Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites
(USC Thesis Other)
Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
PROCESSING AND PROPERTIES OF PHENYLETHYNYL-
TERMINATED PMDA-TYPE ASYMMETRIC POLYIMIDE
AND COMPOSITES
By
Yixiang Zhang
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)
December 2018
Copyright 2018 Yixiang Zhang
i
Dedication
The manuscript is dedicated to my husband and parents who love me without
conditions.
ii
Acknowledgements
With great gratitude, I completed this dissertation. In my PhD study, what I have
obtained is not just professional training but also happy five-year time. Many people
offered help and support to me, and I am sincerely grateful to all of them.
I would first like to like to acknowledge my advisor, Prof. Steven Nutt. He guided
me into the field of composites and educated me to be become an independent researcher.
In addition to professional knowledge and skills he equipped me with, I also appreciate his
trust and protection. In the last five years, whenever I got involved in conflicts or troubles,
he always trusted and protected me without any hesitation. I cannot feel luckier to have
him as my advisor. Moreover, his integrity has a profound influence on me. Through
working with him, I have become a more honest and moral person.
Many colleagues in our research group, M.C. Gill Composite Center, have provided
me with help, assistance and guidance, and many thanks must be offered to them. I would
like to acknowledge William Edwards for microCT measurement, Mark Anders for
specimen machining, Patricio Martinez for sample preparation, Wei Hu for guidance in
data fitting, Yijia Ma for assistance with electron microscopic observation, and Sidharth
Sarojini Narayan and Roberto Postiglioni for long-time assistance in my project. Thanks
also go to Dr. Lessa Grunenfelder, Dr. Atul Jain and Dr. Timotei Centea for valuable
advising to my work, as well as to our lab manager, Yunpeng Zhang, who always gave me
a hand when I need help. In addition to assistance, I am also benefited from scientific
discussions with them, which not only cleared up my confusions but also enlightened me
on my research. I cannot complete my tasks without the help of my colleagues. Our
iii
research group is like a family for me, who always deliver warm to me and never disappoint
me.
Funding for this work was made possible primarily from Kaneka Corporation and
Henkel Corporation. Particularly, Kaneka Corporation offered research materials and
consumables. In addition, I would like to thank Dr. Masahiko Miyauchi of Kaneka
Corporation for his guidance, support, and assistance throughout my graduate research. He
is one of the inventors of TriA X resin system that is my study subject in my PhD training
program. His hands-on training and guidance in polymer chemistry are greatly appreciated.
Finally, I would like to thank my parents and husband. Because of difficulties of
international trip, I had traveled back to my home town only once in the past five years,
although I really miss my family very much. My parents not only understand my situations
but also traveled 8,000 miles across the Pacific to see me. I could not have accomplished
my long study journey without their unconditional love. My love, Dr. Ming Xia,
encouraged me to study abroad and see the outside world. In personal life, he takes care of
me in every possible way. Thank you to my darling for the harbor and home you gave to
me.
iv
Table of content
Dedication ............................................................................................................................ i
Acknowledgements ............................................................................................................. ii
List of Tables ..................................................................................................................... ix
List of Figures ..................................................................................................................... x
Abstract ............................................................................................................................ xiii
Chapter 1. Introduction ....................................................................................................... 1
1.1 Motivation ................................................................................................................. 1
1.2 Polyimide .................................................................................................................. 2
1.2.1 Condensation-type polyimide ............................................................................ 2
1.2.2 Addition-type polyimide .................................................................................... 4
1.2.3 Asymmetric polyimide ....................................................................................... 6
1.3 Polyimide composites ............................................................................................... 9
1.3.1 Applications ....................................................................................................... 9
1.3.2 Processing ........................................................................................................ 10
1.3.3 Environmental effects on polyimide composites ............................................. 13
1.4 Scope of dissertation ............................................................................................... 14
Chapter 2. Structure and properties of TriA X system ..................................................... 16
2.1 Introduction ............................................................................................................. 16
2.2 Materials ................................................................................................................. 17
2.2.1 PMDA di-ester/p-ODA/PEPA mono-ester monomers .................................... 17
2.2.2 PMDA/4,4’-ODA/PEPA imide oligomer ........................................................ 18
2.2.3 TriA X polyimide panels.................................................................................. 19
v
2.3 Experiments ............................................................................................................ 20
2.3.1 X-ray diffraction .............................................................................................. 20
2.3.2 Thermogravimetric analysis ............................................................................. 22
2.3.3 Differential scanning calorimetry .................................................................... 23
2.3.4 Rheological measurements .............................................................................. 23
2.3.5 Mechanical tests ............................................................................................... 23
2.4 Results and discussion ............................................................................................ 24
2.4.1 Material structure ............................................................................................. 24
2.4.2 Thermal processability ..................................................................................... 29
2.4.3 Thermal properties ........................................................................................... 30
2.4.4 Mechanical properties ...................................................................................... 32
2.5 Conclusions ............................................................................................................. 38
Chapter 3. Process development for TriA X composites .................................................. 40
3.1 Introduction ............................................................................................................. 40
3.2 Materials ................................................................................................................. 41
3.2.1 PMDA di-ester/p-ODA/PEPA mono-ester monomers .................................... 41
3.2.2 T650-35 8HS/TriA X prepreg .......................................................................... 41
3.3 Experiments ............................................................................................................ 41
3.3.1 Thermogravimetric analysis ............................................................................. 41
3.3.2 Rheological measurements .............................................................................. 42
3.3.3 Gel Permeation chromatography analysis........................................................ 43
3.3.4 Differential scanning calorimetry .................................................................... 43
3.3.5 Dynamic mechanical analysis .......................................................................... 44
vi
3.3.6 Composite laminate fabrication ....................................................................... 44
3.3.7 Porosity measurement ...................................................................................... 45
3.3.8 Mechanical tests ............................................................................................... 46
3.4 Results and discussion ............................................................................................ 47
3.4.1 Processability ................................................................................................... 47
3.4.1.1 Chemo-rheological properties ................................................................... 47
3.4.1.2. Effect of b-staging temperature ............................................................... 51
3.4.2 Composite fabrication ...................................................................................... 56
3.4.3 Composite properties ....................................................................................... 60
3.5 Conclusions ............................................................................................................. 62
Chapter 4. Effects of thermal cycling on TriA X composites ........................................... 64
4.1 Introduction ............................................................................................................. 64
4.2 Thermal stress analysis ........................................................................................... 67
4.2.1 Material properties ........................................................................................... 67
4.2.2 Thermal stress calculation ................................................................................ 69
4.3 Thermal cycling experiments .................................................................................. 71
4.3.1 Materials........................................................................................................... 71
4.3.2 Thermal cycling conditions .............................................................................. 72
4.3.3 Microcrack Inspection...................................................................................... 72
4.3.4 Oxidized layer inspection................................................................................. 73
4.3.5 Dynamic Mechanical Analysis ........................................................................ 73
4.3.6 Nano Indentation .............................................................................................. 73
4.3.7 Short-Beam Shear Test .................................................................................... 74
vii
4.4 Results and discussion ............................................................................................ 75
4.4.1 Microcrack inspection ...................................................................................... 75
4.4.2 Effects on matrix .............................................................................................. 81
4.5 Conclusions ............................................................................................................. 85
Chapter 5. Effects of moisture on TriA X and composites ............................................... 88
5.1 Introduction ............................................................................................................. 88
5.2 Materials ................................................................................................................. 90
5.3 Experiments ............................................................................................................ 91
5.3.1 Moisture uptake................................................................................................ 91
5.3.2 Fourier-transform infrared spectroscopy ......................................................... 92
5.3.3 X-ray photoelectron spectroscopy ................................................................... 92
5.3.4 Dynamic mechanical analysis .......................................................................... 93
5.3.5 Thermogravimetric analysis ............................................................................. 93
5.3.6 Tensile tests ...................................................................................................... 93
5.3.7 Short-beam shear tests ..................................................................................... 94
5.4 Results and discussion ............................................................................................ 96
5.4.1 Moisture absorption behavior .......................................................................... 96
5.4.2 Hydrothermal aging ....................................................................................... 102
5.4.2.1 Hydrolysis ............................................................................................... 102
5.4.2.2 Thermal properties .................................................................................. 104
5.4.2.3 Mechanical behavior ............................................................................... 107
5.5 Conclusions ........................................................................................................... 111
Chapter 6. Conclusions and recommendations ............................................................... 114
viii
6.1 Concluding remarks .............................................................................................. 114
6.2 Recommendations for future work ....................................................................... 116
References ....................................................................................................................... 119
ix
List of Tables
Table 2-1. TriA X imide oligomers and polyimides. ........................................................ 21
Table 2-2. Specimens for mechanical tests. ...................................................................... 34
Table 2-3. Mechanical properties of TriA X. ................................................................... 35
Table 3-1. Molded TriA X/T650-35 8HS laminates. ........................................................ 45
Table 3-2. Cycles for rheological measurements with ramps from isothermal holds to
400°C at 1°C min
-1
. ........................................................................................................... 54
Table 3-3. Specimens for mechanical tests. ...................................................................... 61
Table 3-4. Mechanical properties of T650-35 8HS/TriA X. ............................................ 62
Table 4-1. CTE of TriA X................................................................................................. 67
Table 4-2. Specimens for SBS tests before and after thermal cycling.............................. 74
Table 5-1. Specimens for tensile tests before and after hydrothermal aging. ................... 94
Table 5-2. Specimens for SBS tests before and after hydrothermal aging. ...................... 95
Table 5-3. Curve fitting results using Equation (5-2) and (5-3). ...................................... 98
Table 5-4. Arrhenius parameters of Equation (5-4) and (5-5). ....................................... 100
x
List of Figures
Figure 1-1. Chemical structures of polyimides: (a) aromatic heterocyclic polyimides and
(b) linear polyimides. .......................................................................................................... 2
Figure 1-2. Synthesis of Kapton polyimide. ....................................................................... 3
Figure 1-3. Synthesis of PMR-15. ...................................................................................... 5
Figure 1-4. Chemical structures of (a) PETI-5 and (b) AFR-PE-4. .................................... 6
Figure 1-5. Chemical structure of TriA X. ......................................................................... 9
Figure 1-6. PMR-15 composite inner cowl for the NASA Quiet Clean Short-Haul
Experimental Engine.
6
...................................................................................................... 10
Figure 1-7. Typical autoclave cure cycle for PMR-15 composites. ................................. 12
Figure 2-1. (a) TriA X and (b) PMDA/4,4’-ODA/PEPA polyimide (control). ................ 17
Figure 2-2. Molding cycle for TriA X resin: (a) oven cycle and (b) hot press cycle. ...... 19
Figure 2-3. XRD patterns of TriA X and control.............................................................. 25
Figure 2-4. DSC curves of TriA X imide oligomers and polyimides. .............................. 27
Figure 2-5. (a) Complex viscosity |η*| profiles of TriA X imide oligomer (red solid) and
the control imide oligomer (blue dash) and images of TriA X imide oligomer (b) and
control imide oligomer (c) before (top) and after (bottom) measurements. ..................... 30
Figure 2-6. Tg’s and Td’s* of TriA X, PMR-15 and AFR-PE-4. ...................................... 31
Figure 2-7. Stress-strain curves of TriA X tensile tests. ................................................... 34
Figure 2-8. TriA X specimen of flexural test: (a) in test and (b) after test. ...................... 35
Figure 2-9. Mechanical properties of TriA X (red), PMR-15 (blue) and AFR-PE-4
(yellow)*. .......................................................................................................................... 37
Figure 3-1. Chemical reactions in TriA X system. ........................................................... 47
xi
Figure 3-2. (a) Dynamic |η*|, (black solid), G’ (red dash) and G” (blue short dash) profiles
for Oligomer 0 at a ramp rate of 1°C min
-1
, and (b) Mn change of Oligomer 0 upon heating
during rheological measurement. ...................................................................................... 48
Figure 3-3. (a) Weight loss (TGA) and (b) complex viscosity of Oligomer 0. ................ 50
Figure 3-4. Imidization and cross-linking relations in PMR polyimides: (a) isolation and
(b) overlap. ........................................................................................................................ 52
Figure 3-5. |η*| profiles (solid lines) and temperature profiles (dash lines) of the B-staged
TriA X resins in Cycle 1 (black), Cycle 2 (red) and Cycle 3 (blue). ................................ 54
Figure 3-6. Heat of reaction and degree of imidization of the B-staged TriA X resins. ... 56
Figure 3-7. Molding cycle for TriA X/T650-35 8HS: (a) Oven cycle and (b) hot press cycle.
........................................................................................................................................... 59
Figure 3-8. Bagging scheme for TriA X/T650-35 8HS. ................................................... 60
Figure 3-9. TriA X/T650-35 8HS laminate with a lay-up of [0/+45/-45/90]2s. ................ 60
*
Determined by image analysis. ....................................................................................... 61
Figure 4-1. Young’s modulus of neat TriA X................................................................... 68
Figure 4-2. (a) Temperature profile of thermal cycles, and (b) corresponding thermal stress
in cross-ply region (solid black) and crimped region (dashed black) of [0]8 laminates and
tensile strength of the matrix (blue). ................................................................................. 70
Figure 4-3. 8HS laminates after 2000 cycles: [0]8 in 0° (a), 45° (b), and 90° (c), and [0/+45/-
45/90]s in 0° (d), 45° (e), and 90° (f). ............................................................................... 76
Figure 4-4. A typical microcrack (boxed in red rectangles) observed in microCT scans of
8HS laminates after 2000 cycles; image sequence from (a) to (h) is corresponding to
xii
direction from a to h indicated in Figure 4-5; (i) fiber morphology in 8HS weave: tow edges
(I and II) at an overlap of orthogonal tows (I-II) and cross-ply region (III). .................... 77
Figure 4-5. Microcrack distribution observed in microCT scans of T650-35 8HS/TriA X
after 2000 cycles; color indicates ply position of microcracks. ........................................ 78
Figure 4-6. Cross-section of neat TriA X after 2000 cycles (transmitted light). .............. 82
Figure 4-7. Tg and Young’s modulus at surface of neat TriA X. ...................................... 82
Figure 4-8. Tg and SBS strength of T650-35 8HS/TriA X. .............................................. 84
Figure 5-1. Moisture gain of neat polyimide in water at different temperatures. The symbols
are experimental values, and the solid lines are curve fits using Equation (5-2).............. 97
Figure 5-2. Diffusivity and moisture saturation in water vs 1/T. The black circle symbols
are Dx listed in Table 5-3, and the blue diamond symbols are experimental results of M ∞.
The black solid line is curve fit using Equation (5-4), and the blue dashed and dot-dash
lines are curve fits by Methods I and II, respectively. ...................................................... 99
Figure 5-3. Moisture saturation vs relative humidity at 35°C. ....................................... 101
Figure 5-4. FTIR spectra of TriA X. ............................................................................... 103
Figure 5-5. Hydrolytic reactions in polyimides. ............................................................. 103
Figure 5-6. Tan 𝛿 curves of TriA X. ............................................................................... 105
Figure 5-7. Tg and Td5% of TriA X. ................................................................................. 107
Figure 5-8. Tensile properties of neat polymer (TriA X). .............................................. 108
Figure 5-9. Stress-stain curves of TriA X in tensile tests. .............................................. 109
Figure 5-10. SBS strength of T650-35 8HS/TriA X composites. ................................... 110
xiii
Abstract
A new type of polyimide, designated TriA X, has been developed for high-
temperature composite applications. TriA X is a polymerized monomeric reactant (PMR)
type polyimide derived from pyromellitic dianhydride (PMDA), 2-phenyl-4,4’-
diaminodiphenyl ether (p-ODA) and phenylethynyl phthalic anhydride (PEPA). In this
dissertation, a TriA X resin (with degree of polymerization n = 7 in the imide oligomer)
was investigated for developing a new polyimide composite and addressing the current
issues associated with conventional polyimide matrices.
To advance understanding of this new imide resin system, comprehensive
characterization was preformed to investigate polymer structure, processability, thermal
and mechanical properties and establish the relationship between the molecular structure
and those properties. TriA X features an asymmetric, irregular, nonplanar backbone. Both
the imide oligomers and the cross-linked polyimides of TriA X exhibited loose-packed
amorphous structures, independent of thermal processing. The peculiar structures were
attributed to the asymmetric backbone, which effectively prevents the formation of closed-
packed chain stacking typically observed in polyimides. The imide oligomers exhibited a
lower melt viscosity than a control imide oligomer (symmetric and semi-crystalline),
indicating a higher chain mobility above the glass transition temperature (Tg). The cured
polyimide exhibited a Tg = 362°C and a decomposition temperature (Td) = 550°C. The
cross-linked TriA X exhibited exceptional toughness and ductility (e.g., 15.1% at 23°C)
for a polyimide, which was attributed to the high molecular weight oligomer and loose-
xiv
packed amorphous structure. The thermal and mechanical properties of TriA X surpass
those of PMR-15 and AFR-PE-4.
The asymmetric and non-planar backbone structure endows cured TriA X with
amorphous structure and high toughness that are attractive properties for composite
applications. Processability of the imide resin system and performance of associated carbon
fiber composites were investigated. Rheological measurements were performed on an
oligomer with a low degree of imidization to understand the chemo-rheology of the resin
system and determine a suitable B-staging temperature. A composite molding cycle was
designed, which yielded fully-consolidated woven carbon fiber laminates. Void contents
in panels produced with this molding cycle were < 0.1% as measured by image analysis of
polished sections, and < 0.2% as measured by X-ray micro-computed tomography. Matrix-
dominated mechanical properties of composites fabricated with TriA X exceeded those of
PMR-15 composites. These mechanical properties and a measured Tg of 367°C indicate
potential for use of this resin system in high-temperature composites.
To further explore the potential of this new polyimide as a composite matrix for
severe service environments. The effects of thermal cycling on TriA X composites were
investigated. Composite specimens were subjected to 2000 thermal cycles between -54°C
and 232°C. At 400-cycle intervals, laminates were inspected for microcracks, and Tg and
short-beam shear (SBS) strength were measured. The composites did not exhibit
microcracks after thermal cycling, although after 2000 thermal cycles, mechanical
properties of the matrix declined slightly. The matrix degradation decreased the resistance
to microcracking upon further loading. No effects of thermal oxidative aging were
observed from thermal cycling, and thermally driven fatigue and creep were identified as
xv
the primary and secondary factors inducing mechanical degradation of the matrix. The Tg
of the composites exhibited no change after 2000 cycles, while the SBS strength decreased
slightly (3-9%). The results highlight the potential for use of TriA X composites as long-
term structural components in high-temperature service environments.
In addition to cyclic temperatures, the effects of moisture on neat TriA X and
associated composites were also investigated. Water uptake tests were performed on the
polyimide at various temperatures and relative humidity levels to investigate moisture
absorption behavior. Two-stage moisture absorption was observed, in which the first stage
was diffusion-controlled, while the second stage was moisture plasticization-controlled. As
exposure temperature increased, the equilibrium moisture content of the polyimide
decreased, indicating an exothermic absorption process. The Arrhenius temperature
dependence and moisture saturation as functions of temperature and humidity in the neat
polymer were determined using curve-fitting based on published mathematical models.
Long-term hydrothermal aging at 95°C was conducted on the neat polyimide and
associated carbon fiber composites. Reversible hydrolytic reactions and a trace of
irreversible hydrolysis were observed in the long-term exposure. The tensile ductility of
the neat polyimide and the short-beam shear strength of the composites decreased with
increasing aging time, while the tensile strength and modulus and thermal properties of the
polyimide exhibited little change after 2000-h aging, demonstrating hydrothermal stability.
The decrease in the ductility of the neat polymer after long-term moisture exposure was
attributed to the network structure change, driven by hydrolysis and moisture plasticization.
1
Chapter 1. Introduction
1.1 Motivation
Carbon fiber reinforced polymer-matrix composites have been used in a wide range
of applications, such as aircraft, automobiles, sports equipment, and medical devices.
Especially, the demand of carbon fiber composites is increasing in aerospace because of
greater strength-to-weight ratio than traditional aero metals. The service temperatures of
composites are matrix-dominated. The most commonly used matrices are epoxies, which
are able meet multiple-functional requirements at prolonged service temperature up to
121°C, and can withstand short-duration spikes up to 204°C.
1
However, this temperature
range is insufficient for aerospace applications in hot zones, such as aeroengines and
supersonic airplanes. Thus, high-temperature resin matrices are needed to push service
temperatures to a greater range − beyond the capabilities of epoxies.
Polyimides possess the highest glass transition temperature
2
(Tg) among all
engineering plastics and exceptional thermal and thermal oxidative stability in severe
conditions. In addition, polyimides maintain high machinal performance at elevated
temperatures. Thus, polyimides have been employed as matrices in high-temperature
composites since 1970s.
3
However, conventional polyimides, e.g., PMR-15, have a serious
limitation – brittleness, limiting the use of polyimide composites in structural applications.
Therefore, the development of new polyimides has never stopped, and meanwhile, more
study on polyimides and associated composites is needed to obtain more accurate
understanding of this group of polymers for more effective designs.
2
1.2 Polyimide
Polyimide is a polymer of imide monomers (Figure 1-1). Depending on the polymer
chain, polyimides can be classified as two major forms: (1) aromatic heterocyclic
polyimides where the imide group is part of a cyclic unit in the polymer chain (Figure 1-
1a), and (2) linear polyimides where the atoms of the imide group are part of a linear chain
(Figure 1-1b). Aromatic heterocyclic polyimides are more widely used in engineering
applications and have attracted more research attention than linear polyimides, because of
superior mechanical and thermal properties.
4, 5
Figure 1-1. Chemical structures of polyimides: (a) aromatic heterocyclic polyimides and
(b) linear polyimides.
1.2.1 Condensation-type polyimide
According to the type of interactions between the main chains, aromatic polyimides
can be classified as condensation-type (non-cross-linked) and addition-type (cross-linked).
Non-cross-linked polyimides are referred to as condensation-type polyimides because
cured polymers are derived from condensation reactions.
6
Based on thermal processability,
condensation-type polyimides are categorized as thermoplastics and non-thermoplastics.
Kapton (developed by DuPont) is the oldest thermoplastic polyimide, which has been used
in a variety of electrical and thermal insulation applications for over 45 years.
3
Thermoplastic polyimides are generally prepared via a two-step synthesis from diamines
and dianhydrides, shown in Figure 1-2 using Kapton an example.
7
In the first step, 4,4’-
oxydianiline (4,4’-ODA) reacts with pyromellitic dianhydride (PMDA) in an anhydrous
organic solvent, typically N,N-dimethylacetamide (DMAc), at room temperature, forming
polyamic acid. In the second step, cyclodehydration reactions of polyamic acid close imide
rings and form polyimide, generating water as by-products.
Figure 1-2. Synthesis of Kapton polyimide.
Like thermosets, some condensation-type polyimides are not thermally remoldable
because of ultrahigh molecular weights or strong interchain integrations, although this type
of polyimide also have linear structures like thermoplastic polyimides. Commercial non-
thermoplastic condensation-type polyimides include Pyralin, Skybond and NR-150B.
Skybond and NR-150B series are based on 3,3’,4,4’-benzophenonetetra-carboxylic
dianhydride (BTDA) and 2,2'-bis(3,4-dicarboxyphenyl) hexafluoropropane dianhydride
4
(6FDA), respectively, supplied by DuPont, while Pyralin resins (developed by Monsanto
Plastics & Resins Company) are based on multiple dianhydrides, such as PMDA, BTDA
and 3,3’,4,4’-biphenyltetracarboxclic dianhydride (BPDA).
6, 8, 9
1.2.2 Addition-type polyimide
Thermoset polyimides are referred to as addition-type polyimides because cross-
linked networks are derived from addition reactions (cross-linking reactions). Thermoset
polyimides are more widely used in aerospace applications than thermoplastic polyimides,
particularly as matrices in high-temperature composites. The most widely known
thermoset polyimide is PMR-15, developed by NASA in 1972.
10
The synthesis of PMR-
15 is achieved via a polymerization of monomeric reactants (PMR) approach (Figure 1-3).
