.87 J/goK and was
shown to decrease with increasing fused silica content, in agreement with the results of
other authors.ll In addition, values for the thermal diffusivity ranged between 3.7 x 10'3

and 4.2 x 10'3 cmZ/s and were shown to increase with increasing ﬁrsed silica content.

123

8.5. NOMENCLATURE

Cp = speciﬁc heat

C I = constant used to measure the effect of the ﬁller on the secondary structure of resin

C2 = constant used to measure the ease with which ﬁller particles begin to form
conductive chains

k = thermal conductivity

kc = thermal conductivity of composite

kt = thermal conductivity of ﬁller

km = thermal conductivity of resin matrix

W = weight fraction of ﬁller

a = thermal diffusivity

¢= volume fraction ﬁller

p = density

pt= density of ﬁller

,0... = density of resin matrix

124

 

0.63

0.62 +

 

 

 

 

0.61

 

 

P
at

 

 

0.59

 

 

 

0.58

 

 

 

0.57

 

 

 

 

 

0.56

 

 

 

 

 

0.55

 

 

 

 

0.54

 

 

 

 

0.53

 

Thermal Conductivity (Wlm K)

 

 

 

 

0.52

 

 

 

 

l0.51

 

 

 

 

 

 

 

 

70 72 74
Wt.% Fused Silica

Figure 8.1. Thermal conductivity of PLEs at 25°C as a function of fused silica loading.

125

 

 

P
on

 

0 Experimental results

 

.°

N
J
I

 

 

 

—Agari and Uno model prediction /

/
/

/
/

o I I T I
0 20 40 60 80 100

Wt.% Fused Silica

Figure 8.2. Comparison of experimental thermal conductivity results to the predicted
values obtained using the model of Agari and Uno from 0 to 80.0 wt.% fused silica (0 to

65.6 vol.%).

 

P
a:

 

P
or

 

P
a

 

.°
00

 

Thermal Conductivity (Wlm K)
O
to

 

P
.s

 

 

 

126

 

0.75

 

 

 

0 Experimental results ’
—Agari and Uno model prediction

 

 

O
a, C
(II N

 

 

\

0.55 /
0.45 ,/

004 I I T T I T T I I
60 62 64 66 68 70 72 74 76 78 80

Wt.% Fused Silica

Figure 8.3. Comparison of experimental thermal conductivity results to the predicted
values obtained using the model of Agari and Uno from 60.0 to 80.0 wt.% fused silica
(41.7 to 65.6 vol.%).

 

 

\

Thermal Conductivity (Wlm K)

 

 

 

 

127

0.9
0.89
0.88
0.87
0.86
0.85
0.84
0.83
0.82
0.81

0.8
0.79
0.78
0.77
0.76
0.75

Specific Heat 0,, (Jlg K)

 

70 72 74
Wt.% Fused Silica

Figure 8.4. Speciﬁc heat of PLEs at 25°C as a function of fused silica loading.

128

0.0044
0.0043
0.0042
0.0041

0.004
0.0039
0.0038
0.0037
0.0036
0.0035
0.0034
0.0033
0.0032
0.0031

0.003

Thermal Diffusivity (cm’ls)

 

70 72 74
Wt.% Fused Silica

Figure 8.5. Thermal diffusivity of PLEs at 25°C as a function of fused silica loading.

129

8.6. REFERENCES
1. P. Procter and J. Solc, IEEE T Compon Hybr, 14, 4, 708, (1991).
2. Y. Agari and T. Uno, J Appl Polym Sci, 32, 5705, (1986).
3. M. Michael and L. Nguyen, Inter-Society Conf Thermal Phenom, 246, (1992).
4. H. Hatta and M. Taya, J Appl Phys, 58, 7, 2478, (1985).
5. Z. Hashin and S. Shtrikman, J Appl Phys, 33, 1514, (1962).

6. LE. Nielsen, Ind Engr Chem Fund, l3, 17, (1974).

\l

. D.M. Bigg, Adv Polym Sci, 119, l, (1995).
8. GP. Wong and RS. Bollampally, IEEE Trans Adv Pack, 22, 1, 54, (1999).
9. Y. Agari, A. Ueda, M. Tanaka, and S. Nagai, J Appl Polym Sci, 40, 929, (1990).

10. L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, New York, (1990).

11. P. Bujard, G. Kuhnlein, S. Ino, and T. Shiobara, IEEE Trans CPMT, Pt A, 17, 4, 527,
(1994).

12. RP. Incropera and DP. DeWitt, Introduction to Heat Transfer, 2’” ed., John Wiley &
Sons, New York, (1990).

130

Chapter 9

ADHESION BETWEEN PHOTOPOLYMERIZABLE LIQUID
ENCAPSULANTS AND ALLOY 194

9.1. INTRODUCTION

The adhesion of an epoxy molding compound to the leadfrarne is a critical
parameter for encapsulated microelectronic devices. The on-going trends of increasing
silicon chip size and thinner packages, as well as the push towards surface mount
operations, have increased the importance of proper adhesion between the molding
compound and leadfrarne."2 Speciﬁcally, improved adhesion between the molding
compound and the leadfrarne leads to enhanced reliability and resistance to package and
chip cracking. In addition, inadequate adhesion leads to popcorn problems in surface
mount devices, as well as delarnination and subsequent thermomechanical failure.2'3
Furthermore, a high degree of adhesion limits moisture-induced corrosion by preventing
penetration of water along the leadfrarne.4

A number of non-destructive and destructive experimental techniques have been
used to characterize adhesion between molding compounds and the leadfrarne. While
non-destructive tests, such as Scanning Acoustic Microscopy (SAM), are applied
primarily for qualitative inspections (to determine whether adhesion exists or not),
destructive tests can provide valuable quantitative information about the adhesion
strength.4 The most widely used destructive tests to measure adhesive strength between
molding compounds and the leadfrarne are the shear (lap and button), tab pull, and peel
test methods.5 Of the many destructive tests available, Kim reported that the 180 degree

peel test is the most sensitive, and provides an accurate measure of adhesion between the

131

2 Therefore, in this chapter, the degree of adhesion

encapsulants and the leadfrarne.
between the photopolymerizable liquid encapsulants (PLEs) and the leadframe metal was
measured by the 180 degree peel test.

