.\.V )s; .

. V . . V anxi. . .
t. V.. «2...: . . V . . V V . . A . V . .summwmflmlgt
enuw‘wgkaf. ‘ V . . . nah: .3: A: . , :thwc. !
r”) .i . , . , . , at}? . .1
a V V q. “$.23.
V . a!

.A, .r .235“.
d

a.

B.
35*»

l
3"

AC 1.

.F, x.
A? t.

, is . .
‘1...“ V...
.23.: 1

5:...
7.... .
Du a

. :Iz. :
4%..th .

Dyna".

“Oink-H

’1‘! int,-
‘J it’ll.

2 in.
.V..; . .

1131111144135
.14; as . . 4 . s. V 3.5%” 3%
21.1.?! . . . a.“ . V «gnaw o ,

11.1.4 gang“:

 

22W
3
goo 6

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled
INTERACTIONS IN MICROWAVE ADHESIVE BONDING OF
POLYMERS AND COMPOSITES

presented by

Shuangjie Zhou

has been accepted towards fulfillment
of the requirements for

Ph.D. . Chemical Engineering and
degree 1n

 

 

Materials Science

XMC’II W3:

Major professor

Date 331% ’7’. LQQ'Z;

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

a.“ 4M _ b _A v .o _A..

b a... _.4 u.

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

aim

 

INTERACTIONS IN MICROWAVE ADHESIVE BONDING OF POLYMERS AND
COMPOSITES

By

Shuangjie Zhou

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Chemical Engineering and Materials Science

2002

ABSTRACT

INTERACTIONS IN MICROWAVE ADHESIVE BONDING OF POLYMERS AND
COMPOSITES

By

Shuangjie Zhou

Microwave processing of advanced materials has been studied as an attractive
alternative to conventional processing. In this dissertation, work was performed in two
sections. The first section was a process study and focused on applying microwave
theories to develop an integrated microwave adhesive bonding system in a single mode
applicator with non-invasive on-line monitoring and precise controlling features. The
second section was a fundamentals study and focused on investigating microwave
heating mechanism to provide explanations for rapid bonding using microwaves.

A single mode microwave heating method was first applied in adhesive bonding
process. The microwave heating mode was chosen based on theoretical computation and
experimental characterization so that the adhesive was placed at the strongest electric
field. Rapid and selective heating of the adhesive was observed. However, only limited
material size could be uniformly heated with single mode microwave method because of
the non-uniform electric field distribution. To solve this problem, a variable frequency
mode switching method (switch between modes with complementary heating patterns)
was used to improve the bonding uniformity of large-size materials. Theoretical heating
modes were applied in this process. A control system was developed to provide rapid,
stable, and uniform bonding by adjusting microwave power and frequency. The control

software was programmed with LabVIEW. To determine the bonding cycle on-line, a

new method (with corresponding software) was invented based on microwave theories
and experimental discoveries. Compared with thermal adhesive bonding, the microwave
method reduced the bonding cycle vastly and obtained equal or even higher bonding
strength for the materials used in this research.

The reduction in bonding cycle resulted from faster curing of the epoxy adhesive
with microwaves. The phenomenon of microwave fast curing of epoxy was investigated
with a new method by studying the effect of carbon additive on microwave curing of
epoxy. It was hypothesized that during microwave curing, carbon absorbed most of the
microwaves and weakened the localized superheating of the epoxy functional groups. If
localized superheating were the main mechanism of rate enhancement in microwave
curing of epoxy, then the curing rate should decrease at the presence of carbon.
Microwave curing experiments were carried out at three temperatures with various
carbon concentrations. New correlations for reaction rate constants as functions of carbon
concentration and temperature were proposed. Results showed microwave curing rate
decreased with increasing carbon concentration. Thus, this study suggested reaction rate
enhancement in microwave curing of epoxy result from localized superheating of the
functional groups. However, the carbon particles used in this study were activated and
might adsorb the amine from the epoxy resin because carbon particles might have higher
temperature than the bulk resin in microwave curing. This will result in decrease in
microwave curing rate. Further study is required to elucidate this problem. In addition, at
higher carbon concentration, the required microwave power was lower to maintain the
same bulk temperature. This might also result in decrease in microwave curing rate at

higher carbon concentration.

Copyright by
Shuangjie Zhou

2002

ACKNOWLEDGEMENT

I would like to thank my advisor Dr. Martin C. Hawley for his invaluable
guidance during this research. I also acknowledge Dr. Leo Kempel, Dr. Lawrence Drzal,
Dr. Greg Baker, Dr. Christian Lastoskie, and Dr. J ianghua Wei, for their many insightful
suggestions. I am also grateful to Mr. Mike Rich and Mr. Kelby Thayer for their very
helpful training in equipment operation and advice in designing material testing
procedures. Acknowledgement is extended to Dr. Richard Schalek and Dr. Per Askeland
for analyzing the surface of carbon particles and carbon filled epoxy resins. Finally, I

thank my husband Kun for his endless love and support throughout this work.

TABLE OF CONTENTS

LIST OF FIGURES .......................................................................................................... xi
LIST OF TABLES .......................................................................................................... xvi
CHAPTER 1 INTRODUCTION ....................................................................................... 1
1.1 Advantages of Microwave Heating over Thermal Process ......................... 2
1.2 Literature Survey ......................................................................................... 5
1.3 Research Topics .......................................................................................... 8
CHAPTER 2 BACKGROUND FOR MICROWAVE HEATING ................................. 11
2.1 Electromagnetic Fields in a Microwave Enclosure ................................... 11
2.1.1 Maxwell’s Equations ................................................................................. 11
2.1.2 Resonant Modes in a Cylindrical Single Mode Cavity ............................. 14
2.1.3 Electromagnetic Fields in a Cylindrical Single Mode Cavity ................... 17

2.2 Microwaves/Materials Interactions ........................................................... 19
2.2.1 Mechanisms of Microwave/Materials Interactions ................................... 19
2.2.2 Dielectric Properties .................................................................................. 21

2.3 Applications of Microwave Energy .......................................................... 25
2.4 Microwave Applicators ............................................................................. 29
2.5 Temperature sensing system ..................................................................... 30
CHAPTER 3 SINGLE MODE MICROWAVE ADHESIVE BONDING ..................... 32
3. 1 Introduction ............................................................................................... 32

vi

3.2

3.2.1

3.2.2

3.2.3

3.2.4

3.3

3.3.1

3.3.2

3.3.3

3.3.4

3.3.5

3.4

Experimental ............................................................................................. 35
Experimental Circuit ................................................................................. 35
Materials .................................................................................................... 37
Method of Dielectric Property Measurement ............................................ 37
Sample Setup and Preparation for Adhesive Bonding .............................. 39
Results and Discussion .............................................................................. 41
Material Dielectric Properties ................................................................... 41
Characterization of Empty Cavity ............................................................. 42
Characterization of Loaded Cavity and Mode Diagnosis ......................... 44
Single Mode Microwave Adhesive Bonding ............................................ 52

Comparison between single mode microwave and thermal adhesive
bonding in terms of bonding time and strength ......................................... 53

Conclusions ............................................................................................... 58

CHAPTER 4 VARIABLE FREQUENCY MODE-SWITCHING MICROWAVE

4.1

4.2
4.2.1
4.2.2
4.2.3

4.3

4.3.1

ADHESIVE BONDING ........................................................................... 60
Introduction ............................................................................................... 60
Experimental ............................................................................................. 62
Circuit for Microwave Adhesive Bonding ................................................ 62
Materials .................................................................................................... 62
Sample Setup and Preparation for Adhesive Bonding .............................. 63
Results and Discussion .............................................................................. 65
Characterization of Loaded Cavity ........................................................... 65

vii

4.3.2

4.3.3

4.3.4

4.3.5

4.3.6

4.4

Mode Diagnosis and Selection of Modes for Mode-switching ................. 67
Algorithm of Mode-switching ................................................................... 69

Heating Profile of Variable Frequency Mode—switching Adhesive Bonding
and Comparison with Thermal Process ..................................................... 73

Study of Adhesive Curing in Mode-Switching Microwave Process and
Comparison with Thermal Curing ............................................................. 78

Comparison of Bond Strength between Mode-Switching Microwave and
Thermal Bonding Processes ...................................................................... 82

Conclusions ............................................................................................... 85

CHAPTER 5 IN SITU MONITORING OF VARIABLE FREQUENCY MICROWAVE

5.1

5.2

5.3

5.3.1

5.3.2

5.4

5.4.1

5.4.2

5.5

5.5.1

5.5.2

PROCESSING IN A SINGLE MODE CAVITY ..................................... 87
Introduction ............................................................................................... 87
Advantages Of This Technique ................................................................. 89
On-line Monitoring of Microwave Adhesive Bonding of Single-Lap

Bexloy W502/ Eccobond A40l-37 ........................................................... 90
Experimental ............................................................................................. 90
Results and Discussion .............................................................................. 9l

On-line Monitoring of Microwave Adhesive Bonding of Sin gle-Lap

Surlyn SG201U / Eccobond A401-37 ....................................................... 94
Experimental ............................................................................................. 94
Results and Discussion .............................................................................. 94

On-line Monitoring of Microwave Adhesive Bonding of Double-Lap

Bexloy W502/ Eccobond A40l-37 ........................................................... 96
Experimental ............................................................................................. 96
Results and Discussion .............................................................................. 98

viii

5.6

Conclusions ............................................................................................. 101

CHAPTER 6 PROCESS CONTROL SYSTEM FOR MICROWAVE ADHESIVE

6.1

6.2

6.3

6.4

6.4.1

6.4.2

6.4.3

6.4.4

6.5

BONDING .............................................................................................. 102
Introduction ............................................................................................. 102

Structure of the Process Control System for Microwave Adhesive Bonding

................................................................................................................. 104
Program for Cavity Characterization ...................................................... 106
Program for Process Control ................................................................... 106
Subprograms for Data Acquisition .......................................................... 107
Subprograms for Mode-Switching .......................................................... 107
Subprograms for Microwave Power Control .......................................... 108
Subprogram for On-line Monitoring ....................................................... 111
Conclusions ............................................................................................. 1 12

CHAPTER 7 INVESTIGATION OF MICROWAVE HEATING MECHANISM VIA

7.1

7.2

7.3

7.3.1

7.3.2

7.3.3

7.4

STUDY OF MICROWAVE CURING OF EPOXY FILLED WITH

CARBON ................................................................................................ 1 13
Introduction ............................................................................................. 1 13
Hypothesis of Carbon Effect on Microwave Curing ............................... 115
Epoxy Curing Kinetics ............................................................................ 116
Kinetics study of thermal curing of neat resin ........................................ 117
Kinetics study of thermal curing of filler-added resins ........................... 121
Kinetics study of microwave curing of neat resin ................................... 124
Kinetics Model Used in this Study .......................................................... 125

ix

7.5 Experimental ........................................................................................... 127
7.5.1 Materials and Sample Preparation ........................................................... 127
7.5.2 Experimental Setup for Microwave Curing ............................................ 128
7.5.3 Dielectric Measurement .......................................................................... 129
7.5.4 Cavity Characterization ........................................................................... 131
7.5.5 Process Control Strategy and Temperature Profile in Microwave Heatinlg32

7.6 Results and Discussion ............................................................................ 134
7.6.1 Microwave Power Deposition During Heating and Curing .................... 134
7.6.2 Comparison of Reaction Rates between Microwave and Thermal Curing

of Neat and Doped Resins ....................................................................... 137

7.7 Conclusions ............................................................................................. 153

CHAPTER 8 CONCLUSIONS ..................................................................................... 155
CHAPTER 9 FUTURE WORK .................................................................................... 163

APPENDIX A. MATLAB PROGRAM FOR CALCULATING THE MODE PATTERN

OF TM022 INSIDE THE EMPTY CAVITY ......................................... 178

APPENDIX B. LABVIEW PROGRAMS .................................................................... 180

LIST OF FIGURES

Figure 1.1 Temperature Profile in Microwave or Thermal Heating [1] ............................. 3
Figure 1.2 Illustration of Microwave Adhesive Bonding ................................................... 6
Figure 2.1 TB Modes in an Empty Cavity with Diameter of 17.78cm ............................ 16
Figure 2.2 TM Modes in an Empty Cavity with Diameter Of 17.78cm .......................... 16
Figure 2.3 Cylindrical Single-Mode Resonant Cavity ..................................................... 30
Figure 3.] Circuit of Microwave Adhesive Bonding ....................................................... 36
Figure 3.2 Sample Setup and Temperature Measuring Points in Single Mode Microwave
Bonding ..................................................................................................................... 40
Figure 3.3 Characterization of Empty Cavity .................................................................. 43
Figure 3.4 Reflectance in Cavity Loaded with Samples (Sample Width of 1") ............... 45
Figure 3.5 Heating Rate at Different Frequencies ............................................................ 46
Figure 3.6 Dimension of the Adhesive and Location of Temperature Probes ................. 48
Figure 3.7 Experimental Temperature Distribution ......................................................... 48
Figure 3.8 Experimental Configuration for Measurement of Half Wave Number Q ...... 49
Figure 3.9 Measurement of the Half Wave Number Q .................................................... 50

Figure 3.10 Electric Field Distribution of TM022 Mode at the Cross-Section of the
Cavity (Lighter Region Represents Higher Electric Field Strength) ........................ 51

Figure 3.11 Temperature Profile during Single Mode Microwave Adhesive Bonding 52
Figure 3.12 Strength of the Assemblies Bonded at Different Conditions ........................ 54
Figure 3.13 Load-Elongation Curves of Assemblies Bonded under Different Conditions

................................................................................................................................... 56

xi

Figure 4.1 Sample Setup and Temperature Measuring Points ......................................... 63

Figure 4.2 Characterization of Loaded Cavity (Sample Size: 93.6 x 76.2 x 6.06mm3)... 66

Figure 4.3 Characterization of Heating Patterns of Modes 1 And 2 ................................ 68
Figure 4.4 Measurement of Half Wave Number Q .......................................................... 70
Figure 4.5 Theoretical E Field Patterns in the Empty Cavity .......................................... 71
Figure 4.6 Dimension of the Adhesive inside the Cavity ................................................. 72

Figure 4.7 Power Deposition in Mode-Switching Adhesive Bonding of Bexloy at 110 °C
................................................................................................................................... 74

Figure 4.8 Power Deposition in Mode-Switching Adhesive Bonding of Bexloy at 120 °C
................................................................................................................................... 74

Figure 4.9 Temperature Profile in Mode-Switching Bonding of Bexloy at 110 °C ........ 75
Figure 4.10 Temperature Profile in Mode-Switching Bonding of Bexloy at 120 °C ...... 75
Figure 4.11 Temperature Profile in Thermal Bonding of Bexloy at 120 °C .................... 76

Figure 4.12 Temperature Measuring Points for the Eccobond/Surlyn Assembly (Top
View) ......................................................................................................................... 77

Figure 4.13 Temperature Profile in Mode-Switching Bonding of Surlyn at 100 °C ....... 78

Figure 4.14 Adhesive Extent of Cure vs. Time in Microwave Process of 6 Minutes of
Heat up and Then Isothermally at 110°C .................................................................. 79

Figure 4.15 Adhesive Extent of Cure vs. Time in Microwave Process of 8 Minutes of
Heat up and Then Isothermally at 120°C .................................................................. 80

Figure 4.16 Adhesive Extent of Cure vs. Time in Thermal Heating of 30 Minutes Heat
up and Then Isothermally at 110°C ........................................................................... 81

Figure 4.17 Adhesive Extent of Cure vs. Time in Thermal Heating of 40 Minutes Heat
up and Then Isothermally at 120°C ........................................................................... 81

xii

Figure 5.1 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Bexloy Substrates at 110°C ............................................................................... 92

Figure 5.2 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Bexloy Substrates at 120°C ............................................................................... 93

Figure 5.3 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Surlyn Substrates at 100°C ................................................................................ 95

Figure 5.4 Sample Setup of Double-Lap Shear Assembly ............................................... 97

Figure 5.5 Temperature Profile in Microwave Adhesive Bonding of Double-Lap Shear
Assembly of Bexloy Substrates ................................................................................. 98

Figure 5.6 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Double-

Lap Bexloy Substrates at 120°C ............................................................................. 100
Figure 6.1 Temperature Oscillation for Determining Ku and Tu ................................... 110
Figure 7.1 Microwave Interactions with Carbon Black and Epoxy ............................... 116
Figure 7.2 Mode Spectrum of Loaded Cavity for Microwave Curing ........................... 132
Figure 7.3 Temperature Profiles during Microwave Heating and Curing at 145, 165 and

185°C ....................................................................................................................... 133
Figure 7.4 Power Curves during Microwave Heating and Curing ................................. 136
Figure 7.5 Total Energy Consumed by Each Resin System in the First 60 Minutes of

Microwave Heating ................................................................................................. 137
Figure 7.6 Thermal Curing at 145°C .............................................................................. 139
Figure 7.7 Thermal Curing at 165°C .............................................................................. 140
Figure 7.8 Thermal Curing at 185°C .............................................................................. 141
Figure 7.9 Microwave Curing at 145°C ......................................................................... 142
Figure 7.10 Microwave Curing at 165°C ....................................................................... 143
Figure 7.11 Microwave Curing at 185°C ....................................................................... 144

xiii

Figure 7.12 Effect of Carbon Concentration on Thermal Curing Rates ........................ 148

Figure 7.13 Activation Energy in Thermal Curing ........................................................ 149
Figure 7.14 Effect of Carbon Concentration on Microwave Curing Rates .................... 150
Figure 7.15 Activation Energy in Microwave Curing .................................................... 151
Figure B.l Hierarchy of the Labview Programs ............................................................ 181
Figure B.2 Front Panel of the Cavity Characterization Program ................................... 182
Figure B.3 Diagram of the Cavity Characterization Program ........................................ 183
Figure B.4 Front Panel for Variable Frequency Mode-Switching Process Control ....... 184

Figure B.5 Diagram for Variable Frequency Mode Switching Process Control (Part 1)
................................................................................................................................. l 85

Figure 8.6 Diagram for Variable Frequency Mode-Switching Process Control (Part 2)

................................................................................................................................. 1 86
Figure B.7 Front Panel for 6t.Vi ..................................................................................... 187
Figure B.8 Diagram for 6t.Vi ......................................................................................... 188
Figure 3.9 Front Panel for Luxtron.Vi ........................................................................... 189
Figure B.10 Diagram for Luxtron.Vi ............................................................................. 189
Figure B.11 Front Panel for Nortech.Vi ......................................................................... 190
Figure 8.12 Diagram for Nortech.Vi ............................................................................. 190
Figure 3.13 Front Panel for Power.Vi ........................................................................... 191
Figure B.14 Diagram for Power.Vi ................................................................................ 191
Figure B.15 Front Panel for Mode-Switch.Vi ................................................................ 192
Figure B.16 Diagram for Mode-Switch.Vi .................................................................... 193

xiv

Figure 8.17 Front Panel for Power Control.Vi .............................................................. 194

Figure B.18 Diagram for Power Control.Vi ................................................................... 195
Figure B.19 Front Panel for Heat Rate.Vi ...................................................................... 196
Figure B.20 Diagram for Heat Rate.Vi .......................................................................... 196
Figure B.21 Front Panel for Pid.Vi ................................................................................ 197
Figure B.22 Diagram for Pid.Vi ..................................................................................... 198
Figure 3.23 Front Panel for Pid Parameters.Vi .............................................................. 199
Figure 3.24 Diagram for Pid Parameters.Vi .................................................................. 199
Figure 8.25 Front Panel for Frequency Diagnosis.Vi .................................................... 200
Figure 8.26 Diagram for Frequency Diagnosis.Vi ........................................................ 201
Figure B.27 Front Panel for F-Write.Vi ......................................................................... 202
Figure 8.28 Diagram for f-write.Vi ............................................................................... 202

XV

LIST OF TABLES

Table 1.1 Comparison of Microwave Heating with Thermal Method ............................... 3
Table 3.1 Dielectric Properties of the Materials ............................................................... 41

Table 3.2 Theoretical and Experimental Values of the Resonant Frequencies of Several

Modes ........................................................................................................................ 44
Table 3.3 Break Pattern of Microwave Bonded Assemblies ............................................ 55
Table 3.4 Break Pattern of Thermally Bonded Assemblies ............................................. 56
Table 4.1 Summary of Mode Diagnosis ........................................................................... 72

Table 4.2 Comparison between Mode-Switching and Thermal Curing at 110°C and
120°C ......................................................................................................................... 82

Table 4.3 Bond Strength of Eccobond/Bexloy Assembly with Microwave or Thermal
Method ...................................................................................................................... 84

Table 4.4 Bond Strength of Eccobond/Surlyn Assembly with Microwave or Thermal
Method ...................................................................................................................... 84

Table 5.1 Microwave Adhesive Bonding Processes to Be Studied with the on-Line
Monitoring Technique ............................................................................................... 89

Table 5.2 Bond Strength of Microwave Bonded Double-Lap Shear Assemblies .......... 100
Table 6.1 PID Controller Parameters for Single Mode Microwave Adhesive Bonding 111

Table 6.2 PI Controller Parameters for Variable Frequency Mode Switching Bonding 1 11

Table 7.1 Total Heat of Reaction per Gram of Resin of the Fresh Samples .................. 128
Table 7.2 Dielectric Constants of the Materials ............................................................. 130
Table 7.3 Dielectric Loss Factors of the Materials ......................................................... 131
Table 7.4 PID Controller Parameters Obtained with Ziegler-Nichols Method .............. 133

xvi

Table 7.5 Average Standard Deviations of the Extent of Cure ...................................... 138
Table 7 .6 Values of M and N ......................................................................................... 138

Table 7.7 Reaction Rate Constants of Microwave and Thermal Curing of Neat Resin. 145

Table 7.8 Coefficients D's for Thermal Curing .............................................................. 147
Table 7.9 Coefficients D's for Microwave Curing ......................................................... 152
Table B.1 Functions of the Labview Programs .............................................................. 180

xvii

CHAPTER 1 INTRODUCTION

In this research, work was performed in two sections. The first section was a
process study and focused on applying microwave theories to develop an integrated
microwave adhesive bonding process in a single mode applicator with non-invasive on—
1ine monitoring and precise controlling features. The second section was a fundamentals
study and focused on the investigation of microwave heating mechanisms to provide
explanations for rapid bonding with microwaves.

Microwave refers to electromagnetic waves in the frequency range from 300MHz
to 3OOGHz or the characteristic wavelength range from 1m to 1mm. Heating is one of the
major non-communication applications of microwave energy. The fundamental
electromagnetic property of nonmagnetic materials for microwave heating and diagnosis
is complex permittivity (8' = 8' - je"). The real part of the complex permittivity is
permittivity a (dielectric constant), which is related to the energy stored in the materials.
The imaginary part of the complex permittivity is loss factor 8", which is related to energy
dissipated as heat in the materials. The loss factor of materials is generally due to
contributions by dielectric loss factor (the motion of dipoles) and conductivity (the
motion of charges). Thermoplastic and thermoset polymers, typical dielectric materials,
have polar groups to interact with electromagnetic fields and exhibit dielectric relaxation
at microwave frequencies, ranging from epoxide groups, hydroxyl groups, amino groups,
and so on. These polar groups can directly absorb microwave energy. The localized

heating on the reactive polar sites can initiate or promote reactions that require heat.

1.1 Advantages of Microwave Heating over Thermal Process

Microwave processing of advanced materials has been studied as an attractive
alternative to conventional thermal processing. Thermal heating is a surface-driven, non—
selective process. The heating efficiency is controlled by the heat transfer coefficient at
the material surface and the thermal conductivity of the materials. During thermal
heating, heat flows from the surface to the interior of the material. This tends to cause
remarkable temperature gradients in thick materials. Residual thermal stresses resulting
from large temperature gradients will reduce the physical and mechanical properties of
the materials. In addition, the production cycle is long because of the difficulty in heating
poor thermal conductors like polymers.

These problems can be solved with microwave radiation. Microwave heating
offers a number of advantages over thermal heating in a wide range of heating
applications. A comparison between microwave and thermal heating is summarized in
Table 1.1. Microwave heating is selective, instantaneous and volumetric with heat loss at
the boundaries whereas thermal heating is nonselective depending on temperature
gradient. Microwave heating can be easily controlled by very fast changes in the applied
electric field whereas thermal heating is characterized with long lag times and difficulty
for composite cure control. In microwave heating, the heat source can be readily removed
to prevent thermal excursion. Figure 1.1 shows a comparison of heating profiles between
thermal and microwave processes. Microwave heating rate is much higher and materials
can be heated to a higher temperature without causing thermal runaway. Microwave
processing has potential for rapid processing of thick-section and complex-shaped

composites.

Table 1.1 Comparison of Microwave Heating with Thermal Method

 

 

 

 

 

 

 

 

 

 

 

 

 

Thermal Microwave
Heating Low, determined by heat transfer High, energy directly coupled
Efficiency coefficients into materials
Selectivity No, controlled by Temperature Yes, determined by material
gradient loss factor
Controllability Low, long-lag High, microwave power level
can be readily adjusted
300
260~
1 0'.

8 220; 230 C microwave

a .,

V 4M"; 200°C thermal

8 130.." 190°C microwave

3 .1

E

a) 140

a

E

,2 100

 

 

r I
10 20 30 40 50

 

60 70 80 90

Time (minutes)

Figure 1.1 Temperature Profile in Microwave or Thermal Heating [1]

 

The advantages of microwave heating over thermal process can be easily
explained by consideration of the microwave power absorption in the material. The
microwave power absorption rate, P, (W/m3), inside a homogeneous material is shown in

Equation 1.1.
.. ~ 2
P=égeqq (LU

where E is the electric field strength inside the material, V/m,
(r) is the radian frequency, rad/sec, m=21tf, f is the frequency in Hz,
80 is the free space permittivity, 80:1/(3671) x 10'9 Flm,
8" is the effective relative loss factor, enzed" + 6/(8000),
ed" is the relative loss factor due to dipolar contributions,

0 is the material conductivity, S/m.

Specifically, E is responsible for the fast heating and controllability as it can be
easily regulated by the input power; 8" is responsible for the selectivity and conversion of
energy into heat. Microwave heating is volumetric because of the penetration of the
electric field into the material. Equation 1.1 can be used with several other equations in
modeling microwave heating. These equations include Maxwell’s equations for solving
the electric field strength, a heat transfer equation for obtaining the temperature
distribution inside the material, and a reaction kinetics equation for calculating extent of

reaction and reaction heat if chemical reaction occurs during the heating. Material

thermal and dielectric properties have non-linear dependence on both temperature and
extent of reaction. These properties change during the processing, which affects the
electrical field strength and power absorption in the materials. Thus the modeling of
microwave heating is a highly coupled non-linear problem. This problem will be

discussed in more detail in Chapter 9.

1.2 Literature Survey

The application of microwave heating in polymer and composite processing has
been shown to be very promising. In polymer processing, both thermosets and
thermoplastics have been studied. The drying process of various thermoplastics with
microwave energy has been investigated [2]. Microwaves have also been used to cure
thermosetting materials, including polyesters [3][4], polyurethanes [5][6], polyirrrides
[7,8,9] and epoxies [1,3, 10-20]. Both pulsed and continuous microwave curing of
epoxies have been studied [1,15]. In composites processing, both batch [21,22,23] and
continuous processes [15.24-28] have been studied. Demonstrated results in microwave
processing of polymers and composites include increased polymerization rate [7,11-14],
reduced drying time for pelletized polycarbonate and polypropylene [2], increased Tg for
cured epoxy [10], enhanced fiber/matrix adhesion in carbon composites [21], and
increased mechanical strength of graphite/epoxy composite [15].

In addition to the applications mentioned above, microwaves have been recently
applied in several adhesive bonding studies for polymers and composites [29-31]. In
adhesive bonding, the adhesive is applied between two thick substrate panels. In

conventional thermal processes, thermal energy has to be transferred through the thick

substrates to reach the adhesive. If the substrates are polymeric materials, the thermal
bonding cycle will be very long due to the low thermal conductivity of polymers. This
problem is solved with microwave method. Figure 1.2 is the illustration of microwave
adhesive bonding. The polymer or composite parts have low loss factors relative to the
adhesive material at the interface. Microwave energy is then directly transferred into the
interface and cures the adhesive extremely rapidly, even though the substrate materials
are very thick. Therefore, microwaves have the potential to be applied in industrial
adhesive bonding process to shorten the bonding time and thus lower the process cost.
This will be greatly beneficial for the automotive, marine and aerospace industries, where
adhesive bonding of polymers and composites has been developed as one of the leading
joining techniques. However, industrialization of a new technique requires design of
processes for specific applications, development of control and on-line monitoring

packages, and adequate understanding of process fundamentals.

Adhesives

 

 

Microwaves

 

 

 

 

 

x I 3t
\———) Polymer or

composite parts

Figure 1.2 Illustration of Microwave Adhesive Bonding

Research and development regarding the application of microwaves in adhesive
bonding is still in its early stage. Several investigators have studied microwave adhesive
bonding in multi-mode applicators [29-31]. These studies demonstrated that reduced
bonding cycle was obtained with adequate bonding strength in microwave process
compared with thermal bonding. However, the mechanism of bonding cycle reduction
was not proposed. In addition, these studies did not investigate the relation of microwave
heating characteristics (e. g. uniformity, selectivity, and rapidity) with the electromagnetic
field patterns and material dielectric properties. Moreover, detailed control and on-line
monitoring issues were not considered in these studies.

