Anda di halaman 1dari 364

This item was submitted to Loughborough University as a PhD thesis by the

author and is made available in the Institutional Repository


(https://dspace.lboro.ac.uk/) under the following Creative Commons Licence
conditions.





For the full text of this licence, please go to:
http://creativecommons.org/licenses/by-nc-nd/2.5/

I
I
LOUGHBOROUGH . ---,
UNIVERSITY OF TECHNOLOGY i
LIBRARY
~ ~
AUTHOR/FILING TITLE
J)Av.J)SOo-l P.
-- -- - ----- ---------------7-- -------- --_______ _
----- - --------------------- ---- --- ----- -- ----- ---
ACCESSION/COPY NO.
(J\t- cro ~ '1. O ~ ~ ,
------------------ ---- ----------------------- ------ ,
VOL. NO. CLASS MARK
!---------r--------r-______
- 3 OCT 1997
i 1. 2 l : ~ m,
i __ _
liflil ffUII 811111' ---- -_ .. -------------------
Loughborough University of Technology
Institute of Polymer Technology and Materials Engineering
Heat-Setting of Biaxially Oriented PET
Paul Davidson
A Doctoral Thesis submitted in partialfulfilment of the requirements
for the award of the Degree of Doctor of Philosophy of the
Loughborough University of Technology.
P. Davidson, (1993)
Loughborough University
of Tech'"-"--,, Library
........... -
... ", , < ~ , ~ '
Date
S"-tAo 9'{'
---,._., .,.", ......
Class
- .. ~ , . , ....
Acc.
(j\.f- CO ~ '2. 0 .:)'6
No.
STATEMENT OF ORIGINALITY
The work presented in this thesis has been carried out by the Author, except
where otherwise acknowledged, and has not previously been submitted to
this University or any other institution for a higher degree.
P. Davidson (1993)
The best moons
are not milky white
ACKNOWLEDGEMENTS
It is traditional to thank all the people who you ought to thank on this page,
regardless of real merit. The people mentioned here are here because they
contributed far and above what would be expected of them.
Firstly, Barry Haworth who, as a supervisor, it would difficult to ask more
of. His technical skills, support, patience and positive outlook were all
crucial to the project. It was his style of supervision that made my three
years in IPTME as enjoyable as they were.
The use of the TMA was by kind permission of Dr Gilbert and a special
note of thanks to Dave Hitt for all his assistance with the TMA. I do not
believe that there exists anyone who could have brought his extraordinary
talents with a microscope to bear quite like' Derek Hemsley. The
contributions by Pete and Ray, as with all projects in IPTME, were
essential. Thanks also to lan Sutherland for the use and advice concerning
the FTIR Both myself and my PC have the good fortune of considering
Muna a very valued and able friend. A special thanks to all the research
students of both IPTME, and dare I say it, Organic Chemistry, for three
entertaining years.
Ken, Gareth and all the Melinar team, past and present, at ICI Wilton who
contributed an absolutely invaluable industrial perspective to the project as
well as making their time and laboratory services freely available. A special
mention must go to Jill Furphy, who helped a very troubled and sceptical
PhD student through the early days of the FED and to Paul, Dane and Anna
who performed the statistical analysis. I would also like to thank ICI for
their very welcome fmancial contribution to the SERC/CASE award.
My family have been as brilliant as ever. Dad, Mum and Louise have all
given me so much, that I would not know where to begin. However, an
extra special mentioned must go to Mum, whose contribution can only be
described as astonishing. Finally, to someone, for whom "just being there
with me " is probably the hardest thing she will ever have to do. Thank
you.
Abstract
The relatively low deformation temperature of biaxially oriented PET excludes its
use in many hot fill applications or hot washing as part of a reuse cycle, due to the
shrinkage which generally occurs on reheating towards Tg.
Isothermal, isometric annealing processes (referred to as heat-setting) can cause
oriented PET to maintain its dimensional stability to higher temperatures.
However, to experiment on a commercial stretch blow moulding machine is a very
costly and time consuming process, justifying the requirement for an accurate
simulation technique. The objective of this project was, therefore, to simulate the
stretch blow moulding of PET with a heat-setting phase, by biaxially drawing PET
sheets on a TM Long Stretcher, then to heat-set them in a separate phase using a
purpose designed jig.
A heat-setting rig was designed and constructed at IPTME. Due to the size of the
proposed investigation window, a factorial experimental design (FED) technique
was employed to characterise the behaviour of two commercial bottle grades of
PET (B90N and B95A Laser) under a variety of draws and heat-setting conditions.
These sheets were then characterised in terms of shrinkage, tensile properties and
resistance to creep. This data was regressed to generate predictive model
equations for both grades of material and in both major and minor draw axes.
Using these equations, it is now possible to define a set of processing parameters
which would give a bottle with lower and more uniform shrinkage (less draw
dependent) with enhanced creep properties and at least comparable mechanical
properties.
To develop a more fundamental understanding of the process of heat-setting in
biaxially drawn PET, as well as to prove the validity of the model equations, a
separate series of conventional experiments was devised to investigate the two
most significant predictor variables from the FED analysis; biaxial draw ratio and
heat-setting temperature (for Laser only).
This material was characterised by TMA, DSC and birefringence measurements.
The DSC and birefringence measurements allow the effect of heat-setting and draw
on crystallinity levels and orientation to be monitored. Orientation of both phases
was studied in heat-set bottles using micro polarised FTIR.
Chapter One
1.1
1.2
1.3
1.3.1
1.3.2
Chapter Two
2.1
2.1.1
2.1.2
2.1.3
2.2
2.2.1
2.2.2
2.3
2.3.1
2.3.2
2.3.2.1
2.3.2.2
2.4
2.4.1
2.4.2
2.4.3
2.5
2.5.1
2.5.2
2.5.3
2.5.3.1
2.5.3.2
2.5.3.3
2.5.3.4
2.5.3.5
CONTENTS
Introduction .......................................................................... 1
General Background ........................................................................ 1
Bottle Production ............................................................................ 1
Objectives and Strategy .................................................................... 3
Project Objectives ............................................................................ 3
Project Strategy ............................................................................... 4
Literature Survey .................................................................. S
Polyethylene Terephthalate Production ............................................ 5
Principle of PET Polymerisation Techniques .................................... 5
Ethylene Glycol and Acetaldehyde ................................................... 6
PET Copolymers ............................................................................. 7
Theory of Shrinkage ........................................................................ 7
Amorphous Phase ............................................................................ 9
Crystalline Phase ............................................................................ 11
Morphology and the Effect of Deformation .................................... 15
Crystallisation ................................................................................ 20
Orientation ..................................................................................... 24
Orientation Functions ..................................................................... 26
Evolution, Classification and Effects of Orientation ........................ 29
Heat-setting ................................................................................... 33
Heat-Setting as a Commercial Process ........................................... 33
Mechanics of Heat-Setting ............................................................. 34
Heat-Setting Structural and Mechanical Considerations .................. 37
Characterisation Techniques and Material Properties ...................... 38
Shrinkage ....................................................................................... 38
Mechanical Testing and Properties ................................................. 39
Orientation Measurement ............................................................... 40
Birefringence ................................................................................. 40
(A) Orientation Birefringence ......................................................... 41
(B) Stress Birefringence ................................................................. 41
(C) Form Birefringence .................................................................. 41
Fourier Transform Infrared Spectroscopy ....................................... 44
X-Ray Diffiaction .......................................................................... 47
Density .......................................................................................... 49
Thermal Analysis ........................................................................... 50
Chapter Three
3.1
3.2
3.2.1
3.2.2
3.2.2.1
3.2.2.2
3.2.2.3
3.2.2.4
3.2.3
3.2.4
3.2.5
3.3
3.3.1
3.3.2
3.4
3.4.1
3.4.1.1
3.4.1.2
3.4.2
3.4.3
3.4.4
3.5
3.5.1
3.5.1.1
3.5.1.2
3.6
3.6.1
3.6.1.1
3.6.1.2
3.6.2
3.6.2.1
3.6.2.2
3.6.2.3
3.6.2.4
Experimental ........................................................................ S3
Material Grade Selection ................................................................ 53
Sample Preparation, Technique Development.. ............................... 53
The TM Long Stretcher ................................................................. 54
Heat-setting ................................................................................... 56
Heat Gun ....................................................................................... 56
Infra Red ....................................................................................... 57
Manual Application ofPreheated Metal Blocks .............................. 57
Electrically Heated Jigged Block .................................................... 58
Design of The Second Heat-setting Rig .......................................... 59
Heat-setting Rig Requirements ....................................................... 59
Heat-setting Methods ..................................................................... 60
Factorial Experimental Design (FED) ............................................. 62
Selected Values For FED Analysis ................................................. 63
Comments on Implications of
the FED on Biaxial Stretching ........................................................ 64
Characterisation of Oriented and Heat-set PET .............................. 65
Mechanical Properties .................................................................... 65
Tensile Testing ............................................................................... 65
Creep Testing ................................................................................ 67
Linear Thermal Shrinkage Determination: ...................................... 69
Thermal Analysis ............................. , ............................................. 71
Density Measurements ................................................................... 72
Conventional Heat-setting Experiments .......................................... 72
Optical Properties .......................................................................... 73
Birefringence ................................................................................. 74
(A) Refractometry and Compensator Methods ............................... 74
(B) Conoscopy .............................................................................. 76
Optical Clarity ............................................................................... 80
(A) Total Transmitted Light Measurements .................................... 80
(B) Micro-Photometry ................................................................... 81
Bottle Trials ................................................................................... 82
Heat-setting of Bottles ................................................................... 82
Plaster of Paris ............................................................................... 83
Polyurethane Foam ........................................................................ 84
Orientation Measurements On Heat-set Bottles .............................. 85
Sample Preparation ........................................................................ 85
Use Of The Spectra-Tech IR-Plan Microscope ............................... 86
Orientation Functions ..................................................................... 89
Dichroic Ratios .............................................................................. 90
Chapter Three Figures ................................................................... 91
Chapter Three Tables ................................................................... 103
Chapter Four
4.1
4.1.1
4.1.1.1
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
(I)
(1)
4.1.1.2
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
(I)
4.1.2
4.1.2.1
4.1.2.2
4.1.3
4.1.3.1
(A)
(B)
(C)
(D)
(E)
(F)
(G)
(H)
4.1.3.2
(A)
(B)
(C)
(D)
(E)
(F)
(G)
Results ................................................................................. 107
FED - Statistical Analysis and Model Predictions ......................... 107
Mechanical Testing Data .............................................................. 109
Homopolymer .............................................................................. 109
Yield stress - major axis ............................................................... 109
Yield stress - minor axis ............................................................... 1\0
Yield strain-major axis ................................................................. III
Yield strain - minor axis ............................................................... III
Ultimate tensile stress - major axis ............................................... 112
Ultimate tensile stress - minor axis ............................................... 113
Ultimate tensile strain - major axis ................................................ 114
Ultimate tensile strain - minor axis ............................................... 114
Strain hardening rate - major axis ................................................. 115
Strain hardening rate - minor axis ................................................. 115
Copolymer ................................................................................... 116
Yield stress - major axis ............................................................... 116
Yield stress - minor axis............................................................... 117
Yield strain - major axis ............................................................... 117
Yield strain - minor axis .... ................ .................. ............ ............. 117
Ultimate tensile stress - major axis ............................................... 118
Ultimate tensile stress - minor axis ............................................... 118
Ultimate tensile strain - major axis ................................................ 119
Ultimate tensile strain - minor axis ... : ........................................... 119
Strain hardening rate - major axis ................................................. 120
Creep Models .............................................................................. 120
Creep strain - Homopolymer.." .................................................... 121
Creep strain - Copolymer ............................................................. 121
Shrinkage Models .................. " .................................................... 122
PET Homopolymer ..................................................................... , 122
Shrinkage at 65C - major axis ..................................................... 122
Shrinkage at 65C - minor axis ..................................................... 122
Shrinkage at 85C - major axis ..................................................... 123
Shrinkage at 85C - minor axis ..................................................... 123
Shrinkage at 100C - major axis ................................................... 124
Shrinkage at 100C - minor axis ................................................... 125
Shrinkage at 110C - major axis ................................................... 125
Shrinkage at 110C - minor axis ................................................... 126
PET Copolymer ........................................................................... 127
Shrinkage at 65C - major axis ..................................................... 127
Shrinkage at 65C - minor axis ................... " ................................ 127
Shrinkage at 85C - major axis ..................................................... 128
Shrinkage at 85C - minor axis ..................................................... 128
Shrinkage at 100C - major axis ................................................... 128
Shrinkage at 100C - minor axis ................................................... 129
Shrinkage at 110C - major axis ................................................... 130
(H)
4.1.4
4.1.4.1
A)
(B)
4.1.4.2
(A)
(B)
4.2
4.2.1
(A)
(B)
(C)
(D)
4.2.2
4.2.3
4.2.4
4.2.4.1
4.2.4.2
4.2.4.3
4.2.4.4
4.3
4.3.1
4.3.1.1
4.3.1.2
4.3.2
4.3.2.1
4.3.2.2
4.3.3
4.3.3.1
4.3.3.2
Chapter Five
5.1
5.1.1
5.1.1.1
(A)
(B)
(C)
(D)
Shrinkage at 110C - minor axis ................................................... 130
FED - Crystallinity Measurements ................................................ 131
Thermal Analysis (DSC) .............................................................. 131
PET Homopolymer ...................................................................... 131
PET Copolymer ........................................................................... 132
Density ........................................................................................ 132
PET Homopolymer ...................................................................... 132
PET Copolymer ........................................................................... 133
Conventional Experimentation Results ......................................... 134
Shrinkage ..................................................................................... 134
Shrinkage at 65C ........................................................................ 134
Shrinkage at 85C ........................................................................ 135
Shrinkage at 100C ...................................................................... 13 5
Shrinkage at 110C ...................................................................... 13 5
Thermal Analysis ......................................................................... 136
Refractive Indices as a Estimate of Degree of Crystallinity ........... 136
Optical Properties ........................................................................ 136
Total Transmitted Light Measurements ........................................ 136
Micro-Photometry ....................................................................... 137
Birefringence ............................................................................... 137
Refractive Index Measurements ................................................... 138
Heat-Set Bottle Results ................................................................ 139
Shrinkage Measurements - Comparisons of
Predictions and Real Data ............................................................ 139
Shrinkage - Hoop axis .................................................................. 139
Shrinkage Axial Axis .................................................................... 139
Crystallinity Measurements .......................................................... 140
Thermal Analysis......................................................................... 140
Density Measurements ................................................................. 140
Orientation Measurements ........................................................... 140
Orientation Functions ................................................................... 140
Dichroic Ratio Plots ..................................................................... 141
Chapter Four Figures ................................................................... 142
Chapter Four Tables .................................................................... 198
Discussion ........................................................................... 200
FED-Statistical Analysis And Model Predictions .......................... 200
Mechanical Properties Models ..................................................... 200
PET Homopolymer ...................................................................... 201
Yield stress .................................................................................. 201
Yield strain .................................................................................. 202
Ultimate tensile stress .................................................................. 202
Ultimate tensile strain ................................................................... 204
(E)
5.1.1.2
(A)
(B)
(C)
(D)
(E)
5.1.2
5.1.3
5.1.3.1
5.1.3.2
5.1.4
5.1.4.1
(A)
(B)
(C)
(D)
5.1.4.2
(A)
(B)
(C)
(D)
5.1.4.3
(A)
(B)
(C)
(0)
5.1.5
5.1.5.1
5.1.5.2
5.1.6
5.1.6.1
5.1.6.2
5.1.7
5.2
5.2.1
(A)
(B)
(C)
(D)
5.2.2
5.2.2.1
5.2.2.2
5.2.3
5.2.4
Strain hardening rate .................................................................... 205
PET Copolymer ........................................................................... 206
Yield stress .................................................................................. 206
Yield strain .................................................................................. 206
Ultimate tensile stress .................................................................. 207
Ultimate tensile strain ................................................................... 207
Strain hardening rate .................................................................... 208
Creep Models .............................................................................. 208
Selected Mechanical Properties as a Function of Crystallinity ....... 212
Yield Stress as a Function ofCrystallinity ..................................... 212
Creep Strain as a Function of Crystallinity .................................... 213
Shrinkage Models ........................................................................ 214
PET Homopolymer ...................................................................... 214
Shrinkage at 65C ....................................................................... 214
Shrinkage at 85C ....................................................................... 215
Shrinkage at 100C ..................................................................... 215
Shrinkage at llOoC ..................................................................... 216
PET Copolymer ........................................................................... 217
Shrinkage at 65C ....................................................................... 217
Shrinkage at 85C ....................................................................... 217
Shrinkage at lOOoC ..................................................................... 218
Shrinkage at llOoC ..................................................................... 218
General Comments and Interpretation .......................................... 219
Comparison of Materials .............................................................. 219
Major Stretch Axis ....................................................................... 220
Minor Stretch Axis ...................................................................... 220
Overall Comments ....................................................................... 221
Selected Shrinkage Models as a Function of Crystallinity ............. 223
Major Axis Shrinkage Models as a Function of Crystallinity ......... 223
Minor Axis Shrinkage Models as a Function of Crystallinity ......... 223
FED Crystallinity Measurements .................................................. 224
Thermal Analysis (DSC) .............................................................. 224
Density Measurements ................................................................. 225
Comments On the Implications Of Using FED Analysis ................ 227
Conventional Experimentation Series ........................................... 229
Shrinkage Measurements ............................................................. 229
Shrinkage at 65C ....................................................................... 229
Shrinkage at 85C ....................................................................... 230
Shrinkage at lOOoC ..................................................................... 230
Shrinkage at IlOoC ..................................................................... 230
Theoretical Volume Shrinkage ..................................................... 232
Volume Shrinkage Calculations .................................................... 232
Volume Shrinkage From FED Models .......................................... 233
Shrinkage Aspect Ratio ................................................................ 235
Thermal Analysis ......................................................................... 238
5.2.4.1
5.2.4.2
5.2.5
5.2.5.1
5.2.5.2
5.2.5.3
5.2.5.4
5.2.5.5
(A)
(B)
(C)
(D)
5.3
5.3.1
5.3.2
5.3.3
5.3.4
5.3.5
5.4
5.4.1
5.4.2
5.5
5.5.1
5.5.2
Chapter Six
6.1
6.1.1
6.1.2
6.1.3
6.1.4
6.1.4.1
6.1.4.2
6.1.5
6.1.6
6.1.7
6.1.8
6.2
DSC Analysis ............................................................................... 238
Degree of Crystallinity from Refractive Index Measurements ....... 238
Optical Properties ........................................................................ 239
Total Transmitted Light Measurements ........................................ 239
Micro-Photometry ....................................................................... 240
Birefringence ............................................................................... 241
Additional Information from Refractive Index Measurements ....... 242
Selected Properties as a Function ofBirefringence ....................... 243
Yield Stress as a Function ofBirefringence .................................. 243
Creep Strain as a Function ofBirefiingence .................................. 244
Degree of Crystallinity as a Function of Birefiingence .................. 244
Shrinkage as a Function ofBirefiingence ...................................... 245
Heat-Set Bottles .......................................................................... 246
Shrinkage Measurements ............................................................. 247
Thermal Measurements ................................................................ 248
Density Measurements ................................................................. 249
Orientation Measurements ........................................................... 250
Shrinkage of Heat-Set Bottles as a Function of Heat-Setting
Temperature and Crystallinity ...................................................... 252
Molecular Interpretations Of Heat-Setting Based On
Experimental Observations ........................................................... 253
Molecular Interpretation of MechaniCal Properties ....................... 253
Molecular Interpretation of Shrinkage Properties """"""""""'"'' 257
Experimental Error Considerations ............................................... 259
Errors Arising From Film Drawing and Heat-Setting Processes .... 259
Errors Arising From Film Property Measurements ........................ 260
Chapter Five Figures .................................................................... 263
Conclusions ......................................................................... 279
Conclusions ................................................................................. 279
New Techniques and Equipment Developed for this Project.. ....... 280
The Use ofBiaxially Oriented Films to Simulate Biaxially
Oriented Bottles ........................................................................... 281
The Reduction of Thermally Induced Shrinkage in PET Bottles ... 282
Mechanical Testing ...................................................................... 283
Tensile Tests ................................................................................ 283
Creep Tests .................................................................................. 284
Shrinkage Measurements ................... """""'''''''''''''''''''''''''''''''' 285
Thermal Analysis and Density Measurements ............................... 287
Optical Measurements .................................................................. 289
Orientation Measurements By FTIR .......... " ............ " "" .......... " ... 290
Suggested Further Work .................... " "'''''''''' """"" " ................ 291
References ............................................................................................................... 294
Appendices .................................................................................................... 304
Appendix AI
Appendix A1
Appendix A3
Appendix A4
Appendix A5
Appendix A6
AppendixA7
Appendix A8
Appendix A9
Appendix AIO
Appendix AI I
Appendix AI2
Appendix A 13
Appendix A14
Appendix A15
Appendix A16
Appendix A17
Appendix A18
Appendix A19
Appendix A10
Appendix A1 I
Appendix A12
Appendix A13
Appendix A14
Raw Force Data From Major Axis Tensile Experiments
Homopolymer ..................................................................... 304
Raw Force Data From Major Axis Tensile Experiments
Copolymer .................................................................. "'''''' 305
Raw Force Data From Minor Axis Tensile Experiments
Homopolymer ..................................................................... 306
Raw Force Data From Minor Axis Tensile Experiments
Copolymer .......................................................................... 307
Raw Extension Data From Major Axis Tensile
Experiments ........................................................................ 308
Raw Extension Data From Major Axis Tensile
Experiments ........................................................................ 309
Raw Extension Data From Minor Axis Tensile
Experiments ........................................................................ 310
Raw Extension Data From Minor Axis Tensile
Experiments "'''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' 31 I
Yield Stress-Major Axis ...................................................... 312
Yield Stress-Minor Axis ...................................................... 313
Yield Strain-Major Axis ""'''''''''''''''''''''''''''''''''''''''''''''''''' 3 14
Yield Strain-Minor Axis ...................................................... 315
Ultimate Tensile Stress-Major Axis ..................................... 316
Ultimate Tensile Stress-Minor Axis "'"'''''''''''''''''''''''''''''''' 317
Ultimate Tensile Strain-Major Axis ..................................... 318
Ultimate Tensile Strain-Minor Axis ..................................... 319
Strain Hardening Rate-Major Axis ....................................... 320
Strain Hardening Rate-Minor Axis .............................. , ....... 321
Creep Strain at 40MPa, 650C after 20 minutes .................... 322
Shrinkage Data at Various Temperatures as measured by
TMA Homopolymer-Major Axis ......................................... 323
Shrinkage Data at Various Temperatures as measured by
TMAHomopolymer-Minor Axis ......................................... 324
Shrinkage Data at Various Temperatures as measured by
TMA Copolymer-Major Axis .............................................. 325
Shrinkage Data at Various Temperatures as measured by
TMA Copolymer-Minor Axis .............................................. 326
Thermal Analysis and Density Data ..................................... 327
Appendix B .................................................................................................... 328
Appendix BI
AppendixB2
Appendix B3
AppendixB4
AppendixB5
AppendixB6
Shrinkage In Conventional Experimentation Films-
Major Axis .......................................................................... 328
Shrinkage In Conventional Experimentation Films-
Minor Axis .......................................................................... 329
Thermal Analysis Data for Conventional Experimentation
Films ................................................................................... 330
Crystallinity Values From Refractive Index Measurements ... 331
Optical Transmission Data for Conventional
Experimentation Films ......................................................... 332
Birefringence Of Conventional Experimentation Films ......... 333
Appendix C ................................................................................................... 334
Appendix Cl
Appendix C2
AppendixC3
AppendixC4
Shrinkage In Heat Set Bottles ............................................. 334
Crystallinity in Heat Set Bottles by Thermal Analysis and
Density ................................................................................ 335
Orientation Functions For Heat Set Bottles """"""""""'"'' 336
Dichroic Ratios For Heat Set Bottles ................................... 336
FIGURES
Chapter Three
Figure 3.1. Biaxial Stretching Rig At IPTME ............................................. 91
Figure 3.2. Biaxial Stretching Rig At IPTME With Heat-Setting
Attachment in Place ................................................................. 92
Figure 3.3. Components For Jaws of Heat-Setting Rig ............................... 93
Figure 3.4. Components For Frame of Heat-Setting Rig ............................. 94
Figure 3.5. Heat-Setting Rig with Film in Place .......................................... 95
Figure 3.6. Heat Setting Block Working Face Showing
Construction Used to allow Variable Film Size to be
Accommodated ........................................................................ 96
Figure 3.7. Cooling of Experimental Steel Block ........................................ 96
Figure 3.8. Cooling of Heat-Setting Block ................................................. 97
Figure 3.9. Load Control Against Time For A Constant Nominal Stress
Operation ................................................................................. 98
Figure 3.10. Metier TMA40 Thermometric Analyser With Film And
Fibre Attachment In Place ........................................................ 99
Figure 3.11. Demonstration of Correction for Expansion In
TMA Experiments .................................................................. 100
Figure 3.12. Optical And IR Paths In A Spectra-Tech IR Plan
Microscope ............................................................................ 101
Figure 3.13 Intensity Distribution At An Edge Demonstrating The Loss
Of Spatial Purity When Using Only The Lower Aperture ........ 102
Chapter Four
Figure 4.1a. Typical FED Output Sheet ..................................................... 142
Figure 4.1 b. Shrinkage Model Correlation as a Function of
Heat-Setting Temperature for PET Homopolymer and
Copoloymer, for Each Draw Area .......................................... 143
Figure 4.2a. Statistical Model For Homopolymer Yield Stress,
Heat-Set For 6 Seconds ........................................................ 144
Figure 4.2b. Statistical Model For Homopolymer Yield Stress,
Heat-Set For 15 Seconds ....................................................... 144
Figure 4.2c. Statistical Model For Homopolymer Yield Stress,
Heat-Set For 30 Seconds ....................................................... 145
Figure 4.3a. Statistical Models For Homopolymer Yield Strain,
Heat-Set For 6 Seconds ......................................................... 146
Figure 4.3b. Statistical Models For Homopolymer Yield Strain,
Heat-Set For 15 Seconds ....................................................... 146
Figure 4.3c. Statistical Models For Homopolymer Yield Strain,
Heat-Set For 30 Seconds ....................................................... 147
Figure 4.4a. Statistical Models For Homopolymer Ultimate
Tensile Stress, Heat-Set For 6 Seconds .................................. 148
Figure 4.4b. Statistical Models For Homopolymer Ultimate
Tensile Stress, Heat-Set For 15 Seconds ................................ 148
Figure 4.4c. Statistical Models For Homopolymer Ultimate
Tensile Stress, Heat-Set For 30 Seconds ................................ 149
Figure 4.5. Statistical Models For Homopolymer Ultimate
Tensile Strain, Heat-Set For 15 Seconds ................................ 150
Figure 4.6. Statistical Model For Homopolymer Strain
Hardening Rate, Heat-Set For 15 Seconds ............................. 150
Figure 4.7a. Statistical Model For Copolymer Yield Stress
In The Major Axis, Heat-Set For 6 Seconds ........................... 151
Figure 4.7b. Statistical Model For Copolymer Yield Stress
In The Major Axis, Heat-Set For 15 Seconds ......................... 151
Figure 4.7c. Statistical Model For Copolymer Yield Stress
In The Major Axis, Heat-Set For 30 Seconds ......................... 152
Figure 4.8. Statistical Model For Copolymer Yield Strain
In the Minor Axis, Heat Set For 15 seconds ........................... 153
Figure 4.9a. Statistical Models For Copolymer Ultimate
Tensile Stress, Heat-Set For 6 Seconds .................................. 154
Figure 4.9b. Statistical Models For Copolymer Ultimate
Tensile Stress, Heat-Set For 15 Seconds ................................ 154
Figure 4.9c. Statistical Models For Copolymer Ultimate
Tensile Stress, Heat-Set For 30 Seconds ................................ 155
Figure 4.10. Statistical Model For Copolymer Ultimate
Tensile Strain, Heat-Set For 15 Seconds ................................ 156
Figure 4.11. Statistical Model For Copolymer Strain
Hardening Rate, Heat-Set For 15 Seconds ............................. 156
Figure 4.12. Statistical Model For Creep Strain As An
Effect Of Time At Two Draw Areas,S and 10 ....................... 157
Figure 4.13. Statistical Model For Creep Strain As An
Effect Of Draw Area .............................................................. 157
Figure 4.14. Statistical Model For Homopolymer
Shrinkage At 65C, Heat-Set For 15 Seconds ......................... 158
Figure 4.15a. Statistical Models For Homopolymer
Shrinkage At 85C, Heat-Set For 6 Seconds .......................... 159
Figure 4.15b. Statistical Models For Homopolymer
Shrinkage At 85C, Heat-Set For 15 Seconds ......................... 159
Figure 4.15c. Contour Plot Of Homopolymer Major Axis
Shrinkage At 85C, Heat-Set For 15 Seconds ......................... 160
Figure 4.15d. Statistical Models For Homopolymer
Shrinkage At 85C, Heat-Set For 30 Seconds ......................... 161
Figure 4.16a. Statistical Models For Homopolymer Shrinkage
At lOOoC, Heat-Set For 6 Seconds ......................................... 162
Figure 4.16b. Statistical Models For Homopolymer Shrinkage
At 100C, Heat-Set For IS Seconds ....................................... 162
Figure 4.16c. Contour Plot Of Homopolymer Major Axis
Shrinkage At lOOoC, Heat-Set For IS Seconds ....................... 163
Figure 4.16d. Statistical Models For Homopolymer Shrinkage
At 100C, Heat-Set For 30 Seconds ....................................... 164
Figure 4.17a. Statistical Models For Homopolymer Shrinkage
At 110C, Heat-Set For 6 Seconds ......................................... 165
Figure 4.17b. Statistical Models For Homopolymer Shrinkage
At 110C, Heat-Set For IS Seconds ....................................... 165
Figure 4.17c. Contour Plot Of Homopolymer Major Axis
Shrinkage At 110C, Heat-Set For IS Seconds ....................... 166
Figure 4.17d. Statistical Models For Homopolymer Shrinkage
At 110C, Heat-Set For 30 Seconds ....................................... 167
Figure 4.18. Statistical Models For Copolymer Shrinkage
At 65C, Heat-Set For IS Seconds ......................................... 168
Figure 4.19a. Statistical Models For Copolymer Shrinkage
At 85C, Heat-Set For IS Seconds ......................................... 168
Figure 4.19b. Contour Plot Of Copolymer Major Axis
Shrinkage At 85C, Heat-Set For IS Seconds ......................... 169
Figure 4.20a. Statistical Models For Copolymer Shrinkage
At 100C, Heat-Set For 6 Seconds ......................................... 170
Figure 4.20b. Statistical Models For Copolymer Shrinkage
At 100C, Heat-Set For IS Seconds ....................................... 170
Figure 4.20c. Contour Plot Of Copolymer Major Axis
Shrinkage At 100C, Heat-Set For IS Seconds ....................... 171
Figure 4.20d. Statistical Models For Copolymer Shrinkage
At 100C, Heat-Set For 30 Seconds ....................................... 172
Figure 4.2Ia. Statistical Models For Copolymer Shrinkage
At 110C, Heat-Set For 6 Seconds ......................................... 173
Figure 4.2Ih. Statistical Models For Copolymer Shrinkage
At 110C, Heat-Set For 15 Seconds ....................................... 173
Figure 4.21c. Contour Plot Of Copolymer Major Axis
Shrinkage At 110C, Heat-Set For 15 Seconds ....................... 174
Figure 4.2Id. Statistical Models For Copolymer Shrinkage
At 110C, Heat-Set For 30 Seconds ....................................... 175
Figure 4.22a. Statistical Models For Degree Of Crystallinity
As Measured By DSC. Heat-Set For 6 Seconds .................... 176
Figure 4.22h. Statistical Models For Degree Of Crystallinity
As Measured By DSC. Heat-Set For 15 Seconds .................. 176
Figure 4.22c. Statistical Models For Degree Of Crystallinity
As Measured By DSC. Heat-Set For 30 Seconds .................. 177
Figure 4.23a. Statistical Models For Degree Of Crystallinity For A
Heat-Set Time Of 6 Seconds As Measured By Density ........... 178
Figure 4.23h. Statistical Models For Degree Of Crystallinity For A
Heat-Set Time Ofl5 Seconds As Measured By Density ......... 178
Figure 4.23c. Statistical Models For Degree Of Crystallinity For A
Heat-Set Time Of 6 Seconds As Measured By Density ........... 179
Figure 4.24. Copolymer Shrinkage In The Major Axis
At 65C, Heat-Set For 15 Seconds.
From Conventional Experimentation Series ............................ 180
Figure 4.25. Copolymer Shrinkage In The Minor Axis
At 65C, Heat-Set For 15 Seconds.
From Conventional Experimentation Series ............................ 180
Figure 4.26. Copolymer Shrinkage In The Major Axis
At 85C, Heat-Set For 15 Seconds.
From Conventional Experimentation Series ............................ 181
Figure 4.27. Copolymer Shrinkage In The Minor Axis
At 85C, Heat-Set For 15 Seconds.
From Conventional Experimentation Series ............................ 181
Figure 4.28. Copolymer Shrinkage In The Major Axis
At 100C, Heat-Set For IS Seconds.
From Conventional Experimentation Series ............................ 182
Figure 4.29. Copolymer Shrinkage In The Minor Axis
At lOOoC, Heat-Set For IS Seconds.
From Conventional Experimentation Series ............................ 182
Figure 4.30. Copolymer Shrinkage In The Major Axis
At I lOoC, Heat-Set For IS Seconds.
From Conventional Experimentation Series ............................ 183
Figure 4.3\. Copolymer Shrinkage In The Minor Axis
At \1 OOC, Heat-Set For 15 Seconds.
From Conventional Experimentation Series ............................ 184
Figure 4.32a. Secondary Melting Peak Temperature As A
Function of Heat-Setting Temperature ................................... 184
Figure 4.32b. Degree Of Crystallinity As Measured By DSC For
Films From Conventional Experimentation Series ................... 185
Figure 4.33. Degree of Crystallinity As Estimated By
Refractive Index Measurements For Films
From Conventional Experimentation Series ............................ 185
Figure 4.34. Optical Absorbtion As Measured By
Hazeguard Haze Meter Of Films From
Conventional Experimentation Series ..................................... 186
Figure 4.35. Plot Ofln(absorption)/thickness As A Correction
For The Systematic Variation In Sample Thickness ................ 186
Figure 4.36. Optical Absorption Measured By Microphotometry
On Films From Conventional Experimentation Series ............. 187
Figure 4.37. Optical Absorption/thickness As Measured By
Microphotometry On Films
From Conventional Experimentation Series ............................ 187
Figure 4.38a. Birefringence Measurements On FilmsWith A Draw
Area Of 5 From Conventional Experimentation Series ............ 188
Figure 4.38b. Birefiingence Measurements On Films With A Draw
Area Of 10 From Conventional Experimentation Series .......... 188
Figure 4.38c. Birefringence Measurements On Films
Heat-Set At 220C From Conventional
Figure 4.39
Figure 4.40.
Figure 4.41.
Figure 4.42.
Figure 4.43a.
Figure 4.43b.
Figure 4.44.
Figure 4.45.
Figure 4.46.
Figure 4.47
Figure 4.48
Figure 4.49.
Experimentation Series ........................................................... 189
Refractive Index Measurements On Films
From Conventional Experimentation Series ............................ 189
Shrinkage In The Hoop Axis Of Heat-Set Bottles
With The Corresponding Predicted Shrinkage ........................ 190
Shrinkage In The Axial Axis Of Heat-Set Bottles
With The Corresponding Predicted Shrinkage ........................ 190
Degree Of Crystallinity Of Heat-Set Bottles
As Measured By DSC ............................................................ 191
Degree Of Crystallinity Of Heat-Set Bottles
As Measured By Density Compared With Model
Prediction, Model Heat-Setting Time of 15 Seconds .............. 191
Degree Of Crystallinity Of Heat-Set Bottles
As Measured By Density Compared With Model
Prediction, Model Heat-Setting Time of25 Seconds .............. 192
Orientation Functions Of Heat-Set Bottles
Measured By P-FTIR (1018cm-
l
) .......................................... 193
Orientation Functions In The Crystalline Phase
Of Heat-Set Bottles Measured By P-FTIR (973cm-
l
) ............ 194
Orientation Functions In The Amorphous Phase
Of Heat-Set Bottles Measured By P-FTIR (1578cm-
l
) .......... 194
Bulk Dichroic Ratio Of Heat-Set Bottles ................................ 195
Dichroic Ratio Of The Crystalline Phase
As A Function of Heat-Setting Temperature
For Heat-Set Bottles .............................................................. 196
Dichroic Ratio Of The Amorphous Phase
As A Function of Heat-Setting Temperature
For Heat-Set Bottles .............................................................. 196
Chapter Five
Figure 5.1. Homopolymer Yield Stress Model In The Major
Axis As A Function Of Degree Of Crystallinity ....................... 263
Figure 5.2. Copolymer Yield Stress Model In The Major
Axis As A Function of Degree of Crystallinity ........................ 263
Figure 5.3. Copolymer Creep Strain Model As A Function
Of Degree Of Crystallinity ...................................................... 264
Figure 5.4. Homopolymer Shrinkage Model At llOoC In The
Major Axis As A Function Of Degree Of Crystallinity ............ 265
Figure 5.5. Copolymer Shrinkage Model At llOoC In The
Major Axis As A Function Of Degree Of Crystallinity ............ 265
Figure 5.6. Homopolymer Shrinkage Model At llOoC In The
Minor Axis As A Function Of Degree Of Crystallinity ............ 266
Figure 5.7. Copolymer Shrinkage Model At i lOoC In The
Minor Axis As A Function Of Degree Of Crystallinity ............ 265
Figure 5.8. Predicted Theoretical Volume Shrinkage At IlOoC
For A 1.51 Bottle With A Draw Area Of 10 ............................ 267
Figure 5.9. Predicted Theoretical Volume Shrinkage At IlOoC
For A 1.51 Bottle With A Draw Area Of 5 .............................. 267
Figure 5.10. Theoretical Volume Shrinkage At 1l0oC For A 1.51
Copolymer Bottle. From Measurements Made
On Conventional Experimentation Series ................................ 268
Figure 5.11. Homopolymer Predicted Shrinkage Aspect Ratio At IlOoC .... 269
Figure 5.12. Predicted Copolymer Shrinkage Aspect Ratio At 110C ......... 269
Figure 5.13. Actual Shrinkage Aspect Ratio Of
Conventional Experimentation Series At 110C ...................... 270
Figure 5.14 Copolymer Major Axis Yield Stress Model As
A Function OfBirefiingence .................................................. 271
Figure 5.15. Copolymer Creep Strain Model As A Function
Of Birefringence .......................................................... .. 272
Figure 5.16. Natural Logarithm Of The Creep Strain Model,
As A Function Of Birefiingence ......... , ................................... 272
Figure 5.17. Copolymer Birefiingence As A Function Of
Degree Of Crystallinity .................................. " ....................... 273
Figure 5.18. Shrinkage In The Major Axis At I JOoC As A Function
Of Birefiingence. From The Conventional
Experimentation Series ........................................................... 274
Figure 5.19. Shrinkage In The Minor Axis At 1 JOoC As A Function
OfBirefringence. From The Conventional
Experimentation Series .......................... , ................................ 274
Figure 5.20. Theoretical Volume Shrinkage For A 1.51 Bottle Based
On Linear Shrinkage Measurements Made On The
Side Walls Of Heat-Set Bottles .............................................. 275
Figure 5.21. Shrinkage At 1 JOoC As A Function
Of Heat-Setting Temperature For Heat-Set Bottles ................ 276
Figure 5.22. Degree of Crystallinity Plotted Against
Heat-Setting Temperature For Heat-Set Bottles ..................... 277
Figure 5.23. Shrinkage Plotted Against Degree of Crystallinity
For Heat-Set Bottles .............................................................. 277
Figure 5.24. Homopolymer And Copolymer Degree Of Crystallinity
Models For Heat-Setting Temperatures Of
JOOoC And 200C As A Function Of Draw Area .................... 278
TABLES
Chapter Three
Table3.1.
rcr 'MelinarI Bottle Grade
Polyethyleneterephthalate Materials ........................................ 103
Table 3.2.
FED Heat Setting Programme ................................................ 104
Table 3.3.
Experimental Values For FED Films ....................................... 105
Table 3.4.
Experimental Plan For Conventional
Experimentation Films ............................................................ 106
Chapter Four
Table 4.1.
Sample Values From Mechanical Models ............................... 197
Table 4.2.
Sample Values From Creep Strain Models .............................. 197
Table 4.3.
Sample Values From Shrinkage Models ................................. 198
Table 4.4.
Sample Values For Crystallinity
From Thermal Analysis and Density Models ........................... 199
1 INTRODUCTION
1.1 GENERAL BACKGROUND
Polyethylene Terephthalate (PET)was originally patented as a fibre forming
polymer in 1941 and commercial production of Terylene (lCI Trade Mark)
commenced in 1955. Continuing development expanded the use of the
material to films, then more recently to thin walled, high strength beverage
containers. The excellent mechanical and gas barrier properties, in
combination with chemical resistance and approval by the FDA for food
use, has given PET complete dominance in the multi-serving carbonated
beverage container market.
1.2 BOTTLE PRODUCTION
For PET to develop sufficient strength and barrier properties, an injection
moulded preform has to undergo significant biaxial deformation. It is this
requirement that dictates that carbonated soft drinks (CSD) containers
caunot be produced in PET using the more common technology of extrusion
blow moulding.
The conversion of resin to artefact is achieved by two separate processes:
one is a melt process and the other is a deformation process, carried out at
typically 20-40C above Tg. Suitably dried (<40 ppm) PET granules are
injection moulded into preforms. Rapid cooling of the melt in the mould is
required to produce high clarity, distortion free, amorphous preforms.
Acetaldehyde is generated by thermal degradation of the molecules during
melt processing and having a distinctive fruity flavour, can diffuse from the
bottle wall and taint the contents. Acetaldehyde generation must be
minimised and this is achieved by adopting the mildest possible moulding
conditions, such as, minimum moulding temperature consistent with clarity;
minimum residence time, minimum screw speed and back pressure.
1
The second stage of the conversion is to stretch blow mould the preforms
into bottles. Apart from imparting the fmal dimensions to the bottle, it is
during this stage that the essential biaxial deformation is achieved.
Longitudinal orientation is established with the aid of a mechanical stretch
rod, inserted into the preform. This differentiates stretch blow moulding
(SBM) from the more common extrusion or injection blow moulding
processes.
PET has a glass transition temperature of about 70C. Therefore containers
fabricated from PET will soften sufficiently allowing the containers to
shrink due to the relaxation of residual orientation. There are two major
reasons why a higher service temperature would be very desirable in PET:
a) For a food to be pasteurised it must be "hot filled", i.e .. filled at
between 85C and 90C, well above the distortion temperature of
PET containers. Therefore, the container must be able to withstand
this excursion to elevated temperature without distortion.
Sterilisation processes are even more demanding, requiring
temperatures of over 100C to be maintained for about 15 minutes.
b) Environmental concern has meant that non-returnable containers are
increasingly unacceptable. Even the rationales of biodegradation
and recycling are being called into question for this type of
packaging. By far the best solution is the returnable, multi-trip
bottle. However, the required washing temperature is well above
the current distortion temperature of existing PET containers.
Increasing the upper service temperature would undoubtedly increase the
market potential for PET. Hence, resin and primary equipment
manufacturers have exerted considerable effort into achieving this. In fibre
and biaxially oriented film production a post-draw anneal, referred to as
heat-setting, is given to PET fibres which can raise the distortion
temperature as well as enhancing some other properties. Transfer of this
fibre technology to bottle production appears the most attractive route to
producing hot fillable, hot washable containers. Examples of heat-setting
production equipment have recently been commercialised by Krupp
Corpoplast,
l
Nissei and SideL
2
The principle of heat-setting is well established. Straining PET will cause
some crystallisation to occur, to an extent detennined by the stretching
conditions and the molecular weight of the resin. The strain-induced
crystallites will act as restraining junction points in the polymer network.
These persist above Tg but being small and imperfect are disrupted below
Tm. Constrained annealing will increase the degree of crystallinity in
drawn PET. These new junction points will persist up to their formation
temperature, which will prevent or substantially reduce shrinkage at the
higher service temperatures. The aim of this project was to generate
fundamental information on this technique, chiefly by process simulation
and microstructural analysis, with a view to possible exploitation in the
SBM markets.
1.3 OBJECTIVES AND STRATEGY
1.3.1 Project Objectives
To reduce thermally induced shrinkage in PET food containers to allow
higher filling or service temperatures. The temperature ranges of interest
may be summarised as:
80C-90C
90C-100C
100C-110C
pasteurised filling
hot filling of jam and bottle washing
sterile filling conditions
3
1.3.2 Project Strategy
The global project strategy is as follows:
1) To simulate injection stretch blow moulding by biaxially drawing
PET sheet under varying conditions.
2) To heat-set the sheet according to pre-set experimental programmes.
3) To analyse this material in terms of mechanical properties,
crystallinity and orientation.
4) To correlate these to thermal shrinkage measurements.
5) To advise on, and/or build equipment for which the principles in part
1 can be exploited.
4
2 LITERATURE SURVEY
2.1 POLYETHYLENE TEREPHTHALATE PRODUCTION
2.1.1 Principle of PET Polymerisation Techniques
PET is usually manufactured by two routes, based on the starting
terephthalate monomers used. Either dimethyl terephthalate (DMT) or
terephthalic acid (TP A) is reacted with ethylene glycol (EG) to yield a fIrst
stage pre-polymer product. In the DMT process the effective monomer, bis
(hydroxyethyl) terephthalate (bis HET), is fIrst formed by the catalysed
transesterification of DMT with two equivalents of EG. In contrast, in the
TP A process, the esterifIcation reaction is self-catalysed by the carboxyl
groups' acidity. Apart from not requiring a first stage catalyst, the initial
esterifIcation product is the same for both processes: viz., a low molecular
weight mixture of PET oligomers. The fmal process involves
poly condensing at elevated temperatures and high vacuum in the presence
of a metallic catalyst (usually an antimony compound). The
polycondensation process involves both esterification and ester interchange
reactions, with the removal of water and EG.
Polyester chemistry involves reversible reactions necessitating the removal
of methanol, water and EG to increase the molecular weight of PET to the
desired level. The molten mass of PET is then extruded from the reactor,
quenched rapidly with water cooling and pelletised. For high molecular
weight grades used in bottle manufacture, the pellets are further
polycondensed in a solid state polymerisation reactor operating below the
melting point of PET. In the solid state polymerisation (SSP) process, the
initially amorphous transparent pellets are thermally crystallised and
consequently increase in density (1.335 to 1.4g1cm
3
) and turn opaque. The
number average molecular weight (Mn) used for most containers (e.g.
carbonated beverages) ranges from 24,000 to 31,000 glmol, corresponding
to intrinsic viscosities of 0.7 to 0.85 dl/g respectively2.
5
2.1.2 Ethylene Glycol and Acetaldehyde
The two bi-products produced during the production of PET are diethylene
glycol (DEG) and acetaldehyde (AA). DEG is an ether-containing glycol
incorporated in the polyester chain. The DEG is effectively a comonomer
glycol which reduces the melting point of the PET resin proportional to the
amount incorporated in the structure. Higher levels of DEG permit lower
extruder temperatures during moulding operations. The incorporation of
DEG also interrupts the regularity of the PET structure and reduces the
crystallisation tendency of the resin. The modulus potential after
orientation and glass transition temperature (Tg) are also lowered, with
increasing tendency for creep, especially at high ambient temperature
exposures.
Acetaldehyde (AA), which affects the taste of many carbonated beverages
at a 60 parts per billion level, is present in all PET resins. It is produced by
thermal decomposition of hydroxyl end groups at temperatures greater than
200C, or their interaction with primary vinyl benzoate from ester bond
thermal scissions in the melt. Thus all PET resin melts contain a certain
amount of AA. Although AA boils at 21C it has sufficient solubility in
PET ( in addition to the melt viscosity retardation influence) such that
approximately 3 parts per million are entrapped in the pelleted resin.
Because of AA's ability to taint foodstuffs, its generation must be strictly
controlled by using effective drying, and mild injection moulding conditions
i.e. avoiding high melt temperature and excessive shear stress 3.
6
2.1.3 PET Copolymers
The introduction of the one piece bottle requires the base to be altered in
,
design from hemispherical to petalloid. The draw ratios in this complex
region are highly variable and so lead to the development of complex
microstructures. The homopolymer grades available at the time had a
tendency to crystallise thermally (forming spheruIities) at the injection
phase whilst, the Iow draws generally encountered in the petalloid base
further increased the level of spherulitic crystallinity in this region. This
manifests itself usually as a thin white ring around the base, centred on the
injection gate, and caused embrittlement and a susceptibility to
environmental stress cracking in this critical region.
By copolymerising PET with a small amount of isophthalic acid (IPA) or
cyclohexanedimethanol (CHDM), 4 the tendency for PET to crystallise can
be sufficiently reduced without detracting from the container's performance.
The reduction in crystallisation rate of copolymers also allows preforms to
be removed from the injection mould warmer and therefore sooner, thereby
reducing cycle times. The copolymer resins have a lower melting point
which require less energy to process and lower processing temperatures and
so less acetaldehyde is generally evolved. Copolymers also have a higher
natural draw ratio and so for a given IV are easier to stretch.
2.2 THEORY OF SHRINKAGE
Shrinkage is of interest to this project for two reasons: frrstly, the overriding
objective of the study is the reduction or eradication of thermally induced
shrinkage in oriented PET containers. Secondly and more subtly, it is a
good diagnostic tool to detect structural changes in partially oriented
polymers such as PET 5. In order to understand shrinkage in thermoplastics
it is important to examine the structure of an amorphous polymer and to be
aware of how this structure is altered by orientation. If orientation is
present in the polymer then shrinkage will occur due to the relaxation of the
stretched carbon-carbon linkages in the molecular chain. This tends to be
greater in the alignment direction than in the transverse direction 6.
7
An unoriented thermoplastic may be considered as a mass of randomly
arranged, intricately intertwined spring-like molecules, frozen in place by
interchain tertiary bonding. When such a material is drawn, two process
occur that alter the internal structure in two major ways:
a) The spatial position of the molecules in the material is altered. Drawing
exerts a rotating force which acts to align the chains (or segments of
chains that are free to move) parallel to the stress direction. Because the
chains are so long and entangled, complete alignment is seldom achieved
in commercial practice. To cause this chain alignment, the rotating force
must be great enough to overcome the sum of the friction-like forces
exerted by side-chain or segment entanglement, and intermolecular
attraction. This energy is non-recoverable and so this deformation mode
does not contribute to thermally induced shrinkage. In rheological terms
it can be described as viscous flow and as such, has a time component.
b) The spring-like molecular chains are extended when the drawing stress
overcomes the intrachain forces (that result in chain coiling) and shape-
forming forces (interatomic forces at fixed bonding angles) within the
chain backbone. Consistent with the spring model of the chain, the
resulting strain is an elastic deformation, and as such, the force is
associated with a potential energy. This is generally recoverable when
the molecular stress is relieved. Therefore only this mode of
deformation can give rise to thermally induced shrinkage
7
.
From this it would seem reasonable to treat PET as a viscoelastic material
and so subject to the same considerations. However, the above discussion
makes no mention of the effect of crystallites, which are formed during
drawing, on the deformation modes and so a purely viscoelastic treatment of
shrinkage in the amorphous phase will yield an incomplete picture.
Peszkin, Schultz and Ling and Biangardi and Zachmann
9
have reported
negative shrinkage (elongation) in oriented PET. Peszkin et al found this to
occur in samples annealed at 160C or above. From X-ray analysis Peszkin
et al found that in these samples, crystallisation began very early so that no
re-coiling could occur.
8
2.2.1 Amorphous Phase
BosleylO concluded via viscoelastic theory of polymers that shrinkage
occurred in the free rotation phase. This only occurs in the amorphous
regions above Tg. Therefore shrinkage depends upon structural factors such
as molecular weight, molecular weight distribution, crystalline structure and
the degree of chain entanglement which will affect the orientation in the
amorphous phase. Furthermore, shrinkage is also controlled by the network
of molecular entanglements.
Samuels
ll
and Wilson
l2
demonstrated that semicrystalline PET fibres can
be treated as a two-phase material and that macroscopic shrinkage occurs by
disorientation of the amorphous phase. Hsuie and Yeh 5 demonstrated that
additions of 40% ethylene isophthalate caused the largest increase in
shrinkage. This was due to the copolyesters having more amorphous phase,
are more loosely oriented and overall, have greater.chain mobility.
If an oriented polymer sample is heated but constrained from shrinking, it
will generate a retractive force which maybe expressed in terms of a
shrinkage stress. The examination of shrinkage stress, as well as the actual
shrinkage, via the rubber elasticity theory can yield very useful information
about the shrinkage mechanisms at work. The starting point for modelling
the shrinkage in a molecular chain system is Kulm and Grun's theory13. By
considering a polymer to be a freely jointed chain of identical links it was
possible to model the behaviour of polymer chains in an amorphous
network. Using this theory, Treloar
l4
obtained fundamental information
about the molecular chains of cross-linked natural rubber. This analysis has
been extended to non cross-linked polymeric systems where the presence of
chain entanglements and crystallites results in an effective molecular
network structure. Alfrey et al
15
studied semicrystalline PVC films, and
Rudd and Andrews
l6
examined polystyrene.
9
Pinnock and Ward
17
investigated the shrinkage forces of PET using the
rubber elasticity theory. They restricted the investigation to draws
(extension ratios) of no greater than 2. Above a draw of 2, strain induced
crystallisation occurs, causing the material to depart from rubber like
behaviour. Bhatt and Bell
18
used the Treloar modification of the Kuhn-
Grun theory to model the full extension range. Their work assumed that the
crystallites act as immobile regions that prevent the shrinkage from
occurring to its fullest extent. Their results concurred with Dumbleton 19 in
that there is negligible disorientation of the crystallites during shrinkage so
the shrinkage is assumed to be totally entropic in nature and no bond
distortions occur in either the amorphous or crystalline phase. They
assumed and subsequently demonstrated that no further crystallisation
occurs in PET during shrinkage. Bhatt and BelP8 also found that the
crystallites reached a high degree of orientation at a relatively low draw
(times 3.5) after which the increases in orientation function were small.
Because they found the average value of the orientation functions
(measured by XRD) of the samples before and after shrinkage force testing
to be essentially the same for all draw ratios, Bhatt and Bell were also able
to conclude that performing shrinkage force measurements (performed at
70C for over 60 minutes) did not significantly alter the orientation of the
crystallites.
However, Long and Ward 20, whilst agreeing that retractive force for a draw
of less than or equal to 1.7 could be modelled successfully using the
Gaussian theory, they considered that the Bhatt and Bell treatment, (see
above), could not model the development of amorphous orientation with
extension ratio very successfully. For higher strains they applied a model
proposed by Edwards and Vilgis21. The major difference with this model is
that the temporary molecular entanglements are modelled as slip links
which are effectively a development of the well-established network theory.
10
Long and Ward
20
concluded that the rubbery region of PET may be
considered in terms of two regimes: the Iow strain Gaussian region,
followed by the strain-induced activation of the slip links, for which the
Edward-Vilgis model provides a good description for draws up to 4.6. As
the draw is further increased there is a clear drop in reduced shrinkage force
that the Edward-Vilgis model cannot accommodate. This deviation at high
draws is probably due to the crystallinity becoming continuous and
hindering the development of the shrinkage force. They also demonstrated
that the peak shrinkage forces are due to the contraction of the shortest
chains and these control the load bearing properties, especially the tensile
modulus.
Perkins
22
demonstrated a relationship between "entropic restoring stress" of
oriented PET and molecular weight, which is somewhat at odds with
Pinnock and Ward
l7
, who found shrinkage force to be independent of
molecular weight. Perkins also reported that the magnitude of the retractive
stresses is not directly related to the observed specimen shrinkage.
De Vries and Bonnebat
23
showed that in amorphous polymers,
birefringence is proportional to the frozen in "internal entropic stress".
However the total internal stress in an oriented glassy polymer may be
significantly larger than this. They then were able to split the two stresses
in PVC due to the considerable differences in relaxation rate of the
distortional stresses, compared to the decay of entropic stress (orientation).
2.2.2 Crysta Iline Phase
AIl the discussions outlined above, have in common that all the shrinkage is
considered to arise from the relaxation of the amorphous phase and that the
crystallites serve to restrict this. However, thermally induced shrinkage can
also be encountered in predominantly crystalline oriented materials e.g.
Polyethylene and isotactic Polypropylene, and so the possibility of induced
shrinkage from the crystalline phase must also be considered. Pinnock and
Ward
l7
found for spin oriented fibres, that a draw ratio above two caused
strain-induced crystallisation to occur. They used birefringence to quantify
the induced orientation and demonstrated a linear relationship between
birefringence and shrinkage force.
11
Peterlin, working with both Polyethylene and isotactic Polypropylene,24
proposed a shrinkage model referred to as the "taut-tie-molecules" model.
In this model, the shrinkage in oriented crystalline polymers is attributed to
tie molecules formed when molecules which, during solidification, were
partially included in the crystal lattices of two different lamellae, so that the
intervening chain is prevented from crystallising. Therefore, this
mechanism can not be applied to amorphous PET preforms.
T
TIE
,
,
I r,<'
l_l/ ....
I
\ \
"
/ \
I
~
\
T \ I
I I
, /
/ /
I /
\
\
\ "l...... I
\ (\ 1 /
,
\
\
\
I
I
E ,X
E' \ '-1./ 1/
\ \,' 1.--/ I T
, I I ,..,.. ..... ",
\t=-7\ / //
"
T
T
,
\
,
,
I
I .1.
V/
/
I
\
\
"
\
I
\
\
\ '
T
E
,/
,/
/
,
I
,
\
I
I
,
)c-A
\
I
\
\
\
I I
/ \
,
,
Crystal.
D Tic .. gmen! block.
Schematic structural model for drawn PET, illustrating the three possible
modes of intercrystalline molecular connection: tie-segments only (T);
entanglements only (E); tie-segments plus entanglements (FIE). A block of
crystals interconected by tie-segments constitutes a 'tie-segment block'.
From Buckley and Salem
25