First, BTDA and 5-norbornene-2,3-dicarboxylic anhydride (NA) are esterified in methanol,
and subsequently methylene dianiline (MDA) is added into the mixture. Then, the mixture
of monomers reacts at ~200°C, forming amide acid oligomers and generating methanol as
by-product, followed by the closing of imide rings, forming imide oligomers and
generating water as by-product. Finally, the imide oligomers cross-link at greater
temperatures, forming cross-linked polyimide. Although PMR-15 is the most common
polyimide in aerospace applications, cured PMR-15 has a critical drawback–the polymer
is notoriously brittle (because of a high cross-link density),
11
leading to unavoidable
microcracking and degradation during thermal cycling.
12, 13
Moreover, one of three
monomers in PMR-15 – MDA – is a suspected carcinogen, causing serious safety concerns
in manufacturing.
3
These shortcomings have historically limited (but not prevented) the
use of PMR-15, and since its inception a suitable alternative has been sought.
5
Figure 1-3. Synthesis of PMR-15.
In the 1990s, NASA Langley Research Center began development of another series
of polyimides with phenylethynyl-terminated imide oligomers (PETI) for a Mach 2.4 high-
speed civil transport program.
14-16
PETI-5, one of the highly touted PETIs, is derived from
3,3’,4,4’-biphenyltetracarboxclic dianhydride (BPDA), 3,4’-oxydianiline (3,4’-ODA), 1,3-
bis(3-aminophenoxy)benzene (1,3-bis(3-APB)), and 4-phenylethynyl phthalic anhydride
(PEPA) (Figure 1-4a).
15, 17
Carbon fiber composites fabricated with PETI-5 can reduce
weight of the fuselage, outboard wing, strake and empennage of a supersonic airliner, and
6
also withstand the high skin temperatures associated with the aerodynamic friction heating
caused by supersonic cruise speeds.
18
AFR-PE-4 is another thermoset polyimide with
PEPA endcap, developed by the U.S. Air Force Research Laboratory during the mid 1990s
in cooperation with the University of Dayton Research Institute.
1, 19
AFR-PE-4, derived
from 6FDA, 1,4-diaminobenzene (p-PDA) and PEPA (Figure 1-4b), has been applied in
advanced propulsion and structural systems.
1
In this dissertation, PMR-15, AFR-PE-4, and
PETI-5 are considered as industrial standard reference materials to compare with our
studied polyimide that is introduced in 1.2.4.
Figure 1-4. Chemical structures of (a) PETI-5 and (b) AFR-PE-4.
1.2.3 Asymmetric polyimide
The exceptional thermal stability and mechanical properties of polyimides derive
from strong interchain attractive forces involving charge transfer complexation between
dianhydride and diamine groups, polar interactions and aromatic 𝜋 -𝜋 stacking.
5
The
intermolecular interactions effectively reduce intersegmental distance and chain mobility
and increase chain rigidity, resulting in high Tg and ultimate strength, but poor
processability (e.g. high viscosity and low solubility) and toughness. Decades of effort have
7
been devoted to development of new polyimides by tailoring molecules and adjusting
interchain interactions to increase processability and toughness.
One way of decreasing chain rigidity and hence increasing processability and
mechanical properties is accomplished at the molecular level by incorporating flexible
segments into backbone structures.
20
PETI-5 is such an example, in that the high number
of flexible ether linkages are induced into the backbone by using 3,4’-ODA and 1,3-bis(3-
APB) as backbone diamines. Although the flexible ether linkages in the repeat units of
3,4’-ODA and 1,3-bis(3-APB) effectively increase the polymer ductility, many flexible
moieties in the backbone reduce the Tg (270°C) significantly compared to that of PMR-15
(341°C).
15, 21, 22
Moreover, the flexible and symmetric backbone of PETI-5 lead to the
formation of a semi-crystalline structure,
23
which is undesirable for composite processing.
Although this approach can increase polymer toughness, thermal properties are
significantly reduced, making polyimides not “high-temperature polymers”. Thus, other
strategies of molecular design have been sought without sacrificing thermal properties.
Over the past few decades, an alternative approach to improving material properties
and processability of polyimides has been explored, based on constructing polyimides from
asymmetric monomers. For example, Hsiao et al. prepared a series of polyimides
containing asymmetric diaryl ether segments derived from 5-(4-aminophenoxy)-1-
naphthylamine, which exhibited enhanced solubility, amorphous (non-crystalline)
structure, and high toughness compared to analogous polyimides based on symmetric
diamines.
24
Using a similar approach, Chern et al. incorporated asymmetric di-tert-butyl
groups into polyimides by imidizing asymmetric diamines and various dianhydrides.
25, 26
The asymmetric repeat units acted to decrease intermolecular forces and reduce chain
8
packing in the resulting polymers, resulting in high solubility, low moisture absorption,
and a low dielectric constant.
A thermoset polyimide (UPILEXTM-AD) derived from 2,3,3’,4’-
biphenyltetracarboxylic dianhydride (a-BPDA), 4,4’-ODA and PEPA was developed by
Yokota et al. and commercialized by Ube Industries Ltd.
27, 28
The distorted/non-planar
structure of a-BPDA inhibits intermolecular interactions, preventing dense chain stacking
and increasing molecular mobility above the Tg, thereby improving processability. The
asymmetric structure, however, suppresses internal rotation around the biphenyl linkage in
the a-biphenyldiimide moiety, and thus the cured polymer also exhibits a high Tg.
Composites fabricated with this polyimide (UPILEXTM-AD) exhibit greater short-beam
shear (SBS) strength and flexural strength than IM7/PETI-5.
28
Taking advantage of the low
melt viscosity and high Tg of a-BPDA based polyimides, NASA Langley Research Center
developed a polyimide for RTM liquid molding, PETI-330. PETI-330 is derived from a-
BPDA, 1,3-diaminobenzene, 1,3-bis(4-aminophenoxy)benzene and PEPA, and exhibits a
low complex melt viscosity (0.01–1 Pa⋅s) at 280°C and is stable for ≥ 2 h at this
temperature.
29
More recently, Miyauchi et al. from Kaneka Corporation developed an asymmetric
polyimide derived from PMDA, 2-phenyl-4,4’-diaminodiphenyl ether (p-ODA) and PEPA
(Figure 1-5).
30-32
This new polyimide system is referred to as “TriA X” for the
characteristics of the polymer: amorphous, asymmetric, addition-type and cross-linked (×).
Because the final cure of thermoset polyimides occurs by means of addition reactions,
thermoset polyimides are usually referred to as addition-type polyimides.
6
TriA X features
an asymmetric, irregular, and nonplanar backbone structure, resulting in an amorphous
9
structure and greater toughness and ductility than conventional polyimides (e.g., εb = 15.1%
at room temperature), surpassing those of conventional polyimides, e.g., PMR-15 and
AFR-PE-4.
33
This dissertation focuses on the fundamental study on TriA X and associated
composites.
Figure 1-5. Chemical structure of TriA X.
1.3 Polyimide composites
1.3.1 Applications
Polyimide composites are widely used in high-temperature aerospace applications.
For example, Pyralin and Skybond polyimide composites are used in secondary structural
applications, such as sound suppressant panels in turbofan engines and radomes, while the
major application for NR-150B polyimide composites is in nozzle flaps for the Pratt &
Whitney F-100 engine.
6
Because the cross-linked networks endow addition-type
polyimides greater thermal properties than condensation-type polyimides, addition-type
polyimide composites are employed in aerospace applications more than condensation-
type polyimides. For instance, PMR-15 is used fairly extensively in military and
commercial aircraft engines components, seeing service temperatures as high as 288°C,
10
including the outer bypass ducts for the F404 engine, the center vent tube for the GE90
engine, the exit flaps for the F100-PW-229 engines and the splitters and inner ducts for the
F110 engine.
34, 35
Figure 1-6 shows the inner cowl for the NASA Quiet Clean Short-Haul
Experimental Engine developed by General Electric.
6
Figure 1-6. PMR-15 composite inner cowl for the NASA Quiet Clean Short-Haul
Experimental Engine.
6
1.3.2 Processing
Because of the intractability of fully imidized polymers, condensation-type
polyimides are usually supplied as precursor solutions of polyamic acids or monomers in
high-boiling point aprotic solvents (e.g., typically N-methyl-2-pyrrolidone, NMP).
6, 9
Condensation-type polyimide prepreg is prepared by pre-impregnating carbon fiber beds
with precursor solutions.
6
Cured polyimide composites are produced by applying heat to
11
remove solvents and facilitate imidization, forming fully imidized polyimides.
9
However,
this processing approach is somehow problematic. First, high-boiling point aprotic solvents
are difficult to remove from prepreg during processing because of their high boiling points
and tendency to enter into complex formation. Moreover, complex formation can interfere
with imidization, decreasing polymer thermo-oxidative stability.
6
During composite
fabrication, continued chain extension and solvent volatilization increase the viscosity of
precursor solutions. Meanwhile, the imidization of monomers or polyamic acids produces
alcohol, water or both as by-products. The increasing viscosity makes volatile by-products
and residual solvents more difficult to remove, resulting in high porosity (> 5 vol%) in
cured composites.
6
For example, void contents of Pyralin and Skybond polyimide
composites are typically > 10 vol%.
Thermoset polyimide composites can also be fabricated via a high-boiling point
aprotic solution approach whereby prepreg is prepared by coating amide acid or imide
oligomer solutions on dry carbon fibers.
36, 37
Similar to condensation-type composites,
thermoset composites are produced by heating to evaporate solvents and initiate
cyclodehydration (if prepreg is based on amide acid oligomer solutions) and cross-linking.
Because high-boiling point solvents are used in this approach, issues associated with
solvent removal from prepreg as described above are also encountered in composite
fabrication.
PMR is an alternative approach of processing thermoset polyimide composites. The
precursor in PMR-type prepreg is a solution of monomer mixture in which diester-diacid
derivatives of dianhydrides, diamines and ester-acid derivatives of end caps are dissolved
in organic solvents. Because monomers are small molecules and hence soluble in low-
12
boiling point solvents, e.g., alkyl alcohols, which are much easier to remove than NMP
(boiling point = 202°C). The molding cycle of PMR-type prepreg typically consists of three
major steps – removal of solvent, imidization and cross-linking. Figure 1-7 shows a typical
autoclave cure cycle for PMR-15 laminates.
3
The first isothermal hold at room temperature
under vacuum is designed to remove solvent (methanol in PMR-15 prepreg). Subsequently,
the temperature is increased to 202°C to initiate the in situ imidization of monomers while
full vacuum is utilized to remove volatile by-products and residual solvent. After
monomers are converted to imide oligomers, the temperature is increased again to trigger
cross-linking of endcaps at 316°C. Because the melt viscosity of fully imidized oligomers
is high, autoclave pressure is applied prior to cross-linking and maintained during cure to
facilitate the consolidation of laminates, yielding void-free high-quality composites.
Figure 1-7. Typical autoclave cure cycle for PMR-15 composites.
13
1.3.3 Environmental effects on polyimide composites
Polyimide composite are used for aerospace structural applications that are exposed
to prolonged, extreme service conditions in military and commercial aircraft and
hypersonic reusable space vehicles, including stress, time, temperature, moisture, chemical
and gaseous environments.
38
For example, high speed commercial transport aircraft require
materials to be able to withstand 120,000 h at ~177°C.
39
At high temperatures, polyimide
matrices are susceptible to oxidization, resulting in weight loss and mechanical degradation.
For instance, 620-h exposure at 316°C caused > 8 wt% loss and ~70% reduction in flexural
strength of T300/PMR-15 laminates.
40
Moreover, thermal oxidation increases the
brittleness of matrices and consequently produces microcracks,
40
adversely accelerating
oxidation by inducing fast oxygen diffusion channels.
In addition to thermal oxidation, thermal cycling is also a factor that can cause
microcracking in polyimide composites. Carbon fibers are dimensionally stable materials,
while polyimides have much greater thermal expansion coefficients (CTE), resulting in
high thermal shrinkage mismatch between matrix and fibers upon cooling. This volume
mismatch creates thermal stress and hence initiates formation of microcracks when the
thermal stress exceeds material strength. Because of the high-temperature manufacturing
and large range of service temperatures, nearly all polyimide composite components
encounter microcracking in thermal cycling tests.
Polyimides are hydrophilic polymers and generally show a maximum moisture
content of 3-5 wt%. The absorption of moisture by polyimide composites causes reduction
in Tg because of the plasticizing effect of moisture on the matrix and consequently
14
adversely affects the strength and elastic properties of the composites at elevated
temperatures.
40
On the other hand, imidization is a reversible reaction, so high
temperate/humidity exposure can drive the reaction (Figure 1-2) in reverse, resulting in the
hydrolysis of imide units in cured polyimides. Moreover, moisture can produce permanent
structural damage to composites, such as microcracks, microvoids and delamination.
1.4 Scope of dissertation
Polyimides are essential to the development of high-temperature composites for
aerospace applications, yet conventional polyimides exhibit critical limitations in
processing and performance aspects of composites. Previous study has demonstrated some
promising properties of TriA X, such as amorphous structure and sufficient solubility of
imide oligomers in low-boiling point solvents. However, no comprehensive research has
been conducted on TriA X system to explore the potential of this polyimide as a high-
temperature composite matrix.
This dissertation presents a fundamental investigation on TriA X system to support
development of a new polyimide composite and address current issues associated with
conventional polyimide matrices. To obtain an accurate understanding of this new imide
resin system, structure-processing-property relationships were determined at a molecular
level, and a material paradigm is presented in Chapter 2. A processing method was
developed to fabricate high quality composites based on TriA X, and the processability of
the imide resin system is described in Chapter 3. To explore the potential of TriA X
composites for extreme service environments, the stability of the neat polyimide and
15
associated composites upon exposure to cyclic temperatures and moisture was investigated
and is described in Chapter 4 and 5.
16
Chapter 2. Structure and properties of TriA X system
2.1 Introduction
TriA X system is investigated to determine polymer structure, processability, and
thermal and mechanical properties. This polyimide exhibits unusual characteristics that
potentially overcome some of the major limitations of conventional polyimides, and thus
may serve to expand service conditions and applications of both neat polyimides and
associated composites. The goal of this work is to determine the relationship between
molecular level structure and macroscopic properties, and thereby reach a better
understanding of the behavior and characteristics of this new polyimide.
In this dissertation, a TriA X polyimide system (Figure 2-1a) with a 3942 g mol
-1
imide oligomer (n = 7) was selected. A distinguishing feature of TriA X is the asymmetric
backbone, derived from the asymmetric monomer, p-ODA. This distinctive feature results
in a disordered chain packing arrangement which profoundly affects the behaviour of the
cured polymer. For example, the loose chain packing imparts high chain mobility above
the Tg, resulting in a low melt viscosity. Furthermore, the pendent phenyl group in the p-
ODA moiety suppresses internal rotation around the biphenyl ether linkage and thus
increases backbone rigidity, resulting in a Tg = 362°C and high material stiffness. To
demonstrate the attributes and properties of TriA X that derive from the asymmetric
backbone, a control material consisting of a symmetric polyimide (Figure 2-1b) was
prepared from PMDA, 4,4’-ODA and PEPA with n = 7 for characterization and comparison
to TriA X.
17
Figure 2-1. (a) TriA X and (b) PMDA/4,4’-ODA/PEPA polyimide (control).
In this work, we will show that the asymmetric/irregular/nonplanar backbone of the
polyimide molecule disrupts the dense chain stacking by decreasing interchain interactions
and thereby increasing intersegmental distances, resulting in a loose-packed, amorphous
structure. The loose chain packing imparts greater processability relative to the control
polyimide, and the cured polymer shows superior thermal stability, ductility, strength, and
toughness, compared to PMR-15 and AFR-PE-4.
2.2 Materials
2.2.1 PMDA di-ester/p-ODA/PEPA mono-ester monomers
A powder blend of PMDA di-ester/p-ODA/PEPA mono-ester (Kaneka, College
Station, Texas, USA) was obtained and used in this study. Briefly, PMDA and PEPA were
esterified in ethanol, based on the method reported by Houlihan et al..
41
After PMDA and
18
PEPA were converted to PMDA di-ester and PEPA mono-ester, p-ODA was added to the
solution with ethanol and heated while stirring, yielding a PMDA di-ester/p-ODA/PEPA
mono-ester solution. The stoichiometric ratios were PMDA di-ester: p-ODA: PEPA mono-
ester = 7: 8: 2, forming an oligomer with a molecular weight of 3945 g mol
-1
via heat-
triggered imidization. Monomer powder was obtained by drying monomer solution at 50°C
in vacuum. The monomer was stored at room temperature prior to use.
2.2.2 PMDA/4,4’-ODA/PEPA imide oligomer
PMDA/4,4’-ODA/PEPA imide oligomer was prepared by a two-step synthesis
method, in which 4,4’-ODA was dissolved and stirred in DMAc at room temperature for
15 min under nitrogen. A stoichiometric amount of PMDA was added to the DMAc
solution, and after stirring at room temperature under nitrogen for 1 h, a stoichiometric
amount of PEPA was added to the solution, and the solution was stirred at room
temperature under nitrogen for 12 h. The stoichiometric ratios were PMDA: 4,4’-ODA:
PEPA = 7: 8: 2, forming an amide acid oligomer with a degree of polymerization n = 7.
The amide acid oligomer solution was coated onto a glass substrate and thermally imidized
at 250°C for 2 h, then at 300°C for 2 min. The imide oligomer was ground using mortar
and pestle, and the powder was stored at room temperature before use.
For clarity, the PMDA/4,4’-ODA/PEPA imide oligomer henceforth will be referred
to as the control imide oligomer, while the polyimide derived from PMDA/4,4’-
ODA/PEPA imide oligomer will be referred as the control polyimide (meaning control
cured imide resin).
19
2.2.3 TriA X polyimide panels
TriA X polyimide panels were produced from TriA X imide oligomer via
compression molding. First, PMDA di-ester/p-ODA/PEPA mono-ester monomer blend
was imidized at 250°C for 2 h in an oven to obtain TriA X imide oligomer (Figure 2-2a).
The monomer blend powder melted at 130°C,
42
and hence a 30-min dwell at 130°C was
added before imidization at 250°C to allow oligomer chains to grow. The imide oligomer
was subsequently ground using an electric grinder and dried at 80°C for 1 h under vacuum.
Figure 2-2. Molding cycle for TriA X resin: (a) oven cycle and (b) hot press cycle.
The imide oligomer powder was placed in a stainless steel frame coated with release
agent on a steel sheet. Both sides of the frame were covered by polyimide release film. The
oligomer powder was cured in a hot press (Figure 2-2b). The oligomer powder was heated
to 325°C at a ramp rate of 3°C min
-1
and equilibrated at 325°C for 5 min to even the
temperature distribution. At the end of the 325°C dwell, a two-minute bump sequence (20
bumps) was performed to evacuate volatiles, and the consolidation pressure (1.72 MPa)
20
was subsequently applied. The panels were heated to 371°C at a rate of 3°C min
-1
, then
cured at 371°C for 2 h. After cure, the panels were cooled to room temperature under
pressure at a rate of 5°C min
-1
.
2.3 Experiments
2.3.1 X-ray diffraction
To investigate material structures, wide-angle X-ray diffraction (XRD)
measurements of the oligomers and polyimides of TriA X and control resin systems were
conducted using an X-ray diffractometer (Ultima IV, Rigaku, Japan) with Cu-Kα incident
radiation (𝜆 = 0.1541 nm). To determine the dependence of the material structure of TriA
X on thermal history, a series of TriA X imide oligomers and polyimides were prepared,
as described in Table 2-1. The thermal processes encompass two types of conditions: (a)
those that favor crystallization at relatively low temperatures (200˚C for imidization and
350˚C for cross-linking) and slow cooling rates (1˚C min
-1
), and (b) those that do not, and
feature high temperatures (250˚C for imidization and 371˚C for cross-linking) and rapid
cooling rates (quench).
43
The control polyimide was prepared by heating the control imide
oligomer at 371°C for 2 h. XRD scans were recorded at 40 kV and 44 mA at a scan speed
of 3° min
-1
. No attempt was made to ensure a constant mass of samples in the X-ray beam,
so absolute intensity comparisons of the patterns from different samples was not warranted.
The XRD patterns were normalized based on baselines.
21
Table 2-1. TriA X imide oligomers and polyimides.
Specimen name Thermal history Cooling rate
Degree of
imidization
(%)
Tg (°C)
Oligomer 1 200˚C/30 min Liquid N2 quench 89.6 225
Oligomer 2 200˚C/30 min 1˚C min
-1
86.3 228
Oligomer 3 250˚C/30 min Liquid N2 quench 93.6 246
Oligomer 4 250˚C/30 min 1˚C min
-1
94.6 245
Oligomer 5 200˚C/2 h Liquid N2 quench 91.4 230
Oligomer 6 200˚C/2 h 1˚C min
-1
89.9 228
Oligomer 7 250˚C/2 h Liquid N2 quench 94.7 247
Oligomer 8 250˚C/2 h 1˚C min
-1
94.8 247
Polyimide 1 250˚C/2 h+350˚C/2 h Ice water quench - 342
Polyimide 2 250˚C/2 h+350˚C/2 h 1˚C min
-1
- 338
Polyimide 3 250˚C/2 h+371˚C/2 h Ice water quench - 360
Polyimide 4 250˚C/2 h+371˚C/2 h 1˚C min
-1
- 362
22
2.3.2 Thermogravimetric analysis
Thermogravimetric analysis (TGA) of TriA X was performed under nitrogen purge
at a flow rate of 5.00 mL min
-1
(Q5000, TA Instruments, USA). TGA data was obtained to
determine degrees of imidization of the oligomers in Table 2-1, as well as the
decomposition temperature (Td) of TriA X cured polyimide. To remove residual solvent
and moisture, the monomer blend and imide oligomers were dried at 60°C for 1 h and
subsequently at 80°C for 1 h in the TGA, prior to each experiment. For TGA experiments,
the monomer blend and imide oligomers were heated from 80°C to 400°C at a rate of 5°C
min
-1
. The monomers were assumed to be fully converted to imide oligomer at 400°C (no
degassing was observed during TGA from 25°C to 400°C after the monomers were heated
to 400°C). The degrees of imidization of the imide oligomers were determined by Equation
(2-1), where WO is the percent weight loss of each oligomer, and WM is the percent weight
loss of the monomer blend, as described by Omote et al.
44
Degree of imidization = (1 −
𝑊 𝑂 𝑊 𝑀 ) × 100% (2-1)
A TriA X polyimide was prepared in the TGA by holding the monomers at 250°C
for 2 h and subsequently at 371°C for 2 h. The polyimide was heated from 25°C to 650°C
at a rate of 10°C min
-1
. The Td of TriA X was determined by the onset of the decomposition
reaction, calculated by the intersection of lines tangent to the knee of the weight loss curve,
using thermal analysis software (Universal Analysis 2000, TA Instruments, USA).
23
2.3.3 Differential scanning calorimetry
The values for Tg and the possibility of phase transitions in TriA X oligomers and
polyimides were investigated using differential scanning calorimetry (DSC; Q5000, TA
Instruments, USA). Tests were performed by heating samples from 25°C to 490°C at 10°C
min
-1
under nitrogen purge at a flow rate of 50 mL min
-1
.
2.3.4 Rheological measurements
Rheological measurements were performed on TriA X and the control imide
oligomers using a parallel plate rheometer (AR 2000EX, TA Instruments, USA). Oligomer
8 of TriA X, which had a similar thermal history to the control imide oligomer, was selected
for rheological measurements. The oligomers were ground and pressed at room
temperature into 25 mm discs, 0.9-1.1 mm thick, using a hand press (YLJ-12T, MTI, USA).
Pressed oligomer discs were placed on the stationary lower plate of the rheometer,
while the upper plate was oscillated at a frequency of 1.0 Hz, and at a strain of 0.1%. The
measurements were performed at a heating rate of 3°C min
-1
from 225°C to 450°C. The
plate gap was adjusted under a normal force control of 1 N to maintain contact with the
samples. Complex viscosity (|η*|) was determined from rheological experiments.
2.3.5 Mechanical tests
The tensile and flexural properties of TriA X polyimide were measured in
accordance with ASTM D638
45
and ASTM D6272
46
standards, respectively. The tests were
24
conducted using a load frame (5567, Instron, USA) with a 5000-N load cell (2525-805,
Instron, USA). Test specimens were stored in a desiccator prior to testing.