Process fundamentals can be more conveniently studied in a single mode resonant
applicator compared with a multi—mode applicator. Unlike multi-mode applicator, where
several modes are excited at the same time [32], a single mode applicator supports only
one mode at a time. The electric field patterns inside the single mode applicator are more
controllable and predictable. In addition, single mode applicators are more efficient in
coupling microwave energy into materials compared with multi-mode applicators. This
has been shown in the processing of crossply and thick-section graphite fiber reinforced
composites [22,23]. If adhesive bonding is conducted in a single mode applicator, the
joint area can be placed at the location with the highest electric field strength. This will
further enhance the selective heating of the adhesive.

In a single mode applicator, uniform heating of large size materials can be
achieved with mode-switching method. Adegbite [33] developed a method to obtain
uniform processing in fixed frequency microwave processing. Fellows et al. [34] used

fixed frequency mode switching technique to process V-shaped polyester/glass composite

parts. However in fixed frequency microwave processing, slow mechanical adjustment of
cavity length for mode switching resulted in instability of temperature and non-uniform
temperature distribution. With the use of a variable frequency microwave power source,
Qiu [35] developed a variable frequency mode switching method for polymer and
composite processing with microwaves. This technology provided more uniform and
stable processing of composite parts compared with fixed frequency mode switching.
However, empirical heating modes were used in the processing. This resulted in
extensive heating characterization and data storage both before and during the processing.
Thus the variable frequency mode switching method needs to be further improved to
solve this problem.

A process control system has been built for microwave variable frequency mode
switching processing in a single-mode cavity [35]. In controlling processing temperature,
both a simple parabolic power controller and a multi-staged PID controller were used for
microwave power control. However, empirical controller parameters were used and
extensive experimental work was needed to find the parameters whenever there was a
change in material setup. In addition, on-line monitoring of the processing was not

considered.

1.3 Research Topics

To summarize the literature survey, research and development in microwave
adhesive bonding is still in its early stage. Several researchers studied this process in
multimode applicators. Results demonstrated that reduced bonding cycle was obtained

with adequate bonding strength in microwave process compared with thermal bonding.

However, the mechanism of bonding cycle reduction was not proposed. In addition, these
studies did not investigate the fundamental relation of microwave heating characteristics
with the electromagnetic field patterns and material dielectric properties. Moreover,
detailed control and on—line monitoring issues were not considered in these bonding
studies. Some recent studies on microwave processing of composites have developed
methods for controlling the temperature distribution. However, the processing and control
methods were developed with empirical trial and error approaches. These methods are
time—consuming and effort demanding when being generalized to different applications.
Inadequate understanding of process fundamentals and inefficient control and monitoring
methodologies are the major obstacles in industrializing a new technique.

These problems will be addressed in this dissertation. Microwave theories will be
used as a guidance to solve real processing problems. A single mode applicator will be
used to study the microwave adhesive bonding process. There are two major reasons for
using a single mode applicator. First, single mode applicators are more efficient in
coupling microwave energy into materials compared with multi-mode applicators. This
has been shown in the processing of crossply and thick-section graphite fiber reinforced
composites. Second, Process fundamentals can be more conveniently studied in a single
mode applicator compared with a multi-mode applicator. The electromagnetic field
patterns inside the single mode applicator are more predictable and controllable. So
microwave theories can be more conveniently used in the computation of field
distribution and the design of processing, control and on-line monitoring methods.

The first focus of this research is to develop a precisely controlled, rapid and

uniform microwave adhesive bonding system with on-line monitoring feature of the

bonding process in a single mode applicator. The second focus is to investigate the rapid
curing of the adhesive components to reveal microwave heating mechanisms, thermal or
non-thermal.

The research efforts include the assembly of experimental components,
implementation and development of uniform heating methods, development of process
controller based on control theory, and invention of an on-line monitoring technique to
determine the bonding cycle. Special microwave effects observed in microwave adhesive
bonding are further studied to improve understanding in microwave/material interactions
and in microwave heating mechanisms (thermal or non-thermal).

Specifically the following tasks are proposed to support the objective:

1. Assemble Experimental Setup for Microwave Adhesive Bonding

2. Explore the Application of Microwaves in Single Mode Adhesive Bonding and
Compare the Results with Thermal Process

3. Implement Variable Frequency Mode-switching Method and Apply It in Adhesive
Bonding to Achieve Uniform Heating for Large Materials

4. Invent an On-line Monitoring Technique for Microwave Processing in a Single Mode
Cavity Based on Microwave Theories and Experimental Discoveries

5. Design the Process Control System lncorporatin g Theoretical Control Algorithms and
On—line Monitoring

6. Investigate Microwave Heating Mechanisms (Thermal or Non-thermal)

10

CHAPTER 2 BACKGROUND FOR MICROWAVE HEATING

Electromagnetic field strength and distribution patterns are essential factors that
influence microwave heating efficiency and uniformity. They are determined by
microwave operating conditions, applicator dimensions, and material properties and
dimensions. To understand microwave heating characteristics, the fundamentals in

microwave heating are reviewed in this chapter.

2.1 Electromagnetic Fields in a Microwave Enclosure

2.1.1 Maxwell’s Equations

The basic laws governing electromagnetic wave propagation are Maxwell's
Equations, which describe the relations and variations of the electric and magnetic fields,
charges, and currents associated with electromagnetic waves. Maxwell's Equations can be
written in either differential or integral form. The differential form is most widely used to

solve electromagnetic boundary—value problems and is shown in the following equations.

V XE = -%f- (Faraday’s law) (2.1)
3D ,

V x H = J + E (Ampere slaw) (2.2)

V -D = p (Gauss law) (2.3)

V - B = O (Gauss law - magnetic) (2.4)

11

where E is the electric field intensity, H is the magnetic field intensity, D is the electric
displacement density or electric flux density, B is the magnetic flux density, J is the

electric current density, and p is the charge density.

D is defined as:

D = 80E + P (2.5)

where 80 is the permittivity of free space, P is the volume density of polarization, the

measure of the density of electric dipoles.

B can be expressed as:

B = [10(H + M) (2.6)

where #0 is the permeability of free space, H is the magnetic field intensity, and M is the

volume density of magnetization, the measure of the density of magnetic dipoles in the
material.

In a simple isotropic medium, the field quantities are related as follows:

D = GB (2.7)

B = uH (2.8)

where 8 is the permittivity and ,u is the permeability.

12

In addition to the Maxwell's Equations, the Equation of Continuity holds due to

the conservation of electric charge:

V-J+—p=0 (2.9)
a:

In the Maxwell's Equations, only two are independent. Usually Equations 2.1 and
2.2 are used with Equation 2.9 to solve for electromagnetic fields.

Maxwell’s equations are first-order differential equations with E and B coupled.
They can be converted into uncoupled second-order wave equations through

mathematical manipulations [36]:

[V #0 (Ti-MF—IB a: (2.10)

S
32 E _ _V(—)+#-‘?——J
at 8:2

—,uV><JS

where 0’ is the conductivity, and J3 is the source current term. In a source free region,

Equations 2.10 become:

a 82
__ __ .1
[V —,u0 t #8 t:=2]{}0 (21)

Equations 2.1-2.4 are the time-domain representation of Maxwell's Equations. If

the source functions. J (m) and p(r,t) , oscillate with a constant angular frequency of a)

13

in the system, then all the fields will oscillate at the same frequency. The Maxwell’s

equations can be written in time-harmonic form:

rV x E(r) = —ia)B(r)

4 V x H(r) = J(r) + ia)D(r)
V - B(r) == .0(r)

(V - B(r) = 0

(2.12)

 

In time-harmonic case, Equations 2.10 become Helmholtz Equations and Equations 2.11

become Helmholtz equations in source-free region:
2 2 3|! E
[V +a) ,as ]{B}=O (2.13)

where

8* = e(1—i-"—) (2.14)
(08

8* is called the complex permittivity.
2.1.2 Resonant Modes in a Cylindrical Single Mode Cavity

In a cylindrical single mode cavity, there are two types of resonant modes,
transverse electric (TE) and transverse magnetic (TM). For TE modes, the electric field
components are transverse and the magnetic field components are parallel to the direction

of wave propagation, which is the axial direction. For TM modes, the electric field

14

components are parallel and the magnetic field components are transverse to the direction
of wave propagation. Three subscripts, n, p, and q, are used to represent the physical
appearance of the correponding mode in an empty cavity, i.e. TEnpq and Tanq. The
subscript n denotes the number of the periodicity in the circumferential direction,
n=0,l,2. ..; p denotes the number of field zeroes in the radial direction, p=1,2,3...; q
denotes the number of half wavelengths of the equivalent circular waveguide, q=0,1,2. ..
for TM modes and q=l,2,3. .. for TB modes.

The relationship between the frequency and cavity diameter and length for a given
resonant mode can be calculated theoretically. The equations for TB and TM modes in an

empty cylindrical single mode cavity are [37]:

 

2
(f),.pq= 2m J—Jx'np 2 {gm} (2.15)

 

2
(f )an 27715 anp2 + [1?] (2.16)

where f rs frequency, a is cavity diameter, h 1s cavity height, xnp and x' up are tabulated

zeros of the Bessel’s function and the derivative of the Bessel’s function, respectively.
Resonant frequency f can be plotted as a function of cavity length in a mode chart, as

shown in Figure 2.1 and Figure 2.2.

15

 

 

Frequency (GHz)

TE modes in an empty cavity with diameter 0117.78cm

45
- A -Xx9
4 in O. ‘;I‘ ”U

 

 

 

 

I . 0
xx! 2 x
a o.. +.AA:::Ԥw:.gggr.ggrrrrrr
35. 0x0- 0.. + AA a“ '_°O99889
XD- 0.; I . - . 0°
‘ XDE- ‘: cg':i'- '- -., °
3 . o x as._ ++ 8AAU-“azeggg.80
. xx DDDU“'-ii AAAA ..
2.5 " . Xxx DmDDUSEE---A6“88
.... XXXXXXXXX
15 . . . 9Ie¢;. . .
4 6 10 12 14 16

Cau'ty length (cm)

 

+TE011
+TEO12
—o-TE013
+TE021
+1E‘I11
-—+—TE1HZ
_._TE113
—+—TEfiN
—e—TE122
—*—TE2H
+TE212
—*—TEZH
+TE311
.- -TE312

 

 

Figure 2.1 TB Modes in an Empty Cavity with Diameter of 17.78cm

 

 

Frequency (GHz)

TM modes in an empty cavity with diameter of 17.780m

4.5

 

 

4 x 'z' D x t _

 

 

 

X A4
3.5 ‘ .i‘x. .’.:‘Q.Q..§X‘A A A A a,.‘:22£25AA
xx ."Ooi..“*1:0¢oo’lg +::
3 ooooooogooooooooeafiOOO ‘on1880.‘0
oooooooooaggqooooooaoagoooo UIQQQ
25~ XXXXX D00
Xxxxxggflfigg
2AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA§§
1.5 4
1 l l l l I T

 

6 8 1O 12

Length of cavity (cm)

14 16

18

 

+TM012
+TM013
+TM020
+TM021
+TM022
+TM110
1—*—TNH11
+TM112
_._TM120
.._TM121
—e—1M210
+TM211
—+—TMZ12
+TM31O

 

 

Figure 2.2 TM Modes in an Empty Cavity with Diameter Of 17.78cm

16

 

 

 

2.1.3 Electromagnetic Fields in a Cylindrical Single Mode Cavity

In a homogeneous, source-free cylindrical single mode cavity with perfectly
conducting walls, The electromagnetic fields inside the cavity can be derived from
Maxwell's equations and boundary conditions [37].

For TE modes, the EM field components inside an empty cavity are [37]:

 

H -1921”an E --;.”an
p A 33sz p ,0 19¢
z
2
1 la V’npq aWnpq
¢=_;__ E¢=—— (2.17)
z)0 B¢BZ 8p
2
1 3 III
HZ =7(—g—’2pl+kzwnpq) E2 :0
z
z

A

where (p, ¢, 2) are the cylindrical coordinates, z = jwfl ,k2 = 021108 , and 1]!an is the

wave potential for TEnpq modes:

: ii Sin 71¢ . 175 l
V’npq Jn( a p){C08n¢}Sm( h Z) (2. 8)

where a and h are the diameter and height of the cavity, respectively.

For TM modes, the field components inside an empty cavity are [37]:

17

2
-13 Wm H -Larnm

 

 

E
p A 3,032 ’0 ,0 8(1)
y
2
=1}.__3 ”’an H =____a”'"”q (2.19)
¢ "p awz ¢ ap
y
2
1 3 r!
Ez--.-< rm>
y dz

A
where y = jams, and V’npq is the wave potential for Tanq modes:

_ 2,: sin n¢ fl
l//npq — Jn( a p){cosn¢}cos( h Z) (220)

When the cavity is loaded with materials, Equations 2.17 to 2.20 are no longer
applicable. For simple materials, analytic methods are useful for calculating the EM field
inside the materials and the cavity. For complex situations (e.g. inhomogeneous or
anisotropic materials, irregular shapes, etc.), numerical techniques are usually used to
solve for the EM field. The most widely used numerical techniques include the method of
moments (MOM), the finite-element method (FEM) and the finite difference method

(FDM).

l8

2.2 Microwaves/Materials Interactions

2.2.1 Mechanisms of Microwave/Materials Interactions

Materials are classified into conductors, semiconductors and dielectrics according
to their electric conductivity. Conductors contain free charges, which are conducted
inside the material under the alternating electric fields so that a conduction current is
induced. Electromagnetic energy is dissipated into the materials while the conduction
current is in phase with the electric field inside the materials. Dissipated energy is
proportional to conductivity and the square of the electric field strength. Conduction
requires long-range transport of charges. In dielectric materials, electric dipoles are
created when an external electric field is applied and they will rotate until they are
aligned in the direction of the field. The polar segments attempt to align up with the
alternating electromagnetic field so that the normal random orientation of the dipoles
becomes ordered. These ordered polar segments tend to relax and oscillate with the field.
The energy used to hold the dipoles in place is dissipated as heat into the material while
the relaxation motion of dipoles is out of phase with the oscillation of the electric field.
Both the conduction and the electric dipole movement cause losses and are responsible
for heat generation in microwave processing. The contribution of each loss mechanism
largely depends on the types of materials and operating frequencies. Conduction loss is
dominant at lower frequencies, while polarization loss is important at higher frequencies.
Most dielectric materials can generate heat via both loss mechanisms. There are mainly

four different kinds of dielectric polarization:

l9

1.

Electron or optical polarization occurs at high frequencies (close to ultraviolet
range of EM spectrum) [38,39]. It refers to the displacement of the electron cloud
center of an atom relative to the center of the nucleus caused by an external
electric field. When no electric field is applied, the center of positive charges
(nucleus) coincides with the center of negative charges (electron cloud). When an
external electric field is present, the electrons are pushed away from their original
orbits and electric dipoles are created.

Atomic polarization is also referred to as ionic polarization. It occurs in the
infra-red region of the EM spectrum. This type of polarization is usually observed
in molecules consisting of two different kinds of atoms. When an external electric
field is applied, the positive charges move in the direction of the field and the
negative ones move in the opposite direction. This mainly causes the bending and
twisting motion of molecules. Atomic polarization can occur in both non-ionic
and ionic materials. The magnitudes of atomic polarization in non-ionic materials
are much less than that in ionic or partially ionic materials.

Orientation or dipole alignment polarization occurs in the microwave range of
the EM spectrum and is the dominant polarization mechanism in microwave
processing of dielectrics. Orientation polarization is usually observed when
dipolar or polar molecules are placed in an electric field. At the presence of
external electric field, the dipoles will rotate until they are aligned in the direction
of the field. The dipolar rotation of molecules is accompanied by intermolecular
friction, which is responsible for heat generation. Orientation polarization is

fundamentally different from electronic and atomic polarization. The latter is

20

because the external field induces dipole moments and exerts displacing force on
the electrons and atoms, while the orientation polarization is because of the torque
action of the field on the pre-existing permanent dipole moments of the
molecules.

4. Interfacial or space charge polarization occurs in low frequency (radio
frequency) range and is a fundamental polarization mode in semiconductors. This
type of polarization is caused by the migration of charges inside and at the

interface of dielectrics under a large scale field.
2.2.2 Dielectric Properties

Most polymers and composites are non-magnetic materials. For non-magnetic
materials, the electromagnetic energy loss in a given material is only dependent on the
electric field. Incident electromagnetic fields can interact with conductive and
nonconductive materials. The fundamental electromagnetic property of nonmagnetic

materials for microwave heating and diagnosis is the complex permittivity:

£=£'—j£ (2.21)

The real part of the complex permittivity is permittivity 8' (dielectric constant), which is
defined as the ratio between the capacitance of a condenser filled with dielectrics and the
capacity of the same condenser when empty. The higher the polarizability of a molecule,

the higher its dielectric constant. The imaginary part of the complex permittivity is loss

21

factor 8 which is related to energy dissipated as heat in the materials. Usually the

relative values with respect to the permittivity of free space are used:
8 2 80(8', — jag/f) (2.22)

where 80 is the permittivity of free space, 8, is the relative dielectric constant, and Eefl

is the effective relative dielectric loss factor. The loss factor of materials consists of both
polarization and conduction loss. The polarization loss is further contributed by all four
polarization mechanisms mentioned earlier. The effective relative loss factor is

expressed as [39]:

.. .. .. .. .. 0'
self (07) = 8d (a7) + £e(w) + Ea + £3 + :6; (2.23)

where a) = 279‘ , the subscripts d, e, a, and 5 refer to dipolar, electronic, atomic and space
charge polarization, respectively. The loss factor is a function of material structures and
compositions, angular frequency, temperature, and pressure.

The ratio of the effective loss factor to the dielectric constant is defined as the loss

tangent, which is also commonly used to describe dielectric losses:

8
tan (Se = —3’,—f— (2.24)
r

22

When introduced into microwave field, the materials will interact with the
oscillating electromagnetic field at the molecular level. Different materials will have
different responses to the microwave irradiation.

For conductive materials such as carbon fibers and acid solutions, microwave
heating is mainly due to the interaction of the motion of ions or electrons with the electric
field. For conductors with high conductivity, the incident microwaves will be largely
reflected and therefore they can not be effectively heated by microwave. The fields
attenuate towards the interior of the material due to skin effect, which involves the

magnetic properties of the material. The conducting electrons are limited in the skin area
to some extent, which is called the skin depth, d5. Defined as the distance into the

sample, at which the electric field strength is reduced to I/e, the skin depth is given by

[37]:

l

(1/ manage)“ 2

 

ds = (2.25)

where a) is the frequency of the EM waves in rad/sec, #0 is the permeability of the free

space, 4 7: x 10‘7 H/m, ,u' is the relative permeability, and o is the conductivity of the
conductor in mhos/m. For graphite, o = 7x104 mhos/m, and d, = 38.4 ,u m at 2.45 GHz

in free space. As frequency increases, the skin depth decreases. For a perfect conductor,
the electric field is reflected and no electric field is induced inside a perfect conductor at
any frequency. Therefore, no electromagnetic energy will be dissipated even though the

conductivity of the perfect conductor is infinite.

23

For nonconductive materials such as polymers, glass fibers, and Kevlar fibers,
dielectric heating is mainly due to the interaction of the motion of dipoles with the
alternating electric field. For microwave processing of thermosets, the process is self-
adjusting. As the crosslinking occurs, the mobility of dipoles decreases because of the
“trapping” or reaction and the dielectric loss factor decreases. Thus microwave
absorption by crosslinked molecules decreases and microwaves are concentrated in
unreacted molecules. For microwave processing of thermoplastics, the dielectric loss
factor usually increases with temperature and thermal runaway is likely to occur.
Thermal runaway can be prevented by decreasing or even turning off microwave power
at a temperature close to thermal excursion. For microwave processing of polymer
composites, microwave heating selectivity depends on the magnitude of dielectric loss
factor of polymers and fibers. When non-conducting fibers (such as glass) are used,
microwaves will selectively heat the polymer matrix. When conducting fibers (such as
graphite) are used, energy is preferably absorbed by the conductive fibers and heat is
conducted from the fiber to the matrix. In this case, loss factor is mainly due to the fiber
conductivity and can not be used to diagnose the curing process of the low loss matrix
materials.

Dielectric measurement of epoxy has shown that both the dielectric constant and
dielectric loss factor increase with temperature and decrease with extent of reaction [1].
This dependence on temperature and extent of reaction is non-linear. During microwave
processing, the dielectric properties change as a result of heating and reaction. This
affects the electrical field strength and power absorption in the materials. The change in

electric field and power absorption in turn affects the temperature and extent of reaction

24

inside the materials. Thus the modeling of microwave heating is a highly coupled non-
linear problem, which involves Maxwell’s equations for solving the electric field
strength, a heat transfer equation for obtaining the temperature distribution inside the
material, and a reaction kinetics equation for calculating extent of reaction. Reference

data in [1] can be used in the model.

2.3 Applications of Microwave Energy

The use of microwave energy can be classified as either a communication or non-
communication application. Since the discovery of electromagnetic waves, it has been
widely used in communications. The earliest reported non-communication use of
microwaves in polymer processing was in 1940 in an attempt to cure plywood cement
[40]. In the 1960s, microwave processing was successfully applied in the vulcanization of
the rubber in the tire industry [41]. By now, the vulcanization of extruded rubber
weather-stripping for the automotive and construction industries has been one of the most
successful applications of microwave heating in industry [42]. Since the mid-19803, there I
has been a resurgence of interest in the microwave processing of polymers and
composites [3,10,16,17,43,44]. Up to now, microwave heating has been used in forest
industry [36], food industry [41,46-51], waste treatment [52-55], organic synthesis
[56,57], material processing [l-28,43,58-33, 81-86], welding [78,79] and adhesive
bonding [29-3 1,80] applications.

The application of microwave heating in material processing includes the
processing of coal [58], ceramics [59] and polymer and composites [l-28,43,60-33, 81-

86]. In polymer processing, both thermosets and thermoplastics have been studied. The

25

drying process of various thermoplastics with microwave energy has been investigated
[2]. Microwaves have also been used to cure thermosetting materials, including
polyesters [3,4], polyurethanes [5,6], polyimides [7,8,9] and epoxies [1,3,10-20]. Both
pulsed and continuous microwave curing of epoxies have been studied [1,15].

In composite material processing, both conducting fiber (such as graphite) and
non-conducting fiber (such as glass) reinforced composites have been processed with
microwave energy. Lee et. al. [22] processed graphite fiber-epoxy laminates up to 32
plies using a waveguide and a multimode applicator (common microwave oven) and the
attempt by this group to process multi-directional graphite-epoxy composites was not
successful. However, Wei et al. [23] used a cylindrical, tunable single-mode cavity and
successfully processed both cross-ply and unidirectional 24-ply 7.62 X 7.62 cm Hercules
AS4/3501-6 graphite/epoxy prepreg laminates with 2.45 GHz microwaves. The
unpressurized microwave processed composites showed higher modulus with shorter
cure time compared with thermal autoclave process. Part of the reason is that microwave
heating environment can substantially increase the amount of chemical interaction
between the surface of the conducting fiber and the epoxy resin and amine components of
the matrix [21]. As a result, the composite performance can be improved. Thick‘section
graphite/epoxy composites were successfully processed or heated using single-mode
cylindrical cavity [23,64]. Microwaves have also been used to process non—conducting
fiber reinforced epoxy composites. A 457 mm long, 127 mm OD epoxy/glass filament
wound tube with a wall thickness of 9.5 mm was processed in one minute using a
rectangular multi-mode cavity at a power level of 20 KW [81]. Different applicators

were used to process planar glass fiber/epoxy laminates [22][82][83]. However, there

26

was no evidence of improved fiber/matrix bonding by microwaves for glass reinforced
composites [21].

Continuous processing has also been studied and developed for composites with
conducting and non-conducting fibers [24-28, 84]. The control of microwave leakage at
the entrance and exit ports is the critical issue in continuous microwave processing.
Usually an anti-leakage structure called a choke is used at the entries. However, for
conducting fiber-reinforced composites, a specially designed microwave leakage jacket
should be used [85]. In recently developed microwave-assisted pultrusion processes, the
length of the process chamber, the processing time and the pulling forces were reduced
significantly [28, 84 ,86].

Demonstrated results in microwave processing of polymers and composites
include increased polymerization rate [7,11-14], reduced drying time for pelletized
polycarbonate and polypropylene [2], increased glass transition temperature (Tg) for
cured epoxy [10], enhanced fiber/matrix adhesion in carbon composites [21], and
increased mechanical strength of graphite/epoxy composite [15].

Microwave energy has been applied in welding thermoplastics in a multi-mode
oven at a fixed frequency of 2.4SGHz [78]. Conductive polyaniline was placed at the
joint interface as microwave absorber. The absorber heats up and conducts heat into the
thermoplastic bulk materials to form molten layers, which are then squeezed by applied
pressure to form the joint. The initial heat generation rate and electric field strength were
estimated. Effect of heating pressure, heating time and welding pressure on weld strength
was investigated. An impedance matching microwave system was developed to increase

the energy transfer efficiency [79]. Two different conductive polyaniline samples were

27

studied in the welding. Compared with multi-mode system, the impedance matching
system reduced the input power as well as the welding cycle.

Radio frequency or microwave adhesive bonding has been investigated. Bernard
et al. [80] applied a 50 £2 radio frequency technology in adhesive bonding process in the
automotive industry. An adhesive formulation was determined to adapt to the 50 52 radio
frequency technology. An applicator was developed to maintain the two parts to be
bonded in position. Compared with the conventional thermal process, the 50 52 radio
frequency process has shown increased throughput, lower power requirement, less space
occupied by the facilities and reduced cost. Microwave adhesive bonding was studied in a
multi-mode applicators for automotive application [29-31] and bonding of flip-chips (FC)
for electronic packaging application [31]. In automotive application, compared with
thermal bonding, microwave techniques reduced the required curing time of the adhesive
significantly while obtained an equal or slightly higher ultimate tensile strength of the
bonded assemblies. In electronic packaging application, both VFM curing of the
underfill and bonding of the chips were performed. The VFM curing cycle was much
shorter than the thermal curing cycle. In Flip-Chip (FC) bonding experiments, high IC
integrity of the bonded chip was obtained with VFM process. In addition, VFM cured
samples had 50% reduction in stress compared with thermal cured assemblies.

Research work throughout the world has shown that microwave techniques have
great potentials in a variety of industrial applications. However, microwave techniques
have not been widely adopted by industry. Part of the reason is because of inadequate
knowledge in microwave/material interactions, design of processes for specific

applications, and intelligent control and automation. Research in this area is expected to

28

elucidate microwave fundamentals and enhance the commercialization of microwave

technique [87].
2.4 Microwave Applicators

The most commonly used microwave applicators include waveguide, commercial
multimode microwave ovens, and single mode applicator. Waveguides are hollow metal
tubes, the hi gh-reflectivity walls of which allow microwaves to propagate. Commercial
multimode microwave ovens have large dimensions compared to the operating
wavelength, allowing the establishment of multi modes at the same time. The EM field
inside multimode ovens is not uniform and shows many peaks and valleys. Turntables are
usually used to rotate the materials to be processed for more uniform heating. Single
mode cavity supports one mode at one time and has well-defined electric field pattern.
The single mode cavity system has higher energy efficiency to transfer microwave power
into the processed materials. A cylindrical single mode cavity will be used in this study.
Since one mode heating is not uniform with high field intensity confined to small regions,
variable frequency techniques can be used to excite several modes with complementary
heating patterns sequentially to obtain more uniform heating. This method will be
discussed in detail in Chapter 4.

A sketch of the cylindrical single-mode cavity used in this study is shown in
Figure 2.3. The cavity is made out of a length of metal circular waveguide with both ends
shorted by the same metal, which is brass in this study. The cavity has an inner diameter
of 17.78cm with cavity length adjustable from 10 cm to 30 cm. Microwave energy is

introduced into the cavity by a coaxial coupling probe. The coupling probe is side

29

mounted 3cm above the base of the cavity. The probe is adjustable in the radial direction
so that the coupling probe depth Lp can be changed for locating critical coupling
conditions. The range of the probe depth is from 0 mm to 50 mm. The top short-plate is
adjustable so that the cavity length Lc can be changed. The bottom plate is removable for
sample loading. Both the top and the bottom plates are shorted with the cavity wall by

metallic finger stocks.

[I I]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Coupling Probe I Shorting Plate I
A
Microwaves '—_'
—_.> I 11- 14C
:' Lp
<-> V

 

 

 

 

Bottom Plate 1 .

Figure 2.3 Cylindrical Single-Mode Resonant Cavity

2.5 Temperature sensing system

Two types of thermometers are used in this study:
1. Multi-channel LUXTRON fluoroptic thermometer: The Luxtron thermometer
uses Decay Time Technology to measure the temperature of the sensor [88].
Luxtron sensors contain a small amount of magnesium fiuorogerrnate. The

sensors are attached at the tip of the optic fiber. The optical system excites the

30

sensor with blue light. In turn the sensor fluoresces a red light, the intensity of
which decays exponentially with time. The time constant of the decay is inversely
proportional to the temperature. So by measuring the decay time, the temperature
can be obtained. The temperature probes are electrically nonconductive. So they
will not perturb or be perturbed by the microwave fields.

. Multi-channel NoEMI-TS fiberoptic thermometer: The working principle of this
type of thermometer is based on the absorption of light by a semi-conducting
crystal bonded to the end of an optical fiber [89]. The crystal is in well contact
with the materials to be processed with microwaves. As the crystal temperature
increases, more low-energy photons are captured and absorbed by the band. The
absorption edge is moved towards the longer wavelengths. Therefore, measuring
the position of the absorption shift gives a measurement of the crystal's
temperature and thus the temperature of the materials. The sensor is immune to

and does not perturb the electromagnetic field.

