rHEs'S

NIVERSITY LIBRARIES

11111111111111111111111111111 1111 2111

3129301710

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

This is to certify that the
thesis entitled

MECHANICAL BEHAVIOR OF EPOXY:
PART A: INFLUENCE OF LOCAL STRUCTURE ON VISCOELASTIC
AND MOISTURE TRANSPORT PROPERTIES
PART B:. INFLUENCE OF REINFORCEMENT
presented by

Peter John Scheuer

has been accepted towards fulfillment
of the requirements for

Master's degree in Materials Science

 

 

/U 1 Q C
M {Vassar .,

 

Date ///<>25’:/?8/

0-7 639 MS U i: an Affirmative Action/Equal Opportunity Institution

 

 

 

 

LIBRARY

Mlchlgan State
Unlverslty

 

 

 

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

 

DATE DUE DATE DUE

DATE DUE

 

 

 

:
I
,‘x
it“.
I“
.9. ‘ Jr"?
" . .l.‘
iii
fig—:i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ma WM“

MECHANICAL BEHAVIOR OF EPOXY:
PART A: INFLUENCE OF LOCAL STRUCTURE ON VISCOELASTIC
AND MOISTURE TRANSPORT PROPERTIES
PART B: INFLUENCE OF REINFORCEMENT
By

Peter John Scheuer

A THESIS

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

MASTER OF SCIENCE

Department of Materials Science and Mechanics

1998

ABSTRACT
MECHANICAL BEHAVIOR OF EPOXY:
PART A: INFLUENCE OF LOCAL STRUCTURE ON VISCOELASTIC AND
MOISTURE TRANSPORT PROPERTIES
PART B: INFLUENCE OF REINFORCEMENT
By

Peter John Scheuer

Viscoelastic and moisture transport properties of modified network epoxy glasses
were examined and probed through stress relaxation and gravimetric sorption
experiments, respectively. The modified networks contained a monofunctional epoxy
which results in “dangling” chains within the network structure. The Viscoelastic
response of these modified networks to physical aging was affected by the local network
mobility. Moisture absorption experiments reveal that the path for diffusion, as a result of
the local network structure, plays a significant role in the transport of small molecules
into glassy materials. In addition, the feasibility of orienting, via a magnetic field,
metallic glass ribbons into complex configurations within an epoxy matrix was
investigated. Magnetic field orientation of reinforcements is possible provided precise
control of the matrix’s viscosity. A low viscosity results in reinforcement sinking, while a
high viscosity prevents the reinforcements from moving into desired configurations when

exposed to the magnetic field.

To my future wife...

iii

ACKNOWLEDGEMENTS

I would like to express my sincerest thanks to Professor Andre Lee and Professor
KN. Subrarnanian for their support, guidance, and leadership throughout my graduate
education. In addition, I would like to thank Professor James Lucas for serving on my
Master’s defense committee. Special thanks goes to the Polymers Division of the
National Institute of Standards and Technology for partial financial support and the
Composite Materials and Structures Center at Michigan State University for providing

usage of the facility during the course of my graduate program.

TABLE OF CONTENTS

List of Tables - -- - - _ _ viii
ListofFigum - - _ - -- - ix

PART A: INFLUENCE OF LOCAL STRUCTURE ON VISCOELASTIC AND
MOISTURE TRANSPORT PROPERTIES

 

CHAPTER 1 INTRODUCTION- - - - - - l
1.1 Polymer Glasses ................................................................................................ 1
1.2 Physical Aging .................................................................................................. 2
CHAPTER 2 MATERIALS AND SAMPLE CHARACTERIZATION ................. 6
2.1 Materials ........................................................................................................... 6
2.1.1 Epoxy ..................................................................................................... 6
2.1.2 Modifiers ................................................................................................. 6
2.1.3 Curing Agents ......................................................................................... 7
2.2 Sample Materials and Preparation ..................................................................... 11
2.2.1 DGEBA/Heloxy 62/D23O ........................................................................ 16
2.2.2 DGEBA/Heloxy 62/mPDA ...................................................................... 18
2.3 Sample Characterization .................................................................................... 21
2.3.1 Glass Transition Temperatures ................................................................ 21
2.3.2 Network Formation ................................................................................. 23

CHAPTER 3 LOCAL STRUCTURE EFFECTS ON THE VISCOELASTIC

 

 

RESPONSE - 25
3.1 Background ..................................................................................................... 25
3.1.1 Viscoelasticity ......................................................................................... 25

3.1.2 Memory Effect ........................................................................................ 29

3 .2 Viscoelastic Response ...................................................................................... 35
3.2.1 Isothermal Aging ..................................................................................... 35

3.2.2 Memory Effect ........................................................................................ 40

3.2.3 Data Analysis Method ............................................................................. 42

3.3 Results and Discussion ..................................................................................... 44
3.3.1 Isothermal Aging ..................................................................................... 44

3.3.2 Memory Effect ........................................................................................ 53
CHAPTER4 MOISTURE ABSORPTION - - - 55
4.1 Background ..................................................................................................... 55
4.2 Water Transport in Epoxy Resins ..................................................................... 56

4.3 Gravimetric Moisture Absorption ..................................................................... 60
4.3.1 Diffusion ................................................................................................. 60

4.3.2 Experimental Procedure .......................................................................... 65

4.3.3 Analysis ................................................................................................... 66

4.4 Results and Discussion ..................................................................................... 66
4.4.1 Cross-link Density Effects ....................................................................... 66

4.4.2 Aging Efi‘ects .......................................................................................... 69
CHAPTER 5 SUMMARY - 73

 

   

 

b ' ‘ ~
I A v
‘ .
‘ .
.
. .
. .
. .
.- ’ v 1a ‘
. ..a
.n ' . l
. , '
, . I‘
..,' _
.
1

5.1 Local Structure Effects on Viscoelasticity ......................................................... 73

5.2 Local Structure Effects on Moisture Absorption ................................................ 74

PART B: INFLUENCE OF METALLIC GLASS RIBBON REINFORCEMENTS

 

CHAPTER 6 INTRODUCTION .............................................................................. 75
6.1 Materials Background ...................................................................................... 76
CHAPTER 7 BONDING CHARACTERISTICS ......... A -- - - ..... 79
7.1 Background ..................................................................................................... 79
7.2 Sample Preparation .......................................................................................... 81
7.3 Results and Discussion ..................................................................................... 85

CHAPTER 8 FEASIBILITY OF MAGNETICALLY ALIGNING

 

REINFORCEMENTS -- - - , - 89
8.1 Background ..................................................................................................... 89
8.2 Sample Preparation .......................................................................................... 89
8.3 Results and Discussion ..................................................................................... 90
CHAPTER9 SUMMARY . ....................... _ AAAAA92

 

LIST OF REFERENCES -- - -- A ...... 94

vii

Table 2-1.

Table 2-2.

Table 2-3.

Table 3-1.

Table 4-1.

Table 6-1.

Table 6-2.

LIST OF TABLES

Stoichiometric mixing ratios of epoxy components with D230. ................ 17
Stoichiometric mixing ratios of epoxy components with mPDA. .............. 19

Bond percolation threshold values for 2-D lattice. Adapted fiom (8) ........ 23

KWW curve fitting parameters 8=0.1°/o; Ta: aging temperature ................. 43
Equilibrium moisture values of Epoxy modified networks ........................ 67
Physical properties of a common metallic glass ribbon. ............................. 76

Magnetic properties of metach glass ribbons. .......................................... 77

viii

Figure 1-1.

Figure 2-1.

Figure 2-2.

Figure 2-3.
Figure 2-4.
Figure 2-5.
Figure 2-6.
Figure 2-7.

Figure 2-8.

Figure 3-1.

Figure 3-2.

Figure 3-3.

Figure 34.

LIST OF FIGURES

Schematic diagram of volume vs. temperature behavior of a glass forming
material isobarically cooled from T1 to T2. Q is the rate of cooling.- ------ 3

 

Schematic diagram of an “ideal” epoxy network. 9

Schematic diagram of an epoxy network modified with a monofunctional

 

 

 

 

epoxy. 10
Diglycidyl ether of bisphenol A (DER 332)—“Epoxy”. 12
O—cresol glycidyl ether (Shell Heloxy—62)-“Modifier”. l3
Polyoxypropylenediamine (J effamine D-230)-“Curing Agent. ------------- 14
m-Phenylenediamine (mPDA)-“Curing Agent”. 15
Curing cycle for DGEBA/mPDA sample processing 20

 

Plot of glass transition temperature as a function of Heloxy content. Solid

line represents the rule of mixtures values. (0) corresponds to
experimentally determined Tg values. 22

 

Response of stress relaxation curves to isothermal aging. 26

 

Thermal treatment involving two consecutive temperature jumps of
opposite sign. (A) Sample directly quenched, then aged, (B-E) Samples
quenched to various temperatures below Tg; (TB>TC>TD>TE) . All
samples were aged for the same time. After aging all samples were
reheated to the same temperature. Adapted fiom [13]. ' 30

 

Volume recovery as a function of time for the samples conditioned with
the thermal treatments provided in Figure 3-2. (A) Direct quench; (B-E)
quench—storage-‘hp-jump”. 8 measures the volume departure from
equilibrium. Adapted from [13]. 31

 

 

Schematic diagram of a generalized Maxwell model. 33

ix

 

r!
'

p.

Figure 3-5.

Figure 3-6.

Figure 3-7.

Figure 3-8.

Figure 3-9.

Figure 3-10.

Figure 3-11.

Figure 3-12.

Figure 3-13.

Schematic diagram of the applied strain history. Applied strain (8:0.1%)- 36

Stress relaxation modulus curves for DGEBA+9S wt% D230+5 wt%
Heloxy-62 quenched from 20°C above Tg and aged at 597°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0 (0)40 (0)8.0
(0)160 (0)320 (0)640. 37

 

Stress relaxation modulus curves for DGEBA+90 wt% D230+10 wt%
Heloxy-62 quenched fi’om 20°C above Tg and aged at 549°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0 (0)40
(0)8.0 (0)160 (0)320 (0)640. 38

 

Stress relaxation modulus curves for DGEBA+80 wt% D230+20 wt%
Heloxy-62 quenched from 20°C above Tg and aged at 418°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0 (0)40
(0)8.0 (0)160 (0)320 (0)640. 39

 

Stress relaxation modulus curves for DGEBA+90 wt%D230+10 wt%
Heloxy-62 quenched from 20°C above Tg and aged at 59°C. Prior to the
mechanical test, the sample was annealed at 468°C for 116 hours.
Applied strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0
(0)40 (0)8.0 (0)160 (A)32.0 (A)64.0. 41

 

Double logarithmic plot of aging time shift factor, are, versus aging time
for DGEBA/D230 samples at different aging temperatures. Reference
aging time is 28 minutes. Applied strain is 0.25%. Aging temperature in
°C: (0)576 (x)61.9 (0)669 (A)71.5 (0)818. u z0.9. 46

 

Double logarithmic plot of aging time shift factor, a“, versus aging time
for DGEBA/D230/5 wt% Heloxy samples at different aging temperatures.
Reference aging time is 30 minutes. Applied strain is 0.1%. Aging
temperature in °C: (0)597 (0)653 (0)704. 47

 

Double logarithmic plot of aging time shift factor, an, versus aging time
for DGEBA/D230/10 wt% Heloxy samples at different aging
temperatures. Reference aging time is 30 minutes. Applied strain is 0.1%.
Aging temperature in °C: (0)503 (0)551 (0)602 48

 

Double logarithmic plot of aging time shift factor, aw, versus aging time
for DGEBA/D230/20 wt% Heloxy samples at different aging
temperatures. Reference aging time is 30 minutes. Applied strain is 0.1%.
Aging temperature in °C: (0)37.l (0)424 (0)463. 49

 

Figure 3-14.

Figure 3-15.

Figure 3-16

Figure 3-17.

Figure 4-1.

Figure 4-2.

Figure 4-3.

Figure 4-4.

Figure 4-5.