Tensile tests were performed at a displacement rate of 1.0 mm min
-1
at -54°C, 23°C,
232°C and 288°C in an environmental chamber (3119-506, Instron, USA). The test
temperatures were monitored by a thermocouple attached to the clamps, adjacent to test
specimens. The strain at 23°C, 232°C and 288°C were measured using a 3D optical
deformation measuring system (ARAMIS; Adjustable Base 2.3M, GOM, Germany).
Speckle patterns were painted on the gauge sections of the specimens prior to the tests. As
the specimens deformed, the measuring system tracked the speckle pattern evolution to
determine dimension change. Due to moisture condensation on specimen surfaces, the stain
at -54°C was measured using an extensometer (2630-121, Instron, USA). Flexural tests
were performed at a strain rate of 0.1 mm mm
-1
min
-1
at 23°C with a load span of 25.5 mm
and a support span of 50.7 mm. Beam deflection was measured using a deflectometer
(2601-093, Instron, USA).
2.4 Results and discussion
2.4.1 Material structure
XRD was performed on TriA X and control samples to characterize the material
structures. The XRD patterns are shown in Figure 2-3, with origins shifted vertically for
clarity. No sharp peaks were observed in the XRD patterns from the eight oligomers
samples or the four polyimides. The twelve samples, prepared according to the thermal
processes listed in Table 2-1, exhibited nearly identical X-ray scattering behavior, each
25
showing broad halos at 2𝜃 values of 6.0~6.2° and 18.0~18.3°. The absence of sharp peaks
indicated amorphous structures in all TriA X oligomer and polyimide samples. In contrast,
the XRD patterns from the control imide oligomer and polyimide displayed multiple sharp
diffraction peaks at 2𝜃 values of 6°, 15°, 22°, and 26° and a scattering halo at ~20°,
indicating semi-crystalline structures.
Figure 2-3. XRD patterns of TriA X and control
The X-ray scattering patterns from the TriA X imide oligomers and polyimides
exhibited two prominent broad halos. Using Bragg’s law,
𝜆 = 2𝑑 sin𝜃 (2-2)
26
where 𝜆 is the X-ray wavelength, 𝜃 is the angle at which the intensity maximum occurs,
and d is the Bragg spacing, the Bragg spacings of the two halos were 0.5 nm and 1.4 nm
(at 2𝜃 = 6.0~6.2° and at 2𝜃 = 18.0~18.3°, respectively). The former spacing corresponds
to a van der Waals (VDW) distance of 0.4~0.5 nm, and the latter one indicates an interchain
distance larger than van der Waals (LVDW).
47
Although the intensity of the LVDW halo
was less than that of the VDW halo, the presence of the LVDW halo indicated that a portion
of the chains manifest intermolecular distances greater than VDW bonds, and overall the
polymers exhibited looser packing than VDW bonding. The amorphous structures with
LVDW packing strongly affected the behavior of TriA X, as described in sections 2.4.2,
2.4.3 and 2.4.4.
DSC curves from the TriA X imide oligomers or polyimides showed no evidence
of crystallization or melting (Figure 2-4), and only glass transition and cross-linking were
observed. The cross-linking reaction peaks of the oligomers occurred at ~390°C, and the
Tg’s of the oligomers and polyimides are listed in Table 2-1. The thermal properties of TriA
X are described further in Section 2.4.3. The DSC results reaffirmed the absence of
crystalline regions, leading to the conclusion that the TriA X imide oligomers and
polyimides were entirely amorphous (non-crystalline). Moreover, the eight imide
oligomers and four polyimides were prepared by different thermal processes, as described
in Section 2.3.1. These processes included conditions favorable to crystallization
(relatively low temperatures and slow cooling rates), as well as conditions that did not favor
crystallization (high temperatures and rapid cooling rates).
43
Thus, the formation of the
amorphous structure in the oligomer and polyimide samples was independent of thermal
process history.
27
Figure 2-4. DSC curves of TriA X imide oligomers and polyimides.
Generally, semi-crystalline imide oligomers or polyimides are undesirable for
processing of polyimide matrix composites − high temperatures are needed to melt
crystalline regions, and this results in narrow processing windows.
48
Moreover,
crystallization inhibits cross-linking, generating defects in the form of unreacted groups in
remaining crystallites if crystal phases are not completely eliminated before or during
cross-linking. Such defects yield reduced material toughness.
23, 49
Although some semi-
crystalline polyimides can be converted to amorphous structures by controlling process
parameters,
23, 49
for large-scale manufacturing of composite components, thermal gradients,
particularly in large parts, are unavoidable and can yield uneven polymer structures. Thus,
28
amorphous imide oligomers and polyimides are preferred over semi-crystalline ones,
because there is no need for high temperature processing to melt crystalline regions.
48, 49
The ability to form completely amorphous structures in TriA X is attributed to the
asymmetric backbones. In polymers, symmetric backbones and strong intermolecular
interactions favor crystallization, because symmetric chains permit the regular, close
packing required for crystallite formation, while strong interchain attractions stabilize
orderly alignments.
43
Consequently, symmetric polyimides, such as AFR-PE-4
50
and
PETI-5
23
, tend to form semi-crystalline structures. We assert that the pendent phenyl group
on the ODA moiety increases backbone irregularity by forming three repeat-unit
configurations: head-to-head, head-to-tail, and tail-to-tail. In addition to the irregular
geometry, the nonplanar structure of p-ODA disrupts the closed-packed arrangements of
polymer chains and increases intersegmental distances, weakening interchain interactions.
The hypothesis that the asymmetric p-ODA affects the chain arrangements can be
tested by investigating the polymer structure that arises when the pendent phenyl groups
are absent from the polymer backbones. The control imide oligomer and polyimide had
formulated chemical structures identical to the TriA X imide oligomers and polyimides
(except for the extra phenyl groups of the diamines in TriA X system). As mentioned
previously, multiple sharp diffraction peaks were present in the XRD patterns of the control
imide oligomer and polyimide (Figure 2-3), indicating that semi-crystalline structures.
Because the pendent phenyl groups were absent from the control system, the symmetric
chains were able to adjust conformations and form orderly arrangements in response to
strong interchain forces. In contrast, the asymmetric/nonplanar p-ODA characteristic of the
TriA X backbone yielded structures that were entirely amorphous.
29
2.4.2 Thermal processability
Rheological measurements were performed on Oligomer 8 of TriA X and the
control imide oligomer (with similar thermal history) to investigate the effects of the
asymmetric structure of p-ODA on the processability of the TriA X system. The |η*|
(complex viscosity) profiles of TriA X imide oligomer and the control imide oligomer are
shown in Figure 2-5a. The control imide oligomer did not soften or melt upon heating up
to 450°C, displaying a constant viscosity of ~10
5
Pa⋅s. The |η*| of the TriA X imide
oligomer began to drop at 258°C, and reached a local minimum at 284°C, a characteristic
of residual imidization.
42
The minimum |η*| of TriA X imide oligomer (2487 Pa⋅s)
occurred at 358°C, which was much lower than the |η*| of the control imide oligomer. In
addition to the viscosity difference, the processability distinction between the two
oligomers was demonstrated by a change in appearance after heating. Both TriA X and the
control imide oligomers appeared as identical orange powders prior to rheological
measurements (Figure 2-5b and 2-5c top). After the measurements, the TriA X imide
oligomer powder deformed and fused to the metal plates, turning darker in color. In
contrast, the control imide oligomer remained as powder and did not fuse to the rheometer
plates. The |η*| of the TriA X imide oligomer was low enough for composite compression
molding under 1.72 MPa,
42
while the control imide oligomer was not processable. This
feature of the TriA X imide oligomer was attributed to the extra phenyl group in the ODA
moiety.
30
Figure 2-5. (a) Complex viscosity |η*| profiles of TriA X imide oligomer (red solid) and
the control imide oligomer (blue dash) and images of TriA X imide oligomer (b) and
control imide oligomer (c) before (top) and after (bottom) measurements.
Because of the asymmetric backbones, TriA X imide oligomers were amorphous
(non-crystalline), and polymer chains were loosely packed, allowing space for
rearrangement above Tg. In contrast, the chain arrangements in the control imide oligomer
were more orderly. The strong interchain attractions held polymer chains so tightly that
even at 450°C, the thermal energy was insufficient for chain movement. Consequently, the
TriA X imide oligomer exhibited much lower melt viscosity than the control imide
oligomer.
2.4.3 Thermal properties
The Tg’s and Td’s of TriA X, PMR-15,
21, 51
and AFR-PE-4
19, 50
are shown in Figure
2-6. The Tg of TriA X was 362°C, 21°C above that of PMR-15, and 15°C above that of
AFR-PE-4, while the Td of TriA X was 130°C above that of PMR-15, and ~30°C above
31
that of AFR-PE-4. Note that all values were determined using TGA at the same heating
rate (10°C min
-1
). Also, the Tg value of TriA X is consistent with the DSC measurements
shown previously in Fig. 4 (TriA X is Polyimide 4). The Tg of the control polyimide was
not detected by DSC because of the high degree of crystallinity.
Figure 2-6. Tg’s and Td’s* of TriA X, PMR-15 and AFR-PE-4.
*All Td’s were measured using TGA at a heating rate of 10°C min
-1
.
The Tg of a polymer is affected by multiple factors. Decreases in intermolecular
interaction or increases in interchain spacing both cause a reduction in Tg, while
suppression of chain rotation increases chain stiffness and thus increases Tg.
43, 47
Although
the increased interchain distances resulting from the asymmetric/nonplanar structure of p-
ODA, tend to decrease the Tg of TriA X, the pendent phenyl group hinders the rotation of
the diphenyl ether linkage in the ODA moiety, raising the internal rotation energy barrier
32
of the backbone. Thus, the enhanced chain stiffness partially counterbalances the adverse
influence of loose chain packing, resulting in a Tg of 362°C.
The superior thermal stability of TriA X compared to PMR-15 and AFR-PE-4 can
be understood in terms of molecular structure. In particular, PMR-15 consists of (a) 4,4’-
methylenedianiline (MDA), (b) 4,4’-benzophenonetetracarboxylic dianhydride (BTDA)
and (c) nadic anhydride (NA; end-cap). In contrast, AFR-PE-4 is derived from (a) p-
phenylene diamine (PPD), (b) 4,4’-hexafluoroisopropylidene diphthalic anhydride (6FDA)
and (c) PEPA (end-cap). In general, PEPA-terminated polyimides exhibit greater thermo-
oxidative stability than NA-terminated polyimides.
52
Moreover, the methylene linkage in
MDA reportedly is thermally less stable than the ether bond in ODA or p-ODA, and
consequently, ODA-PMDA polyimides exhibit greater thermal stability in air and inert
atmosphere than MDA-PMDA polyimides.
53
In addition, the fluorinated methyl group of
6FDA can be easily dissociated by the thermal cleavage of C-CF3 bonds at high
temperatures, and thus the thermal stability of linear aromatic polyimides derived from
6FDA is less than that of PMDA-based polyimides when the diamines are identical.
54-57
Thus, the high Td of TriA X derives from the combined effects of the backbone monomers
and the PEPA end-cap.
2.4.4 Mechanical properties
Mechanical tests were performed on TriA X polyimide at select temperatures, as
given in Table 2-2. The stress-strain curves of TriA X in the tensile tests are shown in
Figure 2-7, and the mechanical property values appear in Table 2-3. After an initial period
of linear elastic deformation, all curves exhibited nonlinear elastic deformation prior to the
33
yield points (zero slope in stress-strain curve). At -54°C, TriA X failed at 9.7% strain prior
to reaching a yield point. Likewise, at 23°C, the polymer broke before reaching a true yield
point. However, the stress-strain curve exhibited a large nonlinear region, and the rupture
point (at 15.1% strain) displayed a near-zero slope (0.05 GPa). Thus, the nonlinear region
at 23°C is believed to contain both elastic and plastic deformation.
Higher test temperatures resulted in lower yield strengths, but broad stress plateaus
and much greater ductility. At 232°C and 288°C, TriA X yielded at 7.6% and 3.6% strain,
with yield strengths of 39.5 MPa and 25.9 MPa, respectively. Beyond the yield points,
stress first dropped slightly, then rose slowly with increasing strain, producing a broad
stress plateau. The stress drop at the yield point at 288°C was more apparent than that at
232°C. Before rupture, stress increased slightly due to alignment of polymer chain
segments. At 288°C, the speckle pattern started to detach from specimens at ~37% strain,
prior to failure. Strain beyond 37% (red dashed line in Figure 2-7) at 288°C was estimated
based on the displacement of the grips, recorded by the load cell. Overall, as test
temperatures increased, the tensile strength and Young’s modulus values decreased, while
ductility increased.
Generally, highly cross-linked thermosets are brittle, displaying only linear elastic
deformation in tensile tests, while the typical tensile stress-strain curve of thermoplastics
exhibits linear and nonlinear elastic deformation, yielding, chain strengthening and large
strain-at-failure. Although TriA X is a thermoset polymer, at high temperatures, the
polymer exhibited thermoplastic-like mechanical behavior, with post-yield stress drops and
extensive ductility.
34
Table 2-2. Specimens for mechanical tests.
Mechanical
test
Test
method
Temperature
(°C)
Length
a
(mm)
Width
b
(mm)
Thickness
c
(mm)
Number of
specimens
Tensile
ASTM
D638
45
-54 65.92 3.18 3.08 10
23 65.96 3.16 3.13 6
232 66.03 3.17 3.11 5
288 66.0 3.16 3.14 5
Flexural
ASTM
D6272
46
23 60.76 12.7 3.15 6
a
Total length of specimens.
b,c
Gauge section dimensions.
Figure 2-7. Stress-strain curves of TriA X tensile tests.
35
Table 2-3. Mechanical properties of TriA X.
Temperature
(˚C)
Tensile
modulus (GPa)
Tensile
strength (MPa)
Strain-at-
failure (%)
Flexural
modulus (GPa)
-54 3.94 ± 0.33 153.3 ± 5.1 9.7 ± 2.1 -
23 3.57 ± 0.04 119.0 ± 1.1 15.1 ± 1.6 3.88 ± 0.05
232 1.97 ± 0.04 47.0 ± 1.8 36.5 ± 2.1 -
288 1.54 ± 0.04 36.9 ± 1.0 > 37* -
*Speckle pattern painting detached from specimens when strain was greater than 37% so
that the measured strain above 37% was not accurate.
In flexural tests, specimens did not yield or break within the 5% strain limit
specified by the test standard.
46
Figure 2-8a shows a beam specimen during a bending test
at a deflection corresponding to ~10% strain. After unloading, the specimen recovered
elastically to a flat beam (Figure 2-8b). The exceptional elastic flexibility was notable,
particularly since most polyimides are brittle.
Figure 2-8. TriA X specimen of flexural test: (a) in test and (b) after test.
36
A comparison of mechanical properties of TriA X, PMR-15,
58-60
and AFR-PE-4
19
is shown in Figure 2-9. TriA X exhibited an elastic modulus similar to PMR-15 and AFR-
PE-4, but with greater toughness. Toughness - the energy a material absorbs during load
prior to failure, can be determined simply by the area under the tensile strain-stress curve.
43
When the elastic moduli of two materials are similar, greater tensile strength and/or strain-
at-failure indicate greater toughness. At 23°C, the ductility of TriA X was ten times of that
of PMR-15 (1.5%)
59, 60
and six times of that of AFR-PE-4 (2.5%),
19
while the tensile
strength was three times that of PMR-15 (37.2 MPa)
58
and 65% greater than that of AFR-
PE-4 (72.0 MPa).
19
At 288°C TriA X exhibited tensile strength 21% greater than PMR-15
(30.5 MPa).
58
The formulated molecular weight of the imide oligomer (3942 g mol
-1
) was greater
than that of PMR-15 (1500 g mol
-1
)
10
and AFR-PE-4 (2634 g mol
-1
).
50
Imide oligomers
with greater molecular weight generally yield reduced cross-link density in the cured
polyimides, resulting in lower stiffness and greater toughness.
48
Indeed, the high molecular
weight oligomer increased the tensile ductility and strength, although the elastic modulus
of the cured polyimide was virtually identical to that of PMR-15 and AFR-PE-4. This
peculiar behavior is attributed to the high backbone rigidity of TriA X. As discussed above,
the pendent phenyl group raises the chain stiffness by hindering the rotation of diphenyl
ether linkage, and in general, stiffer backbones exhibit greater elastic moduli.
43
37
Figure 2-9. Mechanical properties of TriA X (red), PMR-15 (blue) and AFR-PE-4
(yellow)*.
*Only tensile properties at 23°C of AFR-PE-4 are displayed.
The high toughness observed can be attributed not only to the high molecular
weight oligomer, but also the amorphous and loose-packed structure. The segmental chains
between cross-link sites were randomly coiled together, resulting in high chain mobility
upon stretching. Indeed, the uncoiling of the polymer chains was responsible for the
ductility observed at 232°C and 288°C. Beyond the yield points, polymer chains were able
to slip over one other, producing the extended stress plateaus shown in Figure 2-7. If the
polymer chains were straight rather than coiled, few intertwined segments would be
stretched, and negligible plastic deformation would occur in tensile tests.
38
TriA X exhibited a tensile strength of 153.3 MPa and a strain-at-failure of nearly
10%, even at -54°C. This behavior has important implications for polyimide composites,
which are notoriously susceptible to cracking when cycled to low temperatures.
61
A
polyimide matrix with superior low-temperature properties might resist microcracking
upon exposure to cyclic temperature.
2.5 Conclusions
A materials paradigm was presented for an asymmetric polyimide system,
demonstrating the effects of a pendent phenyl group on the polymer structure and
properties at the molecular level. The key findings included: (1) the asymmetric/nonplanar
backbone yielded amorphous structures with packing looser than VDW bonding in both
imide oligomers and polyimides, independent of thermal history; (2) the disordered and
loose molecular arrangement afforded more space for polymer chains to reconfigure above
Tg, resulting in a low melt viscosity and hence improved processability; (3) the pendent
phenyl group in the p-ODA moiety in the backbone hindered the rotation of diphenyl either
linkage, leading to a Tg = 362°C; and (4) the high molecular weight oligomer and loose-
packed amorphous structure of the TriA X polyimide resulted in high toughness, while the
stiff backbone maintained an elastic modulus comparable to conventional polyimides
(PMR-15 and AFR-PE-4). The materials paradigm presented here demonstrates an
approach to eliminating crystallinity in polyimides without sacrificing other desirable
properties. The results provide guidance for the design and development of new high-
performance polymers, especially for the scenarios of diminishing the crystallinity of semi-
crystalline polymers.
39
The exceptional properties of the polyimide reported here have potential to expand
the design space for high-temperature composites and thus lead to new applications for
polyimide composites. The thermal and mechanical properties of the new polyimide
surpassed those of conventional polyimides (PMR-15 and AFR-PE-4), which have been
used effectively in high-temperature composite applications for decades. Composites
fabricated from TriA X should open applications that require greater thermal stability or
matrix-dominated mechanical properties than offered by present polyimides. Moreover,
the unusual flexibility and toughness of TriA X may lead to high-performance applications
for the neat polymer, including adhesives, coatings, and films.
40
Chapter 3. Process development for TriA X composites
3.1 Introduction
In 1972, NASA Lewis Research Center developed the PMR approach to processing
polyimides.
10
This technique improved the processability of polyimides compared to prior
generation resin systems.
3
TriA X is a PMR-type polyimide. Chapter 2 shows that the cured
TriA X resin exhibits a high Tg, thermal stability, and high toughness and flexibility
because of the asymmetric/non-planar unit, p-ODA. Moreover, both the imide oligomer
and cured resin have amorphous structures. The properties and characteristics observed in
the resin system make TriA X suitable as a potential matrix material for high-temperature
structural composites. The manufacturing of such composites, however, poses challenges.
The processing of polyimide composites is difficult not only because of the high
melt viscosity of the resin, but also because of the complex chemical reactions that occur
during molding. Imidization itself involves the production of alcohol and water as reaction
by-products. To obtain void-free parts, these volatile by-products must be removed prior
to consolidation and cross-linking. In this work, we present a polymer science-based
approach for processing polyimide composites, which is applicable to any PMR-type
polyimide resin. To develop this composite molding cycle, chemo-rheology was performed
on the neat resin to understand processing characteristics, as well as the effects of B-stage
temperature on imidization and cross-linking. Based on the rheological properties of TriA
X, a molding cycle for T650-35 8-harness satin (8HS)/TriA X composites was designed
and applied to the fabrication of laminates. The laminates produced showed low void
41
contents (< 0.1%) and matrix-dominated mechanical properties superior to those of PMR-
15 composites.
3.2 Materials
3.2.1 PMDA di-ester/p-ODA/PEPA mono-ester monomers
A powder blend of PMDA di-ester/p-ODA/PEPA mono-ester (Kaneka, Texas,
USA) was utilized in this study. The preparation and storage of the monomer powder blend
are described in Section 2.2.1.
3.2.2 T650-35 8HS/TriA X prepreg
Prepreg was fabricated using an 80wt% PMDA di-ester/p-ODA/PEPA mono-ester
solution in ethanol, which was combined with de-sized T650-35/8HS/3K carbon fabric
(areal weight 368 g m
-2
, Cytec Solvay Group). The prepreg was prepared by calendering
the monomer solution on both sides of the fabric. After fabrication, the prepreg was stored
in sealed bags at -22°C prior to use.
3.3 Experiments
3.3.1 Thermogravimetric analysis
TGA of the monomer powder blend was performed under nitrogen purge at a flow
rate of 5.00 mL min
-1
(TA Instruments, Q5000). TGA data was obtained to investigate
degree of imidization at different B-staging temperatures. Three candidate B-staging
temperatures were selected (220°C, 250°C and 280°C) based on the temperature control
42
accuracy of the heated platens used for laminate fabrication. A 4-h degassing step was
employed during B-staging to remove volatiles and reaction byproducts from the prepreg
stack. 4 hours was selected as a suitable time because of the low gas permeability of the
8HS prepreg.
Monomer powders were dried at 80°C for 1 h prior to each experiment to remove
residual solvent and moisture. The monomers were then imidized for 4 h at 220°C, 250°C
or 280°C. Following the 4-h imidization, samples were heated from the imidization
temperature to 400°C at a rate of 5°C min
-1
. Data acquisition was initiated after the 1-h
drying step at 80°C, and continued until completion of the cycle at 400°C. All monomers
were assumed to be fully converted to imide oligomer at 400° C. The degree of imidization
under each thermal condition was determined as the ratio of the percent weight loss at the
end of isothermal B-staging (WB) and the percent weight loss at the end of the ramp to
400°C (W0), as described in Section 2.3.2.
3.3.2 Rheological measurements
Rheological measurements were performed using a parallel-plate rheometer (TA
Instruments, AR 2000EX). To reduce the influence of bubbling during imidization, the
monomer powder blend was heated to 200°C at a rate of 8.3°C min
-1
, then quenched to
room temperature to obtain an oligomer (Oligomer 0) with a low degree of imidization
(85.5%, determined by TGA). Oligomer 0 was ground and pressed at room temperature
into 25 mm discs, 0.9-1.1 mm thick, using a hand press (MTI, YLJ-12T).
Pressed oligomer discs were placed on the stationary lower plate of the rheometer,
while the upper plate was oscillated at a frequency of 1.0 Hz, and a strain of 0.1%. The
43
plate gap was adjusted under a normal force control of 1 N to maintain contact with the
sample. Complex viscosity (|η*|), storage modulus (G’), and loss modulus (G”) were
determined from rheological experiments.
3.3.3 Gel Permeation chromatography analysis
The molecular weight change of Oligomer 0 during heating was determined by gel
permeation chromatography (GPC) (Shimadzu, Prominence GPC system). Oligomer 0 was
heated in a rheometer at 1°C min
-1
from a starting temperature of 200°C to a final
temperature of 225°C or 250°C, to obtain Oligomer 0-1 and Oligomer 0-2, respectively.
The number average molecular weights (Mn) of Oligomer 0, Oligomer 0-1 and
Oligomer 0-2 were measured using a GPC system with two analytical columns
(Phenomenex, Phenogel 5 µm, 1×10
4
Å, 300 × 7.8 mm and Phenogel 5 µm, 100 Å, 300 ×
7.8 mm) and a UV detector (Shimadzu, SPD-20A). The system was calibrated using
polystyrene standards. Measurements were performed at 40°C using 1-methyl-2-
pyrrolidone as eluent at a flow rate of 1.0 mL min
-1
.
