THE

(«3"

eye.

MICHI IGANS

H lHHHIUNIHI’“Iii!!! Willi1!!!!”in

301686 0789

i!

 

 

 

 

 

This is to certify that the

dissertation entitled

THE IMPROVEMENT OF THE HYGROTHERMAL AND
MECHANICAL PROPERTIES OF BISMALEIMIDE AND K3B/IM7
CARBON FIBER COMPOSITES THROUGH A SYSTEMATIC STUDY
OF THE INTERPHASE

presented by

Mark S. Wilenski

has been accepted towards fulfillment
of the requirements for
Material Science

Ph . D . degree in and Mechanics

Lawrence T. Drzal, Ph.D.

Major professor

DatelmgwAL/QWW {W

MSU is an Affirmatiw Action/Equal Opportunity Institution 0—12771

 

LIBRARY

Michigan State ‘
University

 

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINB return on or before date due.

 

MTE DUE MTE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 cICIRC/Dahoupes-p.“

THE IMPROVEMENT OF THE HY GROTHERMAL AND
NIECHANICAL PROPERTIES OF BISMALEIMIDE AND K3B/IM7
CARBON FIBER COMPOSITES THROUGH A SYSTEMATIC STUDY
OF THE INTERPHASE

By

Mark S. Wilenski

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Materials Science and Mechanics

1997

ABSTRACT

THE IMPROVEMENT OF THE HYGROTHERMAL AND
MECHANICAL PROPERTIES OF BISMALEIMIDE AND K3B/IM7
CARBON FIBER COMPOSITES THROUGH A SYSTEMATIC STUDY
OF THE INTERPHASE

By

Mark S. Wilenski

Advanced aircraft travel at speeds in excess of mach 2 result in external skin temperatures
ranging from -60°C to +190°C with extreme heating and cooling rates in the presence of
moisture. This combination of environments presents a challenge to current materials.
1M7 carbon fiber composites made with either bismaleimide (a thermoset) or K3B (a
thermoplastic polyimide) resin are being considered for this use. The overall objective of
this research was to develop a greater understanding of both the macroscopic response of
these materials to hygrothermal and thermal spike environments. as well as the molecular

level interactions which affect the ability of the resins to bond to the fiber.

Hygrothermal testing of BMI/IM7 composites showed that resin properties are reduced
by the presence of moisture while composite properties are virtually unaffected.
Exposure of composites to a relatively small number of thermal spikes between room
temperature and 250°C was shown to have little effect on BMl/IM7 composites, but

causes excessive blistering of wet K3B/IM7 composites.

A macroscopic microcracking phenomenon is observed in cross-ply laminates of the
BMI/IM7 system. This was shown to be due to the generation of thermal residual
stresses formed during cool down from a "stress-free temperature." Both finite element
and analytical models were utilized to analyze the causes of this microcracking with a
focus on methods for minimizing the resultant stresses. Several methods with potential

were identified and suggestions for future work are made.

Microscopically, tests of the bond strength between both the BMI and K3B resins with
the unsized 1M7 fiber indicated that the adhesion was low. Experimentation on the
BMI/IM7 system was able to identify that the only active adhesion promoting mechanism
was mechanical interlocking and not chemical bonds between functionalities on the fiber
surface and the resin. Various methods for improving the adhesion in the BMI/IM7

system were attempted and some encouraging results were found.

Finally, a set of methodologies for designing an interphase material which will provide an
optimum bond between a fiber and resin is presented. These methodologies have been
incorporated into an artificial intelligence framework and can be used to tailor composite

properties through the addition of an interphase with specific properties.

ACKNOWLEDGMENTS

This dissertation represents three years of contemplation, sweat, and toil. While I will
happily take the blame for everything herein which deserves blame, I must drag a few
others into the picture to accept the proverbial award of gratitude (in whatever shape

manifests itself).

Throughout my encounter with life, I have been encouraged along the paths of my
choosing by my loving family. I have happily ridden along being propelled by their
unique forms of support (wavelike wouldn’t you know) and have found that the summer
dive trips, motor-home sized graduation cards, Christmas Key’s trips, and infinite other
Wilenski things are both who and why I am. I must thank ’Doc’ Nelson for he was the
one who told me in high school that not only could I get a PhD, but that I would get one

whether I knew it or not.

Then there are the folks more directly related to the pages which follow. To my advisor
Dr. Lawrence Drzal I owe many thanks for the freedom with which he allowed me to
work and the many times he helped pull my befuddled brain out of the depths of foolish
misconception. Thanks to Dr. E. Eugene Shin and Dr. Roger Morgan for organizing the
project of which this was but a part, and for all of the help and discussions along the way.
I am very grateful to the other members of my committee who each was chosen for a

iv

particular skill which I knew I would require. Sincere gratitude to: Dr. Ron Averill for
mechanics help and providing me with the MicMec program which saved me from tons
of finite element frustrations; Dr. Greg Baker for doing his best to teach me the many
aspects of chemistry which elude me (I must state that I am personally to blame for any
chemistry related shortcomings herein); and finally Dr. Jim Lucas for providing insight
into moisture related effects. Thanks to Lei Chen and Dr. Gilliland for the statistical
derivation in Appendix C. I am indebted to the Air Force Office of Scientific Research

and people of the United States of America for funding this research.

There were also many people who helped deflect my general Brownian motion toward
appropriate targets. They were Dr. Andre Lee, Mike Rich, Brian Rook, Bob Jurek, Dr.
Cara Weitzsacker, Jason Lincoln, Dr. Sanjay Padaki, Dr. Murty Vyakarnam, Dr. Rik ter
Veen, and soon to be Drs JinKyu Choi, and Ed. Drown (Note there is absolutely no order
to this list!!!). Also, to the wonderful students who endured working with/for me: Dave
Knop (who certainly did more than his fair share of this research), Amanda Centeck,

Mark Niesen, and Shawn Corbin.

Finally great thanks to Brett Wilson (for the liquid beverages, volleyball, and general
traipsing); Michelle Hignite (for Europe, 'real' conversations-whatever they may be, and
similar traipsing); Wendy Reffeor (for plays, wine, beaches - well traipsing); and
finally Cara (see above) for the many meals on short notice, hours of being a concept
bored (sounding board), and all other forms of the mental and physical traipsing and

tripping that were done.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. xii
LIST OF FIGURES .......................................................................................................... xiv
1. INTRODUCTION ........................................................................................................... 1
2. MATERIAL CHARACTERIZATION ........................................................................... 5
2.1 Hercules 1M7 Fiber .................................................................................................... 5
2.2 K3B Resin Description .............................................................................................. 6
2.3 BMI Resin Curing Reaction ...................................................................................... 7
2.3.1 Cure Reaction ..................................................................................................... 9
2.3.2 Reaction Rate Measurement: Isothermal DSC Profiles .......................... 16
2.3.3 Composite Fabrication Issues ........................................................................... 19
2.3.3.1 Resin Viscosity During Prepregging ......................................................... 19
2.3.3.2 Resin Viscosity During Autoclave Curing ................................................ 21
2.3.3.3 The Effect of Resin Batches on the Viscosity Profile ............................... 22
2.4 Material Characterization Conclusions ................................................................... 24
3. EXPERIMENTAL DETAILS ....................................................................................... 25
3.1 Overall Specimen Fabrication ................................................................................. 25
3.1.1 BMI Resin ........................................................................................................ 26
3.1.2 BMI/1M7 Composites ....................................................................................... 27
3.1.2.1 BMI/1M7 Prepreg Manufacture ................................................................. 27
3.1.2.2 BMI Composite Cures ............................................................................... 28
3.1.2.3 BMI Composite Post-Cures ....................................................................... 31
3.1.3 K3B/IM7 Panel History 32
3.1.4 K3B Resin Plaque Manufacture ....................................................................... 32
3.2 Microcracking Characterization Experiments ......................................................... 33
3.2.1 Determination Of The Stress Free Temperature .............................................. 33
3.2.1.] Experimental Procedure ............................................................................ 34
3.2.2 Determination Of Microsc0pic (Fiber Level) Cracking ................................... 38
3.2.3 Determination Of Crack Density And Crack Width ........................................ 39
3.2.3.1 The Effect of Vf and Post-Cures on Microcracks using Optical
Microscopy ............................................................................................................ 39

vi

3.2.3.1 The Effect of Vf and Post-Cures on Microcracks using Optical

Microscopy ............................................................................................................ 39
3.2.4 Identification Of The Effect Of Laminate Features On Crack Propagation ..... 40
3.2.4.1 ESEM Vertical Jaw Design ....................................................................... 40

3.3 Hygrothermal Study Experiments ............................................................................ 45
3.3.1 Hygrothermal Moisture Conditioning ............................................................... 46
3.3.1.1 BMI Composite Moisture Conditioning .................................................... 46
3.3.1.2 BMI Resin Moisture Conditioning ............................................................ 47
3.3.2 Resin F lexure Testing ....................................................................................... 48
3.3.3 Hygrothermal Composite F lexure Testing ........................................................ 51
3.3.4 Isothermal Dma Testing .................................................................................... 59
3.3.5 Interfacial Adhesion Testing ............................................................................. 62
3.3.5.1 Hygrothermal Materials & Conditioning ................................................... 62
3.3.5.2 Specimen Testing ....................................................................................... 62
3.3.5.3 Data Reduction ........................................................................................... 64
3.3.6 The Effect Of Cure Progression On The Physical Properties Of Bmi Resin....65
3.3.7 Moisture Diffusion Study ................................................................................. 69
3.4 Combined Hygrothermal And Thermal Spike Experiments .................................... 72
3.4.1 Iosipescu Shear Testing .................................................................................... 72
3.4.1.1 Specimen Details ....................................................................................... 72
3.4.1.2 Testing and Analysis .................................................................................. 73
3.4.2 Transient Dma Testing ...................................................................................... 76
3.5 Thermal Spike Experiments ..................................................................................... 82
3.5.1 Thermal Spike Conditioning ............................................................................. 83
3.5.2 Resin F lexure Testing ....................................................................................... 85
3.5.3 Resin Tensile Testing ........................................................................................ 86
3.5.4 Thermal Spike Composite Flexure Testing ...................................................... 87
3.5.5 Open-Hole Compression Testing ...................................................................... 89

4. MICROCRACKING OF CROSS-PLY BMI/1M7 COMPOSITES .............................. 91
4.1 Introduction .............................................................................................................. 91
4.2 Cure Induced Microcracking ................................................................................... 91
4.2.1 Review of the Microscopic Microcracking Mechanism ................................... 92
4.2.2 Review of the Macroscopic Microcracking Mechanism .................................. 93
4.3 Finite Element Modeling ......................................................................................... 93
4.3.1 Introduction ....................................................................................................... 93
4.3.1.1 Assumptions ............................................................................................... 95
4.3.1.2 Model Particulars ....................................................................................... 95
4.3.2 Results ............................................................................................................... 98
4.4 Analytical Modeling Using "MicMec" .................................................................. 101
4.4.1 Introduction to MicMec .................................................................................. 101
4.4.1.1 MicMec Assumptions .............................................................................. 101
4.4.1.2 Model Geometry ...................................................................................... 102
4.4.1.3 Generation of a Graphical User Interface for MicMec ............................ 105
4.4.1.4 The Addition of Laminate Calculations in the GUI ................................. 105

vii

4.4.1.5 Derivation of CLT Equations Used ......................................................... 106

4.4.2 MicMec Modeling Test Plan .......................................................................... 110
4.4.2.1 Objectives of Modeling Work ................................................................. 110
4.4.2.2 Analysis Approach ................................................................................... 112

4.4.3 Micromechanical Results ................................................................................ 114
4.4.3.1 Determination of Radial Stress Distribution ............................................ 115

4.4.3.2 The Effect of Fiber Volume Fraction on Residual Stresses and Strains..l 19
4.4.3.3 The Effect of Stress Free Temperature on Residual Stresses and

Strains ...................................................................................................... 122
4.4.3.4 The Effect of Interphase Thickness on Residual Stresses and Strains ..... 125
4.4.3.5 The Effect of Interphase Modulus on Residual Stresses and Strains ....... 129
4.4.4 Discussion of the Macroscopic Modeling Results .......................................... 132
4.4.5 MicMec Modeling Conclusions ...................................................................... 135
4.5 Experimental Verification ...................................................................................... 137
4.5.1 Determination of the Stress Free Temperature ............................................... 137
4.5.2 Determination of Microscopic (Fiber Level) Cracking .................................. 138
4.5.3 Determination of Crack Density and Crack Width ......................................... 139
4.5.3.1 The Effect of Vf and Post-Cures on Microcracks using Optical
Microscopy .......................................................................................................... 1 40
4.5.3.1.] Conclusions ....................................................................................... 144
4.5.3.2 Effect of Vf and Post-Cure on Microcracks using the
Environmental SEM ................................................................................. 144
4.5.3.2.1 Conclusions ....................................................................................... 150
4.6 General ESEM Crack Observations: ...................................................................... 151
4.7 Identification of Microcrack Initiation During the Standard Cure Cycle .............. 154
4.8 Identification of the Effect of Laminate Features on Crack Propagation .............. 157
4.9 Failure Analysis of a Mechanically Fractured K3B/IM7 Specimen ...................... 158
4.10 General Microcracking Conclusions .................................................................... 161
5. HYGROTHERMAL PROPERTY TESTING ............................................................. 163
5.1 Introduction ............................................................................................................ 163
5.1.1 Moisture Degradation ..................................................................................... 164
5.1.1.] General Moisture Degradation ................................................................. 164
5.1.1.2 BMI Moisture Degradation ...................................................................... 165
5.1.1.3 K3B Moisture Degradation ...................................................................... 165
5.1.2 BMI Cure Advancement Effects ..................................................................... 166
5.2 Test Design ............................................................................................................ 168
5.3 Neat BMI Resin Testing ........................................................................................ 170
5.3.1 Resin Testing Conclusions .............................................................................. 172
5.4 Moisture Diffusion Study ...................................................................................... 172
5.4.1 The Effect of the Extent of Cure on D and M¥ for BMI Resin ...................... 173
5.4.2 The Effect of Microcracking on the Diffusivity of BMI Composites ............. 174
5.4.3 Effect of Microcracks on the Ultimate Moisture Content of
BMI Composites ............................................................................................. 175
5.4.4 Effect of Fibers on the Ultimate Moisture Content of BMI Resin ................. 175

viii

5.4.5 Microcrack Formation in Composite Specimens During Saturation .............. 175

5.4.6 K3B Resin and Composite Data ..................................................................... 177
5.5 Interfacial Adhesion Testing .................................................................................. 177
5.6 Mechanical Test Results And Discussion .............................................................. 179

5.6.1 Review of the Experimental Design ............................................................... 182

5.6.2 Sample Results Calculation ............................................................................ 184

5.6.3 Mechanical Testing Results ............................................................................ 186

5.6.3.1 The Effect of Cure on BMI/1M7 Properties ............................................. 186
5.6.3.2 The Effect of Moisture on BMI/1M7 Properties ...................................... 189
5.6.3.3 The Effect of Volume Fraction on BMI/1M7 Properties ......................... 191
5.6.4 Dynamic Mechanical Testing Results ............................................................ 193
5.6.4.1 The Effect of Cure on BMI/1M7 Rheological Properties ........................ 193
5.6.4.2 The Effect of Moisture on BMI/1M7 Properties ...................................... 196
5.6.4.3 The Effect of Volume Fraction on BMI/1M7 Properties ......................... 198

5.7 Hygrothermal Conclusions .................................................................................... 201

5.7.1 Moisture Absorption Behavior ....................................................................... 202

5.7.2 The Effect of Increased Cure .......................................................................... 202

5.7.3 The Effect of Moisture Presence ..................................................................... 203

5.7.4 The Effect of Increased Fiber Volume Fraction ............................................. 204

6. THERMAL SPIKE RESPONSE OF BMI AND K3B RESINS

AND COMPOSITES ................................................................................................... 205
6.1 Introduction ............................................................................................................ 205
6.2 Experimental Summaries ....................................................................................... 206
6.2.1 Test Design ..................................................................................................... 206
6.2.1.1 Post-Conditioning Observations .............................................................. 207

6.3 Combined Results And Discussion ........................................................................ 212
6.3.1 Review of Experimental Design ..................................................................... 215
6.4 Effect Of Moisture Exposure On Final Properties ................................................. 217
6.5 Effect Of Thermal Spikes On Final Properties ...................................................... 221
6.6 Thermal Spike Conclusions ................................................................................... 226
7. IMPROVEMENT OF BMI/1M7 PROPERTIES ......................................................... 228
7.1 Introduction ............................................................................................................ 228
7.1.1 Factors Affecting the Interfacial Shear Strength (IFSS) ................................. 229
7.1.2 Adhesion Improvement In BMI/1M7 And Related Systems .......................... 230
7.1.2.1 Surface Energy Changes .......................................................................... 230
7.1.2.2 Surface Morphology Changes .................................................................. 232
7.1.2.3 Surface Chemistry Changes ..................................................................... 233
7.1.2.4 Resin Chemistry Changes ........................................................................ 235
7.1.2.5 Fiber Sizing .............................................................................................. 237

7.2 Treatment Of Im7 Fibers For Increased Adhesion ................................................ 240

ix

7.2.1 Fiber Surface Treatment .................................................................................. 240

7.2.1.1 Chemical Bonding of K3B monomers During Processing ...................... 240
7.2.1.2 TPD showing chemical species Desorption ............................................. 241
7.2.1.3 Commercial Treatment ............................................................................ 243
7.2.1.4 Air Plasma Treatment .............................................................................. 244
7.2.1.5 XPS Analysis of Treated Fiber Chemistries ............................................ 245
7.2.1.6 Surface Treatment Conclusions ............................................................... 246

7.3 Sizing Of Carbon Fibers ........................................................................................ 247
7.3.1 Use of Thermoplastic Sizings ......................................................................... 247
7.3.1.1 Sizing Chemistries and Properties ........................................................... 248
7.3.1.2 Sizing Method .......................................................................................... 250
7.3.1.3 Sizing Results .......................................................................................... 250
7.3.2 Use of a Compliant Sizing .............................................................................. 256
7.3.2.1 Sizing Method .......................................................................................... 256
7.3.2.2 RTV Sizing Results ................................................................................. 257
7.3.2.3 Composite Preparation ............................................................................. 259
7.3.2.4 Composite Morphology ........................................................................... 261

7.4 Measurement Of Adhesion Levels For Treated And Sized Fiber Composites ...... 263
7.4.1 Unidirectional Specimen Manufacture ........................................................... 264
7.4.2 ITS Testing ...................................................................................................... 264
7.4.3 Mode I Failure Analysis .................................................................................. 267
7.5 Improvement Conclusions ..................................................................................... 277
8. THEORETICAL INTERPHASE DESIGN ................................................................. 279
8.1 Introduction ............................................................................................................ 279
8.2 Definitions .............................................................................................................. 280
8.3 Theoretical Approach - Interphase Design ............................................................ 282
8.3.1 Polymer/Fiber Adhesion ................................................................................. 284
8.3.2 Polymer/Polymer Adhesion ............................................................................ 286
8.4 Implementation ...................................................................................................... 288
8.4.1 Functional Group Reactions ........................................................................... 290
8.4.2 Miscibility Data .............................................................................................. 291
8.4.3 Fiber Surface Treatments ................................................................................ 291
8.4.4 Data Contained Within the Tables. ................................................................. 291
8.5 Future work ............................................................................................................ 294
8.6 An Example and Conclusions ................................................................................ 296
9. CONCLUSIONS AND FUTURE RECOMMENDATIONS ...................................... 299
9.1 Microcracking Conclusions ................................................................................... 299
9.2 Physical Property Conclusions .............................................................................. 300
9.3 Interfacial/Adhesion Conclusions .......................................................................... 301
9.4 Recommendations for Future Studies .................................................................... 301

APPENDIX A - TABULAR BMI RESIN TEST DATA ................................................ 304
APPENDIX B - MOISTURE ABSORPTION STUDY DATA ...................................... 305
APPENDIX C - DERIVATION FOR THE ERROR IN PERCENT

CHANGE VALUES ............................................................................ 306

LIST OF REFERENCES ................................................................................................. 310

xi

LIST OF TABLES

Table 3-1 - BMI Resin Panel Post-Cures ........................................................................... 26
Table 3-2 - BMI Composite Panel Descriptions ................................................................ 31
Table 3-3 - BMI Composite Panel Post-Cure Cycles ........................................................ 32
Table 3-4 Stress Free Temperatures of Various Cure & Post-Cure Cycles ....................... 37
Table 3-5 Post-Cure Cycles Used for Microscopic Cracking Experiment ....................... 38
Table 3-6 Post-Cure Cycles Used for Microscopic Cracking Experiment ....................... 40
Table 3-7 Hygrothermal Testing Scheme ........................................................................ 46
Table 3-8 Hygrothermal Resin Flexural Results .............................................................. 50
Table 3-9 Hygrothermal BMI Composite Flexure Data - 62% Vf ................................... 53
Table 3-10 Hygrothermal BMI Composite Flexure Data - 35% Vf .................................. 53
Table 3-11 Hygrothermal BMI Composite DMA Data — 1 Hz ......................................... 61
Table 3-12 Hygrothermal BMI Resin DMA Data - 1 Hz .................................................. 61
Table 3-13 ITS Results for Hygrothermal Testing ........................................................... 65
Table 3-14 ITS Results for Improvement Attempts ......................................................... 65
Table 3-15 Neat Resin Post-Cure Cycles .......................................................................... 66
Table 3-16 Summary Data for Moisture Absorption Study .............................................. 72
Table 3-17 Hygrothermal BMI Composite Shear Data - 62% Vf ..................................... 75
Table 3-18 Hygrothermal BMI Composite Shear Data - 35% Vf ..................................... 75
Table 3-19 Thermal Spiked BMI Composite Shear Data .................................................. 75
Table 3-20 Thermal Spiked K3B Composite Shear Data .................................................. 75
Table 3-21 Hygrothermal BMI Composite DMA Data - Transient ................................... 79
Table 3-22 Hygrothermal BMI Resin DMA Data - Transient ........................................... 79
Table 3-23 Thermal Spike K3B/1M7 DMA Data - Transient ........................................... 80
Table 3-24 Thermal Spike BMI/1M7 DMA Data - Transient ............................................ 80
Table 3-25 Thermal Spike K3B Resin DMA Data - Transient .......................................... 81
Table 3-26 Thermal Spike BMI Resin DMA Data - Transient .......................................... 81
Table 3-27 Thermal Spike Testing Scheme ...................................................................... 82
Table 3-28 Thermal Spike Resin Flexure Results ............................................................ 85
Table 3-29 Thermal Spike Resin Tensile Results ............................................................. 87
Table 3-30 Thermal Spike BMI Composite F lexure Data ................................................ 88
Table 3-31 Thermal Spike K3B Composite Flexure Data ................................................. 89
Table 3-32 Thermal Spike BMI Open-Hole Compression Data ....................................... 90
Table 3-33 Thermal Spike K3B Open-Hole Compression Data ........................................ 90

xii

Table 4-1 Material Properties Used in the Marc FEM Calculations .................................. 98

Table 4-2 MicMec Test Conditions ................................................................................ 111
Table 4-3 Material Properties Used in MicMec Analysis ............................................... 112
Table 4-4 Stress Free Temperatures of Various Cure & Post-Cure Cycles ..................... 138
Table 4-5 Post-Cure Cycles Used for Microscopic Cracking Experiment ..................... 139
Table 4-6 Post-Cure Cycles Used for the Microscopic Cracking Experiment ............... 140
Table 5-1 Hygrothermal Testing Scheme ....................................................................... 169
Table 5-2 Summary of Resin Testing Results ................................................................ 172
Table 5-3 Summary Data for Moisture Absorption Study .............................................. 173
Table 5-4 Factor Levels Used in Hygrothermal Testing ................................................. 182
Table 6-1 Thermal Spike Testing Scheme ...................................................................... 206
Table 6-2 Factor Levels Used in Thermal Spike Testing ............................................... 215
Table 7-1 Plasma Chamber Experimental Parameters .................................................... 244
Table 7-2 Surface Composition for All Treated Fibers .................................................. 246
Table 7-3 Physical Properties of Sizing Materials .......................................................... 248
Table 7-4 ITS Results for Improvement Attempts ......................................................... 265
Table 7-5 Sizing Improvement Summary ....................................................................... 267
Table 8-1 Functional Group Reactions Used in COMADE ........................................... 292
Table 8-2 Resin Solubility Parameters Used in COMADE ............................................ 293
Table 8-3 Functional Group Surface Free Energies ........................................................ 294
Table A-1 Three Point Flexural Data for BMI Resin ..................................................... 304
Table A-2 Tensile Data for BMI Resin ........................................................................... 304
Table B-1 Summary Data from 100°C Moisture Absorption Experiment ..................... 305
Table B-2 Summary Data from 70°C Moisture Absorption Experiment ....................... 305
Table C-l Factor Level Designations for the Hygrothermal Experiments ..................... 306
Table C-2 Factor Level Designations for the Thermal Spike Experiments ..................... 309
xiii

LIST OF FIGURES

Figure 2-1 - Simplistic Representation of K3B Polymerization Chemistry ....................... 7
Figure 2-2 The Chemical Structure of the Matrimid® 5292 Components ........................ 8
Figure 2-3 The "ene" Reaction ........................................................................................... 9
Figure 2—4 Hom0polymerization of the Allyl Groups ...................................................... 10
Figure 2-5 Homopolymerization of the Maleimide Groups ............................................. 10
Figure 2-6 Dehydration of the Hydroxyl Groups ............................................................. 10
Figure 2-7 Cyclization of the Prepolymer at High Temperatures .................................... 1 1
Figure 2-8 DSC Scans of BMPM and DABPA at 5°C/min ............................................. 12
Figure 2-9 DSC Scan of 1:1 molar BMPM/DABPA at 5°C/min ..................................... 13
Figure 2-10 FT IR Spectra of 1:1 Molar BMPM/DABPA at the Cure Conditions ........... 14
Figure 2-11 Bond Reaction During BMI Curing as a Function of Cure Condition. ........ 14
Figure 2-12 Thermogravimetric Analysis Showing Weight Loss Due to the Hydroxyl
Reaction and Ether Linkage Instability ..................................................................... 16
Figure 2-13 DSC Curves Showing Isothermal Cure Profiles of Matrimid® 5292
Bismaleimide Resin ................................................................................................... 18
Figure 2-14 Viscosity Profile of BMI Resin at the Prepregging Temperature ................. 20
Figure 2-15 Standard Autoclave Cure Cycle for BMI/1M7 Composites .......................... 22
Figure 2-16 Comparison of the Viscosity Profiles of Old and New Batches of
BMI Resin Showing the Higher Cure Rate of the New Resin .................................. 23
Figure 2-17 DSC Scan of Two Batches of BMPM Resin at 5°C/min Showing Lower
Activation Temperature for the "New Batch" ........................................................... 24
Figure 3-1 Autoclave Vacuum Bagging Scheme .............................................................. 29
Figure 3-2 Standard Cure Cycle for BMI/1M7 Composites ............................................. 29
Figure 3-3 Cure Cycle for Low Fiber Volume Fraction BMI/1M7 Composites .............. 30
Figure 3-4 Asymmetric Panel Cure Cycle ......................................................................... 35
Figure 35 Stress Free Temperature Test Set-Up .............................................................. 36
Figure 3-6 Two Typical Strip Deflection Curves Showing the SET ................................ 37
Figure 3-7 ESEM Vertical Jaw - Non-Load Cell End ...................................................... 42
Figure 3-8 ESEM Vertical Jaws - Load Cell End ............................................................ 43
Figure 3-9 ESEM Vertical Jaw Mating Face Plates ......................................................... 44
Figure 3-10 TGA of Dry BMI Resin Showing Negligible Moisture ................................. 49
Figure 3-1 1 TGA of Wet BMI Resin Specimen Showing 4.5% Moisture ....................... 49
Figure 3-12 Failure Micrographs of Dry and Wet Resin Flex Specimens ....................... 51
Figure 3-13 Mid-Plane Shear and Load Nose Shear Failures for Flexure Tests .............. 54

xiv

Figure 3-14 ESEM Micrograph ofa Partially Cured Dry Specimen

Showing "Hackles" .................................................................................................... 56
Figure 3-15 ESEM Micrograph of a Fully Cured Dry Specimen

Showing No "Hackles" .............................................................................................. 57
Figure 3-16 Adhesive Failure of a Fully Cured Dry Specimen ........................................ 58
Figure 3-17 Isothermal DMA Curve for Fully Cured 47% Vf Specimen ......................... 60
Figure 3-18 The Loading of Fiber in the ITS Test ............................................................ 63
Figure 3-19 Stress - Strain Curve for a Typical Resin Tensile Test ................................. 67
Figure 3-20 Modulated DSC Curve Showing Tg Determination ..................................... 68
Figure 3-21 Transient DMA Curve for Partially Cured 47% Vf Specimen ..................... 78
Figure 3-22 Temperature Profile for a Thermal Spiked BMI Composite ......................... 84

Figure 3-23 Moisture Profiles for All Specimen Types during Thermal Spike Testing...84
Figure 3-24 A Tensile Stress-Strain Curve for BMI Resin Showing a Brittle Behavior..87

Figure 4-1 Geometry of the Finite Element Model Showing Resultant Stresses ............. 95
Figure 4-2 The Marc Finite Element Mesh ....................................................................... 97
Figure 4-3 Residual Stresses from Marc FEM Model (Vf = 62%. SFT = 200C) ............. 99
Figure 4-4 Effect of SF T on the Maximum Residual Stress ........................................... 100
Figure 4-5 The Diamond RVE Used in the MicMec Analysis ....................................... 103
Figure 4-6 Location of the RVE in the Model Composite .............................................. 104
Figure 4-7 Location of Analysis Points Used in the MicMec Model ............................. 113
Figure 4-8 Schematic Representation of the Resultant Stresses ..................................... 114
Figure 4-9 Radial Stress Distribution ............................................................................. 117
Figure 4-10 Radial Strain Distribution Showing Large Strains in the Interphase .......... 118
Figure 4-11 Effect of Volume Fraction on Local Stresses .............................................. 120
Figure 4-12 Effect of Volume Fraction on Local Strains ............................................... 121
Figure 4-13 Effect of Stress Free Temperature on Local Stresses .................................. 123
Figure 4-14 Effect of Stress Free Temperature on Local Strains .................................... 124
Figure 4—1 5 Effect of Thickness on Local Stresses ......................................................... 127
Figure 4-16 Effect of Interphase Thickness on Local Strains ......................................... 128
Figure 4-17 Effect of Interphase Modulus on Local Stresses ......................................... 130
Figure 4-18 Effect of Interphase Modulus on Local Strains ........................................... 131
Figure 4-19 The Effect of Fiber Volume Fraction on Lamina Properties ...................... 133
Figure 4-20 The Effect of Sizing Thickness on Lamina Properties ................................. 134
Figure 4-21 The Effect of Sizing Modulus on Lamina Properties ................................... 135
Figure 4-22 Effect of Post Cure Temperature on Crack Density ..................................... 142
Figure 4-23 Effect of Post Cure Temperature on Crack Density ..................................... 142
Figure 4-24 Effect of Stepped Post-Cure on Crack Density ........................................... 143
Figure 4-25 Effect of Stepped Post-Cures on Crack Width ............................................ 143
Figure 4-26 Effect of Vf on Crack Density ..................................................................... 145
Figure 4-27 Effect of Post-Cure on Crack Density ......................................................... 146
Figure 4-28 Effect of Vf on Crack Width ....................................................................... 146
Figure 4—29 Effect of Post-Cure on Crack Width ........................................................... 147
Figure 4-30 ESEM Micrographs of BMI/1M7 Composites Showing The Increase In
Crack Width With Decreasing Fiber Content. ......................................................... 148

XV

Figure 4-31 ESEM Micrographs of BMI/1M7 Composites Showing The Increase In

Crack Width With Increased Post-Cure. ................................................................. 149
Figure 4-32 ESEM Micrographs of BMI/1M7 Composites Showing that
Cracks Form in a Tensile Mode at the Mid—Plane (47% vf, Low Cure) .................. 152

Figure 4-33 ESEM Micrograph of BMI/1M7 Composite Showing Tensile Cracks
Forming in the Mid-Plane and Shear Cracks Forming at the Ply Interface

(47%, High Cure) ..................................................................................................... 153
Figure 4-34 ESEM Micrographs of BMI/1M7 Composites Showing that the Failure

Mode is Interfacial for Both Tensile and Shear Cracks ........................................... 154
Figure 4-35 ESEM Vertical Heating Stage Drawings .................................................... 156
Figure 4-36 ESEM Vertical Heating Stage Chill Plate Drawings .................................. 157
Figure 4-37 Micrographs Showing That Resin Rich Regions Deflect Cracks ............... 159
Figure 4—38 ESEM Micrograph Showing That K3B Resin Has A High Strain To

Failure Which Causes Bridging Of The Interfacial Cracks Which Form. ............... 160

Figure 4-39 ESEM Micrographs Showing Poor Adhesion Between 1M7 and K3B. ..... 160
Figure 5-1 Flexure, Tensile, and Tg Data for BMI Resin with Various Post-Cure ........ 171
Figure 5-2 Absorption Curve for Unidirectional BMI/1M7 Showing a Good Fit to the

Fickian Model, and Non-Fickian Behavior Due to Microcrack Formation ............. 176
Figure 5-3 Hygrothermal IFSS Results ........................................................................... 178
Figure 5-4 A Fully Cured and Saturated Specimen Showing Fiber Debonding

Without Applied Load ............................................................................................. 180
Figure 5-5 Plot of All Hygrothermally Conditioned Mechanical Test Data ................... 181
Figure 5-6 A Cubic Representation of the Experimental Design ................................... 183
Figure 5-7 Graphic Representing the Change in F lexural Strength Due to Cure ........... 185

Figure 5-8 The Effect of Increased Cure on the Physical Properties of BMI/1M7 ......... 188
Figure 5-9 The Effect of Moisture Presence on the Physical Properties of BMI/1M7 ....190
Figure 5-10 The Effect of Increased Volume Fraction on the Properties of BMI/1M7 192

Figure 5-11 Hygrothermal DMA Test Results ............................................................... 193
Figure 5-12 The Effect of Increased Cure on the Rheological Properties of BMI/1M7 .194
Figure 5-13 The Effect of Increased Cure on the Tg of BMI/1M7 .................................. 195
Figure 5-14 The Effect of Moisture Presence on the Rheological Properties of

BIVII Resin and Composites ..................................................................................... 196
Figure 5-15 The Effect of Moisture Presence on the Tg of BMI/1M7 .............................. 198
Figure 5-16 The Effect of Increased Volume Fraction on the Rheological Properties of

BMI/1M7 Composites .............................................................................................. 199
Figure 5-17 The Effect of Increased Volume Fraction on the Tg of BMI/1M7 ................ 201
Figure 6-1 Thermally Spiked K3B Resin Showing Blistering ....................................... 208
Figure 6-2 Close-up of a Blister in a Moist, Thermally Spiked K3B Resin Sample ...... 209
Figure 6-3 A Blister in a Moist, Thermally Spiked K3B/1M7 Composite ..................... 210
Figure 6-4 A Close-Up of a Moist, Thermally Spiked K3B/IM7 Composite

Showing the Viscous Nature of the Failure ............................................................. 21 1
Figure 6-5 All Thermal Cycling Test Data for BMI/1M7 Composites ........................... 213
Figure 6-6 All Thermal Cycling Test Data for K3B/1M7 Composites ............................ 214
Figure 6—7 A Square Representation of the Experimental Design .................................. 216

xvi

Figure 6-8 The Effect of Exposure to Moisture on the Properties of BMI Resin and

Composites ............................................................................................................... 21 8
Figure 6-9 The Effect of Moisture Exposure on Tg for BMI Resin and Composites ..... 219
Figure 6-10 The Effect of Exposure to Moisture on the Properties of K3B Resin and

Composites ............................................................................................................... 220
Figure 6-11 The Effect of Moisture Exposure on the Tg of K3B Resin

and Composites ........................................................................................................ 221
Figure 6-12 The Effect of Thermal Spiking on the Properties of BMI Resin and

Composites ............................................................................................................... 223

Figure 6-13 The Effect of Thermal Spikes on the TE of BMI Resin and Composites ....224
Figure 6-14 The Effect of Thermal Spiking on the Properties of K3B Resin and

Composites ............................................................................................................... 225
Figure 6-15 The Effect of Thermal Spikes on the Tg of K3B Resin and Composites ....226
Figure 7-1 - TPD Profile for 400% Surface Treated 1M7 Fibers ..................................... 242
Figure 7-2 The Chemistries of PEI and PBS .................................................................. 249
Figure 7-3 The Chemistries of PDMS and Thermoplastic Polyimides .......................... 249
Figure 7-4 Outer Skin of PEI Sized 1M7 Tow ................................................................ 251
Figure 7-5 PEI Sizing on an 1M7 Fiber .......................................................................... 252
Figure 7-6 PES Sizing on an 1M7 Fiber .......................................................................... 253
Figure 7-7 Matrimid® 5218 Sizing on an 1M7 Fiber Surface ........................................ 254
Figure 7-8 Matrimid® 5218 Sizing on an 1M7 Fiber Edge ............................................. 255
Figure 7-9 RTV Sizing Between Two 1M7 Fibers ......................................................... 258
Figure 7-10 RTV Sizing Patches on 1M7 Fibers ............................................................ 259
Figure 7-11 Cracking of RTV Sized Composite Caused by Resin Shrinkage ............... 262
Figure 7-12 A Microcrack Within the RTV Sized Cross-Ply Composite ...................... 263
Figure 7-13 Plot of ITS Results for Treated and Sized Specimens ................................ 265
Figure 7-14 General Images of Control, 400% and Air Plasma Treated

Failure Modes .......................................................................................................... 270
Figure 7-15 General Images of Sized Composite Failure Modes ................................... 271
Figure 7-16 100%, Unsized and Air Plasma Treated Mode I Failure Modes Showing

Poor Adhesion Levels .............................................................................................. 272

Figure 7-17 400% Treated 1M7 Mode I Failure Mode Showing Increased Adhesion ...273
Figure 7-18 Images Showing the Adhesion of the Sized Samples

After Mode I Failure ................................................................................................ 274
Figure 7-19 Regions Showing the Failure of PEI and 5218 Sizings Showing Adhesion

Between the Sizing and BMI Matrix ....................................................................... 275
Figure 7-20 Images Showing that the Failure in 5218 Sized Specimens is Between the

Fiber and Sizing ....................................................................................................... 276
Figure 7-21 Beam Damaged Regions of the PBS and RTV Sized Specimens ............... 277
Figure 8-1 Function of the Chemical Coupling Agent ................................................... 283
Figure 8-2 COMADE and Interphase Designer Variable Dependencies ........................ 289
Figure 8-3 Logic Flow for the Interphase Critique ......................................................... 295
Figure 8-4 Logic Flow for the Composite Critique ........................................................ 296
Figure 8—5 Information Flow for COMADE and the Interphase Designer ..................... 298

xvii

Chapter 1

INTRODUCTION

The High Speed Civil Transport (HSCT) is being designed to have a cruise speed of
mach 2.4 and will require materials which can survive exposure to extreme temperatures,
moisture, and thermal shock for lifetimes of 60.000 hours under structural loadings.
Anticipated flight cycles include rapid thermal excursions between temperatures of -60°C
and +190°C. Cooling rates during these excursions are quoted at up to ~500°C/min. A
thermoset, bismaleimide (BMI), and a thermoplastic, AvimidiTo K3B, with 1M7 carbon
fibers are candidate systems for advanced military and commercial aircraft. These
systems have been chosen for their excellent mechanical properties and high temperature

capability.

There are currently several limitations associated with the hygrothermal performance of
these materials. BMI composites experience significant transverse microcracking during
their initial cure cycle which presents a path for moisture diffusion. and causes a
reduction in properties [1]. Because BMI composites do not get fully crosslinked during
standard cure and post-cure cycles [2], they continue to cure during service at the high
use temperatures. This increased cure causes a mechanical property reduction of up to
50% [2]. Prior to this work, a variety of studies were performed to determine the overall

effect of moisture and cure progression on BMI composite properties [2-5]. Few people

2
have, however, studied methods for improving the performance of these systems. Those
methods which have been studied usually cause a reduction in the desirable properties of
the BMI resin (Tg, modulus, processability, and solvent susceptibility) [2]. If these
materials are to be used for these advanced applications. their response to these
environmental conditions must be fully understood, and. if possible, significantly

improved.

The objective of the research presented here is to generate an understanding of the
interactions between the effects of moisture and temperature on the physical properties of
the BMI/1M7 system. Specifically, the research was aimed at relating the macroscopic
response of the material to the microscopic properties at the interface between the fiber
and matrix. The interfacial properties of both the BMI/1M7 and K3B/1M7 systems are
shown to be very poor. By testing the nature of the bond which exists between the fiber
and matrix, methods for improving the performance of the system were found and
investigated. The BMI specimens used in this study were fabricated from initial
components allowing strict control of the processing method and variation of processing
parameters and cure cycles. The K3B specimens were provided in final form by the
Boeing Commercial Airplane Company, and thus less testing flexibility was available.
For this reason, the majority of the study was focused on the BMI system, but when

possible studies of the K3B system are included.

The research has been organized into seven major sections which are represented by

Chapters 2 through 7 in this dissertation. Chapter 2 begins with a review of the materials

b.)

used in the study and investigations into the nature of the cure of the BMI resin. The

experimental details for the testing portions of Chapters 4-6 are given in Chapter 3.

An investigation into the cause of the microcracking phenomenon is presented in Chapter
4. A theory for the mechanism which causes this microcracking is introduced, and
modeling based upon this was performed to show the effects of the various controllable
parameters. The modeling was also used to suggest a method for the relief of the
microcracking. This method entails the addition of an appropriate interphase between the
fiber and matrix. Finally, experimental studies of the microcracking phenomenon were
performed and support the conclusions of the modeling efforts. These studies also

revealed that poor bonding exists between the fiber and matrix in these systems.

A study of the hygrothermal response of the BMI resin and composites is presented in
Chapter 5. The cure is related to the physical properties of the resin, and a study of the
moisture diffusion behavior of both the resins and their composites was performed. The
effects of moisture and extent of cure were studied using a variety of tests on both the
macroscopic and microscopic scale. Composites of varied fiber content were used to
intensify the response of the resin, and to help define the phenomenon encountered.
Testing was organized using a factorial design based on these three factors, and results
are presented in terms of both the main effects: extent of cure. moisture presence, and
volume fraction, as well as an interaction effect between extent of cure and moisture
presence. It was found that both increased cure and moisture have a detrimental effect on

resin properties, but moisture was seen to have little effect on composites.