The most widely used leadfrarne metals for integrated circuit assembly are
copper-based alloys (e.g. Alloys 151, 190, and 194) and an iron-nickel alloy (Alloy 42).
The advantages of copper-based alloys are that they have higher thermal and electrical
conductivity and lower cost compared to Alloy 42. However, molding compounds
generally do not adhere to copper alloys as well as they adhere to other alloys. Copper
and copper-based alloys are difficult metals to bond to for three primary reasons: (i) the
catalytic activity of its surface oxide which alters adhesive curing conversion, (ii) the
activity of copper in the presence of oxidizing and reducing agents in the environment,
and (iii) the poor intrinsic adhesion that is linked to the oxide layer at the oxide/metal
interface.6 Because they represent the most challenging leadfrarne metals for encapsulant
adhesion, a copper alloy was chosen for the studies in this chapter. Speciﬁcally, the
adhesion of the PLEs to Alloy 194 was evaluated using the 180 degree peel test and the
fracture surfaces were characterized using environmental scanning electron microscopy
(ESBM).

Signiﬁcant research has been devoted to improving the degree of adhesion
between copper leadframes and epoxy molding compounds by using adhesion promoters,
metal plating, vacuum deposition, and chemical oxidation.7 Previous research by Song et
al. has demonstrated that both silane coupling agents and azole compounds can improve

peel strength!”9 Therefore, two commercial silane coupling agents and two azole

compounds were evaluated to determine the effect of various surface pretreatrnents on the

132

peel strength. Silane coupling agents improve adhesion by forming covalent linkages
between the metal surface and the polymeric encapsulant. Azole compounds are used as
corrosion inhibitors to improve the bond durability between epoxy resins and copper.8
These compounds have active protons to form a complex with the cuprous hydroxide on
the copper surface and functional groups such as amine and carboxylic acid to interact
with epoxy resins. '0
9.2. EXPERIMENTAL

9.2.1. Preparation of Photopolymerizable Liquid Encapsulant

The photopolymerizable liquid encapsulant (PLE) used for the adhesion
experiments is comprised of a base resin, fused silica ﬁller, photoinitiator, thermal
initiator, and silane coupling agent. The base resin was an epoxy novolac-based vinyl
ester resin (DERAKANE 470-45, Dow Chemical) that was chosen for its low initial
viscosity (40-80 cP at 25 °C) as well as its appropriate thermal and mechanical properties
upon cure. The photoinitiator used was bis(2,4,6-trimethylbenzoyl) phenylphosphine
oxide (IRGACURE 819, Ciba). This photoinitiator, which will henceforth be denoted as
BAPO, has been shown to be especially appropriate for curing relatively thick polymers
and composites due to its efficient photobleaching. The thermal initiator used was
benzoyl peroxide (Aldrich). The ﬁllers used were crushed (angular), untreated fused
silica (MIN-SIL 40 and MIN-SIL 550) obtained from MINCO, Inc. A silane coupling
agent, 3-methacryloxypropyltrimethoxysilane (Z-6030, Dow Corning), was used to
provide a stable bond between the base resin and the fused silica ﬁller, and to improve

the processability of the liquid encapsulant formulation.

133

The PLEs were prepared by adding 0.2 wt.% BAPO photoinitiator (based on resin
weight) to the epoxy novolac-based vinyl ester resin and stirring at room temperature on
a magnetic stir plate until the photoinitiator completely dissolved. The thermal initiator,
benzoyl peroxide, was also added in the amount of 1.2 wt.% (based on resin weight) and,
again the system was stirred until dissolution was complete. The silane coupling agent
was then added dropwise to the desired 1 wt.% (based on ﬁller weight) and the resin
mixture was stirred on the magnetic stir plate. The fused silica (a 50:50 wt.% blend of
MIN-SIL 40 and MIN-SIL 550) was added in the amount of 70.0, 72.0, or 74.0 wt.% and
the system was mixed by using a vortex mixer until the ﬁller was completely dispersed
and a homogeneous mixture was achieved. The PLE was then degassed in a vacuum
oven to remove any air bubbles. Once these steps were completed, the PLE was ready to
use for preparing peel test specimens.

9.2.2. Solvent Degreasing Pretreatment of Alloy 194

Alloy 194 was selected as the leadframe metal for the 180 degree peel test
because of its widespread use in microelectronic packages. It is a copper-based
leadfrarne metal, comprised of 97.5% copper, 2.35% iron, 0.12% zinc, and 0.03%
phosphorus.ll Thin foil sheets of Alloy 194 (C194, Olin Brass) with thickness 0.0127
mm (0.0005 in.) were cut into rectangular strips of dimensions 215.9 mm (8.5 in.) length
and 12.7 mm (0.5 in.) width. The thin foil was necessary for the peel test in order to
permit a turn back at an approximate 180° angle in the expected loading range of the test
without failure. Prior to applying the PLE to the leadframe metal, the copper alloy was
pretreated in order to provide a surface to which the resin can adhere. Typically, metals

available for industrial applications are covered with a contaminant layer that has

134

properties that differ from the bulk metal and, consequently, this alters the degree of
adhesion between the metal and polymer. The contaminant layer may consist of
processing lubricants and oils, water, or other contaminants from the atmosphere which
can adsorb on the high energy metal surfaces.'2 A solvent degreasing pretreatment is
typically the minimum pretreatment used to remove this contaminant layer.