3.3.4 Differential scanning calorimetry
Cross-linking in imide oligomers B-staged at different temperatures was
investigated using DSC (TA Instruments, Q2000). Tests were performed under nitrogen
purge at a flow rate of 50 mL min
-1
. Monomer powder was held for 4 h at 220°C, 250°C
or 280°C, after which the resulting oligomers were cooled to 35°C, then heated to 500°C
at a rate of 10°C min
-1
. Heat of reaction was determined from the area under the cross-
linking reaction peak.
44
3.3.5 Dynamic mechanical analysis
Dynamic mechanical analysis (DMA) was performed to determine the Tg of cured
composite laminates (TA Instruments, Q800). The measurements were carried out in single
cantilever mode at a heating rate of 5°C min
-1
with a fixed frequency of 1 Hz and a strain
of 0.3%. Storage modulus, loss modulus, and tan 𝛿 were recorded to determine the Tg of
T650-35 8HS/TriA X composites (described below).
3.3.6 Composite laminate fabrication
Laminates (300×300 mm) were laid-up at room temperature. The stacking
sequences of the laminates produced for this work are shown in Table 3-1. Fiber volume
fraction (FVF) was determined based on the weight and density of the constituent
components. For FVF measurements a rectangular coupon was sectioned from each
laminate. The dimensions and weight of each coupon were measured by a micrometer and
a scale, respectively, and FVF was determined as follows.
FVF ( 𝑣𝑜𝑙 %)=
𝑛 ×𝐿 ×𝑏 ×𝜌 𝑓 ÷𝜌 𝑐 𝑛 ×𝐿 ×𝑏 ×𝜌 𝑓 ÷𝜌 𝑐 +( 𝑀 −𝑛 ×𝐿 ×𝑏 ×𝜌 𝑓 ) ÷𝜌 𝑟 × 100% (3-1)
where:
n = number of prepreg plies;
L = measured specimen length, m;
b = measured specimen width, m;
𝜌 f = areal density of T650-35 8HS fabric, 0.368 kg m
-2
;
45
𝜌 c = density of T650-35 carbon fiber, 1.76×10
3
kg m
-3
;
M = measured specimen weight, kg, and,
𝜌 r = density of TriA X cured neat resin, 1.29×10
3
kg m
-3
.
Table 3-1. Molded TriA X/T650-35 8HS laminates.
Laminate No. Lay-up sequence Thickness (mm) FVF (vol%)
1 [0/+45/-45/90]s 3.120 55.1
2 [0/90/-45/+45/0/90]s 4.607 55.6
3 [0/+45/-45/90]s 3.097 55.1
4 [0/+45/-45/90]2s 6.237 56.6
After layup, each prepreg stack was dried at 50°C for 24 h in a convection oven,
then vacuum bagged using a polyimide film. Bagged laminates were cured in a hot press.
Details of the cure cycle (oven and hot press) are described in Section 3.4.2.
3.3.7 Porosity measurement
Image analysis (IA) and X-ray micro-computed tomography (micro-CT) were
employed to investigate the porosity of manufactured laminates. Two samples, 20-25 mm
long, were cut from the center of each laminate for IA. The samples were mounted in
potting resin, polished, and imaged using a digital microscope (Keyence, VHX-600). Void
content was determined by the area fraction of voids (visible in the cross section) using
image analysis software (Adobe Photoshop and ImageJ).
46
For validation of microscopy results, high-resolution X-ray tomography was
performed on the thickest laminate, which can be expected to have the highest void content.
A section (20×20×6 mm) was removed from the center of a 16-ply sample with a stacking
sequence of [0/+45/-45/90]2s and scanned (Nikon, XT H 225ST). Mo-Kα incident radiation
(𝜆 = 0.71 Å) was used with 60 kV/220 𝜇 A voltage/intensity settings to achieve a resolution
of 14.596 𝜇 m per pixel. The X-ray attenuation of fiber and matrix is relatively high, while
the signal attenuation of voids (air) is low. Voids were recognized based on attenuation
differential, and the volume fraction of voids was determined using an image analysis
utility (Visual Studio Max).
3.3.8 Mechanical tests
The tensile, flexural, and SBS properties of T650-35 8HS/TriA X composites were
measured at room temperature in accordance with ASTM D3039,
62
ASTM D7264 (four-
point bending procedure),
63
and ASTM D2344
64
standards, respectively. This series of tests
yielded stiffness and strength properties of the composite when subjected to tensile,
bending and shear loads, thus providing holistic information about key mechanical
attributes. While tensile and flexural properties of composites depend on fiber properties,
SBS properties depend primarily on the properties of the matrix material.
Tensile tests were conducted using a high-capacity load frame (Instron, 5585H) at
a displacement rate of 2 mm min
-1
. The flexural and SBS tests were performed on a smaller
load frame (Instron, 5567) at a crosshead displacement rate of 1.0 mm min
-1
. Displacement
was measured using an extensometer (Instron, 2630-109) and a deflectometer (Instron,
2601-093) for tensile tests and bending tests, respectively.
47
3.4 Results and discussion
3.4.1 Processability
3.4.1.1 Chemo-rheological properties
During the processing of PMR-type polyimides, monomers first condense to form
amid acid oligomers, then cyclodehydrate, closing imide rings and forming imide
oligomers. Cyclodehydration occurs at relatively low temperatures (200-300°C), releasing
volatile by-products. Following this imidization reaction, the oligomer softens and then
cross-links at high temperatures (300-400°C), forming the final thermoset polyimide.
Figure 3-1 shows those chemical reactions using TriA X system as an example. Each step
of the cure involves viscosity changes, making viscosity measurements critical to
determination of suitable process parameters. Chemo-rheology of the resin system was
studied to identify relationships between chemical reactions, viscosity changes, and
temperature.
Figure 3-1. Chemical reactions in TriA X system.
48
The dynamic rheological behavior of an oligomer (Oligomer 0) with a low degree
of imidization (85.5%) was evaluated at a heating rate of 1°C min
-1
. To illustrate the
viscosity changes, dynamic rheological data from 200°C to 400°C is shown in Figure 3-
2a. Complex viscosity (|η*|), storage modulus (G’), and loss modulus (G”) are presented.
A viscoelastic fluid will exhibit solid-like or liquid-like behavior under different thermal
conditions. The storage modulus represents the elastic or solid-like component, while the
loss modulus represents the viscous or liquid-like component. When G’ > G”, the material
exhibits viscoelastic solid-like behavior. Conversely, when G’ < G”, the material behaves
as a viscoelastic liquid.
Figure 3-2. (a) Dynamic |η*|, (black solid), G’ (red dash) and G” (blue short dash)
profiles for Oligomer 0 at a ramp rate of 1°C min
-1
, and (b) Mn change of Oligomer 0
upon heating during rheological measurement.
During rheological measurements of TriA X, multiple transitions were observed.
Below 252°C, |η*| was high, and the resin was a solid (G’ > G”). The increase in modulus
observed as the temperature increased to 252°C is a result of both the increase in molecular
weight occurring during imidization (Figure 3-2b), and a gradual and continuous softening
49
of the oligomer, which can effectively enhance the interfacial connection between the
specimen and the rheometer plates.
65
When a resin behaves as a rigid solid, the parallel
plates in the rheometer can slip past the specimen, resulting in a low apparent viscosity.
When the oligomer begins to soften, adhesion to the plates increases, increasing frictional
contact. This behavior was observed in the TriA X system. At 252°C, the viscosity began
to drop, reaching a local minimum at 285°C. In this temperature range, the resin exhibited
liquid-like behavior (G’ < G”), indicating that the oligomer began to soften at 252°C. The
minimum in the |η*| curve at 285°C is attributed to residual imidization, as discussed
below.
TGA and rheological measurements were performed on Oligomer 0 to confirm that
residual imidization was responsible for the minimum |η*| observed at 285°C (Figure 3-
2a). A dynamic TGA measurement of Oligomer 0 at 1°C min
-1
revealed weight loss
between 161°C and 300°C (Figure 3-3a), consistent with off-gassing. This off-gassing
behavior indicates that imidization was taking place in this temperature range. If the drop
in viscosity was due to residual imidization, as hypothesized, the minimum |η*| observed
at 285°C in the initial rheological data would not appear in a secondary heating of a sample.
Indeed, we observed that when a single resin sample was subjected to a second temperature
ramp from 200°C to 400°C (after initial heating to 310°C) the local viscosity minimum at
285°C was no longer observed (Figure 3-3b). The absence of this feature during the second
heating cycle confirmed that the minimum |η*| at 285°C was the result of a chemical
reaction.
50
Figure 3-3. (a) Weight loss (TGA) and (b) complex viscosity of Oligomer 0.
Two possible reactions can occur during residual imidization. The first is a
condensation reaction between ester groups and amino groups, generating ethanol. While
the extent of this reaction is small, the resulting change in the molecular weight of the
oligomer is significant, resulting in an expected increase in complex viscosity. The second
possibility is a cyclodehydration reaction, in which the amide acid oligomer is converted
to the imide oligomer, generating water molecules. The difference in the viscosities of the
two oligomers is assumed to cause the change in the complex viscosity profile at 285°C.
Between 289°C and 314°C, the dynamic rheology measurements of Oligomer 0
revealed that G’ exceeded G”, indicating that the imide oligomer solidified in this
temperature range (Figure 3-2a). Above 298°C, imidization was nearly complete, although
the increasing temperature resulted in re-softening of the resin. Thus, G’ once again
dropped below G”, and the resin behaved as a viscoelastic liquid. The minimum in the
complex viscosity curve (|η*|min = 3.1×10
4
Pa⋅s) was observed at 338°C. This was
attributed to the cross-linking reaction of PEPA. Above 338°C, the viscosity increased
51
sharply, as the imide oligomer cross-linked into a stiff, cured thermoset. Note that the two
modulus curves crossed at 336°C. This crossover point is generally defined as the gel point
of the resin, providing an estimate of the temperature and time at which the resin formed
an infinite network of cross-links.
Overall, the resin began to soften at 252°C, and as temperature continuously
increased, residual imidization was triggered and dominated the viscosity change at 285°C,
resulting in a slight increase in viscosity between 285°C and 298°C. Above 298°C, residual
imidization was nearly complete, and resin softening dominated again, leading to a
decrease in viscosity. The primary minimum viscosity (|η*|min = 3.1×10
4
Pa⋅s) was
observed at 338°C, above which the cross-linking reaction markedly increased the resin
viscosity.
3.4.1.2. Effect of b-staging temperature
The purpose of a B-staging procedure in polyimide processing is to maximize the
degree of imidization of the oligomer prior to the final cross-linking reaction. In PMR resin
systems imidization and cross-linking either occur independently, or the two reactions
partially overlap in an intermediate temperature range. The first scenario (Figure 3-4a) is
generally preferred for the processing of composite materials, as consolidation pressure
can simply be applied after the resin is fully imidized. In the second type of system (Figure
3-4b), of which TriA X is one, imidization and cross-linking cannot be separated
completely. In this case, low B-staging temperatures generate oligomers with a low degree
of imidization and high volatile content which is difficult to remove in subsequent steps of
oligomer softening and laminate consolidation. B-staging at higher temperatures, however,
52
can trigger cross-linking, raising the viscosity of the resin system, and thus increasing the
difficulty of consolidating laminates. Therefore, the B-staging temperature must be tailored
for this type of PMR resin system to achieve an acceptable trade-off between imidization
and cross-linking.
Figure 3-4. Imidization and cross-linking relations in PMR polyimides: (a) isolation and
(b) overlap.
Rheology, DSC, and TGA were used to determine the effects of three potential B-
staging temperatures (220°C, 250°C, and 280°C) on resin viscosity, cross-linking, and
imidization. Rheological measurements were performed on Oligomer 0 (low degree of
imidization at 85.5%) to reduce the effects of volatile-induced bubbling during initial
imidization on viscosity results. Three rheological cycles were investigated, as summarized
in Table 3-2. For each cycle, Oligomer 0 was initially heated from room temperature to the
B-staging temperature and equilibrated for 10 min prior to the start of data collection.
Following this equilibration, a 4-h isothermal hold was performed at each B-staging
53
temperature. At the end of the isothermal hold the resin was heated to 400°C at a rate of
1°C min
-1
. Viscosity data for all three cycles are presented in Figure 3-5. The effect of B-
staging temperature on viscosity changes in each cycle is discussed below.
Cycle 1: Changes in the |η*| profile were observed only in the initial 30 min of the
isothermal hold, a result attributed to rapid imidization. Following the initial increase, |η*|
was approximately constant during the isothermal hold, stabilizing at 4×10
5
Pa⋅s,
indicating that no further reaction (i.e., cross-linking) occurred. The oligomer B-staged at
220°C exhibited a Tg of 240°C (determined by DSC). During the final temperature ramp
to 400°C, |η*| initially increased (up to 251°C) because the oligomer softened, which
increased the interfacial adhesion between the resin and the parallel plates. With continued
temperature increase, the oligomer continued to soften, displaying a local viscosity
minimum at 281°C, characteristic of residual imidization. The minimum |η*| (2.26×10
4
Pa⋅s) in Cycle 1 occurred at 340°C.
Cycle 2: In Cycle 2, |η*| was again approximately constant (1.3~1.4×10
6
Pa⋅s)
during the isothermal hold, although it was greater than observed with the lower
temperature cycle. The Tg of the oligomer (251°C, determined by DSC) was similar to the
B-staging temperature (250°C). As with Cycle 1, a constant |η*| during the isothermal hold
indicated that no cross-linking occurred after the initial imidization. As temperature
increased after the isothermal hold, the resin softened and exhibited signs of residual
imidization in the form of a local viscosity minimum at 290°C. The minimum |η*|
(4.06×10
4
Pa⋅s) in the cycle occurred at 338°C.
54
Cycle 3: The viscosity behavior for Cycle 3 differed markedly from that observed
in lower temperature B-staging cycles. The initial |η*| (1.3×10
5
Pa⋅s) during the isothermal
hold was less than the values for Cycle 1 and Cycle 2, because the dwell temperature was
greater than the Tg of the oligomer (260°C, determined by DSC), and thus the resin was
soft at the start of the dwell. However, during the isothermal hold, |η*| increased
continuously, indicating the contribution of a cross-linking reaction. No residual
imidization was observed during the final temperature ramp in Cycle 3. The minimum |η*|
(5.00 ×10
4
Pa⋅s), occurred at 326°C.
Table 3-2. Cycles for rheological measurements with ramps from isothermal holds to
400°C at 1°C min
-1
.
Cycle name B-staging |η*|min (Pa⋅s) T| η*|min (°C) Tg of oligomer (°C)
Cycle 1 220°C/4 h 2.26×10
4
340 240
Cycle 2 250°C/4 h 4.06×10
4
338 251
Cycle 3 280°C/4 h 5.00 ×10
4
326 260
Figure 3-5. |η*| profiles (solid lines) and temperature profiles (dash lines) of the B-staged
TriA X resins in Cycle 1 (black), Cycle 2 (red) and Cycle 3 (blue).
55
The higher value of minimum |η*| in Cycle 3 relative to Cycles 1 and 2 was
attributed to more extensive cross-linking. While the reaction rate was relatively slow at
280°C, the long heating time generated a non-negligible degree of cross-linking, resulting
in an increase in minimum |η*|. The cross-linking reaction at 280°C was also evident in
DSC results (Figure 3-6 – blue bars). The oligomer B-staged at 280°C exhibited a lower
heat of reaction, while the oligomers B-staged at 220°C and 250°C displayed comparable
heats of reaction.
Heat of reaction was measured via DSC scans of the oligomers from 35°C to 500°C.
Any cross-linking during B-staging causes a reduction in the heat of reaction at the cure
temperature, because some end caps react during B-staging. Therefore, the heat of reaction
for oligomers B-staged at 220°C (37.5 J g
-1
), 250°C (37.2 J g
-1
) and 280°C (24.3 J g
-1
) for
4 h revealed a cross-linking reaction at 280°C, while no cross-linking was observed at
220°C or 250°C.
Finally, TGA was performed on oligomers following each B-staging procedure
(with Oligomer 1 corresponding to Cycle 1, etc.) to determine the degree of imidization for
each isothermal condition. As expected, a lower B-staging temperature resulted in a lower
degree of imidization (Figure 3-6 – red bars). For the B-staging conditions examined here,
Oligomers 2 and 3 exhibited similar degrees of imidization (97.2 and 98.5%, respectively),
while Oligomer 1 showed a slightly lower degree of imidization (94.4%). As discussed
previously, a low degree of imidization is not desirable, because during subsequent heating
steps, partially imidized oligomers generate volatiles which are difficult to remove and can
lead to defects in finished parts.
56
Figure 3-6. Heat of reaction and degree of imidization of the B-staged TriA X resins.
The goal of B-staging is to achieve a high degree of imidization while avoiding the
onset of cross-linking. For the TriA X system, B-staging at 280°C produced a high degree
of imidization but also initiated cross-linking, leading to a high melt viscosity and thereby
reducing processability. No cross-linking was observed in samples B-staged at 220°C or at
250°C, but the degree of imidization was increased to 97.2% at the higher temperature.
Based on the principle of “maximum imidization without cross-linking”, a suitable B-
staging temperature of 250°C was selected (among the three candidate temperatures) for
the fabrication of composite laminates.
3.4.2 Composite fabrication
Relying on the thermochemical data, a molding cycle was developed for unstaged
T650-35 8HS/TriA X prepreg. This molding cycle consisted of two parts − the first carried
out in an air-circulating oven, and the second in a heated press (Figure 2-7). In the oven
cycle, the prepreg stack was heated to 50°C for 24 h to remove solvent (ethanol). At this
57
stage in the process, the viscosity of the monomer solution was low (< 10 Pa⋅s at
50°C~70°C), and thus drying in a vacuum bag assembly would cause unacceptable resin
bleed. For this reason, no vacuum was applied during the drying step. Additionally, the
8HS fabric used in the prepreg had relatively low gas permeability, rendering solvent
removal difficult and slow. When the temperature is close to or above the boiling point of
ethanol, the evaporation rate is rapid, and the prepreg plies exhibit “ballooning”, which can
lead to ply misalignment and wrinkling. To avoid these issues, drying was performed
slowly and at a moderate temperature (50°C).
After drying, the prepreg stack was bagged using a polyimide film (for thermal
stability) and cured in a heated press. An envelope bag assembly was used, with bagging
film on both sides of the laminate stack sealed along the perimeter with vacuum sealant
tape. This approach allowed positioning of the vacuum port and sealant tape outside the
hot zone of the press platens during consolidation and cure. A schematic of the bagging
sequence (Figure 3-8) shows that the lay-up is symmetric, with the prepreg as the mirror
plane.
Process parameters such as vacuum level/time of application, B-staging
temperature and duration, heating rates, and consolidation temperature and pressure, were
selected to obtain high quality laminates. Key steps in the hot-press cycle consisted of:
a. Initial temperature ramp from room temperature to 250°C at 7.5°C min
-1
. The
rate (the maximum heating rate of the hot press) was constant during the ramp. During the
ramp, the monomer mixture began to imidize and volatiles were generated. Three vacuum
58
pressure conditions were evaluated for this step, and a final selection was made based on
experimental observations.
i. Full vacuum: The complex viscosity of the monomer mixture showed a minimum
value of 2.5 Pa⋅s at 130°C. At this temperature, the monomer mixture melted. The pressure
difference between full vacuum and atmosphere yielded extensive resin bleed.
ii. No vacuum: During imidization, water and alcohol by-products formed, but in
the absence of applied vacuum, volatiles could not escape the bag assembly. Elevated
pressure within the bag caused the fabric to balloon, resulting in uneven laminates.
iii. Partial vacuum (10% of -101 kPa): Under conditions of reduced vacuum,
reaction by-products were able to flow out of the prepreg and exit the vacuum bag. With a
low degree of vacuum, neither bleeding nor ballooning occurred.
Based on these observations, partial vacuum was selected for the first ramp from
room temperature to 250°C.
b. B-staging at 250°C. As discussed previously, B-staging was carried out to
increase the degree of imidization. During B-staging, full vacuum was applied to remove
reaction by-products. Because of the low gas permeability of the 8HS fabric, the prepreg
was degassed for 4 h to ensure that all the volatiles were evacuated.
c. Consolidation at 325°C. As temperature was increased following B-staging,
residual imidization occurred with a corresponding release of by-products. A slow ramp
rate (1°C min
-1
) was therefore applied to facilitate evacuation of volatiles. After a 10-min
isothermal hold at 325°C, pressure was applied to consolidate the laminate (1.72 MPa).
59
Prior to applying pressure, a two-minute bump sequence was performed to evacuate
volatiles. To obtain void-free parts, imidization by-products must be removed before the
application of pressure.
65
If by-products remain in the laminate following consolidation,
the only path for volatile escape is through diffusion (which is prohibitively slow), and
violates trapped in the laminate will result in voids. As residual imidization is completed
at 300°C, a consolidation temperature of 325°C was chosen to ensure that no volatiles were
trapped. Given the limits of the heating control of the press, a ten-minute hold at this
consolidation temperature was introduced into the molding cycle to allow time for thermal
equilibration prior to application of pressure.
d. Cure at 371°C. A final isothermal hold was performed to facilitate cross-linking,
yielding a laminate with a Tg of 349°C (onset of storage modulus drop), 358°C (peak of
loss modulus) or 367°C (peak of tan 𝛿 ). The cure temperature/time determines the final
properties of the matrix. 2 h at 371°C yielded the optimal combination of the thermal and
mechanical properties.
Figure 3-7. Molding cycle for TriA X/T650-35 8HS: (a) Oven cycle and (b) hot press
cycle.
60
Figure 3-8. Bagging scheme for TriA X/T650-35 8HS.
The molding procedure described above was used to produce void-free laminates
with T650-35 8HS/TriA X. Figure 3-9 shows the cross section of a 16-ply laminate, c. 6
mm thick. The void content of the laminate, determined by IA, was 0.04%, and the 3-
dimensional void content determined by micro-CT was 0.18%.
Figure 3-9. TriA X/T650-35 8HS laminate with a lay-up of [0/+45/-45/90]2s.
3.4.3 Composite properties
Tensile, flexural, and SBS tests were performed on void-free laminates at room
temperature. Specimen stacking sequences, dimensions, fiber volume fractions, and void
contents are shown in Table 3-3. Prior to mechanical testing, all laminates displayed
uniform FVF, and porosity < 0.01% (as determined by IA).
61
The mechanical properties of T650-35 8HS/TriA X are summarized in Table 3-4.
Tensile and flexural strength and moduli are fiber-dominated properties, affected by
stacking sequence and fiber properties. In contrast, SBS strength depends on matrix
properties and fiber-matrix interfacial shear strength.
66
In this work, mechanical testing
was performed on quasi-isotropic laminates (T650-35 8HS/TriA X). Most mechanical
properties for polyimide composites reported in the literature, however, are from cross-ply
(0/90) laminates. To avoid differences based on fiber properties or lay-up sequence, we
have focused our comparison on SBS strength, a matrix-dominated property. The room-
temperature SBS strength of T650-35 8HS/TriA X (67 MPa) was 16% greater than that of
T650-35 8HS/PMR-15 (58 MPa)
59
and 46% greater than that of T650-35 8HS/AFR-PE-4
(46 MPa).
67
Table 3-3. Specimens for mechanical tests.
Mechanical
test
ASTM
standard
used for
testing
Lay-up
sequence
Length
(mm)
Width
(mm)
Thickness
(mm)
Number
of
specimens
FVF
(vol%)
Void
content
*
(%)
Tensile D3039
18
[0/+45/-
45/90]s
254.10 25.460 3.120 6 55.1 < 0.01
Flexural D7264
19
[0/90/-
45/+45/0/90]s
162.30 25.058 4.607 8 55.6 < 0.01
SBS D2344
20
[0/+45/-
45/90]s
17.54 5.886 3.097 5 55.1 < 0.01
*
Determined by image analysis.
62
Table 3-4. Mechanical properties of T650-35 8HS/TriA X.