4
This is followed by a study on the effect of the interaction between moisture and rapid
temperature spikes in Chapter 6. In this study. only macroscopic composite specimens
were used, so the K38 composites were included. Very few effects were seen in this
testing with the exception of the wet, thermal spiked K3B specimens which showed a
blistering phenomenon after conditioning. The blisters were of sufficient size to cause a

doubling of the composite thickness.

With all of the knowledge gained from the studies in the previous chapters, methods for
improving the interfacial adhesion of the BMI/1M7 system were investigated. Methods
investigated include surface treatment of the 1M7 fiber through additional commercial
treatment, or air plasma treatment. and sizing of the fibers. Four different materials were
used as Sizings on the fiber and their effect on the adhesion was determined. It was found
that strong surface treatments such as 400% of the commercial treatment caused an
increase in adhesion, and this is attributed to an increase in surface roughness. The
addition of a PBS sizing to the fibers was also shown to increase adhesion, and offers

some very promising direction for future research.

As a result of this project, a methodology for designing interphase materials for specific
fiber and matrix combinations was developed based upon surface energetics, miscibility
parameters, and functional group chemistry. This methodology was implemented within
an artificial intelligence framework as an expansion of a currently existing material

designer (COMADE) an provides a first attempt at codification of interphase design.

Chapter 2

MATERIAL CHARACTERIZATION

The materials used in this study were: a thermoset bismaleimide resins, Ciba Geigy's
Matrimid® 5292; a thermoplastic polyimide resin. DuPont's Avimid’fi" K33; and Hercules
1M7 fibers. The resin systems are quite complex, and thus a variety of analyses were
performed to develop a more complete understanding of their processing and
performance. This information was used to help explain the phenomenon encountered,
and steer improvement efforts. This chapter presents the nature of the 1M7 fiber, briefly
reviews what is known about the polymerization of the K38 resin, presents detailed
experimental data about the cure of the BMI resin and its relation to resin properties, and
finally presents the results of a moisture diffusion study performed on all of these

materials.

2.1 Hercules 1M7 Fiber

Hercules 1M7 fiber is a five micron diameter polyacrylonitrile (PAN) based fiber which is
produced as 12,000 filament tows. The fibers used in this study were surface treated, but
unsized. Two different treatment levels (100% and 400% of the standard commercial

treatment) were used to determine the effect of surface treatment level on adhesion levels.

 

6
1M7 fibers are usually provided with a proprietary surface treatment which changes the
surface chemistry and adhesion properties. There is evidence that this surface treatment
increases the adhesion properties of carbon fibers due to the removal of a weak surface
layer, and the addition of active chemical groups to the surface [6]. Most of these
chemical groups are oxygen and nitrogen containing species with the largest majority
being carboxylic [6,7]. Increases in epoxy/carbon fiber interfacial shear strengths of up to
2x have been reported [8], along with the expected increases in matrix dominated

composite properties [9—1 1].

2.2 K3B Resin Description

Avimid® K3B is a melt-fusible polyimide based on pyromellitic dianhydride and 2,2'
bis[3,5 dichloro-4-(diaminophenoxy)-phenylpropane] along with other diamines. Dupont
classifies it as an amorphous thermoplastic polyimide which has good mechanical
properties at temperatures of up to 200°C, and good solvent resistance. Despite requiring
360°C cure temperatures, it can be fabricated in standard autoclaves. Due to its
thermoplastic nature, it possesses the additional advantage of allowing the re-processing

of poorly manufactured parts [12].

The polymer used in the Avimid® K series is an amorphous linear condensation
polyimide produced from monomeric solutions by the reaction of an aromatic diethyl

ester with an aromatic diamine in n-methyl pyrollidone (N MP) as a solvent. The reaction

7
produces water and ethanol as a byproduct as seen in Figure 2-1. Further information

concerning the chemistry of K3B resin is not available due to its proprietary nature.

(I? A
C—OEt
,t H2N- Arm + NMP
C—OH
('5 B
Aromatic Diethyl Aromatic Solvent
Ester of Diacid Diamine
1 Heat
0
II
C\ 2 EtOH (A)
/N-Ar«wv + 2 H2O (B)
if NMP
O

Aromatic Polyimide

Figure 2-1 - Simplistic Representation of K3B Polymerization Chemistry

2.3 BMI Resin Curing Reaction

Bismaleimide resin is a high temperature thermoset polyimide which possesses good high
temperature property retention, processability similar to epoxies. and relatively low cost
[13, 14]. Bismaleimide resins are available in five different forms: 1) BMI mixtures; 2)

blends of BMIs and BMI-diamine oligomers prepared through a Michael addition

8
reaction; 3) BMI and olefinic monomeric and/or oligomeric blends: 4) BMI and epoxy

blends; and 5) BMI and 0,0’ -dicyanobisphenol A mixtures [14].

The Matrimid 5292 system from Ciba-Geigy belongs to the third class. It is a two part
system consisting of 4,4’-bismaleimidodiphenyl methane (BMPM) and (),0’-diallyl
bisphenol A (DABPA). The chemical structures of these components are shown in
Figure 2-2. The BMPM monomer is a crystalline powder which melts at 156°C while the

DABPA monomer is a viscous liquid at room temperature [15].

o o
H u
c c
\ /
c c
u n
o

4.4'-bismaleimidodiphenylmethane (BMPM)

CH,= CH— CH, 0w,— CH: CH,

O,()'-diallyl bisphenol A (DABPA)

Figure 2-2 The Chemical Structure of the Matrimid® 5292 Components

2.3.1 Cure Reaction

At temperatures as low as 100°C the allyl bond of the DABPA and the maleimide bond
of the BMPM react to form a "prepolymer." This reaction has been termed the "ene"

reaction and is shown in Figure 2-3.

CH,———CH—C CH, —=CH CH,

HQ: ‘H .E: .CH‘ Ni]

DA BPA BM PM

CH,=CH—C flbi: ‘51}:- -,-—CH mic “H N:::::lc

Figure 2-3 The "ene" Reaction

With this prepolymer in the resin mixture during the initial stages of cure, reactions are
possible between five active groups. The active groups shown on the prepolymer in
Figure 2-3 are the allyl (A), propenyl (B), maleimide (C), and hydroxyl groups (D). The
possible reactions include the "ene" reaction shown above. homopolymerization of the
allyl groups (Figure 2-4), homopolymerization of the maleimide groups (Figure 2-5),
dehydration of the two hydroxyl groups (Figure 2-6), and a cyclization reaction between

the allyl and propenyl groups of the prepolymer (Figure 2-7).

10

Where R1 and R2 are continuations of the

'prepolymer’ formed in the "ene" reaction,
or the remainder of a DABPA molecule.

Figure 2-4 Homopolymerization of the Allyl Groups

 

 

 

/—\‘ — ("H—CH (PH—(EH
0: \N/ =0 0: [\N/ :0 O//C\N/C=O O//C\N/C=O
CH, + CH, —> C H, 11,
RI R7 R1 R2

Where RI and R2 are maleimide groups, or continuations

of the prepolymer formed in the "ene" reaction.

Figure 2-5 Homopolymerization of the Maleimide Groups

HO—R—OH + HO—R—OH —b HO—R—O—R—OH + H20

Figure 2-6 Dehydration of the Hydroxyl Groups

11

N—C\\
O
CH,
0 O
——N
,C \
9HCH/C§O
CH, 2
/
* CH—CH H
\CH \CH’
/ \
H2C CH2
CH3
|
”° ‘f °”
CH

Where * are radicals which react with other molecules.

Figure 2-7 Cyclization of the Prepolymer at High Temperatures

GPC results of resin mixtures aged at low temperatures (100°C) have shown that the
"ene" reaction occurs before homopolymerization of either the allyl, or maleimide groups
[15, 16]. This is supported by DSC scans of the two separate components (Figure 2-8)
which show that for the BMPM, significant reaction doesn't begin until 175°C, and for
the DABPA, the reaction requires at least 200°C. A DSC scan of the mixed system (1:1

molar) shows that reaction begins occurring as low as 120°C (Figure 2-9).

Heat Flow (N/gl

Heat Flow (N/Q)

12

 

 

 

 

 

 

 

 

 

 

 

1.0
1
0.54 225.95°c
o o- N .
.T
201.25°C
209.9J/g
-0.5fi
l
-1,0—.
155.86°C
'1 .5 v I v I v r V r I
0 50 100 150 200 250 300
Temperature ('C)
0.4
0.2-
0.0-
243.19°C
-O.2-
-0 . 4 ' I ' I ' i ' f r Y I T
0 50 100 150 200 250 300 350 A00

Temperature ('C)

DABPA

Figure 2-8 DSC Scans of BMPM and DABPA at 5°C/min

l3

 

    
 
   
 
     

 

 

 

 

0.3
241.22’C
,
0.2-
.8 0.1—
.5
3
O
lL
u
g 0.0—
x
324.95’C
1
0
-o.1+ 299.10 c
J 179.10’C
239.4u/g
“0.2 v T V f '— 1 ' l ' T V r ' fl '.
CI 50 100 150 200 250 300 350 400

Temperature ('C)

Figure 2-9 DSC Scan of 1:1 molar BMPM/DABPA at 5°C/min

As the cure temperature is increased, the "ene" reaction progresses more rapidly, and the
homopolymerization reactions begin. FTIR studies were performed to show the amounts
of reacted species during the cure cycle. The actual FT IR curves are provided as Figure

2-10, and the reduced data is shown in Figure 2-1 1.

 

REACTION PERCENTAGES

 

1250

14

 

 

 

 

 

 

1200 1:50 noo :oso 1000 9 o 900 use 300 750 7'00
-1
Wavenumber (cm)

Figure 2-10 FTIR Spectra of 1:1 Molar BMPM/DABPA at the Cure Conditions

   

 

 

 

 

uALEIquE CONSUIIPTION
100- 820 cm-t)
ALLYL CONSUMPTION
90~ (915 cut-1)
.HJ
70"
m-
504
/ ETHEP APPEARANCE
40~ (1133 cm-1)
”d
m-
10-
I I 1 I 1 l 1
J1 l1 |1 |1 I3 [9 Inns, HRS

 

   

 

 

I130 I130 [150 [175 Izoo 1225 Jase [an ITEIIP.,'C
' Accuuuumve cues eduomous *

 

Figure 2-11 Bond Reaction During BMI Curing as a Function of Cure Condition.

15
The disappearance of the allyl and maleimide groups show a 1:1 correspondence, lending
support to the theory that the "ene" reaction continues to be favored over either
homopolymerization reaction. The appearance of an ether linkage suggests that the
hydroxyl groups react during the cure, but become kinetically hindered by the glassy
state. Additional cure at a temperature of 300°C for 9 hours resulted in another 10%
reaction of the hydroxyl groups as seen by a decrease in the hydroxyl FT IR band at 3473
cm'l. The concentration of ether linkages, however, did not increase. These ether
linkages have been shown to be unstable at 300°C [17]. The presence of the hydroxyl
reaction, and instability of the ether linkages is further supported by thermogravimetric
analysis (TGA) as shown in Figure 2-12. Rapid weight loss is seen during the initial
ramp to 177°C, followed by an increase in weight loss at each step of the cure cycle.
Finally a rapid weight loss at 300°C is seen which demonstrates the instability of the ether

linkages. Longer TGA experiments showed that the rate of weight loss at 300°C was still

increasing after six hours.

16

99.2 400

 

/~——-———-—- r-300

 

 

 

C3

0

5 m
L

‘c" ”200 3
01 IO
o-c C.
0 (U
I O
E

OJ

’—

"100
99.0 v r ' I ' n ' I ' 1 ' T ' 0
0 2 4 6 8 10 12 14

 

Time (hr)

Figure 2-12 Thermogravimetric Analysis Showing Weight Loss Due to the
Hydroxyl Reaction and Ether Linkage Instability

Reactions involving the propenyl bond are likely to occur only at relatively high
temperatures due to steric hindrance effects. A more complete characterization of the cure
reactions of this system has been published [15]. While the exact nature of the ultimate
cured network is not known, the reactions suggested here provide insight into the
chemical nature and stability of the resin, which is very important when considering long

term hygrothermal effects.

2.3.2 Reaction Rate Measurement: Isothermal DSC Profiles

The reaction of this BMI system, as with all polymers, is controlled by both

thermodynamics and kinetics. Thermodynamically, all of the reactions presented in the

17
previous section are viable, but the rates at which they occur varies with temperature in a
different manner for each reaction. Kinetically, as the cure progresses, the polymer's
chain mobility decreases. This slows the reaction rate by limiting the interaction between
active bond sites. Ultimately, the Tg of the resin will exceed the cure temperature and the
reactions will slow until they appear to have stopped. The reactions are still occurring,

but at a much slower rate.

This presents one of two problems: either additional cure must be provided for increased
properties, or the "as-cured" condition is acceptable. If additional cure is desired, the cure
temperature must be increased to promote a higher reaction rate. If the "as-cured" system
is desired and the use temperature is high (as is generally the case with these materials),

increased cure will occur during use, limiting the life of the component.

This discussion illustrates two factors which must be understood before these materials
can be safely and economically used. First, the cure kinetics must be understood so that a
proper cure cycle can be chosen and the additional cure that takes place in service over
time at use temperatures can be predicted. Second, the effect of the extent of cure on the
material properties must be understood so that a proper cure cycle can be chosen, and the

change in properties over the component lifetime are understood.

To this end, isothermal Differential Scanning Calorimetry (DSC) runs were performed by
others associated with this project [15]. For these runs, a small amount of mixed BMI

resin was placed in the DSC furnace at the desired temperature, and the amount of heat

18
evolved was measured as a function of time. The degree of cure, a , was calculated

using Eqn. 2-1.

._ _t Eqn. 2-]

Where AH, is the partial heat of reaction at time t and AH” is the total heat of reaction.
AH" was found to have a value of 389 1g1 via a dynamic DSC scan with a 5°C/min

ramp rate. The isothermal percent conversion curves are provided in Figure 2-13. From

these curves, the extent of cure for various cure cycles can be approximated.

 

   

 

   

 

 

 

 

1.0 ,
09 -/' ___.——-—-—‘-‘-———— ;
///::';”, ,.-.—‘-——--——--—--—--—---=
08 /-- fl __________
/.‘/ / I/I”
: o7 /// //’
.9 -' /
w /
5 06- / / ..........................
> ,- .........
s /
0 05‘ / /
‘5 /
8 0"“ —1soc
3" = ------ 1500
0' ”fl, ——1aoc :
5 ' —--2ooc j
02‘ ——2soc I
—--2eoc
0'1 —~3ooc
0.0 lrrIIrIT—ITIITIIIIfIITIlIlIlIIIlIIIIIIIIIIIIIIIIIIIIIIIIIII:

 

 

0 60 120 180 240 300 360 420 480 540 600 660

Time (minutes)

Figure 2-13 DSC Curves Showing Isothermal Cure Profiles of Matrimid® 5292
Bismaleimide Resin

19
Unfortunately, the extent of cure itself is not a sufficient description of the resin to truly
predict material properties because the network structure is a function of the full cure
history. Due to competing reactions, different thermal histories will result in different
networks which could have significantly different properties. This effect was analyzed by

testing resin specimens with a variety of thermal histories and is presented in Chapter 5.

2.3.3 Composite Fabrication Issues

The main use for these high temperature resins is as a matrix for high performance
composites. Manufacturing high quality composites with these resins requires an
intimate knowledge of the resin‘s cure mechanism and viscosity profile. The composites
used in this research were manufactured by creating pre-impregnated tape (prepreg) of the
fibers and resin. This tape is then manually laid—up into a part which is autoclave cured.
The full details involved in these processes are provided in Chapter 5. It is absolutely
necessary to know the changes in the resin's viscosity through the thermal cycles involved
in both the prepregging and autoclave cure cycles. The viscosity of the resin during these

cycles was determined and is discussed here.

2.3.3.1 Resin Viscosity During Prepregging

The viscosity of the mixed resin at the prepregging temperature has a great effect on final
composite properties. A constant viscosity during prepregging is essential to create
uniform prepreg. If the resin viscosity increases too much during the process, the resin

content and flow characteristics of the prepreg will change. This in turn can cause

20
fluctuations in the local volume fraction of the final part which will be shown to have a

very significant effect on properties.

A viscosity profile at the prepreg processing temperature of 93°C was performed and is
shown in Figure 2-14. A single run of prepreg took approximately 45 minutes of actual
run time, and approximately two hours elapsed from the initial resin mixing until
completion. It can be seen that the viscosity begins to change significantly toward the
end of the time. To counter this, the resin temperature was generally increased by 5-10°C
during the final 10 minutes of each prepreg run, and the resin was never reused. Good

prepreg quality was achieved with a consistent resin content across the sheet.

 

1600

93°C Isothermal Hold

1soo~~

1400 ~~

Vlscoslty (cps
§ s

d
I
I

 

 

 

 

0:00 2:00 4:00 6:00 8:00 10:00
Elapsed Tlme (hours)

Figure 2-14 Viscosity Profile of BMI Resin at the Prepregging Temperature

21

2.3.3.2 Resin Viscosity During Autoclave Curing

The uncured part is subjected to a controlled autoclave cure cycle to create a void free
composite of high fiber volume fraction. To accomplish this, excess resin and air voids
must be removed from the laminate, remaining voids must be minimized, and the resin
must become gelled. The cure cycle contains essentially three controllable parameters:
temperature, vacuum pressure, and hydrostatic pressure. These parameters must be

optimized to achieve the desired laminate quality.

The removal of air voids and excess resin occurs during the initial stages of the cure cycle
by applying a vacuum while the resin has a relatively low viscosity. If, however, the
viscosity gets too low, the vacuum must be reduced or removed before too much resin is
lost which would result in a poor quality laminate. Once the resin has gelled, further
resin motion is unlikely and any entrapped voids will remain. The hydrostatic pressure is
applied just before gellation to shrink any remaining voids and generally consolidate the

laminate.

Obviously, to apply the vacuum and pressure at the proper times, it is very necessary to
know the resin viscosity throughout the cure cycle. In thermoset systems such as this one,
both the temperature and the extent of reaction control the resin viscosity. Higher
temperatures generally reduce the viscosity of fluids, but in this case, the increased
temperature also increases the extent of reaction, causing a reduction in viscosity. This

competing effect is seen in the viscosity profile shown in Figure 2-15.

22

 

 

—Temperature (F)
I - - - — Vacuum (in Hg) 2000
— -Pressure (psi) 2
Viscosity (cps) 1750

   

 

 

 

‘9‘ i.
I :— 1500
5 A 0 a
”"3 i: a
g e .. 1250 3
:I g " g
3 3 .. a
> ~ 0
£2; I 1000 3
g- I:
.2 3’ 75°
3— 500
I 250

 

 

 

 

100 120 140 160 180 200 220 240 260
11me (min)

Figure 2-15 Standard Autoclave Cure Cycle for BMI/1M7 Composites

From this profile, it is easily understood that the use of an initial vacuum will help
remove excess resin and voids, but the vacuum must be reduced to avoid removing too
much resin during the period of extremely low viscosity. The vacuum can then be
increased once the viscosity increases, followed by the application of hydrostatic pressure

just before complete gellation to shrink the remaining voids.

2.3.3.3 The Effect of Resin Batches on the Viscosity Profile

To further complicate matters, two different batches of resin with significantly different
viscosity profiles were used in this study. Their profiles are shown in Figure 2-16. This

resulted in prepreg and final panels with different fiber volume fractions, despite identical

23
processing conditions. The change in the viscosity profile is attributed to a lower
activation temperature for the second batch of BMPM resin as seen in Figure 2-17.
While this is quite disconcerting for a scientific study, all tests were self-consistent, since
specimens made from only one batch of resin were used. In no case did the effect of the

two resin batches confound the results.

 

 

 

 

 

 

‘30 ' 3500
170 -. 7
.. 3000
160 «~
~~ 2500
150 ~~
55 m a
I 140 ~- ‘* 3
E
Q
8
130 " «~ 1500 .2
h >
120 a.
4» 1000
110 «~
~» 500
100 1~
90 L 1 -——‘—"_f.££.;. ""v 1 1 O
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00
Time (min)

 

L — New Visc. Temp. —Old Visc. Temp. + New BM! Visc + Old BMI Visc 1

 

Figure 2-16 Comparison of the Viscosity Profiles of Old and New Batches of BMI
Resin Showing the Higher Cure Rate of the New Resin

24

 

191.18°C

 

(N/g)
O
p O
l

 

 

 

 

I T "k
3
g . 1
m 167.07°C
3 195.5J/c
m -0.5—
I
J
-l.0~
‘ 157 33°C
-1.5 I V Tr ' 1 V I v —f v 1 v I v I v
o 50 100 150 200 250 300 350 400

Temperature 1°C)

Figure 2-17 DSC Scan of the "New Batch" of BMPM Resin at 5°C/min Showing
Lower Activation Temperature

2.4 Material Characterization Conclusions

The materials used in this study were reviewed, and a reaction mechanism for the BMI
resin was presented with data supporting the theories presented. Cure reaction rates were
presented as a function of temperature and can be used to predict the extent of cure for
various cure cycles. The effect of resin viscosity on processing of the BMI was

discussed, and an optimized autoclave cure cycle was presented.

A variation in the processing response of the resin was shown to be due to a difference in
the activity of the BMPM monomer between two batches used here. This variation will
cause no effect on the results of these studies due to a self-consistent use of the resin from

each batch.

Chapter 3

EXPERIMENTAL DETAILS

The experimental details for all of the testing associated with this research are provided in
this chapter. It is presented in the same order as the data is discussed in the following
chapters with the exception of some overlap between the hygrothermal and thermal spike

testing which is presented together here, and discussed separately later.

3.1 Overall Specimen Fabrication

Three major types of specimens were manufactured. Resin specimens were created to
provide an accurate understanding of the effect of moisture on the resin properties. An
extensive study of the effect of cure on resin properties was presented in Chapter 2. Two
types of composite panels were created. Sixteen ply, [0°/9O°]4S. panels were made for
macroscopic property testing, and four ply, [00/9O°]S, panels were made for tests of the
thermal properties of the system. The detailed methods used for manufacturing these

specimens is provided here.

25

26

3.1.1 BMI Resin

A number of studies on the nature of the pure resin were performed which required
macroscopic specimens. For this resin plaques were created. The details of this BMI

resin system have been described in detail in Chapter 2.

Ciba-Geigy's Matrimid 5292'?” resin Part A (BMPM) and Part B (DABPA) were mixed in
a stoichiometric ratio (100:85 wt%) at 130°C . The resin was degassed in a 29.5 in. Hg
vacuum at 125°C for 30 minutes. The mixed resin was then poured between preheated
(93°C) glass plates treated with the F rekote 700NC release agent. The resin plaques were
cured in a Blue-M, programmable, nitrogen purged oven. The cure cycles for the resin

panels used in these studies are provided in Table 3-1.

Table 3-1 - BMI Resin Panel Post-Cures

 

 

Panels Cure Cycle Testing Performed
AZ 177°C/1 hr, 200°C/1 hr + Hygrothermal Flexure, and
200°C/1 hr, 250°C/6 hrs Isothermal and Transient DMA
G 180°C/1 hr, 200°C/2 hrs, Moisture Diffusion
250°C/1 hr
K 180°C/1 hr, 200°C/2 hrs, Moisture Diffusion
250°C/6 hrs, 300°C/1 hr
TSR & 2TSR 177°C/1 hr, 200°C/2 hrs. Thermal Spike DMA
250°C/6 hrs

 

 

 

 

 

27
These plaques were sectioned into the appropriate specimen sizes on a Bennett 5000 Plate
Saw, using a 1/8” acrylic underlay and the highest traverse speed setting. The specimens’
surfaces were then polished using a Struers Abramin Polisher to remove any possible

release agent effects (a minimum of 0.5 mm was removed per side).

3.1.2 BMI/IM7 Composites

The fiber used was Hercules IM7. This is a PAN (polyacrylonitrile) based carbon fiber
which was provided in 12,000 filament tows. While normally provided with a sizing
which enhances interlaminar shear properties and processability, the fibers used in this

study were obtained in an unsized form to avoid unnecessarily confounding the results.

3.1.2.1 BMI/IM7 Prepreg Manufacture

To fabricate the composites, preimpregnated tape (prepreg) was made from the
constituent materials using a Research Tool hot melt prepregger. The resin was mixed in
a 100285 wt% ratio (BMPM/DABPA) by heating the liquid to 130°C while stirring and
slowly adding the powder. This mixture was then degassed at 130°C in a 29 in. Hg
vacuum. It was then cooled to 93°C and poured into the prepregger's resin pot which was
preheated to 93°C. IM7 fiber had been threaded through the pot. die, flattening pins
(104°C), and guide roller (104°C). The die was 0.22" wide and was shimmed 0.0115"
per side. The original gap size is unknown and there was some wear of the die. The fiber

was taken up on a 2 ft. diameter drum layered with a teflon release film. A drum speed of

28
2.1 rpm was used with a traverse rate appropriate to creating gapless tape. Once a
complete 12" wide tape was created, it was cut off the drum and sectioned into the
desired pieces and stored under refrigeration in nitrogen purged bags until use. Resin
content was measured using fiber spool and resin weight losses. The prepreg contained

an average of 64 vol% resin and good process control was achieved.

3.1.2.2 BMI Composite Cures

Panels were fabricated using standard hand lay-up technique. Plies were manually
oriented as desired. A heat gun and roller were used to press the plies together after
application of each ply. The final laminate was then placed on a steel caul plate with
bagging material as depicted in Figure 3-1. The number of layers of thin bleeder cloth

varied depending upon laminate thickness.

The bagged plate was then placed in a United McGill autoclave and attached to a vacuum
line. The cure cycle used is shown in Figure 3-2. The resin viscosity was determined by
performing viscometry on neat resin using the cure cycle’s temperature profile and was

discussed in Chapter 2.

  

 

 

 

 

 

 

   

Elastomeric Dam

 

 

 

 

<— Vacuum Port

Nylon Vacuum Bag
Thick Bleeder Cloth
Porous Release Film
Porous Aluminum Plate
Porous Release Film
Thin Bleeder Cloth
Porous Teflon

Composite

Porous Teflon
Thin Bleeder Cloth
Non-Porous Teflon
Steel Caul Plate

Figure 3-1 Autoclave Vacuum Bagging Scheme

 

400

 

Temp (F). Vacuum (lan).
Pressure (psi)

 

‘iTemperature (1:) I‘
l

1- ' ' Vacuum (10 H9) l. 2000
‘— - Pressure (psu) ‘

—Vlscosuty (cps) 1 1750

 

 

 

Viscosity (cps)

 

 

0 20 40 60 80 100

240 260

Figure 3-2 Standard Cure Cycle for BMI/IM7 Composites

This cure cycle was varied to create low fiber volume fraction panels. The cycle used to

Create these panels did not contain a vacuum cycle, and is shown in Figure 3-3. Note that

30
for panel V, the prepreg was flattened before lay-up using a Carver press at 20 metric tons
(over an 11" x 11" area). It was also flattened as a panel once completely laid-up. This

helped remove the entrapped air which would have caused voids.

400 2250

 

350 2000

 

1750
300

1500
250
1250

200 ,
1000

    

Pressure (psi)
Viscosity (cps)

150
750

Temp (F), Vacuum (lan),

 

100 «

_._._.-._) ________ _ ...........

1‘8"“) (F) T -

50 \ 1— - Pressure (psi) 1 250
a
\ l—Viscosity (cps) ‘
‘II. . 0

0 a
0 20 40 60 80 100 120 140 160 180 200 220 240

500

 

 

 

Time (min)

Figure 3-3 Cure Cycle for Low Fiber Volume Fraction BMI/IM7 Composites

The lay-up, cure cycle, and final fiber volume fraction of the composite panels used in
this study are summarized in Table 3-2. Panel J was subjected to the cure cycle shown in

Figure 3-3, but the vacuum bag was evacuated and sealed off before the thermal cycle.

The shift in resultant fiber volume fraction seen for panels X and Y was due to the new

batch of resin which was shown to be reactive at a lower temperature (measured via

31
DSC), and to have a higher viscosity profile over the thermal cycle (measured via

viscometry) as discussed in Chapter 2.

Table 3-2 - BMI Composite Panel Descriptions

 

 

 

Panel Lay-Up Cure Cycle Vf Comment

H [0°/90°]S Standard 62%

J [0°/90°]S Evacuation Only 55% Dry Outer Ply Surface
W [0°/90°]S No Vacuum 47%

I [0°/90°]4S Standard 62%

K [0°/90°]48 Standard 63%

V [0°/90°]4S No Vacuum 35%

X [0°/90°]45 Standard 47% New Batch of Resin
Y [0°/90°]45 Standard 47% New Batch of Resin
AA [0°]l6T Standard 55% New Batch of Resin
AB [0°]I6T Standard 51% New Batch of Resin

 

3.1.2.3 BMI Composite Post-Cures

Panels were either directly post-cured after the initial cure. or partially post-cured,
sectioned in half, and then one half was post-cured for comparison of cure effects. All
post-cures were performed either in a nitrogen purged Kapton"r bag. or using a nitrogen

purged oven. The panel post-cures are shown in Table 3-3.

3.1.3 K3B/IM7 Panel History

The K3B composite panel used in this study (# 488) was obtained from Boeing Corp. in

final form. It was made from DuPont's Avimid® K3B and Hercules 1M7 fiber. The panel

 

32

had a [0°/9O°]4S configuration. All machining was performed on the Bennett 5000 Plate

Saw with an acrylic underlay.

Table 3-3 - BMI Composite Panel Post-Cure Cycles

 

 

Panels Post Cure Cycle
HA, JA, WA, 200°C/2 hrs
IA, KA, VA
HB, JB, WB, 200°C/2 hrs +
IB, KB, VB 280°C/6 hrs
X, Y 200°C/2 hrs +

250°C/6 hrs
AA, AB 200°C/2 hrs +
250°C/6 hrs +

280°C/1 hr

 

 

 

3.1.4 K3B Resin Plaque Manufacture

K3B neat resin specimens were cut from a plaque created at Michigan State University.
This plaque was created by taking the K3B monomers, fully imidizing them, and then
grinding the resultant polymer into small particles. These particles were then
compression molded into a plaque at 350°C.

NOTE: The polymerization cycle for the neat resin and composite K3B specimens is
considerably different, allowing for a significant variation in the resultant polymer and

thus final properties.

33

3.2 Microcracking Characterization Experiments

Various experiments were performed which support the modeling work. In particular, 1)
the SFT was determined for various cure cycles; 2) unidirectional composites were
examined to see if the microscopic mechanism alone is capable of causing matrix failure;
and 3) the crack density and width were determined for various SFTs and volume

fractions.

3.2.1 Determination of the Stress Free Temperature

The SF T is caused by a combination of the thermal expansion mismatches from both the
microscopic and macroscopic mechanisms described in the modeling section, and also by
residual stresses due to the shrinkage of the resin. Thus, the SFT itself cannot be simply
predicted, or measured by such properties as T8. T0 accurately determine an appropriate
value for the SF T, unsymmetric laminates were created with the various post-cures. They
were then subjected to a thermal ramp and the temperature at which their deflection
returned to zero was determined and deemed the stress free state. The procedures used,

and results are presented here.

3.2.1.1 Experimental Procedure

Specimens were made at the Composite Materials and Structures Center at Michigan
State University from unsized IM7 fiber, and the two component Matrimid 5292® BMI
system from Ciba Geigy. Prepreg was made using a Research Tool hot melt prepregger.

The resin was mixed at 130°C, degassed at 130°C for 10 minutes under vacuum, and then

34
brought to 93°C and poured into the resin pot. IM7 fiber tow (12k) was fed through the
pot and a shimmed die onto a drum lined with release paper. The final resin content (~62

vol% resin) was determined by fiber spool and resin weight loss.

Panels were created by hand layup of this prepreg. The panel was 10 in2 with a [02/902]T
configuration. The cure cycle used is displayed in Figure 3-4. The warped panel was
then cut into 20mm wide strips which were approximately 8 inches in length. All panel
edges were removed to avoid manufacturing edge defects. Specimens were dried in
vacuum at 80°C until their weight loss was no longer accurately measurable. Specimens
were post cured in either an inert gas oven, or in a nitrogen purged vacuum bag. Ramp

rates were 5°F/min.

 

 

   
   
  

 

 

 

     

 

  

 

 

400 . 7 7 , g f 2250
l—Temperature (F)1
. 1.
350 ‘ f' " " Vacuum (In Hg) 3.. 2000
L— - Pressure (psi) 1;
I VISCOSI c s l 1750
300 _ -, , - 7W! 9).;
i: 250 7' e 1500
'5; a
5 3.. 1250 g
3% 200 7 E
> a a
,; g « 1000 8
' m
9"“ 1501 5
a. .
E _ , 750
100 +5 -- — - — - —I-.---I-I-----I--
500
50 , , 250
0 . z : - — — - O

120 140 160 180 200 220 240 260

Time (min)

Figure 3-4 Asymmetric Panel Cure Cycle

To determine the SFT, warped [03/902]T composite strips were taped in a glass faced oven
such that the degree of warp was measurable as seen in Figure 3-5. Temperatures were
determined with a thin thermocouple taped directly to the strip. The deflection of the
strip was measured as a function of temperature. Extrapolating the deflection vs.

temperature curve to zero deflection provided an accurate SFT.

4
mm,

s

36

 

- . Thermocouple

Oven Tray . Laminate Stip Metal Ruler

. I

 

 

 

 

 

Front View

 

 

 

Figure 3-5 Stress Free Temperature Test Set-Up

Two typical deflection vs. temperature curves from the experiments are provided in
Figure 3-6. The SFTs associated with various post-cure cycles investigated are provided

in Table 3-4.

Strip Deflection (64th In

37

Strip Deflection vs. Temperature

 

 

 

 

60 _.-.,-LL-__-
+ 177°C/90min 1
so . 4.177599%... 30999:!
40
30 .
20 .
10 .
SFT: 105°C SFT= 170°C
0 . - _
20 40 60 80 100 1 20 140 160 1 80 200

Temperature (°C)

Figure 3-6 Two Typical Strip Deflection Curves Showing the SF T

Table 3-4 Stress Free Temperatures of Various Cure & Post-Cure Cycles

 

Stress Free

 

 

Cure & Post-Cure Cycle % Temperature
Cure

177°C/90 min .57 100°C
177°C/90 min + 200°C/2hr .76 170°C
177°C/90 min + 250°C/2hr .84 230°C
177°C/90 min + 300°C/2hr .86 245°C
177°C/90 min + 200°C/2hr + 250°C/2hr .84 210°C
177°C/90 min + 200°C/2hr + 250°C/2hr + .90 240°C

300°C/2hr

 

 

38

3.2.2 Determination of Microscopic (Fiber Level) Cracking

To determine if the microscopic stresses were sufficient to cause microcracking,
unidirectional specimens were created with varied post-cures. These were then sectioned,
and polished. Both optical microscopy and ESEM analysis were performed and showed

that no cracking existed due to this mechanism.

Unidirectional panels were manufactured using the same procedure outlined in Section
3.2.1.1. The panels were 5 in2 and had a layup of [04],. The panel was sectioned into
approximately 1" squares. These were post-cured in either vacuum, or a nitrogen

atmosphere using the post—cure cycles shown in Table 3-5.

Table 3-5 Post-Cure Cycles Used for Microscopic Cracking Experiment

 

Cure & Post-Cure Cycle
177°C/90 min
177°C/90 min + 200°C/2hr
177°C/90 min + 250°C/2hr
177°C/90 min + 300°C/2hr

 

 

 

177°C/90 min + 200°C/2hr + 250°C/2hr + 300°C/2hr

 

To determine if there were any microcracks caused by the fiber level expansion
mismatch, polished sections of the “Mr panel with varied post-cures were examined. The
CMSC's Environmental Scanning Electron Microscope (ESEM) was used to view the
specimens at magnifications of up to 10,000x. No microcracking was evident in any of

the samples.

39

3.2.3 Determination of Crack Density and Crack Width

Two separate experiments were performed to measure crack densities and widths. The
first was designed to determine the effect of post-cure temperature and stepped post-
cures. The second was designed to show the effect of fiber volume fraction (47, 55, and

62% V1) and cure cycle (Low and High).

3.2.3.1 The Effect of Vf and Post-Cures on Microcracks using Optical Microscopy

Initially, panels were created with varied volume fractions and cure cycles. Crack

densities and widths were measured using optical microscopy.

Three cross-ply panels were manufactured using the same procedure outlined in Section
3.1.2. The panels were 5 inz, had a lay-up of [0°/90°]S, and had volume fractions of 45%,
56%, and 65%. The panels were sectioned into 2" x 1" specimens and dried in a vacuum
oven. They were then post-cured ‘in nitrogen atmosphere using the post-cure cycles

shown in Table 3-6.

40

Table 3-6 Post-Cure Cycles Used for Microscopic Cracking Experiment

 

Cure & Post-Cure Cycle
177°C/90 min
177°C/90 min + 200°C/2hr
177°C/90 min + 250°C/2hr
177°C/90 min + 300°C/2hr

 

177°C/90 min + 200°C/2hr + 250°C/2hr
177°C/90 min + 200°C/2hr + 250°C/2hr + 300°C/2hr

 

 

 

The post-cured specimens were then sectioned. mounted. and polished. The polished
specimens were analyzed using a light microscope. Crack densities were obtained by
counting the number of cracks over a measured length of approximately 3A". Crack

widths were measured using a video caliper which was calibrated to i0.01mm before use.

3.2.4 Identification of the Effect of Laminate Features on Crack Propagation

To improve the BMI/1M7 system's toughness, the propagation behavior of cracks must be
understood. To determine this. unidirectional composite specimens were incrementally

fractured in the ESEM while being monitored.

3.2.4.1 ESEM Vertical Jaw Design

To monitor the fracture of composite specimens in the ESEM, the CMSC'S tensile stage
was used. This stage fits within the ESEM chamber, and is capable of loading specimens

up to 1000 lbs. The original grips for the stage were oriented such that the specimen

41
"width" is observable. To watch the propagation of cracks in composites, it is necessary
to view the "thickness" while under load. To allow this. a set of vertical jaws were
designed and fabricated. The designs for these jaws are shown in Figure 3-7, Figure 3-8,

and Figure 3-9. They were made from tool steel with tolerances of i0.010".

Unidirectional ([0°]1(,.,.) panels were fabricated and cut into notched tensile specimens.
They were fabricated according to the procedure outlined earlier in this chapter. The cure
cycle was 177°C/90min with both high and low post-cures (280°C/10 hrs, and
200°C/2hrs). Six specimens were fractured in the ESEM and the resulting cracks were
observed. Only one specimen's propagation was imaged satisfactorily. This specimen

had a high post-cure. A progression of images is provided and discussed in Chapter 4.

42

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O 9 Q
9......
Q 0 O 9
9.9.9.9 _L
1.250" so). to H
°9 9’9’ H 3110"
0500'23'333 ’9’9‘9’ 0170""
9.9.9.9 (11/64 )
T0180" “DIV-7"
' Th reader
I 0.500" I 0.250..
FRONT SIDE
0 200" TOP

 

 

 

 

 

. " T
gpwg i I © 0

 

 

 

 

 

 

 

Figure 3-7 ESEM Vertical Jaw - Non-Load Cell End

43

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.280"
(9/32")
-_ W
0.180"_L L i T LEI: L @0180”
1. 1.250" ’ O+_1| "
0500" 000631 (1' 1,734..)
+.
0 180" 5711“ #1 (21
' —— l l rea ed
i—O'S—(mT-‘i 0.250"
FRONT SIDE
0250" TOP
T r— |
0.150'
0.400' - - |©
_L :1 O
1 0.625" 0925.. ©
@ J- SIDE
|—— 1.000"———| ~
TOP /
FRONT

Figure 3-8 ESEM Vertical Jaws - Load Cell End

0.28021: 3::

 

1:: 0170"
(11/64)

 

 

 

FRONT

 

 

 

 

TOP

44

0.180"l

 

 

 

0.1801"
I 0600' l
SIDE

TOP

 

__S_|%

\

FRONT

 

 

 

 

Figure 3-9 ESEM Vertical Jaw Mating Face Plates

45

3.3 Hygrothermal Study Experiments

The objectives of this hygrothermal study were to identify any effect due to post-cure
cycle, moisture, or fiber volume fraction. These factors were chosen for a variety of
reasons. Cure may change over the life of the part, and has been shown to reduce
mechanical properties over time. Moisture has been shown to be both beneficial and
detrimental to composite properties. Volume fraction was chosen because most laminates
contain local regions of very high resin content which can have a great effect on
composite performance. The properties of these regions can be determined by testing low
volume fraction composites. The properties of interest were modulus, strength, failure

strain, and Tg. The test scheme used to study these is shown in Table 3-7.

A wide variety of experiments were performed to determine the effect moisture, cure
advancement, and volume fraction. The methods used for the fabrication of samples is
provided below, followed by the experimental procedures, data reduction schemes, raw
tabular results, and failure analysis results for each type of test. The results obtained are
analyzed in a method appropriate to the statistical approach taken here and discussed in

Chapter 5.

46

Table 3-7 Hygrothermal Testing Scheme

 

 

 

Test Performed Materials Studied Properties Reported
Tensile BMI Resin Modulus, Strength,
Failure Strain
F lexure BMI Resin (Three Point. Flexure Modulus

Moisture)

BMI Resin (Three Point.

Cure Progression)
BMI/IM7 (Four Point)

F lexure Strength
Flexure Failure Strain

 

 

 

 

Iosipescu Shear BMI/IM7 Shear Modulus
Shear Strength
Shear Failure Strain
Isothermal Dynamic BMI Resin 30°C Storage Modulus
Mechanical - BMI/1M7 30°C Loss Modulus
Three Point Bend 30°C Tan Delta
Tg
Transient Dynamic BMI Resin 155°C Storage Modulus
Mechanical - BMI/IM7 155°C Loss Modulus

Single Cantilever

 

 

155°C Tan Delta
Tg

 

 

3.3.1 Hygrothermal Moisture Conditioning

Both BMI resin and composites were tested. Separate conditioning runs were performed

for these two specimen types, and are reviewed here.

3.3.1.1 BMI Composite Moisture Conditioning

BMI composite specimens from panels H, J, W, I, K, and V were conditioned for testing.

Both saturated ("Wet") and control specimens ("Dry") were tested. The fabrication of

these panels was documented in Section 3.1.2.