12
Peterlin's study is of materials that display a spherulitic or microspherulitic
morphology prior to stretching. He does make the point that this model
would need "radical modification" to be applicable to materials such as
PET, which in the case of bottle manufacture, starts as amorphous preforms.
This means that initially there are no crystallites to be cleaved and
significant strain induced crystallinity is not apparent until draws ratios in
excess of2
17
have been achieved. However, electron microscopy studies on
single Polyethylene crystals have shown that these undergo phase
transformation and twinning, chain slip and tilt when subjected to
increasing plastic deformation
26
, 27, 28. This allowed Peterlin
29
to propose
that the mechanism of crystal fracture into single blocks of folded chains,
partially unfolds the molecules connecting two adjacent blocks. This
incorporates the unfolded sections as tie molecules into the newly formed
microfibril. This mechanism could operate as the strain induced crystallites
are formed and then fractured, thus the deformation of the crystalline phase
gives rise to a small amount of highly stressed amorphous material. This
stress can only be released, and so lead to shrinkage, when the restraining
crystallites are disrupted, which will occur at temperatures above Tg, but
due to the small and imperfect nature of the cleaved crystallites, will occur
over a temperature range below Tm. In this way the crystalline phase can
also contribute to overall shrinkage, in contrast to the viscoelastic theory.
Indeed, recent investigations
30
have demonstrated that shrinkage may occur
in both amorphous and crystalline phases. Sun and Magi11
30
also concluded
that in highly oriented crystalline polymers, shrinkage occurs
simultaneously with melting but in systems with low to moderate draw
ratios, shrinkage often occurs between Tg and Tm. They also concluded
that shrinkage behaviour is primarily dependent upon the extent of drawing,
independent of how it is produced (e.g. uniaxial, or biaxial).
Dumbleton
31
showed that recrystallisation by chain folding near the melting
point had a significant reduction on shrinkage. Statton
32
, in contradiction to
Dumbleton, proposed that the refolding of molecules during thermal
activation is the major contributor to shrinkage, whereas Amano et al.
33
proposed that shrinkage was caused by the accumulation and subsequent
expulsion of defects from the disordered regions of the crystals.
13
Jordan et a/.
34
questioned if the Peterlin "taut tie-molecules" model could
adequately explain the change in long period, i.e. the new thickness of the
lamella crystallites in the drawn material, and the distinct uniaxial
orientation of the lamellae in the microfibrils. They also found it difficult to
attribute high modulus, tensile strength and large shrinkage capacity to only
"tie-molecules". Juska and Harrison
35
, 36 proposed a model by which the
transformation from spherulite to fibre depends only on a phase transition.
In this model regions of the semi crystalline material, under high local stored
elastic energy, become unstable and "melt" at the draw temperature. The
tensile stress present can then cause rapid extension and subsequent strain-
induced recrystallisation. These extended crystallites nucleate the rest of
the melt, which crystallise epitaxially upon the extended chain core. The
resulting fibrous morphology is essentially identical to that found in
commercially spun fibres. This model as outlined by Juska, once again
requires the material to be in a semi crystalline form prior to stretching. It is
unclear from the literature if this model could be applied to a material such
as PET.
Prevorsek, Kwon and Sharma
37
provided a useful summary of the effect of
shrinkage on the morphology of a PET or Nylon fibre, the more significant
aspects of which are summarised here. Shrinkages between 0% and 10%
decreases the long period (the spacing between two adjacent crystallites in
the microfibrils) from 190 to 163 Angstroms and a decrease in amorphous
orientation. Shrinkage increases regular chain folding and the lateral order
of the crystallites. If the shrinkage is between 10% and 20%, the
amorphous orientation is further decreased and the degree of regular chain
folding is increased markedly. There is a small increase in the level of
crystallinity and an improvement in macrolattice lateral order. There is no
change in long period and lateral order of crystallites. For contractions
between 20 and 38% there is a small but significant decrease in the degree
of crystallinity and a decrease in the degree of regular chain folding. There
is no change in the amorphous orientation function and a relatively small
decrease in the crystalline orientation function. There is no significant
change in the long period.
From the literature reviewed in this section it is clear that no one molecular
model exists that can describe the shrinkage behaviour of PET over a large
range of draws.
14
2.3 MORPHOLOGY AND THE EFFECT OF DEFORMATION
The versatility of thermoplastic polyesters stems from the ability to produce
various states of molecular arrangement by using different processing
conditions. Deformation can either occur below or above Tg and is referred
to as cold or hot drawing respectively. Spruiell, McCord and Beuerlein
38
found that PET strained at ooe and above 0.94 min-
I
at 24e fractured in a
brittle manner. If the drawing conditions permit, then deformation below
Tg always results in the formation of a neck in PET. As the neck passes
through the material, it elongates immediately to the equivalent of the
natural draw. Yeh and Geil
39
reported voids several microns long in PET
that had been drawn at 25e which would have made inferring crystallinity
levels from density measurements unreliable, in such circumstances.
Foster and Heap40 and Thompson41 noted that above Tg PET can deform in
two separate ways. At relatively high strain rates PET deformation is
accompanied by the formation of a neck and relatively high draw stress are
generated. This mode of deformation gives rise to a high degree of strain-
induced crystallinity and molecular orientation. At lower strain rates the
second deformation mode, referred to as flow deformation, occurs. It is
characterised by a low deformation stress with little crystallisation and
molecular orientation. The deformation is achieved by a uniform reduction
in cross sectional area without the formation of a neck. As both molecular
orientation and strain-induced crystallinity are required for eSD
applications, it is essential that the conversion from preform to bottle is via
the fIrst, high strain, deformation mode. In fact, Spruiell et ail
8
showed that
above Tg the degree of orientation achieved is markedly dependent on strain
rate. The ability for a material to stabilise a neck reflects strain hardening
properties, as the material's tensile strength must be increased by
proportionately more than the reduction in cross sectional area. This
reduction will equal the natural draw ratio, assuming constant volume
deformation. Spruiell reported the natural draw for PET uniaxially
deformed at 60C to be 4.5. Therefore the strain hardened material has a
tensile strength that has been enhanced by a minimum of 4.5 times by
drawing. This effect can more readily be appreciated by reference to true
stress-strain plots as opposed to nominal stress-strain plots.
15
Foster and Heap40 and Thompson
41
concluded that strain-induced
crystallisation in PET is initiated by a critical stress level which is
independent of the temperature of deformation and this is established when
the strain rate exceeds a critical value. Spruiell, McCord and Beuerlein's38
results indicated to them that a significant degree of crystallisation occurs
when the material is deformed through the neck. The rapid local straining
that occurs in the neck aligns the molecules in the direction of the stretch.
This alignment lowers the configuational entropy, thus creating conditions
which are favourable for crystallisation to take place. Thompson and
Marshall
42
reported a small rise in temperature in the shoulder of the neck,
as it passed a thermocouple embedded in PET. This was interpreted as
further evidence that crystallisation occurs during necking. The
deformational work is nearly all converted to heat4
2
43 which at high strain
rates would have insufficient time to be dissipated, thereby giving a
temperature rise in the material being drawn. If Thompson's work was
performed at low enough strain rates to allow the draw to be considered
isothermal, then any release of energy by crystallisation would dissipate as
it was released, and so would not be manifested as a temperature increase.
This indicates that the work was conducted at relatively high strain rates
which would give quasi-adiabatic conditions. This adiabatic heating by the
release of energy associated with the work of drawing is a probable
explanation of this measured increase in temperature. This does not infer
that crystallisation does not occur in the neck region,) but simply, that the
evidence put forward by Thompson and Marshall does not necessarily
support it. Bonnebat et al
44
stretched a PET film at 95C uniaxially at 30
O/O/min-
I
(0.5s-
I
). They observed a 15C increase in temperature of a sample
within 3 seconds of deformation. Warner4
5
also found similar results in
uniaxial stretching of PET and added that this temperature rise is due to two
factors. One is plastic deformation associated with neck formation and
propagation, and the second is due to crystallisation.
16
Cakmak, White and Spruiell measured the blowing times for a SBM process
of between 0.7 and 4 seconds giving deformation rates between 250s
1
and
616s
1
, 46, which is high enough to give substantial heating due the
deformation. Consequently, the drawing temperature is higher than the
surrounding enviromnent. The extreme case, with no heat conduction to the
enviromnent, occurs at very high strain rates (found in stretch blow
moulding) and can be referred to as adiabatic drawing. This increase in
temperature could either allow more relaxation in the amorphous phase, or
increase the degree of crystallinity of the drawn sample.
The models of microfibrillar morphology proposed by Peterlin for
polyolefms47 and Prevorsek for polyesters
37
provide a good explanation of
mechanical properties, although the models are very different. Peterlin
reasoned that the fibre strength is mainly attributed to the strength of the
microfibrils, while the interfibrillar domains did not contribute a great deal
to fibre strength. According to Prevorsek et a[48, 49 the case for Nylon and
PET fibres is quite different. The extended-chain interfibrillar domains are
the strongest element of the fibre structure, and have an important effect on
fibre strength. The increase in fibre strength on drawing is an effect of
increasing the amount of the extended-chain molecules which are formed as
a result of the relative displacement of the microfibrils. According to this
model, the main function of the microfibrils is to provide dimensional
stability at elevated temperatures, rather than contributing significantly to
fibre strength. Pervorsek does make the point that the disagreement
between the two models is probably due to the fact that Peterlin's study
investigated the properties of polythene and Polypropylene, rather than PET
or Nylon. Pervorsek also found that the crystallites in a low draw ratio fibre
(times 3) are shaped like platelets, oriented perpendicular to the fibre axis,
whilst for a higher draw ratio, the crystallites approach the shape of cubic
particles. Although there is some conflict here with X-ray Analysis, both
methods (mechanical and X-ray) indicate an increase in aspect ratio with
increased draw ratio.
17
Tanka, Nagano and Onogi
50
found a narrow crystallisation temperature and
draw range in which during drawing the spherulitic structure is not
disrupted but the spherulites just move apart. They also found that the
dynamic viscoelastic and optical properties for films with either very low or
high levels of crystallinity were non-linear, while those with an intermediate
level were linear. They explained the non linearity for films with very low
crystallinity by the fact that amorphous chains slip past one another due to a
lack of crosslinking (tie) points. The nonlinearity for films having high
crystallinity is caused by the disruption of the spherulitic structure and the
linearity for films with an intermediate level indicates only one deformation
mechanism is operating; that is, the orientation of the amorphous phase
which is a linear function of strain.
Thompson
41
used simply the lacing order of a filament-drawing machine to
produce amorphous, slightly oriented or strain induced crystalline oriented
fibres. TEM studies of melt spun PET fibres by Petermann and Rieck
51
showed that:
a) As-drawn PET exhibits a micellar structure.
b) Annealing below 150C increases crystallite Size, orientation and
degree of crystallinity, but the micellar structure is maintained.
c) On annealing above 180C the amorphous regions between the side
faces of the micellar blocks begin to crystallise, large crystalline
aggregates are formed but the micellar structure is retained.
Drawing at temperatures below Tg enforces a fringed micelle crystallisation
of the cold oriented chains. Increasing the draw temperature activates the
relaxation and disentanglement of the amorphous chains during the drawing
process
52
, 53. The molecules then undergo chain-folded crystallisation
during the draw process to give rise to imperfect, lamellar crystals, which
are liable to reorganise on reheating.
18
Hennessey and Spartoric0
54
using a TM Long Stretcher to draw branched
PET films, found that there was no correlation between degree of
crystallinity and the maximum stress developed during the stretching
operation. They also found the development of crystallinity in samples
drawn to various extents was not a strong function of strain rate at least in
the range of 54-265s
l
. Koenig and Mele
55
reported that there was no
significant increase in crystallinity above the value of an undrawn sample
until the sample, was drawn by more than 160%.
Most studies of PET consider it to be a two phase system i.e. amorphous
and crystalline. However, Farrow and Ward
56
and Fischer and Fakirov
57
postulated the existence of a third phase. Stem
58
and Linder
59
, amongst
others, found evidence for a smectic mesophase from XRD and density
measurements. Linder did conclude that for amorphous or partially
crystalline PET, for fibres at least, the two phase model is a useful
approximation. In contrast to this Pervorsek et a1
48
49 indicated that with
Nylon and PET fibres, the volume fraction of the extended interfibrillar tie
molecules is so significant that they should be considered as a separate
phase.
A more mathematical approach to deformation is provided by considering
deformation to be either affme or pseudo-affme. At slow rates, the hot
primary stretched film is nearly identical to the cast film. This corresponds
to rubbery behaviour which correlates to an "affme" deformation scheme
which induces very little alignment of the molecules. Conversely, for fast
primary stretching, the deformation follows closely the "pseudo-affme"
mechanism which induces both chain orientation and crystallisation. These
modes of deformations are the same as identified by Foster, Heap40 and
Thompson
41
. However, once the deformation mode has been related to a
mathematical system then the mathematics may be applied to the real
system. Jungnickel
60
has developed this concept, but this lies outside this
reVIew.
19
PET can exist in either of two rotational isomeric forms (trans or gauche) of
the ethylene glycol linkage. The trans isomer is the extended form relative
to the gauche and occurs in both the crystalline and amorphous regions of
the polymer; the gauche isomer, however, occurs only in the amorphous
regions. A film with a given amount of crystallinity may have varying
amounts of trans isomer and a film with a given amount of trans isomer may
have varying amounts of crystallinity. Cast (randomly oriented) amorphous
PET contains a fixed amount of trans-gauche structure. In Heffelfmger and
Schmidt's61 films this was about 13% trans structure and 87% gauche.
Since the trans structure is an extended form, this indicated that the majority
of the molecules are in the relaxed unextended state, as would be expected.
On uniaxial drawing, the molecules become aligned in the stretch direction
and the individual chain segments become extended.
2.3.1 Crystallisation
Unoriented PET will crystallise thermally into spherulitic form throughout
the temperature range between Tg and Tm, but is most rapid between 160C
and 180C ie. midway between Tg and Tm. The PET crystal has a triclinic
unit cell with the c-axis usually inclined at some 6 to the direction of draw
in a fibre
62
. It is known that whilst the amount of orientation will not effect
the amount of crystallinity at equilibrium, greater orientation will lead to
higher crystallisation rates, especially in the early stages of a post draw
anneal.
As already discussed, drawing PET above 150% will lower the entropy of
the polymer by aligning the amorphous chains, thereby creating conditions
that favour strain-induced crystallisation. Cakmak and Wang
63
found that
for biaxial draws of 2 by 2, increasing the draw temperature decreased the
amount of crystallinity formed. This indicates that increasing the draw
temperature reduces the effectiveness of the draw on crystallisation. At a
draw of 3 by 3, however, they found a similar increase in draw temperature
had no effect on induced levels of crystallisation.
20
The kinetics of strain induced crystallisation, which is much faster than
isothennal crystallisation of isotropic samples, are controlled by the initial
orientation of the sample and the annealing temperature. At 120C the half-
time for spherulitic crystallisation is a few minutes but it becomes lower
than O.Ols for strain-induced crystallisation
64
The kinetics of strain
induced crystallisation have been studied by Bouvellec et at6
5
They
reasoned that if crystallisation was very slow compared to the drawing
speed there would be insufficient time for crystallisation to occur, so that no
crystallinity would be detected. As this is not the case
66
, (demonstrated by
strain rate influencing crystallisation rate), then each of these rates must be
comparable. For a given temperature, the kinetics of crystallisation during
stretching are governed by the evolution of the amorphous orientation.
Thereby, crystallisation increases with strain rate. The morphology of
strain-induced crystallinity is of the chain extended type and as such, is far
fmer than its spherulitic counterpart. This accounts for the transparency of
the chain extended morphology as the structure is too fme to act as light
scattering centres.
Most authors agree that melt spun and drawn PE, PP, Nylon and PET fibres
have three phases: amorphous and crystalline domains, both within a
microfibril, and interfibrillar matter. Keller proposed that the microfibril
can be divided into two classes
67
:
Class I comprises molecules that have been synthesised in the chain-
extended fonn, such as biological fibres, crystals fonned by in situ
polymerisation and structures resulting from stress and flow-induced
crystallisation. When these micro fibrils are heated close to the melting
point the dimensions of the sample remain essentially unchanged.
Class 2 microfibrils comprise products of solid state and melt drawing, in
which the original crystalline or amorphous structure is destroyed, and the
chains are aligned in the direction of the orienting force. The fibrils of
class 2 can shrink drastically when heated close to the melting point
21
Prevorsek et af3
7
demonstrated that at temperatures considerably below the
melting point, the shrinkage of fibres is caused primarily by the contraction
of highly extended interfibrillar tie molecules. In this process the
longitudinal structure of the microfibrils remains essentially unchanged,
because the contraction proceeds via lateral displacements of the
microfibrils.
The extended-chain interfibrillar matter formed at high and intermediate
draw ratios, therefore, belongs to class 2. However, the microfibrils
themselves, most likely, belong to class 1. This would indicate that the
microfibrils are products of stress-induced crystallisation. So their
crystalline domains consist of elongated molecules produced by
elongational flow. The microfibrils in Nylon and PET consist of a sequence
of crystalline and amorphous domains whose, dimensions along the fibre
axis are sufficiently regular to act as a microlattice that can give rise to X-
ray scattering. An important characteristic of a microfibril is the very long
period, which represents the spacing between two adjacent crystallites in the
microfibriP7.
Gupte, Motz, and Schultz
68
suggested a two stage transformation process
for PET fibres crystallising with fixed ends. The first stage is the formation
of defective fibrils, and the second is the internal rearrangement of the
fibrils to form more perfect crystals. The crystallisation model of Tiller and
Schultz6
9
for polymers under high tension proposes dendritic growth. Their
model assumes the incorporation of large concentrations of defects into the
crystal filaments. These defects are rearranged in the second stage of
transformation.
22
BosleyIO reviewed three models for the arrangement of individual molecular
segments within crystallites:
(1) The fringe-micelle model considers each molecule to thread its way
through several adjacent crystalline and noncrystalline (fringe)
reglOns.
(2) The paracrystalline model has the molecular segments of atomic
groups occupying crystalline lattice points but with varying degrees
of disorder possible within the lattice.
(3) The lamellar model, in which individual molecules are considered to
fold back on themselves with regularity and form lamellae that are
much thinner in the direction of the chain than in the length of the
chain.
Bosley makes the point that a chain folded lamellar structure cannot shrink.
The fact that SAXS measurements on PET fibres wound at high speed do
not show any diffracted intensity, indicates that the crystallites formed
during spinning are not predominantly of the folded chain type
70
Bosley
made the point that criticisms levelled at the fringed-micelle model are
artificial. Contrary to many authors who describe the model as a rigid inert
crystallite embedded in a rubbery mobile matrix, he felt the model considers
the amorphous and crystalline regions coexist and interact and, in fact, defy
separation. It is interesting to note that Prevorsek et al
37
found that, in cases
they investigated, the structural model consisting of crystallites embedded in
an amorphous matrix (an inaccurate view of the fringed-micelle according
to Bosley) yielded better results than a microfibrillar model, assuming
relatively poor lateral interactions between the microfibrils. This led them
to conclude that the fibre properties do not depend only on the degree of
crystallinity but also on the dimensions and spacings between the
crystallites.
23
2.3.2 Orientation
The concept of "strength through orientation", developed in the mid thirties
is fundamental to modern thinking in plastics artefact design. Orientation is
developed in amorphous polymers by extension in the rubbery state and
may be frozen-in by rapid cooling to below Tg. In general, the molecular
mobility in the glassy state is insufficient for disorientation to occur within
a measurable span of time, even though the chains have an inherent
tendency to revert to the unoriented, most probable, random state. An
already old defmition
71
72. 73 of orientation is the following: A polymer is
called unoriented when its chain segments are randomly oriented in space;
it is called oriented when the spatial distribution is not random but shows
directions of preference. This definition is essentially based on the
anisotropy of the chain segments. If the segments did not have different
properties in different directions, orientation would not exist (spheres
cannot be oriented). Therefore, this defmition is purely geometric. It only
takes into account the distribution of the chain segments in space at a given
moment, completely ignoring the mechanism of orientation. This defmition
can then also cover streaming melts and deformed rubbers. In both of these
the orientation is only preserved by an external mechanical stress.
Struik
74
comments that molecular orientation should not be confused with
mechanical strains or frozen-in-deformations. Orientation is generally the
result of deformation, and frozen-in orientations are generally accompanied
by frozen-in strains. However, the degree of orientation is not simply
determined by the degree of deformation. Following rubber elasticity
theory, for given level of deformation, molecular orientation will increase
with cross link density. Cross link density may increase with decreasing
temperature and increasing strain rate.
24
Because a moulded PET preform is both shorter and narrower than the fmal
shape, (the bottle), it must undergo biaxial deformation to achieve this. In
their introduction to a paper concerned with biaxially oriented PET films
Vallat and Plazek
75
made some useful systematic statements which are
paraphrased here for brevity. "Biaxial orientation of polymer films offers
strength in all planar directions, which is highly desirable for many
applications. Such highly oriented films are extreme examples of non-
equilibrium structures. The orientation and enhanced strength are preserved
by:
(1) the molecular immobility below the glass transition temperature;
(2) the presence of an effective molecular network where crystallites
formed in the oriented state restrain non cryStalline polymer chain
segments from diffusing to random conformations;
(3) both the preceding.
Biaxially oriented polystyrene film is an example of the first kind, and
shrink tubing made from polythene or PTFE are examples of the second
kind. Biaxially oriented PET, which is one of the most dimensionally stable
films available, is an example of the third kind". In PET carbonated soft
drinks (CSD) containers, it is only the successful development of biaxial
orientation in the part that makes this type of container viable, not only in
terms of strength and creep resistance, but also gas barrier properties.
25
2.3.2.1 Orientation Functions
Once the importance of orientation had been established it was then
necessary to quantify it. In the 1930s and 40s Hermans
76
, 77, 78 used the
anisotropy of polarisability to specify an orientation function, i.e.:
where a! and a2 are the mean polarisabilities in the axial and radial
directions and tlaO is the difference in polarisability perpendicular to the
polymer chain.
Hermans et at then went on to show that this was equivalent to:
fH = 3cos
2
<pc! -1
2
where <Pc 1 is the angle between the fibre axis and the polymer chain axis.
In 1941 Muller
79
realised that of fundamental importance was not simply
the cos
2
<pc 1 term but an orientation distribution function about a preferred
axis and expressed this in terms of an expression of even spherical harmonic
functions, Pr (COS<Pcl). This view was also adopted by Ward and
McBrietyS.
SteinS! generalised Herman's orientation. functions to represent the
orientation of the three crystallographic axes in crystalline polymers:
f. = 3cos
2
<p.! -1
2
26
<Pal, <Pa2 and <Pel are the angles between the a, b and c crystallographic axes
and the fibre axis.
For crystals having orthorhombic symmetry:
Classically, biaxial crystallite orientation has been represented by the use of
pole figures. These are stereographic projections showing the spatial
distribution of the density of "poles" or plane normals for a given set of
planes. Stein represented biaxial orientation not with pole figures but
though orientation functions. Alternative methods of achieving this were
proposed by various authors, the most important of which was proposed by
White and Spruie1l
82
. Orientation is defmed in terms of the angles <Pc! and
<Pe2 between the polymer chains and the machine, 1 and the transverse, 2,
direction. They proceeded by generalising Herman's arguments on the
anisotropy of the polarisability tensor. For amorphous polymers:
f B al-a3 2 2 + 2
le =--= cos<Pel cOS<Pe2-1
!J.ao
f B ~ a 3 +
2
= --= 2cOS
2
cne2 cOS
2
cnel_l
e !J.a
o
'Y 'Y
for uniaxial extension a2=a3
from Pythagorean theory:
f B
le
For isotropy fle
B
=f2c
B
= 0
For polymer chains perpendicular to the 1,2 plane fle
B
=f2c
B
= 1
For equal biaxial orientation f
lc
B
= f2e
B
27
Therefore all possible states of biaxial orientation may be presented
graphically as an isosceles triangle with its vertex at (-1,-1) on a plot off
le
B
Vs f2e
B
.
(0,1)
fB
I
UNIAXIAL (MACIflNE DIRECTION)
PLANAR
EQUAL BIAXIAL
UNIAXIAL
____________
(1,0)
IsomOPIC
(-1,-1)
fB
2
Diagram of White-Sprniell Orientation Function Triangle, where the y-axis
is the machine direction which is designated fl and the x-axis, the
transverse direction is referred to as f ,.
28
To generalise, for the jth crystallographic axis:
where </lij is the angle between the crystallographic axis j and the reference
direction i.
Gupta and Ramesh
83
, studying the development of orientation during the
drawing of PET using polarised FTIR, ascribed the 973cm-
i
and the
1578cm-
i
peaks to transitions occurring parallel to the PET chain axis, but
in the crystalline and amorphous phases respectively. They reported their
work not as orientation functions, but rather as dichroic ratios. The dichroic
ratio was defmed as:
where
and
_ (AIB) -I
OF - (AIB)+2
A = IR absorbance measured parallel to the major axis
B = IR absorbance measured perpendicular to the major axis
2.3.2.2 Evolution, Classification and Effects of Orientation
Kashiwagi, Cunningham, Manuel and Ward
84
, using a combination of
nuclear magnetic resonance, optical and X-ray diffraction, concluded that
for uniaxially drawn films the optical anisotropy is governed essentially by
two types of molecular response. In the low draw ratio region the
orthorhombic distribution of chain axes alone is sufficient to characterise
the optical anisotropy, whereas in the high draw ratio region, preferred
planar orientation must also be taken into account.
29
'-----------------------------------------
Ma and Han
8s
commented that the structure of biaxially stretched,
semicIYstalline polymer films can be characterised by studying
(1) the orientation distribution of the cIYstalline Phase,
(2) the orientation distribution of the amorphous phase,
(3) the degree of cIYstallinity.
This is somewhat in contradiction to Prevorsek et al
37
who found for fibres
that the mechanical properties amongst others, do not only depend on the
degree of cIYstallinity, but also on the dimensions and spacings between the
cIYstallites. J abarin 86 concluded that the degree of molecular orientation
and the physical properties of oriented PET are dependent on the strain rate,
in addition to extension ratio, molecular weight and temperature of
orientation and that the influence of the strain rate on the degree of
orientation is extremely significant. He also found that the mechanical and
transport properties of oriented PET were directly related to the degree of
orientation, as measured by birefringence. Ma and HangS also reported a
clear correlation between the amorphous orientation functions (White-
Spruiell type) and the tensile mechanical properties of PET blown film.
They also reported that the cIYstalline orientation functions were relatively
insensitive to stretch ratios for blown films which had been heat-set. For
biaxial orientation, Jabarin
86
found that if the draw was balanced then the
out-of-plane birefringence of the sample increased steadily with the draw
ratio. For an unbalanced draw, the greatest out-of-plane birefringence value
was in the direction of the greatest stretch and that this anisotropy in
birefringence was also reflected in some physical properties such as
mechanical properties (yield stress, modulus, UTS or impact strength). De
Vries et al 23 went further, to report that for unbalanced biaxially drawn
films, the difference in properties in the two, in-plane perpendicular
directions in the film, were (to a first approximation) a linear function of the
difference in the refractive indices in the same directions.
30
For films produced by a sequential biaxial draw, the second draw (referred
to as the transverse draw) causes the chain segments in the amorphous
regions and the crystallites to be shifted from the machine direction towards
the transverse direction. The extent of this movement depends on the
stretch ratio, the temperature and on other process conditions
61
The
movement of the crystallites means that their coaxes move from the machine
direction towards the transverse. Another consequence of biaxial drawing
is that the (100) planes are forced to become more parallel with the surface
of the film. Heffelfinger and Schmidt6
1
also comment that the structural
changes, which take place during the transverse draw, are primarily
associated with the realignment of the structure that has been produced by
the stretch in the machine direction.
Heffelfmger and Burton
s7
modified and extended a set of definitions
produced by Sisson
ss
to give a classification system of crystallite orientation
types, the names of which are derived from the geometry of the reference
system and not the geometry of the crystallite. For example, planar
orientation is defmed in terms of a structural operator (in this case a crystal
axis) referred to a reference plane. The six types of orientation in this
classification system are given below:
Random
Planar
Uniplanar
Axial
In this structure no preferred direction of either crystal
axes or planes exists with respect to reference axes or
planes.
A crystal axis is oriented parallel to a reference plane
from which the name of the structure is derived. With
polymer films, the reference plane mayor may not be
parallel to the plane of the film. This is the least
restrictive type of orientation.
In this type of orientation, a crystal plane is oriented
parallel to a reference plane. This also satisfies the
planar defmition but is a more restrictive case.
In axial orientation, a crystal axis is oriented parallel to
a reference axis, which is usually (but not necessarily)
the direction of stretch or rolling.
31
Plan-Axial In this structure, a crystal plane is oriented parallel to a
reference axis. No restrictions are imposed on the
direction of any crystal axis. This is more easily
visualised as given crystal planes parallel to a reference
axis forming a cylindrical envelope. This type of
orientation is not applicable to film geometry but can
exist in fibres.
Uniplanar-Axial A crystal plane is oriented parallel to a reference plane
and a crystal axis is oriented parallel to a reference
axis. This structure cannot be further ordered except
by decreasing the number of crystallites by joining
them into larger ones. This type of orientation is the
most restrictive.
A clearer explanation of this classification system is given by the set of
diagrams in the reference.
Heffelfmger and Burton
87
make the point that in polymers, orientation is not
complete and distributions of directions of planes or axes will exist and it is
also possible for several orientation types to coexist. For uniaxially drawn
PET they found the (100) planes to be essentially parallel to the surface of
the film and the coaxes oriented parallel to the direction of the stretch.
Using their definitions, this highly-oriented uniaxially-drawn PET sample
had a uniplanar-axial type of orientation. A small fraction of the crystallites
remain at the 90
0
to the stretch direction position, indicating that these
crystallites are extremely difficult to move. Imposing a second, transverse
stretch to give sequential biaxial deformation caused the orientation of the
film to change from uniplanar-axial to uniplanar. They also reported an
increase in parallelism of the (100) plane to the film surface with the second
stretch.
32
2.4 HEAT-SETTING
2.4.1 Heat-Setting as a Commercial Process
The above discussions (sections 2.1 to 2.3) have outlined the achievement
of a PET bottle's structure and some properties, as produced by the stretch
blow moulding route. These properties are seriously diminished as soon as
the temperature approaches and surpasses the material's Tg and therefore
this imposes the upper permissible service temperature. Above this
temperature, the oriented material often undergoes substantial shrinkage.
Heat-setting is a process that has been used to reduce or even eliminate
thennally induced shrinkage in oriented PET systems 89, initially for fibres
and subsequently for films. Constrained annealing of biaxially oriented
plasticised PVC above 90C can also prevent dimensional recovery when
the sample is released
9o
. Because many foodstuffs are required to be hot
filled in the temperature range 85C to 110C, there is considerable
commercial interest in producing a PET container. that can withstand these
temperatures. Heat treatment of twenty seconds or less in the temperature
range between 160C and 220C is sufficient to yield hot-fillable
containers
55
. Although the oriented crystalline regions remain transparent
after heat-setting, amorphous or slightly oriented areas develop spherulitic
crystallites and turn opaque. Container designs including ribbed
reinforcements and flexible panels to resist defonnation when subjected to
the vacuum created on cooling after hot filling have been in existence since
1985. Despite this, the majority of research on the heat-setting of PET has
been conducted on fibres and films.
Heat-setting, under ideal conditions, can be described as an isometric,
isothennal annealing process. It is the necessity to obtain rapid heating and
cooling, whilst preventing the sample from shrinking that makes systematic
experimentation with more complex shapes such as bottles technically more
difficult. Heat-setting is known to increase birefringence
8
, 85 and this is
generally assumed to be a result of improved degree of crystallinity and
structural perfection. Amorphous orientation functions have also been
shown to increase with heat-setting of fibres 91 although the author
commented that this result was rather surprising and conflicted with
conventional opinion.
33
2.4.2 Mechanics of Heat-Setting
The mechanics of heat-setting are still the subject of much debate with most
authors falling into one of two general opinions:
The first position that had, until fairly recently, gained widespread
acceptance, was that heat-setting should be viewed as two competing
processes, meltinglrecrystallisation and disorientation of the amorphous
phase. The evolution of strain induced crystallinity during the drawing
stage has a profound effect on the kinetics of crystallisation in the heat-
setting stage. For example, at 120C the half-time of isothermal
crystallisation is a few minutes
92
, but it becomes less than O.Ols for strain
induced crystallisation
64
and so should be considered as a thermally
activated event. The envisaged mechanism is the melting of the small
defective crystallites, produced during the draw and the rapid
recrystallisation of more perfect and larger crystallites, that should be stable
to just below their formation temperature. Working with drawn fibres and
annealing times between lOms and lOOOms, using dead weights to partially
constrain the fibres during treatment, Peszkin, Schultz and Lin
s
produced
very strong evidence in support of this theory. On reheating, the majority of
their samples did demonstrate shrinkage but some displayed "negative
shrinkage" (expansion), which has also been reported by Biangardi and
Zachmann
9
W AXS patterns for the samples which did shrink during
annealing, showed the initial crystallinity is completely lost within the first
lOms of the heat treatment. For the samples that exhibited negative
shrinkage, the initial crystallinity is not lost, but the crystalline diffraction
peaks become more intense and better defmed. They also found that
maxima in shrinkage curves and minima in birefringence curves coincided
with the beginning of the development of the microstructural features,
which give rise to SAXS and W AXS. Thus they concluded that the
decrease in shrinkage and the increase in birefringence are due to
crystallisation effects. For samples with short annealing times, the
increased shrinkage and decreased birefringence was attributed to the
entropy-driven re-coiling of the chains.
34
Once crystallisation begins and the chains are anchored in the crystals, long
range motion and further shrinkage are hindered. At higher annealing
temperatures (where elongation occurred) crystallisation took place very
early, immediately locking the chains and so preventing re-coiling. The
effect of applied tension was to increase the crystallisation kinetics. XRD
studies of biaxially oriented and annealed plasticised PVC, suggested to
Gilbert and Lui
90
that the formation of small crystallites locked the oriented
chains into position preventing shrinkage. This is in close agreement with
the argument above for PET.
Cakrnak, Spruiell, White and Lin 93 biaxially drew and heat-set PET fihns.
The SAXS patterns from these led them to conclude that fixed annealing
does not cause a major reorganisation of the structure, but merely perfects
the structures developed, primarily by stress-induced crystallisation, during
stretching. This actually supports Peszkin et al
8
as Cakrnak could not have
seen the effects reported by Peszkin as Cakrnak's fihns were heat-set for 10
minutes prior to taking the SAXS patterns. Further evidence that heat-
setting is a process of reorganisation is inferred from the double melting
peaks displayed by heat-set PET. Gupta, Ramesh, and Gupta
94
suggested
that the interpretation, that the double melting peak is a consequence of the
original structure under going reorganisation during the scan, is the accepted
interpretation. The primary endotherrnic peak is ascribed to the melting of
the crystallites formed during heat-setting and the secondary peak is due to
the melting of the less stable fraction, which undergoes reorganisation.
Further circumstantial evidence for this interpretation is provided by Leu
and McCarthy95 and by Haworth and Davidson
96
Both publications report
a close correlation between the peak temperature of the secondary
endotherm and the heat-setting temperature.
The second view on the mechanism of heat-setting of PET broadly suggests
that it is not a process of recrystallisation. Buckley and Salem
25
found for
drawn fibres with a small (0.006) heat-set torsional strain, that heat-setting
time, over a range of many decades of seconds, had as much an effect on
fractional recovery as heat-setting temperature. This time scale is in marked
contrast to those described by Sun and Pereira
64
.
35
Buckley fIrstly discounted the possibilities that heat-setting was associated
with yielding (as in metal forming), or due to hydrolytic attack on the PET
molecules. Due to the time scales involved, Buckley reasoned that if heat-
setting occurred by meltinglrecrystallisation, it would be temperature
triggered; time would only be needed for heat transfer. As the fractional
recovery decreases with heat-setting time up to periods of 24 hours and that
temperature induces the acceleration of the process, they found these to be
characteristic of a viscous flow process. The double melting peaks found in
DSC scans were attributed to non-melting, mechanical relaxation events
such as Tg. Alternatively, they may reflect the time dependent response of
the specimen's enthalpy to the constant heating rate imposed in the DSC
test. This, however, fails to explain the correlation of the secondary peak
with the heat-setting temperature. Buckley does point out that they do not
rule out the possibility that partial melting occurs on re-heating to the region
of the heat-setting temperature, but it simply does not make a signifIcant
contribution to the thermal stability achieved. Buckley and Salem went on
to extend this work to higher setting strains, both torsional and bending,
with similar results
97
.
In direct contradiction to the work of Gupta et af94, further evidence that
meltinglrecrystallisation is not the mechanism of heat-setting is provided by
the DSC analysis, of biaxially drawn and heat-set PET fIlms, as performed
by Vallat and Plazek
7s
Rather than considering the two melting peaks to
have separate origins, they considered them to be the remnants of the usual,
single melting peak that had been split by a superimposed exothermic
process. They assumed that this was due to the exothermic release of high-
energy molecular chain conformations which were formed during the
stretch. They compared the latent heats of fusion of as-drawn and heat-set
PET fIlms and found the as-drawn to be greater. The heat-set fIlm could not
have a lower degree of crystallinity than the drawn fIlm. Hence, the smaller
apparent latent heat of fusion for the heat-set fIlm is evidence against
arguments which invoke melting and recrystallisation to explain the double
melting peaks. Sun and Magill
30
extended this argument to account for
double melting peaks in unheat-set, oriented PET by concluding that
shrinkage in oriented systems (crystalline or amorphous) is exothermic and
that it is responsible for the multiple melting peaks observed in DSC scans.
36
Sun and Magill also reported double melting peaks in unoriented PET
samples, which they interpreted as evidence for two morphologies present
in those samples. This is not common for PET and perhaps reflects on the
validity of some of the conclusions. Even so, this explanation is quite an
attractive one for the case when the secondary endotherm is close to, or is a
shoulder of, the primary endotherm. However, discrete secondary
endotherms can have peak temperatures as low as 115C, for heat-set
oriented PET, which is over 140C lower than the primary peak
7
. After the
secondary peak has occurred, the trace returns to the base line. For an
exothermic reaction to account for this, a continuous endotherm starting at
the onset temperature of the secondary endotherm, with a magnitude that
balances the energy of the exotherm for the temperature range where the
trace returns to the base line, would be required. Sun and Magill's work
was concerned with oriented but unheat-set material, but Vallet and Plazek's
material was heat-set and they reached similar conclusions about the cause
of the secondary melting peak. Unfortunately, their chosen heat-setting
conditions of 190C for 5 minutes did not give rise to secondary
endothermic peaks below 240C. Therefore, the difficulties this theory
would have with lower temperature secondary peaks were not discussed.
2.4.3 Heat-Setting Structural and Mechanical Considerations
As previously mentioned, PET chain structure, can exist in two rotational
forms of which only the trans form is capable of crystallising. The effect of
heat-setting on the relative amounts of the two rotational conformers was
studied by Heffelfmger and Schmidt6
1
During the heat-setting of near-
balanced biaxial films, the total trans content increased in value, reflecting
the increase in the crystalline (trans) content. In contrast to highly
uniaxially strained structures, the molecular axial order remains essentially
unchanged, whilst the amount of molecular uniplanar structure increases.
The authors felt that the greater mobility of the biaxially stretched structure
is such that the near-neighbour gauche segments, having acquired sufficient
thermal energy to isomerise, easily shift into trans segments which are able
to crystallise.
37
Heat-setting can also affect the mechanical properties of the material. Ma
and Han
85
reported a significant increase in the stress at break for oriented
and heat-set samples, compared with oriented samples. Similarly, Cakmak
and Wang
63
found heat-setting to decrease creep strain but this effect was
most pronounced in undrawn material. They attributed this difference to
drawn materials (balanced biaxial draw of 9) possessing high chain
orientation and crystallinity levels. Hence, the increase of crystallinity in
initially amorphous and unoriented films is more substantial than in the
stretched films. In unstretched films, the benefit of chain orientation on
creep behaviour is absent and only the crystallinity effects are observed.
The unoriented samples were heat-set at 170C for between 1 and 720
minutes. Crystallinity measurements indicated that 70% of the achievable
degree of crystallinity is developed within the first minute under these
conditions, and so the effect of draw on the crystallisation kinetics had been
negated.
2.5 CHARACTERISATION TECHNIQUES AND MATERIAL
PROPERTIES
2.5.1 Shrinkage
Methods to quantify thennally induced shrinkage in oriented polymers
broadly fall into one of two types. The first is a 'before and after' type
where the dimension of interest (usually length, but this can be degree of
twist for torsion experiments) of the specimen is recorded prior to heating.
After treatment the sample is cooled, then remeasured. The second group of
techniques are those which continuously monitor the dimension during the
shrinkage process. These two types of techniques will yield different
results, because in the first case, all the measurements are made at the same
temperature whereas in the second case, the temperature can be separated
by over 200C (between Tm and room temperature) and the material's
coefficient of thennal expansion will become significant. Once this is
realised, it may be simply corrected for to make the results of the two types
of techniques comparable. Examples of the first type of measurement were
reported by Long and Ward
20
who used heated silicone oil to heat-set drawn
PET fibres and by Jorden, Juska and Harrison
35
, who perfonned a similar
experiment but on a microscopic scale.
38
Long and Ward reported maximum shrinkage to occur in samples with a
draw of 2.5, which was reduced by over a factor of three when the draw
was increased beyond 4. The second type of measurement can be seen in
the work of Husiue and Yeh
5
and that of Haworth, Dong and Davidson
7
in
which practical considerations of the use of a TMA are explored.
Petermann and Rieck
51
also used TMA to measure shrinkage of oriented
and annealed PET samples but failed to report the test conditions. This is
unfortunate as they report an increase in shrinkage with annealing parallel
to the direction of draw, coupled with the complete elimination of shrinkage
in the perpendicular direction. The increase in shrinkage with annealing is
completely contrary to expectation and in-fact would mean heat-setting
would increase shrinkage, rather than control it.
Shrinkage stress, or retractive stress has also been measured,18, 20, 22, 23.
This is a measure of force or stress developed at constant sample length,
using a suitable force transducer. This is usually performed in a tensile
testing machine
l8
. It has often been assumed that maximum shrinkage
stress is identical to the internal entropic stress
21
, 98, 17 and so its value
should be directly related to orientation birefringence. However De Vries
and Bonnebat
23
failed to find a unique, linear relationship for these,
especially if the samples had been oriented at high strain rates or at low
temperatures, followed by a rapid quench to below Tg.
2.5.2 Mechanical Testing and Properties
The testing of mechanical properties in tension is a fairly standard technique
and reference to the appropriate national or international standard e.g.
99
furnishes the reader with the relevant experimental considerations.
It was noted by Samuels
loo
that the tensile test is in fact a uniaxial extension
experiment performed at relatively low temperature and it may be expected
that if the strain rate is low enough further molecular orientation will take
place. Therefore, the structure at rupture may be more or less independent
of the initial structure.
39
De Vries Bonnebat and Beautemps23 analysed previously published data 100,
101 and their own data on amorphous and crystalline polymers and
confirmed that often, the true ultimate tensile stress varies little with the
initial state of uniaxial orientation, unless the strain rate in the test is very
high, or if the test temperature is very low. Exceptions to this are found for
polystyrene
lOl
, 102 and pVCIOI.
Prevorsek Kwon and Sharma
37
found that a highly drawn fibre (draw ratio
of 5.35) had about double the strength of a low draw ratio fibre (draw ratio
of 1) which was in close agreement with their calculations. They accounted
for the increase in strength on drawing from:
(1) the increase of the volume fraction of the intembrillar extended-
chain domain;
(2) the decrease in stress concentration factor at the tip of the rnicrocrack
which is associated with the decrease in the crystallite width and the
distance between the crystallites.
Creep studies on PET have been made by many authors
63
, 103, 104, 105, 106, 107,
108 Cakmak and Wang
63
found that the increase of crystallinity and
molecular orientation, induced by biaxial stretching results in the decrease
of creep strain. The effect of heat-setting on creep strain, according to these
authors, has been previously mentioned in this section.
2.5.3 Orientation Measurement
2.5.3.1 Birefringence
The presence of orientation and/or residual stress will alter the refractive
index of the material, according to the direction of light propagation and the
orientation of the plane of polarisation. Birefringence is defmed as the
difference between the refractive indices in two orthogonal axes. However,
it is more usual to measure birefringence directly using one of a range of
techniques. Birefringence can arise as a result of anyone of three effects,
namely orientation, stress or form birefringence.
40
(A) Orientation Birefringence
A mediwn alters the propagation of light due to the interaction between the
electric vector of the light and the bonding electrons; a property expressed
by polarisability. The magnitude of the interaction depends on the angle
between the electric vector and the bond axis and is a maximwn when these
are parallel, giving the highest refractive index. Therefore, molecular
orientation will cause a material to be birefringent. The magnitude of the
birefringence will also depend on the nature of the bonds e.g. a double bond
has a much greater polarisability than a single bond. This is referred to as
orientation birefringence
(B) Stress Birefringence
When a bond is stretched or if the bond angle is altered by an applied stress,
the polarisability will change, causing a birefringent effect. This is
sometimes referred to as deformation birefringence, which can be
misleading when referring to polymers, since orientation is often produced
by deformation. The accepted term for this effect is stress birefringence.
(C) Form Birefringence
This may occur in multiphase materials with phases of different
birefringence, and is a result of distortion of the electromagnetic field at the
phase interfaces. Parallel rods, lamellae etc. of one phase within a matrix of
differing refractive index would provide a suitable morphology for this to
occur. The effect is at a maxirnwn when the spacing of these features is
about 'JJ20 (1..= wavelength oflight). Block copolymers quite often separate
on this scale and if they are anisotropic (e.g. through processing), they are
quite likely to show some form birefringence. Birefringence effects caused
by such circwnstances are referred to as form birefringence
109

41
Birefringence has proved to be a veI)' useful tool for examining polymer
structure. However, to correlate mechanical properties with birefringence,
the intrinsic or maximum orientation birefringence of the amorphous and
cI)'stalline regions must be known. Various authors have calculated values
from different approaches, for PET ranging from ~ (the birefringence of
perfectly aligned amorphous phase) from 0.212 to 0.22 and 8Nc ( the
birefringence of a single cI)'stal) from 0.213 to 0.275. Dumbleton liD used
the two phase model proposed by Samuels
lOo
to calculate ~ c at 0.22 and
8Na of 0.275, figures which have been widely used by many authors.
However Konda, Noze and Ishikawa
1l1
used Nc = Na = 0.212. This lack of
agreement for an intrinsic birefringence of PET prompted Gupta and
Kumar
ll2
to conclude that if the Samuels model is applicable then the
values obtained previously will need re examination, and if not, then values
calculated via this theoI)' will have little or no physical meaning. Further,
they also commented that intrinsic birefringence depends on constant lattice
parameters, and as these will change with heat-setting time and temperature,
systematic variation in these values is possible. However the relevance of
birefringence becomes obvious once the following is considered. For any
degree of stretch, various physical properties can be achieved depending on
the temperature of orientation, strain rate and molecular weight.
Birefringence is one single parameter which can specify the physical
properties of oriented PET86. Jabarin also states that:
a) Orientation is a function of at least four variables i.e. temperature, strain,
strain rate and molecular weight.
b) The nature of the orientation achieved is crucial to the physical and
mechanical properties of a PET artefact.
c) Birefringence is a phenomenon most clearly related to molecular
orientation.
42
Pereiia and Benavente
ll3
and others
l8
, 114, llS, ll6 have shown that, under
certain conditions where Gaussian (rubber elastic) conditions prevail,
birefringence is proportional to shrinkage stress, with the constant of
proportionality referred to as the stress optical coefficient. The departure
from linearity generally occurs for draw ratios greater than 3 (corresponding
to birefringence values of between 0.07 and 0.12) but depends on the
drawing conditions. De Vries et aP3 reported that the birefringence in the
plane of biaxially oriented PET film increased with increasing difference
between the stretch ratio in two stretch directions, specifically MD and TD.
J abarin
86
also reported that the birefringence of biaxially oriented PET
sheets increased with decreasing stretch temperature and increasing stretch
ratio. Cakrnak et aP8 noted that the birefringence ofbiaxially oriented PET
bottles increased with increasing inflation pressure. Yokouchi et al
ll7
reported that the birefringence of PET sheets, which had initially been
stretched uniaxially at 80C, increased with increasing stretch ratio.
Biangardi and Zachrnann
9
noted that an improvement in the mechanical
properties of PET films, induced by increased annealing temperatures, was
obtained by an increase in birefringence and by increasing the number of
taut tie molecules. Zachrnann
ll8
also reported that for uniaxially oriented
PET sheets, the orientation of the amorphous phase increased with
increasing film birefringence. He pointed out further, that the fraction of
the amorphous taut tie molecules decreased with annealing temperature and
with decreasing film birefringence.
Orientation birefringence is a composite parameter, made-up of components
from the crystalline and amorphous phases and so care must be exercised
when interpreting the data. Gupta, Sett and Deorukhkar
ll9
reported that at
high spinning speeds, the crystallites formed during the cooling of the melt
under an extensional force field are quite large, ordered and perfect. The
crystalline orientation was also quite high. However, the birefringence was
low and it apparently arose from low amorphous orientation. This has been
explained
l20
on the basis that the stress-induced crystallisation, during high-
speed spinning, takes place in the pre-oriented amorphous phase and
consequently, there is a reduction of orientation in the amorphous phase.
43
Even with the considerations outlined previously the interpretation of
birefringence data is generally far from clear cut. Gupta and Ramesh
121
speculated that the wide range in reported values for the birefringence of the
two phases could be due, not to problems arising in detennining the
intrinsic amorphous value, but because the crystalline phase can exist at
different levels of perfection and this will affect the birefringence result.
Cakmak, White and Spruie1l
63
comment that far more infonnation may be
obtained from refractive index measurements than from birefringence.
Dulmage and Geddes
122
reported problems using birefringence as a measure
of total orientation in their PET films. They found that they had to use an
average transverse refractive index, taking into account the thickness-
direction polarisation (dispersion). Heffelfmger and Burton
61
found the
maximum refractive index to be a more reliable indication of total
orientation, if corrections were applied for the enhancement of refraction
due to the increase in crystallinity, i.e. by dividing the refractive index by
the density. They referred to this as an "orientation index".
2.5.3.2 Fourier Transform Infrared Spectroscopy
FTIR has found considerable application in the literature for studying the
confonnational transitions (trans-gauche) of PET. From such studies,
Moore and O'Loane
123
suggested these confonnational changes were a
molecular basis of embrittlement in PET. IR spectroscopy was one of the
three techniques (density and WAXS being the others) that Aharoni
124
used
to measure the degree of crystallinity in PET samples. After applying a
correction (for the 7% of the trans confonner which exists in the amorphous
phase) they found W AXD and IR measurements to be in close agreement,
but density to be subject to a systematic error. The value of 7% was
confmned by Stokr et a1
125
Sikka and Kausch
126
studied stress-induced
changes in morphology using FTIR. They found that a notable fraction of
the gauche confonner, of the ethylene glycol linkages, is converted to the
trans confonner. Further they showed these transitions accompanied the
transfer of stress to the crystalline regions. They commented that the
stressed PET film behaved as if the degree of crystallinity had been
increased.
44
The application of IR and increasingly FTIR has been extended to
orientation studies by polarising the infrared beam. Gupta and Ramesh
83
defined a dichroic ratio as:
_ (AlIA2)-l
DF - (AlIA2)+2
Al is the absorbency measured with the polariser parallel to the machine
direction and A2 is the measured value in the perpendicular direction.
Uniaxially drawn samples were heat-set, both 'free' and taut-annealed. The
DF values for both phases (amorphous and crystalline) were shown to
correlate with strain. By ratioing the 988cm-! and 1506cm-! bands, chain
unfolding during extension was also examined. Cunningham et al!27
compared orientation functions derived by refractive index and polarised IR
measurements and showed that, for uniaxially drawn PET, good correlation
exists for measurements using the 875 cm-! band. The 875 cm-! and 795 cm-
! bands are both associated with out-of-plane C-H deformation in the
benzene ring. Therefore, they are not directly affected by conformational
changes in the same way as the 975 and 896 cm-! bands, and might be
expected to give an indication of overall chain orientation. Slightly higher
consistency in the results was found for the 875 cm-! band (as opposed to
the 975 cm-! band) which was attributed to the 875 cm-! band's greater
degree of anisotropy in absorbency.
Cuningham assumed that the optical measurement provided an absolute
calibration for the IR measurements and the value of em, the fixed angle an
individual transition moment direction makes with the chain axis, was found
to be 85. This allowed the following identity to be made;
This degree of correlation was only possible when the Lorentz-Lorenz
internal field correction was applied.
45
The orientation behaviour of the 975 cm-
I
and 896 cm-
I
bands associated
with the trans and gauche confonnations respectively, was found to be
much more complicated. From these observations, Cunningham et al
inferred that firstly, the molecular segments which contain the trans
confonner possess(at low overall orientation) much higher orientation than
those segments which contain the gauche confonner. Secondly, up to
<P
2
(9-0.45 the gauche orientation is very close to isotropic. There was
even an indication that the gauche segments for low orientation may be
oriented transversely to the draw direction. Cunningham also stated that the
study of the 896 cm-
I
band quantitatively is very difficult since it overlaps
with the strong 875 cm-
I
band. Finally, he found that for high degrees of
overall molecular orientation there is appreciable alignment of the gauche
segments.
Jarvis et al
128
compared orientation results made by refractive index
measurements, Raman and infra-red spectroscopy: For the uniaxial films
examined, Raman and infra-red agreed well and the three principal
refractive indices also fitted into the overall picture of molecular orientation
obtained. The results indicated that there was a planar orientation of the
benzene rings and that at the rather low degree of overall orientation
obtained at a draw ratio of 3.5: 1, the chain axes are preferentially oriented
away from the plane of the film. This result was contrary to expectation,
but J arvis commented that the degree of this preferential orientation is quite
small.
Chalrners et al
129
demonstrated an approach to quantifying orientation in
PET using polarised FTIR that has now become standard practice in ICI at
Wilton. The absorption band chosen for the study was the 1018 cm-
I
band.
This band is assigned to the vibration of the disubstituted benzene ring. It
was chosen because its resultant dipole moment change is parallel to the
C
1
-C
4
direction in the ring
128