Property T650-35 8HS/TriA X
Tensile strength (MPa) 515±27
Tensile modulus (GPa) 45.5±1.8
Strain-to-failure (%) 1.12±0.07
Flexural strength (MPa) 660.2±26.9
Flexural modulus (GPa) 42.0±0.9
SBS strength (MPa) 67.1±5.3
3.5 Conclusions
A polymer science-based approach was employed to develop and demonstrate a
process suitable for fabricating laminates with a new PMR-type polyimide resin system,
TriA X. The chemo-rheological behavior of the TriA X system was characterized, yielding
an improved understanding of resin flow in a complex reacting system and establishing
connections between viscosity changes and chemical/physical reactions. This
understanding guided the development of a cure cycle suitable for the fabrication of high-
quality laminates. Additionally, a thermal analysis method was designed to determine the
effects of B-staging temperature on resin viscosity, imidization, and cross-linking. A
principle of “maximum imidization without cross-linking” was implemented to determine
a suitable B-staging temperature. The approach we have demonstrated minimizes the
material/time/energy cost of process development, while affording accurate control over
resin properties in different processing conditions.
63
A molding cycle was designed based on the assessment of TriA X processability.
The cycle was implemented, leading to complete consolidation of 8HS prepreg plies into
composite laminates with Tg = 367°C (peak of tan 𝛿 ) and void content consistently < 0.1%.
The consolidated laminates exhibited room temperature matrix-dominated mechanical
properties that exceeded those of conventional polyimide composites. Chapter 2 has
reported exceptional ductility (elongation-at-break of 15.1%) and tensile strength (119
MPa) of TriA X neat resin, properties that far surpass those of conventional polyimides.
The exceptional mechanical properties of TriA X, and the process reported in this chapter
for fabricating composite laminates using this resin system, have the potential to expand
the design space, leading to new applications for polyimide composites. The system
addresses key drawbacks to prior polyimide materials as a result of the intrinsic properties
of the matrix, particularly superior ductility and ability to process without carcinogenic
constituents.
64
Chapter 4. Effects of thermal cycling on TriA X composites
4.1 Introduction
The volumetric shrinkage mismatch between a polymer matrix and carbon fibers
generates thermal stress in composites during cooling. Thermally induced tensile stress on
laminae can cause transverse microcracks, thereby decreasing mechanical performance,
and in extreme cases, cause material failure.
61, 68
Damage introduced by thermal stress is a
major concern in high-temperature polymer composites, because the composites are
processed at high temperatures and experience a wide range of temperatures during service.
In this work, thermal cycling tests were performed on a new polyimide-matrix composite
to investigate the effects of thermal cycling on the mechanical performance and service life
of the composites.
The cure process of thermoset prepreg creates three types of strain: chemical cure
shrinkage, thermal shrinkage and tool-part interation.
69, 70
Tool-part interaction can be
avoided simply by using release films between prepreg and tools. Secondly, the stress
introduced by cure shrinkage is usually negligible if the cure temperature is greater than
the Tg, because the curing resin is in a rubbery state with short relaxation time, high chain
mobility, and low stiffness.
71
Finally, the stress generated by cure shrinkage is typically
released during cure. Thus, it is generally assumed that composites are in a stress-free state
during cure above Tg.
72, 73
After cure, the polymer matrix is vitrified and bonded to the
fibers. CTE of polymers generally are much greater than those of carbon fibers. Thus,
volume change mismatch during cooling induces compressive stress in fibers, and tensile
65
stress in the matrix.
61
In laminates, the mismatch in properties between laminae of different
orientations results in interlaminar stress.
74
The longitudinal direction (0°) of a lamina has
much lower CTE than the transverse direction (90°), and thus 90° plies induce tensile
interlaminar stress on the adjacent 0° plies during cooling.
61, 75
The thermal stress induced during laminate cooling is proportional to the
temperature change (ΔT) from the cure temperature. The lower the minimum service
temperature of a composite, the greater thermal stress generated. When a composite is
cycled between two temperatures, the composite experiences cyclic thermal stress. Thus,
thermal cycling is equivalent to a form of thermally driven fatigue on each lamina.
61
Thermal stress can cause severe damage to composites. When the thermally induced stress
exceeds the transverse tensile strength of a lamina, transverse microcracks can develop,
and when thermal stress is greater than the interlaminar strength, delamination occurs.
74
Polyimide-matrix composites are especially susceptible to defects induced by
temperature variations during processing and service because of the high process
temperatures and the brittle nature. Microcracking and mechanical degradation induced by
temperature fluctuation have been reported for PMR-15 and other polyimide composites.
For example, extensive microcracking in PMR-15 composites was reported in thermal
cycling tests between -18°C and 232°C and between -156°C and 316°C.
13, 74
Thermal
cycling (1500 cycles) between -54°C and 232°C of PMR-15 composites decreased both
compressive strength and interlaminar shear strength by ~60%.
76
Indeed, some polyimide
composites crack during processing. For example, PMR-II-50 polyimide composites
microcracked upon cooling from 371°C (cure temperature) to room temperature during
processing, and additional thermal cycling between room temperature and 316°C
66
proliferated the microcracking.
51
Microcracks were also reported in thermal cycling tests
of thermoplastic polyimide composites, specifically thermal cycling (-54°C and 177°C) of
IM7/PIXA and IM7/K3B.
77
Previous study has suggested that increased matrix strength
reportedly increases composite resistance to thermal cycling, reducing microcracking.
76
Nevertheless, microcracking is a persistent problem for polyimide composites, because
microcracks not only decrease mechanical performance but also accelerate degradation
reactions, particularly thermal oxidation.
78
Thus, microcracking resistance to thermal
cycling is a critical performance metric for polyimide composites.
This chapter describes the thermal cycling test on TriA X composites. Conditioning
up to 2000 thermal cycles between -54°C and 232°C was performed on 8-harness satin
(8HS) T650-35 carbon fiber laminates produced with [0/+45/-45/90]s and [0]8 lay-up
sequences to investigate effects of thermal cycling. Thermomechanical analysis after
conditioning indicated that thermal cycling did not produce microcracks, but indirectly
decreased the microcracking resistance upon loading after thermal cycling because of
matrix degradation. Thermally driven fatigue was the primary mechanism responsible for
matrix degradation, while local creep was a possible secondary effect. The Tg and SBS
strength of the composites were measured after every 400 cycles. The Tg exhibited no
change, while the SBS strength retention was > 90% after 2000 thermal cycles. The high
stability of the polyimide composites upon thermal cycling was attributed to the high
tensile strength and thermal oxidative stability of the matrix.
67
4.2 Thermal stress analysis
4.2.1 Material properties
CTE and Young’s modulus of fiber and matrix are essential for the calculation of
thermal stress in the composites during thermal cycling. Carbon fibers are thermally and
dimensionally stable, and thus, the CTE and Young’s modulus of T650-35 fiber are
assumed to be constant in the temperature range from -54°C to 371°C (cure temperature of
the composites). The properties of T650-35 fiber are taken from reported values, where
Young’s modulus = 241 GPa and CTE = -0.5×10
-6
°C
-1
.
79
The values of CTE and Young’s modulus for TriA X are functions of temperature.
The CTE values of the polymer were measured in different temperature ranges using a
thermomechanical analyzer (TMA; Q400, TA Instruments, USA) and are listed in Table
4-1. The measurement was carried out with a static force of 0.02 N from -55°C to 400°C
at a ramp rate of 5°C min
-1
under a nitrogen flow of 50.0 mL min
-1
. The TMA was
calibrated with a standard aluminium reference from -55°C to 400°C using the same
parameters described above, and the calibration coefficient was determined according to
ASTM E2113.
80
The CTE of the polymer was calculated in accordance with ASTM
E831.
81
Table 4-1. CTE of TriA X.
T (°C) CTE (10
-6
°C
-1
)
-54~-25 36.6
-25~25 43.5
25~75 48.7
68
75~125 64.6
125~175 67.0
175~225 65.7
225~275 68.0
275~325 97.6
325~355 323.7
355~371 217.7
The Young’s modulus values are shown in Figure 1. The modulus values (black
dots in Figure 4-1) at -54°C, 23°C, 232°C and 288°C were measured from the tensile tests
in Section 2.3.5. The Tg of T650-35 8HS composites is 356°C, lower than the cure
temperature (371°C). The stiffness of polymers above Tg is much less than that in the
vitrified state. Thus, the Young’s modulus of the polymer at 371°C is assumed to be zero.
A second order polynomial fitting curve of Young’s modulus vs temperature (red line in
Figure 4-1) was constructed based on the measured and assumed values of modulus. The
fitting curve is described by Equation (4-1) with a R
2
= 0.99, indicating accurate fitting.
𝐸 𝑀 = −1.8 × 10
−5
𝑇 2
− 0.00319 𝑇 + 3.753 (4-1)
Figure 4-1. Young’s modulus of neat TriA X.
69
4.2.2 Thermal stress calculation
Thermal stress in T650-35 8HS composites was calculated to estimate the
magnitude relative to material strength upon exposure to cyclic temperatures. In 0/90/0
cross-ply laminates fabricated with unidirectional plies, the CTE and elastic modulus
mismatch between 0° plies (fiber direction, low CTE and high modulus) and 90° plies
(matrix direction, high CTE and low modulus) generates transverse tensile stress on 90°
plies during cooling after cure, and the stress can be determined by thermo-elastic
calculation.
75
In T650-35 8HS composites, however, the shape of individual fiber tows in
the cross-ply (non-crimp) region could not be distinguished after compression molding,
while fill yarns occupied the entire area between warp yarns. Thus, the fill yarns in the
cross-ply region were considered as a 90° unidirectional ply with a 0° unidirectional ply
(warp yarns) on each side. Consequently, the transverse thermal stress (𝜎 ) in the fill yarns
of the cross-ply region was estimated by thermo-elastic calculation:
61, 75
𝜎 = ∑ ∫
( 𝛼 𝑀 −𝛼 𝐹 ) 𝐸 𝐹 𝐸 𝑀 𝐸 𝐹 +𝐸 𝑀 𝑑𝑇
371°C
𝑇 (4-2)
In this work, only the thermal stress in [0]8 lay-up was determined, because the orientation
mismatch in cross-ply lay-up is the greatest for this case, and hence will produce the
greatest thermal stress.
The modulus and CTE of unidirectional T650-35 composites were determined by
assuming that the longitudinal and transverse properties of T650-35/TriA X were equal to
the longitudinal properties of T650-35 fibers and the properties of neat TriA X,
respectively. The property mismatch between longitudinal and transverse orientations of a
70
unidirectional composite is less than that between fibers and matrix. Thus, this assumption
will present an upper bound of thermal stress. In Equation (4-2), α and E are CTE and
Young’s modulus, while the subscripts M and F represent “Matrix” and “Fiber”,
respectively.
Because T650-35 8HS laminates were cured above the Tg, during cure, polymer
chains were in a rubbery state with high mobility and short relaxation time, allowing
relaxation of stress induced by cure shrinkage.
71-73
Thus, the internal stress at 371°C was
assumed to be zero. The thermal stress at a temperature (T) in a cross-ply region was
calculated by integrating Equation (4-2) from T to 371°C. The stress values at -54°C and
232°C are shown in Figure 4-2b (black solid line) with the temperature profile of cycles
(Figure 4-2a).
Figure 4-2. (a) Temperature profile of thermal cycles, and (b) corresponding thermal
stress in cross-ply region (solid black) and crimped region (dashed black) of [0]8
laminates and tensile strength of the matrix (blue).
71
Numerical calculation of thermal stress in T650-35 8HS/PMR-15 composites
revealed that the thermal stress in crimped region was 12% greater than in the cross-ply
region.
73
Because PMR-15 and TriA X exhibited similar thermal expansion behavior,
T650-35 8HS/TriA X laminates are expected to exhibit thermal stress distribution similar
to T650-35 8HS/PMR-15. Thus, thermal stress in crimped regions of T650-35 8HS/TriA
X laminates was estimated by multiplying the calculated stress in the cross-ply region by
12%, shown as the black dashed line in Figure 4-2b.
The goal of thermal stress calculation is to demonstrate the thermal stress level
relative to the matrix strength. The tensile strength of neat TriA X was measured at -54°C
and 232°C, and results are shown as the blue line in Figure 4-2b. The internal stress induced
by thermal cycles is 51% and 60% less than the tensile strength of the matrix at -54°C and
232°C, respectively. Thus, T650-35 8HS composites were expected to exhibit resistance
to microcracking during thermal cycling between these temperatures (-54°C and 232°C).
4.3 Thermal cycling experiments
4.3.1 Materials
T650-35 8HS/TriA X composites were fabricated in two lay-up sequences – [0]8
and [0/+45/-45/90]s – using the molding cycle described in Chapter 3. A TriA X neat
polymer panel was prepared by molding PMDA di-ester/p-ODA/PEPA mono-ester powder
blend in a hot press using the method described in Section 2.1.3.
72
4.3.2 Thermal cycling conditions
8HS laminates were machined into rectangular coupons for thermal cycling
followed by microcrack inspection and SBS testing. Specimen dimensions are described
in subsequent sections 4.3.3 and 4.3.7. Neat polymer coupons (40×40×3 mm) were
prepared for investigation of effects of thermal cycling on matrix physical properties. The
specimens for thermal cycling test were wet machined, then dried at 100°C for 64-89 h and
stored in desiccators prior to thermal cycling. The specimens were thermally cycled
between -54°C and 232°C with a 10-minute hold at -54°C and a 15-minute hold at 232°C
(Figure 4-2a). Up to 2000 thermal cycles were performed at The Boeing Company,
Huntington Beach. Samples were removed every 400 cycles for microcrack inspection and
testing.
4.3.3 Microcrack Inspection
Composite specimens (100×50×3 mm) with both lay-up sequences were inspected
for microcracks at 400-cycle intervals, and SBS tests were performed. Each specimen was
sectioned in three orientations – 0° (warp), 45°, and 90° (fill) – and mounted in potting
resin. The cross-sections (~20 mm long) were wet polished down to 1 𝜇 m diamond
suspension and inspected for microcracks using a digital microscope (VHX-600, Keyence,
Japan).
High-resolution X-ray tomography (microCT) was performed on the composite
specimens before thermal cycling and after 2000 cycles to detect microcracks in three
dimensions. Specimens were sectioned from thermally cycled laminates and scanned (XT
73
H 225ST, Nikon, Japan). Mo-Kα incident radiation (𝜆 = 0.71 Å) was used with 60 kV/220
𝜇 A voltage/intensity settings to achieve a resolution of ≦4 𝜇 m per pixel.
4.3.4 Oxidized layer inspection
A 2-mm thick cross-section (6×3 mm) was cut from the neat polymer specimen
after 2000 cycles. The cross-section was wet polished on both sides down to 200 𝜇 m using
a disc grinder and sandpapers. The cross-section under transmitted light was imaged using
a digital microscope (VHX-600, Keyence, Japan) to detect discoloration from oxidation.
4.3.5 Dynamic Mechanical Analysis
DMA was performed to determine the Tg values of T650-35 8HS laminates and
neat polymer before and after thermal cycling (Q800, TA Instruments, USA). The
measurements were carried out in single cantilever mode at a heating rate of 5°C min
-1
with
a fixed frequency of 1 Hz and a strain of 0.3%. The peak of the tan 𝛿 curve was recorded
to determine Tg.
4.3.6 Nano Indentation
Nano indentation was conducted on neat polymer samples before and after thermal
cycling to measure the surface Young’s modulus. The indentation was carried out using
continuous stiffness measurement mode with a constant strain rate of 0.01 s
-1
at an
oscillation frequency of 45 Hz, using a nano-indenter (Nano Indenter XP, MTS, USA) with
a Berkovich tip (TB26980, Keysight Technologies, USA). The maximum penetration
depth was 2000 nm, and drift correction was applied during tests. 50 indents were
74
performed on each specimen with an indentation spacing of > 100 𝜇 m. The Young’s
modulus was calculated as the average result between 600 nm and 1000 nm using analysis
software (TestWorks 4, MTS, USA).
4.3.7 Short-Beam Shear Test
The SBS strength of T650-35 8HS/TriA X composites was measured at room
temperature in accordance with ASTM D2344.
64
Specimen thermal history, dimension,
lay-up sequence, and number are summarized in Table 4-2. The tests were performed on a
load frame (5567, Instron, USA) at a crosshead displacement rate of 1.0 mm min
-1
.
Specimens were stored in desiccators for at least 7 days prior to testing.
Table 4-2. Specimens for SBS tests before and after thermal cycling.
No.
of
cycles
Lay-up sequence
[0/+45/-45/90]s [0]8
Length
(mm)
Width
(mm)
Thickness
(mm)
No. of
specimens
Length
(mm)
Width
(mm)
Thickness
(mm)
No. of
specimens
0 18.64 6.20 3.10 8 18.41 6.13 3.08 5
400 18.66 6.21 3.12 6 18.41 6.13 3.11 5
800 18.66 6.21 3.09 7 18.41 6.11 3.08 5
1200 18.65 6.19 3.12 8 18.41 6.12 3.09 8
1600 18.63 6.20 3.10 8 18.42 6.13 3.10 5
2000 18.66 6.21 3.12 8 18.41 6.12 3.09 5
75
4.4 Results and discussion
4.4.1 Microcrack inspection
Composite specimens were removed from the thermal cycle chamber every 400
cycles for microcrack inspection. Specimens were sectioned and polished in three
orientations (0°, 45°, and 90°) for observation. Figure 4-3 shows the polished sections of
specimens after 2000 thermal cycles. No microcracks were observed in any of the
specimens using light microscopy. For further inspection, microCT scans were performed
on specimens after 2000 thermal cycles. Only a few microcracks were observed, and the
morphology and position of microcracks (relative to the fabric architecture) were similar
for all the cracks observed. Figure 4-4 shows a typical microcrack (outlined in red
rectangles), where the image sequence extends from one end of a microcrack (a) to the
other end (h). The sequence is also represented schematically in Figure 4-5, which shows
the three-dimensional distribution of microcracks, using color to indicate the layer position
of microcracks. For example, yellow lines depict microcracks located in the first ply of the
laminate.
76
Figure 4-3. 8HS laminates after 2000 cycles: [0]8 in 0° (a), 45° (b), and 90° (c), and
[0/+45/-45/90]s in 0° (d), 45° (e), and 90° (f).
77
Figure 4-4. A typical microcrack (boxed in red rectangles) observed in microCT scans of
8HS laminates after 2000 cycles; image sequence from (a) to (h) is corresponding to
direction from a to h indicated in Figure 4-5; (i) fiber morphology in 8HS weave: tow
edges (I and II) at an overlap of orthogonal tows (I-II) and cross-ply region (III).
78
Figure 4-5. Microcrack distribution observed in microCT scans of T650-35 8HS/TriA X
after 2000 cycles; color indicates ply position of microcracks.
Microcracking was detected only in the resin-rich regions near tow crimp of 8HS
laminates (Figure 4-4). In these laminates (Figure 4-4i), three types of fiber regions are
present: a cross-ply region (III), two tow edges (I and II) at an overlap of orthogonal tows,
and a crimp region (I-II). In Figure 4-4a, Tow A was at a tow edge (IA) of a tow overlap
region, while Tow B was in a cross-ply region (IIIB) that was near a crimp. A microcrack
terminated in the resin-rich region adjacent to Tow A (Figure 4-4b). In Figures 4-4c and 4-
4d, Tow A extended into the cross-ply region (IIIA) that was yet close to the crimp, while
Tow B reached the first tow edge (IB) of a tow overlap, where the resin-rich region was
larger and the microcrack spanned the entire region. In Figures 4-4e and 4-4f, Tow B was
situated at the tow overlap (I-IIB) and the microcrack was at a boundary between Tow A
79
and an adjacent fiber tow. In Figure 4-4g, Tow B reached the second tow edge (IIB) of the
tow overlap, and the resin-rich region around the microcrack became smaller so that the
microcrack was shorter. In Figure 4-4h, both Tow A and Tow B were in the cross-ply
region, the microcrack had terminated. The microcrack shown in Figure 4-4 is typical of
all microcracks observed in the microCT scans of the 8HS laminates X. The full set of
microCT scans of the microcrack presented in Figure 4-4 is shown as a video and available
at https://youtu.be/cuwALxuYroU.
The features of the microcracks in the microCT scans of thermally cycled laminates
differed from those of thermal-cycling-induced microcracks in polyimide composites. As
described above, microcracking was observed only in the resin-rich regions of laminates
and was attributed to matrix failure. In contrast, microcracking in other polyimide
composites typically occurred within fiber tows (fiber-rich regions), stemming from
debonding at fiber-matrix interfaces.
13, 51, 74, 76, 77
In addition to this difference in microcrack
location, the distribution of microcracks in T650-35 8HS/TriA X was inconsistent with that
of the typical microcracking generated by temperature variations. All microcracks occurred
in one fiber direction in each laminate, apart from just two microcracks in ± 45° plies of
the [0/+45/-45/90]s lay-up. Because of the woven reinforcement, microcracks caused by
internal stress are expected to appear in both fill and warp directions (based on accepted
microcracking mechanics). Moreover, the microcracks here appeared only on one side of
the specimens, while thermal-cycling-induced microcracks normally occur in all fiber
directions and are distributed symmetrically with respect to laminate mid-planes.
13, 74, 77
The unusual features of the microcracks observed in the micro-CT scans indicate
that thermal cycling was not directly responsible for the cracks. The characteristics of the
80
microcracks were consistent with artifactual damage. In particular, the microCT specimens
were machined using a low-speed diamond saw after thermal cycling. Cyclic bending loads
were applied to the specimens during sectioning, producing microcracks, of which ~50%
were near to the cut surfaces. Based on comparison of microcracks in the 8HS laminates
and thermal-cycling-induced microcracks reported in other polyimide composites,
13, 51, 74,
76, 77
we concluded that the additional loading generated by sectioning after thermal cycling
was the cause of observed microcracking.
The microcracking observed in 8HS laminates (attributed to sectioning after
thermal cycling), indicates that the matrix was embrittled by exposure to cyclic
temperatures. Thus, thermal cycling was indirectly associated with the microcracking.
MicroCT scans were also performed on fresh specimens (machined by the same method as
described above), and no microcracks were detected. During thermal cycling, three
possible processes were involved. Because samples were thermally cycled in air, thermal
oxidative aging was possible, causing matrix degradedation.
82
In addition, thermally
driven fatigue and creep (at high temperatures) was possible because of the cyclic stress
induced by temperature change. Both fatigue and creep can induce a process of defect
nucleation before cracking occurs, leading to rupture of atomic bonds and creating stress
concentration sites in forms of chain ends and cavities (clusters of ruptured bonds).
83, 84
Any increase in defects generated by fatigue and creep could result in greater brittleness of
the matrix and make microcracking easier upon additional loading afterward. The effects
of these three factors are discussed next.
81
4.4.2 Effects on matrix
Thermal oxidation of polyimides commonly manifests as surface discoloration and
increased Tg and Young’s modulus.
85-88
To investigate the effects of thermal oxidative
aging on the matrix during thermal cycling, neat polymer specimens were thermally cycled
in the same conditions with the composite specimens. Thermal stress was induced by fiber-
matrix interaction upon temperature change. When the fibers were absent in the system, no
thermal stress occurred. Note that the neat polymer specimens were the same thickness (3
mm) as the composite specimens, giving similar through-thickness temperature gradients
during heating and cooling. Thus, the neat TriA X specimens experienced nearly identical
thermal/oxidative exposure to the composites, but without thermal stress.
A thin section of neat polymer after 2000 cycles is shown in Figure 4-6, imaged
with transmitted light, showing no discolored surface layer (the curved profile at corners
was created by the saw kerf). Discoloration caused by thermal oxidation has been widely
reported for PMR-15,
85, 86, 88
and this has also been reported in long-term thermal oxidative
aging of TriA X at 288°C. The Tg and Young’s modulus (measured on the surface) of neat
polymer after thermal cycling are summarized in Figure 4-7. The Tg and Young’s modulus
exhibited negligible change (~1%) after 2000 cycles. Oxidized polymers typically exhibit
greater Young’s modulus than fresh polymers, and this is commonly observed in PMR-15
and other thermosets.