Figure 4-6

Double logarithmic plot of aging time shift factor, am, versus aging time
for DGEBA/D230/5 wt% Heloxy samples at an aging temperatures of
653°C. 11:033. 50

 

Double logarithmic plot of aging time shift factor, are, versus aging time
for DGEBA/D230/10 wt% Heloxy samples at an aging temperatures of
549°C. u=0.77. 51

 

Double logarithmic plot of aging time shift factor, are, versus aging time
for DGEBA/D230/20 wt% Heloxy samples at an aging temperatures of
418°C. p.=0.56. 52

 

Characteristic relaxation time, t, versus aging time for DGEBA+90 wt%
D230+10 wt% Heloxy-62 at a testing temperature of 59°C. Applied strain
is 0.1%. (0) Sample was quenched directly fiom 80°C to 59°C; (0)
Sample was quenched from 80°C to 46.8 °C and annealed for 116 hours,
then jumped to 59°C; (0) Sample was quenched from 80°C to 387°C
and annealed for 116 hours, then jumped to 59°C. 54

 

Conditioning environment. 61

 

Variation of G with log time. Solid line is based on Eq. 4-4 representing
the exact solution and the dashed line is based on Eq. 4-5 representing the
approximate solution. 64

 

Moisture absorption behavior of unaged DGEBA/mPDA samples
containing: (0)0 wt% Heloxy; (0)5 wt% Heloxy (0)10 wt% Heloxy;
and (A)20 wt% Heloxy. Bath temperature is 60°C. 68

 

Moisture absorption behavior of aged and unaged DGEBA/mPDA
samples containing 0 wt% Heloxy. Bath temperature is 60°C. Samples
directly quenched into the water bath from above Tg are represented by
(I) symbols. Samples aged at a temperature of Tg-l3°C for 120 hours
are represented by (0) symbols. 7O

 

Moisture absorption behavior of aged and unaged DGEB A/mPDA
samples containing 10 wt% Heloxy. Bath temperature is 60°C. Samples
directly quenched into the water bath from above Tg are represented by
(I) symbols. Samples aged at a temperature of Tg-13°C for 120 hours
are represented by (0) symbols. 71

 

Moisture absorption behavior of aged and unaged DGEBA/mPDA
samples containing 20 wt% Heloxy. Bath temperature is 60°C. Samples
directly quenched into the water bath from above Tg are represented by

Figure 7-1.

Figure 7-2.

Figure 7-3.

Figure 7-4.

Figure 7-5.
Figure 7-6.

Figure 7-7.

(I) symbols. Samples aged at a temperature of Tg-l3°C for 120 hours

 

 

 

 

are represented by (0) symbols. 72
Schematic diagram of the mold set-up for preparation of composite
plaques.-- 82
Schematic diagram of (A) cast composite sheet, and (B) compact tension
specimens. 84
Fracture surface of compact tension composite sample. 86

SEM micrograph of the fracture surface of metallic glass ribbon

 

 

reinforced epoxy. 86
Fracture surface indicating failure in the epoxy matrix. 87
SEM micrograph indicating ribbon bonding to the matrix. 88

 

SEM micrograph of a ribbon bonded to the epoxy matrix. 88

 

xii

PART A: INFLUENCE OF LOCAL STRUCTURE ON VISCOELASTIC AND
MOISTURE TRANSPORT PROPERTIES

CHAPTER 1 INTRODUCTION

Epoxies are used in many industrial applications. For example, the aerospace
industry primarily uses epoxy in the form of matrices for composite materials and
adhesives in structural components. The use of epoxies for such applications requires
knowledge of their lifetime performance in extreme service environments. Due to the
severity of conditions in extreme environments, the durability of epoxies is critical
information in the design and development of aerospace components. The durability of
epoxy adhesives and composites is difficult to predict unless a fundamental
understanding of how the chemical structure present in epoxies affects the long-term, in-
service performance of these materials.

There has been increased interest in the structural and property changes that occur
in glassy polymers upon annealing below the glass transition. This “aging” or time-
dependent behavior is a direct result of the non-equilibrium nature of the glassy state. The
importance of this phenomenon in applications stems fi'om the fact that material
performance must be predictable in order to insure compliance with design specifications

throughout the service life of the fabricated polymer product.

1.1 Polymer Glasses
Amorphous polymers upon isobaric cooling from the melt through the glass

transition into the glassy state depart from their thermodynamic equilibrium and exhibit

non-equilibrium characteristics. As the polymer is cooled there occurs a rapid decrease in
molecular mobility which results in a sudden increase in the viscosity and modulus as the
transition temperature is approached. The polymer molecules are unable to reach their
equilibrium conformation and packing, and as the temperature is further decreased the
molecules are essentially “frozen” into a non-equilibrium state of higher energy and
volume relative to the equilibrium state at the same temperature [1]. Figure 1-1 is a plot
of the volume versus temperature behavior for a typical glass forming polymer.

The temperature at which the non-equilibrium structure and behavior begins to
deviate from the extrapolated equilibrium state is one definition of the glass transition
temperature, Tg. The position of the glass transition is dependent upon many factors such
as the processing conditions (i.e. rate of cooling) of the polymer. Cooling a polymer melt
at a high rate will result in a higher energy and volume non-equilibrium state, as
compared to a polymer which has been cooled at a slower rate, because the polymer
molecules have even less time to reach their equilibrium conformation. As a result of this
relatively elevated non-equilibrium state, the glass transition of the polymer moves to

higher temperatures along the equilibrium melting curve (Figure 1-1).

1.2 Physical Aging

The non-equilibrium state of an amorphous polymer appears to be unstable.
Volume relaxation studies have revealed that glasses, at temperatures below Tg, undergo
slow processes which attempt to establish equilibrium. This gradual approach to
equilibrium affects many mechanical and physical properties of the polymer system: (I)

an increase in density, tensile and flexural yield stress, and elastic modulus; (2) a

 

 

 

 

. , ‘ |. . .
r . . . .
, I . . ‘ I u v ‘ ‘
c . : -
. ' ' . .
' I
. j ' ' ,..
I . . . ,. ‘
~
4 ' . -. ‘ ‘ t
, r . r
. , _ rs . .
. l’.
I . 4 ' _ ‘ ,- yA _
. .7 . .,- . ,- .. ,.
' a . A. , a ‘
. . v .. . r
1 I ‘r I: 1 I
‘ . v r . . ‘ v. v A
. A . : . g - ' . . . -
v
. ' . .. A
. ' . . , , . ‘ . .
. , - . , r .
' .. V ' ‘ t ’ ' _ t I a t
, . .. , , .- . .
.
' v V ‘ . .
. | ‘ . -. A » . ..,
. 1 ~ . 4’ f ' I - ,r . ~ 1
l > . f .
4 ‘ . ' . . . . -_ v ,
.._ ‘ . _
. . » y . .
. r ‘. j . _ a t ‘
. . - .. .
b , I .
. u . , . . I u
. - - '1
I r , ‘ ~
1 \
. r . . .
1
1 ‘ , A 1 t ‘ _ ' a < a .
.
, ' .
1v N
‘. r, (it "
. . . . -‘ r
x
a, ' .. . .
v: ‘ ‘ ~. . r ' . . a . ~
v ' ‘ . n » A
> , . . . .
l V 1 I 7 h
, 4 - . .
. , _ ‘ I, I “ l
. , . _ . ‘ I
. , . r .
j . - r " .
, . r, ., _ r 1 I
, - . .. . ‘
.v , 4 ‘ .
v . ‘ ‘ v v ‘7 - .

Volume,
Enthalpy

Liquid

    
    

 

 

 

 

Non-equilibrium
Q2
Q1 Q2>Q1
“Glass” ;
Tgl ng Temperature

Figure 1-1. Schematic diagram of volume vs. temperature behavior of a glass
forming material isobarically cooled from T1 to T2. Q is the rate of
cooling.

decrease in impact strength, fracture energy, ultimate elongation, and creep rate. The time
dependent evolution of these properties is referred to as aging [1]. There are two types of
aging; “Physical aging” is the change in Viscoelastic (mechanical) properties
accompanying volume relaxation, and “chemical aging” which is material degradation of
the polymer through processes such as thermal breakdown and photo-oxidation. Physical
aging, unlike chemical aging, is a thermoreversible process where any previous aging
can be erased by re-heating the aged sample to a temperature above Tg [2].

The effects of physical aging on the mechanical properties of glassy materials are
commonly examined through Viscoelastic experiments. Stress relaxation is a Viscoelastic
experiment which consists of instantly straining a Viscoelastic material and, while
maintaining this deformation, measuring the stress as it decays from an initial value. The
decaying stress in this type of relaxation experiment is impacted by the internal molecular
fi'iction between the polymer molecules [3]. During physical aging the polymer
undergoes densification, thus the internal molecular friction is effected. Therefore, stress
relaxation experiments can be used to probe the change in Viscoelastic properties during
aging.

The Viscoelastic response of polymers affects polymer morphology and structure
[4]. Polymer structure and morphology are important characteristics which contribute to
the mechanical properties of polymers. For example, polymer branches and crosslinks are
structural components which influence the properties and performance of polymers [5].
Thermoset materials, such as epoxies, are highly cross-linked or networked polymeric

systems. Defects in the epoxy network, such as loops and dangling ends, affect the

structure and thus the performance of epoxy polymers. To explore the degradation due to
network defects Viscoelastic experiments are used.

In addition to affecting the mechanical properties of polymer glasses, physical
aging affects the diffusivity and solubility of polymers [2]. The diffusivity and solubility
are important properties which have an impact on the performance of polymers in moist
environments. Hygrothermal effects on epoxies and epoxy composites have been widely
studied. The main hygrothermal effects in epoxy systems are plasticization, swelling,
microcracking, and Tg suppression. The effects may be reversible, as in the case of
plasticization or swelling, or the effects may be irreversible such as hydrolysis or
microcracking. Network defects affect the mechanical properties of thermoset polymers
and thus they may also affect the hygrothermal performance of thermoset polymers.
Local defects affect specific attributes of a polymer network. For example, the presence
of dangling ends leads to a lower cross-link density. On the other hand, chain
entanglements lead to higher values of cross-link density [6,7]. In this case, the affect of
network defects is a change in the cross-link density, which may affect the sorption
behavior of glassy polymers.

Since physical aging is an important phenomenon in the area of polymer glasses
and little attention has been paid to its effects on polymer networks; it is the aim of this
investigation to further understand and clarify the effects of aging on epoxy networks.
Viscoelastic and moisture absorption experiments were conducted in order to determine

the effect of local network structures on physically aged epoxy glasses.

CHAPTER 2 MATERIALS AND SAMPLE CHARCTERIZATION

2.1 Materials
2.1.1 Epoxy
Epoxy resins are characterized by a three-member ring known as the epoxide,

oxirane, or ethoxyline group, and are represented as:

RA

where R is a resin containing aliphatic, cycloaliphatic, or aromatic backbones. Properties
of cured epoxy polymers are influenced by many factors at the molecular level, such as
backbone structures of the epoxy resin and curing agent; nature of the cross-linking bond
formed by the resin and curing agent; and the density and extent of cross-linking. Due to

the ease of processing, epoxy has become an important class of engineering polymers.

2.1.2 Modifiers

In order to facilitate processing and to modify cured network properties, other
substances, such as fillers, solvents, diluents, plasticizers, and toughening agents may be
included in the compositions. For example, rubber particles are added as toughening
agents to improve the fracture toughness of brittle thermoset polymers. A commonly used
modifier is diluents. Diluents affect the prOperties of cured resin systems, and in
particular, lower the viscosity in order to facilitate processing. Compounds containing
only one epoxy or fiinctional group per molecule are known as monofunctional epoxies

and are commonly added to conventional epoxy resin systems as diluents. For example,

7

addition of a monofunctional epoxy to a conventional epoxy resin system reduces the
viscosity of the resin which improves the wetting capability of the polymer when used in
coating applications. In addition, reinforcements, such as metallic glass ribbons, may be
added to a polymer system in order to improve the strength and toughness of polymer

matrix composites.

2.1.3 Curing Agents

In order to achieve desired engineering properties various curing agents, which
react with the epoxy rings to form insoluble and intractable thermoset polymers, are used.
For example, epoxy resin cured with anhydride based curing agents exhibit low viscosity,
low exothermic heats of reaction, and little shrinkage when cured at elevated
temperatures. The cured system exhibits good mechanical and electrical properties along
with good thermal stability which makes this polymer system useful in electrical
applications. Polyamide epoxy curing agents are used extensively and are extreme
versatile. Polyamides are inexpensive, have little color and can be mixed in any ratio.
Epoxies cured with polyamides have good mechanical properties, cure under mild
conditions, and are often used in coating applications.

Polymer molecules linked together through the use of a curing agent form a
network which is, in effect, one giant molecule (Figure 2-1. “Ideal” network structure
diagram). Network polymers, at temperatures below the glass transition temperature, are
hard intractable solids, but network polymers above the glass transition temperature are
rubbery [5]. The characteristics of a network polymer depend on the functionality of the

resin, flexibility of the polymer chain, and the type and concentration of the crosslinking

 

8

agent. Polymer chains with high functionality are capable of forming a crosslink through
a reaction with curing agents or hardeners which react with the epoxide functional groups
on the polymer chain. In order for a network to form, the polymer mixture must contain
an average of three or more functional groups. A monofunctional epoxy is incapable of
forming a crosslink because it has only one epoxide group which can react with the
curing agent. In this manner, addition of a monofirnctional epoxy imparts “dangling”
chains into the network. These dangling chains are considered local network defect
structures. The number of dangling chains present in the network is important because
dangling chains do not contribute to the elasticity and, as network defects, may reduce the
effectiveness of existing crosslinks (Figure 2-2. Epoxy network modified with a

monofunctional epoxy).

NHz—RI—NHz + A— Rz—A
\ J \

 

 

J
Y
Curing Di-functional
Agent Epoxy

 

 

 

 

 

 

 

 

 

 

 

o t t

o b b 5
R2 R2

f r R‘ e 6
R2 R2

k i k b

Figure 2-1. Schematic diagram of an “ideal” epoxy network.