46
Following Spruiell, White et al. 98, they defmed two biaxial orientation
functions fIB and f2B as follows:
fIB = (Ax-Az).C(Ax+Ay+Az)-1
f2B = (Ay-Az).C(Ax+Ay+Az)-1
where Ax was the absorbency obtained whilst the electric vector of the
infrared radiation is polarised parallel to the bottle axis (x). Ay and Az
were similarly defmed relative to the bottle hoop and thickness directions
respectively. C is a constant to correct for the fact that the dipole moment
change for the 1018 cm-
I
band is conically distributed about the polymer
chain at an angle v. C is given by the expression,
C = (1I2(3cos
2
9-1))-1
Following Hutchinson et a/
130
Chalmers et al assumed a value of 20 for 9
and so C = 1.213. The data can be presented as a plot Off
l
B
against f2B to
yield an easily interpreted graph (after White et a[98). Chalmers points out
that this representation does not, however, allow for a description of any
orientation about the chain axis ( e.g. a tendency, or otherwise, for the
benzene rings to lie in the draw plane). If this information is required then a
subsequent series, using another absorption band with its resultant dipole
perpendicular to the chain axis, would be needed. All Chalmers FTIR
measurements were from thin sectioned samples. Sectioning may induce
orientation in polymeric samples and so care must be taken to minimise this.
Although not detailed in the paper above, Everall
l3l
(a contributor of the
Chalmers paper) indicated how this was achieved.
2.5.3.3 X-Ray Diffraction
X-ray diffraction can contribute to the understanding of PET crystalline
morphology at many levels. It provides an alternative to density
measurements for the degree of crystallinity. The reader is referred to a
survey of such X-ray techniques in table I of Linder's paper59.
47
The use of both SAXS and W AXS investigations of orientation and
morphology of PET by Cakmak et af93 demonstrates the potential of this
technique. In the PET triclinic unit cell, the phenyl ring planes are roughly
parallel to (100) planes and the (105) planes are roughly parallel to the
chain axis. Therefore these planes are key to X-ray analysis of orientation.
The (100) planes rapidly_align parallel to the film surface as the transverse
draw is increased. The (105) planes are highly concentrated in the machine
direction, for uniaxially drawn films. Transverse stretching spreads the
concentration of the poles from the machine direction towards the
transverse direction, with the distribution tending to remain always in the
plane of the film. In this way, a sequential biaxial draw converts uniplanar-
axial orientation to uniplanar. Heffelfmger and Burton
s7
examined the
intensity of the (100) planes for uniaxially drawn PET and found that at low
elongations up to 200%, the peaks were broad and not well defined,
indicating that the amount of uniplanar-axial orientation is small. When the
elongation reached 250% the distributions sharpened appreciably and the
orientation of the crystallites in the sample more closely approached the
ideal uniplanar-axial case. However, even at low elongations (100%), they
noted that the crystallites tended to align with the surface of the film and at
higher elongations, the number of (100) planes relative to the film surface
increases rapidly.
Cakmak et al produced SAXS pole figures taken in identical directions in
aunealed and un-aunealed material and found the figures to be very similar,
the only exceptional difference was the extra maximum in the aunealed
sample. This confmned that fixed aunealing (heat-setting) does not cause
gross structural reorganisation, but only perfects the structure developed
primarily by stress induced crystallisation.
48
2.5.3.4 Density
Inferring crystallinity through density measurements is via relationship
u = 100 ( ::!)
c v,c-pa
U
c
= Volume crystallinity
Pa = density, amorphous phase
Pc = density, crystalline phase
It is universally accepted technique by the PET industry as the method for
measuring the degree of crystallinity. However, the literature is far more
sceptical about the validity of the inference. Obtaining density data from a
density column is simple, quick and inexpensive; hence its attraction as a
method of obtaining the degree of crystallinity for industry. For this
technique to be reliable:
a) The densities of the two separate phases ( amorphous and crystalline)
must be known.
b) The density of each phase must remain constant (i.e. crystalline
transformations or densification of the amorphous phase would produce
erroneous results).
c) There must be no internal voiding in the sample.
The standard values of Pc and P. for PET are 1.455 and 1.335 glcm
3
respectively62. Aharoni et al
124
showed that crystallinity measurements
produced by IR and W AXS correlate well with each other, but not with
density values derived using these figures. Altering the p. value to 1.348
glcm
3
gave the best fit. However fully quenched samples having zero
crystallinity by XRD measured 1.337 glcm
3
Therefore the treatments given
to crystallise the PET can also densify the amorphous phase, which is
further supported by work from Miyaka
132
and Fischer et a1
57
. Using
p.=1.348 glcm
3
altered one value of crystallinity from 12.4% to 6.4% in
Aharoni's work, demonstrating its importance. Linder5
9
, through XRD and
density data, characterised three phases, namely crystalline, amorphous and
intermediate (mesomorphic) in PET fibres.
49
Linder states that" as long as no other suitable method is available, the two
phase model has to be regarded as an adequate approximation for PET
fibres with respect to density". However, he also concludes that the
measurement of the amount of crystalline ordered regions by density alone
is unreliable, because of the high degree of variability in density of the non-
crystalline regions.
Kim 133 demonstrated that fifteen minutes is not sufficient time for a sample
to reach its equilibrium point in a density column. This is due to the liquid
penneating the voids caused by drawing, in the surface of the polymer
particularly for low re-heat blow (RHB) times (lower temperatures attained
by the PET prefonns). This also demonstrates that cool drawing gives rise
to voids (and lattice vacancies), many of which will be closed and will
therefore adversely affect the degree of crystallinity value obtained. Kim
recorded a difference between the non equilibrium density and the
asymptotic value of 0.006 gcm
3
.
2.5.3.5 Thermal Analysis
Because of its complex thennal transitions, PET has attracted considerable
interest in its thennal characteristics. DSC curves generally show a small
endothermic transition at about Tg, an exothennic transition at about 140C
(only in previously-amorphous material) and a fmal melting endothenn at
the melt temperature. It has been shown that all these transitions are
processing and environment dependent. PET can also exhibit multiple
melting peaks (also referred to as premelting, secondary or double melting
peaks). The onset temperature of the secondary melting point has been
shown to vary with DSC scan rate!34. The magnitude of these peaks has
also been shown to increase with armealing time and temperature
I35
, 136.
50
Several origins for these have been suggested: disorientation and melting of
oriented crystals 137, and more lately, melting of smaller, less perfect
crystals 138, or the melting of different morphological forms 139 and partial
melting and recrystallisation during the actual DSC run
134
Mechanical
strains have also been shown to induce or prevent multiple endotherms
l40
.
Rao, Kumar and Dweltz
141
demonstrated that the premelting endotherm is
influenced by pre-orientation, crystallite size and distribution of crystallite
dimensions; the small crystallites melt in favour of the formation of larger.
Fontaine et al
142
via SAXS experiments, showed that annealing removes
structural distortions and leaves a better delimitation at the level of the
amorphous-crystalline interface, and an increase in density. It also
increases the perfection and size of the crystallites by forming thin lamellae,
and brings about further crystallisation of the amorphous material. F akirov
et al
134
only found a secondary melting peak when the heat-setting
temperature was in excess of 140C unlike Haworth et aP who reported
secondary endotherms at temperatures as low as 115C. This assignment of
the secondary melting peak is not the only interpretation, but the
alternatives have already been discussed in some detail in section 2.4, and
so need not be repeated here. However, should this view be proved correct,
it could help explain the sometimes poor correlation between crystallinity
measurements made by DSC with other techniques.
Drawing PET past its yield point converts gauche conformer to the trans130.
143 After yielding, both stress and orientation decrease because the
conformational changes can take place to allow the overall network
structure to rearrange. When this occurs, stress relaxation associated with
permanent plastic flow takes place and the network as a whole achieves a
higher state of overall orientation. Peterlin and MeineP44 demonstrated that
drawing increases the free enthalpy of an amorphous system.
51
PET drawn above its Tg has a similar melting behaviour to an undrawn
sample annealed for much longer times. Fakirov et al
134
concluded that,
using W AXS during annealing of drawn PET, crystal thickening occurs
without an increase in long spacing. This means that crystallites once
fonned cannot be reorganised during the annealing treatment, in contrast to
other polymers such as Polyethylene. The hindrance of reorganisation may
be the reason for the relatively low ultimate crystallinity of PET. If
crystallisation occurs before a drawn sample can relax, a higher free
enthalpy, in the amorphous portion, will be frozen in.
PET has numerous sub-ambient transitions; thenno-mechanical experiments
have shown transitions to occur at -130C, -62C and 77
o
CI45. Miller
l45
concluded that this indicated some ordering takes place below 70C ( below
the recognised Tg of PET) Illers and Breuer
l46
presented data illustrating
that low temperature transitions do occur in PET, and the change in the heat
capacity of PET at -130C led Miller to feel that the true Tg of PET to be -
130C. It should be noted that (to date), no other literature has been found
to support this view.
To calculate the crystallinity in a sample prior to the DSC run the
exothermic heat from crystallisation during the scan is subtracted from the
heat(s) of fusion. There does appear some disagreement in the literature
about the correct value for this, for example, Elenga et al
53
used a value of
140 J g-I, whereas Cakmak et a[63 used 126 J g-I.
A literature survey such as this, based on orientation and heat-setting of
PET, gives the overall impression that a considerable amount of work has
been conducted on this material. However, PET is a fairly unique polymer,
having a maximum attainable crystallinity of approximately 50% and so
knowledge of the combined effects of all the phases is required. The long
crystallisation half time, coupled with the dramatic effect on the
crystallisation kinetics that orientation has, only serves to complicate the
problem. Further structural complexity then occurs if heat-setting is
introduced. There is still considerable work to be done before all the issues
and controversies outlined in the review are resolved, and oriented heat-set
PET can be said to be fully understood.
52
3 EXPERIMENTAL
3.1 MATERIAL GRADE SELECTION
The cost of achieving FDA approval for a new material in a food contact
application such as this is extremely high. Hence it would be advantageous
for any gains highlighted by the investigation to be achieved with already
approved grades of PET. From ICI's range of bottle grades (table 3.1), two
materials were chosen; B90N (Trade Mark, ICI Chemicals and Polymers
Ltd), a PET homopolymer, aoIthe recently-launched B95A Laser, (Trade
Mark, ICI Chemicals and Polymers Ltd) a copolymer, containing 1-2%
isophthalic acid. Both grades have similar intrinsic viscosities (O.82dl/g for
the homopolymer and O.81dl/g for the copolymer). - The homopolymer is
well established in the market place, being lCI's best selling grade, whilst
B95A Laser is a recent addition to their portfolio, introduced to provide a
more controlled development of crystallinity, required in one piece bottles
with petalloid bases. The as-extruded films of both grades used in this
investigation have identical intrinsic viscosities, as confmned by melt
viscosity measurements performed by the Melinar (Trade Mark, ICI
Chemicals and Polymers Ltd) analytical laboratory at lCI Wilton, and so
this allows for the investigation of the effect of the copolymer.
3.2 SAMPLE PREPARATION, TECHNIQUE DEVELOPMENT
Sample preparation can be considered as two separate processes: firstly the
as-extruded material is biaxially drawn, then heat-set. The project proposal
intended the use of the biaxial stretching rig at IPTME, figure 3.1. This is
an attachment for an Instron tensile machine and whilst producing rather
small samples, had been successfully used to produce biaxially oriented
samples for projects in both PET and PVC. The samples are raised to, and
maintained at the draw temperature by the whole assembly being placed in
an enviromnental chamber. As the drawing technique was already well
established no further development work on drawing was necessary at this
stage of the project.
53
3.2.1 The TM Long Stretcher
Just as the proving work on the heat-setting attachment for the IPTME rig
was underway, Krupp Corpoplast launched their "Corpotherm" heat-setting
machinery for PET bottles
l
. Its principle is to use a hot mould to impart the
various heat treatments, making the approach adopted by this project
directly relevant to a commercial process. Partially in response to this,
emphasis was shifted from the slightly abstract work planned on orientation
functions/morphology to more detailed experimental design programme
concentrating on heat-setting variables. Mechanical properties were now
also required from the programme and the IPTME biaxial rig simply could
not produce large enough samples for this, whereas the Long Stretcher can.
The Long Stretcher is manufactured by the TM Long Co of New Jersey
USA, and is used to produce laboratory scale samples, by uniaxially or
biaxially drawing specimens of cast sheet. The smallest sample needed for
a test is 6cm x 6cms which is very convenient, being both small enough to
be laboratory scale, but giving drawn samples of sufficient size for
mechanical testing.
The Long Stretcher comprises a hydraulically operated stretching head
mounted inside a heated oven with a liftable lid. The operation of the
stretching mechanism is based upon the relative motion of two pairs of draw
bars mounted normally to each other. The draw bars are attached to
hydraulic rams which control the amount (draw ratio) and speed (draw rate)
of the imposed stretching. On each draw bar are mounted pneumatic
sample clips (jaws) attached to a pantogragh system.
A sample is located symmetrically on a vacuum plate which is mounted on
an arm. This arm allows the plate and sample to be swung into place and
lowered so that the film is between the clips. The sample is then gripped by
actuating the clips using a controlled pressure of compressed gas, usually
nitrogen. The lid is lowered and the upper and lower air heaters rapidly
bring the sample to draw temperature.
54
Force transducers on two of the clips enable the nominal drawing stress in
each axis to be measured during the drawing process. The draw itself can
be monitored using the linear potentiometer on each arm (unfortunately
these were not functioning at the time of our experiments and so the draw
information for this work is not available). The amount of draw is set by
adjusting stops on the right and rear of the draw arms. If they are set the
same then the film is said to be balanced, or if not, unbalanced. The draw
can be carried out either in constant rate, or constant force mode.
In addition to greater sample size, the Long Stretcher has the following
advantages:
a) Tentor jaws: these allow the material to be stretched at the jaw to give
more homogeneous biaxial orientation over a larger area of the sample.
b) Maximum biaxial strain: this is much higher, than the Instron
attachment in IPTME, allowing for wider range of experimental
conditions in which to fully develop strain induced crystallinity.
c) Strain rate: much higher strain rates are possible, (maximum linear
draw speed is 508mms
1
) closer to the real situation found in PET
preform inflation, during bottle production.
d) Instrurnented unit: Stress/strain data are obtainable m both draw
directions.
e) Draw ratio: the total strain in each direction does not have to be equal
or simultaneously achieved.
6cm x 6cm samples were cut using scissors from an as-extruded sheet of
each material (homopolymer and copolymer) and a lcm grid (to assess the
draw) marked on each of them, using a fme indelible pen. The samples,
once secured in the Stretcher's clips using a gas pressure of 410 PSI (2.83
MPa) were allowed 120 seconds to reach the draw temperature of 110C.
The draw speed used was 254mms
1
(10 inches/sec).
55
Taking account of these factors with the overriding constraint of sample
size, the only logical step was to adopt the Long Stretcher for all trials.
However, this meant that firstly, the work developing the heat-setting
equipment in IPTME was made redundant and secondly, as the Long
Stretcher had no suitable heat-setting facilities, more heat-setting equipment
had to be designed and tested.
3.2.2 Heat-setting
Heat-setting is an isothermal, isometric annealing process. Therefore the
ramp times to and from the anneal temperature must be insignificant
compared to the annealing time, and so an enviromnental chamber cannot
be used for the annealing stage. During this heat treatment the material
must be constrained so that it cannot retract under the influence of any
relaxation of orientation imparted by the draw. This is most easily achieved
by leaving the sample in a biaxial rig during heat-setting. Therefore, only
the problem of heating the sample to the annealing temperature remained.
Several different heating methods were considered, (see below).
3.2.2.1 Heat Gun
A heat gun (hot air blower) was tried but both temperature control and heat
distribution were found to be very poor, with the sheet displaying localised
melting.
56
3.2.2.2 Infra Red
Infra red radiation is used to heat PET preforms in production equipment.
However, the heaters have a relatively long-heat up time from powering up,
so the heaters would have to be left on continuously in these experiments.
This would require these radiative heaters to be moved some distance from
the sample, or for a shutter mechanism to be included to prevent premature
heat-setting. Further, temperature control is not by direct feedback but by
calibration of the supplied power and working distance. Whilst this is
acceptable for production, it is not a desirable feature for an experimental
rig, running relatively few trials at many different temperatures, where
accuracy is the most important factor.
3.2.2.3 Manual Application of Preheated Metal Blocks
By heating metal blocks of sufficient size (thermal mass) in an oven and
then applying one surface of them directly to the drawn sample, rapid,
accurate heating can be achieved. This method has been used in uniaxial
stretching work in IPTME, by placing a block each side of the sample.
However, the biaxial stretching rig restricts access to one side, allowing
only one block to be used. Manually placing a heated block with uniform
and reproducible pressure (to ensure uniform and reproducible heat transfer)
is difficult to achieve with consistency, and so block placement must be by
jig. This requires that the block stays in the environmental chamber
throughout the duration of the procedure and so must be heated
independently.
57
3.2.2.4 Electrically Heated Jigged Block
By inserting two cartridge heaters and a thennocouple in a jigged block,
(figure 3.2) the drawn sheet temperature can be rapidly and accurately
raised and maintained for heat-setting. This technique was selected and a
rig built, incorporating the following features:
a) The block can be lowered onto the sheet surface remotely so the
environmental chamber does not have to be opened. This ensures a
more reproducible thennal history.
b) The block pivots on its horizontal axis to ensure even pressure on the
sample.
c) The applied pressure is adjustable by the selection of one of the
multiple fulcrum points on the rig.
d) Heat transfer is from one side only, thereby simulating commercial
heat-setting processes in a hot blow mould.
e) By reducing the power to the heaters, temperature vanatIon was
restricted to O.5C ie. a relatively high degree of accuracy, in
comparison to other techniques.
f) Effective heat transfer to the sample at low pressure is ensured by the
application of silicone oil to the block's working face.
g) The block is sufficiently large that the heaters do not have to respond
during heat-setting, maintaining better temperature control.
However, this was designed for the biaxial stretching attaclunent at IPTME
and so the adoption of the Long Stretcher to produce samples required the
design and construction of a new heat-setting rig based upon the same
working principles.
58
3.2.3 Design of The Second Heat-setting Rig
Two options were now considered: either to modify the Long Stretcher to
give it a heat-setting capability, or to build a separate rig to heat-set the
samples. Technically the first option is the best since when a sample is
removed without cooling, some relaxation can take place. However the
Long Stretcher is an expensive and complex machine; this, coupled with the
sample being obscured during operation meant that the technically, slightly
less-satisfactory route of a separate rig had to be chosen. This also had the
logistical advantage that such a rig could be based in Loughborough,
thereby reducing the amount of time required to be spent at lel Wilton.
3.2.4 Heat-setting Rig Requirements
The objectives associated with the rig design were:
a) To grip a sample with sufficient force, then to apply constant, biaxial
tension to maintain the sample dimensions during heat-setting.
b) To leave sufficient area of the sample exposed, to allow access by the
heat-setting medium.
c) To cope with all the sample sizes planned in the experiment (sizes
ranging from 155mm x lOOmm to 240mm x160mm).
d) To locate a heat-setting device onto one surface of the sample, with
sufficient and uniform pressure, to ensure uniform and continuous heat
transfer.
59
To maintain biaxial tension on a rectangular sample requires four jaws,
which must grip the sample along the entire length of each side. This would
prevent a set of jaws closing down to accommodate smaller samples as the
vertical and horizontal jaws would foul each other. The Long Stretcher
overcomes this problem by mounting the jaws on a pantograph and is the
only solution when large changes in sample size are expected, once the jaws
are closed, as in drawing. This change would not occur in the heat-setting
rig as its role was one of retention not deformation. This obviated the
necessity of a pantogragh system for the jaws, that would have been
extremely time consuming both in design and manufacture.
Separate jaw lengths for each sample size were considered until a far more
elegant solution was found. This involved the design of a set of jaws with a
flush surface which the sample would lie over when gripped. The vertical
pair of jaws were then offset, in the frame with respect to the horizontal pair
so that they could move over each other. With this arrangement any
combination of axial draws, under the maximum size, dictated by the jaw
length, can be put into biaxial tension. These design aspects were included
in the fmal drawings for the heat-setting rig, which was then constructed by
the IPTME workshop (figures 3.3 to 3.6).
3.2.5 Heat-setting Methods
Having designed the frame and clamping arrangements, attention was given
to exactly how heat-setting was to be performed. Once again the variability
of the sample size had to be a major consideration. A heated silicone bath
was considered to overcome this, but it was felt that this would represent
too great a deviation from the objective of simulating a real process. So
having decided to persevere with the hot metal block technique, the problem
was then easily resolved into two components:
I) How to heat the block?
2) How to accommodate sample size variation?
60
At first, electrical heating was favoured as some experience and equipment
had been previously accumulated (section 3.2.2.). However, with a much
larger block required, ensuring a constant temperature across the block's
working surface would be much more difficult with this mode of heating.
Also, if, as it appeared at the time, several blocks of different size would be
required, then each one should be as simple as possible to manufacture.
This could have been overcome by using an electrically heated base plate
and screwing the various sized blocks to it. However, the thermal mass and
lag of such a device would each be rather large, making effective
temperature control more difficult.
Because this rig was not intended to function in an environmental chamber,
there was no longer any need for it to be self heating. Preheating the blocks
in an oven is far simpler and would also be likely to give much better
temperature uniformity, providing they did not cool appreciably during use.
To test this proposal, a thermocouple was surface-mounted onto a steel
block of approximately the smallest size envisaged, 99rnrn x 76rnrn x
24rnrn, ie. the worst case in terms of largest surface to volume ratio. This
was heated to 200C in an oven and its surface temperature was measured
with time, once it was removed and placed in ambient conditions. The
thermocouple was then moved from the centre of the face to a corner of the
same face and the experiment repeated. The low cooling rates of this worst
case experiment (figure 3.7) confirmed that oven heated blocks would be
adequate and the provision of insulation to the other faces could only help.
The arrangement of the jaws (figure 3.5) meant that the length of the heat-
setting block was not restricted by the jaws so that only the width would
have to be reduced as the sample size decreased. This meant that a series of
bars could be clamped together to provide a heat-setting surface. By adding
or removing bars, the width of the surface could be altered to suit the
specimen size, figure 3.6. Heat loss from this block was measured, on the
surface, at the centre and a corner using a thermocouple, figure 3.8. Figure
3.5 show the rig and block together ready for use.
61
3.3 Factorial Experimental Design (FED)
Because of the wide draw capability of the Long Stretcher, the possible
window for the investigation was so large that to investigate by classical
experimentation would consume a disproportionate amount of project time.
Both parties in the project had previous knowledge of using factorial
experimental design techniques and ICI have at their Wilton site a group
dedicated to statistical manipulation. This group were willing to become
involved in the project and so the decision to use FED to examine this area
of interest was taken.
A three variable FED was designed comprising of draw area, (the major
draw multiplied by the minor draw), heat-setting time and temperature, with
draw temperature and strain rate kept constant. The values for each of the
variables were selected to approximate those that would be found in a
commercial stretch blow moulding process. The maximum grip pressure
that could be sustained by the Long Stretcher's gas operated jaws limited the
drawing force that could be generated, as the samples would slip out of the
jaws rather than be drawn. Consequently, the experimental conditions had
to be slightly modified from the ideal for simulation (slightly higher draw
temperature, and lower draw speed and area) to accommodate the
limitations of the Long Stretcher. However, these conditions are still very
relevant to the stretch blow moulding (SBM) process.
62
3.3.1 Selected Values For FED Analysis
FED a=2(p/4) p= number of variables
for p=3 a=1.68
FED -a -1 0 +1
value -1.68 -1 0 +1
Da 4.5 6.4 7.5 9.7
Temo(OC) 80 104 140 176
time(s) 0 6 15 24
Da = draw area (major axis draw * minor axis draw) nominal
Temp = heat-setting temperature
time = heat-setting time
Constant Parameters:
Draw temperature= 110C
Draw ratio between
axes (X:Y) = 1.5:1 nominal
Linear draw speed = 254mms
1
+a
+1.68
11
200
30
The complete FED programme with coded and the corresponding
experimental values can be found in table 3.land how this corresponds to
the films in table 3.3.
Biaxial orientation may be achieved either sequentially or simultaneously.
Although nominally the samples were drawn simultaneously, the draws are
unequal in the X and Y axes but the draw speeds are the same, therefore
the major axis continues to be drawn after the minor has finished. This
infers that the sample under goes a balanced draw to the point where the
minor axis draw has been achieved, at which point the major axis continues
to draw, generating the 1.5:1 desired draw ratio. The exact sequence of the
draws in a stretch blow process will depend upon the relative timing of the
introduction of the inflation gas and the down stroke of the stretch rod.
63
3.3.2 Comments on Implications ofthe FED on Biaxial Stretching
a) The draw temperature is about 10C higher than in typical SBM but
was required to achieve a sufficiently high draw area, with the
homopolymer samples. The draw speed is lower than the ideal, due to
the limitations imposed by the maximum clamping force available.
The draw ratio was intended to be 2:1 (with respect to each principle
axis) but as the inequality of each draw ratio increases, maintaining the
maximum draw became more difficult.
b) The Long Stretcher's method of operation does not lend itself to
producing specific or exact draw areas, as ideally required for an FED
approach. Some variability in the draw area achieved had to be
accepted. However, because SAS (ICI's statistical modelling package)
can accept non FED defined values, the impact of this is virtually
eliminated. If the same data were run on a package where this is not
the case, (e.g. the FED package available at Loughborough University)
it would be a source of considerable error.
c) The samples were removed from the stretcher whilst at a temperature
still above Tg. For highly strained material (Da >7.5) this poses no
problem, as the strain induced crystallites maintain the sheets'
dimensions. However when the draw area is less than 6.4 the samples
could be seen to retract away from the jaws slightly.
It should be borne in mind that despite the comments (a) to (c) above, the
Long Stretcher provides an even biaxial draw at strain rates far higher than
could be achieved using any kind of mechanical driven device (e.g.
attachment to a tensile testing machine). Therefore the Long Stretcher
conditions are about as close to stretch blow moulding conditions as can
achieved, without incurring the high cost of constructing an experimental
blow mould.
64
3.4 CHARACTERISATION OF ORIENTED AND HEAT-SET PET
It was decided to characterise the samples in three different areas, namely
mechanical properties, shrinkage, and thermal properties. The first two
specifically generate information of direct relevance to the performance of
heat-set material in the carbonated soft drinks (CSD) application. The third
provides more depth of understanding into the materials' response to the
drawing and heat-setting conditions.
3.4.1 Mechanical Properties
3.4.1.1 Tensile Testing
Tensile measurements were required to examine the effect of heat-setting on
the basic mechanical properties and to ensure that the process would at least
not be deleterious to these essential properties.
The tensile testing of thin plastic sheets and films is covered by ASTM
D882
99
and was adhered to, except for the number of samples used and the
use of a gauge length. For each of the FED runs, three tensile dumbbell
(gauge length 28mm long, 3.4mm wide) specimens were cut parallel to the
major draw axis and tested to failure. The tests were performed on a JJ
Lloyd R2000 machine, using rubber lined jaws at a cross head speed of
50mm1min, and at ambient temperature. This was then repeated for samples
cut parallel to the minor draw axis. Because of the time required to produce
samples and the size of the tensile specimens, it was only possible to repeat
the test three times and not five, as specified in ASTM D882. This would
not effect the ability to assess the accuracy of the results as there are
sufficient number of repeats in the FED programme to quantify error.
65
In addition, it was decided not to mark a gauge length on the sample waist,
but to use the whole dumbbell gauge length for determination of nominal
strain. This was because of the inaccuracy in trying to measure the gauge
length of a failed film sample. Using the sample as the gauge length meant
that it was possible to use the Lloyd's internal strain gauge, which is
accurate to approximately a micron of cross head movement. This
approach was considered at least as accurate as the alternative in this case,
and had the added advantage of being consistent with the only possible
method available method of strain measurement during the creep tests
(section 3.4.1.2).
The tensile properties chosen for analysis were yield stress, yield strain,
ultimate tensile stress, ultimate tensile strain and strain hardening rate. The
high strains encountered meant that the full scale deflection on the X axis of
the X-Y plotter was too large to allow for accurate Young's Modulus
detennination. The use of computer controlled testing machine which.
became available in IPTME half way through this work removed this
limitation, but constraints on material meant that the work could not easily
be repeated. The selected tensile properties are defmed as:
et -----------------------------
a -
y
STRESS
(MPa)
ay = yield stress
STRAIN(%)
a u = ultimate tensile stress
Ey = yield strain
Eu = ultimate tensile strain
!la/!lE = strain hardening rate
66
3.4.1.2 Creep Testing
Creep data were required, since carbonation of beverages occurs prior to
pasteurisation at elevated temperatures. This means the contents and the
container spend many minutes at a temperature (usually 65C) not far from
Tg, with the internal gas pressure of the bottle also increased due to the
elevated temperature. These quite extreme conditions increase both hoop
and axial stress levels in the bottle, at a temperature when the polymer will
be more susceptible to creep.
Experimentally, "high stress, low duration" creep is quite difficult to
measure directly and creep perfonnance is often calculated from stress
relaxation experiments. However the recently delivered Lloyd tensile
machines included a 'constant stress' specification in the their operating
specification. This made the direct measurement of creep possible,
assuming the samples did not undergo major defonnation, so that constant
load could be approximated to constant stress. The constant load capability
of the machine was tested by first substituting a steel ruler for a sample.
This was chosen as it would display zero creep at this temperature. Also,
having a high Young's Modulus, the transition from loading the sample to
constant load conditions would be far more abrupt than for a PET film
sample, making constant load control more difficult. The fluctuation of
applied load under these conditions was deemed acceptable and can be seen
in figure 3.9. The controlling software was an early release and it soon
became evident that the constant stress mode had not been fully developed.
The software problems encountered at this and, incidentally the tensile
testing stages of the project were used by Lloyds to refme their software but
it caused significant delay to the project.
67
Once the viability of the test method had been established, the test
conditions had to be developed. Returning to the project objective (section
1.31) of simulation it was decided that the test should simulate the
conditions experienced by a pressurised beer container, under going a
typical pasteurisation procedure of 20 minutes at 65C. Details of the
calculation of maximum stress during pasteurisation are given below:
For a typical CO
2
fill of 2.5 volumes, the pressure achieved at 65C is
80 PSI = 0.554 MPa.
For a thin walled container the hoop stress, O"H is twice the axial stress and
is approximated by:
P=Pressure (MPa)
D=Diameter (m)
T=Wall thickness (m)
For the example of a 1.5 litre bottle, (diameter of 75mm and a wall
thickness of O.4mm) this gives a hoop stress of about 51.65 MPa.
From this calculation it was decided to attempt creep testing initially at
65C at 65 MPa for twenty minutes loading time.
However this stress (65 MPa) caused some of the lower draw samples to
fail within the test duration, so the test stress was revised down to 40 MPa.
This inconsistency may be explained by the following argument: In a bottle,
the situation is one of applied load not stress. The areas of the bottle that
have a lower draw and so more prone to creep are considerably thicker and
so the applied stress in these areas is a lot lower than in the thin, highly
oriented and therefore creep resistant, walls on which the calculation is
based.
68
3.4.2 Linear Thermal Shrinkage Determination
ASTM D2732 defmes shrinkage as "the irreversible and rapid reduction in
linear dimension in a specified direction, occurring in film subjected to
elevated temperatures, under conditions where nil or negligible restraint to
inhibit shrinkage is present. It is normally expressed as a percentage of the
original dimension." The test method it prescribes uses a temperature
controlled liquid bath and so is isothermal in nature.
To our knowledge thermo-mechauical analysis (TMA) has not been used to
examine shrinkage in biaxially oriented PET systems, so a preliminary
study was conducted to optimise the experimental parameters associated
with this technique. The study7 was performed using a Mettler TA 40 by
Prof. Z Dong of Jilin University (Department of Packaging Engineering),
China, whilst on a sabbatical at IPTME using sheet drawn on the Long
Stretcher at Wilton.
The principle of the TMA is a very simple one (figure 3.10.). The sample is
clamped using a mounting jig to give a gauge length of lOmm, and is
suspended vertically, the top end being held static in a quartz frame. From
the bottom set of clamps, a small weight is applied to give a force of
between O.OOlN and O.SN. A linear voltage displacement transducer
(L VDT) is used to measure any displacement of the weight support arm. A
computer controlled furnace is then placed over the quartz frame and the
sample. In this way, variations in the sample's dimensions with temperature
(or time for isothermal conditions) can be continuously monitored. This,
combined with the small sample size required, gives the technique such
great advantage over the 'before and after' approach of placing a sample of
known dimensions in an oven and measuring any change in dimensions
after removing the sample.
69
For this technique to be comparable to the 'before and after' type
measurements, the results have to be corrected for thermal expansion,
because the measurements are recorded through an elevated temperature
span. For this correction, accurate temperature-dependent thermal
expansion data for both grades of PET were required, but not found in the
literature. It was decided to measure this using as-extruded material, which
was initially assumed to have no molecular orientation. The results from
the TMA clearly showed this not to be the case, as the material shrank when
tested. To overcome this, extruded samples were annealed at 80C for 30
minutes to remove this orientation. It is only possible to use this correction
up to 11 OOC because above this temperature shrinkage again occurs,
probably due to the material densifying through crystallisation. Figure 3.11
shows raw TMA data, the measured expansion and the true (corrected)
shrinkage obtained by subtracting the first two. The inflection in the
expansion trace is a measure of the material's Tg but is about 10 degrees
lower than would be expected if measured by thermal analysis techniques
e.g. DTA.
Two samples per FED run were cut parallel to the major draw axis and
tested according to the parameters listed below. This procedure was then
repeated for samples cut parallel to the minor draw axis.
The test parameters used were:
Temperature range
Heating rate 20
0
C/min
Load O.OlN
These had been specified as the most appropriate test conditions from the
work performed by Haworth et a(l.
70
3.4.3 Thermal Analysis
Many materials, including semi-crystalline polymers, will show various
thermal transitions either on heating or cooling and these may be monitored
between -lSOC and +725C using Differential Scanning Calorimetry
(DSC). In the DSC cell the sample and a reference (usually air ie. an empty
sample pan) are placed in separate holders with individual heaters and
thermocouples attached. The cell uses a constantan disk as its primary
means of heat transfer to the sample and reference positions and as one
element of the temperature-measuring thermoelectric junctions. Heat is
transferred from the disk to the sample and to the reference via the
aluminium sample pans. The differential heat flow required to maintain
equal temperature between the sample and reference during a temperature
scan is manifested by the different electrical power requirement of each
heater and so can be easily monitored. After various and usually automatic
corrections have been applied by the host microprocessor, the signal is then
used to generate a plot of typically heat flow versus temperature.
This description is not strictly applicable to the. Dupont used as it is a DTA.
Assuming no change of state occurs in the sample in temperature range of
interest, the principle cause of a variation in the heat flow between the
sample and reference is a change in the heat capacity of the sample. The
alterations in the sample's heat capacity measured at constant pressure can
be converted to thermodynamic quantities such as enthalpy and entropy.
DSC has been used extensively in polymer research, being capable of
determining and quantifying the energy associated with thermal transitions
such as glass transition, crystallisation and melting (where conditions
allow). The combination of thermal data for the crystallisation and melting
events has been regularly used to calculate the degree of crystallinity in
samples of semi-crystalline polymers.
71
Thennal analysis by DSC (Dupont DTA 2000) was conducted to measure
the temperature of the secondary melting peak and to quantify the amount
of crystallinity induced in the samples, by both the drawing and heat-setting
procedures. The scan rate was 20C/min to ensure comparable conditions
to the TMA investigation (section 3.4.2). The scans were conducted under
nitrogen to remove water and oxygen from the surrounding environment.
3.4.4 Density Measurements
To provide an independent crystallinity data set, density was measured
using an Arklone-toluene column by the Melinar laboratory at IC!. The
accuracy was quoted as better than 1%. Density measurements were
perfonned on the FED films and the heat-set bottles.
3.5 CONVENTIONAL HEAT-SETTING EXPERIMENTS
The FED work allows rapid assessment of a system's response to a
combination of predefmed variables and their interactions. This approach is
excellent for commercial process development but is rather intractable,
making further treatment of the data, to gain structural infonnation, not
possible. The FED analysis indicated that the two most significant
independent variables for the prediction of the influence of heat-setting are
draw area and temperature. So, to investigate these further a second series
of 38 experiments was designed for a range of heat-set temperatures
between 25C (unheat-set) and 220C for draw areas between 1 (undrawn)
and 10. Non heat-set and undrawn samples were also included as control
specimens. These sheets were tested for shrinkage and crystallinity in the
same way as for the FED samples but in addition, the following properties
listed in subsequent sections were also measured. The complete
conventional experimentation plan is summarised in table 3.4.
72
3.5.1 Optical Properties
The optical properties of polymers can provide a great deal of infonnation
about the underlying structure of the material and are especially sensitive to
molecular orientation. The material in this investigation was relatively thin
and transparent, making these techniques readily applicable. The principle
optic axes were confinned, by examination under cross polars, to be
coincident with the draw directions and are defined as follows:
ex.
/,.L--->Y
This is in keeping with the convention of refractive index nomenclature that
y>l3>a.
73
3.5.1.1 Birefringence
Birefringence is defmed as the difference between the two refractive indices
in a given plane, (section 2.5.3.1). Because the PET samples are biaxially
oriented, two birefringences (the third can be calculated from the other
two), 'in-plane' and 'out-of-plane' are required. Therefore, three
birefringences in a sheet of film material can be defmed:
&1
1
= (y-J3)
where y, J3 and a are the refractive indices in the major, minor and out of
plane axes, respectively (see previous diagram)
(A) Refractometry and Compensator Methods
For samples with high orientation (high draw areas) all three refractive
indices could be directly measured using a refractometer.
The principle of the refractometer is the measurement of the critical angle at
which total internal reflection occurs. This angle is manifested in the
instrument by a demarcation line between light and dark portions of a
telescopic field. This effect is produced by the rays constituting the critical
angle, on emerging from the prism, falling on a mirror where they are
reflected into the field telescope of the instrument. The position of the
mirror required to line the borderline up with the telescopic cross-wires is
indicated by a moving scale observed in the scale telescope. This scale is a
direct function of refractive index.
74
A specimen was cut from each sample using a fresh razor blade to promote
a clean cut (not split) vertical edge. This was then placed on the lower
prism with a drop of a liquid with a refractive index of 1.7455, which is
higher than for a PET fibre and so would be higher than these biaxially
oriented samples. This must be the case for the technique to be successful.
For this work the top prism was not used and so was swung out of place, in
accordance with the manufacturers' recommendations. The specimen was
then illuminated using a sodium discharge lamp to allow measurements to
be made for the sodium D 589.3nm wavelength.
As the orientation of the samples reduced, the refractive indices converged
and the refractometer was unable to resolve the indices of these lower area
draw samples.
For these samples the in-plane birefringence was measured directly, using a
Leitz microscope in transmission under crossed polars using a calcite
'Ehringhaus type' compensator. This type of compensator consists of a
birefringent crystal plate that can be tilted about one of its principal optic
axes. In this way a continuous variation of thickness and refractive indices
can be presented to the light beam, giving a continuously variable relative
retardation. The compensator is used with white light and is adjusted to
bring the black (extinction) fringe to the centre of the field of view. The
amount of tilt required to achieve this with the specimen in the light path is
converted to relative retardation using calibration tables. The relative
retardation values are then divided by the sample's thickness to give the
birefringence. This type of compensator can be used for large relative
retardations (up to 30 wavelengths). Measurements were taken with the
compensator tilted in both directions in order to minimise error.
Oriented PET is a highly optically dispersive material due mainly to the
benzene rings. The implication of this is that the refractive indices vary
with optical path difference (sample thickness) and so the black fringe is no
longer the zero-order fringe. The correct zero-order fringe was identified by
cutting a wedge on the edge of the sample and noting the colour change as
the initially black fringe travels up the wedge, when the compensator is
tilted, to the sample's top surface. This colour is then effectively the colour
of the zero fringe, i.e. after dispersion has occurred and is then used instead
of the black fringe for making measurements.
75
(B) Conoscopy
For orientation studies it is usual to work under crossed polars on an optical
microscope e.g. compensator techniques. If the normal eyepiece is
substituted by a telescopic eyepiece which focuses on the back focal plane
of the objective, then a conoscopic figure (also referred to as an interference
figure) will be seen, when the sample is illuminated with strongly
convergent light. This figure consists of rings of polarisation colours in
white light or dark and bright fringes if monochromatic light is used. The
polarisation colours increase from black, which corresponds to the position
in the black focal plane where the rays passing along the an optic axis
emerge.
1
1
1
1
1
I
1
1
1
1
1
1
1
1
1
1
1
1
I 1
...-,',,",
11
11
11
1
1
1
I
~ 1
\ J / / BACK FOCAL PLANE
--'1-----------
/1 \ B
/ 1 \
\
\
\
\
\
\
\
\
\. OBJECTIVE
1
1
1
1
1
I
1
1
1
1
I
1
SPECIMEN
1
1
For uniaxial orientation the figure is unaffected by rotation, but figures from
biaxially oriented samples change as the sample is rotated through 45.
This provides a method for distinguishing between uniaxial and biaxial
orientation. The fringes are symmetrically arranged around the optic axis
(or axes).
76
OPTIC AXIS
).
I I
tc" 2),
\ ---1,.---'--_+_-,
c=.. \ r J, -=>3J..

" '. I I.' /
'., 1/ 1:;>4)"
'\'1 li
l
/
'\\1 1/1/
" \ \ I I, I //
\ \ I
" \ \ I I , /
, \ \ I I, / //
'\11 "1/
'\\1 lit /
, \\1 I 11/ /
',\'I 11t!/ SAMPLE
T cJ
I
I
1
Diagram of Surfaces of Equal Retardation Around the Optic Axes of a
Uniaxial and Biaxial Crystal. From: Crystals and Polarising Microscope,
Hartshorne. N.H., and Stuart. A, fourth edition p323 and p357, 1970,
Arnold
77
45"
Uniaxial
Biaxial
(constant figure)
, ~ - - ....
: "
'"-",,,,-
~ ~ ~ ' - ~ '
,
', .... __ ;1
9()<>
The fringes fonn concentric rings; the centre of these is referred to as a
melatope. The apparent distance between melatopes is determined by the
optic axial angle (2V) of the sample, the refractive index (13) and the optical
system of the microscope. The angle between the axes which is observed
under the microscope with a dry system of lenses is not the actual angle
(2V) within the crystal, but an apparent angle in air (2E), due to external
refraction of rays at the surface of the section. The relationship between the
real and apparent optic axial angles is given by:
sin E = 13 sin V
where 13 = the intennediate refractive index.
78
It can be shown that half the apparent distance between melatopes of a
biaxial interference figure, as measured by an eyepiece scale, is
proportional to the sine of the angle which the optic axes makes with the
bisectrix, and a constant factor which depends on the lens system of the
microscope. The relationship between the apparent axial angle (E) and half
the apparent distance between melatopes, is given by Mallard's formula:
where:
D = K sinE
D = the separation of the isogyres in the conoscopic figure
~ = the intermediate refractive index, which for mica = 1.593
Once D is found using a standard (e.g.) Mica then 2E may be found
using:
sinE = ~ sinV
v = is the optic axial angle and can be calculated from:
where the principal refractive indices for mica are:
a = 1.560
~ = 1.593
Y = 1.600
The value of K for this microscope set up was calculated as 31.1882
Substituting for ~ the intermediate refractive index for PET = 1.55 and
measuring D allows V for each sample to be evaluated.
79
Mallard's approximation of Wulfing's rule for low birefringence samples
can be written as:
tan
2
v (y-13) = (13-a)
(y-13) is the in-plane birefringence and can be measured directly by the
compensator method (section 3.5.1.1A) and so the out-of-plane
birefringence can be calculated.
3.5.1.2 Optical Clarity
The clarity of a beverage bottle is structurally and aesthetically important.
Excessive haze in PET is generally indicative of spherulitic crystallinity
which causes brittleness and susceptibility to environmental stress cracking,
and therefore must be avoided in the main body of the bottle. Spherulitic
crystallisation is promoted by holding lightly drawn or undrawn PET at
crystallisation temperatures, and occurs at the most rapid rate roughly
midway between Tg and Tm. It is therefore essential that any heat-setting
process should not cause a large degree of spherulitic crystallisation in the
low draw areas of the bottle i.e. the neck and base (especially in petalloid
designs). The copolymer has a slower crystallisation rate and was
introduced specifically to reduce the incidence of spherulitic crystallinity in
such bases.
(A) Total Transmitted Light Measurements
Using a Gardner Hazeguard XL 111, total transmitted light measurements
were made on all of the conventional experimental heat-setting series
samples. To ensure that only loss through absorption and not through
reflection was considered, the effect of reflection was removed by applying
the Fresne1 correction:
%R= 1 0 0 : ~ U
where n is the average refractive index.
80
Reflection occurs sequentially from each face so that the absorbance (A) is
given by:
A = (l-R) - <i'R5
where A is the absorbence
R is the relative reflectivity
y is the total transmission
The absorption measured by the Hazeguard does not vary linearly with
sample thickness. An attempt to model the effect of thickness was tried
using Lambert's Law:
For Transmission
For Absorption
As
Then
(8) Micro-Photometry
I
In-= -kd
10
I
In-=kd
10
I
A=-
10
k _ (lnA)
- d
The results from the Hazeguard indicated a large degree of scatter. On
examination, using a microscope, the sheets were shown to be significantly
scratched and to contain partially melted pellets which would certainly act
as scattering centres. This reduces the effect of the material's innate optical
transmission on the recorded result. It is interesting to note that the
umnelted pellets show no sign of distortion due to drawing. In addition the
Hazeguard measurements cannot be easily compensated for thickness, and
the samples do have a wide range of thicknesses. In an attempt to overcome
these problems micro-photometry was performed on the samples.
81
In this technique the sample to be measured is placed between a microscope
slide and a cover slip and a liquid of higher refractive index than the sample
is introduced to eliminate reflection from the sample. Using a photometer,
light transmission is measured both through the sample and through just the
slide, slip and liquid as a reference. This technique has the advantage of
using a very small sampling area and so the unmelts and the worst of the
scratches could be avoided and the measurements vary linearly with
thickness, making compensation for thickness possible.
3.6. BOTTLE TRIALS
3.6.1 Heat-setting of Bottles
To fully evaluate the shrinkage models generated by the FED work it was
necessary to acquire a selection of heat-set bottles of known thermal and
mechanical history. At ftrst it was hoped that these could be manufactured
to order by Krupp Corpoplast Gmbh as part of a knowledge sharing
agreement. However by Easter 1992 this was no longer an option and so
with no commercial manufacturing facility for heat-set bottles available, an
alternative way of obtaining a series of heat-set bottles was required.
Carters Packaging kindly supplied 50 half litre bottles blown from Melinar
B95A Laser, the PET copolymer used throughout our study. The bottles
were blown at 87C to give approximate hoop and axial draw ratios of 3.7
and 2.5 respectively in the side wall. The draw ratio between the two axes
is hoop:axial, 1.48:1 which is within experimental error of the ratio used for
the ftlms. The hoop draw was calculated directly from comparing the
diameters of the preform and bottle in the side wall areas. The draw along
the length of the bottle is highly variable and so the change in length
between preform and bottle cannot be used to calculate the axial draw in the
side wall area. The reduction in thickness from preform to bottle was used
to approximate the area draw and this was divided by the hoop axis draw to
give the axial draw.
82
For the work on heat-set bottles to proceed, a technique for heat-setting on a
laboratory scale had to be found. The process of heat-setting has two
distinct phases, a rapid rise in temperature to the annealing conditions, then
heat-setting whilst maintaining the bottle in its original shape. It is this last
condition that is the most difficult to achieve in a simulation experiment, as
biaxially drawn polymers are apt to retract at elevated temperature, under
the influence of the relaxing orientation.
3.6.1.1 Plaster of Paris
To retain the bottles' dimensions whilst heat-setting they were filled with a
50% mixture of Plaster of Paris (surgical cast grade) and sand (builders'
general purpose). The sand and Plaster of Paris were first dry mixed by
hand, then water was added, and then the resulting slurry was poured into
the bottle, and allowed to set. The sand was included to reduce the
exotherm and dimensional changes associated with Plaster of Paris. These
bottles were then heat-set using a Tecam fluidised bed (model BFS) with
fme sand as the fluidised medium, and then quenched into water. The
fluidised bed was chosen as it most closely simulates heat-setting in a hot
bottle mould. Sample bottles were heat-set for 3 minutes at various
temperatures. The approximate temperature of the surrounding heat-setting
sand was measured by a thermocouple. This temperature was found to be
rather difficult to control with sufficient accuracy for the requirements of
the experiment.
Previous DSC traces on heat-set sheets from the TM Long Stretcher
confirmed that the peak of the secondary melting endotherm coincides with
the heat-setting temperature used. The DSC traces from the heat-set, Plaster
of Paris filled bottles did not coincide with their heat-setting temperatures.
It was considered that this was due to the high heat capacity and thermal
conductivity of the plaster (as compared to PET), causing the heat from the
sand (in the fluidised bed) to be drawn into the plaster. Therefore, the PET
never achieves the desired temperature. Clearly, Plaster of Paris in this
respect is unsuitable and so an alternative material was sought.
83
3.6.1.2 Polyurethane Foam
Several moisture-curing single part polyurethane foaming systems are
available in aerosol cans on the "do-it-yourself' market, for gap filling. A
can of "Mangers" foam was obtained and used to fill a bottle. The plastic
tube supplied could be attached to the aerosol nozzle and allowed the bottle
to be filled from the bottom upwards. The resulting foam did not reach full
cure and so the cell structure collapsed with time, giving insufficient
support to the bottle. It was felt that this was due to a lack of moisture in
the bottle to induce full cure. A small amount (a few rnls of hand soap) of a
weak soap solution was added to the bottle prior to the foam and this did
improve the cure. The soap was used to wet out the polymer surface.
However, the single part polyurethane system when sprayed has a high
viscosity which made complete filling of the bottles very difficult to achieve
and this, when added to a still incomplete cure, led to insufficient support
for the bottle.
To overcome both the viscosity and cure problems, a two part polyurethane
foam (Isofoam RM118) was obtained from Baxenden Chemicals plc. The
bottles were sprayed inside with release agent, the two components of the
foam were mixed and used to approximately quarter fill bottles and the
bottle cap fitted. The caps had two knife incisions made in them, at right
angles to each other, so when the polyurethane started to foam the air could
be expelled through the incisions but no foam was lost. Once the foam had
set, the caps were removed and the bottles were placed in an air oven at
60C for several hours, to post cure the foam and drive off any excess
foaming agent. This procedure gave a foam of sufficient rigidity to
maintain the bottle dimensions during heat treatment. These bottles were
heat-set using the fluidised bed at various temperatures for 3 minutes, then
immediately quenched into water, except for bottle II which was allowed
to cool in air. The temperature of the fluidised sand adjacent to the bottle
was measured using a thermocouple to provide a rough guide to the heat-
setting temperature of each bottle. The results for DSC analysis of the
conventional experimentation films (4.2.2) demonstrated a high correlation
between J3-peak temperature and heat-setting temperature. DSC was
performed on all the bottles so that for any results analysis, the beta peak
temperature could be used as the indicator of the heat-setting temperature,
and not the sand temperature.
84
-------------------------------------------------------------.......
Table of Bottle Heat-Setting Temperatures
Bottle
No
1
2
3
4
5
6
7
8
9
10
11
Sand Temp
C
130
150
160
180
190
200
205
220
225
160
117
132
147
172
174
194
205
208
220
150
Shrinkage measurements were then made on these bottles, using the
technique described earlier (section 3.4.2).
3.6.2 Orientation Measurements On Heat-set Bottles
3.6.2.1 Sample Preparation
Initially it was planned to perform polarised FTIR measurements on the
PET sheets from the series of conventional heat-setting experiments (section
3.5) using the 1018cm-
1
band, after Chalmers et a/
129
This band is a strong
absorption peak, and the thickness of these sheets was sufficient to cause
this band to be outside the acceptable limits of linearity of the detector in
the FTIR in IPTME. To overcome this, it was necessary to take thin
sections (1OJ.Ull) of the films and to perform micro-polarised-FTIR using a
Spectra Tech IR-PLAN microscope attached to a Nicolet 20DXC bench
FTIR, based in Loughborough University's Chemistry Department.
85
Sectioning was perfonned using freshly prepared glass knives, with the
sample frozen in ice, using CO
2
as the chilling medium, to minimise the
amount of extraneous induced orientation. Unfortunately the Nicolet
Instrument has a lower limit of linearity, and so even samples gave
absorbencies that were too high for accurate quantitative measurements. An
attempt to cut sections was made but because the films were already
quite thin the sections were unmanageable. Therefore, this
part of the work was transferred to the PET bottles (see section 3.6.1.), from
films from the Long Stretcher.
sections were cut from the heat-set bottles in both the hoop (major)
and axial (minor) directions. The bottles are much thicker than the
films, making the sections much wider and so far more manageable.
However, as only one bottle size was heat-set, only the effect of heat-setting
temperature could be studied, as the draw area will be constant for the
bottles. It was not considered wise to obtain samples of different draw by
taking them from different parts of the bottle as the large variations in
thickness could well alter the heat-setting effect.
Once the samples were cut, they were sandwiched between two 2mm thick
KBr discs and checked using the microscope for defects such as holes,
tears, or excessive knife marks. If any of these were unavoidable then the
sample was rejected.
3.6.2.2 Use Of The Spectra-Tech IR-Plan Microscope
The general layout of the IR-Plan microscope is shown in figure 3.12. It
consists of transfer optics to bring the infrared radiation from the
interferometer, imaging both the source and the pupils' image (the beam
splitter image) through the microscope. By setting up the microscope by
viewing through a series of small pinholes and then aligning the infrared
radiation through the same holes, the visible optical train is parfocal and
colinear with the infrared.
86
In use, the area of interest is brought to the centre of the microscope's field
as viewed under visible radiation. Adjustable apertures are then used to
delineate the area to be sampled. Since the optical geometry is the same for
the visible evaluation and the infrared detection, the diffraction caused by
the aperture is worse for the detection step as the wavelength is longer.
From this it can be seen that the area to be studied can be viewed more
clearly than it can actually be measured. For FTIR microscopy, spatial
resolution is defmed as the ability to measure the spectrmn from an object
delineated by the apertures without impurity radiation from neighbouring
objects. An impurity can take the form of either spurious narrow band
impurity from a nearby absorber, or spurious broad band impurity from a
nearby hole. The IR-PLAN microscope uses a technique called redundant
aperturing to improve the spatial purity of the system. Two variable
apertures are used: the lower one is placed between the condenser and the
sample, and the upper one between the objective turret and the binocular
assembly. The upper aperture is used to mask off any overspill light from
the lower aperture caused by the lower aperture diffracting the beam.
The IR-PLAN does have a holder for a polariser to be inserted into the
beam but the only IR polariser available would not fit it. The only suitable
place to enter the polariser in the beam was in place of the upper aperture.
This is in fact nearer the sample in the light path than where the polariser
ought to be, thereby reducing further the opportunity of re-orientation by
the optical elements. The relative loss of spatial purity by removing the top
aperture is shown in figure 3.13. Because the samples are homogeneous in
nature this loss was not considered to significantly affect the spectrmn.
Each bottle had two sections taken from it, one parallel to the hoop axis and
the other parallel to the axial axis. Each section had two spectra recorded
for it, one with the polariser oriented parallel to the section direction and
one with the polariser oriented perpendicular to it. Each spectrmn has a
separate background recorded, to negate any polarisation effects of the
instrmnent. In this way, a spectrmn is recorded for the electric vector
oriented in each of the three Cartesian directions with a repeat in the
thickness direction (X.12 and X.22) to allow the data to be normalised
between sections of the same sample, ie. to remove any thickness effects.
87
Sections were cut and analysed from each bottle wall using the method
outlined below:
HOOP AXIS
( )
AXIAL AXIS
l' SECTION
-It THICKNESS
where X. is the bottle number, and the numerical suffixes indicate the
polariser orientation when the spectrum was collected.
88
3.6.2.3 Orientation Functions
The Spruiell orientation functions may be obtained from P-FTIR
measurements from
l29
:
fIB = (Ax-Az).C(Ax+Ay+Az)-1
f2B = (Ay-Az).C(Ax+Ay+Az)-1
where Ax was the absorbency obtained whilst the electric vector of the
infrared radiation polarised parallel to the bottle axis (x). Ay and Az were
similarly defmed as the bottle hoop and thickness directions respectively.
Therefore Ax = X.21
Ay = X. 11
Az = X.12 and X.22
Using these identities the data gave unexpectedly negative orientation
functions. This implied that the net orientation was out of the plane of the
wall. This is contrary to experience for biaxially oriented parts.
Therefore, to investigate further, a sample of PET copolymer was cold
drawn uniaxially to give a sample of high uniaxial orientation. This was
analysed using P-FTIR, and the result indicated that the orientation was
transverse to the draw direction. This led to the conclusion that either the
polariser was incorrectly marked or a 90 rotation had occurred in the IR.
optic train. The above identities were consequently altered to compensate
for this rotation:
Ax=X.22
Ay=X.I2
Az = X.ll and X.2I
89
Spectra were recorded for each bottle and three peak areas were measured
where possible ie. when the amplitude of the peak was less than 1
absorption unit. The three peaks at 1018cm-
l
, 973cm-
l
, and 1578cm-
1
correspond to overall chain orientation in the bulk material, and orientation
in the crystalline and amorphous phases respectively. The identities were
used to generate orientation functions for each set of peaks and are plotted
in figures 4.44 to 4.46.
3.6.2.4 Dichroic Ratios
Unlike the Spruiell orientation function, the dichroic ratio approach does
not allow for changes in sample thickness. As absorbency does vary with
thickness, these variations, largely caused by the samples warming and
expanding in the microtome, must be corrected for. Gupta and Ramesh
83
normalised their absorption bands to the 1506cm-
1
peak to correct for
thickness and to prove the procedure produced analogous results using the
793cm-
1
peak.
Their work assumed that these peaks would not be dependent on the
orientation of their corresponding dipoles. This assumption is incorrect in
the context of this study and so a weighted average of these two peaks was
required. These two peak areas were measured both parallel and
perpendicular for all the samples and plotted against each other. The data
were then regressed and the resulting coefficient (constant = 0) was used to
weight the contribution of each peak. The result of this is a weighted
average of two dipole transitions measured in two directions. Therefore, the
effect of the dipole transition and orientation have been effectively
removed, leaving only thickness. The data for the three peaks used in the
Spruiell orientation functions could also then be plotted as a set of dichroic
ratios using the same approach as Gupta and Ramesh
83
(section 2.3.2.1).
90
- - . - . ~ - . - ~ - - - - . - .. ------.-----.------
Figure 3.1. Biaxial Stretching Rig At IPTME
91
~
-..j

8 .-
. '.
M -
a ..;
I
r:l
)
,
I I
l
Figure 3.2. Biaxial Stretching Rig At IPTME With Heat-Setting
Attachment in Place
92
SIDE J ~ COMPONENTS
;a--
37.50
112,50
150.00
197,00
262,50
300.00
END J ~ COMPONENTS
75,00
112.50
150,00
Figure 3.3. Components For Jaws of Heat Setting Rig
93
SUPPORT FRAME
SIDE MEMBER TVO OFF
-------------------t------------------------ -

, ....
TOP MEMBER 1 OFF
I
M'
\.
------------'
f-
'I!S.OII
BOTTOM MEMBER 1 OFF
I ..
"'"
I

l-
ll! .... !
"" ...
Figure 3.4. Components For Frame of Heat Setting Rig
94
HEAT SETTING RIG
Figure 3.5. Heat Setting Rig with Film in Place. Supporting Frame Omitted
for Clarity
95
HEAT SETTING BLOCK
ASSEMBLY
i
11
I
I
"'le-
=
_.
J.
t:
.-
...
-
I::: .1 ---
-j-
I
10MM ~ STUD
I
V1TH 15l1EM6 TAIL
I
~
I
./'
. : ~
.-
!==l L--.
I
'"
I
I'DRIL 1 AND TAP M6 I
I
I--
0
L
~
1=
1-
.j.
~
~ .. -
- -
..
0
I ...
. .
d
J ---
. ,.
~ ru
I
-
~
I
~
I
"DRILL
: t:16 CLEAB8
I
I
lkJo
I
100.00
155.00
DRILL AND TAP M5
Figure 3.6. Heat Setting Block Working Face Showing Construction Used to
allow Variable Film Size to be Accommodated.
96
NCE
198
196
194
_192


188
f186
.. 184
i\
"\.
\'''\



"'"

r::::::::
"-!
r--.... ....
152
lBO
17B
.......
.........

r-.........
o 20 40 60 80 100
TIme (s)
Figure 3.7. Cooling of Experimental Steel Block
200 t:::..a......