The Alloy 194 strips were cleaned and degreased by ﬁrst dipping them in 0.1 M
HCl for 30 seconds in order to remove the weak copper oxide, then rinsing with a series
of solvents in the following order: distilled water; methanol; distilled water; acetone. The
strips were then dried gently with compressed air. Environmental scarming electron
microscopy (ESEM) was conducted to determine the surface characteristics of the foils.
The microscope was a Philips Electroscan 2020 environmental scanning electron
microscope (ESEM) equipped with a Lab6 ﬁlament. The accelerating voltage was 20 kV
and the operating pressure ranged between 2-3 Torr. Scanning electron micrographs of
leadfrarne strips before and after pretreatment are shown in Figures 9.1 and 9.2,
respectively. Figure 9.1 shows the dirt and oil (dark-colored circles) on the “as-received”
Alloy 194 strips, conﬁrming that a pretreatment step is necessary in order to clean the
surface and adequately prepare it for adhesion. Figure 9.2 shows a much cleaner surface
after the Alloy 194 strips were pretreated, demonstrating that the pretreatment was
successful in removing the surface contaminants. Following pretreatment, the Alloy 194
strips were primed with silane coupling agents or azole compounds, as described below.
9.2.3. Treatment of Alloy 194 Strips with Silane Coupling Agents

Two different silane coupling agents were used for priming the copper surfaces,

3-methacryloxypropyltrimethoxysilane (Z-6030, Dow Coming) and a silane composed of

I35

an amino-alkyltrimethoxysilane and a polyol diluted with methanol (Z—6026, Dow
Corning). These coupling agents, Z-6030 and Z-6026, will henceforth be denoted as
MPMS and AAMS, respectively. Coupling agent solutions using MPMS were made to a
concentration of 5.0 wt.% in methanol. Alloy 194 strips were dipped in the solution for 1
minute, removed, blown gently with air, and cured at 115°C for 30 minutes.

Coupling agent solutions using AAMS were made to a concentration of 0.5 wt.%
in water. The pH was then adjusted to 4.5 with acetic acid. Alloy 194 strips were dipped
in the solution for 1 minute, removed, gently blown with air, and cured at 115°C for 30
minutes.

9.2.4. Treatment of Alloy 194 Strips with Azole Compounds

The two azole compounds used were benzotriazole-S-carboxylie acid (Aldrich
Chemical) and polybenzimidazole (Aldrich Chemical), and will henceforth be referred to
as BTCA and PBI. The primer solutions using the azole compounds were prepared
according to the procedure speciﬁed by Song et al.8 Solutions using BTCA were made to
a concentration of 3 x 10'3 M in distilled water. Alloy 194 strips were dipped in the
solution for 1 minute, removed, gently blown with air, and cured at 150°C for 30
minutes. Solutions using PBI were made to a concentration of 0.1 wt.% in dimethyl
formarnide (EM Science). Alloy 194 strips were dipped in the solution for 1 minute,

removed, gently blown with air, and cured at 150°C for 30 minutes.

9.2.5. Specimen Preparation for Peel Test
Following pretreatment, the Alloy 194 strips were inserted into rectangular

cavities of a Teﬂon mold, with cavity dimensions of 101.6 mm (4 in.) length, 12.7 mm

136

 

(0.5 in.) width, and 3.5 mm (0.1378 in.) depth. The PLE was then injected, using a
syringe, into the mold cavities and allowed to completely wet the metal strips. The
specimens were photopolymerized for 90 seconds with unﬁltered UV light of 200
mW/cm2 UVA intensity from a 3000 W arcless mercury vapor lamp (Fusion UV
Systems, model F450T) to create a bond length of 101.6 mm (4 in.) between the metal
and photocured encapsulant. Light intensities were measured using a UVICURE Plus
high energy UV integrating radiometer (over the 320-390 nm range). After photocuring,
the samples were allowed to cool to room temperature for 24 hours prior to the peel test.
The peel test was performed by peeling off the thin leadfrarne foil at a 180° angle from
the photocured encapsulant with a United SF M-20 instrument at a rate of 127 mm/min (5
in./min) and a 9.07 kg (20 lb.) load cell. Peel strength was calculated using DATUM 97

software (United Testing Systems).

9.3. RESULTS AND DISCUSSION

9.3.1. Peel Strength of Surfaces Treated with Silane Coupling Agents

Experimental results showing the 180 degree peel strength between the
photopolymerizable liquid encapsulants and the Alloy 194 strips treated with the silane
coupling agents are shown in Figure 9.3. Each bar in the bar chart represents the average
of 5 samples and the error bars indicate one standard deviation above and below the
mean. As shown in Figure 9.3, the peel strength of strips that were not treated with silane
coupling agents ranged from 18 to 40 g/cm. In addition, the peel strength of strips that
were treated with MPMS ranged from 28 to 40 g/cm, demonstrating that the use of

MPMS has a minimal effect in improving the peel strength. The only noticeable

I37

improvement shown by MPMS occurs when the liquid encapsulant is ﬁlled with 74 wt.%
fused silica. However, Figure 9.3 clearly shows that the use of AAMS signiﬁcantly
improves the peel strength, since the peel strength values for strips treated with AAMS
were between 35 and 95 g/cm. At each level of ﬁller loading, the peel strength is more
than doubled for Alloy 194 strips that were treated with AAMS, as compared to strips
that were not treated with any silane coupling agent.