 

10

NH; ——R1—-NH2 + A—R2 --A+ A—R3

 

 

\ x \ Y j
Y O
Curing Di-functional Mono-functional
Agent Epoxy Epoxy

 

 

 

 

IT I
4

1 +
R' o a

R3
k a a

Figure 2-2. Schematic diagram of an epoxy network modified with a mono-fimctional
epoxy.

I
I

 

 

 

 

 

y

 

 

11

2.2 Sample Materials and Preparation

Samples with various network structures were investigated. The epoxide monomers
studied were di-functional pure diglycidyl ether of bisphenol-A (DER 332, Dow
Chemical, USA) mixed with varying amounts of a low viscosity o-cresyl glycidyl ether
monoepoxide (Heloxy 62, Shell Chemical, USA). The epoxide mixtures were crosslinked
with either an amine terminated poly (propylene oxide) (Jeffamine D230, trademark of
Texaco Chemical Company) or m-Phenylenediamine (mPDA, DuPont, USA). The

chemical structures of the components are described in Figure 2-3 through Figure 2-6.

12

Cris A
.AL..2_.L._.@._..._.._ .
CH3

Figure 2-3. Diglycidyl ether of bisphenol A (DER 332)-“Epoxy”.

13

Q ._ ..,..../;\..

Figure 2-4. o-Cresol glycidyl ether (Shell Heloxy 62)-“Modifier”.

l4

 

 

 

 

HzN CH CH2 CH2

0

CH -]——2 6 NH:

CH3 CH3

Figure 2-5. Polyoxypropylenediamine (Jeffamine D23 0)-“Curing Agent”.

15

NH;

NH;

Figure 2-6 m-Phenylenediamine (mPDA)-“Curing Agent”.

16

2.2.1 DGEBA/Heloxy 62/D230

DGEBA, di-functional epoxide monomer, was preheated at 60°C for 2 hours to
melt any crystals present. Stoichiometric portions of the DGEBA epoxide monomer,
monofunctional Heloxy 62 modifier and, D230 amine curing agent were weighed and
thoroughly mixed together. Table 2-1 lists the mixing ratios used in the preparation of
samples to be used for Viscoelastic experiments. The mixture was vacuum degassed at
room temperature for 30 minutes in order to remove entrapped air, then poured into 1/2”
diameter x 7” long glass tubes, which were sealed at one end with silicone plugs, and
cured at 100°C for 24 hours. The epoxy rods were allowed to slowly cool down to room
temperature. Specimens were then machined, on a lathe, from the cast rods into
cylindrical tensile coupons and stored in a desiccator at room temperature until the

experiments were performed.

17

TABLE 2-1. Stoichiometric mixing ratios of epoxy components with D230.

 

 

 

 

 

 

 

 

 

 

Total Wt. fraction Wt. fraction Wt. fraction
Sample Weight of DGEBA of Heloxy 62 of D230
(grams)
100%DGEBA/0%Heloxy/D230 22,750 74.73% 0% 25.27%
95% DGEBA/5%Heloxy/D230 22,795 70.85% 3.93% 25.22%
90% DGEBA/10%Heloxy/D230 22,840 66.98% 7.83% 25.17%
80% DGEBA/20%Heloxy/D230 22,930 59.31% 15.61% 25.07%

 

 

18

2.2.2 DGEBA/Heloxy/mPDA

The DGEBA and mPDA curing agent were preheated at 60°C to melt any crystals
present in the epoxide and to melt the mPDA, which existed as solid flakes. The DGEBA
epoxide monomer, Heloxy 62 modifier, and mPDA curing agent were weighed and
quickly mixed together to prevent the mPDA from crystallizing which can occur at room
temperature. Table 2-2 presents the mixing ratios used in the preparation of samples to be
used for moisture absorption experiments. Afler thorough mixing, the resin was vacuum
degassed at room temperature for 15 minutes to allow air bubbles to escape. The mixture
was then poured into a stainless steel mold and cured for 2 hours at 75°C followed by 3
hours at 120°C. Figure 2-7 depicts the curing cycle used in the sample preparation. Afier
curing, the epoxy plaques were removed fi'om the mold while still warm, placed back in
the oven and allowed to slowly cool to room temperature. The plaques were then cut into
1.5”x 1.5” squares and stored in a desiccator at room temperature until the experiments

were performed. In addition, all samples had an average thickness of 1/32”.

 

 

19

TABLE 2—2. Stoichiometric mixing ratios of epoxy components with mPDA.

 

 

 

 

 

 

 

 

 

 

Total Wt. fraction Wt. fraction Wt. fraction
Sample Weight of DGEBA of Heloxy 62 of mPDA
inn-us)
100%DGEBA/0%Heloxy/mPDA 19,700 86.29% 0% 13.71%
95% DGEBA/5%Heloxy/mPDA 19,745 81.79% 4.53% 13.67%
90% DGEBA/10%Heloxy/mPDA 19,790 77.31% 9.04% 13.64%
80% DGEBA/20%Heloxy/mPDA 19,880 68.40% 18% 13.58%

 

 

20

Temperature

 

 

120°C '11— 3 DEL

 

 

750 Cd... 5 2 hr;

 

 

Curing Time

Figure 2-7. Curing cycle for DGEBA/mPDA sample processing.

21

2.3 Sample Characterization

2.3.1 Glass Transition Temperatures

Differential scanning calorimetry, DSC, is a thermal analysis technique which
measures the input power differential between a sample cell and reference cell. When the
sample undergoes a transition, a characteristic change in power differential is observed.
The nature of the change depends on the order of the thermodynamic transition. Melting
or crystallization is a first order transition which manifests itself as an endothermic or
exothermic peak respectively on the heat flow versus time/temperature curve. The glass
transition is a second order transition which results in an endothermic step change of the
heat flow baseline.

The glass transition temperature, Tg, for the DGEBA/mPDA glasses containing
incremental amounts of Heloxy 62 monoepoxide upto 100 wt% were determined using a
TA Instruments Differential Scanning Calorimeter (Modulated DSC). The heating scans
were performed at 5°C/min immediately afier a cooling run. The glass transition was
taken at the inflection point of the step transition curve. Figure 2-8 is a plot of the glass
transition temperatures as a function of Heloxy content. As can be seen, as the Heloxy
content is increased the glass transition temperature of the polymer decreases, and the
experimentally determined transition temperatures are in good agreement with the
theoretical values of Tg as determined by the rule of mixtures. Another interesting result
was that even though increasing the Heloxy content decreased the Tg, the width of the

glass transition zone was not affected.

 

22

 

o 1201

 

Temperature ( C

 

 

COO

 

20 40 60 80 100
wt% Heloxy

1 0

Figure 2-8. Plot of glass transition temperatures as a function of Heloxy content.
Solid line represents the rule of mixtures values. (0) corresponds to
experimentally determined Tg values.

23

2.3.2 Network Formation

In order to determine the network structure present in the epoxy glasses, a set of
specimens containing a range of Heloxy from 0 wt% to 100 wt% in increments of 10
wt% were fabricated. A portion of each sample was placed in a vial of solvent at room
temperature. Samples containing more than 50 wt% Heloxy dissolved in the solvent
which indicates the lack of an infinite network. Samples containing less than 50 wt%
Heloxy swelled due to solvent uptake by the network. These results indicate that 50 wt%
is the critical concentration necessary to form an infinite network. This value is in
agreement with the value for the critical concentration of bonds needed to form an

infinite network according to bond percolation theory (Table 2-3).

Table 2-3. Bond percolation threshold values for 2-D lattice. Adapted from (8).

 

 

 

 

 

 

 

 

 

Coordination Percolation Critical bond
Lattice Number Threshold number
(Z) (Po) (2pc)
Honeycomb 3 0.6527 1 .96
Square 4 0.5000 2.00
Triangular 6 0.3473 2.08

 

Bond percolation theory applies to a lattice where a fraction p of the bonds are
connected. A function P(p) refers to the probability that a given connected bond is part of
an infinite cluster linked by other such bonds. The percolation probability P(p) is equal to
zero for p<pc, where pc is the critical. concentration necessary to form an infinite network
and is dependent on the lattice type. Above the critical concentration, the percolation
probability rises monotonically from zero at pc to a value approaching unity at p=1. The

critical bond number or the number of bonds needed to form a network, is a firnction of

24

the dimensionality, ‘d’, of the lattice and the coordination number, ‘2’, and gives a result
for pc to within a few percent of that observed experimentally [8].

ZPcPd/(d-l) (2-1)

Therefore, the network structure present in the Heloxy modified epoxy glasses is a

two-dimensional square lattice. Figure 2-2 is a schematic diagram of the epoxy network

modified with a monofirnctional epoxy.

 

CHAPTER 3 LOCAL STRUCTURE EFFECTS ON THE
VISCOELASTIC RESPONSE

3.1 Background

3.1.1 Viscoelasticity

A material may exhibit behavior which combines liquidlike and solidlike
characteristics. For example, a body, which is not quite solid, does not maintain a
constant deformation under constant stress, but slowly continues to deform or creep with
time. When a body is constrained at constant deformation, the stress required to hold it
diminishes gradually, or relaxes. On the other hand, a body, which is not quite liquid,
may, while under constant stress, store some of the energy input instead of dissipating it
like a solid. Materials whose behavior exhibits this energy storage and loss are called
Viscoelastic. In polymeric systems, mechanical behavior is dominated by Viscoelastic
phenomena because of the molecular mobility associated with polymeric structures.
Under a macroscopic deformation, the flexible polymer molecules are able to make
complicated molecular adjustments [9]. Molecular structure is one factor which
determines the Viscoelastic properties of a polymer. Relaxation times are directly related
to the structure of the polymer molecules; thus, the mechanical properties of the material
will be influenced by aging via changes in relaxation times. An increase in aging time
changes all relaxation times by the same factor, thus the shape of creep or stress
relaxation curves are not changed [2]. A change in aging time will only affect the
position of the curve on the log-time scale, and not its shape. Figure 3-1 is a schematic
diagram of stress relaxation curves as a function of log time for isothermally aged

polymer glasses. This is an important factor in the applicability of a classical time-aging

25

 

26

Relaxation
Modulus

 

Aging
Time

 

 

LOG Time

Figure 3-1. Response of stress relaxation curves to isothermal aging.

27

time superposition. According to the superposition principle it is possible to predict the
long-term properties of polymeric materials, for various aging times, using short-term
tests.

Creep and stress relaxation experiments are two direct methods to study the
Viscoelastic properties of polymeric materials. Creep and stress relaxation parameters are
generally improved on cross-linking. In addition, the glass-transition temperature, Tg,
increases by increasing the cross-link density, which is a measure of the total number of
links between chains in a given mass of material. The structure and distance between
crosslink junctions are important characteristics in polymeric networks. Lee and
McKenna [10] investigated the Viscoelastic response of physically aged epoxy networks
of different crosslink densities with homogeneous lengths. In that study, altering the
molecular weight between crosslinks produced various homogeneous network structures.
Utilizing a higher molecular weight curing agent produced a network structure in which
the distance between junctions was longer than the distance between junctions produced
from a low molecular weight curing agent. In other words, the crosslink density increased
with a decrease in crosslink molecular weight. Lee and McKenna also found that when
the crosslink density increased, the glass transition temperature, Tg, increased; however,
the shape of the Viscoelastic spectrum was independent of the crosslink density for the
epoxy networks. In addition, it was possible to superimpose the stress relaxation curves at
different aging times, temperatures, and crosslink densities to form a single master curve,
demonstrating the applicability of a time-aging time-temperature—crosslink density

superposition principle to these types of network structures.

 

 

28

In order to elucidate the effect of the molecular weight distribution between
crosslinks on the mechanical and Viscoelastic responses of epoxy networks, Wang [11]
used a mixture amine terminated curing agents of various molecular weights to form
random networks with a crosslink distribution. Two epoxy systems were studied; one
contained a single diamine (DGEBA/D230) and the other contained a mixture of two
diamines (DGEBA/D230/D400 or D2000). The first epoxy system represents a network
structure which is homogeneous due to crosslinks which are equal in molecular weight;
i.e. the length of the crosslinks are the same. The second epoxy structure, which contains
two diamines of differing molecular weights, forms a heterogeneous network because the
diamines have two different molecular weights which causes the crosslinks to have two
different lengths. In that study it was found that the glass transition is broadened and the
transition temperature decreases with the addition of D400 or D2000. The Viscoelastic
response curves for different aging times and temperatures were able to be superimposed
to form master curves, thus demonstrating the applicability of the classical time-aging
time; time-temperature superposition principle. In addition, the Viscoelastic spectrum was
broadened with an increase in D2000 content and thus the aging process was retarded by

the presence of the long D2000 chains.