47
Specimens were cut from the panels using the Bennett 5000 Plate Saw with a water based
coolant, rinsed in clean tap water, and dried with paper towels. All specimens were then
dried in a vacuum oven (29 in. Hg vacuum) at 80°C for three weeks. After that time, the
specimens were removed and the "Dry" specimens were placed in nitrogen purged zip-

lock bags (doubled for insurance against punctures and leaks).

The "Wet" specimens were placed in the Blue-M hygrothermal chamber on glass racks.
They received 5 weeks of conditioning at 77°C (72°C wet bulb). The temperature was
then dropped to 42°C and the relative humidity was kept at 100% to ensure full
saturation. Subsequent moisture diffusion studies of BMI/IM7 cross-ply laminates have
shown that BMI cross-ply composites require very little time to reach saturation in 100%

humidity (see Chapter 2).

The specimens were removed after approximately 12 months in the chamber. They were
quenched into deionized water, weighed, and tested immediately. Specimens were kept

saturated between moist towels for the short period between removal and actual testing.

3.3.1.2 BMI Resin Moisture Conditioning

Panel AZ was used for hygrothermal testing of BMI resin. Fabrication of this panel was
described earlier. The panel was cut into 3" x 2/5" specimens for flexure testing, and 1" x
2/5" specimens for DMA testing. The Bennett 5000 Plate Saw was used with a water-

based coolant. Specimens were subsequently polished to a thickness of 3.6-3.7 mm using

48
wet polishing techniques. They were then weighed and dried in a vacuum oven for two
days. The "dry" specimens were returned to the vacuum oven and received a thermal
treatment at 80°C for two days to completely dry them before testing. The "wet"
specimens were saturated in a pressure cooker at 98°C and 1 atmosphere for 24 days.
They were quenched in a room temperature bath of deionized water, weighed, and tested.

During testing, untested specimens were kept saturated between soaked towels.

3.3.2 Resin Flexure Testing

Three point flexure tests of neat resin BMI specimens were performed to determine the
change in neat resin properties due to moisture saturation. Specimens measuring 3" x
2/5" were cut from resin plaque AZ. The fabrication of this plaque and conditioning of

these specimens were detailed in earlier.

Testing was performed according to ASTM standard D790 for three point flexure testing.
The test speed was 0.05 in/min with a span of 2 inches. A two pound preload was used.
The moisture content of the specimens was determined both by weight and by Thermal
Gravimetric Analysis (TGA). One piece each of a dry and a wet specimen were run in
the TGA with a ramp rate of 5°C/min in flowing nitrogen. The resultant weight vs.
temperature curves are provided in Figure 3-10 and Figure 3-11. The average weight
gain for the wet specimens (by weight measurement) was 4.6 wt% while the dry
specimens showed no weight change and negligible evidence of moisture in the TGA

CUI'VC.

100

 

 

99-

98—

974

96‘

Height 1%)

 

 

 

c . 50 . 160 V 150 ,. 200 1 250 . 300 . 350 ' aco
Temperature (°C)

Figure 3-10 TGA of Dry BMI Resin Showing Negligible Moisture

100

 

98—

96-

Netght (x)

94H

92—

90 f l V I V I v l v j v I 7 r v
0 50 100 150 200 250 300 350 460
Temperature (’C)

 

 

 

Figure 3-11 TGA of Wet BMI Resin Specimen Showing 4.5% Moisture

50
The load/displacement data from the tests was logged using the UTS test stand’s
computer. The file was then analyzed using an Excel macro written expressly for this
purpose. The macro performed calculations according to the ASTM stande and plotted
the resultant modulus as a straight line shifted from the actual curve with a height
corresponding to the failure stress. Table 3-8 summarizes the results of this testing.

Summary data for the individual tests has been published elsewhere [18].

Table 3-8 Hygrothermal Resin Flexural Results

 

Flex Strength Std. Dev. Flex Modulus Std. Dev. Fail Strain Std. Dev.

 

Condition (MPa) (MPa) (%)
Dry Resin " 151.0 11.84 3571 710.77 71535:: ” “0.51 ”
Wet Resin 116.9 5.25 3613 57.58 3.311 0.14

 

 

 

Both dry and wet specimens were analyzed in the ESEM. No significant differences in
the failure mode was found. Two major morphologies were seen, and are presented in
Figure 3-12. The upper images show a rough failure with surface damage evident. The

lower images show progression of the crack leaving 'strings’ in its path.

51

 

"Dry" Resin "Wet" Resin

Figure 3-12 Failure Micrographs of Dry and Wet Resin Flex Specimens

3.3.3 Hygrothermal Composite Flexure Testing

Four point flexure tests of BMI/IM7 composite specimens were performed to determine
the change in composite properties due to moisture saturation, extent of cure, and fiber

volume fraction.

Specimens were cut from composite panels I, K, and V. Panels I and K were 62% Vf
while V was 35% Vf. All panels had a [00/900143 configuration. Four inch by one inch
specimens were cut from these panels using the Bennett 5000 Plate Saw. The fabrication

and conditioning of these specimens was described in Sections 3.1.2 and 3.3.1

The average weight gain for the wet specimens (by weight measurement) was 1.5 wt%
for low cure 62% Vf specimens. 2.1 wt% for the high cure 62% Vf specimens, 2.1 wt%
for low cure 35% Vf specimens, and 3.4 wt% for the high cure 35% Vf specimens. The

dry specimens showed no weight change.

Testing was performed according to ASTM standard D790 for four point flexure testing.
The test speed for the 62% Vf specimens was 0.15 in/min while the 35% Vf specimens
had a test speed of 0.075 in/min. Quarter-point loading was used with a span of 3 inches

and a two pound preload.

The load/displacement data for each test was logged using the UTS test stand's computer.
The file was then analyzed using an Excel macro written expressly for this purpose. The
macro performed calculations according to the ASTM standard and plotted the resultant
modulus as a straight line shifted from the curve. The height of this modulus line is the
failure stress. These plots and summary data for each test have been published elsewhere.

Summaries of the results for each condition are provided in Table 3-9 and Table 3-10.

53

Table 3-9 Hygrothermal BMI Composite F lexure Data - 62% Vf

 

Flex Strength Std. Dev. Flex Modulus Std. Dev. Fail Strain Std. Dev.

Condition (MPa) (MPa) (%)
lry ow ure . ° . . .
Dry High Cure 765 19.1 87511 1595 0.891 0.011

      

Wet Low Cure 900 29.9 90763 1989 1.050 0.041
Wet High Cure 777 52.0 87801 176 0.920 0.072

 

 

 

Table 3-10 Hygrothermal BMI Composite F lexure Data - 35% Vf

 

Flex Strength Std. Dev. Flex Modulus Std. Dev. Fail Strain Std. Dev.
Condition (MPa) (MPa) (%)

  
     

'1

Dry ow ure

Dry High Cure 460.0 26.7 49543 1385 1.070 0.072
Wet Low Cure 692.2 25.4 53359 501 1.336 0.035
Wet High Cure 512.4 49.8 48929 1433 1.150 0.128

 

 

 

These specimens failed in a few different manners; either a shear mode at the composite
mid-plane, or shear through the thickness under the load nose as shown in Figure 3-13.
These are not desired failure modes, and this fact alone provides a good deal of
information about the system. The mid-plane shear failure mode is similar to that of a
short beam shear test which is considered a good measurement of interlaminar shear

properties.

54

 

 

 

 

 

 

Figure 3-13 Mid-Plane Shear and Load Nose Shear Failures for Flexure Tests

The failure mode of these flex tests was generally either shear in the mid-plane, or shear
through the thickness under the load nose. Some compressive failure on the upper
surface due to the load noses was also seen, but was not the major failure mode. The
failure mode shifted toward load nose failure for the partially cured specimens, but many
partially cured specimens showed mid-plane shear failures, or a combined failure, and
some fully cured specimens showed load nose failures. No trends were noted between

the wet and dry specimens.

The ESEM was used to analyze the failure surface of the specimens which showed mid-
plane shear. The mid-plane shear failure generally traversed from one load nose
completely across the specimen to the opposite side, leaving the specimen intact. To
view the failed surface, it was necessary to complete the failure. This was accomplished
by simply pulling the specimen apart. The section which was failed during the actual test

was marked to ensure that the "opened" section was not analyzed.

55
No differences between the 62% and 35% volume fraction specimens were seen on the
microscopic level. Similarly, there were no differences between the dry and wet

specimens.

The partially cured specimens generally showed a failure surface which consisted of both
adhesive failure of the fiber/matrix interface, and hackled regions where the failure
deviated into the resin as seen in Figure 3-14. The hackles are the small vertical lines
while the horizontal lines represent adhesive failures where fibers used to be. The fully
cured specimens showed only the adhesive failure as seen in Figure 3-15. The large
vertical lines in both figures are the transverse microcracks discussed in Chapter 3. The
adhesive failure seen here is epitomized by the failure shown in Figure 3-16. The fiber

runs horizontally across the center of the image and is devoid of adhering resin.

56

tillm—-3"1i'

_ ,1 {'15. 1:111 11 1 1‘1 1 with“...

 

Figure 3-14 ESEM Micrograph of a Partially Cured Dry Specimen Showing
"Hackles"

57

FLHCI‘lF‘Eil . TIF

 

Figure 3-15 ESEM Micrograph of a Fully Cured Dry Specimen Showing No
"Hackles"

58

F‘hili

 

Figure 3-16 Adhesive Failure of a Fully Cured Dry Specimen

59

3.3.4 Isothermal DMA Testing

Isothermal DMA tests of BMI resin and BMI/IM7 composites were performed at 30°C to
determine the change in properties due to moisture saturation, extent of cure, and fiber

volume fraction.

Specimens were cut from composite panels H, J, and W which had fiber volume fractions
of 62%, 55%, and 47% respectively. All panels had a configuration of [0°/90°]S. The
fabrication and conditioning of these specimens is detailed in Sections 3.1.2 and 3.3.1.
BMI resin specimens were prepared and conditioned identically to those used in the resin

flexure testing.

Testing was performed using the Rheometrics Mark III DMTA in an unclamped three
point bend mode with the large frame. Liquid nitrogen was used to maintain a constant

temperature of 30°C. A strain of 16 um was used and data was taken at many frequencies

(100, 50, 30, 20, 10,5, 3,2, 1, 0.3, 0.2. and 0.1 Hz).

Time, temperature, tan delta, and the log of the storage modulus were exported from the
DMA and summarized using Excel macros. Plots of tan delta and the log of the storage
and loss moduli vs. frequency for each specimen have been published elsewhere. A
typical curve is shown in Figure 3-17 for a dry, 47% specimen. Table 3-1 1 and Table 3-

12 summarize the 1 Hz data for the composite and resin specimens.

Log (Modulus)

10.5 »

10

9.5

8.5

7.5 »

60

 

0.0120

 

0.0115

0.0110

 

, 0.0105

0.0100

 

« 0.0095

. 0.0090

 

0.0085

 

 

 

A ‘ * 0.0080
0 5 10 15 20 25 30 35 40 45 50

Frequency (Hz)

-WO-Log(Ei‘(Pa))7 -o—Log(E"(Pa))7 -I:-Tan Delta ‘

Figure 3-17 Isothermal DMA Curve for Fully Cured 47 % Vf Specimen

Tan Delta

 

61

Table 3-11 - Hygrothermal BMI Composite DMA Data - 1 Hz

 

 

 

 

 

Bending Bending
Storage Loss
Modulus Std. Dev. Modulus Std. Dev. Tan Delta Std. Dev.
Condition (MPa) (MPa) (MPa) (MPa) (x103) (x103)
. ‘.' ow ure,Iry .;. ° I 1 '1. I I.°
62% Vf, High Cure, Dry 96014 3199 475 73 4.970 0.616
62% Vf, Low Cure, Wet 69130 13530 508 93 7.703 2.904
62% Vf, High Cure, Wet 91201 6131 603 130 6.703 1.372
55% Vf, Low Cure, Dry 66527 12345 555 166 8.583 2.535
55% Vf, High Cure, Dry 67143 2244 341 31 5.083 0.376
55% Vf, Low Cure, Wet 66732 7548 516 351 8.307 4.066
55% Vf, High Cure, Wet 87364 10171 517 59 5.967 0.939
47% Vf, Low Cure, Dry 45116 7792 342 20 7.737 1.895
47% Vf, High Cure, Dry 40241 3955 382 49 9.660 2.165
47% Vf, Low Cure, Wet 49621 6359 386 11 7.797 0.777
47% Vf, High Cure, Wet 45082 4578 435 81 9.680 0.847
Table 3-12 Hygrothermal BMI Resin DMA Data - 1 Hz
Bending Bending
Storage Loss
Modulus Std. Dev. Modulus Std. Dev. Tan Delta Std. Dev.
Condition (MPa) (MPa) (MPa) (MPa) (x103) (x103)
esm, an ar ure, ry 4 W
BMI Resin, Standard Cure, Wet 2040 55 63 11 31.453 6.014

 

 

 

 

62

3.3.5 Interfacial Adhesion Testing

Interfacial testing of adhesion (ITS) of BMI/IM7 composites was performed to determine
the change in adhesion levels due to moisture saturation, extent ofcure. and fiber volume

fraction.

3.3.5.1 Hygrothermal Materials & Conditioning

Specimens were cut from composite panels H, J, and W which had fiber volume fractions
of 62%, 55%, and 47% respectively. All panels had a configuration of [0°/90°]S. The
fabrication and conditioning of these specimens is detailed in Sections 3.1.2 and 3.3.1.
The specimens were mounted in epoxy and polished using standard wet polishing
techniques. The "Dry" specimens were then dried in a vacuum oven at 80°C and stored
in nitrogen purged bags until tested. The "Wet" specimens were placed in the

hygrothermal chamber with the other "Wet" samples.

3.3.5.2 Specimen Testing

Specimens were tested using the CMSC's ITS system using a new controlling computer
and new software which retains all load-displacement curves and stores images of each
loading cycle. The ITS system measures the adhesion in real composite specimens by
loading the polished faces of fibers with a hemispherical diamond indentor as depicted in

Figure 3-18.

 

63

 

 

 

 

 

Figure 3-18 The Loading of Fiber in the ITS Test

During the test, the specimen is moved toward the tip using stepper motors in 0.04 pm
increments. The applied load is measured using a Sartorius balance interfaced with the
controlling computer. The balance has a resolution of 0.01g. The debonding load is
determined by using incremental loading steps and visually checking for a debond
between each loading using an attached light microscope at magnifications of up to
2000x. A debond appears as a black ring around the fiber, and for these tests, a failure
was defined as a debond of at least 120° around the fiber. Twelve fibers were tested for

each condition, and the high and low values were rejected before data analysis.

64
For the "Wet" specimens, a plastic bag was placed around the sample holder and coupled
to the indentor/optic collar. A small quantity of water was placed in the bottom of the
sample holder, and the bag was sealed. This caused a saturated atmosphere around the
specimens. Because the bag was now coupling the balance to the indentor, the ITS
program was altered so that the loading cycle could be stopped at various times during
the test initialization to re-tare the balance and thus get a test with little contribution from

the coupling effect. This method was very successful.

3.3.5.3 Data Reduction

The interfacial shear strength (IFSS) is calculated within the ITS software using a
formula obtained from a finite element analysis of this loading condition. The formula
takes into account the distance between the tested fiber and it's nearest neighbor as well
as the tensile modulus of the fiber. shear modulus of the matrix. and failure load. The

formula is shown in Eqn. 3-1.

I ‘

 

L 72 El" 3 D
0.875696 ? —0.018626L0g 3 —0.026496

IFSS(psi)=1811000(d2
,/

Eqn. 3-1

where:

L = max load in (g)

d = fiber diameter in (um)

Em = matrix shear modulus (psi)
Ef= fiber tensile modulus (psi)
D = interfiber distance (um)

65
The IFSS results for the hygrothermal testing are provided in tabular format in Table 3-

13. The results from the improvement attempts are provided in Table 3-14.

Table 3-13 ITS Results for Hygrothermal Testing

 

 

 

 

 

 

 

 

 

 

 

 

DRY th
47% Partial 55% Partial 52% Partial 47% Full 55% Full 52% Full 47% Partial 55% Partial 62% Partial.
Wins—653 91 1 mg 743 146 453 SF‘
Std. Dev. 55 129 172 139 102 115 33 95 123
Table 3-14 ITS Results for Improvement Attempts
DRY
RTV Sized PES Sized PEI Sized 5218 Sized Air Plasma Control 400% ST
ii§§ip$li 30" ‘ 5s '50 ' .
Std. Dev. 230 203 164 231 207 247 260

 

 

 

 

3.3.6 The Effect of Cure Progression on the Physical Properties of BMI Resin

Complex thermoset resins like this BMI system have properties which are highly
dependent upon their full thermal history, not just their extent of cure. To gain insight
into this, resin specimens were made with varied thermal histories and their macroscopic

tensile and flexure properties were determined along with their Tg, as measured using

modulated DSC.

66
Neat resin specimens were made using the procedure outlined in Section 3.1.1. All resin
plaques were initially cured at 177°C for 1 hour. The subsequent post-cure cycles used

are shown in Table 3—15.

Table 3-15 Neat Resin Post-Cure Cycles

 

 

 

 

 

 

 

Post-Cure Post-Cure Cycle
Description (Initial Cure: 177°C/l hr)
Low Cure Short 200°C/1 hr
Low Cure Long 200°C/3 hrs
200°C/2 hrs,
Standard Cure 250°C/l hr
Short
200°C/2 hrs.
Standard Cure 250°C/6 hrs
Long
200°C/2 hrs,
High Cure Short 250°C/6 hrs,
300°C/1 hr
200°C/2 hrs,
High Cure Long 250°C/6 hrs,
300°C/6 hrs

 

 

 

 

Specimens were machined from the post-cured resin plaques using a diamond saw with a
water-based coolant. Since the outer layer of the specimens was contaminated with a
release agent, approximately 0.5 mm of material was removed using wet polishing
techniques. After polishing, the specimens were dried in a vacuum oven at 80°C for one

week to ensure they were equally dry before testing.

67
Three point flexure testing was performed using a 50 mm span and a test speed of 1.3
mm/min. Specimens were nominally 75 mm x 10 mm x 3.5 mm. Tensile tests were
performed using a 'dog bone' specimen with a gauge length of 65 mm and a cross-section
of nominally 10mm. Tests were performed at a cross-head speed of 5 mm/min. A
representative stress-strain curve is shown in Figure 3-19 for a tensile test. The flexure

samples were similarly brittle.

 

STRESS (M Pa)
8
‘3

 

 

 

000 j I I I I I
0.00 0.01 0.02 0.03 0.04

STRAIN (mm / mm)

Figure 3-19 Stress - Strain Curve for a Typical Resin Tensile Test

Tg was measured using a modulated DSC with an overall ramp rate of 5°C/min and a one
cycle/min modulation of i1°C. A typical DSC curve is shown in Figure 3-20. Tg was
taken to be the point at which there was a step change in the reversible heat loss. A step

change in the heat capacity, CD. was also seen at this temperature as expected. While

some curing could have occurred during the actual test, artificially raising the Tg, the shift

68

should be minimal, and the interest here is in the trends, not specific values.

.08

 

 

 

   

 

 

 

0.20
///
/
///' / -0 18
/
\ /
-0.10~ \\ 1.8—
~0.05 \
- A f0.16 A
U 01
'51 160 35°C / °\ i E
2 . '5. , 3 V
v '0.12_‘ E "I 31.6“ 014 i
E V ' / x . 2
S E . > u.
u. “0.00 O / / I: u
.. I ,. g .3
'0 Q h"0.12 I
:°:’ -0.14« E ”50 34° / 8 1.41 >
m
21’ 71 \ , g g
// \d—\ I —0.10 2
“'0 O /’/ \ / I/ll
,/ g/‘f” \ _
‘0 16“ . //——/ \\ 1.2“
.. J"
'0, 18 r I I I m . r r 0.06
0 50 100 150 200 250 300 350

Temperature* (°C)

Figure 3-20 Modulated DSC Curve Showing Tg Determination

Data reduction for these flexure, tensile, and DSC tests was performed at The Advanced
Materials Engineering Experiment Station (AMEES) in Midland, MI by colleagues.

Tabular results for all testing conditions are provided in Appendix A.

69

3.3.7 Moisture Diffusion Study

A major objective of this research has been to understand the effect of moisture on the
properties of the BMI and K3B composites. To effectively utilize the information
obtained from the physical property testing, the moisture absorption behavior of these
systems must be understood. To this end, moisture absorption studies were performed to
determine the ultimate moisture content and diffusion coefficients for BMI and K3B
resins and their composites. The effect of extent of resin cure, fiber presence, and
microcracking was determined for the BMI/IM7 system. Two temperatures were used so
that the diffusivity constants, D” and Q, can be calculated allowing knowledge of the

actual diffirsivity, D.

The methods used to create the resin and composite panels used in this study are detailed
in Section 3.1. Specimens were cut to a nominal size of 2" x 1" except for the K38 resin
which was nominally 2" x 2/5". All composite and neat resin specimens were dried in a
vacuum oven at 80°C under 29.5 in. Hg vacuum for approximately 340 hours, except for

K3B cross-ply composites, which were dried at the same conditions for approximately

240 hours.

Specimens were saturated in a pressure cooker at the desired temperature (100°C or
70°C) on glass racks above the water level. The chamber was open to the atmosphere
through the open pressure release valve, assuring a pressure of one atmosphere. Weight

measurements were made by opening the chamber. quenching the rack and specimens in

70
a large bath of room temperature deionized water. drying the specimens with paper
towels, and weighing them on a Sartorius balance (+/— 0.0001g). The specimens were
then returned to the chamber. Measurement time was removed from calculated exposure

time.

All composite data is provided both as is. and corrected for fiber volume fraction
assuming the fibers to play no role in the diffusion. Absolute weight gain data was
converted to percent weight gain vs. time. This data was then fit to a Fickian diffusion
model suggested by Shen and Springer [19]. The model assumes two dimensional
diffusion through the major faces of a thin sheet and is mathematically represented by
Eqn. 3-2. A correction for "thick" specimens was used which adds the area of the edges
to that of the faces. The model assumes that initially there is no interaction between the
absorption from the edges and faces. Thus. the diffusivity is calculated during the initial
linear region of the curve, and the ultimate moisture content is approached asymptotically
at long times. A plot of the moisture gain of one specimen showing the F ickian nature of
these materials is provided in Chapter 5. The plot also shows a region where the

diffusion becomes non-Fickian due to the onset of microcracking.

, “All. Eqn. 3-2

 

Where:

M, = The moisture content at time t
M a, = the ultimate moisture content (t =00)

71
t = time in seconds
D = the diffusion coefficient
h = specimen thickness
C = an adjustable parameter

The data was fit to this equation using a regression analysis with three adjustable

parameters: Mt, M and C. The resultant fit was generally very good with r2 values

,1) ’

ranging from 0.996 to 0999.

Since experiments were performed at two temperatures, the diffusion coefficient can be
calculated as a function of temperature according to Eqn. 3-3. Q is calculated from the

slope of the ln(D) vs. l/T plot, and ln(D()) is the ordinal intercept.

—()

D = Doe RT Eqn. 3’3

Table 3-16 summarizes the whole moisture absorption study. Values of the adjustable
parameter, "C", and R2 are provided in Appendix B. Plots of the original data and curve

fit for each specimen and condition have been published elsewhere [18].

72

Table 3-16 Summary Data for Moisture Absorption Study

 

 

 

 

 

Maximum Maximum Diffusivity Diffusivity Q 0,,
Condition Moisture % Moisture % % Change (mm’lsec) (mmzlsec) (JlmoIK) (mm’lsec)
70'C 100°C 70°C-100°C 700 1000
mamure 3.3117—‘3'7567—12'5TU7—‘2'350m7—‘7‘ T—T '. —_ —._ "._,— ”.— 47380 :iirg—sisrlri‘=
BMI Resin Full Cure 5.14% 5.00% -2.70% 5.196E-08 1.166E-07 28658 1.203E-03
BMI X-ply 1.55% 1 57% 1.34% 1.879E-06 1.222E-06 -15255 8.927E-09
BMI Unidirectional 2.49% 2 32% -7 09% 3,206E-08 4.258E-08 10060 1.092E-06
K38 Resin 1.53% 1 62% 5.02% 9.998E-07 2.322E-06 29876 3.547E-02
K3B-IM7 X-ply 0.55% O 54% 4.86% 3383508 9.427E-08 36336 1.156E-02
100% Resin Equivalent for Composite Data
WX-ply VfCor 1 05% 1,05% -1 550/:
BMI X-ply VfCor 4.21% 4.27% 1.34%
BMI Unidirectional 5.10% 4 96% -2.72°/o

 

 

3.4 Combined Hygrothermal and Thermal Spike Experiments

Iosipescu shear testing and transient DMA testing for the hygrothermal and thermal spike

studies were performed together. The methods used are presented here.

3.4.1 Iosipescu Shear Testing

Iosipescu shear tests were performed on cross-ply composites to determine the change in

shear properties due to hygrothermal and thermal spike conditioning.

3.4.1.1 Specimen Details

Hygrothermal specimens were cut from composite panels I and V. Panel I was 62% Vf

while V was 35% Vf. All panels had a [0°/90°]_,S configuration. The panel fabrication

and conditioning of these specimens is detailed in Sections 3.1.2 and 3.3.1.

73
Upon removal from the hygrothermal chamber, the "wet" specimens were quenched into
deionized water and kept saturated between moist towels until +45°/-45° strain gauges
were attached to both sides. For these wet specimens, soldering of the leads was done
away from of the specimen to prevent specimen drying. Tests were performed
immediately after application of the gauges for the wet specimens, but the "dry"

specimens were gauged ahead of time and stored under vacuum until tested.

The average weight gain for the wet specimens (by weight measurement) was 1.9 wt%
for low cure 62% Vf specimens. 2.4% Vf for the high cure 62% Vf specimens, 2.3 wt%
for low cure 35% Vf specimens, and 3.8 wt% for the high cure 35% Vf specimens. The

dry specimens showed no weight change.

Thermal Spike BMI/IM7 composite specimens were cut from composite panel Y which
had a configuration of [00/900143 and a 47% vf. K3B/IM7 composite specimens were cut

from panel 488 which was also [00/900145-

3.4.1.2 Testing and Analysis

All specimens were 3" x 3A". Square shear notches were 0.15" deep with a root radius of
0.050". Testing was performed using a hydraulic MTS testing stand and a modified
Wyoming fixture. The test speed was 0.05 in/min and data was stored by the controlling
computer at 10 Hz. Strain gauges were set up using one of two methods: either four

quarter bridges, or two quarter bridges and one half bridge.

74

The data was analyzed using a variety of Excel macros. Ultimately, plots of shear stress
vs. shear strain were produced for each strain gauge. The macro calculated a shear
modulus using the elastic region between 0.2% and 0.6% strain. A failure shear stress
was calculated using a 0.2% strain offset criteria. This was also calculated using a macro,

and the resultant offset curve was plotted on the shear stress vs. shear strain curve.

To get accurate shear moduli and strengths, interpretation of the curves was necessary
because not all of the gauges worked on all specimens. Visual inspection of the curves
was performed, and final values were averages from the gauges which performed
satisfactorily. The calculation used for each test is provided in the summary above the
curves for each specimen. These curves and summaries have been published elsewhere
[18]. The shear moduli, strengths. and failure strains for each hygrothermal condition are
summarized Table 3-17 and Table 3-18. The results for the thermal spike tests are

provided in Table 3-19 and Table 3-20.

Table 3-17 Hygrothermal BMI Composite Shear Data - 62% VI

75

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Shear Strength Std. Dev. Shear Modulus Std. Dev. Fail Strain Std. Dev.
Condition (We) (MPa) (%)

W37 56.20 2.37 5360 163 1.230 0.020
High Cure Dry 45.97 1.07 4470 66 1.198 0.014
Low Cure Wet 38.93 21.82 5273 158 1.155 0.104
High Cure Wet 41.63 2.18 4381 264 1.117 0.033

Table 3-18 Hygrothermal BMI Composite Shear Data - 35% Vf
Shear Strength Std. Dev. Shear Modulus Std. Dev. Fail Strain Std. Dev.
Condition (MPa) (MPa) (% )
l—oWT—u—‘F—myrm 3.64 348$ 14 1.667 021075=
High Cure Dry 35.89 1.41 2878 152 1.462 0.005
Low Cure Wet 45.85 1.80 3782 40 1.384 0.060
High Cure Wet 33.81 0.77 2733 259 1.450 0.083
Table 3-19 Thermal Spiked BMI Composite Shear Data
Shear Strength Std. Dev. Shear Modulus Std. Dev. Fail Strain Std. Dev.
Condition (MPa) (MPa) (%)
ry on ro . 0.98 4230 120 3.279 0.022
Wet Control 46.04 0.57 4206 111 1.272 0.022
Dry Cycled 43.63 0.42 3979 351 1.281 0.100
Wet Cycled 46.38 2.78 4272 365 1.266 0.051
Table 3-20 Thermal Spiked K3B Composite Shear Data
Shear Strength Std. Dev. Shear Modulus Std. Dev. Fail Strain Std. Dev.
Condition (MPa) (MPa) (%)

W01 47.94 0.33 4873 102 1.155 0.019

Wet Control 48.02 2.49 5025 167 1.141 0.030
Dry Cycled 48.22 1.70 4725 214 1.209 0.118
Wet Cycled Blistered

 

 

 

 

76

3.4.2 Transient DMA Testing

Transient (temperature sweep) DMA tests of both resins and composites were performed

to determine the change in TE due to both hygrothermal and thermal spike conditioning.

The hygrothermal specimens used here were the same ones used for the isothermal tests
in Section 3.3.4. After isothermal testing, the specimens were cut to fit the clamped
single cantilever fixture using a Dremel tool with an abrasive blade for the dry specimens,
and a wet saw with a diamond impregnated blade for the wet specimens. The wet

specimens were kept saturated between moist towels until tested.

The thermal spike BMI/IM7 composite specimens were cut from composite panel X
which was [0°/90°]4S and 47% Vf. Sixteen 3 in x '/2 in specimens were machined with the
Bennett saw and dried in a vacuum oven. BMI resin specimens were cut from two
separate panels: TSR and 2TSR. The fabrication and post-cure of these panels is
described in Section 3.1.1. K3B/IM7 composite specimens were cut from panel 488
using the Bennett saw. The K3B resin was cut from the plaque created at MSU. After
conditioning, the thermal spike specimens were cut to fit the DMA's clamped single
cantilever fixture using a Dremel tool with an abrasive blade. No cooling water was

used, and the amount of heat evolved during cutting was negligible.

Testing was performed using the Rheometrics Mark III DMTA in a clamped single

cantilever mode with the large frame. An air atmosphere was used with a strain of 32um

77
and a 1 Hz frequency. The temperature was ramped at 5°C/min from 100°C to above Tg

(between 300°C and 400°C).

NOTE: The temperature measured by this machine is significantly inaccurate. The
measured temperatures differ from part temperature by as much as 30°C due to thermal
lag. An experiment was performed with three thermocouples taped onto a resin
specimen, and this very significant thermal gradient was seen. Fortunately, the readings
are very precise, if not accurate. so the repeatability of the tests is very good, and the data

is acceptable for comparative studies.

Time, temperature, tan delta, and the log of the storage modulus were exported from the
DMA and summarized using Excel macros. Plots of tan delta and the log of the storage
and loss moduli vs. temperature have been provided elsewhere. A representative curve is
provided in Figure 3-21 for a dry 47% Vf specimen. Table 3-21 and Table 3-22
summarize the Tg values and moduli at 155°C for the hygrothermal specimens. Table 3-
23 through Table 3-26 summarize the thermal spike results. The Tg values were
measured by the peak of the tan delta curve (calculated by the macro), and from the

extrapolated onset of property loss (calculated using the DMA software).

Log Modulus

78

DMTA Results - WA26

 

 

 

 

 

 

10.5 0.25
10
0.2
9.5
/ _ 0.15
9 . I . \
e' I . ‘-.\
. I .\~\. .
' ’ \ . "' - ---
_ .o I \
8.5 _______ I \
I ‘\ 0.1
// \\
I \\\
8 I s
I
I
,’ 0.05
I
7.5 ’l
I
’/
7 0
150 170 190 210 230 250 270 290 310
Temperature (°C)

 

Log (E') ..... Log (E") _ _ - Tan Delta

Figure 3-21 Transient DMA Curve for Partially Cured 47% Vf Specimen

Tan Delta

79

Table 3-21 Hygrothermal BMI Composite DMA Data - Transient

 

 

155°C 155°C

Bending Bending

Storage Loss

Modulus Modulus Tan Delta
Condition (MPa) (MPa)
o'Vo,ow ure,Iry . ° ;
62% Vf, High Cure, Dry 36898 317
62% Vf, Low Cure, Wet 27574 994
62% Vf, High Cure, Wet 42855 519
55% Vf, Low Cure, Dry 36141 352
55% Vf, High Cure, Dry 34995 325
47% Vf, Low Cure, Dry 21330 292
47% Vf, High Cure, Dry 23496 301
47% Vf, Low Cure, Wet 21752 1354
47% Vf, High Cure, Wet 15031 354

(x103)
._ i 1
8.580
39.300
12.100
9.750
9.290
13.700
12.800
63.125
23.530

of Tan
Delta)

(°C)
399
243
374
226
358
235
366

206
365

T9 (Peak Tg (Onset

of Loss of
5')
(°C)
I
356
160
337
195
334
200
351
7??
347

 

Table 3-22 Hygrothermal BMI Resin DMA Data - Transient

 

155°C 155°C
Bending Bending
Storage Loss
Modulus
Condition (MPa) (MPa)
Iry - verage . ,.
Dry - Std. Dev. 35.0 0.7
Wet - Average 1312 38.9
Wet - Std. Dev. 36.5 1.5

 

Modulus Tan Delta

(x103)

0.104

29.673

1.845

Tg (Peak

of Tan
Delta)

(°C)
0.6

328
3.9

T9 (Onset
of Loss of
E')
(°C)

2.1

 

 

 

80

Table 3-23 Thermal Spike K3B/IM7 DMA Data - Transient

 

Condition

Wet Control
Dry Cycled

Wet Cycled

 

Std. Dev. 0.63 3 1 1 9

Average

Std. Dev.

Average

Std. Dev.

155°C
Bending
Storage
Modulus

(MPa)

471
8931
154
9090
1 108
Blistered

 

155°C Tg
Bending Tg (Peak (Onset 0
Loss of Tan Loss of
Modulus Tan Delta Delta) E')
(MPa) (x103) (°C) (°C)
2.75 . .
89.81 10.06 253.9 236.8
4.21 0.42 0.7 0.3
94.97 10.47 253.8 237.1
16.37 0.80 0.7 0.7

 

 

Table 3-24 Thermal Spike BMI/IM7 DMA Data - Transient

 

Condition

Wet Control
Dry Cycled

Wet Cycled

 

 

Dry ConIrol Average
Std. Dev.

Average

Std. Dev.

Average

Std. Dev.

Average

Std. Dev.

155°C 155°C Tg J
Bending Bending Tg (Peak (Onset 0
Storage Loss of Tan Loss of
Modulus Modulus Tan Delta Delta) E')

(MPa) (MPa) (x103) (°C) (°C)

1547 6.37 3.60 3.6 4.5

6613 82.99 12.59 328.9 289.7

899 6.17 1.16 2.3 2.9
7277 70.76 9.75 339.2 306.7
828 13.06 0.78 1.1 0.8
5918 70.03 11.88 335.9 301.9
961 8.37 1.15 5.3 5.9

 

 

81

Table 3-25 Thermal Spike K3B Resin DMA Data - Transient

 

155°C
Bending
Storage
Modulus
Condition

 

I ry on ro ‘ verage
Std. Dev 225
Wet Control Average 2443
Std. Dev. 39
Dry Cycled Average 2302
Std. Dev. 110
Wet Cycled Average 2135
Std. Dev. 254

 

    

. 0.97 8.4 1.3

    
   
  
   

155°C Tg
Bending Tg (Peak (Onset of
Loss of Tan Loss of
Modulus Tan Delta Delta) E')
(x103) (°C)

(MPa) (°C)

  

     

3.01

43.65 17.87 266.9 243.2
0.33 0.42 2.3 0.6
50.08 21.76 267.4 243.0
1.62 0.33 0.1 1.5
53.91 25.28 269.3 243.2
3.29 1.45 4.5 1.6

 

 

Table 3-26 Thermal Spike BMI Resin DMA Data - Transient

 

155°C
Bending
Storage
Modulus
Condition (MPa)
Iry on ro verage
Std. Dev. 56
Wet Control Average 1066
Std. Dev. 130
Dry Cycled Average 1412
Std. Dev. 133
Wet Cycled Average 1475
Std. Dev. 22

 

 

155°C Tg
Bending Tg (Peak (Onset of
Loss of Tan Loss of
Modulus Tan Delta Delta) E')
(°C)

  
  

   

(MPa)

(x103) (°C)

    

1.74 1 1 165.5

23.91 22.46 325.8 278.9
1.10 1.70 0.4 5.6
32.75 23.49 330.5 197.1
3.22 4.71 9.3 170.8
40.36 27.50 332.0 290.2
5.16 3.92 9.0 0.5

 

 

82

3.5 Thermal Spike Experiments

The objectives of the thermal spike experiments were to identify any permanent damage
due to moisture exposure and thermal cycling. The properties of interest were modulus,
strength, failure strain, and retention of composite toughness. The test scheme used is

shown in Table 3—27.

Table 3-27 Thermal Spike Testing Scheme

 

 

 

Test Performed Materials Studied Properties Reported
Tensile BMI Resin Modulus, Strength,
Failure Strain
Flexure BMI Resin (Three Point) Flexure Modulus
BMI/IM7 (Four Point) F lexure Strength
K3B/IM7 (Four Point) Flexure Failure Strain
Open Hole Compression BMI/IM7 Initial Stiffness
K3B/IM7 Strength
Iosipescu Shear BMI/IM7 Shear Modulus
K3B/IM7 Shear Strength
Shear Failure Strain
Dynamic Mechanical BMI Resin 155°C Storage Modulus
K3B Resin 155°C Loss Modulus
BMI/IM7 155°C Tan Delta
K3B/IM7 Tg

 

This testing was organized with a 22 factorial design [20] to show both the main effects of
moisture exposure and thermal spikes, as well as their interaction effects. The test
procedures, including data reduction and results, are presented here, and a discussion of

these results including the main and interaction effects is presented in Chapter 6.

 

83
3.5.1 Thermal Spike Conditioning

Both neat resin and composites of BMI and K3B were used. Their fabrication was
presented in Section 3.1.1 and 3.1.2. Four conditions were used: "Dry Control", "Wet
Control", "Dry Cycled", and "Wet Cycled." The Dry Control specimens were kept under
vacuum during conditioning. Wet Control specimens, similarly. were kept saturated with
moisture at 30°C during conditioning. Dry Cycled specimens were kept under vacuum
except during thermal cycling, and Wet Cycled specimens were kept saturated except

during thermal cycling.

Thermal cycling was performed by putting the specimens in an oven at 260°C for 15
minutes and then quenching them to room temperature in air. This was performed 11
times for the cycled specimens. A fifteen minute exposure was chosen because an
experiment with embedded thermocouples showed that the specimen center reaches the
desired temperature (250°C) after 15 minutes of exposure as shown in Figure 3-22.
Moisture profiles of the various specimen types through the conditioning cycle are

provided in Figure 3-23.

84

 

 

    
  

 

 

 

 

 

300 j 1 thermal spike temperature
zso-j f" 4"“
0 I
L. : from upper to lower:
3 1 temp. of surf.
E 1 50 1 temp. of 4th layer
8, I temp. of 8th layer (center)
5 100:
I- I
50: 9— sample temperature
;. L
()1...f,..4.,....,..e.,....,....,....fl
0 200 400 600 800 1000 1200 1400

Seconds

Figure 3-22 Temperature Profile for a Thermal Spiked BMI Composite

 

 

-o— BMI RESIN (1:352 mm)
+ K3B RESIN (t=1.47 mm)
—e— BMI COMPOSITE (1:275 mm)
--8- K38 COMPOSITE (t=2.15 mm)

   
   

 

WEIGHT GAIN (wt °/o)
N
l

 

 

 

 

 

 

 

12 24 36
TIME / THICKNESS ((hr A 1/2) / mm)

 

Figure 3-23 Moisture Profiles for All Specimen Types during Thermal Spike
Testing

85
3.5.2 Resin Flexure Testing

Three point flexure tests of neat resin BMI specimens were performed to determine the
change in neat resin properties due to moisture exposure and thermal cycling. Tests were

not performed on K3B because the resin plaque was not of sufficient size.

Specimens measuring 3" x 2/5" were cut from resin plaque TSR and 2TSR. The
fabrication of this plaque was detailed Chapter 4, and conditioning of these specimens

was detailed in section 3.5.1 above.

Three point flexure testing was performed at AMEES using a 50 mm span and a test
speed of 1.3 mm/min according to ASTM standard D790. Specimens were nominally 75
mm x 10 mm x 3.5 mm. The stress-strain curve for these flex specimens showed a brittle
failure in a manner similar to the resin tensile curve. The data reduction was completed
at AMEES according to the calculations in the standard. Table 3-28 summarizes the

results of this testing.

Table 3-28 Thermal Spike Resin Flexure Results

 

Flexure Strength Std. Dev. Flexure Modulus Std. Dev. Fail Strain Std. Dev.

 

 

Condition (MPa) (MPa) (%)
ry onfror 1B"? 1r—3244 WW
Wot Control 153 22 3270 86 4.87 0.95
Dry Cycled 141 11 3255 35 4.41 0.45

Wet Cycled 125 14 3312 23 3.75 0.42

 

 

 

86
3.5.3 Resin Tensile Testing

Tensile tests of BMI neat resin specimens were performed to determine the change in
neat resin properties due to moisture exposure and thermal cycling. Again, tests were not

performed on K3B because the resin plaque was not of sufficient size.

The specimens were made from the same panels and conditioned in the same manner as
the resin flexure specimens. Tensile tests were performed using a 'dog bone’ specimen
with a gauge length of 65 mm and a cross-section of nominally 10mm. Tests were
performed at a cross-head speed of 5 mm/min according to ASTM standard D638. A
representative stress-strain curve is shown in Figure 3-24 showing a brittle failure. Strain
was measured using the crosshead motion. so the failure strains are artificially high, and
modulus values are thus artificially low. The data reduction was completed at AMEES
according to the calculations in the ASTM standard. Table 3-29 summarizes the results

of this testing.