In addition, Figure 9.3 shows that irrespective of whether the Alloy 194 strips are
treated with a silane coupling agent, the peel strength reaches a maximum when the
photopolymerizable liquid encapsulant is ﬁlled with 72 wt.% fused silica. This is
somewhat unexpected, as other authors, most notably Kim, have shown that peel strength
shows an inverse relationship with viscosity, i.e., the lower the viscosity the higher the
peel strength.2 The viscosity of the liquid encapsulant ﬁlled with 70.0 wt.% fused silica
is 5.92 Paos (5920 cP), while that of the 72.0 wt.% encapsulant is 13.9 Paos (13,900 CF),
and 74.0 wt.% is over 30 Paos (over 30,000 cP). Therefore, we would expect that the
encapsulant ﬁlled with 70.0 wt.% fused silica to have the highest peel strength, followed
by 72.0 wt.%, and then 74.0 wt.%. The reason for the discrepancy between the data
shown in Figure 9.3 and the data of Kim is that Kim has altered the viscosity level by
varying the catalyst levels and processing times, whereas, we have varied the viscosity
levels by solely varying the ﬁller loading. As a result, we have two competing effects on
the peel strength: viscosity and internal stress. It appears that the PLE ﬁlled with 72.0
wt.% fused silica strikes the appropriate balance: it has a low enough viscosity to provide
adequate wetting of the Alloy 194 strips, and enough fused silica to sufﬁciently reduce

the internal stresses (which lower peel strength).

138

Commercial epoxy molding compounds generally exhibit 180 degree peel
strengths of 40 to 200 g/cm when adhered to copper leadfrarne materials.2 Figure 9.3
illustrates that the PLEs ﬁlled with 70.0 and 72.0 wt.% fused silica and that are molded
onto Alloy 194 strips treated with AAMS exhibit peel strength values in this range.
Therefore, these studies indicate that these materials exhibit appropriate adhesion for use
as microelectronic encapsulants. In addition, it is noteworthy that the 180 degree peel
strengths reported by Kim on the commercial encapsulants were performed using
relatively thin samples (1.25 mm thick), whereas, the PLE specimens in this study were
3.0 mm thick (microelectronic encapsulants are generally 1-3 mm thick and we chose the
high end of the range since it places the most stringent requirements upon cure and
adhesion). Other investigators have shown that an increase in the encapsulant thickness
can lead to increased intemal stresses and decreased peel strengths.12 Therefore, if the
peel strengths of the PLEs and the commercial molding compounds were compared based
upon identical specimen thicknesses, the PLEs should compare even more favorably than
the values shown in Figure 9.3.

9.3.2. Mechanism of Bond Failure between PLE and Surfaces Treated with Silanes

Adhesive joints may fail in two ways: 1) adhesive failure, which is interfacial
bond failure between the adhesive and the adherend (in this case the separation occurs at
the adhesive-adherend interface), and 2) cohesive failure, in which a layer of adhesive
remains on both surfaces (in this case separation occurs in the adhesive).l3 Cohesive
failure is the desirable mode of failure since it indicates that the adhesive bond was strong

enough to prevent failure at the interface.

139

In this study, the mode of the failure of the adhesive joint was investigated using
environmental scanning electron microscopy. Speciﬁcally, the Alloy 194 surface of the
failed peel test specimen was examined. Figures 9.4, 9.5, and 9.6 reveal that the surfaces
treated with either silane coupling agent, as well as the control surfaces (strips that were
cleaned and degreased, but were not treated with silanes or azoles) showed cohesive
failure. For these systems, the corresponding electron micrographs indicate that a layer
of the cured PLE remained on the Alloy 194 surface after the peel test.

9.3.3. Peel Strength of Surfaces Treated with Azole Compounds

In contrast to the surfaces treated with silane coupling agents discussed above,
peel test specimens prepared with Alloy 194 surfaces treated with azole compounds
performed quite poorly. In fact, these specimens were signiﬁcantly delarninated and,
therefore, the peel strength between the PLE and Alloy 194 strips treated with azole
compounds were too low to be measured. Not surprisingly, these samples showed
adhesive failure as shown in the scanning electron micrographs in Figures 9.7 and 9.8

(the Alloy 194 surface is essentially bare metal with little, if any, PLE remaining).

9.4. CONCLUSIONS
The adhesive strength of photopolymerizable liquid encapsulants (PLEs) to Alloy
194 (a copper-based alloy) was measured using the 180 degree peel test. Peel strengths
were measured as a function of wt.% fused silica in the PLE and the type of coupling
agent used on the Alloy 194 surface. Speciﬁc observations are as follows:
1. The peel strength of strips that were not treated with silane coupling agents ranged

from 18 to 40 g/cm. In addition, the peel strength of strips that were treated with the

140

 

silane coupling agent, 3-methacryloxypropyltrimethoxysilane (MPMS), ranged from
28 to 40 g/cm, demonstrating that the use of MPMS has a minimal effect in
improving the peel strength.

. Alloy 194 surfaces treated with a silane coupling agent composed of an amino-
alkyltrimethoxysilane and a polyol diluted with methanol (AAMS) exhibited the
highest peel strengths, with values ranging from 35 and 95 g/cm. At each level of
ﬁller loading, the peel strength is more than doubled for Alloy 194 strips that were
treated with AAMS, as compared to strips that were not treated with any silane
coupling agent.

. Irrespective of the coupling agent used on the Alloy 194 surface, the peel strength
reaches a maximum when the photopolymerizable liquid encapsulant is ﬁlled with
72.0 wt.% fused silica.

. All of the surfaces treated with either of the silane coupling agents, as well as, the
control surfaces (strips that were cleaned and degreased, but were not treated with
silanes or azoles) showed cohesive failure.

. The peel strength between PLEs and surfaces treated with azole compounds could not
be measured because of extensive delamination. All of these surfaces showed
adhesive failure.

. The peel strengths of liquid encapsulants ﬁlled with 70.0 and 72.0 wt.% fused silica
and molded onto Alloy 194 strips treated with AAMS are appropriate values for

microelectronic encapsulants.