 

29

3.1.2 Memory Effect

In addition to being nonlinear, glassy state relaxations are also nonexponential and
exhibit the memory effect, that is, relaxation from a particular state depends not only on
what the state is, but also on how that state was reached [12]. Thermal treatments,
involving two consecutive temperature jumps of opposite sign, are depicted in Figure 3-2.
Kovacs [13] studied the response of the system, at a fixed temperature, to these thermal
treatments through their volume recovery behavior. Route A corresponds to a direct
quench from a temperature above Tg to a temperature below Tg. Routes B-E correspond
to temperature jumps from the same temperature above Tg to various T1 temperatures
below Tg, followed by isothermal recovery. The isothermal recovery times, ‘tl’, for each
route were identical. After ‘tl’ time for recovery, the specimens were instantaneously
reheated to T2, where the isothermal recovery is observed as a firnction of elapsed time (t-
t1), where ‘t’ is the time at each volume measurement during the experiment. The curves
displayed in Figure 3-3 are isothermal volumetric recoveries in which curve A represents
a simple approach experiment where the sample is quenched from a temperature above
Tg to a temperature below Tg without any additional thermal treatments. Thus, the
volume approaches the structural equilibrium values as a monotonically decreasing
function and is referred to as a “simple approach” experiment. Curves B through E are
the response curves to two consecutive temperature jumps of opposite sign and are
examples of the phenomena called memory or crossover effects, which have often been

observed experimentally and are attributed to a multiplicity of retardation mechanisms.

VOLUME

Figure 3-2.

 

30

 

 

TEMPERATURE

Thermal treatment involving two consecutive temperature jumps of
opposite sign. (A) Sample directly quenched, then aged, (B-E) Samples
quenched to various temperatures below Tg; (T3>TC>TD>TE) . All
samples were aged for the same time. After aging all samples were

reheated to the same temperature. Adapted from [13].

31

 

1200

 

 

 

 

r ' T F F l I I I r I
1000 [— \A —-§
1 1‘ :
800 1; \ 4
t :
oo 600 :- .1
L r
400 E A:
200 } f
1 :
O 5 r

-2 -1 0 1 2 3 4 5

log (14‘)
Figure 3-3. Volume recovery as a function of time for the samples conditioned with

the thermal treatments provided in Figure 3-2. (A) Direct quench; (B-E)
quench-storage-‘irp-jump”. 6 measures the volume departure fiom
equilibrium. Adapted from [13].

32

Certain conditions are necessary for the occurrence of a peak, which may be a maximum
or minimum depending on the Sign convention used, in the response curves of a recovery
experiment. The first condition is that there is a distribution of relaxation times. If there is
only one relaxation time or mechanism, the recovery as a function of time will vary
monotonically and will not show a maximum. A second condition is that the recovery
time, ‘tl’, should not be too long nor too short; it should be longer than the shortest
relaxation time, and shorter than the longest relaxation time. If the recovery time, ‘tl’, is
longer than the longest relaxation time, the material attains equilibrium during the first
loading period in a stress relaxation experiment. If ‘t;’ is shorter than the shortest
relaxation time, the material will not have sufficient time to respond to the applied strain
during the first stress relaxation period. In other words, the stress relaxation in the first
period must be considerable without the material having attained equilibrium.
Alternatively, a peak in the response curve can be explained more directly through
the use of a generalized Maxwell model, which is diagramed in Figure 3-4. A generalized
Maxwell model consists of an array of individual Maxwell elements connected in
parallel. Since the relaxation times of the different elements in the model are not the
same, rapid units will reach equilibrium during the first strain loading period, and slow
units will not. During the second period, the rapid units which have reached equilibrium,
will have to return (i.e. expand); while the slower units will continue to relax in the same
direction [2]. Therefore, the system departs from its apparent equilibrium, passes through
a maximum or minimum, and then approaches true equilibrium. The peak corresponds to

the point at which the short-time contributions exactly compensate the long-time

33

 

 

A

Smmg

Dashpot

A

Figure 3-4 Schematic diagram of a generalized Maxwell model.

34

contributions. The “memory effect” peak increases in magnitude and occurs at a shorter
time for samples isothermally aged at temperatures the further below Tg [13].

The goal of this work was to study the effects of physical aging on the Viscoelastic
response of an epoxy/diamine network modified with a monofunctional epoxy (Heloxy-
62). Addition of the monofunctional epoxy results in the formation of a network that
contains dangling chains. These chains are only linked at one end and thus do not
contribute to the overall crosslink structure; however, the chains may have an effect on
the relaxation response of the modified networks. Small strain relaxation tests in simple
extension for various aging times and temperatures were be used to develop a

relationship between the Viscoelastic response and the epoxy network.

35

3.2 Viscoelastic Response

3.2.1 Isothermal Aging

The aging behavior was probed using uniaxial extension under stress relaxation
conditions. The mechanical tests were performed using a computer controlled servo-
hydraulic testing machine (Instron Model 1321) equipped with an oven for temperature
control. Cylindrical dumb-bell shaped specimens were first annealed, for 30 minutes, at
20°C above the glass transition temperature of the particular system studied to remove
any previous aging, and then quickly placed in the testing machine at a temperature
where the amorphous glass specimens began to age. Successive deformations were
applied at aging times, t., which approximately doubled with each application, [namely
te=60 min, 120 min, 240 min, etc]. At each interval of deformation, the ratio of the
deformation duration time, ti, to aging time was 0.05. Figure 3-5 is a plot of the applied
strain history. For the stress relaxation tests, the applied strain, a, in these tests was 0.1%
and was measured with an Instron extensometer. The stress, o(t), was measured using a
22001b (900 kg) capacity load cell. Thus the stress relaxation modulus could be
calculated from the following relation:

E(t)= o (t)/e (3-1)

The isothermal stress relaxation curves for the modified epoxy networks are depicted in

Figures 3-6 through 3-8.

36

 

l l l I
Strain/Creep Time =0.05 x t.

' 1
' 1
’ ‘
_ —
’ 1- <
1 1

 

 

 

 

 

 

 

 

 

m
E : 3
5 '_ :
g . -.
g I
(D L.
“a » ".
.9; <
a 1
Q __ .
< _‘.
5???” "“1 "WW
I l l l
0 20 40 60 80 100 '
Aging Time (min.)

Figure 3-5. Schematic diagram of the applied strain history. Applied strain:
e=0.1%

37

5 wt% Heloxy/DGEBAID23O T = 59.7 °C a = 0.1%

4
a 3
D.
9
(D
2
3
O
2 2
C
.9
a
E
G)
D:
1

 

10'1 10° 101 102 103 104 105

Time (Seconds)

Figure 3-6. Stress relaxation modulus curves for DGEBA+95 wt% D230-+5 wt%
Heloxy-62 quenched from 20°C above Tg and aged at 597°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)20 (I)4.0 (0)8.0
(0)160 (0)320 (0)640. '

38

 

 

 

 

10 % Heloxy (DGEBA/0230) T = 54.9 °C a = 0.1 %
I I IIIIIII I IIIIIIII I I IIIIIII I TIIIIIII I I IIIIIII I I IIIT
3.0 _
Ta = :3‘:--- -
D. E=E?§::-“:i£:; ;;;;
g 1 5 ~ . ”:09. NM ‘~.‘\‘
g " ‘ ”“
g 2.0 - -
O .
2
C
.9
iii
g ' o
0 o 0
tr ’ .
1.0 — ° 1 -
0.90 - -
0.80 l IILIIIII l 1 1111111 I l lllllll I 1 1111111 1 1 Illllll l L11L111
10'1 10° 10‘ 102 103 104 105
Time (Seconds)
Figure 3-7. Stress relaxation modulus curves for DGEBA+90 wt% D230+10 wt%

Heloxy-62 quenched fi'om 20°C above Tg and aged at 549°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)20 (l)4.0 (0)8.0
(0)160 (0)320 (0)640.

39

 

20 °/. Heloxy (DGEBA/0230) T: 41.8 °C a = 0.1%
I I IIIIIII I I 111111! I I IIIIIIr I I 1111111 I IIIIIII] I I IIIITI
3'0 11§§§_}§?EEEEE:=‘- —1
r ~~~~~~~~ :: a...
g 200 1- 1 ‘ . . . . ‘ .1
S ‘ ‘ ‘ a
'O .
O
2
C
.9 .0 o 9 g
1‘5 n . o .
3 1 0 A ' 3' ‘ _
CD ' o . .
I 0090 1- . A . .1
0.80 _ . , -
0.70 " a ..
0.60 1 r 1 111111 1 1 1411111 1 1 1111J11 1 1 1111111 1 1 1 Lllll1 1 11 14411

 

 

 

10'1 10° 101 102 103 104 105

Time (Seconds)

Figure 3-8. Stress relaxation modulus curves for DGEBA+80 wt% D230+20 wt%
Heloxy-62 quenched fiom 20°C above Tg and aged at 418°C. Applied
strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0 (0)40 (0)8.0
(0)160 (0)320 (0)640.

3.2.2 Memory Effect

The memory effect phenomena was probed in a stress relaxation experiment using
the same procedures as for the isothermal aging experiments. However, prior to testing
the DGEBA/D230 sample which contained 10 wt% Heloxy was quenched from 80°C
(20°C above Tg) to 468°C and 387°C and annealed at those respective temperatures for
116 hours. Subsequently, the sample was “jumped” to the testing temperature of 59°C
where the glass began to age. The stress relaxation experiment was then conducted at
various aging times. Figure 3-9 is the isothermal stress relaxation curves for the sample

annealed at 468°C and then “jumped” to the testing temperature of 59°C.

 

41

10% Heloxy-62

 

 

 

 

1 01 I TIIIIII] I I IIIIIII I I IIIIIII I TIIIIII] I I IIIIIII I I IIIIIL

I- ..

’5 -1
0.

g .
a:
3
5

8 ..

E 1

.9 ‘

E," 1

g d
an

1: 1

10'1 1 11111111 1 11111111 1 11111111 1 11111111 1 11111111 14111111

1 0'1 1 0° 1 01 1 02 1 03 1 04 1 05
Time (Seconds)
Figure 3-9. Stress relaxation modulus curves for DGEBA+9O wt%D230+lO wt%

Heloxy-62 quenched from 20°C above Tg and aged at 59°C. Prior to the
mechanical test, the sample was annealed at 468°C for 116 hours.
Applied strain is 0.1%. Aging time in hours: (0)0.5 (0)1.0 (0)2.0 (I)4.0
(<>)8.0 (0)160 (A)32.0 ((1)640.

 

42

3.2.3 Data Analysis Method
Each stress relaxation curve at a given aging time was curve fitted with the
Kohlrausch-Williams-Watts [14,15] stretched exponential function:

E<t>=Eoepot/0"1 (3—2)
where E0 is the modulus at t=0, 13 is a characteristic relaxation time, and B describes the
shape of the relaxation curve and the width of the Viscoelastic spectrum. The ability to
perform superposition and form a master curve requires that the value of B at various
aging times must be the same, i.e. the shape of the relaxation spectrum must be invariant.
Thus, horizontal shifts of the relaxation curves, am, along the time axis can be defined in
terms of the KWW function as:

aae=T(te)/ 1: (te(ref)) (3-3)
where the ‘t (te) is the value of t at the relevant aging time and r (t¢(ref)) is the value of r
at the reference aging time. The stress relaxation curves for the isothermally aged
samples were fitted with equation 3-2 and the KWW parameters are presented in Table 3-
1.

The aging process at different temperatures is conveniently characterized by the
double-logarithmic shift rate, 11 ,defined as [2] :

u=d log (eta/d log (te) (3-4)

43

Table 3-1. KWW curve fitting parameters. 8:0.1%', Ta: aging temperature

 

 

Sample t (1112) E0 (GPa) 1 (see) p
DGEBA+95%D230+5% Heloxy 0.5 2.6183 1903.1 0.3
T,: 60°C 1 2.6154 3408.6 0.3
2 2.654 7315.1 0.28
4 2.698 14978 0.27
8 2.697 20691 0.28
16 2.705 41143 0.28
32 2.748 81164 0.285
64 2.755 1 .525+05 0.28
DGEBA+90%D230+10% Heloxy 0.5 2.763 224.3 0.295
Ta: 55°C 1 2.779 492.1 0.292
2 2.806 903.47 0.285
4 2.798 1573.7 0.291
8 2.82 2580.7 0.291
16 2.841 3918.2 0.29
32 2.876 6796.6 0.289
64 2.897 10233 0.285
DGEBA+80%DZ30+20% Heloxy 0.5 3.13 223.1 0.259
Ta: 41 .8°C 1 3.13 274.9 0.274
2 3.21 430.4 0.267
4 3.23 643.7 0.273
8 3.28 999.8 0.269
16 3.3 1471.8 0.265
32 3.31 2125 0.269
64 3.31 2972 0.275

 

3.3 Results and Discussion

3.3.] Isothermal Aging

As seen in Figures 3-6 through 3-8 it is apparent that as the aging time increases,
the relaxation modulus shifts to the right along the time axis. Applying time-aging time
superposition to such data allows one to determine the shift factor, am, for these data sets.
The KWW parameters given in Table 3-1 indicate that the instantaneous modulus, E, and
the shape parameter, B, do not vary with aging time for a given epoxy network, thus each
relaxation curve may be shifted along the time axis to form a single master curve. When
the fraction of Heloxy content in the epoxy network increases, the value of [3 decreases
indicating a broadening of the Viscoelastic spectrum. In addition, since the values of B are
independent of aging time, a classical time-aging time superposition appears to be valid
for these epoxy networks. The instantaneous modulus, E0, did not change by increasing
the Heloxy 62 content. The monofunctional modifier is only crosslinked at one end; i.e.
dangling, thus it does not have an effect on any load bearing properties. Thus the number
of dangling chains present does not affect the instantaneous modulus; however it does
affect the relaxation behavior of the network.