31

CHAPTER 3 SINGLE MODE MICROWAVE ADHESIVE BONDING

3.1 Introduction

Chapter 1 described the advantages of microwave energy over thermal heating.
Chapter 2 reviewed the fundamentals in microwave heating. This chapter applies single
mode microwave method in adhesive bonding to utilize the advantages of selectivity
(selective heating of adhesive if the adhesive is properly formulated), rapidity and high
efficiency of microwave heating. The feasibility and characteristics of using microwaves
in adhesive bonding in a single mode applicator are studied.

Since microwave heating is strongly dependent upon the magnitude of dielectric
properties of the constituents, microwave energy can be applied in different processes to
perform either volumetric, surface or selective heating. The selective heating feature of
microwaves is utilized in adhesive bonding process. The polymer or composite parts have
low loss factors relative to the adhesive material at the interface. Microwave energy is
then directly transferred into the interface and cures the adhesive extremely rapidly, even
though the substrate materials are very thick.

Research & Development regarding the application of microwaves in adhesive
bonding is still in its early stage. Several investigators have studied microwave adhesive
bonding in multi-mode applicators [29-31]. Paulauskas et al. [29] studied fixed frequency
microwave adhesive bonding in a multi-mode applicator. Glass and glass fiber reinforced
urethane-based composite (SRIM-part) were adhesive bonded with Goodrich 582E

(epoxy based) adhesive. The relationship between input power level and bonding time

32

was studied. Additives such as carbon black were added into the adhesives to improve the
lossy characteristics of the adhesive. It was reported that during microwave bonding, the
coupling of microwave energy into the adhesives was highly efficient. The coupling was
further enhanced with the additives. Compared with thermal adhesive bonding, fixed
frequency microwave bonding techniques reduced the required curing time significantly
while obtained an equal or slightly higher ultimate tensile strength of the bonded
assemblies. However, the temperature was not measured or controlled in this study.
Therefore, the comparison between microwave and thermal processes might not be based
on the same heating conditions. In addition, it was unclear whether the reduction of
curing time with carbon black additives was because of increase of heating rate or curing
reaction rate or both.

Paulauskas et al. [30] also applied variable frequency microwave (VFM) in
adhesive bonding in a multi-mode cavity. The frequency was swept over a range to
launch different electric field patterns into the applicator sequentially. An advantage of
VFM bonding over fixed frequency technique was that VFM had a more uniform time-
averaged electric field. Thus VFM did not readily create localized overheating in the
adhesive or the substrates and was not sensitive to sample position. Compared with fixed
frequency microwave bonding, VFM reduced the bonding time slightly.

Wei et al. [31] performed VFM bonding of polymer composites for automotive
application and bonding of flip-chips (FC) for electronic packaging application. A multi-
mode applicator was used. In the automotive application, SRIM (chopped glass fiber
reinforced, urethane-based composite) and Rynite® 5309 (glass fiber/mineral reinforced

thermoplastic polyester composite) panels were studied. A closed-loop feedback control

33

was applied to control the temperature. Reduced bonding cycle was achieved with
adequate bonding strength. Uniform bonding was observed. In electronic packaging
application, both VFM curing of the underfill and bonding of the chips were performed.
The VFM curing cycle was much shorter than the thermal curing cycle. In Flip-Chip (FC)
bonding experiments, high IC integrity of the bonded chip was obtained with VFM
process. In addition, VFM cured samples had 50% reduction in stress compared with
thermal cured assemblies.

These studies on microwave adhesive bonding in multi-mode applicators have
shown the industrial potential of using microwaves in adhesive bonding processes.
However, process fundamentals have not been investigated in these studies. Research
needs to be carried out to relate microwave heating characteristics such as uniformity
with the electromagnetic field patterns. In addition, material dielectric properties largely
determine the heating selectivity in microwave adhesive bonding process. Microwaves
can be selectively coupled into the adhesive if the dielectric properties of the adhesive are
higher than that of the substrates.

In order to study the process fundamentals, a single mode resonant applicator will
be used for microwave adhesive bonding in this research. The electric field patterns
inside the single mode applicator are more controllable and predictable compared with
field patterns inside multimode ovens. In a single mode microwave cavity, two
processing methods can be applied, namely single mode microwave processing and
mode-switching microwave processing. Single mode microwave method refers to the
process that only one mode is used throughout the processing. Only limited material size

can be uniformly heated with single mode microwaves because of the non-uniformity of

34

electric field distribution. Mode-switching microwave method refers to the process that
several modes with complementary heating patterns are excited sequentially to obtain
time-averaged uniform heating. The modes can be switched by either mechanically
changing the volume of the cavity [33] or tuning the operating frequency [35]. Mode-
switching method is based on knowledge and understanding of single mode heating
characteristics. As the starting point, this chapter studies single mode microwave method
in adhesive bonding. The heating selectivity, rapidity, controllability and mode
characterization will be investigated in this chapter. In Chapter 4, the mode switching
method in [35] will be implemented and applied in adhesive bonding process to improve

heating uniformity.

3.2 Experimental

3.2.1 Experimental Circuit

The experimental circuit was assembled for microwave adhesive bonding process.
The circuit directs microwaves into the applicator, allows the measurement of
temperature, incident and reflected powers, and reduces the power reflected back to the
power source to prevent damage to the power source. The microwave circuit is illustrated
in Figure 3.1.

Microwave signal generator is a sweep oscillator (HP83SOB) connected with a RF
plug-in (HP86235A). A variable frequency amplifier (Lambda LT-1000) is used to
amplify the signal. The amplified power signal is in the range from 0 to 200 Watts.

Microwave frequency can be adjusted from 2 GHz to 4 GHz either manually or

35

automatically. A 3-port circulator is used to prevent the reflected power from damaging
the power source. The input and reflected microwave powers are decoupled with 20db
directional couplers (N arda 3043-20) and measured with power meters (HP435B). A
dummy load is used to absorb most of the reflected power. A multi-channel LUXTRON
fluoroptic thermometer and a multi—channel Nortech NoEMI-TS fiberoptic thermometer
are used for sample temperature measurement. The probes are electrically nonconductive.

So they do not perturb or be perturbed by the microwave fields.

Microwave Power

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Source Directional Microwave
Circulator Coupler Applicator
N E +
: V ‘
Direction Pr Pi E; Therrnometry
ECoupler Reflected Incident Power [:3 AID
l _ Power Meter Meter 1
l Dur—nmy '
i .......... £42214 .......................... Em- D W
Controller

 

 

 

Figure 3.1 Circuit of Microwave Adhesive Bonding

A cylindrical single mode cavity with a diameter of 17.78cm is used. The
coupling probe is side mounted 3cm above the bottom of the cavity. The cavity length
(LC) and the probe depth (Lp) are adjusted to be 13.2cm and 2.0cm, respectively. The

sample is loaded at the center of the cavity.

36

3.2.2 Materials

Because the major objectives of this chapter are to study the characteristics of
microwave adhesive bonding in a single mode applicator and to study the relation
between electromagnetic field patterns and heating characteristics, commercial substrate
and adhesive materials were used because of the established knowledge of good adhesion
and good mechanical properties. In adhesive bonding, compatibility between the
substrates and the adhesive is required. Epoxy adhesive was selected because of its
compatibility with a wide range of substrate materials. The epoxy adhesive used for this
study was Eccobond A401-37 (epoxy based) from Grace Specialty Polymers. The
substrate was chosen among Zytel® 726 33L (Glass reinforced Nylon66/6 copolymer),
Hytrel® 5556 (Thermoplastic Polyester) and Bexloy W502 (glass reinforced
ethylene/methacrylic acid copolymer). All three substrate materials were obtained from
DuPont. The selection of substrate was mainly based on material dielectric properties,
because microwaves can selectively heat the adhesive without substantially heating up
the substrates if the adhesive has a higher magnitude of dielectric properties than the

substrates.
3.2.3 Method of Dielectric Property Measurement

The dielectric properties of the adhesive and the substrates were measured with
single mode perturbation method developed by Jinder J ow et a1 [91]. The measurement
was performed at room temperature. The microwave applicator used for the dielectric
measurement was a cylindrical cavity with a diameter of 15.24cm. The mode used was

TM012 at 2.4SGHz. Samples were loaded into a cylindrical Teflon holder and placed at

37

the center of the cavity. Teflon was selected to be sample holder because of its low
dielectric loss factor (0.0003), temperature independent dielectric properties, high
temperature resistance (up to 260°C) and chemical inertness [92]. The inner diameter and
height of the holder were 10mm and 35mm respectively. Liquid samples were filled into
the holder and degassed in a vacuum chamber to remove trapped air during the filling
process. Solid samples were machined into cylindrical shape and then loaded into the
Teflon holder. Unloaded cavity refers to the cavity with the empty Teflon holder inside.
Loaded cavity refers to the cavity with resin-filled Teflon holder. The empty and loaded
cavities were tuned to critically couple the external circuit by adjusting the cavity length,
probe depth and sample position. The cavity length corresponding to the TM012 mode
was within the range of 15.4lcm to 15.45cm at 2.45GHz. In each measurement, the
dimensions of the cavity and the sample, resonant frequency and half-power frequency
bandwidth were measured for both unloaded and loaded cavities. The experimental data
were input into a software to calculate the dielectric properties based on the following

perturbation equations [l]:

——f0 ”f5 = (£'—1)ABGZ‘— (3.1)
0 Vc
L __1_ = 25" A3031 (3.2)
Q5 Q0 VC
Where A=Jo (2.405Rs/Rc)2+ J1 (2.405R5/Rc)2 (3.3)
B=l+[Lc/(21th)] sin (2an/Lc) cos (41tH/Lc) (3.4)
G=0.2718(c/fo/Rc)2 (3.5)

38

In these equations, 8' and e" are the dielectric constant and dielectric loss factor.
Vc, Rc and Lc are the volume, radius and length of the unloaded cavity. Vs, Rs and Ls are
the volume, radius and length of the loaded cavity. H is the distance from the middle of
the sample to the bottom of the cavity and c is the speed of light. f0 and f8 are the resonant
frequencies of the unloaded and loaded cavities. Q0 and Q8 are the Q-factors of the
unloaded and loaded cavities, respectively. The Q factor is the ratio of the resonant

frequency to the half-power frequency bandwidth.
3.2.4 Sample Setup and Preparation for Adhesive Bonding

The sample setup was based on single-lap shear configuration for the convenience
of testing bond strength. Sample setup and temperature measuring points are shown in
Figure 3.2.

The thickness of the substrate was 3mm. The length of the substrate panels was
5.08cm, smaller than that suggested by single lap-shear Standard Test Method in order to
fit into the cavity. Two substrate panels had an overlap of 0.8cm along the length
direction, where the adhesive was applied. Preliminary experiments indicated that
adhesive wider than 2.54cm could not be uniformly heated with single mode microwaves.
Because the width of the adhesive was the same as that of the substrates, the width of the
substrates was determined to be 2.54cm for single mode microwave adhesive bonding.
Several glass beads (diameter 0.6mm) were embedded in the adhesive to get a uniform
thickness of the adhesive. Since the temperature of the adhesive between the two
substrate panels can not be directly measured, an auxiliary panel was used for

temperature measurement to approximate the real adhesive curing environment. Adhesive

39

with the same dimension as the bonding area was applied between the auxiliary panel and
panel 2. Two small holes were drilled through the auxiliary panel, where the temperature
probes were inserted to measure the adhesive temperature. The adhesive dimensions were
1.60m x 2.54cm with a thickness of 0.6mm. The entire assembly was 9.36cm x 2.54cm

with a height of around 6.6mm.

Temperature Auxiliary Panel for
Substrate Panel 1 Probe Holes Temperature Measurement

\ a

Glass L U i i L a
Beads
* / / / fl

Supporting Adhesive Release Films Substrate Panel 2
Panel

 

 

 

 

 

 

 

(a) Side View

 

OK

2 o.\

/
Panel \\ Adhesive

Temperature

 

 

 

 

 

 

Temperatures

(b) Top View

Figure 3.2 Sample Setup and Temperature Measuring Points in Single Mode Microwave
Bonding

Before bonding, the panels were roughened at the joint areas with abrasion paper

in order to enhance bonding. The joint areas were then degreased with acetone. A Teflon

40

mold was used to hold the sample in position. After bonding, the samples were tested

with single lap shear equipment to determine the bonding strength.

3.3 Results and Discussion

3.3.1 Material Dielectric Properties

The results of dielectric measurement are listed in Table 3.1. Three measurements

were made for each sample.

Table 3.1 Dielectric Properties of the Materials

 

 

 

 

 

 

 

Dielectric Dielectric Loss
Constant Factor
8’ 8”
Adhesive Eccobond Uncured 4. 1010.20 0.307i0.015
A401-37 Fully cured 2.58:0.12 0029:0001
Substrate Zytel® 720 33L 3.09:0.15 0034:0002
Substrate Hytrer® 5556 2.60:0.13 0094:0005
Substrate Bexloy W502 2.11:0.10 0.0061000]

 

 

 

 

 

 

Among the three substrates, the dielectric constant and loss factor of Bexloy

W502 (glass reinforced ethylene/methacrylic acid copolymer) were the lowest. In

41

addition, the magnitude of the dielectric properties of the Bexloy substrate was lower
than both the uncured and fully cured adhesive. The dielectric property of Zytel® 72G
33L (Glass reinforced Nylon66/6 copolymer) was lower than the uncured adhesive but
slightly higher than the fully cured adhesive. It can be expected that selective heating of
the adhesive can be obtained at the beginning of curing but will be less significant at the
end of microwave adhesive bonding of the Zytel substrate assembly. The dielectric loss
factor of Hytrel® 5556 (Thermoplastic Polyester) substrate was one third of that of the
uncured adhesive but three times of that of the fully cured adhesive. Therefore after
certain point during the adhesive curing, microwaves will be preferentially absorbed by
the substrates instead of the adhesive. For microwave adhesive bonding of Hytrel
substrate, adhesives with higher dielectric properties should be used to increase the
coupling efficiency of microwave energy into the adhesives.

Because of its low dielectric properties, Bexloy W502 (glass reinforced
ethylene/methacrylic acid copolymer) was selected for the preliminary microwave

adhesive bonding study.
3.3.2 Characterization of Empty Cavity

Empty cavity characterization was performed in the cylindrical single mode
cavity with the diameter of 17.78cm and cavity length of 13.2cm. The characterization
was performed at room temperature. The frequency was swept from 2.0 to 4.0GHz to
study the mode spectrum. The mode spectrums were obtained with measuring the
incident power (Pi) and the reflected power (Pr) as a function of frequency. The

frequency with minimum reflectance (Pr/Pi) is the resonant frequency of a mode. The

42

cavity characterization program was edited with LabVIEW. This program will be
explained in Chapter 6.

Figure 3.3 illustrates the mode spectrum of the empty cavity. The resonant
frequencies of the modes can be determined with mode diagnosis described in next
section. The experimental resonant frequencies in the empty cavity were close to the

theoretical values shown in the mode chart Figure 2.1 and Figure 2.2.

 

 

 

0 . . . ,
2 2.5 3 3.5 4
Frequency (GHz)

 

 

 

 

Figure 3.3 Characterization of Empty Cavity

As an example, the experimental and theoretical empty cavity resonant
frequencies of TM020, TM022 and TM212 modes are shown in Table 3.2. The relative

deviation from the theoretical value was 0.6% for TM020, 0.5% for TM022 and 0.7% for

TM212.

43

 

 

 

 

 

 

Ta

{film
Thes

(life;

30993

SPEC:

Table 3.2 Theoretical and Experimental Values of the Resonant Frequencies of Several

 

 

 

 

Modes
Mode Mode Mode
TM020 TM212 TM022
Resonant Frequency in 2.9830 3.5990 3.756
Empty Cavity (F2) (GHz)
Theoretical Resonant 2.9647 3.5741 3.7356
Frequency in Empty Cavity
(Fr) (GHZ)
Relative Deviation from 0.6% 0.7% 0.5%
Theoretical Value
((F2-Ft)/Ft* 100%

 

 

 

 

 

 

3.3.3 Characterization of Loaded Cavity and Mode Diagnosis

3.3.3.1 Characterization of Cavity Loaded with Materials

The mode spectrum of the cavity loaded with materials was also studied in the
cylindrical single mode cavity with the diameter of 17.7 8cm and cavity length of 13.2cm.
The sample was placed at the center of the cavity. Different material setups should have
different spectrums. Figure 3.4 shows the mode spectrum of the cavity loaded with the
assembly consisting Bexloy W502 substrates (glass reinforced ethylene/methacrylic acid
copolymer from DuPont) and Eccobond A401-37 adhesive (epoxy based from Grace

Specialty Polymers).

 

 

 

 

 

1.5:
E 1—
cl:
0.5~
0 I I r I I 1 I T II
2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4

Frequency (GHz)

 

 

 

Figure 3.4 Reflectance in Cavity Loaded with Samples (Sample Width of 1")

The mode spectrum of the loaded cavity was different from that of the empty
cavity because of the perturbation of materials. As observed from Figure 3.4, many
resonant modes can be excited within the loaded cavity. In order to locate the modes that
could rapidly heat up the sample, the temperature rise was measured as a function of
frequency and then the heating rate at different frequencies was calculated. Both the
temperature data acquisition and the heating rate calculation were performed with a
LabView program. In this experiment, the step increase of the microwave frequency was
1MHz/second. This small step increase was used to assure that the sample, after being
heated up by a heating mode, could cool down before the next heating mode was excited.
Thus the temperature rise of the adhesive at different frequencies was always relative to
the room temperature. The results of heating rate in the frequency range of 2 — 4GHz are

shown in Figure 3.5.

45

 

 

 

ma
char
in If:
a M
“is

the g

‘WE ;

 

E"
u:

 

A 2-
&
U
V1.5—
8
C3
95 14
<3
:i:’
0.54

 

 

 

 

 

O

2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4
Frequency (GHz)

 

 

 

Figure 3.5 Heating Rate at Different Frequencies

From the results shown in Figure 3.4 and Figure 3.5, it can be observed that the
modes at 2.555, 2.771, 3.256, 3.396, 3.492, 3.715-3.730 and 3.837GHz were
characterized with low reflectance and high heating rate. Thus these modes can be used
in the single mode adhesive bonding study. In addition, the mode at 3.715-3.730GHz had
a wide resonant frequency band and the power reflectance within this frequency range
was always zero. In the adhesive bonding process, the temperature rise and the curing of
the adhesive lead to changes in the dielectric properties, which in turn result in the shifts
of the resonant state. The operating frequency has to be frequently tuned to maintain the
resonant state if the cavity dimensions remain constant. However, if the resonant
frequency band of a mode is wider than the shift of the resonant frequency, then the
operating frequency does not need to be tuned and the heating is continuously at its
highest efficiency. Therefore the mode at 3.715-3.730GHz was chosen for the single

mode microwave adhesive bonding study.

3.3.3.2 Mode Diagnosis

To understand the electric field distribution pattern inside the materials, the mode
at 3.715-3.730GHz was diagnosed in the loaded cavity. The resonant frequency changed
slightly in each experimental run due to the slight change in sample size and position.
The same names (Tanq or TEnpq) are used for modes in loaded cavity and in empty
cavity, though the field distributions in loaded cavity are slightly different from that in
empty cavity. The diagnosis includes two parts — characterization of the heating pattern
and measurement of the half wave number q. The major assumption used in mode
diagnosis was that after the sample was loaded into the cavity, the change of resonant

frequency was within 10% of the resonant frequency in the empty cavity [90].

33.3.2.1 Characterization of Heating Pattern

To characterize the heating pattern of the mode, the distribution of adhesive
temperatures was measured as a function of time. Figure 3.6 shows the adhesive
dimensions and the arrangement of the temperature probes in the adhesive. The adhesive
was 48mm x 48mm. T5 was the center temperature of the adhesive. T3 and T4 measured
the temperatures 12mm away from the center point and along the coupling probe
direction. T1 was the temperature 24mm away from the center point and along the
coupling probe direction, and T2 was 24mm away from the center point and
perpendicular the coupling probe direction. The temperature distribution is shown in
Figure 3.7. The temperature decreased from the center to the edge, indicating that the

mode at 3.715-3.730GHz preferentially heated the center of the adhesive.

47

T1
f

 

. T3 24mm
1 2mm

1:
T2 ’1me .
L

T5

T4

 

 

 

H

24mm

 

3 Coupling Probe
i
Figure 3.6 Dimension of the Adhesive and Location of Temperature Probes

 

 

90
80 — Center

70 - T5 4
60 ~
50 ~
40 a
30 - _ ‘
20 5'
10

 

Temperature ( C)

 

 

 

Time (min)

 

 

 

Figure 3.7 Experimental Temperature Distribution

48

3.3.3.2.2 Measurement of Half Wave Number q

In the determination of the half wave number q, the relative radial electric field
strength was studied along the axial direction. The experimental configuration is shown

in Figure 3.8. A small metal probe was inserted through the holes in the cavity wall to

measure the power Pb near the cavity wall, which is proportional to the square of the

radial electric field (szKblEplz) [1]. The diameter of the probe needle was 0.5mm. The

insertion depth was around 1mm. The perturbation of the diagnostic probe to the

microwave field was assumed to be negligible.

 

 

Diagnosis :33:
Probe ::::::1

 

 

 

 

 

 

 

 

 

 

Figure 3.8 Experimental Configuration for Measurement of Half Wave Number Q

At p=a, (p=a, The relation between radial electric field strength EF and q is:

E p cc qsin(qrt%) for TM modes; (3.6 )

49

E p oc sin(q7z;zl-) for TE modes (3.7)

Where p is the radial position, (p is the angle from the coupling probe, 2 is the
axial position from the bottom of the cavity, Ep is the radial electric field strength, q is the
half wave number, and h is the cavity height.

With measuring |Ep|2 as a function of z, the number of half waves gave the value

of q. Figure 3.9 shows the result of the half wave number measurement for the mode at
3.715-3.73OGHz. There are totally two half waves along the axial direction inside the

cavity, thus q is 2.

 

 

 

 

 

Radial Electric Field Along Axial
Direction
12

E

8

o 10 «

m

E 8 ~

>

8

E 6

2

.- 4 ..

8

C

.3 2 ‘

E

O I I T I
0 1 2 3 4 5
Relative Radial Electric Field Strength

 

 

 

Figure 3.9 Measurement of the Half Wave Number Q

50

Therefore, the mode at 3.715-3.730GHz was center-heating and had a half wave
number of 2. With the assumption that the change of resonant frequency after the sample
being loaded into the cavity was within 10% of the resonant frequency in the empty
cavity, the mode was diagnosed as TM022.

The resonant frequency of TM022 mode in the empty cavity was 3.756GHz as
shown in Table 3.2. The resonant frequency in the loaded cavity was 3.723GHz. Thus the
resonant frequency shifted down slightly (around 0.9%) when the small sized materials
were loaded into the cavity. This is consistent with the theory that an increase in the
dielectric constant leads to a decrease of the resonant frequency [37, Chapter 7]. The
mode pattern inside the empty cavity was computed with a Matlab program and shown in

Figure 3.10.

 

Figure 3.10 Electric Field Distribution of TM022 Mode at the Cross-Section of the
Cavity (Lighter Region Represents Higher Electric Field Strength)

51

The Matlab program is shown in Appendix A. In Figure 3.10, the lighter region

represents higher electric field strength, indicating that TM022 is a center-heating mode.

3.3.4 Single Mode Microwave Adhesive Bonding

The center heating mode TM022 was applied in single mode microwave adhesive
bonding. This mode in the loaded cavity was characterized with a wide band of resonant
frequency with zero reflectance. The center frequency in the resonant frequency band
was used for single mode microwave heating, which was around 3.725GHz.

The assemblies were bonded at 100°C, 110°C and 120°C, respectively. The
temperature profiles during microwave adhesive bonding at all the three temperatures

were similar and the profile at 120°C is shown in Figure 3.11.

 

 

 

 

 

 

140
Adhesive l

120 -

100 .. Adhesvie 2
6 80 _
3
1:
{-1 60 I Panel

40 ~

20 -

O T I f I I I
0 5 10 15 20 25 30 35
Time (min)

 

 

Figure 3.11 Temperature Profile during Single Mode Microwave Adhesive Bonding

52

The temperature of the adhesive was measured at two points close to the edges of
the bonding assembly. The adhesives were heated from room temperature to 120°C in
around 4 minutes and then kept isothermal. During the heating up process, the adhesive
temperatures at the two measuring points had very slight difference (with 1°C). In the
isothermal bonding phase, the temperature difference between the two measuring points
increased. However, the temperature difference was less than 5 °C. During the whole
heating process, the temperature of the adhesive was much higher than that of the panel.
The selective heating of the adhesive was because the dielectric loss factor of the
adhesive was much higher than that of the panel and microwaves were concentrated in

the adhesive and the interface.

3.3.5 Comparison between single mode microwave and thermal adhesive bonding in

terms of bonding time and strength

Thermally bonded assemblies were also prepared under normal pressure for
comparison purposes. Autoclave was not used because only contact pressure was
required to cure the adhesive, as indicated in the adhesive MSDS. Preparation of samples

was the same as that used in microwave process. Samples were bonded at 120°C in a

thermal oven for up to 140 rrrinutes. The oven was preheated to 120°C before samples
were put in.

The mechanical properties of microwave bonded and thermally bonded
assemblies were tested with single-lap shear equipment. The crosshead displacement rate
was 0.05 inch/minute (0.127cm/minute). Three assemblies were tested for each data

point. Results are shown in Figures 3.12 and 3.13 and Tables 3.3 and 3.4. Figure 3.12

53

shows that microwave energy reduced the bonding time and enhanced the bonding
strength compared with thermal method. In addition, the bonding time was shorter at
higher temperature for microwave process. At 100°C, 110°C and 120 °C, the isothermal
bonding time was about 40, 20 anle minutes, respectively, in microwave adhesive
bonding. The highest bond strength was close to 6MPa. In thermal heating, the bond
strength increased with time at first and reached the maximum value (around 3MPa) after
100 minutes at 120 °C. The average standard deviation of the bond strength was

0.67MPa and 0.32MPa for microwave and thermal process, respectively.

 

 

 

 

 

 

 

 

 

7

6 ..
’8 +Microwave
E 5 : at 100C
5 4 - -B-Microwave
g0 at 110 C
g 3 - x + Microwave
'2 x at 120 C
33 2 4 x X TThermal at

120 C
1 a
0 r e
0 50 100 150
Time (minute)

 

 

 

Figure 3.12 Strength of the Assemblies Bonded at Different Conditions

The difference in bond strength between microwave and thermal processes is

related to the break pattern in single lap shear test. Table 3.3 summarizes the break

54

pattern of microwave bonded assemblies. With sufficient bonding time, microwave
bonded assemblies at different temperatures all broke within the substrates. This
phenomenon indicates that the bond was stronger than the substrate material itself in
microwave process. Table 3.4 shows the break pattern of thermally bonded assemblies.
The assemblies broke at the interface, indicating that the adhesion between the substrates

and the adhesive was not strong enough in the thermal process.

Table 3.3 Break Pattern of Microwave Bonded Assemblies

 

 

 

 

 

 

 

 

 

 

t(min) T=100 °C T=110 °C T=120 °C
5 not bonded mixed (interface mixed (interface
and panel) and panel)
10 interface mixed (interface shear in panel
and panel)
20 mixed (interface shear in panel shear in panel
and panel)
30 mixed (interface shear in panel shear in panel
and panel)
40 shear in panel shear in panel shear in panel

 

55

 

Table 3.4 Break Pattern of Thermally Bonded Assemblies

 

t(min) T=120 °C

 

Up to 150 min interface

 

 

 

 

Figure 3.13 shows the load-elongation curves of the assemblies in the single lap-
shear test. Microwave bonded assemblies had large elongation before breaking. This was
because the strength of the microwave bonded assemblies was determined by the
substrates. Thermally bonded assemblies had small elongation before breaking, because

the assemblies broke at the interface.

 

 

 

 
  
  

 

 

 

 

 

Load-elongation curves
350 a: Microwave
300 A bonding, 40m'n
A at 110 C
250 . j
.9: 200 . b: Thermal
g bonding, 150min
3 150 I a at 120 C
100 e
50 b
0 r r r k
0 0.5 1 1.5
Elongation (inch)

 

 

 

Figure 3.13 Load-Elongation Curves of Assemblies Bonded under Different Conditions

56

The maximum standard deviation of shear strength was 0.8MPa using the single
lap shear technique in this study. This large experimental error made the design of
bonding cycle uncertain. Further study of extent of cure vs time or on-line monitoring is
necessary to determine the bonding cycle.

In this preliminary microwave adhesive bonding study, observed microwave
effects include reduction of bonding time and enhancement of bonding strength for the
Eccobond A401—37 (epoxy based) / Bexloy W502 (glass reinforced ethylene/methacrylic
acid copolymer) system. The possible reasons causing the reduction of bonding time will
be studied in Chapter 7. The possible reasons causing the enhancement of bond strength
are proposed as follows. First, microwave modification of substrate surface chemistry.
Exposure of materials to microwaves might increase the surface energy for better
adhesion. Second, microwave enhancement of molecular mobility at the interface.
Microwaves were concentrated at the interface and might cause the superheating
phenomenon. The Bexloy substrate was thermoplastics based composite materials. When
thermoplastics are bonded with compatible adhesives at a sufficient temperature (thus
sufficient mobility of materials at interface), diffusion bonding might be one of the
possible bonding mechanisms [93]. The enhancement of molecular mobility of both the
adhesive and the substrate could increase the interdiffusion between the adhesive and the
substrates.