141

a~w-..m .'N__U _‘ .

l
r
I
i

77 31110 T 7 V V ' :3: 1-:lt..l.. _
Philip: EIEL 9395:5213.) TIF 1131] 1'11 131%- 11

 

Figure 9.1. Scanning electron micrograph of Alloy 194 strip before cleaning.

142

1 1‘1 l'l

'IF'FFFil‘l»! TIF

 

Scanning electron micrograph of Alloy 194 after cleaning.

Figure 9.2

143

 

 

120

D No silane
'3 MPMS
I AAMS

 

 

 

 

 

 

 

 

Peel Strength (glcm)

 

 

 

Wt.% Fused Silica

Figure 9.3. Peel strength between photopolymerizable liquid encapsulants and Alloy
194 strips treated with silane coupling agents.

20 iv 1250 E-Et
Pin) 1 1p; 131%.: hr 0'31: 5.

a

[Hill ; '7‘
3' 1717- “.72

NDRMTEHE TIF 15a h

 

Figure 9.4. Scanning electron micrograph of peel test specimen showing cohesive failure
between PLE ﬁlled with 72 wt.% fused silica and Alloy 194 strip without any silane. The
particles are remnants of the PLE adhered to the Alloy 194 surface after the 180 degree
peeltest

145

T 1225 0 >3

7- Ill Ill

' '20 1.0 i 7‘ '"
Philips Electro'Excsl 2.1;, I-lF'i-I‘ST'El'l'it TIF 11313. [Ge-3:11? :-

 

Figure 9.5. Scanning electron micrograph of peel test specimen showing cohesive failure
between PLE ﬁlled with 72 wt.% fused silica and Alloy 194 strip treated with MPMS.
The particles are remnants of the PLE adhered to the Alloy 194 surface after the 180
degree peel test.

146

: J .'H A. 2 ' I I --

‘9 lJE‘lulu _ T
RFlll'E..‘£l‘l-t TlF ll‘lt'l 1'13 1711;. -

 

Figure 9.6. Scanning electron micrograph of peel test specimen showing cohesive failure
between PLE ﬁlled with 72 wt.% fused silica and Alloy I94 strip treated with AAMS.
The particles are remnants of the PLE adhered to the Alloy 194 surface after the 180
degree peel test.

147

‘l
l
l

l

:5 an.
...1_-l....r._,-.. ...-
“"luln

ETERTJHT TIF “(”111

 

Figure 9.7. Scanning electron micrograph of peel test specimen showing adhesive failure
between PLE ﬁlled with 74 wt.% fused silica and Alloy 194 strip treated with BTCA.
Note the lack of any signiﬁcant amounts of the PLE remaining on the copper surface.

148

 

 

 

’ . "4’an" -.--

- .. 'rm . - ~
* - wmvae'bwwmy—m er'ts'y-
. ‘ 1

 

Figure 9.8. Scanning electron micrograph of peel test specimen showing adhesive failure
between PLE ﬁlled with 74 wt.% fused silica and Alloy 194 strip treated with PBI. Note
the lack of any signiﬁcant amounts of the PLE remaining on the copper surface.

149

10.

ll.

12.

13.

9.5. REFERENCES

. Y. Nakamura, M. Yamaguchi, A. Tanaka, and M. Okubo, J Appl Polym Sci, 49, 331,

(1993)
S. Kim, IEEE Trans. CHMT, 14, 809, (1991).

M.G. Pecht, Plastic-Encapsulated Microelectronics, John Wiley & Sons, Inc., New
York, NY, (1995).

E. Suhir and T.M. Sullivan, IEEE Trans. CHI/ff, 911, (1992).

R.L. Walberg and 8. Lion, EEP Electronic Packaging Reliability, ASME, 6, 55,
(1992)

B.J. Love, J. Baborowski, M. Charbonnier, and M. Romand, J Adhesion, 69: (1-2),
165, (1999).

SJ. Cho, K.W. Paik, and Y.G. Kim, IEEE Trans. CPMT Pt. B, 20, (2), 167, (1997).

SM. Song, C.E. Park, H.K. Yun, C.S. Hwang, S.Y. Oh, and J .M. Park, J. Adhesion
Sci. Technol, 12, 5, 541, (1998).

SM. Song, CE. Park, H.K. Yun, S.Y. on, and J.M. Park, .I. Adhesion Sci. Technol,
11, 6, 797, (1997).

H.K. Yuri, K. Cho, J .H. An, and CE. Park, J Mat Sci, 27, 5811, (1992).

L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, New York, (1990).

R.G. Schmidt and J .P. Bell, Adv Polym Sci, 75, 33-71, (1986).

A.H. Landrock, Adhesives Technology Handbook, Noyes Publications, Park Ridge,
NJ, (1985).

150

Chapter 10

RESEARCH SUMMARY AND RECOMMENDATIONS FOR
FUTURE WORK

10.1 SUMMARY OF RESULTS

This research project has demonstrated the tremendous potential of highly ﬁlled
photopolymerizable liquid encapsulants (PLEs) for microelectronic devices. The PLE is
comprised of an epoxy novolac-based vinyl ester resin (~25 wt.%), fused silica (70-74
wt.%), and small amounts of a photoinitiator, thermal initiator, and silane coupling agent.
This research has led to the development of a highly ﬁlled photopolymerizable liquid
encapsulant that has the material properties required by the microelectronics industry for
traditional transfer molding compounds (i.e. ﬂexural modulus and strength, coefﬁcient of
thermal expansion, glass transition temperature, thermal conductivity, and adhesion). In
addition, these PLEs have much lower viscosities than other transfer molding compounds
(in order to minimize wire sweep) and exhibit similar, if not faster, in-mold cure times.
Table 10.1 provides a head-to-head comparison of the material properties between the
PLEs and conventional transfer molding compounds. Table 10.1 shows that the material
properties of the PLEs compare quite favorably to molding compounds. Much more
signiﬁcant is the reduced viscosity of the PLEs and the PLE ﬂexural modulus values that

are near 1000 kgp’mmz.