Figures 3-10 through 3-13 are the double logarithmic plots of the aging time shift
factor, am, versus aging time, te, at different values of T-Tg for each modified network.
These plots reveal that the value of log(%) increases with increasing aging time at a
temperature far below Tg, but shows a more complicated behavior at temperatures closer
to Tg. As the temperature increased, the material showed a regime of rapid aging

followed by one in which aging virtually ceased. Figures 3-14 through 3-16 reveal that as

45

the fraction of Heloxy present in the epoxy network increased the rate of shifiing, i.e. the
slope, u, of the plot decreased.

In addition, for systems which were aged into structural equilibrium at 10°C below
Tg, the time required to reach structural equilibrium decreased with an increase in the

fraction of Heloxy content present in the epoxy network glasses.

 

 

 

 

 

3.0 1 I
2.5 - —
9
2.0 —- [’31 _
l’ D
.2 x?
a I
1.5 - , ’ .....——-O--- -—
3' ,xi ”’4 ...
.1 ’6, 1*
1.0 -— 18',” _
0.5 - ,z” e _.
,3' - °
0 ”II . 1 1
+’
101 102 ' 103 104
Aging Time (Min)
Figure 3-10. Double logarithmic plot of aging time shift factor, 81,, versus aging time

for DGEBA/D230 samples at different aging temperatures. Reference
aging time is 28 minutes. Applied strain is 0.25%. Aging temperature in

°C: (0)576 (x)61.9 (0)669 (A)7l.5 (0)818. 11 20.9.

 

100
5‘2
0‘? 10
E
0:)
1
Figure 3-11.

47

 

 

 

 

5 wt% Heloxy/DGEBAID230 e = 0.1%
1' I I I I I I I I I I r I I r I I IT I I I I I I I I_
: ° T = 59.7 °c :
— ° T = 65.3 °c «
__ 8 a _
E i . ° T = 70.4 °C 3
. a _
:1 I 1# 1 iJ_lJ l l L l 1 L4 11 l 1 I 1 44 1 1
103 104 105 106

Aging Time (seconds)

Double logarithmic plot of aging time shift factor, 8“, versus aging time
for DGEBA/D230/5 wt% Heloxy samples at different aging temperatures.
Reference aging time is 30 minutes. Applied strain is 0.1%. Aging

temperature in °C: (0)597 (0)653 (0)704.

 

 

 

 

 

4s
10 7- HELOXYIDGEBA/D230 e = 0.1%
100 __ I T T I I I I I I I I I I I I I I I I I I I I I I I“
: ° T = 50 0C I
: . T = 55 °c -
l. _
5 , °
8 ° T = 60 °C
“- 10 :- 0 a —:
E i 3 0 ° 1
"’ " :
1 :1 I 1 l l I L 11 I I 1 J l l l 11 l l l 1 l 1 J l
103 104 105 106
Aging Time (Seconds)

Figure 3-12. Double logarithmic plot of aging time shift factor, ac, versus aging time
for DGEBA/D230/ 10 wt% Heloxy samples at different aging
temperatures. Reference aging time is 30 minutes. Applied strain is 0.1%.

Aging temperature in °C: (0)503 (0)551 (0)602.

1‘00
.5
8

I}: 10
E
a)

1

Figure 3-13.

49

 

 

 

 

20 wt% Heloxy/DGEBAIDZ30 e = 0.1% Tg = 56 °c

I I I F r VI I_]' I I I I I I I I I I I I I I I r _

’ T = 37 °c ‘

° ° T = 42 °C
b o 3 a o a T = 46 °C -
h o O ..
= l 4 .4 J I 1 l 1 1 l 1 I l l l I 11 l 1 l I l I l I
103 104 105 106
Aging Time (Seconds)

Double logarithmic plot of aging time shifi factor, a.., versus aging time
for DGEBA/D230/20 wt% Heloxy samples at different aging
temperatures. Reference aging time is 30 minutes. Applied strain is 0.1%.

Aging temperature in °C: (O)37.l (0)424 (0)463.

100 5 wt% Heloxy/DGEBAIDZ3O T = 65.3 °C 8 = 0.1%
I I IIIIIII I I IIIIIIT I I IIII

 

IIIIW

I

  

1 O
Slope = 0.83

    

Shift Factor

 

 

1 1 11111111 1 111114;

103 104 105 106

Aging Time (seconds)

 

Figure 3-14. Double logarithmic plot of aging time shift factor, an, versus aging time
for DGEBA/D2305 wt% Heloxy samples at an aging temperatures of
653°C. u=0.83.

51

1O % Heloxy (DGEBA/0230) T = 54.9 °C 8 = 0.1 %

2
1O IfIIIIIII I I IIIIII' F I IIIII

 

'llll

IIIII

   
 

 

 

 

5‘2
8
“- 1 O r j
E : Slope = 0.769 :
a) » _
1 O : 1 l— I 1 J 1 1 l 1 40 I 1 1 I l 1 L1 1 J J l l 1 I I
103 104 105 106
Aging Time (Seconds)
Figure 3-15. Double logarithmic plot of aging time shifi factor, 8“, versus aging time

for DGEBA/D230/10 wt% Heloxy samples at an aging temperatures of
549°C. u=0.77.

52

20 % Heloxy (DGEBA/0230) T=41.8°C 8 =0.1%

2
10 r I Iij—III I I I IIIIII I I I IIIII

 

IIIIII
Llllll

. -_- —.—1

I
T.
I

I

 
   
  

 

 

 

 

§
8
“- 10 :- __
g _ -
Etc) 3 3
~ Slope =0.558 -
1.0 41 J L 1 1 1 1 111 1 1 1 1 1 1.11
103 104 105 106

Aging Time (SecondS)

Figure 3-16. Double logarithmic plot of aging time shifi factor, 81., versus aging time
for DGEBA/D230/20 wt% Heloxy samples at an aging temperatures of
418°C. u=056.

53

3.3.2 Memory Effect

Figure 3-17 is a plot of the characteristic relaxation time versus aging time for an
epoxy sample containing 10 wt% Heloxy. The curves are similar, though inverted, to the
volume relaxation curves in Figure 3-3 developed by Kovacs where in be defined the
volume departure as a positive quantity. During the step up-jump in temperature, the
characteristic relaxation time crosses over the actual equilibrium, exhibits a minimum,
and then asymptotically approaches the simple approach curve indicated by the bold
circles in Figure 3-17. The position of the minimum on the graph is dependent on the
thermal history applied to the sample. The minimum moves to shorter aging and
characteristic relaxation times as the sample is quenched deeper into the glassy region.
The results, in agreement with Kovacs’ volume recovery experiments, indicate that it is
possible to use the characteristic relaxation time to monitor the kinetics of the glass

forming process.

1 (Seconds)

Figure 3-17.

54

 

 

 

 

4
10 T T f IIIII T I Y I Y TYYYT I f 1 I TYYT
.
C e
L
1000 ,- o —,
l 1
1.
a . o 9 8
a 5 D
O
0
100 :- 0 j
. O U 4
L .
1- O D 4
4
10 4 1 1 L IILIJ L 1 1 1 1 11_Ll 1
1000 10‘ 105 106

Aging Time (Seconds)

Characteristic relaxation time, 1, versus aging time for DGEBA+90 wt%
D230+10 wt% Heloxy-62 at a testing temperature of 59°C. Applied strain
is 0.1%. (0) Sample was quenched directly from 80°C to 59°C; (0)
Sample was quenched from 80°C to 46.8 °C and annealed for 116 hours,
then jumped to 59°C; (0) Sample was quenched from 80°C to 387°C and
annealed for 116 hours, then jumped to 59°C.

CHAPTER 4 MOISTURE ABSORPTION

4.1 Background

Epoxy resins are often utilized as glassy composite matrices in the aerospace and
automotive industries. In polymer based composites, the long-term stability of the matrix
is an important aspect of the performance. Since these materials are used in different
temperature and humidity conditions, knowledge concerning the polymers’ water
sorption and transport behavior in response to glassy structural changes is needed in order
to predict short-term as well as long-term performance in service. Many investigations
have been conducted in an attempt to develop a fundamental understanding of how
moisture transport is affected by epoxy network structures [16,17]. Most of the sorption
kinetic studies have focused on the effect of varying the amine (curing agent) content or
changing the curing conditions [16-18]. Carfagna [19] investigated the effect of
prepolymer composition on the water sorption behavior of amine hardened epoxy resins.
Synthesizing epoxies and hardeners with off-stoichiometric ratios will produce either
excess epoxides or excess hardeners in the system. If epoxide is present in excess, the
secondary hydroxyl groups, could add to the remaining epoxide rings until the glass
transition temperature of the forming network exceeds the curing temperature and the
reaction becomes diffusion-controlled [20]. When the resin is crosslinked with an excess
of hardener there is an increase in the number of hydrogen bonding sites due to the
presence of unreacted amines. Carfagna concluded that resin systems with an excess of
hardener absorbed more water (i.e. higher water equilibrium level) due to the increase in

hydrogen bonding sites which attract and accommodate water molecules [19].

55

56

The other technique utilized in the investigation of sorption kinetics is to alter the
curing conditions [21,22]. The morphology and chemistry of a polymer system is
dependent upon the degree of crosslinking in the resin. Since crosslinking is a thermally
activated process, curing at higher temperatures is expected to increase the crosslink
density and number of hydroxyl groups in the polymer. Moy and Karasz [23] studied the
importance of these effects by examining the water sorption behavior as a function of the
extent of curing time. These researchers found that the equilibrium sorption of water in
the resin was markedly affected by the curing conditions.

In addition, most of the studies have discussed network structure effects due to a
change of the curing agent and their amount. Altering the curing agent, however, alters
the resin polarity. Thus, moisture transport may be affected by the “chemistry” of the
resin. In view of this, it becomes difficult to predict the moisture absorption behavior or
to isolate the contribution of each effect [16]. In an attempt to isolate the structural effects
only, the epoxy networks produced for this study will be altered while maintaining nearly
the same polymer polarity or moisture affinity. The aim of this study is to investigate the

effect of aging on the transport properties of epoxy modified network structures.

4.2 Water Transport in Epoxy Resins

The transport of water in glassy epoxy resins is generally studied by the sorption
method in which a polymer sheet is suddenly exposed to water or water vapor and the
change in weight of the sample is measured as a function of time. Analytical means in
determining the values of the moisture content as a function of time may be obtained

when the following conditions are met [24]:

 

57
1) The temperature inside the material approaches equilibrium faster than the

moisture concentration; thus the energy and mass transfer equations are
decoupled.

2) Moisture diffusion can be described by a concentration-dependent form of
Fick’s law.

3) The thermal conductivity and mass diffusivity depend only on temperature
and are independent of the moisture concentration or stress levels inside the
material.

If the surface concentration of water on the epoxy sheet is constant at all times and the
mass transport process is completely diffusion controlled, then the absorption process
under isothermal conditions is said to be “Fickian”.

Brewis [25] investigated water sorption kinetics of Diglycidyl ether of bisphenol A
cured with various hardeners. Ghorbel and Valentin [26] studied the hygrothermal effects
of pure and glass fiber reinforced polyester and vinylester resins. In both investigations,
the initial moisture uptake as a function of the square root of time were linear implying
Fickian diffirsion.

Due to the complexity of the absorption mechanism, sorption of liquids and vapors
in glassy polymers is not always described by classic Fick’s laws. Apicella has reported
anomalous sorption behaviors for numerous polymer diluent systems [27]. For example,
moisture absorption at high temperatures and relative humidities can cause voids and/or
microcracks in the polymer which, if extensively present, cause a non-Fickian sorption
process. The formed voids/cracks facilitate diffusion by creating new channels for water

penetration and storage and as a result the absorption process is nonlinear (i.e. non-

 

J‘

58
Fickian)[l6]. Glassy polymers respond slowly to changing conditions and most of the

anomalous effects are directly related to this slow response. For example, changing the
polymer structure influences the solubility and diffusional mobility. This leads to
diffusion coefi'rcients which can depend on the previous time-temperature history and
rates of change of the system [28]. Therefore, F ickian and/or non-Fickian transport may
occur depending on the cured polymer properties, thermal and time exposure history of
the polymer, and environmental conditions. Some studies have shown that, for the same
epoxy resin, physical damage occurs while other studies have not observed any damage.
Such varied results have added to the confirsion regarding the already complex nature of
the sorption process in polymers.