190

'i:


!!
-
0
i
170
....
160
ISO
0 10 20 30 40 50 60
TIm. (s)
Figure 3.S. Cooling of Heat-Setting Block.
97
120
....
Centre
....
Corner
70
j
20.00
Sample 1
. . . . . .., . . . . . I . . . . . T . . . . . I' . . . . "
. . . . . + . . . . . + . . . . . + . . . . . +. . . . . .-j
I I , I ,
I ,
. . . . . -t- . . . . . + . . . . . + . . . . . +. . . . . .-j
Load (tl)
I:
...... + ..... + ..... + ..... + ......
, , , . I.
!
-+. -----t--
, .
. .-
!
, .
, .
-1 I :
i .
I,
0.000 -ll--r---r---r---r----ri -----,c---,.--;--.,---;I
0.000 Time : ~ e < 1220.0
11 .!30
Load (N)
:\: CF:EEP
SrF:,u,IN
0.752'+
.'.
.".
r::.
'.'
O.2::::?::
Sample 1
.-.
i.'
. . . .. .., . . ... I . . . . . T . . . . . I' . . . . "
..... + ..... + ..... + ..... + ...... -j
, , , I
, .
..... + . . ... + . . . . . + . . ... + . . . . . .-1
..... + ..... + ..... + ..... + ...... -j
, , I
9.1 00 +---r-----r----r--r----r---r-----r----r--r----1
0.000
Figure 3.9 Load Control Against Time for Con,tant Nominal
Stress Operation
98
1220,0
SAMPLE
SUPPORT
..............
METTLER TMA40
LID
FURNACE
---
FURNACE
SiAGE
HEIGHT ADJUSTMENT RING
CALIBRATION
/WEIGHT
Figure 3.10 MetIer TMA40 Thermometric with Film and Fibre
Attachment in Place
99
20
-
--
EXPANSION
-
-
//'
-
15
1//
V
RECORDEO
-
..... SHRINKAGE
~ 1
,
V
...
I
C)
//

~
6
~
5
~
V
tIl
~
0
-5
30 40 50 60 70 80 90 100 110
TEMPERATURE (oC)
Figure 3.11. Demonstration of Correction for Expansion In TMA
Experiments.
Where "recorded" is the recorded shrinkage, "Expansion" is the
expansion measured by TMA for a quasi-isotropic sample and
"sbrinkage" is the sbrinkage value after correction for expansion
has been applied.
100
FRCN lA
SOt.RCE
lO'IwEA v AAIA8lE APERTURE
Figure 3.12 Optical And IR Paths Of Spectra-Tech IR Plan
Microscope (from Infrared Microspectroscopy Theory
And Applications, P3, Eds. Messerschmidt. R.G.,
And Harthcock, M.A Pub. Marce1 Dekker Inc . 1988)
101
1.00
I i
,
i
I
{I "'Y\.
'"
I
0.80
!
J'
I
I
I\, i , I
;
\J Redundant Apertures
0.60
>-
I ,1_ I I I !
I
ingle Aperture !J

c:
Cl>
0.40
:5
I I
I
_1 I
I
I I !
;

I I

I i
I I I I
1
0.20
I 1
I
I I I
I I I
I
I
I
I
I ! I
I I
,
I
1
I I
, , ,
0.00
80 40 o 40 80 120 160
Spatial Position (arbitrary units)
Figure 3.13 Intensity Distribution At An Edge
The Loss Of Spatial Purity When Using Only The
Lower' Aperture (from Infrared Microspectroscopy
Theory And Applications, P6. Eds. Messerschnddt. R.G .
and Harthcock. M.A . Pub. Marcel Dekker Inc. 1988)
102
Table 3.1. lel 'Melinar'TM Bottle Grade Poly(ethyleneterephthalate)
Materials
Polymer Resin Conversion Applications Typical End Use Food
Identity IV Route
Acetaldehyde Contact
Level Achieved
B90N 0.82 I stage injection Bottles
7-9 ppm in Approved
stretch blow. Jars
preform
Selected 2 stage In-use temperature limit 60C
stretch blow (approx.)
moulding
Laser 0.81 2 stage injection Bottles 7-9ppm in Approved
stretch blow Jars preform
Selected I stage Specially formulated for
injection stretch I piece Petalloid bottles
blow moulding In-use temperature limit 600C
(approx.)
B90S 0.78 2 stage injection Bottles 6-8 ppmin Approved
stretch blow Jars preform
Selected I stage In-use temperature limit 60C
injection stretch (approx.)
blow
B90W 0.75 2 stage injection Bottles 3-4 ppm Approved
stretch blow Jars in preform
moulding Specially formulated for:
Injection blow I) Low acetaldehyde
moulding 2) Thick wall preforms
(max.6mm)
In-use temperature limit 60C
(approx.)
B90C 0.76 I stage injection Bottles 6-8ppm in Approved
stretch blow Jars preform
moulding Specially formulated for
cosmetic and toiletries
containers using very thick
wall preforms (max. 7mm)
In-use temperature limit 60C
(approx.)
B90N 0.74 I stage injection Bottles 8-12 ppm in Approved
colour min stretch blow. Jars preform
blend Selected 2 stage Amber and green colour
stretch blow pre-blended polymer
moulding In-use temperature limit 60C
(approx.)
B90S 0.74 2 stage injection Bottles 8-12 ppm in Approved
Colour min stretch blow Jars preform
Blend moulding Amber and green colour
Selected I stage blended polymer.
injection stretch In-use temperature limit 60C
blow moulding
.(approxt
103
Table 3.2. FED Heat Setting Programme
Run Coded Values Run Real Values
No No
Draw Temp Time Draw Temp Time
(ocj (s)
1 -I -I -1 1 6.4 104 6
2 I -I -I 2 9.7 104 6
3 -I I 1 3 6.4 176 6
4 I I 1 4 9.7 176 6
5 -I 1 I 5 6.4 104 24
6 I -I 1 6 9.7 104 24
7 1 I I 7 6.4 176 24
8 I I 1 8 9.7 176 24
9 a 0 0 9 11.0 140 15
la -a 0 0 10 4.5 140 15
11 0 a 0 11 7.5 200 15
12 0 a 0 12 7.5 80 15
13 0 0 a 13 7.5 140 30
14 0 0 a 14 7.5 140 0
15 0 0 0 15 7.5 140 15
16 0 0 0 16 7.5 140 15
17 0 0 0 17 7.5 140 15
18 0 0 0 18 7.5 140 15
19 0 0 0 19 7.5 140 15
104
Table 3.3. Experimental Values For FED Films
Homopolymer
Run Ideal Actual Major"Minor Heat-Set
Number Draw Draw Axes Temperature
1 6.4 6.20 3.10*2.00 104
2 9.7 9.50 3.80*2.50 104
3 6.4 6.40 3.20*2.00 176
4 9.7 9.69 3.80*2.55 176
5 6.4 6.34 3.25*1.90 104
6 9.7 9.50 3.80*2.50 104
7 6.4 6.36 3.10*2.05 176
8 9.7 9.80 3.80*2.60 176
9 11.0 11.07 4.10*2.70 140
10 4.5 4.33 2.55*1.70 140
11 7.5 8.40 3.50*2.40 200
12 7.5 8.00 3.35*2.40 80
13 7.5 7.58 3.37*2.35 140
14 7.5 7.99 3.40*2.40 140
15 7.5 8.10 3.40*2.40 140
16 7.5 7.99 3.40*2.35 140
17 7.5 7.94 3.45*2.30 140
18 7.5 7.80 3.20*2.30 140
19 7.5 7.99 3.40*2.75 140
Copolymer
Run Ideal Actual Major"Minor Heat Set
Number Draw Draw Axes Temperature
1 6.4 6.20 3.10*2.00 104
2 9.7 9.60 3.85*2.49 104
3 6.4 6.20 3.1*2.00 176
4 9.7 10.10 3.9*2.60 176
5 6.4 6.40 3.2*2.00 104
6 9.7 9.70 3.9*2.50 104
7 6.4 6.40 3.2*2.00 176
8 9.7 9.50 3.8*2.50 176
9 11.0 11.07 4.1*2.70 140
10 4.5 4.42 2.6*1.70 140
11 7.5 7.65 3.4*2.25 200
12 7.5 7.65 3.4*2.25 80
13 7.5 7.48 3.4*2.20 140
14 7.5 7.52 3.5*2.15 140
15 7.5 7.30 3.4*2.15 140
16 7.5 7.30 3.4*2.15 140
17 7.5 7.87 3.5*2.55 140
18 7.5 7.87 3.4*2.25 140
19 7.5 7.87 3.5*2.25 140
105
Table 3.4. Experimental Plan For Conventional Experimentation Films
Run Ideal Actual Major"Minor Heat-Set
No Draw Draw Axes Temoerature
1 11 10.94 4.05*2.70 lOO
2 11 11.06 4.05*2.73 160
3 11 11.07 4.10*2.70 220
4 10 9.85 3.90*2.53 140
5 10 9.82 3.85*2.55 160
6 10 10.19 3.88*2.63 175
7 10 9.82 3.85*2.55 190
8 10 10.13 3.83*2.63 205
9 10 10.01 3.85*2.60 220
10 9 8.74 3.65*2.40 lOO
11 9 9.16 3.70*2.48 160
12 9 9.\3 3.65*2.50 220
13 7 6.66 3.25*2.05 lOO
14 7 6.80 3.40*2.00 160
15 7 7.12 3.35*2.12 220
16 5 4.81 2.75*1.75 lOO
17 5 5.09 2.75*1.85 140
18 5 5.18 2.80*1.85 160
19 5 5.\3 2.85*1.80 160
20 5 5.09 2.80*1.80 175
21 5 4.95 2.83*1.75 190
22 5 5.07 2.85*1.78 205
23 5 4.95 2.75*1.80 220
24 5 4.95 2.75*1.80 220
25 3 2.85 2.00*1.43 lOO
26 3 2.76 1.90*1.45 160
27 3 2.90 2.00*1.45 220
28 I 1.00 1.00*1.00 lOO
29 I 1.00 1.00*1.00 160
30 I 1.00 1.00*1.00 220
31 11 10.87 4.10*2.65 220
32 11 11.20 4.00*2.80 220
33 11 10.60 4.00*2.65 not heat set
34 10 10.05 3.85*2.61 not heat set
35 9 8.69 3.65*2.38 not heat set
36 5 4.95 2.78*1.80 not heat set
37 3 3.16 2.18*1.45 not heat set
38 1 1 1.00*1.00 not heat set
Run 31 was heat set then shock cooled.
Run 32 was heat set, relaxed, then re heat set.
10(
4 RESULTS
The project has generated a considerable amount of data and to aid the
reader it has been presented in the following way. Data From the FED
work has been tabulated in appendix A, the data from the conventional
experimentation series are tabulated in appendix B, and the data from the
bottles are tabulated in appendix C. FED models derived from the raw data
are presented in the main text of this chapter, all graphs and sample values
from the FED models may be found at the end of this chapter. Comments
on the results can be found in this chapter but full interpretation of the
results can be found in chapter 5. Inclusion of variables and combination of
variables is on the basis of improvement of the fit of the model, i.e. omitted
variables if included would not significantly improve the R2 value of the
model.
4.1 FED - STATISTICAL ANALYSIS AND MODEL
PREDICTIONS
If a data set is considered as a set of points in a 3D matrix with the three
predictor variables as the principle axes, then experimental correlation
between the variables will distort the matrix by no longer making the axes
orthogonal. If orthogonality can be restored to the data set, then correlation
between variables will have been effectively removed. Variable correlation
should not be confused with variable interaction: correlation is due to
experimental design, apparatus etc., whereas interaction is a legitimate
experimental effect. The following is an orthogonalising transformation and
was applied to each data set prior to analysis:
x = transformed variable
x = experimental value
it = average experimental value
SD = standard deviation
107
The resulting model coefficients must be transformed back to the original
form. For this to be dimensionally independent, the model must be
hierarchically stable; that is, all lower terms of an included variable must be
present in the model, regardless of significance. The output of the FED
analysis is shown in figure 4.1a. The relevant information and defmitions
are summarised below:
Variable:
A list of predictor variables and combinations used in
the model
Parameter estimate: An estimate of coefficient for each predictor variable
and combination
Standard error:
Prob>F:
The estimate of each parameter is accurate to twice
the standard error, at a 95% confidence level.
This is a measure of the significance of each predictor
variable. For Prob>F=O.OI there is a 1% chance that
the value of the variable could be zero. Therefore the
smaller this value is the 'better known' the variable is.
The highest probability that a variable is not zero that
can be registered is 99.9%, Prob>F=O.OOl.
R-squared (R2): It is measure of the goodness of fit of the model. A
model that will predict 90% of the data from which it
was derived, at a confidence level of 95%, will have an
R2=90%. The R2 for the shrinkage models are plotted
against temperature in figure 4.1b. It can be seen that
increasing the test temperature generally increases R2.
The little shrinkage that occurs at 65C does appear to
increase the R2 value from what would be expected.
Because of the nature of the FED analysis, the results are presented as
model equations in the form of a truncated Taylor series, (see below), and
their corresponding contour plots are also presented. The raw data have
been included in tabulated form (appendices Al to A24) but because the
variables are not varied in a simple fashion they are of only limited use in
interpretation.
108
Truncated Taylor Series are in the general fonn:
Vp = A1X
1
+ A
2
X
2
+ A3X3 + AIIX1X
1
+ A
I2
X
1
X
2
+ A
I3
X
1
X
3
+ A
22
X
2
X
2
+
A
23
X
2
X
3
+ A33
X
3
X
3
where V p is the variable to be predicted, coefficients Aij are experimentally
determined and represented the predictor variables.
4.1.1 MECHANICAL TESTING DATA
Three dwnbbell specimens (28mm gauge length, 3.4 nun gauge width),
were cut parallel to each draw axis (major and minor) and tested to
destruction using a JJ Lloyd R2000 tensile testing machine with a 50N load
cell. The data for the major axis were processed manually, from X-V load-
extension plots whilst the minor axis data were processed by computer. The
raw mechanical data are in appendices Al to A8 and the corresponding
stress - strain data are tabulated in A9 to A19. All the measured variables
have been defmed in section 3.4.1.1.
4.1.1.1 Homopolymer
(A) Yield stress - major axis
O"y(O) (MPa) =
R2= 88%
+ 50.1632
+ 0.130 (Temp)
- 2.278 (Area)
+ 1.1387 (time)
- 0.0062 (Temp*time)
- 0.194 (Temp*Area)
+ 0.0414 (time
2
)
+ 0.8000 (Area
2
)
109
For a heat-setting time of 6 seconds (figure 4.2a) increasing the setting
temperature from 80
0
e to 200
0
e is predicted to give a linear increase in
yield stress from 111 MPa to 125 MPa, for a draw area of 10 and from 68
MPa to 80 MPa for a draw area of 5.
For a heat-setting time of 15 seconds (figure 4.2b) the yield stress is
increased from 107 MPa to 113 MPa for a draw area of 10 and from 74
MPa to 78 MPa for a draw area of 5 when the setting temperature is
increased from 80
0
e to 200
o
e. It should be noted that for equivalent heat-
setting temperatures, increasing the setting time reduces the yield stress for
a draw of 10 and also for all but the highest setting temperatures, for a draw
area of 5.
When the heat-setting time is extended to 30 seconds, figure 4.2c,
increasing the setting temperature from 80
0
e to 200
0
e decreases the
predicted yield stress from 117 MPa to 109 MPa for a draw area of 10 and
from 97 MPa to 90 MPa for a draw area of 5, i.e. heat-setting at lower
temperatures is more effective when the heat-setting time is increased.
(B) Yield stress - minor axis
cry(90) (MPa) =
R2=39%
+6.996
+ 18.369 (Area)
- 0.981 (Area
2
)
This model indicates that heat-setting has no effect on the yield stress in the
minor axis, although with an R2 value of 39%, care must be taken in
reaching this conclusion, as the model is a very poor one.
110
(C) Yield strain-major axis
R2=43%
- 11.276
+ 0.152 (Temp)
+ 1.631 (Area)
+ 0.2159 (time)
- 0.000327 (Temp2)
- 0.006 (Temp*Area)
- 0.03 (time*Area)
The yield strain plots (figure 4.3a to c) indicate that in the major axis, the
more highly drawn sample has a higher yield strain than the lower draw.
This, at first may seem contrary to expectation, but it must be remembered
that the higher drawn sample will have a far higher yield stress. Assuming a
similar modulus for both samples, then the larger yield strain for the higher
drawn sample would be expected. The plot also shows yield strain is
predicted to go through a maximum at 140C, for all cases other than the
major axis at a draw area of 5 (where the maximum is higher, at
approximately 185C). Although neither of the models (major and minor)
are good fits, they do all agree qualitatively and so cannot easily be ignored.
(D) Yield strain - minor axis
Ey(90)(%) =
R2 = 60%
+ 4.3152
+ 0.0864 (Temp)
- 0.193 (time)
- 0.179 (Area)
- 0.0003 (TempZ)
+ 0.0053 (time
2
)
111
The minor axis model shows some dependence on time (figures 4.3a to c).
For a heat-setting time of 6 seconds, both draw areas (5 and 10) behave in
the same manner, except that the draw area of 5 shows about 0.25% less
strain overall. Increasing the setting time increases this difference to about
1%, for both setting times of 15 and 30 seconds. Heat-setting at 30 seconds
also increases the maximum yield strain to over 8.5% for a draw area of 10
and to 7.7% for a draw area of 5. An R2 value of 60% prevents any more
detailed analysis.
(E) Ultimate tensile stress - major axis
O"U(O) (MPa) =
R2 = 56%
+ 158.646
- 0.563 (Temp)
- 6.49 (Area)
- 0.015 (Temp*time)
+ 0.1035 (Temp*Area).
+ 1.89 (time)
From the plot of ultimate tensile stress for a heat-setting time of 6 seconds
(figure 4.4a) it can be seen that heat-setting can either increase or decrease
the ultimate tensile stress depending on draw area (increasing heat-setting
temperature from 80C to 200C increases the UTS from 137 MPa to 181
MPa for a draw of 10, but will also reduce UTS from 129 MPa to 110 MPa
for material with an area draw of 5). It may be unfortunate that for the
lower draw areas, the ultimate tensile stress decreases with increasing heat-
setting temperature, since it is these areas that benefit from heat-setting with
respect to shrinkage. However, a material's maximum useful stress in this
application is not its UTS, rather its yield stress, and the models do show
this to be improved, by varying degrees, in all cases.
By increasing the heat-setting time to 15 seconds (figure 4.4b) the model
predicts that the ultimate tensile stress will increase from 142 MPa to 172
MPa when the setting temperature is increased from 80C to 200C for a
draw of 10, but the UTS decreases from 134 MPa to 100 MPa, for a draw of
5.
112
Heat-setting for 30 seconds (figure 4.4c) causes an increase in setting
temperature from 80C to 200C to have a minimal predicted effect on UTS
for a draw of 10 but for a draw of 5, the UTS falls from 144 MPa to 84
MPa. Once again, the fit is not a good one and so this information can only
be used as a rough guide.
(F) Ultimate tensile stress - minor axis
cru(90) (MPa) =
R2=68%
+51.214
+ 0.1411 (Temp)
+ 0.994 (time)
+ 1.995 (Area)
- 0.007 (Temp*time)
Heat-setting for 6 seconds (figure 4.4a) increases the ultimate tensile stress
from 86 MPa to about 96 MPa for a draw area of 10, and from 75 MPa to
86 MPa for a draw of 5, for an increase in setting temperature from 80C to
200C.
For a heat-setting time of 15 seconds (figure 4.4b), increasing the heat-
setting temperature from 80C to 200C brings about a small increase in
UTS from 88 MPa to 93 MPa for a draw area of 10, and from 79 MPa to 83
MPa for a draw area of 5.
When the heat-setting time is extended to 30 seconds (figure 4.4c), on
increasing the setting temperature from 80C to 200C, the UTS is predicted
to reduce from 96 MPa to 88 MPa for a draw of 10, and from 86 MPa down
to 78 MPa for a draw area of 5.
113
(G) Ultimate tensile strain - major axis
R2=85%
+ 270.47
+ 0.529 (Temp)
- 24.16 (Area)
- 3.1599 (time)
+ 0.015 (Temp*time)
+ 0.444 (time* Area)
- 0.098 (time
2
)
- 0.003 (Temp2)
Ultimate tensile strain in the major axis shows only a small variation with
heat-setting temperature (figure 4.5). The low draw ratio samples (DA=5)
are predicted to exhibited ultimate strains between 140% and 160%, whilst
the high draw area samples (DA=IO) are predicted to show ultimate strains
in the range of 55% to 70%.
(H) Ultimate tensile strain - minor axis
+ 336.505
- 23.697(Area)
R2 = 52%
Ultimate tensile strain in the minor axis is modelled as being independent of
heat-setting temperature (figure 4.5) but the model is not a good one and so
once again caution should be exercised in making such statements. The
lower draw samples again show higher ultimate strains (100% and 218%
respectively). This is reversed as compared to the yield strains,
strengthening the argument that this reversal in yield strain values is due to
the high drawn samples achieving higher yield stresses.
114
(I) Strain hardening rate - major axis
SHR(MPa)=
R2 = 86%
+ 43.101
- 0.344 (Temp)
- 9.094 (Area)
+ 4.900 (time)
- 0.01930 (Temp*time)
+ 0.0805 (Temp * Area)
- 0.4326 (time*Area)
+ 0.0365 (time
2
)
+ 0.968 (Area
2
)
The effect of heat-setting temperature on strain hardening rate is primarily
dependent upon draw area (figure 4.6). An increase in heat-setting
temperature from 80
0
e to 200
0
e reduces strain hardening rate from 53 MPa
to 26 MPa for a draw area of 5, but will increase it from 79 MPa to 100
MPa, for a draw area of 10. It is therefore the highly-drawn samples which
benefit further from increased heat-setting temperature, in this respect.
(J) Strain hardening rate - minor axis
SHR(MPa)=
No model due to poor correlation.
115
4.1.1.2 Copolymer
(A) Yield stress - major axis
O'y(O) (MPa) =
+54.46
+0.163 (Temp)
- 2.394 (Area)
- 0.004 (Temp*time)
+ 0.0266 (time
2
)
+ 0.523 (Area
2
)
- 0.123 (time)
For a heat-setting time of 6 seconds (figure 4.7a), on increasing the setting
temperature from 80C to 200C the yield stress is increased from 94 MPa
to 111 MPa for a draw of 10, and from 67 MPa to 84 MPa for a draw of 5.
For a heat-setting time of 15 seconds (figure 4.7b) increasing the heat-
setting temperature from 80C to 200C is predicted to increase the yield
stress linearly from 68 MPa to 81 MPa and from 96 MPa to 107 MPa, for
draw areas of 5 and 10 respectively. This is very similar to the
homopolymer, but the overall yield stresses are slightly lower for the
copolymer.
When heat-setting is carried out for the duration of 30 seconds (figure 4.7c)
on increasing the setting temperature from 80C to 200C, the yield stress is
predicted to increase from 107 MPa to 113 MPa for a draw of 10, and from
78 MPa to 84 MPa for a draw of 5.
The increase of heat-setting time can be seen to increase the yield stress at
lower setting temperatures, but increasing the temperature has far less effect
at longer heat-setting times.
116
(B) Yield stress - minor axis
O'y(90) (MPa) =
No Model due to lack of correlation and
hierarchical instability.
(C) Yield strain - major axis
R2 = 12%
+ 7.067
- 0.005 (time
2
)
+ 0.153 (time)
This model is too poor to merit further discussion ..
(D) Yield strain - minor axis
E
y
(90) (%) =
R2=43%
+ 3.144
+ 0.095 (Temp)
- 0.3076 (Area)
- 0.000313 (Temp2)
The yield strain in the minor axis (figure 4.8) is predicted to only slightly
increase with heat-setting temperature. This behaviour is very similar to the
homopolymer.
117
(E) Ultimate tensile stress - major axis
O'u(O) (MPa) =
+ 162.164
- 0.864 (Temp)
- 2.4667 (Area)
+ 0.0022 (Temp2)
+ 0.0635 (Temp Area)
R2=63%
The ultimate tensile stress is increased for both draws, from 120 MPa to 130
MPa, and from 135 MPa to 180 MPa for a draws of 5 and 10 respectively,
for an increase in heat-setting temperature from 80C to 200C, (figure 4.9a
to b). The model predicts ultimate tensile stress to be independent of heat-
setting time, i.e. figures 4.9a and b are very similar.
(F) Ultimate tensile stress - minor axis
0'u(90) (MPa) =
R2 = 75%
+ 33.25
+ 0.076 (Temp)
+ 0.2413 (time)
+ 8.4707 (Area)
- 0.497 (Area
2
)
In the minor axis the effect of heat-setting temperature is far less
pronounced but there is a small increase in ultimate tensile stress with heat-
set temperature.
Heat-setting for 6 seconds (figure 4.9a) causes the ultimate tensile stress to
be increased from 78 MPa to 84 MPa and from 71 MPa to 80 MPa for
draws of 10 and 5 respectively, when the setting temperature is raised from
80C to 200C.
118
For a heat-setting time of 15 seconds (figure 4.9b) increasing the setting
temperature from 80C to 200C increases the ultimate tensile stress from
79 MPa to 84 MPa and from 73 MPa to 82 MPa, for draws of 10 and 5
respectively.
Heat-setting for 30 seconds (figure 4.9c) continues this trend, with increases
of 82 MPa to 91 MPa and from 78 MPa to 83 MPa for a draw of 10 and 5
respectively.
(G) Ultimate tensile strain - major axis
Eu(o(%) =
R2 = 87%
+411.655
- 1.255 (Temp)
- 36.646 (Area)
+ 0.13 (Temp)(Area)
+ 2.838 (time)
- 0.096 (time
2
)
For a draw area of 10 the copolymer shows a very small increase in ultimate
strain, from 70% to 78%, by increasing heat-setting temperature from 80C
to 200C (figure 4.10). For a draw area of 5 the same increase in heat-
setting temperature reduces the ultimate strain from just over 200% to
130%.
(H) Ultimate tensile strain - minor axis
No model due to very poor correlation.
119
(I) Strain hardening rate - major axis
SHR(MPa)=
R2=82%
- 38.243
- 0.525 (Temp)
+ 16.229 (Area)
- 0.3995 (time)(Area)
+ 3.1380 (time)
+ 0.0023 (Temp2)
------------
The draw dependence on the type of behaviour seen in the homopolymer is
not reproduced with the copolymer (compare figures 4.6 and 4.11). The
data are quantitatively similar, with greater strain hardening rates apparent
in the major draw axis, but the overall form of the trends (with respect to
heat-setting temperature) is different, in each case. -
4.1.2 Creep Models
Creep properties were measured using the 'constant nominal stress' mode of
the JJ Lloyd R2000 tensile testing machine at 40 MPa, at a temperature of
65C for 20 minutes loading (section 3.4.1.2).
Due to material constraints creep was only measured in the hoop axis
(major draw axis). Treating a bottle as a thin wall cylinder, the hoop stress
is greater than the axial stress and so the test is in the most critical direction.
The raw data used to generate these models are presented in appendix A19.
120
4.1.2.1 Creep strain - Homopolymer
R2 = 89%
+ 119.1223
- 1.3880 (time)
- 23.2778 (Area)
+ 0.15497 (time*Area)
+ 1.11807 (Area
2
)
This model has a high correlation coefficient and predicts that creep strain
in the homopolymer with an area draw of 5 can be reduced from 31% to
12% by heat-setting the material for 30 seconds at any temperature between
80C and 220C, figure 4.12 (the FED experimental temperature limits).
Note that this is the first occasion that time appears to be the most
significant variable, and that temperature is not significant. For a draw area
of 10, creep strain is predicted to be much lower than for a draw of 5, but is
predicted to increase slightly with increasing heat-setting temperature.
4.1.2.2 Creep strain - Copolymer
R2=93%
+ 214.007
- 1.4239 (time)
- 44.451 (Area)
+ 0.1487 (time*Area)
+ 2.2957 (Area
2
)
The model for the copolymer, predicts a reduction of creep strain from 50%
to 29% by heat-setting for 30 seconds (figure 4.12) for material with an
equivalent draw ratio of 5. The high draw creep behaviour is very similar to
the homopolymer.
Sample values for all the mechanical and creep models are tabulated in
tables 4.1 and 4.2.
121
4.1.3 Shrinkage Models
The shrinkage measurements made by the TMA were first corrected for
expansion ( section 3.4.2) and then shrinkage at four reference temperatures
(65
o
e, 85e 100
0
e and 110C) were used to generate statistical models for
shrinkage, in both principal film directions. As in the case of the
mechanical results, the equations and their corresponding contour plots
provide a far more useful summary of the results than the raw data
(appendices A20 to A23). The contour plots for the models for each major
axis have been included to demonstrate the relationships between the
contour plots and their X-Y derivatives.
4.1.3.1 PET Homopolymer
(A) Shrinkage at 65C - major axis
Shrinkage (%) = - 2.19859
+ 0.010508 (Temp)
+ 0.41103 (Area)
- 0.000037017 (Temp2)
- 0.0215 (Area
2
)
At 65e, PET is approximately 10
0
e below its Tg, and so little relaxation
would be expected. This is reflected in the model's rather poor fit
(R2=51%) and the corresponding plot (figure 4.14), shol'iS very low values
of shrinkage, overall.
(8) Shrinkage at 65C - minor axis
Shrinkage (%) = No model due to poor correlation.
122
(C) Shrinkage at 85C - major axis
Shrinkage (%) =
R2 = 88%
- 7.482848
+ 0.030842 (Temp)
- 0.04221 (time)
+ 1.515969 (Area)
- 0.000042431 (Temp2)
- 0.060017 (Area
2
)
- 0.002252 (Temp*Area)
+ 0.001436 (time
2
)
For a draw area of 10, heat-setting reduces shrinkage from about 1.7% to
1.3% (figure 4.15a to d). All of this reduction in the level of predicted
shrinkage occurs at heat-setting temperatures above 120C. The major axis
model for material with a draw area of 5 predicts heat-setting to increase
shrinkage from -0.4% to 0.6%. Whilst the magnitude of the effect is not
important it is of interest that a negative value is recorded in such material
tested in the major axis, but not in the minor axis, which will have lower
linear draw. This can be seen in both materials at all the test temperatures
above 65C.
(D) Shrinkage at 85C - minor axis
Shrinkage (%) =
R2= 61%
+ 2.121292
- 0.016712 (Temp)
+ 0.49287 (Area)
+ 0.000053674 (Temp2)
- 0.000315 (Temp*time)
+ 0.046315 (time)
123
The minor axis model predicts similar reductions in shrinkage for high and
low draw samples to those predicted for a draw area of 10 in the major axis,
figure 4.15. The model predicts that the two draw areas (5 and 10) behave
in parallel fashion with respect to heat-setting temperature, with samples
with a draw area of five exhibit about 0.25% less shrinkage than samples
with a draw area of 10. This model does show some dependence on heat-
setting time; figures 4.15a b, and d. The model predicts that by increasing
heat-setting time from 6 to 30 seconds the minimum shrinkage achievable
can be reduced from 1.1% to 0.6% (for a draw area of 5) and from 1.3% to
0.8% (for a draw area of 10). The heat-setting temperature at which this is
achieved will increase from 170C for a heat-setting time of 6 seconds, to
220C when the heat-setting time is extended to 30 seconds.
(E) Shrinkage at 100"C - major axis
Shrinkage (%) =
R2 =88%
- 22.124391
+ 0.121944 (Temp)
+ 0.040511 (time)
+ 3.377175 (Area)
- 0.016534 (Temp*Area)
+ 0.003129 (time
2
)
- 0.017886 (time*Area)
Increasing the heat-setting temperature from 80C to 220C for a sample
with a draw area of 10 reduces predicted shrinkage at 100C from 6.8% to
0.8%. A similar increase in heat-setting temperature is predicted to increase
shrinkage from -2% to 3.4% for samples with a draw area of 5 (figure 4.16a
to d).
124
(F) Shrinkage at 1000e - minor axis
Shrinkage (%) =.
R2=67%
+ 3.474284
- 0.016595 (Temp)
+ 0.209616 (time)
+ 0.219532 (Area)
- 0.001583 (Temp*time)
This model has only a reasonable fit (R2=67%) and so care should be
exercised when using this model for making predictions and any analysis
must be only ofIimited use. For a draw of 10, heat-setting is now predicted
to reduce shrinkage from nearly 7% to 0% in the major axis and from 3.7%
to -0.8% in the minor axis, for a heat-setting time of 15 seconds. The effect
of heat-setting time is quite complex; an increase 6 to 30 seconds (figure
4.16a to d) increases shrinkage for low temperatures (80C), but the
material becomes more sensitive to heat-setting temperature. Therefore,
heat-setting at high temperatures for a longer period of time is predicted to
reduce shrinkage but under these conditions the prediction is for negative
shrinkage i.e. expansion.
(G) Shrinkage at llOoe - major axis
Shrinkage (%) =
R2=85%
- 38.989445
+ 0.226606 (Temp)
+ 5.836001 (Area)
- 0.031699 (Temp*Area)
125
Heat-setting at temperatures up to 220C is predicted to reduce shrinkage at
110C from 12% to below 0%, for material drawn to an area of 10 in the
major axis, which is a quite dramatic effect (figure 4.17a to d). However,
the model predicts an increase in shrinkage from -4% to 5% for a similar
increase in heat-setting temperature for samples with a draw area of 5. It
should be noted that this is slightly inconsistent with the previous model for
heat-setting temperatures above 190C, as it indicates a lower shrinkage for
a higher test temperature, albeit by less than 1 %.
(H) Shrinkage at 110C - minor axis
Shrinkage (%) =
R2=73%
+ 1.809198
- 0.073392 (Temp)
+ 1.654492 (Area)
+ 0.000414 (Temp2)
- 0.00177 (Temp*time)
- 0.008374 (Temp*Area)
- 0.007168 (time*Area)
+ 0.317987 (time)
This model does show some dependence on temperature. For a heat-setting
time of 6 seconds, (figure 4.17a) increasing the heat-setting temperature
from 80C to 150C reduces shrinkage from 4.5% to 2.2%, for a draw area
of 5. Further, increasing the heat-setting temperature is predicted to
increase shrinkage again. For a draw area of 10, shrinkage can be reduced
from over 9% down to 5% by heat-setting at 220C.
Heat-setting for 15 seconds (figure 4.17b) magnifies the trends seen for the
6 seconds setting period. Importantly, the heat-setting temperature for
minimum shrinkage in samples with a draw area of 5, is increased over
180C which is much closer to the optimum temperature for material with a
draw area of 10.
126
Heat-setting for 30 seconds (figure 4.17d) again continues the trend; the
minimum shrinkages are 1.9% and -0.2% for draw areas of 5 and 10
respectively. Increased heat-setting temperatures now no longer increase
shrinkage, in samples with a draw of 5.
4.1.3.2 PET Copolymer
(A) Shrinkage at 65C - major axis
Shrinkage (%) =
R2=63%
- 0.599205
+ 0.196881 (Area)
- 0.007986 (Area
2
)
(B) Shrinkage at 65C - minor axis
Shrinkage (%) =
R2 = 72%
- 0.738
+ 0.002868(Temp)
+ 0.164053(Area)
- 0.000009966(Temp2)
- 0.007598(Area
2
)
As with PET homopolymer, at 65C there is no significant shrinkage in the
copolymer material (figure 4.18).
127
(C) Shrinkage at 85C - major axis
Shrinkage (%) = - 4.384203
+ 0.029401 (Temp)
+ 0.024529 (time)
+ 0.691053 (Area)
- 0.000171 (Temp*time)
- 0.003392 (Temp*Area)
(D) Shrinkage at 85C - minor axis
Shrinkage (%) =
R2 =33%
+ 1.338341
- 0.004316(Temp)
+ 0.081627(Area)
At 85C (figure 4.19a and b) heat-setting reduces shrinkage in all cases
except for the surprising cases (as discussed earlier), of a draw area of 5 in
the major axis. The overall shrinkage values are similar to those predicted
for the homopolymer. The contrast in fits (91% as compared to 33% for the
homopolymer) between the two axes is of interest.
(E) Shrinkage at 100C - major axis
Shrinkage (%) =
R2=95%
- 18.37236
+ 0.110319 (Temp)
+ 2.556768 (Area)
- 0.013765 (Temp*Area)
- 0.000365 (Temp*time)
+ 0.058245 (time)
128
Increasing the heat-setting temperature from 80C to 220C (figure 4.20a
and c) decreases the predicted shrinkage from 5.5% to below 1% for a draw
of 10 but increases shrinkage from approximately -1.8% to 3.2%, for a draw
area of 5.
(F) Shrinkage at lOOOC - minor axis
Shrinkage (%) = + 6.838637
- 0.072693 (Temp)
+ 0.128144 (time)
+ 0.170166 (Area)
+ 0.000212 (Temp*Temp)
- 0.000844 (Temp)(time)
Heat-setting for 6 seconds (figure 4.20a ) reduces shrinkage from 4.4% to
2.2% when the heat-setting temperature is raised from 80C to 185C (for a
draw area of 10), after which it rises again to 2.4%, when heat-set at 220C.
Samples with a draw area of 5 are predicted to behave in the same way, but
with about 0.75% less shrinkage overall.
Heat-setting for 15 seconds (figure 4.20b and c) magnifies the effect, with
the shrinkage when heat-set at 80C increasing to 5%, but reducing to less
than 2% when the heat-setting temperature is increased from 80
0
C to 200
0
C
(for a draw area of 10). Material with a draw area of 5 is predicted to have
the same behaviour but about 0.8% less shrinkage i.e. for the same increase
in heat-setting temperature, shrinkage is reduced from 4.1 % to 1%.
Heat-setting at 30 seconds (figure 4.20d) gives the same behaviour, but for a
draw area of 10, shrinkage is reduced from 5.8% to 1.2% by increasing
heat-setting temperature from 80C to 220C. Material with a draw of 5 is
predicted to follow this trend but with about 0.8% less shrinkage overall.
129
(G) Shrinkage at 110C - major axis
Shrinkage (%)=
R2=95%
- 33.504142
+ 0.232634 (Temp)
+ 3.93826 (Area)
+ 0.107636 (time)
- 0.000722 (Temp*time)
- 0.029381 (Temp*Area)
+ 0.079811 (Area
2
)
The model for the major axis predicts a decreases in shrinkage from nearly
10% down to below 0% for a draw area of 10, but it also predicts an
increase in shrinkage from -4% to just over 6% for a draw area of 5, when
the heat-setting temperature is increased from 80
0
e to 220
0
e (figure 4.21a
to d).
(H) Shrinkage at 110C - minor axis
Shrinkage (%) =
R2 = 76%
+ 10.313608
- 0.1195 (Temp)
+ 0.358538 (Area)
+ 0.000341 (Temp2)
- 0.001515 (Temp*time)
+ 0.230560 (time)
This model (figure 4.21a to d) behaves very similarly to the 1000e model in
all respects except that, as expected, the overall shrinkage level is higher
(6.5% for a draw area of 10) when heat-set at 80
0
e for 6 seconds.
In addition to these observations it can be seen from these plots that the
predicted shrinkage for materials of differing draws converge as the heat-
setting temperature is increased. Many draw ratios are found in SBM
bottles and so a high heat-setting temperature is predicted to have a lower
and more uniform shrinkage reducing distortion of the bottle shape.
Sample shrinkage values can be found in table 4.3.
130
4.1.4 FED - Crystallinity Measurements
4.1.4.1 Thermal Analysis (DSC)
The raw DSC data are presented in appendix A24.
A) PET
Crystallinity (%) = - 30.5586
R2 = 74%
+ 0.3944 (Temp)
+ 1.5979 (time)
+ 7.2004 (Area)
- 0.0094 (Temp*time)
- 0.0315 (Temp Area) .
- 0.1633 (time* Area)
+ 0.0344 (time*time)
For a heat-setting time of 6 seconds (figure 4.22a to c) increasing the heat-
setting temperature from 80C to 220C increases the predicted crystallinity
linearly from 26% to 52% for a draw area of 5 and from 44% to 47% for a
draw area of 10. When the heat-setting time is increased to 15 seconds,
crystallinity is increased from 33% to nearly 46% for a draw of 5 but the
same increase in setting temperature unexpectedly reduces the predicted
crystallinity linearly from 43% to 35% for a draw of 10. This trend is
continued for a heat-setting time of30 seconds but now the same increase in
setting temperature reduces, linearly, predicted crystallinity for both draw
areas, from 56% to 49% and from 55% to 27% for a draw of 5 and 10
respectively.
131
(B) PET Copolymer
Crystallinity (%) = + 33.1894
+ 0.0470 (Temp)
- 0.4295 (time)
+ 0.0197 (time*time)
This model (figure 4.22a to c) has a poor fit and so is oflimited use. It does
predict that crystallinity is independent of draw area and increasing heat-
setting temperature increases crystallinity linearly from 35% to 42% (heat-
set for 6 seconds), from about 35% to just over 41% (heat-set for 15
seconds), and from 42% to 48% (heat-set for 30 seconds), thereby
indicating that crystallinity is insensitive to heat-setting time.
4.1.4.2 Density
The data used to generate these models are tabulated in appendix A24. In
all cases the crystallinity measured by density is significantly less than that
measured by DSC. The crystallinity level by density for run 7 of the
homopolymer was considered erroneous by inspection and omitted from the
analysis.
(A) PET Homopolymer
Crystallinity (%) = - 28.5348
R2 = 87%
+ 0.1340 (Temp)
+ 10.5735 (Draw)
- 0.6252 (Draw*Draw)
132
The crystallinity model (figure 4.23) generated from the density data is
completely independent of heat-setting time. Increasing the
setting temperature from 80C to 220C increases the predicted crystallinity
linearly from 19% to 38% for a draw of 5 and from 25% to 44% for a draw
of 10. The R2 value is high indicating that this is a useful model for making
predictions.
(B) PET Copolymer
Crystallinity (%) = + 60.0120
R2=92%
- 0.3551 (Temp)
- 0.4018 (time)
- 4.9017 (Draw)
+ 0.0018 (Temp*Temp)
+ 0.0202 (time*time)
+ 0.3322 (Draw*Draw)
The model (figure 4.23a to c) predicts that the crystallinity varies very little
(less than 1%) with draw area in the copolymer which is in agreement with
the DSC model. Increasing the heat-setting temperature from 80C to 220C
increases the predicted crystallinity from 26% to 53%, (but not linearly),
with little increase occurring up to 140C. This model does show some
dependence on heat-setting time; increasing the time from 15 seconds to 30
seconds is predicted to increase the degree of crystallinity by about 10%.
This model has a high R2 value and is far higher than the corresponding
model derived from the DSC data (R2 = 31%).
Sample values from all the crystallinity models can be found in table 4.4.
133
4.2 CONVENTIONAL EXPERIMENTATION RESULTS
The films used in this section of the work were prepared using the same
technique as in the FED experiments, but only using the copolymer. The
chosen experimental values can be found in table 3.4. The purpose of this
set of experiments was to investigate the effect of the two main predictor
variables identified by the FED analysis, that is draw area and heat-setting
temperature.
4.2.1 Shrinkage
The shrinkage measurements for the major and minor axes are tabulated in
appendices Bland B2.
(A) Shrinkage at 65C
Both the major and minor axis plots (figure 4.24 and 4.25) show shrinkage
to increase slightly with draw area, with heat-setting having no appreciable
effect in the major axis.. The major axis plot shows a negative shrinkage
for draws of 5. Neither of these features are in the minor axis plot although
the not heat-set material does appear to behave differently than the heat-set.
The minor axis plot (figure 4.25) does show heat-setting to reduce shrinkage
from nearly 0.9% to 0.4% for samples with a draw area of 10. It should be
remembered that the samples have a gauge length of 10 mm and so 1%
shrinkage corresponds to only 100 J..UIl change in length.
134
(8) Shrinkage at 85C
At 85C the advantage of heat-setting can be seen in the major axis plot
(figure 4.26) with shrinkage reduced from 8% to 2% for a draw area of 10.
The plot also shows shrinkage is far less dependant on draw area when heat-
set above 160C. The increase in heat-setting temperature has increased
shrinkage for samples with a draw of 5 from -0.25% to 1.5% Shrinkage in
the minor axis (figure 4.27) is far lower (1.8% was the maximum recorded).
The effect of heat-setting up to 100C in the minor axis is to increase
shrinkage compared to as-drawn material. This is mainly due to heat-setting
reducing the 'negative shrinkage'.
(C) Shrinkage at 100C
The overall trends seen in the major axis at 85C are continued here but
now heat-setting reduces shrinkage form nearly 15% to 3% for a draw area
of 10 figure 4.28). Once again, for a draw area of 5, increasing the
temperature increases the shrinkage from -2% to 2%. The minor axis
shrinkage values (figure 4.29) are about 3 times lower than in the major
axis, but now the effect of heat-setting can be seen particularly, with the
unheat-set material now having higher shrinkage than the material that has
been heat-set above 160C.
(D) Shrinkage at llOC
The advantages of heat-setting are exemplified further; this time heat-setting
material with a draw area of 10 at 220C will reduce shrinkage from 17% to
3% (figure 4.30). The reduced effect of draw area on shrinkage with higher
heat-setting temperature is also maintained. For draw areas of 5, negative
shrinkages are found for unheat-set and material heat-set at 100e. The
minor axis (figure 4.31) also shows the significant effect on shrinkage
which heat-setting above 160C has, reducing shrinkage from 9% to 2.6%.
135
---------------------------------------------------
4.2.2 Thermal Analysis
DSC was used to measure the degree of crystallinity (figure 4.32b) and 13-
peak values, tabulated in appendix B3. The degree of crystallinity
measurements are related to the heat-setting treatment but are highly
scattered. The 13-peak correlated directly to the heat-setting temperature
(figure 4.32a), regardless of draw area.
4.2.3 Refractive Indices as a Estimate of Degree of Crystallinity
From the refractive index measurements, densities were calculated and
these were then used to calculate crystallites, (appendix B4). These are
plotted in figure 4.33 and show a better correlation between degree of
crystallinity and heat-setting temperature, than that measured by DSC. It is
interesting to note that there is maximum in these crystallinities for draws of
approximately 6 (from calculated data, appendix B4). This draw has
frequently reoccurred as a transition point with respect to shrinkage.
4.2.4 Optical Properties
4.2.4.1 Total Transmitted Light Measurements
The Hazeguard transmitted light measurements, having had reflective losses
removed using the Fresnel correction, are presented in appendix B5 and
plotted in figure 4.34. The two draw areas should not be compared as they
are of different thickness and no thickness correction has been applied. A
logarithmic correction for thickness has been applied in figure 4.35 (see
3.S.1.2.A) but is not fully satisfactory. The plot does indicate that heat-
setting does cause an increase in absorption up to 9% (draw area of 5 and
so material of approximately 100 mm) Losses from reflection are about 8%
and so are comparable to the worst case of the thickest material.
136
4.2.4.2 Micro-Photometry
The micro-photometry results (appendix B5), plotted in figure 4.36, show
even greater scatter than the transmitted light measurements. Both sets of
optical measurements do show that heat-setting has little effect on the
clarity of the material. The data for the two draw areas coincide when
divided by thickness (figure 4.37). This indicates that micro-photometry
vary in a linear fashion with thickness which the Hazeguard results do not.
However this must be only for a limited range of thicknesses as the plot has
a llOo/oImm value. This would mean that no light would pass though PET
of 1 mm thickness which is not in keeping with experience. It is both
surprising, and encouraging that this data point is for unheat-set material.
4.2.4.3 Birefringence
The three principal birefringences were determined for all the samples by
combining the data from the refractometry, conoscopy and compensator
experiments and are tabulated in appendix B6. The principal optic axes
were confirmed to be coincident with the draw directions and so:
&I, = (y-(3)
where y, (3 and a are the refractive indices in the major, minor and out of
plane axes, respectively.
137
The plot of birefringence as a function of heat-set temperature for a draw
area of 5 (figure 4.3Sa) shows that heat-setting only increases &1
3
slightly,
from 0.06 to just over O.OS (the unheat-set data are plotted as heat-set at
25C). The other two plots (the out-of-plane major axis, and the in-plane
birefringences) show a larger increase, even when the temperature is as low
as 100C. It should be noted that the data has some scatter for the lower
draw samples. The corresponding plot for a draw area of 10 (figure 4.3Sb)
exhibits far less scatter. Once again it can be seen that &13 shows the
smallest initial response to heat-setting. Increasing heat-setting temperature
from 140C to 220C increases the birefringences from 0.14 to 0.17, from
0.1 to O.llS and from 0.043 to 0.059 for &1
2
, &1
3
and &1
1
respectively.
The plot of birefringence against draw (figure 4.3Sc) shows that the
majority of increase in birefringence with increasing draw occurs in the
initial stages of draw. The birefringences measured here are high compared
with what would expected in PET injection mouldings (typical &1 10-
3
- 10-
4). this reflects that the major deformation is conducted in the rubbery state
in film stretching as opposed to in the melt phase as in injection moulding.
Also contributing to this is that crystallites, when oriented are more
birefringent than oriented amorphous material, and the overall high
orientation generated by the high strain rates used.
4.2.4.4 Refractive Index Measurements
The refractive indices from which the birefringences were derived are also
plotted in figure 4.39. The nomenclature used in the figure for the
refractive indices is related to the that used for the birefringence plots.
13S
The refractive indices are more dependent on the direction that the index is
measured in, than on draw area. In all cases, the values recorded for N2
were greater than N3 which were greater than NI' For the two in-plane
indices (N2 and N
3
) the refractive indices for samples with a draw of 10
were higher than those with a draw of 5 by about 0.01 to 0.02, and were
increased by heat-setting. Both these trends are reversed for the out-of-
plane (NI) index and the difference between the draws has increased to over
0.002.
4.3 HEAT-SET BOTTLE RESULTS
4.3.1 Shrinkage Measurements - Comparisons of Predictions and
Real Data
4.3.1.1 Shrinkage - Hoop axis
The hoop axis is the major draw axis in the main wall of the bottle
(appendix Cl). The data points in Figure 4.40 show that heat-setting bottles
reduces shrinkage in the same way as in films. In this case heat-setting at
200C reduces shrinkage at 110C from 11% to 3%. The model predictions
from the FED experiments have also been plotted to show how close the
bottles fit the predictions especially for heat-setting temperatures above
140C.
4.3.1.2 Shrinkage Axial Axis
Again, the measurements made on the bottles (appendix Cl and figure 4.41)
echo all the results from the film work, i.e. heat-setting reducing shrinkage
and the close agreement between the bottles and the predicted shrinkage.
This fit could easily be enhanced by increasing the constant in the model
equations slightly as the gradient in most cases is correct.
139
4.3.2 Crystallinity Measurements
4.3.2.1 Thermal Analysis
Samples from all eleven bottles were tested using DSC. The 13-peak was
used to indicate the actual temperature attained by the wall of the bottle
during heat-setting. Crystallinity measurements were also made, appendix
C2 and figure 4.42, which once again shows a poor correlation between
heat-setting temperature and crystallinity, as measured by DSC.
4.3.2.2 Density Measurements
Density measurements were made on samples from each bottle (appendix
C2) and the data yielded figure 4.43. These results show a very close
correlation between heat-setting temperature and crystallinity, discrediting
the DSC crystallinity results. The crystallinity increases with increasing
heat-setting temperature, and as temperature approaches 220DC, the
difference between the measured data and the model prediction becomes
negligible
4.3.3 Orientation Measurements
4.3.3.1 Orientation Functions
The three peak areas at 1018 cm'!, 973cm'!, and 1578cm'! (corresponding to
bulk, crystalline and amorphous phases) were measured for all eleven
bottles and are tabulated in appendix C3. Bottles 9,10 and 11, would not
yield samples that gave a sufficiently low 1018 cm-! value to be within the
linear region of the detector, and so although tabulated, these data were not
plotted.
140
The bulk orientation diagram (figure 4.44) suggests an increase in
orientation from heat-setting but the effect of the temperature is unclear.
The orientation functions of the crystalline phase (figure 4.45) does indicate
that crystalline orientation function does increase with heat-setting
temperature probably due to an increase in the degree of crystallinity with
heat-setting. The amorphous phase orientation diagram shows no tendency
for the amorphous orientation to decrease with heat-setting. This is
contrary to the commonly held view that at least part of the heat-setting
mechanism involves the relaxation of the amorphous phase. The fact that
the functions are not consistently biased on the hoop axis side is unexpected
and may indicate a high degree of experimental error. These orientation
diagrams were informative but inconclusive and so dichroic ratios were also
plotted.
4.3.3.2 Dichroic Ratio Plots
In this fashion the data were corrected for thickness, tabulated (appendix
C4) and plotted to give three dichroic function diagrams, figures 4.47, 4.48
and 4.49.
The plot of dichroic ratio against heat-setting temperature for the bulk
material (figure 4.47) would indicate an increase in dichroic ratio, and
therefore orientation, with increasing heat-setting temperature. The dichroic
ratio plots for the crystalline and amorphous phases are highly scattered
overall, but both would seem to support the observation that increasing the
heat-setting temperature does increase the orientation in both phases.
141
Analysis of Variance Table for the Balanced Model.
Analysis of Variance
Sum of Mean
Source OF Squares Square F Value Prob>F
Model 6 168.15746 2.8.02.62.4 93.360 0.0001
Error 31 9.30604 0.30019
C Total 37 177 .46350
Root MSE 0.54790 R-square 0.9476
Oep Mean 2..77974 Adj R-sq 0.9374
C. V. 19.71051
Parameter Estimates
Parameter Standard T for HO:
Variable OF Estimate Error Parameter-O Prob >
:T:
INTERCEP 1 2..8702.68 0.11594765 2.4.755 0.0001
Xl 1 -0.476765 0.101852.38 -4.681 0.0001
X3 1 1.652.880 0.0852.8195 19.381 0.0001
X2. 1 0.058905 0.112.93143 0.52.2. 0.6057
X1X2. 1 -0.233953 0.10946920 -2.137 0.0406
X1X3 1 -1.586563 0.11834187 -13.407 0.0001
X3X3 1 0.179574 0.06209533 2.892. 0.0069
Converting this equation into the original units gives :
Shrinkage - -33.504142. + 0.232.634*Temp + 3.93826*Area + 0.107636*Time
- 0.000722.*Temp*Time - 0.02.9381*Temp*Area +
A graphical representation of this relationship is given on the following
pages.
The order of selection using the MAX R technique is as follows :
Order of Variable Name Cumulative Percentage
Selection Fit
1 X3 60.26
2. XlX3 89.40
3 Xl 91.96
4 X3X3 93.73
5 X1X2 94.71
6 X1X1 95.15
7 X2.X2. 95.38
8 XZX3 95.52
9 X2. 95.63
Figure 4.1a. Typical FED Output Sheet
142
100.--------------------------------.
H MAJOR AXIS
-

H MINOR AXIS


C MAJOR AXIS
-
70-1---
60+---
C MINOR AXIS
N
; 50-1---fi!!;
'"
40t--
30 -1---5
20t--
10-1----111
0-1---'"
65 85 100 110
SHRINKAGE TEST TEMPERATURE (cC)
Figure 4.1b. Shrinkage Model Correlation as a Function of Heat Setting
Temperature for PET Homopolymer (H) and Copoloymer (C),
for Each Draw Area
143
130
-
MAJOR AXIS 01.=5
120
-
MAJOR AXIS 01.=10
0- 110
r-
, IIINOII AXIS 01.=5

-
III
1
00
wtIOR AXIS OA: 1 0
...
'"

..
90
0
-'
w
>-
80
70
60
80 100 120 UO 160 180 200
lA T SUTtIG TlAP[RATURE (cC)
Figure 4.28. Statistical Model For Homopolymer Yield Stress, Heat-Set For 6
Seconds
115
11 0
105
100

fII
95
fII


90
VI
0

85
>-
80
75
70
80 100 120 140 160 180 200
lOT SETTING TEII'RATURE (oC)
IIAJOR AXIS OA=5
IIAJOR AXIS OA=10
IINlR AXIS DA=5
IItlOR AXIS DA= 1 0
Figure 4.2b. Statistical Model For Homopolymer Yield Stress, Heat-Set For
IS Seconds
144




o 90
...J




80 100 120 140 160 180 200
HEAT SETTING TEMPERATURE (oC)
MAJOR AXIS DA::5
MAJOR AXIS DA:: 10
MINOR AXIS OA=5
IAlNOR AXIS DA= 1 0
Figure 4.2c. Slatislic:al Model For Homopolymer Yield Stress, Heat-Set For
308eJads
145
10.S
10
9.5

9

i' 8.5
c(
Ill:
8
t;
0
7.5
....
101
>-
7
6.5
6
v
..........
t'.... /
V
"-
....-
V
/
./7 I"-----.
/ '/
V /
/
v
5.5
MAJOR AXIS DA=5
MAJOR AXIS OA= I 0
t.tNOR AXIS OA=5
IINlR AXIS OA=IO
80 100 120 140 160 180 200
IAT SETTING TEIoftRATURI: (aC)
Figure 4.3a. Statistical Models For Homopolymer Yield Strain, Heat-Set For
6Se\:onds
10.5
10
9.5

9

i' 8.5
"
Ill:
8
..
/'"
...........
f'"
V " ..-
V
/
MAJOR AXIS 00\=5
MAJOR AXIS OA= I 0
MINOR AXIS OA=5
IINOR AXIS OA=10
fII
0
7.5
Q
>-
7
6.5
6
/
/
/

/
"""
1
5.5
80 100 120 140 160 180 200
KAT SETTtlG TDl'ERATUIE (aC)
Figure 4.3b. Statistic:al Models For Homopolymer Yield Strain, Heat-Set For
IS Sonds
146
10.5
10
9.5
,.... 9
1'1

8.5
Z;

et:
8

VI
Q
7.5
-'
w
>-
7
6.5
6
e--
/'
..........
'-..... /'
/
'-.....
---
/
/
V
.
e--
-.....
/'
V I "
/
V
5.5
MAJOR AXIS DA=5
MAJOR AXIS DA= 10
IotNOR AXIS OA=5
MINOR AXIS DA= 10
80 100 120 140 160 180 200
HEAl SETTING lEIoCPERA lURE (oC)
Figure 4.3c. Statistical Models For Homopolymer Yield Strain, Heat-Set For
30 Seconds
147
200
-
MAJOR AXIS DA=S
180
V
-
....-
V-
IlAJOI! AXIS DA= 1 0
160


I&'lR AXIS DA=S
-
1.0
f-'
IINOI! AXIS DA= 1 0

5
120
100
80
60
80 100 120 1.0 160 180 200
lOT STTtIG TEMPERA T\R (oC)
Fig1ft 4.48. Statistical Models For Homopolymer lJhimate Tensile SIress,
Heat-Set For 6 Seconds
180
110
160
ISO
5
120
110
100
90
80
70
l---
I-
I-----
t------
p
l---
f-'
"
---
=
r---.
80 100 120 1.0 160 ISO 200
lOT SETTNl fDl'(RATlIIE (oC)
WAJOI! AXIS OA=S
WAJOI! AXIS OA: 10
IItQ! AXIS DA=S
IItQ! AXIS OA=10
Figure 4.4b. Statistical Models For Homopolymer Ultimate Tensile SIress,
Heat-Set For IS Seconds
148
160
150
140
130
en 110
....
:::>
100
90
80
70
I'-.,

I'-.,

t'....

t'....