85, 88, 89
The constant physical properties of neat TriA X after thermal
cycling indicates that no thermal oxidative degradation occurred in the matrix of 8HS
laminates during 2000 thermal cycling. As discussed in Section 4.4.1, only three factors,
thermal oxidative aging, thermally driven fatigue and creep, could degrade the matrix in
82
thermal cycling test and reduce the microcracking resistance upon additional loading.
Because thermal oxidative aging was excluded, the other two factors, thermally driven
fatigue and creep, likely caused the matrix degradation.
Figure 4-6. Cross-section of neat TriA X after 2000 cycles (transmitted light).
Figure 4-7. Tg and Young’s modulus at surface of neat TriA X.
To determine the effects of thermally driven fatigue and creep on the matrix of
T650-35 8HS/TriA X, matrix-dominated properties – Tg and SBS strength – were measured
83
(Figure 4-8). The Tg (red solid triangles and green solid diamonds in Figure 4-8) exhibited
negligible change (0.3-0.4% decrease) after 2000 cycles, while the SBS strength of [0]8
(red solid squares in Figure 4-8) and [0/+45/-45/90]s (blue solid circles in Figure 4-8)
decreased slightly, by 3.3% and 9.1%, respectively. Because SBS strength is a matrix-
dominated property, the decrease in SBS strength indicates the mechanical degradation of
the matrix. As described above, only local fatigue and creep induced by cyclic temperatures
contribute to matrix degradation. To investigate the relative importance of these factors,
SBS tests were performed on specimens aged for 500 h at 232°C, equivalent to the total
time at 232°C during 2000 cycles. During aging at 232°C, only local creep could affect the
composites, because the thermal stress was constant at constant temperature. The SBS
strength of aged [0]8 (red hollow square in Figure 4-8) and [0/+45/-45/90]s (blue hollow
circle in Figure 4-8) laminates was 0.6% and 3.1% less than that of fresh specimens,
respectively, and can be attributed to possible local creep or normal measurement variance.
The slight decreases in SBS cannot prove the existence of creep or exclude the possibility.
However, the effects of local creep are negligible or secondary in comparison to thermal
fatigue, because the SBS strength reduction after isothermal aging was far less (82% less
for [0]8 laminates and 67% less for [0/+45/-45/90]s laminates) than after 2000 thermal
cycles, where both fatigue and creep could have occurred. Therefore, thermally driven
fatigue is identified as the primary factor responsible for mechanical degradation of the
matrix, resulting in the reduction in SBS strength after thermal cycling, while creep at high
temperatures is possibly a secondary factor. The slightly greater (6%) SBS strength
retention observed in the thermally cycled [0]8 laminates compared to [0/+45/-45/90]s
84
laminates is attributed to effects of lay-up sequences – the SBS strength of isotropic lay-
ups is more matrix-controlled than that of cross-ply lay-ups.
74
Figure 4-8. Tg and SBS strength of T650-35 8HS/TriA X.
In addition to thermally driven fatigue and creep, fabric geometry also affected the
mechanical degradation of the matrix. The resin-rich regions were adjacent to crimped tows
in the 8HS laminates. The thermal stress in the crimped region is 12% greater than that in
the cross-ply region, because of the tow curvature.
73
Thus, resin pockets in crimped regions
sustain maximum thermal stress. Here, no fiber reinforcement was present to carry the
thermal stress, and thus crimped regions were most susceptible to thermal cycling damage.
Consequently, microcracking occurred upon addition loading. To increase microcracking
resistance after thermal cycling, fabric architecture that minimizes/eliminates fiber crimp
will mitigate effects of resin-rich regions. Thus, non-woven reinforcement, such as non-
85
crimp fabric, unidirectional tape, or spread-tow fabric, should impart increased resistance
to thermal cycling degradation.
Although the matrix degraded slightly from thermal cycling, the retention of SBS
strength was greater than 90% after 2000 cycles. Fiber-dominated mechanical properties
are not as sensitive to microcracks as matrix-dominated properties.
13, 74
SBS strength
(interlaminar shear strength) is one of the common matrix-dominated properties
investigated as an indicator of mechanical performance in thermal cycling tests of
polyimide composites.
61
Note that the interlaminar shear strength of PMR-15 composites
after 1500 cycles between -55°C and 232°C was reported to be ~40% of the initial value
because of extensive microcracking, which arose from the low transverse tensile strength
of laminae (comparable to the predicted thermal stress level).
76
In contrast, high residual
matrix-dominated properties of TriA X are attributed to the absence of microcracks, which
in turn stemmed from the high tensile strength of the matrix.
4.5 Conclusions
Resistance to microcracking during thermal cycling is a critical performance
indicator for polyimide composites because of the brittle nature and high CTE of the
matrix. Indeed, no polyimide composites have been reported to date that withstand thermal
cycling (with similar temperature ranges) without microcracking. Unlike conventional
polyimides, TriA X possesses greater tensile strength, sufficient to withstand the thermal
stresses induced by cyclic temperatures. As reported previously, the intrinsic strength stems
from the distinctive molecular structure. Conventional image analysis did not reveal
microcracks in thermally cycled laminates, although a few microcracks were detected in
86
microCT scans. These cracks were attributed to additional loading (during sample
sectioning) after thermal cycling. The > 90% retention of SBS strength after 2000 cycles
was attributed to the absence of microcracks, while the < 10% reduction was attributed to
slight matrix degradation. After 2000 cycles, laminates were more susceptible to damage
from sectioning than from interlaminar shear loading. Additional investigation is needed
to more fully understand the mechanical behavior of TriA X composites after thermal
cycling.
Three factors – thermal oxidative aging, thermally driven fatigue, and creep – are
considered as possible causes of the SBS strength degradation, and two of these are deemed
unlikely. First, no evidence of thermal oxidation from thermal cycling was detected.
Thermally driven fatigue induced by the cyclic temperatures is the primary cause of the
mechanical degradation of the matrix while creep is the possible secondary factor. In
addition to local fatigue and creep, curved fiber geometry of 8HS fabric also affected the
performance of the matrix and decreased the microcracking resistance of the composites
after thermal cycling. Thus, non-woven reinforcement is recommended for TriA X
composites in the future work.
In the field of polyimide composites, PMR-15 has been the most widely reported
matrix for high-temperature applications. However, all polyimide composites, including
PMR-15 composites, undergo extensive microcracking during thermal cycling, a major
limitation, particularly for oxidizing environments. Thus, polyimide composites are limited
presently to short-term structural components or long-term non-structural components.
13
The results presented here demonstrate that composites based on a newly designed
polyimide (TriA X) can withstand thermal cycling without direct microcracking. The
87
increased resistance to thermal cycling is likely to expand the design space of polyimide
composites for longer-term high-temperature structural applications.
88
Chapter 5. Effects of moisture on TriA X and composites
5.1 Introduction
Because of the high Tg and thermal stability of polyimides, carbon fiber composites
with polyimide matrices are deployed in high-service-temperature applications, such as
aeroengines and supersonic aircraft.
15, 35
However, polyimide-matrix composites are more
expensive than conventional epoxy-matrix composites because of resin costs (200-900
USD kg
-1
) and the complexity of the fabrication/cure process.
90, 91
Thus, long lifetime is
expected for polyimide-matrix composites, and stability issues must be addressed before
adopting polyimide resins in new engineering applications.
Hydrothermal stability of polyimide systems has been a primary concern for
aerospace applications, because polyimides are hydrophilic polymers and thus susceptible
to moisture.
92, 93
The general effects of moisture include the hydrolysis of imide units and
water plasticization.
92-94
Hydrolysis can depolymerize polyimides by opening imide rings
to form polyamic acids, followed by chain scission and regeneration of monomers (or even
de-monomerization), resulting in degradation of dry mechanical properties.
92
For example,
after 1000-h hydrothermal exposure at 160°C and subsequent drying, one polyimide
(AFR700B) manifest ~70% strength loss and ~85% strain-to-failure decrease, while
another (K3B) showed ~18% strength loss and a ~21% decrease in strain-to-failure.
92
In
addition to chemical reactions, water molecules can act as plasticizing agents in polymer
networks, pushing polymer chains apart, and hence reducing Tg.
43
This effect, defined as
plasticization, can not only reduce Tg but also adversely affect mechanical properties. For
89
instance, the wet Tg (256°C) of PMR-15 is 73°C below the dry Tg (338°C).
59
Because the
Tg decreased sharply with moisture uptake, the wet flexural properties at 260°C and 316°C
of HTS-2/PMR-15 composites were 10-20% less than the dry properties at the same
temperatures.
40
Moisture absorption also can lead to permanent damage in polyimide composites.
Microcracking is typically the first form of damage to manifest in composites under
extended hot/wet exposure.
95
For example, long-term moisture exposure reportedly
degraded the “microcracking toughness” of both thermoplastic and thermoset polyimide
composites.
96
When the toughness dropped sufficiently, the composites lost ability to
withstand internal stresses (generated during manufacturing), and microcracks formed.
Microcracking of IM7/K3B and IM7/PETI-5 initiated after water immersion at 80°C for
200 h and 1500 h, respectively.
96
Another type of damage observed in polyimide
composites is blistering. When composites are heated, absorbed moisture can vaporize and
generate internal pressure. When the vapor pressure exceeds the matrix strength, blistering
occurs in the form of macrovoids, microcracks, and/or delamination.
93, 97
Although the effects of moisture absorption on polyimides and polyimide-matrix
composites have been studied for decades, few studies have addressed long-term
hydrothermal aging. This chapter focuses on both short-term moisture absorption of TriA
X and the effects of long-term moisture exposure on the neat polymer and associated
composites. Although the remarkable ductility of TriA X may expand future high-
performance applications of polyimide composites, the stability of TriA X is not yet fully
investigated. Here, we report the long-term hydrolytic stability of TriA X to further explore
the potential of this polyimide as a composite matrix for severe service environments.
90
Short-term moisture uptake experiments were conducted at different temperatures
and relative humidity (RH) to investigate the moisture absorption behavior of neat TriA X.
The polyimide exhibited a two-stage, exothermic absorbing process. Moisture uptake
models were determined as functions of temperature and moisture. The long-term
hydrothermal stability of the neat polyimide and composites were evaluated based on
retention of chemical, thermal and mechanical properties. Long-term hydrothermal aging
caused no deterioration in thermal properties and negligible permanent chemical
degradation. However, ductility decreased with increasing exposure time because of
reversible hydrolysis and moisture plasticization, resulting in associated reductions in
toughness and SBS strength of the associated composites.
5.2 Materials
Neat polyimide panels and films were prepared by molding PMDA di-ester/p-
ODA/PEPA mono-ester powder blend in a hot press using the method decried in Section
2.1.3. The neat polymer samples were machined to designated sizes for measurements
described in the following sections. T650-35 8HS/polyimide composites were fabricated
in two lay-up sequences, [0]8 and [0/+45/-45/90]s, using the molding cycle described in
Chapter 3. The composite laminates were machined to 18×6×3 mm specimens for SBS
tests.
91
5.3 Experiments
5.3.1 Moisture uptake
The moisture diffusion behavior was investigated by short-term water uptake of
neat polyimide specimens (70.0×12.7×3.1 mm) at 35°C, 55°C, 75°C and 95°C. Prior to
absorption experiments, all samples were dried at 95°C for 94 h and then at 151°C for 92
h under vacuum until the weight stabilized. Next, specimens were immersed in deionized
water at different temperatures, and the weight change was monitored as a function of time.
The water temperature was controlled by a digital heating controller (MC810,
Electrothermal, UK). At each temperature, five specimens were tested, and all specimens
were machined from the same panel.
The relationship between maximum moisture content and ambient relative
humidity was determined by conditioning neat polyimide specimens (70.0×12.7×3.1 mm)
at different humidity levels created from selected saturated salt solutions.
98
Saturation was
performed in sealed containers with saturated MgCl2 (38.5% RH), NaBr (54.0% RH) and
NaCl (77.0% RH) solutions, respectively, at 35°C. Humidity and temperature were
recorded using stand-alone data loggers (EL-USB-2, Lascar Electronics, UK). Prior to
moisture exposure, all samples were dried using the method described above. At each
relative humidity, five specimens were tested.
Long-term water uptake at 95°C was conducted to investigate the effects of
moisture absorption on the properties of the neat polymer and associated composites. The
specimens were removed from the water bath every 1000 h for property measurements in
92
wet and dry conditions. Prior to dry tests, the specimens were dried at 95°C under vacuum
for 3 days, then stored in a desiccator at room temperature for at least 4 days, while wet
test specimens were stored in a water bath at room temperature prior to testing.
5.3.2 Fourier-transform infrared spectroscopy
Attenuated total reflectance Fourier transform infrared spectroscopic analysis
(ATR-FTIR; Nicolet 4700, Thermo Fisher Scientific, USA) was performed to investigate
the chemical structure change during long-term hydrothermal aging. The spectra were
recorded by scanning the neat polymer films (~0.1 mm thick) with a diamond single bounce
ATR sampling accessory (Smart Obit, Thermo Fisher Scientific, USA). 128 scans were
performed for each sample.
5.3.3 X-ray photoelectron spectroscopy
The surface chemistry of polyimide films before and after long-term aging was
analyzed using an X-ray photoelectron spectrometer (XPS; Kratos AXIS Ultra, Kratos
Analytical, UK). The instrument featured a monochromatic Al Kα source (1486.6 eV) at 3
mA and 6 kV. Survey spectra were acquired at 160 eV pass energy with 1 eV step, followed
by narrow region spectra (O 1s, N 1s and C 1s) acquired at 40 eV pass energy with 0.1 eV
step. 5 sweeps were conducted for survey spectra and 15 sweeps for each element range.
The operating pressure was ~1×10
-9
Torr, so XPS was performed only on dry specimens.
Four different locations were scanned on each sample, and spectra were analyzed using
software (CasaXPS version 2.3.16, Casa Software Ltd., UK).
93
5.3.4 Dynamic mechanical analysis
DMA was performed to determine the Tg and structure relaxations of the neat
polyimide (Q800, TA Instruments, USA). Measurements were performed in single
cantilever mode at a heating rate of 5°C min
-1
with a fixed frequency of 1 Hz and a strain
of 0.3%. The temperature range for each measurement was -90°C-450°C, achieved using
a gas cooling accessory (Q Series
TM
, TA Instruments, USA). Tg was determined by tan 𝛿
maximum.
5.3.5 Thermogravimetric analysis
TGA of polyimide samples was performed under nitrogen purge at a flow rate of
25.00 mL min
-1
(Q5000, TA Instruments, USA). TGA data were acquired to determine the
decomposition temperature after every 1000-h hydrothermal aging at 95°C. TGA was
performed on dry polyimide films only. To remove residual moisture, the films were dried
at 100°C for 1 h in the TGA, prior to each measurement. Subsequently, the films were
heated from 30°C to 600°C at a rate of 10°C min
-1
. The decomposition temperature was
defined as the temperature of 5% weight loss (Td5%).
5.3.6 Tensile tests
The effects of hydrothermal aging on tensile properties of the neat polyimide were
measured in accordance with ASTM D638 standard.
45
Dimension, condition, and number
of specimens are given in Table 5-1. The tests were conducted using a load frame (5567,
Instron, USA) with a 5000-N load cell (2525-805, Instron, USA). Tensile tests were
performed at a displacement rate of 1.0 mm min
-1
at room temperature. The strain was
94
measured using a 3-D optical displacement measuring system (ARAMIS; Adjustable Base
2.3 M, GOM, Germany). Speckle patterns were painted on the gauge sections of the
specimens prior to the tests. As the specimens deformed, the measuring system tracked the
speckle pattern evolution to determine dimension change.
Table 5-1. Specimens for tensile tests before and after hydrothermal aging.
Aging
time
(h)
Condition
Length
a
(mm)
Width
b
(mm)
Thickness
b
(mm)
Number
of
specimens
0
Dry
66.0 3.16 3.13 6
Wet
66.0 3.19 3.12 6
1000
Dry
66.0 3.18 3.09 5
Wet
66.0 3.17 3.12 5
2000
Dry
66.0 3.15 3.08 5
Wet
66.0 3.20 3.11 5
a
Total length of specimens.
b
Gauge section dimensions.
5.3.7 Short-beam shear tests
The SBS strength of T650-35 8HS composites was measured at room temperature
in accordance with ASTM D2344 standard.
64
Dimension, condition, lay-up sequence, and
number of specimens are summarized in Table 5-2. The tests were performed on a load
95
frame at a crosshead displacement rate of 1.0 mm min
-1
with a support span of 12.5 mm
for all specimens.
Table 5-2. Specimens for SBS tests before and after hydrothermal aging.
Lay-up
sequence
Aging
time
(h)
Condition
Length
(mm)
Width
(mm)
Thickness
(mm)
Number
of
specimens
[0]8
0
Dry
18.41 6.13 3.08 5
Wet
18.44 6.13 3.07 7
1000
Dry
18.44 6.13 3.11 5
Wet
18.43 6.11 3.09 5
2000
Dry
18.44 6.14 3.09 6
Wet
18.45 6.13 3.12 5
[0/+45/-
45/90]s
0
Dry
18.64 6.20 3.10 8
Wet
18.66 6.21 3.12 7
1000
Dry
18.64 6.21 3.12 5
Wet
18.66 6.19 3.10 5
2000
Dry
18.67 6.22 3.09 7
Wet
18.69 6.23 3.13 6
96
5.4 Results and discussion
5.4.1 Moisture absorption behavior
The moisture content (Mt) of neat TriA X is plotted as a function of water
immersion time at different temperatures in Figure 1. The moisture content is defined by
𝑀 𝑡 =
𝑊 𝑤𝑒𝑡 −𝑊 𝑑𝑟𝑦 𝑊 𝑑𝑟𝑦 × 100% (5-1)
where Wwet is the wet sample weight, and Wdry is the dry sample weight. The symbols in
Figure 5-1 are the measured values. In the initial stage of water absorption, Fickian
diffusion behavior was observed, i.e., the moisture content was linearly proportional to the
square root of time. However, after the moisture absorption reached a near-equilibrium
stage (plateaus in Figure 5-1), the moisture content continued to increase, albeit at a much
slower rate compared to the initial stage. The observed moisture absorption behavior
conforms to the two-stage diffusion model reported by Bao et al for a bismaleimide
polymer.
94
The first stage is diffusion-controlled, while the second stage is governed by
network relaxation induced by moisture plasticization, where the absorbed water molecules
create swelling that opens the structural network and renders it more accessible to
additional moisture. This two-stage diffusion model is described by
𝑀 𝑡 = 𝑀 ∞0
( 1 + 𝑘 √ 𝑡 ){1 − 𝑒𝑥𝑝 [−7.3 (
𝐷𝑡
ℎ
2
)
0.75
]} (5-2)
where M∞0 is the moisture content (quasi-equilibrium) at the end of the Fickian diffusion,
k characterizes the relaxation rate in the second stage, t is moisture absorbing time, D is
diffusivity, and h is specimen thickness.
94
97
Figure 5-1. Moisture gain of neat polyimide in water at different temperatures. The
symbols are experimental values, and the solid lines are curve fits using Equation (5-2).
Because of the low surface-to-edge areal ratio (7:2), the diffusion through the edges
of the specimens cannot be neglected, and edge correction must be considered when using
Equation (5-2) to determine diffusivity. In Equation (5-2), D is the apparent diffusivity
determined from diffusion through all six surfaces in three directions. Because the neat
polymer is an isotropic and homogeneous material, the diffusivities in all three directions
are equal. The true one-directional diffusivity Dx is given by
𝐷 𝑥 = 𝐷 (1 +
ℎ
𝑙 +
ℎ
𝑛 )
−2
(5-3)
where l and n are the length and width of specimens.
99
The diffusivity values at different
temperatures were determined by curving-fitting the experimental results using Equation
(5-2). The fitting results are summarized in Table 5-3, and the fitting curves are plotted in
Figure 1 as solid lines. All fits yield R
2
> 0.99, indicating accurate fitting.
98
Table 5-3. Curve fitting results using Equation (5-2) and (5-3).
Temperature (°C) M∞0 (%) k (10
-5
% s
-1/2
) D (10
-6
mm
2
s
-1
) Dx (10
-6
mm
2
s
-1
)
35 3.04±0.02 4.42±0.22 4.82±0.07 2.89±0.05
55 2.97±0.02 4.67±0.20 10.76±0.23 6.51±0.04
75 2.97±0.01 4.95±0.15 22.38±0.29 13.58±0.11
95 2.90±0.02 6.23±0.22 44.74±0.44 26.92±0.40
The diffusivity increased with temperature. Plotting ln(Dx) as a function of 1/T
yielded a linear relationship (solid black line in Figure 5-3) consistent with the Arrhenius
equation:
𝐷 𝑥 = 𝐷 𝑥 0
exp (−
𝐸 𝑎 𝑅𝑇
) (5-4)
where R is the gas constant, and the activation energy (Ea) and pre-exponential factor (Dx0)
were determined by the slope and intercept of the Arrhenius plot. Note that the moisture
saturation (M∞) values (shown in Figure 5-1) decreased as exposure temperature increased,
indicating an exothermic moisture absorption process. The moisture saturation and heat of
absorption (∆H) also followed an Arrhenius relationship:
100
𝑀 ∞
= 𝑀 0
exp (−
∆𝐻 𝑅𝑇
) (5-5)
where M0 is a pre-exponential factor. M0 and ∆H values were calculated using two
methods. In Method I, M0 and ∆H were determined directly from the Arrhenius plot of
experimental M∞ values (blue dashed line in Figure 5-2). In Method II, we applied the
equation:
101
99
ln (
𝑑 𝑀 𝑡 𝑑 √ 𝑡 ℎ
) = 𝑙𝑛 (4𝑀 0
√
𝐷 0
𝜋 ) −
𝐸 𝑎 2
+∆𝐻 𝑅𝑇
(5-6)
where dMt/d√ 𝑡 /h is the initial moisture absorbing rate that was determined by the slope of
the initial linear part in Figure 5-1, and D0 is the pre-exponential factor of the Arrhenius
equation of the apparent diffusivity, which was calculated using the method used for
determining Dx0. The Arrhenius temperature dependence of M∞, determined using
Equation (5-6), is plotted in Figure 5-2 (dash-dot blue line). All calculated values as
described above are summarized in Table 5-4.
Figure 5-2. Diffusivity and moisture saturation in water vs 1/T. The black circle symbols
are Dx listed in Table 5-3, and the blue diamond symbols are experimental results of M∞.
The black solid line is curve fit using Equation (5-4), and the blue dashed and dot-dash
lines are curve fits by Methods I and II, respectively.
100
Table 5-4. Arrhenius parameters of Equation (5-4) and (5-5).
Dx0 (10
-6
mm
2
s
-1
) 2.49
D0 (10
-6
mm
2
s
-1
) 4.03
Ea (kJ mol
-1
) 35.0
M0 (Method I) (%) 2.56
∆H (Method I) (kJ mol
-1
) -0.60
M0 (Method II) (%) 2.05
∆H (Method II) (kJ mol
-1
) -1.33
As shown in Figure 5-2, Method II overpredicts M∞, yielding values 1-10% greater
than those from Method I (from 0°C to 100°C). Using Method I, we assume no moisture
in the specimens prior to moisture exposure. However, the specimens were wet-machined,
and it is difficult to remove all moisture from polyimides, because a small portion of water
molecules are strongly bonded.
93, 94
On the other hand, despite the presence of residual
moisture, the initial absorbing rate (dMt/d√ 𝑡 /h) was controlled by temperature. Therefore,
the Arrhenius parameters determined by Method II are considered more accurate.
Conditioning at different relative humidity levels was performed on neat polymer
samples to determine the relationship between maximum moisture content and the ambient
humidity. The moisture saturation of neat polymer as a function of relative humidity ( ) is
shown in Figure 5-3 and described by
𝑀 ∞
= 𝑎 𝜙 𝑏 (5-7)
101
where a and b are constants.
99
In Equation (5-7), a = 0.0059 and b = 1.4 were determined
by curve fitting experimental results (black circles in Figure 5-3). The fitting yielded a R
2
> 0.99.
Figure 5-3. Moisture saturation vs relative humidity at 35°C.
The measured moisture absorption behavior yields a clearer understanding of TriA
X and how it is likely to behave in service. Equation (5-5) or (5-6) for M∞ describes the
maximum moisture content at a given temperature under 100% RH. However, moisture
saturation of polymers is more sensitive to humidity than to temperature.