The diffusion of penetrant molecules into polymers depends on two factors, namely,
the availability of appropriate molecular size holes in the polymer network and the
attraction forces between the penetrant molecules and the polymer. The presence of holes
is determined by the polymer structure and morphology reflected in its degree of
crystallinity, crosslink density, and molecular chain stiffness. Thus the first factor
affecting diffusion processes is essentially a structural one and is typically referred to as
the “volumetric” approach. The “volumetric” or “free volume” approach presumes that
water diffuses into the epoxy resin and resides in the the volume in an unbound state,
where it is relatively free to travel through the free volume voids [29]. Free volume is
defined as the difference between the measured volume of a polymer and the occupied
volume which is the volume occupied by the actual mass of the molecule plus the volume

a molecule occupies as a result of thermally dependent harmonic vibrations. The free

59

volume manifests itself as “holes” or ‘Voids” caused by packing irregularities in the
occupied volume.

Diamant [ 16] investigated the effect of network structure on moisture absorption of
a diglycidyl ether of bisphenol A resin systems cured with a stoichiometric amount of m-
phenylenediamine/aniline mixtures. The effect of replacing amounts of the original
diamine hardener with aniline was to increase the molecular weight of the chain between
adjacent crosslinks and thus decrease the crosslink density. The moisture absorption
experiments revealed that resin systems with higher chain molecular weights between
crosslink points exhibited lower diffusion parameters. In addition, all samples obeyed
Fickian diffusion laws.

The second factor affecting diffusion processes concerns the chemical nature of the
penetrant versus that of the polymer and is referred to as the “interaction” approach. This
approach states that water molecules are neither distributed randomly in the network nor
concentrated in morphological defects, but rather linked by strong hydrogen bonds to
some hydrophilic loci, mainly hydroxyls or amines [17]. For example, the relatively high
water absorption capacity of epoxy resins results fi'om the presence of hydroxyl groups (-
OH) in the epoxy chains which attract the polar water molecules.

Choi [30] studied network systems, which contained either di-functional, tri-
functional, or tetra-functional epoxide monomers, cured with an amine terminated PPO.
Results fiom the moisture absorption experiment reveal that the diffusion process of the
crosslinked epoxy of higher functionality is hindered by its higher crosslink density. In

addition, the epoxy resin containing the highest number of hydroxyl groups per mole

60

created more hydroxyl sites, which attract and accommodate water, resulting in a higher

moisture weight gain for the tri-functional epoxy network.

4.3 Gravimetric Moisture Absorption

4.3.1 Diffusion

As shown in Figure 4-1 and developed by Shen and Springer [31], a plate of
thickness, ‘h’, exposed on two sides to the same moisture environment is considered to be
infinite in the y- and z- directions such that the moisture content and temperature inside
the plate varies only along the x-direction (one-dimensional). Initially the temperature T1
and the moisture concentration C; inside the sample are zero. Subsequently, the sample is
exposed to a moisture environment where the temperature T. and moisture concentration
C. are constant. The objective is to determine the moisture content, ‘M’, of the sample as
a function of time, ‘t’, in the moist environment. If the diffusion coefficient is constant,
and there is a concentration gradient only along the x-axis, the problem can be described
by Fick’s second law of diffusion as follows:

6C(x,r) 62C(x,t)
_: D ___2__
a: x 6x

(4-1)
C=Ci 0<x<h t=0
C=Ca x=0', x=h t>0
where Dx is the diffusion coelfrcient of the material in a direction normal to the surface.

The solution of Eq. 4-1 for the given boundary conditions is given by lost [3 2].

 

Sample

Water Water x

Ila/4 ”a I .5 '/
[If ‘51, r / -‘,/ /3, /. ,4; 5,2 I, 1’5“ /'j “fit-:1 N ,9 :41
l ,‘I‘ .7," «j ,: 3,926.}, ,1 5g, 7, ’ 31,. .1 3A.,

5 ,g'ioz/
.11, // qr. -"./."/ ,1. , .5 35”.. 5 . ,. ,/ 1',/"/"/V.'/’/3‘A'3’.//"1 '1" 7.

 

Figure 4—1. Conditioning environment

62

 

 

 

4 °° l . 2 +1) 2 +1 Dt
= l — — Z . sm £1—'—7rx_exp[ _ (J ) ] (4_2)

Where C M is the maximum moisture concentration. The total weight of moisture in the

material can be obtained by integrating over the sample’s thickness:
In
M(t) = [C(x,z)dx (4-3)
0

The result of this integration is

M.

I

, Dr
M _ 8 ex131:’(2./ + 02 772 (27)]
M, —M, 2

G:
7r ; (2j-l-l)2

 

 

M
= ‘ =l_ 4'4
Ma, ( )

where ‘G’ is a time dependent parameter of the ratio of moisture content at time, ‘M’, to

the equilibrium moisture content ‘M .0 ’. ‘M ,. ’ is the initial weight of the moisture in the
material and ‘ M m ’ is the weight of moisture in the material when the sample is
completely saturated. In addition M —M,. equals ‘M , ’ which is the moisture content in
the material at time t =t and M m —-M,. equals ‘M .0 ’ which is the saturated moisture

content of the sample at time t = 00.

Equation 4-4 may be approximated by the expression

Dr 0 75
h 2 (4 5)

63
Comparison of Eq. 44 and Eq. 4—5 is shown in Figure 4-2. Since the approximation

curve is in good agreement with the theoretical curve, the equilibrium moisture content

and diffusivity can be obtained by using non-linear least-square curve fitting.

 

 

L 1

 

O
on
fTIIITrIrIIrrITrIIITIII
4111

 

 

0 1 41411111 L 14111111 1 1 ILLIIL

10“ 105 10°
Time (sec)

 

E

Figure 4-2. Variation of G with log time. Solid line is based on Eq. 44 representing
the exact solution and the dashed line is based on Eq. 4-5 representing the
approximate solution.

 

65
4.3.2 Experimental Procedure

Epoxy samples containing various amounts of Heloxy 62 and cured with mPDA
were lightly sanded with a 240 grit sandpaper in order to remove any release agent, which
may have leeched to the samples. All of the specimens were then heated to 20°C above
their respective glass transition temperature in order to remove any residual moisture and
to erase any aging which occurred during storage. A set of specimens from each system
was then physically aged, while the remaining samples were not conditioned prior to
testing. The aging process consisted of placing the samples between two metal plates
which, were in temperature equilibrium inside an oven set at a specific aging
temperature. The samples were allowed to isothermally age for 5 days. Subsequent to the
aging process, the “as-cured” samples and the “aged” samples were weighed on an
analytical balance with 01mg resolution in order to establish a pre-conditioned (i.e. dry)
weight. The specimens were immersed in distilled water chambers which were
maintained at a constant temperature of 60°C with 01°C accuracy. Samples exposed to
the moisture environment were periodically removed from the bath, gently wiped dry to
remove excess surface water, and weighed. Measurements were made every half-hour
during the early stages of the experiment so as to obtain a sufficient amount of data points
during the initial linear portion of the uptake curve. As the experiment progressed
measurements were made periodically until a steady state condition was reached. Three
specimens from each system were used and the average %weight gain at each

measurement time was calculated.

4.3.3 Analysis

The samples used were thin enough, with the ratio of edge area to total surface area
small enough, so as to ignore edge effects, facilitating the validity of the one dimensional
diffusion model developed previously. Gravimetric liquid sorption analysis was used to
determine the moisture content of the samples as a function of time. Assuming that the
sample’s weight gain was due to moisture uptake, percent moisture content, can be

obtained from Eq. 4—6

M= ;, " x 100 (4-6)

I

 

where M is the percent weight gain (%), W is the weight of the moist sample at time t,
and W, is the initial weight of the dry sample. Moisture absorption profiles are plotted in
Figures 4—3 through 4-6 as a function of % weight gain versus time (seconds). Equation
4-5 and non-linear least-square curve fitting was used to fit the experimental data and

obtain values for M.o for the Heloxy modified samples.

4.4 Results and Discussion

4.4.1 Cross-link Density Effects

Figure 4-3 is a plot of the moisture absorption profiles for unaged samples
containing 0 wt%, 5 wt%, 10 wt%, and 20 wt% Heloxy. As evidenced from the curves,
addition of the monofunctional epoxy caused a reduction in the maximum moisture
content of the samples. The diffusion of small molecules, such as water, into glassy
materials is argued to be dependent on the “free volume” of the glassy structure. The
results presented in Figure 4-3 and the values of M“, displayed in Table 4-1 indicate that

the available path for diflirsion plays a more significant role than the amount of fi'ee

 

 

67

Table 4-1. Equilibrium moisture values of Epoxy modified networks.

 

 

 

 

wt% As-cured Aged %
Heloxy Mm Md, Difference
0 3 .36 3 .29 0.07
10 2. 82 2.53 0.29
20 2.69 2.03 0.66

 

 

 

 

 

 

volume available. Samples containing Heloxy have a lower crosslink density do to
network chain scissions. Moy and Karasz [23] investigated the effect of crosslink density
on the transport of moisture in glassy materials. These researchers found that a decrease
in the rate of absorption and equilibrium moisture content occurred when the curing
temperature was increased. Since crosslinking is a thermally-activated process, curing at
a higher temperature is expected to increase the cross-link density. This suggests that
samples with a lower crosslink density should have a comparatively higher rate of
moisture absorption. Thus, in this investigation it was expected that the samples
containing Heloxy would absorb more moisture than the samples containing no Heloxy.
The results, however, did not show this, but instead the samples containing Heloxy
absorbed less moisture. Therefore, these results indicate that morphological features
control the transport of moisture in these epoxy systems. It is believed that moisture
transport is obstructed because of the mobility of the dangling chains present in the
network In addition, the moisture absorption profiles for the samples containing 10 wt%
and 20 wt% Heloxy are nearly identical. This suggests that the reduction in transport

properties is independent of the amount of Heloxy present above some critical amount.

 

 

 

 

 

 

68

T T I T T I I 4
o 0 ~1
o O I
o 0 U '1
A .
1% o A 1
0 g
H -1
'8, 1
g s:
32 .1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 T

s s

110'5 1.5105 210 2.510

 

Time (seconds)

Figure 4-3. Moisture absorption behavior of unaged DGEBA/mPDA samples
containing: (0)0 wt% Heloxy; (0)5 wt% Heloxy (0)10 wt% Heloxy;
and (A)20 wt% Heloxy. Bath temperature is 60°C.

69
4.4.2 Aging Effects

Figures 4-4 through 4-6 compare the absorption profiles of “ideal” and Heloxy
modified networks in aged and as-cured epoxy glasses. The aging behavior of the glasses
was probed at the same temperature relative to the particular sample’s glass transition
temperature. The aging temperature used was Tg-13°C.

Figure 4-4 is a plot of the moisture absorption profile for aged and as-cured samples
containing no Heloxy. It can be seen that there is virtually no effect on the moisture
transport properties of samples which did no contain Heloxy. Since the network present
in this sample is considered to be “ideal”, it is assumed that there are relatively few
network defects. Therefore it is believed the volume and enthalpy recovery response
associated with physical aging in glassy materials occurred homogeneously in the
sample. The small water molecules were not affected by this uniform relaxation and thus
there was no affect on the transport properties.

Figures 4-5 and 4-6 are plots of the moisture absorption profiles for aged and as-
cured samples containing Heloxy. The effect of aging on networks which contain
dangling chains is evident. For networks with large amounts of scissions, the path of
diffusion is obstructed during the isothermal aging process. These results suggest that the
isothermal aging process results in a heterogeneous relaxation of these modified network
glasses. The aging process affected the samples containing 20 wt% Heloxy more
significantly than those containing 10 wt% because of the stability of the network. The
networks containing more dangling chains or chain scissions are less stable and collapse

more readily when subjected to an isothermal volume recovery process such as physical

aging.

 

70

 

.01
01

A N
01 N 0'! (a)
IIIIIIIIIIIT'IIIIIIIII'IIII

% Weight Gain

u—l

 

9
001

0

Figure 4-4.

 

1111111111111

11

1L11111111111111111

 

1110114411 4 141L1 l L1Ll 1111 1 14

510‘ 1105 1.5105 2105 2.5105 3105
Time(seconds)

Moisture absorption behavior of aged and unaged DGEBA/mPDA
sammes containing 0 wt% Heloxy. Bath temperature is 60°C. Samples
directly quenched into the water bath from above Tg are represented by
(I) symbols. Samples aged at a temperature of Tg-13°C for 120 hours
are represented by (0) symbols.

71

 

% Weight Gain

 

11L111111J111111

1111L1

 

 

;
'1
0 1 1 . 1 1 1 1 . 1 1 1 1 L . 1 1 L ‘1
4 5 5
1 2 10 2. 5 10
O 510 Time (secondssl)10;
Figure 4-5. Moisture absorption behavior of aged and unaged DGEBA/mPDA

samples containing 10 wt% Heloxy. Bath temperature is 60°C. Samples
directly quenched into the water bath from above Tg are represented by
(I) symbols. Samples aged at a temperature of Tg-13°C for 120 hours
are represented by (0) symbols.