151

Table 10.1. Comparison of material properties between PLEs and molding compounds

 

 

 

 

 

 

 

 

 

 

 

 

 

PLE Molding Compound
Viscosity (cP) 5,000 - 30,000 15,000 - 100,000
@30°C @170°C
Flexural Modulus (kgp’mmz) 1,000 - 1,200 1,000 - 1,500
Flexural Strength (kgp/mmz) 9 - 11 9 - 15
CTE (pm/m°C) l9 - 22 16 - 25
Tg (°C) 145 -155 140 -180
Thermal Stress Parameter 0.6 - 1.3 0.5 - 1.5
(Rial/Inmz)
Thermal Conductivity (W/m K) 0.56 - 0.62 0.50 - 0.70
180° Peel Strength (g/cm) 20 - 95 40 - 200

 

Based on the results presented in the previous chapters, the PLEs with differing
amounts of fused silica each have their own advantages and disadvantages. Clearly, for
applications with low stress requirements and for devices with high heat outputs, the PLE
ﬁlled with 74.0 wt.% would be best because of the low coefﬁcient of thermal expansion
and high thermal conductivity. If an extremely low viscosity formulation is needed to
minimize wire sweep, then the PLE ﬁlled with 70.0 wt.% fused Silica is most appropriate.
However, if a formulation is needed with a balance set of properties, then the PLE ﬁlled
with 72.0 wt.% is ideal. More importantly, this formulation exhibited the highest degree
of adhesion to Alloy 194, which is imperative to minimize delamination and corrosion.
Speciﬁc research results, as well as, recommendations for future work in this area are
presented in the remainder of this chapter.

10.1.1. Viscosity

For encapsulating microelectronic devices, it is desired to achieve a low enough
viscosity to allow the molding compound to easily ﬂow over the chip, wirebond, and
leadfrarne assembly, thereby minimizing problems associated with incomplete mold

ﬁlling, leadframe displacement, and wire sweep. As a result, we have characterized the

152

viscosity of the highly ﬁlled photopolymerizable liquid encapsulants to investigate the
blending of two different particle size distributions to tailor the viscosity for a given ﬁller
loading. Speciﬁcally, we characterized the viscosity of PLEs containing 70.0, 72.0, and
74.0 wt.% silica, and found that a blend of particle size distributions with a particle size
ratio of 3.13 resulted in the best viscosity reduction. The experimental results for the
effect of the particle size distribution on the PLE viscosity were well correlated by the
Krieger-Dougherty equation to account for the blending of two particle size
distributions.1 In addition, it was found that the viscosity of the PLEs decreased slightly
with increasing concentration of the silane coupling agent 3-
methacryloxypropyltrimethoxysilane.

As the fused silica loading was increased from 70.0 to 74.0 wt.%, the PLE
viscosity increased markedly. This apparently small increase in the fused silica loading
can be important for microelectronic encapsulants because it can lead to a lower
coefﬁcient of thermal expansion and an enhanced thermal conductivity. As the amount
of fused silica was increased from 70.0 to 72.0 wt.%, the viscosity increased by a factor
of 2.3, and as the fused silica loading was increased from 70.0 to 74.0 wt.%, the viscosity
increased more than ﬁve-fold. The current epoxy molding compounds are solid at room
temperature, therefore even the PLE containing 74.0 wt.% fused silica has a much lower
viscosity at ambient temperature than the traditional molding compounds.

10.1.2. Mechanical and Thermomechanical Properties

We have characterized the degree of cure, ﬂexural strength, ﬂexural modulus,

coefﬁcient of thermal expansion, glass transition temperature, and thermal stress

parameter of novel photopolymerizable liquid encapsulants. The effects of fused silica

153

loading, UV illumination time, and postcure time on these thermal and mechanical
properties were also investigated. The results indicate that a photocurable encapsulant
loaded with 74.0 wt.% fused silica is very promising for this application since the
material properties (ﬂexural strength, ﬂexural modulus, coefﬁcient of thermal expansion,
glass transition temperature, and thermal stress parameter) meet or exceed those of
current commercial encapsulants, while the viscosity and in-mold cure time are reduced.

These formulations are liquid at room temperature (current formulations are solid) and

cure to an ejectabie hardness in less than two minutes for an initiating light intensity of

200 mW/cmz. The photocured samples exhibit appropriate values for the above

mentioned material properties. Speciﬁc observations are as follows:

1. The ﬂexural strength and ﬂexural modulus increased with UV light illumination time
until reaching a plateau value in ~90 to 120 seconds.

2. The ﬂexural strength and glass transition temperature increased with increasing
postcure time at 170°C before reaching a plateau in ~4 to 6 hours. Current
encapsulants are typically postcured 4 to 8 hours at 170 or 175°C.

3. The coefficient of thermal expansion below the glass transition decreased with
increasing postcure time at 170°C before reaching a plateau in ~4 to 6 hours.

4. Inclusion of a thermal initiator, benzoyl peroxide, in the photopolymerizable liquid
encapsulant signiﬁcantly increased the ﬂexural strength of the samples, but had little
effect on the other material properties;

5. The ultimate degree of cure attained in a given cure cycle was signiﬁcantly enhanced
by the inclusion of a thermal initiator (in addition to the photoinitiator) and by

postcure at an elevated temperature.