However, adhesion between two polymeric materials with microwaves is a
complex phenomenon that involves multidisciplinary knowledge of microwave
fundamentals, surface chemistry, polymer properties, and so on. Extensive experimental

investigations are needed to improve understanding of microwave adhesive bonding

57

mechanisms. The study of microwave bonding mechanisms is not included in the

research scope of this dissertation.

3.4 Conclusions

This chapter studies the feasibility and characteristics of using single mode
microwave method in adhesive bonding of polymer composites. Adhesive and substrate
materials were selected based on their compatibility and dielectric properties. An epoxy-
based adhesive was used because of its compatibility with a wide range of substrate
materials. The substrate was chosen from three polymer composites based on dielectric
prOperties. With the choice of the materials, selective heating of the adhesive was
observed in microwave adhesive bonding process.

To locate the heating modes for adhesive bonding, empty and loaded cavities
were characterized to find the mode spectrum. Among many available modes, a mode
with a wide resonant frequency range (from 3.715 to 3.730GH2) was selected to use in
the bonding process. This mode was diagnosed to be TM022 by characterizing the
heating pattern and measuring the half wave number q.

The bonding results showed that microwave method reduced the bonding time
and enhanced the bonding strength compared with thermal method. In addition, the
bonding time was shorter at higher temperature for microwave process. The difference in
bond strength between microwave and thermal methods is related to the break pattern in
the single lap shear test. With sufficient bonding time, microwave bonded assemblies at
different temperatures all broke within the substrates. This phenomenon indicates that the

bond was stronger than the substrate material itself in microwave process. Thermally

58

bonded assemblies all broke at the interface, indicating that the adhesion between the

substrates and the adhesive was not strong enough.

59

CHAPTER 4 VARIABLE FREQUENCY MODE-SWITCHING MICROWAVE
ADHESIVE BONDING

4.1 Introduction

Chapter 3 studied microwave adhesive bonding with one mode throughout the
process. The non-uniform electric field pattern of one mode limited the material size that
could be uniformly heated. This chapter implemented the variable frequency mode-
switching method in [35] and applied the method in microwave adhesive bonding of
large-size materials to obtain uniform heating in a single mode applicator.

Uniform adhesive bonding has been achieved in multi-mode applicators with a
frequency-sweeping technique [29-31]. In this technique, the frequency is rapidly swept
over a selected range to launch different electric field patterns into the applicator
sequentially. If the frequency range is well selected and wide enough, then the overall
heating can be uniform. However, this method has poor power efficiency and the
processing frequency range is usually empirically determined because of the difficulty of
characterizing the heating patterns of the modes in a multi—mode applicator. Tremendous
experimental efforts are required and yet heatin g modes with common hot and cold spots
could be used in the same heating process.

In single mode resonant applicators, only one mode is excited at one time and the
field pattern is more defined than that in multi-mode applicators. Frequency-sweeping in
a single mode applicator is not the best approach to create uniform field patterns, because
microwaves are not efficiently coupled into materials at non-resonant frequencies and the

reflectance can be very high. Instead, several modes with complementary heating

patterns can be excited distinctively and alternatively in the single mode cavity to obtain
time-averaged uniform heating. By switching among the selected resonant modes, the
energy efficiency can be highly improved. In a fixed frequency system, mode switching
could only be achieved by mechanically adjusting the volume (most time the length) of
the cavity [33,34]. This mechanical approach affects the rapidity and controllability of the
process. In a variable frequency system, the frequency can be varied to change the modes
electronically. Qiu et al. [35] used this method in composite processing. The results
showed that the speed and the controllability of the process were improved.

In Qiu et al.'s work [35], the mode selection was based on empirical heating
patterns. A large number of modes were involved in the mode-switching and a complex
mode-switching algorithm was needed. For each new material setup, the heating patterns
need to be re-characterized with extensive experimental measurement. This approach
needs to be implemented to reduce the amount of experimental work in heating pattern
characterization, reduce the number of modes, and simplify the mode-switching
algorithm. In this chapter, the variable frequency mode-switching method was
implemented and applied in microwave adhesive bonding of large-size materials to
obtain uniform heating in a single mode applicator. The electromagnetic modes were
diagnosed to obtain the theoretical electric field patterns. Modes with complementary
heating patterns were then selected according to the dimensions of the materials to be
heated. The mode combination with the least number of modes in it was used.

The main objective of this chapter is to design the variable frequency mode-

switching based on theoretical mode patterns, verify the heating uniformity of this

61

approach and compare the variable frequency mode—switching with single mode and

thermal adhesive bonding.

4.2 Experimental

4.2.1 Circuit for Microwave Adhesive Bonding

The circuit configuration used in this chapter was the same as that used in Chapter
3 (Figure 3.1). The microwave power source had variable frequency ability and the
frequency can be adjusted from 2 GHz to 4 GHz either manually or automatically. A
cylindrical single mode cavity with a diameter of 17.78cm was used. The cavity length
(Lc) and the probe depth (Lp) are adjusted to be 13.2cm and 2.0cm, respectively. The

sample is loaded at the center of the bottom plate of the cavity.

4.2.2 Materials

In the preliminary study of microwave adhesive bonding in Chapter 3, the
materials used were Bexloy W502 (major component: glass reinforced
ethylene/methacrylic acid copolymer) as the substrate and Eccobond A401-37 (epoxy
based) as the adhesive. In this chapter, this material system is also used. In addition, a
nylon 6 and ethylene/methacrylic acid copolymer substrate (Surlyn SG201U) is also used
with the Eccobond adhesive. The complex dielectric constant of Surlyn SG201U is 2.30-

j0.008.

62

4.2.3 Sample Setup and Preparation for Adhesive Bonding

The sample setup was similar as that used in Chapter 3. All dimensions were the
same except that the width of the assembly was 7.62cm, three times as that used in

Chapter 3. The sample setup and temperature measuring points are shown in Figure 4.1.

Temperature Auxiliary Panel for
Probe Hole Temperature Measurement

Glass .. .. i I n a
Beads
2' / / /

Supporting Panel Adhesive Release Films Substrate Panel 2

Substrate Panel 1

 

 

 

 

 

 

 

 

(a) Side View

 

X

k Adhesive

H Temperatures

/' /

\

Panel
Temperature

OOOO

 

 

 

 

 

 

 

 

(b) Top View

Figure 4.1 Sample Setup and Temperature Measuring Points

Several glass beads (diameter 0.6mm) were embedded in the adhesive to get a
uniform thickness of the adhesive. Because of the large dimension of the assembly, the

adhesive temperature was measured at 5 points. Since the temperature of the adhesive

63

between the two substrate panels can not be directly measured, an auxiliary panel was
used for temperature measurement to approximate the real adhesive curing environment.
Adhesive with the same dimension as the bonding area was applied between the auxiliary
panel and panel 2. Several small holes were drilled through the auxiliary panel, where the
temperature probes were inserted to measure the adhesive temperature. Two pieces of
microwave transparent release film with a thickness of 2011m were put between the
adhesive and the panels in the temperature measuring region. Adhesive between the
release films was tested with Differential Scanning Calorimetry (DSC) to measure the

residual heat. The extent of cure was then calculated with the following equation:

a=(Htot—Hres)/Htot (4:1)

Where or is the extent of cure, Hm is the total heat of reaction, and Hres is the residual

heat measured with DSC.

The adhesive dimensions were 1.6cm x 7.62cm with a thickness of 0.6mm. The
entire assembly was 9.36cm x 7.62cm with a height of 6.6mm.

Before bonding, the Bexloy panels were degreased with acetone and dried at
77°C. The panels were roughened at the joint areas with abrasion paper in order to
enhance bonding. For the Surlyn substrates, a different preparation procedure was used
because the Nylon component tends to absorb moisture when exposed to air. The
existence of moisture not only affects adversely the bond quality, but also causes the

formation of hot spots in the substrates in microwave heating because of the high

dielectric properties of water. The surface of Surlyn substrates was prepared according to
the procedures for adhesive bonding of Nylon 6 [94]:

1) Degrease with acetone.

2) Scrub with household cleaner, rinse, and dry at 150F for 2 days.

3) Roughen the surface to be bonded with 240-grit sandpaper.

4) Bond the substrates as soon as the materials are prepared.

After adhesive bonding, the assemblies were cut into 9.36cm x 2.54cm coupons

for single lap shear test.

4.3 Results and Discussion

4.3.1 Characterization of Loaded Cavity

The mode spectrum (reflectance vs. frequency) usually changes as the type, size
or position of the material changes in a single mode cavity. Therefore, the cavity loaded
with materials needs to be characterized to locate the resonant frequencies of the modes
whenever there is a change in the factors mentioned above. The mode spectrums were
characterized for the cavity loaded with materials of dimensions shown in Figure 4.1 for
both Bexloy and Surlyn substrates. The frequency was swept from 2.0 to 4.0GHz and the
reflectance was obtained as the ratio of the reflected power (Pr) to the incident power
(Pi). The incident power level was controlled at lSilW. The frequency with minimum

reflectance (Pr/Pi) is the resonant frequency of a mode. The mode spectrums for Bexloy

65

and Surlyn substrates were very similar. The reason might be that the dielectric properties
of the two substrates were close and the size and position of the assemblies were the
same. Results of the mode characteristics for the Bexloy substrates are shown in Figure
4.2 and the mode spectrum for the Surlyn substrates is not plotted here because of its
similarity to Figure 4.2. A large number of modes were available in the cavity at
different frequencies. Most of the modes should have different heating patterns, though
some of them could have similar heating patterns. The availability of modes with various
heating patterns made it possible to select modes with complementary heating patterns to

obtain time-averaged uniform heating.

 

 

 

 

 

 

 

 

 

 

 

 

 

I I

2.9 3.4
Frequency (GHz)

 

 

 

 

Figure 4.2 Characterization of Loaded Cavity (Sample Size: 93.6 x 76.2 x 6.06mm3)

66

4.3.2 Mode Diagnosis and Selection of Modes for Mode-switching

In order to locate modes with complementary heating patterns and to understand
the electric field distribution pattern inside the cavity, several electromagnetic modes
were diagnosed in the loaded cavity with the approach described in Chapter 3. Among
these diagnosed modes, the modes with resonant frequencies of 2.8450GHz and
3.3830GHz formed one of the possible mode combinations with complementary heating
patterns. Therefore these two modes were selected to use in the variable frequency mode-
switching heating. For this reason the diagnosis of these two modes is briefly described
below. The mode at 2.8450GHz is referred to as mode 1 and the one at 3.3830GHz is
referred to as mode 2. The diagnosis includes two parts — characterization of the heating
pattern and measurement of the half wave number q. The major assumption used in
mode diagnosis was that after the sample was loaded into the cavity, the change of

resonant frequency was within 10% of the resonant frequency in the empty cavity.

4.3.2.1 Characterization of Heating Pattern

To characterize the heating pattern of each mode, the temperature distribution in
the adhesive was measured as a function of time. Figure 4.3 (a) and (b) show the heating
profiles of Modes 1 and 2, respectively. The temperature distributions of both modes
were symmetric with respect to the horizontal central axis of the assembly. But the
heating preferences of the two modes were different. In heating with Mode l, the center
temperature T3 was the highest and the edge temperatures T1 and T5 were the lowest. In
heating with Mode 2, the edge temperatures T. and T5 were the highest and the center
temperature T3 was the lowest. Therefore Mode l preferentially heats the center of the

adhesive and Mode 2 preferentially heats the edges.

67

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

time (minute)

 

 

 

 

120
A 100* nuflunaonflanongonfln
(:0 act“""‘““ °T1
'8 807 a .3. ATZ
I: 0 :°
,3 60: a 3 $3 I'13
as) 40~: e“. +T5
E" ..O..

20~

0 r f

0 10 15 20
Time(minute)
(a) Heating Pattern of Mode 1 at 2.8450GHz

120
6100— ++++++++++333¢¢eett
on + °°°°°°°°° .ooxt arr
“3 80“ o 9°:2‘7A
‘0 +0 90‘.
I; +0 0.92:9. ‘TZ
g 604 o 22.815 Dungeon 0T3
‘3 t t: cannon 01.4
g. 40.13. Duncan
0) .9090 +T5
‘" 203°

0 I r I

0 5 10 15 20

 

 

(b) Heating Pattern of Mode 2 at 3.3830GHz

Figure 4.3 Characterization of Heating Patterns of Modes 1 And 2

68

 

 

4.3.2.2 Measurement of Half Wave Number 0

In the determination of the half wave number q, the relative radial electric field
strength was studied along the axial direction. Figure 4.4 (a) and (b) show the results of
the measurement of half wave number q, which is 0 for mode 1 and 2 for mode 2.

With the assumption used in the mode diagnosis, modes 1 and 2 were diagnosed
to be TM020 and TM212 respectively. The theoretical electric field patterns of the two
modes in the empty cavity were computed with electromagnetic field equations [37,
Chapter 5] and shown in Figure 4.5. Lighter region represents higher electric field
strength. These two modes provide complementary heating patterns for the adhesive
dimensions shown in Figure 4.6. Though the introduction of materials into the applicator
will distort the field distributions slightly, the mode patterns remain similar. The results

of mode diagnosis are summarized in Table 4.1.

4.3.3 Algorithm of Mode-switching

Because only two modes were involved in the mode-switching, a simple
algorithm was used: If the center temperature (T3) is 3°C lower than the edge one
(maximum of T1 and T5), then switch the heating to center heating mode TM020. If the
edge temperature is 3°C lower than the center one, then switch the heating to edge

heating mode TM212.

69

 

 

Distance from the Bottom

 

14
12
10
g 8
6 6
e.
o 4
2
o . . ,

 

 

0.2 0.4 0.6 0.8

o -t

Relative Radial Electric Field

 

 

(a) Mode l at 2.8450GHz: Half Wave Number q=0

 

 

Distance from the Bottom

 

l4

lZi O—
10—

of Cavity

 

 

 

0 e r .
0 0.2 0.4 0.6 0.8

Relative Radial Electric Field

 

 

(b) Mode 2 at 3.3830GHz: Half Wave Number q =2

Figure 4.4 Measurement of Half Wave Number Q

70

 

 

(a) Theoretical E field Pattern of Mode l (TM020) at the Cross-section of
the cavity at the adhesive plane in the Empty Cavity

0.8 >
0.8
0.4 -

0.2

414
as .

-0.B

 

 

.1 '
-1 415 U 0.5 1

(b) Theoretical E field Pattern of Mode 2 (TM212) at the Cross-section
of the cavity at the adhesive plane in the Empty Cavity

Figure 4.5 Theoretical E Field Patterns in the Empty Cavity

71

 

vL

Cross- AdheSI VC

section of

 

 

 

 

Coupling Probe

Figure 4.6 Dimension of the Adhesive inside the Cavity

Table 4.1 Summary of Mode Diagnosis

 

Mode 1 Mode 2

 

Resonant Frequency in 2.8450 3.3830

Loaded Cavity (F 1) (GHz) for
the Bexloy Substrates

 

Resonant Frequency in 2.8454 3.3846

Loaded Cavity (F 1) (GHz) for
the Surlyn Substrates

 

 

 

 

 

 

 

Heating Preference Center of Edge of the
the adhesive adhesive
Half Wave Number q 0 2
Mode Name Diagnosed TM020 TM212
Resonant Frequency in 2.9830 3.5990
Empty Cavity (F2) (GHz)
Percentage of Frequency -4.63% -6.00%
Shift ((Fr-F2)/F2*100%) for
the Bexloy Substrates
Percentage of Frequency -4.61% -5.96%

Shift ((Fl-F2)/F2*100%) for
the Surlyn Substrates

 

 

 

72

 

4.3.4 Heating Profile of Variable Frequency Mode-switching Adhesive Bonding and

Comparison with Thermal Process

4.3.4.1 Heating Profile of Eccobond/Bexloy Assembly

The TM020/T M212 mode combination was used in the variable frequency mode—
switching adhesive bonding process. For the assembly with Bexloy substrates, two
different heating cycles of the adhesive were used. One was to heat up the adhesive from
room temperature to 110°C in 6 minutes and then maintain at a constant level. The other
was to heat up the adhesive from room temperature to 120°C in 8 minutes and then
maintain constant. The microwave power profiles of these two heating cycles are shown
in Figures 4.7 and 4.8. A higher power level was required in the heating up phase and a
lower power was used in the isothermal phase. The heating profiles of these two heating
cycles are shown in Figure 4.9 (110°C) and Figure 4.10 (120°C), respectively. Adhesive
temperatures were measured at 5 different locations across the adhesive. The maximum
temperature gradient within the adhesive was less than 10°C. Heating uniformity of the
adhesive was greatly improved compared with single mode heating (Figure 4.3 (a) and
(b)). Panel materials were not substantially heated up and microwaves were concentrated
at the adhesive curing and interfacial bonding region.

To compare variable frequency mode-switching adhesive bonding with thermal
bonding, two thermal heating processes were performed in a thermal oven at 110°C and

120°C, respectively. The temperatures were measured with the same methods as that used

in microwave heating. The thermal heating profile at 120°C is shown in Figure 4.11.

73

 

 

Power (W)

t—n h— N N U)
o Ur o m o
o
o
o

M
1

 

 

O

 

I I

O 10 20 30 40

Time (min)

 

 

 

Figure 4.7 Power Deposition in Mode-Switching Adhesive Bonding of Bexloy at 110 °C

 

 

Power (W)
e=n t—I N N
O U! C VI
0
O
O
O
o
O
O
O

M
L

 

 

O

 

I I I

10 15 20 25

O
Ut

Time (min)

 

 

 

Figure 4.8 Power Deposition in Mode-Switching Adhesive Bonding of Bexloy at 120 °C

74

 

Temperature (deg C)

 

120

 

 

 

 

 

o Substrate T

:1 Adhesive Tl
A Adhesive 12
x Adhesive T3
- Adhesive T4
0 Adhesive T5

 

 

10

I

20

 

30 40

Time (min)

 

Figure 4.9 Temperature Profile in Mode-Switching Bonding of Bexloy at 110 °C

 

Temperature (deg C)

 

 

 

 

 

 

 

 

 

 

 

140
120‘ lxl3fl'iaaghgaa
x§§$"”’¥$9214999
100 3 33 AdhesiveT
.4 .00... OT]
80 ..Q....
..' PanelT ‘T2
60‘ 0 0T3
0
402! “T4
20" +T5
0T6
0 I T I
0 5 10 15 20 25

Time (minute)

 

Figure 4.10 Temperature Profile in Mode-Switching Bonding of Bexloy at 120 °C

75

 

 

Temperature (deg C)

 

140

 

 

 

IZOJ 4:0,Atar or...
O
100 - f
80 fi ’4 /
60 - I
.‘ 4’
40 i A;
20 I
O I I F I
0 20 40 60 80
Time (rrrinute)
I : Range of Adhesive

° 2

9 I

Average Adhesive Temperature
Panel Temperature

 

100

 

Figure 4.11 Temperature Profile in Thermal Bonding of Bexloy at 120 °C

Because the thermal heating mechanism was from outside to inside, the substrates

were first heated up and then heat was transferred to the adhesive via conduction.

Therefore the temperature of the substrates was always higher than that of the adhesive

and the heating of the adhesive was much slower compared with microwave process. The

maximum temperature gradient within the adhesive was around 2°C. For the thin film

adhesive sample, thermal temperature profile was more uniform than mode-switching

with the mode combination used in this study. However, in heating 3 dimensional

adhesive structures with complex shapes, thermal method will very possibly cause large

temperature gradients within the adhesives due to its non-selectivity and surface heating

nature. Microwave energy, on the other hand, can be tailored to perform uniform heating

with proper mode selections.

76

 

4.3.4.2 Heating Profile of Eccobond/Surlyn Assembly

For Eccobond A401-37/Surlyn SG201U, the adhesive temperature could not get
higher than 100°C, otherwise the substrate surface adjacent to the adhesive would
degrade. The adhesive was heated up from room temperature to 100°C in 5.5 minutes and
then maintained at an isothermal temperature of 100°C. The temperature profile during
this process is shown in Figure 4.13. The temperature measuring points for the
Eccobond/Surlyn assembly are illustrated in Figure 4.12. The adhesive temperature was
measured at three points instead of five, because the temperature distribution is

symmetric with respect to the horizontal central axis of the assembly, as shown in Figure

4.3.

 

O
0
T1 Adhesive
/ O V" Temperatures
Panel 6 O /
Temperature T2 /
0

T3

 

 

 

 

 

 

 

 

Figure 4.12 Temperature Measuring Points for the Eccobond/Surlyn Assembly (Top

View)

77

 

 

 

 

 

 

 

 

 

120

100 ~ 1:113 a BEBE
o 8§§83gg 8 88888g8g
"g 80‘ 00000000000°°°°°°°°°°° oAdhesiveTl
93. o DAdhesiveT2
3 6O ‘ B o .
E’. o AAdhesrve T3
0
o. o
.18, 40 o oSubstrateT
E-

20

0 ' I V fi fl 1
0 10 20 30 40 50 60
Time (min)

 

 

 

Figure 4.13 Temperature Profile in Mode-Switching Bonding of Surlyn at 100 °C

4.3.5 Study of Adhesive Curing in Mode-Switching Microwave Process and
Comparison with Thermal Curing

During mode-switching microwave adhesive bonding, the extent of cure of the
adhesive was tested off-line with DSC as the function of heating time at 110°C and
120°C. Three samples at different locations were measured for each time point. The
average extent of cure and the standard deviation were calculated. Figure 4.14 shows the
average extent of cure vs time during the microwave process of 6 minutes of heat up and
then held isothermally at 110°C. Figure 4.15 represents the average extent of cure vs
time during microwave process with 8 rrrinutes of heat up and then held isothermally at

120°C. The maximum incident power level was 35 Watts. The heating rate can be

78

further increased with higher incident power. The microwave curing started at around
70°C. A significant degree of cure (70% for 110°C and 88% for 120°C) was obtained
before the isothermal curing stage was reached. The isothermal curing time was 24
minutes and 12 minutes for microwave at 110°C and 120°C, respectively. The ultimate
extent of cure was around 97% for 110°C and 99% for 120°C. The maximum standard
deviation of the extent of cure was 9.0% for both microwave 110°C and 120°C. The

average standard deviation was 3.9% for 110°C and 3.6% for 120°C.

 

 

0.9 4 °

0.8 ~ ,

0.7 —. ~

0.6 4

0.5 e

0.4 -

0.3 ~ ’

0.2 ~

0.1 ~
0 4’ . a . r

O 10 20 3O 40 50

Time (rrrinute)

Ext. of Cure

 

 

 

 

 

 

Figure 4.14 Adhesive Extent of Cure vs. Time in Microwave Process of 6 Minutes of
Heat up and Then Isothermally at 110°C

79

 

 

1 o f 31
0.9 . . ’
0.8 ~ ,

0.7 4
0.6 a
0.5 ~
0.4 ~
0.3 — °
0.2 —
0.1 —
’ 0 t ° . .

0 10 20 30

Ext. of Cure

 

 

 

Time (minute)

 

 

 

Figure 4.15 Adhesive Extent of Cure vs. Time in Microwave Process of 8 Minutes of
Heat up and Then Isothermally at 120°C

Adhesive curing was also studied in thermal adhesive bonding. Adhesives were
heated from room temperature to 110°C in 30minutes and to 120°C in 40 rrrinutes.
Figures 4.16 and 4.17 show the adhesive extent of cure vs. time at 110°C and 120°C in
thermal heating processes. The curing started at around 110°C during thermal heating up.
Before the isothermal curing stage was reached, a low degree of cure (3% for 110°C and
20% for 120°C) was obtained. The required isothermal curing time was 90 minutes at
110°C and 60 minutes at 120°C. The ultimate extent of cure was around 95% for thermal
110°C and 99% for thermal 120°C. Comparison between mode-switching curing and
thermal curing is summarized in Table 4.2. Mode-switching microwave curing started at
a lower temperature and was much faster than thermal curing. At 110°C, the ultimate
extent of cure of microwave samples was slightly higher than that of thermal samples. At

120°C, the ultimate extent of cure for the two methods was about the same.

80

 

 

0.8 ~

0.6 r
0.5 r
0.4 1
0.3 r
0.2 r

0.1 l
O . I I I

0 50 100 150 200

Ext. of Cure

 

 

 

Time (rrrinute)

 

 

 

Figure 4.16 Adhesive Extent of Cure vs. Time in Thermal Heating of 30 Minutes Heat
up and Then Isothermally at 110°C

 

 

l . h T r—
0.9 a
0.8 a
0.7 ~
0.6 ~ 0
0.5 ~
0.4 «
0.3 — ’
0.2 - o
0.1 a

0 4 r r

0 50 100 150

Ext. of Cure

 

 

Time (minute)

 

 

 

Figure 4.17 Adhesive Extent of Cure vs. Time in Thermal Heating of 40 Minutes Heat
up and Then Isothermally at 120°C

81

Table 4.2 Comparison between Mode-Switching and Thermal Curing at 110°C and

 

 

 

 

 

 

 

 

 

 

120°C
Curing Microwave Microwave Thermal Thermal
Conditions Process of 6 Process of 8 Process of 30 Process of 40
min heat up and min heat up and min heat up and min heat up
then then then and then
isothermally at isothermally at isothermally at isothermally at
l 10°C 120°C 1 10°C 120°C
Total 30 rrrin 20 min 120 min 100 min
Required
Heating
Time
Ultimate 97.4%:1.0% 98.5% 11.0% 95.4%:0.2% 98.8% 10.4%
Extent of
Cure
Maximum 9.0% 9.0% 3.1% 3.8%
Standard
Deviation
Of Extent of
Cure
Average 3.9% 3.6% 1.1% 1.6%
Standard
Deviation
Of Extent
of Cure

 

4.3.6 Comparison of Bond Strength between Mode-Switching Microwave and

Thermal Bonding Processes

4.3.6.1

Bond Strength of Eccobond/Bexloy Assembly

For Eccobond/Bexloy, two assemblies were prepared for each microwave or

thermal process. After bonding, each assembly was cut into 3 coupons and tested with

single-lap shear equipment at a crosshead displacement rate of 0.05 inch/minute

(0.127cm/minute). Results are shown in Table 4.3. For each mode-switching bonded

82

 

assembly, the shear strength of the three coupons was close. This result verified the
heating uniformity of the TM020/T M212 mode-switching process. In addition, the
comparison between the results in Table 4.2 and that in Figure 3.12 shows that the shear
strength of mode-switching bonded large size samples was approximately equal to that of
the single mode microwave bonded small size samples. In addition, the shear strength of
mode-switching bonded samples was significantly higher than that of thermally bonded
ones. In the single lap—shear testing, microwave bonded assemblies all broke in the
substrate, indicating that the interface and the adhesive were stronger than the substrate.
On the other hand, thermally bonded samples all broke at the interface, showing that the
bond strength of thermal samples did not reach the maximum possible value even when

the adhesive was almost fully cured.

4.3.6.2 Bond Strength of Eccobond/Surlyn Assembly

For microwave adhesive bonding of the Eccobond/Surlyn assembly, the heating
up time was 5.5 rrrinutes and the isothermal bonding time was approximately 45minutes
at 100°C. To compare with thermal process, some assemblies were also bonded in a
thermal oven at 120°C. The thermal bonding temperature was 120°C instead of 100°C
because thermal bonding at 100°C was too slow. The approximate time for the thermal
bonding at 120°C was 100 minutes determined by trial and error. The strength of the
bonded assemblies was determined with single lap shear test at a crosshead displacement

rate of 0.05 inch/minute. For each bonding process, three samples were tested. Results

are shown in Table 4.4.

83

Table 4.3 Bond Strength of Eccobond/Bexloy Assembly with Microwave or Thermal

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Method
Bonding Cycles Microwave Microwave Thermal Thermal
6 min heat- 8 min heat- 30 rrrin 40 min
up, 24 min up, 12 min heat-up, heat up,
isothermal isothermal 90 rrrin 60min
bonding at bonding at isothermal isothermal
110°C 120°C at 110°C at 120°C
Break Shear in Shear in Shear at Shear at
Pattern Panel Panel interface interface
Shear Assembly Coupon 5.63 5.95 3.69 3.74
Strength 1 1_1
(MPa) Coupon 6.05 6.12 2.45 3.84
1_2
Coupon 5.60 4.84 2.66 3.12
1_3
Assembly Coupon 7.45 5.77 2.65 3.05
2 2_1
Coupon 5.02 7.06 2.88 2.67
2_2
Coupon 5.46 5.09 2.76 2.94
2_3
Average Shear Strength (MPa) 5.87 5.80 2.85 3.23
Standard Deviation of the 0.84 0.79 0.43 0.46
Shear Strength (MPa)

 

Table 4.4 Bond Strength of Eccobond/Surlyn Assembly with Microwave or Thermal

 

 

 

 

Method
Bonding Conditions Break Pattern Strength of the
Assembly (MPa)
Microwave 5.5 nrin heat up and 45 Shear in Panel 6. 1411.02
min isothermal bonding at 100°C
Thermal at 120°C for 100 min Shear in Panel 6.06:0.48

 

 

 

 

 

Table 4.4 shows that microwave bonded assemblies at 100°C for 45 rrrinutes

obtained the same strength as thermally bonded assemblies at 120°C for 100 minutes.

84

Both microwave and thermally bonded assemblies broke in the substrates in single lap
shear test. Since the interface and adhesive were stronger than the substrates for both
microwave and thermally bonded assemblies, it was uncertain whether there was
microwave enhancement of the adhesion between the adhesive and the substrate for the

Eccobond/Surlyn assembly.

4.4 Conclusions

This chapter implemented the variable frequency mode-switching method to
select modes based on theoretical electric field patterns, reduce the number of modes
used in mode-switching and simplify the mode-switching algorithm. The implemented
mode-switching method is applied in microwave adhesive bonding of large-size materials
to obtain uniform heating in a single mode applicator.