99
Although the
relationship described by Equation (5-7) was determined at room temperature (35°C), the
relation has practical significance. Relative humidity in a given circumstance is readily
measured, so Equation (5-7) can be used as a predictive tool to estimate moisture content
of polyimide components in the field. Moreover, understanding the moisture absorption
behavior is essential when addressing issues governed by moisture diffusion kinetics. For
instance, the Arrhenius parameters of diffusion are critical for prediction of moisture-
102
induced damage, such as blistering
97
and microcracking,
96
using existing mathematical
models.
5.4.2 Hydrothermal aging
5.4.2.1 Hydrolysis
Both reversible and irreversible hydrolysis of imide units were detected after
extended hydrothermal aging of the neat polymer. The FTIR spectra before and after
hot/wet exposure are shown in Figure 5-4. To examine the reversibility of chemical
changes upon absorption or desorption, FTIR was performed on both dry and wet
specimens. After 2000-h aging at 95°C, the spectra remained nearly identical to those
acquired before exposure when comparing either the dry or wet sets only. However, the
intensity of the peak at 1601 cm
-1
was greater in the wet spectra (blue and red curves in
Figure 5-4). This peak corresponds to N-H deformation in amides and/or amines.
101
The
greater intensity of the peak in the wet spectra indicates hydrolysis of the imide rings upon
water absorption (Figure 5-5). After moisture desorption, the intensity decreased to the
original level in the dry spectra (green and black curves in Figure 5-4), providing evidence
that the hydrolysis is reversible. However, the FTIR results did not exclude the possibility
of irreversible hydrolysis upon hot/wet exposure.
103
Figure 5-4. FTIR spectra of TriA X.
Figure 5-5. Hydrolytic reactions in polyimides.
XPS analysis was performed on dry specimens before and after hydrothermal aging
to measure the composition change. After 2000-h hydrothermal aging and subsequent
drying, the O/C mole ratio increased from 0.154±0.006 (dry fresh polymer) to 0.162±0.007
(a 6% increase), indicating moisture remaining in the polymer network in the form of
hydrolytic products. The extent of irreversible hydrolysis was small compared to the
reversible reaction, because no change was detected in the dry FTIR spectrum after 2000-
h aging, but the irreversible hydrolysis might affect the physical properties of the polymer,
discussed next.
104
5.4.2.2 Thermal properties
The tan 𝛿 curves before and after hydrothermal aging are shown in Figure 5-6. Two
major transitions in terms of tan 𝛿 maxima were observed in the measured temperature
range (-90°C-450°C), labeled α and β in decreasing order of temperature in accordance
with conventional nomenclature.
102
Because of the difference in the magnitudes of α and β
transitions, the tan 𝛿 curves are shown in two temperature ranges (-90°C-300°C in Figure
5-6a and 200°C-300°C in Figure 5-6b) to present the complete curve shape for each
transition. α-relaxation represented the glass transition and corresponded to the onset of
long-range mainchain motion (Figure 5-6b). β-relaxation was a secondary transition below
Tg in amorphous polymers − a relaxation process active in the glassy state.
43
Various types
of motion, such as motions of side groups, restricted motion of the main chain, or motions
of end groups, can contribute to the β-transition.
102
Thus, changes in polymer network
structure are manifest through β-relaxation, which is sensitive to physical aging.
94, 103
The
wet α-peaks (dashed lines in Figure 5-6b) were broader than the dry α-peaks (solid lines in
Figure 5-6b), indicating greater heterogeneity
104
in the wet polymers resulting from the
hydrolysis. Meanwhile, the wet α-peaks shifted slightly (2°C) towards lower temperatures,
which can be attributed to both the hydrolytic reactions and to water plasticization effects
on the polymer network. The α-peaks after long-term aging (blue and red lines) were
slightly broader than the fresh peaks (black lines in either wet or dry tests, indicating (1)
increasing extent of hydrolysis with exposure time and (2) the presence of permanent
hydrolytic degradation after long-term exposure and subsequent desorption.
105
Figure 5-6. Tan 𝛿 curves of TriA X.
β-relaxation of neat TriA X split into two peaks, labelled β1 (30-50°C) and β2 (110-
150°C) in Figure 5-6a. This split demonstrates the inhomogeneity in the network. β1-peaks
were presented in the tan 𝛿 curves for both the dry specimens (solid curves) and the wet
fresh specimen (black dashed curve). However, after long-term aging, the β1-peaks
disappeared in the tan 𝛿 curves of the wet specimens, while the α-transition split, forming
a small peak at ~200°C labeled α’ (blue and red dashed lines in Figure 5-6a). This change
indicates that the short-range molecular motions in β1-relaxation were converted into
longer-range motions responsible for the α’-peaks. Although the mechanism of the
structural change is not fully understood, the reversible hydrolytic reactions were clearly
responsible for this conversion, because β1-relaxation reoccurred after moisture desorption,
while α’-peaks were absent. Despite that change in magnitude, β2-relaxation exhibited only
a minor influence from the hydrolytic reactions. The interpretation of the transitions in tan
𝛿 curves can be useful as a tool to identify physical/chemical reactions induced by
environmental exposure, as well as the resulting change in polymer structure, which could
influence other properties, e.g., mechanical properties, as discussed in the next section.
106
The Tg (peak of α transition) and Td5% values of the neat polymer are shown in
Figure 5-7 (Td5% = temperature of 5% weight loss). The Td5% value was unchanged after
aging, while the wet Tg’s exhibited negligible reduction (0.6%) compared to the
corresponding dry values, and after 2000-h aging, the decrease in both dry and wet Tg was
negligible (0.6%). The nearly 100% retention of Tg upon moisture exposure is notable. In
contrast, the effects of moisture plasticization and hydrolysis on conventional polyimides
(e.g. PMR-15, PETI-5, K3B, AFR-700B, etc) are typically much greater, with 15-20%
reduction in the wet Tg’s relative to the dry Tg values.
19, 59, 92
Moreover, significant
permanent hydrolytic degradation of imide units has been reported in water uptake studies
below 100°C, which can cause a permanent decrease in Tg.
92, 101
The effects of plasticization depend on the physical structure of a polymer. The
TriA X polyimide features an asymmetric/nonplanar backbone structure. The pendent
phenyl group hinders rotation of the diphenyl ether linkage in the ODA moiety, raising the
internal rotation energy barrier of the backbone, increasing the chain stiffness. Although
the absorbed water molecules generated swelling stress and tended to make the polymer
network more open, the rigid polymer chains retarded the structure relaxation process, as
evidenced by the nearly flat slope of the second absorption stage in Figure 5-1. Even after
moisture saturation, the stiff chains continued to dominate the Tg. On the other hand, the
~100% retention of Tg upon moisture exposure indirectly demonstrated the degree of
depolymerization in hydrolysis. Chain scissions (the second reaction in Figure 5-5) were
unlikely to participate in the hydrolysis of TriA X, because chain scissions would cause a
more significant Tg knockdown.
92
Thus, polyamic acid linkages were the major hydrolytic
products in the hydrolysis of imide rings (the first reaction in Figure 5-5). Meanwhile,
107
given the negligible Tg decease, the formation of polyamic acid linkages was less than 30-
40%.
93
Figure 5-7. Tg and Td5% of TriA X.
5.4.2.3 Mechanical behavior
Mechanical tests were performed on neat polymer and composites at room
temperature, both before and after moisture exposure, and the tensile properties are
summarized in Figure 5-8. Tensile modulus exhibited no change upon hot/wet exposure.
In tests of the fresh polymer, wet specimens exhibited 1% lower strength, but 28% greater
strain-to-failure than the dry specimens. The increased ductility translated to 30% greater
toughness. Toughness, the energy absorbed during loading prior to failure, was determined
from the area under the tensile strain-stress curve.
43
The tougher mechanical response of
the wet polymer is consistent with the moisture plasticization effects,
105
in that the absorbed
moisture effectively acted as a lubricant, easing the movement of the polymer chains and
pushing them further apart.
43
After 2000 h of hot/wet exposure, the dry strength increased
by 5%, while the wet strength decreased by 4% relative to the fresh group. In contrast, the
108
dry and wet ductility dropped by 29% and 51%, respectively, after 2000-h aging. These
decrements led to corresponding reductions in the dry and wet toughness of 27% and 56%,
respectively. After long-term aging, the difference between the dry and wet toughness was
less significant than that in the fresh group, indicating that hydrolysis might have affected
the mechanical behavior. Thus, the 9% reduction in wet strength after 2000-h aging was
attributed to both reversible hydrolytic reactions and to moisture plasticization.
Figure 5-8. Tensile properties of neat polymer (TriA X).
The decrease in dry/wet toughness and ductility upon aging can be attributed to the
change in β-relaxation. The stress-strain curves (Figure 5-9) of the 2000-h group were
nearly identical to those of the 1000-h group except for a slight difference (9-19%) in
strain-to-failure, which was attributed to normal measurement variance, given the large
scatter of the measured values. This phenomenon indicated that mechanical properties had
stabilized before or at 1000 h. Similar stabilization was also observed in the tan 𝛿 curves,
where the peak positions of the transitions exhibited only minor change after 1000-h aging
109
(Figure 5-6). This stabilization indirectly supports a connection between mechanical
properties and β-relaxation. Johari indicated that the β-relaxation process in a mechanically
rigid glass can be attributed to loosely packed regions with high mobility.
106
Such highly
mobile regions would also contribute to the chain motion in mechanical behavior. After
long-term aging, the decreased area of β-relaxation, especially β1-peaks (Figure 5-6a),
indicated that loosely packed regions were at least partially consumed, resulting in
segments or groups that were less mechanically active. This network structure change can
be attributed to hydrolysis and plasticization. Bao et al indicated that the structural change
in bismaleimide polymers induced by moisture absorption is irreversible.
94
Such
irreversible structural change is consistent with the observation that neither ductility nor
toughness were recovered after drying, although the details of the moisture-induced
physical structure change are not yet fully understood. More investigation is required to
identify the groups and/or motions that were altered by moisture absorption, and how such
changes affect mechanical properties.
Figure 5-9. Stress-stain curves of TriA X in tensile tests.
110
The effects of hydrothermal aging on composites are matrix- and interface-
dominated, because only the matrix absorbs moisture, and matrix degradation occurs
primarily along matrix-fiber interfaces.
95, 107
Matrix- and interface-dominated mechanical
properties are expected to be more affected by moisture than fiber-dominated properties,
and SBS strength is good example of an matrix- and interface-dominated property.
66
Thus,
SBS strength was selected to characterize the change of the mechanical performance of
T650-35 8HS/TriA X composites upon moisture exposure. After 2000-h moisture
exposure, the SBS strength decreased by 7-13%, and also in most cases (except the unaged
[0/+45/-45/90]s group), the wet SBS strength was 1-6% greater than the dry strength
(Figure 5-10). The slight reduction in SBS strength of the composites was attributed to the
decrease in matrix toughness, as discussed above. The slightly (2-5%) greater SBS strength
retention observed in the aged [0]8 laminates compared to [0/+45/-45/90]s laminates was
attributed to effects of lay-up sequences – the SBS strength of quasi-isotropic lay-ups is
more matrix-controlled than that of cross-ply lay-ups.
74
Figure 5-10. SBS strength of T650-35 8HS/TriA X composites.
111
5.5 Conclusions
The effects of moisture on TriA X were investigated to understand the stability of
this polymer in potential high-temperature service environments. Short-term water uptake
study at different temperature and relative humidity was performed to gain an accurate
understanding of the moisture absorption behavior of the neat polymer. The polyimide
exhibited a two-stage moisture diffusion behavior, where the first stage was Fickian
diffusion-controlled, while the second stage was governed by the moisture-induced
plasticization. Mathematical models of moisture uptake as functions of temperature and
humidity were determined and compared to measured behavior. Such models can be used
as tools to estimate the moisture content of a component fabricated with the polyimide in
the field. More importantly, this work built a foundation for future study on moisture-
induced damage in composites, because the Arrhenius parameters determined here are
essential to kinetic models of damage prediction.
Long-term moisture exposure at 95°C caused minimal permanent chemical
damage, although it did cause a reversible hydrolysis reaction of imide rings. The Tg and
Td5% of the neat polymer exhibited no change after 2000-h aging, demonstrating long-term
hydrothermal stability. Furthermore, the Tg was largely unaffected by moisture exposure
(in comparison to the dry Tg values), a finding attributed to the unusual stiffness of the
polymer chains arisen by the asymmetric backbone structure, and effectively suppresses
the plasticization effects of water molecules on Tg. Given the ~100% retention of Tg after
hot/wet exposure, chain scissions appear to be absent or minimal in the reversible
hydrolytic reactions.
112
Moisture exposure had a stronger effect on mechanical properties, although tensile
modulus and strength exhibited little effect. Prior to long-term aging, moisture
plasticization exhibited positive effects, increasing the ductility and toughness. However,
ductility and toughness degradation arose after long-term hot/wet exposure. After 2000-h
hydrothermal aging, the ductility decreased by 30-50%, resulting in a reduction in
toughness of 30%-55%. The decrease in ductility and toughness was attributed to the long-
term effects of hydrolysis and moisture plasticization. The mechanically active regions in
the polymer network were consumed by moisture-induced chemical and physical reactions,
evidenced by the decreased area of the β-transition in tan δ curves. The network structure
change was irreversible, resulting in permanent mechanical degradation. The findings
constitute a case study demonstrating use of tan δ to analyze structural change in polymer
networks and resultant mechanical degradation upon physical/chemical aging. However,
more investigation is required to establish more clearly the relationship between structural
relaxations and mechanical behavior.
Although the ductility of the neat polymer decreased to 11.06% (dry) after 2000-h
of aging, the retained ductility exceeded that of fresh conventional polyimides (e.g., 1.5%
for dry PMR-15
59
and 2.41% for dry AFR-PE-4
19
). Moreover, after 1000-h of aging,
continued exposure did not degrade the tensile properties further, and a steady state was
achieved, indicating limited hydro-degradation on mechanical properties as aging time
increased. In addition, after 2000-h aging, ~90% of the composite SBS strength was
retained. The absence of thermal property deterioration, the negligible changes in chemical
structure and the limited mechanical degradation demonstrate promising stability for long-
113
term performance in humid service conditions, and potential for long service life in hot-
wet conditions.
114
Chapter 6. Conclusions and recommendations
6.1 Concluding remarks
The relationship between macroscopic properties and molecular structure was
established for a new imide resin system – TriA X. The asymmetric and nonplanar p-ODA
units in the system increased the irregularity of the polymer backbones and the interchain
distance, yielding loosely packed amorphous structure and consequently increasing the
processability of the resin system and the ductility of the cured polyimide. The pendent
phenyl group of p-ODA created steric hindrance in the backbone and hence increased the
chain rigidity, partially counterbalancing the adverse influence of loose chain packing on
Tg and elastic modulus and hence resulting in a high Tg at 362°C (measured by DSC) and
comparable Young’s modulus with conventional polyimides. The structure-property-
processing relationship determined in Chapter 1 not only presents a materials paradigm in
an asymmetric polymer for new material design and development in the future but also
advances understanding of this new imide resin system, providing the fundamental
knowledge base for future study on the polyimide and associated applications.
A polymer science-based approach was developed to determine a suitable process
for fabricating high quality composites with TriA X. This approach is based on a
comprehensive thermal characterization on resin flow, volatile generation and cross-
linking, through which connections between processability and complex chemical/physical
reactions were established to select parameters (e.g. B-staging temperature and timing of
consolidation pressure application) for a molding cycle suitable for the imide resin system.
115
This approach is applicable to any PMR resins and especially useful to systems with partial
temperature-range overlap between imidization and cross-linking. The designed molding
process through this approach yielded well consolidated TriA X laminates with < 0.1%
porosity and matrix-dominated properties that surpass those of conventional polyimide
composites.
TriA X composites passed an industrial (Boeing) thermal cycling test with
negligible microcracking. The high reinstate to microcracking in thermal cycling test is
attributed to the high strength and thermal oxidative stability of the matrix. The high tensile
strength of TriA X matrix is sufficient to withstand the thermal stresses induced by cyclic
temperatures, and meanwhile, the matrix exhibited no thermal oxidative degradation
during thermal cycling in air. The > 90% retention of SBS strength after 2000 cycles was
attributed to the absence of microcracks, while the < 10% reduction was attributed to slight
matrix degradation caused primarily by thermal driven fatigue and possibly or secondarily
by creep at high temperatures. The high instance to thermal cycling of TriA X composites
is likely to expand the design space of polyimide composites for longer-term high-
temperature structural applications.
The moisture absorption behaviors and long-term hydrothermal aging of neat TriA
X and associated composites were investigated in Chapter 3. The neat polyimide exhibited
a two-stage exothermic moisture diffusion process, in which the first stage was Fickian
diffusion-controlled, while the second stage was governed by the moisture plasticization.
Mathematical models of moisture uptake as functions of temperature and humidity were
determined. Such models can be used as tools to estimate the moisture content of a
component fabricated with the polyimide in the field. In the long-term hydrothermal aging
116
study, revisable hydrolysis (with a trace of irreversible reactions) of imide units was
detected while the neat polyimide exhibited no thermal property deterioration and limited
mechanical degradation as continued hot/wet exposure. The observed decrease in ductility
and toughness was attributed to the long-term effects of hydrolysis and moisture
plasticization. After 2000-h hydrothermal aging, the residual tensile properties of the neat
polymer still far surpass those of fresh conventional polyimides and the retention of the
composite SBS strength was > 90%. The high hydrothermal stability of TriA X
demonstrates potential for long service life in hot-wet conditions.
Overall, TriA X resin system demonstrates capabilities to resolve many problems
associated with the current polyimide matrices. The high ductility and toughness of cured
TriA X offer a solution to the brittleness of conventional polyimides. The thermal-history
independent amorphous structure addresses the issues (e.g. narrow process window and
uneven polymer structure) of semi-crystalline polyimides in composite processing. More
importantly, the ultrahigh stability of TriA X composites in thermal cycling and humid
environment shows potentials to expand service conditions for composite applications.
6.2 Recommendations for future work
Although systemic investigation has been conducted on TriA X and associated
composites in this dissertation, concerns and questions still remain. For example, in the
process development for TriA X composites, consolidation pressure was not optimized yet.
The designed molding cycle utilized a high compression pressure at 1.72 MPa that is
acceptable for hot press molding but exceeds the capability of most industrial autoclaves.
Meanwhile, we observed a thickness dependence of the minimal consolidation pressure,
117
where thicker laminates required greater pressure. Thus, understanding the relationship
between consolidation pressure and laminate thickness is desired for the optimization of
pressure, especially important in the manufacturing of large-scale parts and complex
shapes.
TriA X composites exhibited high resistance to microcracking upon the exposure
to cyclic temperatures, and conventional image analysis did not reveal microcracks in
thermally cycled laminates, but a few microcracks were detected in microCT scans. These
cracks were attributed to additional loading (during sample sectioning) after thermal
cycling. To avoid the interference of artificial damage in microCT scans, future microCT
samples should be prepared prior to thermal conditioning. On the other hand, the
microcracks observed in microCT scan revealed that laminates after 2000 cycles were more
susceptible to damage from sectioning than from interlaminar shear loading. Therefore,
additional investigation is needed to more fully understand the mechanical behavior of
TriA X composites after thermal cycling. In addition to local fatigue and creep, curved
fiber geometry of 8HS fabric also affected the performance of the matrix and decreased the
microcracking resistance of the composites after thermal cycling. Thus, non-woven
reinforcement is recommended for TriA X composites in the future work.
In Chapter 5, tan δ curves were utilized as a tool to analyze structural change in
polymer networks and resultant mechanical degradation upon moisture exposure. The
current results only led to a conclusion that the decrease in ductility is attributed to the
consummation of mechanically active regions in the polymer network that are responsible
to the β-transition in tan δ curves. Questions, e.g. what groups and/or motions were altered
by moisture absorption, and how such changes affect mechanical properties, still remain.
118
More investigation is required to establish more clearly the relationship between structural
relaxations and mechanical behavior. Future work should also include studies of moisture-
induced damage (e.g. microcracking and blistering) on the polyimide composites. The
Arrhenius parameters determined in Chapter 5 are essential to kinetic models of such
damage prediction, building a good foundation for future study on moisture-induced
damage in composites.
119
References
1. Mcconnell VP. Resins for the Hot Zone, Part I: Polyimides: Composites World;
2009 [Available from: https://www.compositesworld.com/articles/resins-for-the-hot-zone-
part-i-polyimides.
2. Iredale RJ, Ward C, Hamerton I. Modern advances in bismaleimide resin
technology: A 21st century perspective on the chemistry of addition polyimides. Progress
in Polymer Science 2017; 69:1-21.
3. Wilson D. PMR-15 processing, properties and problems—a review. Polymer
International 1988; 20(5):405-16.
4. Polyimides: Polymer Science Learning Center; 2016 [Available from:
http://www.pslc.ws/macrog/imide.htm.
5. Abadie MJ High Performance Polymers: Polyimides Based: from Chemistry to
Applications. Rijeka, Croatia: InTech, 2012.
6. Serafini TT. High-temperature resins. Handbook of Composites. New York, NY:
Van Nostrand Reinhold Company Inc., 1982. p. 89-114.
7. Liaw D-J, Wang K-L, Huang Y-C, Lee K-R, Lai J-Y, Ha C-S. Advanced polyimide
materials: syntheses, physical properties and applications. Progress in Polymer Science
2012; 37(7):907-74.
8. Liou H-C, Willecke R, Ho PS. Study of out-of-plane elastic properties of PMDA-
ODA and BPDA-PDA polyimide thin films. Thin solid films 1998; 323(1-2):203-8.
9. Margolis J Engineering thermoplastics: properties and applications. Boca Raton,
FL: CRC Press, 1985.
120
10. Serafini TT, Delvigs P, Lightsey GR. Thermally stable polyimides from solutions
of monomeric reactants. Journal of Applied Polymer Science 1972; 16(4):905-15.
11. Jang BZ, Pater RH, Soucek MD, Hinkley JA. Plastic deformation mechanisms in
polyimide resins and their semi-interpenetrating networks. Journal of Polymer Science
Part B: Polymer Physics 1992; 30(7):643-54.
12. Mills JS, Herakovich CT, Davis Jr JG. Transverse microcracking in Celion
6000/PMR-15 graphite-polyimide. Technical Report for National Aeronautics and Space
Administration. USA: Virginia Polytechnic Institute and State University; 1979 December.
Report No.: VPI-E-79-35.
13. Owens G, Schofield S. Thermal cycling and mechanical property assessment of
carbon fibre fabric reinforced PMR-15 polyimide laminates. Composites science and
technology 1988; 33(3):177-90.
14. Hergenrother P, Smith Jr J. Chemistry and properties of imide oligomers end-
capped with phenylethynylphthalic anhydrides. Polymer 1994; 35(22):4857-64.
15. Hergenrother P, Connell J, Smith Jr J. Phenylethynyl containing imide oligomers.
Polymer 2000; 41(13):5073-81.
16. Smith Jr JG, Connell JW, Hergenrother P, Criss J. Resin transfer moldable
phenylethynyl containing imide oligomers. Journal of composite materials 2002;
36(19):2255-65.
17. Cano RJ, Jensen BJ. Effect of molecular weight on processing and adhesive
properties of the phenylethynyl-terminated polyimide LARC™-PETI-5. The Journal of
Adhesion 1997; 60(1-4):113-23.
121
18. Wilhite AW, Shaw RJ. An overview of NASA’s high-speed research program.
Hampton, VA: NASA Langley Research Center; 2000 September. Accession No.:
ADA572142.
19. Lincoln JE (2001). Structure-property-processing relationships and the effects of
physical structure on the hygrothermal durability and mechanical response of polyimides.
PhD Thesis, Michigan State University, USA.
20. Abadie MJ, Voytekunas VY, Rusanov AL. State of the art organic matrices for high
performance composites: a review. 2006.
21. Harvey BG, Yandek GR, Lamb JT, Eck WS, Garrison MD, Davis MC. Synthesis
and characterization of a high temperature thermosetting polyimide oligomer derived from
a non-toxic, sustainable bisaniline. RSC Advances 2017; 7(37):23149-56.