154

10.1.3. Reduction of In-Mold Cure Time

Current trends in microelectronic encapsulation will require in-mold cure times
for molding materials to be reduced from 120 seconds to 60-90 seconds. With current
thermally cured systems, this reduced cure time is difﬁcult to achieve. We have
demonstrated that photopolymerizable liquid encapsulants can meet this requirement.
Photopolymerizable liquid encapsulants that were photocured for 60 and 90 seconds had
appropriate ejectabie hardness, and upon postcure, exhibited material properties (ﬂexural
strength and modulus, coefficient of thermal expansion, glass transition temperature, and
thermal stress parameter) that were equivalent or better than those of conventional
encapsulants. It was also shown that the addition of a thermal initiator does not
negatively affect any of the material properties, and leads to improved ﬂexural strength.
10.1.4. Thermal Conductivity, Specific Heat, and Thermal Diffusivity

It is critical that the silicon chip does not overheat during normal operation. The
thermal properties of the encapsulant determine the amount of heat that can be removed
from the device and thus affect operating temperatures of the electronic components. The
thermal conductivity, speciﬁc heat, and thermal diffusivity of photopolymerizable liquid
encapsulants (PLEs) were measured as a function of fused silica loading by the laser ﬂash
method. Photopolymerizable liquid encapsulants loaded with 70.0, 72.0, and 74.0 wt.%
fused silica exhibited thermal conductivities between 0.56 and 0.62 W/m-K, which is
within the 0.5 to 0.7 W/m.K range exhibited by most fused silica loaded molding
compounds. The thermal conductivity values of the PLEs were approximately three

times that of the unﬁlled epoxy novolac-based vinyl ester resin (k = 0.182 W/m-K).

155

The experimental therrnai conductivity data was ﬁt to the second order model of
Agari and Uno in order to predict the thermal conductivity of the PLES at ﬁller loadings
lower than 70.0 wt.% and ﬁller loadings higher than 74.0 wt.%.2 The experimental
thermal conductivity data was well correlated by the model of Agari and Uno; this model
can be used to accurately predict the thermal conductivity of the PLES over a wide range
of ﬁller loadings.

The speciﬁc heat for the PLES ranged between 0.83 and 0.87 J/g-K and was
shown to decrease with increasing fused silica content, in agreement with the results of
other authors.3 In addition, values for the thermal diffusivity ranged between 3.7 x 10'3
and 4.2 x 10‘3 cmzls and were shown to increase with increasing fused silica content.
10.1.5. Adhesion

Proper adhesion between the encapsulant and the ieadframe is critical in order to
prevent delamination and corrosion. The adhesive strength of photopolymerizable liquid
encapsulants (PLES) to Alloy 194 (a copper-based alloy) was measured using the 180
degree peel test. Peel strengths were measured as a function of wt.% fused silica in the
PLE and the type of coupling agent used on the Alloy 194 surface. Speciﬁc observations
are as follows:

1. The peel strength of strips that were not treated with silane coupling agents ranged
from 18 to 40 g/cm. In addition, the peel strength of strips that were treated with the
silane coupling agent, 3-methacryloxypropyltrimethoxysilane (MPMS), ranged from
28 to 40 g/cm, demonstrating that the use of MPMS has a minimal effect in

improving the peel strength.

156

2. Alloy 194 surfaces treated with a silane coupling agent composed of an amino-
alkyltrimethoxysilane and a polyol diluted with methanol (AAMS) exhibited the
highest peel strengths, with values ranging from 35 and 95 g/cm. At each level of
ﬁller loading, the peel strength is more than doubled for Alloy 194 strips that were
treated with AAMS, as compared to strips that were not treated with any silane
coupling agent.

3. Irrespective of the coupling agent used on the Alloy 194 surface, the peel strength
reaches a maximum when the photopolymerizable liquid encapsulant is ﬁlled with
72.0 wt.% fused silica.

4. All of the surfaces treated with either of the silane coupling agents, as well as, the
control surfaces (strips that were cleaned and degreased, but were not treated with
silanes or azoles) showed cohesive failure.

5. The peel strength between PLES and surfaces treated with azole compounds could not
be measured because of extensive delamination. All of these surfaces showed
adhesive failure.

6. Photopolymerizable liquid encapsulants ﬁlled with 70.0 and 72.0 wt.% fused silica
and molded onto Alloy 194 strips treated with AAMS exhibited high enough peel

strengths for the purposes of encapsulating microelectronic devices.

10.2. RECOMMENDATIONS FOR FUTURE WORK
Although this research has demonstrated the tremendous potential of using
photopolymerizations for encapsulating microelectronic devices, further research is

recommended in order to improve the existing formulation and fully establish

157

photopolymerizable liquid encapsulants as a viable alternative to traditional transfer
molding compounds. Speciﬁc research recommendations are described in the following
sections.
10.2.1. Increased Light Intensities

In all of the experiments reported in this work, the light intensity was held
constant at 200 mW/cmz. Increasing the light intensity is an excellent possibility to
further reduce the in-mold cure time. An added beneﬁt of increased light intensities is
that they may be able to signiﬁcantly increase the degree of cure to a point where the
postcure step is unnecessary. Although numerous authors have reported on the beneﬁts
of using higher light intensities to reduce the cure time of composites, higher light

4'6 Therefore,

intensities may also negatively affect some of the material properties.
further experiments should be conducted to determine the effects of higher light
intensities (up to 500 mW/cmz) on the material properties of the photopolymerizable
liquid encapsulants, as well as, the ability of higher intensities to ﬁrrther reduce the in-
mold cure time.
10.2.2. Development of Photomolding Process to Encapsulate Actual Microdevices
As was stated in Chapter 3, the primary goal of this research project was to
develop a photopolymerizable formulation that has the material properties required by the
microelectronics industry to effectively protect semiconductor devices. Now that a
photopolymerizable liquid encapsulant has been created, future research must focus on
developing a process to use this formulation and to demonstrate that the PLES can

minimize problems that are plaguing the conventionally used transfer molding process.