87

 

STRESS (M Pa)

 

 

 

0.0 T I l r T I T
0.00 0.01 0.02 0.03 0.04

STRAIN (mm / mm)

Figure 3-24 A Tensile Stress-Strain Curve for BMI Resin Showing a Brittle
Behavior

Table 3-29 Thermal Spike Resin Tensile Results

 

 

 

 

Tensile Strength Std. Dev. Tensile Modulus Std. Dev. Fail Strain Std. Dev.
Condition (MPa) (MPa) (%)
fintrol 75 9 WE 147—'57 fl
Wet Control 82 7 1779 204 5.6 0.9
Dry Cycled 74 7 1901 68 4.6 0.4
Wet Cycled 81 10 1901 28 5.2 0.7

 

 

 

3.5.4 Thermal Spike Composite Flexure Testing
Four point flexure tests of BMI/IM7 and K3B/IM7 composite specimens were performed
to determine the change in composite properties due to moisture exposure and thermal

spikes.

88

BMI/IM7 specimens were cut from composite panel X. and K3B/IM7 specimens were
cut from panel 488. Panel X was 47% Vf and both panels were [DO/90°]4S. Four inch by
one inch specimens were cut from these panels using the Bennett 5000 Plate Saw. The
fabrication of the BMI/IM7 panel was described in Chapter 4. Conditioning of these
specimens was described in section 3.5.1. Testing was performed according to ASTM
standard D790 for four point flexure testing. The test speed was 0.15 in/min. Third-point
loading was used with a span of 3 inches. The data reduction was completed at AMEES
according to the calculations in the ASTM standard. The BMI specimens showed a
compressive failure between the upper load noses. The K3B specimens showed a
bending failure under both load noses. While these failure modes are not strictly
desirable, they provide some insight into the materials and are discussed later. Table 3-30

and Table 3-31 summarize the composite flexure results.

Table 3-30 Thermal Spike BMI Composite Flexure Data

 

Flex Strength Std. Dev. Flex Modulus Std. Dev. Fail Strain Std. Dev.

 

Condition (MPa) (MPa) (%)
m 100 T 1054 31 1.06 00?:
Wet Control 101 6 1069 59 1.09 0.03
Dry Cycled 99 2 1064 23 1.31 0.24
Wet Cycled 96 4 1058 37 1.32 0.27

 

 

 

89
Table 3-31 Thermal Spike K3B Composite F lexure Data

 

Flex Strength Std. Dev. Flex Modulus Std. Dev. Fail Strain Std. Dev.

 

Condition (MPa) (MPa) (%)
ficntrol 100 6 1 1037 5%?
Wet Control 102 2 1161 44 1.28 0.1
Dry Cycled 100 3 1113 32 1.19 0.05

Wet Cycled Blistered

 

 

 

3.5.5 Open-Hole Compression Testing

Open hole compression tests of BMI/IM7 and K3B/IM7 composite specimens were
performed to determine the change in composite toughness properties due to moisture
exposure and thermal Spikes. BMI/IM7 specimens were cut from composite panel Y, and
K3B/IM7 specimens were cut from panel 486. Panel X was 47% vf and both panels were
[0°/9O°]4S. The fabrication of these panels was described in Chapter 4. Three inch by one
inch specimens were cut from these panels using the Bennett 5000 Plate Saw. A quarter
inch hole was cut in the center of the specimen using a special drill bit for composites.
Conditioning of these specimens was described in section 3.5.1. Testing was performed
at AMEES and results were provided in tabular form. The exact test conditions are

available. Table 3-32 and Table 3-33 summarize the composite flexure results.

90

Table 3-32 Thermal Spike BMI Open-Hole Compression Data

 

 

 

Strength Std. Dev. Modulus Std. Dev.
Condition (MPa) (MPa)
W61 295 30 52% 40
Wet Control 307 26 530 35
Dry Cycled 288 17 560 50
Wet Cycled 290 7 481 46

 

Table 3—33 Thermal Spike K3B Open-Hole Compression Data

 

 

 

Strength Std. Dev. Modulus Std. Dev.
Condition (MPa) (MPa)
W0! 322 16 656 17
Wet Control 343 13 611 48
Dry Cycled 351 5 630 80
Wet Cycled Blistered

 

 

 

Chapter 4

MICROCRACKING OF CROSS-PLY BMI/IM7 COMPOSITES

4.1 Introduction

Current BMI cure cycles cause significant transverse microcracking which has been
shown to greatly increase the moisture diffusion rate as seen in Chapter 2. A reduction in
properties has also been observed [1]. This chapter presents a description of the
microcracking phenomenon using basic mechanistic principles, and then provides results
from both modeling and experimental studies in support of the theory. Finally, the model
was used to evaluate various methods for reducing the microcracking in these models.
Methods investigated centered around the addition of an appropriate interphase between

the fiber and matrix. The ideal properties and size of the interphase were determined.

4.2 Cure Induced Microcracking

Microcracking of continuous fiber composites has been observed in a variety of
fiber/matrix systems. There is an irrefutable body of evidence that this microcracking
results from thermal expansion mismatch strains induced during cooling from cure

temperatures. These thermal strains are proportional to the temperature difference

91

92
between the "Stress-Free Temperature" (SFT) and the final part temperature (room
temperature). The SFT is the temperature at which residual thermal stresses go to zero in
the final laminate. To predict the effect of various cure cycles on the composite's
microcracking, this "Stress-Free Temperature" must be known. Simple tests were

performed to directly determine the SFT as a function of cure cycle.

There are two causes for the thermal stresses which generate microcracking. One is
microscopic and derives from a mismatch between the thermal expansion of the matrix
and the transverse thermal expansion of the fiber. The other is based upon the
macroscopic thermal expansion mismatch between plies in a cross-ply laminate. Thus,
the stress which actually causes microcracking is due to the sum of stresses caused by
both mechanisms. To fully understand this cure induced microcracking, both

mechanisms must be accounted for.

4.2.1 Review of the Microscopic Microcracking Mechanism

Adams [21] has performed a micromechanical two dimensional finite element analysis of
three different carbon fiber/polymer systems. He found that. for an AS4/BMI system,
complete yielding of the matrix was possible during cool down from a 204°C cure
temperature. He also found that the addition of one weight percent moisture could

subsequently cancel almost all of the cure induced stresses.

93

4.2.2 Review of the Macroscopic Microcracking Mechanism

Cross-ply laminates are subjected to more induced stresses than unidirectional laminates
due to the large anisotr0py in the thermal expansion of the plies [22]. The 0° plies act as
constraints to the 90° plies causing significant thermal residual stresses in the 90° plies.
Jones, Mulheron, and Bailey [23] have shown the strain in the 90° ply induced by this
mechanism during cooling to be approximated by Equation

Eqn. 4-1.

8 : EI(w—a¢)AT Eqn. 4_1

EI+E

 

where all properties are in terms of the composite, l=longitudinal, and t=transverse.
Composite property data can be calculated from the constituent materials' properties
using formulas from Whitney et al [24]. and stresses can be determined using Hooke’s

law.

4.3 Finite Element Modeling

4.3.1 Introduction

Initially, a finite element model was used to determine the microscopic stresses caused by
the mismatch between the transverse thermal expansion of the fiber, and that of the

matrix. Ultimately, however, a more accurate and versatile model was used for the

94
majority of the analyses here, but the FEM solution shows the stress distribution nicely

and is thus presented here.

The model assumes a perfectly square packed unidirectional composite under thermal
loading. While the packing (square or hexagonal) and fiber spacing are known to have an
effect on these stresses [25, 26], calculations based on the square array have been shown
to provide good results for this type of analysis [21]. The model used here assumes a
completely stress free condition at the SFT, and predicts the stresses generated upon
cooling to room temperature. To perform this analysis properly, it was necessary to know
both the constituent materials properties and the actual SFT associated with the various

cure cycles under investigation.

Figure 4-1 shows the geometry and resultant stresses of this model. A tensile stress is
generated along the diagonal between fibers, and tangentially between them while a

compressive stress is formed radially between the fibers.

95

 

 

 

 

l

._
9

Figure 4-1 Geometry of the Finite Element Model Showing Resultant Stresses

 

 

 

 

 

 

4.3.1.1 Assumptions

The following assumptions are inherent to this analysis:

1. Fibers are arranged in a perfectly square array.

2. The section under consideration is "far" from the laminate edges.
3. Perfect bonding exists between the fiber and matrix.

4. The matrix material is homogenous.

5. The fiber is cylindrically orthotropic.

4.3.1.2 Model Particulars

The Marc finite element package was used for this analysis. The mesh was generated

using three noded linear elements and is shown in Figure 4-2. The boundary conditions

96
are also shown in this figure. The boundary conditions arise from the fact that in a
unidirectional composite, there are no external constraints. The only requirements are
that the unit cell remain square. Thus, the origin must remain at (0,0), all points along the
"X" and "Y" axes must remain on the axes. Lines 2-3 and 3-4 must remain linear such
that a square mesh is retained. This is accomplished by tying the "X" values for line 2-3,

and the "Y" values for line 3-4.

97

 

/\/V

 

Figure 4-2 The Marc Finite Element Mesh
In the unidirectional case, there is a line of symmetry along 1-3. The material properties
used in the FEM calculations are provided in Table 4-1. The matrix properties were
obtained from the Ciba-Geigy literature, and the fiber properties are from the Hercules

data sheets.

98

Table 4-1 Material Properties Used in the Marc FEM Calculations

 

Constituent Properties

 

 

IM7 Fiber Matrimid 5292 BMI

Tensile Strength 5,518 MPa 82.1 MPa
Modulus 4.276 GPa

Longitudinal, EI 303 GPa

Transverse, El 14 GPa
In-Plane Poisson's Ratio, Vn 20 -34
Coefficient ofThermal Expansion 4 l .9 x 10‘ /°C

Longitudinal, Gt, ‘0-36 X 10‘6 /°C

Transverse, 0tt 140 X 10‘“ I/OC

 

4.3.2 Results

The model was run using a fiber volume fraction of 62%, and various SFTs (130, 177,
200, 250, 300). A typical residual stress diagram for the component of stress in the x
direction is provided as Figure 4-3. It can be seen that the maximum compressive stress
in the x direction occurs along the x axis between fibers. A tensile stress of a similar
magnitude exists between this fiber and the one above it in the y direction. A maximum
shear stress occurs just within the matrix at an angle of 45° to the origin. The effect of
the SFT on the maximum stress is shown in Figure 4-4. Thus, for a sufficiently high
SFT, failure would be expected across the diagonal between fibers. Fortunately, the
calculations show that the residual stresses for a perfect unidirectional composite are

insufficient to cause failure.

 

99

‘3. 4.5211200?

3:150:40?

2211804]?

1337:3417

6.5 ESCvClE

"1.05 DD+UE

-1.275L‘o-l] 7

4'. ”nun?

(1.018130 7

IS. Mica-l] T

|ohl

 

comp II of films

Figure 4-3 Residual Stresses from Marc FEM Model (V f = 62%, SFT = 200C)

100

Thermally Induced Stresses vs. SFT

I

vr = 62% I

120 I
I

I

I

I

 

100 - Total

 

Stress (MPa)
03 on
o 0

.§
0

N
O

 

 

 

 

75 125 175 225 275 325

Figure 4-4 Effect of SF T on the Maximum Residual Stress

The fiber volume fraction in this model is the ratio of the fiber area (nd2/4) to the total
area (sidez). To change the volume fraction, or add an interphase of varied thickness, a
new mesh is required. Analysis of a cross-ply composite could be done by applying a
specified strain to the boundaries at 2-3 and 3-4, but this would require reworking the
boundary elements for each SF T. Because constructing new meshes is time consuming,

an analytical model was used for further analysis.

101

4.4 Analytical Modeling using "MicMec"

4.4.1 Introduction to Micmec

MicMec is a Fortran code created by Averill. It provides an elasticity solution to the
micromechanical stress state in a composite comprised of a periodic array of fibers in a
matrix. The fibers can have any number of interphases with varied properties and
thicknesses. The model allows for thermal and moisture loadings. The material
properties contain temperature and moisture dependent terms. A more complete

description of the model including a derivation of the elasticity solution has been

published [27].

4.4.1.1 MicMec Assumptions

The model uses a diamond shaped periodic element as shown in Figure 4-5 and Figure 4-

6. The basic assumptions of this analytical model as presented in the paper are:

(i) The fibers are continuous and have infinite length in the r] I-direction.

(ii) The fibers have a circular cross-section that is constant along the length of the fiber.

(iii) The fibers are straight, perfectly aligned with the 1], axis. and arranged in a periodic
diamond array.

(iv) Fiber coatings or interphases may be represented by concentric circular cylinders

around the fiber.

102
materials are homogenous and linearly elastic with temperature- and moisture-
endent properties.
material comprising the outermost phase (matrix) is transversely isotropic. All
:r phases (fiber and interphases) are cylindrically orthotropic.
acent constituents (phases) are perfectly bonded.
:hanical loads are applied at infinity. Hygro-thermal effects are uniform within
1 phase of the composite, but may vary from phase to phase.

ds and material properties do not vary along the 111-direction.

Iodel Geometry

mptions made above allow a composite material to be modeled using a
Itive volume element (RVE). The RVE used in this analysis is shown in Figure

location of this element within the model composite is shown in Figure 4-6.

 

103

 

 

 

 

[\LL/

 

Figure 4-5 The Diamond RVE Used in the MicMec Analysis

104

 

 

 

 

 

Figure 4-6 Location of the RVE in the Model Composite

The shape of the element is controlled by the adjustable parameters (12 and d3. For square
packing, dZ/d3 = 1, and for a hexagonally close packed system (HCP). the ratio d3/d2 = \/3.
In Figure 4-5, rl represents the radius of the fiber, and subsequent radii (r2 rn) define
the various interphase thicknesses. Multiple interphases with different properties are
possible. Thus, a diffused interphase with properties which vary from that of the sizing

material, to that of the matrix can be easily approximated.

105

4.4.1.3 Generation of a Graphical User Interface for MicMec

The MicMec code reads a text file which contains the geometric and property data for the
model along with the positions for which solutions are desired. Output can include
stresses, strains, displacements, and macroscopic composite properties. The output is

provided in various logically organized text files.

Because this can become cumbersome, a Graphical User Interface (GUI) was created in
Microsoft Access. The objective of the GUI was to automate the data collection and
storage so that a large analysis domain could be covered quickly and accurately. The data
generated is stored in tables, and results can be graphed simply by selecting the property
of interest, and values for the parameters to be held constant. The GUI interacts with the
original code as any user would, by simply creating the input data files, and then

importing the output files and storing the data in a logical fashion.

4.4.1.4 The Addition of Laminate Calculations in the GUI

The capability of MicMec was slightly enhanced by allowing the user to choose a layup
geometry of either unidirectional, or cross-ply. The choice of a cross-ply laminate
requires that external strains be applied to the RVE. These strains are applied by simply
Stating them in the original MicMec's input file. The magnitude of the applied strains is a
function of the thermal expansion mismatch between the 0° and 90° plies, and the SFT.

These strains were calculated using Classical Lamination Theory (CLT).

106

CLT assumes that composite laminate's properties can be calculated from the properties

of the individual laminae which make up the laminate. Various assumptions are required

for this to be possible.

(i)
(ii)
(iii)
(W)
(V)

The lamina must be homogenous and orthotropic.

All lamina must be perfectly bonded.

The stress transfer between laminae is perfect.

No damage occurs.

In this case, it is also assumed that macroscopic stress distribution across each

lamina is uniform.

4.4.1.5 Derivation of CLT Equations Used

To calculate the thermal strains during cure in a cross-ply laminate, the following

methodology was used.

(0

(ii)

Obtain the lamina level properties. These were obtained by running MicMec with

each specified geometry and material combination.

Calculate the lamina stiffness matrix [Q] using these lamina properties. The

equations used were obtained from Whitney [24].

107

(iii) Calculate the laminate stiffness matrix for use in the governing equation, Eqn. 4-2.

3: =[ili] f: Eqn.4—2
I'I—f BD K

Where A, B, and D are the extensional, coupling, and bending stiffnesses,

respectively. JV and IV are the forces and moments applied, and 5° and K are the
strains and curvatures. Due to the symmetry of a cross-ply laminate, there are no
coupling stiffnesses (Bij=0). The ABD matrix was calculated using equations from

Jones [28].

Given the geometric parameters 11, M. F, and t below, the components of the ABD

matrix are provided in Equations 4-3 through 4-6.

n = the total number of layers (11 = 3).
M = the ratio of the total thickness of odd layers to that of the even layers (M = 1).

F = the ratio of the principle lamina stiffnesses (F= Qn/QI I).

 

108

 

 

 

 

 

 

A22 =1+ M(1+ MF)IQH
A16 Al6 : 0
A66 2 tho
Bil : O
[(F—1)P+1]Q,,I-‘
ll : 12
Q ’3
D12 : 1122
[(1— F)P+ F]Q,,r"
D22 : 12
D16 = D26 : O
Q. ’3
D66 = 162
where
1 M(n - 3)[ M(n — 1) + 2(n +1)]
2 1 + 7 1
(1+ M)‘ (n- — 1)(1 + M)‘

Eqn. 4-3

Eqn. 4-4

Eqn. 4-5

Eqn. 4-6

(iv) Calculate the thermal forces (2V ) and moments ( H ). Again, due to symmetry, the

moments are zero. The thermal forces calculated here are not actual forces, but are

109
representations of the forces which would exist if the laminate were fully

constrained (which it is not).

In terms of global laminate coordinates. these 'thermal forces' are:

NX'T gll QIZ QM at

N y! : Q12 Q22 Q25 a‘ Asz E 4 7
~ - — —.— n. -

N xyl Q16 Q26 Q06 A- u k q

where k represents each layer, and dz denotes integration through the thickness.

[28].

In terms of local lamina coordinates, this is equivalent to:

er =[(Q11a1 + Q12a2)2’0 + (Qua: + Q12a1)2’00 IAT
NY T 2 [(leal + Q22a2)2’0 +(Q12a2 + Qllal )2’90 IAT Eqn. 4-8

NU" = Q,,6a,,(2r0 + 2t.,,)AT = 0

where symmetry provides that On L Q21» Q26 L Q16 = 0. and 01.2 = 0.

If there are no applied forces or moments. these fictitious thermal forces can be

substituted for N in Eqn. 4-2.

110

(i) Invert the AD matrix and solve for the strains,g°. Because of the symmetry, K

will be zero.

(ii) Impose these strains upon the RVE and run MicMec.

4.4.2 MicMec Modeling Test Plan

As with any research, specific objectives must exist, and a logical approach must be used
to meet these objectives. The objectives and approach for this modeling study are

presented here.

4.4.2.1 Objectives of Modeling Work

Microscopic models such as MicMec or FEM are very useful for understanding the local
stress states, but real composites generally violate the modeling assumptions. In
particular, the fibers are never truly periodic and the bond between the fiber and matrix is
not perfect. The true value of models like this lies in their ability to predict the system's
response to geometric or material changes. With this in mind, the objectives for this

modeling work are:

111

(i) Determine the residual stress state and it's response to the SFT, fiber volume

fraction, interphase properties, and interphase thickness in cross-ply laminates of

BMI/IM7.

(ii) Find the most effective interphase material in terms of material properties and

thickness.

(iii) Identify the most practical method for relief of the residual stresses.

The array of test conditions used to accomplish this is given in Table 4-2, and the

material properties for the various phases are given in Table 4-3.

Table 4-2 MicMec Test Conditions

 

 

 

Volume Fraction Stress Free Sizing Thickness Sizing Tensile
(%) Temperature (°C) (um) Modulus (MPa)
35 100 0 2895
55 200 0.0] 182.8
62 300 0.1 18.28
1.0 2.29

 

 

112

Table 4-3 Material Properties Used in MicMec Analysis

 

 

 

Hercules 1M7 Matrimid Matrimid Sizing "A" Sizing "B" GE RTV-31
Fiber“ 5292 BMI 5218
El (MPa) 303,000 1828d 2895b 182.8 18.28 2.29d
E2.3 (MPa) 14,000
G (MPa) 1 1,000 674“ 1069c 67.4c 6.74C 0.845c
v23 0.25
V1213 0.20 0.35 0.35 0.35 0.35 0.35
CTEl -O.36x l O‘°/°C 42x 1 O‘°/°Cd 28x 1 O'°/°Ch 42x10'5/°C 42x10“/°C 200x 1 0'°/°Cb
CTEZI3 14x10'°/°C

 

 

a From a combination of product literature. AS4 data, and oriented graphite data.
b Product literature values

c Estimated based upon tensile modulus

d Experimentally measured

It is important to note that this model does not provide information about the adhesion
level, or failure mode in the system. Failure must be inferred using appropriate failure

criteria such as exceeding the material's tensile or shear strength.

4.4.2.2 Analysis Approach

To gauge the system's response to the various parameters, the appropriate response must
be measured. The responses of interest here are the location and magnitude of the
maximum radial, hoop, and shear stresses. To show these, the radial positions at which
they occur was first determined as shown in Section 4.4.3.1, then results for the ensuing
analyses were calculated for an array of angles along these radial values. The general
location of the analysis points is shown in Figure 4-7. Since the size of the RVE varies

with volume fraction, the outer array of points was kept at the maximum possible.

 

113

 

 

Figure 4-7 Location of Analysis Points Used in the MicMec Model

The results are provided using a cylindrical coordinate system as suggested by the

geometry of the analysis. To help clarify the results. a quick review of the three major
resultant stresses is provided here. The stresses are the radial (GR), theta (hoop, CH), and
shear stresses (0'5). These are shown schematically in Figure 4-8. Tensile radial and

hoop stresses are depicted. It is important to note that the shear stress acts on the outer

surface of the cylinder (the radial plane) in the theta direction and is NOT on the 2 plane.

 

 

 

 

 

Figure 4-8 Schematic Representation of the Resultant Stresses

4.4.3 Micromechanical Results

The results provide some very interesting information which will help steer research into
the use of controlled interphases for relief of residual stresses and are provided in the
following sections in terms of the objectives stated in Section 4.4.2.]. First, the
distribution of stresses and strains in the radial and theta directions are discussed and the
logic used in determining the location of analysis points for the remaining analyses is

provided.

The effects caused by changes in fiber volume fraction, stress free temperature, interphase
thickness, and interphase modulus are then explained. Finally, the effects of volume
fraction, interphase thickness, and interphase modulus on the macroscopic, lamina level,

properties is discussed.

115

4.4.3.1 Determination of Radial Stress Distribution

To determine the position of the maximum stresses and strains. the model was run using
an array of radial values along each of five angles. The resultant stresses and strains are

shown in Figure 4-9 and Figure 4-10. It can be seen that:

For the stresses,

0 The maximum radial stress (44 MPa for this condition) occurs in the bulk matrix as
far from the fiber as possible.

0 The maximum hoop stress (100 MPa) and shear stress (3.2 MPa) occur in the bulk

resin at the interface between the fiber and interphase.

For the strains,

0 The radial strain is a maximum at the interface of the fiber and interphase and is near
zero in the bulk matrix. The magnitude of these strains is very large (90%) showing
that the majority of the strain in the system is absorbed in the interphase.

o The hoop strain is a maximum (3.3%) at the interface between the bulk and the
interphase and diminishes with increasing radius.

0 The shear strain is nearly zero except in the interphase where it reaches a maximum of

almost 18%.

For an interphase with a modulus two orders of magnitude lower than the bulk matrixes

modulus, the interphase material's failure strain must be very high in order to avoid

1 16
failure in the interphase. Fortunately, materials do exist which have failure strains of this
magnitude. Finding the appropriate combination of low modulus, high failure strain, and

temperature stability is challenging as discussed in Chapter 7.

Based upon the stress distributions seen here, it was decided to analyze hoop and shear
stresses on the bulk side of the interface between the interphase and bulk matrix. Radial
stresses were analyzed at the farthest point from the fiber where a complete circle could

be used as depicted in Figure 4-7.

Hoop Stress (MPa) Radial Stress (MPa)

Shear Stress (MPa)

117

 

 

 

 

 

 

 

 

 

 

58 n - ‘
56 1 (@119 / .— __- ;
v '
54 «I .X// / !
524 )7
4. / I
«3 ____.¥ . e i
2 3 4 5
as
801 I\
75] (V: .\
70'] L VT‘V \ TIL \
65 4
60 -I
55-
501 -
.5 - - J
2 3 4 5
0- W — —I— — u— — _-
-1 - _fi
I —o— 0° 1
-2 . [t - 225° I
I —v— 45"
.3 \‘7'. l—v 67.5“
I \7» I + 90"
“v I
-4 . .
2 3 4 5

Radius (um)

 

Vf = 55% - SFT = 200 C - Thickness = 0.1 um - Sizing = 1/100 Modulus

 

Figure 4-9 Radial Stress Distribution

118

as. 526 _ofiom

 

 

 

as 596 ooox

 

 

—I—-—I—-—-——I

_q)

. 1| 1 (I 0| 0|
. .

38 595 59a

675°

 

Radius (um)

0.1 um - Sizing = 1/100 Modulus]

200 C - Thickness

VI = 55% - SFT

 

 

Figure 4-10 Radial Strain Distribution Showing Large Strains in the Interphase

 

119

4.4.3.2 The Effect of Fiber Volume Fraction on Residual Stresses and Strains

As shown in Figure 4-11 and Figure 4-12. the stresses are inversely related to the fiber
volume fraction. Higher volume fractions show lower residual stresses than the lower
volume fractions. This is due to the fact that as the resin content increases, the mismatch
in properties between the 0° and 90° plies increases correspondingly. Similar trends are

seen for the strains.

Based upon this, composites should be made with a high fiber content in order to reduce
the residual stresses. Fortunately, the materials being discussed here are used for high
performance applications where high fiber content composites are desired for their high

modulus and strength properties.

120

 

 

I
R=264um,

Radial Stress

I
I
I
I
I__~_E,__

 

 

0 30 0 35 0 40 0.45 0 50 o 55 o 50
95 .,

] R=264um
90

O
a:
U!

 

I

Hoop Stress (MPa)
3‘.

 

 

0 035 040 045 050 055 060 065

 

R = 264 um

 

Shear Stress (MPa)

 

 

 

-3

 

I- r . . Y .—
0 30 0 35 040 045 0.50 0.55 060 065

Fiber Volume Fraction (%)

 

I SFT = 200 C - Thickness = 0.1 um - Sizing = 1/100 Modumil

 

Figure 4-11 Effect of Volume Fraction on Local Stresses

Radial Strain (%)

Shear Strain (%)

121

 

 

 

 

R = 264 um

 

0.35 0 4O 0 45 0.50 0 55 0 60 0 65

 

Hoop Strain (%)

 

R = 2.64 urn

 

 

0 35 0.40 0,45 0 50 0.55 0.60 0 65

 

R = 2.64 um

 

   
 

I

_._fi I

 

P

 

 

0.35 0 40 0 45 0 50 0.55 0.60 0 65

Fiber Volume Fraction

 

I SFT = 200 C - Thickness = 0.1 um - Sizing = 1/100 Modulus ]

 

Figure 4-12 Effect of Volume Fraction on Local Strains

122

4.4.3.3 The Effect of Stress Free Temperature on Residual Stresses and Strains

As can be seen in Figure 4-13 and Figure 4-14, the stress free temperature has a direct,
linear relationship with the residual stresses and strains. Higher stress free temperatures
cause higher stresses and strains, and thus lower stress free temperatures are desired.
Several methods for reducing the stress free temperature of these materials are discussed

later in this chapter.

90

123

 

80 4
70 <

60 4

Radial Stress (MPa)

 

 

R=2.8 um

 

100 200 300 400

 

Hoop Stress (MPa)

 

44

Shear Stress (MPa)
0.)

 

-5 4L—— T e

100 200 300 400

 

R=2.8 urn

 

 

 

100 200 300 400

Stress Free Temperature (C)

 

[ VI = 55% - Thickness = 01 um - Sizing = 1/100 Modulus

 

Figure 4-13 Effect of Stress Free Temperature on Local Stresses

124

 

 

 

Radial Strain (%)
O
N
O

 

 

0 100 200 300

b

s._____ ’. 4—F# ____._4 o 4

O

 

 

 

 

Hoop Strain (%)

 

200 300 400

SI
0

 

R=2.8 um

é)
s)

b
u

Shear Strain (%)
b c
U! #-

___‘
-‘F—'I45° |

I
I ___ ,L

b
a:

I

 

r_, m.m.i..fin___nn.,mhnnn

:5

\l
O
O
O

200 300 400

Stress Free Temperature (C)

 

[ vr = 55% - Thickness = 01 um - Sizing = 11100 Modulus ]

 

Figure 4-14 Effect of Stress Free Temperature on Local Strains

125

4.4.3.4 The Effect of Interphase Thickness on Residual Stresses and Strains

As can be seen in Figure 4-15 and Figure 4-16, the thickness has a mixed effect on the
stresses and strains shown. The hoop stresses and strains show trends opposite to those
of the radial and shear stresses and strains. The hoop stresses increase with increased
interphase thickness while the radial and shear stresses decrease. All responses are non-

linear with respect to interphase thickness.

To understand this, it is important to remember that very large strains exist in the
interphase for this choice of materials. If we consider that the interphase material can act
in a manner similar to an incompressible fluid in that it can provide high strains, but
volume must be conserved, the results become much more logical. The stresses in the
bulk resin are formed due to its larger shrinkage than that of the fiber. As the thickness of
the interphase becomes very small compared to the fiber radius, the resin's stresses are
directly opposed by the stresses in the fiber. The interphase is incapable of straining

sufficiently to fully relieve any of the stresses.

When the thickness of this compliant interphase approaches the magnitude of the fiber
radius, the interphase is basically capable of deforming sufficiently to relieve the radial
and shear stresses which exist. Again, considering the interphase material to be
incompressible, the shrinking resin will continue to have hoop stresses because of the

conservation of volume.

126
This presents a very interesting possibility. The use of a sufficiently thick compliant
sizing could redistribute the undesirable tensile and shear stresses into hoop stresses
which can help with mechanical bonding which is likely to be responsible for the
majority of the adhesion in systems such as are being investigated here. The use of a
sizing material with a lower CTE than that of the resin would increase the value of this
hoop stress even more. Care must be taken, however, not to generate too large a hoop

stress. as this could cause undesirable microcracking.

The most appropriate interphase thickness appears to be one which is capable of
redistributing the radial and shear stresses. This appears to require an interphase on the
order of 10% of the fiber radius. While this is thicker than commercial Sizings are
generally applied, it should not present a problem from a processing viewpoint. An
interphase of this thickness will represent a full third of the resin content in a 62%
volume fraction composite. Because of this, it is very important to consider the effect of
the sizing on the thermal capability and chemical resistance of the composite in addition

to it's mechanical properties.

Hoop Stress (MP3) Radial Stress (MPa)

Shear Stress (MPa)

127

 

 

 

 

 

 

 

 

 

 

0 i 0.01 0.1 1
Thickness (urn)

 

I vr = 35% - SFT = 200 C - Sizing = 11100 Modulus I

 

Figure 4-15 Effect of Thickness on Local Stresses

128

 

Radial Strain (%)

 

 

I

R 3.51um

 

N

Hoop Strain (%)

 

 

 

 

Shear Strain (%)

 

 

 

 

 

0 0.01 0.1 V 1
Thickness (um)

 

I Vf = 35% - SFT = 200 C - Sizing = 1/100 Modulus

 

Figure 4-16 Effect of Interphase Thickness on Local Strains

129

4.4.3.5 The Effect of Interphase Modulus on Residual Stresses and Strains

Figure 4-17 and Figure 4-18 summarize the effect of the modulus on the stresses and
strains. To understand the effect the interphase modulus is having, it must be considered
in a manner similar to that used in the interpretation of the thickness effect. Interphases
with a sufficiently low modulus are capable of relieving the radial and shear stresses
while the higher modulus interphases cannot. This is again manifested in a

corresponding change in the hoop stress.

To most effectively reduce residual stresses, it is apparent that the use of a lower modulus
interphase material is desired. A modulus of 1-10% of the matrix modulus seems to

present the best balance between radial and hoop stresses.

130

 

70 T
= 28 um

 

60

50

40

WMJ

30

Radial Stress (MPa)

20

 

 

10 100 1000 10000

100 .
R = 2.64 um

 

_i___._I _. T—L._+'___..‘___ _‘L__‘_.at
1

1.0
0

80—4

70 4

Hoop Stress (MPa)

607

 

 

$0-+ . sgr
1 10 100 1000 10000

 

 

_R = 2.8 MEI

 

Shear Stress (MPa)

03
_.I_.___L______J.__l

 

7:

 

 

10 100 1000 10000

Log (Modulus (MPa))

I vr = 55% - SFT = 200 C - Thickness = 0.1 ufl

 

 

Figure 4-17 Effect of Interphase Modulus on Local Stresses

131

 

 

 

 

I R=2.8 umI
I I
A 0 I I
3Q I
E I I
s . I
<73 ‘ 1
e I
U .
(u .
0: -2
I I
I I
-3 I____ f a I
1 10 100 1000 10000
4 _v - _
I R=264 urrTI
. I
a I I
E .
g 2 I I
m I
a I .
O
o I
I I I
1 I I
I I
o ,____ __ ___. _ I
1 10 100 1000 10000

 

Shear Strain (%)

 

 

1 10 100 1000 10000

Log (Modulus (MPa))

I vr = 55% - SFT = 200 C - Thickness = 0.1 um]

 

 

Figure 4-18 Effect of Interphase Modulus on Local Strains

132

4.4.4 Discussion of the Macroscopic Modeling Results

Since there appear to be some benefits to using a compliant interphase to reduce residual
stresses, it is very important to understand how such an interphase will affect the

macroscopic properties of the composite. This is investigated here.

The effect of volume fraction is relatively obvious and is shown in Figure 4-19.
Increased fiber content causes a dramatic increase in the longitudinal modulus, but has a
relatively weak effect on the transverse modulus. Thus, for mechanically loaded parts, it

is generally advantageous to have a high fiber content.

133

 

 

 

 

 

 

 

 

25 3
sass- E1 (GPa)
- E2 (GPa)
20 n
._ 2
7': A
d 3
9. e
,— N
m m
— 1
n 0

 

 

 

I I

 

0.3 0.4 0.5
Fiber Volume Fraction

 

Thickness = 0.1 um - Sizing Modulus = 1/100th BMI I

Figure 4-19 The Effect of Fiber Volume Fraction on Lamina Properties

As commonly expected, the sizing has little effect on the longitudinal modulus as seen in
Figure 4-20. A sizing with a lower modulus than the matrix can, however, reduce the
transverse modulus by as much as 13%. This reduction in off axis modulus may be
acceptable in some applications, but commonly cross-ply or quasi-isotropic laminates are

utilized, and in these cases, the reduction in modulus would affect the overall composites

performance significantly.

134

 

 

  

 

 

 

10 3
_E1(GPa)
-E2(GPa)
8..
—2
a 6' A
n. 3
£5, 9
,— N
1.1.]
I“ 4—
—1
2..
o- —0

0 0.01 0.1 1
Sizing Thickness (urn)

 

Ivr = 55% - Sizing = 1/1001h BMI Modulus

 

Figure 4-20 The Effect of Sizing Thickness on Lamina Properties

As with the thickness, the modulus of the interphase has little effect on the longitudinal
modulus of the lamina as shown in Figure 4-21. It does, however, reduce the transverse
modulus by as much as 20%. It should be noted that a sizing having 10% of the matrix
modulus is much more likely, and would cause a reduction of less than 4%, which would

be quite acceptable in most cases.

135

 

20 3

18 . E1 (GPa)
- E2 (GPa)

16I

 

 

 

 

14-

12-

101

E2 (GPa)

 

E1 (GPa)

 

 

 

 

 

 

 

 

T

100
Sizing Modulus (MPa)

I vr = 55% - Thickness = 0.1 umI

 

0.1 1

 

 

Figure 4-21 The Effect of Sizing Modulus on Lamina Properties

4.4.5 MicMec Modeling Conclusions

The modeling analysis here provides considerable insight into the appropriate methods
for relieving cure induced residual stresses. The most obvious method would be to
reduce the stress free temperature through methods such as reducing the cure temperature,
or utilizing stepped post-cures. A sufficient reduction in the cure temperature is not

acceptable because this would leave the resin partially cured, and subject to property

136

Fortunately, high volume fraction composites have been shown to have the least residual
stresses. These materials are generally used in high performance applications where this

is required.

The use of a sizing as an interphase material to reduce the residual stresses shows good
promise. The results here show that to be most effective. the sizing should be sufficiently
thick, and have a low enough modulus to relieve the shear and radial stresses formed
during the cure cycle. This thickness appears to be on the order of 10% of the fiber
diameter which is considerably larger than Sizings are usually applied. The interphase
modulus should be between one and two orders of magnitude smaller than that of the

matrix to achieve this.

The results also show that the addition of such a sizing will cause a reduction in the
transverse modulus of the lamina by on the order of 10%. This would appear to be
acceptable if the sizing were able to relieve the residual stresses. and still transfer the load
between the fiber and matrix. An additional benefit from the addition of a compliant
sizing is that it will increase the toughness of the composite which is generally a very

desirable trait.

137

4.5 Experimental Verification

Various experiments were performed which support the modeling work. In particular, 1)
the SFT was determined for various cure cycles; 2) unidirectional composites were
examined to see if the microscopic mechanism alone is capable of causing matrix failure;
and 3) the crack density and width were determined for various SFTs and volume

fractions.

4.5.1 Determination of the Stress Free Temperature

The SFT is caused by a combination of the thermal expansion mismatches from both the
microscopic and macroscopic mechanisms described in the modeling section, and also by
residual stresses due to the shrinkage of the resin. Thus, the SFT itself cannot be simply
predicted, or measured by such properties as Tg. To accurately determine an appropriate
value for the SFT, unsymmetric laminates were created with the various post-cures. They
were then subjected to a thermal ramp and the temperature at which their deflection

returned to zero was determined and deemed the stress free state.

The SF Ts associated with various post-cure cycles investigated are provided in Table 4-4.
These temperatures provide insight into the possibility of using a stepped post-cure to
reduce the stresses which cause microcracking, and provided values for use in the

modeling studies.

138

Table 4-4 Stress Free Temperatures of Various Cure & Post-Cure Cycles

 

 

 

Stress Free
Cure & Post-Cure Cycle % Temperature
Cure
177°C/9O min .57 100°C
177°C/90 min + 200°C/2hr .76 170°C
177°C/90 min + 250°C/2hr .84 230°C
177°C/9O min + 300°C/2hr .86 245°C
177°C/90 min + 200°C/2hr + 250°C/2hr .84 210°C
177°C/90 min + 200°C/2hr + 250°C/2hr + .90 240°C
300°C/2hr

 

 

This investigation has shown that the SF T is significantly lower than the cure
temperature, which is often used for modeling work. This greatly affects the magnitude
of the stress calculated by any model based upon thermal expansion mismatch. This has
also shown that while stepped post-cures offer some reduction in SF T, the reduction is

relatively small, and would not be sufficient to eliminate microcracking.

4.5.2 Determination of Microscopic (Fiber Level) Cracking

To determine if the microscopic stresses were sufficient to cause microcracking,
unidirectional specimens were created with varied post-cures. These were then sectioned,
and polished. Both optical microscopy and ESEM analysis were performed and showed

that no cracking existed due to this mechanism.

 

139

Table 4-5 Post-Cure Cycles Used for Microscopic Cracking Experiment

 

Cure & Post-Cure Cycle
177°C/90 min
177°C/9O min + 200°C/2hr
177°C/90 min + 250°C/2hr
177°C/90 min + 300°C/2hr

 

 

177°C/90 min + 200°C/2hr + 250°C/2hr + 300°C/2hr

 

 

To determine if there were any microcracks caused by the fiber level expansion
mismatch, polished sections of the [0411 panel with varied post-cures were examined. The
CMSC's Environmental Scanning Electron Microscope (ESEM) was used to view the
specimens at magnifications of up to 10,000x. No microcracking was evident in any of
the samples, and thus the thermal stresses induced by the microscopic mechanism curing

cure of these composites are insufficient to cause failure.

4.5.3 Determination of Crack Density and Crack Width

Two separate experiments were performed to measure crack densities and widths. The
first was designed to determine the effect of post-cure temperature and stepped post-
cures. Stepped post-cures include post-curing at initially lower temperatures and then
increasing the temperature as required to defeat the hindrances caused by the glassy state
as the Tg of the curing material rises. The second was designed to show the effect of fiber

volume fraction (47, 55, and 62% VI) and cure cycle (Low and High).

140

4.5.3.1 The Effect of Vf and Post-Cures on Microcracks using Optical Microscopy

Initially, panels were created with varied volume fractions and cure cycles. Crack

densities and widths were measured using optical microscopy.

Table 4-6 Post-Cure Cycles Used for the Microscopic Cracking Experiment

 

Cure & Post-Cure Cycle
177°C/90 min
177°C/90 min + 200°C/2hr
177°C/90 min + 250°C/2hr
177°C/90 min + 300°C/2hr

 

177°C/90 min + 200°C/2hr + 250°C/2hr
177°C/90 min + 200°C/2hr + 250°C/2hr + 300°C/2hr

 

 

 

The effect of post-cure temperature on crack density, and crack width are shown in Figure
4-22 and Figure 4-23, respectively. The effect of stepped post-cures is shown in Figure
4-24 and Figure 4-25. Here the stepped post-cures match the last two cycles in Table 4-6

and are compared to post-cures with the same final temperature.

The results seen here are not completely self-consistent, and there are a variety of reasons
for this. First, a single set of experiments was performed, and thus the numbers reflect
single values, not averages. This means that any single data point may be inaccurate and
should not be the basis for any strong conclusions. The emergence of trends does,

however, suggest that the data is revealing the nature of the system.

141
Second, the initiation of cracks in any material is very subject to the presence and
distribution of flaws. By nature, composites are quite inhomogenous materials. A
reasonably thick and varied interply resin rich region is always present. Within each
layer, resin rich regions exist at the outside of the tows, and in other regions due to the
processing method. These resin rich regions can easily act as stress concentrators, or
even cause failure locally due to thermal expansion mismatch or resin shrinkage during

cure.

Finally, the measurement of crack widths is subject to the definition of the crack width.
Crack width was measured using areas where cracks propagated through resin-rich
regions, because these were easily measurable, and seemed the least affected by damage
from local fibers. Measurements were made as Close as possible to the mid-plane of the

ply because crack width decreases near the outer 0° plies due to their constraining effect.