I'-.,
""
I'-.,
""
80 100 120 140 160 180 200
HEAT SETTING TEMPERATURE (oC)
MAJOR AXIS OA=S
MAJOR AXIS OA= 1 0
MINOR AXIS OA=5
MINOR AXIS OA=10
Fig1ft 4.4c. Statistical Models For Homopolymer Ultimate Tensile Stress.
Heat-Set For 30 Seconds
149
220
200
MAJOR AXIS OA=S
180
MAJOR AXIS DA= I 0
160 MINOR AXIS OA=S
-
r----.
MmR AXIS OA=I O



z 140

I! 120
VI
'5
100
80
60
f-
40
80 100 120 140 160 180 200
HEAT SETTING TEIlPERAME (oC)
Figure 4.5. Statistical Models For Homopolymer Ultimate Tensile Strain.
Heat-Set For IS Seconds
110
-
100
MAJOR AXIS DA=S

e-
-
---
MAJOR AXIS DA= I 0
90
----
--

80

70

..
60
i1i
50
......
r------
--
40
---
t---..
30
---- 20
80 100 120 140 160 180 200
HEAT SETTtlG TEIIPERATIIlE (oC)
Figure 4.6. Statislical Model For Homopolymer Strain Hardening Rate.
Heat-Set For IS Seconds
130
liS
-
110
I--
DRAW AREA=S
----
-
105
V
DRAW ARA=IO
-----
'0
1
00
l.-----'

.....-

9S

Is!
90
...
'"
8S
0
---
....I
W
80

----
--
75
-----
-----
70
-----
65
80 100 120 140 160 180 200
HEAT SETTIIG TEMPERA TIR (oC)
Figure 4.7a. Statistical Model For Copolymer Yield Stress In The Major Axis,
Heat-Set For 6 Seconds
110
L----
-
105
DRAW AREA=5
I-------"
I--
-
--
DRAW AREA=IO
100
l.---
I
----
95

'"
90
'"
...
..
...
85
'" 0
Q
80
L---
V
>
75
I-------"
----
70
I---
65
80 100 120 14a 16a 180 2aO
HEAT SETTIIG TDlPERATIJRE (oC)
Figure 4.7b. StatistK:aI Model For Copolymer Yield Stress In The Major Axis,
Heat-Set For IS Seconds
ISI
115
110
DRAW AREA=S
105
DRAW AREA=10

0
100

Cl!
Cl!
Ii!
95
....
Cl!
0
90

""
)0-
85
80
75
80 100 120 140 160 180 200
HEAT SETTING TEMPERATURE (oC)
Figure 4.7c. Statistical Model For Copolymer Yield Stress In The Major Axis,
Heat-Set For 30 Seconds
152
9
8.5
8
7.5

VI
o 7
...J

6.5
6
5.5
/
V
/
V

-------...
/


/

80 100 120 140 160 180 200
tOT SETTING TEt.f'ERUIIlE (oC)
DRAW AREA=5
DRAW AREA: 1 0
Figure 4.1. StatisIic:aI Model For Copolymer Yield Strain In the Minor Axis,
Heat Set For I.S seconds
IS3
180
.-/
V
--
/
r-
-
MAJOR AXIS OA=5
160
MAJOR AXIS DA= I 0
140
WO! AXIS OA=5
-
---
80
60
BD 100 120 140 160 180 200
HEAT SETTI/G TDl'RATUIE (oCl
Figure 4.980 Statistical Models For Copolymer Ultimate Tensile Stress, Heat-
Set For 6 Seconds
180
/
-
/'
IIAJOR AXIS OA=5
160 -
V
IIAJOR AXIS DA='O
--

! '40 r--
IIINOR AXIS OA=5
r--
-
Cl!
---
IIINOR AXIS OA=' 0
I:l 120
a:

Cl!
100
:l
80
60
80
'00 '20 '40
160 180 200
HEAT SETTJjG TIoI>ERATUIE (oC)
Figure 4.9b. Statistical Models For Copolymer UJtim .... Tensile Stress, Heat-
Set For 15 Sec;onds
1S4

i 140 L----I

...
CIl

=>

80 lOO 120 140 160 180 200
H[AT SETTING TEIAPERATUR (oC)
MAJOR AXIS OA=5
MAJOR AXIS DA= 10
MINOR AXIS DA=S
MINOR AXIS OA= IO
-'
Figure 4.ge. Statistical Models For Copolymer Ultimate Tensile Stress, Heat-
Set For 30 Seconds
ISS
220
-
200
IIAJOR AXIS 00\=5

-
-...........
IIAJOR AXIS OA=10
180

160
-........

'---...
140

III
5120
100
80
60
80 100 120 140 160 180 200
IAT SETTN; TEMPERATURE (oC)
Fisure 4.10. Staaistical Model For Copolymer UJtimate Tensile Strain, Heat-
Set For IS Sec:onds
100
90
80
! 70
60
so
40
30
r--
-
./
....----
----
V
---:

80 100 120 140 160 180 200
HEAT SETTING TENPERATIJl( (oC)
1I4JOR AXIS OA=5
IIAJOR AXIS CA= 1 0
Fip-e 4.11. Slatistiad Model For Copolymer Sbain Hudcning Rate. Heat-Set
For IS Sec:onds
1.56
50 ______
-
------
H DA=S
40 -
-
------
r------.
C DA=5
i-
30
---
H 0.=10

r--
-
..
------

C DA=10
Z
20
--
..
--
I---
..
:-------.

'"
10
0
-10
0 5 10 IS 20 25 30
tt:AT SETTING 111( (SECS)
Figure 4.12. Statistical Model For Creep Strain As An Effect Of Time At Two
Draw Areas, 5 and 10
160 1\
-
140 1\ \
H SET=10
-
120 \ \
C SlT=10
H SET=30
100


'\.
-

80
C SET=30
""- ,,'\
:;;
..
60

........
........ "
",- "-
'"
40

0
20

0


-20
1 2 3 4 5 6 7 8 9 10 11
DRAW AREA
Figure 4.13. Statistical Model For Creep Strain As An Effect Of Draw Area
0.6
0.5
0.4
-- 0.3
v
-............
./
V

MAJOR AXIS OA=S
MAJOR AXIS CA= 10

...
C)
0.2 c
I
0.1
III
0
-0. 1

e-
r----,.
V

f',.
'"
-0.2
80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATURE (oC)
Figure 4.14. Statislical Model For Homopolymer SbrinIcagc At 6SOC, Hcat-5ct
For IS Seconds
ISS
2

---
t.4AB AXIS OA=5
----
r--
-......
;:,..--
t.4AJOR AXIS 1),\=.0
1.5
'" IIINOR AXIS OA::5
IIINOR AXIS 04::'0
-

..---
V
V
v
o
-0.5
80 .00 .20 UO .60 .80 200 220
IAT SETTtIC T[II'(RATUR (oC)
Figure 4. 1 Sa. Statistical Models For Homopolymer Shrinkage At &SOC, Heat-
Set For 6 Seconds
2

I
'"
---
----
--...........
----
-......

t.4.uoR AXIS OA::5
t.4.uoR AXIS OA=.O
IIINOR AX.S 0A::5
1.5
IIINOR AXIS OA::'O
..---
/

/
V
o
-0.5
80 .00 120 140 160 180 200 220
KAT SETTING TIIP[RATlII[ (oC)
Figure 4.15b. SIatisIicaI Models For Homopolymer Sbrinkage At 8SOC, Heat-
Set For IS SmInds
IS9
TEMP
% Shrinkage Contour Plot For B90N
Shrinkage at 85 Cegree. C Ma,or Axis
Ti.. held at 15 aecond.
2 0 0 ~ r T ~ ________________ __________ --.
170
140
110
04.50
Fi gure 4. 15c
6.215
SHRINKAG ------ -0.67
0.91
8.00
9.715 11.50
-- -0.27 - .-. 0.12
0.52
1.31 1. 70
2.5
2

MAJOR AXIS OA=5
"-
MAJOR AXIS OA= 1 0

1.5


...
Cl


"'-..
........
--
I---
"'-

-........

I'---........
----
r--
t.tNOR AXIS DA=5
t.MNOR AXIS DA= 10
!
VI
0.5
.....---


0
V
,/'
-0.5
80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATlR (oe)
Figure .t. Ud. Statistical Models For Homopolymer Shrinkage At Mac. Heat-
Set For 30 Seconds
161

-
6
MAJOR AXIS OA=5
-

MAJOR AXIS OA=10
. [-..,
ti' r---

IllNOR AXIS
,..-
3 -
Z
r--.:::::

V
IIINOR AXIS 01.=10
I 1
./
V
0
V
/:
-1
-2
/
-3
80 100 120 ,.0 160 180 200 220
KAT TIIPRATIJlE (oC)
Figure 4.161. SCaIisticaI Models For Homopolymer Shrinkage At lOOOC, Heat-
Set For 6 Seconds

-
6 MAJOR AXIS 01.=5

-
...............
MAJOR AXIS 0.=10




IIINOR AXIS 0.=5

'"
,/
3
...............
"
,/
-
z
'><'"
IINOR AXIS 01.=10
"-
I
1
/>'
........... ...........
1/1
/
"
0
./
- 1
,./
-2
/
-3
80 100 120 UO 160 180 200 220
KAT stTTINC TllPRATIJl[ (oC)
Figure 4.16b. Stalislical Models For Homopolymer Shrinkage At lOOOC, Heat-
Set For IS Seconds
162
% Shrinkage Contour P t For B90N
Shrinkage at 100 Degree. C Ma,or Axl.
Ti held at 15 .econd.
TEMP
200 .-----------------____________________________________________________
170
HO
B O - ~ ~ - - - - - - - - ~ - - r _ - - - - ~ - - - - - - - - - - ~ - - - - - - - - ~ - - - - ~ - - ~ ~ - - - - - - - - ~ ~
".50
SHRINKA6
Fi gure 4.16c
8.25
-- -2.39
5.U
8.00
9.75
U.50
AREA
-- -0.52 - .- - 1.36
3.23
6.99 8.88
8
6


MAJOR AXIS OA=5
MAJOR AXIS OA= 10


"

......
...
<:>
2


en
0
-2


/

V
"-

/
V
/'
/'"
MINOR AXIS OA=5
MINOR AXIS OA= 1 0
-4
80 100 120 1 "0 160 180 200 220
HEAT SETTING TE...rRATURE (oC)
Figure 4. 16cl Statisbcal Models For Homopolymer Shrinkage At lOO'C. Heat-
Set For 30 Seconds
164
14
12
~
MAJOR AXIS OA:5
""'"
~
"-
~
~
"'-
~
"""
""
.-/
-
~
/"
-,
/'"
../
~
/
.-/
.-/
ID
MAJOR AXIS OA: I 0
8
MINOR AXIS OA=5
t;'
~ 6
~ 4
i 2
o
-2
-4
IIINOR AXIS OA:l0
1
80 100 120 140 160 180 200 220
KAT SETffiG TEIIPRATlR (oC)
Figure 4.17a StatislK:al Models For Homopolymer Shrinkage At IIO"C, Heat-
Set For 6 Seconds
14
-
12
~
MAJOR AXIS OA:5
-
10
~
~
MAJOR AXIS OA=10
8
E 6
~
'"
IINOR .XIS DA:5
"- -
IJ
~
"""
I'-,
.-/
IINOR AXIS OA" 1 0
~ 4
! 2
--.
1"-- "
~
./'
V
~
0
V
-2
../
-4
/
1
80 100 120 140 160 180 200 220
KAT stTTING lMPERATlR (oC)
Figure 4.17b. SCatistical Models For Homopolymer Sbrinbge At IIO"C, Heat-
Set For IS Seconds
165
TEMP
% Shrinkage Contour t For B90N
Shrlnkage at 110 Degrees C M.Jor Axle
Tlme held at 15 seconds

:170
140
110
---'--, . .,...-----

4.50 6 . 25 8 .00 9. 75 :11 .50
AREA
SHRINKAG -- - 3.82 -- -0. 64 - .. - 2 . 55 ... -- 5 .73
Figure 4.17c 8.91 ... -- 12. 09 15.27
14
12
10
8
o
-2
-4
-6
~
'"
/'"
~
"'"
~
~
V
/'"
t-....
~
i'-.. /
~
"-.,.
K r-......,
./
V ........
~
~
--.
80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATUlE (oC)
"'AJOR AXIS OA=5
"'AJOR AXIS OA= 10
MINOR AXIS OA=5
MINOR AXIS OA= 1 0
Figure 4.17d Statistical Models For Homopolymer Shrinkage At IIO"C, Heat-
Set For 30 Seconds
167
0.6
MAJOR AXlS OA=5
0.5
MAJOR AXIS OA=10
IH)II "XlS OA=5

... 0.3
----
--.
-
IHlR AXIS OA=10
i
III 0.2
0.1
o
80 100 120 140 160 180 200 220
fAT SETTNl lt.tPERAME (cC)
Figure 4.18. Statistical Models For Copolymer ShriDbge At 65"C, Heat-Set
For IS Seconds
2.5
r----
-----
.......
-----
r-----
--
..............
r--
.........
./
WAJOII AXIS DA=5
MAJOR AXIS DA= 1 0
IItO! "XlS DA=S
2
--

i'--
l/ ----
V
/'"
/
O.s
/
o
80 100 120 140 160 180 200 220
lOT SETTING TDoftRATURE (cC)
Figure 4.191. SIatistical Models For Copolymer Shrinkage At SS"C. Heat-Set
For IS Seconds
168
% Shrinkage Contour Plot For B95A I ase
Shrinkage at as Oegr ees C Major Axis
Ti ffle held at 1S seconds
TEMP

170
140
110

4.50 6 . 25 8 . 00 9.75 11.50
AREA
SHRINKAG -- 0.17 _. -- 0.61 -- 1.05 _. -- 1.49
-- 1.93 ... -- 2.37 --== 2.81
6
-
5
~
IlAJOII AXIS OA=5
""
-
'-...
IlAJOII AXIS OA=10
4
.......
~
~
"'--
~
IINOR AXIS 0A=5
t< 3
'---
~
/ -
-
~ 2
r---. IINOR AXIS OA=10
I
V
"'-
..............
en
I
V
/'

,/
/
-I
/'
-2
80 10O 120 14O 16O 180 200 220
IAT SETTING TtIlPERATURE (oCl
Figure 4.2Oa. SWiSlical Models For Copolymer Shrinkage At lOO"C, Heat-Set
For 6 Seconds
6
-
5
~ IlAJOII AXIS 04=5
~ ~
-
IlAJOII AXIS OA= I 0
4
~
..........
~
"-....
IItQ! AXIS 04=5
t< 3
.......
~
/'
-

...............
~ 2
../
IINOR AXIS OA=10
!
V ~
~ I
/
/

./
/
-I
_2
V
80 100 120 14O 16O 180 200 220
IlA T STTtIG TtlI'ERA TURE (oCl
Figure 4.2Gb. Statistical Models For Copolymer Sbrinbgc At lOO"C. Heat-Set
For IS Secoads
170
TEMP
% Shrinkage Contour Plot For B95A I ase
Shrinkage at 100 Degrees C Major Axle
Time held at 15 seconds

170
140
110

4.50 6 . 25 B.OO
AREA
--- -2.0!5 _. -- -0 . 52
4.06 5.59
Figure 4.20c
SHRINKAG
9.75
1.01 _.--
7.12
11.50
2 . 53
6


"-

MAJOR AXIS OA=5
MAJOR AXIS OA= 1 0
MINOR AXIS OA=5
5
4


./
[7
MINOR AXIS OA=10
L>


./
V
v
o
/
V
-1
-2
80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATlIlE (oC)
Figure 4.2Od. StalislK:al Models For Copolymer Shrit".e At 100"C, Heat-Set
For 30 Seconds
172
10

'"
'-......
I'-....

I'--. /'
V
MAJOR AXIS OA=5
MAJOR AXIS OA= 1 0
IINOR AXIS OA=5
8
6
I
/'
/
IINOR AXIS 04=10
/'
"-,

./
V
/
v
-2
-4
-6
80 100 120 140 160 180 200 220
!AT SETTING TEIftRATURE (oC)
FiS- 4.21a. Statistical Models For Copolymcr Shrinkage At IIOOC, Heat-Set
For 6 Sewnds
10

"
'-......
I'-..
'"

/
/'
MAJOR AXIS OA=5
MAJOR AlCJS OA=10
IINOR AXIS 04=5
8
6
----

/'
IIINOR AXIS OA=10
/' ""'" ',,-
/'
V
/
v
-2
-4
-6
80 100 120 140 160 180 200 220
lOT stTTm TWPERATURE (oC)
Figure 4.2 lb. Statistical Models For Copolymer Shrinkage At IIOOC, HeIt-8et
For IS Sewnds
173
TEMP
% Shrinkage Contour Plot For B95A laser
Shrinkage at 110 Degrees C Ma,or Axis
Time held at 15 seconds
2 0 0 ~ - - - - ~ - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ~ - - - - - - - - - - - - - - - - - - - - - - - - - - ~
170
140
110
8 0 - - - ~ - - - - - - - - - - - - ~ - - - - - - - - - - ~ - - - - - - - . - - ~ - - - - - - - - - - ~ - - . - - - - - ~ - - - - - - - - ~ ~
4.50 6.25 8.00 9.75 11.50
AREA
SHRINKAG
- - -4.38 _. -- -1.37 _. -_. 1.63 _.--
4.64
Figure 4.21c
7.65 ' 10.66 13.67
10
~
[ S ~
~ /'
V
MAJOR AXIS DA=5
MAJOR AXIS DA= 1 0
YNOR AXIS DA=5
8
6
'-.....
~
./
:-....
IiINOR AXIS DA=10
V
~
~
V
./
[7
- 2
-4
/
~
80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATURE (oC)
Figure 4.2 Id. Stlltistal Models For Copolymer Shrinkage At llOOC. Heat-Set
For 30 Seconds
175
55
-
DA--5 H
50
~
-
DA=IO H
-
/'
~ 5
V
DA=5 and 10 C
..
V
....
~ 40
;t
/'/'
t;
/'
3
35
V
/
30
V
25
80 100 120 140 160 180 200 220
KAT SETTING TIIPOIATI.I! (oC)
Figure 4.228. StltistKal Models For DegJee of CrysIaIIinity As Measured By
DSC. HCIt-8et For 6 Secoods
46
/
-
DA=5 H
44
~
/
-
DA=IO H
_42
-.......
V
..
~
DA=5 and 10 C

~ 40
v<

...
~ 3 S
~ /
.,
~
..
I') 36
/
~
34
/
32
80 100 120 140 160 lao 200 220
HEAT SETTIIG TDf'ERATI.I! (oC)
Figure 4.22b. Slllislical Models For DegJee Of Crystallinity As Measured By
DSC. Heat-8et For IS Sccoads
176
60
55
45

..J
..J
40
1/1

0
35
30
25

"'::::
'"




80 100 120 140 160 180 200 220
HEAT SETTING TEt.f'ERATURE (oC)
DA=5 H
DA=10 H
DA=5 and 10 C
Figure 4.22e. SWistical Models For DegRe Of Cry.UUrity As Measured By
DSC. Heat-Set For 30 Secoods
177
55
-
/
50
OA=5 H
1/
-
OA=IO H
45
L-
V

0A=5 C

~ 4
""
-
L5
0 ~
OA=IO C
-'
---
v
0
-----
..
L
...
~ 30
ti
~
,---
~
~
25
---
-----
20
IS
80 lOO 120
"0
160 180 200 220
IAT SETTING Tll>ERAT\II[ (GC)
Figure 4.23a. Slatistic:al Models For Degree or CI}'St8Ilioity For A Heat-8ct
Time Of 6 Seconds As Measured By Density
55
-
/
50
OA=5 H
/
-
45
OA=IO H
~
L-
---

0A=5 C
~ 4
>-
-7
"...
-
...
---
OA=IO C
%
::; 35
-'
V
f-"" /
/
V ..
...
~ 30
----
~
:::.------
ti
~
-
25
-----
20
-----
IS
80 lOO 120 140 160 180 200 220
IAT SETTING TEII'ERAT\II[ (GC)
Figure 4.23b. SlatistiQI Models For Degree Of Crystallinity For A Heat-8ct
Time Of 15 Seconds As Measured By Density
178
65
-
60
/
0.=5 H
-
55
V
0.=10 H
~ 5
/
~
/
0.=5 C

>- 45
/
~
-
!::
0.=10 C
~ 40
..-
-'
V
----
----
<C
--
t; 35
>-
~
I--
----
----
0::
030
----
V-
---- ---
25
----
----
20
15
80 100 120 140 160 180 200 220
HEAT SI:TTING TEMPER. l ~ (oe)
Figure 4.23:. Statistical Models For Degree Of Crystallinity For A Heat-8et
Time Of 6 Seconds As Measured By Deasity
179
0.1
...
...
0
DA:5
...
...
-0. 1
........
1&
010=10
...
~ o . 2
~
-
-0.3
...
...
...
.A.
1-0 ...
-0.5
-0.6
-0.7
-0.8 ""
0 50 100 150 200 250
TOI'RATUI (oC)
Figure 4.24. Copolymer SbriDkage ID The Major Axis At 65OC, Heat-Set For
IS Seconds. From ConvcntioIIaI Expeiimealation Series
0.9
"'"
...
0.1
'\
OA=5
...
0.7
"'"
DA= 10
0.6
~
~

_ 0.5
~

0.4
........
...
I
-
0.3
V - III
.-/'
0.2
V
0.1
/
0
...
-0.1
0 50 100 150 200 250
TEIf'O!A TUI (oC)
Figure 4.25. Copolymer SbriDkage In The Minor Axis At 6SOC, Heat-Set For
IS Sec:onds. From Conventional Experimentation Series
180
9
8
7
6
o
-1
o
lo.
'\
~
---
50
""
""

~
t-.... A
...
~
l--
100 150
T[IftRATlJR (C)
A
...
200 250
FiguR 4.26. Copolymer Shrinkage In The Major Axis At 8Sac, Heat-Set For
IS Seconds. From ConvcutionaI Experimeatation Series
2
~
~
1.8
1.6
~ t 4
~
0.8
0.6
0 .
...
o 50
t-----...
l A ...
...
... ... ...
... ...
...
100 150
T[IIPERATUI: (oC)
200
~
~
250
FiguR 4.27. Copolymer Shrinkage In The Minor Axis At 8Sac, Heat-5et For
15 Seconds. From Conveutional Expetimentldion Series
181
16
14
12
10
g 8
6
I 4
2
o
-2
-4
o
...
'\.
'"
""
./
-------
A'
50
""'"
I'....


I,.----
100 150
TDlPERATlJA (C)
200 250
Figtft 4.21. Copolymer Shrinkage In The Major Axis At 100"C. Heat-8et For
IS Seconds. From Conventional Experimentation Series
5.5
..
5
... OA:5
\
..
4.5
OA: l 0
\

4

i\
.3.5
""
I
3
Cl>
'"
I'....
2.5

.....
2
..
---
.. ..
1.5
0 50 100 150 200 250
TOoftRATIIlE (oC)
Figure 4.29. Copolymer Shrinkage In The Minor Axis At 100"C. Heat-8et For
IS Seconds. From COIMIltionaI Experimentation Series
182
20
15
o
-5
o
~
'\
~
"'"

~
...
~
~
50 100 150
ltNPERA1'UIE (oC)
-::
-
200 250
Figure 4.30. Copolymer Shrinkage In The Major Axis At ll00c, Heat-Set For
15 Seconds. From Conventional Experimentation Series
9
\
..
8
OA=5
\
..
OA=10
7
1\

~ 6
"""
~
~
14
"
t'-,.
...
~
3
......
~
...
~
2
-
1
0 50 100 150 200 250
TEMl'ERAT\JI[ (oC)
Figure 4.31. Copolymer Shrinkage In The Minor Axis At IIOOC, Heat-Set For
15 Smmds. From COIMIltioaal Expetimeutalioo Series
183
,....
240
230
0220
o

210
200
0::
...
'i 190
I:!
:.:: 180

170

tJ 160
m
150
140
1/
[7
,
17-
1/
V
1/
lA
V
140 150 160 170 180 190 200 210 220 230 240
fA T SETTING TEt.f'ERA lURE (oC)
...
DA=5
...
DA=10
Tbeta= Theat set
Figure 4.32a. Secoadary Melting Peak Temperature As A Function of Heat-
Setting Tempaatw'e
184
40
...
...
OA .. 5
38
...
...
DA=10
~ 6
~
~
~ 34
Z
...
...
~ 32
Cl>
>-
3
30
28
26
0 50 100 150 200 250
T[II'ERATlJRE (C)
Figure 4.32b. Degree Of Crystallinity At. Measured By DSC For Films From
CoavCldional Experimentation Series
50
38
36
50
...
I
/'
...
---
V
100 150
TII'ERATlJRE (C)
... ...
.;
...
I
~
DA=5
0:: ,0
/
200 250
Figure 4.33. Degree of CrysIaUinity At. Estjmattd By Reftidive Indcx
Mcasurcmeats For Films From Conventional Experimentaaion
Series
185
10
9
8
3
2
t
.-
I-
.---
-
----
r----
..
l-


--
---
r--
v
20 40 60 80 100 120 140 160 180 200 220
IAT SETtING TEMPER"Tl.llE (oC)
Figure 4.34. Optical Absorbtion As Measwed By Hazeguard Haze Meter Of
Films From Conventional ExperimentIdion Series
-0.02




0.=5
-0.03
...
'"
OA=10
~ -0.04
"-

~
i -0.05
'" lB
i -0.06
~
f-;;
f--
~
-0.07 __
--
~
-0.08 ----
20 40 60 80 100 120 140 160 180 200 220
II:AT SETTt(; TEIIPRATUR (oC)
Figure 4.3S. Plot Of In(absorption)ltbickness As A Comction For The
Systematic Variaaioa In Sample 1'IIickaess
186
7
...
...
OA=5
6
...
...
04=10
~
N
~
~
...
r
c3
...
2
...
...
1
20 40 60 80 100 120 140 160 180 200 220
.AT STTIl T1I'1!ATIJ! (QC)
Figure 4.36. Optical Absorption Measured By Mklophotomeby On Films
From CoaVCDtioDal Experimeatation Series
110
...
100
0.=5
...
90
04=10
~
80
......
...
N
70
~
60 ...
I
...
50
40
30
...
...
20
...
10
20 40 60 80 100 120 140 160 180 200 220
IAT STTIl TOIPERATUIE (QC)
Figure 4.37. Optical Absorption/1bida As Meaued By Mierophotomeby
On Films From Convenaional ExperimeaIation Series
187
0.18 r--
...
0.16
NI
...
0.14

N2

0.12
---'
....--
---...

i 0.1
---
f.--
r-
....-
::; 0.08
1
0
.
06
1
...
0.04
0.02
0
0 20 40 60 80 100 120 140 160 180 200 220
I4T SET TEIof'(lI"TUAE (oC)
Figure 4.388. BirefiiDgencc Measumncnts On FilmsWith A Draw Ala or oS
From Cooveational ExpeaiancPl .. ioIl Series
0.1 8 r--
K
...
0.16
..
NI
..--'

...
0.1 4 N2
I'J 0.12

..-r'
t 0. 1
-
i 0.08
0.06


0.04
0.02
0
0 20 40 60 80 100 120 140 160 180 200 220 240
IA T SET TtIof'(lI" T\II[ (oC)
Figure 4.38b. Bireftinsence MeaswelllCDtS On Films With A Draw Area Of 10
From CooveatioDal Experime
i4
idim Series
188
0. 18 r--
~ ~
0.16
NI
~
~
0. 14
/"
N2
~ 0. 12
~
~
O. I
/ /
i 0.08
V /
0.06
/ /
/- ~ -
0.04
Ij
7
0.02
tf/
0
0 Z -4 6 e ID 12
DRAW AREA
Figure 4.38c:. 8irdiingem:c Measwements On Films Heat-Set AI 22O"C From
Conventional ExperimeaIIIion Series
1.7
~
U8
"'-
OM NI
~ A. ...
~
1.66
--c-
-
OM NZ
'"""

x 1.64
OA5 N3
~ 1.62
"
"
x
..
0

"
~
.

0.. 10 NI
>-
1.6
x
u
0"'10 N2
: 1.58
... x
~ 1.56 0"'10 N3
- ~
1.54
~
.,.
1.52 ~
....
1.5
0 50 lOO ISO 200 250
TDlPERATLR (oC)
Figure 4.39 Refiactive IncIex Measwements On Films From Conventional
Experimencation Series
189
12
....
SHRN< 1It85
10
....
SHRN< .,00
8
~
....
SHRN< itll0
~
~
-
~
6
IIOOEl. liS
-
....
..
r-
-
I
--
"
r-
-
IIOOEl. 100
....
I-;;

h ....
2
~
IIOOEl. 110
..
..
....
..
0
-2
20 40 60 80 100 120 140 160 180 200 220
!AT S T T ~ G TEIf'ERATlII( (oC)
Figure 4.40. Sbrinkage In The Hoop Axis Of Heat-set Bocdes With The
Corresponding Predic:tecl Shrinkage
25
....
SHRN< 085
20
..
SHRN< 0 100
..
15 SIfI1III( 011 0
N
~
-
~ 10
IIOOEl. 8S
I
-
..
IIOOEl. 100
l:
....
..
1/1
5
....
....
1oIOOl. 11 0
....
S

t",
0
-5
20 40 60 80 100 120 140 160 180 200 220
lAT SETmG TDf'ERATIR (oC)
Figure 4.41. SbrinItase In The Axial Axis Of HeatSet Bottles With The
Conespoadins Predicted Sbrinbge
190
...
40
38
...
...
...
...
...
3D
28
26
20 .0 60 80 100 120 140 160 180 200 220
llA T STTtIC TDRRA TIJI( (oC)
Figure 4.42. Degree Of CJystaIliDity Of HeIt-Set BoIdes As Measnd By
DSC
ss
...
OATA
so
...
.J
IIlOEI.
~ 5
"
~
tr
-
. 0
/.
V

"'" 35
'"
...
/'
/

...
./'
'" CJ 30
V
~
V
,L
25
---
~
I'*"
20
20 .0 60 80 100 120 ,.0 160 180 200 220
lOT stTTNG TDftIIATIJI( (oC)
Figure U3a. Desrcc Of Crysaallinity Of Hat-Set BoUIes As McasurecI By
Deasity CompmcI Widl Model Predidioa. Model Heat-Setting
Time of IS Seconds
191
55
/
...
DATA
50
f/
...
IoIOOEL
...... 45
Iff
,-

'-'
40
...
iJ
...J
......
...J
35
VI
V
>-
...
... ........
a:
0
30

......
-----
25
--
c-
l-
20
20 40 60 80 100 120 140 160 180 200 220
HEAT SETTING TEIo4PERATlJlE (oC)
figure 4.43b. Degree Of CtySlallinity Of Heat-8et 80UIes As Meaued By
Density Compared With Model PrecIictioD, Model, Heat-SeItiDg
Time of 25 Sec:oIlds
192
0.9
0.8
0.7
0.6
en
0.5
"-
0.4
0.3
0.2
0.1
o



Ircl
r-l
CJ,
,i:o
EJ
/ /
-;-;l-""
0
Y




fAT S(I
/


/


V
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
F2B
Figure 4.44. Orientation Functions Of Heat-Set Bottles Measured By P-FTIR
(1018cm-
1
)
193
0.9
0.8
0.7
0.6
0.5
u.
0.4
0.3
0.2
0.1
o





'XP

0
o _"
P'

=-


7

"
I-
.......
/ J



/
El


/
150,*
I-


o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
.2B
Figure 4.45. Orientation Functions In The Crystalline Phase Of Heat-Set
Bottles Measured By P-FTIR (973cm-
1
)
m
u.
0.5
0.4
0.3
0.2
0.1
o
o
V
e
--"'0
V

..L.
G
0
l2'EJ


It[
" o-
B
1\ R
V
/
\
11'0" COOlED
1()1 It:AI SI:
1
0.1 0.2 0.3 0.4 0.5
.2B
Figure 4.46. Orientation Functions In The Amorphous Phase Of Heat-Set
Bottles Measured By P-FTIR (1578cm-
1
)
194
0.4
0.3
0.2
....
....
....
....
"-
0.1
Cl
....
....
0
....
-0.1
....
-0.2
20 40 60 80 100120140160180200220
HEAT SETTING TEMPERATURE (oC)
Figure 4.47 Bulk Dichroic Ratio Of Heat-Set Bottles
195
0.4
0.3
0.2
....
....
....
"-
0.1
0
....
....
...
....
0
-0.1
-0.2
20 40 60 80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATURE (oC)
Figure 4.48 Dichroic Ratio Of The Crystalline Phase As A Function of Heat-
Setting Temperature For Heat-Set Bottles
0.4
0.3
....
0.2
I'"
....
....
l's 0.1
....
.... ....
....
o
-0.1
-0.2
o 50 100 150 200 250
HEAT SETTING TEMPERATURE (oC)
Figure 4.49. Dichroic Ratio Of The Amorphous Phase As A Function of Heat-
Setting Temperature For Heat-Set Bottles
196
Table 4.1. Sample Values From Mechanical Models
Draw Area Draw Area
5 10
Heat Set Heat Set Heat Set Heat Set Heat Set Heat Set
Model
Temp Temp Temp Temp Temp Temp
R2%
80C 140C 200C 80C 140C 200C
ay(O)MPaH
74 76 78 108 110 112 88
C
68 74 80 95 101 108 81
ay(9O) MPaH
74 74 74 93 93 93 39
C
-
- - -
- - -
&y(O) % H
6 9 9 9 10 9 43
C
- -
- - -
- -
&y(9O) % H
7 8 7 6 7 6 60
C
7 9 8 6 7 7 43
au(O) MPa H
133 117 100 142 157 172 56
C
120 116 129 133 149 180 63
OU(9O) MPaH
79 81 83 89 91 93 68
C
73 77 82
78 .
83 87 75
&u(O) % H
155 160 144 67 73 57 85
C
201 165 128 70 72 75 87
&u(9O) % H
218 218 218 100 100 100 52
C
- - - - -
-
-
cry yield stress (0) major axis
yield strain (90)
. .
Ey mtnor axIS
cm ultimate tensile stress H homopolymer
EU ultimate tensile strain C copolymer
Table 4.2. Sample Values From Creep Strain Models
Draw Area Draw Area
5 10
Heat Set Heat Set Heat Set Heat Set Heat Set Heat Set
Model
Time(S) Time(S) Time(S) Time(S) Time(S) Time(S) R2%
0 15 30 0 15 30
EC(O) % H
30.68 21.48 12.28 -1.85 0.56 2.98 89
C
49.11 38.91 28.70 -0.99 -0.05 0.90 93
197
Table 4.3. Sample Values From Shrinkage Models
Shrinkage
(%)
65C(O) H
e
6SC(90) H
e
85
0
e(O) H
e
85C(90) H
e
IOOC(O) H
e
IO00C(90) H
e
llOOe(O) H
e
llOC(90) H
e
(0) major axis
(90) minor axis
Heat Set
Temp
sooe
-0.08
0.19
-
0.06
-0.40
0.22
1.69
1.40
-2.13
-1.83
3.66
2.62
-3.36
-4.24
5.62
6.37
H homopolymer
C copolymer
no model
Draw Area
5
Heat Set
Temp
140
0
e
0.06
0.19
-
0.10
0.23
0.82
1.11
1.14
0.23
0.33
1.94
1.12
-0.72
0.24
2.58
2.34
Draw Area
10
Heat Set Heat Set Heat Set
Temp Temp Temp
200
0
e sooe 140
0
e
-0.06 0.37 0.51
0.19 0.19 0.19
- - -
0.07 0.31 0.35
0.56 1.78 1.73
1.41 2.32 1.90
0.92 1.93 1.35
0.88 1.81 1.55
2.60 6.81 4.23
2.49 5.47 3.52
0.22 3.66 1.94
-0.37 3.66 2.16
1.92 11.78 6.73
4.72 9.71 5.40
2.52 10.01 4.46
0.76 8.16 4.13
198
Heat Set
Temp
R2%
200
0
e
0.38 51
0.19 63
- -
0.32 72
1.39 88
1.48 91
1.16 61
1.29 33
1.64 88
1.58 95
0.22 35
0.67 42
1.67 83
1.10 95
1.89 73
2.55 76
Table 4.4. Sample Values For Crystallinity From Thermal Analysis
and Density Models
Draw Area Draw Area
5 10
Heat Set Heat Set Heat Set Heat Set Heat Set Heat Set
Model
Temp Temp Temp Temp Temp Temp
80C 140C 200C 80C 140C 200C
XO/ODSC H
56 52 51 55 39 31
C
41 45 47 41 45 47
X%DENS H
19 30 35 25 36 41
C
33 40 52 34 41 53
ec creep strain H homopolymer
X%DSC % crystallinity from
thermal analysis C copolymer
X%DENS % crystallinity from
density measurements
199
Rl
%
74
31
87
92
5 DISCUSSION
5.1 FED-STATISTICAL ANALYSIS AND MODEL PREDICTIONS
The direct output of an FED analysis is a set of parameters for insertion as
coefficients into a Taylor series. It is common practice at lel Wilton to
graphically represent the results of the analysis as a contour plot. Contour
plots allow the interaction of three variables to be represented on a single plot,
but closer examination demonstrated that they could mask potentially
important features that would be more obvious if expressed using a
conventional X-Y plot. Therefore, conventional plots were generated for all
significant models from the corresponding Taylor series, to facilitate a more
detailed analysis of each model. The X-Y plots are generated by fixing all but
one of the predictor variables in the model, and allowing the unfixed variable
to vary within the experimental range, so making the X-Y plots, in effect, 2-D
sections through the contour plots.
5.1.1 Mechanical Properties Models
The mechanical (tensile) property work was undertaken to ensure that heat-
setting would not detract from a bottle's ability to withstand the stresses
imposed during its service life. Therefore, parity of properties between as-
drawn and heat-set material was sought but, of course, any improvement in
properties with heat-setting would be considered advantageous. Some of the
properties for analysis are of direct practical relevance (e.g. yield stress) whilst
others were selected for their predicted susceptibility to orientation and heat-
setting (e.g: strain hardening rate). Conspicuous by its absence is Young's
Modulus. This had to be omitted because the full scale deflection on the chart
recorder was chosen to allow the ultimate strain to be recorded, so the gradient
of the elastic (low-strain) part of the plot was too great to measure with any
accuracy. Computerised data capture, now standard, would have allowed both
measurements to be made, but was unavailable at the time of testing.
200
5.1.1.1 PET Homopolymer
(A) Yield stress
Yield stress is probably the most important property related to strength, for
although rupture will not have occurred at this point in ductile materials such
as PET, a component will have permanently deformed and so deemed to have
failed.
The model predicts (figures 4.2a to 4.2c) that material with a draw area of 10
will have a substantially higher yield stress than that with a draw area of 5; this
would be expected and so lends support to the model. The effect of heat-
setting in the major axis is predicted to be temperature and time dependent.
Heat-setting from 6 to 15 seconds slightly increases the predicted yield stress
with increasing temperature for both draw areas. Heat-setting at higher
temperatures for 30 seconds is modelled to slightly reduce the yield stress for
both draws. The most significant effect though is the increase in yield stress,
especially for low heat-setting temperatures, when the heat-setting time is
increased above 15 seconds. This increase in yield stress is broadly in keeping
with Elenga et a/
53
although the annealing conditions used (225C for 20
hours) are very different to those used in this investigation. For times below
and including 15 seconds, increasing time reduces the yield stress for a draw
area of. 10. The effect of increasing the temperature at these extended heat-
setting times is predicted to reduce yield stress.
The yield stress model for the minor axis is too poor to merit further
discussion. The data (appendix A12) have an experimental error of
approximately 10% which is acceptable for this type of experiment. Therefore,
the reason for the poor model cannot be the quality of the data. The most
likely possibility is that yield stress in the minor axis cannot be described
solely in terms of the predictor variables used in the analysis.
Homopolymer yield stress in the major axis is not adversely affected, and can
be modestly enhanced by heat-setting (figure 4.2).
201
(B) Yield strain
Yield strain depends upon yield stress and modulus, which are two
independent properties of the material. Increasing the yield stress will increase
yield strain but for a given yield stress, increasing the modulus will decrease
the yield strain. This makes yield strain a very complex parameter to interpret
and is sometimes of only limited practical interest. Neither model (major or
minor axis) could be considered reliable, which could reflect the dependence of
yield strain on two material properties. Another possibility for the poor fits is
that the data do not display great variation and so the model is probably trying
to fit a data set in which the variability is mainly comprised of experimental
scatter.
The features and their explanation of the major axis model have already been
described in 4.1.1.1(C) and as both models fit the data rather poorly further
discussion is not merited.
(C) Ultimate tensile stress
Ultimate tensile stress (UTS) is frequently used as the single parameter to
describe a material's tensile properties, as it is the maximum tensile stress
which the yielded material can sustain without failing. As already mentioned,
the yield stress is of far more practical importance when considering the design
of an artefact.
202
The quality of the model for the major axis prevents an in-depth quantitative
analysis but the general trends are probably of significance (figures 4.4a to
4.4c). Increasing both heat-setting time and temperature is predicted to
increase the UTS for material with a high draw, but this effect is very limited,
for times of 30 seconds and (presumably) above. This is in keeping with the
idea that to reach a degree of heat-setting more time is needed at lower
temperatures, but for a draw area of 10, heat-setting for 30 seconds gives an
effect independent of temperature. This suggests that a maximum degree of
heat-setting, with respect to UTS, has been achieved. For samples with a draw
of 5, increasing the heat-setting time is predicted to increase the UTS, in
keeping with the higher drawn material. However, increasing the temperature
for a given heat-setting time for this material is predicted to reduce the UTS. It
is quite surprising to fmd material with draw ratio as the only difference, to be
predicted to behave in such a contradictory fashion. From this model it can be
deduced that to optimise UTS, the heat-setting cycle should be at relatively low
temperatures for extended times when both draw areas are considered.
The minor axis model (figure 4.4a to c) has a better fit than the major although
still far from ideal as a reliable predictive tool. The effect of heat-setting on
UTS is not very dramatic but as with the major axis, UTS is optimised by heat-
setting at low temperatures for longer periods of time. It is interesting to note
that at 30 seconds, increasing the heat-setting temperature decreases the UTS
for both (5 and 10) draws for the minor axis. This test direction will have a
lower draw than the major axis. The reduction in UTS is not as large as in the
major axis and so the rate of reduction ofUTS with heat-setting temperature is
not simply related to draw area.
UTS in both axes may be optimised by heat-setting for longer periods of time
at relatively low heat-setting temperatures. Heat-setting for shorter times
allows the temperature to be increased, improving the UTS for all but low
draws in the major axis, with only a small penalty in UTS for this case.
From the models discussed so far it can be seen that heat-setting, using the
most appropriate conditions, can improve the bottle's predicted ultimate
breaking strength.
203
(D) Ultimate tensile strain
The model predicts (figure 4.5 ) that material with the lower draw ratio
will show a higher ultimate strain (1400/0-160%) than material with the higher
draw (550/0-70%) This is entirely in keeping with expectation. The majority of
this difference is due to the amount that the material may cold draw. Extension
during cold drawing is made possible by the disentanglement and elongation
(orientation) of unoriented amorphous material. A drawing process (uniaxial
or biaxial) orients the amorphous phase and increases the crystallinity in the
material. Therefore, the amount of unoriented amorphous phase has been
reduced by two processes by drawing. The model predicts that strain only has
a small variation with heat-setting temperature and no effect with time.
The minor axis model predicts similar behaviour (except that ultimate strain is
completely independent of temperature, as well as time) but the fit of the model
is not as good.
Unlike yield strain, ultimate strain is not so strongly influenced by the
magnitude of the corresponding stress. This reversal in behaviour, between
yield and ultimate strain, gives weight to the argument that higher draw
samples have higher yield strains due to them having higher yield stresses.
204
(E) Strain hardening rate
Strain hardening, when it happens, is effectively the defonnation resistance of
the material after yield and cold drawing has occurred. As pointed out by
Samuels
100
, the tensile test will profoundly alter the initial structure. Also, if
large extensions have occurred, particularly if a neck has fonned, the cross
sectional area will have been reduced substantially. So, nominal stress
calculations will be subject to a large error, thereby underestimating the true
stress. Considering the above two points, it is quite surprising that strain
hardening rate is modelled as a function of any of the experimental parameters.
The primary effect is that of the draw area, but possibly more interesting is the
effect of heat-setting temperature and its dependence on draw. Increasing the
heat-setting temperature for a draw of 5 is predicted to decrease the strain
hardening rate, but the same increase is predicted to increase the strain
hardening rate for material with a draw of 10.
Strain hardening is primarily due to strain induced crystallisation taking place,
hence reinforcing the material. It is tempting to speculate that, because the
lower draw material has a lower degree of crystallinity and orientation, the
amorphous phase is more free to relax and increasing the temperature increases
the degree of disorientation. The more highly drawn material will have a
higher degree of crystallinity and a more oriented amorphous phase. The
crystallites will lock the amorphous phase in position
24
making disorientation
more difficult, and act as nuclei for further crystallisation. This would mean
that the same treatment would yield very different results and that strain
hardening rate is a parameter sensitive to these differences. A note of caution
should be sounded about unsubstantiated speculation, especially about a
complex parameter such as this, but collaborative evidence will be sought
elsewhere.
The strain hardening rate model for the minor axis was so poor that it did not
merit listing in the results chapter.
205
5.1.1.2 PET Copolymer
(A) Yield stress
The copolymer yield stress predictions (figures 4.7a to c) are very similar to
those for the homopolymer, except the overall stresses are a little lower and the
model has a limited dependence on heat-setting time. The copolymer is
designed to have longer crystallisation half time and so would be expected to
have a lower degree of crystallinity in these circumstances. Therefore the
copolymer would be expected to be less resistant to yielding.
The minor axis model is too poor to comment upon.
(B) Yield strain
The model predictions for the minor axis, and especially the major axis, model
predictions are very poor as demonstrated by the low coefficient of correlation.
The minor axis model echoes the homopolymer yield strain. Perhaps if a
model is required for the copolymer yield strain, the homopolymer models
could be substituted with acceptable results.
206
(C) Ultimate tensile stress
As expected, the higher draw areas are predicted to have a higher UTS, and
increasing the heat-setting temperature increases this (figures 4.9a to c). The
predicted UTS for the copolymer is of a similar magnitude to that of the
homopolymer. The model also predicts that the UTS is independent of heat-
setting time, whereas the homopolymer does show a time dependency. In
addition. the decrease in UTS with heat-setting temperature for a draw of 5,
which was seen in the homopolymer, is not repeated with the copolymer.
Neither model is a good fit, and so too much importance should not be attached
to these differences.
The minor axis model (figure 4.9a to c) is a better fit than the major axis
equivalent and does show both temperature and a slight time dependency.
Once again the predicted UTS for the high draw is greater than that for the
lower draw and the minor axis is predicted to have a lower UTS than the major
axIS.
(D) Ultimate tensile strain
This model has a very good fit with the data and predicts that heat-set time is
not a significant predictor variable (figure 4.10). The effect of temperature for
a draw of 10 is to slightly increase the strain but for a draw of 5 it significantly
reduces the strain. The behaviour of draw area 10 specimens is similar to that
of the homopolymer but the behaviour seen for a draw area of 5 is not. Under
these circumstances therefore, ultimate tensile strain of the homopolymer does
not vary significantly with heat-set temperature. The model predictions for the
major axis for each polymer are quantitatively similar.
No model is presented for ultimate strain in the minor axis.
207
(E) Strain hardening rate
The model (figure 4.11) has a good fit and is quantitatively similar to the
homopolymer. The fact that the draw dependency seen in the homopolymer
model for strain hardening rate is not duplicated for the copolymer, rather
counts against the theory speculated in 5. I. I. I.(E).
The tensile property models provide a clear indication that heat-setting does
not adversely affect a bottle's ability to function in the same environment as a
standard PET bottle. Where the modelling process has been successful, the
FED analysis has given detailed information on a large experimental window
from less trials than would have been required by conventional
experimentation techniques. Where no useful model has been generated, due
to poor correlation, then no information may be gained, leaving some gaps in
understanding. Unfortunately the intractability of the output rather works
against obtaining reasons for these failures or more fundamental information
about the materials and the effect that the heat-setting process has on them.
The ultimate tensile stress and strain hardening rate for the homopolymer and
the ultimate tensile strain for the copolymer models show a trend. Material
with a draw area of 5 in only the major axis behaves in opposition to that with
a draw of 10 or either draw in the minor axis, where modelled.
5.1.2 Creep Models
The models for each material successfully fit the data from which they are
derived and so can be used for prediction with some confidence. Figure 4.12
demonstrates that for a draw area of 5, whilst the copolymer has a higher
overall level of creep, both materials are predicted to respond to an increase in
heat-setting temperature in an identical way. The predictions for material with
a draw of 10 are also similar.
208
The plot of the effect of draw area on creep, (figure 4.13) emphasises how
dependent creep strain at elevated temperature is on draw and material grade.
However, in the bottle situation it should be remembered that the areas of low
draw (i.e. the neck and base regions), are thicker (thereby reducing the creep
stress, for a given internal pressure). So, this to some extent compensates the
direct effect of creep resistance.
For draw ratios up to approximately 10, the models show that the copolymer is
more prone to creep than the homopolymer under equivalent conditions. The
reason for this could be due to lower crystallinity levels due to either the
copolymer's longer crystallisation half-time or higher natural draw. However,
the crystallinity models from the density measurements (figures 4.23a to c)
show that for heat-setting times of up to 6 seconds, there is little difference in
predicted crystallinity for the two materials. For longer heat-setting times the
copolymer is predicted to have a higher degree of crystallinity. Both materials
exhibit greater creep strains at lower draws. Creep is considered to occur in
the amorphous phase under these conditions. The reduced orientation of this
phase, probably coupled with reduced pinning by a lower degree of
crystallinity at the lower draws, may account for this increased creep. The
models also show that greater heat-setting times can significantly reduce the
creep, particularly at low draws, in both grades of material. This strong
dependence of creep strain on draw ratio and the improvement brought about is
also reported in the literature
63
. Cakmak and Wang also reported that the
reduction in creep strain with heat-setting was greatest for the unoriented
material. The creep models produced from this work are in full agreement with
this.
209
Possibly the most striking feature about both of these models is that heat-
setting time is a very significant predictor variable for creep, and that
temperature is not significant. Increasing the heat-setting time is predicted to
substantially reduce the creep strain of each polymer, until the draw becomes
large enough to eliminate creep strain under these conditions. Creep is the
only measured property where this has been the case. The models are similar
for each grade, with each variable chosen in the same order and they have good
fits with the data set (R2 approx. 90% in both cases). These considerations
make the significance of heat-setting time in this experiment to be far more
than any possible artefact from the FED. Although heat-setting temperature
does not appear to be significant, there must be a minimum temperature that
will actually cause any effect at all. Cakmak and Wang
63
point out that
increasing the stretch ratio decreases the activation energies of PET films at
any stress level. This would mean that highly stretched samples, such as those
used here, would show lower creep strain dependence on heat-setting
temperature.
Heat-setting is generally considered to be a combination of
meltinglrecrystallisation and modification of the amorphous phase, ignoring for
the moment, the work of Salem and Weigmann
91
and Buckley and Salem
25