Fortunately, a photomolding process has been developed by Hanemann et al. that could

158

easily lend itself to the photopolymerizable liquid encapsulant.7 These authors recently
photomolded unﬁlled poly(methylmethacrylate) and found that they could produce high
quality parts more than ﬁve millimeters thick. Basically, the process is adapted from
liquid injection molding and combined with the necessities of photopolymerization.
Liquid injection molding is a reactive polymer processing method where low viscosity
monomers or oligomers are injected into the mold and polymerized to form the hardened
part. Injection molding technology represents an established process for manufacturing
polymer products with high accuracy in large-scale production (which is exactly what is
needed for molding microelectronic devices). The experimental setup used by Hanemann
et a1. consists of two parts: the lower part with the light source and the upper part with the
molding tools. Speciﬁcally, the upper part contains the mold insert, the ejector pins, the
connectors for vacuum pump and external temperature control, and the reservoir with the
reactive mixture. The most distinguishing features of the process developed by
Hanemann et al. compared to conventional injection molding are the use of an evacuated
molding tool made from high accuracy glass (to allow light to shine through the mold and
cure the photopolymerizable formulation) and the integration of a powerful UV light
source. Clearly, in order to effectively utilize the BAPO photoinitiator and to properly
cure the photopolymerizable liquid encapsulant, the selection of the proper light source
for this process is imperative. Medium-pressure mercury lamps have only a few
predominant emission lines at 405 and 436 run that can be absorbed by the photoinitiator.
However, medium pressure gallium-doped lamps have an improved emission over 400

nm and, therefore, it is recommended that this light source be used in conjunction with a

159

conventional medium-pressure mercury lamp in order to generate radiation that can
penetrate into the deeper layers of the encapsulant and increase the degree of cure.8
There are a number of advantages in using a photomolding process. First,

molding at ambient temperature results in only a small temperature rise due to the
released polymerization enthalpy. Second, high ﬁller loadings to increase the thermal
conductivity and to lower the coefﬁcient of thermal expansion can be accommodated
while still achieving lower viscosities than epoxy molding compounds since the base
resin is a low viscosity liquid at room temperature. Third, high thermal conductivity
ﬁllers (which are much more abrasive than fused silica) may be tolerated better in this
process because lower pressures and lower velocities could be used to rrrinirrrize erosion
of the mold.9
10.2.3. Accelerated Testing

Once a photomolding process has been developed that can use the PLE and
effectively encapsulate microelectronic devices, the molded devices must be tested for
their reliability. In the microelectronics industry, extensive accelerated testing is
performed speciﬁcally for this purpose. Accelerated testing is conducted to cause the life
aging of products to occur at a rate faster than would be obtained under normal operating
conditions. Commonly used accelerated tests for microelectronic devices are high
temperature tests, temperature cycling tests, humidity tests, and pressure cooker tests.
10.2.4. Glob-top Encapsulation

Because the PLES are able to meet the stringent requirements of molding

compounds, it is likely that the PLES can meet the needs of lower end encapsulation

methods, such as glob-top encapsulants. A glob-top encapsulant is a moderately viscous

160

liquid or paste and is used to encapsulate a chip-on-board or a hybrid on a substrate.'0
Glob-top encapsulants are typically thermoset epoxy or silicone resin formulations, ﬁlled
to a medium level with inorganic ﬁllers, and are traditionally meant for low-end
disposable electronic products. The glob-top encapsulant is applied through a dispenser
(usually a syringe ﬁtted with a positive displacement valve to minimize the variation in
the volume and rate of encapsulant delivered) to an assembled, cleaned, and tested chip
on a board.ll Epoxies formulated for glob-top applications typically take between 6 to 15
minutes at 120°C to cure.12 For optimum properties, further postcuring is recommended
for at least another 6 hours. It has already been demonstrated that the
photopolymerizable liquid encapsulants can cure in a shorter amount of time and exhibit
a viscosity suitable for this application. However, actual electronic components on
circuit boards must be encapsulated using PLES, as well as the glob-top process, to

demonstrate the suitability of the PLES as glob-top encapsulants.

161

10.

11.

12.

10.3. REFERENCES

. I.M. Krieger and T.J. Dougherty, Trans. Soc. Rheol, 3, 137 (1959).

Y. Agari and T. Uno, J Appl Polym Sci, 32, 5705, (1986).

P. Bujard, G. Kuhnlein, S. Ino, and T. Shiobara, IEEE Trans CPMT, Pt A, 17,4, 527,
(1994)

M. Miyazaki, Y. Oshida, B.K. Moore, and H. Onose, Dental Materials, 12, 5-6, 328,
(1996).

F .A. Rueggeberg, W.F. Caughman, and J .W. Curtis, Operative Dentistry, 19, 1, 26,
(1994).

C. Decker, D. Decker, and F. Morel, Photopolymerization, 673, 63, (1997).

T. Hanemann, R. Ruprecht, J .H. HauBelt, Polym Mat Sci Eng, 77, 418, (1997).
K. Dietliker, Radiation Curing in Polymer Science and Technology, Volume 2,
editors, J .P. F ouassier and J .F . Rabek, Elsevier Applied Science, New York, NY,
155-237, (1993).

M.G. Pecht, Plastic-Encapsulated Microelectronics, JohnWiley & Sons, Inc., New
York, NY, (1995).

L.T. Manzione, Plastic Packaging of Microelectronic Devices, Van Nostrand
Reinhold, New York, (1990).

M. Shinohara, Electronic Materials Handbook, ed. M.L. Minges, ASM International,
Materials Park, OH, 802, (1989).

P.V. Robock and L.T. Nguyen, Microelectronics Packaging Handbook, eds. R.R.

Tummala and E.J. Rymaszewski, Van Nostrand Reinhold, New York, NY, 523-672,
(1989).

162

'l(1:11(lilyl'ljglilgil '