Eccobond A401-37 (epoxy based) was used as the adhesive to bond two substrate
materials, Bexloy W502 (major component: glass reinforced ethylene/methacrylic acid
copolymer) and Surlyn SG201U (Nylon 6 and ethylene/methacrylic acid copolymer). In
the mode-switching process, it was observed that rrricrowaves heated the adhesive rapidly
and selectively. Compared with single mode heating, mode-switching greatly improved
the heating uniformity.

The extent of cure of the adhesive was studied off line as a function of time at
different temperatures for both microwave and thermal bonding processes. Results
showed that the adhesive curing with microwaves was much faster than the thermal

process at the same temperature.

85

For Eccobond A401-37/Bexloy W502, results of microwave adhesive bonding
have shown 75% to 80% reduction in bonding time and at least 2-fold enhancement of
bonding strength compared with thermal process. Microwave bonded assemblies broke in
the substrates, while thermally bonded assemblies broke at the interface even at 99% of
cure of the adhesive.

For Eccobond A401-37/Surlyn SG201U, assemblies bonded with rrricrowaves at

100°C for 45 minutes obtained the same strength as thermally bonded assemblies at

120°C for 100 minutes. Both microwave and thermally bonded assemblies broke in the
substrates in single lap shear test. Since the interface and adhesive were stronger than the
substrates for both microwave and thermally bonded assemblies, it was uncertain whether
there was microwave enhancement of the adhesion between the adhesive and the
substrate.

Therefore, observed microwave effects in adhesive bonding of the two systems
include reduction of bonding time for both systems and enhanced bonding strength for
Bexloy W502 substrates. These results motivate further investigation into microwave

heating and bonding mechanisms.

86

CHAPTER 5 IN SITU MONITORING OF VARIABLE FREQUENCY
MICROWAVE PROCESSING IN A SINGLE MODE CAVITY

5.1 Introduction

Studies on the single mode and mode-switching microwave adhesive bonding are
presented in Chapters 3 and 4. The bonding cycle was determined by measuring either
the bonding strength (with single-lap shear test) or the extent of cure (with DSC) vs. time.
These measurements were carried out off-line and required tremendous amount of
experimental work. In order to reduce the amount of experiments and to realize
intelligent processing, an on-line monitoring technique needs to be developed.

Among several research efforts of monitoring microwave processing on-line,
Compton, et. al andYoung et al. [95,96] have employed Fourier transform infrared
(FTIR) spectroscopy to directly monitor the cure of composite materials. In this
technique, an exposed optical fiber sensor is embedded in the materials to monitor
material spectroscopic properties associated with the curing process. Marand et al. [12]
have applied cavity perturbation technique to on-line monitor the dielectric properties of
epoxy. At the same time, they used the FTIR technique to monitor the spectroscopic
properties of epoxy. The development of these on-line monitoring techniques has led to
improvements in processing materials and studying reaction mechanisms. Without
debating the significance of these techniques, some inherent disadvantages in these
techniques exist. In on-line monitoring with FT IR, the exposed fiber sensor has lower
signal to noise ratio compared with that in the transmission spectra. Repeated usage leads

to degradation and breakage of the fiber sensor, which results in high cost. In on-line

87

monitoring material dielectric properties, the cavity perturbation technique is difficult to
be applied in microwave processing of materials with large sizes and complex shapes.
Under these circumstances, a new method needs to be developed to monitor in
situ microwave adhesive bonding of large samples. In variable frequency mode-switching
microwave processing in a single mode cavity, resonant frequencies shift due to changes
in material dielectric properties. Any decrease in the dielectric property leads to an
increase in the resonant frequency [37, Chapter 7]. The operating frequency needs to be
constantly tuned to maintain the resonant state inside the microwave cavity. The resonant
frequency shifting, which reflects the change in dielectric properties, can be used to
monitor in situ the microwave processing. This provides a new method for determining
the microwave/material interactions on-line. This new on-line monitoring technique was
described in detail in [97] (patent application in process). The method can be applied in
the microwave processes that involve changes of material dielectric properties in a single
mode applicator. In this chapter, the application of this on-line monitoring technique in
several microwave adhesive bonding processes is studied. In these processes, different
substrates or assembly setups were used. These processes are presented in Table 5.1. The

substrates were obtained from DuPont and the adhesive was from Grace Specialty

Polymers.

88

Table 5.1 Microwave Adhesive Bonding Processes to Be Studied with the on-Line

 

 

 

 

 

 

 

 

Monitoring Technique

Process 1 Process 2 Process 3

Substrate Bexloy W502 (major Surlyn SG201U Bexloy W502
component: glass (major components: (major component:
reinforced nylon 6 and glass reinforced
ethylene/methacrylic ethylene/methacrylic ethylene/methacrylic
acid copolymer) acid copolymer) acid copolymer)
Eccobond A401-37 Eccobond A401-37 Eccobond A401-37

Adhesive
(major component: (major component: (major component:
epoxy) epoxy) epoxy)

Assembly Single-lap Single-lap Double-lap

Setup

 

5.2 Advantages Of This Technique

The advantages of this technique over existing approaches are:

- Diagnosis with Non-insertion Method

The measurement of resonant frequencies does not use a probe to be embedded

into material samples. Thus this technique eliminates the disadvantages such as

degradation, breakage and high cost, which are associated with the usage of

probes.

89

 

- Ability to Monitor the Microwave Processing of Materials with Large Sizes and
Complex Shapes
This technique monitors the change in resonant frequencies by scanning the
frequency and measuring the microwave reflectance. The diagnosis can always be
performed regardless of material sizes and shapes. Therefore this technique has
wider applications relative to the method of monitoring dielectric properties with
cavity perturbation technique.

- No Requirement of Separate Diagnostic Circuit
In this technique, the diagnosis uses the same circuit as that of variable frequency
microwave processing. No additional facilities are required. The diagnosis is fast

because no circuit switching is involved.

5.3 On-line Monitoring of Microwave Adhesive Bonding of Single-Lap Bexloy

W502i Eccobond A401-37

5.3.1 Experimental

The material setup was the same as that used in Chapter 4 in the variable
frequency mode-switching adhesive bonding process. Because the bonding cycle for the
Bexloy/Eccobond system had been determined with off-line measurement of the extent of
cure vs. time in Chapter 4, applying the on-line monitoring method to this system can

provide a check-up for the new method.

The two electromagnetic modes TM020 and TM212 used in Chapter 4 were
applied in this study to obtain uniform heating. The starting resonant frequencies for
modes TM020 and TM212 were 2.8450 GHz and 3.3830 GHz, respectively.

In microwave adhesive bonding of the low loss Bexloy W502 substrates, the
substrates are not substantially heated up. The change of substrate dielectric properties is
not as dramatic as that of the adhesive. In the epoxy adhesive curing process, the
dielectric constant and loss factor decrease as cross-linking reaction proceeds. Any
decrease in the dielectric property leads to an increase in the resonant frequency [37,
Chapter 7]. Thus the shifting of the resonant frequency can be used to monitor in situ the
microwave adhesive bonding process. The resonant frequencies were diagnosed
periodically. The diagnosis took about 2 seconds each time and had nearly no influence

on microwave heating.

5.3.2 Results and Discussion

Figure 5.1 shows the resonant frequency shifting during the microwave process of
6 rrrinutes heat up of the adhesive and then isothermally at 110°C. The low loss Bexloy
W502 substrates were not substantially heated up. The resonant frequency shifting curves
take similar shapes as the adhesive curing curve shown in Figure 4.5. At the very
beginning of sample heating up, the cross-linking did not take place and the rising of
temperature should lead to an increase in adhesive dielectric properties. Consequently it
was expected that the resonant frequency should shift down slightly. However, the
resonant frequency of TM020 did not shift up or down at the beginning in most
experimental runs and shifted down slightly in some experiments. The resonant

frequency of TM212 shifted up slightly at the beginning in nearly all experiments. The

91

reason for this phenomenon might be that the increase in the adhesive dielectric
properties at the beginning was too small to be detected because of the small size of the

adhesives in this experimental setup.

 

 

 

 

 

 

 

 

 

A 4.5
g 4* u “‘a “ ‘A‘A A‘ ‘A‘
I: —4
a 3 mmnouucb 0000000001213
2 25 a ‘0 :1
o D
3" 2 " 3
8
[13 1.5 ~
g 1 _ m AMode TM020
t:
O
3:3 0.5 A 0“ uMode TM212
0 £4 . . . 4
O 10 20 30 40 50
Time (minute)

 

 

 

Figure 5.1 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Bexloy Substrates at 110°C

As shown in Figure 5.1, the resonant frequencies shifted up at an increased rate
with the onset of curing, which occurred after 2 or 3 rrrinutes of heating (temperature
around 70°C). The shifting rate remained high for a period of time, corresponding to the
high curing rate of the adhesive. When the curing approached completion, the curing rate
decreased and the shifting of resonant frequencies slowed down as well. After 92% of
cure was obtained, which occurred after the 20th rrrinute, the frequency shifting became

less obvious. This was because the increase in the extent of cure was too small to detect

92

with the shifting of resonant frequency. After this point it took another 10 rrrinutes to
reach the ultimate extent of cure. The final frequency shifting was about 4.0MHz and
2.9MI-Iz for TM020 and TM212 modes respectively.

Similar frequency shifting results were obtained for the process of 8 minutes heat
up and then isothermally at 120°C, as shown in Figure 5.2. The resonant frequencies of
both modes kept shifting and became less obvious after 94% of cure was reached, which
occurred after the 13th minute. After this point it took another 7 rrrinutes to reach the

ultimate extent of cure. The final frequency shifting was around 3.5MHz for TM020 and

 

 

 

 

 

 

 

 

 

 

2.5MHz for TM212.
A 4
g 3 5 ~ 7 A A A A ‘7‘
f2: 3 2 ‘A A
t? 2 - '30
o D
8‘ 1 5 - 30
E 1 _ AMode TM020
... D
t:
g 0.5 - g EiMode TM212
m
g 0 12% . . . . T
—0.5
0 5 10 15 20 25 30
Time (minute)

 

 

 

Figure 5.2 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Bexloy Substrates at 120°C

93

Therefore, monitoring the resonant frequency shifting provided information about
the curing status of the adhesive during microwave adhesive bonding. When the curing
rate was high, the rate of resonant frequency shifting was high too. When the curing
approached completion, the curing rate decreased and the shifting of resonant frequencies

slowed down as well and finally leveled off.

5.4 On-line Monitoring of Microwave Adhesive Bonding of Single-Lap Surlyn

SG201U / Eccobond A401-37

5.4.1 Experimental

For Surlyn/Eccobond system, the two electromagnetic modes TM020 and TM212
were again applied in this study to obtain uniform heating. The starting resonant
frequencies for modes TM020 and TM212 were 2.8454 GHz and 3.3846 GHz,

respectively.

5.4.2 Results and Discussion

Figure 5.3 shows the resonant frequency shifting during the microwave process of
5.5 rrrinutes heating up of the adhesive and then maintained isothermally at 100°C. At
the very beginning of sample heating up, the cross-linking did not take place and the
rising of temperature led to a small increase in the dielectric properties of the adhesive.
However, the resonant frequency shifted down noticeably at the beginning of microwave

heating. This phenomenon was different from that of Eccobond/Bexloy system. The

94

reason might be that the dielectric properties of Surlyn substrates increased noticeably
with temperature because of the N ylon-6 component. After the minimum resonant
frequency was reached, the resonant frequency shifted up at a high rate. This
corresponded to the rapid decrease of material dielectric properties, which resulted from
the high curing rate of the adhesive. Finally the shifting of the resonant frequency slowed
down and became less obvious after a total time of 50 minutes. This indicated that the

curing reaction of the adhesive almost completed.

 

 

 

 

 

 

 

 

 

0.001

A I

N I

e - . t

9 0.0005 g a I 'l

:1:

5 0 a 3

5 I.- I . I

C

a -0.0005 - . I '0 .-

23 on

E: -0001 J O

S O '

g -0.0015 2 o. 0 Mode 1 (TM020)

0‘ I Mode 2 (TM212)
-0.002 r r r r r

0 10 20 30 40 50 60
Time (minutes)

 

 

 

Figure 5.3 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Single-
Lap Surlyn Substrates at 100°C

With the on-line monitoring method, the microwave bonding cycle was

determined to be 5.5 minutes heating up and 45 minutes isothermal bonding.

95

5.5 On-line Monitoring of Microwave Adhesive Bonding of Double-Lap Bexloy

W502] Eccobond A401-37

5.5.1 Experimental

Variable frequency microwave mode switching method was applied in bonding a
double-lap shear assembly of Bexloy W502 / Eccobond A401-37. The sample setup is
shown in Figure 5.4. The size of the substrate panels 1 and 2 was 45mm x 76.2mm. The
size of substrate panels 3 and 4 was 45mm x 11mm. The length of each overlap was
4mm, where the adhesive was applied. The space between panels 1 and 2 was 3mm. Two
supporting panels were used to prevent deformation of the bonds. Several glass beads
(diameter 0.6mm) were embedded in the adhesive to get a uniform thickness of the
adhesive. The length of the entire assembly was 93mm (shorter than that suggested by
double lap shear Standard Test Method to fit into the cavity). The two electromagnetic
modes TM020 and TM212 were again applied in this study to obtain uniform heating.
The starting resonant frequencies for mode TM020 and TM212 were 2.8392GHz and
3.3765GHz, respectively.

Because both the electric field and sample dimensions are symmetric with respect
to the central axis of the sample, the temperature distribution is symmetric too. This
symmetric temperature distribution was observed in previous single lap adhesive bonding
study. In this study the temperatures of the adhesive of the top layer were measured at 3

locations (T r to T3). Panel temperature was also monitored at 1 point (T4).

96

Substrate Adh sive Substrate Panel 3 Substrate
Panel 1 Panel 2

Glass FL _ .. _ 4 L 1 l r J
Beads 1 L J
T T

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Supporting Panel 1 Substrate Panel 4 Supporting Panel 2
(a) Side View
v
4
!
2 ... T.
i Adhesive
‘M 2 Temperatures
i
_____ ,_._._.._._._._.._._.._._ 5.39..-- .-.+X
p—I’ .
T4 “""“ :
Substrate !
Temperature :
i
!
i
l
(b) Top View

Figure 5.4 Sample Setup of Double-Lap Shear Assembly

97

The adhesive was heated up to 120°C in 9 minutes and then maintained
isothermal. The temperature profile during this process is shown in Figure 5.5. The on-

line monitoring method was used to determine the bonding cycle.

 

 

 

 

 

 

 

 

 

140
_ a D D D D 0
film DgSRBogsgfiaanEa
U 100 2‘
w ‘ 1:1
é no xxxxxxxxxxxx
0 80 r 6 X X
*- x
g m« ggxx oAammrr
E. 8 x (:1 AdhesiveT2
[3 4° + x x 8 Adhesive T3
20 x Substrate T J
0 I I f I
0 5 10 15 20 25
Time (min)

 

 

 

Figure 5.5 Temperature Profile in Microwave Adhesive Bonding of Double-Lap Shear
Assembly of Bexloy Substrates

5.5.2 Results and Discussion

Figure 5.6 shows the resonant frequency shifting during the microwave process of
9 minutes heating up of the adhesive and then maintained isothermally at 120°C. At the
very beginning of sample heating up, the cross-linking did not take place and the rising of
temperature led to a small increase in the dielectric properties of the adhesive.
Consequently the resonant frequency shifted down at the beginning of microwave

heating. The degree of frequency shifting down of the double-lap assembly was larger

98

than that of the single-lap assembly. This was because of the larger material size used in
the double-lap assembly. After the minimum resonant frequency was reached, the
resonant frequency shifted up at a high rate. This corresponded to the rapid decrease of
material dielectric properties, which resulted from the high curing rate of the adhesive.
Finally the shifting of the resonant frequency slowed down and became less obvious after
a total time of 19 minutes. This indicated that the curing reaction of the adhesive almost
completed.

With the on-line monitoring method, the microwave bonding cycle was
determined to be 9 minutes heating up and 10 rrrinutes isothermal bonding.

The strength of the bonded assemblies was determined with double lap shear test
at a crosshead displacement rate of 0.05 inch/minute. For each bonding cycle, three
samples were tested. Results are shown in Table 5.2. The assemblies did not break even
after the strain reached 100%. During the testing, the cross-section of the substrate
became narrower and narrower and the testing load dropped slightly after the peak was
reached. Obviously if further tested, the assembly would break within the substrate in a
tensile manner because of the narrowing cross—section of the substrate. The break pattern
in double-lap shear test was different than that in the single-lap shear test. This was
because single-lap shear setup caused deformation of the assemblies during testing and
finally tore the assembly up in a shear manner. The double—lap shear setup, however, did

not result in deformation of the tested assemblies.

99

 

 

 

 

 

 

 

 

 

 

3.50
q; 3.00 i .
m 0 ’ 90
g 2'50 1 0 . ’ ’ n u on
5:; 2.00 - 0’ ° 1: 1:1 1:1
5 1.50 — 01:11:]
5‘ .
§ 1'00 D D 0 Mode 1 (TM020)
8 0'50 L 9 1:1 Mode 2 (TM212)
a 01X) 0 D I I I T
8 -0.50 5 10 15 20 25
i: D
g -1.00 e
04 -1.50 ’

-200

Time (min)

 

 

 

Figure 5.6 Shifting of Resonant Frequency in Microwave Adhesive Bonding of Double-
Lap Bexloy Substrates at 120°C

Table 5.2 Bond Strength of Microwave Bonded Double—Lap Shear Assemblies

 

Bonding Cycles Break Pattern Shear Strength (Peak Value)

 

Microwave 9 min heat- Did not break even at 5.93MPa : 0.39MPa
up, 10 rrrin isothermal 100% strain

bonding at 120°C

 

 

 

 

 

100

5.6 Conclusions

A method has been developed to on-line monitor microwave processing in a
single mode applicator. This method is not intrusive, able to monitor microwave
processing of materials with large sizes and complex shapes, and do not require separate
diagnostic circuit. The new on-line monitoring method was applied in several microwave
adhesive bonding processes, including bonding of single-lap Bexloy IEccobond, single-
lap Surlyn lEccobond, and double-lap Bexloy lEccobond assemblies. During the bonding
process, the curing of the adhesive led to decreases in the dielectric constant and loss
factor. Resonant frequencies of the modes, which reflected changes in material dielectric
properties, were monitored during microwave adhesive bonding. When the curing rate
was high, the rate of resonant frequency shifting was high too. When the curing
approached completion, the curing rate decreased and the shifting of resonant frequencies
slowed down as well and finally leveled off. Thus when the resonant frequencies stopped
shifting, the bonding was considered completed. Bonding cycles were determined with

the on-line monitoring method.

101

CHAPTER 6 PROCESS CONTROL SYSTEM FOR MICROWAVE ADHESIVE
BONDING

6.1 Introduction

In Chapters 3 and 4, single mode and mode-switching microwave adhesive
bonding in a single mode cavity are discussed. A new on-line monitoring technique for
microwave processing in a single mode cavity is presented in Chapter 5. In all the work,
data acquisition and process control was essential to eliminate the extensive recording of
experimental data by hand, reduce the continuous adjustment of the instruments, tailor
microwave energy to perform uniform and stable heating, and enhance the repeatability
of the experiments. The process control system was developed at the beginning of this
research and implemented throughout the dissertation. In this chapter, the functions,
structures and algorithms of the process control system are discussed.

Research has been conducted to build process control systems for microwave
processing in a single-mode cavity [33, 35, 34]. Adegbite et al. [33] developed a process
control system to obtain uniform processing in fixed frequency microwave processing.
Using a fixed frequency microwave power source, the mode switching was realized by
adjusting cavity length and coupling probe depth. To ease the operation of the microwave
processing system, work was done to automate the control of the fixed frequency
microwave power source and the adjustment of the resonant cavity. Relatively uniform
processing was achieved for 3-inch 24-ply square graphite/epoxy composite parts.

However, the mechanical tuning of the cavity affected the rapidity and stability of the

102

temperature control. Large temperature fluctuations and gradients were observed
throughout the heating process.

Fellows et al. [34] used fixed frequency mode switching technique to process V-
shaped polyester/glass composite parts. Slow mechanical adjustment of cavity length for
mode switching resulted in instability of temperature control and non-uniform
temperature distribution.

With the use of a variable frequency microwave power source, Qiu et al. [35]
developed a process control system for variable frequency microwave processing of
polymer and composite materials. This technology provided more uniform and stable
processing of composite parts compared with fixed frequency mode switching. The
variable frequency power source was automated to realize computer control of
microwave power and frequency. Control hardware was implemented based on the fixed
frequency control system. Two processing techniques were studied, including mode
sweeping and intelligent mode switching. The temperature distribution was more uniform
with intelligent mode switching than mode sweeping heating. In mode switching
processing, empirical heating modes were used. The empirical heating characteristics of
each mode, including heating rate and heating pattern, were measured before the
processing and stored in a database. These data were then used for mode selection during
processing to achieve uniform heating. In controlling processing temperature, both a
simple parabolic power controller and a multi-staged PID controller were used for
microwave power control. However, empirical controller parameters were used and
extensive experimental work was needed to find the parameters when there was a change

in material setup.

103

The variable frequency microwave power source and control hardware used in
this chapter are those implemented by Qiu et al. [35]. This chapter is focused on
developing software to perform data acquisition and process control to achieve rapid,

uniform and stable heating.

6.2 Structure of the Process Control System for Microwave Adhesive Bonding

Two bonding processes were studied in this dissertation: single mode microwave
adhesive bonding for small size materials and mode switching microwave adhesive
bonding for large materials. The process control system was first developed for single
mode microwave adhesive bonding and then implemented for mode switching process to
obtain uniform heating for large size materials. The process control system for the
variable frequency mode switching adhesive bonding process will be described as
follows.

Existing hardware configuration in [35] was used in variable frequency
microwave adhesive bonding process. Software was programmed in LabVIEW for cavity
characterization, data acquisition and process control. The software in this dissertation
was developed in parallel with that for variable frequency microwave processing of
composites [35] and further implemented the program in [35]. There are three major
implementations of this work over that in [35]:

First, the software in [35] required extensive input information because empirical
heating modes were used. The empirical heating characteristics of each mode, including

heating rate and temperature distribution, were measured before the processing and stored

104

in a database. These data were then used for mode selection during processing to achieve
uniform heating. Therefore the adaptability of the program to other processes was
affected. If a new material setup is used, then the extensive heating characterization needs
to be performed again before the processing. This problem can be solved with the
theoretical modes, which were used in adhesive bonding process in this research. With
the theoretical modes, the heating rate and temperature distribution were not necessary to
be measured and input into computer before the experiments.

Second, the parameters of the proportional-integral-differential (PID) controller in
[35] were empirical values. The controller parameters in this dissertation were obtained
based on control theory. This implementation eliminated the trial and error procedure of
adjusting controller parameters and also provided sound basis for automatic tuning of the
parameters.

Third, an on-line monitoring technique was developed for microwave adhesive
bonding process in this dissertation. A sub-controller was programmed to perform the on-
line monitoring.

The hierarchy of the software for microwave adhesive bonding process is shown in
Figure B] in Appendix B. The software includes two main programs. One is cavity
characterization before bonding to obtain the mode spectrum, the other is process control
during bonding to achieve rapid, uniform and stable heating with on-line monitoring
feature. LabVIEW was used to program the software, which includes a number of sub-
programs. These programs are shown in Appendix B. The images in Appendix B of

this dissertation are presented in color.

105

6.3 Program for Cavity Characterization

The LabVIEW program for the cavity characterization (cavity characterization.vi)
is shown in Figure B.2 and Figure B.3. Cavity characterization was performed with
measuring the incident power (Pi) and the reflected power (Pr) as a function of frequency
to obtain the mode spectrums. The frequency with minimum reflectance (Pr/Pi) was the
resonant frequency of a mode. The most important caution when editing and using this
program is to make sure that the scale factors for the input power (Pi Scale) and reflected
power (Pr Scale) in the program should be the same as that in the power meters. The

following equations were used to convert the acquired data into power data [98]:

Pi [W] = Pi Scale [W/v] x Acquired Voltage for Input Power [v] (6.1)

Pr [W] = Pr Scale [W/v] x Acquired Voltage for Reflected Power [v] (6.2)

Other parts of this program are straightforward and will not be explained in more

detail in this chapter.

6.4 Program for Process Control

The LabVIEW program for the process control (VFMS.vi) is shown in Figure B.4,
Figure 3.5 and Figure B6. The process control program is composed of a number of sub-
programs, including data acquisition, mode-switching, power and on-line monitoring

controllers.

106

6.4.1 Subprograms for Data Acquisition

In data acquisition, the temperatures of the materials and the incident and reflected
microwave powers are obtained through an AID board and then fed back to the process
controllers. The subprogram 6T acquires temperature data at six points, three of which
are measured with Luxtron thermometer and the other three are measured with Nortech

thermometer. The following equations are used to convert the acquired voltage data into

temperature [88,89]:
For Luxtron: T [°C] = Acquired voltage [v] x 1000 [mv/v] / 20 [mv/°C] (6.3)
For Nortech: T [°C] = Acquired voltage [v] x 40 [°C/v] — 100 [°C] (6.4)

The LabVIEW programs for temperature data acquisition are shown in Figures B.7
to B.12. The microwave power data are obtained with the subprogram power.vi, as

shown in Figures 3.13 and BM.

6.4.2 Subprograms for Mode-Switching

A variable frequency mode-switching controller was programmed with LabVIEW
to achieve uniform heating by switching among modes with complementary heating
patterns. The LabVIEW programs are shown in Figures 8.15 and B.16. The temperature
distribution was analyzed on-line to obtain the heating characteristics and to decide the
complementary heating pattern. The mode-switching controller minimized the

temperature gradient by determining which mode was to be used next. The resonant

107

frequency of the new mode was controlled through a GPIB interface between the sweep

oscillator and the computer.

6.4.3 Subprograms for Microwave Power Control

The objective of the power controller is to keep the heating rate below maximum
void free value during the adhesive heating up and maintain the temperature at constant
value during isothermal bonding. The LabVIEW programs for power control are shown
in Figures B.17 and B.18.

There are two phases during the microwave adhesive bonding process. The first
phase is the heating from room temperature to the isothermal bonding temperature. The
second phase is isothermal microwave bonding. Different algorithms were used in the

two phases to control the heating rate and isothermal bonding temperature, respectively.

6.4.3.1 Control of Heating Rate

During the heating up of the adhesive from room temperature to the bonding
temperature, on-off control was used to keep the heating rate below the maximum void
free value (2°C/s for the Eccobond adhesive). The heating rate was calculated on-line
with the LabVIEW programs shown in Figures 3.19 and B20. If the heating rate was
below the maximum value, then the maximum input power level was used for fast
heating. If the heating rate exceeded the maximum value, then the power was turned off

to prevent bubble formation.

108

6.4.3.2 Control of Isothermal Bonding Temperature

During isothermal bonding, traditional proportional-integral-differential (PID)
was used to control the bonding temperature by adjusting the incident power level. The
highest measured temperature was controlled in the isothermal bonding. The velocity
form of the PID algorithm was used:

At T
Pn - Pn—l = Kc[(en -en—l)+‘T_en +‘th—(en —26n—1 +en-2)] (65)
i

Where Pu and PM are the desired and previous incident powers, Kc is the proportional
control gain, Ti and Td are the integral and differential control time constants, At is the
sampling period, en, em, and en_2 are the temperature offsets at time n, n-1 and n-2,

respectively. The PID controller was implemented with LabVIEW program, as shown in
Figures B.21 and B.22.

The PID parameters Kc, Ti and Td were obtained with Ziegler-Nichols frequency
response method [99]. The proportional control action was used to find the ultimate gain
Ku and the ultimate period Tu. The gain was slowly increased until the process began to
oscillate. Ku is the gain when this oscillation occurs and Tu is the period of the
oscillation. The parameters of PID control were then calculated with Ziegler-Nichols
method based on Ku and Tu. The typical temperature oscillation for determining Ku and
Tu is shown in Figure 6.1.

The PID parameters for single mode adhesive bonding are listed in Table 6.1. In
variable frequency mode switching bonding with TM020 and TM212 modes, PID control
was not as stable as desired. The temperature fluctuation could be as high as :3 °C with

respect to the set point temperature. To stabilize the control, differential action was

109

disabled and the proportional- integral (PI) actions were used together. The PI parameters
were obtained with Ziegler-Nichols method and listed in Table 6.2. The parameters for
the two modes were close but not the same. Therefore a separate PI controller was used
for each of the two modes. With these control parameters, the temperature oscillation can
be controlled within :t1°C with respect to the set point temperature. To simplify the PID
LabVIEW program, an additional subroutine was used to input the PID parameters, as

shown in Figures B23 and B24.

 

 

Temperature (deg C)
8

112~

 

 

1 10 r r r r
200 220 240 260 280 300

Time (sec)

 

 

 

 

Figure 6.1 Temperature Oscillation for Determining Ku and Tu

110

Table 6.1 PID Controller Parameters for Single Mode Microwave Adhesive Bonding

 

 

 

 

 

 

Kc (watt/°C) Ti (sec) Td (sec)
Parameters obtained from 9.0 7.1 1.7
Ziegler-Nichols method (0.6Ku) (0.5Tu) (0.12Tu)

 

 

* Ku=15 watt/°C, Tu=14.3 sec.