Both these processes would lead to a reduction in shrinkage which is the
parameter most often used to investigate heat-setting. As heat-setting
dramatically improves the creep properties of the material, the overall
contribution from any disorientation of the amorphous phase to improved
shrinkage properties must be negligible. This assumes that the contradictory
work
25
, 75, 91, 97 is incorrect and that the creep properties are dependent on the
same microstructural features as shrinkage.
210
It would seem that the network formed by the crystalline regions significantly
reduces the creep strain, by anchoring various portions of the chains in the
amorphous regions. Buckley et a/ 25, 97 considered heat-setting not to be a
melting/recrystaIlisation phenomenon but a high-temperature mechanical
relaxation process, probably composed of entanglement slippage. This lead to
a stiffening of the crystalline regions, acting both as inert reinforcing filler
particles and also as cross-link sites for non-crystalline molecular segments.
The evidence for this was based only on shrinkage recovery work and so no
explanation was given of how heat-setting reduces creep strain. Salem and
Weigmann
91
found, using optical dichroism, that heat-setting increased the
orientation of the amorphous phase, in agreement with the creep data discussed
here. Vallat and Plazek
75
, using DSC, countered the argument of
melting/crystallisation, but as the work was confmed to DSC, the implication
for creep strain is not clear.
However, the strong time dependence and lack of temperature dependence of
the creep models is unusual, and points to creep being influenced by
microstructural features other than those that affect the tensile or shrinkage
properties. SAXS and diffusion studies have shown that, for fibres, the
interfibrillar domains play an important role for the mechanical properties. The
microfibrils themselves were shown to primarily contribute to dimensional
stability at elevated temperatures
37
. This demonstrates the concept of different
microstructural elements affecting different properties.
It is doubtful that the "negative" values of creep predicted for both materials
heat-set for less than 10 seconds (figure 4.12) or for material with a draw area
of approximately 9 (figure 4.13) have any physical significance, since no
negative values were recorded in the experiment (appendix AI9). They are
more likely to have arisen from the models having to cope with quite a
dramatic reduction in strain with draw area and so have overshot in the draw
area region of 7.5 to 10 of figure 4.13. This is supported by the models
predicting an increase in creep for material with an area draw greater than 10.
211
The models generated for the prediction of creep are good quality models, with
a high level of correlation. They demonstrate that far from adversely affecting
the creep properties, heat-setting is very beneficial, particularly for material
with a relatively low draw ratio. The models' time dependency, coupled with
their lack of temperature dependence, poses some interesting questions about
the interaction of creep and microstructure, and the evolution of the latter
during heat-setting. These models indicate that to minimise creep strain, a
bottle design should avoid large areas of low draw in the design, preferably
produced in the homopolymer. The heat-setting which need only be
concentrated on low draw areas should be applied for relatively long periods of
time.
5.1.3 Selected Mechanical Properties as a Function of Crystallinity
Heat-setting has been shown to affect mechanical properties, and also
influences crystallinity (sections 4.1.4.2 and 4.3.2.2). To examine the
relationship between mechanical properties and degree of crystallinity
therefore, the models for yield stress, and creep strain, for each material, have
been plotted against the corresponding model for crystallinity obtained from
the density data.
5.1.3.1 Yield Stress as a Function of Crystallinity
The yield stress model for each material in the major axis has been plotted
against the degree of crystallinity (figures 5.1 and 5.2). This has only been
possible for the major axis due to the poor correlation of the minor axis yield
stress models. The plot was generated by keeping any 'time' component in the
models constant at IS seconds, and allowing the 'temperature' component to
vary from 80C to 220C.
212
The plot for the homopolymer (figure 5.1) shows the models to predict yield
stress as a linear function of degree of crystallinity. However, the predictions
for a draw area of 5 do not coincide with a draw area of 10. Drawing produces
increased crystallinity and orientation, and so the difference in the two draw
areas in this plot is a representation of the contribution to yield stress made by
orientation (approximately 32 MPa). These plots are based on model
predictions and so care must be taken not to extract to much information from
them. However, the homopolymer plot does suggest that yield stress is also
controlled by orientation but in addition degree of crystallinity has a significant
effect. The copolymer plot is slightly complicated by the fact that crystallinity
is not a linear function of temperature as is the case with the homopolymer
(figure 4.23b). However, the roles of orientation and crystallinity in
determining yield stress is maintained. The typical difference between the two
draw ratios for the copolymer is constant with respect to crystallinity, at
approximately 27 MPa.
5.1.3.2 Creep Strain as a Function of Crystallinity
The creep strain versus degree of crystallinity plot (figure 5.3.) generated from
the copolymer creep and crystallinity models by holding the 'heat-setting
temperature' component of the models constant at 200C and allowing 'heat-
setting' time to vary from 0 to 30 seconds. It was not possible to produce a plot
for creep versus crystallinity for the homopolymer as the model for crystallinity
was time independent. The copolymer plot shows for a draw area of 5, that,
although creep strain is dependent on crystallinity, the effect of orientation
prevents the predictions for draw areas of 5 and 10 against crystallinity, from
coinciding. Crystallinity is not predicted to be correlated to crystallinity for a
draw area of 10. Cakmak and Wang63 found that for a film of unbalanced
draw, as the orientation in one axis was reduced, (measured by refractive
index) the creep strain in that axis increased. Clearly, this agrees with this
investigation, in that chain orientation is a major factor in the determination of
creep strain.
213
The predicted slight increase in creep strain with increasing crystallinity is a
consequence of the creep model 'over-shooting' the effect of draw (figure 4.13).
5.1.4 Shrinkage Models
The reduction of thermally induced shrinkage by heat-setting is the primary
objective of the project. The production of reliable models from the FED
analysis would be very desirable for two purposes: firstly to give information
on the success or otherwise of the heat-setting experiments, and secondly, good
models would be of distinct commercial value, as they would make excellent
tools for designing shrink resistant PET containers.
5.1.4.1 PET Homopolymer
(A) Shrinkage at 65C
A temperature of 65C (figure 4.14) is below but close to the Tg of PET. No
significant movement appears to have occurred. The analysis is therefore
trying to fit data that has, as its main source of variance, experimental error.
Both models (major and minor axis) have a poor correlation hence do not merit
discussion.
214
(B) Shrinkage at 85C
The immediate effect of exceeding Tg can be seen (figures 4.15a to d) as a
dramatic improvement in the model quality. The major axis model shows no
significant dependence on time. It predicts that increasing the heat-setting
temperature for a draw of 10 reduces shrinkage at 85C but increases shrinkage
for a draw of 5. It is difficult to explain why the effect of heat-setting
temperature on shrinkage should be draw dependent.
Examination of the minor axis model shows that the same draw dependency
does not exist. The linear draw in the minor axis for an area draw of 5 will be
lower than in the major axis and so it is incorrect to assume that below a
certain draw heat-setting increases shrinkage. The minor axis model does
show some dependence on time. Increasing the setting time decreases slightly,
the minimum achievable shrinkage. For example, increasing the heat-setting
time from 6 to 30 seconds reduces the minimum achievable shrinkage from 1%
to 0.6% and from 1.25% to 0.8% for draw areas of 5 and 10 respectively.
(C) Shrinkage at lOO"C
The major axis model demonstrates no dependence on heat-setting time
(figures 4.16a to d). At this test temperature heat-setting is predicted to have a
substantial impact on shrinkage. The anomalous result of increasing heat-
setting temperature increasing the shrinkage for a draw area of 5 (seen at 85C)
is confirmed here. Also confirmed is that for a given heat-setting time, this
effect is confmed to the major axis.
215
The minor axis model also shows that heat-setting can reduce shrinkage. The
effect of temperature is as-before, for the minor axis. The effect of time is
rather more complex and appears to sensitise the material to the heat-setting
temperature.
(D) Shrinkage at HOoe
The major axis model continues the trends seen at the lower test temperatures,
with an overall increase in the magnitude of shrinkage. It is independent of
time and the draw ratio will determine if heat-setting increases or decreases
shrinkage (figures 4.17a to d).
The minor axis also continues most of the trends, seen in this axis previously.
The effect of time is a little different, compared to the 1000e case, as
increasing the temperature no longer increases the shrinkage for a time of 30
seconds.
The pair of models for llOoe predict that shrinkage will be independent of
draw and time for the homopolymer heat-set at approximately 180
o
e. The
resulting shrinkage will be the same, regardless of which axis it was measured
In.
The considered models (65e to 110C, major and minor axis) for the
homopolymer exhibit some interesting trends. The major axis models are not
dependent on time, whereas the minor axis models are. The major axis models
always have a better fit than their minor axis counterparts (figure 4.lb). The
draw area determines whether heat-setting temperature increases or decreases
shrinkage in the major axis, (as described in section 5.1.3.1A), and is continued
throughout the test range. This continuation makes the effect now difficult to
describe as an anomalous result.
216
The implication of this trend is that the material's behaviour in one direction is
not solely a function of the linear draw in that direction, or the minor axis
models would all predict an increase in shrinkage with heat-setting
temperature. Not only is the magnitude of the linear draw important but if it
constitutes the major or minor axis also appears relevant. Consequently the
two linear draws of a biaxial stretch cannot be considered as independent, with
respect to the effect of heat-setting on shrinkage.
5.1.4.2 PET Copolymer
(A) Shrinkage at 65C
As with the homopolymer at this temperature there is little movement occurring
below Tg and so the model is of little use (figure 4.18).
(8) Shrinkage at 85C
The major axis model fit has been dramatically improved with the increase in
temperature (figure 4.19a and b). The predicted shrinkage for a draw area of
10 is similar to the homopolymer. Another striking similarity in the major axis
is that increasing the heat-setting temperature increases the predicted shrinkage,
for material with a draw of 5.
The minor axis has a disappointing fit but the general trend is for heat-setting
temperature to reduce shrinkage, for both draws.
217
(C) Shrinkage at lOOe
The major axis model is time independent (figures 4.20a to d). It is in keeping
with all the previous major axis shrinkage models in that the draw determines if
increasing the heat-setting temperature will increase or decrease the shrinkage.
The minor axis shows both heat-setting time and temperature-dependent
behaviour. It is similar to the homopolymer model in that at high temperatures,
increasing the time decreases shrinkage, but the reverse happens at lower
temperatures.
(D) Shrinkage at HOe
The models in both axes follow the trends established at the lower test
temperatures, and for the homopolymer. The major axis has an excellent fit
and demonstrates the magnitude of the dual effect that draw area is predicted to
have on shrinkage.
The minor axis also follows the previously stated trends.
218
5.1.4.3 General Comments and Interpretation
(A) Comparison of Materials
It can be seen that the two materials behave in a similar fashion with respect to
shrinkage and heat-setting. The models clearly show that shrinkage may be
significantly reduced by the process of heat-setting but its effects are complex
in nature. The survey of the shrinkage models does highlight some important
similarities and trends. A comparison of the materials shows that the
homopolymer is predicted to shrink slightly more than the copolymer at low
heat-setting temperatures. At higher temperatures there is no overall
discernible trend. The crystallinity results from the FED work (appendix A24)
show that the degree of crystallinity levels in as-drawn samples are very similar
for both materials, the homopolymer has only 3% more. The models for
crystallinity (by density) predict that the two materials respond to drawing and
heat-setting in a very similar way. They also predict that the degree of
crystallinity does not alter substantially when the draw area is increased from
4.5. to 11. "This is because the majority of the crystallinity is already developed
before a draw of 4.5 has been reached. From this it can be seen that
crystallinity cannot alone be used to account for differences between materials
or differences between draw ratios for each material. From these results it may
be concluded that both the homopolymer and copolymer may be processed
under similar conditions. Orientation, as measured by birefringence and
refractive indices (figures 4.38 and 4.39) does show an increase with both draw
ratio and heat-setting temperature. This is developed further in sections 5.2.5.3
and 5.2.5.4.
219
(B) Major Stretch Axis
For the major axis, predicted shrinkage is not a simple function of draw and
heat-setting time and temperature. All the major axis models predict that a
homopolymer or copolymer with a draw area of 10 will shrink less, if the heat-
setting temperature is increased throughout the experimental range. However,
they all predict the converse for material with an overall draw area of 5, for the
same increase in heat-setting temperature. This infers that the comment made
for the homopolymer, about dependence of the two draws in a biaxial system,
must also be applied to the copolymer. The majority of this modelled increase
in shrinkage with increasing heat-setting temperature is in fact the reduction of
negative shrinkage, the origins of which are discussed further in section 5.2.3.
All of the major axis models (both materials) are heat-setting time-independent,
and predict material heat-set in the range 175C-190C to shrink to levels of the
order of 20/0-3%, regardless of draw or test temperature, for test temperatures
of 100C and above. For heat-setting temperatures of 100C and above the
shrinkage in the major axis for draw ratios of both 5 and 10 are predicted to
converge and cross. At 100C the convergence point for both materials is at
approximately 190C and 2% shrinkage. For a temperature 110C the
convergence has risen to 2.5% shrinkage but the two materials now have
different convergence temperatures, separated by about 10C. The major axis
models always have a better fit than their corresponding minor axis models.
(C) Minor Stretch Axis
The minor axis models are also very similar and those with only a reasonable
fit show a time-dependence. This is in complete contrast to the major axis
models. All the minor axis models for a given heat-setting time show a
decrease in shrinkage with increased heat-setting temperature. The draw only
affects the degree of reduction, not the type of effect, as with the major axis.
Furthermore, up to and including a test temperatures of 100C and above, for
the homopolymer, and for all test temperatures for the copolymer, all the
models predict parallel behaviour for the two discussed draws.
220
This means that the rate of change in shrinkage with heat-setting temperature is
independent of draw in the minor axis. The rate of change of birefringence
with heat-setting temperature for draws of 5 and 10 (in the minor axis) is
approximately equal (estimated from figures 4.38a and b). This could account
for the same rate of change in shrinkage with heat-setting temperature.
(D) Overall Comments
To minimise shrinkage in PET bottles heat-setting temperatures should be high
(200C) to both reduce and unify the behaviour of both axes across a wide
range of draw ratios. There is little difference between the two grades of PET
material, with no material constantly showing a higher shrinkage than the
other. Sample values from the shrinkage models are given in table 4.4. The
fundamental mechanical property that determines a materials suitability for an
application is yield stress. Heat-setting at high temperatures to give the
minimum shrinkage is also predicted by the models to modestly increase the
yield stress for both materials (figures 4.2 and 4.7), although information about
the minor axis is limited. This means that no compromise between shrinkage
and yield stress is required in selecting heat-setting conditions for either
material.
From the comparisons of the draw axes two important pieces of information
can be seen. Firstly, in the major axis, increasing the heat-setting temperature
can either increase or decrease shrinkage and this is predominantly determined
by the draw ratio. In the minor axis the same increase in temperature will only
decrease shrinkage. Secondly, the shrinkage models in the minor axis
consistently demonstrate a dependence on heat-setting time, an effect which is
not shown in the major axis.
These two facts provide strong evidence to support the view that in a biaxially
oriented system the properties one direction will be strongly influenced by the
draw ratio in the perpendicular direction. For that is for a given overall draw
ratio therefore, the ratio between each principal draw axis will influence the
properties attained.
221
It will have been noticed that some of the models predict an amount of negative
shrinkage (expansion). This is not an erroneous result from the FED analysis
as negative values were recorded experimentally. Negative shrinkage in
oriented PET samples has been reported in the literature by Peszkin et aiB,
Petermann and Rieck
51
and by Biangardi et aP. Discounting the major axis
low draw samples as "special cases", then all the negative shrinkage values are
predicted to occur in samples with a high heat-setting temperature, in close
agreement with Peszkin
8
Peszkin et al observed that for heat-set PET fibres
which exhibited positive shrinkage, that the initial crystallinity is completely
lost in the first lOms of the heat treatment. For fibres that demonstrated
negative shrinkage the initial crystallinity was retained. Petermann et a/
51
recorded continuous negative shrinkage for uniaxially oriented heat-set PET in
the perpendicular direction, using TMA. The as-spun material did show
shrinkage in both axes. They made no correction for thermal expansion but
their account for the difference in behaviour of-the two axes is interesting.
They proposed that crystallisation caused by heat-setting is enhanced by a
preferred nucleation onto the side surfaces of the micellar blocks whose
crystallographic alignment is nearly coincided with the draw axis. This means
that the material between the original micellar block is now completely
crystallised, forming a continuous lamellar morphology, and so further
crystallisation or relaxation of the amorphous phase in this direction cannot
occur. Regrettably they give no account of their results showing that the as-
spun fibres exhibit far less shrinkage (2.6%) than the heat-set (7%), which is
contrary to expectation.
222
5.1.5 Selected Shrinkage Models as a Function of Crystallinity
Shrinkage models at 110C for both materials in both axes have been plotted as
a function of their corresponding degree of crystallinity models (figures 5.4to
5.7). These plots were generated by keeping any 'time' component in the
models constant at 15 seconds and allowing the temperature components to
vary between 80C and 220C.
5.1.5.1 Major Axis Shrinkage Models as a Function of Crystallinity
The homopolymer plot (figure 5.4) shows that shrinkage is predicted to vary
linearly with degree of crystallinity. The two draw areas are predicted to react
very differently to the increase in shrinkage demonstrating that crystallinity is
not the sole parameter that determines shrinkage in this case. This is echoed in
the copolymer plot (figure 5.5), although the non-linearity of the copolymer
crystallinity model complicates this. This is in agreement with the plots of
yield stress and creep strain when plotted against crystallinity (figures 5.1 to
5.3).
5.1.5.2 Minor Axis Shrinkage Models as a Function of Crystallinity
The minor axis shrinkage models are produced in the same way as the major
axis, and are plotted in figures 5.6 and 5.7. The homopolymer plot does
indicate that for a level of crystallinity above 35%, the two draw areas have
similar shrinkage. The copolymer plot (figure 5.7) shows that for crystallinities
above 25%, shrinkage is constant with respect to crystallinity and that both
draw areas will exhibit very similar, low, shrinkage levels. This plot indicates
that not only is crystallinity not determining shrinkage but, neither is draw area
to any great extent.
223
The shrinkage versus crystallinity plots reinforce the results from the
mechanical properties versus crystallinity plots (figures 5.1 to 5.3), in that
crystallinity, whilst contributing to the property in question, it is not the sole
deterrninator.
5.1.6 FED Crystallinity Measurements
5.1.6.1 Thermal Analysis (DSC)
The model for predicting the degree of crystallinity by DSC for the
homopolymer has quite a good fit with the data. The model does however,
contradict common expectation by predicting a reduction in crystallinity with
increasing heat-setting temperature, for times in excess of 15 seconds (figure
4.22c). This gives the model no credibility for the prediction of degree of
crystallinity. However, the homopolymer model has a good fit and so the
problem appears not to be with the model, but with the raw data.
The literature is in some disagreement about not only the heat of fusion for
PETs3, 63 but more profoundly, the source of the secondary melting peak
94
, 75,
30. The poor model for the copolymer is in contrast to the homopolymer. All
the experiments were performed in a random order, so the reason for this is not
easily understood. When the inconsistencies of the homopolymer model are
added to this, the measurement of the degree of crystallinity by DSC appears
not to be a reliable technique for oriented, heat-set PET. This unreliability is
due to either inaccuracy of the theory or experimental error (either in procedure
or identification of baselines). A survey of the relevant literature may be found
in sections 2.4.2 and 2.5.3.5, and shows that the opinion on the origin of the
secondary endotherm is far from settled. The experimental error can be seen
by the repeat runs in the FED and is considered small enough not to be the
source of the problem. The experimental error associated with DSC is more
fully discussed in section 5.5.2.
224
As a point of speculation, for all the properties that have been plotted against
crystallinity (by density), crystallinity has not been the sole detenninant of the
property. Figure 5.20 shows that shrinkage in the heat-set bottles is strongly
correlated to heat-setting temperature but figure 4.42 shows the relationship
between crystallinity (measured by DSC) and temperature is a very poor one.
This shows that heat-setting modifies the PET morphology, as well as
increasing the degree of crystallinity, in a way that directly affects shrinkage.
It is not impossible that the DSC results are affected by this other
morphological change or its interaction with the crystallisation that occurs
during the DSC run. Figure 5.21. shows that for heat-set bottles, degree of
crystallinity measured by density and heat-set temperature correlate very well.
5.1.6.2 Density Measurements
Because the degree of crystallinity data from the DSC scans were unreliable, it
was necessary to perform density measurements. Density as a crystallinity
measurement technique is not without its criticisms (2.5.3.4) but is the
generally accepted industrial method. Assuming that any change of density of
the amorphous phase may be discounted, because of the high draws used for
the majority of the work, then these models would be far better predictors of
crystallinity.
By comparing the raw density data for the as-drawn homopolymer and
copolymer samples (FED run No 14 in appendix A24) it possible to obtain a
comparison of crystallinity between the two materials at this draw area (7.75).
The homopolymer has about 3% more shrinkage than the copolymer, which is
probably significant. There are two possible reasons for this; either the faster
crystallisation kinetics of the homopolymer give rise to higher crystallinity
levels or, because the homopolymer has a lower natural stretch ratio, for a
given draw the homopolymer will have undergone greater strain induced
crystallisation. It is possible that both of these are occurring, but it should be
remembered that the overall effect is not great.
225
The homopolymer model for crystallinity by density is a good fit and is time-
independent. The model predicts that increasing temperature increases
crystallinity linearly (figure 4.23a to c), and that samples with a high draw area
have approximately 6% more crystallinity than the low draw samples. This is
broadly in agreement with expectation, although the time-independence is
surpnsmg. This infers that for both draws the degree of achievable
crystallinity is determined by temperature and is obtained in a shorter space of
time than the FED analysis can resolve.
The copolymer model (figure 4.23a to c) has an even better fit (R2 = 92%) and
does show some time-dependence for heat-setting times in excess of 15
seconds. This model predicts that the crystallinity levels are independent of
draw area, for all heat-setting temperatures. The FED programme does not
show if strain-induced crystallisation is draw dependent, or if heat-setting
makes the crystallinity levels comparable.
Increasing the heat-setting time to 30 seconds from 6 seconds is predicted to
increase crystallinity by approximately 10%. It is well known that density can
over estimate crystallinity due to the densification of the amorphous phase
during heat-setting
I27
. This over estimation may therefore, become more
significant as the draw area is reduced. Changes in density were mainly
associated with crystallisation by De Vries et aP.3 as the elongation ratio
exceeded 2.5. The draws used in this investigation were predominantly far
greater than this. The rate of densification is unlikely to coincide with the rate
of crystallisation and could be a reason for this time dependency of the
copolymer's density and inferred crystallinity. The results suggest that for the
copolymer, the process of heat-setting could be a two stage process. The first
stage would be a rapid, thermally-activated and controlled process, which
would be substantially completed under, in this case, 6 seconds (the shortest
time interval in FED analysis). The second process, which commences after 15
seconds in these experiments, is time-dependent, possibly similar in nature to
that outlined by Buckley et a[25. 97. In this theory of heat-setting the
mechanism is not meltinglrecrystallisation but a mechanical relaxation. which
was considered time-dependent. This theory has been previously discussed
(section 2.4.2 and elsewhere).
226
The copolymer is designed to have a longer crystallisation half-time than the
homopolymer. This could account for the crystallinity level in the copolymer
being time-dependent, in contrast to the homopolymer. The lack of draw
dependence and the greater overall crystallinity levels at 30 seconds heat-
setting time for the copolymer are not easily explained.
The evidence for a two stage process can be seen in Imai et al
147
who have
recently found that when heat-setting at 80C there is an initial induction
period, prior to the onset of further crystallisation. The induction period is
associated with the formation of a long range ordered structure. For a heat-
setting temperature of 115C148 they found that no local ordering took place.
Further evidence for a two stage process is found in Peszkin et afS. They found
that for samples which did shrink, the initial crystallinity was lost in the fIrst ID
ms of the heat treatment, prior to the commencement of disorientation and
crystallisation.
The literature reviewed in 2.4 presents two mechanisms: one of which is heat-
setting time independent, and one which is time dependent. The literature
presents these as mutually exclusive, but the density results, presented here,
tentatively suggest that both could operate semi-sequentially.
5.1.7 Comments On the Implications Of Using FED Analysis
The use of FED provides an alternative experimental method to the
scientifIcally 'hallowed' approach, of examining the effect of one variable by
keeping all other variables constant. For a given size of experimental window,
the FED allows investigation with fewer trials than with the conventional
technique. This advantage becomes far greater as the number of variables is
increased. The FED includes in its analysis the interaction of variables,
something which is not easily achieved by the conventional approach.
227
The output of the FED is in the fonn of a model constructed from a truncated
Taylor series. Providing the model has an adequate fit, it can be directly used
to control a system or process. FED is not without disadvantages, however. It
is a statistical process, requiring knowledge that most people other than
specialist statisticians would not be aware of, to optimise it. Indeed this project
is indebted to the 'Recite' team at lel Wilton for their help and advice in this
area of the project. The technique is rather 'black box' in nature making
validity of the result difficult to assess. It is quite conceivable that an
investigation that was undertaken using FED without specialist support could
generate apparently good models, which could be flawed by variable
correlation etc. The model makes the relationship of each predictor variable
difficult to assess. In the conventional technique, proportionality (e.g. linear,
quadratic or logarithmic) can usually be identified where present. It is the
identification of these relationships that contributes to the physical
understanding of a system.
From the above considerations it can be seen that FED represents an excellent
method of investigating a process with a view to controlling it. It would be
especially applicable to some industrial problems where 'how' a system reacts
to various inputs, is more important than 'why'. It is also a recommended
technique for a preliminary investigation, as a lot of infonnation about the
significance of each variable may be obtained with relatively few trials. The
intractability of the models makes fundamental infonnation difficult to obtain.
It is interesting to note that, despite the reduction in trials required by FED, the
vast majority of research published in journals, with few a exceptions
l49
, 150,
151, is still conducted using the conventional experimental methodology.
Having used the FED to identify the most significant variables it was decided
to investigate these further for the copolymer, using a conventional
experimentation series.
228
5.2 CONVENTIONAL EXPERIMENTATION SERIES
The design of this series of experiments reflected the results of the FED work
in that the two variables selected for investigation were draw area and heat-
setting temperature; i.e. the two most influential variables on the shrinkage
behaviour of heat-set PET. The copolymer was chosen for this investigation as
it was anticipated that it would be more commercially relevant in the future.
The design of this series allowed for both undrawn and unheat-set material
which had not been included in the FED programme in any detail. The full
experimental design is summarised in table 3.3.
5.2.1 Shrinkage Measurements
(A) Shrinkage at 65C
The major axis results (figure 4.'24) show a large degree of scatter within only a
small percentage of movement. The minor axis shows far less scatter, which is
in contradiction to the model fits from the FED analysis. All the data points
from heat-set material indicate that expansion, and not shrinkage, has taken
place. The coefficient of thermal expansion for isotropic copolymer has
already been subtracted, and so this "expansion" must be an effect of orienting
and heat-setting the material. The samples with a draw area of 5 show more
expansion than those of 10 so expansion is not a simply related to the level of
orientation. The expansion seen cannot be due to the material creeping during
the test, as the same draw area (5) tested in the minor axis (lower linear draw)
displays positive shrinkage, not expansion. The effect of increasing the heat-
setting temperature is to partially converge the shrinkage for both draws. That
is, it decreases shrinkage for a draw of 10 but increases it for a draw of 5. The
FED model does not provide a useful prediction at this temperature.
229
(B) Shrinkage at 85C
Increasing the heat-setting temperature converges the shrinkage of the materials
with draw areas of 5 and 10 (appendix Bland figure 4.26). The sample with a
draw area of 5, as-drawn, still shows negative shrinkage. The FED model
(figure 4. 19a) is in agreement with these trends, although it underestimates the
shrinkage of the as-drawn material. For heat-setting temperatures above 160C
there is good agreement between the model and experimental data for both
draw areas.
The minor axis results (appendix B2 and figure 4.27) are all positive, unlike the
major axis. The model (figure 4.19a) and data are in good agreement,
especially at higher heat-setting temperatures. The minor axis does have one
data point that indicates an increase in shrinkage when heat-set for 100C,
which is seen for all test temperatures in this axis.
(C) Shrinkage at 100C
Increasing the heat-setting temperature causes the shrinkage results to converge
to about 2% for both axes (appendix Bl and B2 and figures 4.28 and 4.29) as
predicted by the models (figure 4.20b). Again the models underestimate the
shrinkage of the as-drawn material but for heat-setting temperatures above
160C the fit is good, especially in the minor axis.
(D) Shrinkage at 110C
The major axis data (appendix BI and figure 4.30) and model (figure 4.2Ib)
are in good agreement. The model prediction (value for the convergence
shrinkage) is accurate but the model then predicts a divergence that is not
shown in the data. The negative shrinkage in the model is seen in the data but
only at lower setting temperatures.
The minor axis model (figure 4.21b) is in good agreement with the data
(appendix B2 and figure 4.31) over the whole FED temperature range.
230
The conventional experimentation series has confmned the majority of the
trends from the FED models, inferring that the models are useful predictive
tools under these circumstances. Both sets of work indicate that minimum
shrinkage is obtained at high heat-setting temperatures. The conventional
experimentation series has shown that, at these temperatures, shrinkage is
independent of draw area and axis. This implies that uniform (distortion free),
low shrinkage in a bottle which has a continuous range of draw ratios may be
obtained by heat-setting at a high temperature.
The expansion and increased shrinkage predicted by the FED for samples with
a draw of 5 in the major axis only, was (surprisingly) confmned by this work.
Because the conventional experimentation series was produced in the same
way as the FED series, then the differences between the model predictions and
the conventional experimentation results (above) that would be expected from
experimental error, must be due to the modelling process.
It was decided to examine the effect of these linear draws in terms of
theoretical bottle volume shrinkage and the ratio of shrinkage between the
major and minor axes. As part of the conventional experimentation series one
sample ( No 31) was shock cooled using liquid nitrogen, and one sample (No
32) was heat-set under relaxed tension then re-heat-set. Neither of these
samples has stood out from any of the trends observed in this work.
231
5.2.2 Theoretical Volume Shrinkage
For commercial acceptance dimensional changes have to be controlled to a
maximum of 5% volume shrinkage. A bottle has a continuous range of draw
ratios from completely undrawn in the screw thread region, through to 11 or
higher in the side wall. For the purposes of the following calculations, a bottle
will be considered to comprise one draw area only. Two such bottles will be
considered, one with a draw area of 5 and one of a draw area of 10, with a
draw ratio of 1.5: 1 between the hoop and axial axes. In this way 'best' and
'worst' case conditions may be approximated.
5.2.2.1 Volume Shrinkage Calculations
If a carbonated soft drinks bottle is approximated to a cylinder then volume
shrinkage is given by:
where:
R = original radius
(. m
2
1
Sv = lltR2L)
L = original length (bottle height - parallel section only)
r = radius after shrinkage has occurred
I = length after shrinkage has occurred
232
Treating a bottle as a cylinder is a close approximation for the wall area of a
bottle, but does not allow for the variation of draw found in the neck and base
regions. Two draw areas, 10 and 5, heat-set for 15 seconds, are discussed in
the following sections. The higher draw is close to the draw found in the wall
panels of the 0.51 bottles used in this project, and the draw area of 5
approximates the lower draws in the neck and base.
5.2.2.2 Volume Shrinkage From FED Models
For a draw area of 10 (figure 5.S), all the models predict that heat-setting can
significantly reduce shrinkage. At S5C, for a rise in the heat-setting
temperature from SOC to 200C, volume shrinkage is predicted to reduce from
5.4% to 3.9% for the homopolymer and similarly, from 6% to 4.2% for the
copolymer. At 100C the same increase in settIDg temperature is predicted to
reduce volume shrinkage from 18% to 4% and from 15% to 5% for the
homopolymer and copolymer respectively. At 110C, volume shrinkage can be
reduced by a factor of 5 by the same increase of heat-setting temperature, with
homopolymer shrinkage dropping from 31% to 4.1% and the copolymer
dropping from 25% to 5%. Both materials are predicted to behave very
similarly at temperatures up to 100C, but at 110C, the copolymer is predicted
to have 5% less shrinkage when heat-set at SooC. It is interesting to note that
all the models converge at about 4%, for both materials when heat-set at
200C.
233
For a draw of 5 (figure 5.9) the models are less clear and less encouraging
about heat-setting at higher temperatures. However, the overall shrinkage
levels are quite low and are lower than for a draw area of 10 (figure 5.8).
Increasing the heat-setting temperature from 80C to 200C is predicted to
increase volume shrinkage at 85C slightly from 1% to 2% for the
homopolymer and from 1.9% to 3.7% for the copolymer. At 100C, the same
rise in setting temperature increases predicted shrinkage from 0.4% to 4.8%,
and from 0.6% to 5.9% for the homopolymer and copolymer. At 110C,
increasing the heat-setting temperature from 80C to 200C increases shrinkage
from -2.8% to 9.8%, and from -1.8% to 9.9% for the homopolymer and
copolymer respectively.
The data from the conventional experimentation series has been used to
calculate volume shrinkage using the same theory used for the models (figure
5.10). For samples with a draw of 10 heat"setting can reduce volume
shrinkage from 34% to 9% at 110C. This dramatic reduction in shrinkage is
seen for all test temperatures at this draw ratio. For material with a draw of 5
the trend is less clear. The increase in shrinkage predicted (figure 5.11), is not
as pronounced, mainly due to the expansion in the conventional experiments
being considerably less than predicted by the model. The predicted increase in
shrinkage with heat-setting temperature is due entirely to the major axis
models.
This work clearly confirms the earlier fmdings that heat-setting at a high
temperature can substantially improve a PET bottle's thermal stability. From
figure 5.10 it can be seen that heat-setting can keep shrinkage in real bottle
(comprised of many draw ratios), below 5% at 85C and possibly up to 100C
and beyond, depending on the bottle design.
234
5.2.3 Shrinkage Aspect Ratio
The expansion predicted and seen in the major axis for material with a low
draw area was unexpected. "Creep" in the TMA could not be an explanation.
as this expansion was not observed in the minor axes that have lower linear
draws. To see if a systematic variation exists between the shrinkage in each
axis, shrinkage aspect ratios were plotted. The shrinkage aspect ratio was
defmed as:
A . _ shrinkage in major axis
spect RatIo - shrinkage in minor axis
The shrinkage models for each material at 110C were used to calculate aspect
ratios for a range of draw areas and heat-setting temperatures. The
homopolymer aspect ratio plot (figure 5.11), clearly indicates that for low
draws and heat-setting temperatures, the minor axis is predicted to shrink more
than the major. The actual draw ratio between the principal axes is 1.5: 1. This
ratio is only reflected by an equivalent shrinkage aspect ratio for the highest
draw (10) heat-set between 140C and 160C and for a draw of 5 heat-set at
200C. For heat-setting temperatures below 160C, the aspect ratio increases
with draw ratio, but with what appears to be a regular convergence. This
relationship is inverted by heat-setting at temperatures in excess of 185C. The
significance of a convergence point for all the models is not known.
The copolymer aspect ratio predictions are shown in figure 5.12. For heat-
setting temperatures up to 140C the homopolymer and copolymer aspect ratio
predictions are very similar. Increasing the heat-setting temperature causes the
predicted aspect ratio to increase at a far greater rate for the lower draws than
for the homopolymer. Unlike for the case of the homopolymer, there is no
distinct convergence point in the copolymer aspect ratio plot.
235
The shrinkage results for the conventional experimentation senes (PET
copolymer) at 110C are plotted in figure 5.13. For an area draw of 10, all the
aspect ratios are positive, indicating that the major axis shrinkage was greater
than the minor axis (no negative shrinkage in the minor axis was observed).
The aspect ratios for this draw are reduced from 2 to about 1.4 by heat-setting.
It is unclear if the outlying point at 140C is rogue or not. As the draw ratio in
all cases was 1.5: 1, this range of aspect ratios would be expected. The aspect
ratios for a draw area of 5 are very interesting. They increase linearly with
heat-setting up to a temperature of 200C where the trend becomes less clear.
This linearity is strong evidence for a systematic relationship between
shrinkage in the two orthogonal axes and heat-setting temperature, for low
draw material. The data for this draw area were regressed to give the
following:
Aspect Ratio = (0.00963 * Temperature) - 0.993
This regression has an R2 of 92%.
It is important to establish if the trends in the aspect ratio, above, are real or
occur as a result of an experimental artefact. At its most extreme, the aspect
ratio is characterised by the large expansions measured for samples with a draw
area of 5 and low heat-setting temperatures. The key to source of the aspect
ratio relationship is this expansion. There are two experimental possibilities
for these results, creep during the run and a manifestation of the Poisson effect.
Creep in the TMA instrument was proposed as a possible reason for the
expansion recorded for samples with a low draw area measured in the major
axis. This was easily discounted because samples with the same or lower
draws measured in the minor axis did not show any expansion. Section 5.2.1A
provides a fuller explanation.
236
The Poisson ratio is the ratio of elongational and transverse strain, and so,
perhaps shrinkage in one axis could cause an expansion in the other.
Previously published work concerning the use of TMA on oriented polymers
51
have considered the two axes independent i.e. Poisson ratio is not applicable to
this case. Further evidence is supplied by this work, as it is based on an
unbalanced draw. For the films with a high draw area the shrinkage in the
major axis is higher than in the minor. Therefore, any or greater expansion
would be expected when the major axis is perpendicular to the test direction
i.e. during shrinkage tests for the minor axis. However, in our experiments, no
expansion was recorded in the minor axis. The draw dependence of this effect
(expansion or aspect ratio) counts against the Poisson effect being a likely
explanation. Some of the mechanical properties (UTS and strain hardening
rate for the homopolymer) also show the effect of heat-setting to be draw area
dependent in the major axis. Recent un-published work in IPTME on balanced
biaxially oriented PVC also lends evidence to this. Oriented PVC completely
recovers its initial dimensions on heating. Full recovery has been measured for
both draw axes by TMA. If contraction in one axis caused an expansion in the
other, then this recovery would not be possible. From these considerations it
can be seen that no evidence has yet been found to account for the expansion
and aspect ratios found. Therefore, these surprising results must, at present,
stand.
237
5.2.4 Thermal Analysis
5.2.4.1 DSC Analysis
Figure 4.32a shows a very good correlation between heat-setting temperature
and the secondary endotherm peak. The source of this secondary endotherm
has been discussed at length, both in the literature and in earlier sections, but
will be assumed for this analysis to represent energy absorbed by the melting
of smaller imperfect crystals formed during the heat-setting process, i.e. the
melting and recrystallisation theory. The temperature of the peak appears to be
independent of draw area. This increases the usefulness of this relationship for
analysing heat-setting in PET e.g. a possible quality control method for heat-
setting temperature.
The crystallinity measurements made by DSC for the conventional series of
experiments (figure 4.32b) show a high degree of scatter. The trend of
increasing heat-setting temperature increasing the degree of crystallinity can be
loosely discerned. Surprisingly, the degree of crystallinity appears to have
little or no dependence on draw area. For comments on the scatter of the DSC
results the reader is referred to section 5.1.6.1.
5.2.4.2 Degree of Crystallinity from Refractive Index Measurements
The refractive index measurements were used to calculate density and therefore
crystallinity (figure 4.33). The results are less scattered and crystallinity values
are generally about 10% higher than those from DSC. The refractive index
results do support the DSC work in so far as the crystallinity level is shown to
be independent of the two draws plotted. The reduction in scatter of the
refractive index results, compared with the DSC results, contributes to the
conclusion from the FED DSC work, that DSC is not a reliable method to
obtain crystallinity in these circumstances. The FED model predicted a
decrease in crystallinity with increasing heat-setting temperature. Neither the
conventional DSC or refractive index measurements have repeated this.
238
5.2.5 Optical Properties
5.2.5.1 Total Transmitted Light Measurements
PET is such an attractive material for packaging, not least because of its high
level of optical transparency. Heat-setting is known to increase the degree of
crystallinity and this could give rise to a reduction in transparency. It is
important in CSD applications that, except for the neck and base, a bottle
should remain transparent after heat treatment. Transmission losses through a
film arise through two mechanisms, reflection and absorption/scattering. The
Hazeguard transmitted light measurement technique is an industry standard for
this type of measurement. To make the results more sensitive to any intrinsic
change in the material, the reflection from the films' two surfaces was
removed, using the Fresnel Correction (3.5.1.2A). Figure 4.34 shows that the
material with a draw area of 5 has a greater absorption than a material with a
draw of 10. These results are not corrected for thickness and so the material
with the lower draw (and therefore greater thickness), would be expected to
have a greater absorption. The plot does show that heat-setting does not reduce
the transmission of PET films, on the basis of the low level of optical
transmission losses.
Absorption does not vary linearly with thickness. In an attempt to compensate
for thickness, the 'Lambert' part of the Beer-Lambert correction was applied to
give figure 4.35. The Lambert law is used to compensate for sample thickness
in spectrometry and so was assumed to be applicable here, at least to a first
approximation.
239
For Absorption
As
Then
where A is the absorbence
I
In-= kd
10
I
A=-
10
k = (1nA)
. d
10 is the incident beam intensity
I is the emerging beam intensity
d is the sample thickness
k is a constant
From the corrected plot (figure 4.35), it can be seen that either the lower draw
material has an intrinsically higher absorption or the correction is unsuccessful.
PET does have a greater tendency for "clouding" at lower draws when heated
but comparison of two plots suggests that in fact the correction is not complete.
5.2.5.2 Micro-Photometry
The micro-photometry results, whilst scattered, do not show any significant
variation with the two draws plotted (figures 4.36 and 4.37). This strengthens
the argument that the correction applied to the light transmission experiments
was not sufficient to correct for thickness effects.
240
Both the total transmitted light and micro-photometty experiments indicate that
heat-setting under these conditions does not adversely affect the clarity of PET.
Undrawn material, when heat-set, has no strain induced crystallinity to act as
nucleation sites and so crystallise in a spherulitic fashion. Spherulites are
sufficiently large to act as efficient light scattering centres, leading to an
opaque material. This opacity can be observed in the neck and base regions of
a bottle that has been heat-set via Krupp Corpoplast's Corpotherm route, for
example.
5.2.5.3 Birefringence
For both draw ratios (figures 4.38a and b) the birefringences are as expected,
i.e. the two out-of-plane birefringences are larger than the in-plane
birefringence, and the out-of-plane birefringence in the major axis direction is
the largest of all three. Heat-setting increases all three principal birefringences,
roughly by the same proportion. Samples with a draw of 10 have higher out-
of-plane birefringences than those with 5, as expected. The in-plane
birefringence does not show the same increase with the increase in draw area;
it is defined as the difference between the refractive indices in the major and
minor axes. As the draw ratio is constant between the draw axes, this would be
expected. The increasing birefringences indicate an increase in orientation (or
crystallite orientation) with heat-setting. Disorientation of the amorphous
phase could be occurring as well, but if it is, then its effect is swamped by
further crystallisation and orientation. Ma and Han
8s
concluded that the
increase in birefringence with annealing was due to an improvement in degree
of crystallinity and structural perfection. The plot of birefringence against
draw area (figure 4.38c) indicates that the bulk of the birefringence increase
occurs during the initial draw. The equivalent plots for lower heat-setting
temperatures (not included) displayed very similar trends in this respect. The
stability of the in-plane birefringence commented on before is confIrmed in this
plot for all draw ratios above approximately 3.5. A draw ratio of about 2.5 is
generally accepted to be the upper limit of Gaussian deformation
2o
, which
corresponds with the rapid increase with birefringence seen here for draw
ratios up to this point.
241
5.2.5.4 Additional Information from Refractive Index Measurements
The majority of the birefringence values discussed are calculated from
refractive index measurements. Several authors
87
, 122 have found that
additional information can be obtained by considering the actual refractive
index measurements (figure 4.39). The fact that the test direction is more
significant than draw area is interesting. The two draw areas are each typically
composed by the two linear draws given below:
Draw Area 10 = (3.85*2.60)
Draw Area 5 = (2.75*1.80)
i.e. a draw ratio of approximately 1.5: 1 between draw axes
The minor axis draw (N3 direction) for an area draw of 10 is larger than that of
the major axis (N2 direction) draw for a draw of 5, but only marginally so. If
the refractive index was dependent on only linear draw then it would be
expected that the draw area 5 N2 index should be just above the draw area 10
N3 index and not just below the draw area 10 N2 index. In other words, whilst
the order follows the linear draw values, their grouping, which is very distinct,
does not. This confirms that the majority of the molecular orientation occurs
up to an area draw of 5 and further drawing does not substantially increase the
orientation. Also, this clearly demonstrates that the linear draw should not be
taken in isolation when discussing biaxially oriented systems. The decrease of
both NI indices with heat-setting temperature, along with draw area 5 N I index
being higher than the draw area 10 NI suggests that both drawing and heat-