22. Ratta V (1999). Crystallization, morphology, thermal stability and adhesive
properties of novel high performance semicrystalline polyimides. PhD Thesis, Virginia
Polytechnic Institute and State University, USA.
23. Lincoln JE, Morgan RJ, Shin EE. Effect of thermal history on the deformation and
failure of polyimides. Journal of Polymer Science Part B: Polymer Physics 2001;
39(23):2947-59.
24. Hsiao SH, Lin KH. Polyimides derived from novel asymmetric ether diamine.
Journal of Polymer Science Part A: Polymer Chemistry 2005; 43(2):331-41.
25. Chern Y-T, Twu J-T, Chen J-C. High Tg and high organosolubility of novel
polyimides containing twisted structures derived from 4-(4-amino-2-chlorophenyl)-1-(4-
aminophenoxy)-2, 6-di-tert-butylbenzene. European Polymer Journal 2009; 45(4):1127-
38.
122
26. Chern Y-T, Tsai J-Y. Low dielectric constant and high organosolubility of novel
polyimide derived from unsymmetric 1, 4-bis (4-aminophenoxy)-2, 6-di-tert-butylbenzene.
Macromolecules 2008; 41(24):9556-64.
27. Yokota R, Yamamoto S, Yano S, Sawaguchi T, Hasegawa M, Yamaguchi H, et al.
Molecular design of heat resistant polyimides having excellent processability and high
glass transition temperature. High Performance Polymers 2001; 13(2):S61-S72.
28. Ogasawara T, Ishikawa T, Yokota R, Ozawa H, Taguchi M, Sato R, et al.
Processing and properties of carbon fiber reinforced triple-A polyimide (Tri-A PI) matrix
composites. Advanced Composite Materials 2002; 11(3):277-86.
29. Connell JW, Smith Jr JG, Hergenrother PM, Criss JM. High temperature transfer
molding resins: laminate properties of PETI-298 and PETI-330. High Performance
Polymers 2003; 15(4):375-94.
30. Miyauchi M, Kazama K-i, Sawaguchi T, Yokota R. Dynamic tensile properties of
a novel Kapton-type asymmetric polyimide derived from 2-phenyl-4,4′-diaminodiphenyl
ether. Polymer journal 2011; 43(10):866-8.
31. Miyauchi M, Ishida Y, Ogasawara T, Yokota R. Novel phenylethynyl-terminated
PMDA-type polyimides based on KAPTON backbone structures derived from 2-phenyl-
4, 4 ′-diaminodiphenyl ether. Polymer journal 2012; 44(9):959-65.
32. Miyauchi M, Ishida Y, Ogasawara T, Yokota R. Highly soluble phenylethynyl-
terminated imide oligomers based on KAPTON-type backbone structures for carbon fiber-
reinforced composites with high heat resistance. Polymer journal 2013; 45(6):594-600.
123
33. Zhang Y, Miyauchi M, Nutt S. Structure and properties of a phenylethynyl-
terminated PMDA-type asymmetric polyimide. High Performance Polymers 2018. DOI:
10.1177/0954008318762592.
34. Dunsky C, Tacheron P, Hashish M. Waterjet Techniques for Composite Material
Jet Engine Component Repair. Technical Report. Kent, WA: Quest Integrated Inc.; 1996
June. Report No.: 716.
35. Kool G. Current and future materials in advanced gas turbine engines. ASME 1994
International Gas Turbine and Aeroengine Congress and Exposition, The Hague,
Netherlands,June 13-16 1994.
36. Connell JW, Smith Jr JG, Hergenrother PM, Rommel ML. Neat resin, adhesive and
composite properties of reactive additive/PETI-5 blends. High Performance Polymers
2000; 12(2):323-34.
37. Ishida Y, Ogasawara T, Yokota R. Development of highly soluble addition-type
imide oligomers for matrix of carbon fiber composite (I): imide oligomers based on
asymmetric biphenyltetracarboxylic dianhydride and 9, 9-bis (4-aminophenyl) fluorene.
High performance polymers 2006; 18(5):727-37.
38. Morgan R, Shin E, Lincoln J, Zhou J. Overview of polymer matrix composites
performance and materials development for aerospace applications. SAMPE Journal 2001;
37(2):102-7.
39. Wilson D. Polyimide matrix composites: candidates for high speed commercial
aircraft or not. High Performance Polymers 1991; 3(2):73-88.
124
40. Serafini T, Hanson M. Environmental effects on graphite fiber reinforced PMR-15
polyimide. Composites for Extreme Environments. West Conshohocken, PA: ASTM
International, 1982.
41. Houlihan F, Bachman B, Wilkins Jr C, Pryde C. Synthesis and characterization of
the tert-butyl ester of the oxydianiline/pyromellitic dianhydride polyamic acid.
Macromolecules 1989; 22(12):4477-83.
42. Zhang Y, Jain A, Grunenfelder LK, Miyauchi M, Nutt S. Process development for
phenylethynyl-terminated PMDA-type asymmetric polyimide composites. High
Performance Polymers 2017; 30(6):731-41.
43. Cowie JMG, Arrighi V Polymers: chemistry and physics of modern materials. Boca
Raton, FL, USA: CRC press, 2007.
44. Omote T, Yamaoka T, Koseki K. Preparation and properties of soluble and
colorless fluorine-containing photoreactive polyimide precursors. Journal of applied
polymer science 1989; 38(3):389-402.
45. D638. Standard Test Method for Tensile Properties of Plastics. West
Conshohocken, PA: ASTM International, 2014.
46. ASTM D6272. Standard test method for flexural properties of unreinforced and
reinforced plastics and electrical insulating materials by four-point bending. West
Conshohocken, PA: ASTM International, 2017.
47. Miller RL, Boyer RF, Heijboer J. X-ray scattering from amorphous acrylate and
methacrylate polymers: evidence of local order. Journal of Polymer Science: Polymer
Physics Edition 1984; 22(12):2021-41.
125
48. Connell JW, Smith Jr JG, Hergenrother P. Oligomers and polymers containing
phenylethynyl groups. Journal of Macromolecular Science, Part C: Polymer Reviews
2000; 40(2-3):207-30.
49. Li Y, Murphy LA, Lincoln JE, Morgan RJ. Phenylethynyl End-Capped Fluorinated
Imide Oligomer AFR-PEPA-N: Morphology and Processibility Characteristics.
Macromolecular Materials and Engineering 2007; 292(1):78-84.
50. Li Y, Obando N, Tschen F, Morgan R. Thermal analysis of phenylethynyl end-
capped fluorinated imide oligomer AFR-PEPA-4. Journal of thermal analysis and
calorimetry 2006; 85(1):125-9.
51. Gentz M, Armentrout D, Rupnowski P, Kumosa L, Shin E, Sutter J, et al. In-plane
shear testing of medium and high modulus woven graphite fiber reinforced/polyimide
composites. Composites Science and Technology 2004; 64(2):203-20.
52. Takekoshi T, Terry J. High-temperature thermoset polyimides containing
disubstituted acetylene end groups. Polymer 1994; 35(22):4874-80.
53. Sroog C, Endrey A, Abramo S, Berr C, Edwards W, Olivier K. Aromatic
polypyromellitimides from aromatic polyamic acids. Journal of Polymer Science Part A:
General Papers 1965; 3(4):1373-90.
54. Wieneke JU, Staudt C. Thermal stability of 6FDA-(co-) polyimides containing
carboxylic acid groups. Polymer Degradation and Stability 2010; 95(4):684-93.
55. Matsuura T, Hasuda Y, Nishi S, Yamada N. Polyimide derived from 2,2'-bis
(trifluoromethyl)-4,4'-diaminobiphenyl. 1. Synthesis and characterization of polyimides
prepared with 2,2'-bis (3,4-dicarboxyphenyl) hexafluoropropane dianhydride or
pyromellitic dianhydride. Macromolecules 1991; 24(18):5001-5.
126
56. Mehdipour-Ataei S, Sarrafi Y, Hatami M. Novel thermally stable polyimides based
on flexible diamine: synthesis, characterization, and properties. European polymer journal
2004; 40(9):2009-15.
57. Gibbs HH, Breder C. High-temperature laminating resins based on melt fusible
polyimides. Copolymers, Polyblends, and Composites. Washington, DC: American
Chemical Society, 1975. p. 442-57.
58. McClung AJ, Ruggles-Wrenn MB. Strain rate dependence and short-term
relaxation behavior of a thermoset polymer at elevated temperature: experiment and
modeling. Journal of pressure vessel technology 2009; 131(3):031405.
59. CYCOM
®
2237 polyimide resin system. Woodland Park, NJ: Cytec Solvay Group,
2012.
60. Scola DA, Vontell JH. High temperature polyimides, chemistry and properties.
Polymer Composites 1988; 9(6):443-52.
61. Hancox N. Thermal effects on polymer matrix composites: Part 1. Thermal cycling.
Materials & Design 1998; 19(3):85-91.
62. ASTM D3039. Standard Test Method for Tensile Properties of Polymer Matrix
Composite Materials. West Conshohocken, PA: ASTM International, 2014.
63. ASTM D7264. Standard test method for flexural properties of polymer composite
materials. West Conshohocken, PA: ASTM International, 2015.
64. ASTM D2344. Standard test method for short-beam strength of polymer matrix
composite materials and their laminates. West Conshohocken, PA: ASTM International,
2016.
127
65. Hou T, Wilkinson S, Johnston N, Pater R, Schneider T. Processing and properties
of IM7/LARCTM-RP46 polyimide composites. High Performance Polymers 1996; 8:491-
506.
66. Ahmed KS, Vijayarangan S. Tensile, flexural and interlaminar shear properties of
woven jute and jute-glass fabric reinforced polyester composites. Journal of materials
processing technology 2008; 207(1-3):330-5.
67. AFRPE
®
-4 Prepreg Data Sheet. Moraine, OH: Performance Polymer Solutions
Inc.,
68. Naebe M, Abolhasani MM, Khayyam H, Amini A, Fox B. Crack damage in
polymers and composites: A review. Polymer reviews 2016; 56(1):31-69.
69. Kravchenko OG, Kravchenko SG, Pipes RB. Chemical and thermal shrinkage in
thermosetting prepreg. Composites Part A: Applied Science and Manufacturing 2016;
80:72-81.
70. Khoun L, Hubert P. Investigation of the dimensional stability of carbon epoxy
cylinders manufactured by resin transfer moulding. Composites Part A: Applied Science
and Manufacturing 2010; 41(1):116-24.
71. White SR, Hahn H. Process modeling of composite materials: residual stress
development during cure. Part II. Experimental validation. Journal of composite materials
1992; 26(16):2423-53.
72. Takagaki K, Hisada S, Minakuchi S, Takeda N. Process improvement for out-of-
autoclave prepreg curing supported by in-situ strain monitoring. Journal of Composite
Materials 2017; 51(9):1225-37.
128
73. Rupnowski P, Kumosa M. Meso-and micro-stress analyses in an 8HS
graphite/polyimide woven composite subjected to biaxial in-plane loads at room
temperature. Composites science and technology 2003; 63(6):785-99.
74. Tompkins S, Williams SL. Effects of thermal cycling on mechanical properties of
graphite polyimide. Journal of Spacecraft and Rockets 1984; 21(3):274-80.
75. Datoo MH Mechanics of fibrous composites. London, England: Elsevier Science
Publishers, 1991.
76. Wilson D, Wells J, Hay J, Lind D, Owens G. Preliminary investigations into the
microcracking of PMR-15/graphite composites. I. Effect of cure temperature. SAMPE
Journal 1987; 23:35-42.
77. Shimokawa T, Katoh H, Hamaguchi Y, Sanbongi S, Mizuno H, Nakamura H, et al.
Effect of thermal cycling on microcracking and strength degradation of high-temperature
polymer composite materials for use in next-generation SST structures. Journal of
composite materials 2002; 36(7):885-95.
78. National Research Council. US Supersonic Commercial Aircraft: Assessing
NASA's High Speed Research Program. Washington, D.C: National Academies Press,
1997. pp. 72.
79. Benedikt B, Kumosa M, Predecki P, Kumosa L, Castelli M, Sutter J. An analysis
of residual thermal stresses in a unidirectional graphite/PMR-15 composite based on X-ray
diffraction measurements. Composites Science and Technology 2001; 61(14):1977-94.
80. ASTM E2113. Standard test method for length change calibration of
thermomechanical analyzers. West Conshohocken, PA: ASTM International, 2013.
129
81. ASTM E831. Standard test method for linear thermal expansion of solid materials
by thermomechanical analysis. West Conshohocken, PA: ASTM International, 2014.
82. Lafarie-Frenot M, Rouquie S. Influence of oxidative environments on damage in
c/epoxy laminates subjected to thermal cycling. Composites Science and Technology 2004;
64(10-11):1725-35.
83. Sauer J, Richardson G. Fatigue of polymers. International journal of fracture 1980;
16(6):499-532.
84. Vujosevic M, Krajcinovic D. Creep rupture of polymers—a statistical model.
International journal of solids and structures 1997; 34(9):1105-22.
85. Johnson LL, Eby R, Meador MAB. Investigation of oxidation profile in PMR-15
polyimide using atomic force microscope (AFM). Polymer 2003; 44(1):187-97.
86. Tandon G, Pochiraju K, Schoeppner G. Thermo-oxidative behavior of high-
temperature PMR-15 resin and composites. Materials Science and Engineering: A 2008;
498(1-2):150-61.
87. Scola D, Vontell J. Mechanical properties and mechanism of the degradation
process of 316°C isothermally aged graphite fiber/PMR-15 composites. Polymer
Engineering & Science 1991; 31(1):6-13.
88. Lu Y, Tandon G, Putthanarat S, Schoeppner G. Nanoindentation strain rate
sensitivity of thermo-oxidized PMR-15 polyimide. Journal of materials science 2009;
44(8):2119-27.
89. Barjasteh E, Bosze E, Tsai Y, Nutt S. Thermal aging of fiberglass/carbon-fiber
hybrid composites. Composites Part A: Applied Science and Manufacturing 2009;
40(12):2038-45.
130
90. Red C. The Outlook for Thermoplastics in Aerospace Composites, 2014-2023:
Composites World; 2014 [Available from: https://www.compositesworld.com/articles/the-
outlook-for-thermoplastics-in-aerospace-composites-2014-2023.
91. Hou T-H. Processing robustness for a phenylethynyl-terminated polyimide
composite. Journal of applied polymer science 2006; 100(4):3212-21.
92. Shin EE, Morgan RJ, Zhou J. Hydrolytic degradation mechanisms and kinetics of
polyimides for advanced composites. Society for the advancement of material and process
engineering conference proceedings, Long Beach, CA,21-25 May 2000.
93. Shin EE, Morgan RJ, Zhou J, Lincoln J, Jurek R, Curliss DB. Hygrothermal
durability and thermal aging behavior prediction of high-temperature polymer-matrix
composites and their resins. Journal of Thermoplastic Composite Materials 2000;
13(1):40-57.
94. Bao L-R, Yee AF, Lee CY-C. Moisture absorption and hygrothermal aging in a
bismaleimide resin. polymer 2001; 42(17):7327-33.
95. Burcham LJ, Vanlandingham MR, Eduljee RF, Gillespie Jr JW. Moisture effects
on the behavior of graphite/polyimide composites. Polymer composites 1996; 17(5):682-
90.
96. Han M-H, Nairn JA. Hygrothermal aging of polyimide matrix composite laminates.
Composites Part A: Applied science and manufacturing 2003; 34(10):979-86.
97. Czabaj MW, Zehnder AT, Chuang KC. Blistering of moisture saturated
graphite/polyimide composites due to rapid heating. Journal of Composite Materials 2009;
43(2):153-74.
131
98. Greenspan L. Humidity fixed points of binary saturated aqueous solutions. Journal
of research of the national bureau of standards 1977; 81(1):89-96.
99. Shen C-H, Springer GS. Moisture absorption and desorption of composite
materials. Journal of composite materials 1976; 10(1):2-20.
100. El-Sa'ad L, Darby M, Yates B. Moisture absorption by epoxy resins: the reverse
thermal effect. Journal of Materials Science 1990; 25(8):3577-82.
101. Bao L-R, Yee AF. Effect of temperature on moisture absorption in a bismaleimide
resin and its carbon fiber composites. Polymer 2002; 43(14):3987-97.
102. Bower DI An introduction to polymer physics. New York, NY, USA: Cambridge
University Press, 2002. pp. 212-213.
103. Mikols W, Seferis J, Apicella A, Nicolais L. Evaluation of structural changes in
epoxy systems by moisture sorption-desorption and dynamic mechanical studies. Polymer
Composites 1982; 3(3):118-24.
104. Yu Q, Zhou M, Ding Y, Jiang B, Zhu S. Development of networks in atom transfer
radical polymerization of dimethacrylates. Polymer 2007; 48(24):7058-64.
105. Immergut EH, Mark HF. Principles of Plasticization. Plasticization and Plasticizer
Processes. Advances in Chemistry. 48. Washington, DC, USA: American Chemical
Society, 1965. p. 1-26.
106. Johari G. Localized molecular motions of β-relaxation and its energy landscape.
Journal of non-crystalline solids 2002; 307:317-25.
107. Bao L-R, Yee AF. Moisture diffusion and hygrothermal aging in bismaleimide
matrix carbon fiber composites—part I: uni-weave composites. Composites Science and
Technology 2002; 62(16):2099-110.
Abstract (if available)
Abstract
A new type of polyimide, designated TriA X, has been developed for high-temperature composite applications. TriA X is a polymerized monomeric reactant (PMR) type polyimide derived from pyromellitic dianhydride (PMDA), 2-phenyl-4,4’-diaminodiphenyl ether (p-ODA) and phenylethynyl phthalic anhydride (PEPA). In this dissertation, a TriA X resin (with degree of polymerization n = 7 in the imide oligomer) was investigated for developing a new polyimide composite and addressing the current issues associated with conventional polyimide matrices. ❧ To advance understanding of this new imide resin system, comprehensive characterization was preformed to investigate polymer structure, processability, thermal and mechanical properties and establish the relationship between the molecular structure and those properties. TriA X features an asymmetric, irregular, nonplanar backbone. Both the imide oligomers and the cross-linked polyimides of TriA X exhibited loose-packed amorphous structures, independent of thermal processing. The peculiar structures were attributed to the asymmetric backbone, which effectively prevents the formation of closed-packed chain stacking typically observed in polyimides. The imide oligomers exhibited a lower melt viscosity than a control imide oligomer (symmetric and semi-crystalline), indicating a higher chain mobility above the glass transition temperature (Tg). The cured polyimide exhibited a Tg = 362℃ and a decomposition temperature (Td) = 550℃. The cross-linked TriA X exhibited exceptional toughness and ductility (e.g., 15.1% at 23℃) for a polyimide, which was attributed to the high molecular weight oligomer and loose-packed amorphous structure. The thermal and mechanical properties of TriA X surpass those of PMR-15 and AFR-PE-4. ❧ The asymmetric and non-planar backbone structure endows cured TriA X with amorphous structure and high toughness that are attractive properties for composite applications. Processability of the imide resin system and performance of associated carbon fiber composites were investigated. Rheological measurements were performed on an oligomer with a low degree of imidization to understand the chemo-rheology of the resin system and determine a suitable B-staging temperature. A composite molding cycle was designed, which yielded fully-consolidated woven carbon fiber laminates. Void contents in panels produced with this molding cycle were < 0.1% as measured by image analysis of polished sections, and < 0.2% as measured by X-ray micro-computed tomography. Matrix-dominated mechanical properties of composites fabricated with TriA X exceeded those of PMR-15 composites. These mechanical properties and a measured Tg of 367℃ indicate potential for use of this resin system in high-temperature composites. ❧ To further explore the potential of this new polyimide as a composite matrix for severe service environments. The effects of thermal cycling on TriA X composites were investigated. Composite specimens were subjected to 2000 thermal cycles between −54℃ and 232℃. At 400-cycle intervals, laminates were inspected for microcracks, and Tg and short-beam shear (SBS) strength were measured. The composites did not exhibit microcracks after thermal cycling, although after 2000 thermal cycles, mechanical properties of the matrix declined slightly. The matrix degradation decreased the resistance to microcracking upon further loading. No effects of thermal oxidative aging were observed from thermal cycling, and thermally driven fatigue and creep were identified as the primary and secondary factors inducing mechanical degradation of the matrix. The Tg of the composites exhibited no change after 2000 cycles, while the SBS strength decreased slightly (3-9%). The results highlight the potential for use of TriA X composites as long-term structural components in high-temperature service environments.❧ In addition to cyclic temperatures, the effects of moisture on neat TriA X and associated composites were also investigated. Water uptake tests were performed on the polyimide at various temperatures and relative humidity levels to investigate moisture absorption behavior. Two-stage moisture absorption was observed, in which the first stage was diffusion-controlled, while the second stage was moisture plasticization-controlled. As exposure temperature increased, the equilibrium moisture content of the polyimide decreased, indicating an exothermic absorption process. The Arrhenius temperature dependence and moisture saturation as functions of temperature and humidity in the neat polymer were determined using curve-fitting based on published mathematical models. Long-term hydrothermal aging at 95℃ was conducted on the neat polyimide and associated carbon fiber composites. Reversible hydrolytic reactions and a trace of irreversible hydrolysis were observed in the long-term exposure. The tensile ductility of the neat polyimide and the short-beam shear strength of the composites decreased with increasing aging time, while the tensile strength and modulus and thermal properties of the polyimide exhibited little change after 2000-h aging, demonstrating hydrothermal stability. The decrease in the ductility of the neat polymer after long-term moisture exposure was attributed to the network structure change, driven by hydrolysis and moisture plasticization.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Processing, thermal-oxidative stability and thermal cyclic fatigue of phenylethynyl-terminated polyimides
PDF
Sustainable manufacturing of out-of-autoclave (OoA) prepreg: processing challenges
PDF
Defect control in vacuum bag only processing of composite prepregs
PDF
In situ process monitoring for modeling and defect mitigation in composites processing
PDF
Material and process development and optimization for efficient manufacturing of polymer composites
PDF
Chemical recycling of amine/epoxy composites at atmospheric pressure
PDF
Characterization, process analysis, and recycling of a benzoxazine-epoxy resin for structural composites
PDF
Void evolution in vacuum bag-only prepregs
PDF
Atomistic modeling of the mechanical properties of metallic glasses, nanoglasses, and their composites
PDF
Mechanical behavior and deformation of nanotwinned metallic alloys
PDF
High temperature creep behaviors of additively manufactured IN625
PDF
Ageing and mechanical failure of fiber reinforced polymers
PDF
Low-dimensional asymmetric crystals: fundamental properties and applications
PDF
The effect of lattice structure and porosity on thermal conductivity of additively-manufactured porous materials
PDF
Processing, mechanical behavior and biocompatibility of ultrafine grained zirconium fabricated by accumulative roll bonding
PDF
Shock wave response of in situ iron-based metallic glass matrix composites
PDF
Metallic syntactic foams synthesis, characterization and mechnical properties
PDF
Long range internal stresses in cyclically deformed copper single crystals; and, Understanding large-strain softening of aluminum in shear at elevated temperatures
PDF
On the dynamic fracture behavior of polymeric materials subjected to extreme conditions
Asset Metadata
Creator
Zhang, Yixiang
(author)
Core Title
Processing and properties of phenylethynyl-terminated PMDA-type asymmetric polyimide and composites
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Materials Science
Publication Date
04/23/2019
Defense Date
08/20/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
amorphous,asymmetric backbone,hydrothermal aging,imidization,mechanical properties,OAI-PMH Harvest,polyimide,processing,structure-property relationship,thermal cycling,thermal properties,thermoset
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Nutt, Steven R. (
committee chair
), Centea, Timotei (
committee member
), Grunenfelder, Lessa K. (
committee member
), Williams, Travis J. (
committee member
)
Creator Email
zhangyix@usc.edu,zhangyixiang007@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-96472
Unique identifier
UC11675182
Identifier
etd-ZhangYixia-6905.pdf (filename),usctheses-c89-96472 (legacy record id)
Legacy Identifier
etd-ZhangYixia-6905.pdf
Dmrecord
96472
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Zhang, Yixiang
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
amorphous
asymmetric backbone
hydrothermal aging
imidization
mechanical properties
polyimide
processing
structure-property relationship
thermal cycling
thermal properties
thermoset