Table 6.2 PI Controller Parameters for Variable Frequency Mode Switching Bonding

 

 

 

Ku Tu Kc Ti
(watt/°C) (sec) (sec)
Mode 1 12.0 42.4 4.8 33.9
(TM020) (0.4Ku) (0.8Tu)
Mode 2 11.0 37.4 4.4 29.9
(TM212) (0.4Ku) (0.8Tu)

 

 

 

 

 

 

 

6.4.4 Subprogram for 0n-line Monitoring

The algorithm of on—line monitoring was described in detail in Chapter 5. Material
dielectric properties changed with temperature and extent of reaction during heating and
curing. This led to a shifting of resonant frequencies. The resonant frequencies were
diagnosed (LabVIEW program Figures B.25 and B26) and updated (LabVIEW program
Figures B.27 and B28) periodically to maintain the resonant state inside the single mode

applicator. With the on—line monitoring, the bonding cycle can be directly determined.

111

6.5 Conclusions

A process control system was developed for single mode and variable frequency
mode switching microwave adhesive bonding processes. Compared with previous work
in microwave process control, the control system developed in this study has several
improvements. First, theoretical modes were used instead of empirical heating modes.
This eliminated extensive heating characterization and data storage. Second, the PID
controller parameters were obtained based on control theory instead of empirical values.
This eliminated the trial and error procedure of adjusting controller parameters and also
provided sound basis for automatic tuning of the parameters. Third, an on-line monitoring
technique was developed to obtain the bonding cycle directly.

The software was programmed with LabVIEW for cavity characterization, data
acquisition and process control. Cavity characterization was performed before the
bonding process to obtain the mode spectrum. In data acquisition, material temperatures
and microwave powers were obtained. Process control included the control of heating
rate, isothermal temperature and on-line monitoring of the resonant frequency. Heating
rate was adjusted with an on—off control algorithm. Isothermal temperature was controlled
with the traditional PID algorithm. The PID parameters were obtained based on control
theory. The resonant frequencies of the modes were monitored with the on-line
monitoring sub-controller to obtain the bonding cycle. This process control system was
successfully used in the microwave adhesive bonding process to provide rapid, uniform

and stable heating with the ability to. determine the bonding cycle on-line.

112

CHAPTER 7 INVESTIGATION OF MICROWAVE HEATING MECHANISM

VIA STUDY OF MICROWAVE CURING OF EPOXY FILLED WITH CARBON

7 .1 Introduction

The experimental results in Chapters 3 and 4 have shown that microwaves
reduced the curing time of the epoxy based adhesive with equal or even higher bonding
strength. These results motivate further investigation into microwave heating
mechanisms to provide explanations for reaction rate enhancement with microwaves. The
enhancement of polymer curing rate with microwaves has been demonstrated in a number
of studies [7,10,12-14, 100]. Some investigators suggested that the reaction rate
enhancement was because of microwave thermal effect, which is localized superheating
[7 , 10]. Some other investigators attributed the rate enhancement to specific microwave
non-thermal effects such as accelerated reaction of the secondary amine group [12] and
improved diffusion rate of reactive species [100]. Whether the reaction rate enhancement
was mainly because of microwave thermal or non-thermal effect requires further
investigation.

Fu et al [101] studied the microwave thermal or non-thermal effect by comparing
continuous-power and pulsed-power microwave curing of epoxy resins. In pulse-power
curing, as power is turned on and off, the system is only partially processed with
microwave irradiation. If the microwave enhancement of reaction rates is due to thermal

effect, which is localized superheating, energy will be transferred from those “hot spots”

ll3

to the surrounding area. As a result, the local temperature of those “hot spots” will
decrease when the power is turned off. Therefore, the reaction rates of the system will
decrease. However, since the bulk temperature is kept stable, this decrease is limited. If
the microwave enhancement of reaction rates is due to non—thermal effect, there is no
microwave effect when the incident power is turned off. Therefore, the power-off state
will lead to significant decrease of reaction rates. It is expected that continuous-power
curing will have faster reaction rates than pulsed-power curing if non-thermal effect is the
main reason of microwave enhancement of reaction rates. Results showed that
continuous-power microwave curing had only slightly higher reaction rates and ultimate
extents of cure than pulsed-power processing. The results seemed to support the theory of
thermal effect. But non-thermal effect could not be disproved because the power level in
pulsed-power curing was much higher than that in continuous-power curing.

The problem encountered in previous study [101] was that microwave power has
large influence on both microwave thermal and non-thermal effects. If microwave
thermal effect could be preferentially modified without altering the non-thermal effect,
then the changes in reaction rate could be attributed to rrricrowave thermal effects only
and microwave heating mechanism could be revealed. It has been pointed out that
microwave heating of materials depends largely on dielectric properties [39].
Microwaves can be more efficiently coupled into components with higher dielectric
properties. Fillers with high dielectric properties can be added into resins in microwave
curing to modify microwave thermal effect without significantly affecting the non-

thermal alignment of polar groups in the electromagnetic field.

114

In this chapter, microwave heating mechanism is investigated via studying
microwave curing of carbon filled epoxy. Carbon black was used as the additive because
of its high dielectric properties. The effect of carbon black concentration on microwave
heating efficiency and curing kinetics of epoxy is studied and the reason of reaction rate

enhancement with microwaves — thermal or non-thermal, is discussed.

7.2 Hypothesis of Carbon Effect on Microwave Curing

The effect of carbon black on microwave absorption by polymer reactants is
illustrated in Figure 7.1. In microwave curing of the neat resin (Figure 7.1(a)),
microwaves are absorbed by the functional groups, dissipated into heat and then the heat
is transferred to the entire molecules. Therefore the localized temperature of the
functional groups should be higher than that of the bulk, though the degree of localized
superheating is uncertain. When carbon black is filled into the resin (Figure 7.1(b)), much
less microwaves are absorbed by the functional groups because most microwaves are
absorbed by carbon owing to its much higher dielectric property. Heat is then transferred
from carbon to the resin molecules and the functional groups of the resin. Thus the
heating mechanism of the functional groups was actually similar to a thermal heating
process when carbon black is present. Carbon black weakens the localized super-heating
effect of the functional groups. Therefore, if localized super-heating was the main reason
of curing rate enhancement with microwaves, then the curing rate should decrease with

increasing carbon concentration.

115

 

Polar group

W Microwaves
Heat transy

Polymer molecule

    

 

 

 

(a) Microwave interaction with neat epoxy

 

Polar group Carb
on

\ OW
‘/ Microwaves
Heat transfer

Polymer molecule

 

 

 

(b) Microwave interaction with epoxy filled with carbon

Figure 7.1 Microwave Interactions with Carbon Black and Epoxy

7.3 Epoxy Curing Kinetics

Epoxy resins are the most widely used matrix materials for advanced composites.
A large amount of work has been performed in the curing of the general class of epoxy
resins. A variety of models were proposed for thermal curing of neat epoxy resins. These
models have been further applied to thermal curing of doped resins and microwave
curing of neat resins. Research efforts on the curing kinetics of epoxy and other

commonly used resins (such as vinylester and polyester) are reviewed as follows.

116

7.3.1 Kinetics study of thermal curing of neat resin

There are mainly two categories of kinetics models for the curing process [102].
One is mechanistic model, which is obtained based on reaction mechanisms. The other is
phenomenological model, which is developed without considering the details of cure
reactions. Mechanistic models offer the advantages of better prediction and interpretation
without conducting cure experiments for each new variable in the cure system. However,
the mechanistic models usually take more complex forms with more kinetic parameters
than the phenomenological models. In addition, the complexity of cure reactions
sometimes makes the derivation of mechanistic models very difficult or even impossible.
On the other hand, phenomenological models have the advantages of relatively simpler
forrrr than the mechanistic models. Therefore phenomenological models have been used
in most studies of cure kinetics. A summary of mechanistic and phenomenological

models for cure reactions is presented as follows.

7.3.1.1 Mechanistic Models

The proposed reaction kinetic mechanism for epoxy-aromatic diarrrine system is

[103]:

K1
a1+e—>a2+0H

K1'

K2
a2+e—-)a3+0H (7.1)

K2'
K3

0H +e—)et+0H
K3'

117

Where a, a2, a3, e, and et are primary amine, secondary amine, tertiary amine, epoxide,
and ether group, respectively. Ki and Ki, i=1,2,3, are specific reaction rate constants for
the catalytic and non—catalytic reactions, respectively. From the kinetic mechanism,
mechanistic models for the curing process can be derived.

For the simplified case of no etherification, steric hindrance or OH impurity, a

cure kinetics expression for epoxy has been derived as follows [104]:

d
71:1 =(k,+k2a)(1—a)(B —a) (7.2)

where B is the ratio of the initial hardener equivalents to epoxide equivalents. For a
stoichiometric mixture, B=1.
0: is the extent of cure,

k; and k2 are the catalytic and non-catalytic polymerization reaction rate constants,

respectively.

The above equation holds well up to the gelation point. To model the whole
curing reaction, the following kinetics model has been proposed [105], which models the
reaction kinetics before gelation with equation 7.2 and models the reaction kinetics after

gelation with an equivalent first order reaction:

da

gt— : (k1 + kza'Xl - Q'XB " a) , when a < agel
da
:1;— = k3 (1 — a). when a > age, (7-3)

118

Where k3 is the first order reaction rate constant with Arrhenius temperature dependency
and age] is the extent of cure at the gelation point.

It has been shown that for a stoichiometric mixture of epoxide and amine, the
etherification can be neglected at low curing temperatures [106-111]. However, the
etherification can no longer be ignored at high curing temperatures or with excess
epoxide [110,111]. In addition, the reaction rate constant for primary and secondary
amine is not always the same. For the generalized case of epoxy curing with
etherification, steric hindrance and an OH impurity, the curing kinetics has been derived
for a stoichiometric mixture of epoxide and amine [111]. The kinetics models are shown

in the following equations:

 

 

 

 

d 21— "’2
—9’-=1( "MW” +LF(¢)1(1-a>1k1+k2F(¢)1 (7.4)
dt 2—n
n/2
W) =1+ [OHIO _(1-n)¢+¢ (75)
e0 2-n
(1—¢)<1-n)(2-L)+2(1—¢"’2)(1--’1>—(2—n)L(1+[0H]°>1n¢
_ n 80
“7 2(2—n)
(7.6)

Where n is the reaction rate constant ratio between the secondary amine-epoxy reaction
and the primary amine-epoxy reaction, n = K2/K1 = K2'lKr',
L is the reaction rate constant ratio between the etherification and the primary

amine-epoxy reaction, L= K3/K1 = K3'/Ki',

119

[OH]0 is the initial concentration of OH impurity,
¢=a]/Co,
e0 is the initial epoxide concentration,

k 1 =60Kj ' and k2=602K1.

If L=0 (i.e. no etherification), n=1 (i.e. no steric hindrance) and [OH]0=0 (i.e. no
OH impurity), then the above reaction kinetics simplifies into the following equation for

a stoichiometric epoxy-amine mixture:

id? = (k1 + kzaXl -a)2 (7.7)

This kinetics equation is consistent with equation 7.2, because in equation 7.2 B=1 for a

stoichiometric mixture.

7.3.1.2 Phenomenological Models

The simplest phenomenological model is the nth order reaction kinetics model

[112,113,114], which assumes that the kinetics can be expressed as:

195: lemma (18>
dr

where or is the extent of cure, t is the time, the function flu) is expressed as (1-a)“, and

k(T) is the overall reaction rate constant which obeys the Arrhenius relation:

120

k(T) = Aexp(—E—;R-) (7.9)

The nth order reaction kinetics is computationally simple. According to this

model, the maximum reaction rate should occur at the beginning of the reaction.
However, in real cases, 0t=0.3 ~ 0.4 at maximum reaction rate, which is better explained
by the autocatalyzed reaction mechanism [115,116]. The reactions between amines and
epoxide are autocatalyzed by the hydroxide group formed in the reactions. The initial rate
should be slow due to lack of catalytic hydroxide groups. The cure kinetics expression of

autocatalyzed reaction for a stoichiometric reactant mixture is given by:

d
It? = (k1 + kzam )(1 - a)" (7.10)

where k1 is the non-catalytic polymerization reaction rate constant, k2 is the autocatalytic
polymerization reaction rate constant, or is the autocatalyzed polymerization reaction
order, and n is the non-catalyzed polymerization reaction order. This model has been
widely used to represent adequately the cure kinetics of epoxy and unsaturated polyester

cure systems [115-125].
7.3.2 Kinetics study of thermal curing of filler-added resins

In commercial thermosetting polymer systems, fillers are generally used in the

resin formulation to improve the thermal, mechanical and/or electrical properties of the

121

cured product. Knowledge of the effects of fillers on the curing kinetics is necessary for
designing cure cycles and optimizing part properties. The kinetic models developed for
neat resin curing have been applied to the curing of filler-added resins. A variety of
filler/resin systems have been investigated, including calcium carbonate filler/styrene and
polyester resins [126] [127], aluminum/polyester [126], glass/polyester [126][128][129],
kaolinite clay/polyester [129], chopped glass fiber/vinylester [130], and carbon black and
silica fillers/epoxy resin [131]. The effects usually depend on specific filler and resin
systems studied.

It was reported that poly(vinyl acetate) (PVAc), when used as low profile
additives in curing unsaturated polyester resin, decreases the rate of cure and ultimate
extent of cure [132] and decreases the total heat of cure [133]. It was indicated that the
major effect of PVAc was the reduction of the radical generation rate, which was
possibly because that PVAc acted as a radical scavenger.

Experimental evidence [134] has shown that both the thermal conductivity and
the viscosity of the resin increased at the presence of a high quantity of filler. These
effects enhanced the curing reactions and the cure started at a lower temperature with
increasing filler content. However, fillers reduced the concentration of functional groups
per unit volume. This effect resulted in a decrease in the total heat of cure with increasing
filler content.

It was observed that the addition of CaCO3 to the styrene/polyester system led to a
decrease in the induction period without any significant effect on the subsequent curing
rate [127]. The possible reason for this effect was that the filler-rich phase preferentially

adsorbs inhibitors, including dissolved oxygen that acted as a co-inhibitor.

122

A model to describe the effect of particulate fillers on the heat transfer and cure
process was developed with the assumption that the presence of filler did not alter the
reaction kinetics and that the thermal characteristics of the composite are adequately
described by the "spherical inclusion model" [126]. A simple 11'11 order model was used to
examine the effect of fillers (glass, calcium carbonate and aluminum) on the time-
temperature profiles, peak temperatures, extent of cure, and gel time of a polyester resin
system. It was shown that the filler content significantly affected the progression of the
cure fronts through the thickness of the mold. In addition, the type of filler affect the ratio
of polymerization rate to the rate of heat transfer, which resulted in the dependence of the
minimum gel time on the type of filler.

The influence of kaolinite clay and glass fiber on the curing of a polyester resin
was investigated [129]. It was reported that the fillers did not affect the total heat of
reaction per unit mass of the resin or the peak temperature. In addition, the fillers had
very little effect on the progression of the curing for high temperature peroxide initiation.
However it was observed that for room temperature curing of polyester system, glass
fillers had a severe inhibiting effect on the curing [135].

The effect of chopped glass fiber on the curing of a polyester system was
investigated [128]. It was reported that the glass fibers resulted in some changes in the
overall reaction rate of the polyester cure system but had little effect on the reaction
exponents. In addition, the glass fibers did not result in significant changes in the ultimate
heat of cure per unit mass of the resin or final degree of conversion. However, in the
curing of vinylester resin, the glass fibers significantly decreased the ultimate extent of

cure [130]. The reason for this inhibition was partly because of the chemical structure of

123

the vinylester resin and the sizing agent of the reinforcement which was formulated for a
polyester resin.

The carbon black filler had a significant effect on the thermal curing reaction rate
of DGEBA/mPDA [122]. It was observed that the sum of the kinetic exponents is
approximately 2, independent of temperature and filler concentration. Carbon black
influenced both the catalytic and non-catalytic reaction rate constants. For all the
temperature studied, the catalytic and non-catalytic reaction rate constants first decreased
to a minimum value at around 2wt% of carbon concentration. As carbon concentration
further increased, the rate constants increased and finally leveled off. The possible reason
for the carbon effect on epoxy curing was attributed to the chemical complexes on the
carbon black surface. These chemical groups, such as phenolics, carboxylics, quinones,

hydroquinones, and lactones, might catalyze the curing of epoxy.
7.3.3 Kinetics study of microwave curing of neat resin

Thermal cure kinetics models have been used in modeling the reaction kinetics of
microwave cured epoxy resins [15,111,101]. It was demonstrated that the cure kinetics
of DGEBA/mPDA and DGEBA/DDS systems could be described by the autocatalytic
kinetic model up to vitrification in microwave curing process [15, 111]. In the
comparative study of continuous-power and pulsed-power microwave curing of epoxy

resins [101], a semi-empirical kinetic model was used:

da
E =(k1+k2am)(au -a)" (7.11)

124

where or is the extent of cure, k1 and k2 are rate constants, m and n are constants, and 01.,
is the ultimate extent of cure. This model is similar to equation (7.10) except that the I
ultimate extent of cure or“ is included in the equation. This is because at certain stage of
the reaction, gelation and vitrification take place, and the reaction rates are controlled by

physical deposition. Due to these physical transitions, the ultimate extent of cure is

usually less than 100%.

7.4 Kinetics Model Used in this Study

In this study, neat and carbon-filled epoxy systems are cured with both thermal method
and microwaves. The phenomenological kinetics model with the form of equation 7.10 is

used. The reaction rate constants RI and k2 obey the Arrhenius relation:

Er
ki (T) = At CXp(-'E) (7.12)

where i = 1 (for non-catalytic polymerization reaction),

2 (for autocatalytic polymerization reaction).

In Equation 7.12, A, and E, are the Arrhenius frequency factor and the activation
energy, respectively. E; and A, can be obtained from the reaction rate constants at
different temperatures. If plotting Ln(ki) vs. —1/RT and performing linear regression, then

the slope is E and the intercept is Ln(Ai).

125

When carbon black filler is used in the resin, k] and k2 are the overall reaction rate
constants of the filled resin. Let km and k20 represent the reaction rate constants of the
neat resin. The relationship between k1 and km, and k2 and k20 can be described in the

following equations:

k1: kw * f. (7.13.1)

k2 = k2o * 13 (7.13.2)

where f1 and f2 govern the effect of carbon black on the resin curing. If the effect of
carbon black varies at different temperature, then f 1 and f2 are functions of both carbon

concentration and temperature:

f1 =f1(c, T) (7.14.1)

f2 = f2(c, T) (7.14.2)

where c is carbon concentration and T is temperature.

The reaction rate constants and the kinetic exponents can be obtained with a least-
square fit of experimental data (extent of cure vs. time) to the kinetic model equation.
Then from the rate constants at different carbon concentration, the expressions for MC, T)

and f2(c, T) can be obtained.

126

7.5 Experimental

7.5.1 Materials and Sample Preparation

The epoxy resin used in this study was diglycidyl ether of bisphenol A (DGEBA)
/ diaminodiphenyl sulfone (DDS). The DGEBA used was DER332 from Dow Chemical
with an epoxy equivalent weight of 173. The curing agent DDS was from TCI America
with an amine equivalent weight of 62. Conducting additives used were carbon black
powder from Mallinckrodt Baker Inc. The average particle size of the carbon black was
75pm. All of the materials were used as received without further purification. In
preparing the neat epoxy resin, stoichiometric DGEBA/DDS (2.79 DGEBA: 1 DDS by
weight) were mixed at 130°C. The mixture was well stirred at 130°C until DDS was
completely dissolved (in approximately 5 rrrinutes). Doped resins were also prepared with
carbon black concentrations of up tolOwt%. The resins were degassed at 0.02bar and
100°C for 5 rrrinutes. Fresh samples were kept in a -20°C freezer and used within 2
weeks. The fresh samples were analyzed with Differential Scanning Calorimetry (DSC)
to determine the total heat of reaction per gram of resin. Results are shown in Table 7.1.
As the carbon black concentration increased from 0 to 10wt%, the total reaction heat per
gram of resin had a slightly decreasing trend. This was because an increase in additive

concentration corresponded to a decrease in the concentration of the reacting species

DGEB A/DDS.

127

Table 7.1 Total Heat of Reaction per Gram of Resin of the Fresh Samples

 

Materials Neat Resin Doped Resin Resin Resin Doped

DGEBA/ with 1wt% of Doped with Doped with with 10wt%

 

DDS Resin Carbon 2wt% of 5wt% of of Carbon
Black Carbon Carbon Black
Black Black
Total Heat
of

Reaction 413.9:5.1 403.4i13.2 398.5:9.l 403.1i12.6 381.9:t6.3
perGram

of Resin

(J/g)

 

 

 

 

 

 

 

 

7. 5.2 Experimental Setup for Microwave Curing

The microwave curing circuit is the same as that used previously in adhesive
bonding experiments, as shown in Figure 3.1. A cylindrical single mode cavity with a
diameter of 17.78cm was used for microwave curing. The coupling probe was side
mounted 3cm above the bottom of the cavity. The cavity length and the probe depth were
adjusted to be 13.2cm and 2.0cm, respectively. The sample was loaded at the center on
the bottom plate of the cavity.

For each microwave curing experiment, 0.1:t0.003 grams of resin was loaded into

a cylindrical Teflon holder, the inner diameter of which was 1.25cm. The thickness of the

128

resin was around 0.6mm. Because of the small dimensions of the sample, the temperature
within the resin was assumed as uniform. The temperature of the resin was measured
with a Nortech NoEMI-TS fiberoptic thermometer. A LabVIEW program was developed
for data acquisition of temperature, incident power and reflected power and to perform
process control. For thermal curing experiment, 0.1:0003 grams of resin was applied on
a cylindrical Teflon disk with a diameter of 1.25cm. The samples were put into the oven

after the oven was preheated to the curing temperature.

7.5.3 Dielectric Measurement

The dielectric properties of the carbon black, uncured neat and doped resins were
measured with single mode perturbation method at room temperature. The dielectric
measurement method was described in detail in literature [1]. The diagnosis mode used
was TM012 at 2.45GHz. The microwave applicator for the dielectric measurement was a
cylindrical single mode cavity with a diameter of 15.24cm. A cylindrical Teflon holder
was used to contain the resins. The inner diameter and height of the Teflon holder were
1.0cm and 3.5cm respectively. The cavity length corresponding to the TM012 mode was
around 15.43cm at 2.45GHz. Results of the dielectric constant and loss factor are shown
in Tables 7.2 and 7.3, respectively. The dielectric properties increased noticeably with
carbon concentration. Compared with the dielectric properties of the neat resin, the
dielectric constant and loss factor of the resin doped with 10wt% of carbon black
increased by around 100% and 50%, respectively. With the measured values of carbon
black and neat resin, the dielectric constants and loss factors of doped resins are predicted
with mixing rule and also shown in Tables 7.2 and 7.3. For lightly doped resins (e.g. 1%

and 2%), the mixing rule holds with a relative error of less than 10%. But for heavily

129

doped resins (e.g. 5% and 10%), the relative error can be as high as 30% and the mixing

rule does not hold any more.

Table 7.2 Dielectric Constants of the Materials

 

 

 

 

 

 

 

Materials Dielectric Dielectric Constant Relative
Constant 8’ 8’ Predicted with Error
Measured Mixing Rule
Carbon Black 26.631054 \ \
Neat DGEBA/DDS Resin 4.211025 \ \
Resin Doped with 1wt% of 4.761018 4.4342 -6.8%
Carbon Black
Resin Doped with 2wt% of 4.781024 4.6584 -2.5%
Carbon Black
Resin Doped with 5wt% of 6.761031 5.331 -21.1%
Carbon Black
Resin Doped with 10wt% 8.601027 6.452 -25.0%
of Carbon Black

 

 

 

 

 

 

130

Table 7.3 Dielectric Loss Factors of the Materials

 

 

 

 

 

 

 

Materials Dielectric Dielectric Loss Relative
Loss Factor 8” Factor 8” Predicted Error
Measured with Mixing Rule
Carbon Black 2.351005 \ \
Neat DGEBA/DDS Resin 02110.03 \ \
Resin Doped with lwt% 02110.01 0.2314 10.2%
of Carbon Black
Resin Doped with 2wt% 02410.03 0.2528 5.3%
of Carbon Black
Resin D0ped with 5wt% 0.261002 0.317 21.9%
of Carbon Black
Resin Doped with 10wt% 0.321003 0.424 32.5%
of Carbon Black

 

 

 

 

 

 

7. 5.4 Cavity Characterization

Before microwave curing, the loaded cavity was characterized to locate the
heating modes. The mode spectrum, as shown in Figure 7.2, was obtained with
measuring the incident power (Pi) and the reflected power (Pr) as a function of
frequency. The frequency with minimum reflectance (Pr/Pi) was the resonant frequency

of a mode. Among many available electromagnetic modes, a center heating mode TM020

131

was selected because the material sample was loaded at the center of the cavity. For the

experimental setup in this study, the resonant frequency of TM020 was around 2.89 GHz.

 

0 I 1 I I
2.4 3.9

 

2.9 3.4
Frequency (GHz)

 

 

 

Figure 7.2 Mode Spectrum of Loaded Cavity for Microwave Curing

7.5.5 Process Control Strategy and Temperature Profile in Microwave Heating

In this study, the resins were cured at 145, 165 and 185°C. Traditional
proportional-integral—differential (PID) method was used to control the curing
temperature by adjusting the incident power level. The PID controller was programmed
with labview as a subroutine. The three parameters Kc, Ti and Td were obtained with
Ziegler-Nichols frequency response method. The details of the method are described in

literature [99]. The controller parameters are listed in Table 7.4.

132

Table 7.4 PID Controller Parameters Obtained with Ziegler-Nichols Method

 

 

 

 

 

 

 

 

Ku Tu Kc Ti (0.5Tu) Ti (0.12Tu)
(second) (0.6Ku) (second) (second)
Mode TM020 3.0 11.5 1.8 5.75 1.38

 

The typical temperature profiles during microwave heating and curing of both

neat and filled resins are shown in Figure 7.3. With the PID parameters shown in Table

7.4, the curing temperature was controlled within 1°C of the set point temperature.

 

 

 

 

 

 

 

 

 

 

A 200 A AAAAAAAAAAAAAAAAAAAAAAA AA 1
L50 . . . , , , . . ,
3
8
g 100 A 185 degree C
E- 50 1:1 165 degree C
E2 0 145 degree C
0 r r r
0 20 40 60
Time (rrrinutes)

 

 

 

Figure 7.3 Temperature Profiles during Microwave Heating and Curing at 145, 165 and

185°C

133

 

7.6 Results and Discussion

7. 6.1 Microwave Power Deposition During Heating and Curing

During microwave heating and curing, data acquisition of incident and reflected
powers was performed every second. The power deposition was calculated as the
difference between the incident and reflected powers. The reflected power was close to
zero for the experimental setup in this study.

The power depositions during microwave curing at 145°C are shown in Figure
7.4. The vertical axis was the average power deposition in every 5 minutes. Each curve
was the average of 3 sets of experimental data. The shapes of the power deposition curves
for all the resin systems were similar. At the beginning, high power levels were required
to heat up the materials from room temperature to the isothermal curing temperature. The
heating up of the resins took approximately 1.4 rrrinutes. After the isothermal curing
temperature was reached, the power level dropped at the beginning of curing and then
started increasing after 15 minutes (for neat resin and resin doped with lwt% of carbon)
or 20 minutes (for resins doped with 5wt% and 10wt% of carbon black). This
phenomenon can be explained with the exothermal heat generation and dielectric
property change during the curing process. It has been shown that the curing of
DGEBA/DDS was governed by an autocatalyzed reaction mechanism and the maximum
curing rate usually occurred at the extent of cure of 0.3-0.4. Therefore at the beginning of
curing, the curing rate kept increasing until the maximum rate was reached.
Correspondingly the rate of heat generation from the exothermal curing reaction

increased. This tended to lower the power requirement. However, the decrease in

134

dielectric properties in cross-linking process tended to increase the power requirement.
Therefore at the beginning of curing, the trend of power requirement was determined by
the dominating factor of the two causes. As observed from Figure 7.4, the power level
decreased at the beginning of curing, showing that the dominating factor was the
exothermal heat generation. The minimum power might occur at the maximum reaction
rate. After the maximum curing rate was reached, the curing slowed down and the
decrease in the rate of exothermal heat generation caused an increase in the power
requirement. At the same time, the decrease in dielectric properties also required higher
power level. This explained why the microwave power deposition increased at later time
of the curing process.

The shapes of the power deposition curves at 165 and 185°C were similar to that
at 145°C. But the power level was higher at higher temperature.

The total energy consumed during microwave heating of each resin system (0.1
gram of neat or doped resins) was computed with numerical integration of the power
data. Figure 7.5 shows the total energy consumed by each resin system at different
temperatures in the first 60 minutes of microwave heating.

As observed from Figure 7.5, the increase in the carbon concentration led to a
decrease in the energy requirement at all the three curing temperatures. One possible
reason of this phenomenon was the aggregation of carbon particles in the epoxy resin.
The carbon density is higher than the epoxy density, so the carbon particles tend to
aggregate into the bottom of the resin, as shown by Environmental Scanning Electron
Microscope (ESEM) analysis. Though this was not a serious problem because the density

difference was not very large, some carbon particles were surrounded by others and

135

might not absorb microwave energy. This might result in decrease in power requirement.

Another possible reason for lower energy consumption at higher carbon concentration

was that the conducting additives improved the dielectric properties of the resin and thus

enhanced the microwave heating efficiency. Consequently lower power level was

required to maintain the same isothermal curing temperature.

 

20

 

18*

l6~

14.