142

Crack Density vs. Post Cure

 

50.00
45.00 . - ; Ego/2.5 i '1
I
g 40.00 7I I 56%,“ ‘ :
a 35.00 0 '
g I Ana/ow . .2 I
e 30.0045 L -.z --
9, .
g. 25.00 4 .
'3 .1!
: 201M) .
i8
g 15.00 .
e 9 ‘
0 10.00
5.00 . A A' -'
0.00 __.—£ . e

 

 

 

150 170 190 210 230 250 270 290 310

Post Cure Temperature (C)

Figure 4-22 Effect of Post Cure Temperature on Crack Density

Crack Width vs. Post Cure

 

0.26
024 I ; 557/04.
I +56%vf
I‘ k45%vf
0.2 . , 7 _

   

0.22 I

0.18 .

0.16

Crack Width (mils

0.14 .

0.12

 

 

0.1

 

150 170 190 210 230 250 270 290 310
Post-Cure Temperature (C)

Figure 4-23 Effect of Post Cure Temperature on Crack Density

Crack Density (cracks/in)

50.00

143

Crack Density vs. Post Cure Type

 

45.00
40.00
35.00
30.00
25.00
20.00 1
15.00
10.00
5.00

 

0.00

 

- Jigsaw}- - -- .
’ "rl45%VfL L LL 7 . » 7- .7;—

    

 

 

250°l2hr 200°C/2hr+ 300°l2hr 200°C/2hr+
250°C/2hr 250°C/2hr+
300°C/2hr
Post Cure Type

Figure 4-24 Effect of Stepped Post-Cure on Crack Density

Crack Width (mlls)

Crack Width vs. Post-Cure Type

 

0.300

0.250

0.200

0.150

0.100

0.050

 

0.000

U 56% vf
I45% vf ‘ T

 

 

250°l2hr 200°C/2hr+

250°C/2hr

300°l2hr 200°C/2hr+
250°C/2hr+
300°C/2hr

Post Cure Type

Figure 4-25 Effect of Stepped Post-Cures on Crack Width

144

4.5.3.1.] Conclusions

Despite the scatter in the data. it can be stated that:

0 Standard volume fraction (64%) composites are microcracked after the 177°C cure.

o Post-cures at any of the 200°, 250°, and 300°C temperatures result in significantly
increased crack densities.

- Higher post-cure temperatures may result in a lower crack density, but the data is
inconclusive.

o The effect of stepped post-cures on crack density varied. but the data supports the
previous sections conclusion that stepped post-cures do not reduce the SFT

sufficiently to prevent or significantly reduce microcracking.

4.5.3.2 Effect of Vf and Post-Cure on Microcracks using the Environmental SEM

Since the previous experiments contained a fairly wide scatter, new specimens were
made, and the analysis of crack widths was performed in the CMSC‘s Environmental
Scanning Electron Microscope (ESEM). Optical microscopy was still used for the crack

density measurements.

Panels H, J, and W were used in this experiment. Their fabrication was discussed in
Chapter 4. The "High" post-cure was 280°C for 6 hours. "Low" post-cure was 200°C for
2 hours. The specimens were 4 cm x 1 cm and were polished without mounting. or the

use of any solvent.

The effects of volume fraction and post-cure on crack density are seen in Figure 4-26 and
Figure 4-27. Their effects on crack width are plotted in Figure 4-28, and Figure 4-29. It

is more convincingly illustrated by the ESEM images shown in Figure 4-30 and Figure 4-

31

After fabrication, the 55 vol% panel (MSW-J) was visually different than the other

panels, thus it is felt that the absolute values for crack density and width obtained from it

145

are not meaningful, but the trends appear valid.

70.00

60.00

50. 00 .

h
_o
o
0

Crack Densit
(Cracks/in)
01)
_o
O
O

20.00

10.00

0.00

BMI Crack Density vs. Fiber Volume Fraction

 

 

High Cure

Low Cure

 

 

0.4

0.45 0.5 0.55
Fiber Volume Fraction

Figure 4-26 Effect of Vf on Crack Density

06

0.65

Crack Densit

(Cracks/in)

Crack Width (um)

146

BMI Crack Density vs. Resin % Cure

 

62 WI 0/0

60.00 I P

 

.I.

40.00

30.00

1947 \Dlo/o
20.00

10.00

I... 55 \Ol°/o
0.00

0.7 0.75 0.8 0.85 0.9
Resin % Cure

 

 

 

Figure 4-27 Effect of Post-Cure on Crack Density

Crack Width vs. Fiber Volume Fraction

 

1.8 . J.High Cur?

1.6 .LowCure‘

 

1.4

1.2

0.8

0.6

0.4

 

0.2

0 c . .
0.45 0.47 0.49 0.51 0.53 0.55 0.57 0.59 0.61 0.63

Fiber Volume Fraction

 

 

 

Figure 4-28 Effect of Vf on Crack Width

Crack Width (urn)

1.8 »
1.6 .1
1.4 .

1.2

0.8 -
0.6
0.4 .

0.2

-0.2

147

Crack Width vs. Resin °/o Cure

 

.75

 

{fame/.1
I . '55 vol%'
I .517 \Ol°/o'

 

 

— _.4—

0.77 0.79 0.81 0.83 0.85

 

 

 

.37

 

 

Resin °/o Cure

Figure 4-29 Effect of Post-Cure on Crack Width

148

WB17
47 Vol% Fiber
High Post Cure
5000X

Specimen JB17
55 vol% Fiber
High Post Cure
5000x

Specimen HB17
62 vol% Fiber
High Post Cure
5000x

 

Figure 4-30 ESEM Micrographs of BMI/IM7 composites showing the increase in
crack width with decreasing fiber content.

I49

WB17
47 Vol% Fiber
High Post Cure
5000X

WA17
47 vol% Fiber
Low Post Cure
5000x

 

Figure 4-31 ESEM Micrographs of BMI/IM7 composites showing the increase in
crack width with increased post-cure.

150

4.5.3.2.] Conclusions

Crack density increases with increasing fiber volume fraction as shown in Figure 4-

26.

0 Crack density increases with increasing post-cure as shown in Figure 4-27

0 Both crack width and density were insensitive to post-cure cycle for the 62 vol%
composites (see Figure 4-27 and Figure 4-29). This is undoubtedly due to the fact
that the contribution from the resin diminishes drastically as the fiber packing
becomes nearly perfect, and while 62% is not yet perfect, the inhomogenous nature of
the laminate causes a majority of the fibers to be nearly perfectly packed by 62 vol%.

0 Crack width increases with decreasing fiber volume fraction as seen in Figure 4-28
and shown graphically in Figure 4-30.

0 Crack width increases with increased post-cure as seen in Figure 4-29 and shown

graphically in Figure 4-31.

151

4.6 General ESEM Crack Observations:

A variety of observations were made using the ESEM which support the microcracking

theory presented here, or provide further insight into the nature of this system. They are:

0 Cracks form in a tensile mode at the ply mid-plane (Figure 4-32) or in a shear mode
near the ply interface (Figure 4-33). This supports the theory that stresses are caused
by the mismatch between the properties of the 0° ply and the 90° ply. Stresses which
are formed in this manner would be transferred between the plies through shear and
would result in a tensile load in the center of the constrained ply.

o The crack failure mode is interfacial for both tensile and shear cracks (Figure 4-34).
This suggests that an improvement in the overall composite system could be achieved
by designing an interphase region between the fiber and matrix which is capable of
better adhesion and/or increased strain relief. Another method which has been
attempted for this system is to add a tough interleave between the plies to relieve the

residual stresses by allowing shear strain at the ply interface [2].

152

 

Figure 4-32 ESEM Micrographs of BMI/IM7 Composites Showing that Cracks
Form in a Tensile Mode at the Mid-Plane (47% vf, Low Cure)

153

 

 

Figure 4-33 ESEM Micrograph of BMI/IM7 Composite Showing Tensile Cracks
Forming in the Mid-Plane and Shear Cracks Forming at the Ply Interface
(47%, High Cure)

154

 

47% vf, Low Cure 47% vf, High Cure

Figure 4-34 ESEM Micrographs of BMI/IM7 Composites Showing that the Failure
Mode is Interfacial for Both Tensile and Shear Cracks

4.7 Identification of Microcrack Initiation During the Standard Cure Cycle

Because microcracking in BMI/IM7 cross-ply laminates occurs during the initial 177°C
cure, it is difficult to determine the point of crack initiation, and correspondingly difficult
to measure improvements aimed at preventing this initiation. Thus, a method for

monitoring the crack initiation throughout the cure cycle was desired.

To accomplish this, one needs to be able to detect crack formation during the cure cycle.

This requires the ability to observe the formation of cracks at the cure and post-cure

155
temperatures of up to 300°C. A method for detecting these cracks has been devised, and

the required hardware has been manufactured and tested.

The method consists of laying up a small cross-ply laminate (3/4" x 1/2") and placing it
in a vertical heating stage which can be controlled within the ESEM. The ESEM can
then be used to monitor the edge of the laminate for cracks while the laminate is
subjected to a controlled thermal cycle. This can be used to determine what changes to
the cure cycle may reduce the severity of the microcracking. Changes which can be
investigated include different cure temperatures/times, varied ramp rates, or the addition

of a lower temperature dwell either before or after the main dwell, to name the obvious.

The Vertical Heating Stage with a chill plate was designed and manufactured as seen in
Figure 4-35 and Figure 4-36 to achieve thermal cycles with specified ramp rates and hold
times. Heating rates in excess of 50°C/min and cooling rates of over 30°C/min have
been achieved within the temperatures of interest using 25°C cooling water. A cooling
rate of 25°C/min was achieved between 50°C to 30°C. Cooling rates are achieved
through the use of the water cooled base plate, thus faster cooling rates could be achieved

by chilling the water with the attached water chiller.

To ensure accurate part temperature, a thermocouple mounted directly to the specimen is
used to control the thermal cycle. The system is controlled by a remote computer which

records real time specimen temperatures which will certify that the desired thermal cycle

156
was actually achieved. In addition, a video record of the experiment can be kept via a

direct cable connection to the ESEM.

 

 

 

1/8 1/8
”I I~+
1/8
O “8 O
1
1/4' 3/4"

 

 

 

II OTI/a I I
1+ III

 

 

 

 

 

 

 

 

 

314"
Side View Front View
_ _. 1., _L _ _
_ __ ._'_\ _I r _
l |
_ 4 __ _ _
1 |
_ _ ._| l _ _
I |
_ _, L,)_L- _ -
_ _ ._I_\ 1 ,_ _
Top View
NOTE: All Threads are 4/40 Image of Assembled Vertical Heater Stage
Unless otherWIse noted. (with Chill Plate)

Figure 4-35 ESEM Vertical Heating Stage Drawings

157
I 1/4 I 1/4
1
1/4 I 3f
i . O '_;:
:1: m I. \)

1 11/16 ---- Ti J, U
@ GD 1 ' :-
Countersunk 4/40 H0 esi... :‘I: ' "‘ I

:1: 3,4. I it \ 10-32 threaded

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

r
t)
04 I1:
Gasket Seal
I— . .-I ' [_-_-_j J — NOTE: All Threads are 4/40
:1 _ .1 1:1 I: _ _ _ _ _ .11 | I _ Unless otherwise noted.
Side View

Figure 4-36 ESEM Vertical Heating Stage Chill Plate Drawings

4.8 Identification of the Effect of Laminate Features on Crack Propagation

To improve the BMI/IM7 system’s toughness, the propagation behavior of cracks must be
understood. To determine this, unidirectional composite specimens were incrementally

fractured in the ESEM while being monitored.

It was observed that resin rich regions appear to act as crack deflectors/stoppers. This is
demonstrated by the progression of the crack in Figure 4-37. Initially, the crack initiates
at the right edge of the specimen and propagates to the left until it hits the resin region
shown in (A). With additional load, the crack widens, but does not propagate as seen in

(B) and (C). Finally, with sufficient applied load, the crack deflects to a region above the

158
current crack (D) and continues (E). This behavior suggests that the use of a tough
interleave between plies may help to slow crack formation, and thus increase the overall

toughness of the composite.

4.9 Failure Analysis of a Mechanically Fractured K3B/IM7 Specimen

A tensile test of a K3B/IM7 composite ([0°/90°]4S) was performed to determine the locus
of failure in this system. A specimen was cut from K3B/IM7 panel 488. It was
approximately 6.5 cm x 1 cm and one edge of the specimen was polished for analysis
while the other was notched to promote cracking away from the grips. The specimen was
loaded in a tensile frame until both a significant load drop and audible cracking were

observed.

Failure analysis was performed in the ESEM and a few interesting observations were
made. The resin has a high strain to failure (Figure 4-3 8) which causes bridging of the
interfacial cracks which formed. It is known that the resin has a failure strain of between
8% and 12%, fitting the failure mode seen here. It is also obvious that there is poor

adhesion between the fiber and matrix (Figure 4-39).

159

       

(B) Crack widens with additional load
but does not progress.

(E) Deflected crack widens and propagates
with additional load.

Figure 4-37 Micrographs Showing That Resin Rich Regions Deflect Cracks

 

Figure 4-38 ESEM Micrograph Showing That K3B Resin Has A High Strain To
Failure Which Causes Bridging Of The Interfacial Cracks Which Form.

 

Figure 4-39 ESEM Micrographs Showing Poor Adhesion Between IM7 and K3B.

161

4.10 General Microcracking Conclusions

Two models were utilized to provide insight into the microcracking mechanism, and to

determine if an appropriate interphase could be found to help relieve the residual curing

stresses. Experiments were then performed to verify the results of the model, and provide

some information about the formation of microcracks in real composites. The following

conclusions were drawn:

A mechanism for microcracking was proposed which includes contributions from
both microscopic and a ply-level mechanisms due to thermal expansional mismatch
strains during cool-down from cure temperatures.

Finite Element Analysis using a square packed array of fibers suggests that the
stresses from either the microscopic or macroscopic mechanisms alone are
insufficient to cause microcracks, but the combination of stresses will cause
microcracking in any sufficiently cured composite.

Analytical modeling using MicMec shows that the stress free temperature has a
direct, linear relationship with the residual stresses.

The fiber volume fraction was shown to have an inverse relationship with the residual
stresses, thus a high volume fraction is desired.

The addition of an interphase around the fiber was shown to be an effective method
for relieving residual stresses.

The appropriate interphase thickness was found to be on the order of 10% of the fiber

diameter, and the modulus should be a fraction of that of the bulk matrix. Such an

162
interphase was shown to cause approximately a 10% reduction in the transverse
modulus of the lamina, but shows little effect on the longitudinal properties.
Stress free temperatures were experimentally found to be over 200°C for a
sufficiently cured system.
Microcracking due to the microscopic mechanism was not found as would be
expected from the modeling.
Cross-ply composites are microcracked after the initial cure, and subsequent post-
cures cause an increase in both crack density and width.
Stepped post-cures are at best marginally successful at limiting crack formation.
Higher cures were shown to produce larger crack densities and widths.
Higher fiber volume fractions were shown to produce smaller crack widths.
High volume fraction composites were shown to be insensitive to changes in the cure.
Tensile cracks form at the mid-plane of the transverse plies, and shear cracks form
near the ply interface.
A vertical heating stage for the ESEM was created to allow detection of the point of
crack initiation during the cure cycle.
Resin rich regions were shown to act as crack deflectors.
Failure in both the BMI/IM7 and K3B/IM7 systems were shown to be interfacial

suggesting poor adhesion.

Chapter 5

HYGROTHERMAL PROPERTY TESTING

5.1 Introduction

The mechanical and thermal properties of BMI and K3B composites are the final judge of
their utility as high performance materials. The applications for these materials require
temperatures of up to 190°C with moist environments. Thus, their response to heat and
moisture must be understood. During use at elevated temperatures, BMI composites may
undergo further curing and/or degradation. Either of these can adversely affect
performance, so the cure mechanisms were studied and presented in Chapter 2, and the
effect of extent of cure and moisture on material properties have been tested and are

presented here.

Composite properties are a strong function of the fiber volume fraction both globally, and
locally. To help understand the effect of local volume fraction fluctuations and to
magnify the effects caused by matrix changes. composites with varied volume fractions

were studied.

163

164
The current state of knowledge concerning the effects of moisture and cure advancement
for various systems are reviewed in the following sections both in a general sense, and

then specifically for the systems of interest here.

5.1.1 Moisture Degradation

The moisture degradation of polymer matrix composites is of great concern because most
of these materials are used in humid environments. The effect of moisture on composite
properties is complex and varies dramatically from system to system. In the following
sections, the general effect of moisture on composites will be presented, and then specific

knowledge associated with the degradation of BMIs and K3B will be reviewed.

5.1.1.1 General Moisture Degradation

When exposed to a humid enviroment. thermoset polymers absorb moisture on the order
of 1-3% by weight [29, 30]. This moisture can have a variety of effects on the polymer's
properties which will cause very significant effects on the properties of their reinforced

composites.

The rate of diffusion and equilibrium moisture content in these systems is a function of
temperature and environment. While attempts have been made to describe moisture
diffusion in epoxies throught the use of Fickian diffusional laws, this method is

incomplete, and thus potentially inaccurate in that it ignores the possibility of damage

165
accumulation which may significantly alter the diffusion of moisture [1]. Absorbed
moisture will plasticize resins causing a drop in any matrix controlled composite
properties [30], a lowering of it's Tg, an increase in impact properties above Tg [31],
swelling stresses, and a change in the deformation and failure modes [1]. It weakens the
fiber/matrix interface in graphite/epoxy systems [32]. Moisture has no measurable effect
on the thermal expansion coefficients for epoxies [33,34], but an effect is seen in

polyesters which can cause larger thermal residual stresses [34].

5.1.1.2 BMI Moisture Degradation

Composites made from Hexcel's F655 BMI reinforced with T650-35 carbon fibers have
been shown to attain an equilibrium moisture content of 1.1-1.5% by weight after 30 days
which resulted in a reduction in Tg of approximately 80°C [35]. Varma, Needles,
Kourtides, and Fish [36] studied the effect of moisture and found that the moisture had
little effect on the ILSS, and even found an increase in some cases. Very few other
studies of the effect of moisture content on the properties of BMI/carbon fiber composites

have been published.

5.1.1.3 K3B Moisture Degradation

Composites made from Avimid‘i‘ K3B attain an equilibrium moisture content of 0.4% by
weight after approximately 20 days with an associated reduction in TE of only 20°C [3 5].

A recent study has found that the Tg of saturated K3B/IM7 composites can be as much as

166
15°C lower than the dry baseline. but that the Tg is recovered after drying [37]. The same
study showed the intralaminar fracture toughness. GIC, to be essentially unaffected by

moisture.

5.1.2 BMI Cure Advancement Effects

BMI composites do not reach full cure during their standard cure and post-cure cycles as
shown in Chapter 2. Completely cured BMI exhibits a Tg of ~350°C while the standard
post-cure (200°C/2 hr; 250°C/6 hr) has a Tg of only ~325°C. The cure reaction shifts
from being kinetically controlled to being diffusionally controlled as the Tg of the resin
increases past the cure temperature. To complete the cure, the addition of a 300°C/2 hr
cycle is necessary [2]. Unfortunately, these high post-cure temperatures cause the
inherent, cure-induced, microcracks to widen to a point where they are visible without the

aid of either a microscope or penetrant [2. 3].

The increase in cure associated with the Tg increase from ~325°C to ~350°C did not
cause a decrease in Glc toughness values of Matrimid 5292 BMI/A84 composite
specimens. This fact is thought to be due to a toughening of the resin due to a more fully
cured and "defect-free" crosslinked network where unreacted chain sites are considered
defects [2]. This suggests that increasing Tg to 350°C by performing the additional post
cure at 300°C will not directly reduce the room temperature physical properties of the
composite. Unfortunately, this investigation did not determine the moisture sensitivity,

or thermal aging effect of the increased Tg.

167
A reduction in interlaminar shear strength with increase in T, or moisture content has
been reported for the Hexcel F665/T650-35 carbon fiber system [35]. There is no
measurable effect of Tg on IITRI compression strengths, but moisture is shown to cause a
significant decrease. Other investigations (using 0,0'-dicyanobisphenol A, BMI) have
shown a notable reduction in impact strength after thermal aging at 190°C. All
specimens in this study failed in shear due to the small span-to—depth ratio (12:1) and the
aging times were sufficient (1000 hours) to cause a significant rise in Tb, [13]. All of these
investigations show the importance of performing a detailed study of combined

hygrothermal exposures.

A careful study of the relationship between crosslinking and composite properties for the
PMR-Pl/Celion 6000 system was recently performed [4]. The study has shown that the
Tg, initial weight loss and density, and elevated temperature interlaminar shear strength
increase linearly with increases in crosslink density. Elevated temperature flexural
strength and modulus were found to increase in a non-linear manner with an increase in
crosslink density. Initial moisture absorption and coefficient of thermal expansion were

found to decrease linearly with an increase in crosslink density.

It is suggested that a similar set of relations for bismaleimide resins could be used to
predict the effect of crosslink density on properties [4]. Another paper, relating to a high
Tg amine/epoxy system, also states that there is a direct, one-to-one relation between TE

and percent conversion, independent of cure temperature [3 8]. A relationship between Tg

168
and percent conversion has already been determined for the BMPM-DABPA BMI system

[15].

5.2 Test Design

The objectives of this hygrothermal study were to identify any effect due to post-cure
cycle, moisture, or fiber volume fraction. These factors were chosen for a variety of
reasons. Cure may change over the life of the part, and has been shown to reduce
mechanical properties over time. Moisture has been shown to be both beneficial and
detrimental to composite properties. Volume fraction was chosen because most laminates
contain local regions of very high resin content which can have a great effect on
composite performance. The properties of these regions can be determined by testing low

volume fraction composites.

The majority of the hygrothermal testing performed was organized into a designed
experiment. The list of tests used in this experiment is provided in Table 5-1. In addition
to the testing shown here, a more complete set of resin tests was performed which
included flexure tests, tensile tests, and Tg determination for a wide variety of cure
histories. A study of the moisture absorption behavior of these materials was also

conducted.

169

Table 5-1 Hygrothermal Testing Scheme

 

 

 

Test Performed Materials Studied Properties Reported
Tensile BMI Resin Modulus, Strength,
Failure Strain
Flexure BMI Resin (Three Point. F lexure Modulus

Moisture)

BMI Resin (Three Point.

Cure Progression)
BMI/IM7 (Four Point)

Flexure Strength
Flexure Failure Strain

 

 

 

 

Iosipescu Shear BMI/IM7 Shear Modulus
Shear Strength
Shear Failure Strain
Isothermal Dynamic BMI Resin 30°C Storage Modulus
Mechanical - BMI/IM7 30°C Loss Modulus
Three Point Bend 30°C Tan Delta
Tg
Transient Dynamic BMI Resin 155°C Storage Modulus
Mechanical - BMI/IM7 155°C Loss Modulus
Single Cantilever 155°C Tan Delta
Tg

 

 

 

This testing was organized with a 23 factorial design to show both the main effects of

cure, moisture, and volume fraction, as well as their interaction effects.

procedures including data reduction and tabular results are presented in Chapter 3. The

discussion of these results including the main and interaction effects are presented in

separately in Section 4.6. Here the data is presented with respect to changes in the

properties of interest, encompassing the data from all pertinent tests. First, a detailed

study of the effect of cure history on the physical and thermal properties of the BMI

system is presented, and then an investigation into the moisture absorption behavior of

the system is provided.

 

The test

170

5.3 Neat BMI Resin Testing

The results of the tensile, flexure, and DSC tests are provided in Figure 5-1. It can be
seen that there is a distinct relationship between T8 and the cure time and temperature as

expected. It is also apparent that the ultimate Tg for this system is slightly over 350°C.

The results of these studies have been interpreted in terms of the changes brought about
by longer cures at the same temperature, and increased cure temperatures. A summary of

the findings is in Table 5—2.

171

 

Modulus (MPa)

 

Strength (MPa)

 

 

 

   
 
 
 
 
  

 

 

 

 
        

 

 

,3 7 Flexure Data
t 6 — — Tensile Data ‘_ 300
c .
E 5 - + T9 ( C) 280 8
5 4 — - " — 260 L,
U)

2 3 _ , , mi— 240 '-
5 2 . g i I- 220
u. - 200

1 “ T i — 180

o z ‘— 160

Low Short Low Long Std. Short Std. Long High Short High Long

Post-Cure Condition

 

Low Short = 177°C/90min + 200°C/1 hr

Low Long = 177°C/90min + 200°C/3hr

Std. Short = 177°C/90min + 200°C/2hr + 250°C/1 hr

Std. Long = 177°C/90min + 200°C/2hr + 250°C/6hr

High Short = 177°C/90min + 200°C/2hr + 250'C/6hr + 300°C/1 hr
High Long = 177°C/90min + 200°C/2hr + 250°C/6hr + 300°C/6hr

 

 

 

Figure 5-1 Flexure, Tensile, and Tg Data for BMI Resin with Various Post-Cure

172

Table 5-2 Summary of Resin Testing Results

 

 

 

Longer Cures at Increasing Temp Increasing Temp

Same Temperature from 200°C-250°C from 250°C-300°C
Tensile Modulus -20% -20% (long cures) mixed
Flexural Modulus -8% -1 4% +9%
Tensile Strength +l2-61% -29% (long cures) -38-62%
Flexural Strength -21% (250°C) -29% (long cures) ~0
Tensile Failure +15-120% +80% (long cures) -36-72%
Strain
Flexural Failure ~23% (250°C) -23% (long cures) ~0
Strain

 

5.3.1 Resin Testing Conclusions

Generally, low temperature post-cured specimens show the best combination of
properties. A reduction in most properties is seen as the post-cure temperature is
increased, and longer exposures at each temperature seem to decrease the modulus while
the effect on strength is mixed. Thus, low cured BMI resin is desired due to it's superior
properties, but continued cure during component use is likely, and will cause a change in
the material's properties. This change must be understood and minimized, so that proper

design criteria can be utilized.

5.4 Moisture Diffusion Study

A major objective of this research has been to understand the effect of moisture on the

properties of the BMI and K3B composites. To effectively utilize the information

 

173
obtained from the physical property testing, the moisture absorption behavior of these
systems must be understood. To this end, moisture absorption studies were performed to
determine the ultimate moisture content and diffusion coefficients for BMI and K3B
resins and their composites. The effect of extent of resin cure, fiber presence, and
microcracking was determined for the BMI/IM7 system. Two temperatures were used so
that the diffusivity constants, D0 and Q, can be calculated allowing knowledge of the

actual diffusivity, D. The results of these experiments are provided in Table 5-3 and

discussions follow.

Table 5-3 Summary Data for Moisture Absorption Study

 

 

Maximum Maximum Diffusivity Diffusivity Q Do
Condltion Moisture % Moisture % % Change (mm’lsec) (mm’lsec) (JlrnolK) (mm’lsec)
70°C 100°C 70°C-100°C 7°C 10°C
eSIn ar Ia ure BMW‘
BMI Resin Full Cure 5.14% 5.00% -2.70% 51965-08 1.166E-07 28658 1.203E-03
BMI X-ply 1.55% 1.57% 1.34% 18795-06 1.222E-06 —15255 8.927E-09
BMI Unidirectional 2 49% 2.32% -7.09°/o 3.206E-08 4.258E-08 10060 1.092E-06
K38 Resin 1.53% 1.62% 5.02% 9.998E-07 2.322E-06 29876 3.547E-02
K3B-IM7 X-ply 0.55% 0.54% -1.86% 33835-08 9.427E-08 36336 131565-02

100% Resin Equivalent for Composite Data
‘RS'BTIIW X-ply VfCor 1.08% 1.06%
BMI X-ply VfCor 4.21% 4.27%
BMI Unidirectional 5.10% 4.96%

 

 

-1 .W/o
1.34%
-2.72°/o

 

 

 

5.4.1 The Effect of the Extent of Cure on D and Mm for BMI Resin

BMI resin with a full cure showed an increase in ultimate moisture content, Mg, by an
average of 44% over the partially cured resin. Unfortunately, there was a significant
difference between the partially cured values for the two experiments. It is unlikely that

this was caused by an increase in cure of the resin. because the exposure at 100°C is fully

174
100°C below the post-cure temperature for the partially cured system. An interaction due
to both the temperature and water is also unlikely as this would have been evident in the

thermal spike data shown in Chapter 6.

The increase in cure also caused a very significant increase in the moisture diffusivity.
For the 70°C experiment, the diffusivity increased by 119%, while the 100°C experiment
showed an increase of only 29%. Despite the anomaly in the data, it is obvious that resin

with a higher extent of cure will absorb moisture faster and to a higher ultimate content.

5.4.2 The Effect of Microcracking on the Diffusivity of BMI Composites

By comparing the cross-ply and unidirectional specimens, it is easily seen that the
presence of microcracks increases the moisture diffusivity by over 50x (5700%). This is
expected because the microcracks allow the moisture access to a tremendously larger
surface area, thus requiring the moisture to diffuse over a much smaller distance to

achieve saturation.

It should be noted that the diffusivity calculated here is not a valid material property, even
for the microcracked composite. The model used here assumes two dimensional Fickian
diffusion with negligible diffusion through the exposed edges (due to their smaller
surface area). It is obvious that this is not the case for these materials because the
microcracking is most prevalent in the central plies. and thus the increased diffusion is

through the microcracks in the edges.

175

5.4.3 Effect of Microcracks on the Ultimate Moisture Content of BMI Composites

The ultimate moisture content, Mg of the resin in microcracked composite specimens is
an average of 16% lower than that of the fully cured neat resin. This difference is
believed to be due to the incredible speed with which moisture can diffuse into and out of
this specimen. It is known that this specimen lost weight during the measurement process

despite efforts to avoid this.

Another contributing factor may be that this composite panel was created with the "old"
batch of resin which has been shown to have a lower reaction rate. Thus this may be less

cured resin which should show a lower Mm.

5.4.4 Effect of Fibers on the Ultimate Moisture Content of BMI Resin

The ultimate moisture content of the resin in the unidirectional composite specimens
matches the Mo, of the fully cured resin to within 2% error. This suggests that the resin in
the composite has a similar morphology to that of the neat resin. Thus, there is no

evidence of a difference in the resin cure due to the presence of the fibers.

5.4.5 Microcrack Formation in Composite Specimens During Saturation

Careful examination of the absorption profiles of the unidirectional BMI/IM7 and cross-
ply K3B/IM7 specimens shows a shift in the diffusivity and apparent ultimate content

after 225 hours of exposure as seen in Figure 5-2 for a unidirectional BMI/IM7 sample.

176
Failure analysis of the saturated specimens in the ESEM at magnifications of up to 2000x
after polishing did not showed the presence of microcracks in either the K3B/IM7 or the
unidirectional BMI/IM7, but a recent study of the K3B/IM7 system has shown that ply-
level microcracking is observed after approximately 400 hours of immersion at 80° [37].
This suggests that microcracking of these specimens may have initiated, but that the

width of the microcracks was still extremely small due to the short duration of the test.

Zhou and Lucas [39, 40] have seen a similar crack formation in a T300/934
graphite/epoxy system after 4300 hours in water at temperatures of 75°C and 90°C.
Crack initiation was detected in the absorption curve as early as 800 hours which is

essentially at the end of the F ickian region.

 

 

 

6.00%
5.00% _ +Measured i
I _ Model
400%
.E
Q
0
E 3 00%
.9 '
o
3 Fickian Region I Non Fickian Region
3‘ No Microcracks . Microcracks
2.00% { .. .-L,
1 00%
0.00% .
o 200 400 600 800 1000 1200 1400

 

time (sec)"2

Figure 5-2 Absorption Curve for Unidirectional BMI/IM7 Showing a Good Fit to
the Fickian Model, and Non-Fickian Behavior Due to Microcrack Formation

177

5.4.6 K3B Resin and Composite Data

As previously mentioned, the resin used in these studies was fabricated using a
significantly different processing cycle than the composites received, and thus is very
likely to have significantly different properties. The diffusivity and ultimate moisture
content of the resin and composite are obviously different, and this is undoubtedly due
largely to the difference in their manufacture. It is possible, however, that there are other
factors contributing to the differences, such as a steric interaction between the fiber and
the polymerizing resin. These factors cannot be ruled out from this experiment, but it is

felt that they are unlikely.

5.5 Interfacial Adhesion Testing

Interfacial testing of adhesion (ITS) of BMI/IM7 composites was performed to determine
the change in adhesion levels due to moisture saturation, extent of cure, and fiber volume

fraction. The IF SS results are shown in Figure 5-3.

178

 

1200

1000 - L - .2 ._ 7. . _. .- _. _ _ _ _ _ .,.___ _-- _z

IFSS (psi)

 

400 E. In . . . _ _.. L

200 4 ,7, _. i 7. _— ya
I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 A.

47% 55% 62% 47% 55% 62% 47% 55% 6296

Partial Partial Partial Full Full Full Partial Partial Patid
Dry Dry Dry Dry Dry Dry Wet Wet Wet

Figure 5-3 Hygrothermal IFSS Results

The failure loads for this system were generally between 0.2 and 0.6 grams. Failure loads
this low are approaching the limit of accurate testing with this machine, and have
undoubtedly caused an increase in the scatter seen here. Despite this, a number of things
are apparent. First and foremost is the fact that the adhesion in this system is very low.

IFSS values for a carbon/epoxy system are generally on the order of 10,000 psi while the

results here are less than 1,000 psi.

There appears to be little variation due to fiber volume fraction. as would be expected
because the testing here is performed by choosing fibers within a specific local volume

fraction range. No change is apparent due to the increase in cure from partial to full.

179
Saturation with moisture appears to reduce the IF SS. The magnitude of this effect is
rather uncertain considering the low adhesion of the system in general, and the difficulties
in the test method. The reduction in IFSS due to moisture is supported by the fact that
the fully cured specimens were debonded after saturation without the addition of any
load. A fiber from a saturated, fully cured specimen showing this debonding is shown in

Figure 5-4.

5.6 Mechanical Test Results and Discussion

All of the testing was performed to determine the effect which advancing cure, moisture,
and fiber content have on the physical properties of the BMI/IM7 system. All tests
provided a measure of the modulus of the system, and some provided failure strengths
and strains. Because we are interested in these properties of the system regardless of the
test method, all mechanical test methods will be examined together. The dynamic
mechanical tests will be discussed separately due to the functionally different nature of
their result. As mentioned at the beginning of the chapter, the testing was designed using
a statistical design so that interaction effects could be studied. To ensure clarity in this
presentation of results, the statistical design is first reviewed with a sample set of
calculations, and the method utilized for data presentation is provided, but first the

complete set of data is plotted without manipulation in Figure 5-5.

180

 

Figure 5-4 A Fully Cured and Saturated Specimen Showing Fiber Debonding
Without Applied Load

181

 

       

       

           

 

 

 

 

 

 

 

    

 

     

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3%: 583m 3:. 333: $3 526 Egan.
.mmmm 5 4 1. 2 1 0
an _ :2....;.:;.:;::

mm =3. an

m m “3...“ 5.0413480. tun. b
mm
“T
Rm
1U
mm
:-
m-
mm
08
C
m
.mm
Wm
mm
0...-
c

amen». .4
5.05904.»
_ _ _ _ _ I I II _ _ . e
o o o m. .e w w w m. o. s o 5. n r
mmmmmm mmwmmm. . i .. .mmw
1 0 6 4 2 c 0
3%: £825 in: .238: as 52.» 95:: M

 

 

Figure 5-5 Plot of All Hygrothermally Conditioned Mechanical Test Data

182

5.6.1 Review of the Experimental Design

This testing was performed using a 23 factorial design with extent of cure, moisture, and
volume fraction as the factors. This was a two level design, so "low" and "high" levels of

each factor were chosen. These levels are presented in Table 5-4.

Table 5-4 Factor Levels Used in Hygrothermal Testing

 

Factor Level

 

 

 

 

 

Factor Low High
Partially Cured Fully Cured
Extent of Cure 177°C/90 min; 200°C/2 hr 177°C/90 min; 200°C/2 hr;
280°C/10 hr
Moisture Dry Wet
Volume Fraction 35% or 47% 62%

 

 

 

An easy way of viewing this type of statistical design is to consider that each factor
represents one side of a cube as depicted in Figure 5-6. Viewing the data in this manner
makes it easy to see that there are four sets of data which provide a measure of the effect
of the extent of cure. This change is represented by the change in response from the left
side of the cube to the right side. Since interaction effects are present in this analysis, the
data for each change is provided instead of presenting just the average change from face

to face.

 

183

 

 

   

62%

 

 

 

Volume Fraction

35%

 

 

 

 

 

(Partial) Extent of Cure (Full)

Figure 5-6 A Cubic Representation of the Experimental Design

From the original data, the percentage change in response for each edge (like condition 1
to condition 2 on the cube) was calculated as the difference in the two values divided by
the overall average of all of the data (the grand mean). In this testing, each condition
(comer of the cube) is represented by between 3 and 5 tests. Thus, each condition has a
standard deviation of it's own. A derivation of the proper standard deviation for the

percent changes is provided in Appendix C. These standard deviations represent the

184
actual deviation in the percent change value, not just the average of the standard deviation

of the two conditions.

All data has been analyzed in this manner, and will be presented in terms of the change in
physical properties caused by the change in each factor. The transient DMA test is the
only test which provides the T, of the system, and thus is the only test present in the
discussion of the thermal response to each factor. Interaction effects will be discussed

whenever they are present.

5.6.2 Sample Results Calculation

An example of the calculations use to generate the plots found in the following sections is
provided here. The calculations for the percent change in flexural strength which results
from increasing the cure from partial (177°C/90min + 200°C/2hr) to full (177°C/90min +
200°C/2hr + 280°C/ 10hr) are reviewed. All other calculations were performed in a

similar manner.

First, the original data from Tables 3-9 and 3-10 is plotted on the eight corners of the
cube as seen in Figure 5-7. The overall (grand) average is 726 for this case. Using this,
the change in strength as the cure increases is calculated for each combination of volume
fraction and moisture. These calculations are shown in Equation 5-1. Finally, the results
for all related testing are plotted in one figure allowing quick. easy visualization of the

meaning of the test results. These results are plotted in Figure 5-8 and it is easily seen

185
that all test methods show a reduction in strength due to an increase from partial to full

cure (because all 'percent change' values are negative).

 

 

'-17.0 \
909 I——/ 777

1000 I -32.4 x‘ \765

 

 

 

 

 

o\°
53
C:
.9
5
LI-i
g 692 1'24-8 /
s?

 

 

698 W
Lfi/

(Partial) Extent of Cure (Full)

 

Figure 57 Graphic Representing the Change in Flexural Strength Due to Cure

186

D L 'l’f—I——46O—698I10 — ‘28
I’y, ()l'l ‘ — 726 _ 3 .
D H' th (765-1000I100 324
r )9 1r : —— = — _
)' A ‘ 726
Eqn. 5-1
Wt L 'Vf'—I———512-692I100- 276
c , 011 ‘ — 726 — .
Wt L 17' (777-900I100 170
2 ~ 711’ ’ : —— = — .
( ‘ 726

5.6.3 Mechanical Testing Results

Both flexure and Iosipescu shear tests were performed on composite specimens for all
conditions. Flexure testing on resin specimens was performed to show the effect of cure
and moisture. Tensile tests of resin specimens with varied cures were performed to show
the effect of cure as discussed in Chapter 2 and will be used here as well to support these

results.

5.6.3.1 The Effect of Cure on BMI/IM7 Properties

The results of the physical testing in terms of the change from partial cure to full cure is
presented in Figure 5-8. Increasing cure can be seen to decrease the modulus of both the
resin and composite samples by between 5 and 30% with the flexure properties being the
least sensitive. Their is an interaction effect in the flexure data which shows that the
lower volume fraction composites show the effect of cure more than the high volume

fraction composites.

187
The fillly cured resin shows a lower strength for all tests by between 15 and 35%. As
with the flexure modulus response, the Iosipescu shear strength changes more for the low
volume fraction specimens than the high volume fraction specimens. The flexure
response in terms of volume fraction is clouded by an effect from the moisture. For both
the flexure and Iosipescu shear tests, the strength of the dry specimens is reduced by a
larger amount than the wet specimens. It will become apparent that this is due to a

reduction in the partially cured wet specimens strength due to the moisture.

Finally, the failure strain data is quite scattered, but the flexure specimens show a loss of
approximately 15% while the Iosipescu shear specimens are generally unaffected. The
tensile and flexure resin tests show rather large changes in the failure strain, but also have
very large scatter, so the meaning of the data is uncertain. More information concerning
the effect of cure on the resin's response is available in Chapter 2. The interactions seen

for the failure strain are basically the same as those seen for the strength.

188

Composite Composite . BMI R '
. . BMI Resm esm
Four Pomt IOSIpeSCU Tensile Three Point
Flexure Shear Flexure

.1 44

 

 

    

Percent Change in Modulus

 

 

 

Percent Change in Strength

 

 

 

 

 

 

0 ES

.
&1
mm

 

45m"
am
an
—E‘1
m

Percent Change in Failure Strain

 

 

 

{see . .
‘30 °N 33 ‘81 Dry Resm Dry Resm
mtomco
@3656

Dry 35%
Dry 62%
Wet 35%
Wet 62%

Moisture

 

 

 

 

 

 

 

 

Plots showing the percent change in
strength, modulus, and failure strain values
of BMI resin and composites as the cure varies
from a partially cured to a fully cured condition.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-8 The Effect of Increased Cure on the Physical Properties of BMI/IM7

189

5.6.3.2 The Effect of Moisture on BMI/IM7 Properties

The results of the physical testing in terms of the change from partial cure to full cure is
presented in Figure 5-9. The resin flexure data shows a loss in modulus of almost 10%,
but the composite data is quite insensitive to the presence of moisture. The high volume
fraction flexure and low volume fraction shear specimens even show an increase in
modulus of approximately 5%. The fact that the resin modulus falls while the
composite's modulus is either insensitive to moisture, or shows a slight increase suggests
that another mechanism is active in the composites. It is likely that the swelling of the
resin due to moisture presence has caused an increase in modulus which completely
counters the loss in resin modulus. The resin strength shows a drop of approximately
25% with the addition of moisture. The flexure strength was generally insensitive to the
presence of moisture with the exception of the partially cured high volume fraction
specimens which showed a decrease of approximately 10%, as did the Iosipescu shear

specimens in general.

There were a few actual interaction effects present in the Iosipescu shear strength data.
The high volume fraction specimens were more affected by moisture than the low volume
fraction specimens, and the partially cured specimens were more affected than the fully
cured specimens. The resin specimens showed a 25% loss in failure strain while the
flexure data was generally insensitive, and the Iosipescu shear specimens showed only a

slight loss.