 

% Weight Gain

72

 

 

 
 

 

   
   

 

 

3 1 T I I I T I 1 T 1 1
: I _
2.5— /—&-F“‘—‘“' L
: i/ 'l :
/ -4
2 r / I I o O ‘2
s . r
1.5 :- I ‘:
0.5 —
0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
o 510‘ 1 105 1.5105 2105 2.5105
Time (seconds)
Figure 4-6. Moisture absorption behavior of aged and unaged DGEBA/mPDA

samples containing 20 wt% Heloxy. Bath temperature is 60°C.
Samples directly quenched into the water bath fi'om above Tg are

represented by (I) symbols. Samples aged at a temperature of Tg-
13°C for 120 hours are represented by (0) symbols.

CHAPTER 5 SUMMARY

Polymer glasses utilized at temperatures below their glass transition are susceptible
to a process known as physical aging. With the growing importance of long-term
durability of polymeric materials, it is particularly important to determine the effects
physical aging will have on glassy polymers subjected to various mechanical and
environmental conditions. Therefore, the goal of this study was to determine the effects

of physical aging on modified network glasses.

5.1 Local Structure Effects on Viscoelasticity

Stress relaxation experiments were conducted on epoxy glasses which were
modified with a monofunctional epoxy in order to determine the effect of these
modifications on physical aged networks. The following results were achieved:

0 Classical time-aging time superposition was observed as a result of the
relaxation modulus, E0, and shape parameter, B, being independent of aging
time.

a The value of the shape parameter, [3, decreased as the fraction of
monofimctional epoxy, Heloxy 62, in the epoxy network increased.

0 The slope, ‘u’, of the Log(aging time shift factor) versus Log (aging time)
decreased with an increase in the Heloxy content. Thus, the aging process was

slowed down due to the presence of the “dangling chains”.

73

 

74

0 For systems which were aged into structural equilibrium at 10°C below their
glass transition, the time required to reach equilibrium decreased as the
fraction of Heloxy 62 was increased.

:1 The characteristic relaxation time demonstrates the “memory effect” after two

consecutive temperature jumps of opposite sign.

5.2 Local Structure Effects on Moisture Absorption

Moisture absorption experiments were conducted in order to determine the role of

 

network structure on the transport properties of aged epoxy glasses:

:1 The rate of moisture absorption in unaged epoxy glasses decreased as the
amount of Heloxy present in the network increased. A critical weight fraction
of Heloxy was achieved, above which the reduction of transport properties
became independent of the amount of Heloxy present.

0 Physical aging had no affect, as compared to unaged specimens, on the rate of
absorption of samples without Heloxy. However, aged samples containing
Heloxy showed a decrease in moisture transport properties. These decreases
became magnified as the fraction of Heloxy content increased.

:1 Therefore, the diffusion of small molecules into glassy materials is argued to

be dependent on the available paths for diffusion.

PART B: INFLUENCE OF METALLIC GLASS RIBBON REINFORCEMENT S

CHAPTER 6 INTRODUCTION

Initial studies have been carried out on the use of metallic glasses as a reinforcing
phase for ceramic and thermoplastic polymers. These studies have indicated the
tremendous potential of such composites for structural, aerospace, electrical and medical
applications [33,34]. Metallic glass ribbons are potential reinforcements because of the
unique properties which they possess. These properties include an ultra high tensile
strength (upto 4000 MPa), high surface to volume ratio [35], favorable surface
morphology [36], and good ductility and corrosion resistance [35]. In addition, metallic
glass ribbons have soft magnetic properties which provides the possibility of orienting
them into specific configurations, during composite fabrication, with the use of either a
magnetic or electric field. It is the intent of this study to determine the feasibility of
orienting metallic glass ribbon reinforcements with an external magnetic field during
composite fabrication so as to enhance the physical and mechanical properties of
complex-shaped polymer matrix composites along critical directions.

The initial studies conducted on the use of metallic glass ribbons as reinforcements
in thermoplastic polymer matrix composites have shown that such composites possess
relatively weak interfacial strengths [37]. The weak interface is the result of poor
adhesion or bonding between the matrix and metallic glass ribbons reinforcements.
Another aim of this study is to determine the bonding characteristics of a thermoset

polymer, namely epoxy, reinforced with metallic glass ribbons.

75

 

 

76

6.1 Materials Background

Commercially available metallic glasses are a relatively new class of
reinforcements that provide unique advantages over conventional fiber reinforcements.
Metallic glass ribbons are potential reinforcements because of the unique properties
which they possess. These properties include an ultra high tensile strength, high surface-
to-volume ratio, favorable surface morphology, good ductility and corrosion resistance.

Some physical properties of a typical metallic glass ribbon are presented in Table 6-1.

Table 6-1. Physical properties of a common metallic glass ribbon.

 

Material Chemical Crystallization Elastic Yield Density
Composition Temperature Modulus Strength
Metglas Fe-78%
26053-2®* 13-13% 550°C 85 GPa 1200 MPa 7180 Kg/m’

Si-9%
‘Metglas® is a registered trademark of Allied Signal Inc. for amorphous metallic alloys and brazing alloys

 

 

 

 

 

 

 

 

 

Of all of the properties of metallic glasses, it is their soft magnetic properties that
have led to the most significant applications. The microstructural homogeneity of the
glasses and the absence of magnetocrystalline anisotropy produces glasses which have a
high relative permeability and a narrow hysteresis loop [38] facilitating easy
magnetization of and requiring only a small external magnetic field to move the ribbon
pieces. The magnetic properties of the metallic glass ribbons used in this study are

presented in Table 6-2.

 

Table 6-2. Magnetic properties of metallic glass ribbons.

 

 

 

Coercive Remanent Relative Resistivity
Material Force Induction Permeability Ohm/cm x 10‘
Hc (Oersteds) Br (Gausses) (pmax)
Metglas 26055-2 .03 15,000 300,000 130

 

 

 

 

 

Metallic glasses are produced by rapidly quenching metallic melts. Cooling rates of
the order of 105 to 106 Ks'l are required in order to avoid crystallization. Because of the
requirement for rapid liquid quenching, metallic glass samples are available only in thin
cross sections (sheets or ribbons) which have a thickness of 25-100 um, and a width of l-
15 mm. The cross-sectional geometry is an important characteristic of metallic glasses.
Ribbon shaped reinforcements offer improved off-axis composite properties because
ribbons, unlike fibers, possess the ribbon width as an additional geometrical parameter
[39]. This produces a large interfacial surface area which aids in effective load transfer
from the matrix to the reinforcing phase. Therefore, attractive physical properties, in
addition to the unique geometry of metallic glasses, have led to use of metallic glass
ribbons as reinforcements in composites.

So far studies have been carried out on the use of metallic glasses as a reinforcing
phase for ceramic and polymer matrices. These studies have indicated the tremendous
potential of such composites for structural, aerospace, electrical and medical applications
[33,34].

Metallic-glass ribbons have been found to be successful as reinforcements in

ceramics, especially for glassy matrices. In particular, even at a very low volume fraction

 

78

of reinforcement, metallic glass ribbons significantly improved the elastic properties,
fracture strength and fracture toughness of brittle glass-ceramic matrices [40].

Metallic glasses have been studied as reinforcements for brittle polymer matrices by
Hornbogan et a1 [41]. The investigation showed that small volume fractions of metallic

glass ribbons have a considerable beneficial effect on the strength of a brittle glassy resin.

 

 

 

CHAPTER 7 BONDING CHARACTERISTICS

7.1 Background

In general, an interface is a critical point in a composite system. Therefore,
adhesion or bonding between the matrix and ribbon is an important issue in composite
fabrication. No matter what matrix material is chosen, good adhesion between the matrix
and reinforcing ribbon is one of the most important factors which determines the
mechanical and physical properties of a composite [35]. Various polymer matrices
reinforced with metallic glass ribbons have been fabricated in order to investigate the
interfacial characteristics of metallic glass ribbon reinforced polymer matrix composites.

Studies carried out on thermoplastic polymer matrix composites reinforced with
metallic glass ribbons have shown that they possess relatively weak interfacial shear
strengths [37]. The weak interface can be attributed to poor adhesion, or bonding,
between the thermoplastic polymer matrix and the metallic glass ribbon reinforcements.
A thermoplastic matrix reinforced with a ductile material has a weak mechanical bond
and the ductile matrix flows more easily than the reinforcement. For a thermoplastic
polymer composite reinforced with continuous ribbons this results in the reinforcements
failing first since the ribbons are unable to keep up with the stretching of the matrix.
However, for discontinuous ribbon reinforced composite systems, the matrix fails in
regions where there is no reinforcement. Due to poor adhesion between the matrix and
reinforcement the highly stressed matrix can not effectively transfer its load across the

interface to the stronger reinforcing phase.

79

 

80

This low interfacial shear strength leads to failure behaviors in which a crack,
initiated in the matrix, propagates to the interface and causes debonding between the
matrix and ribbons. The crack then propagates around the ribbon at the matrix/ribbon
interface and when it reaches the same plane on which it arrived it continues on along
that plane into the matrix. If the crack propagates to the other end of the sample any
further loading will cause a phenomenon known as ‘pull-out’. Because the bonds
between the matrix phase and reinforcing phase have been broken, the only force holding
the matrix and reinforcing phase together is fiiction. If the loading has sufficient energy
to overcome these fiictional forces the matrix will be pulled away from the ribbons.
However, if there is excellent bonding between the matrix and reinforcement a higher
interfacial shear strength is necessary in order to cause debonding.

A brittle matrix composite, such as thermoset polymer composites, reinforced with
a ductile material will not fail in the same manners as ductile matrix composites
reinforced with a ductile material. For continuous metallic glass ribbon reinforced
composite systems the matrix will fail first because it will not be able to stretch as much
as the reinforcements. A portion of this study is going to focus on the bonding
characteristics of a thermoset polymer matrix composite reinforced with a ductile
material. Specifically, an epoxy matrix reinforced with discontinuous metallic glass
ribbons will be examined in order to characterize the fracture behavior of these composite

systems.

 

81

7.2 Sample Preparation

The composite system examined in this study consisted of a thermoset polymer
matrix, specifically Epon 828 epoxy resin (Bisphenol A/Epichlorohydrin based) cured
with a stoichiometric amount of Jeffamine D400, and amorphous metallic glass ribbons
as the reinforcement. Metglas 26058-2 was the specific metallic glass reinforcement
used. A closed glass mold was used to cast reinforced plastic composite sheets with the
following dimensions 16.5 cm x 3.5 cm x .4 cm (Figure 7-1). The Epon 828 epoxy resin
was heated in an oven for 2 hours at approximately 60°C in order to decrease the
viscosity so as to allow for uniform and thorough mixing with the curing agent. The
epoxy resin and curing agent were weighed separately in aluminum weighing dishes, then
mixed by hand until uniform. The mixture was then degassed at room temperature for 20
minutes in a vacuum chamber. In the meantime, the Metglas ribbon, which came on a
spool fiom the manufacturer, was carefully cut into small pieces of various sizes. A
portion of the degassed mixture was then poured into the mold and allowed to settle.
Some ribbons were then added to the mold which was partially filled with the
epoxy/curing agent mixture. This process was repeated until the mold was completely
filled. After sealing the mold it was placed in an oven set at 100°C and rotated frequently
in order to minimize ribbon sinking and clumping. Afier approximately 30 minutes the
gel point of the polymer mixture had been reached and the ribbons were, in a sense,
“frozen” in position; thus, no further turning of the mold was required. The sample was
then allowed to fully cure at this temperature for 24 hours. The cast composite sheets

were cut, with a diamond saw, into rectangular pieces in order to make compact tension

 

 

82

K 7 cm )i
Glass \ 18.25 cm

 

 

Silicone

Rubber \ JD. 476 cm

 

 

 

 

Glass

 

 

 

 

J-

Figure 7-1. Schematic diagram of the mold set-up for preparation of composite
plaques. .

83

specimens (Figure 7-2). A notch was cut, in line with at least one metallic glass ribbon,
into the edge of the rectangular pieces. Next, two holes were drilled on either side of the
notch. The holes were loading points where upon application of the load the crack would
propagate in the direction of the ribbons. A tabletop Instron, model 43 02, with a
crosshead speed set at its minimum value of 0.5 min/min was used to test the specimens.

Polarized sheets placed in front and back of the sample and a light source positioned
behind the Instron were used in order to obtain the isochromatic stress birefiingence
patterns of these specimens. In addition, a video camera secured to a tripod was
positioned in front of the Instron to videotape the compact tension tests. The fractured
specimens were then gold coated so that specimen charging would not occur when

viewing the fracture surfaces with a scanning electron microscope (SEM).

84

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7r 1/ If 1
3 El
1:]
3.5
111. 1:1 [] [:1
AL 4' ”J \
l< >1\o.1 .111
16.5 cm
(A)
Ribbon
ll
0 Pre-crack
Loading
Holes
(3)

Figure 7-2. Schematic diagram of (A) cast composite sheet, and (B) compact
tension specimen.