setting reduce the out-of-plane molecular orientation. This is probably due to
the improvement of the in-plane orientation.
242
5.2.5.5 Selected Properties as a Function of Birefringence
The investigation of the effect of crystallinity on selected properties (sections
5.13 and 5.15) suggested that whilst in most cases the degree of crystallinity
was important, it was not the only determining factor on physical properties. In
an attempt to identify any other factors, similar properties have been plotted
against birefringence in figures 5.14. to 5.19. Properties in the major axis have
plotted against the out-of-plane birefringence in the major axis direction (&1
2
)
and against the out-of-plane birefringence in the minor axis direction (&1
3
) for
properties considered in the minor axis.
(A) Yield Stress as a Function of Birefringence .
The copolymer major axis yield stress model has been plotted against
birefringence data, for the conventional experimentation series, in figure 5.14.
This plot clearly indicates that a linear relationship exits between modelled
yield stress and birefringence, for data from both draw area samples. This
linear relationship between yield stress and birefringence is in complete
agreement with Jabarin
86
. His data allowed the conclusion that the mechanical
and transport properties of oriented PET are directly related to the degree of
orientation as measured by birefringence. This confrrms the good evidence
presented here that yield stress in the major axis is determined principally by
orientation of the PET semi-crystalline morphology.
243
(B) Creep Strain as a Function of Birefringence
Figure 5.15 shows. creep strain to be strongly but not linearly dependent on
birefringence. Figure 5.16 shows the naturallogarithrn of creep strain plotted
against birefringence. This is good evidence that creep strain is a logarithmic
function of birefringence. The data has been regressed to give a straight line
fit:
In (creep strain) = (-21.53 * birefringence) + 5.23
R2 = 92%
The draw area trend is as-expected i.e. lower draws showing larger creep strain
but increasing draw area rapidly reduces the creep strain (figure 5.15).
Because in both these plots the effect of draw area is continuous with
birefringence. it can be concluded that creep strain is almost entirely
determined by orientation. This is supported by figure 5.3 which shows creep
strain to be poorly correlated with degree of crystallinity with respect to draw
area.
(C) Degree of Crystallinity as a Function of Birefringence
The birefringence data have been plotted against the copolymer degree of
crystallinity model (by density) in figure 5.17. Only the major axis out-of-
plane birefringence has been plotted, as it has previously been established
(figure 4.38) that the minor axis out-of-plane birefringences follow the same
trend as the major axis, but have lower overall values. Figure 5.17 shows that
although crystallinity and birefringence are related, for an equivalent degree of
crystallinity samples with differing draw ratios will have different
birefringences. This difference is assumed to be the contribution made by
orientation (most probably of the amorphous phase). The data point
corresponding to a birefringence of 0.083 is considered rogue as it relates to an
unheat-set sample which the crystallinity model is known not to perform well.
244
(D) Shrinkage as a Function of Birefringence
Both the shrinkage at 11 Doe and birefringence results are plotted directly from
the conventional experimentation series. Figure 5.18 shows shrinkage against
birefringence in the major axis. Although shrinkage is strongly affected by
birefringence, especially for the higher draw area, the two draw areas do not
coincide totally, suggesting that another factor must also be involved. For a
draw area of 10 the trend shows an increase in birefringence dramatically
reducing shrinkage in the fIrst instance, becoming asymptotic at approximately
4% shrinkage when birefringence values of 0.17 are exceeded. For a draw area
of 5 the trend is reversed with shrinkage increasing from initially negative
values with increasing birefringence. These values, also, appear to be reaching
an asymptotic value very similar to value reached by samples with a draw area
of 10. Therefore, once again, birefringence is not the sole factor controlling
shrinkage in the major axis.
The minor axis plot (fIgure 5.19) exhibits both lower shrinkage and
birefringence values than its major axis counterpart, as expected. Both plotted
draw ratios show a trend of increasing birefringence reducing shrinkage but
this is nowhere near as strong as in the major axis. The data point indicating
9% shrinkage for a draw area of 10 with a birefringence value of 0.1, is
considered to have an erroneous birefringence point as reference to fIgure 4.31
indicates that the shrinkage value is correct.
Peszkin and Schultz
8
reported a decrease in shrinkage with increasing
birefringence on heat-setting and attributed this to increasing crystallinity.
They also found the shrinkage to go through a maximum when the
birefringence went through a minimum, which they interpreted as a
manifestation of the two competing processes of crystallisation and relaxation.
245
There is a considerable amount of work reporting a linear relationship between
birefringence and shrinkage stress
l8
, 23, lB. However De Vries
23
found that for
PVC two samples with identical amounts of recoverable strain, internal
entropic stress and therefore birefringence had different shrinkage stresses.
This was due to the samples having different distortional stress (due to local
distortions i.e. segmental rotation against rotation energy barriers) which can
give rise to distortional birefringence which may be either negative or positive
depending on the chemical structure of the polymer. They concluded that a
knowledge of the state of orientation (as characterised by birefringence) did not
allow a prediction of shrinkage stress or shrinkage rate.
These plots indicate that although birefringence is not the only factor
controlling shrinkage, it does have a very strong influence, greater than the
degree of crystallinity.
From this work it can be seen that due to its speed and strong correlation with
properties, birefringence would make a good quality control technique. It
could be possible to develop this further, with lasers as the light source and
computer data handling to be an on-line technique.
5.3 HEAT-SET BOTTLES
The bottle heat-setting time of 3 minutes was decided on because the work on
oriented films indicated that they had reached the maximum achievable heat-
setting for a given temperature, in about 15 seconds. It was decided to
replicate this maximum heat-setting condition by increasing the time to 3
minutes for the bottles to allow for the increased thickness. Despite the
increased heat-setting time, 15 seconds was still used to generate the model
predictions. The longest time in the FED analysis was 30 seconds, and so 3
minutes is too far outside the experimental window to be modelled.
246
The actual side-wall draw area of 9 for the bottle was used in the models. The
bottles have a far smaller surface area to volume ratio than the films and so
would take far longer to cool. The bottle cooling times would increase with
heat-setting temperature and so could affect the results. To prevent this, the
bottles were immediately quenched into water after heat-setting.
5.3.1 Shrinkage Measurements
The shrinkage results (figures 4.40 and 4.41) show that shrinkage can be
drastically reduced in bottles, in both axes, by heat-setting. Both the hoop
(major) and axial (minor) results are in close agreement with the model
predictions. This is of major importance as it implies that biaxially drawn PET
films can be used to model the behaviour of biaxially oriented PET bottles, at
least in the side walls. The TMA procedure, using the film and fibre method,
could not be used to measure shrinkage in the bottle neck or base regions as the
samples would be too thick and would not be flat. .. The area shrinkage could be
obtained using the TMA by measuring the increase of thickness in a sample
during heating.
The hoop axis results and models have negative shrinkage values but only at
the higher end of the heat-setting temperature. As with the films, this negative
shrinkage is not seen in the axial (minor) axis. The hoop axis displays far
lower shrinkage than the axial axis, at lower heat-setting temperatures. This is
quite surprising as the hoop axis has a higher draw ratio (3.7) than the axial
(2.5). In addition, the axial (minor) axis model predicts greater shrinkage at
lower draws than the major at these lower heat-setting temperatures. It can
only be assumed that the major axis has an element of expansion (not thermal
expansion), in keeping with the film results at these lower temperatures. The
aspect ratio relationship discussed in 5.2.3 was not found in this case. As the
bottle side walls have a high draw ratio (approx. 9) this is in agreement with
the high draw area aspect ratio data, also discussed in 5.2.3. The axial (minor)
axis displays significantly less scatter for the higher heat-setting temperatures
than the hoop axis which is unexpected.
247
Theoretical volwne shrinkage has been plotted for the heat-set bottles in figure
5.20. It shows that increasing the heat-setting temperature can significantly
reduce shrinkage to levels under the industry requirement of 5%. This appears
to be obtainable even at 110C, if the highest heat-setting temperatures are
used. It should be remembered that the volwne predictions are based only on
shrinkage measured in the side walls of the container. Figure 5.10 shows that
lower draw ratios give lower volwne shrinkage values. Therefore, based on the
bottle measurements a real bottle which will have a variety of lower draws
would exhibit even lower overall volwne shrinkage. The models do not
provide a reliable indication of shrinkage for draw ratios below 4.5. These
draw ratios occur in the neck and base regions of a bottle and so these areas
may require extra attention. Krupp's Corpotherm process for producing heat-
set PET bottles does concentrate on crystallising these areas.
5.3.2 Thermal Measurements
The use of a fluidised bed to heat-set the bottles ensured rapid heat-transfer,
due to the fluid nature of the transfer mediwn (sand). This is due to the
fluidised sand being in good contact with the bottle and also being continually
agitated by the compressed air. This would be a fair simulation of heat-setting
in a hot mould, because the bottle/mould surface contact would be good, the
mould would be large and heated and so would have a large thermal mass.
However the fluidised bed's temperature control was only approximate,
confirmed by thermocouple readings from the sand. The correlation of the p-
peak with heat-setting temperature found for conventional experimentation
films was used to measure the temperature of the heat-set given to the bottles.
The crystallinity levels obtained from the DSC scans do not correlate well with
heat-setting temperature. This could be due to DSC providing an unreliable
method of measuring crystallinity and/or heat-setting temperature. The
previous work on films (sections 5.1.6.1 and 5.2.4.1) has shown the problem to
be the quantification of the degree of crystallinity by DSC, but to verify this,
crystallinity for the bottles was also obtained by density measurements (see
over).
248
5.3.3 Density Measurements
Figure 4.43a shows that a very close correlation exists between PET
crystallinity (as measured by density) and heat-setting temperature. It is
assumed that, the predominant cause of variation in density is a change in
crystallinity level and not densification of the amorphous phase due to the high
draw in the bottle wall. This correlation discredits the crystallinity levels made
by DSC, but validates DSC as a method of measuring heat-setting temperature.
The data fit the model prediction fairly well at low heat-setting temperatures,
but the correlation is much improved as the temperature is increased. The
model underestimates the crystallinity at these lower temperatures. The choice
of 15 seconds for the model to simulate the bottles is fairly arbitrary and so can
be altered to improve the fit. Figure 4.43b has the time component of the
model changed to 25 seconds. The overall fit is much better, especially at the
lower heat-setting temperatures. There is some loss of fit at the highest-setting
temperatures. The improvement could reflect that the bottles were heat-set for
3 minutes and the films may not have achieved full heat-set at the lower
temperatures in 15 seconds. Neither version of the model can predict the
crystallinity for the unheat-set material, which was at the extremity of the
experimental window for the FED analysis.
249
5.3.4 Orientation Measurements
The Spruiell orientation functions for the 1018cm-
i
, 973cm-
i
and 1578cm-
i
absorption bands have been calculated for the bottles, and are plotted in figures
4.44 to 4.46. The 1018cm-
i
dipole is assigned to a vibration of the
disubstituted benzene ring and has its resultant dipole moment change parallel
to the C
i
-C
4
direction in the ring. Hence this dipole moment change is
approximately parallel to the chain axis. This dipole moment is irrespective of
the phase
l29
and so is an indication of overall orientation. The hoop axis has
the greater draw and so it would be expected that the orientation functions
would be biased towards the f2B axis, but this is not always the case. Also,
there is no discernible trend with respect to heat-setting temperature. If heat-
setting has the combined effect of increasing crystallinity and disorienting the
amorphous phase, then its effect on overall chain orientation would not be a
simple one.
In an attempt to distinguish the orientation in the two phases measurements
were made for the 973cm-
1
[u(C-O)] and 1578cm-
1
[u(AI)] bands. These
correspond to dipoles that are approximately parallel to the chain axis in the
crystalline and amorphous phases respectively83. The 973cm-
1
plot (figure
4.45) again shows no bias to the f2B axis. There is however, a trend for the
samples with a higher heat-setting temperature to have higher fiB and f2B
values, although not in a systematic fashion. The unheat-set sample does have
the lowest fiB and f2B values. The 1578cm-
1
plot (figure 4.46) again shows no
bias towards the f2B axis, or any systematic variation with heat-setting
temperature.
250
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - ~ --
The amorphous orientation is the most likely to coincide with the draw
directions and so this is unexpected. The overall values are considerably lower
than in the other two plots. However, the unheat-set sample has the lowest
amorphous orientation functions, implying that the low amorphous orientation
functions are not a result of disorientation of the amorphous phase due to heat-
setting. The points are close together which suggest that heat-setting has little
effect on amorphous orientation. There is no evidence to suggest that
substantial relaxation of the amorphous phase orientation occurs, during heat-
setting.
It was hoped that by plotting the data as dichroic ratios (after Gupta et aJ83),
would highlight any trends in the FTIR results (figures 4.47 to 4.49). All three
plots are scattered but do show an increasing dichroic ratio with heat-setting
temperature. It is interesting to note that no evidence has been supplied by the
FTIR analyses, to suggest that heat-setting has the disorientation as part of its
mechanism. The negative values seen on the 1018cm
1
dichroic ratio indicate
that the orientation is greater in the axial axis than in the hoop axis.
The FTIR results are far from conclusive. The 10 18cm-
1
band is a strong
absorber, thereby requiring thin sections to ensure the absorbence was within
the linear region of the detector. Taking sections of oriented PET is
notoriously difficult to carry out and can easily induce further orientation.
Every care was taken to avoid this during the section preparation, but with such
thin sections 5 ~ ) , the induced orientation must be significant. If the
investigation was restricted to the 973cm
1
and 1578cm
1
bands and the FTIR in
IPTME, with its extended linear range used, then the full film thickness would
be permissible. This would remove the greatest source of experimental error
associated with this technique.
Bottle No II was allowed to cool in air, as opposed to the others, which were
shock cooled. There was no significant difference between these, for any of
the properties measured, suggesting that the air cooling is still sufficiently fast
to be considered a quench.
251
5.3.5 Shrinkage of Heat-Set Bottles as a Function of Heat-Setting
Temperature and Crystallinity
The comparisons between yield stress, creep strain and shrinkage against
crystallinity have been performed by comparing the FED models in each case.
The heat-set bottles allow a direct comparison using 'real' shrinkage and
crystallinity (from density measurement) data. Figure 5.21 is of shrinkage at
110C plotted against heat-setting temperature and it can be easily seen that for
both axes, these are strongly correlated. Therefore in this case of constant
draw area, the heat-setting temperature is the sole factor for determining
shrinkage at 110C. Linear regression of the data for both axes gives:
Hoop Axis: Shrinkage (%) = (Temp*-0.056) + 13.1
R2 = 80%
Axial Axis: Shrinkage (%) = (Temp*-O.llO) + 24.1
R2 = 90%
Degree of crystallinity is plotted against heat-setting temperature in figure 5.22.
This figure shows a strong correlation between heat-setting temperature and
degree of crystallinity. Heat-setting at 220C raises the degree of crystallinity
from the as-drawn level of 23% to nearly 50%. Figure 5.23 shows that this
increase in crystallinity is accompanied by a large decrease in shrinkage at
1l0oC. This is linear for the hoop axis and appears asymptotic for the axial
axis.
From these observations, coupled with those made in sections 5.13, 5.15 and
figure 5.24 (which shows that crystallinity is not a strong function of draw ratio
for the draws investigated), it can be seen that crystallinity will account for
differences in shrinkage caused by heat-setting but not those due to a difference
in draw ratio.
252
5.4 MOLECULAR INTERPRETATIONS OF HEAT-SETTING
BASED ON EXPERIMENTAL OBSERVATIONS
Heat-setting has the possibility of producing a number of morphological
changes in biaxially oriented PET. The possibilities are:
1) increase in the degree of crystallinity,
2) alter the type of crystallinity,
3) increase orientation,
4) tension tie molecules.
To present a single molecular model to account for all property changes
associated with heat-setting would appear to be an incorrect approach, if as
stated in the literature, different microstructural features control different
properties
37
. Therefore, interpretations dealing with the effect of heat-setting
on mechanical properties and shrinkage, will be attempted separately.
5.4.1 Molecular Interpretation of Mechanical Properties
The mechanical properties were only investigated using the FED and so some
of the trends are unclear. However, the creep results have good correlations
with both the data here and with literature
63
, and so it is intended that this
should provide the basis for an interpretation. It is accepted that the unusual
time dependence of the models suggests different morphological mechanisms
to the rest of the mechanical properties, but these models do not provide clear
enough trends to interpret.
253
The creep models (figures 4.12 and 4.13) show that creep strain is strongly
influenced by draw ratio, and is dramatically reduced by heat-setting for
samples with lower draw ratios. It was subsequently shown that birefringence
is predicted to vary with the natural logarithm of creep strain. Both materials
behave in a similar fashion, but with the copolymer exhibits higher creep
strains. These observations allow the four morphological alteration
possibilities to be discussed.
Heat-setting is known to increase the degree of crystallinity of PET and the
crystallinity models reflect this. Crystallinity levels are not predicted to alter
significantly with draw ratio, in the limits investigated (figure 4.23a to c). This
is not unexpected, since Hennessey et al
54
reported that the development of
strain induced crystallinity occurred only for strains between 150% and 300%
(draws of between 2.5 and 4). This is below the smallest draw studied. This
means that an increase in crystallinity cannot be the reason for the two draw
ratios exhibiting different creep strains.
Changes in the type of crystallinity are difficult to monitor. This becomes
impossible if the only reliable method of crystallinity measurement is density,
which requires a known difference in the phases. Without a detailed X-ray
diffraction study this possibility cannot be realistically commented on.
The best indication of orientation in this investigation are the birefringence
data. Figures 4.38a to c show that both the out-of-plane birefringences
increase with increasing draw area and heat-setting temperature. Therefore, an
increase in orientation in one or both phases accompanies the reduction of
creep strain through heat-setting. The additional information from examining
the refractive index data (section 5.2.5.4) confirmed that the majority of the
orientation is developed before a draw area of 5 is reached. As creep strain is
so strongly related to birefringence (figure 5.15 and 16), a relatively small
increase in orientation from increasing the draw from 5 to 10 will significantly
reduce the creep strain.
254
The tensioning of taut tie molecules has been proposed as a mechanism by
which non-crystalline material may become more oriented by heat-setting
25
.
This mechanism is combined with the mechanism of tie molecule formation,
and further developed below to account for the draw dependence on creep
strain. Perfect uniaxial orientation of the crystalline phase is depicted for
clarity.
When PET is drawn the strain induced crystallites are fractured with the
chains common to both crystal fragments forming the tie molecules.
DRAW AREAS DRAW AREA 10
/
TAUT TIE MOLECULES
The tie molecules are modelled to be the primary load bearing structures and,
so a morphology with taut tie molecules (draw area JO) will show less creep
strain.
255
Buckley et al proposed that heat-setting can tension the non taut tie molecules
in the following way:
Ir r
A
D
c
I
B
For a draw area of 5 both the non taut tie chain (e) and the "free" amorphous
chain (D) are incorporated into the crystal (B). For a draw area of 10 only the
amorphous chain (D) isfree to be incorporated into the crystal.
By this mechanism, a similar increase in crystallinity will significantly alter the
condition of the reinforcing network for low draw area samples. In this way
the reinforcing network of this material becomes more similar to that of a
material with a higher draw area. This would mean that the two draw areas
will behave more similarly under creep stress conditions. The increase in
crystallinity and increase in amorphous orientation with increasing heat-setting
temperature, could both account for the observed increase in birefringence. It
would be easy to envisage how the modifications made to the microstructure,
by the mechanism out lined, would increase yield stress. Yield stress has been
shown to be determined entirely by birefringence, which again supports the
idea that heat-setting does not substantially rely on relaxing the amorphous
phase.
256
5.4.2 Molecular Interpretation of Shrinkage Properties
The trends observed in the shrinkage experiments may be summarised as:
1) both materials behave similarly,
2) as-drawn material shrinkage increases with draw area,
3) heat-setting reduces shrinkage and diminishes the difference between
draw areas,
4) for the major axis with low draw ratios, unheat-set exhibits negative
shrinkage.
The model developed above for creep strain has a very interesting property, in
that heat-setting is modelled to reduce the morphological difference between
the draw areas. The shrinkage measurements also show heat-setting to remove
the effect of draw ratio on shrinkage. This would indicate that the
interpretative model for creep strain would also fit some of the aspects of the
shrinkage data. Therefore, this model must then be examined to see if it will
account for the other observed trends.
The mechanism by which heat-setting reduces shrinkage is often considered to
be by the establishment and improvement of a reinforcing network. This is a
very convenient model as it accounts for the reduction in shrinkage combined
with the reduction in creep strain. However, such a network comprising tie
molecules will have only tensile qualities and no 'compressive' strength. Once
the material is above Tg all the crystallites are free to move closer together,
under the influence of the amorphous orientation. Hence on closer
examination, it is difficult to see how such a network could reduce shrinkage.
This is exacerbated if the tie molecules are considered to be tensioned by heat-
setting. From the previous arguments, the taut tie molecules will be considered
to contribute to shrinkage.
257
The as-drawn material with a higher draw ratio shrinks more, due to the action
of taut tie molecules, as opposed to the non-taut tie molecules in the material
with the lower draw ratio. Prevorsek et aJ37 concluded that the highly extended
(taut) interfibrillar tie molecules are the main factor in fibre shrinkage
properties, and their volume fraction is so significant that they should be
considered as a separate phase. The relaxation resulting from the contraction
of the tie molecules leads to inclusion of substantial sections of the initially
non crystalline ties into the crystalline lattice, thus increasing the crystallinity.
During heat-setting (constrained annealing) the relaxation of the taut tie
molecules may occur by partial pulling through the crystal lattice and inclusion
of properly located sections into the crystal lattice of the folded chain blocks.
A difficulty arises because the heat-setting was viewed in the creep model to
tension non-taut tie molecules and increase amorphous orientation. Both these
could be mechanisms by which shrinkage could be increased by heat-setting
5
!.
In the case of shrinkage heat-setting was assumed to relax the taut tie
molecules, thereby reducing the tendency to shrink. Therefore, it is unlikely
that the taut tie theory can explain how heat-setting can improve creep
properties and reduce shrinkage. The key to this could be the sensitivity of
creep strain to orientation. It has been established here that the reduction in
shrinkage due to heat-setting is directly related to the increase of crystallinity.
The increase in density of crystallites will produce a more discontinuous
amorphous phase, possibly disrupting the amorphous phase's effect on
shrinkage. If heat-setting improves both orientation and crystallinity, then
creep strain will be reduced by both these factors. The increase in crystallinity
is in part due to the incorporation of the tie molecules into the crystallites and,
so they may no longer contribute to shrinkage.
258
5.5 EXPERIMENTAL ERROR CONSIDERATIONS
Experimental error has two basic sources in this investigation; from film
drawing and heat-setting processes, and from the techniques used for property
measurement.
5.5.1 Errors Arising From Film Drawing and Heat-Setting Processes
The draw ratio was quantified by marking a lcm grid on the sample and
measuring the resulting grid separation after drawing. The change in grid
separation for the centre four squares was measured to the nearest half
millimetre. The drawing increased the line thickness and so the separation was
taken between the centres of each line. The achievement of specified draw and
its reproducibility may be quantified by the repeat runs in the FED programme.
For a target draw of 7.S, the samples' average draw was 7.77. The
reproducibility is indicated by the standard deviation, which for these samples
was 0.29. This is not of concern in the FED work as the actual draw for each
sample was used in the analysis, but is relevant in the conventional
experimentation series, where two nominal draws were considered.
The draw temperature was controlled by the Long Stretcher to SoC. As the
work of deformation is nearly all converted to heat4
2
, 43, the actual material
deformation temperature would be higher than this. The strain rate was within
10%, as calibrated by rcI.
Heat-setting was performed usmg a metal block previously heated in a
calibrated oven. The cooling curves (figure 3.S) demonstrate that in the worst
case (heat-setting for 30 seconds at 200C), the block will be about SaC-SaC
(depending on position) cooler than intended, at the end of the treatment. As
the block is initially at the desired temperature and heat-setting time appears
less significant, this small error is even less significant for this material. This is
confmned by the correlation between the ~ p e a k temperatures in the DSC
scans for the conventional experimentation series (figure 4.32a).
2S9
5.5.2 Errors Arising From Film Property Measurements
The tensile properties were generally measured using ASTM D882
99
, apart
from the exceptions outlined in 3.4.1.1. The load cells were selected to ensure
that the bottom 5% of their range was not used, in accordance with the
manufacturer's instructions. Using yield stress as an example, the combined
process of drawing, heat-setting and measurement gave an average yield stress
in the major axis of 91 MPa with a standard deviation of 5.18 for the
homopolymer, and 85.4 MPa with a standard deviation of 2.92 for the
copolymer (appendix A9). The decision not to use a specific gauge length for
extension measurement has not detracted from the accuracy of the results. The
yield strain in the major axis for the homopolymer is 9.1% with a standard
deviation of 1.26, and 8.4% and a standard deviation of 0.75 for the copolymer.
It is possible that the absence of a gauge length may have introduced a constant
error from using the whole test length as the g a u g ~ but this will have reduced
the experimental scatter. The same considerations apply to the creep
measurements, except they were performed at an elevated temperature. The
environment chamber used for this controlled the temperature to 2C.
The shrinkage measurements using the TMA have two components; l.e.
changes in length and temperature. The initial gauge length of IOmm
0.25mm is achieved using a jig. The change in length is measured by a linear
voltage displacement transducer, which is calibrated and is accurate to below 1
micron and so will not give rise to a significant error. The temperature
measurement again is subject to regular calibration, and even at the relatively
high heating rate of 20
0
C/min used in this investigation, is considerably better
than 5C. Using the FED repeat runs in appendices A20 and A21 for the
homopolymer and copolymer respectively, the mean and standard deviation for
shrinkage at 110C was 3.11% and standard deviation of 0.16 for the
homopolymer and 2.85% and 0.19 (respectively) for the copolymer.
260
The accuracy of the DSC measurements was ensured by calibration against an
Indium standard. There are insufficient repeat runs in the FED DSC work to
produce meaningful statistics, but the repeat homopolymer data show a ~ p e a k
temperature range of 1C and a variation in crystallinity of3% (appendix A24).
The equivalent copolymer results have a temperature range of 6C and a
variation in crystallinity of 6%. The difference in accuracy between the two
grades would indicate that the DSC is not the source of the bulk of the error.
Both grades have a crystallinity range of 17% and so the copolymer variation
of 6% is very significant. This probably accounts for the poor fit (R2 =31%) of
the copolymer model for crystallinity by DSC ,relative to the homopolymer (R2
=74%).
The quoted accuracy for the density measurements by ICI is better than 1%.
The density results for the FED repeat runs give an average crystallinity for
these samples (Appendix A24) of 34.2% for the homopolymer and 26.9% for
the copolymer. The associated standard deviation (using low sample number
version) for these results was 0.95 and 1.44. It should be remembered this does
not just reflect the experimental accuracy of the density work, but also the film
drawing and heat-setting process.
The experimental error associated with the optical investigations performed on
the conventional experimentation is hard to quantify, due to the lack of repeat
runs. The Hazeguard used to make the total transmitted light measurements
was allowed to settle, zeroed, then calibrated against ground glass standards of
known transmission, prior to use. The repeatability of the measurement for the
standards was better than 1%. The photometer used for the micro-photometry
was calibrated for 100% and 0% transmission in-situ, in the optical train of the
microscope. The birefringence, and refractive index measurements were
performed on regularly calibrated equipment and in accordance with the
manufacturer's recommendations.
261
The bottle heat-setting trials used the DSC scans to measure the effective heat-
setting temperature. The correlation between the j3-peak and heat-setting
temperature had been established with the conventional experimentation series
and so could be used with some confidence. All the considerations for each
characterisation technique discussed for the films are equally applicable for the
bottle investigations, where the technique was also used.
The polarised FTIR results were the ouly set of results that were predominantly
unsuccessful. The necessity of taking thin, strain free sections was, in all
probability, the source of the error, despite all the precautions taken. The
sectioning of highly oriented PET is very difficult, due to the material either
splitting and extra, spurious orientation is thought to be induced during the
sectioning process.
262
115
110
105
1
100
lE
95

III
III
...
90

III
85
0
80
75
70
65
-
e-
l-

OA=5
OA=IO
-
15 20 25 30 35 40 45 50 55
DEGREE or CRYSTALlNTY (70)
Figure 5.1. Homopolymer Yield Stress Model In The Major Axis As A
Function Of Degree Of Crystallinity
115
110
105
I'OO

95
UI
UI
...
90
Cl<
..
UI
85
0
-'
!o/
80
>-
75
70
65
.-/
---
V
---
V
-
f--
-


15 20 25 30 35 40 45 50 55
OEGAD: or CIIYSTAUNTY (70)
Figure 5.2. Copolymer Yield Stress Model In The Major Axis As A Function
of Degree of Crystallinity
263



iI:
"
a:
....
11>
...
...
...
a:
0
50
40
,.----
f--
(
'--
---
r--

OA=5
OA=10
30
20
10
0
-10
42 43 44 45 46 47 48 49 50 51 52
OEGREE or CRYSTALlNTY (%)
Figure S.3. Copolymer Creep Strain Model As A Function Of I>egRe Of
Crystallinity
264
20
18
16
,.
r:::J
OA=5
OA=10
12
~
!!. 10
~
a

l
6
~
4
2
:/
/
v
0
/'
1/
-2
;7
-4
-6
10 15 20 25 30 35 40 45 50 55
OEGR[ Of CRYST AUNTY ~ )
Figure 5.4. Homopolymer Shrinkage Model At IIOC'C In The Major Axis As A
Function OfDcgRe OfCry"linity
20
18
16
,.
12
~ 10
~
~ 8
i :
2
o
-2
-4
-6
/'
1\
~
1/
1/
-- P
--::::::::::
=
~
~
10 15 20 25 30 35 40 45 50 55
OEGR[ Of CRYST AI.lNTY ~ )
Figure H. Copolymer Shrinkage Model At IIOC'C In The Major Axis As A
Function Of Degree Of Cry"linity
265
20
18
16
U l:J
010=5
010=10
12

!!. 10
"-
"

8
..

6

4
2
'"
'-.,
"-
I--- ..><
0
-2
-4
-6
10 15 20 25 30 3S 40 45 50 55
1lEGR[ or CIIYST AUNTY (:r.)
Figure 5.6. Homopolymer Shrinkage Model At IIO"C In The MinoFAxis As A
Fuaction Of Degree Of Cryillallinity

20
18
16
14
12
!!. to
8
i :
2
o
-2
-4
-6
l:J
010=5
010=10
."---
-
10 15 20 25 30 35 40 4S 50 55
1lEGm:[ or CIIYST AUIN1TY (:r.)
Figure S.7. Copolymer Shrinkage Model At IIO"C In The MinoFAxis As A
Function Of Degree OfCryaallmity
266
3!>
-
30
8!>(oC) H
100(oC) H
2!>
~

110(oC) H
N
i'--
2
l!l ~
8!>(oC) C
...
~
i 15
~
100(oC) C
-
110(oC) C
10
:::

!>
0
80 100 120 140 160 180 200
lOT STTtIG TDlPRATlIIE (oC)
Figwe S.8. Predictecl1'heorctK:al Volume Sbrinbge At IIO"C For A I.S.
BouIc With A Draw Area Of 10
10
/
-
8
8S(OC) H
V
/
100(oC) H
6
~
V
110(oC) H
N
/'

4
~
l!l
~
-
8S(oC) C
...
/
i 2
100(oC) C
./"
-
...
110(oC) C
0
~
,/
-2
-4
80 100 120 140 160 180 200
lAT STTlNG TDlPRATlIIE (oC)
Figwe '.9. PredictecI TbcoreticaI Volume Shrinkage At 11O"C For A I.,.
BouIc With A Draw Area Of'
267
35
1
___
---......
...
30
OA=10 850C
"'"
+
""
OA=10 1000C
-
--
25
lIE
'"'
1\
OA=10 1100e


.......
...
20

OA=5 850C

:.:
+
115


OA=5 1000C
VI
r--..
.

"-
OA=5 1100C
10
-.....
'-- ---.......
r-..
5
...


+
0
20 40 60 80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATURE (oC)
Figure S.IO. Theoretical Volume Sbriokage At IIOOC For A I.SI Copolymer
8oUIe. From Measurements Made On Conventional
Experimentation Series.
268
2
-
OA=5
1.5
<
Q
-............
0.=6
... r--
'"
"/
It
I
'\
OA=1
...
<>
~
DA=8
'"
0.5
~
/
OA=9
'"
-
I
0
-
OA=IO
/
-0.5
~
-I
80 100 120 140 160 180 200
IAT SETTING TEIl'ERATIJRE (oC)
Figure 5.11. Homopolymer Predicted Shrinkage Aspect Ratio At 11OC'C
2
/
-
OA=5
1.5
I1
2
OA=6
...
---
~
-
'"
It
I
-I
OA=1
...
~
<>
le 0.=8
III
0.5
'"
/

01.=9
-
i
0
/
01.=10
-0.5
./'
--
-1
80 100 120 140 160 180 200
IA T SETTING TOI'(RATtR: (oC)
Figure 5.12. Predieted Copolymer Sbriokage Aspect Raaio At IIOC'C
269
2.5
A
OA=10
2
I
A
Q
OA=5
~
1.5
...
...
Cl<
A j
~
..)
u
1 ...
V
~ CL
Cl!
~
,

...
0.5
P
(!)

/"
I
../'
:I:
0
V Cl!
/
-0.5
...-/

A
-1
20 40 60 80 100 120 140 160 180 200 220
HEAT SETTING TEMPERATURE (cC)
Figure S.13. Aetual Shrinkage Aspect Ratio Of Conventional Experimentation
Series At 11O"C
270
130

J J ~
DA=5
120
~


. /
DA=10
'0
110
V


III 100
~
III
~
....
90 III
V
0
./
...J
III
>-
80
V
70
. /
60
0.06 0.08 0. 1 0.12 0. 14 0. 16 0.18 0.2
BIlEfRINGENC[
Figure 5.14 Copolymer Major Axis Yield Stress Model As A Function Of
Bircfringcnc:e
271
100
90
....
80
~ 70
I<
~
~
60
..
'"
50 >-
VI
"-
40 w
Is!
u
30
20
10
\
\
\
~
~
~
~
t--- _A
...
...
o
o 0. 020.040.060.08 0. 1 0.120. 140. 160.18
8IlEFRlNGENCE
Figw-e 5. 15. Copolymer Creep Strain Model As A Function Of Birefringence
4.5 ...,
~
...
OA=3
4
...
~
CA=5
i"
p .5
~
CA=7
0: ...
>-
CA=11 VI
Q. 3
~
...
Is!
u
"-
S 2.5
""
'"
2
... ",
A
...
""
1.5
0.02 0.04 0.06 0.08 0.1 0. 12 0.14 0. 16 0. 18
B1RFRtIGNC
Figw-e 5. 16. Natural Logarithm Of The Creep Strain Model, As A Function
Of Birefringence
272
0. 18
0.17
...

0. 16
/
0. 15
7
...
If
0.14

...
..

0.13
e: - /
....
i 0.12
/'

0.11
"
0.1
0.09
..
0.08
0 10 20 30 40 50 60
eRYST AUINITY (%)
Figw-e S.l? Copolymer 8ireftingence As A Function OfDegrec Of

273
..
DA=5
..
DA=IO
20
~
15
~
~
~
~
~
1/
o
-5
0.06 0.08 0.1 0.12 0. 14 0. 16 0. 18 0.2
81REFRINGENC(
Figure 5.18. Shrinkage In The Major Axis At IIO"C As A Func:tioIl Of
Birefrinsence. From The CoavCDtiooal Experimentation Series.
9
&
a
OA=5
..
7
0A=10
~ 6
..

~
I_
3
"""
'"
--
....
..
2 A
1
0.06 0.07 0.08 0.09 0. 1 0.11 0.12
IRTRING[NC[
Figure 5.19. Shrinkage In The Minor Axis At IIO"C As A Function Of
BirefiiDgeacc. From The Conventioaal Expcrimcntation Series.
274
40 I.

35
SHRINK 650C

30
SHR1I'f( 850C
~
~
A
~ 5
SHRINK 1000C
~
A

" 20
~
A
SHRINK 11 OoC
~ 15
A
~
A
d 10
A
.r
I ~

~ > .. ..
5
--. -.
-
A
A
A
~
i
:i:
~
0
-f
-5
20 40 60 80 100 120 140 160 180 200 220
HEAT SETTtIG TEt.f>ERA TIAlE oC
Figure S.20. Theoretical Volume Shrinkage For A UI BoaIe Based On
Linear SbriDkage Measurements Made On The Side Walls Of
Heat-5et Botdes
27S
25
...
...
HOOP AXIS
20
...
AXIAl AXIS
,.... 15
~
~
...
A--- ..
...
~ 1
i---
!o<
---
...
~
---
1---
!
,"---
t
VI
5
...
t--
~
-...........
~ ...
0
-5
20 40 60 80 100 120 140 160 180 200 220
HEAT -SETTING TEMPERATURE (oC)
Figure S.21. SbriDbge At IIO"C AsA Function OfHeat-8eUing TempaBhR
For Heat-8et Boa1es
276
SO


10
5
o
I .....
&
:--
----
-
./
./
&


v
20 60 80 100 120 160 180 200 220
lOT STTING IDf'ERATUl (oC)
Figure 5.22. Degree ofCrysIaIlinity Plotted Against Heat-Sening Temperature
For Heat-Set Bottles
25
& \
&
HOOP AXIS
20
&
f\
AXIAL AXIS
15



10
&
t
I


5
t
--...-



-
0
-'lI.
-5
20 25 30 3S 50
CRYSTAllHTY (70)
Figure '.23. Shrinkage Ploned Against Degree ofCrysIaIlinity For Heat-Set
Bottles
50
1000C H
045