190

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

     

 

 

 

 

 

 

Composite Composite BMI Resin
Four Point Iosipescu Three Point
Flexure Shear Flexure
a
,,, e .
3 .
.8 4 a '.
O
2 2
c . T
o 0 Li; m “E
o, .
C .
I“ Q -I I
J:
‘3 4
I:
3 6 ~
0)
o. ,3 z
.10 I -
10 7 e v
‘ . .
.5 5 I I I
2" I . :.' m
9 0 v up 7.7
to”
.E :5 I I
a) I .1
g 10
is“
0 -15 I
E L
3 ~20 I
an .
CL 25 I
30 I—
10 I’ T4 ’ fl—‘T ghdl‘
.E
g 5
U)
2
2 0 2»:
E IE
E 5 I L
CD 11’." I
8’ -10 I >1; ‘
(U -.
‘E I
U -15
E
0)
e 20
1b
'1 I
Vf (‘0 is co (0 co co m to Standard Cure
tt== tt== NeatReSln
Cure (U (U 3 D N (U 3 3
CL CL LL u. o. o. u. u.

 

 

 

Plots showing the percent Change in strength,
modulus, and failure strain values of BMI
composites associated with a change from
dry to moisture saturated specimens.

 

Part Cure = 177°C/90min + 200"C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr
Standard Cure = 177°C/90min + 200°C/2hr + 250°C/6hr

 

 

 

Figure 5-9 The Effect of Moisture Presence on the Physical Properties of BMI/IM7

191

5.6.3.3 The Effect of Volume Fraction on BMI/IM7 Properties

The results of the physical testing in terms of the change from low to high volume

fractions is presented in Figure 5-10.

Increasing the fiber content of composites is known to have a very strong affect on the
mechanical properties of the system. The results seen here are no different. The modulus
values are increased by between 40 and 50%. The strength values increase by 10 to 30%,
and the failure strain shows a decrease on the order of 15%. One interaction effect is
apparent in these results. The fully cured Iosipescu shear specimens appear to show a
greater increase in strength than the partially cured specimens when the volume fraction

is increased.

192

Composite Composite
Four Point losipescu
Flexure Shear

 

Percent Change in Modulus

 

Percent Change in Strength

 

Percent Change in Failure Strain

 

 

 

 

 

 

. 563-5 rise-'65
Morsture 030g DEC);
tt== tt==
Cure (unis: torus:
(Lawn. [LCLLLLL

 

 

Plots showing the percent change in
strength, modulus, and failure strain values
of BMI composites associated with a change
in volume fraction from 35% to 62%.

Pan Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200“C/2hr + 280°C/10hr

 

 

 

Figure 5-10 The Effect of Increased Volume Fraction on the Properties of BMI/IM7

193

5.6.4 Dynamic Mechanical Testing Results

Dynamic mechanical tests were performed to show the system's rheological response to
the factors. In addition to the information about the rheological response of the system,
Tg has been determined. The results of the dynamic mechanical testing is shown in

standard form in Figure 5-11.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 5-11 Hygrothermal DMA Test Results

5.6.4.1 The Effect of Cure on BMI/IM7 Rheological Properties

The results of the dynamic mechanical testing in terms of the change from partial cure to

full cure is presented in Figure 5-12.

Storage Loss
Modulus Modulus Tan Delta
50000 $299 0.06
"l
_ "I
.7 1000— —o.05
A 40000— a
a T &
~ 2
g . 800- V _o.04
£3 30000— _3 cu
_ 3 =
3 'u 0
8 500- o ~o.03 D
2 "m a 2 S
q, 20000- I" 5; — F ’-
0') 0
ca 400— _. —o.02
5 —~ —’i
m 000 mm 1 F“
10 ". 200-« - [. ”7 —o.o1
v . Ag . 3 l 0.00
mm Esfisfisfis E§E=E§E= §=t=§=t=
3 3
unmanned dudmauau aigidigi
. EEEEEBEE aafifieefifi 3:55:253
MmMeooggoogg 0033003; ooggooig
$$$$$$$$ o\°o\°o\°o\°o\°o\°o\°o\° o\°o\°o\°o\°o\°o\"o\°o\°
Vf [\NNNNNNN [\NNNNNNN NNNNNNNN
VVVVCDCOOCD VVV'VCDQCDCD VY'Q'VCOQDCDCD

194

 

   

 

 

 

 

Storage Loss
Modulus Modulus Tan Delta
to
O)
C
(U
.C
0
H
C
a:
2
a)
o.
29 i . ,

\° \° \° \° \° \° \° \° \° \° \c‘ \°

w sass ea°° %&°°

co co co co co co 3 “g co to g g

-o-a o—I ..a ..a H -o-

Moisture 33‘1“” 22‘0“” BEHIND

00;; 00;; 00;;

 

 

 

 

 

Plots showing the percent change in
the rheological properties of BMI composites
associated with an increase in cure from
partial to full.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-12 The Effect of Increased Cure on the Rheological Properties of
BMI/IM7
The isothermal DMA test results show an increase in both the storage and loss moduli.
The value of tan delta which is the ratio of the loss modulus to the storage modulus

(E"/E') shows a slight increase with increased cure. This suggests that the damping

195
characteristics of the fully cured resin may be slightly higher than that of the partially

cured resin.

The effect of increased cure on the T5, of this system as measured by transient DMA
testing is shown in Figure 5-13. As would be expected, increased cure causes a

significant increase in the Tg of the system.

 

 

 

 

 

 

 

 

 

 

 

 

60
50— “T
8’ "— F7
.5 40-
0)
U)
C
2 30—
o
‘E
8
5 20-
0.
10-
%} '
vr § § ff f:
V (D V (O
- a a 76 “
Morsture o D g ;

 

 

 

 

 

Plot showing the percent change in
the glass transition temperature of BMI
composites associated with an increase

in cure from partial to full.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-13 The Effect of Increased Cure on the Tg of BMI/IM7

196

5.6.4.2 The Effect of Moisture on BMI/IM7 Properties

The results of the physical testing in terms of the change from partial cure to full cure is

presented in Figure 5-14.

 

 

 

 

 

 

   

 

 

 

Resin Resin . Composite Composite .
Storage Loss T aieg'erlta Storage Loss gomréfslite
50 Modulus Modulus Modulus Modulus an e a
40 a
(D
2 30 -
3
'0
o
2 20 - -.
.E E.
8: J ‘
r: 10 g
‘u a:
...- 0
c g
a) a
o .
5 -10 ~ g
D.
_L
-20 a _
an
'W I I 1 fi I
Vf ssss ssss ssss
[x N N N Ix N N N N N N N
Y (O V (D V (O Y (O V (D V (D
t t = = t t = = t t = =
Cure $83.3 £855 £6338.

 

 

 

 

 

 

 

 

 

 

Plots showing the percent change in
the rheological properties of BMI composites
associated with an increase in moisture
content from dry to wet.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-14 The Effect of Moisture Presence on the Rheological Properties of BMI
Resin and Composites

197
The resin specimens show a decrease in both the storage and loss moduli as well as the
value of tan delta with increased moisture content. This suggests that the resin becomes
less viscous and more elastic in nature and would have lower damping capability. The
composite data shows little change in the modulus values for the low volume fraction
specimens. The high volume fraction specimens, however, show some decrease in the
storage modulus and some increase in the loss modulus. These changes cause the tan
delta values to rise suggesting that the composite system is becoming more viscous in

nature with the addition of moisture.

The fact that the high fiber content specimens show this change more than the low fiber
content specimens, and the resin specimens actually show the opposite effect suggests
that the increase in the damping characteristic (viscous nature) of the composite is caused
by an interaction between the fibers and resin. This is likely to be due to a change in the

adhesion level as has been seen with the ITS tests.

The effect of moisture presence on the Tb, of this system as measured by transient DMA
testing is shown in Figure 5-15. There is an interaction effect present here between the
moisture and extent of cure, and mentioned previously. The partially cured system is
subject to changes in Tg due to the presence of moisture, while the fully cured system

appears insensitive to the presence of moisture.

198

 

 

Percent Change in T9
.L

 

 

 

 

 

 

 

 

'8 ‘ _4
411
IV I
o\° o\° o\° a"
W N N N N
v co v to
g = g =
3 3
Cure 0. LL Q LL

 

 

 

Plot showing the percent change in
the glass transition temperature of BMI
composites associated with an increase

in moisture content from dry to wet.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-15 The Effect of Moisture Presence on the Tg of BMI/IM7

5.6.4.3 The Effect of Volume Fraction on BMI/IM7 Properties

The results of the physical testing in terms of the change from low to high volume

fractions is presented in Figure 5-10.

199

 

 

 

 

 

 

 

 

 

 

Storage Loss
Modulus Modulus Tan Delta
100
80 « i :E
(D 60 —1
3
g I
O 40 ‘l
S.
.E
a) 20 ~ I
C»
c:
2 0 T
0 T
§ -20 +
(T)
n. .40 s
89 r I fl '
Moisture 2‘75 6‘5 2‘5 2‘55 2‘33 E5
ago; ago; 0:03
C t t = = t t = = t t = :-
ure cums: toms: toms:
{LQLLLL 0.0.LLLL. 0.0.LLLL.

 

 

 

 

 

 

 

Plots showing the percent change in
the rheological properties of BMI composites
associated with an increase in volume fraction
from 47% to 62%.

 

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

Figure 5-16 The Effect of Increased Volume Fraction on the Rheological Properties
of BMI/IM7 Composites

200
Increasing the volume fraction of BMI/IM7 composites produces an increase in both the
storage and loss moduli. This is manifested in a loss in the value of tan delta and
suggests that the presence of more fibers makes the specimen more elastic in nature as

would be expected.

The effect of increased volume fraction on the Tg of this system as measured by transient
DMA testing is shown in Figure 5-17. No specific trends can be seen in terms of
interaction effects. A general increase in the value of Tg is seen here, but is believed to be
due to an experimental effect. The lower fiber content specimens will lag the actual test
temperature by a larger amount than the higher content specimens because of their lower

thermal conductivity. Thus, the effect is believed to be an artifact of the test method.

201

 
   
 

Percent Change in T9

 

Plot showing the percent change in
the glass transition temperature of BMI
composites associated with an increase

in volume fraction from 47% to 62%.

Part Cure = 177°C/90min + 200°C/2hr
Full Cure = 177°C/90min + 200°C/2hr + 280°C/10hr

 

 

 

 

Figure 5-17 The Effect of Increased Volume Fraction on the T: of BMI/IM7

5.7 Hygrothermal Conclusions
The hygrothermal response of the BMI resin and BMI/IM7 composite systems have been
characterized using a wide variety of tests. A number of general conclusions can be

drawn in terms of the effect of extent of cure, moisture presence, and increased volume

202
fraction. They are presented here organized in terms of these factors after a summary of

the conclusions from the moisture absorption study

5.7.] Moisture Absorption Behavior

A moisture absorption study was performed, and the diffusivity and ultimate moisture
content for both the resins and composites of the BMI and K3B systems were determined.
It was found that for the BMI system, increased cure causes a higher ultimate moisture
content as well as a higher diffusion rate. The addition of fibers to the BMI resin does
not appear to affect the cure, but the microcracking caused by the cross-ply laminate
caused a tremendous increase in the diffusion rate as expected. The diffusion was
generally seen to be Fickian in nature, but a deviation from this was seen for the
unidirectional BMI/IM7 and cross-ply K3B/IM7 composites after approximately 225
hours. This has been attributed to the formation of microcracks after this amount of

exposure as has been seen by other researchers on this and similar systems.

5.7.2 The Effect of Increased Cure

Increasing the cure of BMI has been shown to decrease both the modulus and strength for
both composite and resin specimens. Generally, the lower volume fraction specimens
showed this effect more than the higher volume fraction specimens because of the

increased contribution of the resin.

203
A detailed study of the progression of BMI resin properties with advancing cure was
presented, and it was found that lower cure temperatures produce the best combination of
properties, but with partially cured resins, cure will continue during service life, causing a
reduction in part performance. Thus, this progression must be understood and accounted

for when designing with this system.

Increased cure was seen to have little effect on adhesion as measured by the ITS test, but
failure analysis of the failed flexure specimens showed the fully cured system to have
lower adhesion. Finally, increasing the cure was seen to cause a very significant increase

in Tg.

5.7.3 The Effect of Moisture Presence

Moisture has been shown to decrease the resin's modulus, but the composite's modulus is
insensitive to moisture presence. It is suggested that this effect is most likely due to a

resin swelling mechanism in the composites.

Moisture decreases the strength of resin specimens, and the composite specimens show a
decrease as well, although the effect is much smaller. A general trend showing that
moisture causes a larger decrease in the properties of the partially cured resin than the
fully cured resin is seen. This suggests that the unreacted chain sites in the resin may

interact with moisture causing a reduction in strength.

204
Both ITS and DMA testing have shown that the adhesion is decreased by the presence of
moisture. Moisture has also been shown to decrease the Tg of the system as has been

seen in many other systems.

5.7.4 The Effect of Increased Fiber Volume Fraction

Increasing the fiber content of these composites causes the expected increases in strength
and modulus as well as the associated drop in failure strain. The use of low volume
fraction specimens has proven very useful in differentiating the effects of resin property

changes from the composite property changes.

No effect of volume fraction was seen on the ITS adhesion values and increased fiber
content was shown via DMA testing to cause the composite materials to become more
elastic in nature. An apparent change in the Tg due to changing volume fraction has been

attributed to an artifact of the experimental method.

Chapter 6

THERMAL SPIKE RESPONSE OF BMI AND K3B RESINS AND COMPOSITES

6.1 Introduction

Thermal excursions were known to increase the equilibrium moisture content of many
resin systems as early as the mid 19705 [41]. This has been attributed to a variety of
causes ranging from chemical changes in the resin to physical damage. Recently, there
has been interest in the effect of more rapid temperature changes because high
temperature composite systems are being used for components which see a larger range
of temperatures, and are subjected to very rapid excursions between the temperature
extremes. The BMI and K3B resin systems are used for aerospace applications where
components are often moist due to outdoor storage and then called upon to reach the
extremes of their design envelope within extremely small time periods. Previous work on

the BMI/IM7 system has shown only a small effect when specimens are cycled [42].

205

206

6.2 Experimental Summaries

6.2.1 Test Design

The objectives of this study were to identify any permanent damage due to moisture
exposure and thermal cycling. The properties of interest were modulus, strength, failure
strain, and retention of composite toughness. The test scheme used to study these is

shown in Table 6-1.

Table 6-1 Thermal Spike Testing Scheme

 

 

 

Test Performed Materials Studied Properties Reported
Tensile BMI Resin Modulus, Strength,
Failure Strain
Flexure BMI Resin (Three Point) F lexure Modulus
BMI/IM7 (Four Point) Flexure Strength
K3B/IM7 (Four Point) Flexure Failure Strain
Open Hole Compression BMI/IM7 Initial Stiffness
K3B/IM7 Strength
Iosipescu Shear BMI/IM7 Shear Modulus
K3B/IM7 Shear Strength
Shear Failure Strain
Dynamic Mechanical BMI Resin 155°C Storage Modulus
K3B Resin 155°C Loss Modulus
BMI/IM7 155°C Tan Delta
K3B/IM7 Tg

 

This testing was organized with a 22 factorial design to show both the main effects of
moisture exposure and thermal spikes, as well as their interaction effects. The test
procedures, including data reduction and results, are presented in Chapter 3. A discussion

of these results including the main and interaction effects is presented in Section 6.3.

 

207
These sections will focus on the properties of interest, encompassing the data from all

pertinent tests.

6.2.1.1 Post-Conditioning Observations

To determine the effect of the thermal spikes on the morphology of the specimens, the
ESEM was utilized to analyze the specimens after conditioning. After thermal spiking,
the wet K3B composites were visually blistered to the point that the thickness had
doubled. The BMI specimens showed no effect due to the thermal cycling. Figure 6-1
shows the blisters present in a moist, thermally spikes K3B resin specimen. The
specimen was approximately 1 cm3 and has been sectioned in half for analysis. Note that
no blisters occur near the edges where moisture is able to escape during the early stages
of the exposure. Figure 6-2 provides a higher magnification showing the general

morphology of the blisters in the resin specimen.

Figure 6-3 shows a blister in the K3B/IM7 composite. Figure 6-4 shows the viscous
nature of the failure in this specimen which suggests that the failure occurred above the
resin's Tg. This suggests that the Tg of wet K3B should be accurately determined and any
applications in the presence of moisture should utilize a design which remains well below

this value.

208

p I“! I
ll 1 I: h 1

 

Figure 6-1 Thermally Spiked K33 Resin Showing Blistering

209

Cl‘i'E-II l‘11-: ‘- _ '-.-'>Er"=.1tg ‘E-F'FF'IE‘IE TIF

 

Figure 6-2 Close-up of a Blister in a Moist, Thermally Spiked K3B Resin Sample

210

 

Figure 6-3 A Blister in a Moist, Thermally Spiked K3B/IM7 Composite

211

. 13$ '

EU ‘
Universitg SPKPIB44 TIF

 

Figure 6-4 A Close-Up of a Moist, Thermally Spiked K3B/IM7 Composite Showing
the Viscous Nature of the Failure

212

6.3 Combined Results and Discussion

All of the testing was performed to determine the effect moisture exposure and thermal
spiking have on the physical and thermal properties of the BMI and KBB resin and
composite systems. All tests provided a measure of the modulus of the system, and some
provided failure strengths and strains. As mentioned at the beginning of the chapter, the
testing was designed using a statistical design so that interaction effects could be studied.
To ensure clarity in this presentation of results, the statistical design is reviewed and the
method utilized for data presentation is provided, but first the complete set of data is

plotted without manipulation in Figure 6-5 and 6-6.

350

. 300
— 250
» zoo
. 150
—» 100
o
no
a» 3000
.. zooo
,. 1000

 

BMI Resin
Tensile

8.05.02,
865 to
, 63.80.25
.2200 >5

is

 

Flexure

BMI Resin
Three Point

2.”. v. 1.. m. U03>o a”;

.. 820 >5
. .. 6:505;
. _ozcoo >5

 

 

2 1 3
Composite
Open-Hole

Compression

“.226 ~m>>
3.95 >5
.9250 ES
.2250 to

 
 

 

Composite
Iosipescu
Shear

8.30.03
8.05 an.
,. _obcooam;
. 6:80 to

 

Compogite
Four Point
Flexure

 

. 8.0852,
8.26 >5
_oscooao;
.9580 ED

 

 

 

 

 

 

350

mmm....:-.

2 3s; EEG SEEK

100
50 <-

0

000
4000 4.

m w

300 a
250 o

Aunts Emcohw swag 3.3.82

Thermal
Spike
Morsture

 

 

 

 

Figure 6-5 All Thermal Cycling Test Data for BMI/IM7 Composites

214

 

 

 

 

 

 

 

 

 

 

 

 

 

Composite Composite Composite
Four Point Iosipescu Open-Hole
Flexure Shear Compression
400 400
73
n.
E
5
8’
2
:7:
’6
n.
E
n
_g
3
8
2
2 2
E
.s
1% ‘-r -—I
9 '0
2 0 '0
E 9 9
“ :2 a g
2 a N/A 79'
co m E
n o
_ — '0 _ — '0 T) " ‘0 '0
Thermalgggg gg28€.§%2
. _ c _
Sp'keogggogggoo>g
O O O O O O O O U Q 0 O
... ... ..- o-v 0" “
Moisture 2‘0?” 3’0?” broad)
a E o E o E o E D E D E

 

 

Figure 6-6 All Thermal Cycling Test Data for K3B/IM7 Composites

215

6.3.1 Review of Experimental Design

This testing was performed using a 22 factorial design with moisture exposure and
thermal spiking as the factors. This was a two level design, so "low" and "high" levels of

each factor were chosen. These levels are presented in Table 6-2.

Table 6-2 Factor Levels Used in Thermal Spike Testing

 

Factor Level

 

 

 

Factor Low High
Moisture Exposure Dry Wet
Thermal Spiking Control Cycled

 

 

 

An easy way of viewing this type of statistical design is to consider each factor to
represent one side of a square as depicted in Figure 6-7. Viewing the data in this manner
makes it easy to see that there are two sets of data which provide a measure of the effect
of the thermal spiking. This change is represented by the change in response from the left

side of the square to the right side.

 

216

 

   

(Wet)

Moisture Exposure

(Dry)

 

 

 

 

 

(Control) Thermal Spiking (Cycled)

Figure 6-7 A Square Representation of the Experimental Design

From the original data, the percentage change in response for each edge (like condition 1
to condition 2 on the square) was calculated as the difference in the two values divided by
the overall average of all of the data (the grand mean). In this testing, each condition
(comer of the square) is represented by between three and five tests. Thus, each
condition has a standard deviation of it's own. A derivation of the proper standard
deviation for the percent changes is provided in Appendix C. These standard deviations
represent the actual deviation in the percent change value, not just the average of the

standard deviation of the two conditions.

All data has been analyzed in this manner, and will be presented in terms of the change in

physical properties caused by the change in each factor. The transient DMA test is the

217
only test which provides the Tg of the system, and thus is the only test present in the
discussion of the thermal response to each factor. Interaction effects will be discussed

whenever they are present.

6.4 Effect of Moisture Exposure on Final Properties

It is very important to keep in mind that all testing was performed on dry specimens. The
moisture saturated specimens were fully dried before testing. and thus any effects due to
moisture are due to irreversible changes resulting from the exposure to moisture, not

simply its presence.

The effect of saturating the specimens with moisture during the experimentation for BMI
resin and composites is shown in Figure 6-8. A slight increase in the shear modulus and
shear strength for the cycled specimens is seen and there are some effects evident in the
resin tensile and flexure results. The change in properties in the flexure and tensile resin
results show opposite trends and contain a high amount of scatter, so it is felt that there is
little or no real effect here. There is also a loss in modulus of approximately 15% for the
wet, cycled open-hole compression specimens. This result is discussed further in the

thermal spiking section.

The Tg was measured using transient tests in the DMA, and the results are provided in

Figure 6-9. No effects of trends of significance can be seen.

218

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Composite Composite Neat Resin Neat Resin Composite
losipescu Four Point Tensile Three Point Open Hole
Shear Flexure Flexure Compression
10
in
3 5 _
E 0 ii? T . - T ET-t i PI,
,2 T T I i I
8.
c ‘5 T
N
.C
9 -10 -
C
§
G) . .i
o. 15
-20
10
a _.
5
0: 6 —<
C
‘2 4 —
a r is
r: 2 T ‘ T
a) o » —£I:l +
U) A
C
m -2 _i I
.C
o -4 I
E‘
-e
i 1
q, -a
o.
,10 .4
-12 17 r
15
.E
S
‘7, 10 ~ 1:: Control
9 . _ - Cycled
::
.5 5 _ .' .
u. , .
'E T &l -:l
“’ ° E T
O)
i I I I
.C -
o 5
E
g .10 a
O
n.
-15 .

 

Plots showing the percent change in strength,
modulus, and failure strain values of BMI resin and
composites associated with the change from a

 

 

 

control specimen to a moisture saturuated specimen.

 

Figure 6-8 The Effect of Exposure to Moisture on the Properties of BMI Resin and
Composites

219

 

 

 

 

 

 

 

     

BMI Resin BMI Composite BMI Composite
(Peak of (Peak of (Onset of
Tan Delta) Tan Delta) Property Loss)
1
Control
- Cycled
.9
.5 0
a)
0)
C
m
.C
0
E
8
tr: '1 ‘
CL
-2

 

 

 

 

 

 

Plot showing the percent change in
the glass transition temperature of BMI
composites associated with a change
from dry to moisture saturated specimens.

 

 

 

Figure 6-9 The Effect of Moisture Exposure on Tg for BMI Resin and Composites

The effect of saturating the specimens with moisture during the experimentation for K3B
resin and composites is shown in Figure 6-10. No significant effects are seen for the
control specimens, but the wet, cycled K3B specimens show a very large loss in
properties. This is due to the blistering effect which has been discussed in Section

6.2.1.1. No effect on T3 was seen for either the resin or composites as shown in Figure 6-

ll.

220

 

 

 

 

 

 

 

 

 

 

Composite Composite Composite
Iosipescu Four Pornt Open Hole
Shear Flexure Compression
5
2
3
o "5 _
E
.E '10 '
S
c .15 —
E
0 -2o —
‘E
g -25 —
53
IL -30 —
-35
10
5 0 +
U!
5
.e .10 —
a)
.E
0 .20 —
O)
{is
5 -30 ~
0
E .40 _
i3
{‘1’ -5o —
~60
8
6 -1

[:1 Control
_ Cycled

 

 

Percent Change in Failure Strain

 

 

 

 

° 3

_. F,

Plots showing the percent change in strength,
modulus, and failure strain values of K3B

composites associated with the change from a
control specimen to a moisture saturated specimen.

 

 

 

 

 

Figure 6-10 The Effect of Exposure to Moisture on the Properties of K3B Resin and
Composites

221

K38 Resin K3B Composite K3B Composite
(Peak of (Peak of (Onset of
Tan Delta) Tan Delta) Property Loss)

 

.....
''''''

......

Blistered
Blistered

 

 

 

 

......

..s‘
' at!

.. ,
,...

 

.....
.....
.......

- .
.......
. zzz
~ .-\

-----
.....

Percent Change in T9

.....
‘_~~

 

.....
.......
..........

...... [2:] Control
_._ - Cycled

 

 

 

 

 

 

 

 

 

 

 

Plot showing the percent change in
the glass transition temperature of K38
composites associated with a change
from dry to moisture saturated specimens.

 

 

 

Figure 6-11 The Effect of Moisture Exposure on the Tg of K3B Resin and
Composites

6.5 Effect of Thermal Spikes on Final Properties

The effect of thermally spiking specimens in a 250°C oven for 15 minutes for BMI resin
and composites is shown in Figure 6-12. While some individual effects are apparent

within the various tests. no overall conclusions for the properties of the system can be

drawn.

222

The only effect which appears to be real is the loss in modulus for the wet, cycled open-
hole compression specimens. It is felt that this may be due to the formation of a
permanently damaged region around the hole due to the combination of moisture and
cycling. Since the hole will magnify the effects of any edge damage, it is a better

indicator of this than the other tests performed.

A small increase in the T1, of the system is seen in Figure 6-13 and is easily attributed to

increased cure of the resin.

223

Composite Composite Neat Resin Composite

IOSIpescu Four Point NeratResin Three Point Open Hole
Flexure ensr e Flexure Compression

; a _

 

 

 

 

    

 

 

1f

Percent Change in Modulus
c» s. L w 0 Pa b m m

 

0

U!

 

 

 

 

I
g 0 l —CU L l
II 1 1
a 511 2d L I
8
20 ,__I, L_ #

 

 

 

 

 

 

 

 

Percent Change in Failure Strain

.30 ;

 

 

Plots showing the percent change in strength.
modulus, and failure strain values of BMI resin and
composites associated with the change from a
control specimen to a thermally cycled specimen.

 

 

 

Figure 6-12 The Effect of Thermal Spiking on the Properties of BMI Resin and
Composites

224

 

BMl Resin BMI Composite BMI Composite
(Peak of (Peak of (Onset of
Tan Delta) Tan Delta) Property Loss)
6
5 _ E: Dry
- Wet

Percent Change in T9 (C)

 

 

 

 
 

 

 

 

 

Plot showing the percent change in
the glass transition temperature of BMI
composites associated with a change
from control to thermal cycled specimens.

 

 

Figure 6-13 The Effect of Thermal Spikes on the T3 of BMI Resin and Composites

The effect of thermally spiking specimens in a 250°C oven for 15 minutes for K3B resin
and composites is shown in Figure 6-14. No effect is seen for the dry specimens, but the
wet cycled specimens showed a catastrophic failure during conditioning as shown and

discussed in Section 6.2.1.1. No effect on the Tg of K3B was seen as shown in Figure 6-

15.

225

Composrte Composite Composrte
Iosrpescu Four Pornt Open Hole
Shear Flexure Compressmn

 

Percent Change in Modulus
O
i

 

 

|

 

 

 

 

 

 

 

 

 

3 ‘7
g .
o -10 ’
2 I
C
a, -20 4 I
a I
g I
z .0, 1
o
E, -40 1'
2 ~ I
8 -50 I I

so I

s
1

U

 

A
_ _._._____._._ _1'

I

 

“V“. __v_ .- _——.— _ -
r__ _—

 

Percent Change in Failure Strain

 

 

 

 

l A iii,

 

 

Plots showing the percent change in strength,
modulus. and failure strain values of K3B
composites associated with the change from a
control specimen to a thermally cycled specimen.

 

 

Figure 6-14 The Effect of Thermal Spiking on the Properties of K3B Resin and
Composites

226

 

 

 

 

 

 

 

 

 

K3B Resin K38 Composite K3B Composite
(Peak of (Peak of (Onset of
Tan Delta) Tan Delta) Property Loss)
5
4 ..
3 _
Q 2
*9 r s
.E 1 - .03 93
.2 9
g: a a
co 0 trim
.C
33 -1 —
C
a)
a. Dry
-3 - - Wet
.4 —i
-5

 

 

 

 

 

 

Plot showing the percent change in
the glass transition temperature of K38
composites associated with a change
from control to thermal cycled specimens.

 

 

 

Figure 6-15 The Effect of Thermal Spikes on the Tg of K3B Resin and Composites

6.6 Thermal Spike Conclusions

The thermal spike testing performed here was designed to determine the effect of
exposure to moisture, and thermal cycling as is likely to occur during the life cycles of

these materials. All tests were performed in a dry condition because the effect of the

227

presence of moisture has been studied in Chapter 4. The following conclusions are

apparent from the testing:

Exposure to moisture and subsequent drying of the specimens has very little
discemable effect on Tg or the mechanical properties of BMI specimens, or dry K3B

specimens.

Thermal spiking causes a small increase in the Tg of the BMI specimens. This would
be expected because an exposure to a temperature of 250°C represents enough of a
thermal event to increase the cure of these specimens. Thermal spiking has no effect

on the Tg of K3B resin or composites.

The wet, cycled BMI open-hole compression samples showed a modulus loss of
approximately 15% suggesting that a large damaged region around the hole may be

formed during this exposure.

Finally, wet, cycled K3B specimens showed a catastrophic blistering failure during
the thermal spike conditioning. The specimens showed blisters which caused the
composite specimen to double in thickness. Since the 250°C exposure is above the Tg
of this resin, it is very apparent that the wet TE needs to be known, and all designs

using this material must remain well below this temperature.

Chapter 7

IMPROVEMENT OF BMI/IM7 PROPERTIES

7.1 Introduction

The cure cycle of BMI/IM7 composites has been shown to cause tensile stresses in the
transverse direction of cross-ply laminates. Calculations (FEM and Analytical) show in
some cases, these stresses are insufficient to cause cracking in the plies, but cracking does
exist suggesting that the presence of the weak interfacial bond between fiber and resin is
partially to blame for the microcracking. Relief of cracking could, therefore, be

accomplished through either increased adhesion, or a strain absorbing mechanism.

The adhesion in the BMI/1M7 system has been measured using a variety of techniques.
Direct evidence of poor adhesion was obtained by ESEM imaging of Mode I crack
opening failure surfaces. The revealed fibers were completely devoid of resin which is
seen in adhesive failures. Images of both thermally and mechanically induced cracks also
show a completely adhesive failure. Four point flex testing of composites revealed very
poor interlaminar shear strength as evidenced by the mid-plane shear failure of flexure

samples with a 28:1 span to depth ratio. IFSS values measured using the Interfacial

228

229
Testing System (ITS) were less than 10% of the values for standard epoxy/carbon fiber

systems.

All of these observations verify the need for increased adhesion in this system. The
following section describes the factors which affect the IFSS and reviews work
performed on various related systems. This forms the basis for the testing approach taken

here.

7.1.1 Factors Affecting the Interfacial Shear Strength (IF SS)

Since this effort concentrates on the adhesion in these systems, it is important to
understand all of the factors which can effect the IFSS. Rao and Drzal used extensive
single fiber fragmentation testing of epoxy/A84 to show that the IFSS decreases with
decreasing matrix modulus [43], and decreases as temperature rises with a large decrease

as the Tg of the matrix is approached [44].

Similarly, Varma, Needles, Kourtides, and Fish [36] showed a decrease in interlaminar
shear strength (ILSS) with increased temperature and attributed it to a reduction in IFSS.
Pegoraro and Landro [45] used single fiber fragmentation to show a decrease in adhesion
with increased temperature for the PEI/carbon fiber system. They attributed this to a
reduction in the residual compressive stress formed during cure. Unfortunately, their
analysis did not include the effect of nearest neighbor fibers or ply level interactions

which will severely change the resultant stress magnitudes and patterns [46].

230
Carman and Averill [46, 27] have shown that an optimum interphase can be designed to
increase the transverse strain to failure by a factor between three and six. The interphase

is optimized in terms of modulus and thickness.

7.1.2 Adhesion Improvement In BMI/IM7 And Related Systems

There has been much interest in the adhesion of BMI to carbon fibers. While the focus
and results of the available literature vary, it is generally acknowledged that the adhesion
is poor. Due to this a variety of approaches have been employed to improve the system's
performance. These fall into four major categories: (I) changes in fiber surface energy,
(11) changes in functional group concentration, (III) resin chemistry changes

(toughening), and (IV) fiber sizing.

7.1.2.1 Surface Energy Changes

Fiber wetting is a requirement for most composite manufacture. Without it composites
generally have poor adhesion and porosity. The wetting between two materials can be
predicted using thermodynamic surface energy concepts. Generally liquids will wet
solids if the liquid's surface energy is lower than that of the solid. A relatively weak but
functionally important bond can be formed between surfaces due to dipole-dipole,

dispersive, or acid-base interactions.

231
Krekel, Huttinger, and Hoffman [47] calculated the work of adhesion for fibers with
varied surface treatments and correlated these directly to the concentration of carboxylic
acid groups as measured by thermal desorption of CO2 species. They suggest that the
work of adhesion resulting from carboxylic acid groups is a better predictor of adhesion
to thermoplastics than CO2 desorption losses because CO2 can be produced by other

surface species.

In a series of articles. Nardin et al characterized the reversible work of adhesion (W) as
the sum of the dispersive and acid-base components of PEEK and carbon fibers of
various surface treatments [48]. They then found a linear correlation between W and the
interfacial shear strength as measured with the single fiber fragmentation test [49].
Finally [50], they related W to the transverse tensile strength of composites. Here they
found a weak correlation because the fractures were cohesive through the resin, not
adhesive in nature. The correlation which was seen was explained as an effect of the

improved surface bond on the nature of the resin in the failure region.

Bucher and Hinkley [51] related transverse fiexural strength to wetting angles in carbon
fiber/PEKK composites. They found that the 030-500 fiber exhibited the lowest contact

angle of fibers tested. and thus best transverse strength.

Jangchud et a1 [52] found that the ILSS of PMR-lS/carbon fiber composites correlated

well with the polar nature (% weighted dipole moment) of the various fibers tested. They

232
suggest that increased bonding can be achieved by increasing the polar components in the

resin.

Heisey et al [53] attributed an increase in the adhesion of bismaleimide resin (Matrimid
5292‘“) and a phosphorus containing BMI) to treated A84 and AU4 carbon fibers to
surface energy increases of the fibers. Unfortunately, the microbond single fiber pull-out
technique used to measure the adhesion caused fiber failure more often than not, and the

results were too scattered for any conclusive observations.

7.1.2.2 Surface Morphology Changes

Many surface treatments also change the morphology of the fiber surface as seen by
SEM, AFM, STM, and surface area analysis (Kr adsorption). Increased surface area has
been proposed to aid in adhesion by promoting mechanical interlocking of the resin and

fiber surface.

O'Kell and .I ones [54] found that air and nitrogen plasma both roughen the surface on the
nanometer level. Jangchud et al [52] were not able to correlate surface morphology as
measured by AFM to ILSS for PMR-lS/carbon fiber systems. They suggest that

correlations might exist with dimensional scales below 25nm.

Dzenis et al [55] performed AF M analysis of Kevlar”, graphite, and polyimide fibers and

determined that the surface roughness is fractal in nature, and thus the interphase should

233
be fractal in nature. They propose that studying this fractal nature will aid in the
understanding of stress transfer between fibers and resins and ultimately help in the

design of appropriate topographies for efficient stress transfer.

7.1.2.3 Surface Chemistry Changes

Surface treatments have been used widely to increase the adhesion between carbon fibers
and matrices. There are generally four types of surface treatment: gas phase, liquid
phase, electrochemical, and plasma treatment. Donnet [56] provides a review of these
methods and their effect on fiber and composite properties. The most characterized
change is in the concentrations of surface functional groups, which has been readily

correlated to adhesion in many systems.

It is well known that nitric acid treatment of carbon fibers increases the concentration of
acidic surface functional groups, decreases the fiber tensile strength, and causes fiber
weight loss [57]. This change in surface chemistry has been readily correlated with

increased adhesion in many composite systems [6].

Testing of BMI/IM7 fiber composites with electrolytic anodization in ammonium
bicarbonate treatment varying from 50% to 400% showed no significant affect on
adhesion as shown by the purely adhesive failure of Mode I fractured composites
(internal work). This suggests that an increase in the surface functionality is insufficient

to cause significant property improvement.

234
Plasma treatment of carbon fibers has been performed with a variety of gasses, including
air [7, 53, 58, 59], Argon [7, 60], Nitrogen [7], Ammonia [53, 60], and oxygen [53, 60].
The results generally conclude that plasma treatment increases the functionality and/or
energy of the fiber surface. These increases have been correlated with adhesion
improvements in many cases [7, 53, 58, 60]. O'Kell and Jones [54] determined that
nitrogen plasma can only chemically interact with edge sites of graphite, but may activate

the basal planes, allowing oxygen adsorption when exposed to air.

Oxygen and air appear to be the most effective at increasing the oxygen containing
functionalities which are generally considered to be responsible for increases in adhesion.
It has also been noted, however, that oxygen plasma tends to etch the fiber surface more

dramatically, and thus reduce fiber tensile strength[60].

Chang [60] proposed that Ammonia/Argon plasma to be the most effective in increasing
adhesion between carbon fibers (AU4) and BMI (Matrimid 5292) without decreasing
fiber strength. His data, however, shows higher transverse tensile strength for oxygen
treated fibers with only a 20% reduction in fiber tensile strength after a full 20 minutes of
treatment. While the plasma treatments here may have been effective, 20 minutes of
plasma treatment is not commercially viable, and others suggest that most active sites are

oxidized after just 60 seconds.

He suggested that the effectiveness of the Ammonia/Argon plasma is due to a Michael

Addition reaction between the amines formed on the surface, and the maleimide groups in

235
the BMI. It should be noted that this study was performed using type AU4 fibers, and
Drzal has shown [6] that AU4 fibers have a weak boundary layer which limits apparent
adhesive strength. The removal of this layer by long time oxygen plasma may account

for the increase in transverse tensile strength seen here.

Heisey et al [53] found that 15 seconds of air plasma treatment caused a twofold increase
in the surface oxygen content of the fibers (9.3% to 19.8%) with only a 7% loss in fiber
strength. A one minute NH3 plasma treatment caused an almost twofold increase in
nitrogen content (2.7% to 5.2%) with no significant loss of fiber strength. The air plasma
treatment caused an increase in Na content from 0.0026% to 0.084% after five minutes
suggesting an etching of the surface. Both treatments caused a significant increase in the
overall surface free energy which was manifested in the polar component. SEM images
at 50,000x showed that the air plasma treatment had roughened the surface, but the NH3
had not. Unfortunately, the adhesion measurements to the BMI resin were inconclusive

due to experimental difficulties.

7.1.2.4 Resin Chemistry Changes

As noted previously, changing the modulus of the matrix material can have a very
significant effect on the IFSS which can, in turn, have a major effect on macroscopic
composite properties. Many people have studied methods for improving the toughness of

high temperature resins.

236
Lin and Chen [61] mixed BMI with TGDDM and DDS (as a curing agent) and found an
almost 3x increase in resin fracture toughness due to a simultaneous interpenetrating

network, but sacrificed thermal stability.

Morgan et al [2] added PEI and Kevlar!" 49 whiskers to the Matrimid 5292® BMI/IM7
carbon fiber system and found that the PEI formed a 1 pm diameter second phase within
the BMI which caused an ~30% increase in Glc due to crack bridging. The Kevlar‘D fibers
caused an increase in Gk, but with a lower crack initiation stress. The increased
toughness is due to a more torturous crack path while the lower initiation stress is most

likely due to a lack of adhesion between the Kevlar® and resin.

Cartwright et al [62] added maleimide terminated polysulfone to the Matrimid 5292‘11D
system and studied the cure using FTIR. They cite literature which suggests that the size
and morphology of a second phase is due to competition between the thermodynamic
driving force for phase separation and kinetic inhibition due to cure advance. They also

cite work which has used microwave processing to produce novel morphologies.

The BMI system being used in this study (Ciba-Geigy's Matrimid 5292®) is a two part
system consisting of 4,4' bismaleimidodiphenyl methane (BMI) and 0,0' diallyl
bisphenol A (DABPA). Guozheng, Aijuan, and Liwen [63] showed that the addition of
four different allyl compounds (including DABPA) improve the processability and

toughness of BMI while retaining thermal capability and wet performance. Continuing

237
work at MSU has shown that simply varying the ratio of the BMI and DABPA monomers

can have a significant effect on final neat resin properties.

Wilkinson, Ward, and McGrath [64] mixed five thermoplastics into the Matrimid 5292
system and reported the thermal, mechanical, and morphological changes for both neat
resins and composites. They found that the Tg of the modifier controls the onset of
decline in storage modulus of the cured resin and thus should be considered carefully
when choosing a modifier. Higher molecular weight modifiers showed further increased
fracture toughness (KIC). A phase-inverted morphology was noted for all modifiers, but
was not shown to be required for the increased toughness seen. Resin toughness
translated into composite toughness well, and the addition of modifiers to the melt did not

impede the use of standard composite prepreg and cure methods.

7.1.2.5 Fiber Sizing

Since the probability of significantly increasing adhesion in this system via fiber surface
treatment appears low, an alternate method is required. The addition of an interphase has
great potential because its composition and properties can be tailored to provide adhesion
to the fiber, miscibility with the resin, and overall composite property improvement. The
property improvements can come about through an adhesion improvement and/or a strain

absorption mechanism.

238
Jenkins et al [65] showed that the use of a thermoplastic sizing on carbon fibers reduced
the transverse microcrack density, increased the transflexural strength, and improved the
thermo-oxidative stability of PMR-IS composites. The thermoplastics used were

polyisoimide, polyarylsulphone, and an imidised polyarylsulphone.

Since one of the requirements for a good interphase material is bonding to the fiber, the
work on adhesion between thermoplastic matrices and carbon fibers provides a wealth of

useful information.