85

7.3 Results and Discussion

Figures ‘7-3 through 7-5 presents a schematic diagram and the scanning electron
micrographs of the fracture surface of a thermoset polymer reinforced composite
specimen. One noticeable feature in all of the pictures is that the fi'acture occurred in the
matrix. The fact that the approaching crack did not propagate through the ribbons can be
attributed to the superior strength and good ductility of metallic glasses. Typically,
metallic glasses have ultimate tensile strengths on the order of 4000 MPa [3 5]. The
ductility of the metallic glass ribbons retards the crack front’s propagation, causing crack
deflection around the ribbon reinforcement. In the compact tension sample, from the pre-
crack upto the far end of the ribbon’s width the surface is smooth due to stable and
continuous crack growth. From the end of the metallic glass ribbon through the remaining
matrix, the surface is rough because the crack has propagated around the ribbon and does
not meet on the same plane. In addition, failure in the matrix is an indication of strong
interfacial adhesion between the epoxy and metallic glass ribbons. Weak bonding at the
interface would be revealed by different fi'acture surface features, due to crack
propagation through to the interface causing failure along this weak interface path.

Another feature indicating strong interfacial adhesion is observed in the
micrographs of Figures 7-6 through 7-7. These SEM micrographs show a magnified
section of ribbon which is well-bonded to the matrix. Near the ends of the ribbon, the
ribbon attempted to pull away from the matrix but the bonding between the polymer
matrix and metallic glass ribbon was strong enough to prevent complete separation of the
matrix fi'om the ribbon. The result of this strong interface is a fibril polymer structure

formed in those regions where the ribbon and matrix tried to separate.

i
i
. .
. ‘ '
l' t
1 ,
1 ,
1
J 1
v
I‘ I
l ' I
~ .
f -
1‘ 1 '3 ‘
p
A '1
- e
,
a- ‘
1.
A. . ' .‘ .
I '1 1
. ,l . , .~ . l
I
. ‘ > ‘
.1 l ‘ ’ l I
' '1 1 ‘
“ l .I
' I
e l '
‘ .' ' '1 .J 1
I ' ’
v 1 ‘ ‘ ' . '
a o i. S ‘1 ‘ ‘ '
. a
. - 1. ._.r .3 _ . N" '-
P a
a» ‘ '
. ' .
. ‘ ‘ ‘ '

86

 

 

 

 

 

 

Smooth
Notch surface (See Figure 5'4)
A I — ?
Pre-crack Rough
Ribbon Surface

Figure 7-3. Fracture surface of compact tension composite sample.

I - '
I / '
’

., r
1 . .

I ‘
,._-- -../,—-_-/ — -—

 

(A) (13)

Figure 7-4. SEM micrograph of the fracture surface of metallic glass ribbon
reinforced epoxy

87

 

Figure 7-5. Fracture surface indicating failure in the epoxy matrix.

 

88

 

Figure 7-6. SEM micrograph indicating ribbon bonding to the matrix.

 

Figure 7-7. SEM micrograph of a ribbon bonded to the epoxy matrix.

CHAPTER 8 FEASIBILITY OF MAGNETICALLY ALIGNING
REINFORCEMENTS

8.1 Background

Polymer composites reinforced with metallic ribbons offer very attractive
properties. They increase the strength and stiffness over pure polymer structures. The
ribbons are the main load-bearing components and the polymer acts as a binder,
transferring the load to the ribbons. Ribbon reinforcements are normally distributed with
a random orientation. This is acceptable from a macroscopic viewpoint. However, it
would be beneficial to be able to orient the ribbons in such a way as to counteract the
forces which could possibly cause the failure of a particular component. For example, a
rivet is susceptible to shear failure. Therefore, reinforcement, via ribbon alignment, along
these critical shear directions could remarkably strengthen those weak areas. The
potential use of polymers reinforced with metallic glass ribbons in strengthening
applications lies in the ability to disperse the ribbons uniformly and in positioning the

ribbons along critical directions.

8.2 Sample Preparation

The same epoxy/curing agent system used for the compact tension specimens was
used as the matrix material for the aligned composite samples. A stoichiometric amount
of Jeffamine D400 was added to Epon 828 and thoroughly stirred. The mixture was
vacuum degassed at room temperature for 20 minutes in order to remove the air bubbles.
Metglas 26053-2 metallic glass ribbon was purchased from Allied Signal and was used as

the reinforcing material. The Metglas ribbon came on a spool fiom the manufacturer and

89

 

 

90

was carefully cut to size. The ribbons ranged in size from 6.99 to 14.30 mm. in length
and 7.01 to 10.40 mm. in width and all ribbons had an average thickness of 25m. A low
volume fraction of reinforcements (less than 5%) was used. The ribbons were cleaned
with acetone, dried and added to the liquid epoxy mixture. A magnetic field, generated by
permanent magnets, was applied across the mold by placing the magnets at various
positions along the perimeter of the mold. A heat lamp, and in some instances a hot plate,
was used to cure the sample. A thermocouple attached to the surface of the mold was

used to monitor the temperature during curing.

8.3 Results and Discussion

Two problems were encountered during sample fabrication. The first problem is
related to the viscosity of the epoxy mixture prior to curing. The viscosity of the matrix
has to be sufficiently high enough to overcome gravitational forces acting on the ribbons.
When the ribbons were added directly after mixing, the viscosity of the epoxy was very
low and the ribbons settled to the bottom of the mold. In addition, when the ribbons were
added alter the epoxy mixture had cured to a more viscous state, strong drag forces
restricted the ribbons motion.

Various curing cycles were utilized in an attempt to control the viscosity of the
epoxy matrix. For example, one cycle consisted of allowing the polymer to cure under a
heat lamp for approximately 1 hour. This preliminary cure was an attempt to increase the
viscosity of the polymer so that gravitational forces would not affect the positioning of
the ribbons. This allowed the ribbons to be placed into approximately ideal positions with

minor magnetic orientation adjustments still feasible.

91

To date, the best solution to the ribbon settling problem was to incorporate a two-
step curing cycle into the composite fabrication process. In this procedure, the mold was
partially filled with the uncured matrix material and then cured between 1 and 1.5 hours
at 100°C. After partially curing the sample, the remaining uncured mixture and metallic
ribbons were added on top of the partially cured resin. This allowed the ribbons to be
moved through the use of permanent magnets which were positioned around the mold.
Since the partially cured resin had become rubbery, the ribbons only sank approximately
half—way through the thickness of the sample.

Despite the problems experienced in this portion of the investigation it has been
determined that it is possible to orient metallic glass ribbons. Complex shaped ribbon
configurations may be possible provided that the viscosity of the matrix is sufficiently
high to prevent both magnetic attraction and settling of the ribbons. The viscosity needs
to be low enough, however, to be able to move the ribbons into the desired
configurations. Future investigations which need to be conducted include, producing a
controllable magnetic alignment field so as to allow for better control over the presiding
forces. In addition, analytical models of the viscous elastic drag forces and external
magnetic forces should be constructed in order to aid in the alignment. of metallic glass

ribbons into complex patterns.

CHAPTER 9 SUMMARY

Metallic glass ribbons have unique properties which makes them potential
reinforcements for polymer matrices. The ribbon’s width is an additional parameter
providing a greater surface area to volume ratio aiding in effective load transfer from the
matrix to the reinforcing phase. This will hold true if there is adequate adhesion or
bonding between the two phases. Compact tension tests were conducted in order to test
the bonding characteristics of metallic glass ribbons in an epoxy matrix. The results
achieved are:

D Compact tension tests revealed that an approaching crack did not propagate
through the ribbon due to the ribbons’ superior strength and good ductility.

D Failure in the matrix indicates strong interfacial bonding between the metallic
glass ribbons and the epoxy matrix.

a SEM photographs reveal a polymeric fibril structure, between the ribbon and
matrix, where the ribbon unsuccessfully tried to pull away from the matrix.

In addition to the unique mechanical properties metallic glasses possess, they also
have usefiil magnetic properties which has lead to interesting applications. The feasibility
of utilizing these magnetic properties was investigated. An external magnetic field was
used during composite fabrication so as to align the metallic glass reinforcements along
critical directions in order to enhance the physical and mechanical properties of the
polymer composites. The following results were achieved:

0 Metallic glass ribbons become magnetized when exposed to an external

magnetic field and as a result, ribbon clumping occurred.

92

 

93

u The ribbons settled to the bottom of the samme because the viscosity of the
epoxy was too low prior to curing. The viscosity of the matrix has to be high
enough to counteract the gravitational forces acting on the ribbons so as to

achieve the required distribution of reinforcement in an epoxy matrix.

LIST OF REFERENCES

10.

11.

12.

13.

14.

15.

LIST OF REFERENCES

Tant, MR and Wilkes, G.L. Polym. Eng. Sci, 21, 874, (1981).

Struik, L.C.E. ‘Physr'cal Aging in Amorphous Polymers and Other Materials.’
Elsevier, Amsterdam, (1978).

Matsuoka, S. ‘Relaxation Phenomena in Polymers.’ Hanser Publishers, New
York, (1992).

Parsons, W.F. and Pochan, I.M. Polymer, 25, 1690 (1984).

Miller, M.L. ‘Ihe Structure of Polymers. ’ Reinhold Book Corporation, New
York, (1966).

Priss, L.S. J.Polym.Sci. Part C, 53, 195 (1975).
Dusek, K. and Prins, W. Adv.Polym.Sci., 6, 1 (1969).
Zirnan, I.M. ‘Models of Disorder. ’ Cambridge University Press, New York, 1979.

Ferry, JD. ‘ Viscoelastic Properties of Polymers. ’ John Wiley and Sons, Inc., New
York, 1980.

Lee, A. and McKenna, G.B. Polymer, 29, 1812 (1988).

Wang, Y. and Lee, A. ‘Effect of Crosslink Density Distribution on Mechanical
Responses of Physically Aged Epoxy Glasses.’ MS Thesis, Michigan State
University, 1995.

Hodge, I.M. Science, 267, 1945 (1995).

Kovacs, A.J., Aklonis, J.J., Hutchinson, I.M. and Ramos, AR. J .Polym.Sci.Phys.
Ed, 17, 1097 (1979).

Kohlrausch, R. Pogg.Ann.Physik, 12, 393 (1847).

Williams, G. and Watts, D.C. Trans. Faraday Soc, 66, 80 (1970).

94

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

95

Diamant, Y., Marom, G. and Broutman, L]. J. Appl. Polym. Sci, 26, 3015
(1981)

Bellenger, V., Verdu, J. and Morel, E. J. Mater. Sci, 24, 63 (1989).

Wong, TC. and Broutman, L. J. Polym. Eng. Sci, 25, 521 (1985).

Carfagna, C., Apicella, A. and Nicolais, L. J.Appl.Polym.Sci., 27, 105 (1982).
Dusek, K., Bleha, M. and Lunak, S. J.Polym.Sci.Chem.Ed., 15, 2393 (1977).
Enns, JD. and Gillham, J.K. J. Appl. Polym.Sci., 28, 2831 (1983).

Wong, TC. and Broutman, L. J Polym.Eng.Sci., 25, 529 (1985).

Moy, P. and Karasz, F .E. Polym.Eng.Sci., 20, 315 (1980).

Zhou, J. and Lucas, J.P. ‘Water-Environment Effects on Anomalous Absorption
Behavior in Graphite-Epoxy Composite.’ PhD. Thesis, Michigan State
University, 1996.

Brewis, D.M., Comyn, J ., Shalash, R.J.A. and Tegg, J .L. Polymer, 21, 357
(1980)

Ghorbel, I. And Valentin, D. Polym.Comp., 14, 324 (1993).

Apicella, A. and Nicolais, L. Ind.Eng.Chem.Prod.Res.Dev., 20, 138 (1981).
Cohen, D.S. J. Polym. Sci. Phys. Ed, 21, 2057 (1983).

Adamson, M.J. J. Mater. Sci, 15, 1736 (1980).

Choi, J. and Lee, A ‘Hygrotherrnal Effects on the Epoxy Resins Influenced by
the Network Topology.’ MS Thesis, Michigan State University, 1996.

Shell, CH. and Springer, G. S. ‘Environmental Effects on Composite Materials.’
Ed. G. S. Springer, 15, (1981).

Jost, M. ‘Diffusion in Solids, Liquids, and Gases.’ Academic Press, 1960.

Bhattacharya, S.K. ‘Metal-Filled Polymers: Properties and Applications. ’ Marcel
Dekker, Inc., New York, 1986.

Vaidya, R.U., Norris, C. and Subramanian, K.N. J .Mater. Sci, 27, 4957 (1992).

35.

36.

37.

38.

39.

40.

41.

96

Blank-Bewersdorff, M., Koster, U. and Steinbrink, G. J.Mater.Sci.Lett., 8, 796
(1989)

Tio, T.K., Friedrich, K., Hornbogen, E., Koster, U. and Fels, A. J .Mater.Sci.Lett.,
3, 415 (1984).

Varin, RA. and Ruutopold, A. Polym.Comp., 12, 411 (1991).

Greer, A.L. Science, 267, 1947 (1995).

Vaidya, R.U. and Subramanian, KN. J.Am.Ceram.Soc., 73, 2962 (1990).
Vaidya, R.U. and Subramanian, K.N. J .Mater.Sci., 25, 3291 (1990).

Fels, A., Friedrich, K. and Hornbogen, E. J .Mater. Sci.Lett., 3, 569 (1984).