12 «o

10-

Power (W)

 

o Neat Resin

0 Resin Doped with lwt% Carbon
A Resin Doped with 5wt% Carbon
x Resin Doped with 10wt% Carbon

 

 

 

20

I I I

40 60 80 100

Time (min)

 

 

Figure 7.4 Power Curves during Microwave Heating and Curing

136

 

 

 

 

 

80 ,
g 70
g 60 I
DO
1;; 50 i I D 145 degree C
in a
‘2 40 __ § g I A 165 degreeC
g 30 _ o 185 degree C
g 20 a D 1: Error Bar
3 10 .
0 T l I I I
0 2 4 6 8 10 12
Carbon Concentration (wt%)

 

 

 

Figure 7.5 Total Energy Consumed by Bach Resin System in the First 60 Minutes of
Microwave Heating

7. 6.2 Comparison of Reaction Rates between Microwave and Thermal Curing of

Neat and Doped Resins

Microwave and thermal curing of the neat and doped resins were performed at
145, 165 and 185°C. The extent of cure of the resins was tested with DSC as a function
of curing time. Three samples were measured for each time point. The average standard
deviations of the extent of cure for thermal and microwave curing processes are shown in
Table 7.5. The average extent of cure was used in data regression to determine the
kinetic parameters. In the data regression, the kinetic parameters were first assigned
initial values and Equation 7.10 was integrated numerically with the fourth-order Runga-
Kutta method to regenerate the extent of cure as a function of time. The least square
method was then used to minimize the difference between the experimental data and the

calculated values by solving for the reaction kinetic parameters. The experimental and

137

calculated values of the extent of cure are shown in Figures 7.6-7.8 for thermal curing
and in Figures 7.9-7.11 for microwave curing. Markers and lines represent experimental
and calculated values, respectively. It can be seen that both microwave and thermal
curing curves have the typical shape of autocatalytic reaction. The initial curing rate was
slow due to lack of catalytic hydroxide groups. As the reaction proceeded, hydroxide
groups were generated and the maximum curing rate occurred at the extent of cure of
around 0.3 to 0.4.

In both thermal and microwave curing, no obvious trend was found for the
reaction order constants m and n as function of filler concentration. The average values of

m and n for thermal and microwave curing are shown in Table 7 .6.

Table 7.5 Average Standard Deviations of the Extent of Cure

 

 

Curing Conditions Thermal Curing Microwave Curing
Average Standard Deviation 3.4% 5.5%
of the Extent of Cure

 

 

 

 

 

Table 7.6 Values of M and N

 

Thermal Curing Microwave Curing

 

m 0.831015 0.94101 1

 

n 1.211010 1.571021

 

 

 

 

 

138

 

 

Thermal curing
T=145 degree C

 

Extent of Cure

 

 

 

 

I I

 

 

0 20 40 60 80 100
Time (min)
I O neat resin-exp - - - - neat resin-reg
X 0.5wt% carbon-exp - '- 0.5wt% carbon-reg
+ lwt% carbon-exp ------- lwt% carbon-reg
X 2wt% carbon—exp 11.11 2wt% carbon-reg
A 5wt% carbon-exp - ~ - - - 5wt% carbon-reg
0 7wt% carbon-exp - - - - - 7wt% carbon-reg
D 10wt% carbon-exp 10wt% carbon-reg

 

 

 

 

exp: experimental data

reg: regression data

Figure 7.6 Thermal Curing at 145°C

139

 

 

 

Extent of Cure

Thermal curing, T=l65 degree C

 

 

 

 

 

I I I

 

 

20 40 60 80
Time (rrrin)

O neat resin-exp - - - neat resin-reg

x 0.5wt% carbon-exp 0.5wt% carbon-reg
+ lwt% carbon-exp ----- lwt% carbon-reg

x 2wt% carbon-exp ~ 2wt% carbon-reg

A 5wt% carbon-exp - - — - 5wt% carbon-reg

o 7wt% carbon-exp - - - - 7wt% carbon-reg

D 10wt% carbon-exp -—-10wt% carbon-reg

 

 

exp: experimental data

reg: regression data

Figure 7.7 Thermal Curing at 165°C

140

 

 

 

 

Extent of Cure

Thermal curing
T=185 degree C

 

 

 

 

 

 

10 20 30 40 50
Time (min)

neat resin-exp — — - — neat resin-reg
0.5wt% carbon-exp 0.5wt% carbon-reg
lwt% carbon-exp ------- lwt% carbon-reg
2wt% carbon-exp -—-——- 2wt% carbon-reg
5wt% carbon-exp - - - - - 5wt% carbon-reg
7wt% carbon-exp — - - — - 7wt% carbon-reg
10wt% carbon-exp —— 10wt% carbon-reg

 

 

exp: experimental data

reg: regression data

Figure 7.8 Thermal Curing at 185°C

141

 

 

 

 

Microwave curing
T=145 degree C

 

Extent of Cure

 

 

 

 

 

 

0 r r r r
0 20 40 60 80 100
Time (min)

0 neat resin-exp - - - - neat resin- reg
+ lwt% carbon-exp ------- lwt% carbon-reg
X 2wt% carbon-exp -~-~——-—- 2wt% carbon-reg
A 5wt% carbon—exp - - - - - 5wt% carbon-reg
D 10wt% carbon-exp —- 10wt% carbon-reg

 

 

exp: experimental data

reg: regression data

Figure 7.9 Microwave Curing at 145°C

142

 

 

 

Microwave curing
T=165 degree C

 

 

 

 

 

 

 

0 r r r
0 20 40 60 80
Time (rrrirr)
O neat resin-exp - - - - neat resin-reg
+ lwt% carbon—exp ------- lwt% carbon-reg
X 2wt% carbon-exp 2wt% carbon-reg
A 5wt% carbon-exp - - - - - 5wt% carbon-reg
D 10wt% carbon-exp —— 10wt% carbon-reg

 

 

 

exp: experimental data

reg: regression data

Figure 7.10 Microwave Curing at 165°C

143

 

 

Extent of Cure

Microwave curing
T=185 degree C

 

 

 

 

I I I I I

 

 

10 20 30 40 50
Time (min)
0 neat resin-exp - - - - neat resin-reg
+ lwt% carbon-exp ------- lwt% carbon-reg
X 2wt% carbon-exp -----— 2wt% carbon-reg
A 5wt% carbon-exp - - - - - 5wt% carbon-reg
El 10wt% carbon-exp —- 10wt% carbon-reg

 

 

exp: experimental data

reg: regression data

Figure 7.11 Microwave Curing at 185°C

144

 

 

 

 

7.6.2.1 Comparison of Reaction Rates between Microwave and Thermal

Curing of Neat Resin

The calculated rate constants of neat resin curing with microwaves and thermal
method are shown in Table 7.7. For both microwave and thermal curing, the magnitude
of the rate constant k2 was higher than that of the reaction rate k]. This verifies that the
epoxy curing is mainly governed by autocatalytic reaction mechanism. Compared with
thermal curing, microwaves enhanced both k1 and k2, but preferentially enhanced the
non-catalytic reaction rate k]. This result indicated that microwaves preferentially
enhanced the activity of amine and epoxide groups and accelerated the non-catalyzed

reaction between epoxide and amine.

Table 7.7 Reaction Rate Constants of Microwave and Thermal Curing of Neat Resin

 

 

 

 

 

 

 

 

 

 

 

N on-catalytic Autocatalytic k1 / k2

reaction rate reaction rate

constant k. constant k2
Microwave 145°C 0.016 0.084 0.19
Microwave 165°C 0.040 0.16 0.26
Microwave 185°C 0.089 0.24 0.37
Thermal 145°C 0.0024 0.065 0.037
Thermal 165°C 0.0040 0.10 0.040
Thermal 185°C 0.0068 0.18 0.039

 

145

 

7.6.2.2 Effect of Carbon Black Concentration on Thermal Curing Rates

The reaction rate constants k. and k2 at different carbon concentrations were
obtained with the data regression method mentioned earlier. The rate constants for
thermal curing are shown as the markers in Figure 7.12 (a) and (b). The rate constants
increased with increasing carbon concentration. This might be explained with the carbon
surface chemistry. X-ray Photoelectron Spectroscopy (XPS) analysis of carbon surface
has shown the presence of functional groups such as carboxide, hydroxide, carboxyl, and
ester groups. These groups might have catalytic effect on the curing.

From the reaction rate constants at different temperatures, the activation energy E,
of thermal curing was calculated with linear regression, as shown in Figure 7.13. Since
the activation energy changes with carbon concentration, the effect of carbon black on
resin curing depends on both carbon concentration and temperature. In modeling of

thermal curing, the following expression was used:

 

k- A E- E-
fr(c,T) = ~4— = -—'—-exp(—-—'——'9-)
[‘10 10 RT
_ 5-5.
:(I‘I'dilC'I'dizcz+di3C3)CXp( ( [RT I0)) (7.15)

where i = l (for non-catalytic polymerization reaction),

2 (for autocatalytic polymerization reaction).

In this equation, f;(c, T) is the ratio of the rate constant of the doped resin to that

of the neat resin. The rate constants obey the Arrhenius relation. A, and A10 are the pre-

146

 

exponential factor in the curing of doped and neat resins, respectively. E and Bio are the
activation energy in the curing of doped and neat resins, respectively. For thermal curing,
f,(c, T) had an exponential dependence on the temperature, shown by the Arrhenius
relation. The ratio of the pre-exponential factor Ai/Aio only depended on the carbon
concentration and a polynomial function was used to correlate the ratio AilAio. The
coefficients d's in the polynomial were obtained with least-square method and are shown
in Table 7.8. The calculated coefficients were then used to regenerate the reaction rate

constants k] and k2 as functions of carbon concentration, as shown in Figure 7.12 (a) and

(b) (lines).

Table 7.8 Coefficients D's for Thermal Curing

 

 

 

d1 1 0.0035 d21 0.0008
d12 -0.2077 d22 -0.0244
d13 0.0802 d23 0.0092

 

 

 

 

 

 

147

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 7 I I I
0 2 4 6 8 10
carbon concentration (wt%)

 

(a) Non-catalytic reaction rate kl vs. carbon concentration

 

 

 

 

 

 

 

0 2 4 6 8 10
carbon concentration (wt%)

 

 

 

(b) Catalytic reaction rate k2 vs. carbon concentration
Markers: reaction rate constants obtained from experimental data regression

Lines: reaction rate constants calculated with kinetic model Equation (7.15)

Figure 7.12 Effect of Carbon Concentration on Thermal Curing Rates

148

 

 

 

 

 

 

 

m 14 .
3612‘ . o '
810 l I
8 8h“
.5. 6—
2 4“
2'3 21
< 0 . .
0 5 10
Carbon concentration

15

 

 

0E1
IE2

 

 

Figure 7.13 Activation Energy in Thermal Curing

7.6.2.3 Effect of Carbon Black Concentration on Microwave Curing Rates

The reaction rate constants k; and k2 for microwave curing at different carbon
concentrations are shown as markers in Figure 7.14 (a) and (b), respectively. Both k1 and

k2 decreased with increasing carbon concentration. This effect was more obvious at

higher temperatures.

The activation energy of microwave curing is shown in Figure 7.15 as a function

of carbon concentration. The activation energy changed as a function of carbon

concentration.

149

 

 

 

 

 

Microwave Curing

 

0.1
0.08 3
~ 0.06 — ‘ A 185C

.2
004 145C 165C
0.02 W T
0 2

      
 

 

 

 

 

I I I I

4 6 8 10
Carbon Concentration (wt%)

 

 

 

(a) Non-catalytic reaction rate k1 vs. carbon concentration

 

Microwave Curing

0.25
A
0.2 L
015 185C
2 - 65C
0'1‘ ° " 145C _:

0.05 a

O I I f I

0 2 4 6 8 10
Carbon Concentration (wt%)

 

 

 

 

 

 

 

 

 

(b) Catalytic reaction rate k2 vs. carbon concentration
Markers: reaction rate constants obtained from experimental data regression

Lines: reaction rate constants calculated with kinetic model Equation (7.16)

Figure 7.14 Effect of Carbon Concentration on Microwave Curing Rates

150

 

 

 

 

 

 

 

 

 

 

 

20
a l
E6 15 r .9 .
LE 10 . 9E1
c. I
.2 I . I 3 E2
*5 I
.g 5 a
O
<
0 r r
0 5 10 15
Carbon Concentration (wt%)

 

Figure 7.15 Activation Energy in Microwave Curing

In modeling microwave curing, Equation 7.16 is used. The function f;(c, T) had an
exponential dependence on the temperature. This was sirrrilar to the thermal curing.
However for the ratio of the pre-exponential factor Ai/Aio, a polynomial expression alone
could not sufficiently model the data. The trend of the data also showed an exponential

dependence on the carbon concentration.

k- A- E- _ E-
f.- (61) = —‘- = —'—exp(——'——'9)
km A“) RT

= exp(—d,-0c)(l + dilc + di2c2)exp(_ (BEEN) (7. 16)

 

The coefficients d's are shown in Table 7.9.

151

Table 7.9 Coefficients D's for Microwave Curing

 

 

 

dro 1.05 d20 0.52
d” -O.55 d2] -O.59
drz 0.30 (122 0.12

 

 

 

 

 

 

If localized superheating is the main reason of rate enhancement in microwave
curing of epoxy, then the curing rate should decrease with increasing carbon
concentration. Therefore the microwave curing results suggest that the reaction rate
enhancement in microwave curing of epoxy result from localized superheating of the
polar functional groups. However, there were also other possibilities that have not been
ruled out. The carbon particles used in this study were activated and might adsorb the
amine from the epoxy resin because carbon particles might have higher temperature than
the bulk resin in microwave curing. This will also result in decrease in microwave curing
rate. In addition, the power deposition during the curing process also could affect the
curing rate. The microwave power curves in this study showed that the power level was
lower at higher carbon concentration to maintain the same isothermal curing temperature

of the bulk resin. This might also result in decrease in curing rate with microwaves.

152

 

7.7 Conclusions

The effects of carbon black concentration on microwave curing of DGEBA/DDS
were studied. The magnitude of the dielectric properties increased noticeably with

increasing concentration of carbon black. Compared with the dielectric properties of the

V. .‘t

neat resin, the dielectric constant and loss factor of the resin doped with 10wt% of carbon

black increased by around 100% and 50%, respectively. Microwave curing experiments

 

were carried out at three different temperatures with different carbon black
concentrations. Parallel thermal curing was also performed for comparison. Correlation
for reaction rate constants as functions of carbon concentration and temperature was
proposed. For the neat epoxy resin, the curing rates with microwaves were much higher
than that with thermal method. In microwave curing of the resins with different carbon
concentrations, the reaction rate constants decreased with increasing carbon
concentration. In thermal curing process, the rate constants increased with increasing
carbon concentration. These trends were more obvious at higher temperature. According
to the hypothesis made at the beginning of the paper, the presence of carbon black
weakened the localized super-heating of the functional groups with microwaves because
the conducting carbon absorbed most microwaves during the curing process. If localized
super-heating were the main reason of curing rate enhancement with microwaves, then
the curing rate should decrease with increasing carbon concentration. This study suggests
that reaction rate enhancement in microwave curing of epoxy result from localized

superheating of the polar functional groups. However, there were also other possibilities

153

that have not been ruled out. The carbon particles used in this study were activated and
might adsorb the amine from the epoxy resin because carbon particles might have higher
temperature than the bulk resin in microwave curing. This will also result in decrease in
microwave curing rate. In addition, the power deposition during the curing process also
could affect the curing rate. The microwave power curves in this study showed that the
power level was lower at higher carbon concentration to maintain the same isothermal
curing temperature of the bulk resin. This might also result in decrease in curing rate with

microwaves.

154

Vim: .
l
.

 

 

CHAPTER 8 CONCLUSIONS

In this research, work was conducted in two sections. The first section was the
development of a precisely controlled, rapid and uniform microwave adhesive bonding
system with on-line monitoring features of the bonding process in a single mode
applicator. The second section was the investigation of microwave heating mechanisms
to provide explanations for reaction rate enhancement with microwaves. i’

A single mode applicator was used throughout this study because of its high

energy efficiency, good controllability, and convenience in studying process

 

fundamentals. In a single mode microwave cavity, several processing methods can be
applied, including single mode microwave processing and mode switching microwave
processing. Single mode microwave method refers to the process that only one mode is
used throughout the processing. Mode switching microwave method refers to the process
that several modes, with complementary heating patterns, are excited sequentially to
obtain time-averaged uniform heating. The mode switching method is based on
knowledge and understanding of single mode heating characteristics.

The single mode microwave method was first applied in an adhesive bonding
process to explore the characteristics of using microwaves for adhesive bonding in a
single mode applicator. Adhesive and substrate materials were selected based on their
compatibility and dielectric properties. An epoxy-based adhesive (Eccobond A401-37)
was used because of its compatibility with a wide range of substrate materials and its
high dielectric loss properties. A substrate (Bexloy W502, major component: glass

reinforced ethylene/methacrylic acid copolymer) with very low dielectric loss properties

155

was chosen from three polymer and composite materials. With the choice of the
materials, rapid and selective heating of the adhesive was observed in the microwave
adhesive bonding process. The microwave heating mode with the desired heating pattern
was determined from cavity characterization, theoretical resonant frequency, and
theoretical electric field distributions. The bonding results showed that the bonding time
was reduced at higher temperatures for the microwave process. Compared with the
thermal method, the microwave method reduced the bonding time and at the same time
enhanced the bonding strength for the selected materials. The difference in bond strength
between microwave and thermal methods was related to the break pattern in the single
lap shear test. With sufficient bonding time, microwave bonded assemblies at different
temperatures all broke within the substrates. This phenomenon indicated that the bond
was stronger than the substrate material itself if processed with microwaves. Thermally
bonded assemblies all broke at the interface, indicating that the adhesion between the
substrates and the adhesive was not sufficiently strong. Therefore single mode microwave
method was successfully applied in adhesive bonding to obtain rapid heating with
enhanced bonding strength. However, only limited material size could be uniformly
heated with single mode microwaves because of the non-uniformity of the electric field
distribution.

To solve this non-uniformity problem using single mode heating, a variable
frequency mode switching method was implemented based on existing method and then
applied in microwave adhesive bonding of large-size materials to obtain uniform heating
in a single mode applicator. Implementations were performed with modes selected by

studying theoretical electric field patterns to reduce the number of modes used in mode

156

 

switching to simplify the mode switching algorithm. In mode switching microwave
bonding, Eccobond A401-37 (epoxy based) was again used as the adhesive to bond two
substrate materials, Bexloy W502 (major component: glass reinforced
ethylene/methacrylic acid copolymer) and in a separate process, Surlyn SG201U (Nylon
6 and ethylene/methacrylic acid copolymer). Experimental temperature profiles showed
that uniform heating of the large size materials were obtained with the mode switching

method. The bonding cycles were determined with measuring the extent of cure of the

. J 9": ”I!" A."

adhesive as a function of time for both microwave and thermal bonding processes. The

bonding strength was determined with single-lap shear test. The results were slightly

 

different for the two types of substrates. For Eccobond A401-37/Bexloy W502,
microwave process had a 75% to 80% reduction in bonding time and at least 2-fold
enhancement of bonding strength compared with thermal process. Microwave bonded
assemblies broke in the substrates in single lap shear test, while thermally bonded
assemblies broke at the interface even at 99% of cure of the adhesive. For Eccobond
A401-37/Surlyn SG201U, assemblies bonded with microwaves at 100°C for 45 minutes
obtained the same strength as thermally bonded assemblies at 120°C for 100 minutes.
Both microwave and thermally bonded assemblies broke in the substrates. Since the
interface and adhesive were stronger than the substrates for both microwave and
thermally bonded assemblies, it was uncertain whether there was microwave
enhancement of the adhesion between the adhesive and the substrate. Therefore, observed
microwave effects in adhesive bonding of the two systems include reduction of bonding

time for both systems and enhanced bonding strength for Bexloy W502 substrates.

 

 

157

In the previous bonding study, the bonding cycle had been determined by
measuring either the bonding strength (with single-lap shear test) or the extent of cure
(with DSC) vs. time. These measurements had been conducted off-line and required
tremendous amount of experimental work. A successful process calls for the development
of an on-line monitoring technique to determine the bonding cycle on-line. In this study,
a new method (with corresponding software) was developed for on-line monitoring of
variable frequency microwave bonding of large materials. This method used a high
power processing circuit to monitor the resonant frequency shifting, which resulted from
changes in material dielectric properties during microwave processing in a single mode
applicator. This method was not intrusive, able to monitor microwave processing of
materials with large sizes and complex shapes, and did not require separate diagnostic
circuit (thus avoiding hardware switching between circuits). The new on-line monitoring
method was applied in several microwave adhesive bonding processes, including bonding
of single-lap Bexloy /Eccobond, single-lap Surlyn /Eccobond, and double-lap Bexloy
/Eccobond assemblies. During the bonding process, the curing of the adhesive led to
decreases in the dielectric constant and loss factor. Resonant frequencies for the modes,
which reflected changes in material dielectric properties, were monitored during the
bonding process. When the curing rate was high, the rate of resonant frequency shifting
was also high. When the curing approached completion, the curing rate decreased and the
shifting of resonant frequencies slowed down as well and finally leveled off. Thus when
the resonant frequencies stopped shifting, the bonding was considered completed and the

bonding cycles were determined.

158

Throughout the microwave adhesive bonding study, process control was essential
for realizing uniform and stable heating and repeatable processing. A process control
system was developed in this study for single mode and variable frequency mode
switching microwave adhesive bonding processes. Compared with previous work in
microwave process control performed by other researchers, the control system developed
in this study has several advantages. First, theoretical modes were used instead of
empirical heating modes. This simplified the controller by eliminating extensive heating
characterization and data storage. Second, the, PID controller parameters were obtained
based on control theory insteadof empirical values. This eliminated the trial and error
procedure of adjusting controller parameters and also provided sound basis for automatic
tuning of the parameters. Third, an on-line monitoring technique was developed to obtain
the bonding cycle directly. The software was programmed with LabVIEW for cavity
characterization (to locate the microwave modes), data acquisition (to obtain material
temperatures and microwave powers) and process control (to control heating rate and
isothermal temperature and implement the on-line monitoring method). This process
control system was successfully used in the microwave adhesive bonding process to
provide rapid, uniform and stable heating with the ability to determine the bonding cycle
on-line. I

The obtained results in microwave bonding process study have shown that
microwave processing reduced the curing time of the epoxy based adhesive with equal or
even greater bonding strength. In this study, the mechanism of microwave fast curing of
epoxy was investigated by studying the effect of carbon additive on microwave curing of

epoxy - diglycidyl ether of bisphenol A (DGEBA) / diarninodiphenyl sulfone (DDS).

159

Carbon was used as the additive because of its high dielectric 1088 properties, which
modified microwave thermal effect without significantly affecting the non-thermal
alignment of polar groups in the electromagnetic field. During the microwave curing
process, carbon absorbed most microwaves and weakened the localized superheating of
the epoxy functional groups. If localized superheating were the main mechanism of rate
enhancement in microwave curing of epoxy, then the curing rate should decrease at the
presence of carbon. Microwave curing experiments were carried out at three different
temperatures with various carbon black concentrations. Parallel thermal curing was also
performed for comparison. Correlation for reaction rate constants as functions of carbon
concentration and temperature was proposed. In thermal curing of the epoxy with
different carbon concentrations, the rate constants increased with increasing carbon
concentration. The reason might be the functional groups adsorbed on the carbon surface
had catalytic effect on the curing reaction. In microwave curing process, the reaction rate
constants decreased with increasing carbon concentration. These trends were more
obvious at higher temperature. Thus, this study suggested reaction rate enhancement in
microwave curing of epoxy result from localized superheating of the functional groups.
However, the carbon particles used in this study were activated and might adsorb the
amine from the epoxy resin because carbon particles might have higher temperature than
the bulk resin in microwave curing. This will result in decrease in microwave curing rate.
Further study is required to elucidate this problem with the use of passivated carbon
particles or carbon with smaller surface area. In addition, at higher carbon concentration,
the required microwave power was lower to maintain the same bulk temperature. This

might also result in decrease in microwave curing rate at higher carbon concentration.

160

 

In summary, this study successfully developed a microwave adhesive bonding
process in a single mode applicator with precise control and on-line monitoring features
and at the same time contributed to better understanding of process fundamentals.

The major accomplishments of this study are itemized as follows:

(1) Assembled experimental setup for microwave adhesive bonding in a single

mode applicator.

(2) Explored for the first time the application of microwaves in adhesive bonding

in a single mode applicator.

(3) Implemented existing variable frequency mode switching method and applied

 

it in adhesive bonding of large materials.

(4) Invented an on-line monitoring technique for microwave processing of large
materials with complex shapes in a single mode cavity.

(5) Designed the process control system for microwave adhesive bonding
incorporating theoretical control algorithms and on-line monitoring of
bonding cycle.

(6) Proposed new correlations for reaction rate constants as functions of carbon
concentration and temperature in microwave curing of carbon filled epoxy.

(7) Studied for the first time the relation between microwave power levels and
curing process.

(8) Investigated microwave heating mechanisms with new approaches and
suggested that reaction rate enhancement in microwave curing of epoxy result

from localized superheating of the functional groups.

161

(9) Quantified advantages of microwave processed bonding for a few

adhesive/substrate systems.

162

CHAPTER 9 FUTURE WORK

Throughout this research, several interesting problems have emerged. In this
dissertation, the effect of material dielectric properties on the bonding process was not
quantitatively studied. To achieve selective heating of the adhesives, the dielectric
properties of the adhesives should be higher than that of the substrates. However, one
question remains to be answered - what should be the minimum difference between the
adhesive and substrate dielectric properties to obtain desired selective heating profiles?
This problem can be studied with extensive experiments, but can also be quantitatively
estimated by modeling microwave heating.

The importance of modeling microwave heating is also shown in another process.
In Chapter 7, the power deposition into the cavity was experimentally measured during
microwave curing of the epoxy resin. The shape of the power curve was qualitatively
related with the curing reaction and heat transfer. However, if the power level can be
predicted with a model during microwave curing, then smoother control can be realized.

Development of mathematical models for microwave heating is beneficial for the
process in a number of ways. In addition to setting criteria for material selection and
realizing smooth and precise control, the occurrence of hot spots and thermal runaway
can be predicted with the model and procedures can be taken to avoid these undesired
phenomena.

Microwave heating modeling has been studied by a number of investigators [137-
143]. In many cases, the problems are simplified by decoupling the Maxwell’s equations

with the heat transfer equation to obtain analytical solutions. For example in some

163

“fi‘. -— v.1... - .-

 

studies, thin materials are considered so that the electric-field can be assumed to be
constant within the material and the heat transfer equation can be solved independently
[137,138]. In some other studies, the electric—field is incorporated but is assumed to decay
exponentially with distance [139,140,141]. But for complex microwave heating systems
such as microwave drying of porous materials, analytical solutions cannot be obtained
and numerical methods have to be used [142, 143]. Jolly and Turner [142] developed a
one-dimensional model for the heating of a dielectric slab and used a finite-difference 'r"
scheme to solve the problem numerically. Later Turner and J olly [143] extended the one-

dimensional model to study microwave drying of porous materials numerically.

 

The modeling of microwave adhesive bonding is a more comprehensive problem.
The material system consists of two different materials: the adhesive and substrates.
Curing reaction takes place in the adhesive and has to be incorporated in the modeling.
The material thermal and dielectric properties have non-linear dependence on both
temperature (T) and extent of reaction (a). Material properties that affect microwave
heating include specific heat CP(T, or), thermal conductivity k(T, or), dielectric constant
8'(T, a), dielectric loss factor 8"(T, or), and electrical conductivity 0(T, 01). Because data
in handbook are only for limited temperature range for a few materials, accurate
experimental determination of these material properties over the heating temperature
range at different extent of cure need to be performed prior to the modeling.

The energy balance within a unit volume of the materials is given by:

Heat Absorbed Heat Heat from
Build-up Microwave Power Transfer Reaction
p 91 P + KVZT + pd—a(—AHr) (9.1)
3: dt

 

164

Where p is density (kg/m3), CP is heat capacity (J/kg/K), T is temperature (K), t is time

(second), K is thermal conductivity (W/K/m), or is extent of cure, and -AHt is reaction
heat (J/kg).

In Equation (9.1), the heat generation source terms include chemical reaction heat .
and microwave power absorption. The reaction heat can be obtained from the reaction
kinetics. Microwave curing kinetics of neat resin has been extensively studied in
literature and microwave curing kinetics of carbon-doped resin is investigated in Chapter
7 of this dissertation. The obtained results can be used in the heating model. The local

microwave power absorption rate P (W/m3) is:
.. - 2
P = £808 mlEl (92)

where E is the electric field strength inside the material (V/m), (o is the radial frequency,
(rad/sec), 80 is the free space permittivity, and 8' is the effective relative loss factor. The
electric field can be determined by solving the Maxwell’s equations described in Chapter
2.

With a well-designed numerical approach, Equations (9.1), (9.2) and the
Maxwell’s equations can be solved together to determine the microwave heating and

curing profiles provided that the material properties are known.

165

 

 

Many experimental results obtained in this research can be used to verify the
heating model, such as the temperature profiles in Chapters 3 and 4 and the power
deposition curves in Chapter 7.

In addition to the work mentioned above, the mechanisms responsible for bond
strength enhancement need to be investigated for the Eccobond A401-37/Bexloy W502
system. Adhesion between two polymeric materials with microwaves is a complex
phenomenon that involves multidisciplinary knowledge of microwave fundamentals,
surface chemistry, polymer properties, etc. More exploratory experiments need to be
carried out to provide explanations for the enhanced adhesion. Microwaves might modify
the substrate surface chemistry. Microwaves might also enhance the molecular mobility
at the interface to result in better adhesion.