Denault [66] found that the interfacial strength between thermoplastics and carbon fibers
was primarily due to a chemisorption mechanism. This was supported by the fact that
polypropylene did not form a good bond until degraded, while polyamide-12 readily
formed a good bond. Polypropylene has many fewer reactive groups than polyamid-12.
It was also supported by the large loss in short beam shear strength of PA-12 composites
after cure cycles above 350°C where a large number of the carbon fiber surface functional
groups have thermally evolved from the surface. Unfortunately, degradation of the resin
may have contributed to this result in addition to the evolution of the surface

functionalities.

A large amount of work has been focused on understanding the way interphases affect
composite properties, and what the optimum interphase should be. For epoxy sizes in

epoxy matrices, it was seen that a low molecular weight sizing led to the highest

239
interfacial shear strength [67]. This was attributed to the ability of the sizing to diffuse

into the resin.

Carman and Averill [46] showed that the modulus of the interphase plays a large roll in
its ability to increase the composites ability to absorb strain. This would strongly suggest
a molecular weight dependence. Unfortunately, it is also known that for high molecular
weight materials chemical bonding with the surface is unlikely. Pittman et a1 [68] also
showed that low modulus interphase materials provide the greatest increase in impact
strength (toughness) and that interphase thicknesses show a maximum effect on impact

strength in the range of 1000-1500 angstroms [69].

Raghavendron and Drzal [70] have shown that lower molecular weight materials have a
thermodynamic tendency toward the fiber surface, and thus graduated interphase

properties may be readily achievable.

Rajagopalan et al [71] showed that the amount of diffusion of polysulfone into epoxy had
a strong effect on adhesive properties. A wider diffusion zone was achieved with higher
initial processing temperatures and increased the fracture toughness of the joint by 35%.
The size of the diffusion zone and thus toughness of the joint can be seen to be a function
of the miscibility of the two polymers. Carman and Averill’s work showed that the size
of a distinct interphase must be tailored to provide a maximum strain to failure for

transversely loaded composites.

240

7.2 Treatment of IM7 Fibers for Increased Adhesion

With all of this background available, attempts were made to increase adhesion through

both fiber surface treatments, and the addition of various Sizings.

7.2.1 Fiber Surface Treatment

To determine the effect of changing the fiber surface chemistry and/or morphology, two
different surface treatments were studied. IM7 fibers with varied levels of commercial
treatment were provided by Hercules, Inc. Air plasma treatment of the unsized IM7 fiber
was performed and evaluated due to the promising results seen in the literature. Before
discussing these treatment methods. it is imperative to discuss a few experiments on the
thermal and chemical stability of the surface functional groups which are formed by these

processes.

7.2.1.1 Chemical Bonding of K3B monomers During Processing

A study has been performed in which monofunctional counterparts of the K3B
constituent materials were reacted with the fiber surface to determine the type and extent
of chemical reactions between K38 and IM7 [72]. These model compounds were
phthalic anhydride, pyromellitic dianhydride, APB-133, and 2-phenyl-APB-l44. Fiber
bundles were sealed in containers with a solution of each component and n-methyl
pyrrolidone (NMP, a solvent used in the processing of K3B). The containers were then

subjected to the 350°C thermal cycle used in the processing of K3B. The fiber bundles

241
were then washed in hot NMP to remove all physisorbed species, leaving only those

bonded to the surface.

No evidence of the model compounds was found after processing, and it was also found
that in all cases, the amount of oxygen containing species (which would account for the
majority of the chemical bonding in this system) decreased by 50%. The fact that this
occurred in the control specimens (N MP only) suggests that the combination of NMP and
a 350°C thermal cycle will reduce the number of active bonding sites. NMP without the

350°C thermal cycle was shown to have little effect.

7.2.1.2 TPD showing chemical species Desorption

Since the previous study showed that a temperature of 350°C and an NMP environment
strips surface functionalities from the IM7 fiber, a temperature programmed desorption
(TPD) was performed to determine whether this was due to the temperature alone, or if
both the temperature and the NMP were required. The fiber used for this study was

400% surface treated IM7. The resultant desorption profile is provided in Figure 7-1.

242

0.25

 

0.2

O

.15

 

 

(CC/9)

0.1

Desorbed Material

0.05

 

 

o 200 400 600 800 1000 1200 1400 1600
Temperature (C)

Figure 7-1 - TPD Profile for 400% Surface Treated IM7 Fibers

The TPD was performed by Zhigang Zhang in the Chemical Engineering Department of
MSU. Experimentally, a section of fiber was placed in a ceramic tube in a tube furnace.
The tube was attached to a mass spectrometer which analyzed the gasses which evolved
from the surface as a function of furnace temperature. The gasses detected were CO,
C02, and N2. The CO2 is generally attributed to the decomposition of carboxylic groups
while the CO is attributed to carbonyl and hydroxyl groups. It is generally believed that
the carboxylic species are more active and thus have a greater contribution to adhesion.
In this case, it is shown that temperatures of 300°C will cause a loss of less than 10% of
these groups, and at 350°C, less than 20% are gone, but the species lost at lower

temperatures are likely to be the more chemically active species.

243
This may suggest that the combination of temperature and NMP removes more surface
functionality from the carbon fiber surface then the temperature alone. TPD results for
similar fibers [56] also show that the amount of functionality lost solely due to thermal

desorption is lower than found in the NMP solvent results.

7.2.1.3 Commercial Treatment

Carbon fibers are generally provided with a surface treatment which alters it's surface in a
variety of ways. It has been shown that the treatment removes an outer "skin" or weak
boundary layer, and increases the surface functionality of the fibers. The IM7 fibers used
here were received from Hercules with treatment levels of 100% and 400%. These levels
refer to a percentage of the standard commercial treatment which is an anodic
electrolization in an ammonium bicarbonate bath. The amperage was varied to produce
the different levels. Unfortunately. this surface treatment process is proprietary, and

further information is unavailable.

The effect of treatment series like this has been studied previously for a different resin
system. Waterbury showed that the concentration of oxygen containing species increases
linearly with the amount of treatment after an initial jump. This trend was also seen in
the polar surface free energy. There was a direct correlation between these factors and
the adhesion between the fiber and matrix as measured by the single fiber fragmentation

test. Thus, once the initial treatment is performed, large changes in adhesion are not seen.

244

7.2.1.4 Air Plasma Treatment

As shown at the beginning of this chapter, some success has been reported for increasing
the adhesion of high temperature resins to carbon fibers through the use of air plasma
treatments. Unfortunately, the adhesion data is consistently suspect, and the plasma
treatments used generally cause a significant reduction in fiber tensile strength. The high
energies and long treatment times used in these studies should not be necessary to cause

very significant increases in surface oxygen containing species.

A Plasma Sciences plasma chamber was used. The unit was run under experimental
conditions for approximately five minutes before treatment of these fibers to clean the
chamber of any possible contaminates, and check for irregularities in the plasma field.
Fiber tows of approximately 8" in length with aluminum foil holding the ends together
were laid in the chamber between glass bars which held the tows off the chamber floor.

The parameters used for the plasma run are shown in Table 7-1.

Table 7-1 Plasma Chamber Experimental Parameters

 

 

Parameter Level
Power 55 Watts
Process Time 30 seconds
Base Pressure 0.05 torr
Gas Flow Rate 100 cc/min
Temperature 26°C
Purge Slow Purge to Air over 40
minutes
Vent Time 3 min maximum

 

 

 

245
After processing, the fiber tows were removed and placed in clean aluminum foil until
analyzed with the XPS, and manufactured into a unidirectional composite for adhesion

measurements.

7.2.1.5 XPS Analysis of Treated Fiber Chemistries

To quantify the difference in the surface chemical nature of the fibers used here, XPS was
performed on a baseline fiber tow, as well as the air plasma and 400% commercially
treated fibers. The XPS experiment was carried out using a Physical Electronics 5400
ESCA spectrometer. The instrument was equipped with a monochromatic Al x-ray
source (15kV, 40 mA, 600W). Signal collection utilized a 180° hemispherical analyzer
operating in the fixed analyzer transmission mode at a pass energy of 89.45 eV for survey
scans and 35.75 eV for high resolution scans. and a position-sensitive detector for signal
detection. The signal was digitized and processed using the Physical Electronics V4.0
ESCA software. Spectra were collected over an analysis area of 1.5 mm x 5 mm with the
sample surface oriented at 65° relative to the analyzer lens (sampling depth of the order

of 70A). The samples were oriented parallel to the x-ray source and analyzer aperture.

The instrument was calibrated using Mg K,I x-ray radiation where the Ag 3d,,2 peak had a
binding energy of 368.2 i 0.1 eV with a full width at half maximum (FWHM) of 1.06 eV
at 35.75 eV pass energy and 515,856 counts per second, with a rectangular analysis area
of 1.5 x 5 mm in the small-area, large solid angle aperture. The difference between the

Cu 2p3,2 and Cu 3p.)2 was 857.75 eV at 17.9 eV pass energy. The binding energy of Au 4f

246
was 84.0 eV. Using the monochromated Al KO, x-radiation, the F WHM of sputtered Ag
3d,,2 was 1.06 eV at 35.75 eV pass energy and 1.5 x 5 mm analysis area, with a count rate

of 45,476 counts per second.

Surface atomic concentrations were used to calculate the surface composition and thus,
the surface oxygen/surface carbon ratios. The results of the XPS testing for the 100%,
400%, and plasma treated samples are provided in Table 7-2. It can be seen that the
400% commercially treated fibers had the highest surface oxygen content, and the air

plasma treatment provided only a small increase in the functionality.

Table 7-2 Surface Composition for All Treated Fibers

 

Carbon Oxygen Nitrogen Sodium

 

100% Baseline 83.0 11.6 4.8 0.5
400% Commercial 73.] 20.2 3.5 0.3
30 sec Air Plasma 79.7 16.2 3.6 0.5

 

 

 

 

 

 

 

7.2.1.6 Surface Treatment Conclusions

Both the higher levels of commercial treatment, and the plasma treatment have been
shown to increase the amount of oxygen containing surface groups. The thermal stability
of these species has been demonstrated by thermally programmed desorption. The ability

of this increased functionality to increase the adhesion in the BMI/IM7 system was

247
investigated using the Interfacial Testing System, and Mode I failure analysis as

discussed in Section 7.4.

7.3 Sizing of Carbon Fibers

A thermally stable, compliant, adhesion promoting sizing has been sought with the goals
of improved adhesion, reduced microcracking, and thermal stability. Unfortunately this
combination of properties is rather self-limiting in that thermal stability of polymeric
carbon chains is generally directly related to their modulus and the available high
temperature resins generally show very little adhesion to carbon fibers. Several high
temperature thermoplastic polymers and a silicone resin have been investigated as Sizings

here. The specific reasons for their investigation are provided below.

7.3.1 Use of Thermoplastic Sizings

As discussed in Chapter 2, the microcracking stresses can be relieved through the
addition of a compliant sizing. The thickness of this sizing was shown to be most
effective when on the order of 0.5 microns for this system. It was also shown that the
effect of Sizings which have approximately the same moduli as the bulk matrix are
relatively insensitive to thickness. The ability of the thermoplastic Sizings to bond with

both the fiber and matrix is of much greater importance.

248

7.3.1.1 Sizing Chemistries and Properties

Three thermoplastic materials were chosen as candidates for Sizings in this system. They
are polyetherimide (PEI), polyethersulphone, and Matrimid® 5218, a thermoplastic
polyimide. All materials were chosen due to their ability to meet the service and

processing temperature requirements. The properties of these resins are provided in

 

 

Table 7-3.
Table 7-3 Physical Properties of Sizing Materials
Matrimid Matrimid Ultem 1000 Victrex GE RTV-31
5292 BMI 5218 PEI PES
E(MPa) 1828° 2895T 30005 2410b 2.29cl
v 0.35 0.35 0.44b 0.35c
CTE| 42x10"‘/°Cd 28x10'°/°Cb 56x 10‘°/°Cb 55x10‘°/°C 200x10*‘/°Cb
Tensile 94.1 MPa 86 MPa 105 MPa 70.3 MPa 6 MPa
Strengthb
Elongation 6.66% 48.6% 60% 75% 170%
@ Breakb
Tg 350°Cd 326°Cd 226°Cd 229°Cd
MW - 5200 -

 

 

 

 

b Product literature values
c Estimated based upon tensile modulus
d Experimentally measured

The Chemistries of the PEI and PBS polymers are provided in Figure 7-2. The general
chemical structure of thermoplastic polyimides such as Matrimid® 5218 and the backbone

chemistry of PDMSO are provided in Figure 7-3.

249

polyether-imide chemical structure

polyethersulfone chemical structure

Figure 7-2 The Chemistries of PEI and PES

 

polydimethylsiloxane chemical structure

Figure 7-3 The Chemistries of PDMS and Thermoplastic Polyimides

250

7.3.1.2 Sizing Method

The thermoplastic Sizings were applied to the carbon fiber tow using a solution chemistry
method. First, solutions of 1% polymer in dichloromethane were prepared.
Approximately one yard of carbon fiber tow was wrapped around a glass rack and kept in
place with nickel wire. The sizing solution was poured into a clean polypropylene basin
and the rack was immediately immersed in the solution. The rack was removed after one
minute and slowly spun in air to dry the tows. All three sizing solutions showed a very
good ability to spread the fiber tow and impregnation was not an issue. After drying in
air for fifteen minutes, the sized fibers were placed in an air oven at 60°C for one hour to

ensure removal of the solvent.

The thickness of the sizing was approximated using geometric calculations based on the
weight gain of the fibers, density of the resin, density and radius of the fiber. This

estimate showed that sizing thicknesses were on the order of 0.1 pm.

7.3.1.3 Sizing Results

The sized fibers were analyzed for morphology using the ESEM. Each sizing had a skin
on the outer surface of the tow as shown for the PEI sizing in Figure 7-4. The sizing
penetrated the tow and coated the fibers as shown in Figure 7-5, again for the PEI sizing.
Figure 7-6 shows a case where the sizing formed a bridge between fibers, and still coated
the whole fiber. A similar image for the Matrimid 5218 sizing shows the presence of the

sizing on the fiber surface (Figure 7-7) while it can also be seen when looking at the edge

251
of a fractured fiber (Figure 7-8). The conclusion to be drawn from the sizing
morphologies is that the sizing method was sufficiently successful at coating the fibers
for this analysis, but better process is achievable and should be used if these methods are

found to be of interest.

 

Figure 7-4 Outer Skin of PEI Sized IM7 Tow

252

 

Figure 7-5 PEI Sizing on an IM7 Fiber

253

i ‘8
. ‘4: .. ,f,
itiiééi"

 

Figure 7-6 PES Sizing on an IM7 Fiber

254

 

Figure 7-7 Matrimid® 5218 Sizing on an IM7 Fiber Surface

255

 

Figure 7-8 Matrimid® 5218 Sizing on an IM7 Fiber Edge

256

7.3.2 Use of a Compliant Sizing

A compliant material which was capable of withstanding the use temperatures of 200°C
was desired. The only materials found which meet this description are high temperature
silicones based on the polydimethylsiloxane (PDMSO) backbone shown in Figure 7-3. A
sample of General Electric's RTV-31 was obtained from GB. The exact contents of this
material are unknown, but it's general contents are given as 30-60 wt% PDMSO, 10-30
wt% diatomaceous earth, 1-5 wt% ethyl silicate, and 30-60 wt% red iron oxide. The
material is crosslinked with the aid of a dibutyl tin dilaurate (DBT) curing agent. This is

a moisture activated catalyst.

7.3.2.1 Sizing Method

Initially, the "base compound" of silicone is a thick red viscous liquid. The DBT curing
agent is a liquid which is provided in a small tube. To get the RTV onto the fibers and
then cured into a lightly cross-linked network, the base compound was put into a 1.25
wt% solution with toluene as the solvent. Three drops of the DBT was added to the
solution. This is the appropriate amount of DBT for the amount of base compound in
solution. The solution was immediately poured into the clean polypropylene beaker and
fibers were pulled through the solution as quickly as possible and hung to dry. The fibers
were handled only using the nickel wire which was used to tie the ends. After drying for
eighteen hours in air, the fiber tows were subjected to a thermal cycle of 150°C for 1 hour

and 200°C for 1 hour because there is an evolution of formaldehyde reported at 150°C

257
and the evolution of volatiles during composite processing causes voids and generally

poor laminate quality. A maximum weight loss of 1.3% was seen.

A macroscopic piece of RTV was created at the same time the sized fibers were made to
help ensure that the silicone became crosslinked, and to allow the determination of a
modulus value. This was made by mixing the compound and curing agent in the same
ratio as used for the sizing. After this mixing, 40% by weight of the sizing solution was
added to the mixture. This was done to simulate the effect of the solvent on the
processing of the sizing. The specimen was dried in air until solid. The specimen was

fully solid and the crosslinking reaction was successful.

Small strips of this piece of silicone rubber were cut and tested on a tensile testing
machine to provide accurate modulus values for use in the modeling conducted in
Chapter 2. Three specimens were tested, and the average modulus value was 2.29 MPa.
Specimens were not tested to failure, so strength and elongation at break values were not

obtained.

7.3.2.2 RTV Sizing Results

Small sections of the sized fiber were analyzed in the ESEM and the resultant sizing
morphology was determined. It was found that in general, the RTV did not provide a
uniform coating on the fiber, but adhered to the fiber in small patches as shown in Figure

7-9 and Figure 7-10. While this is not the ideal coating desired to determine the

258
effectiveness of a compliant sizing at reducing strain in composites. the use of a silicone
sizing has already broken one of the assumption of the analysis used in Chapter 2.
Silcones are well known for providing very poor adhesion to surfaces, and thus failure

would be expected at either the fiber/sizing interface. or at the sizing/matrix interface.

., 3.423,».- -. «0972/31

1-' In

FIT'u'SZTFJE.

 

Figure 7-9 RTV Sizing Between Two IM7 Fibers

259

 

Figure 7-10 RTV Sizing Patches on IM7 Fibers

7.3.2.3 Composite Preparation

A cross-ply composite was made from the RTV sized fibers to determine if the sizing was
capable of reducing the strains either through it's compliant nature. it's poor adhesion, or a
combination of the two. The morphology of the sizing was not known at the time. Since

only a small amount of sized fiber had been created, a small (34 ml) cross ply laminate

260
was fabricated. To ensure that any changes in the microcracking behavior were not due

to the processing method, a control laminate was also fabricated using the unsized IM7

fiber.

Both the sized and unsized fibers were "prepregged" by being pulled through a bath of
BMI resin and laid on telflon release film. The unsized fiber showed good wet-out of the
tow, while the RTV sized fiber showed very poor wet-out. The cross-ply was fabricated
by cutting 3%: in lengths of tow and placing them together to form a unidirectional sheet.
This was done on release film on a hot plate so that the resulting % in2 prepreg sheet could
be removed. Four sheets were made and they were then stacked in a [0°/90°]S
configuration. The sized fiber tows were spread out on the hot plate to promote

impregnation of the resin.

The two laminates were then vacuum bagged in the same manner as presented in Chapter
2 for the macroscopic panels. This vacuum bag was then placed in the programmable
Nitrogen purged oven and a vacuum was pulled on the specimens for the same cycle as
used in the creation of the macroscopic panels. The only difference in the processing of
these laminates is that the 100 psi of pressure was not applied. The cure cycle was
5°C/min to 177°C for 90 minutes and 5°C/min to room temperature. The standard post-
cure cycle of 200°C for 2 hours and 250°C for 6 hours was performed in a Nitrogen

atmosphere after the specimens were removed from the vacuum bag.

261
The RTV panel quality was poor in terms of high performance composites, but more than
sufficient for the objectives here. A section of this panel was mounted in epoxy and
polished for analysis of panel quality and ITS testing. The remainder of the specimen

was kept for Mode I failure analysis.

7.3.2.4 Composite Morphology

The polished section of RTV sized cross-ply composite was examined both using an
optical microscope and in the ESEM. Optically, it was apparent that the laminate
contained a fairly large number of voids and some regions where the matrix resin did not
impregnate the tow. The laminate did show good bonding between the layers, and had
some evidence of both resin shrinkage stresses, and ply level expansion mismatch

stresses.

ESEM analysis of this specimen was difficult, but images of the pertinent features were
obtained. A region showing a resin rich region where the resin has pulled away from the
surrounding fibers is shown in Figure 7-11. Another region shows the formation of 3 ply-
level microcrack (Figure 7-12). This was one of the two such cracks found on the whole

'/2" length.

262

 

Figure 7-11 Cracking of RTV Sized Composite Caused by Resin Shrinkage

263

F: T U E [I H131 T" . T I F

 

Figure 7-12 A Microcrack Within the RTV Sized Cross-Ply Composite

(The outer plies run in the direction of the 100 micron bar)

7.4 Measurement of Adhesion Levels for Treated and Sized Fiber Composites

The effect of increased levels of commercial surface treatment was measured by use of
the interfacial testing system (ITS), and by ESEM analysis of a Mode I crack opening

failure.

264

7.4.1 Unidirectional Specimen Manufacture

Unidirectional specimens were made using the three types of thermoplastic sized fiber,
the 400% treated fiber, and the air plasma treated fiber, as well as the baseline 100%
treated and unsized fiber. The specimens were made by placing three inch lengths of the
fiber tows in a silicone mold. The mold had previously been post-cured at 200°C and
thus was stable for the 177°C cure used here. BMI resin was then heated to 130°C,
degassed in a vacuum oven, and poured over the pre-heated molds. The molds were then

placed in the vacuum oven and degassed more to facilitate wetting of the fibers.

The molds were then placed in the programmable Nitrogen purged oven an subjected to 3
177°C cure for 90 minutes. The thermal cycle was verified using a thermocouple
attached to a data logging computer. After this cycled, the specimens were removed from
the molds and were subjected to the standard post-cure cycle of 200°C for 2 hours and
250°C for 6 hours. All ramp rates were 5°C/min. Finally, the specimens were removed
and '/2" pieces were cut for analysis. One piece of each was mounted in epoxy for ITS

testing, while a similar piece was set aside for Mode I fracture and analysis.

7.4.2 ITS Testing

The ITS testing method was identical to that presented in Chapter 4. The results are

provided in Table 7-4 and are plotted in Figure 7-13.

265

Table 7-4 ITS Results for Improvement Attempts

 

 

RTV Sized PES Sized PEI Sized 5218 Sized Air Plasma Control 400% ST
TFSS‘(—‘)psr 804 1314 867—986—1117_955—r323—
Std. Dev. 230 203 164 231 207 247 260

 

 

 

 

1600

 

1400 . _

1200 ,7, 7.7 . 7 7 i<i

1000iiii if

 

IFSS (pal)
I

 

60er 7 7

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

RTV Sized PES Sized PEI Sized 5218 Sized Air Plasma 400% ST Control

Figure 7-13 Plot of ITS Results for Treated and Sized Specimens

It can be seen that the adhesion levels are still very poor. The data scatter prevents any
definite conclusions, but it is apparent that the PES sized fibers and the 400% treated

fibers aided in the adhesion of the IM7 fiber to the BMI resin. The increase seen for the

266
400% treated fibers is attributed to an increase in the surface roughness because no
chemical bonding is expected in this system as discussed earlier. The air plasma showed
a slight increase which appears to track with the slight increase in oxygen content, but
also can be explained as a surface roughness phenomenon. Air plasma treatment of

carbon fibers is known to etch the fibers significantly.

Finally, the RTV sized specimens showed a small drop in the adhesion. This is expected

because of the poor bonding generally experienced with silicones.

267

7.4.3 Mode I Failure Analysis

Small pieces of the unidirectional specimens were used for this analysis. Specimens were
failed in a Mode 1, crack-opening manner by initiating the crack with a razor blade and
then forcing the specimen apart. This was performed by hand, so no mechanical data are
available. One fractured surface was then analyzed in the ESEM. No surface coating or

preparation was performed.

The objectives of the analysis were to determine the locus of failure around the fibers and
note any differences in failure morphology which may provide information about the
level of adhesion in the system. The conclusions of this study are provided in tabular

form in , and then images supporting the conclusions are provided.

Table 7-5 Sizing Improvement Summary

 

 

 

Adhesion Level Failure Locus Interdiffusion?
100% Unsized Poor Fiber/Matrix ---
400% Unsized Improved Fiber/Matrix ---
Air Plasma Poor Fiber/Matrix ---
Matrimid 5218 Poor Fiber/Sizing Yes
Ultem PEI Poor Sizing/Matrix ???
Victrex PES Improved Sizing/Matrix Yes

GE RTV-31 Poor F iber/Sizing No?

 

The failure of all of specimens, both sized and unsized, showed the same general

appearance at relatively low magnifications as seen in Figure 7-14 and Figure 7-15. The

 

268
poor adhesion of the 100%, unsized control specimen and the air plasma treated specimen
is shown in Figure 7—16. Figure 7-17 shows two images of 400% commercially treated

fibers which exhibit better adhesion, although the adhesion is still generally poor.

The adhesion levels for the sized specimens are shown in Figure 7-17. It can be seen that
there is poor adhesion in the Matrimid® 5218, PEI, and RTV sized systems. The PES
system, however, exhibits evidence of better bonding. It is important to understand
whether the failure was between the fiber/sizing interface, or the fiber/matrix interface,

and to what level interdiffusion of the sizing occurred.

Regions in which the sizing existed as a second phase were found for the PES and
Matrimid® 5218 systems. Close examination revealed good adhesion between the sizing
materials and the BMI matrix as shown in Figure 7-18. The ductile failure mode of these

materials is also obvious from the images.

To determine locus of failure of the sizing, beam etching was used. The beam was
focused on the seemingly bare fibers at 80,000x, and allowed to heat the surface for
approximately five minutes, or until damage was seen. Images of this for the various
systems are provided in Figures 7-19 and 7-20. In Figure 7-19, it can be seen that the
etched fiber showed no evidence of the presence of any polymer on it's surface. A region
from which a fiber had been removed was etched next, showing a viscous flow of the
polymer due to the beam. This suggests that the failure was between the fiber and sizing

because the BMI resin itself will not flow, it will only thermally degrade.

269
In Figure 7-20, images of the regions before etching have been provided as part of Figure
7-17. It is seen that the PES sizing has adhered to the fiber, and there is some viscous
nature to the matrix surrounding the fiber, suggesting interdiffusion of the polymers. The
RTV sized specimen shows a region believed to be only sizing. No evidence of polymer
on the fiber surface was seen after etching. When the sized region was etched, it
exhibited shrinkage characteristics normally attributed to rubbers, and kept it's shape like

a lightly cross-linked material would.

270

 

400% Commercial Treated IM7 Air Plasma Treated IM7

Figure 7-14 General Images of Control, 400% and Air Plasma Treated Failure
Modes

271

u. «u 2..rr.'..._..‘.... .muu.

".

 

      

RTV Sized IM7

PEI Sized IM7

Figure 7-15 General Images of Sized Composite Failure Modes

272

  
    

   

100%. Unsized IM7

 

11‘

Air Plasma Treated IM7

    

Air Plasma Treated IM7

Figure 7-16 100%, Unsized and Air Plasma Treated Mode I Failure Modes
Showing Poor Adhesion Levels

273

_ _
_ 31‘.

 

400% Treated IM7 400% Treated IM7

Figure 7-17 400% Treated IM7 Mode I Failure Mode Showing Increased Adhesion

274

      

Matrimid 5218 Sized Failure Mode

 

Ultem PEI Sized Failure Mode RTV Sized Failure Mode

Figure 7-18 Images Showing the Adhesion of the Sized Samples After Mode I
Failure

275

 

    

— 4

 

1.11. .._ .

Victrex ® PES Sizing Failure

 

Victrex® PES Sizing Failure

Figure 7-19 Regions Showing the Failure of PEI and 5218 Sizings Showing
Adhesion Between the Sizing and BMI Matrix

276

 

    

 

  

Beam Damage Showing 5218
Sizing Adhered to BMI

 

(Before Beam Damage)

Figure 7-20 Images Showing that the Failure in 5218 Sized Specimens is Between
the Fiber and Sizing

277

  

    

Beam Damaged PES Sized Fiber Beam Damaged RTV Sizing Showing
Showing Good Adhesion Rubber-Like Properties

Figure 7-21 Beam Damaged Regions of the PES and RTV Sized Specimens

7.5 Improvement Conclusions

All of the efforts presented in this chapter were designed to help improve the adhesion
and reduce the microcracking observed in cross-ply laminates. The nature of the
adhesion in the BMI/IM7 system was determined and discussed, and various attempts

were made to improve the system. The following conclusions can be drawn.

0 There is very little chemical bonding in this system as shown by the K38 model
compound study. This is supported by TPD data which shows the evolution of a

significant amount of the surface oxygen at processing temperatures.

278
IM7 fibers subjected to 400% of the standard commercial treatment showed a very
significant increase in surface oxygen content while the air plasma performed here

showed only a marginal increase.

IM7 fibers were successfully sized with PES, PEI, and the thermoplastic polyimide

Matrimid® 5218.

An RTV silicone was sized onto the IM7fibers, but the resulting morphology was
poor. A cross-ply laminate made from the RTV sized fibers showed an ability to
reduce microcracking, but the mechanism for this is believed to be due to both the
presence of the sizing, and the formation of cracks due to resin shrinkage and the

extremely poor adhesion in the sample.

ITS testing showed an increase in adhesion for both the 400% surface treated and the

PES sized fibers. while all other improvement attempts showed no significant effect.

Failure analysis of Mode-I fractured specimens provided evidence supporting the ITS
testing results. The failure locus for the PEI and PES was found to be between the
sizing and matrix while the failure in the 5218 sized specimens showed failure
between the fiber and sizing. lnterdiffusion of the PBS and 5218 was seen and a

ductile failure mode was seen for the sizing material.

Chapter 8

THEORETICAL INTERPHASE DESIGN

8.1 Introduction

The experimental work in the previous four chapters has shown that the interfacial
properties of composites has a dramatic effect on it's performance. It has been shown that
the addition of an interphase in the form of a fiber sizing has great potential for improving
both the adhesion of the system, and relieving thermally induced stresses. The ideal
interphase will vary depending upon the fiber and matrix material being used, and the
service environment for the final component. The interphase should be able to bond with
the fiber and the matrix in addition to having specific physical properties. The choice of
a sizing material which meets all of these requirements is a challenge for even the most
experienced scientist. Thus, a systematic approach for designing these interphases is
needed and has been created here and implemented within an artificial intelligence
framework. The purpose of this work has been to illustrate the relationships involved in
the engineering of an interphase and to introduce a methodology which can be utilized.
The methodology is not meant to provide a prediction of interphase properties or
performance. The purpose is not to review or comment on the artificial intelligence

technique.

279

280

The original artificial intelligence framework was created as a tool for the design of basic
composite materials based on service requirements. This intelligent decision support
system, termed COMADE (Composite Material Designer), uses a combination of
theoretical calculations and heuristic knowledge to select composite systems which meet
the essential performance requirements. These resulting systems are defined by a Matrix
Material, Chemical Agent, Fiber Type and Fiber Length [73]. While COMADE provides
theoretically viable systems, there is no insurance that the chosen matrix
material/chemical agent combination and fiber are sufficiently compatible to form a

composite with acceptable properties.

In an effort to bridge the gap between the theoretical material combinations provided by
COMADE and actual materials which meet the specified requirements, an Interphase
Designer has been created to design the interfacial region (interphase) between the chosen
fiber and matrix combination. With the proper specification of the interphase, a complete
composite material system is defined. This specification can be used to point to actual
commercial systems. or to steer research activity toward promising material

combinations.

8.2 Definitions

There is a large amount of ambiguity in the terminology used to describe composite
interphases. The following definitions are presented to clarify the exact meanings used

here.

281

Binder - a material applied to the fiber directly after the manufacturing process. It can
contain such things as starch, oils, polyvinyl alcohol or acetate, wetting agents, dispersing
agents, cationic softeners, etc. Its purpose is to aid handling in subsequent operations,

and to improve the wetting between the matrix and fiber.

Chemical Coupling Agent - any chemical substance designed to chemically react with
both the fiber and matrix. For the purposes of this report, it is defined as a combination
of three elements: the fiber functional group, the matrix functional group, and the
connecting polymeric chain. Figure 8-1 shows the function of the chemical coupling
agent. Note that the chemical coupling agent is one part of a sizing which will also
contain a binder to promote wetting on the fiber, and various other components. The

chemical coupling agent usually represents about 3-5% by weight of the sizing.

Finish - a finish commonly refers to the coating applied to carbon fibers at the time of
manufacture. It is a film former whose purpose is to bind the fibers together, protect
them during future operations, and promote wetting. In this report, the overall coating
applied to carbon fibers will be called a sizing. The binder portion of the sizing

represents what is commonly referred to as a finish.

Functional group - any chemically reactive species. It may be attached to either an

organic chain (polymer), or a fiber surface.

282
Interphase - the material between the bulk fiber and bulk matrix. It has a different

composition than the matrix, and thus different properties. It usually measures anywhere
from 3 - 1000A in thickness. By tailoring the properties of the interphase, the properties

of the overall composite can be changed dramatically.

Sizing - a material applied to fibers during the final stages of manufacture. It contains a
variety of components which are often proprietary. The most important of these
components are: lubricants, antistatic agents, binders, and chemical coupling agents. The
binder and chemical coupling agent are of greatest importance for this project. A sizing
usually refers to the coating applied to glass fibers. For this report, a sizing will be
defined as the combination of binder, chemical coupling agent, lubricant, and antistatic
agents used on any fiber. Sizings are usually applied to fibers to about 1% of the fiber

weight.

8.3 Theoretical Approach - Interphase Design

The interphase in a composite material is defined as the material which exists between the
fiber surface and bulk matrix. The role of this interphase in composite materials has been
well documented by various authors [6, 8-11, 43, 49, 74-77] It may be a sizing which has
been applied to the fiber, or just an extension of the bulk matrix material which has varied
properties due to it's proximity to the fiber. The Interphase Designer builds a definition

of a specific interphase material by providing a set of characteristics which should cause

283
it to adhere well to both the fiber and matrix suggested by COMADE. The interphase

designed here is based upon the model shown in Figure 8-1.

 

“R

 

 

 

/

FF G = Fiber Functional Group
MFG = Matrix Functional Group
R = A Polymeric Chain

Figure 8-1 Function of the Chemical Coupling Agent

The interphase consists of three major components which are the binder, the coupling
agent, and a surface treatment for the fiber. The role of the binder is to transport the
functional groups of the coupling agent to both the fiber surface and the matrix material.
It also provides any mechanical properties required of the interphase. The coupling agent
is responsible for chemically adhering to the fiber surface and matrix material. Surface

treatment is required in many cases to alter the concentration and types of fimctional

284

groups on the fiber surface. The importance of utilizing appropriate surface treatments

was shown in Chapter 7.

Since the most challenging part of defining an interphase is determining what
combination of binder, coupling agent, and surface treatment will adhere to both the fiber
and matrix, the adhesion between polymers and fiber surfaces is reviewed as well as the

adhesion between two polymers.

8.3.1 Polymer/Fiber Adhesion

The adhesion of a polymer to a fiber is governed by both chemical and physical
interactions. The chemical interaction is brought about by either the chemical bonding of
the polymer to functional groups present on the fiber surface, or secondary interactions
between the two species. Both covalent and ionic chemical bonds are extremely strong
(39-111 kcal/mol). Secondary interactions are much weaker (2-6 kcal/mol), but can play
a significant role[78]. These interactions include non-polar dispersive forces (van der
Waals forces) and polar dipole interactions. In the Interphase Designer created here,
functional group reactions are predicted by using references to cases where the reactions
have been seen to occur. A degree of confidence is assigned to each reaction based upon
how closely it fits the kinetic and thermodynamic environment of it's intended use. More

detail of this approach is provided in Section 8.3.1.

285
The physical interaction is the anchoring of the polymer into the topographical features of

the fiber surface. When the fiber surface is rough, the polymer can surround asperities on
it's surface. When the composite is stressed, load is transferred to the fiber by the
mechanical interlocking of the asperities and the polymer. In order for either the
chemical, or the mechanical interactions to take place, the fiber and sizing must be in

intimate contact.

The degree of intimacy between two surfaces can be gauged using the degree of wetting,
which is a measure of the thermodynamic compatibility between the two surfaces and is a
described using the surface free energies (y) of the two materials. The surface free energy
is a measurable quantity which can be broken into various components. A few different
sets of components have been suggested. The most common components used are the
polar (yp) and the dispersive (yd); where ym, = 7,, + yd. Other researchers have added a third
component to the system, the hydrogen bonding component yH. Yet other researchers
suggest that the surface energy should be measured in terms of it's Lewis acid-base
character. While all methods have their benefits this knowledge based system requires
data to be available on a large variety of materials, and since the majority of the available

data is for the polar and dispersive components. they have been utilized here.

Once the two components of the surface energy are known, a specific set of logic must be
applied to try to ensure the wetting of the two components. For the best wetting in a

fiber/matrix system, the total surface free energy of the fiber should be greater than that

286

of the matrix (ym, “be, > ym,‘ ,,,,,,,~,), and the ratio yp/yd for the two materials should be as

nearly equivalent as possible [79].

8.3.2 Polymer/Polymer Adhesion

The adhesion of two polymers also involves both chemical and mechanical interactions.
The chemical interactions are similar to those described above and are dominant in the
bonding of thermosetting polymers. physical interactions arise from the entanglement of
the polymeric chains and are the dominant mechanism in the adhesion of thermoplastics.
To properly define the adhesion between two polymers, one must define both the reactive

functional groups, and the capability of the polymers' chains to entangle.

The intermingling of two polymers can be described by their miscibility. Two polymers
which are highly miscible will have well entangled chains, and thus have a high degree of
mechanical integrity. The miscibility of polymers can be estimated using the concept of a
miscibility parameter. The concept of such a parameter, and the solubility parameters

used to calculate it has been reviewed by David and Sincock [80].

Solubility parameters of a polymers can be estimated using various methods [80]. These
method can be used to calculate a solubility parameter for a polymer by using a group
contribution theory which assumes additivity of the molar attractive forces of the various
groups which make up the polymer structure. Because this contribution theory does not

allow for the steric effects which occur when the groups are bonded to one another, it is

287

important to calculate the polymer solubility parameter based upon the largest

contributing groups for which data exists.

The general relationship for determining the solubility parameter of a compound is given
in Equation 8-1 where F i is the molar attraction constant of group i, and V is the molar

volume.

F.
5 = Z? Eqn. 8-1

The solubility parameter may have contributions from various components, in a manner
similar to the various surface free energy components. These components are combined

according to Equation 8-2, giving a solubility parameter for the polymer as a whole.

5122 = 5.12 + 5,12 + 51.2 Eqn. 8-2

where d, p, and h represent the dispersive, polar, and hydrogen bonding components of

the solubility parameter.

Once the solubility parameters for two materials are known, the enthalpy of mixing per

unit volume is calculated according to Equation 8-3.

AH... = rim — )2 Eqn. 8-3

288

where (I), and (1)2 are the volume fractions of components 1 and 2, and 5, and 62 are their
solubility parameters. The Gibbs free energy of mixing is given by Equation 8-4, and

when negative is indicative of a spontaneous event.

AG!” : AH”) — TASIH

Eqn. 8-4

where T is temperature and AS”, is the entropy of mixing. Since the mixing of two
components increases the entropy of the system as a whole, ASm is positive and there is
an entropic driving force for mixing. Thus, if AHm is sufficiently small, mixing will be

thermodynamically favored, and given no kinetic hindrances, will occur.

The concept of a miscibility parameter based upon the enthalpy of mixing is therefore

utilized, as shown in Equation 8—5.

MP = ((2 — (32)2 Eqn. 8-5

For mixing to occur, MP must be as small as possible [80].

8.4 Implementation

The Interphase Designer uses a hierarchical knowledge structure based upon matching

tables. The logic used to translate the knowledge from the individual materials into the

289

definition of an appropriate sizing is reviewed here. A schematic of the variable

dependencies used in COMADE and the Interphase Designer is provided in Figure 8-2.

 

 

Figure 8-2 COMADE and Interphase Designer Variable Dependencies

The matching tables used for the addition of the Interphase Designer were created using
data found in reference books and published literature for the surface energetics,
miscibility parameters, and functional group reactions discussed in the previous section.
Since the data available in the literature is often lacking, or only partially applicable to the
system of interest, a method was introduced which takes into count the confidence of the

data desired by the user.

290

8.4.1 Functional Group Reactions

The reaction between the functional groups of the coupling agent, and the functional
groups present on the fiber, or in the resin are accounted for using a set of confidence
levels based upon the similarity between known reactions and the desired reaction in
terms of chemical and physical environment. Four confidence levels were chosen:
Commercial, High, Medium, and Low. Commercial confidence will provide the user
with results based upon reactions which are known to exist commercially, with available
references cited. The High confidence level will provide the user with systems which are
based upon reactions between the two species in the form of interest (solid surface or
polymer) and in environments very similar to those used in the processing of the
composite. Thus, extrapolation to the composite system of interest should be safe. The
Medium confidence level allows for data in which the reactions have been shown to be
successful in the form of interest, but the environment was not the same as the
environment of interest. For example a reaction could be known to occur at a
temperature of 200°C, but the composite fabrication is formed at only 120°C, so the

viability of the reaction is uncertain.

The Low confidence level allows data for which the reactions have been shown to occur
in solution chemistry, or under very different conditions. In this case the composite
processing conditions would be quite different, and there is a good chance that the

reaction would not occur.

291

8.4.2 Miscibility Data

The miscibility data is used to design an interphase by determining the solubility
parameter for the matrix material and requiring the solubility parameter of the binder to
be as close to this as possible. This results in the smallest miscibility parameter, and thus

the best intermingling of polymer chains.

8.4.3 Fiber Surface Treatments

The fiber surface treatments are implemented within the framework by the
thermodynamic description of their surface energies, and the functional groups which are
present upon the surface after treatment. The surface energies ultimately provide a range
for the total surface energy of the binder. The fiber functional groups are used to suggest

coupling agent functional groups which will react with the species on the fiber.

8.4.4 Data Contained Within the Tables.

The matching tables used within the program of COMADE were generated using data for
gathered for reactions between the utilized functional groups. Miscibility parameters
were gathered for the polymers, and information concerning the functional groups present
after the various surface treatments was found. The functional groups used within this

study are provided in Table 8-1 which also presents the references upon which these were

based.

292
Table 8-2 shows the solubility parameters for the various resins as calculated using group

contribution theory. Finally, the surface free energies for the various functional groups is

provided in Table 8—3.