-
2000C H
-
>- 040
i
1000C C
...J
:ci 35
....
2000C C
C/l
>-
'"
(J 30
----
-----
...
-------'-..
0
V ""'--
25
//
0
:5 20
/'
15
04 5 6 7 8 9 10 11
DRAW AREA
Figure 5.24. Homopolymer And Copolymer Degree OfCrystaIlinity Models
For Hcat-8ettiDs Temperatures Of lOOOC And 2000C As A
Function Of Draw Area
278
6 CONCLUSIONS AND SUGGESTED FURTHER
WORK
6.1 CONCLUSIONS
The objective and strategy of the investigation was to discover if PET bottle
shrinkage behaviour could be approximated by PET films. It had previously
been established that heat-setting can reduce shrinkage in oriented PET
systems. However, the quantification of this process, under simulated
industrial and not laboratory conditions had not been considered in much
detail. It was intended for the project to produce statistical models for the
behaviour of heat-set PET on biaxially oriented films, so that a bottle
manufacturer could predict the result of heat-setting an ICI commercial grade
of PET. It is against the project objectives that the success of an industrially
motivated project, such as this, must ultimately be judged .
. The primary project objective was stated as:
"To reduce thermally induced shrinkage in PET food containers to allow higher
filling or service temperatures". The temperature ranges of interest may be
summarised as:
80C-90C
90C-100C
lOooC-ll OC
pasteurised filling
hot filling of jam and bottle washing
sterile filling conditions
The project strategy of using biaxially oriented films to simulate the heat-
setting of PET bottles was new and untried. Therefore, this technique had to
be established as a viable method of attaining the objective. It was also
planned that, at least the initial stages of the investigation should be conducted
using a Factorial Experimental Design package, available for use at IC! Wilton.
This was another technique that had to be familiarised before experimental
progress could be made.
279
'---------------------------------- - ----
6.1.1 New Techniques and Equipment Developed for this Project
Several techniques and new equipment were developed for the successful
completion of this investigation. These are summarised below:
a) The whole project approach of simulating PET bottles by using PET films
has never before been attempted, to our knowledge.
b) The isothennal heat-setting rig used was specifically designed and built
for this project. This was a critical link in the experimental chain as
almost all of the samples would be heat-set and so its design would affect
the feasibility and accuracy of the whole investigation.
c) The use of polyurethane foam to allow constrained heat-setting of PET
bottles has, to the investigators knowledge, never been used before. This
allowed the assessment of the validity of models generated for the films to
be tested for bottles, without the need of expensive heat-setting trials on
commercial processing equipment.
d) This project introduced to IPTME the use of TMA, quantitative
conoscopy, and P-FTIR to characterise biaxially oriented PET.
e) The use of computer controlled tensile machines to provide constant
stress conditions for creep experiments is sufficiently rare that the
supplied control software was under-developed when initially supplied.
From this it can be seen that this project required a substantial amount of
practical development to make it feasible.
280
6.1.2 The Use of Biaxially Oriented Films to Simulate Biaxially Oriented
Bottles
The ftrst area for consideration is the ability to model bottle behaviour by the
use of ftIms. Bottles are produced by injection stretch-blow-moulding and
ftlms by tentor drawing of extruded cast sheets. The injection phase in the
bottle production can lead to appreciable levels of orientation in the prefonn
that will be distributed though the wall thickness. The casting phase of sheet
production generally produces less orientation, and as the sheet is thinner, a
lower orientation distribution through the sheet thickness is likely to be
produced. The biaxial orientation in the bottles was induced by stretch blow
moulding at 87C, and the actual draw ratio between the hoop and axial axes
during defonnation is unspecifted. The draws were equal for both axes during
the ftlm stretching, until the minor axis draw had been achieved, where the
defonnation stopped and the draw in the major axis continued. The drawing
was perfonned at 1 10C.
From these considerations it can be concluded that bottle and ftlm materials
have fairly different processing histories, and so a similarity in behaviour
between the two biaxially oriented PET systems could not be assumed. The
principal shrinkage fmdings are sununarised below:
a) Figures 4.41 and 4.42 show that models derived from ftlm shrinkage
measurements, predict the heat-setting behaviour of bottles well. The
accuracy of the prediction improves with increasing heat-setting
temperature.
b) Shrinkage measurements were only possible in the side wall of the bottle,
so the correlation between model and bottle for lower draws was not
possible.
281
6.1.3 The Reduction of Thermally Induced Shrinkage in PET Bottles
a) The key objective of reducing thermally induced shrinkage has also been
achieved using heat-setting. As a general trend, increasing heat-setting
temperature reduces the predicted shrinkage.
b) The minor axis bottle data show that for heat-setting temperatures above
160C, the shrinkage is virtually independent of service temperature up to
the maximum test temperature of 110C. This indicates that in the minor
axis, a maximum shrinkage stability exists and has been attained under
these conditions. Therefore heat-setting cannot be used to reduce
thermally induced shrinkage below about 3% in the axial axis.
c) The bottle hoop axis results show that no such limit exists for the major
axIS.
d) This reduction in shrinkage translates to a calculated minimum volume
shrinkage of below 5% which is the maximum tolerable by the industry.
This level of volume shrinkage appears to be sustainable up to 110C if
the heat-setting temperature is above 200C.
Therefore the primary objective of the project has been achieved, i.e. the
identification of process conditions to control shrinkage in PET bottles to an
acceptable level. Apart from the these two key objectives the investigation has
yielded a considerable amount of information concerning the heat-setting
behaviour ofbiaxially oriented PET.
282
6.1.4 Mechanical Testing
6.1.4.1 Tensile Tests
It was vital to ascertain if heat-setting had any deleterious effects on the
strength of the polymer. The critical tensile parameter for controlling the point
when an artefact (as opposed to an experimental sample) is deemed to have
failed is the yield stress. Ultimate tensile stress and the strain hardening rates
are of less commercial importance but convey significant information about the
material.
a) The yield stress for both polymers investigated was not adversely affected
by heat-setting and under some conditions was moderately enhanced.
This contributes to heat-setting being commercially viable, as it does not
require the bottle to be thicker to reduce the service stress. On, increasing
the heat-setting temperature from 80
0
e to 200
o
e, the yield stress of the
homopolymer in the major axis is predicted to increase from III MPa to
123 MPa for a draw area of 10.
b) The copolymer was predicted by the FED analysis to have a slightly
lower yield stress than the homopolymer overall, but to respond to heat-
setting in a qualitatively similar fashion.
c) The copolymer yield stress model in the major axis has been shown to be
a linear function of birefringence.
d) The correlation between yield stress and degree of crystallinity models
predicts that crystallinity has only a secondary affect on yield stress.
e) Ultimate tensile stress and the strain hardening rates for the homopolymer
(in the major axis) only display a dependence on draw area. For a draw of
5 the effect of increasing the heat-setting temperature is the reverse of that
for a draw area of 10.
283
f) This effect is less pronounced for the copolymer ultimate tensile stress
with the draw only affecting the degree of increase with heat-setting
temperature.
g) The strain hardening rate for the copolymer shows no such dependence on
draw area but unlike the homopolymer, the ultimate tensile strain does.
h) None of the considered variables demonstrated this type of draw
dependence when measured in the minor axis.
The mechanical property results outlined here indicate that there is a transition
in behaviour when the draw area is increased from 5 to 10 for many tensile
parameters, and that the behaviour of any linear draw is dependent on which
axis it is situated.
6.1.4.2 Creep Tests
The high stress, low time creep properties at 65C are important for
applications where the a bottle is filled pressurised and then passed through a
pasteurisation tunnel. The FED work on creep properties yielded good quality
models and allowed several conclusions to be drawn:
a) The copolymer was shown to creep significantly more than the
homopolymer (e.g. 30% for a draw area of 4.5) especially at low draw
ratios.
b) Material that has a high draw ratio creeps far less than that with a low
draw ratio.
c) The natural logarithm of the copolymer creep strain model was shown to
be linearly correlated with orientation, as measured by birefringence.
Creep strain was not strongly influenced by crystallinity.
284
d) Heat-setting can substantially reduce creep strain in either material e.g.
from 50% to 30% for copolymer with a draw area of 5.
e) The time component of the heat-setting treatment is far more significant
than the temperature; this is the only occasion this has occurred in this
investigation.
The fact that heat-setting improves creep resistance provides good evidence
that the mechanism for the reduction of shrinkage, by heat-setting, does not
significantly disorient the amorphous phase. A less oriented material would be
prone to higher creep levels.
6.1.5 Shrinkage Measurements
The primary conclusions on the ability of heat-setting to control thermally
induced shrinkage in bottles is sununarised in section 6.1.3.
a) Shrinkage data on biaxially oriented PET samples have been generated
using thermo-mechanical analysis (TMA). The effects of thermal
expansion have been accounted for, and eliminated, by parallel tests on
annealed, isotropic samples.
b) Both the PET material grades behave very similarly, with the
homopolymer demonstrating slightly higher shrinkage at lower heat-
setting temperatures than the copolymer.
c) Heat-setting can substantially reduce shrinkage, typically from a 'worst
case' of 17% to 3%, but for the major axis, this appears to be confmed to
draw areas of in excess of approximately 7.5.
d) Below a draw area of 7.5 heat-setting reduces the negative shrinkage,
observed in the major axis, until positive shrinkage values of about 3%
were recorded.
285
e) Heat-setting at higher temperatures causes the shrinkage to be far less
dependent on draw. In the heat-setting of PET bottles, where there is a
complex series of draw ratios in different regions of the containers, high
temperature heat-setting is predicted to give not only lower but also more
uniform shrinkage.
f) For both materials below a draw of about 7.5 there is a competing process
of expansion, resulting in an overall increase in length for the lower
draws. This only occurred in the major axis, ruling out the possibility of
the sample creeping during the TMA test. This effect was fIrst seen in the
FED work but then subsequently confIrmed in the conventional heat-
setting series of fIlms.
g) For fIlms with a draw area of 5, there is a clearly defmable linear
relationship between the shrinkage aspect ratio in the two draw axes and
heat-setting temperature.
h) The fIlms with a draw ratio of 10 did not show this relationship, neither
did the shrinkage data derived from the PET bottle results. As the draw
area in the side wall of the bottles, where the samples were taken was
high, (9), these results concur.
i) At IIOoe, increasing the heat-setting temperature from 80C to 220C can
reduce shrinkage in the hoop axis from 11% (to displaying 1% expansion)
and from 23% to 3% in the axial axis.
j) The shrinkage observed in the bottles was found to be a linear function of
heat-setting temperature, particularly in the hoop axis.
k) For the heat-set bottles, crystallinity was to be strongly correlated heat-
setting temperature and shrinkage.
I) Shrinkage in the fIlms was found to be strongly, but not wholly dependent
on orientation, as measured by birefringence.
286
The work on heat-set bottles clearly demonstrates that heat-setting can be used
to substantially reduce thermally induced shrinkage in PET stretch blown
containers. The models generated by the film experiments fit the bottle data
well, especially at heat-setting temperatures above 140C. From this it may be
concluded that the predictive models from experiments on biaxially oriented
films, may be used to simulate the behaviour of stretch blown bottles, with
respect to heat-setting and shrinkage, at least to a first approximation.
6.1.6 Thermal Analysis and Density Measurements
There are two quite distinct elements to the DSC analysis; one was the
measurement of the temperature for the 13-peak (or secondary endotherm) and
the other was the measurement of the overall degree of crystallinity in the PET
sample prior to the DSC scan.
a) The correlation between the 13-peak and heat-setting temperatures was
found to be excellent, and the two temperatures were directly equivalent
for these test conditions. From this it may be concluded that DSC
provides an accurate measurement of the 13-peak temperature and that the
equipment was heat-setting the films at the desired temperature.
b) The degree of crystallinity measurements were loosely correlated with
heat-setting temperature, for the PET films or bottles. The source of this
anomaly has not been identified.
c) The estimation of crystallinity by refractive index proved to have a better
correlation with the independent variables, and less scatter than the DSC
measurements.
d) The best of the methods used by this investigation, for obtaining the
degree of crystallinity was the density measurement technique.
287
e) The models generated from the density data predict increasing heat-setting
temperature to increase the degree of crystallinity, e.g. from 25% to 60%
by increasing the heat-setting temperature from 80C to 220C and by
increasing the heat-setting time from 6 seconds to 30 seconds.
f) Increasing the draw ratio from 5 to 10 only increases the predicted
crystallinity by 10% for the homopolymer, and even less for the
copolymer.
g) The homopolymer crystallinity is not modelled to be a function of heat-
setting time whereas the copolymer is. The converse is true of the effect
of draw.
h) The models generated by the FED analysis fit the bottle crystallinity data
very well, for heat-setting temperatures above 80C, and this fit can be
substantially improved by increasing the 'time' value in the equation from
15 to 25 seconds.
i) The agreement of crystallinity results from the film and bottle
experiments is strong supportive evidence that bottle behaviour may be
modelled by films.
Even by taking into account the assumptions made about the constant density
of the amorphous phase, from this work it must be concluded that DSC cannot
be relied upon to produce meaningful degree of crystallinity data on oriented
heat-set PET, using the experimental technique applied in this investigation.
The effect that degree of crystallinity has on yield stress, creep strain, and
shrinkage are summarised in sections 6.1.4.1d, 6.1.4.2c, and 6.1.5j and k.
288
6.1.7 Optical Measurements
The clarity of the heat-set films was measured by two techniques and neither
suggested that heat-setting PET under these experimental conditions gave rise
to a significant loss of clarity. The Fresnel correction was applied to the
Hazeguard light transmission data to account for reflection at the films'
surfaces. The Lambert correction was also applied to the data but proved
insufficient to correct for the effect of sample thickness. The micro-
photometry results were too scattered to derive further information from them.
a) The birefringence and refractive index measurements yielded substantial
information about the material's behaviour during drawing and stretching.
The birefringence data are a function of several factors, the most relevant
here are orientation in the amorphous phase and orientation due to
crystallisation.
b) The increase in the two out-of-plane birefringences through drawing is
most rapid up to draw area of approximately 3, which coincides with the
general transition point from Gaussian to non-Gaussian deformation for
PET. IV, draw temperature and strain rate will affect this transition point.
c) The in-plane birefringence is largely unaffected by changes in draw ratios
greater than approximately 3. This is assumed to reflect the constant ratio
of draws between the two principal axes.
d) Increasing the heat-setting temperature increases all three birefringences.
This counts against the possibility of the mechanism by which heat-setting
reduces shrinkage by reducing the orientation of the amorphous phase, as
this would reduce birefringence. It is possible that disorientation of the
amorphous could be occurring but the effect must be less than a
competing mechanism (i.e. crystallisation and/or increase of orientation)
as birefringence does increase during heat-setting. This is in agreement
with the observations made about the creep results.
289
e) The increase in in-plane birefringence implies that heat-setting increases
the structural imbalance between the two draw axes. The refractive index
data confinn that the maximum rate of structural development occurs
during the initial draw stages. They show that both drawing and heat-
setting reduce the out-of-plane orientation, giving rise to a more planar
structure.
f) The refractive index measurements provide structural evidence confmning
that the linear draws cannot be considered in isolation, when discussing
biaxially oriented PET. This effect was first seen in the shrinkage data.
The effect of birefringence on yield stress, creep strain and shrinkage has been
summarised in sections 6.1.4.1c, 6.1.4.2c, and 6.1.51.
6.1.8 Orientation Measurements By FTIR
The data derived from the polarised FTlR experiments are far from conclusive,
due to a high degree of scatter, but the following have been noted:
a) There is some evidence that the bulk orientation functions are increased
with heat-setting temperature, but not in a systematic way.
b) The amorphous orientation functions showed little variation with heat-
setting temperature, indicating that disorientation does not occur in this
phase during the heat treatment. The dichroic ratios are also highly
scattered but do tend to increase with heat-setting temperature.
290
6.2 SUGGESTED FURTHER WORK
This investigation has been, from the outset, industrially-led and so has been
extremely broad in both objectives and experimental content. There was no
prior work which had investigated the heat-setting of PET in this way and so
this project could, in some areas, be viewed almost in terms of a pilot or
feasibility study. However, the project's primary objective concerning the
identification of processing conditions to control shrinkage below
commercially acceptable levels has been achieved. In addition, a considerable
amount of information has been generated on mechanical, creep and optical
properties and their interaction. As with any investigation of this nature, there
is considerable possibility for further work, and given here is a brief outline of
key areas where further work is most likely to yield interesting results.
a) The use ofTMA to measure shrinkage in biaxially oriented PET, the heat-
setting rig, the statistical models and the heat-setting process used for the
bottles are all new. As is usual with new techniques, a stage of
refmement would undoubtedly improve them.
b) The relationships between birefringence, creep and shrinkage indicate that
orientation is of central importance to the properties of PET and their
modification by heat-setting. The development of P-FTIR to study this in
more detail should give more detail about the operative mechanism(s). It
is felt that concentrating on the weak absorbing bands in the PET IR
spectrum would remove the need for sectioning, and therefore reduce the
errors induced here and in the management of a complex optic train.
c) The inability of DSC to quantify crystallinity in PET in this project is
interesting, as it has gained wide acceptance in this role. It was wished
that XRD could have been used to provide a third alternative technique in
an attempt to resolve between DSC and density measurements, but this
was not possible due to time constraints.
291
d) The unexpected behaviour of PET ("negative shrinkage") with a relatively
low draw ratio in the major axis has not been successfully explained.
Despite significant efforts to fmd experimental reasons for this trend,
none have been found and so these results still stand. The linear
relationship found between shrinkage aspect ratio and heat-setting
temperature has not to our knowledge been reported in the literature, and
any linear relationship should evoke some interest. Obviously, further
work is needed here to establish the validity, or otherwise, of this trend.
This could include some accurate 'before and after' type shrinkage
measurements on biaxially oriented PET.
e) An investigation of different grades and drawing conditions (temperature,
strain rate and ratio between draw axes) would give a more complete
understanding of the effect of heat-setting PET. This could usefully be
extended to include pre-oriented and pre-crystallisedJnucleated PET.
f) For commercial acceptance, the shrinkage models would have to be tested
using an industrial heat-setting process. The models would probably
undergo some refmement by this process.
g) Once the models have been verified and refmed where necessary, they
could be incorporated into software to allow accurate design of shrinkage,
mechanical and creep strain property behaviour in heat-set PET bottles.
h) The statistical models were generated for biaxially oriented PET films.
PET films are commercially very important and so these models can be
applied here as w l ~ to enhance mechanical or creep properties.
Depending on the application for the films, shrinkage is sometimes
desirable e.g. shrink-wrap films. The models could be used to either
increase or decrease shrinkage, depending on the application.
i) PET is fmding increasing use in extrusion blow moulding applications.
This process imparts uniaxial orientation. The models were developed for
stretch blown containers which will be biaxially oriented. The models
should need little, if any modification to be applicable to uniaxial
deformation and so aid in the design of PET blow moulded containers.
292
j) This approach could then be extended to other materials such as PVC,
which undergoes secondary crystallisation on annealing and is often used
in an oriented condition. There is considerable interest in heat-setting
PVC and at least some of infonnation given by this investigation should
be of use.
h) The relationships between birefringence, yield stress and creep strain
could be developed in to a non-destructive on-line quality control
technique for PET bottles.
293
REFERENCES
1 PET Pointer, Issue 4, Technical Publication, Krupp Corpoplast Gmbh.,
June 1990, p4.
2 ICI Technical Information "Melinar No 6"
3 rCI Technical Information "Melinar No 10"
4 Meehan, T.P. and McBride, P., High Performance Plastics Packaging,
9th International S.P.E. Conference, London, Qct 1988.
5 Hsuie, G., Yeh, T., Chang, S., J. Appl. Polym. Sei., 1989,37,2803.
6 Crawford, RJ., Plastics Engineering, Pergamon Press, New York,
1981, p21.
7 Haworth, B., Dong, Z. and Davidson, P., Polym. Int., 1993,32, 325.
8 Peszkin, P. and Schultz, J.M., J. Polym Sci: Polym. Phys.Edn.,1986, 24,
2591.
9 Biangardi, HJ. and Zachmann, H.G., Prog. Colloid. Sci., 1977,6,271.
10 Bosley, D.E., J. Polym. Sei.: C, 1967,20,77.
11 Samueis, R.T., J. Polym., Sci. A-02, 1972,9,781.
12 Wilson, M.P.W., Polymer, 1977, 15, 277.
13 Kuhn, W. and Grun, F., Kolloid Z, 1942, 101,248.
14 Treioar, L.RG., Trans. Faraday Soc., 1947, 43, 284.
15 Alfrey, T., Wiederholm, N., Stein, R.S. and Tobolsky, A.V., J. Colloid
Sci., 1949,4,211.
16 Rudd, H. and Andrews, RD., J. Appl. Phys., 1958,29, 1421.
17 Pinnock, P.R and Ward, I.M., Trans. Faraday Soc., 1966, 62, 1308.
294
18 Bhatt, G.M. and Bell, J.P., J. Polym., Sci.: Polym. Phys.Edn., 1976, 14,
575.
19 Dumbleton, J.H., J. Polym. Sci.: C, 1966, 13, 133.
20 Long, S.D. and Ward, I.M., J. Polym. Sci., 1991,42, 1921.
21 Edwards, S.F. and Vilgis, Th., Polymer, 1986,27,483.
22 Perkins, W.G., 41st Annual Technical Conference, Society of Plastics
Engineers, 1983.
23 De Vries, AJ., Bonnebat, C. and Beautemps, J., J. Polym. Sci: Polym.
Symp., 1977, 58, 109.
24 Peterlin, A, J. Mat. Sci, 1971,6,490.
25 Buckley, C.P. and Salem, D.R., Polymer, 1987,28,69.
26 Geil, P.H., Polymer Single Crystals, (Interscience, Pub!., J. Wiley and
Sons, New York), 1963,421.
27 Geil, P.H., J. Polym. Sci., 1964, A2, 3813.
28 Kiho, H., Peterlin, A. and Geil, P.H., J. Appl. Phys., 1964, 35, 1599.
29 Peterlin, A, J. Polym., Sci:C., 1965, 9, 61.
30 Sun, D.C. and Magill, J.H., Polym. Eng. and Sci., 1989,29, No 21,
1503.
31 Dumbleton, J.H., J. Polym. Sci., A2, 1969, 667.
32 Statton, W.O., Koenig, J.L. and Hannon, M., J. Appl. Phys., 1970, 41,
4290.
33 Amano, E.W. and Hinrichson, G., J. Macromol. Sci-Phys, 1969, D3,
209.
34 Jordon, M.E., Juska, T.D. and Harrison, I.R., Polym. Eng. and Sci.,
198626, No 10, 690.
295
35 Juska, T.D. and Harrison, I.R., Polym. Eng. Rev., 1982,1, 13.
36 Juska, T.D. and Harrison, l.R., Polym. Eng. and SCi., 1982 12, 766.
37 Prevorsek, D.C., Kwon, Y.P. and Shanna, R.K., J. Mat. Sci., 1977, 12,
2310.
38 SpruieIl, J.E., McCord, D.E. and Beuerlein, R.A, Trans. Soc. of Rh eo. ,
1972, 16:,3, 535.
39 Yeh, G.S.Y. and Geil, P.H., J. Macromol. Sci., 1967, B3, 251.
40. Foster, E.L. and Heap, H., The Rheology ofElastomers, 1957,
Pergarnon Press, New York, 190
41 Thompson, AB.,J. Polym. Sci., 1959,34,741.
42 Thompson, AB., MarshaIl, l., Proc. Roy. Soc. (London), 1954, A211,
541.
43 Bonnebat, C., RouIlet, G. and DeVries, AJ., Polym. Eng. Sci., 1981,21,
189.
44 Warner, S.B.,J. Appl. Polym. Sci., 1984,29,219.
45 Cakmak, M., White, J., SpruieIl, J., J. Appl. Polym. Sci., 1985,30,
3679.
46 Vincent, P., Polymer, 1960, 1,7.
47 Peterlin, A, Colloid Polym. Sci., 1987,265,357.
48 Prevorsek, G.A., Harget, PJ., Shanna, R.K. and
Reimischuesse1, AC., J. Macromol. Sci., 1973, B8, 1-2, 127.
49 Prevorsek, D.C., Tirpak, G.A., Harget, PJ., Shanna, RK and
Reimischuessel, AC., J. Macromol. Sci., 1974, B9(4), 733.
50 Tanaka, A, Nagano, H. and Onogi, S., Polymer Journal, 1988,20,
No 11,1003.
296
51 Petennann and Rieck, J. Polym. Sci.: Polym. Phys.Edn, !987, 25, 274.
52 Engelaere, lC., Cavrot, lP. and Rietsch, F., Polymer, 1982,23, 766.
53 Elenga, R., Seguela, R. and Rietsch, F., Polymer, 1991,32, No 11,
1975.
54 Hennessey, and WJ., Spatorico, A.L., Polym. Eng. and Sci., 1979, 19,
No 6, 462.
55 Koening, J.L. and Mele, M.D., Makromol. Chem., 1968, 118, 128.
56 Farrow, G. and Ward, LM., Polymer, 1960, 1,330.
57 Fischer, E.W. and Fakirov, S., J Mat Sci, 1976, 11, 104
58 Stem, P.G., Kolloid Z. Polymer 1967215, 140.
59 Linder, W.L., Polymer, 1973, 14, 9.
60 Jungnickel, BJ., Polym. Eng. and Sci., 1987,27, No 14, 1021.
61 Heffelfmger, CJ. and Schmidt, P.G., 1. Appl. Polym. Sci., 1965, 9,
2661.
62 Daubeny, R. de P., Bunn, C.W. and Brown, CJ., Proc. Roy. Soc.
(London), 1954, A226, 531.
63 Cakmak, M. and Wang, Y.D., J. Appl. Polym. Sci., 1990, 41, 1867.
64 Sun, T., Pereira, J.R.C. and Porter, R.S., l Polym. Sci.: Polym. Phys.
Edn., 1984,22, 1163.
65 Bouvellec, G. le, Monnerie, 1. and Jarry, lP., Polymer, 1987, 28, 1712.
66 Bouvellec, G. le, Jarry, J.P. and Monnerie, L., Polymer, 1986, 27, 856.
67 Barham, P. and Keller, A., 1. Polym. Sci. Letters, 1975, 13, 197.
68 Gute, K.M., Motz, H. and Schultz, 1. Polym. Sci.: Polym. Phys. Edn.,
1983, 21, 1927.
297
69 Tiller, WA and Schultz, J.M., J. Polym. Sci.: Polym. Phys. Edn., 1984,
22, 143.
70 Dismore, P.F. and Statton, W.O., J. Polym. Sci.:C, 1966, 13, 133.
71 Wohlisch, E., Kolloid Z, 1939,89, 239.
72 Muller, F.H., Kolloid Z, 1941,95, 138,306.
73 Kuhn, W., Grun.F., Kolloid Z, 101, 1942,248.
74 Struik, L.C.E., Internal Stresses, Dimensional Instabilities and
Molecular Orientations in Plastics, J.Wiley 1990, p302.
75 Vallat, M.F. and Plazek, P., J. Polym. Sci.: Polym. Phys. Edn., 1988,
26,545.
76 Hermans, P.H. and Vallat, M.F., Kolloid Z, 1939, 88, 68.
77 Hermans, J.J., Hermans, P.H., Vermaas, D. and Weidinger, A, Rec.
Trac. Chim., 1946,65,427.
78 Hermans, P.H., Hermans, J.J., Vermaas, D. and Weidinger, A, J.
Polym. Sci., 1948,3, 1.
79 Muller, F.H., Kolloid Z, 1941,95, 138,207.
80 Ward, I.M. and Mc Briety, V.J., J. Appl. Phys., 1968, 1, 1529.
81 Stein, R.S., J. Polym. Sci., 1958,31,327.
82 White, J.L. and Spruiell, J.E., Polym. Eng. Sci., 1981,21,859.
83 Gupta, V.B. and Ramesh, C., J. Polym. Sci.: Polym. Phys. Edn., 1985,
23,405.
84 Kashiwagi, M., Cunningham, A, Manuel, AJ. and Ward, I.M.,
Polymer, 1973, 14, 111.
85 Ma, T.e. and Han, C.D., J. Appl. Polym. Sci., 1988, 35, 1725.
86 Jabarin, SA, Polym. Eng. and Sci., 1984, 24, No 5, 376.
298
87 Heffelfinger, CJ. and Burton, R.L., J. Polym. Sci., 1960, XLVII, 289.
88 Sisson, W.A. and Clark, G.L., Ind. Eng.: Chem. Anal. Edn., 1933,5,
296.
89 Selwood, A., Ward, LM. and Craggs, G., Plast. and Rubber Processing
and Appl., 1988, 10, No 2, 93.
90 Gilbert, M., and Lui,Z., Plast. and Rubber Processing and Appl., 1988,
9, No 2,67.
91 Salem, D.R. and Weigmann, H.D., J. Polym. Sci.: Polym. Phys.
Edn.,1991, 29, 765.
92 Przygochi, W., Acta Polym., 1982,33, 729.
93 Cakmak, M., Spruiell, J.E. and White, J. Polym. Eng. and Sci., 1987,
27, No 12, 893.
94 Gupta, V.B., Ramesh, C. and Gupta, A.K., J. Appl. Polym. Sci., 1984,
29,3727.
95 Leu, S.H. and Mc Carthy, S., 133 Kim, Antec 90, Annual Technical
Conference. Society of Plastics Engineers, 1990, 1000.
96 Haworth, B., Davidson, P., Jones, K.M., Sykes, G., Proc Europak 92
Ryder ConJ26-28 Nov 1992 Dusseldorf, 1992,485.
97 Buckley, C.P. and Salem, D.R., J. Appl. Polym. Sci., 1990,41, 1707.
98 Cakmak, M., Spruiell, J.E. and White, J. Polym. Eng. and Sci., 1984,
24, No 18, 1390.
99 American Society for the Testing of Metals, ASTM D882.
International Standards Organisation, ISO-R1184-1970.
100 Samuels, RJ., Structured Polymer Properties, John WHey and Sons,
New York, 1974.
101 Retting, W., Colloid Polym. Sci., !975, 253, 852.
299
102 Jones, T.T., Pure andAppl. Chem., 1976,45,39.
103 Ward, I.M., Polymer, 1964,5,59.
104 Murayarnarn, T., Dumbleton, J.H. and Williarns, M.L., 1. Polym. Sci.,
A-2,6, 1968, 787.
105 Gupta, V.B., Rarnesh, C. and Gupta, A.K., 1. Appl. Polym. Sci., 1984,
29,4203
106 Hawthome, J.M., J. Appl. Polym. Sci., 1981,26,3317.
107 Bhusharn, B., Heinrich, J.C. and Connolly, D., in Bhusharn, B., Bogg,
D. Eiss, N.S. Jr. and Talke, F.E., Eds., Trihology and Mechanics of
Magnetic Storage Systems, American Society of Lubrication
Engineering, ASLE SP-16, Park Ridge, IL, 1984, p158.
108 Wallach, W.L., 1. Polym. Sci.: C, 1966,13,69.
109 Carnpbell,D. and White, J.R., Polymer Charaterisation Physical
Techniques, Chapman and Hall, 1989, p272.
110 Dumbleton, J.H., 1. Polym. Sci., 1968,6, A2, 795.
111 Konda, A., Nose, K. and Ishikawa, H., 1. Polym. Sci., 1976, 14, A2,
1495.
112 Gupta, V.B. and Kuma, S., 1. Polym. Sci.:Polym. Phys. Edn., 1979,17,
1307.
113 Perefia, J.M. and Benavenk,R., Polym. Eng. and Sci., 1988,28, No 19,
1260.
114 Rietsch, F., Duckett, R.A. and Ward, LM., Polymer, 1979,20, 1133.
115 Garg, S.K., 1. Appl. Polym. Sci., 1982,27,2857.
116 Engelaere, J.C., Cavrot, J.P. and Rietsch, F., Europ. Polym. 1., 1980,
16, 721.
300
117 Yokouchi, M., Hiromoto,Y. and Kobayashi, Y., J. Appl. Polym. Sci.,
1979,24, 1965.
118 Zachmann, H.G., Polym. Eng. Sci., 1979, 19,966.
119 Gupta, V.B., Sett, S.K. and Deorukhkar, P.D., Polymer
Communications, 1989,30,341.
120 Zaibicki, A. and Jarecki, L., High-speed Fibre Spinning, (Eds Zaibicki,
A. and Kawai), H. Wiley Interscience, New York, 1985, p263.
121 Gupta, V.B. and Ramesh, C., Polymer Communications, 1987,28,43.
122 Dulmage, W.J. and Geddes, A.L., J. Polym. Sci., 1958,31,499.
123 Moore, R.S., O'Loane, J.K. and Shearer, I.C., Polym. Eng. and Sci.,
1981, 21, No 14, 903.
124 Aharoni, S.M., Sharma, R.K., Szobota, J.S., Vemick, D.A., J. Appl.
Polym. Sci., 1983,20,2177.
125 Stokr, I., Schneider, B., Doskocilova, D., Lovy, J. and Sedlacek, P.,
Polymer, 1982, 23, 714.
126 Sikka, S.S. and Kausch, H.H., Colloid and Polym. Sci., 1979,257,
1060.
127 Cunningham, A., Ward, I.M., Willis, H.A. and Zichy, V., Polymer,
1974, 15, 749.
128 Jarvis, D.A., Hutchinson, I.I., Bower,D.1. and Ward, I.M., Polymer,
1980,21,41.
129 Chalmers, I.M., Croot, L., Eaves, J.G., EveraII, N., Gaskin, W.F.,
Lumsdon, J. and Moore, N., Spectros. 1nt. J., 1990,8, 13.
130 Hutchinson, I.J., Ward, I.M., Willis, H.A. and Vichy, V., Polymer,
1980, 21, 55.
131 EveraII, N. Private Communication, 1992.
132 Miyake, A., J. Polym. Sci., 1959,38,479.
301
133 Kim, S.L., 41st Annual Technical Conference, Society of Plastics
Engineers, 1983.
134 Fakirov, S., Fisher, E.W., Hoffmann, R. and Schmidt, G.F., Polymer,
1977, 18, 1121.
135 Nearly, D.L., Davis, T.G. and Kibler, C.J., Polym. Sci., 1970, 8, A2,
2141.
136 Wunderlich, B., in: Thermal Characterisation of Polymeric Materials,
Chpt 2, Turi, E.A., Ed, Academic Press, 1981.
137 White, T.R., Nature, 1955,175,895.
138 Yoshimoto, T. and Miyagi, A., Kogyo Kagaku, Zashi (J. Chem. Soc.
JapanindChem. Sect.), 1966,69,1771.
139 Bell, J.P. and Dumbleton, J.H., J. Polym. Sci., 1969,7, A2, 1033.
140 Todoki, M. and Kawagunhi, T., J. Polym. Sci.: Polym. Phys. Edn.,
1977, 15, 1067.
141 Rao, M.V.S., Kumar, R. and Dweltz, N.E., J. Appl. Polym. Sci., 1986,
32,4439.
142 Fontaine, F., Ledent, J., Groeninckx, G., Reynaers, H., Polymer, 1982,
23, 185.
143 Avramova, N., and Fakirov, S., J. Appl. Polym. Sci., 1986,31, 1631.
144 Peterlin, A., Meinel, J. Polym. Sci., B2, 1965, 751.
145 Miller, G.W., in Analytical Calorimetry, Voll, Porter, R.S., and
Johnson, J.F., Eds, Plenum, New York, 1969, p71.
146 Illers, K.H. and Breuer, H., J. Colloid Sci., 1963, 18, 1.
147 Imai, M., Mori, K., Mizukami, K., Kaji, T. and Kanaya, K., Polymer,
1992,33, No 21,4457.
302
148 Imai, M., Mori, K., Mizukami, K., Kaji, T. and Kanaya, K., Polymer,
1992,33, No 21, 4451.
149 Birley, A.W., Haworth, B., and Ola, A., 1nt. Polym. Process., V, 1990,
2.
150 Basir, K.B., and Freakley, P.K., Kauts. und Gummi. Kunst., 35, 1982,
305.
151 Hatt, B.W., J. Vinyl. Tech., 6, 1984, 120.
303
Appendix A1 Raw Force Data From Major Axis Tensile Experiments
Homopolymer
Run Sample Thickness Yield Force Ultimate Force
No
(Ilm)
(N) (N)
A B C A B C A B C
1
85 90 85 23.22 23.82 2.80 31.67 35.66 42.08
2
55 55 55 22.1 21.3 21.0 25.3 29.0 25.0
3
80 80 75 22.8 24.8 23.3 33.0 40.6 38.5
4
60 60 60 22.4 23.3 23.4 34.3 33.8 34.3
5
70 70 75 21.8 22.6 22.4 35.0 36.9 39.5
6
60 60 60 20.95 21.45 21.31 31.27 31.25 28.24
7
80 80 85 23.8 23.8 24.8 33.5 27.0 38.4
8
55 60 55 21.24 21.08 20.56 31.62 33.35 27.28
9
50 50 50 20.30 20.8 21.5 26.6 29.8 28.8
10
115 115 110 27.1 28 27.4 37.4 39.3 40.0
11
65 70 60 20.5 22.0 20.0 31.5 32.8 28.5
12
70 65 70 20.0 19.6 19.3 27.5 27.8 30.0
13
60 65 60 20.71 20.29 20.04 27.24 26.60 28.59
14
65 60 65 19.28 18.96 19.62 28.59 27.72 32.49
15
60 65 70 20.58 20.68 20.64 30.04 26.66 31.20
16
60 60 70 19.3 19.0 20.0 31.5 28.1 33.5
17
70 65 65 20.5 21.1 19.9 29.0 31.3 30.5
18
70 65 70 20.8 21.0 20.9 30.9 36.5 31.0
19
65 60 70 20.1 20.3 20.3 30.5 32.6 34.0
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
304
Appendix A2 Raw Force Data From Major Axis Tensile Experiments
Copolymer
Run Sample Thickness Yield Force Ultimate Force
No (Ilm)
(N) (N)
A B C A B C A B C
1
80 91 85 20.57 20.62 21.38 30.49 31.64 30.90
2
55 55 50 15.5 14.8 15.0 23.4 22.0 22.8
3
80 75 75 23.0 22.9 22.9 37.1 35.8 31.8
4
50 50 50 18.3 18.4 18.1 28.9 28.0 27.8
5
75 75 75 21.1 20.9 20.8 32.3 34.4 33.1
6
50 50 55 17.88 17.82 17.74 21.81 23.28 26.48
7
75 80 75 23.3 23.3 22.4 37.6 37.0 38.8
8
55 50 50 18.09 17.89 18.02 30.27 26.86 29.16
9
50 50 45 18.9 19.0 18.8 28.4 23.5 24.3
10
110 105 110 24.9 24.6 25.8 39.3 37.1 41.5
11
70 70 70 20.8 20.5 20.4 37.9 36.3 30.3
12
70 70 70 19.0 18.4 18.4 31.5 30.9 31.4
13
70 70 70 20.05 20.65 20.15 27.95 27.60 30.53
14
70 70 70 19.67 19.96 20.97 31.05 30.58 33.05
15
70 65 65 21.02 19.33 18.75 35.42 30.23 28.45
16
70 70 75 20.8 20.5 21.1 28.4 30.9 33.8
17
60 65 65 18.5 18.9 19.0 29.5 28.8 30.0
18
70 70 70 19.5 19.3 19.6 31.5 29.0 32.3
19
70 65 60 20.0 18.3 18.3 35.0 33.5 33.1
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
305
Appendix A3 Raw Force Data From Minor Axis Tensile Experiments
Homopolymer
Run Sample Thickness Yield Force Ultimate Force
No ~ m ) (N) (N)
A B C A B C A B C
1
80 90 85 21.59 22.73 21.77
-
24.06 20.65
2
55 55 55 16.04 15.99 16.23 19.58 19.20 11.93
3
70 65 70 21.20 21.10 21.18 22.47 22.28 19.40
4
55 55 55 17.66 17.69
-
21.67 13.56
-
5
70 70 70 20.34 20.05 20.40 24.39 20.50 22.45
6
55 50 50 15.42 16.06 15.80 17.74 11.52 16.89
7
75 75 80 14.83 22.75 23.26 23.53 20.73 24.43
8
50 50 50 17.05 15.39 16.92 10.77 17.54 20.79
9
50 50 50 15.21 15.23 14.99 17.87 12.14 17.46
10
110 105 100 29.41 29.28 27.86 22.92 22.18 23.01
11
65 65
-
19.23 19.43
-
22.76 20.29
-
12
65 65 65 18.96 19.24 18.90 12.53 21.34 22.41
13
65 60 60 18.65 18.96 18.87 20.43 21.44 19.77
14
65 60 60 16.75 16.83 17.05 19.17 20.05 18.92
15
70 70 65 19.89 20.39 19.54 20.72 14.51 48.72
16
60 60 60 17.79 17.86 17.76 18.72 14.43 20.02
17
60 60 60 18. \3 18.21 18.42 20.30 21.09 22.11
18
65 65 60 19.14 18.96 18.87 19.76 14.57
-
19
60 65 65 18.16 19.07 19.31 21.20 17.98 19.45
Premature sample failure is indicated by a '-'.
306
Appendix A4 Raw Force Data From Minor Axis Tensile Experiments
Copolymer
Run Sample Thickness Yield Force Ultimate Force
No
(Ilm)
(N) (N)
A B C A B C A B C
1
85 90 75 20.50 20.20 19.54 21.51 24.27 18.48
2
50 50 50 13.16
-
14.01 4.56
- 16.87
3
95 80 80 24.52 23.08 23.27
-
21.14 13.45
4
50 50 50 15.04 15.08 15.24 15.24 15.29 11.53
5
70 65 70 19.27 19.11 19.87 22.59 10.09 14.69
6
50 50 50 14.14 13.81 13.88 5.21 15.45 12.55
7
80 80 75 23.12 23.05 22.63 19.55 18.80 22.21
8
50 55 50 15.35 16.95 15.13
-
16.48 11.73
9
45 45 45 13.05 13.12 13.12 18.51 15.29 13.97
10
100 105 125 27.98 28.47 30.26 21.04 20.81
-
11
65 65 70 19.86 20.27 20.59 18.10 16.94 10.46
12
65 60 60 17.91 17.46 17.35 15.72 16.08 18.13
13
60 65 60 18.00 20.54 18.05 13.00 23.23 13.99
14
65 75 65 18.14 18.86 18.36 19.63 4.8 3.87
15
65 60 70 18.64 18.26 19.13 18.69 19.70 14.23
16
65 70 60 19.08 18.96 17.87 20.09 9.82 6.96
17
60 60 60
-
18.41
- -
21.18
-
18
70 65 65 18.85 18.96 19.12 8.46 14.57 18.56
19
65 60 60 18.54 18.23 17.61 17.10 17.27 11.23
Premature sample failure is indicated by a '-'.
307
Appendix AS Raw Extension Data From Major Axis Tensile Experiments
Homopolymer
Run No Yield Extension mm) Ultimate Extension (mm)
A B C A B C
1 2.11 2.13 2.14 32.75 40.14 47.54
2 2.4 3.1 3.1 13.8 19.5 14.3
3 3.1 2.1 2.6 33.0 33.0 36.0
4 2.5 3.0 2.9 18.9 16.8 17.0
5 2.1 2.5 2.4 29.0 29.5 32.5
6 2.30 2.38 2.54 22.52 22.43 18.96
7 2.4 2.8 2.8 32.0 33.3 37.8
8 2.36 2.48 2.52 18.28 20.82 14.27
9 2.8 3.0 2.9 16.1 18.9 16.5
10 2.5 2.0 2.0 50 50 47.3
11 2.0 2.6 2.5 23.6 23.6 19.6
12 1.8 2.0 2.4 26.5 27.3 32.8
13 2.43 2.34 2.45 19.93 21.89 25.28
14 2.07 2.15 2.01 28.95 28.20 34.89
15 2.43 2.35 2.23 27.15 22.65 30.32
16 2.4 3.0 2.5 36.8 25.0 33.5
17 1.9 2.3 2.9 28.0 29.0 33.0
18 2.1 3.0 2.6 28.5 32.1 32.9
19 2.5 3.1 2.9 28.5 33.4 34.9
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
308
Appendix A6 Raw Extension Data From Major Axis Tensile Experiments
Copolymer
Run No Yield Extension mm) Ultimate Extension (mm)
A B C A B C
1 2.00 1.99 2.01 37.71 46.34 44.32
2 2.4 1.6 2.4 20.1 20.3 21.0
3 2.4 2.4 2.4 35.6 34.8 29.0
4 2.0 2.4 2.5 16.8 19.5 18.6
5 2.4 2.4 2.4 34.4 38.4 35.1
6 2.15 2.41 2.15 14.82 17.21 21.45
7 2.4 2.5 2.4 27.8 30.3 34
8 1.99 2.26 2.09 22.29 18.51 20.97
9 2.4 2.5 2.5 20.4 13.4 14.9
10 1.5 3.0 2.8 54.5 47.8 53.5
11 2.4 2.1 2.4 25.3 32.8 21.8
12 2.5 1.9 1.9 40.8 35.9 39.8
13 2.04 2.06 2.06 27.75 29.15 37.07
14 1.89 1.94 2.06 32.32 30.99 29.52
15 2.07 2.06 1.99 37.87 31.91 32.71
16 2.5 2.6 2.4 28.6 35.9 36.0
17 2.5 2.4 2.5 30.4 28.1 28.8
18 2.5 2.1 2.4 35.3 29.3 35.9
19 2.4 2.5 1.8 33.9 34.8 35.4
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
309
Appendix A 7 Raw Extension Data From Minor Axis Tensile Experiments
Homopolymer
Run No Yield Extension mm) Ultimate Extension (mm)
A B C A B C
1 2.30 2.28 2.32 58.86 64.99 59.08
2 2.01 1.96 1.95 40.49 41.33 20.58
3 2.65 2.56 2.63 58.86 61.04 55.10
4 2.17 2.30
-
47.21 3.54 4.23
5 1.89 2.07 2.Il 68.74 58.86 59.02
6 1.81 1.90 2.02 40.82 5.06 36.15
7 1.17 2.60 2.51 50.88 47.46 57.60
8 2.12 2.01 2.11 4.92 39.48 46.73
9 2.04 1.99 2.07 31.84 5.33 30.18
10 2.34 2.18 2.31 46.82 67.54 43.48
11 2.00 1.85
-
42.31 28.92
-
12 1.96 1.96 1.90 15.85 38.81 40.88
13 2.35 2.51 2.66 35.31 36.82 32.28
14 2.36 2.02 2.16 44.07 49.59 43.09
15 1.88 1.93 2.03 57.54 41.89 53.31
16 1.95 2.10 2.23 43.04 33.60 49.34
17 2.41 2.30 2.34 41.47 51.02 51.35
18 2.09 2.09 2.02 30.07 8.28
-
19 2.27 2.52 2.42 66.81 57.96 60.70
310
Appendix A8 Raw Extension Data From Minor Axis Tensile Experiments
Copolymer
Run No Yield Extension 'mm) Ultimate Extension (mm)
A B C A B C
1 2.28 2.24 2.20 70.95 81.28 67.37
2 2.03
-
1.8 3.63
- 51.41
3 2.74 2.49 2.49 51.16 73.28 9.00
4 1.95 1.95 1.97 33.94 32.42 5.47
5 1.79 1.75 1.91 76.61 55.47 38.25
6 1.72 1.76 1.62 3.63 39.56 32.31
7 2.49 2.60 2.59 11.51 55.16 69.50
8 1.76 1.78 1.81 35.87 23.33 24.89
9 1.93 1.93 1.97 42.50 31.16 26.36
10 2.09 2.28 2.24 20.33 21.64 4.23
11 1.88 1.76 1.76 61.12 43.23 26.00
12 1.73 1.79 1.62 52.61 53.42 56.34
13 2.51 2.51 2.30 35.11 36.82 38.56
14 2.03 2.01 2.10 58.55 12.27 5.17
15 1.98 1.90 2.00 59.47 30.58 18.49
16 1.90 1.96 3.05 59.75 42.11 4.53
17
-
2.41
- -
62.97
-
18 1.96 2.10 2.14 3.85 8.28 62.86
19
-
2.35 2.32 60.70 60.28 3.39
311
Appendix A9 Yield Stress-Major Axis
Homopolymer Copolymer
Run No Stress I Stress 2 Stress 3 Run No Stress I Stress 2 Stress 3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
I 80.35 77.84 82.35 I 75.63 67.39 73.98
2 118.34 113.64 112.30 2 82.89 79.14 88.24
3 83.82 91.18 91.37 3 84.56 89.80 89.80
4 109.80 114.22 114.71 4 107.65 108.24 106.47
5 91.60 94.96 87.84 5 82.75 81.96 81.57
6 102.70 105.15 104.46 6 105.18 104.82 94.87
7 87.50 87.50 85.81 7 91.37 85.66 87.84
8 113.58 103.33 109.95 8 96.74 105.24 106.00
9 119.41 122.35 126.47 9 111.18 111.76 122.88
10 69.31 71.61 73.26 10 66.58 68.91 68.98
11 92.76 92.44 98.04 11 87.39 86.13 85.71
12 84.03 88.69 81.09 12 79.83 77.31 77.31
13 101.52 91.81 98.24 13 84.24 86.76 84.66
14 87.24 92.94 88.78 14 82.65 83.87 88.11
15 100.88 93.57 86.72 15 88.32 87.47 84.84
16 94.61 93.14 84.03 16 87.39 86.13 82.75
17 86.13 95.48 90.05 17 90.69 85.52 85.97
18 87.39 95.02 87.82 18 81.93 81.09 82.35
19 90.95 99.51 85.29 19 84.03 82.81 89.71
312
Appendix AlO Yield Stress-Minor Axis
Homopolymer Copolymer
Run No Stress I Stress 2 Stress 3 Run No Stress I Stress 2 Stress 3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1 79.38 74.27 75.33 1 70.93 66.01 76.63
2 85.79 85.51 86.79 2 77.39
-
82.44
3 89.07 95.47 89.01 3 75.92 84.85 85.54
4 94.43 94.58
-
4 88.50 88.70 89.67
5 85.46 84.23 85.71 5 80.96 86.47 83.50
6 82.48 94.45 92.92 6 83.18 81.26 81.65
7 58.17 89.23 85.51 7 85.00 84.75 88.76
8 100.30 90.52 99.5 8 90.03 90.65 88.98
9 89.48 89.56 88.19 9 85.29 85.77 85.73
10 78.64 82.03 81.93 10 82.30 79.76 71.19
11 87.01 87.94
- 11 89.87 91.70 86.52
12 85.80 87.07 85.5 12 81.02 85.59 85.06
13 84.38 92.92 92.48 13 88.23 92.92 88.50
14 75.78 82.48 83.60 14 82.10 73.97 83.08
15 83.59 85.68 88.42 15 84.33 89.50 80.37
16 87.22 87.54 87.05 16 86.35 79.68 87.60
17 88.86 89.27 90.28 17
-
90.25
-
18 86.62 85.77 92.50 18 79.20 85.77 86.53
19 89.04 86.31 87.37 19 83.88 89.37 86.34
Premature sample failure is indicated by a '-'.
313
Appendix All Yield Strain-Major Axis
Homopolymer Copolymer
Run No Strain 1 Strain 2 Strain3 Run No Strain 1 Strain 2 Strain 3
(%) (%) (%) (%) (%) (%)
I 7.54 7.6\ 7.64 I 7.14 7.11 7.19
2 8.5 II II 2 8.5 5.5 8.5
3 11 7.5 9 3 8.5 8.5 8.5
4 9 10.5 10.5 4 7 8.5 9
5 7.5 9 8.5 5 8.5 8.5 8.5
6 8.21 8.51 9.06 6 7.66 8.62 7.68
7 8.5 10 10 7 8.5 9 8.5
8 8.43 8.86 9.00 8 7.11 8.07 7.46
9 10 10.5 10.5 9 8.5 9 9
10 9 7 7 10 5.5 10.5 10
11 7 9.5 9 11 8.5 7.5 8.5
12 6.5 7 8.5 12 9 7 7
13 8.66 8.37 8.74 13 7.28 7.36 7.34
14 7.39 7.66 7.16 14 6.75 6.92 7.34
15 8.69 8.40 7.96 IS 7.39 7.37 7.09
16 8.5 10.5 9 16 8.5 9.5 8.5
17 7 8 10.5 17 8.5 8.5 9
18 7.5 10.5 9.5 18 8.5 7.5 8.5
19 9 11 10.5 19 8.5 9 9.5
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
314
Appendix Al2 Yield Strain-Minor Axis
Homopolymer Copolymer
Run No Strain 1 Strain 2 Strain3 Run No Strain 1 Strain 2 Strain 3
(%\ (%\ (%\ (%\ (%) (%)
1 8.22 8.13 8.29 1 8.15 8.01 7.87
2 7.18 7.01 6.96 2 7.26
-
6.44
3 9.47 9.16 9.38 3 9.78 8.91 8.91
4 7.76 8.22
-
4 6.97 6.98 7.05
5 6.76 7.38 7.54 5 6.39 6.24 6.83
6 6.47 6.80 7.23 6 6.15 6.28 5.80
7 4.16 9.29 8.96 7 8.89 9.30 9.26
8 7.57 7.19 7.52 8 6.27 6.36 6.48
9 7.29 7.11 7.40 9 6.88 6.89 7.04
10 8.37 7.77 8.27 10 7.45 8.15 8.00
11 7.16 6.60
-
11 6.70 6.30 6.29
12 7.00 6.99 6.80 12 6.17 6.40 5.77
13 8.40 8.95 9.50 13 8.98 8.95 8.23
14 8.43 7.20 7.70 14 7.26 7.16 7.50
15 6.70 6.90 7.25 15 7.06 6.80 7.15
16 6.97 7.49 7.97 16 6.80 7.00 10.09
17 8.60 8.21 8.35 17
-
8.60
-
18 7.45 7.46 7.22 18 7.00 7.50 7.63
19 8.12 9.01 8.66 19
-
8.40 8.30
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chart recorder or by the more accurate
computer.
315
--------------
Appendix Al3 Ultimate Tensile Stress-Major Axis
Homopolymer Copolymer
Run No Stress 1 Stress 2 Stress 3 Run No Stress 1 Stress 2 Stress 3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1 109.58 116.54 145.61 1 112.10 103.40 106.92
2 135.29 155.08 133.69 2 125.13 117.65 134.12
3 121.32 149.26 150.98 3 136.40 140.39 124.71
4 168.14 165.69 168.14 4 170.00 164.71 163.53
5 147.06 155.04 154.90 5 126.67 134.90 129.80
6 153.28 153.19 138.43 6 128.29 136.94 141.60
7 123.16 99.26 132.87 7 147.45 136.03 152.16
8 169.09 163.48 145.88 8 161.87 158.00 171.53
9 156.47 175.29 169.41 9 167.06 138.24 158.82
10 95.65 100.51 106.95 10 105.08 103.92 110.96
11 142.53 137.82 139.71 11 159.24 152.52 127.31
12 115.55 125.79 126.05 12 132.35 129.83 131.93
13 133.53 120.36 140.15 13 117.44 115.97 128.28
14 129.37 135.88 147.01 14 130.46 128.49 138.87
15 147.25 120.63 131.09 15 148.82 136.79 128.73
16 154.41 137.75 140.76 16 119.33 129.83 132.55
17 121.85 141.63 138.01 17 144.61 130.32 135.75
18 129.83 165.16 130.25 18 132.35 121.85 135.71
19 138.01 159.80 142.86 19 147.06 151.58 162.25
316
Appendix A14 Ultimate Tensile Stress-Minor Axis
Homopolymer Copolymer
Run No Stress I Stress 2 Stress 3 Run No Stress I Stress 2 Stress 3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
I 79.38 78.62 71.46 I 74.43 79.32 72.46
2 104.70 102.70 63.81 2 26.84
-
99.26
3 94.43 100.80 81.52 3
-
77.71 49.44
4 115.90 72.52 95.85 4 89.67 89.95 67.84
5 102.50 86.13 94.32 5 94.93 45.66 61.71
6 94.86 67.76 99.34 6 30.67 90.89 73.83
7 92.28 81.30 89.82 7 71.89 69.10 87.09
8 63.34 103.20 122.30 8 95.11 88.15 69.02
9 105.10 71.40 102.70 9 121.00 99.94 91.29
10 61.28 62.13 67.69 10 61.88 58.30
-
11 103.00 91.79
- 11 81.92 76.66 43.93
12 56.70 96.58 101.40 12 71.15 78.80 88.85
13 92.43 105.10 96.90 13 63.71 105.10 68.60
14 86.74 98.27 92.73 14 88.84 18.83 17.49
15 87.05 60.98 84.69 15 84.57 96.54 59.81
16 91.78 70.72 98.14 16 90.92 41.28 34.10
17 99.50 103.40 108.40 17
-
103.8
-
18 89.39 65.91
- 18 35.55 65.91 84.00
19 103.90 81.38 88.07 19 77.38 84.67 55.04
Premature sample failure is indicated by a'-'.
317
Appendix A15 Ultimate Tensile Strain-Major Axis
Homopolymer Copolymer
Run No Strain 1 Strain 2 Strain3 Run No Strain 1 Strain 2 Strain 3
(%) (%) (%) (%) (%) (%)
1 116.96 143.36 169.79 1 134.5 165.50 158.29
2 49.5 69.5 51 2 72 72.5 75
3 118 118 128.5 3 127 124.5 103.5
4 67.5 60 60.5 4 60 69.5 66.5
5 103.5 105.5 116 5 123 137 125.5
6 80.43 80.11 67.71 6 52.93 61.46 76.61
7 114.5 119 135 7 99.5 108 121.5
8 65.29 74.36 50.96 8 79.61 66.11 74.89
9 57.5 67.5 59 9 73 48 53
10 178.5 178.5 169 10 194.5 170.5 191
11 84.5 84.5 70 11 90.5 117 78
12 94.5 97.5 117 12 14S.5 128 142
13 71.18 78.18 90.29 13 99.11 104.11 132.39
14 103.39 100.71 124.61 14 IIS.43 110.68 105.43
15 96.96 80.89 108.29 15 135.25 113.96 116.82
16 131.5 89.5 119.5 16 102 128 128.5
17 100 107 118 17 108.5 100.S 103
18 102 114.5 117.5 18 126 104.S 128
19 102 119.5 124.S 19 121 124.S 126.S
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by ch-art recorder or by the more accurate
computer.
318
Appendix A16 Ultimate Tensile Strain-Minor Axis
Homopolymer Copolymer
Run No Strain I Strain 2 Strain3 Run No Strain I Strain 2 Strain 3
(%) (%) (%) (%) (%) (%)
I 210.2 232.1 211. I 253.4 290.3 240.6
2 144.6 147.6 73.5 2
- -
183.6
3 210.2 218 197 3 182.7 261.7
-
4 168.6
- -
4 121.2 115.8
-
5 245.5 210.2 210.8 5 273.6 198.1 136.6
6 145.8
-
129.1 6
-
141.3 115.4
7 181.7 169.5 205.7 7
-
197.0 248.2
8
-
141.0 167.0 8 128.1 83.3 88.9
9 113.7
-
107.8 9 151.8 111.3 94.15
10 167.2 241.2 155.3 10 72.6 77.3
-
11 151.1 103.3
- II 218.3 154.4 92.84
12 56.6 138.6 146.0 12 187.9 190.8 201.2
13 126.1 131.5 115.3 13 125.4 131.5 137.7
14 157.4 177.1 153.9 14 209.1
- -
15 205.5 149.6 190.4 15 212.4 209.2
-
16 153.7 120.0 176.2 16 213.4 150.4
-
17 148.1 182.2 183.4 17
-
224.9
-
18 107.4
- - 18
-
-
224.5
19 238.6 207.6 216.8 19 216.8 215.3
-
Premature sample failure is indicated by a '-'.
319
Appendix A17 Strain Hardening Rate-Major Axis
Homopolymer Copolymer
Run No SHRl SHR2 SHR3 Run No SHRl SHR2 SHR3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1 41.9 36.7 40.9 1 18.8 21.8 22.0
2 87 102 75 2 87 79.5 94
3 46.5 59.5 60.5 3 54.5 55 51.5
4 99 104.5 110 4 95.5 99 102
5 76.5 81 79 5 57 59.5 59.5
6 83 81.2 73.1 6 77.0 78.5 75.2
7 44.5 54.5 47.5 7 70.5 62 62.5
8 97.6 92.4 80.02 8 87.0 89 90.8
9 99 110.5 105.5 9 95.5 82.5 97
10 26.5 32 33.5 10 36
-
36.5
11 68.5 66 71.5 11 66 66 64.5
12 59 63.5 63.5 12 59 57.5 57.5
13 61.1 50.9 56.5 13 46.7 43.3 41.2
14 54.5 63.7 49.6 14 53.5 63.7 67.2
15 56.6 53.2 58.4 15 53.0 52.6 46.7
16 62 57.6 57.5 16 54 53 48.5
17 59 66 58.5 17 70 68.5 65.9
18 60 68.5 61 18 59 56.5 57.5
19 64.5 71.5 62.5 19 68 64.5 63
The variation in decimal places in the tabulated results reflects the method by
which the data were obtained i.e. by chiut recorder or by the more accurate
computer.
320
Appendix A18 Strain Hardening Rate-Minor Axis
Homopolymer Copolymer
Run No SHRl SHR2 SHR3 Run No SHRl SHR2 SHR3
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
1 -
34.5
- 1 27.2 27.1
-
2 45.6 34.5
-
2 -
-
37.5
3 40.5 36.5
-
3
-
-
-
4 45
- -
4
- - -
5 36
-
42.5 5 33.5
- -
6 37.4
-
37.9 6 -
28.1
-
7
- - -
7 -
- -
8
-
30.3 42.7 8 48.5
-
-
9 36
-
39 9 45 37
-
10
- - -
10
- -
-
11 29.5
- - 11 - - -
12
-
44 50.5 12
- - -
13 -
44.4 54.0 13
-
44.9
-
14 - - - 14 -
-
-
15
- - - 15 -
30.4
-
16
- -
39 16 33
- -
17 35 40.5 37 17 36.5
- -
18
- -
- 18 - -
-
19 35.5
- -
19 - -
-
Premature sample failure is indicated by a '-'.
321
Appendix A19 Creep Strain at 40MPa, 65C after 20 minutes
Homopolymer Copolymer
Run No Strain I Strain 2 Strain3 Run No Strain I Strain 2 Strain 3
(%) (%) (%) (%) (%) (%)
I
-
-
-
I 16.98
- -
2 1.06 1.27 1.44 2 2.09 2.18 1.73
3
- - - 3 - -
-
4 1.07 1.06 0.88 4 1.13 1.45 1.22
5 1.98
- - 5 4.95
-
-
6 1.03 1.78 1.87 6 1.25 1.25 1.23
7 1.84
-
-
7 2.84
- -
8 1.38 1.30
-
8 1.12 1.22
-
9 1.41 0.93 0.79 9 1.64 1.57 1.59
. ....
i.1O
:23.95
.. .i
37
,74 .
25.39 lil
ll

53:61
//57.93 <>L.i
II -
- - II 2.81 2.13
-
12 3.12 3.45 4.7 12 4.57 5.03 4.85
13 1.84 1.85 2.3 13 2.07 2.61 2.96
14 3.30 3.24 3.29 14 9.63 8.44 6.43
IS 1.57 1.79 2.23 IS 2.93 3.97
-
16 1.93 1.87 2.47 16 3.40 4.73 3.33
17 2.35 1.79 2.08 17 2.35 2.87 2.27
18 1.86 I.61 2.01 18 3.04 2.10 3.19
19 2.10 3.02 1.78 19 2.61 2.90 3.08
Premature sample failure is indicated by a 'M'.
322
Appendix A20 Shrinkage Data at Various Temperatures as measured by
TMA
Homopolymer-Major Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
C at 85
0
C at 100
0
C at 1l0
o
C
1 2 1 2 1 2 1 2
1 0.26 0.26 0.73 0.73 0.93 0.93 0.68 0.68
2 -0.06 0.52 1.92 1.96 6.96 6.93 11.72 12.03
3 0.23 0.28 1.15 1.31 2.06 2.28 2.49 2.73
4 0.50 0.49 1.63 1.61 2.94 2.92 3.76 3.73
5 0.27 0.25 1.07 1.00 1.77 1.70 2.24 2.06
6 0.52 0.50 1.97 1.78 5.96 4.60 10.04 7.76
7 0.26 0.19 1.22 1.I1 2.10 2.01 2.48 2.39
8 0.50 0.52 1.64 1.68 2.78 2.74 3.75 3.91
9 0.52 0.54 1.33 1.94 4.16 4.18 6.85 7.05
10 0.15 -0.35 0.12 -0.79 -0.04 -U8 -0.55 -1.92
11 0.35 0.33 1.42 1.37 2.49 2.39 3.03 2.90
12 0.27 0.26 1.18 1.07 2.81 2.82 4.29 4.43
13 0.45 0.48 1.54 1.58 2.69 2.80 3.30 3.50
14 0.47 0.43 1.51 1.47 3.30 3.44 4.65 5.12
15 0.44 0.46 1.52 1.51 2.63 2.57 3.25 3.33
16 0.45 0.48 1.41 1.57 2.36 2.71 2.81 3.31
17 OJI 0.29 1.36 1.33 2.43 2.43 2.97 3.01
18 0.55 0.55 1.44 1.47 2.46 2.55 3.13 3.14
19 0.31 0.82 1.36 1.29 2.49 2.39 3.13 2.99
-
323
Appendix A21 Shrinkage Data at Various Temperatures as measured by
TMA
Homopolymer-Minor Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
C at 85
0
C at 100
0
C at llOoC
1 2 1 2 I 2 1 2
1 0.04 0.10 1.60 2.01 3.96 4.69 5.55 6.34
2 0.28 0.26 1.89 1.77 5.69 5.57 9.19 9.12
3 0.13 0.20 1.06 1.14 2.01 2.10 2.53 2.60
4 0.24 0.26 1.24 1.23 2.31 2.27 2.91 2.85
5 0.19 0.19 1.72 1.84 4.68 5.01 6.99 7.52
6 0.27 0.28 1.70 1.73 5.52 5.54 8.81 8.72
7 0.14 0.15 1.00 1.02 -1.78 1.88 2.28 2.31
8 0.25 0.28 1.26 1.27 2.35 2.32 2.95 2.90
9 0.28 0.26 1.34 1.27 2.71 2.62 4.22 4.02
10 0.02 -0.11 1.05 0.83 1.95 1.97 2.82 2.60
11 0.16 0.21 1.04 1.10 -1.95 2.02 2.41 2.48
12 0.30 0.25 1.90 1.61 4.40 3.86 6.65 5.99
13 0.25 0.26 1.19 1.24 2.20 2.33 2.81 3.00
14 0.22 0.20 1.25 1.15 3.39 3.16 5.33 4.98
15 0.17 0.18 1.15 1.17 2.16 2.28 2.71 2.91
16 0.21 0.24 1.17 1.23 2.23 2.32 2.84 2.97
17 0.23 0.21 1.21 1.23 2.29 2.32 2.90 2.98
18 0.24 0.25 1.23 1.25 2.35 2.37 3.06 3.08
19 0.12 0.12 1.05 1.06 2.02 2.07 2.52 2.61
324
Appendix A22 Shrinkage Data at Various Temperatures as measured by
TMA
Copolymer-Major Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
e at 85
0
e at 1000e at 1I0
o
e
1 2 1 2 1 2 1 2
1 0.23 0.10 0.87 0.67 0.24 -0.08 -0.50 -0.87
2 0.59 0.60 2.02 1.98 4.35 4.19 7.14 6.70
3 0.28 0.36 1.21 1.31 1.59 1.69 1.92 3.46
4 0.59 0.58 1.81 1.73 2.74 2.54 3.56 3.25
5 0.32 0.24 1.01 0.74 0.87 0.62 0.89 -0.20
6 0.60 0.60 2.06 2.07 4.82 4.96 8.38 8.53
7 0.36 0.28 1.33 1.20 1.78 1.68 2.20 2.10
8 0.49 0.57 1.34 1.69 2.51 2.37 3.21 2.95
9 0.60 0.51 1.96 2.10 3.54 3.92 5.74 6.18
10 0.04 0.31 0.50 0.86 0.32 0.53 0.29 0.36
11 0.42 0.36 1.55 1.39 2.15 1.95 2.64 2.42
12 0.30 0.44 1.25 1.76 1.71 2.27 1.97 3.04
13 0.48 0.48 1.42 1.41 1.79 1.80 2.13 2.18
14 0.41 0.50 1.04 1.09 1.37 1.47 1.79 1.83
15 0.50 0.45 1.44 1.35 1.77 1.71 2.10 2.04
16 0.50 0.47 1.42 1.44 1.79 1.82 2.12 2.18
17 0.38 0.37 1.39 1.43 1.88 2.01 2.31 2.53
18 0.63 0.62 1.48 1.42 1.87 1.68 2.23 2.18
19 0.40 0.37 1.42 1.35 1.93 1.83 2.38 2.27
325
Appendix A23 Shrinkage Data at Various Temperatures as measured by
TMA
Copolymer-Minor Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
C at 85
0
C at IOOoC at 110
0
C
I 2 I 2 I 2 I 2
I 0.18 0.21 2.06 1.76 4.32 3.84 6.20 5.83
2 0.34 0.31 1.87 1.83 4.47 4.49 7.00 6.97
3 0.17 0.22 1.11 1.15 1.62 1.66 2.12 2.17
4 0.34 0.34 1.38 1.39 2.01 2.04 2.60 2.65
5 0.24 0.24 2.11 1.90 4.96 4.47 7.38 6.86
6 0.33 0.32 1.64 1.65 4.30 4.31 7.40 7.22
7 0.24 0.22 1.23 1.15 1.78 1.67 2.28 2.17
8 0.32 0.35 1.33 1.35 1.91 1.88 2.46 2.40
9 0.38 0.37 1.51 1.45 2.73 2.56 4.81 4.48
10 0.19 -0.11 0.96 0.27 0.65 1.39 1.98 1.26
11 0.25 0.21 1.20 1.11 1.63 1.55 2.07 1.99
12 0.25 0.22 1.43 1.43 3.59 3.72 5.79 5.84
13 0.27 0.28 1.26 1.32 1.85 1.97 2.44 2.63
14 0.21 0.23 1.18 1.18 3.02 3.00 4.77 4.72
15 0.28 0.27 1.31 1.33 1.96 2.04 2.61 2.73
16 0.25 0.26 1.31 1.27 1.94 1.88 2.57 2.45
17 0.27 0.27 1.39 1.42 2.16 2.21 2.89 2.94
18 0.25 0.25 1.31 1.27 2.03 1.93 2.76 2.57
19 0.27 0.24 1.34
1.26-
2.03 1.95 2.73 2.67
326
Appendix A24 Thermal Analysis and Density Data
H I omopOlymer
Run Beta Onset Beta Peak Crystallinity Crystallinity
No
(0C) (oC) (%)DSC O / ~ ) Density
1 168 182 31 27.2
2 166 182 44 28.9
3 89 110 48 40.5
4 89 107 45 39.8
5 170 185 48 29.2
6 158 175 43 30.3
7 104 132 43 10.2
8 II9 144 40 39.7
9 202 2II 42 31.7
10 90 IIO 38 22.7
11 120 148 36 42.6
12 118 142 41 28.1
13 122 144 45 35.6
14 I 17 143 39 29.2
15 117 142 38 34.2
16 118 143 39 34.4
17
- -
37 35.1
18
-
-
38 32.4
19
-
-
40 34.9
Copolymer
Run Beta Onset Beta Peak Crystallinity Crystallinity
No
(0C) (OC)
(%)DSC (%) Density
1 87 105 29 25.2
2 163 182 37 27.9
3 165 177 45 37.0
4 86 105 43 34.7
5 89 106 40 26.7
6 169 184 37 29.6
7 164 181 46 39.1
8 94 136 42 41.6
9 204 2II 41 30.3
10 86 102 36 29.2
11 122 146 33 42.3
12 II4 146 40 21.6
13 121 146 44 33.5
14 126 146 36 26.3
15 123 147 35 24.3
16 II8 144 37 26.5
17
- -
37 27.8
18
- -
37 27.8
19
-
-
41 28.3
327
Appendix HI Shrinkage In Conventional Experimentation Films-Major
Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
e at 85
0
e at 1000e at 110
0
e
1 0.00 7.09 13.53 16.64
2 -0.05 2.57 4.77 7.42
3 -0.09 2.00 2.83 3.47
4 -0.06 2.50 4.46 6.76
5 -0.16 2.06 3.19 4.18
6 -0.10 2.12 3.24 4.17
7 -0.10 1.99 2.89 3.63
8 -0.11 1.94 2.76 3.42
9 -0.09 1.97 2.74 3.35
10 -0.09 3.84 7.36 9.10
11 -0.14 1.96 2.89 3.70
12 -0.41 1.46 2.04 2.52
13 -0.50 0.79 0.99 1.04
14 -0.33 1.34 1.78 2.19
15 -0.20 1.63 2.09 2.47
16 -0.75 0.50 0.20 0.10
17 -0.38 1.04 1.05 0.96
18 -0.39 0.99 0.24 -0.14
19 -0.34 1.31 1.63 1.91
20 -0.43 1.24 1.46 1.66
21 -0.40 1.45 1.83 2.12
22 -0.39 1.41 1.84 2.21
23 -0.49 1.24 1.51 1.76
24 -0.37 1.61 2.04 2.35
25 2.35 6.61 0.49 -1.47
26 -1.07 0.05 -0.12 -0.23
27 0.11 2.66 3.00 3.14
28 -1.41 -1.11 -1.82 -1.57
29 -0.63 0.72 0.80 1.51
30 -1.07 -0.33 -0.06 0.61
31 -0.09 1.92 2.63 3.21
32 -0.11 1.91 2.64 3.23
33 0.08 6.44 12.24 14.57
34 0.07 8.10 14.66 17.37
35 -0.19 4.80 8.92 11.24
36 -0.79 -0.45 -2.23 -3.38
37 -0.68 1.74 -3.40 -5.82
38 -0.77 0.66 0.80 1.09
328
Appendix B2 Shrinkage In Conventional Experimentation Films-Minor
Axis
Run Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 650C at 85C at 100C at 110C
1 0.41 1.52 4.06 7.48
2 0.42 1.54 2.39 3.34
3 0.39 1.442 1.98 2.52
4 0.39 1.45 2.22 3.10
5 0.39 1.42 2.03 2.66
6 0.41 1.44 2.06 2.68
7 0.38 1.40 1.96 2.51
8 0.40 1.43 2.05 2.56
9 0.40 1.45 2.03 2.58
10 0.38 1.79 4.62 7.90
11 0.40 1.44 2.07 2.70
12 0.38 1.42 1.96 2.47
13 0.31 1.63 3.68 6.02
14 0.35 1.44 2.16 2.83
15 0.36 1.38 1.88 2.36
16 0.27 1.76 3.85 6.16
17 0.29 1.35 2.02 2.69
18 0.26 1.24 2.93 4.66
19 0.31 1.31 1.89 2.45
20 0.30 1.29 1.78 2.26
21 0.31 1.27 1.73 2.20
22 0.30 1.33 1.86 2.36
23 0.33 1.35 1.85 2.34
24 0.29 1.16 1.51 1.92
25 0.08 0.94 2.76 6.03
26 0.20 1.28 0.43 0.66
27 0.20 0.58 0.63 0.92
28 0.14
-
- -
29 0.09
-
- -
30 0.09
- -
-
31 0.39 1.45 2.04 2.58
32 0.38 1.38 1.87 2.35
33 0.38 1.66 4.68 8.10
34 0.86 1.92 5.09 8.97
. 35
0.34 1.36 3.64 6.53
36 -0.02 0.45 1.91 3.85
37 -0.06 -0.39 0.90 3.87
38 0.01
- -
2.75
329
Appendix B3 Thermal Analysis Data for Conventional Experimentation
Films
Run Beta Onset Beta Peak Crystallinity
No
(DC) (DC)
(%)
1
-
-
33
2 133 142 38
3 150 224 45
4
- -
34
5 117 160 39
6 157 173 38
7 177 193 38
8
-
205 34
9
-
220 40
10
- -
36
II 148 161 37
12
-
225 38
13 109 -
29
14 146 167 36
15
-
222 41
16
-
-
30
17 117 141 36
18
- -
31
19 151 166 38
20 153 171 32
21 180 197 38
22 180 200 37
23
- -
38
24 112 215 36
25
-
-
14
26
- -
29
27 109 150 25
28
- -
0
29 124 -
3
30 123
-
4
31
-
219 43
32
-
223 42
33
-
-
31
34 109
-
33
35 109
-
31
36 100
-
27
37 100
-
19
38 138
-
1
330
Appendix B4 Crystallinity Values From Refractive Index Measurements
Run NI N2 N3 Density Crystallinity
No (%)
I 1.5910 1.6544 1.6170 1.379 35
2 1.5152 1.6646 1.6236 1.387 42
3 1.5127 1.6843 1.6234 1.397 51
4 1.5193 1.6619 1.6174 1.384 39
5 1.5147 1.6619 1.6219 1.836 39
6 1.5168 1.6707 1.6278 1.394 48
7 1.5124 1.6744 1.6260 1.393 47
8 1.5123 1.6772 1.6272 1.395 49
9 1.5111 1.6819 1.6248 1.396 49
10 1.5249 1.6483 1.6148 1.377 34
11 1.5168 1.6618 1.6212 1.384 40
12 1.5178 1.6771 1.6245 1.397 50
13 1.5404 1.6486 1.6030 1.380 36
14 1.5248 1.6612 1.6167 1.386 42
15 1.5229 1.6736 1.6279 1.400 53
16 1.5408 1.6468 1.6019 1.378 35
17 1.5398 1.6516 1.6104 1.386 41
18 1.5539 1.6403 1.6013 1.382 38
19 1.5363 1.6618 1.6130 1.391 46
20 1.5458 1.6562 1.6178 1.397 50
21 1.5366 1.6638 1.6117 1.392 46
22 1.5342 1.6677 1.6118 1.393 47
23 1.5328 1.6630 1.6152 1.391 46
24 1.5321 1.6699 1.6146 1.395 49
25
-
- - - -
26
-
- - - -
27
- - - - -
28
-
- - - -
29
- - - - -
30
- - - - -
31 1.5121 1.6831 1.6208 1.394 48
32 1.5102 1.6847 1.6233 1.396 49
33 1.5208 1.6574 1.6158 1.381 37
34 1.5192 1.6539 1.6196 1.380 36
35 1.5261 1.6499 1.6143 1.379 35
36
- - - -
-
37
- - - -
-
38
-
- - - -
331
Appendix B5 Optical Transmission Data for Conventional
Experimentation Films
Run Total Sample Absorbtion Absorbtion
No Transmission Thickness per mm
(%) (I1m) (%)
1 87.6 44 0.032 0.072
2 87.1 44 0.037 0.084
3 85.3 48 0.056 0.117
4 87.0 50 0.038 0.076
5 86.2 48 0.047 0.097
6 86.6 48 0.042 0.088
7 86.7 50 0.Q41 0.083
8 86.6 58 0.042 0.073
9 87.0 52 0.038 0.073
10 86.9 56 0.039 0.070
11 87.0 52 0.038 0.073
12 86.1 60 0.048 0.079
13 86.6 58 0.042 0.073
14 86.4 70 0.044 0.063
15 85.9 52 0.050 0.096
16 86.0 102 0.049 0.048
17 84.5 94 0.064 0.069
18 83.5 96 0.075 0.078
19 83.7 90 0.073 0.081
20 83.8 102 0.072 0.070
21 82.8 112 0.082 0.073
22 84.3 92 0.066 0.072
23 84.0 106 0.070 0.066
24 84.0 124 0.070 0.056
25 80.0 182 0.112 0.061
26 73.4 152 0.181 0.119
27 60.0 178 0.322 0.181
28 88.1 516 0.027 0.005
29 39.6 508 0.536 0.106
30 37.8 552 0.555 0.101
31 87.1 48 0.037 0.077
32 86.9 48 0.039 0.082
33 86.7 46 0.041 0.090
34 88.8 50 0.019 0.038
35 87.6 54 0.032 0.059
36 81.6 lOO 0.095 0.095
37 83.5 154 0.075 0.049
38 86.0 506 0.049 0.010
Absorbtion
per mm
(%)
0.068
0.045
0.010
0.040
0.029
0.027
0.060
0.078
0.096
0.034
0.115
0.030
0.034
0.019
0.027
0.069
0.039
0.069
0.024
0.018
0.055
0.011
0.054
0.051
0.009
0.047
0.072
0.005
0.033
0.078
0.031
0.033
0.054
0.110
0.019
0.023
0.046
0.012
The % transmission data in the last column were obtained via micro-
photometry. The rest of the transmission data were obtained using a
Hazeguard XL211.
332
Appendix B6 Birefringence Of Conventional Experimentation Films
Run
(eJ)
~ ) ~ ) No
1 0.0374 0.1354 0.0980
2 0.0410 0.1494 0.1084
3 0.0609 0.1716 0.1107
4 0.0445 0.1426 0.0981
5 0.0431 0.1503 0.1072
6 0.0429 0.1539 0.1110
7 0.0484 0.1620 0.1136
8 0.0500 0.1649 0.1149
9 0.0571 0.1708 0.1 137
10 0.0335 0.1234 0.0899
11 0.0406 0.1450 0.1044
12 0.0526 0.1593 0.1067
13 0.0456 0.1082 0.0626
14 0.0445 0.1364 0.0919
15 0.0457 0.1507 0.1050
16 0.0449 0.1060 0.0611
17 0.0412 0.1118 0.0706
18 0.0422 0.0916 0.0494
19 0.0511 0.1291 0.0780
20 0.0384 0.1 104 0.0720
21 0.0521 0.1272 0.0751
22 0.0559 0.1335 0.0776
23 0.0511 0.1335 0.0824
24 0.0553 0.1378 0.0825
25 0.0081 0.0160 0.0079
26 0.0228 0.0353 0.0126
27 0.0398 0.1042 0.0644
28 0.0003 0.0006 0.0002
29 0.0002 0.0003 0.0001
30 0.0002 0.0003 0.0000
31 0.0623 0.1710 0.1087
32 0.0614 0.1745 0.1131
33 0.0416 0.1366 0.0950
34 0.0343 0.1347 0.1004
35 0.0356 0.1238 0.0882
36 0.0228 0.00831 0.0603
37 0.0098 0.0209 0.0111
38 0.0002 0.0003 0.0000
333
Appendix Cl Shrinkage In Heat Set Bottles
Hoop Axis
Bottle Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
C at 85C at IOOOC at ll00C
1 1.06 0.95 6.87 11.18
2 0.73 1.41 2.67 4.64
3 0.86 2.66 5.42 9.37
4 1.33 2.62 3.85 5.56
5 0.49 1.25 1.94 2.76
6 0.83 1.83 2.60 3.41
7 0.63 0.73 0.61 0.95
8 0.93 1.81 2.21 2.83
9 0.98 1.66 1.72 2.03
10 0.08 -0.71 -1.11 -0.85
11 1.04 2.53 3.44 3.98
Axial Axis
Bottle Shrinkage (%) Shrinkage (%) Shrinkage (%) Shrinkage (%)
No at 65
0
C at 85C at IOOOC at ll00C
1 0.40 5.84 16.20 22.97
2 0.78 3.04 6.26 11.06
3 0.95 3.88 7.98 11.74
4 0.66 2.50 4.34 6.20
5 0.53 1.89 2.66 3.41
6 0.68 1.99 2.70 3.41
7 0.52 1.67 2.13 2.61
8 0.64 1.90 2.41 2.93
9 0.47 1.51 1.93 2.39
10 0.50 1.26 1.43 1.78
11 0.35 1.37 2.25 3.32
334
Appendix C2 Crystallinity in Heat Set Bottles by Thermal Analysis and
Density
Run ~ P e a k Crystallinity (%) Crystallinity (%)
No (oC) DSC Density
1
-
26 22.6
2 117 37 31.7
3 132 38 31.8
4 147 32 36.8
5 172 34 36.1
6 174 38 38.7
7 194 34 42.9
8 205 37 45.2
9 208 38 44.6
10 220 37 49.0
11 150 41 35.6
335
Appendix C3 Orientation Functions For Heat Set Bottles
Bottle Bulk Crystalline Phase Amorphous Phase
No 1018cm-
1
973cm-
1
1578cm-
1
F1B
F,U F1B F,B F1B
F,B
1 0.286 0.00 0.316 0.283 0.128 0.175
2 0.311 0.420 0.291 0.457 0.267 0.232
3 0.381 0.215 0.481 0.233 0.172 0.180
4 0.443 0.277 0.459 0.318 0.363 0.157
5 0.322 0.332 0.300 0.418 0.235 0.198
6 0.398 0.271 0.391 0.364 0.259 0.177
7 0.269 0.368 0.821 0.189 0.192 0.211
8 0.299 0.413 0.425 0.389 0.180 0.211
9
- -
0.470 0.404 0.279 0.188
10
- -
0.313 0.489 0.143 0.263
11
- -
0.301 0.332 0.163 0.184
Appendix C4 Dichroic Ratios For Heat Set Bottles
Bottle Bulk Crystalline Phase Amorphous Phase
No 1018cm-
1
973cm-
1
1578cm-
1
1 -0.151 0.121 0.081
2 0.135 0.183 0.087
3 0.007 0.051 0.126
4
- -
-
5 0.151 0.200 0.139
6 0.121 0.139 0.087
7 0.180 0.108 0.057
8
- - -
9
-
0.217 0.196
10
-
0.305 0.269
11
-
0.161
-
336

Anda mungkin juga menyukai