Finally, the effect of carbon surface complex on microwave curing of epoxy
needs to be studied. A number of studies have shown microwaves enhanced the curing
rate of epoxy compared with thermal energy. In Chapter 7 of this dissertation, carbon
particles were added into epoxy to study the mechanism of rate enhancement in
microwave curing of epoxy. Carbon absorbs most microwaves and weakens the localized
superheating of the epoxy functional groups. Experimental results showed that the
addition of carbon enhanced thermal curing rate but decreased microwave curing rate of
epoxy. These results suggested reaction rate enhancement in microwave curing of epoxy
result from localized superheating of the functional groups. However, there were also
other possibilities that have not been ruled out. The carbon particles used in this study
were activated and might adsorb the amine from the epoxy resin because carbon particles

might have higher temperature than the bulk resin in microwave curing. This will result

166

 

 

in decrease in microwave curing rate. In future studies, carbon particles can be passivated
to remove the surface complexes. Then the passivated carbon particles can be used as
additives in epoxy curing to eliminate the effect of surface complexes on the curing.
Alternatively, carbon particles with smaller surface area can be used as additives in the
epoxy to reduce the effect of carbon surface complexes on the curing. Experiments can
be performed to study how these carbon particles affect thermal and microwave curing
and the results can be compared with that obtained in this dissertation. The comparison
will contribute in the elucidation of the microwave heating mechanisms and improve
understandings in microwave/materials interactions.

These future studies will generate new methods for process design, improve
understanding in process fundamentals, and enhance the industrialization of microwave

processing techniques.

167

 

 

REFERENCES

168

 

 

REFERENCES

1. J ow, J ., Microwave Processing and Dielectric Diagnosis of Polymers and Composites
Using a Single-Mode Resonant Cavity Technique, Ph.D. Dissertation, Michigan State
University (1988).

2. George, C. E., Lightsey, G. R., and Wehr, A.G., Materials Research Society
Symposium Proceedings, 124, 189 [Microwave Processing of Materials] (1988).

3. Gourdenne, A., Maassarani, A. Monchaux, H., P., and Aussudre, S., Polym. Prepr.,
Am. Chem. Soc., Div. Polym. Chem., 20(2), 471 (1979).

l!"
4. Hottong, U., Wei, J ., Dhulipala, R., and Hawley, M. C., Ceram. Trans, 21, 587
(1991).
5. Mashida, T., Matsushita, S., and Ikegarni, 1., Japanese Patent 50 138 067 (1975). I

 

6. Jullien, H. and Vant, H., Polymer 26, 505 (1985).

7. Lewis, D.A., Summers, J. D., Ward, T. C., and McGrath, J. E., Journal of Polymer
Science: Part A: Polymer Chemistry, 30, 1647 (1992).

8. Gagliani, J., Lee, R., and Wilcoxson, A. L., US Patent 4 305 796 (1981).

9. Lewis, D., Proceedings of the 39171 International SAMPE Symposium and Exhibition,
pp.l-12 (1994).

10. Wei, J., Delong, J. D., DeMuse, M., and Hawley, M. C., Polym. Eng. Sci, 33, 1132
(1993).

11. Beldjoudi, N. and Gourdenne, A., Eur. Polym. J., 24, 265 (1988).
12. Marand, E., Baker, K. R., and Graybeal, J. D., Macromolecules, 25, 2243 (1992).
13. Boey F.Y.C., and Lee, W. L., J. Mater. Sci. Lett., 9, 1172 (1990).

14. Delong, J .D., Jow, J ., and Hawley, M. C., AIChE Second Topical Conference on
Emerging Technologies in Materials, San Francisco, CA, Nov. 6-9 (1989).

15. Wei, J ., Microwave processing of epoxy resins and graphite fiber/epoxy composites in
a cylindrical tunable resonant cavity, Ph.D. Dissertation, Michigan State University
(1992).

16. Lewis, D., Hedric, J. C., Ward, T. C., and McGrath, J. E., Polym. Prepr., Am. Chem.
Soc., Div. Polym. Chem. 28(2), 330 (1987).

17. low, J ., Delong, J .D., and Hawley, M. C., SAMPE Q., 20, 46 (1989).

169

18. DeMuse, M.T. and Johnson, A. C., Proceedings of the 29'” Mcirowave power
Symposium, pp. 168-171 (1994).

19. Thuillier, FM. and J ullien, H., Makromol. Chem. : Macromol. Symp., 25, 63 (1989).
20. Mijovic, J ., Fishbain, A., and Wijaya, J ., Macromolecules, 25, 986 (1992).

21. Agrawal, R. and Drzal, L. T., J. Adhes., 29, 63 (1989).

22. Lee, W.I., and Springer, G.S., J. Composite Materials, 18(4), 357 (1984).

23. Wei, J., Jow, J., Delong, J. D., and Hawley, M.C., SAMPE J., 27(1), 33 (1991).

24. Lind, A.C., Medgyesi-Mitschang, L. N ., Kurz, J. E., McKinney, H. F., and Wear,
F.C., Mater. Res. Soc. Symp. Proc., 189, 461 [Microwave Processing of Materials II]
(1991).

25. Ippen, J ., Rubber Chem. Technol., 44, 294 (1971).
26. VanKoughnett, AL. and Dunn, J. G., J. Microwave Power, 8, 101 (1973).
27. Wang, C.S., Rubber Chem. Technol., 57, 134 (1984).

28. Methven, J .M., Proceedings of the 39'h International SAMPE Symposium and
Exhibition. PP. 25-33 (1994).

29. Paulauskas, F.L., Meek, T.T., and Warren, C.D., MRS Symposium Proceedings, 430,
193 (1996).

30. Paulauskas, F.L., Mcmillan, A.D., and Warren, C.D., MRS Symposium Proceedings,
430, 493 (1996).

31. Wei, J ., Paulauskas, FL, and Johanson, W.G., “Industrial Processing Via Variable
Frequency Microwaves. Part I: Bonding Applications”, (1998).

32. Asmussen, J ., Lin, H. H., et al., Rev. Sci. instrum, pp.1477-l486 (1987).

33. Adegbite, V., Automation and control of the microwave processing of composite
materials, Ph.D. Dissertation, Michigan State University (1995).

34. Fellows, L. A., Delgado, R., and Hawley, M. C., 26th International SAMPE
Technical Conf, Atlanta, GA, Oct. (1994).

35. Qiu, Y., Variable frequency microwave processing and microwave process control
for polymer composites, Ph.D. Dissertation, Michigan State University (2000).

36. Chen, K.M., Electromagnetic Theory 1, Class Notes, Department of Electrical
Engineering, Michigan State University (1995).

170

*1”.____j

 

37.
38.

39.
40.

41.
42.

43.

45.
46.
47.
48.

49.
50.

51.
52.
53.
54.
55.
56.
57.

Harrington, R.F., Time-harmonic electromagnetic fields, Publisher: McGraw-Hill,
New York (1961).

Ku, CC. and Liepins, R., Electrical Properties of Polymers: Chemical Principles,
Macmillan Publishing Company (1987).

Saltiel C. and Datta, A. K., Advances in Heat Transfer, 33, l (1999).
White, W. C., Proc. Inst. Radio Electron. Eng., August, 50, 1129 (1962).

Chabinsky, I. J ., Materials Research Society Symposium Proceedings, 17, 124 (1988).

Krieger, B., Proc. Am. Chem. Soc., Div. Polym. Mater. : Sci. Eng., 66, 339 (1992).

Lewis, D., Mater. Res. Soc. Symp. Proc., 269, 21 [Microwave Processing of Materials
HI] (1992).

. Methven, J .M. and Ghaffariyan, S. R., Principles of Microwave Heating (Binner,

J .G.P., ed.), pp.56-77, Abington Publishing.

Gault T.W. and Wilson, J .B., J. Microwave Power, p. 179 (1987).
Smith, R.D., EPRI Report EM-3645 (1984).

Nelson, 8.0., J. Microwave Power, p.65 (1985).

Vetsuypens J .P. and Loock, W.V., J. Microwave Power, p.110, (1986).

Li ghtsey, G., et. al., J. Microwave Power and Electromagnetic Energy, 23(1), 11
(1988).

Thiebaut, J .M., Akye, C., and Roussy, G., IEEE Trans. Instrum. Meas., 37(1), 114
(1988).

Li ghtsey, 6., George, C., and Russell, L.D., J. Microwave Power, p. 86, (1986).
Dauerman, L., MRS Symp. Proc., 189, L33 (1990).

Suzuki, J ., et. al., J. Microwave Power, 25(3), 168 (1990).

Oda, S.J., Inter. Microwave Power Inst. 26"I Symp. Proc., Buffalo, NY (1991).
Oda, S.J., MRS Symp. Proc., 269, 453 (1992).

Giguere, R.J., MRS Symp. Proc., 269, 387 (1992).

Gedye, R.N., Smith, RE, and Westaway, K. C., Can. J. Chem. 66, 17 (1988).

171

 

 

58. Bruce, R.W., Material Research Society Symposium Proceeding, 124, 3 (1988).
59. Sutton, W.H., MRS Symp., Proc., 269, 3 (1992).
60. Hawley MC. and Wei, J ., MRS Sym. Proc., 189, 413 (1990).

61. Gourdenne, A., Proceedings, International Conference on Reactive Proceeding of
Polymers, p.23, (1982).

62. Gourdenne, A. and Le, V.Q., Polymer Preprints, 22, 125 (1981).
63. Strand, N.S., Modern Plastics, 54, 64 (1980).

64. Wei, J ., Chang, Y., Thomas, B., and Hawley, M.C., Proceeding of ICCM/VIII, 1, 10- f1
L (1991 ). i

65. DeMuse, M.T., Polym. Mater. Sci. Eng., 66, 428 ( 1992).

 

66. Hawley, M.C., Wei, J ., Adegbite, V., Chang, Y., Michigan State University, Patent l
number 5,406,056, 1995.

67. Hawley, M.C., Asmussen J ., and Wei, J ., Michigan State University Invention
Disclosure Number 93-038, 1993.

68. Wei, J ., Thomas, B., and Hawley, M.C., 37th Intern. SAMPE, Anaheim, CA, March
(1992).

69. Fellows LA. and Hawley, M. C., Michigan State University Invention Disclosure
Number 93-066, 1993.

70. Wei, J ., Delgado, R., Hawley, M. C., and DeMeuse, M., MRS Spring Meeting, CA,
San Francisco, April (1994).

 

71. Delgado, R., Wei, J ., DeMeuse, M., and Hawley, M.C., AIChE 1993 Summer
National Meeting, Seattle, Washington (1993).

72. Wei J. and Hawley, M.C., 39th Intern. SAMPE, Anaheim, CA, April (1994).

73. Delgado, R., MS. Thesis, Michigan State University (1995).

172

 

74.

75.

76.

77.

78.

79.

80.
81.
82.

83.

84.

85.
86.

87.

88.

89.

90.

91.

Hawley, M.C., Wei, J ., Adegbite V., and Lin, M., US. Patent, No. 5,470,423, Nov.
28, 1995.

Lin, M., Continuous processing of polymers and composites with microwave
radiation, MS. Thesis, Michigan State University (1993).

Fellows, L.A., Lin, M., and Hawley, M. C., Tenth Annual Polymer Processing
Society Meeting, Akron, Ohio, April 3-7 (1994).

Hawley, M.C., Wei, J ., Fellows, L. A., and Adegbite, V., Michigan State University
Invention Disclosure Number 93-043, 1993.

Wu, B. and Benatar, A., Polymer Engineering and Science, 37(4), 738 (1997).

Wu, C., Staicovici S., and Benatar, A., Journal of Reinforced Plastics and
Composites, 18(1), 27 (1999).

Bernard, J.P., Sabran, and Collet, L., MRS Symposium Proceedings, 430, 37 (1996).
Willams, N.H., Journal of Microwave Power, 2, 123 (1967).
Gourdenne, A. and Le, V.Q. Eur. Polym. J., 23(10), 777 (1987).

Fellows, LA. and Hawley, M. C., Proceedings of the 29th Microwave Power
Symposium, p.91, Chicago, IL (1994).

Lin, M. and Hawley, M. C., Proceedings of the 38m SAMPE International
Symposium, Anaheim, CA (1993).

Wei, J ., Thomas, B., and Hawley, M. C., SPE Proc.: ANT EC ’92, 1, 1170 (1992).

Qiu, Y., Smith, A., Wei, J ., Lin, M., and Hawley, M.C., Composites in
Manufacturing, SME Quarterly, 1st Quarter, (2000).

Clark, D. B., Ceramic engineering and science proceedings, 14 (7-8), 3 (1993).

The LUXT RON model 750 F luoroptic Thermometrym System Service and Technical
Reference Manual, LUXTRON FluoropticTM Temperature Sensing, California
(1986).

The NoEMI-TSTM Family F iberoptic Thermometer Systems User’s Guide, Nortech
Fibronic, Inc., Quebec, Canada (1996).

Zhang, J ., Interaction of Electromagnetic Fields with a Material Sample Placed
within an Energized Cavity, Ph.D. Dissertation, Michigan State University (1998).

Jow, J ., Hawley M.C., and Finzel, M.C., Review of Scientific Instruments, 60 (1), 96
(1989).

173

 

 

92. Jow, J., Finzel, M., Asmussen, J ., and Hawley, M.C., Proceedings 1987 IEEE MTT-S
I International Microwave Symposium, Las Vegas, June 9-11 (1987).

93. Fourche, G., Polymer Engineering and Science, 35 (12), 957 (1995).

94. Cagle, C.V., editor. Handbook of Adhesive Bonding. Chapter 19. New York, McGraw
Hill (1973).

95. Compton, D.A.C., et. al. Applied Spectroscopy, 42(6), 972 (1988).
96. Young, P.R., et. al. SAMPE J., 25(2), 11 (1989).

97. Hawley, M.C., Zhou, S., Qiu, Y., and Wei, J. B., Patent application in preparation,

 

2000. ti
98. HP 4353 Power Meter Operating and Service Manual, Hewlett Packard, Printed in §

UK (1993). '1
99. Levine, W.S., The control handbook. Publisher: Boca Raton, Fla.: CRC Press (1996). J

100. Hedrick, J .C., Lewis, D.A., Lyle, G.D., Wu, S.D., Ward, TC, and McGrath, J E
Polymeric materials : science and engineering, proceedings of the ACS Division of
Polymeric Materials, Science and Engineering, 60, 438 (1989).

101. Pu B. and Hawley, M. C., Polymer Engineering & Science, 40(10), (2000).

102. Yousefi, A., Lafleur, P. G., and Gauvin, R., Polymer Composites, 18, 157, (1997).
103. Schechter, L., Wynstra, J ., and Kurkjy, R. P., Ind. Eng. Chem, 48, 94 (1956).
104. Sourour, S. and Kama], M. R., Thermochimic Acta, 14, 41 (1976)

105. Lee, W.I., Loos, A. C., and Springer, G. 8., Journal of Composite Material, 16,
510 (1982)

106. Dannenberg, H., SPE Trans, 3, 78 (1963).

107. Horie, K., Hiura, H., Sawada, M., Mita, I., and Kambe, H., J. Polym. Sci., A-I, 8,
1337 (1970).

 

108. Dusek, K., Ilavsky, M., and Lunak, S., J. Polym. Sci. Polym. Symp., 53, 29
(1976).

109. Riccardi, C. C., Adabbo, H. B., and Williams, R. J ., J. Appl. Polym. Sci., 29, 2481
(1984).

110. Finzel, M., Bulk-phase kinetics and physical properties of epoxy/amine systems,
PhD dissertation, Michigan State University (1991).

174

111. Wei, J., Hawley M. C., and Demeuse, M. T., Polym. Eng. Sci., 35, 461 (1995).
112. Acitally, M. A., Prime, R. B., and Sacher, B., Polymer, pp. 335-342 (1971)

113. Hagnauer, G. L., Pearce, P. J ., LaLiberte, B. R., and Roylance, M. B.,
Thermorheology of Thermosetting Polymers, pp. 25-47 (1983)

114. Prime, B., Thermal Characterization of Polymeric Materials, E. A. Turi, ed.,
Academic Press, New York (1981).

115. Karnal, M. R., Polym. Eng. Sci., 14, 231 (1974).
116. Kamal, M. R. and Sourour, S., Polym. Eng. Sci., 13, 59 (1973).
117. Progelhof, R. C. and Throne, J. L., Polym. Eng. Sci., 15, 690 (1975).

118. Gonzalez, V. M., Castro, J. M., and Macosko, C. W., The World Congress of
Chemical Engineering, Montreal (1981).

119. Chan, A. W. and Hwang, S. T., Polym. Eng. Sci., 31, 1149 (1991).

120. Lin, R., Lee, L. J ., and Liou, M., Intern. Polym. Process, Vi, 356 (1991).

121. Gebart, B. R., J. Appl. Polym. Sci., 51, 153 (1994).

122. Dutta, A. and Ryan, M. B., Journal of Applied Polymer Science, 24, 635 (1979)
123. Mijovic, J ., Journal of Applied Polymer Science, 31, 1177 (1986)

124. Moroni, A., Mijovic, J ., Pearce, E. M., and Foun, C. C., Journal of Applied
Polymer Science, 32, 3761 (1986)

125. Sichina, W.J., Dupont Company Applications Brief, No. TA-93

126. McGee, S. H., Polym. Eng. Sci., 22, 484 (1982).

127. Lucas, J. C., Borrajo, J ., and Williams, R. J. J ., Polymer, 34, 1886 (1993).
128. Yousefi, A., Lafleur, P. G., and Gauvin, R., J. Vinyl Additive Technol. (1996).
129. Ng, H. and Manas-Zloczower, I., Polym. Eng. Sci., 29, 302 (1989).

130. Yousefi, A., Lafleur, P. G., and Gauvin, R., Polym. Compos (1996).

131. Ryan, M. E. and Dutta, A., Polymer, 20, 203 (1979).

132. Lem, K.W. and Han, C. D., Polym. Eng. Sci., 24, 175 (1984).

133. Tollens, F. R. and Lee, L. J ., Polymer, 34, 29 (1993).

175

 

. . J1 . TI‘WA'

 

 

 

134. Kubota, H., J. Appl. Polym. Sci., 19, 2279 (1975).
135. Plueddemann, E. P., Silane Coupling Agents, Plenum Press, New York (1982).

136. Zhou, S., Hawley M. C., and Wei, J ., Proceedings of the 35'h Annual Microwave
Symposium, Montréal, Canada, July 16-19, 2000.

137. Hill, J. M. and Jennings, M.J., Appl. Math. Model. 17, 369 (1993).

138. Roussy, G., Bennani, A., and Thiebaut, J ., J. Appl. Phys, 62, 1167 (1987).
139. Smyth, N.F., J. Aust. Math. Soc. B, 33, 403 (1992).

140. Coleman, C. J ., Appl. Math. Model, 14, 439 (1990). E
141. Marchant, T. R., J. Eng. Math. 28, 379 (1994).

142, Jolly P.G. and Turner, 1. W., J. Microwave Power Electromag. Energy, 25, 3 I
(1990).

 

143. Turner I.W. and Jolly, P.G., J. Microwave Power Electromag. Energy, 25, 211
(1990).

176

APPENDICES

 

 

177

 

APPENDIX A. MATLAB PROGRAM FOR CALCULATING THE MODE
PATTERN OF TM022 INSIDE THE EMPTY CAVITY

 

% - - - - - - - - we - - - - - -
%

% This program computes the electric field pattern of a Tanq mode

% and returns a 2-D view of any specified cross—section of an empty

% cylindrical single mode cavity.

 

%

% In this example, the electric field pattern of TM022 is computed.
%

% Variables that need to be specified are:

% n, p, q: parameters associated with the name of the mode

% xnp: solution of Jn(x)=0, where Jn is bessel function of

% the first kind.

% z: axial position in the cylindrical cavity, unit mm

% h: cavity height, unit mm

% a: radius of the cylindrical cavity, unit mm

%

(70.. - - W- .. - - u H-
%

% Other variables and functions are decribed before each executable

% sentence.
%

 

n=O;p= ;q=2;xnp=5.520;

 

z=13;a=88.9;h=132;

% rhogrid: dimensionless radial position, from O and 1
% phigrid: dimensionless angular position, from 0 to 2*pi
rhogrid=0:0.01:1;phigrid=-pi:2*pi/100:pi;

 

% "meshgrid" transforms the domain specified by vectors phi grid
% and rhogrid into arrays specified by phi and rho

% phi: dimensionless angular position, which specifies the rows
% of the array.

% rho: dimensionless radial position, which specifies the columns
% of the array.

[phi,rho]=meshgrid(phigrid,rhogrid);

% "polanrt" transforms polar to Cartesian coordinates
[x,y]=p0120art(Phi,rh0);

178

 

% space: spacing between points in both x and y direction
space=xnp*0.0l ;

% "besselj" is the Bessel function of the first kind
Jn=besselj(n,xnp*rho);

% "gradient" returns the numerical gradient of the matrix J n
[Fx,Fy]=gradient(Jn,space);

% Jngrad: d(Jn)/dy, gradient of Jn in the y direction
J ngrad=Fy;

% Erho: rho component of electric field strength
Erho=xnp/(a*0.001)*(-l *q*pi/(h*0.001))*Jngrad.*cos(n*phi).*sin(q*pi*z/h);

% Ephi: phi component of electric field strength
Ephi=1./(rho*(a*0.001)+eps).*n.*(-1*q*pi/(h*0.001)).*Jn.*cos(n*phi).*sin(q*pi*z/h);

% E2: 2 component of electric field strength
Ez=xnp"2l(a*0.001)"2*Jn.*cos(n*phi)*cos(q*pi*z/h);

% E: magnitude of electric field strength
E=sqrt(Erho."2+Ephi."2+Ez."2);

% "mesh" plots 3-D mesh surface
mesh(x,y,E);

% specify the view point to directly overhead 2-D view
view(0,90);

 

% set axis scaling and appearance
axis equa1;axis tight;

179

 

APPENDIX B. LABVIEW PROGRAMS

The functions of each LabVIEW program are listed in Table B.1.

Table 8.] Functions of the Labview Programs

 

 

Function
Main Program cavity Obtain the mode spectrum before
characterization .vi bonding

 

VFMS.vi

Perform data acquisition, process
control, and on-line monitoring during
variable frequency mode switching
bonding

 

Sub-programs

 

 

 

 

6T.vi Acquire temperature data at 6 points

luxtron .vi Convert voltage to temperature for
LUXTRON thermometer

nortech .vi Convert voltage to temperature for
NORTECH thermometer

power.vi Acquire incident and reflected power

data

 

mode-switch .vi

Achieve uniform heating

 

power control .vi

Control heating rate and temperature

 

pid .vi

Control the temperature with PID
algorithm

 

PID parameters .vi

Input PID parameters

 

heat rate .vi

Calculate heating rate

 

 

 

frequency Monitor the variable frequency adhesive
diagnosisvi bonding process on-line
f-write .vi Write the operating frequency to

 

hardware

 

180

 

 

 

The hierarchy of the LabVIEW programs is shown in Figure B.1.

 

Data acquisition and process control software for
microwave adhesive bonding process

A

Cavity VFMS.vi
characterization .vi .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6T.vi Power. mode- power frequency
vi switch .vi control .vi diagnosis .vi
luxtron nortech pid .vi heat f-write
.vi .vi rate .vi .vi

 

 

 

 

 

 

 

 

 

 

 

 

PIDi

parameters .vi

 

 

 

Figure B.l Hierarchy of the Labview Programs

 

181

 

 
  

 

1L ~

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

        

ran-n1 «a» w
Parametkl' lung Monitoring
9932! NW) ma "1'51
5311 I |0.00 ] [0.00 [0.00 lmfl
EEL Frequen reflectam [0.00 1&1 .7"
2.3 1
___110m [WEI ,
Press this i ,
n Scar . -
5 300 00 Chart ornerrectance :35? l-
' _———J‘"m’ Elm—U: procram. l.
PrScaI 1 ’—“ E .
$300.00 0.9- ‘
ElStarting «0|er 0.0-
3 2.0000 0.7- 3
Incremeng 0.6- '
5 0.0002 ‘
Endlng “GM: ”'5'
[M-
$40000
Forma 03‘
LJ “2'
Loop Delay “-1“
(mllllseconds 0.0-, ' ' | . . . .
31 10mm 0 10 40 50 60
I -" l’ .mmzfi,” ‘ .. Nikita}; “Olai-A-ri‘3r‘1-JI‘WE“?{‘9‘ _— ; warhead-:7

Figure B.2 Front Panel of the Cavity Characterization Program

182

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure B.3 Diagram of the Cavity Characterization Program

183

 

 

|0.00 sec] 'v
1103;; 3"...“ l4,

 

j 3‘25
,1,
ii"
if
C H gflme

19'” l—IluTfin— min

 

 

 

 

 

to stop program
"(0
72(0
13(0
"(0
T5“)
76“)

 

m ctr-l V
031 ..o .
Li
a, r
Press this butto ’

 

 

 

 

 

 

WLIPE WIN!“
Phat

 

r

WNW

1M1
. r
" r
r
00.0
00.0
40.0
20.0

0 1000

LI_1_1

 

 

 

 

fig;
SEEEIEE
311327053

Figure 8.4 Front Panel for Variable Frequency Mode-Switching Process Control

I I
219:!!! mm

 

 

 

Salaam | glam] gm]

 

dignosls Interval lfie—lfl

Model: Inmmj "lowed Huang]

ifii =

 

 

 

DEMMEHE
. ~ A r ‘ vrmsl

 

 

 

 

Mode 2:

lifll
Hm

 

 

 

 

Filename :

File-name of
1 liftxt]

File-name- of
‘2 (itxt)

éUFMSNi Diagram *

l4pt Dialog Font

 

185

Figure B.5 Diagram for Variable Frequency Mode Switching Process Control (Part 1)

 

 

 

 

 

 

 

Initial 'v'
on

 

Curin- time

 

 

Figure B.6 Diagram for Variable Frequency Mode Switching Process Control (Part 2)

186

 

 

El

6T.vi

 

in;
33::

:3:
..

r'ur
. .

Scanned Data

::c:: -

 

_-F9§ "I mp1 Dialog Font

Index 2:1' lug
[EDT—l

Index 3:1 Ind
[0.00

lndex5zTnor ’ A

16730—1

M "

no:

 

 

 

 

 

 

 

Figure B.7 Front Panel for 6t.Vi

187

 

Wm.

 

61m Diagram *

l
Elumlfillflm' [ 12pt Dialog Font V“ :5 VINE—1

 

 

 
 

.—-—-—f:]
lndex3:T1ux
. D ,:, "f“"j
t |____;J

 

 

 

 

 

 

 

Figure B.8 Diagram for 6t.Vi

188

luxtron .vi

 

4) . II mm

as sigma! 119
Slim—- W

 

 

 

 

Figure 8.9 Front Panel for LuxtronVi

El Emm mum E4!!!“

    

convert V scale factor
to m9

 

Figure B.10 Diagram for Luxtron.Vi

189

nortech.vi
Ifilfiifllm7 ““1
eslgnaj Hg
32E] IE]

 

    

 

 

 

 

Figure 8.1] Front Panel for Nortech.Vi

W

 

 

 

 

 

nortech.“ Diagram E

 

     

     

azimllonlmmlm

   

 

Figure 3.12 Diagram for Nortech.Vi

El

 

          
 
 
    
  
     

 

'/\‘
c. 4

IE
n Sung]
Pr Seals]

3111.00

 

 

 

      

 

EEC—1‘

 

 

 

Figure B.14 Diagram for Power.Vi

l9]

 

 

mode-switch.vi

 

 

 

 

 

 

Figure B.15 Front Panel for Mode-Switch.Vi

 

 

 

ma

 

 

 

 

 

..Pln. Pin—P

E 5.5 a: _ E Eh EHHSEWEI

 

£9595 ism-«573%.ilfiii. .. a.

Figure B.16 Diagram for Mode-Switch.Vi

I93

 

 

 

 

 

 

 

 

 

Figure B.17 Front Panel for Power Control.Vi

I94

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.8“. 3233:“ when—EA a. mum.“

  

 

m a WI. Elhafla E_§°U 5030-— W

 

 

Figure 3.18 Diagram for Power Control.Vi

195

 

El? i‘;:;::heatratevl::fa

 

 

 

)WWW fl

 

 

 

 

 

lfin- I] lhatflate (Us)
um um

 

 

 

 

 

Figure 3.20 Diagram for Heat Rate.Vi

196

 

pid.vi

 

 

 

 

 

 

 

Figure B.21 Front Panel for Pid.Vi

197

 

 

 

 

 

 

 

 

 

 

 

 

 

 

       

 

“so... 0205 «E E L W

 

 

 

Figure B.22 Diagram for Pid.Vi

198

E ENNIS.“ E‘IE

 

 

 

 

  
  

  

ill

‘I. u
4'“—

 

E‘EIZ‘EDIJIWIE‘ FIE

 
       
 

 

 

 

 

Figure 8.24 Diagram for Pid Parameters.Vi

 

 

 

 

 

 

Figure B.25 Front Panel for Frequency Diagnosis.Vi

200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

59.5.! 5.233.... mucus-.3...

 

Figure B.26 Diagram for Frequency Diagnosis.Vi

201

 

This program wrltes {requencles to the
? sweep osclllatorvla GPIB.

Frequengj

 
   
 

 

Figure B.27 Front Panel for F-Write.Vi

f-write.vi Diagram

   

 

   

 

 

 

 

 

 

 

Figure B.28 Diagram for f-write.Vi

202

i, m, __

m

l