Table 8-1 Functional Group Reactions Used in COMADE

 

 

 

 

 

 

 

 

Amine Vinyl Carboxylic Acid Mercaptan Chloroalkyl
Amine N
Vinyl Y [86,p689] Y
Carboxylic Acid Y [83, p92] N N
Mercaptan N N [86, p687] ? Y
Chloroalkyl Y [83, p104] N N [86, p353] Y [86, p362] N
Alcohol N 7[87, p11] A N Y [81, p6] N [86, p361] Y [82, p258;281]
Nitrile N [86, p803] B N Y [86, p870] Y Y [86, p872]
lsocyanate Y N Y [84, p125] ? ?
Epoxy Y [83, p134] N Y [88, p47] Y [89, 86 p360] Y [89, 88 p37]
BMI Y [83, p161] Y [84, p255]
Keytone Y [87, p11]

Alcohol Nitrile lsocyanate Epoxy BMI
Amine
Vinyl
Carboxylic Acid
Mercaptan
Chloroalkyl
Alcohol N
Nitrile Y N
lsocyanate Y [83, p136] N N
Epoxy Y [85, p438] ? Y [89, 88 p48] Y 9
BMI Y
Keytone Y [83, p3]

 

 

 

 

293

Table 8-2 Resin Solubility Parameters Used in COMADE

RESINS

Semi-Crysta_l1ne Thermoplastics

PBT 10.83
PA66 10.28
PET 11.72
PPO 8.59
PPS 10.60
PEEK 10.39

Amorphous Thermoplastics
ABS

PC 9.39

PSU 9.53

PAS

PEI 11.45
PAI

Blend Thermoglastics
ABS/PBT

PC/PBT 1 0.1 1
PC/ABS

Thermoset Epoxies

DGEBA 8.32
EpoxPhenNov

TGMDA 9.06
TGETPE 8.42

Phenolics
Novolac
Resole

Polyesters
Orthophthalic

Isophthalic
Vinyl Ester
BPA Fumarate
Chlorendic

Bismaleimides

44MDA BMI 13.19
Polyimides

PMR15 10.70
Thermid 600 10.26

All values expressed in cal'acm

CHEMICAL AGENTS

For Therrnoset Epoxles
DEAPA

Dicyandiamide

MPDA

Chlorendic Anhydride
DDS/BF3-MEA

For Phenolics
HMTA

For Polyesters
DAP

Vinyl Toluene
Styrene

For Bismaleimides
Dicyanate BPA
OODiaIIyIBPA

MABA

1,4Tolyl BMI

For Polyimides
MeOH

EtOH

10.06

10.44

10.74

17.64

9.88
8.56
8.47

21.03

9.70
12.86
14.18

16.17
13.68

Table 8-3 Functional Group Surface Free Energies

294

 

 

 

8.5 Future work

Temperature (°K) 298 Temperature (°K) 398
dH (18 (10 dH dS dG

Alcohol -4 1.56 1.28 -45.3744 Alcohol -4 1.56 1.28 -46.6544
Amines Amines

Primary 2.82 2.71 -5.2558 Primary 2.82 2.71 -7.9658

Secondary 12.93 3.16 3.5132 Secondary 12.93 3.16 0.3532

Tertiary 19.46 3.82 8.0764 Tertiary 19.46 3.82 4.2564
Carboxylic Acid -98.39 2.86 -106.9128 Carboxylic Acid -98.39 2.86 -109.7728
Chloroalkyl -8.25 0 -8.25 Chloroalkyl -8.25 0 -8.25
Epoxy 12.86 -0.63 14.7374 Epoxy 12.86 -0.63 15.3674
Mercaptan - l 0.68 l .07 - 1 3.8686 Mercaptan - 10.68 1.07 - 1 4.9386
Nitrile 30.75 -0.72 32.8956 Nitrile 30.75 -0.72 33.6156
Vinyl (H2C=CH-) 13.737 1.655 8.8051 Vinyl (HZC=CH-) 13.737 1.655 7.1501
Water -58.076 1.154 -61.51492 Water -58.076 1.154 -62.66892
Ester -92.62 2.1 -98.878 Ester -92.62 2.1 -100.978
Ammonia -1 1.606 2.556 - 1 9.22288 Ammonia -1 1.606 2.556 -21.77888
Methane - 1 8.948 2.225 -25.5785 Methane -18.948 2.225 -27.8035
Methyl -10.943 2.215 -17.5437 Methyl -10.943 2.215 -l9.7587
Ether -15.79 -0.85 -l3.257 Ether -15.79 -0.85 -12.407
Aldehyde -29.28 0.77 -31.5746 Aldehyde -29.28 0.77 -32.3446
Ketone -28.08 0.91 -30.7918 Ketone -28.08 0.91 -31.7018
Carboxylic Acid -87.66 2.473 -95.02954 Carboxylic Acid -87.66 2.473 —97.50254

All values in dyn/cm

 

The presentation of specific materials which meet the users requirements has been
alluded to a number of times in this discussion. The implementation of this requires the
addition of essentially two more logical modules to the system as it exists. The function
of the modules would be to match the description of the sizing to actual Sizings, and
having done this, to match the fiber type, fiber length, matrix, chemical agent, and sizing
to commercially available composite materials. The two modules would be

prograrnmatically quite different than the Interphase Designer in that they would

essentially be an interface to a large database using a SQL (Structured Query Language)

295

request for records matching the criteria. It is envisioned that ultimately, a very large,
encompassing database will exist, and that this type of information will be readily

available.

These two modules have been termed the "Interphase Critique" and "Composite

Critique." Logic flows for them are provided in Figures 8-3 and 8-4.

 

 

[ Interphase Critique

 

 

 

Trial Sizing Properties

Fiber Functional Group Amine I Record # 666

 

 

 

 

 

 

 

 

Matrix Functional Group [Epoxy Sizing A User Criteria J
Surface Free Energy .67 ,
Solubili Ratio .89 Manufacturer: ChemEnvrronment: No
_ty__ 1D # FlameRetardance: No
HumidEnvironmcnt: No
Use Temp: _ to _ Use Temperature: 270

 

 

 

 

Promrties:

 

 

 

 

 

 

 

 

 

 

 

 

 

— Stable in:
. . . — Flame
SFE, Solubility Ratio's 1 Humidity
Matrix Functional Group I Hvdrocaibons

g Alcohol Epoxy Amine ' '
E _
D g Unstable in:
g Lu Alcohol ll
.9 A _ w
:5: -—
LL. LL!
5 -
L: <

 

 

 

 

 

 

 

Next Combination

Figure 8-3 Logic Flow for the Interphase Critique

296

 

 

L Composite Critique

 

 

 

 

 

 

 

 

 

 

     
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  
 

Trial Combination
FiberTvgc': A84
Matrix/Chem em: BMI —» Record # 316 User Criteria |
S‘ .‘ : S A
w ”mg As4; BMI. Sizing A ChemEnvironment: No
Author: FlarneRctardane: No
Title' lliimidlinvrronment: No
(‘it'ition' ReqFlexuralModulus: 75
ll ‘ ‘ RechnsileStrength: 1000
r Sizing C Testing Performed: iSItreslsEnvrronment: 5:85!
1 Sizing B Tensile. DSC "“3 “mpmture'
Sizing A r —
Matrix/('hemAgent 45”!“—
. . . UTS:
PB T Epoxy BMI
I _ Modulus:
E Use Temp: _ to _
L —
e a J
4.. < I—
§ I
6 i
Li

 

 

 

 

 

 

 

No

 

Next Combination

Figure 8-4 Logic Flow for the Composite Critique

8.6 An Example and Conclusions

With the understanding of the interphase's adhesion mechanisms with both the bulk
matrix and fiber provide here, it follows that an appropriate interphase is sufficiently
defined by surface energetics (ym. yp/yd), a solubility parameter. and a chemical coupling
agent containing a fiber functional group to react with the fiber and a matrix functional

group to react with the matrix.

297

A diagram showing the information flow for a simple example is shown in Figure 8-5. In
this figure, the Interphase Critique, and Composite Critique have been lumped into a
black box called the Current Material Database. The user needs a material which has a
minimum tensile strength of 1000 MPa, a minimum flexural modulus of 200 GPa, at a
maximum temperature of 80°C in a constant stress environment. Amongst other
combinations, COMADE suggests an epoxy matrix of DGEBA (diglycidylether of
bisphenol A) with a DEAPA (diethylaminopropylamine) chemical agent and continuous

AS4 carbon fiber.

One of a multitude of the interphase options suggested by the Interphase Designer is
specified as having no fiber surface treatment, ym, >56 dyn/cm, yp/yql 5 1.154, 6 E 8.32
cal'/2cm3/2, and a coupling agent consisting of carboxylic acid and amine functional
groups. This information will then be used to probe a material database for commercially
available systems shown to actually meet the original requirements. The interested

researcher will be able to find the material combinations which have not been researched.

298

 

Fiber Treatment = None
YIN 56 dyn/cm

Yr lit 2 1 154

5 -:- 8 32 call’zcmm
FFG: Carboxylic Acid
MFG =‘ Amine

 

  
     
     

   

  

(.‘urrent
Material
Database

 

Chemical Environment = None
Flame Retardance ' No
Humid Emironment '— No
Flexural Modulus = 200 (il’a

» Tensile Strength = 1000 MPa
Stress Environment = (‘onstant

\

Use Temperature 80 “(

 

 

 

  

Commercial Materials
or Literature Citations

Figure 8-5 Information Flow for COMADE and the Interphase Designer

The COMADE and Interphase Designer system is currently operational for most of the
common composite fibers and matrices. Data for some of the less common materials still
needs to be researched. A future expansion will utilize the current results to provide
industrial users with commercially available materials which meet their requirements.
Academic or industrial researchers will be provided with literature citations related to the

specific composite systems suggested, and will be able to quickly determine if there is a

need for further research.

Ii
5‘:
a"
:.:. a ’ _.
a

.‘¥ ‘3!
Interphase 3
Designer

4 .-
r .
'alt ‘.‘°¢.'|¥

 

 

   
     
  

Matrix Material = DGEBA
(‘hemicnl Agent = DEAPA
Fiber Type ~. AS4 (‘arbon

Filter Length Continuous

 

 
 

Chapter 9

CONCLUSIONS AND FUTURE RECOMMENDATIONS

This research has provided a great deal of insight into the hygrothermal properties of the
Matrimid® 5292 BMI system. This chapter will attempt to present the conclusions to be
drawn from the research as a whole. Conclusions concerning specific details of the

studies were presented at the end of each chapter, and are not necessarily repeated here.

9.1 Microcracking Conclusions

At the macroscopic level, it was seen that advancing cure in the BMI system causes a
general reduction in properties and partially cured resin will continue to cure at
anticipated use temperatures. The increase in cure was also shown to cause an increase in

the moisture diffusion rate and ultimate moisture content.

A mechanism for the cure induced microcracking was presented. It is shown that
microcracking is due to a combination of two mechanisms. One results from a ply level
thermal expansion mismatch between the 0° and 90° ply orientations in a cross-ply
laminate. Other stresses are caused by a thermal expansion mismatch between the

transverse expansion of the fiber, and the expansion of the matrix. Two models were

299

300
utilized to show that a reduction in the residual stresses induced by the cure cycle can be
obtained by a variety of methods. These methods include a reduction in the stress free
temperature; use of high volume fraction composites; and the addition of a compliant
interphase. The addition of an interphase has the added potential of increasing the

adhesion of the system which was shown to be inherently very poor.

9.2 Physical Property Conclusions

Both increased cure and the presence of moisture were shown to be detrimental to the
performance of BMI resin. In addition, it was seen that the moisture had a more
detrimental effect on the performance of partially cured resin specimens. The decreases
in performance of the resins due to the presence of moisture were also found to be fully

recoverable. Composites, on the other hand, were insensitive to the presence of moisture.

Thermal spiking of both wet and dry BMI specimens between room temperature and
250°C was shown to have very little effect on properties, at least for short term tests
involving a relatively small number of cycles. Dry K3B resin and composites showed the
same lack of sensitivity to thermal spiking, but wet K3B specimens cycled to 250°C were
shown to fail a catastrophically. This is attributed to stresses caused by the escaping
moisture. Since this temperature is above Tg for K3B, the stresses were easily able to
deform the material causing excessive damage. It is felt that this should not occur below

the wet Tg of the resin.

301

9.3 Interfacial/Adhesion Conclusions

The interfacial adhesion in the BMI/IM7 system was seen to be very poor both in failure
analysis of flexure and Mode I fractured specimens and through direct testing of the
interfacial properties via ITS. The presence of moisture showed a detrimental effect on
adhesion both through a reduction in the measured interfacial shear strength of partially
cured resin specimens, and by causing interfacial failure in unstressed samples for fully
cured and saturated specimens. The adhesion that does exist in this system is believed to
be due to mechanical interactions between the fiber surface and matrix, not because of

chemical bonding.

Attempts at improving the adhesion of the BMI/IM7 system were made by both fiber
surface treatment and the addition of Sizings. Fibers treated with 400% of the
commercial treatment and those sized with PBS were shown to cause an improvement in
adhesion. The improvement due to the commercial treatment is attributed to an increase

in fiber surface roughness which aids in the mechanical bonding of the system.

9.4 Recommendations for Future Studies

This research is part of an ongoing study of these materials, so the efforts begun here will

be very useful for steering this continuing effort.

302
First, the microcracking in this system was seen to be a strong function of the stress free
temperature, and stepped post-cures were seen to be at least moderately successful at
reducing it. It is felt that significant reductions in the SFT may be achieved by using
stepped cure cycles. By creating a gelled network at a lower temperature, the stresses
caused during the subsequent, necessary, post-cures may be reduced. The vertical heating

stage which was manufactured will allow the efficient investigation of this.

The thermal spiking here showed very little effect, but the number of cycles was very
small so longer term testing is required. The open-hole compression tests showed some
response to wet cycling for the BMI system suggesting cumulative damage at the edge of
these composites. More testing should be performed in which the holes are cut into the
composite both before and after conditioning. This will reveal whether the loss in

properties is an edge effect.

Since the K3B specimens failed during the tests above Tg, testing should be performed
below Tg to determine if this catastrophic failure is possible for wet specimens below the

resin's wet Tg (in the region of component service).

The dynamic mechanical tests on wet BMI/1M7 specimens showed an increase in the
viscoelastic nature of the system while the resin alone showed a decrease in its
viscoelastic nature. This suggests either a mechanical constraint caused by the fibers and

expansion of the resin. or a change in the adhesion of the system. This should be

303
investigated because this may be a useful method for characterizing composite interfacial

bonding.

The 400% surface treated and PES sized fibers showed an increase in the adhesion of the
system and should be investigated further. A more controlled sizing method should be
used and should not be difficult to achieve. A sizing using a lower molecular weight of
PES should be investigated. This may allow increased interdiffusion with the BMI resin
which should further increase the adhesion. A combination of the 400% fiber treatment

and the PES sizing should also be attempted.

Finally, the knowledge base for the expanded COMADE system should be filled further,
and the concepts of the Interphase Critique and Composite Critique should be
implemented. The methodology presented within this study should be investigated
further, and expansions should be made to include such appropriate things as fiber
morphology, processing temperatures/environments, and many of the other things known

to affect the adhesion at the interphase.

 

APPENDICES

APPENDIX A

Appendix A

TABULAR BMI RESIN TEST DATA

Table A-1 Three Point Flexural Data for BMI Resin

 

 

Flexural Tg From

Flexural F lexural Failure Peak of

Cure Cycle Strength Std. Dev. Modulus Std. Dev. Strain (%) Std. Dev. Tan Delta
m 12 4368 159 3.8 0.558 180
Low Long 159 13 4009 11 4.34 0.424 197
Std. Short 143 7 3665 193 4.09 0.342 262
Std. Long 113 7 3573 28 3.15 0.181 313
High Short 138 21 3965 23 4.13 0.902 343
High Long 107 34 3846 97 2.98 1.188 351

 

 

 

 

Table A-2 Tensile Data for BMI Resin

 

 

Tensile

Tensile Tensile Failure
Cure Cycle Strength Std. Dev. Modulus Std. Dev. Strain (%) Std. Dev.
Low Short 69- 15.8 2293 201 3.43 0.732
Low Long 77 5 2274 120 3.96 0.269
Std. Short 76 4.7 2275 185 3.99 0.587
Std. Long 94 6.1 1828 23.6 6.66 0.5
High Short 36 3.8 2024 46.6 1.893 0.258
High Long 58 2.6 1683 12.7 4.23 0.085

 

 

 

Where:

Low Short = 177°C/90min + 200°C/ 1hr

Low Long = 177°C/90min + 200°C/3hr

Std. Short = 177°C/90min + 200°C/2hr + 250°C/ 1hr

Std. Long = 177°C/90min + 200°C/2hr + 250°C/6hr

High Short = 177°C/90min + 200°C/2hr + 250°C/6hr + 300°C/1hr
High Long = 177°C/90min + 200°C/2hr + 250°C/6hr + 300°C/6hr

304

APPENDIX B

 

Appendix B

MOISTURE ABSORPTION STUDY DATA

Table B-1 Summary Data from 100°C Moisture Absorption Experiment

 

 

 

 

 

ummary 0 me sor n xper n
Maximum
Condition Moisture °/. Std. Dev. Diffusivity Std. Dev. C Std. Dev. Raquarod Std. Dev.
1000 BMI Resin Low Cure 3.783% 0.006% 9.014E-08 5 08E-09 0.677 0.002 0.9990 0.0001
1000 BMI Resin High Cure 5.003% 0.021% 1.166E-07 1.35E-08 0.712 0.004 0.9990 0.0001
1000 BMI X-ply 1.573% 0.009% 1.222E-06 2.39E-07 0.331 0.021 0.9961 0.0004
1000 BMI Undirectional 2.324% 0.133% 4.258E-08 1.80E—08 0.580 0.103 0.9984 0.0003
1000 K38 Resin 1.615% 0.026% 2.322E-06 2.39E-07 0.701 0.012 0.9985 0.0003
100C K3B-1M7 X-ply 0.538% 0.011% 94276-08 8.95E-09 0.667 0.025 0.9975 0.0009
100% Resin Equivalent for Composite Data
1000 K38-1M7 X-ply 1.063% 0 025% 9.430E-08 8935-09 0.667 0.025 0.9975 0.0009
1000 BMI X-ply 4.266% 0.024% 1.220E-06 2.42E-07 0.330 0.021 0.9961 0.0004
100C BMI Undirectional 4.964% 0.068% 42586-08 1.80E-08 0.580 0.103 0.9984 0.0003

 

 

Table B-2 Summary Data from 70°C Moisture Absorption Experiment

 

 

 

 

 

u ry o u n r
Maximum

Condition Moisture % Std. Dev. Diffusivity Std. Dev. C Std. Dev. quuared Std. Dev.
70C BMI Resin Low Cure 3.306% 0.031% 2.369E-08 1.725E-09 0.653 0.004 0.9995 0.0000
70C BMI Resin High Cure 5.138% 0.027% 5.196E-08 2.743E-09 0.721 0.000 0.9989 0.0000
70C BMI X-ply 1.552% 0.019% 1.879E-06 1.200E-07 0.612 0.006 0.9933 0.0009
700 BMI Unidirectional 2.488% 0.062% 3.206E-08 5.275E-10 0.656 0.056 0.9984 0.0008
70C K38-1M7 X-ply 0.548% 0.003% 3.383E-08 1.731E-09 0.672 0.003 0.9988 0.0002
70C K3B Resin 1.534% 0.011% 9.998E-07 3.688E-07 0.707 0.006 0.9986 0.0003

100% Resin Equivalent for Composite Data

70C K38-IM7 X-ply VfCor 1.084% 0.007% 3.383E-08 1.732E-09 0.672 0.003 0.9988 0.0002
700 BMI X-ply VfCor 4.210% 0.007% 1.879E-06 1.178E-07 0.612 0.006 0.9933 0.0009
700 BMI Unidirectional 5.099% 0.030% 3.206E-08 5.275E-10 0.656 0.056 0.9984 0.0008

 

 

305

A -'.r ‘K'

 

APPENDIX C

Appendix C

DERIVATION FOR THE ERROR IN PERCENT CHANGE VALUES

The calculation of appropriate measures of the error associated with the percent
difference in properties associated with the results of the Hygrothermal and Thermal
Spike designed experiments is not simply the average of the two condition’s standard
deviations. A proper calculation of the error in this measure is provided here. In all

cases, x represents a property such as the modulus, strength, or failure strain.

For the Hygrothermal study, the designations for the different conditions are shown in

Table C-l , where all x,‘ are assumed to be independent.

Table C-l Factor Level Designations for the Hygrothermal Experiments

 

 

 

Designation Moisture Volume Fraction Extent of Cure
xl Dry Low Vf Partial Cure
x2 Dry Low Vf Full Cure
x3 Dry High Vf Partial Cure
x4 Dry High Vf Full Cure
x5 Wet Low Vf Partial Cure
x6 Wet Low Vf Full Cure
x7 Wet High Vf Partial Cure
x8 Wet High Vf Full Cure

 

 

For the Hygrothermal experiment, the percent differences are calculated using Equation

C-l.

 

8 Eqn C-1

306

 

307

To approximate the appropriate error associated with this, a Taylor series expansion
(Equation C-2) is used. The series is truncated at the linear term for mathematical

simplicity, but higher order terms should be negligible.

f(xis---xs): - f'(x109' 3‘30) + 209—x Z('”x10"”‘x80)+ Eqn C-Z

&I

The derivative terms from Equation O] are:

 

( 8 )3— Eqn C-3

 

= 8—-——=——8" 8 2 Ean-4

 

flit—3.). = g x, —x,2 for 3Sj$8
(3C, ( 3 ] Eqn C-5
. 2x.-
i=1

Using the designations in Table C-l, the percent change in property by moving from

Partial to Full Cure for the Dry, Low Vf specimens is represented by

firm-‘w'm

308

E. — E3
8 s Eqn. C-6

1 -
8E)“

 

°/o(.'hange =

and the variance of this is given by

 

 

_ ._ 2 2 2
xi —X2 64 S2 8 — — — s72 8 _ _ _ _ __ 2 8 s.
—————————— _———— 1 - J
War 1 8 = 8 4 — ZI,+X2—X1 +"— in+xr—x2 +(xl—X2) _
{a (2;) "i "2

- 81:1 ~ i=1"
Eqn. C-7

2

I

where Si and n,’ are the sample dev1atton and srze for condition 1, respectively. — is the
n

variance. The errors shown for the percentage changes reported in this study all represent

the square root of this variance which is approximation of it's standard deviation.

This same analysis can be easily adapted for the Thermal Spike experiments. For this
case, instead of a three factor designed experiment, a two factor designed experiment
was used. Equation C-8 results from a derivation very similar to that shown above.
Again, the square root of this variance is used as an approximation of the standard

deviation of the percent change.

309

F' q

— _-3 ,2 4 __ _ _ 2
Var LELTi—J =—4—1—6—I[%l—[zx,+X3‘-th +
._ 1 1:1

474. (W

(2; .4}. 5r.) +(;, 5.724351]

 

2
”2 j=3 "j

 

 

Eqn. C-8

The designations for the different factors in the Thermal Spike experiments are shown in

Table C-2.

Table 02 Factor Level Designations for the Thermal Spike Experiments

 

 

 

Designation Moisture Cycling
x, Dry Control
x2 Dry Cycled
x3 Wet Control
x4 Wet Cycled

 

 

LIST OF REFERENCES

LIST OF REFERENCES

Roger J. Morgan, James E. O'Neal, and Dale L. Fanter, "The Effect Of Moisture
On The Physical And Mechanical Integrity Of Epoxies," Journal of Materials
Science, 15 (1980), pp. 751—764.

Roger J. Morgan, Robert J. Jurek; Anna Yen; and Tom Donnellan, "Toughening
Procedures, Processing And Performance Of Bismaleimide-Carbon Fibre
Composites," Polymer, 34 (4) (1993), pp. 835-842.

H.H. Man, R.J. Morgan; and P.J. Hogg, "The Effects of Impact and Moisture
Content on the F lexural Properties of Bismaleimide-Carbon Fiber Composites,"

Advanced Composite Materials: New Developments and Applications Conference
Proceedings, Detroit, MI, Sept. 30-Oct. 3, 1991.

R. H. Pater, K. Whitley; C. Morgan and A. Chang, "Crosslinking-Property
Relationships in PMR Polyimide Composites. Part I," Polymer Composites, 12
(2) (1991), pp. 126-132.

Dick Cornelia, "Hygrothermal Performance Of Polyimides," 39th International
SAMPE Symposium, April 11-14, 1994, pp. 917-929.

Lawrence T. Drzal, "Adhesion of Carbon Fibers to Epoxy Matrices," Treatise on
Adhesion and Adhesives Volume 6, ed. Robert L. Patrick, Marcel Dekker, Inc,
1989, pp. 187-211.

Nirmala P. Vaidyanathan, Vinayak N. Kabadi, Ranji Vaidyanathan, and Robert L.
Sadler, "Surface Treatment of Carbon Fibers Using Low Temperature Plasma,"
Journal of Adhesion, 48 (1995), pp. 1-24.

Madhu S. Madhukar And Lawrence T. Drzal, "Fiber-Matrix Adhesion and Its
Effect on Composite Mechanical Properties: 1. Inplane and Interlaminar Shear

Behavior of Graphite/Epoxy Composites," Journal of Composite Materials, 25
(1991), pp. 932-957.

310

10.

11.

12.

13.

14.

15.

l6.

17.

18.

19.

20.

311

Madhu S. Madhukar And Lawrence T. Drzal, " Fiber-Matrix Adhesion and Its
Effect on Composite Mechanical Properties: 11. Longitudinal (O) and Transverse

(90) Tensile and Flexure Behavior of Graphite/Epoxy Composites," Journal of
Composite Materials, 25 (1991), pp. 958-991.

Madhu S. Madhukar And Lawrence T. Drzal, " F iber-Matrix Adhesion and Its
Effect on Composite Mechanical Properties. III. Longitudinal (0) Compressive

Properties of Graphite/Epoxy Composites," Journal of Composite Materials, 26
(1992), pp. 310-333.

Madhu S. Madhukar And Lawrence T. Drzal, "Fiber-Matrix Adhesion and Its
Effect on Composite Mechanical Properties: IV. Mode I and Mode II Fracture
Toughness of Graphite/Epoxy Composites," Journal of Composite Materials, 26

(7) (1992), pp. 936-968.

Dupont Publication, "Properties and Characteristics of Avimid® K Composite
Materials."

Lin Li, "Thermal Aging Properties Of Graphite Fiber Reinforced Peek And
Graphite Fiber Reinforced BMI Composites," Polym.-Plast. Techno]. Eng, 29
(5&6) (1990), pp. 549-562.

R.H. Pater, "Thermosetting Polyimides: A Review," SAMPE Journal, 30 (5)
(1994), pp. 29-38.

Roger J. Morgan, E. Eugene Shin, Boris Rosenberg, and Ann Jurek,
“Characterization of the Cure Reactions of Bismaleimide Composite Matrices,”
Polymer, 38 (3) (1997), pp. 639-645.

M. Chaudhari, T. Galvin, and J. King, “Characterization of Bismaleimide System,
XU 292,” SAMPE Journal, July/August 1985.

Roger J. Morgan and E. T. Mones, J. Appl. Polym. Sci, 33 (1987), p. 999.

Mark S. Wilenski, MSU internal publication, "Test Data in Support of AFOSR
Grant No. F49620-95-1-0129", 1997.

Chi-Hung Shen and George S. Springer, "Moisture Absorption and Desorption of
Composite Materials," in Environmental Effects on Composites, volume 2, ed. G.
S. Springer, Technomic Publishing Co., 15 (1981).

G. E. P Box, W. P. Hunter, and J. S. Hunter, Statistics for Experimenters, John
Wiley & Sons, New York, 1978.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

312

Donald F. Adams, "A Micromechanics Analysis of the Influence of the Interface
on the Performance of Polymer-Matrix Composites," Journal of Reinforced
Plastics and Composites, 6 (1987), pp. 66-88.

P. M. Jacobs, M. Simpson and F R. Jones, "Generation Of Thermal Strains In
Carbon Fibre-Reinforced Bismaleimide (PMR-15) Composites: Part 1: The
Determination of Residual Thermal Strains in Cross-Ply Laminates," Composites,
22 (2) (1991).

F. R. Jones, M. Mulheron, J. E. Bailey, "Generation Of Thermal Strains In GRP
Part 2 The Origin Of Thermal Strains In Polyester Cross-Ply Laminates," Journal
ofMaterials Science. 18 (1983). pp. 1533-1539.

J. M. Whitney, I. M Daniel, and R. B. Pipes, Experimental Mflmics of Fiber
Reinforced Composite Materials, Revised Edition, Prentice-Hall, Inc., Englewood
Cliffs, New Jersey, 1984.

B. Cunningham, J. P. Sargent, K. H. G. Ashbee, "Measurement Of The Stress
Field Created Within The Resin Between Fibres In A Composite Material During

Cooling From The Cure Temperature," Journal of Materials Science, 16 (1981),
pp. 620-626.

Catherine A. Bigelow, "The Effects of Uneven Fiber Spacing on Thermal
Residual Stresses in a Unidirectional SCS-6/Ti-15-3 Laminate," Journal of
Composites Technology and Research, 14 (4) (1992), pp. 211-220.

R. C Averill. and G. P. Carman, “Analytical Modeling of Micromechanical Stress
Variations in Continuous Fiber-Reinforced Composites,” IUTAM Symposium on
‘Local Mechanics Concepts for Composite Material Systems, '(1991), pp. 28-31.

R. M. Jones, Mechanics of Composite Materials, Scripta Book Co., Washington,
1975,p176.

Alfred C. Loos and George S. Springer, "Effects of Thermal Spiking on Graphite-
Epoxy Composites." Journal of Composite Materials, 13 (1979), pp. 17-34.

E. L. McKague, Jr, J. E. Halkias and J. D. Reynolds, "Moisture in Composites:
The Effect of Supersonic Service on Diffusion," Journal of Composite Materials,
9 (1975), pp. 2-9.

Mark L. Karasek, Larry H. Strait, Maurice F. Amateau and James P. Runt, "Effect
of Temperature and Moisture on the Impact Behavior of Graphite/Epoxy
Composites: Part I-Impact Energy Absorption," Journal of Composites
Technology and Research, 17 (1) (1995), pp. 3-10.

 

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

313

T.K. Tsotsis, Y. Weitsman, "Energy Release Rates for Cracks Caused by
Moisture Absorption in Graphite/Epoxy Composites," Journal of Composite
Materials, 24 (1990), pp. 483-496.

Q. Zheng and R. J. Morgan, "Synergistic Thermal-Moisture Damage Mechanisms
Of Epoxies And Their Carbon Fiber Composites," Submitted to Journal of
Materials Science, April 28, 1993.

F. R. Jones, M. Mulheron, J. E. Bailey, "Generation Of Thermal Strains In GRP
Part 1 Effect Of Water On The Expansion Behavior Of Unidirectional Glass
Fibre-Reinforced Laminates," Journal of Materials Science, 18 (1983), pp. 1522-
1532.

Dick Cornelia, "BMI Mechanical Properties and Performance," Presented at High
Temple Workshop XV.

D. S. Varma, H. L. Needles, D. A. Kourtides, and R. H. Fish, “Interlaminar Shear
Properties of Graphite Fiber, High-Performance Resin Composites,” Polymer
Composites, 4 (2) (1983), pp. 98-103.

Loyd J. Burcham, Mark R. Vanlandingham, Rushad F. Eduljee, and John W.

Gillespie, Jr., “Moisture Effects on the Behavior of Graphite/Polyimide
Composites,” Polymer Composites, 17 (1996), pp. 682-690.

G. Wisanrakkit and J .K. Gillham, "The Glass Transition Temperature (Tg) As an
Index of Chemical Conversion For a High-Tg Amine/Epoxy System: Chemical

and Diffusion Controlled Reaction Kinetics," Journal of Coatings Technology, 62
(783) (1990), pp. 35-50.

Jiming Zhou and James P. Lucas, “The Effects of a Water Environment on the
Anomalous Absorption Behavior in Graphite/Epoxy Composites,” Composites
Science and Technology, 53 (1995), pp. 57-64.

Jiming Zhou and James P. Lucas, “Effects of Water on a Graphite/Epoxy
Composite,” Journal of Thermoplastic Composite Materials, 9 (1996), pp. 316-
328.

Alfred C. Loos and George S. Springer, "Effects of Thermal Spiking on Graphite-
Epoxy Composites," Journal ofComposite Materials, 13 (1979), pp. 17-34.

E. E. Shin and R. J. Morgan, "Effects of Thermal Cycling on Mechanical
Properties of High Temperature Bismaleimide (BMI)-Carbon Fiber Composites",
ANTEC '93, pp. 1357-1363.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

314

Venkatesh Rao And Lawrence T. Drzal, "The Dependence of Interfacial Shear
Strength on Matrix and Interphase Properties," Polymer Composites, 12 (1)
(1991), pp. 48-56.

V. Rao and L. T. Drzal, "The Temperature Dependence of Interfacial Shear
Strength for Various Polymeric Matrices Reinforced with Carbon Fibers,"
Journal of Adhesion, 37 (1992), pp. 83-95.

M. Pegoraro and L. Di Landro, “Influence of Components and Interface

Conditions on Mechanical Properties of Composites and Blends,” Macromol.
Symp, 70/71 (1993), pp. 193-212.

Gregory P. Carman, Ronald C. Averill, Kenneth L. Reifsnider, and J. N. Reddy,
“Optimization of Fiber Coatings to Minimize Stress Concentrations in Materials,”
Journal of Composite Materials, 27 (6) (1993), pp. 589-612.

G. Krekel, K. J. Huttinger, W. P. Hoffman, " The Relevance Of The Surface
Structure And Surface Chemistry Of Carbon Fibres. In Their Adhesion To High

Temperature Thermoplastics Part 11 Surface Chemistry," Journal of Materials
Science, 29 (1994), pp. 3461-3468.

M. Nardin, E. M. Asloun, and J. Schultz, "Study of the Carbon Fiber-Poly(Ether-
Ether-Ketone) (PEEK) Interfaces, 1: Surface Characterization of Fibers and

Matrices, and Interfacial Adhesion Energy," Polymers for Advanced
Technologies, 2 (1991), pp.109-114.

M. Nardin, E. M. Asloun and J. Schultz, "Study of the Carbon Fiber-Poly(Ether-
Ether -Ketone) (PEEK) Interfaces, 2: Relationship Between Interfacial Shear

Strength and Adhesion Energy," Polymers for Advanced Technologies, 2 (1991),
pp. 115-122.

M. Nardin, E. M. Asloun, J. Brandt, H. Richter, and J. Schultz, "Study of the
Carbon Fiber—Poly(Ether-Ether-Ketone) (PEEK) Interfaces, 4: Influence of Fiber-
Matrix Adhesion on the Mechanical Properties of Unidirectional Composites,"
Polymers for Advanced Technologies, 2 (1991), pp.171-176.

R. A. Bucher and J. A. Hinkley, “F iber/Matrix Adhesion in Graphite/PEKK
Composites,” Journal of Thermoplastic Composite Material, 5 (1992), pp. 2-13.

Ittipol Jangchud, R. K. Eby, A. M. Serrano, and M. A. Meador, “Studies of PAN-
Based Carbon Fiber Surfaces: Their Influence on Interfacial Bonding with PMR-
15 Polymide and Composite Therrno-Oxidative Stability,” Journal of Advanced
Materials, October (1996), pp. 19-25.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

315

C. L. Heisey, P. A. Wood, J. E. McGrath, and J. P. Wightman, “Measurement of
Adhesion Between Carbon Fibers and Bismaleimide Resins,” Journal of
Adhesion, 53 (1995), pp. 117-147.

S. O’Kell, and C. Jones, “A Study Of The Effects Of Low Power Plasma
Treatment On Graphite And Highly Orientated Pyrolytic
Graphite(HOPG)Surfaces,” Proceedings of the Eighteenth Annual Adhesion
Society Meeting, 1995, pp. 75-76.

Yu A. Dzenis, D. H. Reneker, V. V. Tsukruk, and R. Patil, “Fractal Analysis of
Surfaces of Advanced Reinforcing Fibers by Atomic Force Microscopy,”
Composite Interfaces, 2 (4) (1994), pp. 307-319.

Jean-Baptiste Donnet, Carbon Fibers, Marcel Dekker, New York, 1990.

Zhihong Wu, Charles U. Pittman, Jr., and Steven D. Gardner, “Nitric Acid
Oxidation of Carbon Fibers and the Effects of Subsequent Treatment in Refluxing
Aqueous NaOH,” Carbon, 33 (5) (1995), pp. 1-9.

Yue-Zhen Wei, Zhi-Qian Zhang, Yin Li, Zi-Hai Guo, Bai-Ling Zheng, " A Study
Of The Interfacial Bonding Between Carbon Fibre And PMR 15 Resin,"

Controlled Intgphases in Composite Materials, Hatsuo Ishida Editor, Elsevier
Science Publishing Co., Inc., 1990.

K. E. Atkinson, and C. Jones, “A Study of the Interphase Region in Carbon
F ibre/Epoxy Composites using Dynamic Mechanical Thermal Analysis,” Journal
of Adhesion, 56 (1996), pp. 247-260.

Tao Chou Chang, “The Effects of Fiber Surface Treatments on the Mechanical
Properties of Graphite Fiber/Bismaleimide Composites,” PhD Dissertation,
Auburn University, 1991.

King-Fu Lin and Jyh-Chien Chen, “Curing, Compatibility, and Fracture
Toughness for Blends of Bismaleimide and a Tetrafunctional Epoxy Resin,”
Polymer Engineering and Science. 36 (2) (1996), pp. 211-217.

J. M. Cartwright, S. P. Wilkinson, J. Hellgeth, T. C. Ward, and D. E. Kranbeuhl,
“Toughened Bismaleimide Resins: Kinetics, Morphology and Processing, Part 1,”
pp. 51-52.

L. Guozheng, G. Aijuan, and L. Liwen, “Modification of BMI and Allyl
Compounds,” Journal of Advances Materials, January (1996), pp. 61-64.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

316

S. P. Wilkinson, T. C. Ward, and J. E. McGrath, “Effect of Thermoplastic
Modifier Variables on Tougheninga Bismaleimide Matrix Resin for High-
Performance Composite Materials,” Polymer, 34 (4) (1993), pp. 870-884.

S. D. Jenkins, G. T. Emmerson, P. T. Mcgrail And R. M. Robinson,
"Thermoplastic Sizing of Carbon Fibres in High Temperature Polyimide
Composites," Journal of Adhesion, 45 (1994), pp. 15-27.

J. Denault, “Interfacial Interaction in Carbon Fiber/Thermoplastic Composites,”
Composite Interfaces, 2 (4) (1994), pp. 275-289.

T. H. Cheng, J. Zhang, S. Yumitori, F. R. Jones, and C. W. Anderson, “Sizing
Resin Structure and Interphase Formation in Carbon Fiber Composites,”
Composites, 25 (7) (1994), pp. 661-669.

Charles U. Pittman, Jr., Steven D. Gardner, Guoren He, Lichang Wang, Zhihong
Wu, C. Singamsetty, Biahua Wu, and Glyn Booth, “Reactions of Defined
Oxidized Carbon Fiber Surfaces with Model Compounds and Polyurethane
Elastomers.” MRS, April 1995.

V. Raghavendran, Effects of Molecular Weight of Thermoplastic Matrix aLd
Processing Conditions of Interfacial Adhesion in Carbon Fiber Composites, MS
Thesis, Michigan State University, 1994.

G. Rajagopalan, K. M. Immordino, S. H. McKnight, and J. W. Gillespie, Jr.,
“Characterization and Fracture Toughness of Diffuse Adhesive Interphases,”
Advances in Polymers and Fibers, pp. 539-547.

C. L. Weitzsacker, and L. T. Drzal, SPE Conference Proceedings: ANT EC ’96,
May 5-10, 1996, pp. 3656-3660.

T. Lenz, J. K. McDowell, B. Moy, J. Sticklen, and M. C. Hawley, “Intelligent
Decision Support for Polymer Composite Material Design in an Integrated Design

Environment,” Proceedings of the American Society for Composites Ninth
Technical Conference, Newark, DE, September 20-22, 1994, pp. 685-691.

M. Nardin, E. M. Asloun, F. Muller, and J. Schultz, “Study of the Carbon Fiber-
Poly(Ether-Ether-Ketone) (PEEK) Interfaces, 3: Influences and Properties of
Interphases,” Polymers for Advanced Technologies, 2 (1991), pp. 161-169.

A. R. Sanadi and R. V. Subramanian, V.S. Manoranjan, "The Interphasial
Regions in Interlayer Fiber Composites," Polymer Composites, 12 (6) (1991), pp.
377-383.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

317

J. G. Williams, M. E. Donnellan, M. R. James, and W. L. Morris, “Properties 0f
the Interphase in Organic Matrix Composites,” Materials Science and
Engineering, A126 (1990), pp. 305-312.

John F. Maguire, Peggy L. Talley, and Mark Lupkowski, “The Interphase in
Adhesion: Bridging the Gap,” Journal of Adhesion, 45 (1994), pp. 269-290.

Javad Kalantar, “The Bonding Mechanism of Aramid Fibers to Epoxy Matrices,”
MS Thesis, Michigan State University, 1988.

Ryutoku Yosomiya, Kiyotake Morimoto, Akio Nakajima, Yoshito Ikada, and
Toshio Suzuki, Adhesion and Bonding in Composites, Marcel Dekker, Inc., New
York, 1990.

D. J. David and T. F. Sincock, “Estimation of Miscibility of Polymer Blends
Using the Solubility Parameter Concept,” Polymer, 33 (21), 1992, pp. 4505-4514.

Sidney H. Goodman, Handbook of Thermoset Plastics, Noyes Publications, Park
Ridge, New Jersey, 1986.

R. T. Morrison, and R. N. Boyd, Organic Chemistry, Allyn and Bacon, Boston,
MA, 1966.

George G. Odian, Principles of Polymerization, 3rd Ed., John Wiley & Sons, New
York, 1991.

Materials Handbook: Plastics

F. Rodriguez, Principles of Polymer Systems, Hemisphere Pub. Corp., New York,
1989.

J. March, Advanced Organic Chemistry: Reactions, Mechanisms, and Structure,
_4"‘_Eg, John Wiley & Sons, New York, 1992.

R. Yosomia et al, Adhesion and Bonding in Composites, Marcell Dekker, New
York, 1990.

E. Plueddeman, Silane Coupling Agents, 2nd Ed., Plenum Press, New York,
1982.

 

Petrarch Systems - Silane Coupling Agent Selector